Frost Action in Concrete -...

Proceedings of the International RILEM Conference

Materials, Systems and Structures in Civil Engineering 2016Segment on

Frost Action in Concrete

Edited byMarianne Tange Hasholt, Katja Fridh and R. Doug Hooton

ProceedingsPRO 114

International RILEM Conference on

Materials, Systems and Structures in Civil Engineering 2016

segment on Frost Action in Concrete

II

International RILEM Conference on Materials, Systems and Structures in Civil Engineering Conference segment on Frost Action in Concrete

22-23 August 2016, Technical University of Denmark, Lyngby, Denmark

Published by RILEM Publications S.A.R.L. 4 avenue du Recteur Poincaré 75016 Paris - France Tel : + 33 1 42 24 64 46 Fax : + 33 9 70 29 51 20 http://www.rilem.net E-mail: [email protected]

2016 RILEM – Tous droits réservés. ISBN: 978-2-35158-182-7 e-ISBN : 978-2-35158-183-4

Printed by Praxis – Nyt Teknisk Forlag, Ny Vestergade 17, 1471 København K, Denmark Photo 1st cover page: Snowflake. Picture was taken by Kenneth G. Libbrecht using a specially

designed snowflake photomicroscope (SnowCrystals.com).

Publisher's note: this book has been produced from electronic files provided by the individual contributors. The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility or liability for any errors or omissions that may be made. All titles published by RILEM Publications are under copyright protection; said copyrights being the property of their respective holders. All Rights Reserved. No part of any book may be reproduced or transmitted in any form or by any means, graphic, electronic, or mechanical, including photocopying, recording, taping, or by any information storage or retrieval system, without the permission in writing from the publisher. RILEM, The International Union of Laboratories and Experts in Construction Materials, Systems and Structures, is a non profit-making, non-governmental technical association whose vocation is to contribute to progress in the construction sciences, techniques and industries, essentially by means of the communication it fosters between research and practice. RILEM’s activity therefore aims at developing the knowledge of properties of materials and performance of structures, at defining the means for their assessment in laboratory and service conditions and at unifying measurement and testing methods used with this objective. RILEM was founded in 1947, and has a membership of over 900 in some 70 countries. It forms an institutional framework for co-operation by experts to: optimise and harmonise test methods for measuring properties and performance of building

and civil engineering materials and structures under laboratory and service environments, prepare technical recommendations for testing methods, prepare state-of-the-art reports to identify further research needs, collaborate with national or international associations in realising these objectives.

RILEM members include the leading building research and testing laboratories around the world, industrial research, manufacturing and contracting interests, as well as a significant number of individual members from industry and universities. RILEM’s focus is on construction materials and their use in building and civil engineering structures, covering all phases of the building process from manufacture to use and recycling of materials. RILEM meets these objectives through the work of its technical committees. Symposia, workshops and seminars are organised to facilitate the exchange of information and dissemination of knowledge. RILEM’s primary output consists of technical recommendations. RILEM also publishes the journal Materials and Structures which provides a further avenue for reporting the work of its committees. Many other publications, in the form of reports, monographs, symposia and workshop proceedings are produced.

III



International RILEM Conference on


Segment on

Frost Action in Concrete

Lyngby, Denmark

August 22-23, 2016

Edited by R. Doug Hooton, Katja Fridh and Marianne Tange Hasholt

RILEM Publications S.A.R.L.

Marianne Tange Hasholt, Katja Fridh and R. Doug Hooton

IV



Sponsors:

Knud Højgaards Foundation Larsen & Nielsen foundation Ingeborg og Leo Dannis Legat for Videnskabelig Forskning

Hosted by:

Technical University of Denmark

Department of Civil Engineering

International Organization by: Marianne Tange Hasholt (Technical University of Denmark) Katja Fridh (Lunds University, Sweden)

Scientific Committee: R. Doug Hooton (University of Toronto, Canada) Katja Fridh (Lunds University, Sweden) Marianne Tange Hasholt (Technical University of Denmark)

V



RILEM Publications The following list presents the latest offer of RILEM Publications, sorted by series. Each publication is available in printed version and/or in online version.

RILEM PROCEEDINGS (PRO) PRO 94 (online version): HPFRCC-7 - 7th RILEM conference on High performance fiber reinforced cement composites, e-ISBN: 978-2-35158-146-9, Eds. H.W. Reinhardt, G.J. Parra-Montesinos, H. Garrecht PRO 95: International RILEM Conference on Application of superabsorbent polymers and other new admixtures in concrete construction, ISBN: 978-2-35158-147-6; e-ISBN: 978-2-35158-148-3, Eds. Viktor Mechtcherine, Christof Schroefl PRO 96 (online version): XIII DBMC: XIII International Conference on Durability of Building Materials and Components, e-ISBN: 978-2-35158-149-0, Eds. M. Quattrone, V.M. John PRO 97: SHCC3 – 3rd International RILEM Conference on Strain Hardening Cementitious Composites, ISBN: 978-2-35158-150-6; e-ISBN: 978-2-35158-151-3, Eds. E. Schlangen, M.G. Sierra Beltran, M. Lukovic, G. Ye PRO 98: FERRO-11 – 11th International Symposium on Ferrocement and 3rd ICTRC - International Conference on Textile Reinforced Concrete, ISBN: 978-2-35158-152-0; e-ISBN: 978-2-35158-153-7, Ed. W. Brameshuber PRO 99 (online version): ICBBM 2015 - 1st International Conference on Bio-Based Building Materials, e-ISBN: 978-2-35158-154-4, Eds. S. Amziane, M. Sonebi PRO 100: SCC16 - RILEM Self-Consolidating Concrete Conference, ISBN: 978-2-35158-156-8; e-ISBN: 978-2-35158-157-5 PRO 101 (online version): III Progress of Recycling in the Built Environment, e-ISBN: 978-2-35158-158-2, Eds I. Martins, C. Ulsen and S. C. Angulo PRO 102 (online version): RILEM Conference on Microorganisms-Cementitious Materials Interactions, e-ISBN: 978-2-35158-160-5, Eds. Alexandra Bertron, Henk Jonkers, Virginie Wiktor

In relation to the International RILEM Conference on Materials, Systems and Structures in Civil Engineering, MSSCE 2016 which the present proceedings belongs to, the following RILEM proceedings will be issued:

PRO 108: Innovation of Teaching in Materials and Structures PRO 109 (two volumes): Service life of Cement-Based Materials and Structures PRO 110: Historical Masonry PRO 111: Electrochemistry in Civil Engineering PRO 112: Moisture in Materials and Structures PRO 113: Concrete with Supplementary Cementitious materials PRO 114: Frost Action in Concrete PRO 155: Fresh Concrete RILEM REPORTS (REP) Report 45: Repair Mortars for Historic Masonry - State-of-the-Art Report of RILEM Technical Committee TC 203-RHM, e-ISBN: 978-2-35158-163-6, Ed. Paul Maurenbrecher and Caspar Groot

VII



Contents Page

Preface R. Doug Hooton, Katja Fridh and Marianne Tange Hasholt

IX

Welcome Ole Mejlhede Jensen

X

1. The use of image analysis to quantify the orientation of cracks in concrete Einar N. Andreassen, Andreas B. Elbrønd and Marianne T. Hasholt

1

2. Non-destructive evaluation of concrete subjected to freeze-thaw cycles Sofía Aparicio, Javier Ranz, Margarita G. Hernández and José Javier Anaya Velayos

11

3. Frost resistance of concrete – Experience from long term field exposure Dimitrios Boubitsas, Peter Utgenannt, Luping Tang and Elisabeth Helsing

21

4. The influence of the freeze-thaw loading cycle on the ingress of chlorides in concrete Miguel Ferreira, Markku Leivo, Hannele Kuosa and David Lange

31

5. Frost damage of concrete subject to confinement Marianne Tange Hasholt

41

6. The salt-frost resistance of concrete with supplementary cementitious materials (SCM) Elisabeth Helsing and Peter Utgenannt

51

7. Foam index measurements on mixes of air entraining agents, super plasticizers and fly ash-cement-filler blends Stefan Jacobsen, Henrik Nordahl-Pedersen, Hawar Omer Rasol, Øyvind O. Lødemel, Lori Tunstall and George W. Scherer

61

8. Freezing induced stresses in concrete-steel composite beams and effect of air voids Stefan Jacobsen and George W. Scherer

71

9. Correlation between characteristic distances of air voids as point processes and conventional spacing factors in mortars Hidefumi Kotou, Takuma Murotani and Shin-Ichi Igarashi

81

10. The influence of carbonation and age on salt frost scaling of concrete with mineral additions Ingemar Löfgren, Oskar Esping and Anders Lindvall

91

11. Modeling freezing of cementitious materials by considering process kinetics Francesco Pesavento and Dariusz Gawin

101

12. Experimental studies on frost-induced deterioration of concrete in Swedish hydroelectric structures Martin Rosenqvist, Katja Fridh and Manouchehr Hassanzadeh

111

13. The influence of air void characteristics on freeze-thaw-salt resistance of pavement concretes Aljoša Šajna and Lado Bras

121

, Takuma Murotani and Shin-Ichi Igarashi

Marianne Tange Hasholt, Katja Fridh and R. Doug Hooton

VIII



14. Identification of optimal condition for the de-icing salt scaling resistance of concrete Samindi Samarakoon, Samdar Kakay, Pål Lieske Tefre, Mats Buøen and Vikrant Kaushal

131

15. Towards an adequate deicing salt scaling resistance of high-volume fly ash (HVFA) concrete and concrete with superabsorbent polymers (SAPs) Didier Snoeck, Philip Van den Heede and Nele De Belie

141

16. Freeze-Thaw-Attack on concrete structures – laboratory tests, monitoring, practical experience Frank Spörel

151

17. Methodology to analyse the salt frost scaling mechanism(s) in concrete with different binders Martin Strand and Katja Fridh

161

18. Mitigation of deicer damage in concrete pavements caused by calcium oxychloride formation – Use of ground lightweight aggregates Prannoy Suraneni, Naomi Salgado, Hunter Carolan, Chang Li, Vahid Jafari Azad, O. Burkan Isgor, Jason H. Ideker and Jason Weiss

171

19. Deicer-salt scaling of concrete containing fly ash Michael Thomas and Huang Yi

181

20. Linking surfactant molecular structure to mortar frost protection Lori E. Tunstall, George W. Scherer and Robert K. Prud'Homme

191

21. Percolation in cementitious materials under freeze-thaw cycles investigated by means of electrical resistivity Zhendi Wang, Ling Wang and Yan Yao

201

22. Application of air entrained concrete in tollways constructions in Liaoning Province of China Wencui Yang, Xiaoping Cai, Yong Ge and Jie Yuan

211

23. Influence of ductility and microcracking on the frost durability of cementitious composites with high volumes of fly ash

221

24. Water penetration into frost damaged concrete Peng Zhang, Yuan Cong, Wanyu Zhao, Wenchao Geng, Zhengzheng Dai and Tiejun Zhao

231

Author Index 239

IX



Preface The conference “Materials, Systems and Structures in Civil Engineering – MSSCE 2016” is part of the RILEM week 2016, which consists of a series of parallel and consecutive conference and doctoral course segments on different topics as well as technical and administrative meetings in several scientific organizations. The event is hosted by the Department of Civil Engineering at the Technical University of Denmark and the Danish Technological Institute and it is held at the Lyngby campus of the Technical University of Denmark 15-29 August 2016. This volume contains the proceedings of the MSSCE 2016 conference segment on “Frost action in concrete”. Despite research in this field has been ongoing since the 1930’es, the mechanism(s) leading to frost damage is not fully understood. Therefore, there is still a need for both basic research and practical solutions to the challenges encountered in the field. When we first decided to have a conference segment on “Frost action in concrete”, our aim was to make an international forum for presentations of recent research and discussions. With 24 papers from 14 countries and 3 different continents, and from both industry and academia, we think we have reached this goal. Within the overall theme “Frost action in concrete”, the contributions deal with many different topics, for example: the relation between mix design and frost resistance, modelling of frost action, combined action when concrete is exposed to freeze/thaw load together with other types of load, air void analysis, novel non-destructive test methods, and experience gained from monitoring of structures as well as from field exposure sites. All contributions have been peer reviewed. Some of the papers in these proceedings origin from the work in two COST actions: TU 1404 Service Life of Cement-based Materials and Structures and TU 1301 NORM for Building Materials. The papers from COST TU 1404 are spread across several MSSCE 2016 conference segments and it is planned that a separate set of electronic proceedings for this COST action will be published after the conference. R. Doug Hooton Katja Fridh Marianne Tange Hasholt August 2016, Lyngby, Denmark

Marianne Tange Hasholt Katja Fridh R. Doug Hooton

X



Welcome Were you aware that a part of your daily language is likely to be in Danish? A thousand years ago the Danish word “Vindue” came along with the Vikings to England. Several hundred years later it reached North America, and from there – just two to three decades ago – almost every person in the world learned to understand and pronounce this word: “Windows”, which etymologically means “an eye to the wind”. As a child your career as construction professional may have started with LEGO, and before you went to bed, your mother told you the unforgettable fairytales of H.C.Andersen. You may have grown up with the delicious taste of Lurpak butter on your bread, and though you might find it strange that “God plays dice with the Universe”, hopefully your school teacher told you that on this topic Einstein was flat out wrong and Niels Bohr was right. Right now you may prefer to be sitting in the sun with a chilled Carlsberg beer in your hand, enjoying the iconic view of the Sydney Opera House. All of it is Danish made, and many things around you at home, if not made in Denmark, were probably brought to you by Maersk, the world’s largest shipping company, the modern Danish Viking fleet. Though Denmark is one of the world’s smallest countries, yet it stands – along with your country – among the greatest. On top of a thousand years of outreach from Denmark, your visit to the Danes is most welcome. On your approach to Copenhagen airport you had a view to wind turbines harvesting green energy, you saw record breaking bridges, and perhaps you got a glimpse of the island Ven where the nobleman Tycho Brahe literally speaking changed our view of the world through perfection of astronomical observations with his naked eye. In Copenhagen you may appreciate a walk in the fairytale amusement park TIVOLI, and in the Copenhagen harbour you may have a rendezvous with a Little Mermaid. Of all things in Denmark you will surely enjoy the conference and doctoral courses Materials, Systems and Structures in Civil Engineering, MSSCE 2016 which are held in conjunction with the 70th annual RILEM week. On this occasion RILEM celebrates its 70 years birthday and thus maintains generations of experience. However, new activities and the in-built diversity keep RILEM fresh and dynamic like a teenager. The event takes place in northern Copenhagen, Lyngby, at the campus of the Technical University of Denmark, 15-29 August 2016. MSSCE 2016 aims at extending the borders of the RILEM week by including doctoral courses, by involving a palette of RILEM topics in the conference and workshop activities, and by collaborating with other scientific organizations. The insight and outlook provided by this event make it RILEM’s technical and educational activity window. It is a pleasure to share with you what is unique to RILEM and Denmark! Ole Mejlhede Jensen, Technical University of Denmark Honorary president of RILEM 2016, Chairman of MSSCE 2016

1



THE USE OF IMAGE ANALYSIS TO QUANTIFY THE ORIENTATION OF CRACKS IN CONCRETE Einar N. Andreassen (1), Andreas B. Elbrønd (1) and Marianne T. Hasholt (1) (1) Technical University of Denmark, Lyngby, Denmark Abstract Cracks formed in concrete due to frost action (or other expansive reactions) can lead to further damage e.g. because they increase moisture transport. The extent of the consequential damage in concrete is likely influenced by the orientation of the initial cracks. Traditional quantification of the crack orientation is a time consuming manual process. In this paper, a method using automatic image analysis is proposed. The method is based on using image gradients to detect cracks and their orientation. The method produces results that concur with visual observation and manual counting in addition to being substantially quicker. 1. Introduction Ice formation during frost action (or other expansive reactions) may result in crack formation in concrete. As a result, the concrete will have reduced mechanical strength and the cracks will lead to increased moisture and chloride transport [1]. Therefore, it is likely that the extent of further damage occurring in the concrete is related to the orientation of the cracks. Presently, the common way to analyse the crack orientation in concrete is by using the human eye, but it is difficult and time-consuming to precisely measure the orientation. It is therefore important to develop a method that can quickly and reliably give an assessment of the crack orientations. An ideal method would be easy to use and able to produce results independent of the operator. The results should be in agreement with what can be achieved by manual counting without requiring extensive calculations. Lastly, it should require a minimum of preparation of the test specimen and the image. In this paper, an investigation was undertaken to determine if image analysis based on a so-called gradient method can fulfil these requirements.

2



2. Theory A digital image is a matrix of pixel values representing the colour intensity, or in the case of a greyscale image, the brightness. This pixel value ranges from 0 to 255. Edges between areas with different colours can be identified, because there is a relatively large difference in pixel values between neighbouring pixels in the border region. This results in the largest gradients to occur orthogonal to the border. For each pixel in the image, the direction and magnitude of the largest gradient is calculated. The magnitude of the gradient is used to decide which pixels are parts of a border and the direction determines the orientation of the border. The gradients are determined through a so-called correlation process. 2.1 Correlation of image To determine the magnitude and direction of the largest gradient between each pair of adjacent pixels, it is first necessary to determine the x- and y-components of the gradients. Furthermore noise in the image must be reduced to get the optimal results from the gradient analysis. This is done via a so-called correlation process through the application of so-called kernels. Kernels are square matrices applied to the input image, producing two new matrices. In this case, two matrices Gx- and Gy-matrix are produced. Each entry in the new Gx- and Gy-matrix is a combination between the weights of the kernel applied, and the corresponding pixels in the input image. The entry in the G-matrices is then the sum of the weights in the kernel times the affected pixel values in the input image. This can be explained mathematically as shown in equation (1): ( , ) = ( , ) ( + , + ) (1)

Where ( , ) is the value stored in the Gx-matrix that contains the x-component of the gradients, R is the radius of the kernel, is the applied kernel and is the pixel value of the input picture. and are pixel coordinates in the input picture. The same is the case for the Gy-matrix. This process is repeated for all the pixels in the input image.

Figure 1: Coefficients in the Prewitt and Sobel kernels after [2] The two kernels shown in figure 1 are often used kernels for edge detection [2]. Both have the radius R = 1. There are many different kernels, but these two are highlighted due to their simplicity and ability to produce satisfactory results. The Sobel and Prewitt operators are

3



quite similar. The main difference between the two is that the Sobel kernel weighs the centre pixel more than the rest, and the Prewitt kernel weighs all the pixels equally [2]. The gradient operator implemented for the method described in this article, is the Prewitt operator due to its efficiency and simplicity [3]. The Prewitt kernel is furthermore well suited to eliminate low-frequency noise [4]. The difference between the kernels is demonstrated in [5]. The Prewitt method does not register edges, where there is a small difference in pixel grey level, in contrast to the Sobel method. Since the edges of the cracks (see section 3 Method) analysed in this paper are clearly defined, it is assessed that the Prewitt method is better suited for this task. In addition, the Sobel method sharpens some transitional phases that can distort the results. 2.1 Gradient calculation The magnitude and direction of the steepest gradient is determined for each pixel. The kernel operators result in two values, one in the x-direction and one in the y-direction. The magnitude of the gradients is the difference in the pixel grey level and the direction of the gradients is the combination of the x- and y-gradients. The orientation of the cracks will then be the orthogonal direction of the gradient. The magnitude and direction of the gradient can be expressed as in equations (2) and (3) respectively, with ( , ) being the pixel position in the gradient matrices: ( , ) = ( , ) + ( , ) (2)

( , ) = arctan ( , )( , ) (3)

3. Method Cracks in concrete can be identified and quantified by image analysis, if the cracks have clearly defined edges. To get optimal contrast between the cracks and their surroundings, the image analysis procedure is applied on plane sections of fluorescent epoxy-impregnated concrete. The surface is photographed under UV-light, causing the epoxy-filled cracks to light up with a green hue. The rest of the concrete remains black or dark blue. The image is photographed with a high-resolution camera and stored in a lossless file format (e.g PNG) for an optimal result. The image file is then cropped to remove the image background and loaded into the image analysis program. The program was developed in Matlab, because it is easy to use and well suited for matrix calculus. The program is described in the following steps. Because the gradient method only registers difference in brightness, it is not possible to apply it to an RGB image, which constitutes of a red (R), green (G) and blue (B) layer, and it is therefore necessary to convert the image to a greyscale image, which contains one layer.

4



Step 1: Convert image to greyscale. Rather than using the default method for conversion of RGB to greyscale, it turned out to be more effective to simply use the green layer of the RGB image, because the cracks in the image primarily are green. This has the added benefit of removing the risk of detecting a false edge from transition between aggregate and cement paste. Step 2: Apply horizontal and vertical kernels to the greyscale image. This yields two matrices, Gx and Gy, containing the and components of the steepest gradient for each pixel in the image. Step 3: Determine gradient magnitude and direction. This is done for each pixel in the image using Gx and Gy as described earlier. It is noteworthy that the gradients will point towards the cracks, since this is where the increase in pixel values is. Step 4: Select threshold for the analysis. This is done by analysing selected regions with distinct geometric features, e.g. an approximately vertical or horizontal crack. This makes it easy to establish a ground truth. The threshold is then set by comparing the control image for different threshold values with the original image. Step 5: Determine crack edges. This is done based on the magnitude of the gradient. If the magnitude is lower than the specified threshold, the pixel is not considered as being part of a crack edge, and therefore ignored. Note that the gradient between pixels located in the same crack is small, so only the edges of the cracks are above the threshold. Step 6: Quality control. To ensure that the threshold has a value that produces a satisfactory result, a control image is produced. This shows the pixels where the gradient is above the threshold and should show the same cracks as can be identified in the original photo. Step 7: Display the results regarding the direction of the gradients. Orientations (angles) are rounded to the nearest integer. Finally, the direction of the cracks are calculated as the direction orthogonal to the gradient and converted to a range between 1 and 180 degrees 4. Validation of method Validation has been carried out on images of concrete damaged by alkali-silica reactions (ASR), as they show very distinct crack orientations. The analysis presented in figure 2-3 is conducted on sections of figure 4a. These sections contain simple geometrical figures and it is therefore easy to evaluate the accuracy of the applied method. First, the influences of threshold values are analysed, see figure 2. The best threshold for this image is found to be = 150. Next, the method's ability to detect different orientations in the same image is tested. Figure 3 shows the results from the image analysis when run on two images with distinct geometrical features. Lastly, the analysis is conducted on two full images, ASR1 and ASR2, and compared to results obtained from manual counting. The manual counting is done by dividing the specimen into 10x10 mm sections and then identifying each crack in the section using an optical microscope. The direction of the identified crack is then calculated using a computer. [6] In figure 2-4, the horizontal line = = 0.556 in the histograms represents an image with evenly distributed orientations.

5



(a) Original image displaying horizontal crack

(b) Control image for threshold = 50 (c) Crack orientation histogram for threshold = 50

(d) Control image for threshold = 150 (e) Crack orientation for threshold = 150

(f) Control image for threshold = 250 (g) Crack orientation for threshold = 250 Figure 2: Test of different threshold values

6



(a) Image with horizontal and approximately 45° crack (b) Image with a vertical and two horizontal cracks.

(c) Control image (d) Control image

(e) Crack orientation histogram (f) Crack orientation histogram Figure 3: Test of ability to recognize distinct geometrical features

7



(a) ASR1 original image. The image is 100 x 200 mm and 1835 x 3725 pixels

(b) ASR2 original image. The image is 100 x 200 mm and 1777 x 3713 pixels

(c) ASR1 control image (d) ASR2 control image

(e) Crack orientation histogram for ASR1 (f) Crack orientation histogram for ASR2

(g) Manual count of ASR1 orientations (g) Manual count of ASR2 orientations Figure 4: Samples analysed using gradient method and manual counting

8



5. Discussion The method presented in this paper is evaluated on the criteria presented in the introduction. The method requires a thorough specimen preparation as it is necessary to embed it in fluorescent epoxy and polish the surface that is going to be analysed. The described method is very dependent on that cracks have a significantly different colour than the rest of the concrete, and therefore it cannot be used on samples without preparation. It is clear that the usage of the green value to determine the gradient would not work for specimens where the cracks are not green. The photographing of the specimens is uncomplicated; it only requires a dark room, a UV-lamp and a camera. The resolution of the camera must be high enough to ensure that the smallest crack has a width of at least two pixels and preferably higher. If the crack only has a width of 1 pixel, a gradient cannot be properly identified. It is important, however, that the specimen surface is parallel in regards to the camera lens to prevent distortion in the image. With the right setup, a large number of specimens can be photographed quickly. The only preparation needed for the images is the manual removal of background. While a lossless file format should be used, a lossy format (such as JPEG) can still produce adequate results. The need to choose a threshold is the weakest part of the method and the only instance where the operator can influence the results, because the threshold is selected manually based on the control image. This is problematic as the method is somewhat sensitive to different thresholds, as can be seen in figure 2. A lower threshold will lead to smoothing of the results, as more noise is included in the results, whereas a higher threshold will omit parts of the crack. This becomes even more problematic as different light conditions and even the composition of the specimen can influence the magnitudes of the gradients, making it difficult to determine a universal threshold. The method is able to identify cracks with a clear orientation as seen in figure 3. As can be seen in the figure, the Prewitt kernel identifies the edge of the crack but not the transition between the aggregate and the cement paste. When compared to manual counting (see figure 4), the method produces similar results. The difference in the results is because the gradient method registers the orientation of every pixel that is part of a crack (and unavoidably some noise), while the manual count does not register small changes in crack orientation, such as a deflection by aggregate particle or very small cracks. This will cause the results from the manual counting to be grouped more closely around dominant crack orientations. While both methods produce similar results, the gradient method is considerably faster. An image with a size of 1835x3725 pixels or 200x100 mm takes less than 1 minute to analyse on a standard laptop whereas it can take an experienced operator 2-3 hours to do a manual count. The fast analysis for the gradient method derives from the fact that it does not require any detailed calculations. For each pixel in the image, only a few simple calculations are performed. In addition to being substantially faster, the gradient method can also be used to analyse specimens with a dense crack pattern consisting of thin, short cracks, something that is very difficult to do manually. Examples of such crack patterns can for instance be found when analysing specimens heavily damaged by frost action, as shown in figure 5.

9



Figure 5: Specimen subjected to freeze/thaw cycles [7] 6. Conclusions The developed method is able to produce results that are in accordance with visual evaluation and manual counting. The method has a substantial speed advantage in comparison to the manual count, as it takes less than one minute to perform the analysis on a regular laptop, in comparison to the 2-3 hours it takes an experienced operator to do a manual count. As discussed earlier, the method has its weakness in relation to the manner how the threshold is determined. The manual configuration is strongly dependent on the operator, whose experience and insight in the method will decide, how fast the process is. Acknowledgements The authors would like to thank Ricardo Barbosa, Ph.D. student at DTU Department of Civil Engineering for providing the samples used for validation and for sharing his results of manual counting (figure 4g). The authors would also like to thank Amanda McNair, MSc student at DTU Department of Civil Engineering for providing results of manual counting (figure 4h). A special thanks to Rasmus Reinhold Paulsen, DTU Compute, for valuable discussion and for suggesting the use of image gradients to determine crack orientation. References [1] Mehta, P.K. and Monteiro, P.J.M., Concrete - Microstructure, Properties, and Materials,

Fourth Edition, McGraw-Hill Education, 2014 [2] Paulsen, R.R. and Moeslund, T.B., Introduction to Medical Image Analysis, DTU

Compute 2014 [3] Seif, A. et al, A Hardware Architecture of Prewitt Edge Detection, IEEE Conference on

Sustainable Utilization and Development in Engineering and Technology, Malaysia (2010), pp. 5686999

10



[4] Dong, W. and Shisheng, Z, Color Image Recognition Method Based on the Prewitt Operator, International Conference on Computer Science and Software Engineering, China (2008), pp. 170-173

[5] Pereira, O. et al, Edge Detection based on Kernel Density Estimation, Cornell University Library, 2015

[6] Powers, L.J. and Schrimer, F.T., Quantification of ASR in Concrete: An Introduction to the Damage-Rating Index Method, 29th International Conference on Cement Microscopy (2007), pp. 345-353

[7] Elbrønd, A.B. and Andreassen, E.N., Interplay between the Development of Inner and Outer Frost Damage of Concrete, Bach Thesis, Technical University of Denmark, 2015

11



NON-DESTRUCTIVE EVALUATION OF CONCRETE SUBJECTED TO FREEZE-THAW CYCLES Sofía Aparicio (1), Javier Ranz (1), Margarita G. Hernández (1), José Javier Anaya Velayos (1) (1) Instituto de Tecnologías Físicas y de la Información “Leonardo Torres Quevedo”, ITEFI

(CSIC), Madrid, Spain Abstract The concrete infrastructure requires recovering the maximum information about the quality and state of the structural and constitutive materials. The evaluation of cement-based materials by non-destructive is still a necessary requirement. The use of embedded sensors and wireless sensor networks could be an attractive solution. This paper reports a study of the behaviour of concrete during the degradation process by freeze-thaw cycles. The temperature, humidity, and strain gauge sensors were embedded in concrete during its manufacturing to monitor the deformations produced in this process, while ultrasonic sensors were adhered to the outer parallel surfaces of the specimens after the curing process. The temperature, relative humidity, and ultrasonic signals were registered by wireless sensor networks. As a result, a diagnostic of the sample status during this process was obtained. Two specific mixtures were made and monitored to investigate the ordinary concrete mix proposed during the Cost Action TU1404. 1. Introduction Concrete structures located in cold climates suffer damage by freeze-thaw (F-T) cycles during their life time. The freezing and melting of water with deicing salt in porous structure causes serious damage and requires large investments in the repair and/or replacement of such structures. This deterioration process has widely been studied and, as a consequence, different theories and standards have been proposed to evaluate the resistance of concrete exposed to F-T cycles, using accelerated tests. Among the standards developed to evaluate the resistance of concrete subjected to accelerated F-T cycles are: UNE CENT/TS 12390-9 [1], ASTM C666/C666M-03 [2], prENV-9 [3], JIS A 1148-2001 [4], and Rilem TC 176-IDC [5,6]. However, these standards differ in the testing method used, and the methodology employed for damage evaluation. In addition, none of these standards has been designed to perform measurements as a real-time continuous

12



monitoring process. The non-destructive testing methods commonly used to assess damage in concrete are the ultrasonic velocity [2,7] and the fundamental transverse frequencies measurements [3,5,6]. This paper deals with the study of the concrete behaviour during the degradation process by accelerated F-T cycles. Two different mixtures were compared varying the water/cement ratio and taking as a reference the ordinary concrete mix used during the Round Robin Test of the European Cost Action TU1404 [8]. The deformation was recorded using strain gauges; this system provided satisfactory results in previous studies [9-10]. The temperature (T), humidity (RH) and strain gauge sensors were embedded in the concrete samples during its manufacturing, while the ultrasonic sensors were adhered to the outer parallel surfaces of the specimens after the curing process. The embedded sensors, ultrasonic transducers and the multisensorial system for monitoring the specimens, were designed by our research group [11]. The temperature, relative humidity and ultrasonic signals were recorded by a wireless sensor network. One advantage of this monitoring system is the large number of measurements obtained along the F-T cycles with no need to remove the specimens from the climatic chamber. The quality of the specimens was evaluated by automated ultrasonic inspections that provided maps of attenuation before and after the F-T cycles [12, 13]. 2. Experimental Procedure A description of the materials and the experimental protocol used in this work is presented. 2.1 Materials Two different mixtures, A and B, were prepared using the materials supplied by the Cost Action TU1404 [8] and varying the w/c ratio. The proportions of the mixtures are shown in Tab. 1. Both mixtures are analogue to the one used in the context of the Vercors project [14], which consists in an experimental mock-up of a reactor containment-building at 1/3 scale which is being built at Renardières near Paris by Électricité de France S.A. (EDF). Mix A presents a concrete mix with a water (effective)-to-cement ratio of 0.52, high-strength Portland cement, and addition of chemical admixture in the form of a plasticizer. For mix B the water (effective)-to-cement ratio was increased to 0.57. Table 1: Mix proportions and characteristics of the concrete specimens. Basic Material Type of the material Mix A and B

(kg/m3) Cement CEM I 52,5 N CE CP2 NF Gaurain 320

Sand 0-4 mm, REC GSM LGP1 (13 % of CaO and 72 % of SiO2) 837

Gravel

4-11mm, R GSM LGP1 (rounded, containing silicate and limestone) 457

8-16 mm, R Balloy (rounded, containing silicate and limestone) 563

Admixtures Plasticizer SIKAPLAST Techno 80 (water content 80%) 2.4

13



The materials were conditioned and the specimens were manufactured in accordance with the instructions of the Cost Action TU1404. For each mixture, three cubic molds of 150x150x150 mm size were used following the standard EN 12390-1:2012 [7]. Specimens were prepared and cured following the procedure described in standard EN 12390-2:2009 [15]. After 28 days of curing the F-T experiment was initiated. For each mixture, one specimen was kept under water as a reference specimen, and the other two specimens were monitored, one of them with an embedded strain gauge. 2.2 Freeze-Thaw Cycles The applied F-T cycle is shown in Fig. 1, the test carried out is a non-standardized test method. The specimens were subjected to repeated F-T cycles of 12 hours of duration, up to 28 cycles in total.

Figure 1: Freeze-thaw cycle. 2.3 Monitoring of Freeze-Thaw Cycles The concrete resistance to frost was evaluated after monitoring the temperature, relative humidity, deformations, and ultrasonic parameters along all F-T cycles. The corresponding sensors were embedded in the concrete samples, while the ultrasonic transducers were adhered to the outer side surface using a mechanical system, designed and fabricated by our group, to overcome decoupling between transducers and specimen, see Fig. 2. Therefore, a large number of measurements along the F-T cycles were obtained in-situ in the climatic chamber. These measurements were made in transmission mode with longitudinal wave transducers of 54 kHz frequency.

14



Figure 2: Specimen with the ultrasonic and temperature sensors inside the climatic chamber. Commercial sensors, SHT15 from Sensirion Company, were used to measure T/RH. Since it has been proven that the T/RH values do not vary significantly when embedding the sensors inside the specimens, they were placed on top of the concrete specimens using an insulated material, see Fig. 2. Additionally, the T/RH inside the climatic chamber was monitored. In one specimen, both parameters were measured using a wireless network system, WilTempUS, developed by our research group [16]. This system allows the simultaneous acquisition of two channels for T/RH and generates and receives the A-scan signals. The ultrasonic monitoring in the other specimen was performed using a commercial Pundit system, see Fig. 3.

Figure 3: WilTempUS and Pundit systems employed to monitor the ultrasonic signals. The ultrasonic velocity (V) was determined by the travelling time of the ultrasonic pulses in the specimen (t), the thickness (d = 150 mm) considered constant along all cycles, and the

15



reference travelling time (t0), in order to calibrate the ultrasonic acquisition system using a reference aluminium specimen:

0

dVt t

(1)

The travel time, t, from the A-scan signals was calculated using a threshold algorithm [17]. A PMFL-50 strain gauge was used with an integrated temperature sensor to measure internal deformations of the material during the F-T cycles. The strain gauge was connected to a data acquisition system based on Labview to recording their values. These gauges were embedded in the middle of the cube mold using a transparent wire in one specimen of each mixture, see Fig. 4.

Figure 4: Cube mold with a strain gauge before the filling. 2.3 Ultrasonic Automated Inspections Automated ultrasonic inspections, providing amplitude maps of all specimens, were made before and after the F-T cycles to evaluate their quality. The maps obtained after the F-T cycles serve as “diagnostic” of the damage in the specimens. Two types of inspections were conducted simultaneously, in through-transmission and pulse-echo mode, to provide amplitude measurements [13]. The specimens were scanned on two faces: width and length, with a standard automatic system of three Cartesian axes, see Fig. 5. The specimens were aligned at the bottom of the tank and two ultrasonic transducers scanned the parallel surfaces of the specimens with a spatial resolution of 2 mm in the horizontal and vertical directions, with the same frequencies that were used in the continuous monitoring (500 kHz) and a frequency sampling of 10 MHz. Approximately 80 and 160 A-scans were obtained in the length and width directions, respectively, on each specimen.

16



Figure 5: Automated inspections of concrete specimen [13]. It is necessary to point out that all manufactured specimens (three for each mix) were inspected, although not all of them were subjected to accelerated cycles in the climatic chamber. 3. Experimental Results 3.1 Temperature, Relative Humidity and Strain Monitoring The temperature behaviour inside the specimen measured with the embedded strain gauge temperature sensor and with the WilTempUS system on the sample surface along three F-T cycles is represented in Fig. 6. It can be observed that the results obtained with both sensors are very similar. There was a delay on the temperature values of the specimen with respect to the ambient temperature in both cases.

Figure 6: Temperature values using the strain sensor and the WilTempUS system during three F-T cycles.

17



The RH behaviour of the specimen measured with the WilTempUS system on the sample surface and in the climatic chamber along three F-T cycles is represented in Fig. 7. It is observed that when the RH of the climatic chamber decreases, the specimen absorbs water and its RH increases.

Figure 7: Relative humidity values using the WilTempUS system during three F-T cycles. The continuous internal strain measurement of the samples was performed in one specimen of each mixture. This measurement enables to study the deformation of the samples. The strain evolution, maximum and minimum values, of the two types of mixtures can be seen in Fig. 8. This figure shows that the strain of the mix A samples grew faster than those of mix B.

Figure 8: Internal strain evolution of both mixtures during the F-T cycles. 3.2 Monitoring of Ultrasonic Parameters The maximum and minimum values of the ultrasonic relative velocity (UV) were measured in two concrete specimens of each mixture during the F-T cycles, one using the Pundit system

18



and the other one with the WilTempUS, see Fig. 9. Both ultrasonic systems led to very similar results. A comparison of both mixtures shows that the maximum and minimum UV values decrease along the F-T cycles, except for the maximum values of the mixture with more water, Mix B.

Figure 9: Maximum and minimum ultrasonic velocity values during the F-T cycles. 3.3 Ultrasonic Images of Concrete before and after the F-T Cycles The maps of amplitude before and after the F-T cycles are shown in Fig. 10. Three different specimens from each mixture were inspected, two of them were subjected to the F-T cycles and the other one served as the reference specimen (right specimen). It shows that mix B was slightly more affected than mix A by the F-T cycles.

Figure 10: Amplitude maps of concrete specimens before and after the F-T cycles.

19



4. Discussion and conclusions It has been observed that the temperature behaviour using an embedded temperature sensor does not vary significantly with the results obtained placing the sensors on top of the concrete specimens with an insulated material. During the F-T cycles, there was a delay on the temperature measurements of the specimen with respect to the ambient temperature of the climatic chamber, and these values did not reach the temperature programmed values of the climatic chamber. The deformation process measured with the embedded strain gauges was very similar for both mixtures. The strain of the mix A samples grew faster than those of the other mixture. In mix A, this deformation process has occurred primarily during the first cycles, while in mix B occurred from cycle 15. The coefficient of thermal expansion in mix A samples has remained almost constant but this coefficient increased slightly for mix B specimens, probably produced by their further deterioration. It has been observed that the two systems used to monitor the ultrasonic velocity have obtained similar results. The initial ultrasonic velocity is lower for the specimens of mix A, than those of mix B. During the freezing cycle the velocity decreases in mix A, but it remains almost constant in mix B, indicating that the loss of velocity in mix B may be compensated with higher water absorption. During the thawing cycle, the velocity loss is small in both mixtures, approximately 2% in mix A and about 4% for B, indicating a further deterioration in mix B. This small deterioration process, probably only on the samples surface, caused a drop of the amplitude of the ultrasonic signals greater than 20dB as it was shown in the amplitude images. Finally, it can be concluded that all monitored parameters have shown changes in the structure of the specimens during F-T cycles. After the F-T cycles, the two mixtures were damaged leading to small microcracks on the sample surface. Specimens of the mixture with more w/c ratio were slightly more affected by the F-T cycles. Acknowledgments The Spanish Economy and Competitiveness Ministry supported this research under grant number TEC2012-38402-C04-03. The authors will also like to thank the Cost Action TU1404 for supplying the material of this work.

References [1] UNE-CEN/TS12390–9EX, Testing Hardened Concrete-Part 9: Freeze-Thaw Resistance-

Scaling, AENOR, Spain (2008) [2] ASTM C 666/C 666M-03, Standard Test Method for Resistance of Concrete to Rapid

Freezing and Thawing, ASTM International, USA (2003) [3] prENV 12390–9, Testing Hardened Concrete. Part 9: Freeze-thaw Resistance. Scaling,

European Committee for Standardization, Hungary (2003) [4] JIS A 1148:2001, Method of Test for Resistance of Concrete to Freezing and Thawing,

Japanese Industrial Standards, Japan (2001)

20



[5] Tang, L. and Petersson, P.E., Slab test: freeze/thaw resistance of concrete—Internal deterioration, Mater Struct 37 (2004), 754–759.

[6] Setzer, M.J. et al, Test methods of frost resistance of concrete: CIF-Test: Capillary suction, internal damage and freeze thaw test—Reference method and alternative methods A and B, Mater Struct 37 (2004), 743–753

[7] EN 12390-1:2012 Testing hardened concrete -- Part 1: Shape, dimensions and other requirements for specimens and moulds (2012)

[8] http://www.tu1404.eu/ [9] Enfedaque, A. et al, Durabilidad del hormigón frente a los ciclos hielo-deshielo.

Evaluación de dos tipos de hormigón, Anales de la Mecánica de la fractura XVIII (2011), 675–680

[10] Romero, H.L. et al, Deterioro del hormigón sometido a ciclos de hielo-deshielo en presencia de cloruros, Anales de la Mecánica de la fractura XVIII (2011), 669–674

[11] Aparicio, S. et al, Procedimiento y Sistema Inalámbrico de Medida del Grado de Fraguado y Endurecimiento de Materiales Cementicios Para la Predicción de Resistencias Mecánicas, Spain Patent PCT/ES2012/070439 (2012)

[12] Molero, M. et al, Evaluation of freeze-thaw damage in concrete by ultrasonic imaging, NDT E Int 52 (2012), 86–94

[13] Ranz, J. et al, Monitoring of Freeze-Thaw Cycles in Concrete Using Embedded Sensors and Ultrasonic Imaging, Sensors 14 (2014), 2280-2304

[14] http://fr.amiando.com/EDF-vercors-project.html [15] EN 12390-2:2009, Testing hardened concrete -- Part 2: Making and curing specimens for

strength tests (2009) [16] Ranz, J., Caracterización no destructiva del proceso de curado en materiales cementicios,

PhD thesis, UPM (2015) [17] Molero, M. et al, Caracterización de materiales mediante la dispersión ultrasónica:

Aplicación a los materiales cementicios (in Spanish), Académica Española: Editorial Académica Española, Germany (2011)

21



FROST RESISTANCE OF CONCRETE – EXPERIENCE FROM LONG TERM FIELD EXPOSURE Dimitrios Boubitsas (1), Peter Utgenannt (1), Luping Tang (2), Elisabeth Helsing (1)

(1) CBI Swedish Cement and Concrete Research Institute, Borås, Sweden

(2) Chalmers Univ. of Technology, Gothenburg, Sweden Abstract Concrete specimens made with different binder types/combinations, including supplementary cementitious materials, have been exposed at an exposure site in a highway environment for over nineteen years. The resistance to internal and external frost damages have been regularly evaluated by measurements of changes in volume and in ultrasonic pulse transmission time. This paper presents the results after nineteen years of exposure. The results from this study show that the existence of entrained air and the water-binder ratio are the main parameters influencing the resistance of concrete to external salt-frost damage. Furthermore, the concrete mixes with CEM I, CEM I + 5 % silica, CEM II/A-LL, CEM II/A-S and CEM I + 30 % slag as binder with entrained air and a water/binder ratio of 0.4 or below, has good resistance to internal and external frost damage. Results show that concrete containing large amounts of slag in the binder (CEM III/B) have the severest scaling, irrespective of if it contains entrained air or not. 1. Introduction Resistance to external and internal frost damage is of great importance in determining the durability of concrete in the Scandinavian countries, as well as in other European countries with a cold climate. To acquire experience from representative outdoor environments an investigation was started in Sweden in the mid-nineties. Three field exposure sites were established in the south-west of the country: one in a highway environment, one in a marine environment, and one in an out-door environment without salt exposure. The results after ten years of exposure at the three field test sites are presented in [1] and after fourteen years at the highway test site in [4]. The results after ten years of exposure clearly indicated the highway environment as being the most aggressive with regard to external frost damage. In this paper, results after nineteen years of exposure at the highway exposure site are presented.

22



The highway field exposure site is located alongside highway 40, 60 km east of Gothenburg. The specimens are mounted in steel frames at road level, and a guard rail separates the exposure site from the traffic. The specimens are placed in such a way that they are fully exposed to splash water from the traffic. The air temperature during the winter season fluctuates widely, at the beginning and at the end of the winter season these fluctuations are between +15 °C to about -5 °C. At mid-winter the fluctuations are from 0 °C to about -10 °C, with occasional peaks down to -20 °C. In this region a de-icing agent is normally used between October and April, in average about 1.7 kg/m2 a year. The de-icing agent used is sodium chloride. The climate at the field site is estimated to correspond to exposure class XD3 / XF4 in EN 206-1. A more detailed description of the field site can be found in [2]. A large number of concrete specimens with different binder types/combinations, water-binder ratios (w/b) and air contents have been exposed for more than nineteen years at this test site. The external and internal frost damages have been evaluated regularly by measurements of the volume change of, and the change in ultrasonic transmission time through, each specimen. 2. Materials and methods 2.1 Materials The binder types/combinations studied in this investigation are shown in Table 1 and the chemical compositions of the binders are shown in Table 2. For a more complete presentation of the used materials, see [3]. Table 1: Binder types/combinations investigated.

Binder type/combination Comments 1 CEM I 42,5N MH/SR/LA Low alkali, sulfate-resistant 2 CEM I + 5 % silica by binder weight Silica in the form of slurry 3 CEM II/A-LL 42.5R Cement with 15 % limestone filler 4 CEM II/A-S Finnish cement with ~15 % slag 5 CEM I + 30 % slag by binder weight Ground blast furnace slag added in the mixer 6 CEM III/B Dutch slag cement, ~70 % slag Table 2: Chemical composition of the binders

(%) CEM I CEM II/A-LL CEM II/A-S CEM III/B Silica slurry Slag CaO 65.1 62.0 60.2 42.1 0.02 35.5 SiO2 22.6 18.9 20.1 30.3 93.3 36.0 Al2O3 3.41 3.36 4.8 13.6 10.3 Fe2O3 4,37 2.49 2.4 1.18 0.07 MgO 0.75 2.31 3.4 8.45 0.42 14.9 K2O 0.58 1.14 0.84 0.66 0,84 0.60 Na2O 0.07 0.17 0.76 0.31 0.18 0.50 Na2O - equ 0.45 0.92 1.31 0.74 0.73 0.89 SO3 2.13 3.59 3.0 5,2 0,4 3.7 Cl- 0.01 <0.01 0.02 0.063 <0.01 Loss on ignition 0.52 6.08 3.8 0.5 1.9 1.0

23



For all binder type/combinations, several concrete qualities were produced. Five different w/b-ratios (0.30, 0.35, 0.40, 0.50, 0.75) were used and mixes with and without entrained air were produced. For all mixes 0–8 mm natural sand and 8–16 mm crushed aggregate were used. A naphthalene-based plasticizer (Melcrete) was used for mixes with w/b-ratio of 0.40 or lower. The air-entraining agent used, L16, is a tall oil derivative. Data about the composition and some properties of all the mixes studied in this investigation are compiled in Table 3. Table 3 - Concrete constituents and properties. Binder type w/b-

ratio Equiv. w/c-ratio(1

Cement (kg/m3)

SCM(2 (kg/m3)

aea(3 Air content fresh (%)

Slump (mm)

Compressive strength (MPa)

SS(4 Recalc(5

Scaling (kg/m2)(6 28 56

CEM I 0.30 0.30 500 - Yes 4.8 240 95 87 0.02 0.04 0.35 0.35 450 - Yes 4.8 190 95 87 0.05 0.09 0.40 0.40 420 - Yes 4.6 125 67 60 0.01 0.02 0.50 0.50 370 - Yes 4.6 90 49 44 0.02 0.02 0.75 0.75 260 - Yes 4.7 100 21 18 0.13 0.14 0.30 0.30 500 - No 1.1 120 102 93 0.16 0.26 0.35 0.35 450 - No 1.2 140 91 83 1.94 4.39 0.40 0.40 420 - No 0.8 130 87 79 3.11 7.92 0.50 0.50 385 - No 0.8 70 56 50 5.09 14.5 0.75 0.75 265 - No 0.9 60 31 27 4.34 >15 95 % CEM I + 5 % silica

0.30 0.29 475 25 Yes 4.6 100 103 94 0.04 0.12 0.35 0.33 427.5 22.5 Yes 4.5 90 91 83 0.02 0.04

0.40 0.38 399 21 Yes 4.8 105 72 65 0.02 0.04 0.50 0.48 361 19 Yes 4.6 70 57 51 0.02 0.03 0.75 0.71 237.5 12.5 Yes 4.3 70 25 22 0.19 0.20 0.30 0.29 475 25 No 1.1 125 121 111 0.12 0.20 0.35 0.33 427.5 22.5 No 1.1 90 105 96 0.36 0.89 0.40 0.38 399 21 No 0.5 100 84 76 1.67 3.25 0.50 0.48 370.5 19.5 No 1.2 60 67 60 1.86 4.61 0.75 0.71 256.5 13.5 No 0.3 75 35 31 3.45 6.58 CEM II/A-LL

0.30 0.30 520 - Yes 4,4 110 89 81 0.03 0.05 0.35 0.35 450 - Yes 4,3 110 85 77 0.14 0.25

0.40 0.40 420 - Yes 4,6 90 72 65 0.11 0.12 0.50 0.50 390 - Yes 4,8 70 52 46 0.17 0.20 0.75 0.75 260 - Yes 4,7 70 34 30 2.92 5.26 0.30 0.30 530 - No 2,3 100 95 87 0.16 0.28 0.35 0.35 470 - No 2,4 115 86 78 0.52 1.05 0.40 0.40 420 - No 2,4 110 76 69 3.04 5.67 0.50 0.50 400 - No 1,8 70 64 57 5.05 11.6 0.75 0.75 280 - No 1,2 75 36 32 TD TD CEM II/A-S 0.30 0.30 520 - Yes 4,4 70 62 56 0.09 0.11 0.35 0.35 450 - Yes 4,5 100 62 56 1.40 2.23 0.40 0.40 420 - Yes 4,5 80 53 47 1.75 2.45 0.50 0.50 380 - Yes 4,8 100 39 34 0.15 0.16 0.75 0.75 260 - Yes 4,7 100 25 22 1.12 1.25 0.30 0.30 540 - No 2,2 30 69 62 0.32 0.42 0.35 0.35 450 - No 2,6 35 64 57 2.37 4.06 0.40 0.40 420 - No 2,0 90 58 52 3.93 6.71 0.50 0.50 400 - No 1,7 65 48 43 5.28 7.65 0.75 0.75 275 - No 0,5 100 30 26 TD TD

24



Binder type w/b- ratio

Equiv. w/c-ratio(1

Cement (kg/m3)

SCM(2 (kg/m3)

aea(3 Air content fresh (%)

Slump (mm)

Compressive strength (MPa)

SS(4 Recalc(5

Scaling (kg/m2)(6 28 56

70 % CEM I + 30 % slag

0.30 0.34 350 150 Yes 4.8 230 90 82 0.03 0.05 0.35 0.40 315 135 Yes 4.8 130 86 78 0.09 0.14

0.40 0.45 294 126 Yes 4.4 110 65 58 0.04 0.07 0.50 0.57 259 111 Yes 4.8 80 49 44 0.02 0.03 0.75 0.85 175 75 Yes 4.4 100 20 17 0.49 0.54 0.30 0.34 350 150 No 0.7 220 101 92 0.10 0.13 0.35 0.40 315 135 No 1.1 140 91 83 1.78 3.61 0.40 0.45 294 126 No 0.9 120 78 71 1.78 3.83 0.50 0.57 273 117 No 1.3 80 52 46 0.86 1.94 0.75 0.85 185.5 79.5 No 0.5 80 25 22 1.58 4.37 CEM III/B (~70 % slag)

0.30 0.30 520 - Yes 4.8 200 78 71 0.25 0.36 0.35 0.35 460 - Yes 4.7 200 74 67 0.43 0.62

0.40 0.40 420 - Yes 4.3 120 61 55 0.57 0.85 0.50 0.50 380 - Yes 4.5 70 46 41 0.99 1.66 0.75 0.75 255 - Yes 4.4 90 26 23 2.05 3.16 0.30 0.30 520 - No 0.8 200 99 90 0.22 0.28 0.35 0.35 470 - No 0.7 200 80 72 0.49 0.65 0.40 0.40 420 - No 0.9 125 68 61 0.84 1.14 0.50 0.50 400 - No 1.0 65 54 48 1.21 1.61 0.75 0.75 265 - No 0.1 100 31 27 3.94 6.89 (1 Equiv. w/c-ratio=water/(cement + 2 silica + 0.6 slag). (2 SCM – Supplementary Cementitious Materials. (3 AEA – Air Entraining Agent. (4 Dry stored cubes tested in accordance with SS 13 72 10 [5] at the age of 28 days. (5 Recalculated to wet stored cubes in accordance with fwet,cube = 0.76 (fdry,cube)1.04. (6 In accordance with the ‘Slab test’, SS 137244 [6]. TD = Totally disintegrated All concrete batches were produced in the autumn of 1996, and a number of 150 mm cubes were cast from each batch. The cubes were demoulded 24 hours after casting, and stored in lime-saturated water for six days. They were then stored in a climate chamber (50 % RH at 20 C) for a period of between one and a half and three months. Between eight and twelve days before the specimens were placed at the field exposure sites, the cubes were cut along the casting direction into two specimens with the shape of a half 150 mm cube with one cut surface and the rest mould surfaces. After cutting, the specimens were stored in a climate chamber (50% RH at 20 C) until placed at the exposure site. During this second conditioning period, the volume of and the transmission time through each specimen were measured. Two specimens of each mixture were then placed at the exposure site. The specimens were placed in steel frames close to the road, so that they could receive the water and slush splashed by the passing traffic. The specimens were exposed with the cut surface facing upwards. 2.2 Methods The volumes of the specimens are calculated from results obtained from measuring the weight of the specimens first in water and then in air. The ultrasonic pulse transmission time through the specimen (150 mm) is measured and the mean of three measurement positions is calculated, where possible, on each specimen. In addition, a number of laboratory tests were carried out in order to determine the concrete characteristics (results are partly given in Table 3 above and fully in [3]). One of these was testing for salt-frost resistance in accordance with the Swedish Standard SS 137244 which is a laboratory test at the age of 31 days (the ‘Slab test’) [6]. The ‘Slab test’ is in agreement with the reference procedure in CEN/TS 12390-9.

25



3. Results The results from measurements of the changes in volume (%) and ultrasonic pulse transmission time (%) after nineteen years of exposure at the highway exposure site for concretes with different binder combinations, water-binder ratios and with or without entrained air are shown in Figures 1 and 2 and Tables 4 and 5, respectively. The reference value is the initial volume or the transmission time before exposure, and each point is the mean value of measurements on two specimens. In Tables 4 and 5, no value is presented for w/b of 0.75 because for most of these concrete mixes the damage on the concrete surfaces was so severe that measurements of ultrasonic pulse transmission time were not possible.

Figure 1. Volume change after nineteen years at the highway exposure site. Concretes with different binder combinations and water/binder ratios, with entrained air (4–5%).

Figure 2. Volume change after nineteen years at the highway exposure site. Concretes with different binder combinations and water/binder ratios, without entrained air.

26



Table 4: Change in transmission time (%) after nineteen years. Concrete with entrained air. Entrained air (4-5) Water/binder ratio Concrete mix 0.30 0.35 0.40 0.50 CEM I -2 0 -2 -2 CEM I + 5 % SF 1 0 1 0 CEM II/A-LL -4 -5 -2 -2 CEM II/A-S -2 -1 -2 1 CEM I + 30 % slag -2 -2 0 -4 CEM III/B -4 -1 0 - 1) 1) No detection could be made because of to severe surface damage

Table 5: Change in transmission time (%) after nineteen years. Concrete without entrained air. Without entrained air Water/binder ratio Concrete mix 0.30 0.35 0.40 0.50 CEM I -3 -2 -2 -2 CEM I + 5 % SF 1 -1 5 15 CEM II/A-LL -2 -3 -1 2 CEM II/A-S -4 -3 -3 0 CEM I + 30 % slag -3 0 1 -5 CEM III/B -1 -2 2 - 1) 1) No detection could be made because of to severe surface damage

Figure 3 shows results from the laboratory test (SS 137244) as a function of the decrease in volume for specimens exposed in the highway environment for nineteen years.

Figure 3. Scaling results from the laboratory test (SS 137244) as a function of the decrease in volume for specimens exposed in the highway environment for nineteen years.

27



4. Discussion 4.1 Performance after nineteen years exposure External frost resistance of concretes with entrained air: As can be deduced from Figure 1, all mixes with CEM I, CEM I + 5 % silica, CEM II/A-LL with a water/binder-ratio of 0.5 or below show good resistance to external frost damage. Concrete mixes with CEM II/A-S and CEM I + 30 % slag as binder with a w/b-ratio of 0.40 or below, has good resistance to internal and external frost damage but has a somewhat greater volume decrease when the w/b-ratio is 0.50. Concrete with CEM III/B with a w/b-ratio of 0.35 or below, also has good resistance to internal and external frost damage. For w/b-ratio 0.40, however, the volume loss after nineteen years is somewhat higher for concrete with CEM III/B as binder and at a w/b-ratio of 0.50, shows significant damage. For all concrete mixes with entrained air and with a w/b-ratio of 0.75, the specimens were more or less damaged after nineteen years of exposure. External frost resistance of concretes without entrained air: For concrete without entrained air all concrete qualities with w/b-ratio 0.75 show severe scaling, see Figure 2. For concretes with w/b-ratio 0.50, the qualities with CEM I and CEM III/B as binder show the highest scaling. For concrete with w/b-ratio 0.40 and below, the differences in scaling are relatively small. However, the concrete with CEM III/B and w/b-ratio 0.40, shows a markedly higher scaling compared to the other concrete mixes with the same w/b-ratio. Also the concrete quality CEM II/A-S and w/b-ratio 0.35 shows a higher scaling than do other concrete qualities. The major part of this scaling occurred during the first winter, after which the scaling was very small. One explanation for this unexpected behaviour could be problems with the compatibility between the cement and the plasticizer used, resulting in a poor air void structure. The plasticizer was used for all mixes with w/b-ratio 0.40 and lower. From Figure 2 it can be seen that, for the qualities with CEM II/A-S and with w/b-ratio 0.40 and 0.35, the scaling is higher than for the quality with w/b-ratio 0.50. The same tendency can be seen for concrete with entrained air and with w/b-ratio 0.35 in Figure 1. In addition, the results from testing in the laboratory indicate lower scaling resistance for concretes with entrained air and with w/b-ratios 0.35 and 0.40 compared to 0.50 (see Table 3), indicating a possible compatibility problem between the cement, plasticizer and air-entraining agent.

In general all concrete qualities without entrained air with w/b-ratio 0.40 and below, and for some binder types even with w/b-ratio 0.50, shows relatively limited surface damage. A higher degree of damage for the concretes without air entrainment was anticipated. A drawback of the volume measurement procedure is that the measured volume is the net volume of both a negative and a positive element, i.e. it may be the result of a volume decrease due to surface scaling and a volume increase due to internal cracking. Concrete qualities with an apparent volume loss might, therefore, also have a small increase in volume caused by internal cracking without this being observed. It is therefore important to complement volume measurements with measurement with other techniques, such as the ultrasonic pulse transmission time, in order to detect internal damage. Internal frost resistance: Table 4 and 5 shows the percentage change in transmission time after nineteen years of exposure for concrete with and without entrained air, respectively. A negative value (decrease in transmission time through the specimen with age) is expected for

28



sound, undamaged materials due to the densification of the paste as a result of continued hydration. A clear positive value indicates possible internal damage. The results presented in Table 4 show that for the concretes with entrained air there is no clear indication of internal damage for any of the concrete mixes with w/b-ratio 0.50 or below. For the concretes without entrained air the results in table 5 show that only concrete mixes with w/b-ratio 0.40 and 0.50 and with CEM I + 5 % silica shows a marked increase in transmission time indicating internal damage. These concrete mixes show only a small volume loss after nineteen years of exposure, see figure 2. However, results after five years of exposure, presented in [2], and after ten years of exposure, presented in [1], show an increase in volume. Combining these results from volume measurements after five and ten years with the transmission time measurements gives clear indications of internal damage for the concrete qualities with CEM I + 5% silica as binder and without entrained air, also for concrete with w/b-ratios down to 0.40. Microscopic techniques, such as the analysis of polished sections or thin sections, has been used to look for further evidence of internal damage, and the results of that study substantiated the indications of internal damage observed in this study for these qualities after five years exposure at the exposure sites [2]. For concrete with CEM III/B as part of the binder and with w/b-ratio 0.50, both with and without entrained air, the surfaces of the specimens were too severely damaged to be able to measure the ultrasonic transmission time. Possible internal damage could therefore not be evaluated for these concrete mixes. In general, the volume changes for concrete with high w/b-ratios and entrained air are less than for concrete without entrained air (compare Figure 1 with Figure 2). However, this is not valid for concrete qualities with CEM III/B as the binder. The volume change for mixes with CEM III/B as binder and with entrained air is of the same order of magnitude as that for concrete without entrained air. For these qualities, entrained air does not seem to improve the scaling resistance. This behaviour is confirmed by the freeze/thaw testing in the laboratory, Table 3. For concrete with CEM III/B as the binder, the air-entrained qualities show damage of the same order of magnitude as for concrete without air. 4.2 Correlation between laboratory and field tests In the present investigation, each concrete was tested at the age of 31 days in accordance with SS 137244, the ‘Slab test’. The acceptance criterion in this laboratory test is 1 kg/m2, illustrated by a horizontal line in Figure 3. An acceptance criterion of 2–3 % volume loss, shown by the vertical zone in Figure 3 after nineteen years of exposure has been chosen for the field exposed specimens, corresponding to a scaling of approximately 1 kg/m2. The filled in symbols in Figure 3 represent concrete with entrained air, while the not filled symbols represent concrete without entrained air. The results presented in Figure 3 show that only three qualities clearly fall into Quadrant IV, which is the worst case, where they are accepted by the test method but fail in field exposure. These are, however, air-entrained qualities with high a w/b-ratio (0.75). The standard test method is primarily intended to be used for bridge concrete, with entrained air and with a w/b-ratio up to 0.50. Three concrete mixes accepted in the laboratory; CEM III/B (w/b-ratio 0.40), CEM II/A-S (w/b-ratio 0.50), CEM I + 30 % slag (w/b-ratio 0.50), all with entrained air

29



are showing a volume loss in the field of around 3 % and is moving from Quadrant III to Quadrant IV. All of these qualities contain slag as part of the binder. This is an indication that the laboratory test method may overestimate the scaling resistance of concrete containing a medium to high content of slag as part of the binder. Most qualities fall into Quadrants II or III, which means that the test method and ‘reality’ correspond. Some concrete qualities fall into Quadrant I, which means that they are rejected by test method but withstands field exposure. However, since the concrete in Quadrant I show only limited damage in the field, the test method results are on the safe side. All qualities in Quadrant I are with w/b-ratio 0.5 or lower and without entrained air (or for CEM II/A-S with a poor air-void structure), which makes them particularly susceptible to frost damage. During the first nineteen years, the climate has not been aggressive enough to damage these qualities significantly. However, one winter season with a more aggressive climate may cause internal damage as well as scaling on these qualities, moving them into Quadrant II. Two concrete qualities show an increase in volume (after five and ten years of exposure) and an increased transmission time, indicating internal frost damage. These qualities also fail the acceptance criterion when tested in the laboratory. From the results it can be concluded that the laboratory test, ‘Slab test’, classifies most concrete qualities in this investigation correctly. For some qualities the slab test gives results on the safe side (Quadrant I) indicating a positive effect of ageing for these concrete qualities. For some concrete qualities with medium to high contents of slag as part of the binder results from the field indicate more damage than could be expected from the slab test results, indicating a negative effect of ageing. Possible explanations to the effect of ageing is discussed in [2 and 4] where carbonation is pointed out as the most important factor. Carbonation, however, influences the scaling resistance differently depending on the binder type. For concrete with CEM I and with low amount of additions the effect of carbonation is positive with regard to scaling resistance. For concrete with high content of slag as part of the binder the effect is negative. The importance of the effect of carbonation needs to be taken into consideration when testing and evaluating scaling resistance using laboratory test methods. 5. Conclusions The following conclusions can be drawn after nineteen years of exposure at the highway exposure site:

Concrete mixes with CEM I, CEM I + 5 % silica, CEM II/A-LL, CEM II/A-S and CEM I + 30 % slag as binder with entrained air and a water/binder ratio of 0.4 or below, has good resistance to internal and external frost damage. For concrete mixes with CEM I, CEM I + 5 % silica, CEM II/A-LL, with entrained air and a water/binder ratio of as high as 0.5 show good resistance to internal and external frost damage. For concrete mixes with CEM II/A-S and CEM I + 30% slag as binder a somewhat greater volume loss can be seen for concrete with entrained air and a w/b-ratio of 0.5 compared to the other mixes.

30



Concrete with CEM III/B, suffers from severe scaling, even with a w/b-ratio of 0.5 and with entrained air. The concrete mix with w/b-ratio 0.4 and entrained air shows a somewhat higher scaling compared to the other binder types with the same w/b-ratio.

In general, internal damage is observed only for concrete qualities without entrained air and, furthermore, in most cases for concrete qualities with high w/b-ratios. However, for concrete qualities with CEM I + 5 % silica as binder, internal damage is found at lower w b-ratios, down to w b-ratio 0.4.

Comparing results from laboratory testing in accordance with SS 137244 (the ‘Slab test’, essentially in accordance with CEN/TS 12390-9) with results after nineteen years of exposure shows that the laboratory test classifies most concrete qualities correctly. Some concrete qualities with w/b-ratios of 0.50 or below and without air or with poor air-void structure were rejected by the laboratory test but show only low levels of damage in the field, indicating that the slab test gives results on the safe side for these qualities.

Field results for three concrete qualities containing medium to high contents of slag as part of the binder and w/b-ratio between 0.4 and 0.5 indicate more damage than could be expected from the slab test results.

For concrete with CEM III/B as the binder, entrained air does not seem to improve the scaling resistance. This is seen for specimens exposed in the field and it is confirmed by freeze/thaw testing in the laboratory.

The effect of ageing and in particular carbonation needs to be taken into consideration when testing and evaluating scaling resistance using laboratory test methods.

6. Acknowledgement This project is financed by the Swedish Transport Administration within the BBT-programme and the cement producer Cementa AB.

References [1] Utgenannt, P., Frost resistance of concrete - Experience from three field exposure sites,

Proceedings from a workshop on Nordic exposure sites, Nordic Concrete Federation, Hirtshals, Denmark, 2008.

[2] Utgenannt, P., ‘The influence of ageing on the salt-frost resistance of concrete’, Doctoral thesis, Division of Building Materials, Lund Institute of Technology, Report TVBM-1021, Lund, Sweden, 2004.

[3] Utgenannt, P., The effect of binder on the frost resistance of concrete - Test specimens produced in 1996 - Material and production data and results from life testing in the laboratory, Swedish National Testing and Research Institute, BTB report no 1, Borås, Sweden, 1997, (in Swedish).

[4] Utgenannt, P. and Petersson, P.-E., Frost Resistance of Concrete Containing Secondary Cementitious Materials - Experience from Field and Laboratory Investigations, Nordic Concrete Federation, Oslo, Norway, 2012.

[5] SS 137210, ‘Concrete testing – Hardened concrete – Cube strength’, Swedish Standards Institution (SIS), first edition, Stockholm, Sweden, 1978.

[6] SS 137244, Concrete testing – Hardened concrete- Frost resistance, Swedish Standards Institution (SIS), 3rd edition, Stockholm, Sweden, 1995.

31



THE INFLUENCE OF THE FREEZE-THAW LOADING CYCLE ON THE INGRESS OF CHLORIDES IN CONCRETE Miguel Ferreira (1), Markku Leivo (1), Hannele Kuosa (1), David Lange (2) (1) VTT Technical Research Centre of Finland Ltd., Kemistintie 3, Espoo 02044, Finland (2) University of Illinois, Department of Civil Engineering, Urbana, 61801, Illinois, USA Abstract Recent research has drawn attention to the interaction of degradation mechanisms such as frost attack and chloride penetration occurring in the typical environmental conditions of the Nordic countries. Frost attack of concrete affects the chloride penetration by reducing the concrete cover, and more importantly, by changing the characteristics of the surface and internal concrete due to cracking. Literature shows that a long period of freezing conditions will slow chloride ingress. New research results, however, show that the rate of chloride ingress can remain identical regardless of whether the tests are exposed to daily freeze-thaw cycling. This paper presents these finding, which suggests that the interaction between the various transport mechanism (capillary water uptake, water and vapour diffusion and micro ice-lens pumping) is complex, and that no single mechanism consistently explains the test results. Even if freezing slows bulk transport, other mechanisms counteract the slowdown. Their combined interactions result in profiles identical to those of pure ponding under constant ambient temperature. 1. Introduction Reinforced concrete structures (RCS), exposed to extremely harsh winters such as those experienced in Nordic countries, are expected to perform in such difficult conditions despite the unique combinations of degradations mechanisms that can occurs, such as freeze-thaw damage, chloride ingress, carbonation, among others. In the coastal areas, chlorides can originate from the sea, otherwise, the most common source of chlorides is from the use of de-icing salts in the winter. Research has recently focused on freeze-thaw and carbonation as the predominant degradation mechanisms [1-3], but more attention is now being given to

32



coupling degradation mechanisms such as freeze-thaw and chloride penetration. The phenomena of freeze-thaw has been extensively studied with many significant contributions by [4-19], however, little attention has been given to the interaction between freeze-thaw and chloride ingress. Freeze-thaw reduces the concrete cover due to scaling (in the presence of salts), and by changing the characteristics of both surface and internal concrete due to cracking [17]. As a result, it has been shown [1] that frost attack affects chloride penetration, but research has noted that this phenomenon is not yet understood. The purpose of the testing was to investigate the effect of selected freeze-thaw exposures on the chloride ingress in concrete. To minimize freeze-thaw induced damage (scaling and internal cracking), a preliminary study was undertaken to optimize the air entrainment for each concrete. 2. Materials and methods 2.1 Mixture proportions and specimen preparation Two concrete compositions were chosen with a w/c ratio of 0.42 and 0.55 (referred to as B42 and B55, respectively). CEM I 42,5 N-SR3 was used to minimise chloride binding. A plasticizer (VB-Parmix) was used in the B42 mix, and an air entrainment agent (Ilma-Parmix) in both mixes. The concrete compositions and their workability and fresh concrete air content are summarised in Table 1, and the characteristics of the air entrainment measured on hardened specimens are summarised in Table 2 Table 1: Concrete compositions and workability

Series w/c Water (l/m3)

Cement (kg/m3)

Aggregate - Total (kg/m3) & fractions (%) Slump (mm)

Air (%) Total <0.125 <0.250 <4.0 <10.0

B42 0.42 175 420 1695 5.8 12.1 58.7 89.1 100 5.0 B55 0.55 195 355 1716 4.3 9.7 59.4 89.2 180 6.0 Table 2: Characteristics of the air pores of hardened concrete.

Series Protective AP 0.02-0.80 mm (%)

Compaction AP > 0.800 mm (%)

(%)

SS area for protective AP mm2/mm3)

Spacing factor for protective AP (mm)

B42 2.9 1.2 4.1 33 0.21 B55 5.1 0.8 5.9 21 0.24

AP – air pores; SS – specific surface The results in Table 2, determined by thin section microscopy analysis, show that very good air entrainment was achieved in both concretes mixes, as shown by the low spacing factor and the specific surface. 150 mm cubic specimens were cast in moulds and compacted using a vibrating table. 24 hours after casting the specimens were removed from the moulds and

ly 8 months to minimise the influence of microstructure changes due to continuous hydration on the test results.

33



Ten days prior to testing, the specimens were prepared in accordance with the procedure defined in the CEN/TS 12390-9 [20], except with regard to the age of the specimens. A test specimen constitutes a concrete specimen with 150x150x50 mm (See Figure 1).

(a) (b) (c)

Figure 1. Views of the test specimen: (a) prior to preparation for freeze-thaw test, (b) after being prepared for freeze-thaw test, (c) during freeze-thaw test. The internal moisture condition of the specimens at the start of testing strongly influences the results. Published research literature often neglects to mention this aspect, resulting in misleading or incomplete interpretation of results. Prior to testing the specimens were ponded with a layer of deionised water for three days, resulting in a large capillary uptake of water without chlorides. The influence of capillary uptake is therefore minimized for the freeze-thaw exposures. At the start of freeze-thaw exposure, the layer of water was replaced with a 3 % by weight of NaCl solution. 2.2 Testing Procedure Four different freeze-thaw exposures cycles were chosen to investigate the effect of the minimum temperature reached during the freeze-thaw cycle, and the length of time at which the minimum temperature was held during the freeze-thaw exposure cycle. The first three exposure cycles, where the minimum temperature reached is studied, follow closely the reference test procedure curve for freeze-thaw scaling [20], varying only in the minimum temperature reached: -20 °C (reference), -10 °C and -5 °C, as shown in Figure 2a. Each of these cycles has a 24 hour duration. The rate of freezing was kept as close as possible to the reference test (see Figure 2b). The fourth exposure follows the reference exposure (-20 °C), except that the lowest temperature is maintained for 60 hours, so that the total length of one exposure cycle is 84 hours. The choice of different freeze-thaw exposure intensity cycles (i.e. minimum negative temperature) and different durations of intensity was to promote varying freeze-thaw behaviour of the brine solution in the pore structure of the concrete [21]. The total number of freeze-thaw cycles the concrete samples were subject to was 112 cycles for the 24 hour exposure, and 41 cycles for the 84 hour exposure. As a reference for measuring the chloride ingress due to diffusion, additional exposures were conducted with full immersion at constant +5 °C, and +20 °C. At regular intervals during exposure cycles, the following measurements were conducted: scaled material mass, mass

34



variation (water uptake – scaled material) and fundamental frequency according to CEN/TR 15177 [22]. Chloride profiles were determined on specimens removed from the freeze-thaw cambers at 28, 70 and 112 cycles for the 24 hour exposures, and at 14, 30 and 41 cycles for the 84 hour exposures

(a) (b)

Figure 2. Reference temperature curves for the 24 hours exposures to determine the effect of minimum temperature reached (a), and example of actual temperature reached (refer to that of the solution on the surface of the specimens) for one set of specimens (b). 3. Results So as to minimise the effect of certain factors on the test results, special considerations were taken: CEM I 42,5 N-SR3 was used to minimise chloride binding; good air entrainment was assured to minimise damage (scaling and internal cracking); long wet curing was used to minimise the influence of microstructure changes as a result of continuous hydration; and low surface carbonation degree was achieved to minimize the changes to the pore structure. Concrete surface scaling measurements made were 0.063 kg/m2 for B42 and 0.049 kg/m2 for B55, both for the -20 °C/24 hour exposure cycle. The minimum relative dynamic modulus obtained (using fundamental frequency) after 112 freeze-thaw exposure cycles tests was 99.5 %. This indicates that the effect of surface scaling and internal cracking was minimised and not considered to interfere with the transport mechanisms. Complete results of the initial characterization of concrete performance and exposure testing are presented in Ferreira et al. [23, 24]. 3.1 Water uptake In Figure 2 the average water uptake for concrete B42 and B55 are presented, for the concrete subject to both 24 h and 84 hour exposure cycles, and to ponding at 20 °C. Specimens show a greater water uptake during freeze-thaw exposure when compared to only to ponding. This is due to the ice-lens pumping mechanism [15], which in addition to concrete quality, depends on the total number of exposure cycles. In ponding conditions, as a result of the coarser pore structure, the B55 water uptake is greater and requires more time to stabilize when compared to the B42. These curves serve as a reference for comparison of the freeze-thaw exposure uptake. Water uptake values were not corrected for weight of scaled concrete since it was

35



determined to be < 0.7 % of the weight of water uptake.

(a) (b)

Figure 3. Water uptake as a function of exposure for (a) B42 and (b) B55 concretes. The results for the concretes subject to varying 24 hour exposure cycles show that water uptake is higher for B55 concrete than for B42 (approximate upper limit B55 – 2.0 kg/m2; B42 – 1.0 kg/m2), which is expected based on the characteristics of the concrete composition (i.e. w/c ratio) and the results of capillary uptake. For the concretes subject to -20 °C/84 hour exposure cycles, the water uptake is greater than that observed for ponding. When comparing results based on number of cycles to the 24 hour freeze-thaw exposure, a similar uptake is observed for the B42 concrete, and a lower uptake is observed for the B55 (yet higher than ponding uptake). This contribution from the diffusion of non-freezable water from the wet surface via the hardened cement paste to the voids (greater amount in B55) [25]. More cycles are required to exceed the ponding limit for the B55 concrete than the B42. This could be due to the larger pore structure that requires filling, but also because the B55 concrete had greater amount of entrained air. 3.2 Chloride profiles Figures 4, 5 and 6a present the average chlorides profiles of the specimens subject to 24 hour exposure cycles measured after 28, 70 and 112 cycles, respectively. Figure 6b presents the average chloride profiles of the specimens subject to 84 hour exposure cycles measured after 14, 30 and 41 cycles. The chloride profiles reveal clearly the difference in the quality between the concretes. B55 concrete shows deeper chloride profiles than the B42 concrete, and younger concrete samples had shallower chloride profiles than older samples. The chloride profiles reveal minor differences between the ponding and the freeze-thaw profiles. This is an unexpected finding as the water uptake in Figure 2 is clearly larger for the concretes with freeze-thaw exposure than ponding only. It is thought that this water uptake is mainly salt free as it occurs due the micro ice-lens pumping effect. As such, one would expect a dilution of the chloride ions solution, affecting the profiles accordingly. The chloride profiles, however, show no such effect. This confirms the complexity of the interaction of the transport mechanism in action.

36



(a) (b)

Figure 4. Chloride profiles measured in water after 28, 70 and 112 days of exposure cycles for

(a) (b)

Figure 5. Chloride profiles measured in water after 28, 70 and 112 days of exposure cycles for both B55 and B42 concretes, for freeze-thaw cycle until (a) - -

(a) (b)

Figure 6. Chloride profiles measured in water after 28, 70 and 112 days of exposure cycles for both B55 and B42 concretes, for freeze-thaw cycle until (a) - -3.5 day length.

37



The chloride surface concentration of the chloride profiles of both quality concretes shows an increases exposure cycle (i.e. time). The average surface chloride concentration for all

-thaw tests is 0.35

freeze-thaw is slightly higher than that of the specimens subject to ponding. Furthermore, for B55 concrete, a “convection zone” effect is observed. The effects of surface ageing (carbonation, wetting/drying, leaching, etc.) have been minimised as a result of the adopted curing regime and specimen preparation for testing. From the chloride profiles subject to 24 hour exposure cycles (Figures 4-6) it is possible to observe that the inner portion of the chloride profiles depths > 10 mm is where ponding specimens reveal the highest average values. Furthermore, there seems to be no clear distinction between the freeze-thaw chloride profiles as a function of the intensity of the freeze-thaw exposure (i.e. minimum negative temperature). A comparison of chloride profiles after 41 cycles of 84 hour exposure to the 112 cycles of 24 hour exposure show relatively similar chloride profiles despite the large difference in number of exposure cycles. Considering that the uptake of water is proportional to the number of cycles, potentially another transport mechanism is influencing the results. 4. Discussion An analysis of the test procedures and results suggest that the main transport mechanisms contributing to the ingress of water and chlorides are: capillary action (minimised but still present), diffusion (both water vapour and chlorides in the pore solution) and micro ice-lens pumping due to the freeze-thaw loading [3, 14, 17]. Freeze-thaw cycles affect chloride ingress due to micro ice-lens pumping of the brine solution. Test results confirm this, but indicate that no clear difference is noticed in water uptake as a result of the intensity of the freeze-thaw exposure (i.e. minimum negative temperature) suggesting that it does not have a significant influence on the chloride ingress. Consider a hypothetical concrete, subject to pure ponding, with a gradient chloride distribution in the capillary pore system. Consider another identical concrete, but subject to periods of freezing and thawing, and assume that the freezing and thawing result only in localised ice formation, and that no additional water uptake occurs. If the chloride profile of both concretes were to be measured, the concrete subject to ponding would be expected to have a deeper chloride profile as a result of longer time for diffusion at higher temperatures. In addition, the second concrete’s reduced chloride profile may also be expected due to the physical obstacle presented by ice formation in the capillary pore system. Nevertheless, actual test results show that the profiles are similar, despite long periods of freezing, so water uptake due to micro ice-lens pumping has some role. Both concretes start testing in identical conditions, and the test results show that there is significant water up take from the concretes subject to freeze-thaw exposure. If this water up take would include the diluted salt, then we would observe some sort of increase in the chloride profile as a result. This is not observed, and literature indicates that the water migration across the ice lens occurring inside the concrete is mainly salt free. However, ice-lens suction at the surface, especially during thawing, draws chlorides into the concrete. This explains the higher chloride concentration in the surface layer of concrete subject to freeze-

38



thaw, and the lower concentration in chloride in the deeper concrete, in comparison to the ponding concrete. This implies that water uptake due to ice-lens pumping controls transport in the surface layer. This effect is not felt in depth, where a comparison to chloride profiles for ponding profiles show consistently lower amount of chlorides for the same. This does not explain how the chloride profiles of the 84 hour exposure cycles match the other despite fewer cycles and smaller water uptake. Larger pore sizes could have resulted from the prolonger freezing pressures, and some near surface micro-cracking (despite the relatively low scaling degree and high relative dynamic modulus of elasticity). Could this imply there to be other mechanisms influencing chloride transport in concrete? One possible explanation resides in the expulsion of insoluble ions from a liquid volume transforming into ice [26]. Although brine pockets might occur within the ice in the pore network of concrete, due to the long freezing period at -20 C (approximately 60 hours), the ice formation in concrete has time to accommodate extensively to the porous network resulting in the salt brine expulsion from the growing ice crystal interfaces. Assuming the main direction of ice crystal growth to coincide with the direction of the temperature gradient across the concrete specimens, this implies the direction chloride expulsion to follow, i.e., towards the concrete interior. Once the concrete thaws, the ice crystals melt resulting in pockets of clear water and highly concentrate brine, which would be subject to intense localised diffusion as a result of concentration gradients. However, the effect of the freezing of ice crystals in the pores has a greater effect than the diffusion process, resulting in a movement of chlorides inwards from the concrete surface. 5. Conclusion The study of the effect of multiple transport mechanisms are complex because they combine capillary water uptake with diffusion at the same time, and overlaying these with the action of freeze-thaw. This research has looked at the direct relationship between freeze-thaw exposure and chloride ingress for concrete continuously covered with water (which does not reflect the situation for most concrete structures) while attempting to minimise other possible interfering factors such as the surface scaling and internal cracking of concrete due to freeze-thaw loading. The results of this study support the following conclusions:

Concrete specimens subject to freeze-thaw exposure showed a greater water uptake when compared to ponding, indicating that micro ice-lens pumping was active.

It is generally thought that concrete subject to freeze-thaw in the presence of chlorides shows less ingress due to the freeze-thaw loading. However, chloride profiles for ponding and freeze-thaw exposure were similar. Micro ice-lens pumping at the surface is possibly contributing.

The chloride profiles of concrete exposed to freeze-thaw show consistently higher surface chloride concentrations than ponding profiles. This difference may be explained by the contribution of a freeze-thaw pumping mechanism, and possibly by the larger pore sizes that could have resulted from the prolonged freezing pressures, and some near surface micro-cracking (despite the relatively low scaling degree and high relative dynamic modulus of elasticity). At a greater depth from the surface, the chloride profiles of ponding specimens exhibit higher values indicating the dominance of diffusion, which is suppressed in the freeze-thaw exposure chloride profiles.

39



For the concrete subject to 84 freeze-thaw exposure cycles, despite the lower water uptake, the chloride profiles were similar to the other freeze-thaw tests and ponding. Micro ice-lens pumping alone cannot explain this behaviour. A possible explanation is the contribution of the ice formation induced brine expulsion that aids the transport of chloride in the direction of ice growth.

Acknowledgements The authors gratefully acknowledge funding from the TEKES FiDiPro Program grant for the project CSLA - Concrete Service Life Assessment: modelling frost attack degradation in the presence of chlorides (TEKES-40083/12), and from the Fulbright Scholar Program which supported D.A. Lange during his sabbatical collaboration with researchers at the VTT Technical Research Centre of Finland. References [1] Leivo, M., Sistonen, E., Al-Neshawy, F., Piironen, J., Kuosa, H., Holt, E., Koskinen, R.,

Nordqvist, C., ‘Effect of interacted deterioration parameters on service life of concrete structures in cold environments. Laboratory results 2009 – 2010’. VTT Research Report. VTT-R-04799-11, 2011. 57 p.

[2] Kuosa, H., Leivo, M., Holt, E., Ferreira, R.M., ‘Effect of varying surface ageing on frost-salt scaling’. Proc. 6th Int. Conf. Bridge Maintenance, Safety & Management. Italy, Stresa. July 2012. 974-981.

[3] Ferreira, R.M., Kuosa, H., Makkonen, L., ‘Performance & Durability of Concrete in Extreme Cold Environment. CSLA Project – Task 1. Literature Review’. VTT Research Report. VTT-R-073643-12. (2012). 108p.

[4] Kukko, H. ‘Frost effects on the microstructure of high strength concrete, and methods for their analysis’. Technical Research Centre of Finland, VTT Publications 126. 1992. 133p.

[5] Kukko, H. ‘Concrete and its constituents: image analysis in characterizing concrete and its constituents’ 7th EuroSeminar Microscopy applied to Building Mat. TNO. Delft (1999), 521-530.

[6] Kukko, H., Paroll, H. ‘Round Robin tests on concrete frost resistance.’ 1st Int. Wrkshp Resist. of concrete to scaling due to freezing in the presence of deicing Salts, CRIB, Laval, 1993. 263-272.

[7] Kukko, H., Tattari, K., ‘Durability of high strength concrete.’ VTT Publication 808. 1995. 33.

[8] Penttala, V.E., ‘Freezing-induced strains and pressures in wet porous materials and especially in concrete mortars’, Advanced in Cement Based Materials 7 (1998) 8-19.

[9] Fagerlund, G., ‘The international cooperative test of the critical degree of saturation method of assessing the freeze/thaw resistance of concrete’, Mater. Constr. 10 (58) (1977) 230–251.

[10] Fagerlund, G., ‘The required air content of concrete’, Workshop on ‘Mass Energy Transfer and Deterioration of Building Components’, Paris, January 1995.

[11] Fagerlund, G., ‘Predicting the service life of concrete exposed to frost action through

40



modelling of the water absorption process in the air-pore system’, Lund IT., TVBM- 7085, 1994.

[12] Fagerlund, G., ‘Moisture uptake and service life of concrete exposed to frost’, Pro. Int. Conf. on Concrete under Severe Conditions, Sapporo, Japan, Vol. 1, E & FN Spon, Tokyo, Aug. 1995.

[13] Jacobsen, S., ‘Scaling and cracking in unsealed freeze/thaw testing of Portland cement and silica fume concretes’, Thesis report 1995: 101, NTNU, Trondheim,(1995).

[14] Setzer, M.J., Proceedings of the International RILEM Workshop on Resistance of Concrete to Freezing and Thawing with or without Deicing Chemicals, Proceedings, vol. 34, 1997, 157.

[15] Setzer, M.J., ‘Micro-Ice-Lens Formation in Porous Solid’. J.Colloid Interface Sci. 243(1) (2001) 193-201 Marchand, J., Sellevold, E.J., Pigeon, M., ‘The Deicer Salt Scaling Deterioration of Concrete-An Overview’, Am. Concr. Inst. SP vol. 145-1 (1994) 1–46.

[16] Valenza, J.J., Scherer, G.W., ‘A Review of Salt Scaling: II. Mechanisms’. Cem. Concr. Res. 37 (2007) 1022–1034.

[17] Valenza, J.J., Scherer, G.W., ‘A review of salt scaling: I. Phenomenology’. Cem. Concr. Res. 37 (2007), 1007-1021.

[18] Sun, Z., and Scherer, G.W., ‘Measurement and Simulation of Dendritic Growth of Ice in Cement Paste’. Cem. Concr. Res. 40 (2010) 1393–1402.

[19] Li, B., ‘Chloride transport in concrete under frost action – An experimental study’. Master Thesis. TCH. Göteborg. 2009, 156p.

[20] CEN/TS 12390-9:2006. Testing hardened concrete - Part 9: Freeze-thaw resistance – Scaling. CEN. 24p.

[21] Vesikari, E., Ferreira, R.M. 2011. Frost Deterioration Process and Interaction with Carbonation and Chloride Penetration. In DuraInt Project. VTT. Research Report VTT-R-02782-11. 45.

[22] CEN/TR 15177 (2006) Testing the freeze-thaw resistance of concrete. Internal structural damage. CEN. 34p.

[23] Ferreira, M., Kuosa, H., Leivo, M. (2014a) ‘Characterization of testing concrete. CSLA Project – Task 5. Testing’. VTT Research Report. VTT-R-01620-14

[24] Ferreira, M., Kuosa, H., Leivo, M. (2014b) ‘Study of the effect of freeze-thaw on chloride ingress. CSLA Project – Task 6. Test setup and results’. VTT Research Report. VTT-R-01621-14.

[25] Jacobsen, S., (2005) Calculating liquid transport into high-performance concrete during wet freeze/thaw, Cement and Concrete Research, Volume 35, Issue 2, 213-219, ISSN 0008-8846,

[26] Makkonen, L. (2012) Ice adhesion – Theory, measurements and countermeasures. J. Adhesion Sci. Technology. 26, 4-5, 413-445.

41



FROST DAMAGE OF CONCRETE SUBJECT TO CONFINEMENT Marianne Tange Hasholt (1) (1) Technical University of Denmark, Lyngby, Denmark Abstract When internal frost damage is observed in real concrete structures, the usual pattern is cracks with a preferred orientation parallel to the exposed surface. When exposing concrete with poor frost resistance to a standardised freeze/thaw test in the laboratory, the orientations of the resulting cracks are more or less random. The present study is an experimental study, which aims at investigating the influence of confinement during freeze/thaw action on the developed crack pattern. Confinement is established by mounting hose clamps on cylindrical test specimens, using similar test specimens without hose clamps as reference. The results show that confinement can change the outcome of a freeze/thaw test as regards extent of internal cracking, crack orientations, and amount of surface scaling. Thus it seems likely that the difference in confinement (and therefore also in stress state) can explain the different crack patterns observed in the field and in the laboratory. 1. Introduction The study presented in the following emanated from a discussion at a meeting in 2014 in the Danish Society for Microscopy of Building Materials (FMIB). In Denmark, the reference method of CEN/TS 12390-9 [1] is the most widely used method for testing concrete frost resistance. According to this test method, the frost resistance is quantified by measurement of surface scaling during 56 standardised freeze/thaw cycles for concrete specimens exposed to 3% NaCl solution on the test surface. It is not part of the test method to inspect the inner frost damage in the form of cracking. However, if test specimens after testing are impregnated with fluorescent epoxy, it is possible to observe the crack patterns. For concrete that is not frost resistant, this normally leads to a diffuse crack pattern with no distinct crack orientation.

42



If concrete from the field (walls, pavements, etc.) suffers from inner frost damage due to natural freeze/thaw exposure, and this concrete in a similar way is treated with fluorescent epoxy, the usual pattern is cracks with a preferred orientation parallel to the exposed surface. Therefore, the discussion at the FMIB meeting concerned to what extent the laboratory test represents what takes place in concrete that is being exposed to a natural climate with freezing temperatures. The opinions could more or less be summarised in the following two conflicting standpoints: A. The difference in crack patterns indicates that it may be different mechanisms that trigger

internal frost damage in the laboratory and in the field. If it is not the same mechanism that causes damage in the laboratory as in the field, it puts strong limitations on what can actually be deduced from laboratory results about the frost resistance of concrete in service.

B. Ice formation is an expansive reaction. For other expansive reactions in concrete it is known that the crack pattern depends on the initial stress situation of the concrete. Alkali-silica reaction (ASR) is an example of this [2]. For concrete elements that are free to move in all directions, ASR leads to randomly orientated cracks, also known as map cracking. If the concrete is restrained in one direction, e.g. a loadbearing bridge column or a pre-stressed girder, the expansion due to ASR takes place in other directions, and the orientation of the resulting cracks is determined by the confining stress direction. Thus, it is expected that crack patterns from frost damage are different for specimens tested in the laboratory and in the field, as the small test specimens are unrestrained during frost action, whereas the concrete in field typically experiences some kind of confinement. The difference in crack patterns cannot be interpreted as a difference in mechanism leading to frost damage.

The aim of the present study is to investigate if the confinement during freeze/thaw action influences the developed crack pattern. The study is carried out as an experimental study, where concrete specimens with and without confinement are subject to accelerated freeze/thaw testing in the laboratory. 2. Materials and methods When choosing a concrete composition for the present study, there were two things to consider. On one hand, the concrete for testing should not be frost resistant; if no damage evolves during the tests, it will not be possible to examine if there is a difference in crack patterns. On the other hand, the concrete should not be too frost susceptible, because if the concrete completely disintegrates during testing, it will not be possible to investigate the crack patterns either. Air void structure and w/c ratio are two important factors for concrete frost resistance [3]. It was decided to test concrete with w/c ratio 0.45 without air entrainment. A similar mix had been used several times for freeze/thaw testing in our laboratory. Here, the omission of air entrainment resulted in concrete with poor frost resistance, but the concrete was still coherent

43



after 56 freeze/thaw cycles. Moreover, in Denmark, concrete with w/c ratio 0.45 is a realistic concrete for outdoor structures; 0.45 is the maximum w/c ratio for concrete in environmental class A (aggressive environment), which includes combined salt and frost exposure [4]. The mix design is shown in table 1: Table 1: Mix design for concrete (w/c = 0.45). Aggregates are saturated and surface dry. Constituent Type Density

[kg/m³] Mass

[kg/m³ concrete] Cement CEM I 52.5 N 3160 475 Water Tap water 1000 214 Sand Sea dredged sand (class E) 2640 841 Coarse aggregate, 4-8 mm Crushed granite (class E) 2710 168 Coarse aggregate, 8-11 mm Crushed granite (class E) 2720 673 The concrete was mixed in a pan mixer with a batch size of 45 l. The air content in the fresh concrete was 0.7% (measured with a pressure-meter). This corresponded well with the expected natural air content of approximately 1%. Three Ø150 x 300 mm concrete cylinders were cast (labelled I, II, and III). From time of casting to time of testing, the curing conditions followed the curing regime of the reference method in [1]. 21 days after casting, 2 cylindrical discs, 50 mm thick, were cut from the middle part of each concrete cylinder. For each disc, a rubber sleeve was glued on the disc perimeter to make it possible to establish a liquid reservoir on the top of the specimen. A rubber sheet was glued to the bottom of the disc. The discs were labelled REF and CLAMP, respectively: REF: The specimen was externally unrestrained during freeze/thaw testing, as is the

normal test condition. This specimen served as reference of the experiment. CLAMP: 2 hose clamps were placed around a concrete disc, each clamp being 25 mm

wide. In this way the concrete was subjected to moderate external compression, and concrete expansion became restricted in the radial direction, see figure 1 (left).

44



Figure 1: Specimen with hose clamps before (left) and after (right) all surfaces except the test surface had been covered with thermal insulation. The clamps were mounted 28 days after casting, just before the test surfaces were covered with 3 mm de-ionised water. The clamps were tightened by hand with a hex key wrench. The compression enforced by the hose clamps on the concrete specimens was not measured. It was anticipated that the resulting compressive stress in the concrete was small relative to the concrete compressive strength. 31 days after casting, all surfaces except the test surface were covered with thermal insulation, and the de-ionised water was replaced with 3% NaCl solution, before the specimens were placed in a freezing cabinet. The clamps were also covered with insulating material to prevent the metal clamps from forming thermal bridges that would spoil the one-dimensional heat transport through the specimen, see figure 1 (right). In the freezing cabinet, the specimens were exposed to repeated freezing and thawing with temperature cycles complying with [1]. Scaling was collected at intervals as specified in the test standard. After 33 freeze/thaw cycles, two specimens, one from each of the test series REF and CLAMP, were removed from the freezing cabinet to prepare epoxy impregnated plane sections. 3. Results Results from the freeze/thaw scaling tests are shown in figure 2.

45



Figure 2: Development in accumulated amount of scaled material. Specimens REF-II and CLAMP-III were removed from the freezing chamber after 33 freeze/thaw cycles. After 28 cycles, the results become more uncertain. After removal of specimens for epoxy impregnation, the areas of test surface for each series REF and CLAMP no longer fulfil the Danish requirement for minimum test surface for the test (minimum 42,000 mm²). Moreover, problems with leakage were discovered for CLAMP-I (detected when collecting scaling after 28 cycles), and REF-III (detected after 42 cycles). This may explain why the accumulated scaling for CLAMP-I is the lowest of the CLAMP series, and REF III shows the lowest scaling of the REF series; if the concrete surface is not covered with NaCl solution in every freeze/thaw cycle, it reduces the amount of scaling. Figure 3 shows the epoxy impregnated plane sections for the specimens REF-II and CLAMP-III. The pictures are scaled to the same size. The reason why the height of CLAMP-III is slightly larger than the height of REF-II is because several millimetres of the test surface of REF-II had scaled off. Figure 3 also shows analyses of crack orientation made with the method described in [5]. REF-II is cracked throughout the specimen, but for CLAMP-III, a large area at the centre of the specimen is uncracked. To make the analyses comparable, they are for both specimens based on the upper 300 pixel rows (approximately 12 mm), corresponding to the cracked area of CLAMP-III beneath the test surface. The orientation 0° / 180° corresponds to cracks parallel to the test surface.

46



REF-II CLAMP-III

test surface

test surface_

Figure 3: Top: Epoxy impregnated plane sections under UV light (width of each plane section is 150 mm). Bottom: Diagrams showing distribution of crack orientations. Crack orientation 0° / 180° corresponds to cracks parallel to the test surface. Horizontal lines at 1/180 = 0.56% correspond to random crack orientation. 4. Discussion Though the REF-II specimen had not fallen to pieces during the 33 freeze/thaw cycles, the epoxy impregnated plane section showed that it was completely cracked from top to bottom. Based on a purely visual inspection, the crack orientation was judged to be “random”. However, the image analysis (figure 3, left) shows that there tend to be more cracks parallel to the test surface than perpendicular to the surface. Some of the cracks follow the interfacial transition zones of the coarse aggregates, so one possible explanation is that this outcome is due to the casting direction. The coarse aggregate was crushed granite, where some of the particles are flaked; if they systematically take bearing in a certain direction, e.g. when the moulds are vibrated during casting, this may induce a systematic trend in crack orientation. Another possible explanation may be the geometry of the test specimens. The REF specimens are not subject to an outer constraint on specimen movement. But the geometry where the specimen is wider than it is tall may restrict movements more in one direction than in others. When comparing the observed distribution of crack orientations for REF and CLAMP, there are more surface parallel cracks and less cracks perpendicular to the surface for the CLAMP

47



specimen than for the REF specimen. If even the very moderate confinement that is achieved by the manual tightening of the hose clamps has an effect on the resulting crack orientation, this supports hypothesis B in preference to hypothesis A. However, the most striking difference between the two specimens is not the difference in crack orientation but the extent of damage. The CLAMP specimen is not as heavily damaged as the REF specimen. The CLAMP specimen is cracked up to 10-15 mm from all free surfaces of the specimen, but further away from the surfaces, no cracks are seen. This is probably related to the availability of free water for ice formation inside the concrete. Cracks perpendicular to the test surface contribute to water transport into the concrete, whereas cracks parallel to the surface do not. Moreover, it may also play a part that for the REF specimen, there are more favourable conditions for widening the cracks during frost action, so they can transport more water, because the movement of the specimen is unrestricted. According to [6], mechanical stress has only a small effect on concrete permeability, as long as the stress level is less than 40% of the concrete strength. At higher stress levels, the mechanical action may lead to cracking, thereby increasing the permeability several orders of magnitude. The crack width also has a pronounced effect. During the freeze/thaw test, surface scaling was regularly collected, though surface scaling was not the primary focus of the study, as hypotheses A and B do not relate to scaling. The collection of surface scaling was mainly seen as a quality control to identify if one specimen behaved very different from the others e.g. due to casting errors. The results presented in figure 2 show a significant difference in scaling for specimens with and without hose clamps. In the first part of the test period (collection after 7 and 14 freeze/thaw cycles), the amount of scaling of the CLAMP specimens is 20-25% less than the scaling of the REF specimens. It is likely that the confinement imposed by the hose clamps is reduced over time, because the clamps were not tightened during the freeze/thaw test period. If the confinement is relieved over time, the effect of the hose clamps should also level off, but this does not seem to be the case. Actually, the scaling of the REF specimens seems to accelerate, whereas for the CLAMP specimens, the amount of scaling per unit of time is almost constant, and after completion of the 56 freeze/thaw cycles, the cumulated amount of scaling for the CLAMP specimens are less than 50% of the cumulated amount for the REF specimens. The difference in scaling for specimens with and without hose clamps is surprising, and there is no obvious explanation for it. It may be because surface scaling is not a pure surface phenomenon; the surface scaling may depend on e.g. the general moisture state of the concrete, and it is likely that the REF specimens are more wet than the CLAMP specimens, as outlined above. If the experiment is repeated in the future, the hose clamps should be tightened with a torque wrench, or other measures should be taken to ensure better control of the stress situation. The stress situation may change during the experiment. If for example the concrete expands due to frost action, the clamping stress will increase, and the use of a torque wrench cannot prevent this. However, using a torque wrench set at a certain torque would ensure a more uniform start situation for all specimens, and the tightness of the clamps can be adjusted during the test period.

48



It is thought provoking that a small change such as mounting two hose clamps on a test specimen and thereby introducing a modest confinement can change the test result 25% or more. Luckily, it is normally on the safe side to perform freeze/thaw testing on unconfined specimens, as most concrete in real structures is placed under compression, which this study points to will reduce the extent of damage. In recent years, it has become popular to study combinations of load, for example the effect of freezing and thawing on fracture energy [7], shear resistance [8], pre-stress loss [9], and fatigue [10] (and other examples can also be found in [11-13]). The present study urges caution when interpreting and extrapolating these results, especially if the loads are not imposed simultaneously. If the concrete is exposed to freezing and thawing before it is subjected to mechanical load, the result probably differs from what would have been obtained, if freeze/thaw action had taken place at the same time as the mechanical load. 5. Conclusion In the present study, hose clamps were mounted on 3 of 6 cylindrical specimens prior to a standardised freeze/thaw test. The hose clamps introduced compressive stress in the radial direction (the stress level being small relative to the compressive strength of the concrete) and confined specimen expansion during the test. The results show that confinement can change the outcome of a freeze/thaw test. The outcome is changed as regards extent of internal cracking, crack orientations, and amount of surface scaling. The results can explain why the crack patterns of concrete specimens tested in the laboratory differ from crack patterns that can be observed in structures in the field made of concrete that is not frost resistant. The difference in crack patterns does not arise from different frost damage mechanisms in the two situations. The difference is probably due to different stress situations, because concrete in a structure is confined during frost action whereas concrete tested in the laboratory is virtually unconfined. Acknowledgement The experimental work presented in this paper was carried out by ERASMUS student Adrien Faugeras, École d’ingénieurs de l’université de Nantes, who visited Technical University of Denmark for a 3 month internship during the summer 2014. The thorough and hard work carried out by Adrien Faugeras is much appreciated. References [1] CEN/TS 12390-9, Testing hardened concrete – Part 9: Freeze-thaw resistance – Scaling,

Danish Standards Association, Denmark (2006) [2] Thomas, M.D.A., et al., Alkali-silica reactivity field identification handbook, FHWA-

report FHWA-HIF-12-022 (2011)

49



[3] Powers, T.C. Freezing effects in concrete, proceedings on Durability of Concrete, ACI SP-47, American Concrete Institute, Detroit (1975), 1-11

[4] DS 2426: Concrete - Materials - Rules for application of EN 206-1 in Denmark (in Danish), Danish Standards Association, Denmark (2011)

[5] Andreassen, E. N. et al, The use of image analysis to quantify the orientation of cracks in concrete, International RILEM Conference on Materials, Systems and Structures in Civil Engineering, Conference segment on Frost Action, Denmark (2016)

[6] Hoseini, M. et al, The effect of mechanical stress on permeability of concrete: A review, Cem Concr Comp (2009), 213-220

[7] Kosior-Kazberuk, M., Variations in fracture energy of concrete subjected to cyclic freezing and thawing, Archives of Civil and Mechanical Engineering (2013), 254-259

[8] Martin, A., and Rivard, P., Effects of freezing and thawing cycles on the shear resistance of concrete lift joints, Can J Civ Eng (2012), 1089-1099

[9] Cao, D.-F. et al., Evaluation of prestress losses in prestressed concrete specimens subjected to freeze-thaw cycles, Structure and Infrastructure Engineering (2016), 159-170

[10] Qiao, Y. et al., Damage process of concrete subjected to coupling fatigue load and freeze/thaw cycles, Constr Build Mater (2015), 806-811

[11] Hanjari, K. Z. et al., Modelling the structural behavior of frost-damaged reinforced concrete structures, Structure and Infrastructure Engineering (2013), 416-431

[12] Berto, L. et al., Constitutive model of concrete damaged by freeze-thaw action for evaluation of structural performance of RC elements, Constr Build Mater (2015), 559-569

[13] Shang, H. et al., Bond behaviour between steel bar and recycled aggregate concrete after freeze-thaw cycles, Cold Regions Science and Technology (2015), 38-44

51



THE SALT-FROST RESISTANCE OF CONCRETE WITH SUPPLEMENTARY CEMENTITIOUS MATERIALS (SCM)

Elisabeth Helsing (1), Peter Utgenannt (1) (1) Swedish Cement and Concrete Research Institute, CBI, Borås, Sweden Abstract The experience in Sweden with the slab method (reference method in CEN/TS 12390-9) for determining the salt-frost resistance of concrete is that it is successful in predicting the salt-frost resistance of Portland cement concrete. However, doubts have been raised whether the same can be said when used on concrete with supplementary cementitious materials SCM. Results from a research project which is carried out in order to validate the method for concrete with SCM or propose appropriate adjustments to it is presented in this paper. Six different variations of the pre-conditionings in the slab test are tested, the age of the specimens is increased, the time in 65 % RH is prolonged and exposure to 1 % CO2-environment is included. The test programme covers mixes with up to 35 % fly-ash and 65% slag and two different Portland cements, all with a water-to-binder ratio =0.45. For air-entrained concrete with SCM both prolonging the exposure to 65 % RH and exposure to CO2 diminishes the salt-frost resistance. The influence increases with increasing amount of SCM. However, the type of Portland cement also has a certain influence, and the type which has the highest salt-frost resistance when used alone exhibits a lower salt-frost resistance when combined with SCM than the other cement. 1. Background and purpose In Sweden requirement levels with regard to the salt-frost resistance of concrete is set with the use of the slab method in CEN/TS 12390-9[1]. This method was originally developed for Portland cement (PC) concrete, and according to more than 20 years of experience, has proven to be a useful tool to distinguish PC-concretes with a good salt-frost resistance from those which perform poorly. However, pure PC concrete is becoming less frequently used, since the use of SCM to replace PC in concrete is steadily increasing. The question then arises, whether this method for evaluating the salt-frost resistance also gives results relevant for concretes with SCM. Some test results have shown that for instance aging and CO2 exposure influences the salt-frost resistance of concrete containing ground granulated blast furnace slag (GGBS) to a much larger degree than PC concrete [2], [3] & [4]. A project was initiated in order to evaluate if the slab method gives reliable results also with concretes containing primarily GGBS and Fly Ash (FA), or whether changes to the test procedures are

52



needed. In the project the method and five modified versions of it has been used on a number of concrete mixes, containing different amounts of GGBS and FA. The salt-frost tests are combined with a study of structural and chemical changes in the surface layer of the specimens prior to and after the salt-frost exposure. The test results are to be compared to the deterioration of these mixes when subjected to a natural road environment where de-icing salts are used. However, as the project is still ongoing, this paper only includes results from three variations of the salt-frost test procedure and only results from selected test methods, not from the entire test programme. 2. Experimentals

2.1 Materials Since this project is mainly aimed at validating the use of the method for exposure class XF4, i.e. marine environment and a road environment where de-icing salts are used, the main Portland cement used in this study is one commonly used in Swedish bridges, a CEM I 42.5 N - SR3 with a maximum alkali content of 0.6 % (trade name Cementa Anläggningscement Degerhamn), in the following named "Infra". In some mixes an ordinary CEM I, 52.5 R (trade name Cementa Snabbhårdnande Slite) was used, here called "Rapid". The GGBS was Merit 5000 from Merox conforming to EN 15167-1 [5] and the FA came from Norcem and was conforming to EN 450-1, Category A [6]. The coarse aggregate consisted of 8/16 crushed granite, and the fine aggregate of natural graded deposits 0/8. The superplasticizer used was Sika Visco Crete RMC-520PB and the air entraining agent (AEA) was Sika AirPRO. According to the Swedish application standard to the European concrete standard [7] the requirement on maximum water to cement ratio in exposure class XF4 is 0.45. This was therefore chosen as water to binder ratio for all the mixes, without taking into account any k-factors for the additions. The proportions between cement and additions were chosen to reflect the limits between different cement types according to EN 197-1[8]. In all cases the PC and the GGBS or FA were mixed in the concrete mixer. The total binder content in all mixes was 400 kg/m3. An air content of 5.0 % was aimed at for mixes with entrained air. Three mixes without air entraining agent were also included. A superplastizising agent was used in all mixes to obtain a slump around 150 mm. The test specimens, 50 mm slabs sawed out of 150 mm cubes, were prepared according to CEN/TS 12390-9. Four parallel specimens were used for each mix. Data for the mixes are given in Table 1. The 28-day strength was measured on water cured specimens.

2.2 Salt-frost test As a basis the test procedure according to the reference method in CEN/TS 12390-9 was used, in the following called the basic method. The first variation of this test was that after the 7 days in 65 % RH of the sawed specimens, they were subjected to 1 % CO2 and 65 % RH for an additional 7 days, here referred to as the CO2 exposure. The evaporation rate in the CO2-chamber was about 90 g/(m2 h). The CO2 -level in the climate chamber used for 65 % RH

53



only was normally around 0.04 % and the evaporation rate around 40 g/(m2 h). The second variation of the procedures was that the period in 65 % RH after sawing was prolonged from 7 days to 21 days, here referred to as prolonged conditioning. All specimens, irrespective of conditioning procedure, were sawed at an age of 21 days. Table 1: Binder composition, air content and average 28-day strength of the mixes.

Mix PC % of

binder

GGBS % of

binder

FA % of

binder

AEA PC type Air content in

fresh concrete

28 day strength

MPa

1 100 Yes "Infra" 5.9 41.3 2 100 No "Infra" 2.01) 68.4 3 80 20 Yes "Infra" 5.8 43.0

4 80 20 No "Infra" 2.2 54.1 5 80 20 Yes "Infra" 5.4 36.0 6 80 20 No "Infra" 2.3 45.6 7 65 35 Yes "Infra" 5.5 44.0 8 65 35 Yes "Infra" 5.0 30.0 9 100 Yes "Rapid" 4.92) 59.3

10 65 35 Yes "Rapid" 5.2 50.0 11 65 35 Yes "Rapid" 4.5 47.0 12 35 65 Yes "Rapid" 5.5 41.0 1) Density measurements indicate that the air content in this mix is around 1%. 2) Pore structure analysis and density measurements indicate that the air content in this mix

is around 3%.

Figure 1: Schematic representation of the three test procedures (numbers = weeks)

1 2 3 4 5 6 7 8 - 20 21 22

Basic CO2 Prol. Water curing Exposure to 1 % CO2

65 % RH, 20 °C Saturation of the surface during 3 days Sawing Salt-frost cycling

54



After the conditioning, the surface of the specimens was water saturated for three days. Subsequently the specimens without entrained air were then exposed to 56 frost-cycles with 3 % NaCl-solution, which is the normal number of cycles according to [1]. All specimens with entrained air were exposed to 112 frost-cycles, in order to make sure that no sudden changes in the behaviour occur after 56 cycles. The general out-line of the preparation and the test procedures for the 3 variations, from casting to the end of the frost cycling is showed in Figure 1.

3. Results

When evaluating the salt-frost resistance in the following the Swedish traditional classification [9] is used, which in a simplified form can be described as follows:

Very good salt-frost resistance: scaling at 56 days (m56) < 0.1 kg/m2

Good salt-frost resistance: m56 < 0.2 kg/m2 or m56 < 0.5 kg/m2 and m56/ m28 < 2 or m112 < 0.5 kg/m2 Acceptable salt-frost resistance: m56 < 1 kg/m2 and m56/ m28 < 2 or m112 < 1 kg/m2 Not acceptable salt-frost resistance: when the criteria above are not met

3.1 Scaling of concretes without entrained air Figure 2 shows the scaling of all the mixes without air entraining agent when the three conditionings were used.

Figure 2: Accumulated scaling after 56 cycles of the mixes without entrained air.

For all the mixes without air entraining agents and for all the three pre-conditioning procedures used, the salt-frost resistance is far from satisfactory. In order to have a good performance the total scaling after 56 days should not exceed 0.5 kg/m2. The ranking of the

55



mixes is the same irrespective of which pre-conditioning is used; the mix with 20 % GGBS has the best salt-frost resistance, then comes the pure PC concrete and the one with the worst performance is always the mix with 20 % FA. Nevertheless, the different mixes react differently on CO2 exposure and prolonged conditioning. However, the scatter between parallel specimens is large at these high levels of scaling. Smaller variations in the result should therefore be ignored. The only two observations which appears to be more than marginal is that the PC concrete enhances its salt-frost resistance somewhat when exposed to CO2 and that prolonged conditioning causes a substantial increase of scaling of the FA concrete. 3.2 Scaling of concretes with air-entraining agent - basic method - influence of binder composition The accumulated scaling after 56 and 112 frost cycles of the mixes with entrained air when the basic method is used can be found in Figure 3.

Figure 3: Accumulated scaling obtained with the basic method for the mixes with entrained air. Numbers over bars are the scaling after 56 cycles.

From this figure it can be deduced that 65 % of GGBS increases the scaling considerably. There is a difference between if the GGBS is combined with the "Infra" cement or the "Rapid" cement. When the proportion of GGBS in the combination with "Infra" cement is 20 % there is no increase in scaling, but at 35 % GGBS the scaling is about four times higher than without GGBS. Nevertheless, since this cement has a very low scaling when used alone the salt-frost-resistance of this binder combination can still be classified as good (scaling < 0.5 kg/m2). The "Rapid" cement shows a scaling a little bit higher than the "Infra" cement when used alone. This may however be due to the lower air-content of "Rapid" mix (around 3 %). But up

56



to 35 % GGBS can be used with this cement without increasing the scaling more than marginally. When it comes to replacement of cement with FA, the influence differs from what happens with GGBS. Up to 20 % FA can be used without more than marginal changes of the scaling. When the amount is 35 % the increase in scaling become dramatic, for both cement types, and is far beyond what is acceptable for a salt-frost resistant concrete.

3.3 Influence of CO2 exposure and prolonged conditioning on the scaling of air entrained concretes The scaling of the pure cement concretes with the three varied test procedures are shown in Figure 4 (Note the small vertical scale!). The scaling is approximately doubled after CO2 exposure for both cements. Prolonged conditioning before the frost-cycling starts have increased the scaling somewhat for the "Rapid" cement. For the "Infra" cement the influence is marginal. However, in all cases the mixes will be classified as having a very good or good salt-frost performance.

Figure 4: Accumulated scaling of the pure cement concretes obtained with the basic method, the test with prolonged conditioning and the one with exposure to 1 % CO2.

The scaling after cycles of the mixes containing GGBS and FA obtained with the basic method, after CO2 exposure and after prolonged conditioning are given in Figure 5. From this figure it can be concluded that addition of a CO2 exposure to the method results in substantially larger scaling when GGBS is present in the mix. For all these mixes the accumulated scaling at 112 days is increased by a factor between 5 and 7.5 when they have been exposed to CO2 compared to what is obtained with the basic method. When it comes to the FA concretes, CO2 exposure only affects the scaling marginally, and sometimes even positively. It must though be pointed out that the higher evaporation rate in the CO2 chamber may also have had an influence of these results. The prolonged conditioning before the frost cycling take place, increases the accumulated scaling after 112 cycles of all the mixes with GGBS with between 30 % and 60 %. The results

0

0,1

0,2

0,3

0,4

0,5

0 14 28 42 56 70 84 98 112

Scal

ing,

kg/

m2

Cycles

Infra Basic

Infra CO2

Infra Prol.

Rapid Basic

Rapid CO2

Rapid Prol.

57



obtained on the mixes with FA are a bit erratic and not easily interpreted. On the mixes with the "Infra" cement the scaling is increased, but on the mix with the "Rapid" cement the scaling is decreased with about 40 %. These results need to be looked into more thoroughly. However, all the mixes with only 20 % GGBS or FA show a good salt-frost performance, even after CO2 exposure and prolonged conditioning.

Figure 5: Accumulated scaling after 112 cycles for the mixes containing GGBS and FA, obtained with the three varied test procedures. 4 Discussion

4.1 Why is not the efficiency factors, k-values, of the GGBS and FA used? According to the concrete standard EN 206 [10] GGBS and FA are treated differently if they are incorporated in the cement or if they are used as additions added in the concrete mixer. When used as additions their contribution to the durability parameter, the effective water-to-cement ratio, should be reduced by a k-factor. When incorporated in the cement they are regarded as equivalent to the PC, without reduction. Since the k-values may differ for various reasons it was decided to disregard the k-value in this study, even though the GGBS and FA were used as additions. One should then be aware of that the mixes with GGBS and FA do not formally meet the Swedish requirement on effective water-to-cement ratio for exposure class XF4 (0.45). However if factory made blended cement with the same proportions were used, the requirement would be met. The k-value, which is used for durability requirements, is normally based on the contribution of the GGBS and FA to the compressive strength. A consequence of not using k-values would then be that the 28 day compressive strength of the mixes with GGBS and FA should be lower than in the ones with only PC. However, the strength of the GGBS mixes with the

58



"Infra" cement is quite comparable to the strength of mix 1. In Table 1 it can be seen that the 28 day strength of the FA mixes combined with the "Infra" cement is though clearly below the strength of mix 1. Since the air content of mix 9 is too low it is not possible to compare the strength of mixes with the "Rapid" cement. 4.2 Influence of GGBS and FA on the salt-frost resistance - Basic method When no air entraining agent is used, incorporation of 20 % GGBS enhances the salt-frost resistance and 20 % FA decreases the salt-frost resistance. When the mixes contain entrained air this image is changed. If up to 20 % of GGBS and FA are used with both types of PC-cements used in this study the salt-frost resistance is hardly influenced at all. When 35 % of the cement is replaced by GGBS there is a difference in scaling depending on the type of cement. With the "Rapid" cement the scaling is only increased marginally while it is doubled with the "Infra" cement. When used alone the "Infra" cement performs better than the "Rapid" cement. However, with 35 % GGBS it is still possible to obtain results with the basic method indicating a good salt-frost resistance with both cements. With 65 % GGBS the scaling is increased with a factor around 8-10, but the results will still be classified as acceptable. When FA is used the influence of cement type is small. There is hardly any difference at all in scaling if the binder contains 20 % FA or not. If however 35 % FA is used the increase in scaling is dramatic and is far beyond any acceptance limits. 4.3 Influence of CO2 exposure and prolonged conditioning Also when it comes to exposure to CO2 prior to the salt-frost cycling, the effect on air entrained concrete is not the same as on concrete without entrained air. Without entrained air the largest effect of CO2 exposure is a positive effect on the scaling of the PC concrete. The effect on mixes with GGBS and FA is almost none. However if the mixes contain entrained air the CO2 exposure has a large effect (factor 5 - 9) on GGBS concrete. The effect on FA concrete is small or none. Prolonged conditioning does not have any significant effect on the scaling if the FA content does not exceed 20 % and if the GGBS content does not exceed 35 %. The development of the structural properties and the strength with time is altered when some of the PC is replaced with FA and GGBS, especially at early ages (< 28 days), while at later ages the strengths are often comparable or even higher. In this study this has been observed on the FA mixes, which all have rather low early strength. It has therefore been argued that the salt-frost test, which is carried out on rather young concrete, is disadvantageous for mixes with GGBS and FA, and does not reflect the true long term salt-frost resistance of such concrete. In order to evaluate this effect, the three remaining conditioning procedures within the project consist of delaying the start of the salt-frost cycling to around 90 days of age.

59



4.4 What does the differing influence on air entrained concrete and concrete without entrained air indicate? In all the results it was observed that the influence of the tested parameters differs depending on if the concrete has entrained air or not. In concrete the entrained air is present as discrete fine air-filled bubbles embedded in the paste phase of the concrete. Replacement of PC with GGBS or FA changes the chemical structure of the hydrated phases and the gel-capillary pore-structure of the paste phase. The same can be said about CO2 exposure, which changes the chemistry, the gel-capillary pore structure and the freezable water content [11], [12]. A prolonged exposure to 65 % RH may also induce coarsening of the capillary pore structure in the exposed surface. Neither of these affects primarily the entrained air pore system. However, the chemical changes and physical changes in the gel- capillary pore system have a significant effect on the capability of the air pore system to provide an adequate salt-frost resistance. Thus, a certain parameter may have a positive influence on the salt-frost resistance of the paste phase but a detrimental effect on the capability of the air pore system to provide protection against salt-frost damage. Since the contribution of the air pores to the salt-frost resistance is much greater than the contribution of the properties of the gel-capillary pore system at the water-to-binder ratios and air contents used in this study, the influence on the capillary pore structure will be totally overshadowed by the detrimental effect of the inadequate air pore system. The question to be answered then becomes: What is the explanation to the changing effect of air pores induced by the changes in the chemical structure of the hardened cement paste and the alterations of the gel-capillary pore structure? Some clarifications on this point will hopefully be the result of the complementing tests in the project. 5 Conclusions The following conclusions are drawn from testing the salt-frost resistance in the laboratory: The influence the replacement of PC with GGBS and FA, of CO2 exposure and of

prolonged conditioning on the salt-frost resistance differs between air entrained concretes and concretes without entrained air, when tested with the slab method according to CEN/TS 12390-9.

20 % of any PC may be replaced by GGBS or FA in concrete with w/c-ratio =0.45 and 5 % air with maintained salt-frost resistance. Replacement of PC with 35 % FA in such a concrete leads to a very poor salt-frost resistance when tested according to CEN/TS 12390-9

The cement type has an influence on the salt-frost performance of a concrete with GGBS but not on a concrete with FA.

Exposure to CO2 prior to salt-frost testing has a large negative effect on the salt-frost resistance of air entrained concrete with higher amounts of GGBS, while it has no significant effect on a concrete with FA.

Concretes with higher amount of FA or GGBS are sensitive to prolonged conditioning at 65 % RH prior to the salt-frost cycling.

60



The correlation of these results to what is obtained in reality will be investigated on field-exposed samples of the same mixes. 6 Acknowledgement This project is financed by the Swedish Transport Administration within the BBT-programme.

References [1] CEN/TS 12390-9, Testing hardened concrete - Part 9: Freeze-thaw resistance - Scaling,

European Committee for Standardization, CEN, (2005) [2] Utgenannt, P., The influence of aging on the salt-frost resistance of concrete, Report

TVBM-021, Division of Building Materials, Lund Institute of Technology (2004) [3] Gunter, M. et.al., Effect of curing and type of cement on the resistance of concrete to

freezing in salt solutions, Concrete Durability, American Concrete Institute, ACI SP-100, pp 877-899, Detroit, USA (12987)

[4] Stark, J. and Ludwig, H.-M., Freeze-thaw and freeze-deicing salt resistance of concrete containing cement rich in granulated blast-furnace slag, ACI Materials Journal, Vol. 94, No. 1 (1997)

[5] EN 15167-1:2006 Ground granulated blast furnace slag for use in concrete, mortar and grout - Part 1: Definitions, specifications and conformity criteria, European Committee for Standardization, CEN, (2006)

[6] EN 450-1:2012 Fly ash for concrete - Part 1: Definitions, specifications and conformity criteria, European Committee for Standardization, CEN, (2012)

[7] SS 137003:2015 Concrete - Application of EN 206 in Sweden, Swedish Standards Institute, SIS, (2015)

[8] EN 197-1:2011 Cement - Part 1: Definitions, specifications and conformity criteria for common cements, European Committee for Standardization, CEN, (2011)

[9] SS 137244:2005 - Concrete testing - Hardened concrete - Scaling at freezing, Swedish Standards Institute, SIS, (2005)

[10] EN 206:2013, Concrete - Specification, performance, production and conformity, European Committee for Standardization, CEN, (2013)

[11] Bier, T. A., Influence of type of cement and curing on carbonation progress and pore structure of hydrated cement paste, Proceedings of the Materials Research Society symposium - Microstructural Development During Hydration of Cement, vol 85., pp 123-134. Boston, USA, (1987)

[12] Parrott, L. J., Variations of water absorption rate and porosity with the depth from an exposed concrete surface; Effects of exposure conditions and cement type. Cement and Concrete Research, vol. 22, No. 6 pp. 1077-1088. (1992)

61

FOAM INDEX MEASUREMENTS ON MIXES OF AIR ENTRAINING AGENTS, SUPER PLASTICIZERS AND FLY ASH-CEMENT-FILLER BLENDS S.Jacobsen(1), H.Nordahl(1), H.Rasol(1), Ø.Lødemel(1), L.Tunstall(2) & G.W.Scherer(2) (1) Norwegian University of Science and Technology, Trondheim, Norway (2) Princeton University, Princeton, N.J., USA Abstract Foam Index (FI) measurements were made to investigate the foaming efficiency of different AEAs with varying cements (OPC, blended), fly ashes (FA) and fillers, and the effect on foaming of mixing sequence of AEA and SP. Ranking of the FI among the seven AEAs is the same for different OPC and FA binders (including different carbon contents). FA mixes with higher carbon show higher FI (i.e. more AEA needed to get stable foam) and FI in pure OPC is very low and not very different among the 7 AEAs as expected. Adding a low dosage of copolymer SP before both resin and tenside AEA could reduce the negative effect on foaming of carbon content in one of the FA blends. The low FIs of AEAs in pure OPC seemed unaffected when adding SP before AEA. Adding SP after and with the AEA reduced the efficiency of most AEAs markedly, seen as foam bursting after adding/shaking for both OPC and FA blends. Replacing parts of blended (FA) cement with limestone- and quartz filler reduced FI, presumably due to reduced carbon in the mix. 1. Introduction Concrete structures exposed to freezing and thawing with or without deicing salts can suffer deterioration by scaling and/or cracking unless properly composed with air entrainment and also cast and cured properly. With fly ash, which is probably the most common pozzolan used in most concrete nowadays, obtaining the right air entrainment (air void volume, air void characteristics) can be difficult. This is usually due to variable and/or excessive carbon content in the fly ash from incomplete combustion in the power plant [1,2]. A widely accepted mechanism is that some air entraining agents adsorb on carbon [1-3] and thereby are immobilized from air entraining action; however, recent work [4,5] shows that the AEA actually adsorbs onto the glassy ash, and the surrounding carbon shields it from interaction with air bubbles. It was proposed [2] that the problem can simply be solved by increasing the AEA dosage when needed. Still the variability of carbon in the fly ash [1-3] makes it difficult



FOAM INDEX MEASUREMENTS ONMIXES OF AIR ENTRAINING AGENTS, SUPERPLASTICIZERS ANDFLY ASH-CEMENT- FILLER BLENDS

62

to estimate the optimum type and dosage of air entraining agents and to obtain quality concrete. The Foam Index (FI) test [1,3,7,8] measures the necessary AEA dosage to obtain stable foam on top of a w/b = 2.5 slurry of the actual binder and is used as a quick test of AEA requirement of different binders in many concrete laboratories. However, variable fly ash composition is not the only challenge to obtaining desired air entrainment in practice. Normally AEA is used in combination with superplasticizing (SP) admixtures, to retain workability of concrete. This may affect air entrainment both due to the effect of workability on air entrainment [9] and the interaction between AEA and SP [10-12]. Today there is also an increasing use of other fines such as limestone and other crushed fillers in concrete production that can affect the required dosage of AEA due to their high specific surface [11]. Also variation in production (concrete composition, workability, mix equipment and sequence, transport in rotating drum trucks, placement by pumping etc.) can affect air entrainment. It is therefore of interest to understand how AEAs work, both alone and with copolymer SP [7,9-14]. According to [9,15] there are two working mechanisms of anionic AEAs. The first is formation of insoluble Ca-salts that concentrate at the air-water interface. This stabilizes air voids and makes the air void surface film dense against air leakage to the surrounding fresh paste, reducing dissolution and coalescence. Free surfactants are here less important [9,15]. Coalescence of air voids in fresh concrete needs some description. According to [16] air voids can be lost by rising to the surface, and smaller voids can dissolve so that air then diffuses through water in the fresh paste, which makes large voids grow to various degrees. This depends on void size, workability, time and thickness of the film at the void surface [16]. The second working mechanism of anionic AEAs is that surfactant molecules soluble in the presence of Ca-ions orient across liquid-air interface and bubbles adhere onto cement. Here free surfactants play a main role both forming micelles [17] and orienting at the water-air interface [9]. Non-ionic and cationic surfactants (which do not form insoluble Ca-salts) are in this view forming less stable air voids, though they can entrain larger air void volumes than anionic AEAs.

In [13] both mechanisms were recognized based on measured solubility of Ca-salts and air entrainment in concrete for different types of AEA. Increasing concrete air entrainment at prolonged mixing was related to free surfactants for non-ionic AEAs. Constant and moderate air entrainment, even at long concrete mixing periods (up to 15 minutes) and high AEA dosages for ionic AEAs, was related to measured precipitation of Ca-salts. Based on this it was claimed that AEAs that precipitated, i.e. had a limited capacity to dissolve and stabilize air voids after prolonged mixing, also were less sensitive to excessive air content at over-dosages. The existence of Critical Micelle Concentrations (CMC) [3,4,9,10,17] for AEA is probably a measure of AEA effectiveness. Particularly for the second working mechanism described above [9,10], lower CMC means reduced air-water surface tension at lower concentration. In [4,5] CMC variation depending on AEA and ionic strength was investigated as part of the work [18]. That work indicated that air entrainment depends on adsorption of AEA onto cement and fly ash, in agreement with the conclusions of Bruere [15] whereas reduction of surface tension most likely has a minor influence.

Workability also plays a role in air void formation and stability. Ref. [9] pointed to the important and difficult self-regulating relationship between air entrainment and workability. Entrained air volume usually follows the pattern: paste < concrete < mortar. Here aggregate



63

plays a role, presumably by its shearing action and by accommodating air voids in the fine part (~0.125-0.5 mm) of the sand. The workability is however usually reduced in paste and increased in mortar and concrete by air. In paste this could be due to the composite effect of air voids, whereas in concrete and mortar the air could mainly cause increased paste volume fraction and a “roller bearing effect”.

As mentioned, the combination of AEA and SP affects air void formation. Lignosulphonate, hydrocarboxylic and alkyl-sulphate AEA are compatible with SP [12]. In the systematic study in [19], melamine SP was found to work with both synthetic tenside and vinsol resin-type AEAs, whereas naphthalene-type SP worked only with vinsol resin. These 3 AEA-SP combinations were the only ones that gave both good foaming in a foaming test and good air void systems in hardened concrete [19]. In [20] foam meter studies were made of AEA with low-alkali cement containing 6% silica fume, 12% fly ash, and different SP/AEA combinations. Compared to when these admixture combinations were used with pure OPCs or pure slag, the SF and FA binders showed medium foaming and some loss of foam during 90 minutes in line with other findings of increased AEA requirements for blended binders. For copolymer (3rd generation) SP and AEA, suppliers usually recommend which AEAs and SPs are compatible in their product sheets and advise trial mixing. In [21] the FI test was used to measure effect on foaming of different AEA-SP combinations and mix procedures on a 75/25 OPC/FA blend with moderate carbon. Two different polycarboxylate SPs were investigated with a “tenside” AEA. Four different FI-test procedures were applied: 1) only AEA (i.e. standard FI test), 2) AEA first, then adding SP 3) SP first, then adding AEA, 4) AEA+SP simultaneously. Procedure (2) showed less and less stability of the foam as SP was added. Procedure (4) decreased FI slightly compared to (1). Procedure (3) reduced FI most. There were no clear effects of the two types of co-polymeric superplasticizers. In practice, addition of SP before AEA has proven to give more stable air void content during production of ready mixed concrete with fly ash [22].

The FI test is of course not reflecting reality directly (high w/b, no aggregate, foam instead of air voids). In [9] it was also claimed that FI is not reflecting the two main working mechanisms of anionic surfactants, since insoluble Ca-salts do not foam whereas they entrain air in concrete. However, all AEAs tested in our lab could foam with OPC. Furthermore, FI has been found to correlate to AEA dosage for 6% air in concrete [1, 8] and to correlate to air void content in concrete with varying fly ash carbon content at constant AEA dosage [11,21]. Also the observed relation between BET-surface of fly ash and AEA dosage required for 6% air void content in mortar and concrete [23,24] indicates that adsorption plays a role in air entrainment and that FI measurements can capture this effect. Furthermore, the FI is usually of similar magnitude as the AEA dosage needed in concrete, somewhat depending on test details, such as size and filling of container, method of shaking and adding ingredients, and observation procedure [8]. This study therefore uses this quick test to investigate the effect of type of AEA and sequence of AEA-SP addition in various binders (OPC, OPC+ FA low/high FA, OPC/FA/mineral filler). The scope is two-fold: to study the effect on foaming of different AEAs combined with a copolymer SP in different sequences (AEA alone, AEA before SP, AEA after SP and AEA with SP) and to do this on different OPC/fly ash/filler mixes.



64

2. Experiments 2.1 Materials Tables 1 and 2 give properties of admixtures (AEA and SP) and powders. The admixtures AEA1-4 and AEA5-7 are by two different manufacturers with more detailed information on the compositions of AEA1-4 [18] than of AEA5-7, see Table 1. The recommended dosages are based on manufacturer datasheets and for AEA5-7 we assumed cement content 380 kg/m3 since dosage was given per m3 of concrete for these. All AEAs are anionic. Table 1: Air Entraining Admixtures (AEA) and Superplasticizer (SP), R:Resin, T:Tenside Admixture Description Recomm. dosage/g binder

From data sheet [6] Min ( l/g) Max ( l/g)

AEA1 R Saponified rosin Resipal 55K K-abietate 0.32 1.95

AEA2 R Saponified tall oil Dresinate TX-60W Na soap of tall oil

0.16 0.65

AEA3 R Neutralized vinsol resin Vinsol resin flakes Carboxylates

0.32 0.97

AEA4 T Olefin sulfonate Ninol 40-CO Alkanolamide 0.1 1.95

AEA5 T+R Based on synthetic tensides and tall oil derivatives

0.13 1.32

AEA6 R Based on tall oil 0.13 1.32

AEA7 T Based on tensides 0.13 1.32

SP Acrylic polymer 3 12 Table 2: Cement, fly ash and filler. Powder Density

(kg/m3) Carbon

(%) LOI (%)

Blaine (m2/kg)

Eq.alkal. (%)

OPC(NO) Norcem std - CEM I 3150 0 2.35 396 1.0

OPC(US) Unicem - Type I/II (3150) 0 0.9 388 1.0

FACem (No) NorcFA CEM II 2990 0.35 1.21 461 1.4

fly ash Norcem 1.74 2.27 334 2.0

fly ash USA 2340 2.06 1.91 428 1.0

Limestone filler 2730 37.66 37.66 362 -

Quartz filler 2670 - - 3001) - 1): calculated from PSD measured with sedigraph [25] All AEAs had densities 1000 kg/m3 whereas the SP had 1090 kg/m3. AEAs 1-4 were diluted 1:9 (AEA:water) before application to have the same dilution as AEAs 5-7 had from



65

the manufacturer. Amount of “active ingredient” was thereby equal in all drops. In all results, AEA dosages are given as active ingredient per unit mass of powder. The term “active ingredient” is admittedly not well defined, but presumably very low based on [15] and compared to in SP. Table 2 shows the powders: a Norwegian and a US OPC, a Norwegian FA cement with 20 % fly ash ground with the clinker, a “Norwegian” and a US FA and two fillers. Carbon was measured by combustion and infrared detection (Leco 230-series) and LOI by TG. Densities were measured by helium pycnometry and blaine according to CEN-NS 196-6. Specific surface of quartz filler was calculated from PSD by X-Ray sedigraph [25]. 2.2 Measurements of Foam Indices The "standard" FI test is done by first shaking a container with water and binder powder with w/b = 2.5 for 60 seconds, removing the lid and adding AEA stepwise, closing lid (always quickly), 15 seconds shaking, then removing lid, 45 seconds observation while recording the time with stable foam on the liquid. We used an automated Griffin flask shaker with 10 Hz frequency adjusted using an accelerometer on the container. The agitated Plexiglas container had 70 ml volume and 40 mm inner diameter fixed to an arm giving amplitude of 20 mm and an O-ring type gasket around the lid and a rubber band holding it closed. AEA and SP were added with precision pipettes with disposable tips. The foam forming on top of the slurry was considered stable if the liquid surface remained covered with foam > 45 seconds after AEA addition and 15 seconds shaking. The number of adding/shaking/observing-repetitions and the time then varied depending on the actual mix. At stable foam, the FI was recorded as concentration of AEA per unit mass of powder. To study the effect of sequence of addition on FI, the following 4 different measurement procedures were used: 2.2.1 AEA without SP. This is the “standard” procedure [1,7,8,11 etc] with initial shaking of 10 g powder and 25 g water for 60 seconds. Then a 20 l drop of 1:9 diluted AEA was added, shaking 15 seconds, observing and recording stable time with a stop watch and repeating adding/shaking/observing; when stable foam remained after > 45 sec, FI was recorded. The 20 l drop of diluted AEA added (= 2 l per g of binder powder) is practically average of the minimum recommended dosage in Table 1 (0.1 – 0.32 l /g) for all 7 products: 0.2 l active ingredient per g of powder. 2.2.2 AEA addition before SP. This was done with the sample from 2.2.1 after FI was recorded. Then a 20 l drop of SP was added immediately, lid closed, then shaking for 15 seconds and then remove lid and observe foam. If still stable then repeat procedure with one drop of SP/shake/observe until foam bursts. A 20 l drop of SP to 10 g binder is 0.2 mass-% SP of binder whereas recommended dosage of this copolymer SP is 0.4 – 1.2 mass-% of cement. 2.2.3 SP addition before AEA. This was done by initial shaking of 10 g of powder and 25 g water for 60 seconds (as 2.2.1). Then 40 l of SP was added, corresponding to 4 l / g = 0.4 % SP by mass of binder, shaking for 15 seconds and waiting 45 seconds (no foaming with only SP). Then AEA was added stepwise: 20 l drop of AEA, shaking 15 seconds, observing 45 seconds while recording stable time and repeating adding AEA/shaking/observing until stable foam > 45 sec was observed and then FI was recorded. 2.2.4 SP with AEA. This was done by initial shaking of 10 g of powder (cement, cement+FA, cement+FA+filler) and 25 g water for 60 seconds (as 2.2.1 and 2.2.3), then adding 40 l of SP



66

together with the AEA dosage giving stable foam obtained in 2.2.3 (= FI2.2.3), shake 15 seconds and observe the stability of the foam during 45 seconds while recording stable time. 3. Results and discussion All measurements are collected in Table 3 with all dosages given as “active ingredient”. Figure 1 shows the results of 5 parallel FI-measurements with SP added before AEA5 in a mix of 70 % cement and 30 % fly ash from Norway. There is no or very little foaming (stable only for 5 seconds) at increasing AEA dosage the first 6 – 7 drops. Then relatively sharp nick points are reached by adding just one extra 20 l drop of diluted AEA to the 10 g of binder as stable foam (>45 seconds) is obtained at the end point of the test.

Figure 1: AEA dosage vs Foam stable time, 5 parallel samples of 70/30 OPC(NO)/FA(NO), Proc.2.2.3: 0.4 mass-% copolymer SP before adding AEA5 (tall oil-tenside mix) Figure 1 shows average FI = 1.2 l/g for the 5 measurements. This is a bit lower than the recommended maximum dosage for the actual product (1.32 l/g, see Table 1). We also see that the nick points where foaming propagates and the end points of the test where the foam is stable > 45 sec have low scatter. From Table 3 the results in Figure 1 can be compared with 5 parallel measurements of FI without SP on the same powder and with the same AEA5, i.e. with the standard FI procedure (2.2.1). This measurement gave higher FI = 3.2 l/g (similar scatter to Figure 1 – not shown). Therefore the addition of SP before AEA5 reduces the required AEA-dosage to obtain stable foam for this mix. Figure 2 shows the effect of adding SP before AEA in the 70/30 OPC(NO)/FA(NO) blend for 4 different AEAs. For AEA1, AEA5 and AEA7 there is a positive effect with reduced FI from 2.6 – 3.2 l/g without SP, to 1 – 1.8 l/g by adding SP first. For AEA4, which had the lowest FI = 1.4 in the “standard” procedure without SP, there is a small increase of FI (from 1.4 to 1.8 l/g is seen) when SP is added first. Table 3 shows that this reduction effect of SP on required AEA dosage only worked for the high carbon 70/30 OPC/FA(NO) blend. It was not



67

present for the pure OPC with AEA5 and AEA6 (FI very low) nor for the 30/70 US OPC/FA blend with AEA1 and AEA4 (FI low to moderate).

Figure 2: AEA dosage vs stable foam time. Effect of only AEA (procedure 2.2.1) vs 0.4 mass-% SP before AEA (procedure 2.2.3), 70/30 OPC(NO)/FA(NO) blend Table 3 shows that the rank of FI is very similar for the 7 AEAs for the 3 different powders in the “standard” procedure 2.2.1. It is also seen that adding SP after AEA (procedure 2.2.2) has a negative effect on foaming for all materials, also for pure OPC-mixes: all 2.2.2-measurements show that foaming disappears after adding 2 - 4 l/g = 0.2 - 0.4 % SP of binder powder. This is a quite moderate SP dosage, below or at the recommended minimum dosage for this SP. The reason for this effect is presumably related to the above mentioned AEA-SP compatibility question [10,12]. It could also be that the SP, when added after the foam-stabilizing effect by the surfactant is established, further de-flocculates cement particles. This can increase surface available for surfactants to adsorb on, affecting AEA efficiency. The SP could also modify interface properties (like surface tension) that the surfactant works on. Table 3 also shows that the standard Foam Indices without SP (2.2.1) are lower for all the other binders / powders than for 70 OPC(NO)+30FA(NO). The Norwegian fly ash FA(NO) is the likely cause of this. Lowest FI values are seen for OPC as expected. Table 1 shows that the US fly ash had slightly higher carbon content than the NO fly ash. However, LOI was highest for NO fly ash and there can be other differences (structure, BET specific surface) that could explain why FA(NO) requires more AEA than FA(US) [3,4,6]. Comparing the results in Table 3 for procedures 2.2.1 and 2.2.3 for the other mixes than those with OPC(NO)+30FA(NO) there seems to be no clear effect on Foam Index of adding SP before AEA. These other mixes obtain FI within or close to recommended dosage. The last column of Table 3 shows the effect of adding SP together with AEA (proc 2.2.4). The AEA dosage was the same as when adding SP before AEA. The results show that 0.4 % SP dosage then reduces the stability of the foam for 7 of the 8 combinations investigated in this way. These 7 had stable foam only 4 – 27 seconds whereas one AEA (AEA4) in the



68

moderate carbon US 70/30 mix seemed to be unaffected with stable foam > 45 seconds. Possibly there is some kind of competitive adsorption. This could cause AEA to become less available for foam stabilisation compared to when AEA is added without SP (proc 2.2.1). AEA4 which in all cases has a low/moderate FI has the longest foaming time with the 70/30 NO high carbon fly ash and foams > 45 seconds with the moderate carbon US fly ash mix. AEA4 is also the AEA with lowest CMC of AEA1-AEA4 for the solutions used in [4]. Table 3: All FI measurements – admixture concentration ( l/g) (1 l/g = 0.1 mass-%) Powder AEA 2.2.1 2.2.2

(AEA first) 2.2.3

(SP first) 2.2.4

(together) no. AEA SP as foam bursts AEA stable time

OPC(NO)

1 2 3 4 5 6 7

0.6 0.8 0.6 0.4 0.6 0.6 0.4

4.0 2.0 4.0

0.6 1.0

5s 4s

OPC(NO)70% + FA(NO)30%

1 2 3 4 5 6 7

2.6 4.0 1.8 1.4

3.21) 5.4 2.8

2.0 2.0 2.0 2.0 2.0 4.0 4.0

1.8

1.8 1.21)

1.0

7s

27s 4s

5s OPC (US)70% + FA(US)30%

1 2 3 4 5 6 7

1.0 2.0 0.6 0.6 1.4 2.6 0.8

2.0 2.0 2.0 2.0 2.0 2.0 2.0

1.4

1.2

11s

>45 s

FaCem(NO) 80/20 5 2.0 OPC(N)+FA(N) OPC(N)+FA(US)

80/20 80/20

5 5

1.8 0.8

OPC(NO)56%+ FA(NO)24%+ Filler20%

limest limest quartz

5 6 6

2.4 4.8 5.2

1): average of 5 parallel measurements Negative effect of late addition (and adsorption) of SP (procedures 2.2.2, 2.2.4) could be related to de-flocculation of cement- or binder particles making more carbon surface available for AEA and hence increasing the required AEA dosage. Positive effect of early SP addition (and adsorption) could be SP adsorption on the carbon so that less of the added AEA is adsorbed and hence more of it is available for air void stabilisation. A rule of precaution could therefore be to add SP before AEA when entraining air in fly ash Concrete [22].



69

In general the “tensides” AEA4, AEA5 and AEA7 have moderate or low FI when used alone (proc. 2.2.1). There are not enough measurements to conclude on how they perform in combination with SP compared to the resin-types (AEA1, AEA2, AEA3, AEA6). Earlier foam tests and concrete air entrainment measurements [19] showed that the stability of tenside depended strongly on the kind of SP. Therefore more studies on this are needed for optimization with copolymer. Probably trial mixing in each case would be the safest in practice. The FI of the last 6 materials in Table 3 were only measured with AEA without SP (proc.2.2.1). The first two 80/20 mixes with FA(NO) and AEA5 have similar FI = 2 and 1.8

l/g showing little effect of grinding the fly ash with the clinker (FaCemNO). The third 80/20 mix with the US fly ash gives a lower FI = 0.8 l/g as for the 70/30 blends. The last 3 materials show the effect of replacing 20 % of the 70/30 OPC(NO)/FA(NO) blend with filler. Limestone gave a slight reduction of FI from (3.2 - 5.4) to (2.4 - 4.8) for AEA5 and 6, respectively. The effect is of similar magnitude for limestone- and quartz filler based on the last two measurements with AEA6. The effect of filler presumably is to replace a part of the high surface of the carbon in the fly ash with solids with lower surface reducing the required AEA for stable foam. 4. Conclusions Foam Index (FI) measures AEA concentration needed to obtain stable foam in a w/b = 2.5 slurry. Different OPC/FA/Filler powder blends showed similar ranking of FI for seven different AEAs. Fly ash mixes with carbon showed highest FI, but adding SP before AEA could reduce the negative effect of carbon on foaming. The FI of some (presumably non-adsorbing) AEAs is not affected by SP. For those AEA/FA combinations where FI is affected by SP, adding SP with AEA and adding SP after AEA reduces the efficiency of the AEA. Replacing blended (fly ash) cement with limestone filler and quartz reduced FI slightly. References [1] Gebler S, Klieger P. (1983) Effect of fly ash on the Air-void Stability of Concrete,

American Concrete Institute Special Publication SP 79-5, pp.103-142 [2] Malhotra V.M, Ramezanianpour A. (1994) Fly Ash in Concrete, 2nd Ed, Report Canmet

MSL 94-45, pp.67-72, 153-166 [3] Külaots I., Hsu A, Hurt R.H., Suuberg E.M. (2003) Adsorption of surfactants on

unburned carbon in fly ash and development of a standardized foam index test, Cement and Concrete Research 33(12) 2091-2099

[4] Tunstall L., Prud’homme R., Scherer G.W (2016) Adsorption of Air Entraining Agents Quantified by Tensiometry, submitted to Cem.Conc.Res

[5] Tunstall L., Prud’homme R., Scherer G.W (2016) Mechanism of void formation by Air Entraining Agents, to be submitted to J. Am. Ceram. Soc.

[6] Tunstall L., Scherer G.W. (2011) Influence of fly ash on air entrainment, Int. Congr. Durab. Concr. (ICDC2012), Trondheim, June 18-21 Norwegian Concrete Assoc. 8 p

[7] Dodson V. (1990) Concrete admixtures, Van Nostrand Reinhold, 211 p. [8] Harris N., Hover K., Folliard K., Ley M. (2008a,b,c) The use of the foam index to predict

AEA dosage in concrete containing fly ash, Part I,II,III, J. ASTM Int 5 (7) 15p, 15p, 11 p,



70

[9] Joliceur C., To T.C., Nguyen T.S., Hill R., Pagé M. (2003) Investigation of Physico-Chemical aspects of Air Entrainment in Cementitious systems, ACI SP-217 pp. 595-620

[10]Du L. & Folliard K.J. (2005) Mechanism of air entrainment in concrete, Cement & Concrete Research 35 pp.1463-1471

[11] Jacobsen, S., Ollendorff, M., Geiker, M., Tunstall, L., Scherer, G.(2012) Predicting AEA dosage by Foam Index and adsorption on Fly Ash. Nordic Concrete Federation Workshop Proceedings no.10, Oslo Norway ISBN 978-82-8208-034-7 pp. 103-120

[12] Rixom R., Mailvaganam N. (1986) Chem Adm for Concrete 2nd Ed., E&FN Spon 306p. [13] Eickschen E.(2012) Working Mechanisms of Air Entraining Agents and their subsequent

activation potential, ACI Special Publication 288, pp. 305-315 [14] Aitcin P.C., Flatt R. (2016) Science and techn. of concrete admixtures, Woodhead, 666 p. [15] Bruere G.M.(1971) Air-Entraining Actions of Anionic Surfactants in Portland Cement

Pastes, J.Appl.Chem.Biotechnol. V.21, 61-64 [16] Fagerlund G. (1990) Air-pore instability and its effect on the concrete properties, Nordic

Concrete Research Publication No.9 pp. 34-52 [17] Rosen M. Kunjappu J.(2012) Surfactants and interfacial phenomena, 4th ed Wiley, 600 p. [18] Tunstall L. (2016) A study of surfactant interaction in cement-based systems and the role

of the surfactant in frost protection, Ph.D. thesis, Princeton University, 222 pp. [19] Eriksen K., Andersen P.J. (1986) Foam stability experiments on solutions containing

superplasticizing- and AEA for concrete, Nord.Conc.Res Publ.No.4, pp.45-54 [20] Sørensen R., Geiker M.(1995) Foam stability measurements, High performance concrete

for exposed concrete structures (In Danish) Dansk Betoninstitut A/S, 23 p. [21] Vestgarden J.(2006) Air entrainment in fly-ash concrete, MSc, NTNU (Norwegian Univ.

of Sci. & Tech., Dept. of Structural engineering) 141 p. [22] E.Mørtsell (2014) Heidelberg-Norbetong, Pers comm [23] McCarthy M.J., Sadiqul Islam G.M., Csetenyi L.J., Jones M.R.(2012) Colorimetric

evaluation of admixture adsorption by fly ash for use in air-entrained concrete, Mat&Str (45) 1793-1803

[24] Ley M.T, Harris N.J., Folliard K.J., Hover K.C.(2008) Investigation of Air-Entraining Admixture Dosage in Fly Ash Concrete, ACI Mat. Jour. (Sept-Oct) 494-498

[25] Cepuritis R. (2016) Development of Crushed Sand for Concrete Production with Micro-proportioning, PhD thesis NTNU 2016:19, 381 p.

Acknowledgment This paper is based on the BSc work of Henrik Nordahl-Pedersen, Hawar Rasol & Øyvind Olsen-Lødemel: Foam Index Test – influence of AEA and SP, Gjøvik College/NTNU, 96 p, 2015 (in Norwegian) and supported by the DACS project (Durable Advanced Concrete Structures). The financial contribution of the Norwegian Research Council is gratefully acknowledged. The DACS partners are: Kværner Concrete Solution AS (project owner), Axion AS (Stalite), AF Gruppen Norge AS, Concrete Structures AS, Mapei AS, Multiconsult AS, NorBetong AS, Norcem AS, NPRA (Statens vegvesen), Norges teknisk-naturvitens-kapelige universitet (NTNU), SINTEF Byggforsk, Skanska Norge AS, Unicon AS and Veidekke Entreprenør AS.



71



FREEZING INDUCED STRESSES IN CONCRETE-STEEL COMPOSITE BEAMS AND EFFECT OF AIR VOIDS Stefan Jacobsen (1) and George W. Scherer (2) (1) Norwegian University of Science and Technology, Trondheim, Norway (2) Princeton University, Princeton, N.J., USA Abstract The thermal expansion match between concrete and steel is lost when freezing occurs, owing to enhanced shrinkage caused by cryosuction in the pore solution and the high coefficient of thermal expansion (CTE) of ice. Freezing dilatometry performed on steel and on well-cured mortar samples with and without air entrainment showed a sharp transition to damage by very slight increase of water content, and revealed how cryosuction increased the CTE of air entrained mortar after nucleation. Freezing-induced warping was measured on mortar-steel composite beams. After nucleation, beam warping in both directions (concave down or up) was observed depending on whether the mortar was air entrained or not. Comparison of beams with different quality of mortar-steel bond shows loss of bond and internal damage to various degrees for non-air entrained mortar, whereas air entrainment protected the bond while preventing internal frost damage. Poromechanical modelling of the bi-material and hydrodynamic effects could simulate most of the pre-freezing warping, whereas post-freezing modelling is still in progress. 1. Introduction The volume or length changes measured during freezing of concrete is normally measured on unrestrained samples; hence the significance of such tests for reinforced and/or loaded concrete structures can be questioned. In particular, the effect of constraint on structural damage from freezing could be significant. The present study investigates the stresses to be expected in reinforced concrete by performing dilatometry on the isolated components and by measuring the deflection of a composite plate of mortar and steel. The influence of air entrainment, degree of saturation, and quality of the interfacial bond are explored.

72



Studies on how external loads affect frost damage have been performed. In [1,2] it was found that air entrainment reduced damage from tensile loading during freezing. However, these results may be misleading with regard to the behavior of reinforced concrete, because air entrainment results in strong contraction during freezing, so the concrete will shrink relative to the reinforcement. Therefore, whereas air entrainment reduces the sensitivity to externally applied tensile loads during freezing, it will induce tensile stresses in reinforced concrete. In the absence of freezing, dilatometry is used to measure the Coefficient of Thermal Expansion (CTE) from the length change, L, recorded during heating and/or cooling. Freezing strains and frost damage can be expressed as deviations from linear thermal contraction behavior and as residual strain, res= Lres/L, after a complete cooling-heating loop. In [3] the freezing expansion was completely suppressed by a compressive load equal to ~40% of the compressive strength, which indicates that internal cracking in frost tests of non-reinforced concrete does not represent frost damage in reinforced structures. Furthermore, the CTE of steel (1.1 - 1.2 x 10-5 °C-1) is generally higher than that of concrete (~0.7 – 1.3 x 10-5 °C-1, depending on aggregate type, water content, etc.) The slight incompatibility often brings the concrete into mild compression during cooling (before any pore water freezes). Any expansion in reinforced concrete due to pore water freezing will compress it further. However, in air-entrained concrete this is not necessarily the case. DMA (Dynamic Mechanical Analysis) measurements during freezing and thawing of capillary-saturated and air-entrained mortar showed that thermal contraction increased to approximately 3.5 times that of the concrete pre-freezing contraction [4] in line with the observations in [5] on air entrained paste. Therefore, in reinforced air-entrained concrete the concrete should go into tension during freezing, potentially affecting its structural behavior. The scope of this study is to measure and analyze, by experiment and modeling, the stresses that can arise in concrete and the concrete/steel bond zone during freezing/thawing of reinforced concrete with and without air entrainment. 2. Experiments and modelling 2.1 The warping box An experimentally simple way of studying the effect of restraint during freezing is to investigate the bi-material bending caused by differential strain during cooling, freezing and heating. The experiments are done with mortar, rather than concrete, to allow the use of samples of a convenient size; of course, the qualitative trends would be the same for concrete/steel composites. Figure 1 shows the experimental set-up previously used to study frost-induced warping of paste and mortar with deicer salt solution/ice on top [6]. The beam is arranged with the steel on top and the mortar layer in contact with low-expansion invar supports. The deflection, , is detected by an optical probe that is focused on a reflector on the underside of the beam; deflection is reported as negative when the beam bends downward (toward the probe), thus becoming concave upward.

73



Figure 1: Experimental set-up for warping during freezing. Composite sample has steel plate (thickness LS) on top and mortar plate (thickness Lm) below.

Two different temperature – time cycles were used: one similar to the temperature cycle in the DMA apparatus and one with stepwise cooling to obtain thermal equilibrium between steel and mortar in several periods of the cooling-heating cycle.

2.2 Materials and specimens Mortars having w/c = 0.55 were prepared with a Hobart 1.5 L table mixer using Norwegian OPC (CEM I), gneiss-granitic sand according to NS3099 with Dmax = 4 mm, Sika polycarboxylate water reducer and Sika Aer-S tenside type AEA. Consistency was measured with a mini cone with height 120 mm, lower diameter 80 mm, and upper diameter of 40 mm. The mixes had stable edges on the slump-flow cake. Density and air content were measured in fresh and hardened mortars by air pressure meter, density measurements, and image analysis. Two final mortar mixes were used: one without and one with air entrainment. Steel plates 0.5 x 20 mm were prepared for making of composite specimens in 20 x 20 x 250 mm steel moulds for mortar beams. After cutting to suitable lengths, the plates were sandblasted; in some cases, anchorage points were made by inwards penetration of a sharp tip through the steel before placing at the bottom of the moulds. Mortar beams 20 x 20 x (120 or 230) mm were cast in the steel moulds, demoulded after 24 h, water cured to 28 d, moist cured with cloths and wrapped in plastic for 6 - 12 months before drilling of cylinders for DMA measurements and cutting of plates for warping experiments. Air void analysis was done as recommended in [7]: 4 x 16 cm cut-and-polished surfaces were colored black with a permanent marker (Eddington 800) and the air voids were filled with 1 – 4 micron BaSO4 powder. The sample was imaged with a flat-bed scanner (CanoScan 8400F) and the image was analyzed with Matlab-based software. Table 1 shows properties of the mortars.

Specimens were cut/drilled with a fine diamond saw/coredrill 5 – 7 days before testing and stored submerged in deionized water until testing. The plates for warping were 6 x 20 x 100 mm (sandblasted), and 5 x 20 x 100 mm (sandblasted + inward anchorage points). The thickness includes the 0.5 mm thick steel plate. The cylinders for DMA had diameter 8.0 mm and height 15.0 mm. DMA measurements were also made on samples of the steel. Table 1 shows that air entrainment was clearly increased by AEA and densities indicate that the wrapped curing and/or the sawing had allowed some drying before the specimens were capillary saturated prior to DMA and warping tests.

74



Table 1: Mortar properties No AEA AEA Paste volume fraction (excl air) 0.467 0.421 Mini-slump (mm) 88 87 Mini slumpflow diameter (mm) 142 132 Fresh density at mixing 2226 2005 Air content at mixing (vol-%) 4.0 12.1 (+8.1) Density of specs. demould. 24 h 2314 2041 Density of DMA specs * 2258 2015 Air content hardened mortar 4.6 14.3 Air void spec surface (mm-1) 15 20 Spacing factor (mm) 0.42 0.12 *: after drilling/cutting, before water storage

2.4 Freezing dilatometry – DMA, E-modulus A Perkin Elmer DMA7 was used with a cooling cycle designed to reflect one of the two temperature cycles used in the warping box. The cooling cycle for the mortar samples was: (1): hold for 20 min at +20 °C; (2): cool from +20 °C to -25 °C at 0.33 °C/min; ( 3) hold for 60 min at -25 °C; (4) heat from -25 °C to +20 °C at 0.33 °C/min. After water curing, metaldehyde powder as nucleating agent [8] was sprinkled on the surface of the specimens; tape was then wrapped around their lateral surfaces before they were immersed in kerosene within the temperature-controlled sample chamber, and the glass probe tip was lowered onto the upper surface. In addition some DMA tests were run on samples of the steel. Dynamic E-modulus (PUNDIT, 54 kHz) was determined by ultrasonic pulse velocity (UPV) measurements at 10 months age on beam specimens (without steel) and found to be 40.0 and 27.8 GPa for non-air- and air-entrained mortars, respectively.

2.5 Modelling warping strain Timoshenko [9] analyzed the deflection (see Figure 1) in a composite in which both plates have uniform properties. That solution does not apply to a saturated plate of non-air-entrained mortar, because the pore pressure is relieved by flow to the exposed surface, resulting in a gradient in the strain from the metal/mortar boundary to the free surface. That problem is addressed using the general solution for a composite whose properties vary arbitrarily, which is given in [10]. In the case of air-entrained mortar, positive pore pressure is relieved by flow into the air voids, so no gradient is present. Using the analysis for a bi-material plate [9,10] with the geometries, CTE, and E-moduli of our experiments, the relationship between temperature change and warping deflection, , is shown in Table 2. Before freezing, the effects of pore water thermal expansion and contraction during heating and cooling can be modelled using the method employed in [11]. After freezing, the main contributions from pore water freezing are expansion from hydraulic pressure and crystallization pressure, and contraction from cryosuction in air-entrained concrete plus the enhanced contraction owing to the high CTE of ice in the matrix. Thermal contraction of the mortar/ice composite is partially offset by expansion caused by hydraulic and crystallization pressure, and the net strain of the mortar can be determined from the deflection. If the mortar is sufficiently air entrained and the air voids are not saturated, then cryosuction by ice

75



formation in the air voids contributes to contraction. Again, the net strain in the mortar is revealed by the deflection of the composite beam. The elastic analysis also allows calculation of the stresses at the interface between steel and mortar in the composite. As will be seen in section 3.2 the warping in Table 2 is very close to warping measured before pore water freezing sets in. Table 2: Bi-material warping, (microns) as function of temperature change, T, for steel/mortar composite beam with thickness ratio m = Ls/Lm = 1/9 and span W = 0.1 m.

T (K) –1 –5 –10 –20 –40 (microns) (air-entrained) –0.9 –5 –9 –19 –37 (microns) (no air entrainment) –0.6 –3 –6 –12 –24

3. Results and discusssion 3.1 DMA - freezing dilatometry and CTE Figures 2a-c show typical DMA dilation curves of capillary-saturated mortar cylinders with and without air voids and Table 3 gives the CTE. Figure 2a shows the behavior of the non-air-entrained mortar after 5 days of storage in water under atmospheric conditions. It is seen that ~4% entrapped air voids are capable of protecting the mortar against frost damage. Figure 2b shows a companion specimen that after 5 days immersion was kept a few minutes under slight vacuum pumping, still under water, and then stored for another day in water at atmospheric pressure. This increased the water content only very slightly, with sample density increasing only 1 kg/m3 as detected by weighing the sample in the same kind of surface dry state. The capillary porosity of these samples measured from saturation to drying at 105 °C was 187 L/m3 mortar. Thus the increased water content as a fraction of the estimated capillary porosity was < 1 % of the pore volume. Figure 2b therefore clearly shows that the sample in Figure 2a was very close to the so-called critical degree of saturation [12], since the slight additional absorption increased the freezing expansion so as to create permanent frost damage. The maximum freezing dilation shown in Figure 2b is about 0.0002 (200 μ ) above the extrapolated linear pre-freezing contraction curve. This dilation is slightly larger than the tensile strain capacity of concrete (around 150 μ ) so that a residual expansion is seen in 2b.

Fig. 2: Freezing dilation a): non-air entrained capillary saturated, where there is enough entrapped air to provide protection b): non-air entrained slight vacuum to increase saturation, c) air-entrained capillary saturated

76



Figure 2c shows that the air voids increase the freezing contraction after nucleation, thereby protecting against frost damage. The reason is the cryosuction by ice crystals formed in air voids that draw water from the capillary pores [5,13].

Table 3. CTE of mortar and steel before freezing [x 10-5 °C-1] (R2) sample Virgin Non air entr 0.86 x 10-5 °C-1 (0.998) Air entrained 0.75 x 10-5 °C-1 (0.999) Steel 1.11 x 10-5 °C-1 (1.000) 3.2 Freeze-warping experiments 3.2.1 Pilot tests with transient freeze/thaw cycle – smooth bond zone Figure 3 shows temperature and deflection in warping tests with samples with and without air entrainment and with sandblasted bond zones. These first runs were made with a similar freeze/thaw cycle as in the DMA to check the magnitude of warping and whether the composite warping could go in different directions with and without air voids.

Figure 3: Steel-mortar composite, pilot tests with sandblasted steel surface and similar cooling-heating as in DMA ( > 0: sample warps up); circles indicate onset of freezing.

Figure 3 shows a much larger warping of the non-air-entrained samples than predicted by the bi-material effect in Table 2. This indicates that pore water thermal expansion and then pore water freezing starting at around -5 °C contributed to warping. The specimen with air voids in Figure 3 shows upward warping as pore water begins to freeze, due to cryosuction causing contraction of the mortar (see Figure 2c). For the air-entrained specimens, the pore water freezes at around -5 ºC and the combination of hydraulic pressure and crystallization pressure expand the mortar, so that it deflects strongly downward. The downward deflection prior to freezing is a bit more than 30 microns for a temperature change of ~30 °C, which is in quite good agreement with the pure bimaterial bending predicted in Table 2. The large permanent deflections after freezing and thawing, particularly in the non-air entrained specimen in Figure

77



3, indicate bond failure and/or permanent frost damage in the concrete. Modelling the strain over the cross-section at this deflection gave a neutral axis 1.5 mm from the top of the steel, 50 microstrains in the steel on top, and maximal tensile strain in the mortar of 200 microstrains at the bottom (presumably cracked).

3.2.2 Cooling cycles with equilibrium temperatures - beams with improved bond / anchorage. After the pilot tests, two specimens, a non-air-entrained and an air-entrained mortar cast on sandblasted steel with additional 6 anchorage points to improve bond, were run in the warping box. This time stepwise cooling was used to obtain equilibrium temperature between steel and mortar in regular intervals during cooling. Non-air entrained: Figure 4a shows the deflection of non-air entrained mortar before and after freezing. The deflection before the first pore water freezing event in Figure 4a is similar to the corresponding bimaterial bending calculated in Table 2 for the actual temperature change. (A disturbance is seen as a peak or glitch after 4.5 h, because the lid of the box was opened and insulation was put on top of the sample.) Figure 4a also shows that freezing of the pore water (near 9 h) causes a large deflection, as for the non-air-entrained sample in Figure 3.

a) b)

Figure 4: (a) Measured deflection of non-air-entrained mortar; (b) calculated deflection when T > 0 (Solid curve = data, dashed and dotted curves calculated using various permeabilities)

Figure 4b shows the results of modelling of the warping before pore water freezing (Figure 4a) for different permeability values of the mortar based on the poromechanical contribution of hydrodynamics of expanding pore water [11]; details of the analysis will be presented in a future publication [14]. When the temperature drops, the pore water contracts and develops a negative pressure; this can be relieved by retreat of the liquid into the free surface, so the suction is greater near the interface with the metal, and the resulting pore pressure gradient in the mortar tends to bend it concave upward. At the same time, the pore suction enhances the linear contraction of the plate, which tends to make the composite become concave downward. Thus, the effects of pore pressure on contraction and bending of the mortar plate are offsetting, so the net poromechanical contribution is modest. Figure 4b compares the measured data (solid black curve) to curves calculated assuming k = 1.e-12 (dashed), 1.e-13 (dash-dot), and 3.e-14 (dotted) m/s. During cooling, the negative pore pressure (which is higher when k is smaller) tends to reduce the deflection of the composite

78



by making the CTE of the mortar closer to that of the steel. As k decreases, the simulations begin to capture the brief reversal of deflection near 6.3 h. The large deflection beyond 7 h appears to be an artifact, as the temperature change in that time period was not very large. So the sample might have shifted. Modeling of the post-freezing deflection will be presented in [14].

Air entrained: Figure 5a shows the warping of an air-entrained mortar-steel composite at T > and < 0°C. During the first 4 hours (before freezing) the warping follows the stepwise cooling due to the higher thermal contraction of the steel compared to mortar. Before nucleation, a short heating and cooling was performed, while remaining above T = 0 ºC, from 4 to 5 hours . This caused a fast downward deflection, but then a rapid recovery before cooling was resumed at 5 hrs. The explanation for this behavior is a rapid expansion of liquid by the quick heating causing the mortar to expand more than the steel, resulting in the fast downward deflection. As the expanding water reaches the air voids and flows into them, the pore pressure drops and the mortar contracts, reducing the warping until cooling resumes near 5 h. According to [15] the permeability of the air-entrained samples should be around 25% lower than that of the non-air-entrained, which could slightly offset the rapid relaxation of pore pressure as the pore water flows into the voids. The simulation in Figure 5b, is based on the poromechanical contribution of the hydrodynamics of expanding porewater [11] including flow through paste shells around the air voids. It captures the qualitative features of the data, including the reversal in strain when the temperature suddenly increases near 4 h.

The difference in warping between non-air-entrained (Fig. 4a) and air-entrained (Fig. 5a) mortar before freezing needs some further clarification. A rapid change in temperature causes dilatation of the pore liquid that creates pore pressure, so the strain of the mortar is greater than would be expected from its CTE alone. One might expect that this effect would be negligible in the air-entrained mortar, because the liquid can flow into the voids to relieve its pressure. Nevertheless, the deflection during rapid changes in temperature is greater for the air-entrained sample. The reason for this counter-intuitive result is that the non-air-entrained sample develops a bending moment when the pore pressure rises, because that pressure is only relieved on the free surface, not near the interface with the steel. Expansion of the mortar tends to deflect the composite downwards, but the pore pressure gradient tends to deflect it upwards, and the effects offset. In the air-entrained mortar, the pressure is relieved by flow into the voids, so there is no macroscopic gradient in pore pressure; consequently, only the linear expansion affects the deflection. When the air-entrained mortar (still above freezing) is heated rapidly, it expands more than the steel, so the composite deflects downward.

a)

79



b)

Figure 5: a) Measured temperature (right ordinate) and deflection (left ordinate) of air-entrained mortar/steel composite; (b) measured deflection (solid black curve) compared to simulated deflection at T > 0ºC, assuming k = 1.e-12 (dash-dot), and 3.e-13 (dotted) m/s.

After pore water freezing (Figs. 4a and 5a) the same phenomena as already observed in Figure 3 are seen again with continued bending downward for the non-air entrained mortar (Fig. 4a) due to hydraulic and crystallization pressure. Upon nucleation (near 9 h), there is a huge downwards warping (concave up), presumably from ice crystallization pressure, then a relaxation (from creep of the mortar) when holding the temperature at -20 ºC, and then it warps up again at melting. Finally, the mortar stiffness drops due to microcracking when all ice has melted. For the air entrained mortar (Fig. 5a), the strong increase of CTE after nucleation due to cryosuction (Fig. 2c) warps the plate upward again (becoming concave down) and there is less permanent damage.

Figure 5a also shows the warping of the air-entrained mortar-steel composite at T < 0 °C. After freezing (~7 h), the ice in the air voids causes contraction (cryosuction causing high CTE according to the DMA data shown in Fig. 2). Melting of ice just before 16 h reduces the CTE of mortar again as cryosuction stops, so that the specimen warps down again. Finally, heating causes the steel to expand more than the mortar and hence finally warp the beam back up again (becoming concave down). The final loss of stiffness due to frost damage in the non-air-entrained specimen (Fig. 4a) causes a larger permanent downwards warping than the air-entrained mortar shows after freezing (Fig. 5a).

4.Conclusions Air entrainment protects concrete from frost damage by providing a sink for the expansion of pore water as it crystallizes, and the ice growing in air voids compresses the concrete by creating suction in the unfrozen pore solution. However, this contraction induces a mismatch in contraction between concrete and steel that might lead to damage. This study indicates that the strain mismatch is significant, and that the simple warping test provides valuable insight into the magnitude of the strains. A poromechanical analysis (to be presented more completely elsewhere) provides quantitative evaluation of the factors contributing to the stresses during thermal cycling and freezing of reinforced concrete.

80



Acknowledgement We thank Priscilla Fonseca for assistance with DMA and air voids measurements, Joe Vocaturo for lab assistance, Jon Spangenberg for dynamic modulus measurements, and Princeton University for hosting SJs visit. The work is part of the DACS project (Durable Advanced Concrete Structures). The financial contribution of the Norwegian Research Council is gratefully acknowledged. The DACS partners are: Kværner Concrete Solution AS (project owner), Axion AS (Stalite), AF Gruppen Norge AS, Concrete Structures AS, Mapei AS, Multiconsult AS, NorBetong AS, Norcem AS, NPRA (Statens vegvesen), Norges teknisk-naturvitenskapelige universitet (NTNU), SINTEF Byggforsk, Skanska Norge AS, Unicon AS and Veidekke Entreprenør AS. References [1] Zhou Y.X., Cohen M.D., Dolch L.W. ACI Mater. J. 91 [6] (1994) 595-601 [2] Sun W., Mu R., Lia X., Miao C. Cem Conc Res 32 (2002) 1859-1864 [3] Planas J., Corres H., Chueca R, Elices M., Sanchez-Galvez, V. (1983) Influence of load on thermal deformation of concrete during cooling down, 2nd Int.Conf Cryogenic Concrete, Amsterdam [4] Sun Z. & Scherer G. Effect of Air Voids on Salt Scaling. CemConcRes 40 (2010) 260-270 [5] Powers T.C & Helmuth R.A. Proc.Highway Research Board 32 (1953) 285-297 [6] Valenza J., Scherer G. Mechanism for Salt Scaling. J.Am Cer Soc. 89[4] (2006)1161-1179 [7] Fonseca P.C., Scherer G.W. An image analysis procedure to quantify the air void system of mortar and concrete. Mater. Struct. 48(2015) 3087-3098 [8] Fukuta N. Ice nucleation by metaldehyde. Nature 199 (1963) 475-476 [9] Timoshenko S. J. Analysis of bimetal thermostats. Opt. Soc. America 11(1925) 233-255 [10] Scherer G. Drying Gels: III. Warping Plate J. Non-crystalline Solids 91 (1987) 87-100 [11] Ai, H. Young, J.F. and G.W. Scherer, Thermal expansion kinetics: Method to measure permeability of cementitious materials: II, Application to hardened cement paste. J. Am. Ceram. Soc. 84 [2] (2001) 385-391; Erratum, J. Am. Ceram. Soc. 87 [8] (2004) 1611 [12] Fagerlund G. Mater. Struct. 4 (1971) 271-285 [13] Scherer, G.W. and Valenza, J.J. Mechanisms of Frost Damage, pp. 209-246 in Materials Science of Concrete, Vol. VII, eds. J. Skalny and F. Young (American Ceramic Society, 2005) [14] Jacobsen, S. and Scherer, G.W. Thermal mismatch stresses in reinforced concrete, to be published [15] Wong H., Pappas A., Zimmerman R., Buenfeld N. Cem Conc Res 41(2011) 1067-1077

81



CORRELATION BETWEEN CHARACTERISTIC DISTANCES OF AIR VOIDS AS POINT PROCESSES AND SPACING FACTORS IN MORTARS Hidefumi Koto (1), Takuma Murotani (1) and Shin-ichi Igarashi (1) (1) Kanazawa University, Kanazawa, Japan Abstract Images of air voids were acquired with a flatbed scanner. Distribution of air voids in the images was regarded as a spatial point process. The nearest neighbour distance function was calculated for the point process. The median distance determined by the function was proposed as a characteristic average distance between air voids. The median distance in a point process was compared with the conventional spacing factor. In addition, taking account of the presence of aggregate particles, air-void systems in real mortars were simulated as interrupted point processes. The median distances between air voids in real mortars had a strong correlation with the spacing factors. Furthermore, there also exists a good correlation between the median distances of the simulated point processes and the spacing factors of real voids. Thus, conversion of air-void systems to point processes is useful for evaluating actual spatial arrangement of air voids. The point process function enables to estimate the spacing factor easily. 1. Introduction Adequate frost resistance of concrete is usually ensured by entraining a proper amount of air. An amount of air is measured at a fresh state of concrete. However, it is empirically known that the measured initial air content is not always kept in hardened concrete. Therefore, in addition to the air content, spacing of air voids is considered as a crucial parameter to ensure the frost resistance of concrete since it is correlated to a durability factor of concrete. The spacing is generally expressed using a spacing factor. The spacing factor is obtained by following the test procedures prescribed in ASTM C457. However, the procedures are quite time consuming. These days, an easy and fast apparatus to measure in accordance with the ASTM procedures has been commercially available, but not spread as a general tool. It

82



Table 1 Mix proportion of specimens

W/C C:S NameSuper

plasticizer(C%wt)

AE waterreducing

admixture(C%wt)

AEadmixture(C%wt)

P-Ref -P-AE[1] 0.10P-AE[2] 0.25P-AE[3] 0.50P-AE[4] 0.25 0.01P-AE[5] 0.25 0.02M1-Ref -M1-AE[1] 0.06M1-AE[2] 0.1M2-Ref -M2-AE[1] 0.002M2-AE[2] 0.006

--

0.1

0.45

-

-

Cementpaste 1:0

Mortar

1:1

1:2

0.40

0.40

should be noted, however, that the spacing factor of ASTM C457 is a parameter of virtual spatial arrangement where air voids with the same size are regularly arranged in bulk cement paste matrix of concrete. Therefore, it doesn’t specify a real distance between air voids in concrete. If void spacing is important in determining the frost resistance of concrete, it should be evaluated from real distribution of air voids in hardened concrete [1]. This is the simple motivation behind this study. These days development of imaging technology and data analysis has made possible to describe spatial arrangements of objects that are irregularly or randomly dispersed in 2D space. It is generally assumed that air voids in concrete are dispersed randomly and uniformly. Strictly speaking, however, complete random distribution of air voids may not be realized in concrete. For example, aggregate particles prevent air bubbles from dispersing randomly. Therefore, it is necessary to evaluate various patterns of spatial arrangements of air bubbles. To describe the geometrical structure of spatial point patterns formed by objects such as air bubbles, point process statistics has been used for recent decades [2]. Each point in a point process shows positions of objects of interest. When free distribution of points is not allowed, their spatial structure can be treated as an interrupted process. Based on second-order stereology, features of the spatial structure can be expressed by means of several statistical functions and probability. Fortunately, there is a useful statistical function which has a similar viewpoint to the spacing factor as a general parameter for the air-void system. In this study, air voids in images scanned at a low magnification are treated as a spatial point process. Taking account of the fact that the spacing factor is a distance parameter, the nearest neighbour distance function is calculated for those point processes, which represent exact positions of each air void in mortars. A characteristic distance of the point process is defined by a realization probability in the nearest neighbour distance function. Correlation between the characteristic distances and the spacing factors is discussed. Furthermore, spatial distribution of air voids in mortars is simulated as an interrupted point process. The characteristic distances for the simulated point patterns are calculated as well. They are again compared with the corresponding distances in real point patterns of mortars and with the spacing factors. The procedure to apply the point process statistics to air voids distribution is proposed as an easy way to estimate the spacing factor from images of mortars. 2. Experimental 2.1 Materials The cement used was an ordinary Portland cement with a Blaine fineness value of 331m2/kg. River sand was used as a fine aggregate. The water/cement ratio was 0.40. The mass ratios of sand to cement in mortar were 1.0 and 2.0. A superplasticizer and an air-entraining (AE) agent were also used. The total air content was measured with a small air meter for mortar. Prism specimens of 40×40×160mm were produced in

83



Fig. 1 Binary segmentaion of aggregate particles and air voids

Fig. 2 Schematic drawings:(a) the nearest neighbour distance function and definition of R50 (b) characteristic distance L’ derived from R50

1.0

G(r

)

Distance r

0.5

0.0R50 R100

R50

D/2

Air

Air

accordance with JIS R5201. They were demolded at 24h after casting. They were then cured in water at 20ºC for 7d. Mix proportions of cement pastes and mortars are given in Table 1. 2.2 Image acquisition At the age of 7d, slices about 10 mm thick were cut from the prisms. They were finely polished with abrasive SiC papers. To distinguish between aggregate particles and cement paste matrix, polished surfaces of mortars were treated with a solution of phenolphthalein. Ten images in each specimen were acquired with a conventional flatbed scanner. Then the treated surfaces were dyed with black ink. Air voids on the surfaces were filled with a white powder of calcium carbonate (diameter 12- The surfaces were scanned again to obtain images of air voids. The resolution of images was

2.3 Image analysis Various images used in the image analysis process are shown in Fig. 1. All the relevant image information was extracted from an original scanned image (Fig. 1(a)). For instance, aggregate particles were segmented on the basis of brightness and chromaticity (Fig. 1(b)). Air voids were directly segmented from images of cross sections treated with black ink (Fig. 1(c)). It is known that the size of entrained air bubbles is usually in the range between a few tens and several hundred Thus small

the images. Residual voids after deleting the noise were regarded as air bubbles. The area fraction of them was simply obtained by image analysis. Centroids of each air bubble xi (i=1, n) was also obtained by image analysis. Then all the air bubbles were converted to a set of points of which positions are the centroids of original air bubbles (Fig. 1(d)). The set of those points was regarded as a spatial point process X={ xi :i=1, n}. 2.4 Calculation of point process statistics (1) Point intensity The point intensity p of the point process X was determined by the following equation.

)1()()(

WAWpN

p

84



Fig. 3 Procedure to generate thinned point processes; (a) basic process Xb (b) aggaregate phase

a (c) thinned point process Xs

Where Np(W) is the total number of points in an observation window W. A(W) is the area of the window. (2) The nearest neighbour distance function (G-function) The nearest neighbour distance function (G-function) was calculated using Eq.(2) [3].

)2())(

1 ()(1

))(1 ()(1)(1

)(ˆWN

i iswibiS

WNi iswibiSriS

rG

Where 1( ) is the indicator function, i.e. equal to one if its argument is true and zero otherwise. r is distance. bi is the shortest distance from each point xi to edges of the window W. si is the distance to the nearest other point in the point process. w(si) is a weighting factor, and given as the inverse of the window area eroded by a circle of radius si [3]. A schematic drawing of the nearest neighbour distance function is shown in Fig. 2. It is a distribution function of distance from a point xi in X to the nearest neighbour xj j) in X. The function is expressed as a cumulative probability function. To represent a continuous distribution function by a single characteristic parameter, the median distance R50 is defined as a distance corresponding to the cumulative probability of 50% (Fig.2 (a)). (3) Thinning points lying on aggregate particles Using the point intensity p for the whole area and the area fraction of cement paste matrix in mortar, a point intensity b of air bubbles in the cement paste matrix was calculated. A point process with the intensity b was simulated as a Poisson process for the whole area of W. This simulated point pattern was used as a basic point process Xb (Fig. 3(a)) where all the air bubbles were not disturbed by aggregate particles. The simulated basic process was superimposed with real images of the aggregate phase a (Fig. 3(b)). Then points on a (xi Xb a) were deleted. The points in cement paste matrix as a random field were survived to make a different point process with less points (Fig.3(c)). This thinned process Xs (=Xb a

c W was regarded as a virtual point process of air bubbles in mortar (Fig. 3(c)). Simulation to generate the basic process Xb was repeated ten times. For each simulated pattern, the thinned process Xs was obtained by deleting points, using the segmented images of aggregate particles. The whole procedure to obtain the thinned point process is schematically shown in Fig. 3. (4) Spacing factor of air voids Using the images acquired in 2.2, the spacing factor L of ASTM C457 was calculated by Eqs. (3) and (4).

)3(13 14.13 )33.4( APforA

PL

85



)4(6a

Where is the specific surface area of air voids. P is the volume fraction of cement paste. A is the volume fraction of air. a is the mean area of air bubbles [4]. These values of A, P and a are also easily obtained from images by means of an image analysis software. 3. Results and discussion

3.1 Properties of air voids in images Properties of air voids in the samples are summarized in Table 2. Table 2 Fresh properties of mixtures and air void properties obtained from scanned images

Ref AE[1] AE[2] AE[3] AE[4] AE[5] Ref AE[1] AE[2] Ref AE[1] AE[2]Flow value 195 213 222 249 228 227 198 197 198 202 217 218

Point density (/mm2) 0.46 0.89 2.02 2.43 2.89 3.06 0.60 2.18 2.91 0.83 1.66 2.80Average air void diameter 123 138 120 118 100 126 144 145 142 175 158 157

Air (fresh) (%) 2.4 3.4 4.9 6.4 6.0 8.5 2.0 5.5 7.2 5.7 8.6 11

Air (hardened) (%) 1.0 2.1 3.5 4.4 3.7 7.3 1.6 5.3 6.6 4.7 6.5 9.5

Coefficient of variationof the air volume(hardened)

0.13 0.05 0.06 0.12 0.04 0.06 0.11 0.10 0.11 0.14 0.08 0.06

Residual rate of air void(%)

41.7 63.1 72.2 69.0 62.0 85.9 81.8 96.4 91.7 82.7 75.6 86.4

Cement paste (C:S=1:0) Mortar (C:S=1:1) Mortar (C:S=1:2)

There are differences in the air contents between fresh and hardened states. They are attributed to losses during casting and compacting since the volume ratio of voids deleted as noise in the images was quite low. The residual ratio of air in cement paste specimens without AE agents is the lowest among the mixtures. This suggests that entrapped air in the cement paste is easy to be lost compared to entrained air. In view of the residual ratios of the cement pastes and the mortars, air bubbles in the mortars are more stable than those in the cement pastes even if any AE agents are not used. The amount of AE agents necessary to increase the unit content of air is decreased with increasing the amount of aggregate. Aggregate affects the efficacy of the AE agents. Furthermore, the coefficients of variation of the air volumes are smaller in the air entrained mixtures. Entrained air bubbles are clearly more stable than entrapped air. 3.2 Variation of spacing between air voids Fig. 4 shows the nearest neighbour distance functions G(r) of air voids in cement pastes and mortars. They intersect the horizontal axis at certain distances. This suggests air voids have their own sizes so that other points cannot be present in the vicinity. The distances at which the function curves converge on the unity decrease with the increase in air content. This is because the increase in the point intensity results in a closer neighbour point. The median

86



distance R50 which corresponds to a probability of 50% in the cumulative function can be considered as an average spacing between air voids.

Fig. 4 Nearest neighbour distance function of air voids in cement pastes and mortars

Fig. 5 Median distance vs. point intensity in cement pastes and mortars Fig. 5 shows the relationship between the median distance and the point intensity. There exists a negative association between them. When the point intensity is greater than about 1.5, the rate of change in the median distance is decreased with the increase in the point intensity. This may be attributed to coalescence of air bubbles that have their own sizes. In other words, when two air bubbles approach each other below a certain distance, some interaction between them could form a bigger bubble. Thus the median distance does not decrease proportionally to the point intensity. Furthermore, it should be noted that the median distance in mortar is shorter than in cement paste at the same point intensity. Presence of aggregate particles reduces room for air bubbles. The occupation by aggregate particles simultaneously produces small local regions where large air bubbles cannot exist. As a result, many small air bubbles get closer each other in the local regions while the total air content is increased. 3.3 Correlation between the spacing factor and a characteristic distance of the points The median distance mentioned above is defined as a representative distance between points in a point process. However, actual sizes of air voids in real mortar specimens are ignored. On the other hand, the spacing factor of a traditional parameter is defined as the mean farthest distance from any point in cement paste matrix to the surface of a bubble. The mean radius of air bubbles is implicitly taken into account in the calculation of the spacing factor. Therefore, to compare the median distances with the spacing factors, the median distance is modified to

87



consider the sizes of air voids in the point processes. A characteristic spacing L’ is simply defined by subtracting half of the average diameter of air voids from the median distance (Eq. (5), Fig. 2(b)).

)5(250' DRL

Where D is the average diameter of voids. The spacing factors L and the characteristic spacing L’ are given in Table 3. Table 3 Air parameters in the point process obtained from images

Ref AE[1] AE[2] AE[3] AE[4] AE[5] Ref AE[1] AE[2] Ref AE[1] AE[2]

Median distance R50 710 508 349 324 293 296 572 314 273 490 344 267

648 439 289 265 243 233 500 242 202 402 265 189

558 432 298 276 250 251 441 244 214 358 266 199

90 7 9 11 7 18 59 2 12 44 1 10

Cement paste (C:S=1:0) Mortar (C:S=1:1) Mortar (C:S=1:2)

Relationship between L and L’ in cement pastes and mortars is shown in Fig. 6. There exists a strong correlation between them. Furthermore, they are almost the comparable values. Compared to time-consuming and boring procedure to obtain L, the derivation of L’ is much easier. Therefore, instead of obtaining L directly following the specified procedure, the spacing factor L could be estimated from the characteristic spacing L’ of point processes in scanned images. In other words, the spacing factor used generally corresponds well with an average distance of a point process, which is realized in actual mortar.

Fig. 6 Relationship between characteristic distances L’ and conventional spacing factors L 3.4 Simulation of air voids distribution as a thinned point process Fig. 7 compares the two nearest neighbour distance functions of air voids. The function GS(r) is calculated for thinned point patterns, in which points on aggregate particles are deleted from a basic point process. This basic process is simulated as a random process with the same point intensity as in the original image of air voids. The function GM(r) is directly obtained from the original images. There is a difference between the two functions in the range of short

88



distances. The curves of GS(r) are plotted upper than those of GM(r) in the short range. This is because, as mentioned before, two points can get closer in the simulated patterns since the size of air voids is ignored. Therefore, a probability of finding other points in the short range is greater in the simulated process than in the real distribution. However, there is little difference between the two functions in the range beyond certain distances.

Fig. 7 Comparison of nearest neighbour distance function of air voids in mortars (GM:real distribution in mortars GS:point patterns simulated by thinning points from basic distribution) Fig. 8 shows the relationship between the median distance RS50 of the thinned patterns and the one R50 of the real distribution of air voids in mortars. There again exists a strong correlation between them. Furthermore, they are almost similar values. These results suggest an alternative way to estimate the spacing factor. Once random points with a given point intensity are generated as to simulate air voids distribution, and then they are thinned to leave points in a cement matrix, we can represent spatial distribution of air voids in real mortars. Then following the procedure mentioned above, the median distance of the thinned pattern can be calculated. Fig. 9 shows relationship between the median distances RS50 in the thinned patterns (i.e. simulated distribution) and the spacing factors L in real mortars. As expected, there exists a strong correlation between them.

89



Fig. 8 Relationship between median distances in thinned and real processes of air voids

Fig. 9 Relationship between spacing factors and median distances in simulated processes for air voids

At the present, the spacing factor is still regarded as an important parameter to evaluate the air-void system in concrete. However, it seems that the procedure specified in ASTM C457 is tedious and is built on an old-fashioned technique. Advanced equipment for measuring it automatically is not widely used. However, if a simulation technique of point processes is used for generating air voids as random points, and if those points are thinned to retain air voids in cement paste matrix, then it is possible to represent real distribution of air voids. Using the simulated point patterns, it is also possible to estimate the spacing factor L. From the stereological viewpoint, the number of points in a cross section of concrete cannot be simply interpreted as air volume. However, in practice, there is a good correlation between the point intensity and the air content in concrete as long as air-entrained concrete is produced properly [1, 5]. Thus, once a necessary air content is specified, for example 5% or more, the corresponding point intensity is easily determined from the correlation with air contents, which is obtained in advance. Then the specified point intensity can be used as input data to simulate air bubble spatial dispersion in mixtures. This simulation can be executed for mortars and concrete since the both mixtures can be treated by the same procedure of interrupted point processes. Effectiveness of AE agents must be examined when frost resistance or durability of concrete is an important concern. The spacing factor must be evaluated as a routine parameter to judge the resistance. Following the procedure mentioned above, a characteristic distance that is closely correlated with the spacing factor is easily obtained using simple scanned images and image operations. In particular, images scanned with a commercial flatbed scanner seem to have sufficient resolution, which is comparable to images specified in the ASTM C457 method. Application of the point process statistics to low magnification images of air voids is a useful way to evaluate spacing between air voids. 4. Conclusions Air voids distribution in mortars was analyzed as a spatial point process in 2D space. Air voids in scanned images were converted to the set of points, which stood for centroids of each void. The nearest neiggbour distance function was calculated to compare different air-void

90



systems in mortars with different air contents. A characteristic distance in the point process was defined by a probability specified in the nearest neighbour distance functions. The characteristic distances in mortars were compared with the traditinal spacing factor of air void. Futhermore, aiming at easy estimation of the spacing factors, the procedure based on random simulation and image operation for points of interest was proposed. The simulated patterns of air voids were also compared with real distribution of air voids in cast mortar specimens. The major results obtained in this study are as follows; (1) The median distances in mortars are shorter than those in cement pastes at the same point

intensity. The presence of aggregate particles makes air voids approach each other. (2) The median distance between air voids in real mortars has a strong correlation with the

spacing factor that has been used as a parameter for air-void system. (3) Spatial distribution of air voids in mortar is easily simulated by the thinning operation of

random point processes. Simple deletion of points on aggregate particles results in almost the same distribution patterns as in real mortars.

(4) The spacing factor is estimated by means of the median distance of simulated point patterns. Treating air voids as a spatial point process is a useful way to characterize air void system in mortars.

Acknowledgment This study was supported by JSPS KAKENHI(Grant Number 24560564). References [1] Mayercsik, N.P. et al., A probabilistic technique for entrained air void analysis in

hardened concrete, Cem. Concr. Res., 59 (2014), pp.16-23 [2] Stoyan, D., Interrupted point processes, Biom. J. , 21(7) (1979), pp.607-610 [3] Stoyan, D. et al., Stochastic geometry and its applications, John Wiley and Sons,

Chichester, UK (2008) [4] Konagai, N. et al., Determination of air void parameters in hardened concrete from

measurement of intercepted air void’s area, Monthly report of research institute of civil engineering, No.396 (1986) pp.2-8 (in Japanese)

[5] Murotani, T. et al., Quantitative evaluation of spatial distribution of air voids in cement pastes by point process statistics, Proceedings of the Japan Concrete Institute, 37(1) (2015), pp.493-498 (in Japanese)

91



THE INFLUENCE OF CARBONATION AND AGE ON SALT FROST SCALING OF CONCRETE WITH MINERAL ADDITIONS Ingemar Löfgren (1 & 2), Oskar Esping (1) & Anders Lindvall (1) (1) Tomas Concrete Group, Gothenburg, Sweden (2) Chalmers University of Technology, Gothenburg, Sweden Abstract Resistance to salt frost scaling is tested by accelerated methods such as CEN/TS 12390-9 which originally were developed for Portland cement concrete. However, it has been shown that ageing and coupled deterioration mechanisms, like carbonation or leaching, alter the frost resistance. An example is concrete with high amount of slag where the frost resistance is reduced when the concrete is carbonated. Hence, modifications to the test methods have been proposed to take these effects into account and often an accelerated carbonation at an early age have been used. Though, it has been found that the accelerated tests show a much more negative effect than what is experienced in field conditions. This paper presents results from a laboratory study of concrete with mineral additions at different dosages and water/binder ratios which have been exposed to accelerated carbonation at 1% CO2-concentration at different ages. The results show that exposing the specimens to accelerated carbonation at a young age will result in an increased scaling but that the carbonation depths corresponds to 10 year natural exposure. By increasing the age before the accelerated carbonation exposure the scaling is significantly reduced and the salt frost scaling resistance seems to correlate better with field observations. 1. Introduction The resistance of concrete to salt frost scaling is tested by accelerated methods such as CEN/TS 12390-9 [1] and SS 137244 [2], which originally were developed based on the experience of Portland cement concrete [3] [4]. The testing regime is being under review, partly due to that it does not consider ageing effects, such as changes to pore structure, micro cracking, leaching and the effect of carbonation [3] [4] [5]. With increasing use of mineral additions, such as slag (GGBS) and fly ash (FA), for reducing the environmental footprint and improve resistance to reinforcement corrosion this type of test methods need to be modified so that it can safely and adequately be used for concrete with mineral additions. Moreover, the test results also need to be correlated with the performance in field conditions [6] [7].

92



2. Experimental procedure 2.1 Materials and mix designs The concrete mixture proportions are listed in Table 1; where the aggregate was a granitic type, the superplasticizer (SP) was polycarboxylate-based (PCE) with a dry content of 18%, and the air entraining agent was a synthetic surfactant with a dry content of 1.1%. For the experiments, two different Portland cements (CEM I, C2 & C3), one Portland-fly ash cement (CEM II/A-V, C5), one Portland-slag cement (CEM II/B-S, C6), one blast furnace cement (CEM III/A, C4), and two different GGBS (S1 & S2) were used. See Table 2 for properties of the cement and GGBS used. For the mixes with GGBS an efficiency factor (k-value) of 1.0 was used, i.e. comparison is made at equal water/binder (w/b) ratios. Table 1: Concrete mix proportions, kg/m3 (if not otherwise stated). Binder (see Table 2) w/b Cement GGBS Aggregates SP AEA Air content

C2 0.45 400 1 705 2.00 0.040 5.2 % C3 0.45 400 1 703 2.00 0.040 5.6 %

C2+20%S1 0.45 320 80 1 699 2.00 0.040 4.9 % C2+30%S1 0.45 280 120 1 696 2.00 0.040 5.1 % C2+40%S1 0.45 240 160 1 693 2.00 0.052 5.0 % C2+60%S1 0.45 160 240 1 687 2.00 0.040 5.8 % C2+20%S2 0.45 320 80 1 687 2.00 0.040 5.8 % C2+40%S2 0.45 240 160 1 703 2.00 0.040 5.5 %

C6 0.45 400 1 679 2.00 0.040 5.0 % C4 0.45 400 1 688 2.00 0.040 5.1 % C5 0.45 400 1 704 2.00 0.040 5.8 % C5 0.40 425 1 708 2.55 0.053 5.7 % C2 0.40 425 1 709 2.55 0.053 5.5 %

C2+20%S1 0.40 340 85 1 703 2.55 0.043 6.0 % C2+30%S1 0.40 298 128 1 700 2.55 0.043 6.2 % C1+40%S1 0.40 255 170 1 697 2.55 0.043 6.2 %

C4 0.40 425 1 690 2.98 0.036 6.3 % C6 0.40 425 1 700 2.98 0.036 6.2 %

93



Table 2: Materials. ID Type

Acc. to EN 197-1 Density kg/m3

Blaine m2/kg

CaO M.-%

SiO2 M.-%

Al2O3 M.-%

Fe2O3 M.-%

Na2Oeqv M.-%

C1 CEM I 42,5 N SR3 MH/LA 3 200 330 64 22 3.7 4.5 0.51 C2 CEM I 42,5 N SR3 MH/LA 3 160 330 64 22 3.3 4.6 0.45 C3 CEM I 52,5 N 3 140 420 63 19 4.3 3.1 0.90 C4 CEM III/A 42,5 N/NA 3 000 450 52 28 8.9 1.2 0.70 C5 CEM II/A-V 42,5 N MH/LA 3 040 370 0.85 C6 CEM II/B-S 52,5 N 3 060 460 56 25 6.3 2.1 0.80 S1 GGBS 2 900 420 40 35 12 1.20 S2 GGBS 2 920 500 31 34 13 0.90 C4: Contains about 49% GGBS. C5: Is a Portland-fly ash cement with app. 14% and with the clinker of C1. C6: Contains about 33% GGBS 2.2 Freeze-thaw test procedure The standard slab test procedure for freeze-thaw testing as described in CEN TS 12390-9 [1] and SS 137244 [2] were used to assess the salt-frost scaling resistance on a cut surface. Four different preconditioning regimes have been studied: Standard procedure - 31d Std: From demoulding (24±2 h) the cube is stored in water

until the age of 7 days, and then stored in climate chamber (20±2°C and RH 65±5%) until a 50 mm thick specimen is cut at an age of 21 days. The slab is placed in a climate chamber (20±2°C and RH 65±5%) until it is 28 d old. At 28 d, 3 mm de-ionized water is poured on the top surface and the specimen is saturated for 72±2 h. At the age 31 d the freeze thaw cycles are started.

Modified standard procedure – 31d C: As standard procedure, but from the age 21 d until 28 d the cut specimen is placed in a climate chamber with 1.0 % CO2-concentration (20±2°C and RH 65±5%). At 28 d, 3 mm de-ionized water is poured on the top surface and the specimen is saturated for 72±2 h. At the age 31 d the freeze thaw cycles are started.

45 d curing regime – 45d C: From demoulding (24±2 h) the cube is stored in water until the age of 21 days. Then the cube is stored in a climate chamber (20±2°C and RH 65±5%) until specimen is cut at an age of 35 days. At the age 35 d the cut specimen is placed in a climate chamber with 1.0 % CO2-concentration (20±2°C and RH 65±5%) until it is 42 d old. At 42 d, 3 mm de-ionized water is poured on the top surface and the specimen is saturated for 72±2 h. At the age 45 d the freeze thaw cycles are started.

87 d curing regime – 87d C: From demoulding (24±2 h) the cube is stored in water until the age of 63 days. Then the cube is stored in a climate chamber (20±2°C and RH 65±5%) until specimen is cut at an age of 77 days. From the age 77 d until 84 d the cut specimen is placed in a climate chamber with 1.0 % CO2-concentration (20±2°C and RH 65±5%). At 84 d, 3 mm de-ionized water is poured on the top surface and the specimen is saturated for 72±2 h. At the age 87 d the freeze thaw cycles are started. This procedure was only used for the mixes with GGBS.

94



For all specimens, the weights were recorded immediately before and after the saturation with de-ionized water to determine the water uptake. Moreover, on accompanying specimens, cured in the same manner as the specimens for freeze-thaw testing, the carbonation depth was determined after 7, 14 and 28 d exposure to 1.0% CO2 by spraying a phenolphthalein solution on a freshly split concrete surface. In addition, the compressive strength of the concretes were determined at 28, 56, and 180 days. 3. Results 3.1 Compressive strength The compressive cube strength (at age 28, 56, and 180 days for water cured specimens) for the concrete mixes are presented in Figure 1 (average of two specimens). For the mixes with higher amounts of GGBS the compressive strength at 28 and 56 days are lower. However, at 180 days the differences becomes smaller.

Figure 1: Compressive strength (water cured cubes) at 28, 56, and 180 days. 3.2 Carbonation, water uptake & pre-conditioning In Figure 2 the carbonation depth for the different concrete mixes and curing conditions are presented. The carbonation depths are, as expected, influenced by the curing conditions and age at exposure and are lower for more mature specimens. Moreover, with increasing GGBS content the carbonation depth increase and with decreased w/b ratio it decreases and is considerably lower, especially for the more mature specimens (curing condition 87d C). In Figure 3 the water uptake (after 72±2 h saturation) for the different concrete mixes and curing conditions is presented. Compared with the standard curing condition (31d Std), almost all concrete mixes show an increased water uptake for the early carbonated specimens (31d C, start of carbonation at 21 d). For the curing condition 45d C (age 35 d when start of carbonation) the water uptake is reduced or almost unaffected for most of the mixes. The only deviation is for cement C5 (CEM II/A-V) which show an increased water uptake. Finally, for the curing condition 87d C (age 77 d when start of carbonation) the water uptake is reduced for all mixes. From these results it can be seen that the influence of hydration (age) is much more apparent for high GGBS contents.

25

30

35

40

45

50

55

60

Com

pres

sive

stre

ngth

cub

e [M

Pa]

28d 56d 180d

95



Figure 2: Comparison of carbonation depth after 1 week in 1% CO2 for the different curing conditions.

Figure 3: Comparison of the measured water uptake after 72±2 h surface saturation for the different curing conditions. 3.2 Surface scaling & pre-conditioning The surface scaling after 56 and 112 cycles are presented in Figure 4 and Figure 5. The acceptance criteria for “good” frost resistance according to SS 137244 [2] is shown in the figures. As can be seen in the figures, in all cases the scaling compared with the standard procedure increased for the carbonated specimens independent of curing condition, with the exception of mix C2+20%S1 0.40 for 45d C and 87d C curing. The increase in scaling is generally highest for 31d C, i.e. when carbonation starts on specimens at an age of 21 d.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Car

bona

tion

dept

h [m

m]

31d C45d C87d C

0.00.10.20.30.40.50.60.70.80.91.0

Wat

er u

ptak

e [k

g/m

2 ]

31d Std 31d C45d C 87d C

96



Figure 4: Comparison of the surface scaling (log scale) after 56 cycles for the different curing conditions. The ratio of the mass of scaled material after 56 cycles to 28 cycles (m56/m28) were for all tested concretes <2.0 except for C4 0.40 and C6 0.40 with curing condition 87d C.

Figure 5: Comparison of the surface scaling (log scale) after 112 cycles for the different curing conditions. For some of the tested concretes the scaling for curing condition 31d C after 112 cycles could not be recorded due to large scaling and leakage problems.

0.001

0.01

0.1

1

Surf

ace

scal

ing

afte

r 56

cycl

es [k

g/m

2 ] 31d Std 31d C45d C 87d C

0.2

0.5"Good" frost resistance [2]

"Good" frost resistance if m56/m28<2 [2]

0.001

0.01

0.1

1

Surf

ace

scal

ing

afte

r 112

cyc

les [

kg/m

2 ] 31d Std 31d C45d C 87d C

0.5"Good" frost resistance [2]

97



4. Discussion As can be seen in Figure 4 and 5, the carbonation and age of the specimens has a big influence on the surface scaling. It has been known for a long time that with high amounts of GGBS the salt-frost scaling resistance is low and is negatively affected by carbonation [8] [3]. At what amount this negative effect starts to have influence have been said to be at about 30 to 40% GGBS of total binder content based on field data [9] [10]. Based on the results from this study, for the tested concrete mixes and materials, at approximately 30 to 40% GGBS of the total binder the negative effect starts to have influence; but it is dependent on w/b ratio and the properties of the GGBS. In Figure 6 the difference in behaviour between 20 and 40% GGBS for w/b ratio 0.45 is shown. In both cases there is a big difference for the early carbonated specimens (31d C). At the higher GGBS addition also the 45d C specimens show a large scaling. What is interesting is that the major part of the scaling, for 40% GGBS, occurs during the first 7 days. Moreover, for the more mature specimen, 87d C (carbonation started at 77 days) there is only a moderate increase in the scaling. For the mixes with 20% GGBS, all meet the requirement for “good” frost resistance according to [2]. With 40% GGBS it is only the standard procedure (31d Std) and the 87d C that meets the requirement for “good” frost resistance according to [2] with GGBS S1. For GGBS S2 the criteria is met at 30% GGBS but not at 40%. However, there is a difference between GGBS S1 and S2, where S2 have a larger carbonation depth at dosage 40% compared to S1. This could be due to the different chemistry of the GGBS with a lower Ca/Si ration for S2 which could promote more rapid carbonation.

(a) (b) Figure 6: Example of surface scaling and the effect on curing conditions at different amounts of GGBS (note the different scales on the y-axis). (a) For mix C2+20%S1 w/b = 0.45 and (b) for mix C2+40%S1 w/b = 0.45. In Figure 7 a comparison is made for a GGBS addition of 30% at w/b ratio of 0.45 and 0.40. Also here it can be seen that at w/b 0.45 that the scaling is much higher for 31d C and 45d C. But at w/b 0.40 difference are smaller and all meet the requirement for “good” frost resistance, while at w/b 0.45 it is only the standard procedure (31d Std) and the 87d C that meets the requirement. Figure 8 shows the results for two mixes with CEM II cements (C6 and C5) where an accelerated scaling occurs after a number of cycles for the more mature specimens. The reason for this type of behaviour is not clear but it could either be due to the prolonged water curing, giving rise to an increased water saturation, or due to that the concrete becomes more dense with prolonged curing and age, which could lead to increased hydraulic pressure.

0.0

0.1

0.2

0.3

0.4

0.5

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d C45d C

87d C31d Std

mix C2+20%S1 w/b 0.45

0.00.20.40.60.81.01.2

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d Std

87d C

45d C31d C

mix C2+40%S1 w/b 0.45

98



(a) (b) Figure 7: Example of surface scaling and the effect on curing conditions and w/b ratio (note the different scales on the y-axis). (a) For mix C2+30%S1 w/b = 0.45 and (b) for mix C2+30%S1 w/b = 0.40.

(a) (b) Figure 8: Example of surface scaling and the effect on curing conditions on late accelerated scaling for mature specimens. (a) For mix C6 w/b = 0.40 and (b) for mix C5 w/b = 0.45. It has been suggested that the amount of scaling after carbonation correlates to the carbonation depth [8] and also that the carbonation will cause a coarsening of the pore structure leading to a higher water absorption for concrete with GGBS [3]. In Figure 9(a) the scaling at 56 cycles have been plotted against the carbonation depth for all mixes with GGBS and in Figure 9(b) the water uptake has been plotted against carbonation depth. The general trend is that scaling increase with increasing carbonation depth, but the correlation is only high (R2 0.7 to 0.9) for the scaling and carbonation depth for 31d C (R2 = 0.83) and 45d C (R2 = 0.75). For 87d C the correlation is much lower. For the water uptake, there is trend indicating a higher water uptake with increased carbonation depth, but the correlation is moderate (R2 0.5 to 0.7) or low (R2 0.3 to 0.5). With respect to the measured carbonation depths for the accelerated carbonation (7 days in 1% CO2) the measured depths (see Figure 2 and 9) should be compared to what is expected in field conditions (exposure conditions corresponding to XF4). Data from specimens exposed to atmospheric CO2 during 11 year exposure at a field stations [10] showed a carbonation depth of about 1.1 mm with 30% GGBS at w/b 0.50 and 3.8 mm for a CEM III/B at w/b 0.50. At w/b 0.40 the carbonation depths were approximately 0.7 mm respectively 2.1 mm.

0.00.10.20.30.40.50.6

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d Std

87d C

45d C31d C

mix C2+30%S1 w/b 0.45

0.0

0.1

0.2

0.3

0.4

0.5

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d C45d C

87d C31d Std

mix C2+30%S1 w/b 0.40

0.0

0.1

0.2

0.3

0.4

0.5

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d C45d C

87d C

31d Std

mix C6 (CEM II/B-S) w/b 0.40

0.0

0.1

0.2

0.3

0.4

0.5

7 14 28 42 56 70 84 98 112Surf

ace

scal

ing

[kg/

m2 ]

Cycles [days]

31d C

45d C

31d Std

mix C5 (CEM II/A-V) w/b 0.45

99



(a) (b) Figure 9: Correlation between scaling after 56 cycles and carbonation (a) and between water uptake and carbonation (b) for all mixes containing GGBS. In general, the coupled effect of carbonation and frost and ageing is complex. In this study all the tested concretes hade a larger scaling for specimens exposed to carbonation, this was also the case for the pure Portland cement and the fly ash cement, but the increase was rather small. For the concrete mixes with GGBS the increased scaling for carbonated specimens was dependent on the amount of GGBS and at low dosages (20%) and at low w/b the increase was small. In the literature there are different opinions and results reported with respect to ageing, [3] [5] [11], both positive and negative effects of ageing have been reported for concrete with Portland cement as well as for concrete with fly ash and GGBS. In light of the results from this study and what has been reported in the literature it seems plausible that the main mechanism(s) causing a change in the salt-frost scaling resistance is related to the coarsening of the pore system caused by carbonation at high GGBS amounts [3]. Prolonged curing or lowering the w/b ratio reduces the thickness of the carbonated layer and the scaling. However, it cannot be ruled out that there also could be a chemical effect. It has been reported that decomposition of monsulfate to ettrengite, due to carbonation or due to partly transformation into monochloride, can occur during the freeze-thaw cycles [12]. 5. Conclusions The effect of accelerated carbonation, 1 week in an atmosphere with 1% CO2-concentration, on salt-frost scaling and the influence of different curing conditions has been studied. With the materials used and w/b ratios investigated it is clear that at 20% GGBS of total binder there is very little effect of carbonation on the frost resistance. Even with early accelerated carbonation at age of 21 days (31d C) all the tested concretes meet “good” scaling resistance or better after 112 cycles. For higher amounts of GGBS the early accelerated carbonation lead to increased scaling and especially at high amounts of GGBS (>40%). With prolonged curing the scaling generally decreased, but at higher GGBS content there was not that big difference between 31d C and 45d C (21 or 35 days when starting carbonation). For the longest curing time 87d C (77 days when starting carbonation) the scaling was lower and GGBS contents of 30% achieved good frost resistance and this was also the case for some of the mixes with 40% GGBS, but the scaling was higher. But it should also be pointed out that for the mixes with a higher slag content the compressive strength was much lower as an efficiency factor (k-value) of 1.0 was used. The following conclusions can be made:

R² = 0.83

R² = 0.75

R² = 0.590.0

0.5

1.0

1.5

2.0

2.5

3.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Scal

ing

56 c

ycle

s [kg

/m2 ]

Carbonation depth [mm]

31d C45d C87d C

R² = 0.62

R² = 0.54

R² = 0.400.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Wat

er u

ptak

e [k

g/m

2 ]

Carbonation depth [mm]

31d C45d C87d C

100



The correlation between carbonation depth and scaling was high for specimens exposed early at the age of 21 and 35 days but lower for the specimens exposed at 77 days.

With 20% GGBS of the total binder it was found that carbonation did not have a significant effect on the salt-frost scaling, even at early age carbonation.

A GGBS content of about 30 to 40% is reasonable with respect to the salt-frost scaling resistance, but the testing and carbonation should not be done to early as this will produce a carbonation depth corresponding to more than 10 years natural carbonation.

A prolonged curing regime before commencement of carbonation is needed if the test results should be realistic. In this study an age of 77 days when starting carbonation and 87 for start of frost cycles seems to have given reasonable results.

Acknowledgment: This project has been financially supported by the Swedish Transport Administration. References [1] CEN/TS 12390-9, Testing hardened concrete – Part 9: Freeze/thaw resistance – Scaling,

CEN, 2006. [2] SS 137244, Concrete testing- Hardened concrete- Frost resistance, Swedish Standards

Institution (SIS), 4th edition, Stockholm, Sweden 2005. [3] Utgenannt, P., The influence of ageing on the salt-frost resistance of concrete, Report

TVBM-1021, Division of Building Materials, Lund Institute of Technology, Lund, 2004. [4] Rønning, T, Concrete Freeze-Thaw Scaling Resistance Testing Experience and

Development of a Testing Regime & Acceptance Criteria, In Workshop Proc. ‘Durability aspects of fly ash and slag in concrete’ Nordic Miniseminar 15-16 Feb. 2016, Oslo.

[5] Ferreira, M., Leivo, M., Kuosa, H., Holt, E., The effect of by-products on frost-salt durability of aged concrete, In Workshop Proc. ‘Durability aspects of fly ash and slag in concrete’ Nordic Miniseminar 15-16 Feb. 2016, Oslo.

[6] Boos, P, Eriksson, B.E., Giergiczny, Z., & Härdtl, R., Laboratory Testing of Frost Resistance – Do the Tests Indicate the Real Performance of Blended Cements? In proc. from the 12th Int. Congress on the Chemistry of Cement, Montreal 8-13 July, 2007.

[7] Boyd, A.J. & Hooton, R.D., Long-Term Scaling Performance of Concretes Containing Supplementary Cementing Materials, J. of Mat. in Civil Eng., Vol. 19, pp 820-825, 2007.

[8] Stark, J. & Ludwig, H.-M., Freeze-thaw and freeze-deicing salt resistance of concretes containing cement rich in granulated blast furnace slag, ACI Mat. J., Vol. 94, pp. 47-55.

[9] Hooton, R.D., & Boyd, A., Effect of finishing, forming and curing on de-icer salt scaling resistance of concretes. Int. RILEM Workshop on Resistance of Concrete to Freezing and Thawing with or without De-icing Chemicals, Essen, Germany. pp. 174–183, 1997.

[10] Utgenannt, P., Frost resistance of concrete - Experience from three field exposure sites. In Workshop proc. no. 8: Nordic Exposure sites - input to revision of EN 206-1, Hirtshals, Denmark, Nov. 12-14, 2008.

[11] Rønning, T.F., Freeze-Thaw Resistance of Concrete. Effect of: Curing Conditions, Moisture Exchange and Materials. PhD-thesis, The Norwegian Inst. of Tech., 2001.

[12] Ludwig, H.M., Zur rolle von Phasenumwandelung bei frost- und frost-tausalz-belastung von beton, Diss., Hochsch. Archit. Bauwes. Weimar Univ.

101



MODELING FREEZING OF CEMENTITIOUS MATERIALS BY CONSIDERING PROCESS KINETICS Francesco Pesavento (1), Dariusz Gawin (2) (1) University of Padova, Padova, Italy

Abstract In this work we present a new numerical model for the simulation of the behaviour of partially saturated porous materials, exposed to temperatures below the freezing point. Usually in modelling such a kind of phenomenon, the assumption that there exists thermodynamic equilibrium between the ice and liquid water, is applied. In this way, one can use the equilibrium curve, which describes liquid water content in the pores being in equilibrium with ice at a given temperature T. Here the water freezing / melting process in the material pores is modelled as a non-equilibrium process, the kinetics of which is governed by an evolution law, obtained from the second law of Thermodynamics. As far as the effective stress principle is concerned, during freezing of water at thermodynamic equilibrium, the pressure difference on the interface between ice and liquid water in the pores can be calculated by means of the Gibbs-Thomson law. With these measures of pressure it is possible to evaluate the effective stress on the solid skeleton of the porous medium due to the simultaneous presence of gas, water and ice. Some relevant numerical cases will be presented to show the effectiveness of the formulated model. 1. Introduction In this work we present a new numerical model for the simulation of the behaviour of porous materials under cooling, with particular regard to cementitious materials. The latter ones are characterized by a fine inner microstructure, formed of C–S–H gels, mineral crystals of various compositions and pores of different sizes which can be fully or partially saturated with water. Unlike bulk water which normally freezes at temperature T= 0 °C (the freezing point of water), phase change of pore water - ice in building materials, due to the pore confinement,

102



occurs in a range of temperatures below 0 °C, dependent on both the material pore size distribution and the actual moisture content. Pore water freezing is initiated by heterogeneous nucleation and then followed by a progressive penetration of the ice front from big pores to small pores. Moreover in the case of pore water, different ice content may be observed at the same temperature during the phase transitions (i.e. freezing and melting) due to different curvature of water – ice interface, what causes a sort of hysteresis [1,2] (which is also related to the so-called ink bottle effect). In modelling of water freezing in porous building materials both the water undercooling and freezing-thawing hysteresis are usually not considered, and thermodynamic equilibrium between water and ice is assumed. On the contrary, here the water freezing / melting in the material pores is modelled as a non-equilibrium process (but close to thermodynamic equilibrium), the kinetics of one is governed by a first order evolution law, obtained from the second law of Thermodynamics. During freezing of water in a partially saturated porous material (i.e. filled partly with gas phase), liquid water occupies the whole volume of pores with the smallest radii, while ice is formed at the interface between liquid and solid water (gradually progressing into finer pores) and on the ice / gas interface. The liquid water can flow through a thin film of physically adsorbed water between the skeleton and ice crystal [1,2], hence water pressure does not increase during freezing in a partially saturated material. The model defined in such a way, it is able to take into account several phenomena leading to the deformation of the porous material and to the internal microcracking: (i) the expulsion of unfrozen water from the freezing sites towards the airvoids, which is usually related to the 9% expansion of water upon freezing, (ii) the crystallization process which is strictly related to the interactions between the phases/constituents and the microstructure of the material; (iii) the transfer of liquid water from the unfrozen pores to already frozen sites (i.e. cryo-suction process); (iv) the thermo-mechanical interactions between the solid skeleton and the water in its various forms (liquid, gaseous, solid), [3-5]. About the mechanical effects of the interfacial interactions between the constituents, the model uses the effective stress principle: at temperatures when only liquid water and moist air are present in the pores (i.e. no ice is present), a usual relation between the water pressure and gas pressure, considering also disjoining pressure. During freezing of water at thermodynamic equilibrium, the pressure difference on the interface between ice and liquid water in the pores can calculated by means of the Gibbs-Thomson law [6]. The crystallization pressure, exerted on material skeleton by a crystal of ice situated in a pore can be derived by the relationships proposed by Scherer [2]. With these measures of pressure it is possible to evaluate the effective stress on the solid skeleton of the porous medium due to the simultaneous presence of gas, water and ice. The model describing the development of freezing/melting process is defined in the framework of the Porous Media Mechanics, which allows for considering concrete as a multiphase porous material. So, the freezing evolution equations are introduced in a more general model obtained in the context of the Thermodynamically Constrained Averaging Theory (TCAT), which considers several phases and components: the solid skeleton of the material, the water species in its three different phases (liquid, vapour and solid, i.e. ice) and the dry air, [7]. A detailed description of the model here proposed and a comparison with the existing ones in literature, can be found in [9].

103



2. Kinetics of water freezing in porous materials Water freezing in a capillary-porous material is a very complex physicochemical process, rather difficult to model numerically. For modelling such a kind of process, usually a basic assumption is made by considering the thermodynamic equilibrium between the constituents and the phases involved. This allows for defining a sort of “equilibrium curve”, which describes liquid water content in the material pores (or alternatively the saturation degree with liquid water eq

wS T ) being in equilibrium with the solid water (ice) at a given temperature T during very slow freezing:

/2 w gweqw w eq

frS T S

r T (1)

with

/*

2 w iceeqfr

fr mr T

T T (2)

where 2

/ 0.0757 J/mw gw is the liquid water - vapour interface energy (surface tension of water) at the melting temperature of free water, *

mT = 273.15 K, 2/ 0.0409 J/mw ice the liquid

water - ice interface energy at T= *mT , and 1.2 MPa/Kfr the freezing entropy at T= *

mT . eqfrr T is the entrance radius of pores, which are accessible for crystal of frozen water

entering from larger pores at temperature T. The equilibrium law in the form of eq. (1) is due to [3-5] and can be obtained both from experimental tests or from thermodynamics (e.g. 1-5,8). Here the water freezing/melting in a capillary-porous material is modelled as a non-equilibrium process, the kinetics of which is governed by a linear evolution law, obtained from the second law of Thermodynamics in the following form [6],

, (3) where is the rate of internal production of molar entropy due to the freezing, the rate of freezing, and the process affinity is given by the following equation [6],

, ,w icefr w iceA T p T p , (4)

with w and ice being chemical potentials of liquid water and ice (solid water), respectively. By considering: i) the definition of the chemical potentials and equilibrium condition for the transition of bulk phases at temperature T0 and pressure p0;

104



ii) that in porous materials, due to curved interfaces between the phases and interaction of pore walls, the water and ice pressures are different; iii) a simplified form of the affinity Afr; iv) the relationships at equilibrium, eqs. (1,2); it is possible to formulate the equations describing the kinetics of the freezing/melting processes taking place in the pores of the material (see [9] for further details). These evolution equations have the form described in the following. Water freezing proceeds when the actual liquid water saturation degree is greater than the equilibrium saturation degree corresponding to the actual temperature of water, eq

w wS S T , which is given by equation (1). This condition may be expressed in terms of capillary pressure as, ,

c cfr fr eqp p T , with c

frp - the actual value of the capillary pressure during freezing process (far from equilibrium) and ,

cfr eqp - the corresponding value of capillary pressure under

equilibrium. Hence the evolution of the ice mass source due to freezing of liquid water, , can be described through the following law:

, (5)

where, n is the porosity of the material, w and vw are the density and the molar volume of the water respectively, R is the universal gas constant and, finally, fr is the characteristic time of freezing which is related to the affinity of the process and to its reaction rate. Melting of ice may be described in a similar way, but considering different (smaller) curvature of the ice / water meniscus during melting of ice (i.e. intrusion of the wetting phase into the non-wetting one, see [6]), , than during freezing of water, hence one may

express the equilibrium value of pressure difference during melting of ice, , and the

ice mass source due to melting as follows,

, , /c cm eq fr eq wp p S , (6)

(7)

where m is the characteristic time of melting, and , ( )cm eqp T is the capillary pressure at

equilibrium during the melting process (different than the one during freezing). Therefore, this formulation takes into account that water freezing starts at higher value of water saturation degree (corresponding to 2 times higher pore radius entrance, rpore ) than melting of ice. Hence freezing of water in the pores of given entrance radius (i.e. water saturation degree) starts at lower temperature than melting of ice in these pores, what is also confirmed by several experiments, see e.g. [6] or Fig. 1. This effect is considered through the function wS taking into account that the interface liquid water/ice has different curvature depending on we are considering a freezing or a melting process. In modelling, the function can be identified either in a simplified way, as done above (i.e. by assuming spherical shape

105



of liquid water / ice meniscus during freezing and cylindrical one during melting, e.g. [6]), or using directly experimental data, for example those obtained with the Differential Scanning Calorimetry, Fig. 1.

Figure 1: Volume of ice in the pores of a fully water saturated cement mortar at different temperature,

during freezing and melting, measured by means of Differential Scanning Calorimeter (DSC)

Figure 2: Difference of ice and water pressures at thermodynamic equilibrium, during freezing of water and melting of ice shown as function of undercooling ( )

Figure 2 shows the equilibrium values of pressure difference between water and ice at different temperatures, during freezing - , ( )c

fr eqp T (denoted as “freezing”), and during melting - , ( )c

m eqp T following simplified approach (“melting =1/2”) and using experimental data (“melting =exp.”). For further details the reader is referred to [9].

106



3. General model (governing equations) The model proposed here, is obtained from microscopic scale by applying the so-called Thermodynamically Constrained Averaging Theory (TCAT, [7]). This approach allows for taking into account both bulk phases and interfaces of the multiphase system and involves averaging established microscale thermodynamic principles to the macroscale. In doing so, it inherently assures consistency between microscale and macroscale forms. Moreover, one can obtain some important thermodynamic restrictions imposed on the evolution equations describing the material deterioration. For the sake of brevity hereafter only the final form of the governing equations for the multiphase system is shown, taking already into account the constitutive relationships. The model is based on four governing equations and the following primary variables: pg-gas pressure, pc- capillary pressure, T-temperature, u-displacements. Depending on the chemical-mechanical degradation processes one wishes to consider, the basic model can be supplemented also with a set of evolution equations and the related internal variables. The procedure for obtaining these sets of equations can be found in [7,9]. - Water mass balance equation (solid+vapour+liquid) Taking into account the diffusion of water vapour, the darcian flow of liquid water, the deformation of the solid skeleton, and the thermal expansion of the multiphase system, the final form of this equation is:

,

=0,

gw gwg w icegw w ice gw

g gw g

rw rgw w w gw g g

swg icew g

w gw ice sw g ice

S S SnS n n nt t t t

k k Tp pt

S S S

D

k kg g

v

�

� �

�

(8)

where S is the saturation of the -phase/constituent ( =g,w,ice), k the intrinsic permeability of the porous material, ,swg ice the thermal expansion coefficient of the whole multiphase medium, the density of the -phase/constituent ( =gw,ga,w,g), n the porosity, Dgw the diffusion tensor of water vapour in the dry air and finally is the Biot coefficient. - Dry air mass balance equation It reads:

=0

ga gwgga ga s gw

g g gw g

rgga g g

g

SS n n S

t tk p

v D

k g

� �

�

(9)

in which the darcian flow of the gas phase, the diffusion of dry air in water vapour and the deformation of the solid matrix are considered (Dga is the diffusion tensor of the dry air).

107



- Energy balance equation , ,

, / / , / , ,

w w s g g sp w p g p effeff

vap w vap w fr w fr w fr vap fr vap

TC C C T Tt

m H m H m H

v v � (10)

where Hi/j are the enthalpy variations due to the phase changes between phase i and j and the related rate of mass exchange.

The mass and heat source terms related to desublimation of vapour, and ,

respectively, are very small in comparison to the sources due to water freezing , and will be omitted in the following developments. - Linear momentum balance equation: After introduction of effective stress in the form valid for a material with no ice present in its pores, may be written as follows:

/ 0s g w s ce sp p I g� +s (11)

while for a material containing frozen water (ice), with considering effective stress accounting for crystallization pressure pcryst, may be written as,

/ / / 0s g s g w s w ice s cryste s s sp p p I g� +s (11)

where is the surface fraction of the fluid i in contact with the solid matrix s. 4.Numerical examples In the following, two numerical applications of the model will be shown and their results discussed. The first application deals with a sensitivity analysis that we have performed in order to assess the impact of some relevant coefficients of the constitutive relationships and evolution equations on the response of the model. The second example is a validation of the model carried out by comparing some experimental and numerical results. 4.1 Sensitivity analysis The case selected for performing this analysis consists of a wall, made of standard concrete, exposed to a cooling process on the left side. The variation of temperature can be depicted in Fig. 3; the material is considered sealed, i.e. no mass exchange is allowed, while the heat exchange is of convective type with an exchange coefficient equal to 8 W/m2K. Several factors have been considered for assessing their influence in the behaviour of the numerical model, among them: the characteristic time for freezing and melting, the form of isotherms, the initial value of the relative humidity in the pores (and the related water saturation level), ect.

108



Figure 3: Cooling process adopted for the sensitivity analysis

A

B Figure 4: results of the sensitivity analysis for two different values of initial RH, 97% (dashed line)

and 98% (solid line): ice saturation (A) and total strain (B) evolution in the domain of temperature at 0.5 cm from the cooled surface

109



Herein it is not possible to describe all the possible combinations, so it was decided to only show the case related to the initial degree of saturation. A different of 1% in the initial relative humidity (97% and 98%) corresponds to a 10% in term of liquid water saturation (65% versus 75%) taken into account the isotherm selected for this case. Figure 4 shows the main results of the model at 0.5 cm from the cooled surface of the wall, in terms of ice saturation (4A) and total strain (4B). The different value of the initial saturation degree, leads to not only a different peak of ice content, but also to a different freezing point, even if the material is obviously the same (see Fig. 4A). In Fig. 4B it is possible to observe that the hysteresis between freezing and melting due to the different shape of the interfaces, results in a similar behaviour in term of strain. 4.2 Validation of the model The results obtained by Sun & Scherer [2] in laboratory by combining calorimetric measurements (DSC) with dilatometry (DMA), were used for the validation of the model. From the work [2] it was possible to deduce only part of the material properties needed for the numerical simulations. From this set of experimental results it has been possible to calculate the contributions of thermal expansion, pore pressure, and crystallization pressure of ice to the strain observed in a mortar during freezing/thawing cycles and to perform a comparison with similar numerical results obtained with the model described in the previous sections. In particular the case of a material with 0% of entrained air and subject to the temperature variation shown in Figure 5, was considered. Figure 5 shows also the evolution of ice content in time. Figure 6 shows a comparison between experimental and numerical results in term of strain evolution. The agreement is quite good even if only a part of the material properties were available.

Figure 5: temperature and ice saturation evolution for the case of a mortar with 0% of entrained air.

Conclusions In this work a new numerical model for the analysis of freezing of porous media, in particular partially saturated cementitious materials, is presented. Differently than what is present in literature, the evolutive model for freezing is based on a non-equilibrium approach taking into

110



account the thermodynamics of phase changes, and it is combined into a more general model for the analysis of multiphase porous materials. The good response of the model is encouraging for further developing the proposed approach, taking into account also some additional damage processes with a potentially very wide field of applications.

Figure 6: strain evolution for the mortar specimen; experimental vs numerical results

References [1] Sun, Z. and Scherer, G.W., Measurement and simulation of dendritic growth of ice in

cement paste, Cem Concr Res 40 (2010), 1393-1402 [2] Sun, Z. and Scherer, G.W., Effect of air voids on salt scaling and internal freezing, Cem

Concr Res 40 (2010), 260-270 [3] Coussy, O., Poromechanics, John Wiley & Sons, Chichester, (2004) [4] Coussy, O., Poromechanics of freezing materials, Journal of the Mechanics and Physics

of Solids 53 (2005), 1689-1718 [5] Coussy, O, Monteiro, P. J.M., Poroelastic model for concrete exposed to freezing

temperatures, Cem Concr Res 38 (2008), 40-48 [6] Setzer, M.J., Micro-Ice-Lens Formation in Porous Solid, Journal of Colloid and Interface

Science 243 (2001), 193-201 [7] Gawin, D., Pesavento, F., Schrefler, B.A., Hygro-thermo-chemo-mechanical modelling

of concrete at early ages and beyond. Part I. Hydration and hygro-thermal phenomena, Int J for Num Meth in Eng 67(3) (2006), 299-331

[8] Bronfenbrener, L., A non-instantaneous kinetic model for freezing in porous media, Chem Eng and Processing 47 (2008), 1631-1646.

[9] Gawin, D., Pesavento, F., Schrefler, B.A, Modeling freezing of partially saturated cementitious materials with considering the process kinetics and hysteresis - Part 1: Theoretical model, in preparation.

111



EXPERIMENTAL STUDIES ON FROST-INDUCED DETERIORATION OF CONCRETE IN SWEDISH HYDROELECTRIC STRUCTURES Martin Rosenqvist (1,2), Katja Fridh (1), Manouchehr Hassanzadeh (1,3) (1) Lund University, Lund, Sweden (2) Vattenfall AB, Älvkarleby, Sweden (3) Sweco Energuide AB, Stockholm, Sweden Abstract This article presents field observations and results from experimental studies on the moisture conditions and frost resistance of concrete in Swedish hydroelectric structures. The experimental work has been carried out by subjecting concrete specimens to exposure conditions similar to those at existing structures. It was shown that surface deterioration at the waterline is caused by interactions between leaching, frost action and abrasion. Moreover, macroscopic ice lens growth can occur in hardened concrete under certain conditions. Measurements and microstructural analyses were conducted on concrete samples from existing hydroelectric structures to verify the results from the laboratory-cast concrete. 1. Introduction Many hydraulic structures in cold regions suffer from concrete deterioration due to frost action in winter. Hydroelectric structures, harbours, canals and bridge foundations are all examples of hydraulic structures. Constant exposure to water leads to an increasing water content of concrete over time. Since the volume of water increases by about 9% upon freezing, concrete with a high degree of saturation may become susceptible to freezing. Hence, the frost resistance is an important characteristic of concrete to be used in hydraulic structures. There are generally two types of frost damage; scaling of the surface but minor losses in strength, and minor or no scaling but large losses in strength [1]. It is well known that the presence of de-icing agents can have a detrimental effect on the durability of concrete structures subjected to frost action. However, freezing of concrete in the presence of soft water can also cause severe damage to concrete structures and thus reduce their expected service life. Field observations of superficial and internal damage have raised questions about the long-term behaviour of concrete in hydraulic structures in contact with soft water in cold regions. Swedish hydroelectric power stations and dams are examples of such structures.

112



This article is based on field observations and experimental work conducted as part of a PhD study on the moisture conditions and frost resistance of concrete in Swedish hydroelectric structures. The experimental work has been carried out by subjecting concrete specimens to exposure conditions similar to those at existing structures, including exposure to multiple degradation mechanisms. To verify the results, measurements and microstructural analyses have been conducted on concrete samples extracted from existing hydroelectric structures. 1.1 Superficial damage at the waterline Many hydroelectric structures in contact with soft water in cold regions suffer from concrete deterioration at the waterline [2,3]. Progressive disintegration of the concrete surface results in exposure of coarse aggregate, see Figure 1. In a long-term perspective, also the structural integrity can be reduced if the reinforcing steel is exposed to the elements and starts to corrode. The greatest amount of damage is generally found at the waterline. Deterioration rates of about 1 mm per year have been reported in some cases. Superficial damage to concrete can be caused by chemical, physical and mechanical processes or a combination of them. However, surface deterioration at the waterline is commonly assumed to be caused by ice floes and driftwood. Abrasive wear of the concrete surface may occur if drifting objects push against the structures. This scenario is probably true regarding structures where the river current leads objects towards the structures, i.e. the upstream side of water intakes and spillways. On the downstream side, however, the water flows away from the structures and carries away the objects. In spite of this fact, superficial damage at the waterline can be observed on the downstream side. Such observations indicate that damage to the concrete surface is not exclusively caused by abrasion. River water can be more or less aggressive regarding the ability to dissolve components of the cement, i.e. leaching. The concrete surface is affected differently due to varying water chemistry of different fresh water bodies. Leaching results in increased porosity and reduced mechanical strength of concrete. However, leaching kinetics is generally slow but the impact can be substantial in a long-term perspective.

Figure 1: Coarse aggregate exposed at the waterline on the upstream face of a concrete dam. The reservoir water level was about 0.3 m lower than normal when the picture was taken.

Water

Dam

113



Knowledge about the deterioration processes is important in order to improve the efficiency of maintenance of existing hydraulic structures, but also to develop new materials for repairs. Even though the resistance of concrete to a certain degradation mechanism is good, the effects of other mechanisms may significantly reduce the resistance to the first mechanism. Since hydraulic structures often are subjected to a number of degradation mechanisms, synergy may occur. Synergy is defined as the interaction of two or more elements, which together produce an effect greater than the sum of their individual effects. In this article, results from experiments on the effects of interactions between leaching, frost action and abrasion on the surface deterioration of concrete will be shown and discussed. 1.2 Concrete spalling far below the water level Concrete spalling is another type of damage that has been observed in different types of water retaining concrete structures exposed to long periods of freezing weather. Several cases of severe damage have been reported far below the water level on the water side of thin concrete dams, such as buttress and arch dams [4]. As a result of concrete spalling, pieces of concrete can come loose and fall off the dams, see Figure 2. None of the damaged dams had been provided with heat insulating walls during construction to protect them from freezing. In one case, 50 areas of damage were found on the upstream face of a buttress dam about 35 years after its commissioning. The damaged areas varied between 0.5 and 10 m2 in size and were up to 200 mm in depth [5]. In another case, a hole appeared in the flat slab about 70 years after its commissioning, see Figure 3. The size of the hole was approximately 0.8 m in width and 1.2 m in height. The water escaping through the hole flushed away all loose pieces of concrete. Afterwards, only the reinforcing steel bars were left in the hole. During long periods of freezing weather, a freezing front can be assumed to move towards the water side of thin concrete structures, see Figure 2. Concurrent water transport towards the freezing zone may facilitate macroscopic ice lens growth and subsequent damage to the structure. The hypothesis is that poor quality concrete, inadequate compaction or aging make hardened concrete susceptible to macroscopic ice lens growth, i.e. ice segregation. In this article, results from experiments on the conditions to facilitate ice segregation in hardened concrete are shown and discussed.

Figure 2: Temperature distribution within a thin concrete dam in winter (A). Spalling of concrete has exposed reinforcement bars (B). The photo was taken during the repair works.

Water

AirConcrete dam

+–

Free

zing

front

Heat flow+Water flow Heat flow

A B

Upstream face

Loose piece of the concrete cover

Exposed and corrodingreinforcement bars

Tem

pera

ture

114



Figure 3: Concrete buttress dam commissioned in the 1940s (A). A hole appeared in the flat slab about 70 years after commissioning. Downstream face in (B) and upstream face in (C). 2. Materials and methods 2.1 Concrete technology development in Sweden During the early stages of the Swedish hydropower development, it was common to use mass concrete of very dry consistency with cement content less than 200 kg/m3 [6]. Wet concrete with water to cement-ratio (w/c-ratio) up to 0.9 was used in reinforced structures. Due to high porosity and permeability, deterioration of the lean concrete occurred within a few years. Since the beginning of the 1930s, concrete with cement content between 275 and 350 kg/m3 and w/c-ratio between 0.5 and 0.6 was used to improve the durability of hydroelectric structures. However, increasing the cement content increased the risk of cracking due to the evolution of heat during the hydration process. As a countermove, new cements were developed in Sweden. To be mentioned is the low heat Portland cement (Limhamn LH) developed in the early 1930s [7]. This cement became the most used cement during the major part of the Swedish hydropower development between 1945 and 1975. To improve the workability of concrete with the low heat cement, globular proteins from egg white (called albuminoids) were added to the fresh concrete mix. As a side benefit, normal dosage of albuminoids increased the air content of the concrete to about 3-5% [8]. Hence, the frost resistance of concrete was improved. Albuminoids were used for the first time in Sweden during the heightening of the multiple-arch dams at Suorva in 1937. Since the late 1940s, albuminoids were commonly used for the construction of hydroelectric power stations and dams in Sweden. During the 1950s and 1960s, the introduction of plasticisers and improved air entraining agents further improved the durability of concrete to be used in hydroelectric structures. 2.2 Studies on surface deterioration Three concrete mixes were used to assess the freeze/thaw resistance of concrete at the waterline of hydraulic structures. The three mixes represent concrete used in the construction of hydroelectric structures in the period prior to 1930 and the periods 1930-1950 and 1950-1975. The w/c-ratio was 0.70, 0.62 and 0.54, respectively, while the cement content was 285, 325 and 300 kg/m3, respectively. Only the latter concrete mix was air entrained. The mix

A CB

115



proportions were selected in accordance to guidelines published by the Swedish State Power Board between 1942 and 1972 [4]. Standard cubes of size 150x150x150 mm were cast. Specimens of size 150x150x70 mm were later sawn out from the cubes. The climate in northern situated countries varies greatly over the year where summers are warm and winters are cold. Air temperatures down to -30 °C occur frequently, and can last several days or weeks, during the winter months. Hence, most fresh water bodies become ice covered. Even in the case of open water-conditions due to turbulent water, a band of ice frozen solid to the concrete surface can be seen at the waterline [4]. This band of ice indicates freezing temperatures in the concrete at and below the waterline. In order to simulate the temperature conditions at the waterline, four concrete specimens were placed in a plastic box and the lower halves of the specimens were submerged in tap water. With the exception of the surfaces facing air and water, the remaining surfaces were insulated. A fan was used to keep the surface thermal resistance at minimum. In reality, the wind does the same thing. To make sure that the water did not freeze, a heating coil kept the water temperature at about +2 °C and a pump was used to keep the water circulating. During the 13-hour freeze/thaw cycle, the air temperature alternated between +10 and -36 °C. The test setup is shown in Figure 4 and the method is further described by Rosenqvist et al. [9].

Figure 4: The test setup used to assess the frost resistance of concrete at the waterline. Similar concrete mixes with w/c-ratio 0.62 and 0.54 were used in the study on the effects of interactions between leaching, frost action and abrasion on the surface deterioration of concrete. The size of the specimens was 150x150x50 mm. Several combinations of the three degradation mechanisms were tested. Leaching of calcium compounds from the concrete surface was facilitated when the specimens were submerged in deionised water at pH 4. The desired pH value was maintained by automatic addition of nitric acid (HNO3). The scaling resistance of the concrete was assessed in the presence of deionised water according to the Swedish test method SS 13 72 44 [10]. Abrasion to the concrete surface was caused by steel brushes of size 50x10 mm which were carefully hand-brushed ten times back and forward. The test method and experimental design are further described by Pham and Terzic [11]. 2.3 Studies on concrete spalling Eight concrete mixes were used to assess the risk of macroscopic ice lens growth in hardened concrete. In order to determine the w/c-ratio over which ice segregation may occur, the w/c-

116



ratio ranged from 1.4 to 0.5. The cement content ranged from 150 to 375 kg/m3. Specimens of size 150x150x70 mm were sawn out of standard cubes of size 150x150x150 mm. In some cubes, sheets of standard copy paper of size 110x110 mm were cast into the concrete to represent cavities or other imperfections of the material. Three types of specimens were produced; (1) specimens of mechanically sound concrete, (2) specimens with internal damage due to frost action and (3) specimens with sheets of paper cast into the concrete. Specimens with internal cracking due to frost action were produced to represent concrete at the water side of water retaining structures where the water content of concrete increases over time. Frost damage may occur if the water content exceeds the critical degree of water saturation and the structure is exposed to freezing weather [12]. The test setup was designed to correspond to conditions present at thin uninsulated water retaining concrete structures; one side of the structure subjected to freezing and one side in contact with unfrozen water. The specimens were placed in a box filled with water, letting the lower surface stay in contact with water. To ensure one dimensional heat flow, the specimens were insulated on the remaining surfaces, see Figure 5. The air temperature was about -17 °C and the water temperature +3 °C. The test method is further described by Rosenqvist [4].

Figure 5: The test setup used to subject concrete specimens to unidirectional freezing. 2.4 Studies on concrete from existing structures In order to study the alteration and moisture conditions of concrete in existing hydroelectric structures, concrete cores were taken at four vertically different locations on the upstream face of a buttress dam. The cores were taken perpendicular to the upstream. In addition to the waterline, the other locations were 1 m above the normal waterline and at 10.5 and 18.5 m water depth. Cores were also taken at the corresponding locations on the downstream face. Construction of the buttress dam took place between 1954 and 1958. The low heat Portland cement mentioned in section 2.1 was used as binder. The cement content of the concrete in the upstream face was 330-350 kg/m3 and the w/c-ratio was about 0.5. The aggregates are natural sand and gravel with maximum aggregate size of 80 mm. Firstly, the concrete cores were used to determine the degree of water saturation according to the distance to the upstream face. Secondly, chemical and mineralogical modifications of the cement paste were analysed with an electron microprobe (EPMA), an X-ray diffractometer (XRD), a thermogravimetric analyser (TGA) and a scanning electron microscope (SEM) combined with an energy dispersive X-ray spectrometer (EDS). Thirdly, the concrete cores were subjected freeze/thaw tests in the presence of deionised water according to the Swedish test method SS 13 72 44 [10].

0 °C

+

-

Water

Air

Insulation

117



3. Results 3.1 Surface deterioration at the waterline When the concrete mixes with w/c-ratio 0.70, 0.62 and 0.54 were subjected to freeze/thaw tests simulating the temperature conditions present at the waterline [9], only minor damage to the concrete surface was observed for the non-air-entrained mixes with w/c-ratio 0.70 and 0.62. Air-entrained concrete with w/c-ratio 0.54 did not suffer from any superficial damage at all. These results did not correspond to reality where progressive deterioration of the concrete surface occurs at the waterline of hydraulic structures in contact with soft water. When the concrete mixes with w/c-ratio 0.62 and 0.54 were subjected to interactions between leaching, frost action and abrasion, also air-entrained concrete with w/c-ratio 0.54 suffered from superficial damage. The final appearance of the concrete surface of specimens with w/c-ratio 0.54 is shown in Figure 6. The specimens had been subjected to four repetitions of one week of leaching in deionised water at pH 4, seven freeze/thaw cycles in the presence of deionised water and abrasive wear by steel brushes. The final appearance of concrete with w/c-ratio 0.62 was similar to concrete with w/c-ratio 0.54. However, the amount of damage to the concrete surface was somewhat greater.

Figure 6: Concrete surface after exposure to freeze/thaw cycles in the presence of deionised water (A). Surface after exposure to leaching, frost action and abrasion in (B) and (C). The amount of damage caused by interactions between the three degradation mechanisms leaching, frost action and abrasion exceeded the total amount of damage caused by the mechanisms separately. Further, leaching significantly amplified the effects of abrasion and particularly frost action for both concrete mixes. Based on the analyses of chemical and mineralogical modifications of the cement paste in concrete extracted from the buttress dam, decalcification of the cement paste occurs in two steps. In the first step, mainly calcium hydroxide, ettringite and calcium-rich C-S-H dissolve. Hence, the porosity of the concrete is increased. Further loss of calcium from the C-S-H occurs in the second step. Figure 7 gives the percentage values of CaO, SiO2 and total oxide content of cement paste in concrete at 10.5 m water depth according to the distance to the upstream face. The analysis was conducted by using an electronic microprobe. The method used to conduct the analysis and to process the data is described by Bertron et al. [13]. Freeze/thaw tests of the concrete surface revealed that concrete which is about to undergo, or has undergone, the second step of decalcification is highly susceptible to freezing in the presence of deionised water. The amount of loose materials collected from the specimens was greater for concrete from far below the water level compared to concrete at the waterline.

A CB

118



Figure 7: Percentage values of CaO, SiO2 and total oxide content of cement paste in concrete at 10.5 m water depth according to the distance to the upstream face of the buttress dam. 3.2 Concrete spalling far below the waterline Macroscopic ice lens growth occurred in specimens of sound concrete with w/c-ratio 0.9 and higher. The ice formation was uniform over the cross section and the ice lenses reached a thickness of up to 15 mm before growth ceased. The higher the w/c-ratio the shorter the time of freezing required to facilitate ice segregation. No signs of ice segregation were observed after 235 days in the specimens with w/c-ratio below 0.8 when the tests were terminated. The specimens with three paper layers cast into the concrete were subjected to freezing for 291 days. By then, all specimens had failed due ice segregation. The fracture surface coincided with the middle paper layer, see Figure 8. Also in this case, the time to ice segregation was generally increased with decreasing w/c-ratio. In the case of concrete specimens with internal cracking due to frost damage, the specimens failed due to ice segregation within only a few days of freezing. All specimens failed within three days except the specimen with w/c-ratio 0.5, which failed after seven days. Measurements of the water content of concrete taken out from the upstream face of the buttress dam indicated that it probably exceeds the critical degree of saturation in concrete from far below the water level. At the waterline, the concrete surface absorbs water whereas it dries out above the maximum water level. On the downstream side of the dam, the moisture content exceeds the hygroscopic range, thus indicating that water transport through the lower part of the dam is driven by hydrostatic pressure in combination with capillary suction.

Figure 8: Macroscopic ice lens growth in a specimen with paper sheets cast into the concrete to represent imperfections of the concrete. Two sides of the specimen are shown.

01020304050607080

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Oxi

des:

Tot

al, C

aO, S

iO2

[%]

Distance to the upstream face [ m]

Total CaO SiO2SiO2

Dissolution of Ca(OH)2

Macroscopic ice lens

119



4. Discussion Considering the exposure conditions at the waterline of Swedish hydroelectric power stations and dams, the results reported in this article show that leaching of calcium compounds increases the susceptibility of concrete to abrasion and particularly frost damage. Since the normal pH values in Swedish lakes and rivers vary in the range 6 to 8, leaching of calcium from the concrete surface should be a continuous process over time. This assumption was confirmed by the microstructural studies on chemical and mineralogical changes to the cement paste in concrete from the buttress dam commissioned in 1958. A possible scenario of surface deterioration at the waterline is that leaching of the concrete surface takes place during the summer. The surface is thus more susceptible to frost action. During the following winter, the surface layer is damaged by frost action and later removed due to ice abrasion. During the next year, the process starts all over again. The experience so far is that the deterioration rate is low. However, when structural reinforcement bars are exposed, the need for rehabilitation is obvious. Detailed knowledge about the deterioration processes is consequently important in order to achieve durable repairs. Results obtained from experimental studies on the risk of ice segregation in hardened concrete showed that macroscopic ice lens growth occurs under certain conditions. Macroscopic ice lens growth may occur in concrete with cavities or other imperfections of the material due to poor workmanship, such as inadequate compaction. Also the effects of various types of aging, such as internal frost damage and leaching, can make concrete susceptible to ice segregation. However, ice segregation is not likely to occur in sound concrete with w/c-ratio below 0.8. The overall risk of ice segregation in hardened concrete depends on the duration of a stationary freezing front. The lower the w/c-ratio the longer time the freezing front has to remain stationary to attract a sufficient amount of water in order to facilitate macroscopic ice lens growth. Nevertheless, the risk of concrete spalling in thin uninsulated water retaining concrete structures cannot be overlooked since unfavourable temperature and moisture conditions may exist in winter. Field observations confirm this assumption since there are numerous examples of concrete spalling far below the water level in water retaining structures subjected to unidirectional freezing in winter. Concrete spalling can ultimately lead to the appearance of holes and leaks in the structures and endanger their structural safety. 5. Conclusions This article presents field observations and results from experimental work conducted as part of a PhD study on the moisture conditions and frost resistance of concrete in Swedish hydroelectric power stations and dams. The results may also be useful for concrete in other types of hydraulic structures in contact with soft water in cold regions. It was shown that concrete mixes performing well in freeze/thaw tests sustain frost damage when subjected to exposure conditions similar to those at the waterline of existing structures. Interactions between leaching, frost action and abrasion cause progressive deterioration of the concrete surface at the waterline. Especially calcium leaching significantly reduces the frost resistance of the concrete surface. Since long-term exposure to soft water in rivers and lakes causes leaching, the frost resistance of the concrete surface decreases over time. Results from tests on laboratory-cast concrete correspond to results from concrete in existing structures.

120



Concrete spalling in thin uninsulated water retaining concrete structures can be caused by macroscopic ice lens growth. Ice segregation is facilitated by internal frost damage or various types of imperfections of concrete with w/c-ratio as low as 0.5. However, the lower the w/c-ratio the longer time the temperature distribution over the specimen has to remain stationary to facilitate macroscopic ice lens growth. Ice segregation is not likely to occur in sound concrete with w/c-ratio below 0.8. Acknowledgements The authors express their gratitude to Energiforsk AB (Swedish Energy Research Centre), SBUF (The Development Fund of the Swedish Construction Industry), SVC (Swedish Hydro Power Centre) and Vattenfall AB for funding this study. References [1] Powers, T.C., A working hypothesis for further studies of frost resistance of concrete, J

Am Concr Inst 16 (1945), 245-272 [2] Heggestad, R., Myran, R., Investigations on 132 Norwegian concrete dams, Proceedings

of the 9th International Congress on Large Dams, Turkey (1967), vol. 3, 491-517 [3] Lehtinen, P., On the deterioration of concrete observed in dams and hydraulic structures

in Finland, Proceedings of the 13th International Congress on Larges Dams, India (1979), vol. 2, 83-90

[4] Rosenqvist, M., Moisture conditions and frost resistance of concrete in hydraulic structures, Licentiate thesis, Lund University (2013)

[5] Eriksson, H., Investigation and rehabilitation of the Storfinnforsen dam, Proceedings of the 18th International Congress on Large Dams, South Africa (1994), vol. 1, 247-259

[6] Löfquist, B., Notes on Swedish experience of concrete performance in hydraulic structures, Proceedings of the 5th International Congress on Large Dams, France (1955), vol. 4, 177-184

[7] Forsén, L., Swedish cements for hydraulic structures, Proceedings of the 2nd International Congress on Large Dams, USA (1936), vol. 2, 181-199

[8] Lalin, G.S., Admixtures for the purpose of improving the workability of concrete, Proceedings of the 3rd International Congress on Large Dams, Sweden (1948), vol. 2, 1217-1221

[9] Rosenqvist, M., Oxfall, M., Fridh, K., Hassanzadeh, M., A test method to assess the frost resistance of concrete at the waterline of hydraulic structures, Mater Struct 48 (2015), 2403-2415

[10] SS 137244:2005, Concrete Testing – Hardened Concrete – Scaling at Freezing, Swedish Standards Institute, Sweden (2005)

[11] Pham, L.W., Terzic, A., Effects of leaching and abrasion on the scaling resistance of concrete, Report TVBM-5093, Lund University (2013)

[12] Fagerlund, G., The international cooperative test of the critical degree of saturation method of assessing the freeze/thaw resistance of concrete, Mater Struct 4 (1977), 231-253

[13] Bertron, A., Escadeillas, G., de Parseval, P., Duschesne J., Processing of electron microprobe data from the analysis of altered cementitious materials, Cem Concr 39 (2009) 929-935

121



THE INFLUENCE OF AIR VOID CHARACTERISTICS ON THE FEEZE-THAW-SALT RESISTANCE OF PAVEMENT CONCRETES Aljoša Šajna (1), Lado Bras (1) (1) Slovenian National Building and Civil Engineering Institute Abstract As part of renovation works being presently carried out in the centre of Ljubljana, Slovenia, some concrete pavements with additional architectural requirements were laid. According to the valid Slovenian technical specifications, wearing layers of concrete pavements have to be made from freeze-thaw-salt (FTS) resistant concrete. This resistance has to be proved according to the provisions of the Slovenian supplement to the standard EN 206, i.e. SIST 1026. Since testing according to SIST 1026 is time-consuming, complementary test methods, based on the determination of air void size characteristics, were used during the concrete mix optimization process. Beside an air entraining agent, a superplasticizer, a shrinkage reducing admixture and microsilica were used to fulfil all the requirements for concrete. In the paper the results of FTS performance tests carried out on selected pavement concretes are presented. These results are compared to the measured entrained air void characteristics, determined according to EN 12350-7 and EN 480-11, and also by means of an Air Void Analyser. Although all the admixtures were produced by the same manufacturer, in some cases their unexpected incompatibility resulted in an inadequate air void structure and hence inadequate FTS resistance of the concrete. 1 Introduction As part of renovation works being presently carried out in the centre of Ljubljana, Slovenia, for architectural reasons concrete was laid at some locations, instead of asphalt. The challenge which the contactor and his team had to solve was how to fulfil the additional architectural requirements, as well as the usual technical requirements: a bright, bush-hammered surface, with specified proportions of white and black aggregates of different sizes. Numerous trial tests were planned, in which different concrete compositions were used. Test slabs having an area of approximately 1 m2 were prepared and bush-hammered before the architects gave their

122



final approval to one of the trial concrete surfaces. After this approval had been obtained it was necessary to experimentally verify whether the Slovenian technical requirements for concrete pavements could be fulfilled. According to the Slovenian Technical Specification for Roads TSC 06.420 [1], wearing layers of concrete pavements have to exceed defined compressive and bending strengths, and also have to be freeze-thaw-salt (FTS) resistant. FTS resistance has to be proved by means of a test which is defined in the Slovenian supplement to the standard EN 206, i.e. SIST 1026 [2]. Since testing according to SIST 1026 is a time-consuming process, other indirect and complementary test methods were used during the mix design process. Since FTS resistance was achieved by the addition of an air entraining agent, all the applied indirect test methods used are based on the determination of the characteristics of the entrained air voids. Beside the air entraining agent, other admixtures were used to improve the workability and rheology of the fresh concrete, i.e. a superplasticizer and a shrinkage reducing admixture. In the paper the result of the FTS performance tests carried out on the investigated pavement concretes are presented. These results are compared to the characteristics of the entrained air voids, which were determined according to EN 12350-7 [3] and EN 480-11 [4], and also by means of an Air Void Analyser (AVA) [5]. Although all the admixtures were produced by the same manufacturer, in some cases their incompatibility resulted in an inadequate air void structure, and hence inadequate FTS resistance of the concrete. 2 Experimental program 2.1 Concrete mixtures In this investigation four different concrete mixes, with different compositions, were prepared and investigated. The basic concrete mix proportions are given in Table 1. Table 1: Basic proportions of the investigated concrete mixes [kg/m3] Raw material CM1 CM2 CM3 CM4 Cement CEM II A-S 42,5R 380 380 380 380 Water, added (w/c = 0.42) 176 159 159 159 Microsilica MS 11.4 11.4 11.4 11.4 Air entraining agent AEA 0.33 0.5 0.5 0.5 Superplasticizer SP1 3.3 3.7 3.7 3.7 Plasticizer P1 2.4 1.2 1.2 1.2 Shrinkage reducing admixture SRA 19 19.6 19.6 19.6 0/4 limestone aggregate, white 1117 870 0/4 limestone aggregate, grey 870 870 8/16 limestone aggregate, white 599 871 8/11 basalt aggregate, black 347 347 11/16 basalt aggregate, black 520 520

123



The concrete mixes vary in their aggregate composition and also in the dosage of the admixtures, what turned out to be important for the characteristics of the entrained air voids and thus also for the FTS resistance of the concrete. The mix CM1 was mixed at a concrete batching plant, and the samples were cast at the construction site, whereas the mixes CM2 to CM4 were mixed and cast in the laboratory. 2.2 Test methods For concrete characterisation, the following tests were performed: slump (EN 12350-2 [6]), density of fresh concrete (EN 12350-6 [7]), air content (EN 12350-7 [3]), 28-day compressive strength (EN 12390-3 [8]), resistance to wear (EN 13892-3 [9]), and density of hardened concrete (EN 12390-7 [10]). FTS resistance of the concrete was ensured by means of the selected adding air entraining admixture. It is well known that air-void parameters (in particular, the spacing factor) have a significant impact on the efficiency of the air entraining admixture, and thus the durability of concrete. The incompatibility of raw materials, in particular between the used cement and the mineral additions on the one hand, and the different chemical admixtures on the other, can result in an undesirable air-void structure, so that the FTS resistance is inadequate. The entrained air void structure of the fresh concretes was determined by means of AVA [5], and that of the hardened concretes according to EN 480-11 [4]. The FTS resistance of the hardened concrete mixes was evaluated according to SIST 1026, Annex 5 [2], which is similar to the slab test which is defined in CEN/TS 12390-9 [11]. These tests were performed on both saw-cut and bush-hammered surfaces of the same concrete. 3 Test results 3.1 Concrete characterization The concrete characterization test results are presented in Table 2. The slump and air content tests were performed on fresh concrete samples, whereas the compressive strength and density tests were performed on hardened concrete samples. The highest compressive strength was achieved by the CM2 mix, and the lowest by the CM4 mix, but all the obtained results exceeded the required values for compressive strength class C30/37 [1, 2]. For all the mixtures, the achieved air content (according to EN 12350-7 [3]) was slightly lower than the designed value, but higher than the value set by SIST 1026 [2], i.e. 4 %. Table 2: Characterisation test results

Mix designation

Slump

[mm]

Air content

[%]

Compressive strength [MPa]

Density

[kg/m3] CM1-1 200 5.7 66.6 2370 CM1-2 180 5.5 59.1 2350 CM2 140 6.0 67.0 2320 CM3 130 6.0 65.3 2400 CM4 180 6.9 55.3 2360

124



3.2 FTS resistance 3.2.1 FTS performance of the concrete mixes The FTS resistance of the hardened concrete mixes was evaluated according to SIST 1026 [2]. The tests were performed on slabs (half of 150 mm cubes). The moulded surfaces were exposed to freezing-thawing. The test results are presented in Table 3. Table 3: FTS resistance test results obtained on cubes

Mix designation Average scaling [kg/m2]

Max scaling [kg/m2]

CM1-1 0.42 0.48 CM1-2 0.58 1.10 CM2 0.00 0.00 CM3 0.00 0.00 CM4 0.02 0.02

According to SIST 1026 [2], for an exposure class of XF4 the maximum permitted scaling of the tested concrete is 0.20 kg/m2 (the average of 3 samples), whereas individual results may not exceed a value of 0.25 kg/m2. All of the concrete mixes except CM1 can be ranked as suitable for exposure class XF4. 3.2.2 FTS resistance of the surface of the concrete pavement surface As bush-hammering can have a negative effect on the FTS resistance of the surface of concrete pavements, additional tests were performed on cores which were taken from the trial concrete slabs. The tests were performed on cores having a diameter of 150 mm, according SIST 1026, Annex 5 [2]. The bush-hammered surfaces were exposed to freezing-thawing. The test results are summarized in Table 4. As the CM1 mix was not resistant to FTS, no bush-hammered test samples were prepared for this mix. Table 4: Results of the FTS resistance tests performed on cores

Mix designation Average scaling [kg/m2]

Max scaling [kg/m2]

CM1 n.a. n.a. CM2 0.06 0.06 CM3 0.10 0.10 CM4 0.08 0.08

3.3 Analysis of the air-void parameters The air void parameters were determined using two methods: firstly, by means of the German Instruments Air Void Analyzer AVA-3000 (AVA GI [5]) for the fresh concretes, and secondly according to the standard EN 480-11 [4] for the hardened concretes. 3.3.1 AVA test results The air void structure characteristics determined by means of the AVA GI test method are summarized in Table 5, and in Figures 1 to 4.

125



Table 5: AVA test results

Mix designation

Total air voids in the paste

[%]

Spacing factor

[mm]

Total air voids in the concrete

[%]

Air voids in the concrete, d 300 m

[%] CM1-1 2.0 0.510 0.6 0.3 CM1-2 2.5 0.529 0.8 0.3 CM2-1 7.7 0.300 2.5 1.1 CM2-2 14.4 0.250 4.6 1.8 CM3 6.6 0.258 2.2 1.2

Figure 1: Average cumulative air void distribution in the investigated concrete mixes CM1, CM2 and CM3

Figure 2: Distribution of the air void content in the the concrete mix CM1

126



Figure 3: Distribution of the air void content in the the concrete mix CM2

Figure 4: Distribution of the air void content in the the concrete mix CM3 From the AVA test results it can be seen that there are significant differences in the sizes of the air voids. The total air void content is the lowest in the case of the concrete mix CM1, and highest in the case of the concrete mix lower in the case of the concrete mix CM1 than in the mixes CM2 and CM3. The air void distribution of the concrete mix CM2 is similar to that of CM1, but the total amount of air voids is lower. The concrete mixes CM2 and CM3 contain a higher proportion of small voids

, and have a lower spacing factor than the concrete mix CM1. Based on the assumption that concrete is FTS resistant if the spacing factor has a value which

mixes CM2 and CM3 can be considered to be FTS resistant, whereas CM1 isn’t.

127



3.3.2 Results obtained according to the EN 480-11 test method The air void structure characteristics as determined according to the EN 480-11 test method are summarized in Table 6, and in Figure 5. Clear differences can be seen in the air void percentage, and in the air void distribution. The concrete mix CM1 has fewer air voids and, which is more important for FTS resistance, it has fewer small air voids ( 0.300 mm) than the concrete mixes CM2, CM3, and CM4. Table 6: Results obtained according to the EN 480-11 test method

Mix designation

Total air voids in the paste

[%]

Spacing factor

[mm]

Total air voids in the concrete

[%]

Air voids in the concrete,

m [%]

CM1-1 6.71 0.378 4.94 0.79 CM1-2 6.85 0.445 4.85 0.38 CM2-1 6.56 0,167 4.61 2.02 CM3 5.32 0.140 5.67 2.53 CM4 4.59 0.133 6.57 3.25

Figure 5: Distribution of the measured air void content in the cement pastes (for

128



4 Discussion and conclusions In this paper the results of an optimization process of concrete mixes for pavements with additional architectural requirements are presented. The emphasis is on the freeze-thaw-salt (FTS) resistance of these concrete, which was achieved by using an air entraining agent. As was shown by the results of tests which were performed according to the Slovenian standard SIST 1026, the concrete mix proposed by the contractor did not satisfy the required FTS performance, but the optimised mixes with optimised admixture dosages did. Additionally, when the admixture dosage was optimized adequate FTS resistance of the bush-hammered concrete pavement surface, too, was achieved. In addition to the FTS performance tests required by the corresponding Slovenian standard, the air void parameters in the fresh and hardened concrete mixes were determined using a German Instruments Air Void Analyzer AVA-3000 (AVA GI), and the EN 480-11 test method, respectively. Based on the results of the investigations described in this paper, the following conclusions can be drawn: Adequate FTS resistance can be achieved by proper selection of the necessary chemical

admixtures. Using more admixtures does not always result in better performance of the concrete. The compatibility of admixtures has to be checked by various tests.

When the AVA and the EN 180-11 test methods were used, correct predictions of the FTS resistance of the concrete mixes was achieved, but this was not the case when the standard EN 12350-7 was used.

As expected, when the standard EN 12350-7 was used, the highest air voids percentage in the concrete was measured, since this method is based on the total air content in the fresh concrete (Figure 6). The total voids percentage defined when using the EN 480-11 test method gave comparable results, although slightly lower amounts of air voids in the concrete were measured. A far lower air voids percentage was measured when using the AVA test method, since in this case only air voids smaller than 2.0 mm are taken into account.

The air content in the cement paste, which is very important for the FTS resistance of concrete, should only be measured by the AVA and EN 480-11 test methods, assuming that the concrete composition is known. These test methods can result in either comparable (e.g. concrete mix CM3) results or incomparable results (e.g. concrete mixes CM1-1a and CM2-2) (Figure 7). The comparability of results depends on the air void sizes which are taken into account, and on the air voids distribution.

The AVA test method consistently results in a larger spacing factor than that obtained in the EN 480-11 test method (Figure 8). But since the recommended values for FTS resistant concrete are different, too, differ (0.300 mm for AVA, and 0.200 mm for the EN 480-11 test method according to SIST 1026), both of the test methods provide consistent and comparable conclusions with regard to the FTS resistance of concrete.

129



The conclusions regarding the FTS resistance of concrete based on the use of a spacing factor according to the AVA and EN 480-11 test methods are comparable to the conclusions which are based on the performance of the test according to SIST 1026.

Optimization of the dosage of admixtures leads to a higher air voids volume and, in particular, to an improved size distribution of the air voids, resulting in lower spacing factors and in FTS resistant concretes for wearing layers, even when the surface is "damaged" by bush-hammering.

Figure 6: Air voids content in the different concrete mixes obtained by using different test methods

Figure 7: Content of air voids in the investigated cement pastes determined by using different test methods

130



Figure 8: Spacing factors determined using different test methods

References

[1] TSC 06.420:2005, Technical Specifications for Roads, Wearing layers, Concrete [2] SIST 1026:2005, Rules for the implementation of SIST EN 206 [3] EN 12350-7:2009, Testing fresh concrete - Part 7: Air content - Pressure methods [4] EN 480-11:2005, Admixtures for concrete, mortar and grout - Test methods - Part 11:

Determination of air void characteristics in hardened concrete [5] Instructions and Maintenance Manual for Air Void Analyzer AVA-3000, German

Instruments, 2006 [6] EN 12350-2:2009, Testing fresh concrete - Part 2: Slump-test [7] EN 12350-6:2009, Testing fresh concrete - Part 6: Density [8] EN 12390-3: 2009, Testing hardened concrete - Part 3: Compressive strength of test

specimens [9] EN 13892-3:2014, Methods of test for screed materials - Part 3: Determination of wear

resistance - Böhme [10] EN 12390-7:2009, Testing hardened concrete - Part 7: Density of hardened concrete [11] CEN/TS 12390-9:2006, Testing hardened concrete - Part 9: Freeze-thaw resistance -

Scaling

131



IDENTIFICATION OF OPTIMAL CONDITION FOR THE DE-ICING SALT SCALING RESISTANCE OF CONCRETE Samindi Samarakoon(1), Samdar Kakay(1), Pål Lieske Tefre(1), Mats Buøen(1), Vikrant Kaushal(1) (1) University of Stavanger, Stavanger, Norway

Abstract Scaling of concrete surfaces becomes a major durability issue on concrete roads, sidewalk pavements and bridge decks, especially in the case of a salt environment (i.e. because of de-icing salt) generated in cold climates. There are a number of factors affecting salt scaling resistance in concrete; this study has identified air entrainment, water/cement ratio, curing method and silica content as the significant parameters that affect concrete scaling resistance. An engineering robust design approach has been deployed to investigate the optimal condition that results in optimal salt scaling resistance. Parameter design has been performed using orthogonal arrays and lab experimentations to determine the best parameter combination to provide optimal scaling resistance. A comparison is made, using results of the lab experimentation and the predicted amount of scaled material at optimal conditions. 1. Introduction

In cold climates, salts such as NaCl and CaCl2 are often used to de-ice concrete roads, sidewalk pavements and bridge decks. Salt scaling is one of the major durability concerns facing concrete structures in such climates, resulting in high maintenance cost during the service life of concrete structures. Moreover, salt scaling can be defined as superficial damage caused by the freezing and thawing of a saline solution on the surface of a concrete body [1]. In this paper, resistance to the removal/scaling of material due to the aforementioned phenomenon can be defined as de-icing salt scaling resistance. When the level of de-icing salt scaling resistance is high, the level of removal/scaling of material is low. Similarly, there is no reliable remedial measure to protect concrete structures from such deterioration [2]. Many laboratory experiments and field investigations have been carried out to understand the mechanism of salt scaling and the factors affecting salt scaling resistance [2, 3]. For example, factors such as air entrainment dosage, water/cement ratio, curing method and silica content can be considered as significant among them. To control the deterioration of concrete surfaces due to de-icing salts, it is vital to determine the influence of potential factors on the de-icing scaling resistance of concrete. The

132



engineering ‘robust design’ approach, or Taguchi method, provides a systematic way to determine such an influence, leading to an enhancement of the de-icing scaling resistance of concrete. This approach is more beneficial than the conventional experimental design methods that demand higher experimental cost [4, 5]. In addition, this approach minimizes the variability around the investigated parameters/factors when bringing the performance value to the target value [6]. This manuscript aims to use the Taguchi method to determine optimal levels of significant factors, which give the best de-icing scaling resistance. It uses the problem type “smaller the better”, taking into consideration the fact that when the amount of scaled material is lower, de-icing salt scaling resistance is higher. It also predicts the scaled material at the optimal condition to compare with experimental findings. 2. Methodology for experiments

In this experimental study, Taguchi’s approach is used to determine the influence of controlling factors, which affect the de-icing salt scaling resistance of concrete. This method is widely used in parameter design; it provides the design engineer with a systematic and efficient method for determining near optimal design parameters [4, 5]. It involves the following steps:

Definition of the problem: In this case, the problem is the determination of the optimal levels of the controlling factors of the concrete mixture, which give the lowest scaling of material due to de-icing salts

Determination of the performance characteristic(s) and the measuring system: The de-icing salt scaling resistance is evaluated by measuring the scaled material (g/m2) from the concrete surfaces after exposure to a number of freezing/thawing cycles. It is expected that, when the scaled material is smaller, the scaling resistance to de-icing salt is better.

Determination of the factors (parameters) affecting the performance characteristic(s): There are various factors affecting the de-icing salt scaling resistance of concrete surfaces, such as air entrainment, cement content, silica content, curing condition, surface treatment, water/cement ratio, etc. [2, 3]. Considering the available literature [1, 2, and 3] and expert knowledge, air entrainment, water/cement ratio, curing method and silica content can be considered as factors that significantly affect performance.

Determination of the number of levels and values of the controllable variables (parameters): The approach uses standard tables known as orthogonal arrays (OA) for the design of the experiments [5]. Only nine experiments are needed to study the entire list of parameters using the L9 orthogonal array. This is shown in Figure 1, which also illustrates the steps involved in the overall parameter design process.

133



Factors influencing the de-icing salt scaling resistance

Significant influencing

factors

Levels of major influencing factors

Run the experiment at

laboratory scale

Calculate factor effects and find out

the best settings of

control factors

Use the additive formula to estimate

theoretical value of de-icing salt

scaling

Is the improvement satisfactory?

Run verification experiment to investigate the

accuracy of the optimum settings

Yes

No

Revise the factor levels

Number of experiments

Factor levels f1 f2 f3 f4

1 2 3 4 5 6 7 8 9

1 1 1 2 2 2 3 3 3

1 2 3 1 2 3 1 2 3

1 2 3 2 3 1 1 2 3

1 2 3 3 1 2 2 3 1

Factors Level 1 Level 2 Level 3 f1 f2 f3 f4

l11 l21 l31 l41

l12 l22 l32 l42

l13 l23 l33 l44

Figure 1: Framework of Taguchi method [4]

Determination of loss function and the performance statistics: the “smaller-is-better”

(SMB) loss function is selected because the smaller the scaled material, the better the scaling resistance to de-icing salt is. The signal to noise ratios (S/N ratios) can be calculated as given in Eq. (1) [5]. / ( ) = 10 (1)

where, n = number of replications yi = performance indicator value (i = 1,2…n). In this case, the mass of scaled material per surface area of concrete samples after freeze/thaw cycles is measured in g/m2.

Conducting of experiments and recording of results. Analysis of data and selection of the optimal value of the controllable variables. Testing of results. Evaluation, implementation and observation: Under the optimal set points, the

corresponding opt is calculated using the additive model for factor effects using Eq. (2) [5].

)( ratio S/N ijklDl

Ck

Bj

Aiopt (2)

where, = overall mean S/N ratio for the amount of scaled material over all the possible

combinations

i,j,k,l = particular levels of each of the factors which were selected (so in this model i,j,k,l must all take on one of the values 1, 2 or 3)

134



= deviation from caused by setting factor “A” at level i (similarly, other terms can be defined)

= error term.

The calculated “ opt” represents the theoretical level of scaled material at the optimal parameter combination investigated by the experimentation and subsequent calculations. Eq. (3) is used to convert the theoretically calculated S/N ratio (i.e. in dB) of the scaled material at the optimal parameter combination (i.e. the anticipated scaled material at the optimal parameter combination) to the corresponding real-world value (g/m2), where, “y” is the anticipated scaled material at optimal parameter combination in g/m2. It is also possible to perform a verification experiment to investigate the accuracy of the optimal parameter combination.

= 10 (3)

3. Material and experiment

3.1 Material The concrete mixtures were designed using Portland cement CEM I, which has 52 MPa of 28 days’ strength, silica fume, sand, gravel and air entrainment admixture. Tab. 1 shows the chemical composition of cement, and Tab. 2 shows the properties of other materials used. Table 1: Chemical composition

SiO2 (%) Al2O3 (%) CaO (%) MgO (%) SO3 (%) Fe2O3 (%)

Standard Cement (CEM1) 20 5.2 61.5 2.3 3.4 3

Table 2: Properties of materials used in the experiment

Density (kg/m3) Water absorption (%) Sand (0-8 mm) 2670 0.5 Gravel (8-16 mm) 2670 0.5 Cement (CEM 1) 3150 Not applicable Silica fume 2200 Admixture (air entrainment) 1020 Not applicable

135



3.2 Mixture design and preparation of concrete samples

In this study, nine concrete mixtures were prepared by using L9 orthogonal arrays, according to the methodology described in Section 2. To prepare the nine concrete mixtures, the particle matrix model [7] was used. Moreover, the matrix volume was selected as 350 L/m3 ( particle size is less than 0.125 mm) and the particle volume was selected as 650 L/m3( particle size large than 0.125 mm). In addition, instead of water/cement ratio, equivalent water/cement((W/C)eq) ratio was used in this study due to usage of silica that is a supplymentry cementing material. Taking into consideration the literature on the scaling of concrete due to de-icing salts [1, 2, 3], three different (W/C)eq ratios (0.55, 0.5 and 0.60), silica contents (0%, 5% and 7% by weight of cement), air entrainment admixture dosages (0%, 0.1% and 0.05% by weight of cement) and three different curing methods (open to air, water and covering concrete cubes with plastic sheet) were used, as shown in Tab. 3. The levels of each factor were found from the literature [2, 3] and expert knowledge. Details of each experiment are provided in Tab. 4. Tab. 5 provides the composition of each mixture used in the nine experiments. The air content was not measured in all the mixtures. In concrete mixtures 1 to 3 the air content was not measured. Mixtures 4 to 6 consisted of average 9 % air content, and mixtures 7 to 9 consisted of 7 % air content and slump 14-140 mm. Table 3: Factors and their levels used in this experiment

Table 4: Details of experiments

Factor Level 1 Level 2 Level 3 Air entrainment admixture dosage (A) 0% 0.1% 0.05% Silica content (B) 0% 5% 7% W/Ceq(C) 0.55 0.60 0.50 Curing method (D) Water Plastic sheet Open to air

Experiment number A B C D

1 2 3 4 5 6 7 8 9

0% 0% 0%

0.1% 0.1% 0.1% 0.05% 0.05% 0.05%

0% 5% 7% 0% 5% 7% 0% 5% 7%

0.55 0.60 0.50 0.60 0.50 0.55 0.50 0.55 0.60

Water Plastic sheet Open to air Open to air

Water Plastic sheet Plastic sheet Open to air

Water

136



Table 5: Composition of nine mixtures used in nine experiments

Mixture no

Composition of concrete mixtures (kg/m3)

Sand Gravel Silica fume Water

Air entrainment Cement

1 967 760 0 211 0 383 2 967 760 17 220 0 333 3 967 760 25 206 0 362 4 967 760 0 218 0.4 363 5 967 760 19 206 0.4 374 6 967 760 24 214 0.3 341 7 967 760 0 203 0.2 407 8 967 760 18 213 0.2 352 9 967 760 23 220 0.2 322

3.3 Test method The CDF (capillary suction of dei-cing solution) test method was used in this experiment, in accordance with NS-CEN/TS 12390-9:2006 [8]. Three concrete cubes of 100 mm*100 mm*100 mm were cast at each experiment, considering the acceptable height of a specimen varies between 50 mm and 150 mm. A climate chamber with a temperature of 20 ± 2°C and a relative humidity of 65 ± 5% was used. The concrete cubes were protected by lateral sealing, using epoxy resin. The test liquid consisted of de-icing agent solution, e.g. 97% by weight of demineralized or distilled water and 3% by weight of NaCl. The capillary suction period is seven days at a temperature of 20 ± 2°C. Figure 2 shows the freeze-thaw test for a concrete cube. After capillary suction, test specimens were placed in an incubator under temperature variation shown in Figure 3.

Figure 2: Freeze-thaw test for a concrete cube

137



Figure 3: Temperature variation in the climate chamber [8] 4. Results According to NS CEN/TS 12390-9:2006 [8], after 28 freeze/thaw cycles, scaled material from concrete specimens was measured, and the results are given in Tab. 6. In every experiment, three specimens of 100 mm*100 mm*100 mm were cast and the average amount of scaled material after 28 freeze/thaw cycles was measured. Very light scaling was observed with the air entrainment admixture. Using Eq. (1), the S/N ratio was calculated; this is also given in Tab. 6. The primary objective in conducting the experiment was to determine the optimal level of each factor that significantly affects the de-icing salt scaling resistance of concrete. The optimization was performed, taking into consideration the fact that the amount of scaled material is low, the level of de-icing scaling resistance is high and the plot of factor effect is given in Figure 4. In this Figure, A1 means that the dosage of air entrainment admixture is 0% (level 1) as given in Tab. 3. The optimal value of each factor is the level that has the highest value of “ the lowest amount of scaled material from the concrete surface. Thus, Using Figure 4, the best dosage of air entrainment admixture is A2 (0.1%), the best silica content is B1 (0%), the best (W/C)eq ratio is C3 (0.5) and the best curing method is D1 (water curing). The results obtained in this experiment show that 0% of silica content is the optimal amount to achieve the highest frost resistance. In this experimentation approach, the influence of combined effect of all the parameters have been taken into consideration. However, previous research demonstrates results with high variability in silica content vs frost resistance, as they may have carried out the optimization of one parameter (i.e. silica) in an isolated fashion without considering the combined effect.

138



Table 6: Average scaled material after 28 cycles Experiment no

/mixture no Average scaled material after 28

cycles (g/m2) S/N ratio(dB) i)

1 41.23 -29 2 127.9 -40 3 49 -33 4 33.3 -30 5 17 -23 6 24 -26 7 18 -25 8 24 -28 9 57 -33

Figure 4: Plot of average value of factor effect Using Eq. (2), the amount of scaled material under the optimal condition can be predicted.

opt can be estimated at the optimal condition as -21dB and the corresponding predicted scaled material is estimated as 11 g/m2 using Eq. (3); this is smaller than the amount of scaled material found from nine experiments. 5. Conclusion In this experiment, a parameter design approach has been used to investigate the optimal parameter combination that provides the highest de-icing scaling resistance of concrete. As the scaling resistance corresponds to minimum scale material, the analysis has been treated as the “smaller the better” type of problem. A L9 orthogonal array, together with four

-40

-35

-30

-25

-20

-15

-10A1 A2 A3 B1 B2 B3 C1 C2 C3 D1 D2 D3

Aver

age

for e

ach

leve

l (dB

)

Factors with levels A-air entrainment, B-silica content, C-(W/C)eq ,

D-curing method

139



parameters (i.e. taking into account that air entrainment dosage, silica content, W/Ceq ratio and curing method significantly influence the de-icing scaling resistance of concrete), is used. From the experiment, it has been found that the optimal condition for the best de-icing scaling resistance is as follows: the best dosage of air entrainment admixture is A2 (0.1%), the best silica content is B1 (0%), the best (W/C)eq ratio is C3 (0.5) and the best curing method is D1 (water curing). Scaling resistance at the optimal parameter combination is theoretically calculated. It has been revealed that scaling resistance at the optimal parameter combination is 11 g/m2 (i.e. the lowest level of material removal). Compared with the scaled material from the nine experiments, the theoretically calculated scaling resistance at optimal condition gives the lowest amount of scaled material. Future research needs to investigate the possibility of fine-tuning the optimal level to enhance the results. References [1]. Valenza, J. J. and Scherer, G. W., A review of salt scaling: II. Mechanisms, Cem Concr

Res 37 (2007), 1022–1034 [2]. Sahin, R., et al., Determination of the optimum conditions for de-icing salt scaling

resistance of concrete by visual examination and surface scaling, Construct Build Mater 24 (2010), 353–360

[3]. Valenza, J. J. and Scherer, G. W., A review of salt scaling: I. Phenomenology, Cem Concr Res 37 (2007), 1007–1021

[4]. Samarakoon, S. M. S. M. K. and Ratnayake, R. M. C., Use of engineering robust design approach to improve the surface quality of pre-cast concrete elements: An experimental approach, IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), Bangkok, Thailand, 10-13 December (2013)

[5]. Phadke, M. S., Quality engineering using robust design, Prentice Hall, Englewood Cliffs, New Jersey (1989)

[6]. Turkmena et al., Taguchi approach for investigation of some physical properties of concrete produced from mineral admixtures, Build Environ 43 (2008), 1127–1137

[7]. Jacobsen, S., Concrete technology - 1, Compendium, Norwegian University of Science and Technology, Trondheim, Norway (2009)

[8]. Norwegian standard: NS-CEN/TS 12390-9:2006, Testing hardened concrete: Part 9: Freeze-thaw resistance scaling, Norway (2009)

141



TOWARDS AN ADEQUATE DEICING SALT SCALING RESISTANCE OF HIGH-VOLUME FLY ASH (HVFA) CONCRETE AND CONCRETE WITH SUPERABSORBENT POLYMERS (SAPS) Didier Snoeck (1), Philip Van den Heede (1,2), Nele De Belie (1) (1) Magnel laboratory for Concrete Research, Ghent University, Ghent, BELGIUM (2) Strategic Initiative Materials (SIM vzw), project ISHECO within the program ‘SHE’,

Ghent, BELGIUM Abstract The deicing salt scaling resistance has been investigated for two types of concrete, i.e., air entrained high-volume fly ash (HVFA) concrete with a 50% cement replacement and non-air entrained concrete containing superabsorbent polymers (SAPs). A full characterization of their air void systems from the moment of casting until the freeze/thaw test was also done. Due to the presence of the highly AEA adsorptive fly ash an increased AEA dosage (7.0 ml/kg binder) was needed to achieve an adequate air void system in terms of air content and spacing factor to keep salt scaling within acceptable limits. For the novel non-air entrained concrete type with SAPs, which are able to absorb up to 500 times their weight in fluids, the salt scaling resistance is surprisingly high. The microstructural analysis revealed the formation of macro-pores due to these SAPs, creating an air void system as can be found in air-entrained concrete. Another advantage is that the strength of concrete with SAPs is much higher than for a conventional air-entrained concrete. This substantiates the further use of these SAPs as admixture in precast concrete road elements. 1. Introduction Exposure to freeze/thaw cycles in combination with deicing salts can cause scaling or flaking of the concrete surface. Although most researchers agree upon the general definition of this damage form, there is still a lot of discussion about the underlying deterioration mechanism. Boel gives a good summarizing overview of the existing theories to explain the phenomenon [1]. In accordance with the theory of Powers salt scaling damage would be induced by occurring hydraulic and osmotic pressures in combination with gradients in salt concentration [2]. On the other hand, the theory of Hansen states that the damage is due to an oversaturated salt solution in the larger pores [3]. Furthermore, there is the theory of Snyder which is based on the variation in freezing point for the water in the concrete due to a salt concentration gradient [4]. More recently, the glue-spalling theory [5, 6, 7] has been introduced. There the

142



damage phenomenon is explained by the fact that the ice layer tends to contract much more than the concrete layer. The cracks in the ice layer penetrate into the underlying concrete and propagate according to a path parallel to the concrete surface [8]. Also note that there are some differences between concrete standards regarding the required minimum air content in the fresh state to achieve an acceptable salt scaling resistance. According to the European standard NBN EN 206-1, concrete exposed to freeze/thaw in combination with deicing salts should be air entrained with a minimum air content of 4% in the fresh state. However, according to the Belgian standard NBN B 15-001 artificial air entrainment is not mandatory and it should relate to the maximum aggregate size. For a maximum nominal aggregate size of 16 mm, the standard specifies an air content of at least 5%. The American standard ACI 201.2R also takes into account the higher air requirements of concrete mixtures with higher paste contents due to smaller nominal maximum aggregate sizes. However, also the severity of the exposure is of importance. A distinction is made between moderate and severe exposures. Exposure conditions are considered to be severe whenever deicing salts are present. In such an environment, an air content of 6 to 7% is recommended for a maximum nominal aggregate size of 16 mm; 5 to 5.5% of air is only sufficient in the case of moderate exposure (without deicing salts) for the same aggregate size. Nevertheless, the recommendations of ACI 201.2R are not binding. Local conditions and experience with specific mixtures and procedures could still warrant other values. Both the ambiguity regarding the precise deterioration mechanism and the existence of different criteria in terms of air content make it not evident to properly design novel concrete compositions, such as High-Volume Fly Ash (HVFA) concrete and concrete containing superabsorbent polymers (SAPs), and make them resistant to salt scaling. With respect to HVFA concrete, a potentially sustainable concrete type in which at least 50% of the ordinary Portland cement is replaced with pozzolanic fly ash, Malhotra and Mehta reported an adequate salt scaling resistance of HVFA concrete pavements in Wisconsin (US) and sidewalk sections in Halifax (Canada) [9]. On the other hand, they also reported higher deterioration rates for the same material under laboratory conditions. Since the applicable European standards mainly focus on performance criteria for concrete that has been exposed to accelerated laboratory testing conditions, it is quite logic that many concrete manufacturers remain for the moment sceptical about using this HVFA concrete on a larger scale. Therefore, more research on that matter is still needed. SAPs are a new kind of material to be used in building applications. They have the ability to absorb a significant amount of liquid from the environment (up to 500 times their own weight) and to retain it without dissolving. In concrete they are used to decrease the autogenous shrinkage and self-desiccation due to internal curing [10] and to promote self-sealing and self-healing [11, 12]. Due to their swelling capacity SAPs can also be used to change the pore structure. They can extract water of the fresh concrete mixture causing the stiffening of the paste which is accompanied by a reduction of the capillary porosity. Cavities will hereby remain as empty pores afterwards. SAP particles introduce a system of fine, evenly distributed air voids after release of their absorbed water. SAPs can thus also function as air-pore entraining agents, increasing the freeze-thaw resistance and the durability [13, 14]. In this research, a HVFA concrete was developed by carefully controlling the air content in fresh and hardened state with appropriate dosages of air entraining agent. Moreover, the effects of SAPs on concrete’s air content and air void system were examined. Also, the influences on the strength and freeze-thaw resistance were evaluated and compared.

143



2. Materials and Methods 2.1 Concrete mixtures Regarding the research on HVFA concrete, six concrete mixtures were manufactured (Table 1). Mix T(0.45) is a non-air entrained ordinary Portland cement (OPC) concrete with the required minimum cement content (340 kg/m³) and the maximum water-to-cement ratio (W/C: 0.45)) for freeze/thaw environments according to NBN B15-001. An air entrained version of the reference concrete mix (T(0.45)A) was manufactured as well. The incorporation of a fatty acid/polyglycol based air entraining agent (dry matter mass percentage: 4%) is indicated within the concrete mix name by means of the letter ‘A’. Regarding the inert fraction per m³ of mixtures T(0.45) and T(0.45A), the concrete contained 715 kg sand 0/4, 515 kg gravel 2/8 and 671 kg gravel 8/16. Table 1. Studied HVFA and corresponding reference concrete mixtures with nomenclature. T(0.45) T(0.45)A F(1)50 F(1)50A F(2)50 F(2)50A W/C or W/B 0.45 0.45 0.35 0.35 0.35 0.35 AEA x x x Since a traditional concrete for freeze/thaw environments does not necessarily have to be air entrained according to NBN B15-001, both non-air entrained HVFA mixes (F(1)50 and F(2)50) and air entrained HVFA mixes (F(1)50A and F(2)50A) were evaluated in this research. The mixing time for each mix was 5 min: 1 min of dry mixing, 2 min of mixing after adding the mixing water and another 2 min of mixing after adding the polycarboxylic ether-based superplasticizer (SP) (dry matter mass percentage: 35%). In case of air entrainment, the AEA was added to the mixing water before being added to the binder, sand and aggregates. The AEA dosage was fine-tuned experimentally with the aim of achieving an air content in the fresh state of 6-7% cf. ACI 201.2R. Each HVFA composition was made twice: once with a fine fly ash F(1) (45 μm fineness: 13.2% retained) and once with a coarser fly ash F(2) (45 μm fineness: 26.6% retained). Note that all HVFA mixtures were characterized by a higher binder content B (cement + fly ash: 450 kg/m³) and a lower water-to-binder ratio (W/B: 0.35) compared to the references T(0.45) and T(0.45)A. The higher binder content and lower W/B ratio also explain the different proportioning of the inert fraction per m³ (sand 0/4: 645 kg/m³, gravel 2/8: 465 kg/m³, gravel 8/16: 606 kg/m³). The basic composition used in the SAP concrete investigation is based on ordinary concrete in road construction with a water-to-cement ratio of 0.45; 350 kg/m³ CEM I 42.5 N, 157.5 kg/m³ water, 690 kg/m³ quartz sand 0/4, 188 kg/m³ gravel 2/8, 1003 kg/m³ gravel 8/16 and an amount of superplasticizer. The concrete was made in the framework of a round-robin test of the technical committee 225-SAP from RILEM. As a superplasticizer, a commercial product

-naphthalene sulphonate (BNS) was used (Woerment FM30/BV30 from BASF; 1.2 g/cm³). All concrete mixtures showed a consistency class F3 (soft) as specified in EN 1045-2. Bulk-polymerized SAPs (two types) were added on top at 0.15 m% of cement weight. Both SAPs are able to take up approximately 300 g demineralized water and 33.3 g mixing water per g SAP. Their sizes are 180 ± 45 μm (SAPa) and 70 ± 21 μm (SAPb). The amount of additional water was 17.5 kg/m³ to receive a total water-to-cement ratio of 0.50. To retain consistency class F3, the dosage of superplasticizer was increased. Additionally, mixtures with SAPs but without additional water and a mixture with the additional water on top but

144



without SAPs were made. A comparative reference incorporating a conventional air-entraining agent (AEA) was used as well. The air-entraining agent was a commercial product (LP75 from BASF) and was simply added on top by first homogeneously mixing it with the mixing water. The amount of air entrainer was 0.025 m% (by weight of cement). The different mixtures are shown in Table 2. Table 2: Studied SAP and corresponding reference concrete mixtures with nomenclature.

REF I REF II SAP 1 SAP 2 SAP 3 SAP 4 AEA (W/C)tot 0.45 0.50 0.45 0.45 0.50 0.50 0.50 (W/C)add - - 0.05 0.05 0.05 0.05 - (W/C)eff 0.45 0.50 0.40 0.40 0.45 0.45 0.50 SAPa x x SAPb x x AEA x REF I and REF II have water-to-cement ratios of 0.45 and 0.50, respectively. SAP1/2 is a SAP3/4 mixture without additional water, containing SAPs. AEA is a mixture containing the air entraining agent. The total, additional and effective water-to-cement ratios are given for comparative reasons. The mixing procedure started with a 2 min homogenization of the dry powders and the dry SAPs in the respective mixtures, at low speed (Rotating pan mixer Zyklos 50 l). In the following 30 s, water (and air-entrained for the AEA mixture) was added. The whole was mixed for 1 minute and the superplasticizer was added in the next 15 s, followed with an additional 2 min of mixing at high speed. 2.2 Air content measurements in the fresh and hardened state The air content in the fresh state was determined using the Standard EN 12350-7 with the pressure gauge method. A known volume of air at a known pressure is hereby merged in a sealed container with the unknown volume of air in the concrete sample. The dial of the pressure gauge gives the percentage of air for the resulting pressure. The observed microstructures in hardened state were investigated by means of air void analysis based on the Standard EN 480-11. Two specimens (100 × 150 mm² area) were hereby cut perpendicular to the longitudinal axis and the surface was then ground and polished to produce a smooth flat surface finish. The specimen surface was treated to produce a better contrast between the air voids and the cementitious matrix. First, black ink was applied on the surface. Then, the surface was covered with fine barium sulphate powder which was pressed in the air voids. The excess white powder was then removed and the specimens could be microscopically studied. The air void structure was hereby examined by scanning along a series of traverse lines parallel to each other. The air voids intersected by the traverse lines were recorded and the total porosity could be determined. 2.3 Mechanical properties and salt-scaling test The compressive strength of the materials was tested according to the Standard EN 12390-3 and the test to determine the freeze-thaw resistance was the slab test according to the Standard CEN/TS 12390-9 (Slab test) on cylindrical specimens of 100 mm in diameter and 80 mm high.

145



3. Results and discussion 3.1 Fresh state properties and air void system in hardened state The use of fly ash in concrete has an impact on the required AEA dosage (Table 3). For the reference T(0.45)A, a dosage of 2.0 ml/kg binder was sufficient to have an air content (6.8%) conforming to all standards. With respect to the air entrained HVFA mixtures F(1)50A and F(2)50A, a much higher AEA dosage (5.0 ml/kg binder) resulted in an air content that was considerably lower than the 6.8% of the reference: 4.9% for mix F(1)50A and 5.2% for mix F(2)50A. As a consequence, the air content criterion imposed by ACI 201.2R was not met. Table 3. AEA and SP dosage [ml/kg B] and fresh concrete air content [%] for HVFA and the corresponding reference concretes. T(0.45) T(0.45)A F(1)50 F(1)50A(-2) F(2)50 F(2)50A(-2) AEA (ml/kg B) - 2.0 - 5.0 (7.0) - 5.0 (7.0) SP (ml/kg B) 2.0 4.0 7.0 7.0 (4.0) 5.0 5.0 (4.0) Air content (%) 2.8 6.8 2.6 5.2 (6.9) 2.8 4.9 (7.3) Two additional air entrained HVFA mixes with an increased AEA dosage (7.0 ml/kg B instead of 5.0 ml/kg B) were made: F(1)50A-2 and F(2)50A-2. Their fresh air contents amounted to 6.9% for mix F(1)50A-2 and 7.3% for mix F(2)50A-2, respectively. Within the framework of the SAP concrete research, the fresh concrete air content was approximately 3% for all mixtures made (Table 4). The AEA mixture shows 6-7% of air content, needed for an adequate salt scaling resistance. Table 4: Fresh concrete air content [%] for SAP and the corresponding reference concrete.

REF I REF II SAP 1 SAP 2 SAP 3 SAP 4 AEA Air content [%] 3.0 2.3 3.3 3.5 3.7 3.3 6.6 The air content and spacing factor in hardened state are important parameters to be considered when studying the salt scaling resistance. For the research related to the HVFA concrete, the obtained automated air-void analysis results are shown in Table 5. Compared to T(0.45), the air entrained reference T(0.45)A was characterized by a much higher air content (7.5-8.0%). Note that the higher values for the T(0.45)A reference mix are somewhat different from its initial air content (6.8%, Table 3) in the fresh state. Regarding the HVFA compositions, air entrainment resulted in an increase of the air content, though the values obtained were lower than 6-7%. Compared to F(1)50 and F(2)50, the average hardened air contents of F(1)50A and F(2)50A were increased by only 1.8% and 1.5%, respectively. The air contents in hardened state were more or less in accordance with the measured air contents in the fresh state (F(1)50A: 4.9%, F(2)50A: 5.2%). Not only a sufficient total air content is of importance, but also an adequate distribution of the artificial air bubbles. In other words, their spacing should be close enough to prevent the development of pressures from freezing which would fracture the concrete. The fulfilment of this requirement is usually evaluated through the calculation of a spacing factor for the concrete. This is the maximum distance from any point within the concrete matrix to the edge of the nearest air bubble. According to NBN B15-001 and ACI 201.2R, a spacing factor of

146



200 μm is recommended as the maximum for concrete exposed to freeze/thaw attack. The spacing factors of all air entrained concrete compositions (T(0.45)A, F(1)50A, F(2)50A) do not exceed this maximum value, although the HVFA mixes are characterized by somewhat higher values (100-200 μm) when compared with the OPC reference (± 100 μm). Without air entrainment, the spacing factors of T(0.45), F(1)50 and F(2)50 range between 250 and 300 μm. Table 5: Air content and spacing factor of HVFA and the corresponding reference concrete after automated air-void analysis. T(0.45) T(0.45)A F(1)50 F(1)50A(-2) F(2)50 F(2)50A(-2) Air content [%] 2.7±0.6 8.0±0.4 3.5±0.6 5.2±0.9

(7.3±0.7) 3.0±0.7 4.6±0.6

(7.9±0.2) Spacing factor [μm] 291±36 86±3 282±71 119±12

(82±8) 252±39 130±21

(77±1) Increasing the AEA dosage from 5.0 ml/kg B to 7.0 ml/kg B, resulted in a more adequate air void system in hardened state for HVFA concrete. The total air contents measured for F(1)50A- 2 and F(2)50A-2 slightly exceeded the 6-7% criterion recommended by the ACI 201.2R guideline. Spacing factors decreased to values around 100 μm cf. T(0.45)A. All specimens examined in view of obtaining a salt scaling resistant SAP concrete showed an air content of 5.5-11.2 % and a spacing factor of 91-277 μm (Table 6). Table 6: Air content and spacing factor of HVFA and the corresponding reference concrete after automated air-void analysis.

REF I REF II SAP 1 SAP 2 SAP 3 SAP 4 AEA Air content [%] 9.4±0.8 8.9±1.1 5.5±0.7 6.5±1.6 10.0±0.4 10.1±0.3 11.2±1.9 Spacing factor [μm] 97±4 143±3 277±21 272±48 162±11 91±1 115±8 The cumulative air content for all studied mixtures is given in Figure 1 (average of two 100 × 150 mm² samples). If one would start from the initial size of the superabsorbent polymers and the amount of additional water (33.3 g mixing water/g SAP for both SAPs with a SAP bulk density of 700 kg/m³), one could calculate the final swollen size of the SAP particles in the concrete. SAPa with an initial size of 180 ± 45 μm attains a swollen size of 670 ± 167 μm μm. For SAPb, this is 70 ± 21 and 260 ± 78 μm, respectively. These values correspond with the results of the automated air void analysis. The AEA mixture shows typical additional pores with pore sizes of approximately 250 μm. The fact that SAP macro pores are fluid-filled upon the determination of the air content in the fresh state and empty in the hardened state, explains the higher values found in Table 6 compared to Table 4. One can calculate the total number of SAPs in the concrete, based on the m% of cement weight, the initial dry size and the bulk density of the SAPs. Assuming a uniform distribution (primitive cubic order arrangement) of the SAPs, the spacing between the centres of the SAPs can be compared. If one subtracts the found swollen sizes, an estimation of the spacing factor can be calculated, hereby neglecting all other voids. These are 925 μm and 359 μm for SAPa and SAPb, respectively. The values in Table 6 are lower due to the neglected air voids in the calculation, the random distribution of the voids and the use of aggregates. The trend for the

147



smaller SAPb is found when comparing SAP 3 versus SAP 4; a lower spacing factor. The standard deviation in SAP 2 is too large to make such analogous qualitative comparison for SAP 1 versus SAP 2.

Figure 1: Cumulative air content of the studied samples.

3.2 Mechanical properties At 28 days, the HVFA concrete mixtures have a lower compressive strength than the corresponding OPC references (Table 7). This behaviour can mainly be attributed to the slow pozzolanic reaction of the fly ash which first requires the presence of Ca(OH)2, one of the main hydration products of the Portland cement reaction. When looking at the effect of the applied AEA dosages, it is clear that adding 5.0 ml AEA/kg B resulted in rather limited strength reductions for the HVFA mixtures (-8 MPa). Given the rather low air contents achieved as such (see Table 5) this is not very surprising. Applying the increasing dosage of 7.0 ml AEA/kg B to achieve a more satisfactory air void system had a much more pronounced effect in terms of strength reduction (-23 MPa). Table 7: Mechanical properties [MPa] of the HVFA and corresponding reference mixtures. The results show averages and standard deviations on the single results.

T(0.45) T(0.45)A F(1)50 F(1)50A(-2) F(2)50 F(2)50A(-2) Compression 69.3±1.4 53.2±2.0 59.7±0.3 51.3±0.4

(36.8±3.9) 51.7±0.8 43.6±0.4

(29.1±5.6) When using SAPs and additional water, the strength decreases within an acceptable range (Table 8). Increasing the water-to-cement ratio without adding SAPs, the strength also decreases within the range of the SAP mixtures without additional water. Using an air-entraining agent decreases the strength to a high extent, especially due to the entrained air content. When using SAPs and no additional water, the effective water-to-cement ratio decreases, thus increasing the strength. The combined effect with the strength reduction by the SAPs leads to a material with approximately the same strength as the REF I mixture.

-2

0

2

4

6

8

10

12

0-10

15-2

025

-30

35-4

045

-50

55-6

065

-80

85-1

0010

5-12

012

5-14

014

5-16

016

5-18

018

5-20

020

5-22

022

5-24

024

5-26

026

5-28

028

5-30

030

5-35

035

5-40

040

5-45

045

5-50

050

5-10

0010

05-1

500

1505

-200

020

05-2

500

2505

-300

030

05-4

000

Cumulative air content [%]

Void diameter [μm]

AEA SAP 3 SAP 4 REF I REF II SAP 2 SAP 1

148



Table 8: Mechanical properties [MPa] of the SAP and corresponding reference mixtures. The results show averages and standard deviations on the single results.

REF I REF II SAP 1 SAP 2 SAP 3 SAP 4 AEA Compression 57.8±0.7 52.0±0.9 58.0±1.0 50.0±2.1 49.3±1.0 48.6±0.8 33.8±1.4 3.3 Salt scaling resistance The salt scaling results are given after 7, 14 and 28 cycles (Figure 2) for all studied mixtures. From Figure 2a it is clear that ordinary Portland cement concrete does not necessarily require air entrainment to ensure limited salt scaling. The addition of a limited amount of AEA (2.0 ml/kg B) to the same OPC concrete mixture was found to be very effective, since the mass loss per unit area was negligible (0.07 kg/m²). In contrast with reference mixture T(0.45), the performance of the non-air entrained HVFA mixtures subjected to the same experiment was far from acceptable. Using an AEA significantly improved the salt scaling resistance. However, the effectiveness of the admixture was less than in reference T(0.45)A due to the typical partial adsorption of the AEA by the fly ash. Although the AEA dosage was more than doubled (5.0 ml/kg B versus 2.0 ml/kg B), higher mass losses per unit area were recorded for the HVFA mixtures F(1)50A and F(2)50A. An acceptable salt scaling resistance was confirmed for mixtures F(1)50A-2 and F(2)50A-2 with a more adequate air void system. A more in-depth characterization of their air void systems in relation to their salt scaling performance can be found in Van den Heede et al [15].

(a) (b)

Figure 2: The amount of scaled material as a function of the number of cycles by conducting freeze-thaw resistance tests for the HVFA (a) and SAP (b) series. The REF II mixture showed the highest scaling as the mixture was especially susceptible due to the high water-to-cement ratio. The next mixture is REF I. Both SAP 3 and SAP 4 show a lower scaling but not as good as the AEA mixture, and they are not significantly different in terms of scaling. The SAP 1, SAP 2 and AEA behave best in terms of reduced scaling. The difference between SAP 1/2 and SAP 3/4 is due to the lower apparent water-to-cement ratio. This causes a further increase in the salt scaling resistance as the strength is increased. The rate of scaling showed that the reference mixtures deteriorated more rapidly compared to the mixtures with SAPs, again proving their increased salt scaling resistance. The changes in mass loss for SAP-containing mixtures in comparison to the SAP-free concrete (REF I) after

012345

0 14 28# cycles

REF I REF IISAP 1 SAP 2SAP 3 SAP 4AEA

Scaling [kg/m²]

012345

0 14 28

T(0.45) T(0.45)AF(1)50 F(1)50AF(1)50A-2 F(2)50F(2)50A F(2)50A-2

Scaling [kg/m²]

# cycles

149



28 freeze/thaw cycles are 60% for SAP 3, 50% for SAP 4, 95% for SAP 1, 92% for SAP 2, 97% for AEA and -26% for REF II. Thus, there was a considerable improvement in the materials performance due to the use of SAPs in terms of decreasing mass loss after a given number of freeze/thaw cycles. 4. Conclusions

The overall results of the first series of HVFA concrete indicate that it is possible to design a HVFA concrete composition with an adequate salt scaling resistance under laboratory conditions. To achieve this, a lot of attention must be paid to the applied AEA dosage and the resulting air void system.

Air void analysis proved to be useful to determine the evenly distributed void system created by the superabsorbent polymers. The macro-pore sizes were as expected.

The compressive strength is mostly governed by the formation of macro pores by the SAPs when using a high water-to-cement ratio and additional water.

Superabsorbent polymers are able to increase the freeze-thaw resistance as they caused less scaling due to the formation of an evenly-distributed pore system, when using additional water. When SAPs without additional water were used, the performance even further improved considerably. The overall improvement was hereby similar to that obtained with conventional air entrainment. However, when using an air-entraining agent, the strength decreased significantly, even more compared to using SAPs with additional water in a mixture.

Acknowledgements As a Research Assistant of the Research Foundation-Flanders (FWO-Vlaanderen), D. Snoeck wants to thank the foundation for the financial support. As postdoctoral researcher, P. Van den Heede gratefully acknowledges the funding by SIM (Strategic Initiative Materials in Flanders) and IWT (Agency for Innovation by Science and Technology). References [1] Boel, V., Microstructure of self-compacting concrete related to gas permeability and

durability aspects (in Dutch), PhD thesis, Ghent University (2006). [2] Powers, T. C., The mechanisms of frost action in concrete. Stanton Walker Lecture Series

on the Material Science (1965). [3] Hansen, W. C., Crystal growth as a source of expansion in Portland cement concrete. Am

Soc Testing Mats, Proc 63 (1963), 932-945

150



[4] Snyder, M. J. (Ed.), Protective coatings to prevent deterioration of concrete by deicing chemicals, National Cooperative Highway Research Program Report 16, Transportation Research Board, USA (1965)

[5] , O., The characterization, improvement and modelling aspects of frost salt scaling of cement-based-materials with a high slag content, PhD thesis, TU Delft (2006)

[6] Valenza II, J. J. and Scherer, G. W., A review of salt scaling: I. Phenomenology, Cem Concr Res 37 (2007), 1007-1021

[7] Valenza II, J. J. and Scherer, G. W., A review of salt scaling: II. Mechanisms, Cem Concr Res 37 (2007), 1022-1034

[8] Gruyaert, E., Effect of blast-furnace slag as cement replacement on hydration, microstructure, strength and durability of concrete, PhD thesis, Ghent University (2009)

[9] Malhorta, V.M. and Mehta, P. K., High Performance, high-volume fly ash concrete: Materials, mixture proportioning, properties, construction practice, and case histories, Second edition, Supplementary Cementing Materials for Sustainable Development Inc., Canada (2005)

[10] Jensen, O. M. and Hansen, P. F., Water-entrained cement-based materials. I. Principles and theoretical background, Cem Concr Res 31 (2001), 647-654.

[11] Snoeck, D. et al, Visualization of water penetration in cementitious materials with superabsorbent polymers by means of neutron radiography, Cem Concr Res 42 (2012), 1113-1121

[12] Snoeck, D. et al, Self-healing cementitious materials by the combination of microfibres and superabsorbent polymers, J Intel Mat Syst Str 25 (2014), 13-24.

[13] Mechtcherine, V. et al, Effect of superabsorbent polymers (SAP) on the freeze-thaw resistance of concrete: results of a RILEM interlaboratory test, Mat Str (2016), accepted

[14] Hasholt, M.T. et al, Superabsorbent polymers as a means of improving frost resistance of concrete, Adv Civ Eng Mat 4 (2015), 237-256

[15] Van den Heede, P. et al, Influence of air entraining agents on deicing salt scaling resistance and transport properties of high-volume fly ash concrete, Cem Concr Compos 37 (2013), 293-303.

151



FREEZE-THAW-ATTACK ON CONCRETE STRUCTURES – LABORATORY TESTS, MONITORING, PRACTICAL EXPERIENCE Frank Spörel (1) (1) Federal Waterways Engineering and Research Institute, Karlsruhe, Germany Abstract Waterway structures like locks are exposed to severe freeze-thaw attack. Sufficient resistance of the concrete to this exposure has to be assured. In Germany the concrete used for federal waterway structures in exposure class XF3 has to undergo a CIF test in the laboratory in addition to meeting descriptive requirements. In the meantime, experience has been gained with this procedure and several structures based on it have been in operation for up to ten years. In order to establish the procedure, research was conducted concerning the transferability of laboratory tests to practical experience. One important aspect here was a service life study on the degree of water saturation of the concrete under practical conditions in combination with temperature exposure. Taking the example of a lock, this paper presents the results of laboratory tests as well as monitoring data for the degree of water saturation and temperature exposure and undertakes an assessment of the condition of the structural concrete after ten years in operation. 1. Introduction Waterway structures are solid structures that place special requirements on concrete properties. A low heat of hydration of the concrete is necessary to minimize restraint. This results in certain limitations concerning cement properties and content. To achieve a sufficient freeze-thaw resistance and meet the requirements resulting from limitations on the heat of hydration, consideration must be given to the concrete technology when choosing the raw materials and mix design. In addition to European and national standards [1, 2], requirements regarding concrete for the construction of waterway structures under the responsibility of the Federal Ministry of Transport and Digital Infrastructure (BMVI) have been regulated for more than 3 decades

152



now in the “Supplementary Technical Contract Conditions – Hydraulic Engineering” [3]. Concerning freeze-thaw exposure in exposure class XF3, the concrete has to undergo CIF testing according to the BAW Code of Practice “Frost Resistance Tests for Concrete” [4]. The procedure is similar to that described in [5]. When testing slowly hardening concrete, the test starts at an age of 56 days. After casting, the specimens are stored for 14 days in water and subsequently for 42 days in a climate chamber at 20°C / 65% relative humidity. One question which has been discussed ever since the CIF test became mandatory in 2004 is the transferability of the test results to the performance of the exposed concrete during operation of the structure. In particular, very little has been known until now about seasonal variations in the degree of water saturation as an important aspect of a freeze-thaw attack; information has also been lacking on additional water adsorption, which can be described by the micro-ice lens pump according to Setzer [6]. For this reason, a monitoring system was installed on several structures [7] to contribute to new findings concerning temperature exposure and the degree of water saturation by the concrete under different conditions. The data was evaluated to receive an impression of the intensity of freeze-thaw attack. This paper describes the results for a lock taking account of the actual condition of the structure. 2. Structure and Measuring System 2.1 General To investigate freeze-thaw attack on concrete structures, a lock was equipped with sensors. Several measuring points were installed, so that observations were possible at parts of the structure with different moisture and temperature exposure. A data logger with a remote control allowed an immediate analysis of the data which was collected over a period of several years. A non-destructive determination of the degree of saturation of concrete is only possible using indirect measurement methods. A continuous, depth-dependent measurement of the resistivity was transferred to the degree of saturation by means of a calibration in the laboratory. The resistivity measurement was conducted using a multiring electrode (MRE). The MRE is a sensor consisting of several rings of stainless steel, each with a thickness of 2.5 mm, with an insulating plastic ring between two steel rings. It enables AC resistance measurements of the concrete between two adjacent steel rings in eight steps at a frequency of 10.8 Hz and at a depth of 7 to 42 mm from the concrete surface. The measuring depth can be increased to 87 mm using two MREs and a distance piece [8]. A multitemperature probe (MTP) is installed near the MRE in order to monitor temperature exposure. The MTP is equipped with eight PT 1000 sensors to facilitate temperature measurements at eight different distances from the concrete surface. The example in Figure 1 shows a sectional view of a lock with four measuring points installed inside the northern side wall of the northern lock chamber. The concrete surface is orientated in a southerly direction. Measuring point MP1 is permanently submerged whereas MP2 and MP3 are located between the head water and the tail water, and are thus in frequent contact with water (XF3). MP4 is unsheltered and exposed to rain and freezing (XF1). The concrete with blast furnace slag cement and fly ash had an equivalent water-to-cement ratio of 0.47 (k=0.4) without an air entraining agent (Table 1). To demonstrate the typical effects in exposure classes XF3 and XF1, this paper will focus on MP3 and MP4.

153



Fig. 1 Measuring points in the side wall of the lock Table 1: Concrete mix design

Material Content [kg/m³]

CEM III/A 32.5 N LH 270 Fly ash 80 Water 140

Plasticizer 3 Aggregate 1876

The investigations were conducted by the Institute of Building Materials Research, RWTH Aachen University, by order of the Federal Waterways Engineering and Research Institute. 2.2 Calibration To enable the resistivity to be converted to the degree of water saturation of concrete, a calibration provided a relation between these two parameters. This relation was determined by simultaneously measuring the degree of saturation and the resistivity using the two-electrode method (TEM) at a concrete temperature of 20°C. Different degrees of saturation were set in 28 concrete disks with a diameter of 80 mm and a height of 20 mm by capillary suction and drying at a temperature of 60°C. The disks were sealed and stored for two weeks prior to the resistivity measurements, to allow a uniform moisture distribution inside them. To minimize the influence of hydration on the resistivity, the calibration was started on concrete at least two years old. The water content is the most dominant parameter influencing the resistivity of concrete. However, besides the concrete-specific parameters like the w/c ratio, degree of hydration, cement type, content of additions or carbonation, resistivity is also influenced by environmental conditions such as temperature variations or the chloride content due to marine exposure or the use of de-icing agents [9]. Investigations aiming to determine the degree of saturation indirectly by means of resistivity measurements have to consider these influencing parameters to avoid misinterpretations of the data. 2.3 Influence of temperature on resistivity Apart from the degree of water saturation, concrete resistivity is also influenced by temperature, among other things. The calibration functions are only valid for the temperature at which the tests were conducted, in other words the influence of temperature on resistivity

154



has to be considered when the calibration function is applied to measurements at a structure. Whereas concrete-specific parameters were considered by the calibration of the tested concrete, the temperature was taken into account by a compensation routine which bears in mind the simultaneously measured resistivity as well as the temperature measured at the structure. The Arrhenius equation (Equation 1) was used to compensate the influence of temperature on the resistivity of concrete. This equation enables the resistivity to be calculated at a temperature of 20°C, as in the calibration test, from the measured temperature and resistivity values at the structure. The constant b depends on technological aspects and the degree of saturation [9, 10].

0T1

T1b

e0,elel (1) el Electrolytic resistivity at temperature T in m el,0 Electrolytic resistivity at temperature T0 in m

T, T0 Absolute temperature in K b Constant in K

Owing to the variation range of the constant b, data from the structures was evaluated and laboratory tests carried out to consider the concrete mix design, degree of water saturation and conditions on site. In addition to TEM measurements and an evaluation of the data from the structures according to [11], MRE tests were carried out on partially saturated and sealed concrete cubes with an edge length of 200 mm. Inside these cubes, MREs and MTPs enabled continuous resistivity measurements during a 12-hour air temperature cycle in the range from +20°C to -15°C at a heating and cooling rate of 6.3 K/h. The air temperature was kept constant at -15°C for one hour. The heating and cooling rate of 6.3 K/h was based on the average extreme values observed at different structures during freezing and thawing. Additional tests were performed at a heating and cooling rate of 1.6 K/h to investigate whether this rate has an influence on the constant b. This was the range most frequently observed during freeze-thaw cycles at the structures. The correlation of the constant b to the degree of saturation was determined by transferring the measured resistivity into the degree of saturation with the aid of the calibration functions. The influence of the concrete mix design, concrete age and degree of saturation was considered in this way. The measurements were carried out at a high concrete age of several years because the data recording for the long term measurement also started at a high age. That way the influence of hydration was generally negligible. The effectiveness of compensating the influence of temperature on resistivity were mainly investigated for temperatures > 0°C, and there were pointers to a distinct influence of temperature in the temperature range from 10 to 50°C [12]. Investigations in temperature ranges below 0°C are described in [13]. Temperatures below 0°C are of special interest for freeze-thaw attack, which is why investigations were carried out in that direction. Two different states of concrete saturation were studied; the test surface was sealed with cling film during the test to minimize evaporation or water uptake during the temperature cycle. The first test was performed after storage in a climate chamber at 20°C / 65% relative humidity. After a ten-day period of capillary suction, the test was repeated with a higher degree of saturation. The gradient of saturation in both tests enabled a measurement over a wide saturation range.

155



3. Results 3.1 Calibration The resulting calibration function of the concrete used for the lock is shown in Figure 2 for concrete stored in laboratory conditions and for cores taken from the structure at an age of about 5 years. Differences were observed in the degree-of-saturation range below about 50%. As the concrete of the cores is most representative, the calibration function for the core concrete was used to convert the data from the structure to the degree of saturation. The range of resistivity measured at the structure was covered by the calibration function.

Fig. 2: Calibration functions of the concrete used for the lock 3.2 Influence of temperature on resistivity In order to use the calibration function, the data from the structure has to be transferred to a temperature of 20°C according to Equation 1. It was possible to determine the constant b at temperatures below 0°C by means of MRE measurements as described in section 2.3. In addition to the lock concrete, 6 cubes of different concretes were tested and their resistivity transferred to the degree of saturation using the various calibration functions. The concretes used for a tunnel, a bridge and a quay wall all had water-to-cement ratios of about 0.5 but different types of cement as well as different cement and fly ash contents [7]. The constant b decreased as the degree of saturation increased. Depending on the 7 different concretes investigated, the constant b varied from about 7000 K at low degrees of saturation of approximately 0.1 to 0.2 to about 2500 K under water saturated conditions (degree of saturation approximately 0.9). The results enabled sufficient account to be taken of the influence of temperature on resistivity. Furthermore, the test revealed a special effect which occurred at high degrees of saturation and temperatures below about -2°C. Figure 3 shows the temperature compensated resistance during the temperature cycle described in section 2.3. Figure 3 (top) shows an almost linear development of the concrete resistance during the test after storage in laboratory conditions (20°C / 65% relative humidity). No temperature effects can be detected, proving the efficiency of the compensation routine. Figure 3 (bottom) depicts the test results for the same specimen following a capillary suction period of about 10 days. An almost linear development is likewise observed at a measuring depth of 82 mm. Closer to the concrete surface an abrupt increase in the resistance is visible when the temperature falls below about -2°C. The compensation routine according to Equation 1 is not able to eliminate these temperature effects at low temperatures for highly saturated concrete. This increased resistance is accompanied by a deviation of the otherwise linear temperature decrease at a distance of 7 mm from the surface. This is a sign of a phase transformation from water to ice.

156



In other words, a phase transformation from water to ice can be detected using resistance measurements. The fact that the dry concrete (Fig. 3 (top)) does not exhibit these effects indicates that freezing of water in concrete is dependent on the degree of saturation. In rather dry conditions of the concrete, it is mainly fine pores that contain water. Freezing, or at least freezing as detectable by resistivity measurements, does not occur at temperatures relevant for practical conditions. The minimum degree of saturation at which freezing occurs was determined in further tests by means of the calibration function. The evaluation showed that freezing of water in the pore structure occurred at a degree of saturation in the region of hygroscopic saturation of the different concretes at a relative humidity of about 95%. This was deduced by transferring the resistivity values to the degree of saturation using the calibration functions. These results were compared with the hygroscopic saturation of the concrete, which was determined by storing concrete disks according to section 2.2 in exsiccators with a relative humidity between 65 and 95%. Similar degree-of-water saturation ranges at which frost problems can occur were observed for Portland cement pastes [14].

Fig. 3: Effectiveness of compensating the temperature influence on concrete resistance In order to describe the effect of freezing water in the pore structure on the resistivity in Equation 2, a factor FG(T) is introduced [7] which describes the ratio of resistivity with a compensated temperature influence in the frozen state of the water to the liquid state. The data in Figure 3 was analysed and the factors FG(T) determined at a temperature of -11°C. The results are documented in Table 2. FG(T)=Rg(T) / Rf (2)

157



Rf Temperature-compensated resistivity (liquid pore water) before freezing Rg(T) Temperature-compensated resistivity (frozen pore water) at a temperature T Table 2: Factor FG(T) for the test results in Figure 3.

Depth FG(-11°C)

7 mm 200 17 mm 45 27 mm 2 82 mm 1

It becomes obvious that at a measuring depth of 82 mm no change in resistivity occurs (FG=1), whereas closer to the surface the factor FG increases. This can be explained by the higher degree of saturation closer to the surface after the period of capillary suction. This higher degree of saturation implies that a higher amount of water is available, which is freezable in the tested temperature range. It is assumed that at higher degrees of saturation more water is present in the capillary pores than in a dryer state. This method of analysing the resistivity data seems to be suitable for investigating water freezing in the pores of concrete. 4 Evaluation of Freeze-Thaw Exposure at the Structure 4.1 Degree of saturation The degree of saturation refers to the water absorption of the concrete at a pressure of 150 bar. The degree of saturation is 0.91 for the investigated concrete at atmospheric pressure. This specific value provides a basis for evaluating the calculated saturation degree of the concrete in the structure during operation. High degrees of saturation in the capillary saturation region of the concrete were calculated at the measuring point for class XF3 exposure during almost the entire observation period (Fig. 4).

Fig. 4: Calculated degree of saturation of the lock for class XF3 exposure The concrete only dried during lock inspection intervals. No significant changes in the degree of saturation were observed during normal operation of the lock. A lower, mainly constant

158



degree of saturation was observed at measuring point MP4 at a distance of more than 40 mm from the surface. Higher average degrees of saturation were determined closer to the surface during winter periods and lower average degrees of saturation during summer periods (Fig. 5). Degrees of saturation in the order of saturation at atmospheric pressure also occured for a short period during single events at this XF1 measuring point close to the surface.

Fig. 5: Calculated degree of saturation of the lock for class XF1 exposure 4.2 Freezing of water in the pore structure In order to determine the transferability of laboratory test results to practical conditions, the monitoring data was analysed with regard to the effects described in Figure 3. An example is given in Figure 6. If the concrete temperature decreases below about -5°C in exposure class XF3 the resistance rises abruptly. A factor FG=6 at a distance of 87 mm from the surface can be deduced by analysing the data at a temperature of -11°C. According to [7], a tendency was observed for the factor FG(T) to increase when the degree of saturation increases.

Fig. 6: Freezing of water in the pore structure as detected by resistance measurements A factor FG(T)=1 was determined in exposure class XF1. Regarding the lower degree of saturation compared to exposure class XF3 at this measuring point, the different resistivity effects can be explained by the different amounts of water in the pore structure. According to the results of the laboratory tests described in section 3.2, a phase transformation from water to ice hardly seems possible for this concrete at the structure in exposure class XF1.

159



4.3 Surface condition of the structure After ten years in service, the condition of the structure was investigated during an inspection. There were large areas of the concrete surface where the hardened cement paste and mortar were eroded and the coarse aggregate was visible. The deterioration process was not yet complete (Fig. 7).

Fig. 7: Condition of the lock surface after 10 years in service This indicates that freeze-thaw attack has been taking place during the last ten years, leading to damage to the concrete surface. The requirement for damage – freezing of the water in the pore structure – was observed under practical XF3 conditions, accounting for the results of the monitoring system. This does not necessarily mean that damage has to occur, because the critical degree of saturation of this concrete is unknown; however, the freezing and thawing of water can induce the micro-ice lens pump effect according to [6], which may ultimately result in concrete damage if the pore structure has insufficient freeze-thaw resistance. Carbonation of the blast furnace slag cement concrete could also be responsible for the scaling, as the pore structure and thus the freeze-thaw resistance of carbonated concrete are different. The XF1-exposed parts of the structure showed no degradation of the surface. The monitoring data at these parts revealed no freezing of water in the pore system, meaning that the requirements for damage were not met. 5 Summary The actual freeze-thaw exposure of concrete – comprising temperature exposure, degree of water saturation and freezing of water in the pore structure – was investigated at a lock for exposure classes XF1 and XF3. The results were obtained by monitoring data in addition to laboratory tests. The main difference between XF1 and XF3-exposed concrete was that with XF3 exposure a mainly constant, high degree of saturation was observed in the complete 90 mm measuring zone whereas with XF1 exposure, seasonal and short-time changes were visible close to the surface. Fairly constant degrees of saturation were also determined above a concrete surface depth of about 40 to 50 mm, though at a lower level than with XF3 exposure. Furthermore, a decisive difference was noted between XF1 and XF3 exposure: resistivity measurements revealed effects which can be explained by water freezing in the

160



pore structure. To enable these effects to be described and quantified, a factor was introduced. These effects were only observed with XF3 exposure. An inspection of the structure after ten years in service revealed large areas of the concrete surface in exposure class XF3 where the hardened cement paste and mortar were eroded and the coarse aggregate was visible. This combination of an indirect degree-of-saturation measurement and a typical resistivity effect occurring when pore water freezes seems suitable for evaluating the intensity of a freeze-thaw attack. References [1] DIN EN 206-1:2001-07, Concrete – Part 1: Specification, performance, production and

conformity; German version EN 206-1:2000 [2] DIN 1045-2:2008-08, Concrete, reinforced and prestressed concrete structures – Part 2:

Concrete – Specification, properties, production and conformity – Application rules for DIN EN 206-1

[3] ZTV-W LB 215 – Supplementary technical contract conditions – hydraulic engineering, Federal Ministry of Transport and Digital Infrastructure, Germany, 2012

[4] BAW Code of Practice: Frost Resistance Tests for Concrete (in German), Federal Waterways Engineering and Research Institute (BAW), Karlsruhe, Germany, 2012 (http://www.baw.de/EN/service_wissen/publikationen/merkblaetter_empfehlungen_richtlinien/merkblaetter_empfehlungen_richtlinien.html)

[5] Setzer, M.J. et al.: RILEM TC 176-IDC: Internal damage of concrete due to frost action. Final recommendation: Test methods of frost resistance of concrete: CIF Test: Capillary suction, internal damage and freeze-thaw test – Reference method and alternative methods A and B. Materials and structures 37 (2004), 743-753

[6] Setzer, M.J.: CDF / CIF Freeze-Thaw Test Procedures – an adoption of the micro-ice lens pump model to testing and simulating practical conditions. Proceedings – Concrete under severe conditions: Environment and loading, Vancouver (2001), 428-438

[7] Spörel, F.: Frostbeanspruchung und Feuchtehaushalt in Betonbauwerken. Berlin, Beuth Verlag GmbH. Schriftenreihe des DAfStb, Nr. 604 (2013) (in German), PhD thesis, RWTH Aachen University

[8] Raupach, M.: Zur chloridinduzierten Makroelementkorrosion von Stahl in Beton. Berlin: Beuth Verlag GmbH. Schriftenreihe des DAfStb, Nr. 433 (1992) (in German), PhD thesis, RWTH Aachen University

[9] Elkey, W., Sellevold, E.J.: Electrical resistivity of concrete, Norwegian Road Research Laboratory, Publication No. 80, 1995

[10] Chrisp, T.M. et al: Temperature-conductivity relationships for concrete: An activation energy approach. Journal of material science letters 20 (2001), 1085-1087

[11] Schiegg, Y.: Online-Monitoring zur Erfassung der Korrosion der Bewehrung von Stahlbetonbauten. PhD Thesis, ETH Zürich (2002) (in German)

[12] McCarter, W. J.: Effects of Temperature on Conduction and Polarization in Portland Cement Mortar. In: Journal of the American Ceramic Society 78 (1995), 411-415

[13] Sato, T.; Beaudoin, J. J.: Coupled AC Impedance and Thermomechanical Analysis of Freezing Phenomena in Cement Paste. Materials and Structures 44 (2011), 405-414

[14] Bager, D. H.; Sellevold, E. J.: Ice formation in hardened cement paste, Part I. Room temperature cured pastes with variable moisture contents. Cement and Concrete Research 16 (1986), 615-792

161


22-23 August 2016, Lunds University, Lund, Sweden

METHODOLOGY TO ANALYSE THE SALT FROST SCALING MECHANISM(S) IN CONCRETE WITH DIFFERENT BINDERS Martin Strand (1), Katja Fridh (1)

(1) Building Materials, Lunds University, Lund, Sweden Abstract The purpose of this article is to present a methodology which can be used to evaluate how the salt frost scaling mechanism(s) affects concrete with different air void systems. First, a test is made where the binder is mixed with various combinations of superplasticizer and air entraining agent to find one combination which performs well. The combination is then used when casting concrete to produce four air void systems, from approximately 1.5% to 4.5% in total air content, which are quantified with an air void analysis made on the hardened concrete. The concrete is sealed and hydrated until for over 300 days which results in a high degree of hydration for any binder. Then a salt frost scaling method is used with a temperature cycle which increases the salt frost scaling mechanism(s) load. The results present scaling from concrete with a high degree of hydration, regardless of hydration rate, and the effect the air void system has on the salt frost scaling damage in various binders given an even microstructure since each sample is fully hydrated. This enables a study of how various microstructures are affected by a constant load from the salt frost scaling mechanism(s) and of the salt frost scaling process. 1. Introduction The purpose of this article is to present a methodology for a salt frost scaling analysis of close to fully hydrated binders with an analysis of the effect various air void systems has on each binder. The degree of hydration has a significant influence when analysing salt frost scaling resistance for different binders [1]. The binders with a slow rate of hydration will produce more scaling when testing after the same amount of time given equal conditions during the hydration. However, by allowing any binder to hydrate in sealed conditions for over 300 days salt frost scaling test results will give information about how the salt frost scaling mechanism affect various binders with fully developed microstructures. The air void system also has a significant effect on the salt frost scaling in concrete containing various binders, therefore it is interesting to study the effect from slight changes in the air void system for various binders.

162


22-23 August 2016, Lunds University, Lund, Sweden This paper presents a description of the methodology used, starting with a method to find a combination of superplasticizer (SP) and air entraining agent (AEA) for each binder to enable the creation of various air void systems with small differences. These differences were then quantified by an air void system analysis (linear traverse). Then the article presents the hydration process and briefly mention the salt frost scaling test method which is presented in [2] together with some results from four casts of CEM I concrete. Lastly the methodology is discussed and conclusions are drawn. 2. Test of superplasticizer and air entraining agent combinations The purpose of this test was to get indications of the difference in performance and learn about the risk of water separation for various combinations of admixtures (one AEA and one SP) given the same recipe (binder and aggregate fractions). The results then enable a choice of a AEA and SP combination (for each binder) which perform well and have a low risk for bleeding, which in turn enables a creation of various air void systems with only small differences. The test measured workability, the air content inside the fresh mortar, and a foam test which gives an idea of the risk for air voids to merge. The mortar recipe contained 2.000 kg binder, 6.000 kg of “0-8” aggregate fraction, and 0.800 kg water. 2.1 Method for the mortar tests To enable an attempt to produce an acceptable workability for all binders, the AEA was mixed with the water, and then the superplasticizer was added. The following mixing procedure was used:

The binder and aggregates were blended for 2.5 min ±5 sec Water was added and mixed for 2.5 min ±5 sec, Mixing continues for 2.5 min ±5 sec and a pipette was used to drip the superplasticizer in

to the mix and stop when one of the following occurred o the workability looked acceptable. o bleeding started (before an acceptable workability was acquired).

The average recommended mass for each AEA was used. When mixing the concrete for the salt frost scaling test a smaller mass of AEA were used to get close to 5% air content. However, at this stage it was assumed that if a larger (average recommended) mass of AEA work well with a specific superplasticizer, then a smaller mass also will perform well. 2.2 Results from the mortar tests After the mixing procedure was done the workability was measured with a Hägermann cone which is presented in Figure 1a. When the workability was measured the edge of the mortar was studied to look for indications of bleeding (Figure 1b and 1c). The indication of risk for separation was graded “clear”, “slight”, and “no” from visual observation.

163


22-23 August 2016, Lunds University, Lund, Sweden 1a

1b

1c

Figure 1. 1a) The Hägermann cone filled with mortar. 1b) When the edge of a mortar is considered to indicate no risk for water separation according to the grading system. 1c) A combination with clear risk for water separation, which happened before the mortar acquired an acceptable workability. Then the air content was measured on 1 dm³ mortar 5 min after the mixer had stopped. Then circa 0.5 dm³ of mortar were put in to a bucket of water, slowly mixed and dispersed with a large spoon which enabled a comparison between the foam created at the surface when the air inside the mortar ascend. Two different results are shown in Figure 2. In Figure 2b less foam is created since the bubbles come up to the surface and bursts quicker than the bubbles in Figure 2a. The foam created for each mix was graded from zero to two from visual observation in combination with photographs. 2a

2b

Figure 2. Both figures show an area of approximately 15x15 cm². 2a shows thick foam created at the surface with the foam rating 2. 2b shows less foam created with the foam rating 1. The results from the admixture combination tests should be compiled according to Table 2. The results presented are from the variation where each of three AEAs has been combined with each of five different superplasticizers for the same binder. Since the recommended average mass of each AEA was used for all mixes, the total air content in the mixes with different SPs should only vary by a few percent if the SPs do not affect the air content. According to the examples of results in table 2 there is clearly a wide spread of how well different combinations seems to work together with the binder. When evaluating the results a system was made to find the combination which seems to perform best out of the combinations which have been tested. First considerations were taken to the estimated risk for bleeding; the combinations that showed a clear risk were disregarded and focus was put on combinations considered to have no risk for water separation. Then the air content was studied by comparing the air contents for each AEA. As an example, when looking at AEA1 (in table 2) the combination with SP5 resulted in no risk for water separation, therefore approximately 12.5% was considered to be a minimum (or target) air content that the specified mass of AEA1 should result in when mixed with a SP with an acceptable interaction

164


22-23 August 2016, Lunds University, Lund, Sweden (i.e. the approximate minimum air content was decided depending on the results from a combination without bleeding). This means that SP4 was also considered since it only had a slight indication of water separation and air content above 12.5%. The third factor which was considered was the workability. Since the SP is added until it looked like the mortar had a workability which was approximately 150 mm (an approved flow-measurement was set to be 130 mm or more). The addition of SP was also stopped when the water started to separate from the paste, hence the mass of SP could become more than the maximum recommended amount. When focusing on the combinations which performed well, the results presented a strong correlation between no indication of separation, high air content and a larger volume of stable foam. Table 2. Compiled results from the mortar test of one binder.

Combination Indication of water separation

Air content [%]

Workability [mm]

Superplasticizer1)

[%] Foam

SP12)+AEA13) Slight 9,6 170 112,0 1 SP22)+AEA1 Clear 5,2 170 126,8 1 SP32)+AEA1 Clear 7,2 175 126,5 1 SP42)+AEA1 Slight 15,1 155 257,8 2 SP52)+AEA1 No 12,5 170 196,7 2 SP1+AEA23) Clear 4,3 175 417,3 0 SP2+AEA2 Clear 6,0 165 244,8 1 SP3+AEA2 Clear 18,5 120 103,8 1 SP4+AEA2 Slight 5,2 150 318,0 1 SP5+AEA2 No 11,5 180 125,7 2 SP1+AEA33) Slight 13,5 225 65,0 2 SP2+AEA3 Slight 18,5 140 37,0 2 SP3+AEA3 Slight 15,0 185 33,0 2 SP4+AEA3 No 20,0 120 69,5 2 SP5+AEA3 No 21,5 130 68,3 2 1) Mass of superplasticizer used divided by mass of maximum recommended amount. 2) SP4 is based on a sulphonate melamine-formaldehyde condensate. The rest of the SPs is based on different modified polycarboxylates. 3) AEA1 and AEA3 are based on a synthetic detergent. AEA2 is based on Vinsol resin. When a combination which performed well had been found from evaluating the results another complementing test was made to get an indication if this combination would release air over time from the fresh concrete. This was done by measuring the air content after 5 min (same as before), 30 min and 60 min on the same recipe as before. Results from these tests were also considered before a combination of AEA and SP for each binder was determined for the concrete recipe which was used for the salt frost scaling test. The measurements made on the combination SP5+AEA3 showed the air content between 17.0 and 19.0% for the three measurements and not indicating a decrease over time. This was considered a good result, therefore the combination was chosen without doing the same test with other combinations.

165


22-23 August 2016, Lunds University, Lund, Sweden 3. Concrete for salt frost scaling test 3.1 Recipe and casting the concrete To keep the binder content close to 430 kg/m³ the mass of aggregates was adjusted with consideration to the anticipated air content. The anticipated air content was 1.5, 2.5, 3.5 and 4.5 % and the mass of AEA was adjusted with the goal for the measured air content in the fresh concrete to be close to the anticipated air content for each cast. The air content was measured in a 8 dm³ of the fresh concrete, which was approximately 35% of the total mass of concrete mixed. Considering that the air content measurement was made on a large fraction of the cast, the measurement should be a good indication for the entire cast. When casting the concrete the mass of superplasticizer was constant for the four recipes which resulted in a slump of more than 150 mm (when no AEA was added) to facilitate the casting. 3.2 Curing To enable the tested binder to come as close as possible to fully hydrated conditions, the concrete was cured for 1 day inside a covered steel mould, 7 days submerged in water (approximately 20°C), and 300 days sealed inside a climate room with 20.0±0.1°C. The purpose of the relatively short hydration in water counteracted any early drying shrinkage at the same time as should have added a negligible additional mass of water to the concrete surfaces. Therefore the long hydration in the climate room was constantly 100% RH with a limited mass of water at the set temperature. 3.3 Air void analysis The air void analysis was done with the linear traverse methodology, according to the ASTM C 457 standard method [3], with one sample per cast. Linear traverse measurements are done by scanning a sample with a small light beam and at the same time measure the reflection. It requires thorough sample preparations to get accurate results from the reflections. First the sample surface is polished, then it is coloured with a black paint and the air voids are filled with zinc paste. When the light beam hits the zinc there is a strong reflection and the machine starts to draw a cord (line), when the light beam has past the air void filled with zinc and is back on the surface painted black there is much less reflecting light and the cord is finished. When approximately 3.5 meters has been measured on a sample (100x150 mm²), calculations are made to estimate the air content, spacing factor and specific surface from all of the cords. In table 4 results from linear traverse measurements are presented for four casts of concrete with 100 mass% CEM I as binder. Cast #4 contained ca 0.015 mass% of the air entraining agent. Table 4. Results from four casts made with 100% CEM I. The variables presented are calculated from the linear traverse analysis. Cast Fresh

Concrete [%]1)

Air void content Air void content in paste Specific surface [mm-1]2)7)

Spacing factor [mm]2)8)

Total [%]2)

<2 mm [%]2)3)

<0.35 mm [%]2)4)

<2.00 mm [%]2)5)

<0.35 mm [%]2)6)

#1 1,10 3,6 0,8 0,3 2,3 0,9 15 0,76 #2 2,40 5,8 2,5 1,2 7,0 3,4 19 0,37 #3 3,50 5,2 3,9 1,7 11,1 4,9 19 0,30 #4 4,80 2,2 2,2 1,5 6,7 4,6 33 0,22

166


22-23 August 2016, Lunds University, Lund, Sweden The results show that the specific surface increase and the spacing factor decrease when an increased mass of AEA is added to the mix. This trend is consistent when comparing the results from all voids and when looking at voids less than 2 mm. When all results was considered a general conclusion was drawn that the air void system was improved when the mass of AEA was increased. 4. Salt frost scaling test 4.1 Method The salt frost scaling method is developed for the methodology described in this paper where concrete with a high degree of hydration and with various air void systems are tested. During development some factors were chosen to enhance the effect from the salt frost scaling damage e.g. the temperature cycle which is used has been shown to increase the salt frost scaling damage in comparison to the CDF-cycle [1, 4]. This enabled the present methodology to test how the salt frost scaling mechanism(s) is affected by different factors and the effect the mechanism(s) has on concrete with favourable conditions (high degree of hydration and various air void systems) with a high resolution while using climate chambers with air as a thermal conductor. This enables a study as to which properties have the biggest impact on concrete salt frost scaling resistance. These must be considered when varying the binders used for the concrete in structures which is exposed to de-icing salts and freezing temperatures. A detailed description is given in [2]. 4.2 Scaling results The figures below show scaling results from four CEM I concrete casts presented in table 4. Figure 3 presents non-cumulative results from each separate cast including the mean scaling from all (six) samples together with the standard deviation. In [2] an argument is made that the salt frost scaling method contributes to a relatively constant salt frost scaling load. Considering this with the fact that Figure 3 shows non-cumulative results with measurements of the total scaling per seven cycles, the general gradient of the curve highlights information about an acceleration or deceleration of the scaling damage.

167


22-23 August 2016, Lunds University, Lund, Sweden

Figure 3. Scaling results measured at each measurement occasion (not cumulative). The line presents the mean scaling from the six samples from each cast after 112 cycles. The error bars present the standard deviation for the six samples at each occasion measurements were done. Upper left show the results from cast #1. Upper right show the results from cast #2. Lower left show the results from cast #3. Lower right show the results from cast #4. Figure 4 presents a compilation of all mean values of the results, 4a non-cumulative and 4b cumulative. Both highlight the decrease in salt frost scaling damage when improving the air void system, however, they should be presented together to complement each other. 4a shows indications of acceleration and the rate of the salt frost scaling process, while 4b highlights distinctions between each cast at the same time as it presents the total average scaling.

Figure 4. The average mass of scaling for one sample for each cast. 4a shows scaling measured at each measurement occasion. 4b shows the cumulative mass of scaling.

168


22-23 August 2016, Lunds University, Lund, Sweden Figure 5 shows a general summary of the results by presenting the total average mass of scaling for all samples, and the error bars show the total mass of scaling from the sample with the most and least mass of scaling after 112 cycles from each cast. This highlights the decrease in salt frost scaling damage, but a slight con is that there is no information provided about the salt frost scaling process for each material.

Figure 5. The numbers present the average mass of scaling per sample (6 samples per cast) after 112 cycles. The error bars present the total mass of scaling for the sample with the most and least mass of scaling after 112 cycles. 5. Discussion When analysing the results from the mortar test, air void analysis and salt frost scaling test it is clear that they seem to agree with three main ideas. First, the mortar test showed that the combination of AEA and SP is vital for minimizing the risk of bleeding and to enable the creation of a relatively stable air void system. Secondly, the results from the air void analysis show that it is also possible to create slightly different air void systems by changing the amount of AEA when using the combination from the mortar test. Finally, by testing concrete which have been sealed for a long time all samples from the same batch will have a high degree of hydration with little variation. This means that the microstructure will not change due to hydration during the testing which contributes to more constant conditions for the material during the test. By using the salt frost scaling method with a high precision and presenting the results according to the diagrams used in chapter 4, the air void systems effect on the salt frost scaling damage on the concrete can be observed for the binder (or binder combination) of interest.

5.940 ± 0.566 4.799 ± 0.302 4.116 ± 0.501 3.082 ± 0.372

1 2 3 4

Accu

mul

ated

scal

ing

[kg/

m²]

169


22-23 August 2016, Lunds University, Lund, Sweden Considering results presented in the current article, the methodology is considered to contribute information of the salt frost scaling mechanism(s) and how it affects various binders given favourable conditions. When a binder is hydrated according to this methodology it will be as close to fully hydrated as can be, given the water binder ratio used. When testing cast #1 (which is the fully hydrated concrete, only containing a SP which provides an excellent workability) with a salt frost scaling method which contributes a relatively constant salt frost scaling load, a non-accumulative and cumulative baseline is acquired for the binder’s (or concrete’s) salt frost scaling resistance. When this baseline for the binder is known, the impact in salt frost scaling resistance from adding AEA and thereby improving the air void system can then be measured for the specific binder. The salt frost scaling method used also contributes information about the salt frost scaling process over time for the binder. 5.2 Future studies To obtain more information about the salt frost scaling mechanism(s) and how it varies for concrete with various properties (e.g. concrete containing different SCMs) a grading analysis should be made on the scaled material. The grading will contribute more information about how the average size of the scaled material varies from the beginning of the test until the end of the test, which will give information about the salt frost scaling process. This would also enable an analysis of how the average scaling size varies when the air void system changes. When considering different additives it would be interesting to compare the difference in average scale size between different additives. Additional salt frost scaling tests should be made with AEA and SP combinations which do not seem to perform well according to the first mortar test results. This would acquire more information about the effectiveness and reliability of the mortar test used. It would also give a better comprehension about the difference in scaling for the salt frost scaling method which increases the mass of scaling damage and enhance knowledge about the salt frost scaling mechanism(s). The results from these complementary tests could also give information about the robustness of a certain binder to various admixtures combinations. Since the salt frost scaling test method which has been used differs from the standard methods the following should be noted. This method is designed to contribute to a large amount of scaling damage from the salt frost scaling mechanism(s) and provide a higher resolution in the results when testing concrete with a water binder ratio of 0.40 (which could concrete considered when building e.g. a bridge). This means that any claim of a limit for the mass of scaling for a concrete to be considered salt frost scaling resistant cannot be made without much more testing preferably with a lot of field tests to verify the lab results. Another option could be a large round robin test where concrete with various salt frost scaling resistances is tested according to the method presented in [2] together with some of the standard methods [5, 6]. This would provide information about how the salt frost scaling method used in this methodology compares to the other methods, considering relative mass of scaling when testing concrete with the same binder/binder combination. The results would also show if all methods give the same indications about how the salt frost scaling damage increase or decrease when testing various factors, e.g. when varying binder combination, binder content, or air content in the same binder.

170


22-23 August 2016, Lunds University, Lund, Sweden 6. Conclusions

The methodology enables a study of the salt frost scaling mechanism(s) in concrete with a high degree of hydration which contributes to an optimal microstructure for any binder regardless of hydration rate. Therefore more constant material conditions during the salt frost scaling test in comparison to tests of young concrete.

This methodology enables an analysis of how differences in the air void system affects the mass of salt frost scaling damage from a relatively constant salt frost scaling mechanism(s) load for a concrete with a given microstructure.

References [1] Sjöbeck H (2015) The time dependency of salt-frost damage at low temperature on

concrete with SCMs. Lund University, (in Swedish), Master of Science thesis, https://lup.lub.lu.se/student-papers/search/publication/7454864

[2] Strand MJ (2016), Salt frost scaling in uncarbonated concrete containing fly ash and slag with various air void contents, Licenciate thesis, Lunds University, TVBM 3181

[3] ASTM (2012) ASTM C457 / C457M - 12, Standard Test Method for Microscopical Determination of Parameters of the Air-Void System in Hardened Concrete.

[4] Jacobsen S, Saether DH, Sellevold EJ (1997) Frost testing of high strength concrete: Frost/salt scaling at different cooling rates Materials and Structures/Materiaux et Constructions 30:33-42

[5] Setzer MJ, Fagerlund G, Janssen DJ (1996) CDF Test - Test method for the freeze-thaw resistance of concrete - tests with sodium chloride solution (CDF) Materials and Structures 29:523-528 doi:10.1007/bf02485951

[6] SS137244 (2005) Concrete Testing – Hardened Concrete – Scaling at freezing, Swedish Standards Institute, (in Swedish) 10 p.

171



MITIGATION OF DEICER DAMAGE IN CONCRETE PAVEMENTS CAUSED BY CALCIUM OXYCHLORIDE FORMATION – USE OF GROUND LIGHTWEIGHT AGGREGATES Prannoy Suraneni (1), Naomi Salgado (1), Hunter Carolan (1), Chang Li (1), Vahid J. Azad (1), O. Burkan Isgor (1), Jason H. Ideker (1), Jason Weiss (1) (1) Oregon State University, Corvallis, USA Abstract Deicing salts have long been used to improve the safety of pavement surfaces by melting ice and snow. These salts are typically chloride based and therefore may contribute to the corrosion of the reinforcing steel in concrete infrastructure elements. Much less known however is their tendency to react chemically with calcium hydroxide in concrete leading to the formation of calcium oxychloride salt that can cause premature deterioration, especially at pavement joints. It is only recently that the magnitude of this problem has been realized. Mitigation strategies are important, as post damage repair tends to be expensive, disruptive to the travelling public, and puts construction crews in harms way. Prior work has shown that supplementary cementitious materials (SCMs) help mitigate the potential extent of expansive salt formation. Unfortunately, however, SCMs can often be in short supply locally. This paper expands on the previous work by evaluating the use of ground lightweight aggregates (GLWAs) to mitigate calcium oxychloride formation. The effect of sample age on calcium oxychloride formation is studied for neat cement pastes and cement pastes with SCMs and GLWAs. Fitted equations are developed which can be used to estimate the potential for expansive salt formation knowing the mixture design and sample age. 1. Introduction Deicing salts are extensively used in cold locations in North America to improve pavement safety by melting ice that forms on the surface of the pavements. Currently, the most commonly used deicers by United States Department of Transportations are chloride-based: NaCl, CaCl2, MgCl2, or combinations thereof [1]. Although formate-based and acetate-based deicing alternatives exist, these are more expensive and therefore, not as frequently used in

Prannoy Suraneni (1), Naomi Salgado (1), Hunter Carolan (1), Chang Li (1), Vahid J. Azad (1), O. Burkan Isgor (1), Jason H. Ideker (1), Jason Weiss (1)

172



highway applications [1]. Although they are effective at melting ice, chloride-based deicers can lead to corrosion of the steel if the concrete is reinforced [2, 3]. Chloride-based deicers can also interact with the concrete physically, causing damage in the concrete [4, 5]. This damage leads to the flaking or scaling of small pieces of concrete and is thus known as salt scaling damage. Additionally, in cases of fluid pooling or high fluid saturation, conventional freeze-thaw damage can occur [6], in spite of the presence of deicers. Corrosion, salt scaling, and freeze-thaw damage have been extensively studied and are not covered here. Instead, we will focus on chemical interactions between chloride-based deicers and the concrete, leading to an expansive reaction product that swells and damages concrete pavements [7]. Chemical interactions between chloride-based deicers and concrete have been studied for over thirty years [8-10]. Unfortunately, comparatively little research has been done on this topic, and mitigation methods have not been understood or employed. Consequently, premature damage that is both costly and expensive to repair [7] has been noted in numerous concrete pavements in the United States, often within the first few years of service [11]. This damage is typically observed in zones of high fluid saturation, such as joints, and is characterized by the formation of damage along the length of joints. The damage is caused by a chemical reaction between chloride-based deicers and calcium hydroxide in the concrete, according to the Equation below [10]: CaCl2 + 3Ca(OH)2 + 12H2O 3Ca(OH)2.CaCl2.12H2O (1) The reaction product is known as calcium oxychloride. It is expansive and swells during a phase change. Calcium oxychloride forms with high concentrations of CaCl2 [12, 13] or MgCl2 [14], but not with NaCl [15]. Since the calcium oxychloride is formed because of a reaction with the calcium hydroxide, an obvious mitigation strategy is the reduction in the amount of calcium hydroxide in the paste. It has been shown that supplementary cementitious materials (SCMs) can be used for this purpose and that addition of volume replacements of 60 % of fly ash or slag almost completely eliminated calcium oxychloride formation [16]. In this study, fly ash and slag replacements are compared with ground lightweight aggregates (GLWAs) in terms of their ability to reduce calcium oxychloride formation. Since lightweight aggregates are typically aluminosilicate materials [17, 18], it is expected that finely ground lightweight aggregates should behave in a pozzolanic manner and reduce the calcium hydroxide amounts in paste over time. Since lightweight aggregates are made from widely available raw materials such as clay, slate, and shale, they may be a useful alternative pozzolanic material, especially when more commonly used pozzolans difficult to obtain locally. There seems to very little research on the possible pozzolanicity of GLWAs [18, 19]. This paper is primarily focused on the mitigation of calcium oxychloride using ground lightweight aggregates. The objectives of this study are to compare SCMs and GLWAs in terms of their ability to reduce calcium hydroxide and calcium oxychloride contents as a function of the sample age and to develop equations to describe calcium hydroxide and calcium oxychloride contents as a function of sample age and SCM/GLWA replacement level.

173



2. Materials and methods

A Type I ordinary portland cement (OPC) was used in this study. According to the Bogue calculation, the cement had a potential phase composition of 52 % C3S, 18 % C2S, 7 % C3A, and 10 % C4AF. A Class F fly ash (FA) and slag (SL) were used as supplementary cementitious materials. A GLWA based on expanded slate was ground to two different fineness levels with mean particle diameters being 8 and 5 manufacturer), these are denoted as GLWA1-8 and GLWA1-5, respectively. A second GLWA, based on expanded shale was also used, denoted as GLWA2. A full powder characterization program to determine the particle size distribution (using laser diffraction), oxide contents (using X-ray fluorescence), and crystalline and amorphous phases (using X-ray diffraction) is currently underway for all the materials listed above. Reagent grade CaCl2 was used; a 20 % mass CaCl2 solution was prepared by dissolving it in deionized water. Pastes with a water-to-cementitious material ratio of 0.36 were prepared using neat OPC and OPC with 20 %, 40 %, and 60 % nominal volume additions of SCMs or GLWAs. The following density values (g/cc) were used: cement 3.15, fly ash 2.6, slag 2.9, GLWAs 1.7; for the GLWAs, these values are assumed, and the actual density values are currently being measured. Subsequent analysis in this study is done using both mass and volume contents. Pastes were prepared by mixing water and cementitious material in a vacuum mixer at 400 rpm for 90 seconds, scraping material off the mixer and paddle, and then mixed again at 400 rpm for 90 seconds. Samples were cast in cylindrical plastic molds and sealed cured under ambient conditions (23 ± 1 °C) till testing (performed at 3, 7, 28, and 49 days). For each mixture, one sample was sealed cured at ambient conditions (23 °C) for three days, and then oven cured at 50 °C for 25 days. This accelerated curing regime is designed to increase the degree of hydration extent in samples. For neat OPC pastes, this is equivalent (in terms of degree of hydration) to about 91 days curing at 23 °C, according to maturity calculations. We make the assumption that this curing is also equivalent to 91 day curing at 23 °C for the SCMs and GLWAs, though it should be noted that SMCs (and presumably GLWAs) have rather complex effects on cement degree of hydration, though these effects are lower at later ages [20]. Prior to testing at the designated age, samples are ground using a lathe grinder and passed through a 75- sieve. Two tests were done on the ground powder. The first test is thermogravimetric analysis (TGA), in which the ground cement paste is heated to 1000 °C under an inert nitrogen atmosphere. The amount of calcium hydroxide in the paste can be estimated based on the weight loss during its decomposition between 380 °C and 460 °C. A sample size of around 30 mg was used and the heating rate was 10 °C/min. Under these conditions, tested samples (of neat cement pastes) had a coefficient of variability (COV) of about 1 %. The second test is low-temperature differential scanning calorimetry (LT-DSC), in which 10 ± 0.5 mg of ground paste was mixed with 10 ± 0.5 mg of 20 % CaCl2 solution (by mass) in a 1 to 1 mass ratio in a LT-DSC pan, which is subsequently sealed, placed in the LT-DSC and run through the following temperature cycle: isothermal at 25 °C for one hour; 3 °C/min cooling till – 90 °C; low temperature loop from – 90 °C to – 70 °C and back to – 90 °C at 3 C/min; 0.25 °C/min heating till 50 °C As the mixture is heated, calcium oxychloride melts (typically between 25 °C and 40 °C, under these conditions). The amount of calcium oxychloride that forms is

174



found by normalizing the heat release on melting with the heat release of melting of a pure calcium oxychloride system. Under these conditions, tested samples (of neat cement pastes) had a coefficient of variability (COV) of about 5 %. Additional details are presented elsewhere [16], we only note here that these conditions ensure all the calcium hydroxide in the paste reacts and the formation of the maximum amount of calcium oxychloride that can be formed in the system.

3. Results and discussion

Figure 1 shows calcium hydroxide contents for all mixtures (OPC, OPC with 20, 40, and 60 % of FA, SL, GLWA1-8, GLWA-2, and GLWA2) for mixtures cured at an equivalent of 91 days (23 °C for 3 days and 50 °C for 25 days). These values are presented as g calcium hydroxide/100 g paste. From the figure, it is clear that different SCMs and GLWAs decrease the calcium hydroxide contents in different manners. If one assumes that the SCMs and GLWAs are acting only by dilution, calcium hydroxide values should be between 11.3 – 12.3 g/100 g paste, 8.66 – 10.27 g/100 g paste, and 5.9 – 7.7 g/100 g paste for 20, 40, and 60 % volume replacements, respectively. For the SCMs and GLWAs, the corresponding values are between 7.4 – 10.3 g/100 g paste, 4.1 – 8.0 g/100 paste, and 2.7 – 4.8 g/100 g paste, clearly showing the pozzolanicity of all tested materials. In a similar manner, Figure 2 shows calcium oxychloride contents for all mixtures cured at 91 days, presented as g calcium oxychloride/100 g paste. Calcium oxychloride and calcium hydroxide results follow broadly the same trend. If one assumes that the SCMs and GLWAs are acting only by dilution, calcium oxychloride values should be between 26.8 – 29.0 g/100 g paste, 20.4 – 24.2 g/100 g paste, and 13.8 – 18.2 g/100 g paste for 20, 40, and 60 % volume replacements, respectively. For the SCMs and GLWAs, the corresponding values are between 14.0 – 23.5 g/100 g paste, 6.8 – 15.8 g/100 paste, and 2.8 – 6.3 g/100 g paste, clearly showing the pozzolanicity of all tested materials, and that SCMs and GLWAs can be effective in reducing the calcium oxychloride induced pavement damage. The pozzolanicity of these materials can be better visualized in Figure 3, which shows the calcium hydroxide contents for the different mixtures plotted against the mass replacement level of SCM/GLWA (it should be noted that all SCM and GLWA have same volume replacement but not same mass replacements due to differing densities). The decrease in calcium hydroxide amounts in the case of pure dilution is also plotted, and it is clear that all the materials fall significantly below this line (and are thus pozzolanic). The data are reasonably well fitted to a line, enabling the prediction of calcium hydroxide contents given only the mass replacement of SCM or GLWA. A very similar analysis can be done for calcium oxychloride, as shown in Figure 4, and the results are similar: these materials are pozzolanic, reduce the calcium oxychloride contents more than they should by pure dilution, and the data are reasonably well-fit to a line, enabling the prediction of calcium oxychloride contents given only the mass replacement of SCM/GLWA.

175



Figure 1. 91-day equivalent calcium hydroxide values for all mixtures tested (0, 20, 40, 60 % volume replacements). Calcium hydroxide values clearly decrease as the amount of SCM or GLWA increases.

Figure 2. 91-day equivalent calcium oxychloride values for all mixtures tested (0, 20, 40, 60 % volume replacements). Calcium oxychloride values clearly decrease as the amount of SCM or GLWA increases.

176



Figure 3. 91-day equivalent calcium hydroxide values for all mixtures tested plotted against SCM or GLWA mass replacement level. Calcium hydroxide values clearly decrease as the amount of SCM or GLWA increases. The line of best fit for all mixtures and a line for dilution (dashed) are also shown.

Figure 4. 91-day equivalent calcium oxychloride values for all mixtures tested plotted against SCM or GLWA mass replacement level. Calcium oxychloride values clearly decrease as the amount of SCM or GLWA increases. The line of best fit for all mixtures and a line for dilution (dashed) are also shown.

177



The time evolution of calcium hydroxide contents is plotted in Figure 5. Individual data for SCM or GLWA are not plotted on this chart, instead, the average calcium hydroxide value for each volume replacement level and its standard deviation are plotted against time for FA, GLWA1-8, GLWA-5, and GLWA2. The time evolutions are fit using a logarithmic curve for the neat OPC and a linear fit is used for the SCMs and GLWAs. Although these are simplifying assumptions, it is noted that the data is fit reasonably well and the standard deviations are not particularly large (typically less than 1 g/100 g paste for calcium hydroxide), and thus, these fits can be used as a numerical model to estimate the time evolution of calcium hydroxide contents with SCMs and GLWAs (though further work must be done to establish that this behavior is a general one and not specific to the cements, SCMs and GLWAs tested). An interesting point is that calcium hydroxide contents with SCMs and GLWAs do not show a large decrease, this is likely due to the system running out of water, due to the low w/c used which can limit hydration. Another interesting observation is that in some cases, the calcium hydroxide contents increase at 7 days, most likely due to a combination of the filler effect and slow reaction of the SCMs and GLWAs. A similar analysis is shown for calcium oxychloride in Figure 6. As with calcium hydroxide, data is fit reasonably well and the standard deviations are not particularly large, and thus, these fits can be used as a numerical model to estimate the time evolution of calcium oxychloride contents with SCMs and GLWAs. Results with calcium hydroxide and calcium oxychloride are broadly similar. Since calcium oxychloride forms from calcium hydroxide, it is reasonable to assume that they follow similar trends [21, 22], though in some cases, such as carbonation, results may be different [23].

Figure 5. Time evolution of average calcium hydroxide contents for OPC and SCM or GLWA replacements. OPC data is fit logarithmically and SCM and GLWA data is fit linearly.

178



Figure 6. Time evolution of average calcium hydroxide contents for OPC and SCM or GLWA replacements. OPC data is fit logarithmically and SCM and GLWA data is fit linearly. 4. Conclusions

The following conclusions can be drawn from this study: 1. Ground lightweight aggregates are pozzolanic and both ground lightweight aggregates and supplementary cementitious materials can be used to reduce the calcium hydroxide and calcium oxychloride contents (and therefore damage due to calcium oxychloride formation). 2. Simple equations can be developed to fit the calcium hydroxide and calcium oxychloride contents for neat cement and cement with ground lightweight aggregates or supplementary cementitious materials. Acknowledgements

The authors thank Expanded Shale, Clay, and Slate Lightweight Aggregate Industry (ESCSI) for providing support that helped perform some of the work reported here.

179



References [1] Shi, X., et al, Evaluation of alternative anti-icing and deicing compounds using sodium

chloride and magnesium chloride as baseline deicers - phase I, Report CDOT-2009-1 (2009)

[2] Vu, K. A. T. and Stewart, M. G., Structural reliability of concrete bridges including improved chloride-induced corrosion models, Struct Saf 22 (2000), 313-333

[3] Shi, X., et al, Corrosion of deicers to metals in transportation infrastructure: introduction and recent developments, Corros Rev 27 (2009), 23-52

[4] Valenza, J. J. II and Scherer, G. W., Mechanism for salt scaling, J Am Ceram Soc 89 (2006), 1161-1179

[5] Valenza, J. J. II and Scherer, G. W., Mechanisms of salt scaling, Mater Struct 38 (2005), 479-488

[6] Li, W., et al, Water absorption and critical degree of saturation relating to freeze-thaw damage in concrete pavement joints, J Mater Civ Eng 24 (2012), 299-307

[7] Suraneni, P., et al, Deicing salts and durability of concrete pavements and joints: Mitigating calcium oxychloride formation, Concr Int 38 (2016), 48-54

[8] Chatterji, S., Mechanism of the CaCl2 attack on portland cement concrete, Cem Concr Res 8 (1978), 461-467

[9] Berntsson, L. and Chandra, S., Damage of concrete sleepers by calcium chloride, Cem Concr Res 12 (1982) 87-92

[10] Collepardi, M., et al, Durability of concrete structures exposed to CaCl2 based deicing salts, Proceedings of the 3rd CANMENT/ACI International Conference, France (1994), 107-120.

[11] Jones, W., et al, An overview of joint deterioration in concrete pavement: Mechanisms, solution properties, and sealers, Purdue University Report (2013) doi: 10.5703/1288284315339

[12] Galan, I., et al, Impact of chloride-rich environments on cement paste mineralogy, Cem Concr Res 68 (2015), 174-183

[13] Farnam, Y., et al, The influence of calcium chloride deicing salt on phase changes and damage development in cementitious materials, Cem Concr Compos 64 (2015), 1-15

[14] Farnam, Y., et al, Damage development in cementitious materials exposed to magnesium chloride deicing salt, Constr Build Mater 93 (2015), 384-392

[15] Farnam, Y., et al, Acoustic emission and low-temperature calorimetry study of freeze and thaw behavior in cementitious materials exposed to sodium chloride salt, Transport Res Rec 2441 (2014), 81-90

[16] Monical, J., et al, Reducing joint damage in concrete pavements: Quantifying calcium oxychloride formation for concrete made using portland cement, portland limestone cement, supplementary cementitious materials, and sealers, Transport Res Rec (2016, in press)

[17] Lo, Y., et al, Microstructure of pre-wetted aggregate on lightweight concrete, Build Environ 34 (1999), 759-764

[18] Erdem, T. K., et al, Use of perlite as a pozzolanic addition in producing blended cements, Cem Concr Compos 29 (2007), 13-21

[19] Mouli, M., and Khefali, H., Performance characteristics of lightweight aggregate concrete containing natural pozzolan, Build Environ 43 (2008), 31-36

180



[20] Feng, X., et al, Estimation of the degree of hydration of blended cement pastes by a scanning electron microscope point-counting procedure, Cem Concr Res 34 (2004), 1787-1793

[21] Suraneni, P., et al, Calcium oxychloride formation in pastes containing supplementary cementitious materials: Thoughts on the role of cement and supplementary cementitious materials reactivity, RILEM Tech Lett 1 (2016), 24-30

[22] Suraneni, P., et al, Calcium oxychloride formation potential in cementitious pastes exposed to blends of deicing salt, ACI Mater J (2016, submitted)

[23] Ghantous, R.M., et al, The influence of carbonation on the formation of calcium oxychloride, Cem Concr Compos (2016, submitted)

181



DEICER-SALT SCALING OF CONCRETE CONTAINING FLY ASH Michael Thomas and Huang Yi (1) (1) University of New Brunswick, Fredericton, NB, Canada Abstract This paper presents results and observations from the laboratory, field-exposure sites and real pavements on the performance of concrete containing fly ash exposed to freezing and thawing in the presence of deicing salts. Results from accelerated laboratory tests indicate variable performance of concrete containing fly ash especially at higher replacement levels. However, fly ash concrete generally performs well in the field even with frequent applications of salt and at relatively high replacement levels. Scaling of fly ash concrete in the field is largely restricted to hand-finished flatwork (e.g. sidewalks and driveways) as opposed to formed (e.g. barrier walls or bridge columns) or slipformed (e.g. pavements, curb and gutter) surfaces and is possibly the result of improper placing, finishing or curing practices. It is shown that concrete containing high levels of fly ash (and slag) maybe be more sensitive to poor practices. 1. Introduction Numerous research studies [1-10] have indicated that supplementary cementing materials (SCM) such as fly ash and slag, decrease the salt scaling resistance of blended-cement concrete as compared with straight Portland cement concrete tested in the laboratory. There are many potential explanations for the inferior scaling resistance of SCM concrete: 1) reduced early-age strength due to comparatively slow reaction of SCM; 2) increased porosity of SCM concrete due to carbonation; 3) weakened surface strength due to excessive bleeding; 4) affected air-void systems due to SCM and others. However, there does not appear to be a consensus on which cause is the most directly responsible for the reduced resistance to salt scaling. Furthermore, it has been demonstrated [10-14] that the inferior performance of SCM concrete may be limited to accelerated-laboratory testing and that SCM concrete that is properly proportioned, placed, finished and cured can be expected to have satisfactory scaling resistance even at relatively high levels of placement.

182



This paper presents data from laboratory and field studies in an attempt to determine the discrepancy between the behaviour of SCM concrete in accelerated tests versus field performance. The sensitivity of concrete containing fly ash (and slag) to poor practices is also evaluated. 2. Field Performance of Fly Ash Concrete In 1997, Thomas [11] published the results of a survey of concrete structures (pavements, sidewalks, bridge decks and barrier walls) containing various levels of Class F (low-calcium) and Class C (high-calcium) fly ash that had been exposed to freezing and thawing in the presence of deicing salts for different periods of time. Although some of the structures were relatively young at the time of inspection damage due to scaling is usually evident after the first winter. Figure 1 shows the results of this survey and has been modified to include a number of new observations that have been made since 1997; these are:

“HVFA concrete” sidewalks placed in Halifax, Nova Scotia, in 1994 [16]. The fly ash concrete had a total binder content of 400 kg/m3, a fly ash content of 55% and water content of 110 kg/m3, resulting in w/cm= 0.275. Figure 2 shows the visual appearance of the control and HVFA concrete mix in the summer of 2006, after 12 winters; it is compared with the a mix without fly ash that was placed adjacent to the HVFA at the same time. The concrete has scaled heavily (significantly more so than the control) but is still serviceable. It should be noted that this concrete receives frequent applications of deicer salt and is exposed to more than 100 cycles of freeze-thaw per year [17].

Concrete containing PLC and SCM [18]. Three pavement field trials (in Quebec, Nova Scotia and Alberta) were conducted in 2008 and 2009 using portland limestone cement (PLC) containing approximately 12% interground limestone. Concrete mixes contained anywhere between 15 to 50% fly ash. After 6 to 7 winters all of the concrete appears to be performing satisfactorily.

Fly ash concrete parking lot placed in Wisconsin (see Figure 3). The parking area was constructed by placing alternate strips of concrete using a slipform paving machine one day and then filling in the strips between the slipformed concrete sections one week later using concrete consolidated and levelled by vibrating screed. Figure 3 shows that the slipformed concrete has performed very well with no signs of scaling whereas the “in-fill” concrete has severe scaling damage after just 5 winters. According to the specifications all of the concrete was to be slipformed and the mixture proportions were prescribed to be 183 kg/m3 Portland cement, 225 kg/m3 (55%) Class F fly ash, and 116 kg/m3 water (requiring a heavy dose of superplasticizer). The slump was specified “as required for slip form paving but not in excess of 3 inches” (75 mm). The minimum compressive strength was 28 MPa at 28 days. However, it is suspected that the “in-fill” concrete was placed at a higher w/cm to achieve the slump required to place the material using a vibrating screed (typically > 100 mm slump). Further evidence for this is found in strength results for cylinders tested at 28 days which, although more than adequate for code requirements for this

183



type of exposure (30 MPa minimum in Canada for this class of exposure), are significantly lower in the concrete placed by vibrating screed (36 MPa) than by slipforming (43 MPa).

Figure 1 Performance of Fly Ash Concrete Structures Exposed to Deicing Salts

Collectively the observations from the field studies indicate that fly ash concrete can be produced to have a high resistance to deicer salt scaling even when the replacement level is 50% or more. However, the relatively poor performance of the fly ash concrete in Halifax and Wisconsin, both of which contain 55%, compared with slipformed concrete with the same

184



level of fly ash in the parking lot and even higher levels in other pavements does raise the question of the role of the placement method. Slipformed concrete is placed at a very low slump and is usually characterized by having a low water content; furthermore very little finishing is required after the concrete is extruded from the machine (typically bullfloating and sometimes texturing or tining). On the other hand, concrete consolidated and finished by hand using either an immersion vibrator or vibrating screed will have a significantly higher slump and, consequently, water content (and w/cm). Furthermore, use of a vibrating screed with high slump concrete can lead to increased w/cm at the surface and, potentially, reduced air content. Given these considerations, the superior performance of the slipformed concrete is not surprising. Having said this, hand-finished sidewalks and driveways produced with Portland cement do not generally result in scaling problems. Is it possible that concrete produced with high levels of fly ash and slag are less robust and more sensitive to placing, finishing and curing practices? This question was explored in a laboratory study discussed in Section 4 below.

Figure 2 Appearance of Concrete Sidewalks with 55% Fly Ash (Left) and without Fly Ash (Right) after 12 years Exposure in Halifax, NS

3. Laboratory versus Field Performance In five of the field placements shown in Figure 1 small scale slabs were cast in the field for laboratory scaling tests. This permits a direct comparison of the field and laboratory performance. Full details of the concrete mixes used, methods of construction, field exposure conditions, and laboratory test data have been presented elsewhere [6, 19-22]. In four of the five cases, control placements with plain Portland cement concrete are also available for comparison. Brief details of the placements, mixes used and the results from laboratory scaling tests have been presented by Thomas [11]. In all cases, the fly ash concrete showed inferior performance in laboratory tests compared with the control concrete. Some of the fly ash concretes failed to meet the commonly-used acceptance limit of a maximum 0.80 to 1.0 kg/m2 mass loss after 50 cycles [6]. An example is given in Figure 4 which shows the excellent condition of a slipformed pavement after six winters; slabs from the same mix tested in the laboratory gave a visual rating of 4 [20]; a rating of 5 is the most severe damage. The

185



poor performance of the fly ash concrete in the accelerated laboratory tests is not consistent with the field performance of the same concrete mixes. All of the fly ash concretes have given satisfactory performance in the field and are generally in good to excellent condition. Furthermore there is little consistent difference in the surface appearance of the fly ash and control concretes at each location.

Figure 3 Construction and Visual Appearance (after 5 Winters) of Parking Lot in Wisconsin – Concrete Placed by Slipform Paving Machine (Left) and Using Vibrating Screed (Right)

186



Figure 4 Slipformed Concrete Pavement with 50% Class C Fly Ash after Exposure to Deicing Salts over Six Winters

4. Laboratory Study on “Robustness” of SCM Concrete The impact of poor construction practices on the scaling resistance of concrete with and without SCM is illustrated in Figure 5, which shows the visual rating after 50 cycles of a salt scaling test (similar to ASTM C 672 but with 3% NaCl instead of 4% CaCl2) of concrete slabs from three different concrete mixes. The mixes all started with an initial W/CM of 0.42 and a slump of approximately 25 mm; one mix contained 30% Class F fly ash, the second 50% slag and the third was a control mix without SCM. Ten slabs were produced from each mix using the following placing, finishing and curing procedures:

1. Slab cast with low-slump concrete, heavy (steel) troweling and water spray used to finish the surface; a curing compound was then applied.

2. Portion of the concrete transferred to a smaller mixer and a high-range water-reducing admixture added to the mix to produce a 150-mm (6-inch) slump (final W/CM = 0.42), surface struck with wood board and textured with a broom and a curing compound applied.

3. Water added to the mixer to provide a 150-mm (6 inch) slump (final W/CM = 0.50), concrete finished with steel trowel and a curing compound applied.

4. Placement delayed by 3 hours, water periodically added to the mixer to maintain 150-mm (6-inch) slump (final W/CM = 0.60), concrete finished with steel trowel, and a curing compound applied.

5. As for previous (# 4) but no curing compound was used.

The results in Figure 5 show that both the fly ash and slag concrete showed good scaling resistance provided procedure 2 was followed. However, when improper practices, such as

187



the addition of water, over finishing with a steel trowel, extended delays or inadequate curing, were used the scaling resistance was reduced significantly. The control concrete with no SCM was more robust in that it was less sensitive to improper practices.

Figure 5 Scaling of Slabs Produced with SCM Concrete using Different

Mixing, Placing and Finishing Procedures

5. Effect of Strength on Scaling of SCM Concrete Figure 6 shows data from an ongoing comprehensive study on the scaling resistance of SCM concrete. Concrete slabs were produced with w/cm = 0.45 and with either no SCM (M1), 50% Class F fly ash (M2) or 60% slag (M3). The standard scaling test procedure was to wet-cure for 14 days and then store in laboratory air for 14 days prior to ponding the slab with salt solution and initiating freeze-thaw cycles. Some slabs were cured for extended periods of time (28, 91 and 182 days) before the 14-day air-storage period. Figure 6 shows the compressive strength of cylinders at the end of the wet-cure period versus the mass loss after 50 freeze-thaw cycles. It is considered that the strength at the end of the wet-curing period is representative of the surface strength of the slabs as the surface will dry rapidly in laboratory air halting hydration/reaction on the cementitious materials in the surface layer almost immediately. The data in Figure 6 indicate that the poor performance of the concrete with high levels of fly ash and slag in the standard test is due to the low strength of the surface after the 14-day curing period. The low strength is probably a contributor to the apparent disconnect between performance in the laboratory versus the field. Unless concrete is placed very late in the Fall (Autumn), it will likely have the opportunity to mature and gain significant strength prior to the first exposure to freezing and deicing salts.

188



Figure 6 Effect of Compressive Strength (f’c) after Wet-Curing on Salt Scaling of Concrete Containing SCM (Mass Loss after 50 Freeze-Thaw Cycles)

6. Discussion It is evident from the field survey of structures that fly ash concrete can be highly resistant to scaling when exposed to freezing and thawing in the presence of deicing salts even at relatively high levels of replacement. It would appear that slipformed concrete (pavements, curb and gutter) is more resistant than “handfinished flatwork” (sidewalks and driveways) and this is probably a result of differences in the mixture proportions used and finishing practices for these types of concrete. Indeed, in the authors’ experience salt scaling problems in the field are rarely encountered in properly proportioned slipformed or formed concrete surfaces and are largely confined to hand-finished flatwork. Such concrete, which includes driveways and sidewalks, may be particularly prone to scaling if improper placing, finishing and curing practices are used. Using a vibrating screed on high slump concrete that is prone to bleeding will tend to draw water to the surface creating a weak layer with, perhaps, an inferior air-void system. Finishing prior to the end of bleeding and/or providing no curing (e.g. curing membrane) may also create a weak surface layer. It is the authors’ contention that concrete containing relatively high levels of SCM replacement may be less robust than straight portland-cement concrete and be more sensitive to poor placing, finishing and curing practices.

0

1

2

3

4

5

0 20 40 60 80

Mas

s Los

s (kg

/m2 )

f'c (MPa)

M1-14 M1-28 M1-91M2-14 M2-28 M2-91 M2-182M3-14 M3-28 M3-91 M3-182

(suffix indicates wet-curing period in days)

0% SCM 50% Fly Ash 60% Slag

189



Many specifications limit the amount of SCM that can be used in concrete exposed to deicing salts and cyclic freezing and thawing. For example, the ACI 318 Building Code limits the amount of SCM in such environments to 25% fly ash and natural pozzolans, 50% slag, 10% silica fume; the total for all pozzolans (fly ash + silica fume + natural pozzolan) is limited to 35% and the total for all SCM (pozzolans + slag) is limited to 50%. While some limit seems prudent for certain concrete, such as “hand-finished flatwork” like driveways and sidewalks, it is the authors’ opinion that such limits are not appropriate for all concrete as they limit the beneficial effect of increased levels of SCM in terms of other durability properties (e.g. chloride resistance and prevention of alkali-silica reaction). With regards to accelerated laboratory testing, it is suggested that the period of wet-curing be extended beyond 14 days unless it is intended to determine the performance of the concrete when it is to be placed shortly before the first salt application and freezing event.

References 1. Bilodeau, A. and Carette, G., 1983. “Resistance of Condensed Silica Fume Concrete to

the Combined Action of Freezing and Thawing Cycling and Deicing Salts.” ACI SP 114-47, American Concrete Institute, Detroit, pp. 945-970.

2. Sorensen, E., 1983. “Freezing and Thawing Resistance of Condensed Silica Fume (Microsilica) Concrete Exposed to Deicing Chemicals.” ACI SP 79, American Concrete Institute, Detroit, pp. 709-718.

3. Gebler, S. and Klieger, P., 1986. “Effect Fly Ash on the Durability of Air-Entrained Concrete.” ACI SP 91-23, American Concrete Institute, Detroit, 1986, pp. 483-519.

4. Johnston, C. 1987. “Effects of Microsilica and Class C Fly Ash on Resistance of Concrete to Rapid Freezing and Thawing, and Scaling in the Presence of Deicing Agents.” ACI SP-100, American Concrete Institute, Detroit, pp.1183-1204.

5. Whiting, D. 1989. “Deicer Scaling Resistance of Lean Concretes Containing Fly Ash.” ACI SP 114-16, American Concrete Institute, Detroit, pp. 349-372.

6. Afrani, I. and Rogers, C. "The effects of different cementing materials and curing on concrete scaling." Cement, Concrete, and Aggregates, Vol. 16 (2), 1994, pp. 132-139.

7. Zhang, M., Bouzoubaa, N. and Malhotra, V., 1999. “Resistance of Silica Fume Concrete to Deicing Salt Scaling –A Review.” ACI SP172-5, American Concrete Institute, Detroit, pp.67-102.

8. Talbot, C., Pigeon, M. and Marchand, J., 2000. “Influence of Fly Ash and Slag on Deicer Salt Scaling Resistance of Concrete.” ACI SP192-39, American Concrete Institute, Detroit, pp. 645-657.

9. Marchand, J., Jolin, M. and Machabee, Y., 2005. “Deicer salt scaling resistance of supplementary cementing material concrete: laboratory results against field performance.” Proceedings of the international Conference on cement combinations for durable concrete. pp. 579-590.

10. Bouzoubaâ, N., et al., 2008. “Deicing salt scaling resistance of concrete incorporating supplementary cementing materials: Laboratory and field test data.” Canadian Journal of Civil Engineering, 35(11):1261-1275.

190



11. Thomas, M., 1997. “Laboratory and Field Studies of Salt Scaling in Fly Ash Concrete.” Frost Resistance of Concrete, (Ed. Setzer, M.J. and Auberg, R.), E&FN Spon, London, UK, pp. 21-30.

12. Bleszynski, R., Hooton, D., Thomas, M., and Rogers, C., 2002. “Durability of Ternary Blend Concrete with Silica Fume and Blast-Furnace Slag: Laboratory and Outdoor Exposure Site Studies.” ACI Materials Journal, 99(5): 499-508.

13. Bouzoubaâ, N., Bilodeau, A., and Foumier, B., 2004. “R&D consortium on deicing salt scaling resistance incorporating supplementary cementing materials.” Natural Resources Canada, Ottawa, Ont. MTL 2004-15 (TR-R).

14. Bouzoubaâ, N., et al., 2011. “Deicing salt scaling resistance of concrete incorporating fly ash and (or) silica fume: laboratory and field sidewalk test data.” Canadian Journal of Civil Engineering, 38(4): 373-382.

15. Boyd, A. and Hooton, D., 2007. “Long-term scaling performance of concretes containing supplementary cementing materials.” Journal of Materials in Civil Engineering, 19(10): 820-825.

16. Langley, W.S. and Leaman, G.H. 1998. “Practical Uses of High-Volume Fly Ash Concrete Utilizing a Low Calcium Fly Ash.” Proceedings of the Sixth CANMET/ACI/JCI International Conference on Fly Ash, Silica Fume, Slag and Natural Pozzolans in Concrete, ACI SP-178, Vol. 1, American Concrete Institute, Detroit, MI, pp. 545 - 574.

17. Malhotra, V.M. and Mehta, P.K., High-Performance, High-Volume Fly Ash Concrete. Supplementary Cementing Materials for Sustainable Development Inc., Ottawa, Canada, 2005, 124p.

18. Hossack, A., Thomas, M.D.A., Barcelo, L., Blair, B. and Delagrave, A. 2014. “Performance of Portland Limestone Cement Concrete Pavements.” Concrete International, Vol. 36 (1), pp. 40-45.

19. Gifford, P.M., Langan, B.W., Day, R.L., Joshi, R.C. and Ward, M.A. 1987. "A study of fly ash concrete in curb and gutter construction under various laboratory and field curing regimes." Canadian Journal of Civil Engineering, Vol. 14, 1987, pp. 614-620.

20. Naik, T.R, Ramme, B W. and Tews, J.H. 1995. “Pavement construction with high volume Class C and Class F fly ash concrete." ACI Materials Journal, Vol. 92 (2), pp. 200-210.

21. Bilodeau, A., Sivasundaram, V., Painter, KE. and Malhotra, V.M. "Durability of concrete incorporating high volumes of fly ash from sources in the U.S." ACI Materials Journal, Vol. 91 (I), 1994, pp. 3-12.

22. Chojnacki, B. and Northwood, RP. 1988. “Fly ash concrete - laboratory and field trials in Ontario." MTO Report EM-86, Engineering Materials Office, Ministry of Transportation, Downsview, Ontario.

191

International RILEM Conference on Materials, Systems and Structures in Civil Engineering Conference segment on Frost action in Concrete


LINKING SURFACTANT MOLECULAR STRUCTURE TO MORTAR FROST PROTECTION

Lori E. Tunstall (1), George W. Scherer (1), Robert K. Prud’homme (1) (1) Princeton University, Princeton, New Jersey, USA Abstract Using surface tension to investigate two commercial surfactants, a saponified tall oil and a saponified rosin, we find that the tall oil interacts more strongly with calcium ions in solution than does the rosin. When mortars are made using these surfactants, we show that the pore size in the air void shells decreases with an increase in surfactant until it reaches a pore size (< 2.3 nm) where the ice in the voids cannot escape. We also demonstrate that the tall oil surfactant is able to reach this minimum pore size at about half the molar concentration of the rosin surfactant, which supports what has been suggested by others, that surfactant interaction with calcium ions is vital to the frost protection of concrete. 1. Introduction

It is well established that the control and distribution of air voids in concrete is essential for protection against frost damage [1]. As ice grows in the mesopores (Figure 1.2A), it generates crystallization pressure as the ice growth is restrained by the pore walls. It also generates hydraulic pressure, as pore water is displaced by the ice volume expansion (Figure 1.1). The air voids offset these two effects, by providing closely spaced locations into which water can flow and then crystallize. As the macroscopic ice grows in the air void and consumes the pore water (Figure 1.2B), it generates suction in the pores that causes a net contraction of the concrete body.

192



Figure 1. Expansion and Compression of Mortar Generated by Ice Growth: 1.1 Ice growth in the mesopores displaces pore liquid, which flows into the air void; 1.2 Ice in the mesopores (A) results in local crystallization pressure (expansion of the mortar) but when the liquid in the air void freezes, it generates suction (B) that offsets the crystallization pressure (net contraction of the mortar); 1.3 If the ice in the air void escapes, the suction is lost and only crystallization pressure (expansion of the mortar) is observed.

The mortar will remain in compression so long as the ice cannot escape from the void, which will happen once the ice can overcome its surface energy barrier to growth. In other words, once the ice can achieve a small enough radius to penetrate a pore of the air void shell (Figure 1.3), the ice will be able to grow throughout the pore network, until it encounters an even smaller radius. The radius that the ice can sustain is a function of the temperature as given by the Gibbs-Thomson equation [2]. Ice can enter a pore with radius rp at a temperature T given approximately by [3]

(1)

where is the thickness of the unfrozen layer of water on the solid surface, which is about T is the freezing point depression due to pore size. Therefore, if the pores in the

shell around the air void had radii of 7.2 nm, ice could escape from the void when the temperature fell below about -10 ºC. In this paper, we discuss the influence of the chemistry of the air-entraining agent (AEA) on the microstructure of the shell around the air void, and on the frost resistance of the mortar. Tensiometry is used to quantify the interaction of the AEA with calcium ions, which is expected to favor formation of a shell [4]. Dilatometry reveals the temperature at which ice escapes from the air voids, so that the pore size in the shells can be inferred, and indicates whether freezing results in damage.

193



2. Methodology 2.1 Materials DI water used for cement mortar preparation and surface tension experiments was always filtered using a HYDRO PicoTech 2 ultrapure water filtration system to avoid potential contamination. Data for two types of surfactants are presented here, a saponified rosin and a saponified tall oil. Although the exact concentration of surfactant in the commercial AEA is unknown, thermogravimetric analysis (TGA) was used to determine the weight percent of solids and total organic carbon (TOC) analysis was used to determine the carbon’s contribution to this weight. In TGA, each AEA was first heated to 80 °C at 5 °/min and held for one hour. Samples were then heated to 120 °C at 1 °/min. until the weight stabilized. Fourier transform infrared spectroscopy confirmed that the TGA weight loss was due to the evaporation of water. These results are summarized in Table 1. Carbon accounted for the bulk of the solids weight of the rosin surfactant, but only about half of the solids weight of the tall oil surfactant. In a previous study [5], NMR confirmed that the commercial air-entraining agent (AEA) was nearly all active ingredient (provided by Sika), so the difference in TOC and TGA analysis indicates that the evaporation process at 120 ºC was incomplete.

Table 1. Summary of AEA Details

Description Recommended

Dosage (mL / 100 kg)a

Weight of Solidsb (g / L)

Total Organic Carbon (g / L)

Sika Air Saponified rosin

32 – 195 60 51

Sika AEA-15 Saponified tall oil

16 – 65 160 97

a 100 kg cement b

Using the total organic carbon, the molecular weight of the surfactants were calculated using the empirical relationship

(2)

where CMC is the critical micelle concentration, A and B are constants for a specific type of linear surfactant, and N is the number of carbon atoms in the hydrophobic tail [6]. Since the tall oil is linear and the rosin is only slightly branched, this should serve as a fair approximation. Since the CMC is in terms of moles of surfactant per L of solution, if the weight concentration of surfactant in solution at the CMC is known, then the CMC can be written in terms of N and eq. (2) can be used to solve for N. For instance, the tall oil CMC occurs at 0.786 g / L (grams of organic carbon per liter of AEA, which is appromximately the surfactant weight) so we write the CMC in eq. (2) as (0.786 g / L) / MWtall oil where MWtall oil

194



is the molecular weight of the tall oil. Although the molecular weight is unknown, its general structure can be written as

(3)

Usi

eq. (3) for N and solve eq. (3) for the molecular weight of the tall oil surfactant. Using this method, the molecular weights of the tall oil and rosin surfactants were estimated as 274 and 242 g / mol, respectively. 2.2 Dynamic Mechanical Analysis Dynamic mechanical analysis (DMA) is used to measure the length change of mortars as they expand and contract when subject to freezing temperatures. The mortar samples are cylinders that are ~9.8 mm wide and range from 10 to 15 mm in height. Immediately before analysis, the saturated samples are taken from their limewater bath, rinsed with DI water, and surface dried with a kimwipe. The mortar is then wrapped in tape that has been lightly coated with vacuum grease and sprinkled with metaldehyde, which encourages ice nucleation [7]. Before use, the metaldehyde is ground with a mortar and pestle to promote fracture along the cleavage plane, since this is the plane that is similar to the crystal structure of ice [7]. Once the mortar is wrapped, the tape is loosely secured by a piece of wire. The exposed ends of the sample are briefly left to dry to reduce the risk of ice extrusion at the caps, which could be mistaken for sample expansion. The sample is dried at least long enough for the mortar color to lighten. No variation was seen for samples that had been dried for seconds or for several minutes, so the drying time was not controlled.

During data collection, the sample is immersed in a cup of kerosene to prevent drying and to help with heat transfer. The low temperatures are achieved through a bath of ethanol and dry ice, which can reach a minimum temperature of about - A static force of 110 mN is applied to the sample and no dynamic force is used. Each sample is evaluated with the following temperature program: Cool from 20 to 5 isothermally hold for 30 min (or until a stable probe reading is obtained), cool from 5 to -from -50 to - -

Conventional tests, such as ASTM 666, use a minimum temperature of - thaw cycles to evaluate durability. In the present work, we are evaluating the mechanisms of frost protection (as opposed to the durability of mortars), so we use one cycle, but cool to -The isothermal hold at - is included to allow the system to stabilize so kinetic effects could be observed. This was done at -showed that suction was re-established. The DMA samples were taken from larger mortars that ranged in age from about 2 to 6 months old. Data from samples that were tested after 3 months and 6 months had excellent agreement. Each mortar was made with filtered DI water, Ottawa sand conforming to ASTM C778, Type I/II ordinary Portland cement (OPC), and AEA from Sika Corp. Three dosages

195



were used for each type of AEA, taken as the maximum recommended dosage provided by the manufacturer, and two and three times this amount. The mortars were prepared using a Kitchen Aid stand mixer as follows. First, the AEA was added to the DI water and mixed for 30 seconds. Then the pre-mixed dry ingredients were added and mixed at 135 rpm (speed 4). After 1 minute, the mixer was turned off and the mortar was left to rest for 2.5 minutes. During this time the bowl was quickly scraped down and manually recombined so that no dry cement was visible. Mechanical mixing resumed at 280 rpm (speed 10) for 30 seconds. After mixing, the mortar was cast into a pre-moistened cylindrical mold, covered, and left to harden for 24 hours. Immediately after the mortar was removed from the mold, it was placed into a limewater bath to cure. 2.3 Surface Tension Measurements Surface tension measurements were used to examine each AEA’s response to calcium coagulation. Since surface tension decreases with an increase in surfactant concentration, we can measure the amount of surfactant that is coagulated by filtering out the precipitated surfactant and measuring the surface tension of the filtrate. AEA that is coagulated by calcium is removed during the filtration process, so the filtrate has a reduced concentration of surfactant, which is reflected by an increase in surface tension. This is discussed in detail in section 3. Surface tension is measured using the drop weight method, which correlates surface tension with drop size. Using a syringe pump to control the speed of drop ejection, drops were collected, counted, and weighed. The speed of the syringe pump was adjusted until the drop detached by gravity, not by force from the syringe pump. This was usually obvious, since inconsistent drop weights were evident when the drop did not detach by gravity. Once consistent weights were measured, the surface tension was calculated as

(4)

where W is the weight of the drop, r is the outer radius of the syringe capillary tip, V is the volume of the drop, and f(r/V1/3) is a corrective factor for drop shape [8]. The volume of the drop is calculated from the drop weight and density, which was measured using a DMA 35 density meter from Anton Parr USA, sensitive to 1.999 g/cc. The temperature of the room was not controlled, but usually is 20 is about 72.8 ± 0.3 mN/m [9]. This paper discusses data from two solutions, AEA with DI water and AEA with limewater (2 g of calcium hyrdoxide per 1 L of DI water). Both solutions were mixed with a magnetic stirrer at the highest rate not resulting in foaming. Mixing time was established by increasing the time until the solution surface tension was the same as a solution that had been mixed for 4 hours, since that is the maximum time that a cement slurry would remain fluid. This usually occurred at about 1 minute. After mixing, solutions with limewater were filtered using an AcetatePlus, supported, plain 0.22-micron filter from Maine Manufacturing. Occassionally

196



this filter would not yield much filtrate, so alternatively a 1-to-5-micron filter (Q2 quantitative grade from Fisher Scientific) was used. For each solution, the surface tension was plotted against the log of the surfactant concentration. When this is done for a solution of DI water and surfactant, it is called a critical micelle concentration (CMC) curve, because of its distinct shape: a steeply sloped linear region followed by a plateau that begins around the CMC. The CMC is determined as the intersection of the tangents of the two linear regions of the CMC curve. This is the concentration at which the surfactant molecules spontaneously aggregate. 3. Results and Discussion

Several authors have commented on the significance of surfactant interaction with calcium in real cement systems [2,4,10]. To evaluate this we use surface tension measurements to determine the amount of surfactant in solution before and after exposure to limewater (saturated calcium hydroxide solution). Since cement continues to replenish calcium ions in solution as it dissolves during hydration, a saturated calcium hydroxide solution is used so that it too can replenish calcium ions that are precipitated by the surfactant molecules. Since surface tension increases with decrease in surfactant concentration, the amount of surfactant lost to coagulation with calcium ions can be calculated as the difference between the limewater curve and the DI water curve, which serves as a baseline (Figure 2 and Figure 3). Focusing on Figure 2, this would mean that at a starting surfactant concentration of 5 g / L, there is only 0.12 g / L of surfactant left after precipitation with calcium (and removal by filtration). In other words, even thought we start with an initial concentration of 5 g / L, after filtration the limewater curve shows a surface tension of 62.5 mN / m, which means that there is only 0.12 g / L of surfactant in solution, as indicated by the baseline curve. Comparing Figure 2 and Figure 3, the tall oil surfactant apparently interacts more strongly with calcium ions than does the rosin surfactant, since the gap between the limewater curve and the baseline CMC curve is larger for the tall oil. The positions of the curves are also influenced by the ionic strength of the solution, which reduces electrostatic repulsion between the head groups of the AEA. This phenomenon is responsible for the crossing of the curves in Fig. 3, because the higher ionic strength of the limewater more than compensates for the coagulation by calcium. Clearly, the shifts in the curves have to be interpreted with care; however, experiments at higher ionic strength (to be presented elsewhere) confirm that the tall oil interacts more strongly with calcium ions.

197



Figure 2. Tall Oil Surfactant Coagulation with Calcium Ions

Figure 3. Saponified Rosin Surfactant Coagulation with Calcium Ions

198



Mortars were made with the tall oil and saponified rosin AEAs to examine whether the difference in calcium affinity would affect the shell structure of the air void. As discussed previously, in the presence of air voids, a mortar body will show a net contraction due to ice growth in the air voids. The body will remain in compression so long as the ice cannot escape from the voids, which is a function of shell microstructure. So, using DMA, we can compare the microstructures of the air void shells made with different AEAs and dosages by determining the temperature at which the ice escapes from the voids, as evidenced by an expansion in the body. Figure 4 shows the strain of mortar samples made with three different dosages of tall oil AEA (0.65, 1.30, and 1.95 mL per kg of cementitious material) as they are cooled to - . The mortar with the lowest dosage begins expansion at about -35 strain during the cooling cycle, and has the largest residual strain (indicating damage). With an increase in dosage, the temperature at which the contraction stops decreases, as does the maximum strain, and the residual strain. For all the AEAs tested (including ones not shown here), the mortar body begins to expand by - before. This might indicate that the pore size in the shell decreases with an increase in dosage, but has a limiting pore size of about 2.3 nm, according to eq. (1), corresponding to an escape temperature of - More likely, however, is that the pores in the shell are small enough to keep the ice contained inside, and the increase in strain results from homogeneous nucleation of ice in the mesopores. This would result in increased crystallization pressure in the pores, but the strain remains negative because of the suction that is sustained by the ice in the voids. Figure 5 shows the strain response of mortars made with 1.95, 3.90, and 5.85 mL / kg of rosin surfactant. In this case, the lower dosages show almost identical behavior and have an escape temperature of about - At a dosage of 5.85 mL / kg, the temperature of maximum contraction again converges to - Since this seems to be the best performance these entrained mortars can achieve, the AEAs were evaluated by comparing the minimum dosages required to achieve the optimal microstructure (corresponding to an escape temperature -42

). This occured at a dosage of 1.95 for the tall oil surfactant (equal to 6.9 x 10-4 M) and 5.85 mL / kg for the rosin surfactant (equal to 1.24 x 10-3 M), indicating that the tall oil can obtain the same microstructure as the rosin with half the molar concentration of surfactant. This confirms that the tall oil surfactant’s affinity for calcium ions does affect the microstructure of the shell formed aroun the air void, allowing it to form a denser microstructure than surfactants that do not interact as strongly with calcium. This does not mean that the same densification cannot be achieved by the other AEA, but rather that more AEA is required to achieve the same densification. During reheating, the expansion stops at very nearly the same temperature at which it began, because the ice passing through the very small pores of the shell melts and the pore suction is re-established. For example, in Figure 4, the mortar with the lowest dosage begins to expand on cooling below about -35 ºC, because the ice can grow from inside the void through the shell into the surrounding paste at that temperature. As the temperature drops further, the body continues to expand, but when it is reheated above -35 ºC the expansion stops and contraction resumes, because the ice is again trapped inside the void and the pore suction offsets the crystallization pressure. This phenomenon will be analyzed quantitatively in a

199



future publication. At the highest dosages the strain remains negative, because the negative pore pressure is sustained by the ice trapped in the voids.

Figure 4. DMA of Mortar Made with Saponified Tall Oil Surfactant

Figure 5. DMA of Mortar Made with Saponified Rosin Surfactant

200



4. Conclusions This work shows that the air content and distribution are not the only factors that need to be controlled. The microstructure of the air void shells is also an essential consideration. Surface tension experiments show that the tall oil interacts more strongly with calcium ions in solution than does the saponified rosin. This interaction is shown to affect the microstructure of the air void shells that are formed in mortars. At similar molar concentrations, the tall oil is able to form a finer shell microstructure than the rosin surfactant. This is illustrated by using DMA to determine the temperature at which the ice escapes (the mortar expands). Since the ice escape temperature is related to the pore size of the air void shell by the Gibbs-Thomson equation, we can use DMA to probe the difference in shell microstructure for various AEA and dosages. We can use this to determine the appropriate dosage and AEA type for mortars, since a shell with a smaller pore size will provide better frost protection. 5. Acknowledgements This work was funded by NSF grant CMMI-1335320. The information presented in this paper does not necessarily reflect the opinion or policy of the federal government and no official endorsement should be inferred. We would like to thank Sika Corp. for providing the commercial AEAs and their active ingredients.

References

1. Pigeon, M. and Pleau, R. Durability of Concrete in Cold Climates. (E & FN Spon, London, 1995) 2. Scherer, G.W. and Valenza II, J.J. Mechanisms of Frost Damage. pp. 209-246 in Materials Science of Concrete, Vol. VII, eds. J. Skalny and F. Young (American Ceramic Society, 2005) 3. Sun, Z. and Scherer, G.W. Pore size and shape in mortar by thermoporometry. Cement Concr. Res. 40 (2010) 740–751 4. Mielenz, R.C. Volkodoff, V.E. Backstrom, J.E. and Flack, H.L. Origin, Evolution, and Effects of the Air Void System in Concrete - Part 1: Entrained air in unhardened concrete. Proceedings, ACI, 55 (1958) 95-121 5. Tunstall, L.E. and Scherer, G.W. Influence of fly ash on air entrainment, paper C3-1 in Proc. Int. Cong. Durability of Concrete, Trondheim, Norway, June 18-21, 2012, Eds. Harald Justnes, Stefan Jacobsen, ISBN 978-82-8208-031-6 6. Rosen, Milton J., and Kunjappu, Joy T. Surfactants and Interfacial Phenomena, 4th ed. (John Wiley & Sons, Hoboken, 2012) 7. Fukuta, N. Ice nucleation by metaldehyde, Nature 199 (1963) 475-476 8. Harkins W.D., and Brown, F.E. The Determination of Surface Tension (Free Surface Emergy), and the Weight of Falling Drops: The Surface Tension of Water and Benzene by the Capillary Height Method. J. A. Chem. Soc. 41(6) (1919) 970-992 9. N.R. Pallas and Y. Harrison. An Automated Shape Apparatus and the Surface Tension of Pure Water. Colloid. Surf. 43 (1990) 169-194 10. Ley, M.T. et. al. The Physical and Chemical Characteristics of the Shell of Air-Entrained Bubbles in Cement Paste. Cement Concr. Res. 39 (2009) 417-425.

201



PERCOLATION IN CEMENTITIOUS MATERIALS UNDER FREEZE-THAW CYCLES INVESTIGATED BY MEANS OF ELECTRICAL RESISTIVITY Zhendi Wang (1), Ling Wang (1) , Yan Yao(1) (1) State Key Laboratory of Green Building Materials, China Building Materials Academy, Beijing, China Abstract Percolation in cement-based materials under cyclic freezing and thawing was observed by electrical resistance measurement. Wenner method was employed to determine the electrical response, reference mortar and mortar with 0.5wt.‰ of cement mass of analytically pure AgI were investigated in this study. Plots of electrical resistivity in terms of time and temperature were used to analyse the percolation process. Results show that ice played a decisive role in the percolation of electric current. Two phases of electrical resistivity in terms of temperature during freezing and thawing was picked out. Depercolation during freezing and repercolation during thawing occurred, and the percolation electrical resistivity remains mortar with AgI. The percolation in reference mortar was not observed due to the water content and high degree of surpecooling of pore water. Because depercolation and repercolation occur at certain diameter of pores, the findings may provide a new perspective for in-suit pore structure determination. 1. Introduction

The percolation theory deals with emergent properties related to the connectivity of large numbers of disordered objects. A simple percolation problem can be described by a good electrical conductivity system, which contains random emplacement of metallic and plastic shperes. If two plastic shperes touch each other, a current can’t pass from one to the other. If the fraction of plastic shperes exceeds a critical value, a continuous conducting pathway will be blocked, the larger the fraction of plastic spheres, the poor connected the path will be and the greater the electrical resistivity of the system. The nucleation and growth of ice in pores of cement-based materials (CBM) can be understood to be a percolation process in which the ice crystals (regard as plastic spheres) that form in the free space of aggregate and cement

202



hydration products (regard as metallic spheres), breaking the conductible paths for electrical current, see figure 1. Generally, the electrical conductivity is contributed from both the solid and liquid phases, while solid phases (cement clinkers and hydration products) and ice crystals are several orders of magnitude lower than the liquid phase [1-3]. Therefore, the electrical conduction in cement paste takes place through its liquid phase (see Fig. 1). This permit to use electrical method to assess the percolation phase transition in CBM. The percolation process in fresh cement paste has been investigated using electrical resistivity method, a few of representative tests of which was done by Li and co-workers [4-6]. Their results showed that the setting and hardening behaviour of fresh CBM can be regard as hydration products form on the clinker particles and lead to the formation of clusters and eventually join into a continuous elastic network. Based on the fact that the electrical resistivity of the CBM increases when damage breaks percolation paths for conduction, Chung and coworkers studied the evolution of damage in CBM subjected to fast and slow freeze-thaw cycles [7]. Boundoin et al. found more continual patterns of increases of electrical resistivity with temperature decreasing by ac impedance spectroscopy [1, 8]. Because the values of electrical resistivity are sensitive to nucleation and growth of ice crystals in pores of CBM, it has been suggested that the curve of -T may provide the information to assess the amount of ice formed in pores and the effect of frost action to durability of CBM [8]. Both depercolation point and repercolation point were observed and determined in partly saturated cementitious materials by Wang and co-workers [9-11]. They also found that supercooling of pore solution was highly impacted on the percolation process [11, 12]. However, percolation process in saturated mortar has not yet been addressed clearly. Therefore, the aim of this study is to characterize the percolation phase transition in vacuum saturated CBM by means of electrical resistance measurement. Direct current (DC) electrical resistivity of two types of mortars was measured using a four-probe method to monitor the electrical resistivity response during freeze-thaw cycles. The hysteresis in the electrical resistivity between freezing and thawing was observed and analysed. Electrical resistivity of cement hydration products and pore water and the role of ice in percolation were also discussed.

(a) (b)

Fig.1 Schematic sketch of patterns of electrical conducting in saturated porous materials (a)

above 0°C; (b) Ice formed

203



2. Experiments and methods 2.1. Materials and mixture ASTM Type I Portland cement was used in this study. Water to binder ratio of 0.5 was adopted. Two kinds of mortars, reference mortar and mortar with silver iodide (AgI), were designed for the investigations. The reference mortar with the mix proportion of cement: water: sand equal to 1:0.5:3.8 was prepared. AgI is often used for rain-inducing. To minimize supercooling of pore water, an additional 0.5wt.‰ of cement mass of analytically pure AgI was added to the fresh reference mortar, instead of sprinkling AgI powder on the external surfaces of the hardened mortar. 2.2. Samples preparation and experimental procedures The raw materials with the mix proportions presented above were mixed and casted into prisms with dimensions of 25 mm x 25 mm x 280 mm. After casting, the specimens were demoulded and stored in a dark room maintained at 20°C and about 95% relative humidity (RH) for further curing. At an age of 28 days, the specimens were taken out and each long specimen was cut into three shorter prisms (25 mm x 25 mm x 70 mm). The short prisms were then vacuum saturated for 24 hours. After that, the vacuum saturated specimens were taken out of water, and the surfaces of the selected mortars were polished gently to remove free laitance. The four-probe method was used for DC electrical resistivity measurement. Copper wires were applied circumferentially around the short prisms. Electrically conducting carbon powder was placed between the specimens and the copper wires to ensure good connection and electrical conduction. The specimens were then covered by several layers of plastic film to avoid moisture exchange during F-T cycling. The prepared specimens were then assembled as illustrated in Fig. 2 and placed into an environmental chamber. The temperature of the freeze-thaw courses was cycled was between 7 0°C, and the number of the freeze-thaw cycles were 45. The electrical resistivity was recorded automatically using a Keithley 2001 multimeter.

Fig.2 Schematic illustration of electrical resistivity measurement setup

3. Results and analysis 3.1 Measured electrical resistivity during freeze thaw cycles The measured electrical resistivity of reference mortar and the temperature at the center of sample as a function of testing time are illustrated in Fig. 3. In the beginning, the vacuum saturated reference mortar has electrical resistivity of around 150 , which consistent with

204



the results for OPC concrete under very wet, submerged, and splash zone environment [13]. The resistivity increases with the decreasing of temperature on freezing process, and reaches the peak value around at the minimum temperature. Then, the electrical resistivity decreases with the increasing of temperature upon thawing. It can be seen that the cycle repeats, and the electrical resistivity varies continuously with the temperature during the whole testing, which implies that the conductible paths for electrical current of the vacuumed saturated reference mortar aren’t blocked by the ice crystal in the current temperature range. In addition, the upper envelope (at the highest temperature) upshifts more than the lower envelope (at the lowest temperature), implying that the damage occurs more significantly on freezing than on thawing [7].

Fig.3. Electrical resistivity of reference mortar and temperature in terms of testing time

Fig.4 shows the measured electrical resistivity of AgI mortar and the temperature at the center of the sample in terms of testing time. The initial value of electrical resistivity is basically close to the results got from OPC concrete in very wet circumstance but a little bit higher than that of reference sample. Similar to the reference sample, the electrical resistivity increases first with the decreasing of temperature upon freezing. Note that the electrical resistivity suddenly rose to essentially infinity, which is also observed by Cao and Chung [7]. This phenomenon repeats for the whole experiment. Cao and Chung understood that the interruption of electrical resistivity was attributed to the freeze-thaw failure [7]. To compare with the reference mortar, we are more inclined to believe that the interruption attributes to the adding of AgI and action of ice formation. The water in fine pores isn’t solidified in the existing temperature of reference mortar due to the high degree of supercooling of pore water, while the supercooling of pore water in fine pores of AgI mortar is reduced significantly after

205



AgI added. With the aid of AgI, ice crystals nucleate and grow with the temperature, and eventually block the conductible paths for electrical current.See section 4.2 for more detailed discussion

Fig.4. Electrical resistivity of AgI mortar and temperature in terms of testing time

3.2 Electrical resistivity-temperature hysteresis Fig.5 shows the results of electrical resistivity versus temperature for reference mortar. To better illustration, freeze-thaw cycle at 1st, 10th, 20th, 30th, and 40th are selected. Fig.5 shows that the hysteresis of electrical resistivity between freezing and thawing is not obvious; this may be due to the less water migration of the saturated samples during one cycle. It can be seen from enlarged part of fig.5 that the electrical resistivity curves move up progressively with the number increase of freeze-thaw cycles. This can be attributed to the water redistribution and damage accumulating. It also can be found that the electrical resistivity can be divided into two less obvious stages. From the room temperature to 0°C the electrical resistivity increases slowly, while the electrical resistivity increases more rapidly as the temperature decrease further because the water is solidified in macro-pores and blocks part of the conductible paths for electrical current.

206



Fig.5 Electrical resistivity in terms of temperature for reference mortar

The electrical resistivity in terms of temperature for AgI mortar at the1st, 10th, 20th, 30th, and 40th cycle are shown in fig.6. Obviously, a significant hysteresis of electrical resistivity between freezing and thawing are observed for all cycles. The thermal gradient [14], pore shape and connectivity [15], and the curvature-induced metastability of liquid confined in pores [16, 17] are responsible for the can account for the freezing thawing hysteresis.

(a) (b)

207



(c) (d)

(e)

Fig.6 Electrical resistivity in terms of temperature for AgI mortar Again, the electrical resistivity curves upon freezing can be divided into two less obvious phases. The augment of electrical resistivity is unremarkable when the temperature above 0°C, while the electrical resistivity increase significantly in the temperature range of -17~ 0°C. It can be found from fig.6b-

paths for electrical current is cut off. In the former study [9], we defined the first not detectable point as depercolation point. In the thawing stage, the ice crystal melts with the increase of temperature, and correspondingly brings out more free water and makes the current running. The point of the first measurable electrical resistivity is defined as the repercolation point. 4. Discussion

4.1 Electrical conductivity of cement hydration products and pore water

contributions from both the solid and the liquid phases. The relationship between the conductivity and the solid and liquid phases is [18]:

208



(1) Where s term is a solid, such as hydrations products and aggregates. Where a, b, , t and denotes the tortuosity at saturation, liquid phase, water content, and the critical volume for percolation. The solid phase, such as cement hydration products and aggregate, is assumed to have zero conductivity, so the first term can be neglected. In other words, the current flow of the vacuum saturated specimen is mainly contributed by the pore water. Thus, the phase transition from liquid to solid will be impacted the conductivity behaviour significantly. 4.2 The actions of ice The plots of electrical resistivity against temperature indicate important information of ice action in pores. Take fig.6c as an example, the slight change in electrical resistivity around zero can be attributed to the ice formed in big pores of AgI mortar. According to the Gibbs–Thomson equation, the solidification temperature of pore water depends on the ’throat’size percolating the pore, Assuming that the pores are cylindrical, and the ice front penetrating in pores have a hemispherical surface. In the freezing stage, the electrical resistivity before ice nucleation increases slightly to decrease temperature; here the mobility of pore water is lowered by decreasing temperature. Ice then forms in and occupies large pores, which partially breaks the routes for electrical conduction; see step (a) in Fig. 7. As the temperature decreases further, more pores are occupied by ice, more routes for electrical conduction are broken, and the electrical resistivity increases significantly. Once ice occupies the critical pores, which are the smallest pores creating a connected path in a pore structure, depercolation occurs; see step (b) in fig.7. In the thawing stage, ice melts at the side of cylindrical pores, and the temperature required for melting is half of that for freezing. Once the ice in critical pores melts, it repercolates; see step (c) in Fig. 7. As temperature increases further, more ice melts, and the electrical resistivity decreases significantly; see steps (d) in Fig. 7. After ice has melted completely, the electrical resistivity decreases slightly. On the other hands, because depercolation and repercolation occurs when the critical pores occupies or not, the electrical resistivity-temperature curve also provides a possibility to assess the in suit pore structures of porous materials.

209



Fig.7. the effect of ice on the percolation process

5. Concluding remarks

This study measured the electrical resistivity of reference mortar and AgI mortar under the action of freeze-thaw loading. The electrical resistivity varies continuously with the temperature during the whole testing for reference mortar, while the electrical resistivity was interrupted by the ice formation for AgI mortar. The plots of electrical resistivity as a function of time and temperature were used to analysis the percolation process. Ice played a decisive role in the percolation of electrical resistivity. Both depercolation and repercolation electrical resistivity remain around 800~900 m for AgI mortar. The percolation in reference mortar was not observed due to the water content and high degree of surpecooling of pore water. Further studies need to be conducted to illustration the relationship between electrical resistivity-temperature and pore structure. Acknowledgements The authors acknowledge the financial support from National Natural Science Foundation of China (Grant No. 51402278).

210



References [1] Perron, S., and Beaudoin, J. J. “Freezing of water in Portland cement paste—an ac imped

ance spectroscopy study.” Cem. Concr. Compos. 24 (2002) , 467–475. [2] Rajabipour, F. “In situ electrical sensing and material health monitoring of concrete struct

ures.” Ph.D. dissertation, Purdue Univ., West Lafayette, Indiana (2006).. [3] Weiss, J., Snyder, J., Bullard, J., and Bentz, D. “Using a saturation function to interpret th

e electrical properties of partially saturated concrete.” J. Mater. Civil Eng., (2013), 10.1061/(ASCE)MT.1943-5533.0000549, 1097–1106.

[4] Li Z, Wei X, Li W, Preliminary interpretation of Portland cement hydration process using resistivity measurements. ACI Mater J 100 (2003), 253–7.

[5] Xiao L, Li Z. Early-age hydration of fresh concrete monitored by non-contact electrical resistivity measurement. Cem Concr Res 38(2008), 312–9.

[6] Wei X, Li Z. Study on hydration of Portland cement with fly ash using electrical measurement. Mater Struct 38(2005), 411–7.

[7] Cao JY, Chung DDL. Damage evolution during freeze–thaw cycling of cement mortar, studied by electrical resistivity measurement. Cem Concr Res 32(2002), 1657–61.

[8] Sato, T., and Beaudoin, J. J. “Coupled AC impedance and thermomechanical analysis of freezing phenomena in cement paste.” Mater. Struct., 44(2011), 405–414.

[9] Z. Wang, Q. Zeng, L. Wang, Y. Yao, K. Li, Characterizing blended cement pastes under cyclic freeze–thaw actions by electrical resistivity, Construction and Building Materials 44(2013), 477-486

[10] Z. Wang, Q. Zeng, L. Wang, Y. Yao, K. Li, Electrical Resistivity of Cement Pastes Undergoing Cyclic Freeze-Thaw Action, Journal of Materials in Civil Engineering 27 (2015), 04014109

[11] Zhendi Wang, Qiang Zeng, Ling Wang, Yan Yao, Kefei Li, Effect of moisture content on freeze–thaw behavior of cement paste by electrical resistance measurements, Journal of Materials Science, 49(2014), 4305-4314

[12] Zhendi Wang, Qiang Zeng, Yuekai Wu, Ling Wang, Yan Yao, Relative humidity and deterioration of concrete under freeze–thaw load, Constr. and Build. Mater. 62(2014), 18–27

[13] Polder R, Andrade C, Elsener B, Vennesland O, Gulikers J, Weidert R, et al. Test methods for on-site measurement of resistivity of concrete. Mater Struct; 33(2000):603–11.

[14] Bishnoi S, Uomoto T, Strain-temperature hysteresis in concrete under cyclic freeze–thaw conditions. Cem Concr Compos 30(2008), 374–380

[15] Sun Z, Scherer GW Pore size and shape in mortar by themoporosimetry. Cem Concr Res 40 (2010), 740–951

[16] Petrov O, Furo´ I, A study of freezing–melting hysteresis of water in different porous materials. Part I: porous silica glasses. Microporous Mesoporous Mater 138(2011), 221–227

[17] Petrov O, Furo´ I, Curvature-dependent metastability of the solid phase and the freezing-melting hysteresis in pores. Phys Rev E 73(2006), 011608

[18] Hunt A., Ewing R. Percolation theory for flow in porous media, Springer, (2009) Doi10.1007/978-3-540-89790-3

211



APPLICATION OF AIR ENTRAINED CONCRETE IN TOLLWAYS CONSTRUCTIONS IN LIAONING PROVINCE OF CHINA Wencui Yang (1), Xiaoping Cai (1), Yong Ge (1) and Jie Yuan (1) (1) Harbin Institute of Technology, Harbin, China Abstract Liaoning Province is located in northeast China, from 38°43' to 43°26' north latitude. The analysis of 30 years weather data showed that there were many annual freeze-thaw cycles in this area, which caused severe frost deterioration of concrete. The survey of tollways and roads in Liaoning Province found that scaling damage on affiliated concrete constructions and bridges concrete was common and severe. In a rebuilding project of tollways from Shenyang City to Siping City, we recommended to use air entraining agent, fly ash and slag to improve the frost resistance of concrete. Specimens were selected from the construction sites randomly and the air voids structure and the frost resistance were tested, and the results showed that the durability of most air entrained concrete was improved. However, two concrete mixtures with the air content of 4% failed in the frost test, which could attribute to their unsatisfactory air voids structure and greater spacing factor. In addition, some suggestions for air entrained concrete construction and quality control are described. 1. Introduction Cement-based materials deteriorate due to the action of freeze-thaw cycles in cold areas. Under the freeze-thaw cycles, micro cracks appear and widen gradually, leading to the degradation of cement concrete. Different damage mechanisms about this phenomenon have been proposed [1, 2]. The hydraulic pressure theory proposed that the damage effect was caused by the formation of ice with 9% volumetric expansion of original water in pores, while the osmosis pressure attributed it to the vapour pressure difference and concentration difference between frozen site and unfrozen water [3~5]. Moreover, the damage of cement-based materials becomes more serious and complicated when salts present. Researches showed that deicing chemicals, such as sodium chloride and calcium chloride, caused severe scaling on concrete surfaces, especially when the solution concentration was 3%~4% [6~8].

212



Researches also showed that air entraining improved the frost resistance of concrete remarkably [9, 10]. On one hand, the air bubbles created by air entraining agent could reduce the capillarity and saturation degree of pores. On the other hand, the entrained air voids could release the hydraulic pressure and the osmosis pressure. The key is to produce enough air bubbles and lower the air voids spacing factor in hardened concrete. However, in construction sites, the air content of fresh concrete could lose during the construction process, such as transport, cast, compaction and curing, leading to the failure of air entraining. So quality control of air entrained concrete during the construction is very important. In this paper, a survey of the frost deterioration and scaling damage of concrete in tollways and roads in Liaoning Province, China, is reported as well as the environmental and climate conditions. And in a rebuilding project of tollways from Shenyang City to Siping City, we recommended the contractors to employ air entraining agent, fly ash and slag to improve the frost resistance of concrete. The frost resistance and air voids structure were tested using the samples from the construction field to see the effect. In addition, some suggestions for air entrained concrete construction and quality control are also described. 2. Frost deterioration of concrete in tollways in Liaoning Province Liaoning Province locates in northeast China, from 38°43' to 43°26' north latitude, the climate of where is temperate continental monsoon climate. There are 3 to 4 months in a year that the lowest temperature in a day is below 0°C, and the temperature fluctuates at 0°C frequently in autumn and spring, which could cause freeze-thaw. Table 1 showed that some weather data (average values of 30 years data) of 5 typical cities in northeast China and the annual freeze-thaw cycles calculated. It is should be noted that the pure water freezes at 0°C, but the water in concrete pores freezes at lower temperature, since the freezing point of a solution in small capillaries decreases. Thus, the annual freeze-thaw cycles were calculated when the minimum land surface temperature in one day was lower than 0°C, -3°C and -5°C respectively, as showed in Table 1. The results showed that Shenyang had the most annual freeze-thaw cycles in 5 cities, even more than that of other northern ones. Therefore, concrete structures are subjected to many freeze-thaw cycles in this area. Table 1: Annual freeze-thaw cycles of 5 cities in northeast China.

Cities North latitude

Average air temperature in January

(°C)

Lowest land surface

temperature (°C)

Annual freeze-thaw cycles (Days in one year that the maximum land

surface temperature >0°C and the minimum land surface temperature <

T°C) T=0°C T=-3°C T=-5°C

Mohe 53°48' -29.6 -53.9 107.8 80.2 65.9 Harbin 45°44' -18.3 -40.9 124.2 101.2 87.1 Changchun 43°54' -15.1 -41.5 132.2 110.6 96.8 Shenyang 41°48' -11 -36.2 151.3 119.9 101.2 Dalian 38°55' -3.9 -21.9 131.7 102.5 79.6

213



In addition, Liaoning Province has lots of rain and snow. The average annual precipitation is 600-1100 mm, and the period of snow in each year lasts at least 3 months. Deicer salts, basically chlorides, are massively used on urban roads and tollways in winter. It is found that the scaling damage due to the action of freeze-thaw cycles and deicer salts is common for concrete in Liaoning. Although the pavements of highways and roads in Liaoning Province were usually built by asphalt concrete, many affiliated constructions were made of cement concrete, such as barriers, trench cover plates and bridge pavement expansion joints [11]. Concrete barriers with the height of 0.8m to 1.3m are widely used, making it difficult for vehicles to cross into lanes and directly into the path of oncoming traffic. The concrete for those barriers has medium compressive strength, 30MPa-40MPa, and is usually non-air entrained. The survey found that the concrete of barriers scaled severely, as shown in Fig. 1, Fig. 2 and Fig. 3. The scaling started from the base of the barriers where deicer salts accessed first and then progressed and scattered to the whole surface of barriers, even up to the back surface. To protect the concrete form the scaling deterioration, expensive granite stone plates were used to cover the concrete barriers at the bottom in several newly built highways. However, the scaling still occurred above the stone plates after several years (Fig. 4). Trench cover plates also showed scaling deterioration, and it occurred even faster than those barriers (Fig. 5). Some cement concrete expansion joints between asphalt pavements on bridges showed very bad performance, as shown in Fig. 6. The concrete cracked, scaled and small pieces and aggregate peeled off and were taken away by vehicles gradually. The deterioration was fast, and the grooves were almost empty after only one winter. The impact of vehicles wheels and the vibration of bridges may cause original cracks, but the freeze-thaw action and deicer salts accelerated the deterioration of concrete. It is noticed that there is damage on the bottom surface of bridges beams. Fig. 7 showed that the concrete on inside walls of a box beam scaled seriously and aggregates and steel bars are exposed. This is caused by the freeze-thaw action and deicer salt solution which leaked from the pavement above. The damage to this extent is not common, but the small scale breaks on beams concrete which was patch repaired are found cross the province, as given in Fig. 8.

Fig. 1 Scaling deterioration of concrete barriers.

Fig. 2 Scaling on the back surface of concrete barriers.

214



Fig. 3 Damage at the bottom of concrete barriers.

Fig.4 Scaling of concrete above the stone plates covered the bottom of barriers.

Fig. 5 Deterioration of concrete on trench cover plates.

Fig. 6 Damage of concrete expansion joints on a bridge.

Fig. 7 Scaling of concrete inside of a bridge box beam.

Fig.8 Small scale damage and patch repaired of concrete beam.

215



3. Application of air entrained concrete in tollways constructions 3.1 Materials, proportions and test methods Based on the survey results in Section 2, we recommended using air entraining concrete for piles, piers and beams of bridges and barriers in a rebuilding project of tollways from Shenyang City to Siping City. Local raw materials were used. Cement with the strength grade

A compound chemical admixture with the key component of SJ-2 saponin air entraining agent and polycarboxylate superplasticizer was used to adjust workability and air content of fresh concrete. The properties of all the raw materials could meet the requirements of Chinese standards. Table 2 shows the proportions of concrete produced by different contractors. Table 2: Proportions of concrete (kg/m3).

Contractor Mix W/B Cement Fly ash

Slag powder Sand Crushed

Stone Water Chemical admixture

A A1 0.39 30 788 1088 3.69 A2 321 62 33 1078 A3 0.31 373 73 39 1086

B B1 263 761 169 3.76 B2 709 1108 B3 0.31 103 721 1037

C C1 0.39 1099 C2 321 728 1093 C3 0.31 387 39 671

The specimens with size of 100mm×100mm× 00mm were cast from each mixture. All

°C and e-thaw

cycles. Three specimens were tested for each mixture. The frost resistance of concrete was tested by a fast freeze-thaw method in water -thaw cycle involved freezing at (-

°C of sample center of sample center hours. The sound velocity and weight of the specimens were measured at different freeze-thaw cycles. The relative dynamic modulus of elasticity and weight loss of specimen were used to evaluate the frost resistance. The relative dynamic modulus of elasticity (RDEM) is the ratio of the dynamic modulus of elasticity for a certain cycles to the initial value before subjected to freeze-thaw cycles. The RDEM is calculated by the following expression: RDEM =En/E0=(Vn/V0)2, where En and E0 are the dynamic modulus of elasticity of n cycles and initial value, respectively. Vn and V0 are the sound velocity of n cycles and initial value, respectively. The weight of specimens was measured by electronic balance of precision of

The parameters of voids system of concrete mixtures were tested by the linear traverse

-2001 “Test code for hydraulic concrete”[12]. Before testing, the samples were cut into 10mm-width slices, and polished with successively finer abrasives until suitable for testing.

216



3.2 Frost resistance and air voids structure The results of frost test using the concrete specimens prepared in the construction field are shown in Fig. 9. The concrete from contractor A showed satisfactory frost resistance, as the relative dynamic modulus of elasticity was higher than 90% and the mass loss was lower than 2% even after 300 freeze-thaw cycles. For concrete from contractor B, the maximum freeze-thaw cycles at failure of the concrete with the water-binder ratio of 0.34 and 0.31 were up to 300 cycles. However, the relative dynamic modulus of elasticity of concrete with the higher water-binder ratio, 0.45, decreased fast, and it failed at 60 freeze-thaw cycles. The concrete from contractor C showed the similar results to that from contractor B. The concrete with the water-binder ratio of 0.39 damaged early, as the relative dynamic modulus of elasticity dropped to 20% and the mass loss increased to 5% as the freeze-thaw cycles increased from 100 to 150. The other two mixtures showed fairly good frost resistance and their relative dynamic modulus of elasticity were still more than 90% after 300 freeze-thaw cycles.

217



0 50 100 150 200 250 3002030405060708090

100

Relat

ive d

ynam

ic ela

stic m

odul

us (%

)

Freeze-thaw cycles

W/B=0.39 W/B=0.35 W/B=0.31

0 50 100 150 200 250 300

-2

-1

0

1

2

3

4

5 W/B=0.39 W/B=0.35 W/B=0.31

Mas

s los

s (%

)

Freeze-thaw cycles (a) Contractor A

0 50 100 150 200 250 3002030405060708090

100

Relat

ive d

ynam

ic ela

stic m

odul

us (%

)

Freeze-thaw cycles

W/B=0.45 W/B=0.34 W/B=0.31

0 50 100 150 200 250 300

-2

-1

0

1

2

3

4

5

W/B=0.45 W/B=0.34 W/B=0.31

Mas

s los

s (%

)

Freeze-thaw cycles (b) Contractor B

0 50 100 150 200 250 3002030405060708090

100

Relat

ive d

ynam

ic ela

stic m

odul

us (%

)

Freeze-thaw cycles

W/B=0.39 W/B=0.35 W/B=0.31

0 50 100 150 200 250 300

-2

-1

0

1

2

3

4

5

W/B=0.39 W/B=0.35 W/B=0.31

Mas

s los

s (%

)

Freeze-thaw cycles (c) Contractor C Fig. 9 Relative dynamic elastic modulus and mass loss of concrete during freeze-thaw cycles.

218



The spacing factor of air voids system in concrete could be used to evaluate the potential frost resistance of concrete. Researches have showed that the concrete would have enough frost resistance if the spacing factor was less than a critical value [13, 14]. The parameters of the air voids structure of concrete mixtures are given in Table 3. The results showed that all nine mixtures had proper air content which was supposed to provide enough protection from frost action. However, the group B1 and C1 had a greater spacing factor, 0.392 mm and 0.340 mm respectively. And the frost resistance of these two groups concrete was poor too. So the relationship between the air voids structure and the frost resistance was consistent with the results of researches mentioned above. The air voids structure and the water-binder ratio were the main factors influenced the frost resistance of concrete. The results from frost test indicated that the air entraining agents could fail when they were used to improve the frost resistance of construction concrete. So not only the air content should be controlled, but also the air voids structure should be considered when using the air entraining in field. There are some factors affected the stability of air voids which remained in the hardened concrete. Therefore, measures should be employed during the construction process to assure the function of air entraining. Table 3: Parameters of the voids system of concrete.

Contractor Mix Average

chord length (mm)

Specific surface

(mm2/mm3)

Average radius (mm)

Air content (%)

Spacing factor (mm)

A A1 0.187 21.4 0.140 6.1 0.205 A2 0.173 23.2 0.129 4.9 0.217 A3 0.151 26.4 0.113 4.3 0.209

B B1 0.283 14.2 0.212 4.1 0.392 B2 0.168 23.7 0.126 4.1 0.230 B3 0.180 22.2 0.135 4.0 0.262

C C1 0.263 15.2 0.197 4.3 0.340 C2 0.172 23.3 0.129 5.3 0.209 C3 0.139 28.7 0.105 3.9 0.205

219



3.3 Suggestions on concrete quality control in constructions

Using air entraining agents with stable quality and good compatibility with other materials. According to the frost test results, the proper air content of fresh concrete couldn’t assure the good frost resistance of concrete, while the air voids structure of hardened concrete is more important. To gain the satisfactory air voids structure, the air entraining agent with stable quality should be used, and its compatibility with other materials also should be tested.

Prolonging the mixing time of air entrained concrete. Air entrained concrete needs more mixing time to produce air bubbles, and it also needs more time for the concrete with fly ash and slag to become homogeneous. More 15-20 seconds could be prolonged for mixing of air entrained concrete.

Monitoring and controlling the workability of fresh concrete. Bleeding or segregation of fresh concrete could lead to bad performance of hardened concrete, while the suitable workability is beneficial for durability and appearance quality of concrete. For example, the field monitoring data of the rebuilding project of tollways from Shenyang City to Siping City showed that strength and durability of concrete were better when the slump was controlled as 180mm.

Controlling proper vibration time during cast and compaction of concrete. If the vibration time is too long, the sand line and stratification line may occur on concrete surface and the air content may decrease. On the other hand, if the vibration time is not enough, big air bubbles and honeycomb could appear on concrete surface. The proper vibration time and method ought to be determined by field test.

Curing the finished concrete structures in time. The hydration of cement needs appropriate temperature and relative humidity. Therefore, after casting, concrete structures need to be covered by plastic film and supplemented curing water in time.

4. Conclusions (1) The 30 years weather data of two main cities of Liaoning Province showed that there were many freeze-thaw cycles in each year, even more than that in the cities with higher latitude, which caused frost deterioration of concrete in this area. (2) The survey of tollways and roads in Liaoning Province found that frost deterioration and scaling damage of concrete was common and severe. (3) Air entrained concrete with fly ash and slag was used in a rebuilding project of tollways from Shenyang City to Siping City, and the test on the specimens from the construction sites showed that the durability of concrete was improved, although two concrete mixtures had unsatisfactory frost resistance and air voids structure. (4) When the air entrained concrete was used, the quality of concrete could be assured by using air entraining agents, controlling the proper mixing time and vibration time, monitoring slump and curing in time.

220



5. Acknowledgement This work was supported by the Natural Science Funds of China (Project No. 51278157) and the Fund of “The Consulting for Concrete Construction Technique in the Rebuilding Project of Highways from Shenyang to Siping”. References [1] M. Pigeon, J. Marchand and R. Pleau. Frost Resistance Concrete. Construction and

Building Materials. 1996, 10(5): 339-348 [2] Vesa Penttala. Surface and internal deterioration of concrete due to saline and non-saline

freeze–thaw loads. Cement and Concrete Research. 2006,36: 921-928 [3] T. C. Powers. A Working Hypothesis for Further Studies of Frost Resistance of Concrete.

Journal of ACI. 1945,16(4): 245-272 [4] T. C. Powers. Void Spacing as A Basis for Producing Air-Entrained Concrete. Journal of

ACI. 1954, 50(9): 741-760 [5] T. C. Powers, R. A. Helmuth. Theory of Volume Changes in Hardened Portland Cement

Paste During Freezing. Proceedings, Highway Research Board, 1953, 32: 285-297 [6] J. J. Valenza, G. W. Scherer. Mechanism for Salt Scaling. Journal of American

Ceramisite Society. 2006, 89(4): 1161-1179 [7] W. Kejin, E. N. Daniel, A. N. Wilfrid. Damaging Effects of Deicing Chemicals on

Concrete Materials. Cement and Concrete Composites. 2006, 28: 173-188 [8] John J. Valenza II, George W. Scherer. A review of salt scaling: I. Phenomenology.

Cement and Concrete Research. 2007, 37: 1007-1021 [9] Zhenhua Sun, George W. Scherer. Effect of air voids on salt scaling and internal freezing.

Cement and Concrete Research. 2010, 40: 260-270 [10] S. Chatterji. Freezing of Air-entrained Cement-based Materials and Specific Actions of

Air-entraining Agents. Cement and Concrete Research. 2003, 25: 759-765 [11] Wu Liang. Investigation and Study on Typical Structure of Rural Highway Pavement in

Northern Region of China. Northern Communications. 2010. (6): 20-23 (In Chinese) [12] Test code for hydraulic concrete (DL/T 5150-2001). Electric Power Industrial Standard of

China. 2001. China Electric Power Press. (In Chinese) [13] HUANG Xiao-heng, XU Cai-hong, WANG Li-wen. Experiment and Study of Influence

of Property of Air-Bubbles in Hardened Concrete upon Frost Resistance of Concrete. China Harbour Engineering. 2006, (3):14~17 (In Chinese)

[14] Christiane Foy, Michel Pigeon, Nemkumar Banthia. Freeze-Thaw Durability and Deicer Salt Scaling Resistance of a 0.25 Water-Cement Ratio Concrete. Cement and Concrete Research. 1988, 18(4): 604~614

221



INFLUENCE OF DUCTILITY AND MICROCRACKING ON THE FROST DURABILITY OF CEMENTITIOUS COMPOSITES WITH HIGH VOLUMES OF FLY ASH

(1), (2) (3) (4) (1) Adana University of Science and Technology, Adana, Turkey (2) Selçuk University, Konya, Turkey (3) Gazi University, Ankara, Turkey (4) Ryerson University, Toronto, Canada

In the present paper, effects of utilizing high volumes of fly ash (FA) and poly-vinyl-alcohol (PVA) fibers on the freeze-thaw resistance of Engineered Cementitious Composites (ECC) have been investigated. Three different ECC mixtures with FA-cement (FA/C) ratios of 1.2, 2.2 and 4.2 were manufactured at a constant water to cementitious materials ratio of 0.27. For comparison, identical mixtures without PVA (matrix) were produced as well. Frost resistance was evaluated after exposing the mixtures up to 300 freeze-thaw cycles following ASTM C666, Procedure A. Performance assessment has been made by residual flexural properties (flexural strength, mid-span beam deflection and flexural stress - deflection curve), ultrasonic pulse velocity and mass loss. The air-void characteristics of mixtures were also studied using linear transverse method. Test results confirm that ECC mixtures with high volumes of FA remain durable, and show a tensile strain capacity of more than 2% even after 300 freezing and thawing cycles. On the other hand, before completing 300 freezing and thawing cycles, matrix (ECC without fiber) specimens have severely deteriorated. Thus, results indicate that the addition of micro PVA fibers into the ECC matrix substantially improved the frost resistance. The results of freeze-thaw tests also showed that the reduction of residual physical and mechanical properties with increasing number of freeze-thaw cycles increases with the increase in FA/C ratio. 1. Engineered cementitious composites (ECC) became popular recently as one of the special types of High Performance Fiber Reinforced Cementitious Composites (HPFRCCs). What makes ECC different is its superior ductility and resistance to damage (cracking) under

222



different mechanical loading types [1]. Application of micromechanical material design allows designers to achieve superior tensile strain with moderate fiber volume (2%, by volume for standard ECC) [2]. High ductility comes with the formation of hair-like tiny microcracks on the brittle matrix and is a direct result of strain-hardening behavior. Crack openings generally stay below 100 μm level in a controlled way and are free of tensile rebar ratio. multiple microcracking behavior brings about outstanding durability properties to the ECC material itself [3]. As mentioned above, micromechanical material design is utilized in proportioning ECC mixtures for the satisfaction of strength and energy criteria in favor of attaining high ductility [4]. In order to obtain cracking in a controlled manner and multiple microcracking behavior in ECC fiber, matrix and interface properties are carefully tailored. Current ECCs are designed with a limited amount of very fine aggregates. It was stated in the literature that increased aggregate amounts and size results in higher matrix fracture toughness in ECC design delaying the initiation of cracks and lowering overall ductility in tension [5]. Furthermore, when coarser aggregates with greater amounts are added into ECC matrices, fibers show a tendency to clump and non-uniformity [6]. These considerations therefore lead aggregates small in size and amount to be used in ECC mixtures. However, by excluding the coarse aggregates in ECC mixtures, demand for cement usage increases. Thus, to partially replace the high cement contents in ECC mixtures with the supplementary cemetitious materials could be significantly advantageous in terms of environmental and economical reasons. Fly ash (FA) is more widely available all around the world in larger quantities among other supplementary cementitious materials. Beneficial effects of using FA in ECC production have been realized formerly thus it is now commonly used in ECC mixture designs [7]. Substitution of high amounts of cement with FA in ECC mixtures can significantly reduce the negative environmental effects. Moreover, it is stated that utilizing FA in ECCs nourishes tensile ductility by lowering matrix fracture toughness and contributes to robustness [8]. In addition, unhydraous FA particles having small size and smooth morphological characteristics contributes filler effect, more compact fiber-to-matrix interface increasing interfacial frictional bond [8] which is beneficial for multiple microcracking response and ultimate durability. Repair, maintenance and rehabilitation of the deteriorated infrastructures are a growing concern. In the presence of frost action and/or cold climatic conditions deterioration could be much faster as in the case of most of the transportation infrastructures serving under such conditions. It is therefore important to design ECC material to be used in such infrastructures to be frost resistant. By properly designing the air-void system of conventional concrete, resistance to freezing-thawing can be achieved. Although ECCs serving in field is reported to have good resistance to frost action [9], laboratory data on the freeze-thaw performance of ECC are lacking in literature. Additionally, studies related to frost durability of ECCs incorporated with high volumes of FA (HVFA) are even more limited in the current literature. Given the widespread usage of FA in the production of ECC mixtures nowadays, it is relatively important to obtain detailed understanding on the utilization of high volumes of fly ash in ECCs and its effects on frost durability.

223



As the main objective of the current paper, influence of FA and poly-vinyl-alcohol (PVA) fibers on the freeze-thaw durability of non-air-entrained ECC mixtures was of focus. Three different ECC mixtures with FA to cement ratios of 1.2, 2.2 and 4.2 were produced. Along with the ECC mixtures, mortar mixtures (without PVA fiber) with the identical proportions were prepared for comparison. Assessment of frost resistance was made following ASTM C666 Procedure A standard. The evaluated parameters were the changes in mass, pulse velocity and flexural parameters (flexural strength, mid-span beam deflection and flexural stress – deflection curve) after frost action. Air-void parameters were also studied according to ASTM C457 modified point count method. 2. 2.1 The materials used in the production of ECC mixtures were Type-I Portland cement (C), Class-F fly ash (FA) with a lime content of 5.6%, micro silica sand with a maximum aggregate size of 250 m, water, synthetic poly-vinyl-alcohol (PVA) fibers with a diameter of 39 μm, length of 8 mm, tensile strength of 1620 MPa, elastic modulus of 42.8 GPa and maximum elongation of 6.0% and a polycarboxylic-ether type high-range water reducing admixture (superplasticizer – SP). As mentioned previously, three different ECCs with FA to cement ratios (FA/C) of 1.2, 2.2 and 4.2 were produced herein and the proportions were given in Table 1. Mortar mixtures identical to ECC mixtures were manufactured for comparison purposes as well. To produce the mixtures, a standart mortar mixer was used and water to cementitious materials ratio (W/CM) for all of the mixtures was kept constant at 0.27. Since different amounts of FA were used in ECC mixtures it was not possible to keep the SP amount constant thus SP addition was continued until favorable fresh paste properties were observable. All of the mixtures were produced with no air-entraining admixture. Average compressive strength tests results of ECC and ECC matrix mixtures obtained from three different 50 mm cubic specimens were shown in Table 1 for 14 and 28 days. Specimens were cured under environmental conditions at 23±2 °C and 95±5% relative humidity until the pre-determined testing ages. As clearly seen from this table, there were continuous decrements in compressive strength with the increased FA amounts and PVA fiber addition was not that of influential in increasing the overall values. Despite the variations with the usage of different amounts of FA however, even at almost 80% replacement of portland cement with FA (FA/C = 4.2 – ECC3), 28-day-old average compressive strength of ECC could be more than 35 MPa. Table 1: Mixture properties of ECC mixtures.

ID.

W/CM

FA/C

PVA

3 SP

3 w/o PVA

14- 28- 14- 28- ECC1 0.27 1.2 0.36 26 2.3 36.1 60.3 39.2 62.5 ECC2 0.27 2.2 0.36 26 2.0 25.8 52.4 27.7 54.1 ECC3 0.27 4.2 0.36 26 1.5 18.2 35.1 19.1 36.4

224



2 - To be used in frost durability tests and analysis for determining air-void parameters, eight prism specimens measuring 400×100×75 mm from each mixture were manufactured. Air contents of ECC and matrix specimens were measured following ASTM C231 as well. All of the specimens used for frost durability were produced without appliying any compaction, removed from their molds after 1 day and further moist curing was continued for an additional 13 days at 23±2 oC. After completing 14 days, prisms were put into freeze-thaw cabinet following the ASTM C666 Procedure A standard. The cabinet consisted of 18 chambers where specimens were entirely submerged in water with thermocouple embedded in the center of each prism and specimens were exposed to cycling freezing and thawing for five or six times in one day. Each cycle composed of lowering the temperature of specimens to -18 oC and increasing it to 4 oC in 4 to 5 hours. The average flexural parameters (ultimate deflection and flexural strength) of the specimens were obtained by testing four companion samples just prior to the freezing and thawing cycles at the age of 14 days, and the result was used as the basis for determining the preloading flexural performance (control). After each 30 freeze-thaw cycle, specimens were taken out of the cabinet when they were thawed to measure the changes in mass and pulse velocity. Before putting back the specimens inside the cabinets chambers were cleaned and drinkable tap water was freshly filled in. In addition, the cycle numbers causing the failure of specimens were noted. With the completion of 300 cycles, specimens that kept their integrity were subjected to four-point bending loading from a span length of 355 mm for the evaluation of residual flexural parameters. Freeze-thaw resistance was determined in triplicate on all ECC mixtures with and without fiber. The air-void content and spacing factor of hardened ECC and matrix mixtures were determined by modified point count method according to ASTM C457. 3. 3.1 - Frost durability of concrete is closely related with the air-void parameters. As can be seen from Table 2, although no air entrainment was used in ECC and ECC matrix mixtures, fresh and hardened air contents of different mixtures were 6 to 7% and more than 7%, respectively. The recorded air content values were reported to be adequate for a reasonable frost resistance [11]. The reason for the high air content values recorded from these mixtures could be attributed to the absence of coarse aggregates and high viscosity of matrices at the fresh state [12]; while applying placement, rising of air bubbles to the surface of fresh mixture can be interrupted by fine particles and increased viscosity. Additionally, this became more pronounced when PVA fibers were incorporated to the mortar mixtures (Table 2). Addition of PVA fibers to the matrix mixtures significantly increases the viscosity of fresh ECC mixtures leading to the occurence of higher amounts of entrapped air voids inside the cementitious matrices. Another possible reason for the higher amounts of air voids in the case of ECC mixtures compared to mortars could be due to the effect of proprietary hydrophobic oiling agent covering the surface of PVA fibers to tailor fiber-to-matrix interfacial properties for the strain-hardening response.

225



Hardened air content of 8.2, 8.9 and 9.4%, spacing factors of 0.241, 0.149 and 0.132 mm and specific surfaces of 25.6, 39.5 and 44.1 mm-1 were determined for ECC1, ECC2 and ECC3 mixtures, respectively. Although ECC1 mixture exhibited spacing factor (0.241 mm) slightly higher than the generally accepted value of 0.200 mm, this did not negatively affect the frost resistance of ECC1 mixture as detailed in the forthcoming sections. One can see from Table 2 that with the increments in FA content, marked reductions in the spacing factor values of hardened ECC mixtures took place. Nearly same air content along with the reduced spacing factor means that increased amounts of FA led average bubble sizes to be smaller. This behavior obtained when high volumes of FA were used can be attributed to increments in paste volume due to lower density of FA and pore refinement effect. As in the non-air entrained mortar incorporating silica fume postulated by Feldman [13], after consuming calcium hydroxide as a result of pozzolanic reactions FA can create its own micro air void network around the fine aggregate particles which creates voids. On the other hand, since the pozzolanic reaction involving FA is slower than regular cement hydration, there is a significant amount of unhydrated FA in the high-volume FA ECC after short time curing (14 days in this study), and therefore the addition of FA in ECC still leads to higher porosity in the matrix. Table 2: Air-void parameters of ECC mixtures. ECC1 ECC2 ECC3

PVA PVA PVA

PVA PVA

PVA Fresh air content (%) 7.3 5.9 7.1 6.2 6.9 6.4 Hardened air content (%) 8.2 7.3 8.9 7.9 9.4 8.3 Specific surface (mm-1)* 25.6 53.0 39.5 91.3 44.1 110.2 Spacing factor (mm)* 0.241 0.129 0.149 0.069 0.132 0.057 Average chord length (mm) 0.156 0.125 0.101 0.117 0.088 0.110 *:For freeze/thaw resistant concrete, the American Concrete Institute (ACI) recommends that [14]: Min. specific surface = 24 mm-1, and Maximum spacing factor = 0.2 mm. Except for the spacing factor value of ECC1, air-void parameters of all hardened ECC and ECC matrix mixtures conform to specifications ASTM C 457 for frost durability. In the case of hardened ECC mixtures, it is interesting to note that the addition of the PVA fiber can result in a significant increase in spacing factor (see Table 2). Moreover, specific surface of the air void system is significantly higher for ECC matrixes (w/o PVA fiber) when compared with ECC mixtures, which indirectly implies that the average bubble size is smaller in ECC matrices. This is likely due to the fact that the randomly distributed PVA fibers form a network that provides a path for the air bubbles to coalesce, thus creating large entrapped air voids instead of finely distributed air voids (Figure 1). Other than this difference, the air-void distributions are similar in ECC and ECC matrix mixtures. This is not in agreement with the results of the frost durability study described in forthcoming sections which concludes that ECC and ECC matrix mixtures essentially behave differently when subjected to freezing and thawing cycles.

226



Figure 1: Typical air voids in a non-air-entrained ECC and matrix specimens.

3.2 The addition of FA and micro-PVA fiber on freeze-thaw durability of ECC and ECC matrix mixtures were assessed by the calculation of mass loss, which is a measure of scaling. To measure the internal damage caused by freezing and thawing cycles, the changes in pulse velocity through a prism were also measured.

Figure 2: Relative pulse velocity and mass loss changes as a function of number of freezing and thawing cycles.

Figure 2 shows the reduction in the average relative pulse velocity (Vi/V0) and relative mass change (Mi/M0) with the number of freezing and thawing cycles. Vi and Mi are the pulse velocity and mass, respectively, after a specific number of freezing and thawing cycles, and V0 and M0 are initial pulse velocity and mass, respectively, prior to any freezing and thawing cycles. As seen in Figure 2, the mass and pulse velocity losses of ECC matrices increase with the number of freeze-thaw cycles and FA content. This trend is similar to but less severe than that for the corresponding ECC mixtures. Figure 2 indicates that a higher amount of air content and significantly lower spacing factor did not benefit ECC matrices in meeting the minimum durability factor requirement of ASTM C666. After 60 cycles, ECC2 and ECC3 matrix and after 210 cycles, ECC1 matrix specimens had severely deteriorated, requiring

227



removal from the freeze-thaw machine (typical normal strength, non-air-entrained concrete prisms usually fail in 60 to 95 cycles in this apparatus). Figure 2 demonstrates that, ECC mixtures showed excellent performance when exposed to freezing and thawing cycles, even after 300 cycles. The addition of PVA fiber to ECC matrices improved the freeze-thaw resistance considerably; and the improvement was more than enough to make the mixture acceptable according to ASTM C666, Procedure A. Note that at the end of 300 cycles, ECC2 and ECC3 have a lower relative pulse velocity and mass loss values than ECC1 at the same number of freeze-thaw cycles though ECC2 and ECC3 had a higher hardened air content and significantly lower spacing factor than ECC1. Therefore, it is not possible on this figure to determine the critical spacing factor for ECC with high volume FA, since they were all found to be frost resistant. In general, ECC specimens exhibited some surface scaling at the conclusion of the freeze-thaw cycling. Reduced surface scaling was observed on the ECC1 specimens compared to ECC2 specimens. Compared to ECC1 and ECC2, surface scaling was much more evident for ECC3 specimens. This was probably due to the greater matrix density (higher compressive strength at the time of testing; see Table 1) of ECC1 and ECC2 specimens because of lower FA content. The scaling was, however, clearly confined to the surface layers of the test specimens, and had no effect on the integrity and mechanical properties of the ECC mass (see Figure 2). A number of reasons could be responsible for excellent frost performance of non-air entrained ECC samples. A proper air-void system is needed in normal concrete to avoid internal cracking due to freezing and thawing. The greater frost resistance of ECC mixtures could be due to larger amounts of coarse pores in ECCs compared to matrix mixtures [15]. Another possible reason can be attributed to high ductility under tensile loadings. It is well known that upon freezing, water in capillary pores expands. If the required volume is greater than the available space, the pressure build-up could reach the tensile strength of material, resulting in local microcrack formation, brittle rupture and scaling. Thus, high tensile strength (fracture resistance) of ECC could lead to higher frost resistivity. When PVA fiber is used in ECC matrix, both the pressure-releasing effect (due to larger pore sizes) and the crack-resisting effect contribute to the ability to resist disintegration during freezing and thawing cycles. 3.3 A summary of the flexural test results before and after freezing and thawing deterioration is given in Table 3. Typical flexural stress – mid-span beam deflection curves of ECC and ECC matrix (without fiber) specimens before and after freezing and thawing deterioration are shown in Figure 3. As seen in Table 3 before freezing and thawing. All ECC mixtures show more than two times higher ultimate flexural strength than that of ECC matrices. This may be due to the fact that micro fibers inhibit the localization of microcracks into macrocracks and consequently, the flexural strength of the ECC matrix increases with the formation of multiple microcracks during inelastic deformation. In all of the ECC mixtures with/without freezing and thawing deterioration, prismatic specimens showed multiple cracking behaviors with small crack spacing and tight crack widths (less than 70 m). On the other hand, because of their low

228



tensile properties and brittle nature, the ECC matrix beam specimens failed catastrophically with a single crack under the four-point bending test. Table 3: Flexural properties of ECC prisms before and after 300 freeze-thaw cycles.

-

ECC1 w/o fiber 0.68 4.42 - with fiber 5.23 11.44 ~61



ECC1 with fiber 4.91 (93.9)* 9.70 (84.8) ~70 ECC2 with fiber 4.88 (81.2) 6.93 (74.3) ~78

ECC3 with fiber 4.57 (71.2) 5.12 (71.5) ~52 *Numbers in parentheses are percent residual flexural properties (deflection or strength).

Figure 3: Effect of freezing and thawing cycles on flexural behavior of ECC mixtures.

The complete flexural stress – mid-span deflection curves for ECC specimens before and after 300 freezing and thawing cycles are shown in Figure 3. The effect of 300 cycles of freezing and thawing was evident on the stress-strain curve, especially for ECC2 and ECC3 specimens. Compared to control specimens cured in laboratory air, the ascending portion of the flexural stress - deflection diagram indicated less stiffness (the slope of the load-deflection curve) with a smaller first crack and maximum stress, and smaller inelastic deformation. The typical flexural stress-mid-span beam deflection curves of ECC specimens after frost deterioration show that the influence of 300 freezing and thawing cycles on the flexural stress-mid-span deflection curves is minor especially for ECC1 specimens. The ultimate mid-

229



span beam deflection capacity of ECC specimens, which reflects material ductility, exposed to frost deterioration is 4.91 mm for ECC1 (93.9% of its original ductility), 4.88 mm for ECC2 (81.2% of its original ductility) and 4.57 mm for ECC3 (71.2% of its original ductility). The 4.0 mm deformation is nearly equivalent to 2% strain on the tensile face of the beam. Increasing levels of FA lead to higher reductions in the tensile strain capacity compared with standard ECC mixture (ECC1). This result is consistent with the earlier results of mass and pulse velocity losses of ECC specimens after 300 freezing and thawing cycles. The reduction in flexural performance of ECC mixtures may be attributed to the effects of damage on the fiber/matrix interface and matrix micro-cracking. Despite a reduction in stiffness, ductility and flexural strength, all ECC mixtures after 300 cycles of freezing and thawing exposure are found to retain ductility and flexural strength significantly more than that of normal concrete and FRC with no environmental exposure. For this reason, it is expected that the ECC mixtures investigated in this study are suitable for long-term application under severe freeze-thaw environments if the structure is designed for long term mechanical properties. It should also be noted that the residual ultimate flexural load-deflection curves of frost deteriorated ECC beams obtained following induced accelerated freeze-thaw cycling (up to six freezing and thawing cycles were achieved in a 24-hour period) provide a conservative estimate of their residual flexural properties in actual structures. These accelerated deterioration periods are equivalent to a time span of many years in real structures, even those located in regions with harsh winters. This difference in accelerated and normal frost deterioration periods should have a significant influence on the residual flexural properties of ECC because in the long term, deterioration in ECC as a result of freezing and thawing cycles can easily be closed due to a self-healing process [7]. By neglecting this self-healing capability of ECC, thus, a conservative estimation for the flexural performance of the material is presented. 4. The objective of this research is to assess the effect of fly ash (FA) and micro poly-vinyl-alcohol (PVA) fibers (ductility) on the microstructure and frost durability of the non-air-entrained ECCs. Following conclusions were drawn based on the experimental studies: Although the control ECC matrix mixtures specimens rapidly failed in freezing and

thawing cycles (after 60 cycles, the ECC2 and ECC3 matrix (FA/C=2.2 and 4.2) specimens and after 210 cycles, the ECC1 matrix (FA/C=1.2)), both ECC mixtures showed excellent performance when exposed to freezing and thawing cycles, even after 300 cycles. The addition of PVA fiber to ECC matrixes improved the freeze-thaw resistance considerably.

Increasing levels of FA leads to higher reductions in the mechanical performances compared with standard ECC mixture (ECC1) due to the lower matrix density – lower compressive strength at the time of testing. Apart from the slight reductions in ductility and strength capacities and higher residual crack width, the results presented in this study largely confirm the durability performance of ECC material incorporating high volume of FA under frost exposure.

The presence of micro synthetic PVA fibers critically contributes to the higher crack resistance and larger pore volume, and resulting pressure-releasing effects under freeze-

230



thaw conditions. In addition to superior ductility and high tensile strength properties, a large pore volume in the presence of a PVA fiber is assumed to be responsible for the improved frost resistance of ECC mixtures. Recall that this high frost durability was achieved without deliberate air entrainment into the ECC. Normally, concrete durability remains very sensitive to the amount of air entrainment, air-void parameters, strength and microstructural density of concrete. Since this sensitivity can be overcome through an ECC material solution even with high volume fly ash and without air-entraining admixture, the ultimate result will be overall improved structural durability.

[1] Li, V. C., On engineered cementitious composites (ECCs): a review of the material and its applications, J Adv Concr Technol 1 (2003), 215-230.

[2] Lin, Z. et al, On interface property characterization and performance of fiber reinforced cementitious composites, Concr Sci Eng 1 (1999), 173-184.

[3] and Li, V. C., Durability properties of micro-cracked ECC containing high volumes fly ash, Cem Concr Res 39 (2009), 1033-1043.

[4] Li, V. C. et al, Tensile strain-hardening behavior of PVA-ECC, ACI Mater J 98 (2001), 483-492.

[5] Li, V. C. et al, Matrix design for pseudo strain-hardening fiber reinforced cementitious composites, Mater Struct 28 (1995), 586-595.

[6] Soroushian, P. et al, Optimization of the use of lightweight aggregates in carbon fiber reinforced cement, ACI Mater J 89 (1992), 267-276.

[7] ahmaran, M. et al, Self-healing capability of cementitious composites incorporating different supplementary cementitious materials, Cem Concr Compos 35 (2013), 89-101.

[8] Wang, S. and Li, V. C., Engineered cementitious composites with high-volume fly ash, ACI Mater J 104 (2007), 233-241.

[9] Lepech, M. D. and Li, V. C., Long term durability performance of engineered cementitious composites, Int J Restor Build Monument 12 (2006), 119-132.

[10] Yang, E. et al, Rheological control in the production of engineered cementitious composites, ACI Mater J 106 (2009), 357-366.

[11] Pigeon, M. and Pleau, R., Durability of concrete in cold climates, London: Chapman & Hall, (1995). [12] Powers, T. C., Frost resistant concrete, J PCA Res Dev Lab (1964), 6-19. [13] Feldman, R. F., Influence of condensed silica fume and sand/sement ratio on pore

structure and frost resistance of portland cement mortars, Amer Conc I 91 (1986), 973-989.

[14] Siebel, E., Air-void characteristics and freezing and thawing resistance of superplasticized air entrained concrete with high workability, Amer Conc I 119 (1989), 297-320.

[15] et al, Frost resistance and microstructure of engineered cementitious composites: Influence of fly ash and micro poly-vinyl-alcohol fiber. Cem Concr Compos 34 (2012), 156-165.

231



WATER PENETRATION INTO FROST DAMAGED CONCRETE Peng Zhang (1), Yuan Cong (1), Wanyu Zhao (2), Wenchao Geng (1), Zhengzheng Dai (1), Tiejun Zhao (1),(3)

(1) Department of Civil Engineering, Qingdao Technological University, Qingdao 266033, PR China (2) Department of Civil Engineering, Ocean University of China, Qingdao 266000, PR China (3) Center for Durability & Sustainability Studies of Shandong Province, Qingdao 266033, PR China Abstract

In service life, concrete can be damaged either by mechanical or environmental loads or by combined ones. These damages will strongly influence water movement in concrete which could later lead to more serious deteriorations. This paper used neutron radiography to investigate the influence of frost damage on water penetration into concrete. In addition, the improvement of frost resistance by addition of air entrainment was investigated. The results indicate that it is possible to visualize penetration of water into the porous structure of concrete by neutron radiography. Further evaluation of the test data allows determining time-dependent moisture profiles quantitatively with high resolution. After concrete is damaged by freeze-thaw cycles water penetration into ordinary concrete is accelerated. It can be shown that frost damage is not equally distributed in specimens exposed to freeze-thaw cycles. Thermal gradients lead to more serious damage near the surface. The beneficial effect of air entrainment on frost resistance has been demonstrated. After 50 freeze-thaw cycles, air entrained concrete showed no measurable increase in water absorption. But layers near the surface of concrete absorbed slightly more water after 200 freeze-thaw cycles although the dynamic elastic modulus remained constant. Results presented in this paper help us to better understand mechanisms of frost damage of concrete. 1. Introduction Cracks are always preferential paths for water flow. Cracks may be caused by mechanical load or by thermal and hygral gradients or by local swelling processes. Many authors have studied the influence of cracks on penetration of water and aggressive compounds into

232

cement-based materials in the past [1-8]. Jacobsen et al. [1] studied the effect of freeze-thaw cycles on chloride transport into concrete and they found that internal cracking increased the chloride penetration rate by a factor of 2.5 to 8 when compared with undamaged specimens. Yang et al. [5] studied water transport in concrete damaged both by tensile loading and frost cycles and reported that the presence of freeze-thaw damage increased both the initial sorptivity and total water absorption of concrete.

It is obvious that if one uses material properties measured on undamaged concrete in prediction models the performance of a structure in aggressive environment may be overestimated. There is an urgent need to modify existing prediction models in such a way that they can take into account mass transport properties modified by damage due to mechanical or environmental loads. But necessary experimental data to feed this new generation of models are scarce. By means of neutron radiography it is possible to visualize and quantify water penetration into concrete. Up to now, this technique has been successfully applied already to study water movement in different porous building materials such as concrete, mortar, stone, and bricks by several researchers [9-14].

The main aim of this paper is to investigate the influence of freeze-thaw cycles on water penetration into concrete. To which extend is capillary water absorption increased by damage induced by frost action? From the raw data obtained by neutron radiography time-dependent spatial water distributions in concrete during water penetration can be determined in a quantitative way. Frost damage usually is assumed to be equally distributed in the volume. It should be possible to observe additional damage near the surface due to hygral gradients during cooling. All results obtained will be presented and discussed. 2. Materials and Methods 2.1 Materials and Preparation of Test Specimens Prismatic specimens with the following dimensions were prepared with two types of concrete, both with water cement ratio of 0.6: 100 × 100 × 400 mm. Ordinary Portland cement 42.5, local crushed aggregates with a maximum diameter of 20 mm, river sand with a maximum grain size of 5 mm, all from Qingdao area, were used. Part of the specimens was produced with the addition of 0.017 % of an air entraining agent related to the mass of cement into the fresh concrete to improve its frost resistance. The exact compositions of the two types of concrete C and CA used in this project, and their air content and compressive strength at an age of 28 days are given in Table 1. Table 1: Composition, air content and compressive strength of two types of concrete used in this project.

Type Cement (kg/m3)

Sand (kg/m3)

Gravel (kg/m3)

Water (kg/m3)

Air entraining agent (g/m3)

Air content (%)

28-day compressive

strength (MPa) C 300 699 1191 180 — 2.0 32.2

CA 300 699 1191 180 51 5.2 28.9



233

All specimens were demoulded after one day and then stored in a moist room at a temperature of 20 ± 3 and relative humidity higher than 95 % until an age of 24 days. At that time all specimens were taken out of the curing room and further stored in water for another four days. At an age of 28 days, specimens were then ready for the following frost test. 2.2 Freeze-thaw cycles After curing concrete specimens were exposed to freeze-thaw cycles following a Chinese standard method [15]. The samples were exposed to pure water during the freeze-thaw cycles. In this case one freeze-thaw cycle lasts for about three hours. The temperature at the centre of specimens varied between -17 ± 2 oC and 5 ± 2 oC. On concrete samples without addition of air entraining agent (Concrete C), the dynamic elastic modulus has been measured after 10 and 25 freeze-thaw cycles. On concrete samples prepared with addition of air entraining agent (Concrete CA), the dynamic elastic modulus has been measured after 10, 50, 100, 150, 200, 250 and 300 freeze-thaw cycles. In the following, specimens which had suffered freeze-thaw cycles are identified by the appropriate concrete type followed by the number of frost cycles, as for instance, the concrete with air entrainment exposed to 200 freeze-thaw cycles will be designated “CA-200”. As can be expected concrete made with air entraining agent has a significantly improved frost resistance. Change of the related dynamic elastic modulus of both types of concrete, without and with air entraining agent, as function of number of frost cycles is shown in Figure 1.

0 50 100 150 200 250 3000

20

40

60

80

100

Relat

ed d

ynam

ic ela

stic

mod

ulus

/ %

Number of freeze-thaw cycles

C CA

Figure 1 Related dynamic elastic modulus of two types of concrete C and CA as function of

the number of freeze-thaw cycles

After exposure to a certain number of freeze-thaw cycles, selected specimens were taken out of the frost testing machine and cut in parts for water penetration tests. First a centre cube with dimensions with an edge length of 100 mm was cut off. Then two opposite layers with a thickness of 25 mm each were cut off the cube. Finally the remaining block which had the following dimensions: 100 mm × 50 mm × 100 mm has been separated into five thin slices with a thickness of approximately 20 mm. Slices from the surface were designated as -1 , the intermediate slices with -2 and the centre slice with -3 . The schematic representation for the cutting scheme of the samples is shown in Fig. 2. The test sample were then dried in a ventilated oven at 50 for four days until constant weight was achieved. Then four surfaces were covered with self-adhesive aluminum foil. Two opposite surfaces with the dimensions 20 mm × 100 mm remained free for the capillary absorption test. After the first image had been taken by means of neutron radiography in the dry state of the



234

specimens the aluminum container was filled with water. At this moment water started to penetrate into the frost damaged concrete.

Fig. 2. Schematic representation for cutting scheme of test specimens.

2.3 Neutron radiography All tests of neutron radiography were performed at the thermal neutron radiographic facility called NEUTRA of Paul Scherrer Institute (PSI) in Switzerland [10]. After the reservoir was filled with water neutron images were serially taken every twenty seconds for up to four hours. In order to visualize more clearly the process of water penetration into concrete, differential images from any time related to the initial time were then processed. For the quantitative evaluation of the digitized neutron radiographs, software programs QNI and IDL were utilized. More details about quantitative calculation of water content in porous materials can be found in references [9,11,16]. The results shall be further analyzed. However they can be used to indicate the difference of different concrete in this paper. 3. Results and Discussion 3.1 Water Penetration into Concrete without Air Entrainment after Frost Damage Neutron images of water penetration into concrete without air entrainment and without frost action are shown in Figure 3. It can be seen immediately that by means of neutron radiography we can follow qualitatively the process of water penetration into porous cement-based materials. After a contact time of about five minutes the penetrating water front becomes visible. Then this irregular front gradually moves deeper into concrete with increasing contact time. The aggregates of the composite material are marked with the lighter areas as they do hardly absorb water. The water moves around the aggregates in the porous cement-based matrix. The time-dependent moisture profile along a vertical axis in the marked rectangular area as shown in Figure 3 can be analyzed quantitatively. Obtained results are also shown in Figure 3.

100 × 100 × 400 mm3 100 × 100 × 100 mm3 100 × 50 × 100 mm3 100 × 50 × 20 mm3

-1 -2

-3



235

Figure 3 Neutron images of water penetration into neat concrete without air entrainment before frost damage after different durations of contact with water and the corresponding

moisture distribution along the rectangular are in concrete After exposure of the initial concrete prisms to 10 freeze-thaw cycles, concrete slices were cut off at different distances from the surface of the damaged specimens as described above. The process of water penetration into these samples was then measured. A few selected neutron images are shown in Figure 4 for the three different slices. It can be seen that in comparison with results shown in Figure 3 water penetrates deeper into the frost damaged concrete at the same time of contact. Quite obviously freeze-thaw damage increased the rate of water absorption into concrete. This has been observed by Yang et al. before [5]. The corresponding water distribution profiles in different layers from the surface to centre of concrete were calculated and the resulting profiles are shown in Figure 3. It can be clearly seen that the closer a sample was positioned to the concrete surface, the deeper water penetrates. This is a clear indication that there exists a damage gradient. And damage decreases with increasing distance from the surface.

Figure 4 Water penetration into layers of concrete C, having different distances from the

surface, after 10 freeze-thaw cycles after 60 and 240 minutes of contact with water, and the moisture distributions along the rectangular area on the images.

3.2 Water Penetration into Air Entrained Concrete after Frost Damage Direct observation of water penetration into concrete prepared with air entraining agent by neutron radiography is shown in Figure 5 and Figure 6. The water front moves into this type of concrete with slightly lower rate as compared with normal concrete (see Figure 3). This tendency has been observed before by ordinary capillary suction tests and it can be explained by the fact that the artificially introduced spherical air pores break the capillary force locally. The artificial air pores will be filled with water very slowly and this is probably the reason

-0.005

0.010

0.025

0.040

0.055

0 5 10 15 20 25 30 35 40 45Depth / mm

5 min 15 min 30 min60 min 120 min 240 min

-0.005

0.010

0.025

0.040

0.055

0 5 10 15 20 25 30 35 40 45Depth / mm

C10-1, 60 min C10-1, 240 minC10-2, 60 min C10-2, 240 minC10-3, 60 min C10-3, 240 min

Wat

er c

onte

nt in

sam

ple

Wat

er c

onte

nt in

sam

ple



236

why we observe two levels of water content in the distribution curves. The pores, which remain empty for a long time even if concrete is in contact with water, assure increased frost resistance. Water in the smaller pores in hardened cement paste around the comparatively big artificial pores is under high capillary under-pressure and therefore it cannot enter the bigger spherical pores.

Figure 5 Water penetration into layers of concrete CA, having different distances from the surface, after 50 freeze-thaw cycles

Figure 6 Water penetration into layers of concrete CA, having different distances from the surface, after 200 freeze-thaw cycles

The obtained digital images on concrete type CA after 50 and 200 freeze-thaw cycles were also evaluated. Results along the vertical direction of the rectangular area marked in the images are shown in Figure 7. Water profiles measured in concrete type CA from the surface to the centre after 50 freeze-thaw cycles are nearly the same as results found in specimens without frost action. Even after 200 freeze-thaw cycles, water profiles have not strongly increased except for the surface near layer.



237

Figure 7 Moisture profiles in layers of concrete CA, having different distances from the

surface, after 50 and 200 freeze-thaw cycles after 60 and 240 minutes of contact with water.

4. Conclusions Investigations into the water penetration into undamaged and frost damaged concrete by eutron radiography are described in this paper. Based on the results obtained, the following conclusions can be made: (1) Water penetration into ordinary concrete is increased significantly by frost damage. From the surface to the centre of the tested concrete samples, the rate of water penetration decreases. A damage gradient can be observed. (2) After 50 freeze-thaw cycles, water penetration into air entrained concrete did not vary significantly as compared to values obtained on undamaged concrete. After 200 freeze-thaw cycles, however, the surface near zone absorbed more water although the dynamic elastic modulus as measured on the bulk material has not decreased. Damage concentrated into a thin surface near layer. (3) In addition to the common frost damage induced by freezing of water in narrow gaps and in nano-pores, differential deformations between hardened cement paste and aggregates may also induce damage into the interfaces. References [1] Jacobsen, S., Marchand, J. and Boisvert, L. Effect of cracking and healing on chloride

transport in OPC concrete. Cement and Concrete Research, 26(6) (1996), 869-881. [2] Aldea, C. -M., Shah, S. P., and Karr, A. F. Permeability of cracked concrete. Materials

and Structures, 32(6) (1999), 370-376. [3] Win, P. P., Watanabe, M. and Machida, A. Penetration profile of chloride ion in cracked

reinforced concrete. Cement and Concrete Research, 34 (7) (2004), 1073-1079. [4] Kato, E., Kato, Y. and Uomoto, T. Development of simulation model of chloride ion

transportation in cracked concrete. Journal of Advanced Concrete Technology, 3(1) (2005), 85-94.

[5] Yang, Z. F., Weiss, W. J. and Olek, J. Water transport in concrete damaged by tensile



238

loading and freeze-thaw cycling. Journal of Materials in Civil Engineering, 18(3) (2006), 424-434.

[6] Wang, L. C., Soda, M. and Ueda, T. Simulation of chloride diffusivity for cracked concrete based RBSM and truss network model. Journal of Advanced Concrete Technology, 6(1) (2008), 143-155.

[7] Kanematsu, M., Maruyama, I., Noguchi, T., Iikura, H. and Tsuchiya N. Quantification of water penetration into concrete through cracks by neutron radiography. Nuclear Instruments and Methods in Physics Research Section A, 605(1-2) (2009), 154-158.

[8] Picandet, V., Khelidj, A. and Bellegou H. Crack effects on gas and water permeability of concretes. Cement and Concrete Research, 39(6) (2009), 537-547.

[9] Pleinert, H., Sadouki, H. and Wittmann, F. H. Determination of moisture distributions in porous building materials by neutron transmission analysis. Materials and Structures, 31(4) (1998b), 218-224.

[10] Lehmann, E. H., Kuhne, G., Vontobel, P. and Frei, G. The NEUTRA and NCR radiography stations at SINQ as user facilities for science and industry. In: P. Chirco and R. Rosa, Eds. Proc. 7th World Conference of Neutron Radiography, Rome, Italy, Aedificatio Publishers, (2002), 593-602.

[11] Hassanein, R., Meyer, H. O., Carminati, A., Estermann, M., Lehmann, E. H. and Vontobel, P. Investigation of water imbibition in porous stone by thermal neutron radiography. Journal of Physics D: Applied Physics, 39(19) (2006), 4284-4291.

[12] Cnudde, V., Dierick, M., Vlassenbroeck, J., Masschaele, B., Lehmann, E., Jacobs, P. and Hoorebeke, L. V. High-speed neutron radiography for monitoring the water absorption by capillarity in porous materials. Nuclear Instruments and Methods in Physics Research Section B, 266(1) (2008), 155-163.

[13] Abd, A. El, Czachor, A. and Milczarek, J. Neutron radiography determination of water diffusivity in fired clay brick. Applied Radiation and Isotopes, 67(4) (2009), 556-559.

[14] Zhang, P., Wittmann, F. H., Zhao, T. J. and Lehmann, E. H. Neutron imaging of water penetration into cracked steel reinforced concrete. Physica B: Condensed Matter, 405(7) (2010), 1866-1871.

[15] Ministry of Housing and Urban-Rural Development of China. GB/T 50082—2009, Standard for Test Methods of Long-term Performance and Durability of Ordinary Concrete. Beijing: China Architecture & Building Press, 2010.

[16] Zhang, P., Wittmann, F. H., Zhao, T. J., Lehmann, E., Vontobel, P. and Hartmann, S. Observation of water penetration into water repellent and cracked cement-based materials by means of neutron radiography. Int. J. Restoration of Buildings and Monuments, 15(2) (2009), 91-100.



239



Author Index Einar N. Andreassen 1 Sofía Aparicio 11 Vahid Jafari Azad 171 Nele De Belie 141 Dimitrios Boubitsas 21 Lado Bras 121 Mats Buøen 131 Xiaoping Cai 211 Hunter Carolan 171 Yuan Cong 231 Zhengzheng Dai 231 Andreas B. Elbrønd 1 Oskar Esping 91 Miguel Ferreira 31 Katja Fridh 111 161 Dariusz Gawin 101 Yong Ge 211 Wenchao Geng 231 Marianne Tange Hasholt 1 41 Manouchehr Hassanzadeh 111 Philip Van den Heede 141 Elisabeth Helsing 21 51 Margarita G. Hernández 11 Jason H. Ideker 171 Shin-Ichi Igarashi 81 O. Burkan Isgor 171 Stefan Jacobsen 61 71 Samdar Kakay 131 Vikrant Kaushal 131 Hidefumi Kotou 81 Hannele Kuosa 31 Mohamed Lachemi 221 David Lange 31 Markku Leivo 31 Chang Li 171 Anders Lindvall 91 Øyvind O. Lødemel 61 Ingemar Löfgren 91 Takuma Murotani 81 Henrik Nordahl-Pedersen 61 Francesco Pesavento 101 Robert K. Prud'Homme 191 Javier Ranz 11 Hawar Omer Rasol 61 Martin Rosenqvist 111 Mustafa Aljoša Šajna 121

Naomi Salgado 171 Samindi Samarakoon 131 George W. Scherer 61 71 191 Didier Snoeck 141 Frank Spörel 151 Martin Strand 161 Prannoy Suraneni 171 Luping Tang 21 Pål Lieske Tefre 131 Michael Thomas 181 Lori E. Tunstall 61 191 Peter Utgenannt 21 51 José Javier Anaya Velayos 11 Zhendi Wang 201 Ling Wang 201 Jason Weiss 171 Wencui Yang 211 Yan Yao 201 Huang Yi 181

Jie Yuan 211 Peng Zhang 231 Wanyu Zhao 231 Tiejun Zhao 231

Koto 81


Frost Action in ConcreteEdited byMarianne Tange Hasholt, Katja Fridh and R. Doug Hooton

RILEM Proceedings PRO 114ISBN: 978-2-35158-182-7e-ISBN: 978-2-35158-183-4

This volume contains the proceedings of the MSSCE 2016 conference segment on “Frost action in concrete”. Despite research in this field has been ongoing since the 1930’es, the mechanism(s) leading to frost damage is not fully understood. Therefore, there is still a need for both basic research and practical solutions to the challenges encountered in the field.

The present conference segment comprises 24 papers from all over the northern hemisphere. Within the overall theme “Frost action in concrete”, the contributions deal with many different topics, for example: the relation between mix design and frost resistance, modelling of frost action, combined action when concrete is exposed to freeze/thaw load together with other types of load, air void analysis, novel non-destructive test methods, and experience gained from monitoring of structures as well as from field exposure sites.

The event “Materials, Systems and Structures in Civil Engineering 2016” 15-29 August 2016, Lyngby, Denmark, is scientifically sponsored by RILEM. The event is hosted by the Department of Civil Engineering at the Technical University of Denmark and is financially sponsored by a number of independent foundations and organizations.

RILEM Publications S.a.r.l.157 rue des Blains, F-92220 Bagneux - FRANCETel: +33 1 45 36 10 20 Fax: +33 1 45 36 63 20E-mail: [email protected]

Frost Action in Concrete -...

Documents

Transcript of Frost Action in Concrete -...