European Commission

Research Fund for Coal and SteelHigh quality acoustic and vibration

performance of lightweight steel constructions

J. Kesti (1), S. Hicks, J. Rackham (2), J. Widman (3), M. Villot, C. Guigou (4), A. Rodríguez-Ferran, J. Poblet-Puig (5), P. Sipari, A. Talja (6),

F. Ljunggren, A. Ågren (7)

(1) Ruukki — Laajamäentie 1, FI-13430 Hämeenlinna(2) SCI — Silwood Park, Ascot SL5 7QN, United Kingdom

(3) SBI — Banérgatan 54, S-115 92 Stockholm(4) CSTB — 84, avenue Jean Jaures, Champs sur Marne, F-77420 Marne-la-Vallée Cedex 2

(5) UPC — Calle Jordi Girona 31, E-08034 Barcelona(6) VTT Technical Research Centre of Finland — PO Box 1000, FI-02044 VTT

(7) Luleå University of Technology — S-971 87 Luleå

Contract No RFSR-CT-2003-00025 1 September 2003 to 31 December 2006

Final report

Directorate-General for Research

2008 EUR 23319 EN

LEGAL NOTICE

Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of the following information.

A great deal of additional information on the European Union is available on the Internet. It can be accessed through the Europa server (http://europa.eu). Cataloguing data can be found at the end of this publication. Luxembourg: Office for Official Publications of the European Communities, 2008 ISBN 978-92-79-08304-4 ISSN 1018-5593 © European Communities, 2008 Reproduction is authorised provided the source is acknowledged. Printed in Luxembourg PRINTED ON WHITE CHLORINE-FREE PAPER

Europe Direct is a service to help you find answers to your questions about the European Union

Freephone number (*):

00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.

CONTENTS

FINAL SUMMARY 7 Introduction and objectives 7 Requirements and testing procedures (WP1) 7 Acoustic modelling and testing (WP2 + WP4) 8 Vibration modelling and testing (WP2 + WP5) 10 Product development (WP3) 12 Design Guide (WP6) 13

SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS 15

1 OBJECTIVES OF THE PROJECT 15

2 COMPARISON OF INITIALLY PLANNED ACTIVITIES AND WORK ACCOMPLISHED 15

3 REQUIREMENT AND TESTING PROCEDURES FOR ACOUSTIC AND VIBRATION PERFORMANCE OF STRUCTURES (WP1) 17 3.1 Introduction and objectives 17 3.2 Current and future requirements for acoustic performance of structures (Task

1.1) 17 3.2.2 Special consideration of traffic noise 19 3.2.3 Relevant International Standards for acoustics 19 3.2.4 Glossary of acoustic terms 21

3.3 How the requirements are met with current solutions (Task 1.2) 22 3.4 Harmonisation of the testing procedure for vibration performance (Task 1.3) 24

3.4.1 Modal testing 24 3.4.2 Dynamic performance assessment 25 3.4.3 Subjective evaluation 25

3.5 Harmonisation of the rating of the annoyance of vibrations (Task 1.4) 26 3.5.1 Review on current practice 26 3.5.2 Harmonisation of criteria for the acceptability of vibrations 31

4 ACOUSTIC MODELLING AND TESTING 35 4.1 Component level 35

4.1.1 Introduction 35 4.1.2 Modeling 35 4.1.3 Testing 42 4.1.4 Calibration/validation and comparison between models 47 4.1.5 Parametric studies and conclusions 50

4.2 Whole building level, including junctions 52 4.2.1 Introduction 52 4.2.2 Modeling 52 4.2.3 Testing 57

4.3 Conclusion 65 4.3.1 Acoustic performance of lightweight building elements 65 4.3.2 Acoustic performance of lightweight buildings 66

4.4 Summary table of the acoustic tests performed 67

5 VIBRATION MODELLING AND TESTING 69 5.1 Introduction and objectives 69 5.2 Modeling and simulation of vibration performance of structures (Task 2.2) 69

5.2.1 FE - Model A 70 5.2.2 FE - Model B - including orthotropic elements 71 5.2.3 Prediction of walking response 72

5.3 Laboratory tests on the vibration performance of lightweight floors (Task 5.1) 74 5.3.1 Dynamic test 74

3

5.3.2 Subjective evaluations 75 5.3.3 Analysis 77

5.4 Field tests on the vibration performance of lightweight floors (Task 5.1) 78 5.4.1 Brixton 78 5.4.2 Derby 79 5.4.3 Daventry 80 5.4.4 Kauklahti 82

5.5 Vibration models in comparison with experimental results (Task 2.4) 85 5.6 Role of connections and boundaries of a lightweight floor (Task 5.2) 88

5.6.1 Line loads with different mountings 88 5.6.2 Mass loaded support 91 5.6.3 Reinforced support 92 5.6.4 The connections’ relation to a real building 92

5.7 Subjective assessment of floor vibrations with a motion simulator (Task 5.3) 93 5.7.1 Experimental set-up 94 5.7.2 Study I: Threshold values from single frequency vibrations 94 5.7.3 Study II: Threshold values for a second frequency component 95 5.7.4 Study III: Annoyance of dual sinusoidal vibrations 96 5.7.5 Study IV: Annoyance and acceptance of multi-frequency vibrations 97 5.7.6 Further analysis and prediction model 100

5.8 Discussion and conclusions 104 5.8.1 Horizontal vibrations 104 5.8.2 Parameters affecting the floor response 104 5.8.3 Vibration design criteria 104 5.8.4 Multiple frequency vibrations 105

6 PRODUCT DEVELOPMENT 107 6.1 General objectives 107 6.2 Façade and outer wall structures (Task 3.1) 107

6.2.1 Requirements for facades 107 6.2.2 Development process 107

6.3 Floors (Task 3.2) 111 6.3.1 Requirements for floor structures 111 6.3.2 Lightweight floors 111 6.3.3 Semi-heavy floors 114

6.4 Partition walls (Task 3.3) 115 6.4.1 Requirements for partition wall structures 115 6.4.2 Development process 116

6.5 Conclusions 119

7 DESIGN GUIDE FOR ACOUSTIC AND VIBRATION PERFORMANCE OF LIGHTWEIGHT CONSTRUCTION 121 7.1 Introduction and objectives 121 7.2 Design Guide Chapter 2: Typical Lightweight Steel Construction 122

7.2.1 Lightweight steel construction systems 122 7.2.2 Building parts 125

7.3 Design Guide Chapter 3: Acoustics 131 7.3.1 Component level 131 7.3.2 Whole building level (including junctions) 135 7.3.3 Case studies 137

7.4 Modelling methods 137 7.4.1 Wave approach combined with SEA 137 7.4.2 Finite element method 140

7.5 Design Guide Chapter 4: Vibrations 142 7.5.1 Introduction 142 7.5.2 Deflection-Based Model 143 7.5.3 Level 2: Acceleration-based method 144 7.5.4 Finite Element Modelling of Floor Vibrations 148

4

7.6 Design Guide Chapter 5: Recommendations for Design and Detailing 151 7.6.1 Acoustic performance of the structure and componenets 151 7.6.2 Designing for good floor dynamic performance 155

8 CONCLUSIONS 157

9 EXPLOITATION AND IMPACT OF THE RESEARCH RESULTS 161 9.1 Actual applications 161 9.2 Technical and economic potential for the use of the results 161 9.3 Patent filings 161 9.4 Publications resulting from the project 161 9.5 Dissemination of results 162

10 LIST OF FIGURES AND TABLES 163 10.1 List of figures 163 10.2 List of tables 168

11 REFERENCES 170

APPENDIX A TEST FLOOR DESCRIPTIONS 174 A.1 Laboratory tests 174 A.2 Field tests 176 A.3 Old laboratory tests used in verifications 179

5

FINAL SUMMARY

Introduction and objectives

The use of light steel and modular construction is increasing in housing, residential buildings and commercial buildings, where the benefits of lightweight, improved quality and speed of construction are realised. In these applications, the performance of the building in service is very important. These performance characteristics include many aspects, but the Acousvibra project concentrates on the following topics:

• Acoustic behaviour of lightweight structures

• Vibration performance of lightweight floors

The commercial objective of the project is to increase the use of the light gauge steel framed buildings especially in the urban areas, where noise insulation and vibration of structures are known to be a problem. The prime technical objective is to develop guidance for codified design rules, guidance on analysis methods and guidance on detailing. This will promote the use of light gauge steel construction with high quality vibration and sound insulation properties. The main scientific and technical objectives of the project are: 1) improved modelling of acousticl and vibration behaviour of lightweight structures, 2) better understanding of acoustic behaviour of lightweight structures in respect of flanking, 3) standardisation of testing methods for vibration of lightweight floors, 4) introduction of new construction methods and 5) introduction of a design guide for designers.

The work has been carried out in six different technical work packages:

• WP 1 Requirements and testing procedures for acoustic and vibration performance of structures • WP 2 Modelling and Simulation • WP 3 Product development • WP 4 Laboratory and field tests on the acoustic performance of structures • WP 5 Laboratory and field tests on the vibration performance of structures • WP 6 Design guide for acoustic and vibration performance of lightweight construction

The main activities and results are summarised below. Acoustic and vibration issues are considered separately: Chapter “Acoustic modelling and testing” includes the acoustic part of WP2 and WP4 as a whole, and chapter “Vibration modelling and testing” includes the vibration part of WP2 and WP5 as a whole. Otherwise, the reporting follows the work package sequence.

Requirements and testing procedures (WP1)

The general objectives of the WP1 were to carry out a comprehensive review of current and future sound insulation requirements. A study was carried out of the current acoustic requirements and design principles adopted for the design of façade structures. A further objective was to review how these acoustic requirements are met with current solutions and to consider the improvement possibilities. It was intended to harmonise the testing procedure for vibration performance of lightweight floors and to introduce a standard way of making vibration measurements on the basis of previous experience. The final objective in WP1 was to harmonise the rating of the annoyance of vibrations.

The acoustic requirements for dwellings in different European countries were collected and put in table format. It was found that they vary both in terms of measurement used, and in terms of absolute values. Study covered airborne sound between separating walls and floors, impact sound (for separating floors), and limits for the attenuation of external noise. Acceptability is expressed sometimes in laboratory

7

values, in field values, and in other cases by calculation. Spectrum adaptation terms are sometimes applied to cater for traffic or low frequency noise. Limits for the attenuation of external noise are expressed as a facade noise reduction in some countries, but in others as an absolute level in the dwelling. Current design practice to meet acoustic requirements often involves the application of standard solutions which have been developed and approved by testing. Typical methods for façade, roof, floor and wall construction in different countries were collected and presented in tables. The tables show, an image and description of the make up, and their acoustic performance.

One task in WP1 concentrated on harmonisation of testing procedures. The purpose of dynamic testing is to: 1) measure the actual dynamic responses of floors at the dynamic loading specified in the design, 2) post-process these responses in accordance with the specified national (or international) Standard to obtain a relevant response parameter, and 3) to rate this parameter against an acceptance criterion specified for the floor design. The type of dynamic testing is depending on the floor type and the measurement results needed. Methods for modal testing with or without measuring the excitation force are presented for heel drop tests and for mass shaker tests.

Although a floor’s modal properties are very important, its vibration performance depends most directly on the actual vibration response due to a given dynamic excitation. Therefore it is necessary to create the relevant dynamic excitation, then, after measuring and post-processing floor vibration responses due to the excitation, the obtained response parameter can be rated. For residential and office floors, it was considered that floor responses due to a single person walking should be measured, as this kind of excitation occurs frequently in floors and is difficult to isolate. Human sensitivity to vibrations is frequency dependent, and so a method of frequency weighting is employed to account for this. There are two parameters which are typically used in modern codes of practice for assessing the amount of vibration and its effects: these are the RMS acceleration, and the recently established vibration dose value (VDV). The VDV is a cumulative measure of the vibration transmitted to a human receiver during a certain period of interest.

The human body is a very sensitive vibration instrument, and many objects and some furniture are also very sensitive to floor vibrations. Human perceptions are often more authentic, and also more accurate, than many theoretical calculations. For that reason it was concluded that it is important to carry out subjective tests for new types of floors. During subjective tests, observers are asked to give their opinion of the intensity and acceptability of the vibrations induced by a walker. The harmonised test procedure to the subjective evaluations was proposed.

The final task, harmonisation of the rating of annoyance of vibrations, was prepared in parallel with work in other work packages, following laboratory and field tests and analysis. The task was challenging, because there was lack of specific information regarding lightweight floors. Furthermore, even new versions of the ISO 2631 Standard have been revised with a clause stating “Guidance values above which adverse comments due to building vibration could occur are not included anymore since their possible range is too widespread to be reproduced in an International Standard”. However, two different methods have been selected in the Acousvibra-project for the assessment of vibration acceptance of light-weight floors. The first method is based on deflection criteria and is valid for high-frequency floors with natural frequency over 9 to10 Hz. The second method is based on acceleration criteria and vibration dose values (VDV).

Acoustic modelling and testing (WP2 + WP4)

Acoustic modelling and testing correspond to tasks spread over two work-packages of the project: WP2 on modelling and WP4 on testing, with the aim of better understanding and mastering of both the acoustic performance of lightweight building elements, such as floors or walls, which were considered separately as well as collectively in the form of the whole building. The following tasks are considered here:

8

Task 2.1 Modelling and simulation of acoustic performance of structures Task 2.3 Calibration of acoustic models and comparison with experimental results Task 4.1 Laboratory acoustic tests Task 4.2 Field acoustic tests Task 4.3 Laboratory characterisation

Acoustic performance of lightweight building elements

The acoustic performance of building elements considered separately, and expressed in terms of the airborne sound reduction index R and impact sound level Ln, can be either measured in the laboratory or estimated using models. In this project, existing models were used, but also new models were developed in order to better understand the role of the different components (boards, cavity, studs, rails etc.) in sound transmission through lightweight building elements. Three different modelling methods were used and compared: an energy based method (SEA), a simple wave based analytical method and different numerical methods (finite elements – FEM; spectral finite elements – SFEM; and boundary elements - BEM). An accurate and fast model has been obtained by combining the wave approach and SEA. The numerical methods allow a finer level in the geometrical description, but are more time consuming. Most of the modelling work has been focused on understanding the role of metal frames (studs and rails) in airborne sound transmission, particularly in the case of single frame double walls where the frame plays a dominant role. The models show that the key parameter is the section stiffness of studs and rails; therefore, effort was focused on developing laboratory methods for estimating this key parameter. A modelling method has been developed successfully, and the rail or stud section stiffness can be measured and then used as input data in the model for calculating the R index of the wall considered. Moreover, a 3D numerical model has been developed to reproduce the laboratory method for characterizing studs; the model can estimate the section stiffness of new stud profiles numerically.

All the models developed or improved during the project have been validated and compared to each other, particularly in order to find out their limits. The conclusions are that: (i) it seems that in the case of a double wall, the wave approach is more accurate than SEA for modelling the transmission path through the wall air cavity, dominant at low frequencies; SEA is however well adapted for modelling the structural transmission paths through all the mechanical contacts between the two leaves of the wall at mid and high frequencies; the best model seems therefore to be an hybrid model combining both wave approach and SEA; (ii) when a finer level in the geometrical description is needed (optimizing stud profile for example), numerical models can then be used; in this case, 2D models seem to be a good compromise between computing time and accuracy; the absolute results given can be inaccurate, but at least 2D models are capable of giving the right ranking between different solutions.

Two parametric studies have also been performed using two different models; the results are comparable and clearly show the relative role of the stud translational and rotational stiffness in the sound transmission through the studs.

Most of these results have been presented at the international Euronoise 2006 conference in Tampere (Finland).

It should be noted that many building elements (mainly outer walls and separating walls) have been tested in the laboratory during the project for the following different reasons: (i) because the building element was new and its acoustic performance unknown, (ii) in order to validate models, (iii) in order to compare the performance of the element tested separately in the laboratory to its performance in the building on site.

9

Acoustic performance of lightweight buildings

Two problems have been considered: (i) the problem of airborne and impact sound transmission between dwellings, which include both direct transmission through the separating element (wall or floor) and flanking transmission through other elements and their junctions (one of the main goals of the project was to understand, quantify better and if possible reduce flanking transmissions in the case of steel frame lightweight construction); (ii) the problem of airborne sound transmission through façades, which includes the transmission through all the elements comprising the façade (wall, windows, doors, air inlets etc.).

For the first problem, a European standardized prediction model for calculating the building performances, which is only valid for heavy concrete structures, has been adapted to lightweight structures. The model input data are mainly the performance of the building elements (index R and impact level Ln) and two less known parameters: (i) the performance of the junctions between elements, which can be measured either in laboratory or on site, and (ii) the radiation efficiency of the building elements, which can be measured in the laboratory. During the project, this prediction model was validated on an existing small building which was thoroughly tested on site; moreover, the performance of all the building elements was tested separately in the laboratory. The prediction model was able to estimate the sound insulation corresponding to the different transmission paths in the building accurately. In particular, it was shown that for the lightweight building tested, which had a separate load bearing column-beam frame (which allows loose connections between walls and load bearing frame), flanking transmissions were still present, particularly in the case of impact noise, and for both vertical and horizontal transmissions.

A successful attempt has also been made in estimating two of the rather numerous input parameters of the prediction model using the finite element method; (i) a 2D FEM/BEM model has been developed for estimating the radiation efficiency of lightweight stiffened panels; the results are good, showing the extra radiation provided by the stiffeners; (ii) a 2D SFEM model has been developed for estimating the vibration level difference at junctions between building elements; this 2D model is capable of giving the correct ranking (compared to measured results) between the different flanking paths.

Other tests have been performed on site during the project: (i) several buildings carefully studied, in order to reduce flanking (avoiding any structural continuity at junctions between elements), showed that flanking can be almost suppressed; (ii) heavy concrete buildings with lightweight façade have been tested in order to evaluate the importance of flanking through the façade in the airborne sound insulation between apartments; the results show that overall flanking is important, but of the same order as for heavy concrete façades.

For the second problem of airborne sound transmission through façades, a European standardized model for calculating the façade performance from the performance of all the façade elements (wall, windows etc.) has been used and validated on site; it seems that the European model is too optimistic for lightweight structures and that a safety margin has to be taken into account.

Vibration modelling and testing (WP2 + WP5)

Objectives

The modelling, simulation, dynamic testing and subjective evaluation of vibration in lightweight floor structures are covered within this chapter.

The main objective of modelling and simulation of vibration performance of structures is to obtain an understanding of which structural parameters have the most significant influence on vibration properties of floors. The modelling is performed using finite elements where the connections and support have been incorporated.

10

A group of field- and laboratory tests have been carried out using a (virtually) unified testing procedure. The dynamic properties of the floors have been measured, together with subjective evaluations, where the latter concerns both body perception and vibration induced noise and shaking of articles. The test results have then been compared to vibration models, which allowed an update and refinement of the models afterwards.

The role of connections in vibration performance of structures has been studied through a number of laboratory tests in order to increase the knowledge about how connected structures influence the vibration properties of lightweight floors.

A laboratory test with a special motion simulator has been performed to evaluate subjectively phenomena related to floor vibrations. Mainly artificial vibration signals with specific characteristics have been used, but also some vibration signals measured in the field have been reproduced (with slight modification) thorough the simulator.

Horizontal vibrations

The results from one specific series of laboratory tests showed the importance of considering not only vertical but also horizontal vibration in floor construction. The horizontal contribution from walking might significantly affect the perception inside a real building, depending on how the floor is installed. Even though the response in the horizontal direction should be low in comparison to the vertical, it should not be neglected since interaction effects, e.g. beating, could occur in conjunction with the vertical response.

In addition, in the presence of horizontal vibration, although the body perception was unacceptable, the vibration of articles was accepted (for the concrete layer). Thus, the vibrating articles seem not to be affected by the low frequent horizontal vibration. It could also be the case that vibrating articles in general should be of considerable less concern compared to body perception, although this proposal is rejected by other researchers.

Vibration design criteria

There has been a lack of specific guidance relating to the vibration design of lightweight floors. Two different methods have been partly developed and evaluated in the Acousvibra project. A common design criterion for both methods concerns the floor’s fundamental frequency. Owing to the special vibration nature of lightweight floors, they should be designed in such a way that the fundamental frequency exceeds 10 Hz.

The first method studied in the project was based on deflection limits. The floors can be classified in different vibration classes with specified deflection limits. For typical residential buildings, the deflection limit due to 1 kN load is 0,5 mm. Based upon earlier tests, together with tests carried out within the project, this design criterion has worked well for different kind of floors and is therefore recommended for wider use. An addition advantage of this method is its simplicity.

The second method studied and developed is based upon weighted floor acceleration. The method originates from the ISO-2631 Standards, where no specific limits are given. For heavy weight floors, the acceleration and the ‘so called’ response factor usually predicts the floor performance quite well, but for lightweight floors it is harder to find limits that distinguish acceptable floors from those that are unacceptable. Limits have been introduced in the project concerning response factors of lightweight floors, but it seems that the method does not predict some aspects accurately, e.g., the acceptance of the floating floors.

Multiple frequency vibrations

One important reason that makes it so hard to find a robust, but at the same time simple, design criterion is the fact that, depending on the floor’s nature, different properties are in focus. For floors with

11

resonant behaviour, it should be adequate just to look at the fundamental frequency if, and only if, higher natural frequencies can be neglected.

However, many floors can not be treated that simple. It might therefore be needed to include natural frequencies other than the fundamental i.e., to consider the vibration response from multiple frequencies. The beating effect – an interaction effect of two adjacent natural frequencies – is according to the presented study a main factor, but, whether the corresponding frequency separation is best treated as a design parameter of its own - or complemented/combined by others, is a task that hopefully future research will answer.

A model that predicts the human annoyance from floor vibration is presented. It is based upon the ISO Wm weighting, and it constitutes a valuable evaluation method that seems to handle the difficulties with single vs. multiple frequency response well. It seems inadequate to use the ISO weighting directly for prediction – or even for comparison between signals - since the Wm weighting underestimates the importance of multiple frequency components. The ISO weighting handles single frequencies well, but, for signals that comprise two to three well defined discrete frequencies – a situation where the beating effect is significant but also a situation that might occur for real floors, it is not accurate enough.

Product development (WP3)

Objectives of this work package were to develop high quality acoustic and vibration performance solutions for lightweight outer walls/facades, partitions and floors. In particular, external noise of urban locations was taken into consideration in the development of the envelope of the building. New solutions were developed in order to reduce the vibration sensitivity of light weight floors. The following tasks are considered here:

Task 3.1 Development of façade and outer wall structures for tightened sound insulation requirements taking into account the flanking

Task 3.2 Development of new floor structures for improved sound insulation and vibration performance

Task 3.3 Development of partitions for improved sound insulation performance

Two types of light-weight façade elements were studied and developed: 1) for residential and office building fulfilling high traffic noise insulation requirements and 2) for industrial building fulfilling high airborne sound insulation value. Both applications were based on perforated light-gauge steel thermal studs which also fulfilled high thermal insulation requirements. In both cases, prediction models developed in the WP2 were utilized in the design of walls, and later, products were also tested in the laboratory. In the first case, field tests were carried in order to validate the acoustic performance in actual buildings and to validate calculation methods. Connection details were also verified by field measurements. The results showed that it is possible to fulfill high acoustic requirements for traffic noise and for airborne sound insulation in general by using lightweight steel facades. For flanking, it can be concluded that in the buildings with lightweight steel facades the sound insulation between dwellings is quite similar compared to the concrete outer wall system. Generally, the measured sound insulation indexes were good and fulfilled the Finnish sound insulation requirement R’

w ≥ 55 dB very well.

Two types of floor structures were studied, and more were developed in the project. Light-weight floors comprised light steel joists, gypsum board ceiling and the covering that was built from trapezoidal sheeting and either two layers of gypsum boards or thin concrete slab (~50 mm). Special attention was paid to the composite action between different components in order to improve floor stiffness properties in both transverse and perpendicular directions to the steel joists. Vibration tests were carried out in the laboratory as well as in situ. Laboratory tests indicated the importance of connection details on horizontal vibrations of the floors due to walking activities. Finally, subjective tests with proper floor connections showed acceptable vibration behaviour of the light-weight floors. Similar results were

12

received from field tests. The subjective tests also indicated that the deflection-based design criterion predicted the floor performance well. Furthermore, some field acoustic tests were carried out on a semi-heavy type of floor that was developed and tested in laboratory earlier. The acoustic field tests showed good airborne sound insulation properties, but some minor development is still needed concerning impact sound insulation.

The third group of developed products consisted of partition walls. Firstly, some modifications were made to the typical partition wall. The material saving was possible by introducing thinner studs and rails than used previously. The studs were embossed in order to increase the section stiffness of the stud profile, and so fulfil the stiffness requirements. Different embossed types were studied and finally bending tests and acoustic tests were carried out to the modified partition wall type. Furthermore, some preliminary design of a totally new type of partition wall was started. The study for the optimum shape of the wall stud and rail was made in close cooperation with other partners during research on work packages 2 and 4. It was possible to define some simple parameters to the partition wall studs in order to be able to predict the airborne sound insulation analytically. Those parameters for wall studs can be determined by simple laboratory tests or by fe-modelling. Some preliminary proposals for new section types were given.

Design Guide (WP6)

The objectives of this work package were to compile a European design guide for practical engineers and architects and to provide guidance on how to use modelling as a design tool. It was produced during WP6, and relied on work carried out in the other work packages. It involved the collaboration of all the participants in the project.

The main tasks identified in the package consisted of writing guidance on the following topics:

1. introduction

2. general aspects

3. design rules for lightweight structures with good acoustic performance

4. design rules for lightweight structures with good vibration performance

5. testing procedures and classification criteria for vibrations of lightweight floor structures

6. utilisation principles and possibilities of FEM- and SEA- models in product development

7. presentation of generic solutions

8. good design and assembly practice, and recommended connections

The design guide provides practical information for the design of lightweight steel structures so that they may have an adequate acoustic and vibration performance. A set of design rules for acoustic performance (Task 6.1) and vibration performance (Task 6.2) have been developed, and are included in the guide. It illustrates many examples of buildings using lightweight construction and their components and connection details. It demonstrates that lightweight steel construction is perfectly capable of meeting modern acoustic and vibration serviceability requirements.

The guide outlines the current requirements for acoustic performance in many European countries for floor, separating wall and façade design. It illustrates many examples of floor and wall construction, together with their insulating properties, and makes recommendations for best practice (Task 6.3). Major parameters that affect the performance at the whole building level are discussed, as well as for component design and junctions. Sound insulation against traffic noise is also covered. The basic principles of the modelling methods used to predict sound transmission are provided, including the

13

‘wave’, energy (SEA) and finite element approaches, and guidance is given on their specific use for lightweight structures.

Specific proposals for the dynamic assessment of lightweight floors are presented in the guide (Task 6.4): these include a deflection-based model and an acceleration-based model. Advice on finite element modelling of lightweight floors is also given. Case studies are presented, whereby field measurements of dynamic behaviour of floors are compared with predictions using these methods. The key parameters influencing floor vibrations are also discussed, and current methods of on-site testing and measurement, data processing and subjective evaluation are given. The guide also presents proposals for harmonization of annoyance criteria.

14

SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS

1 OBJECTIVES OF THE PROJECT

The commercial objective of the project is to increase the use of the light gauge steel framed buildings especially in the urban areas, where noise insulation and vibration of structures are known to be a problem. The prime technical objective is to develop guidance for codified design rules, guidance on analysis methods and guidance on detailing. This will promote the use of light gauge steel constructions with high quality vibration and sound insulation properties.

The main scientific objectives are:

• standardisation of testing methods for vibration of lightweight floors • improved modelling of acoustical and vibration behaviour of structures • better understanding of acoustical behaviour of lightweight structures in respect of flanking • more comprehensive concept of acoustical behaviour of lightweight structures at low

frequencies flanking

The main technological objectives are:

• standardised detail principles • improvement of sound insulation capacity of facades and other structures • introduce new construction and solutions • improvement of solutions for structures and junctions • design manual

The final objective of the project is to develop standard type of structures and details with classified (or known) sound insulation properties so that the structure and details can be easily picked up by designer in order to design a properly functioning building taking into account the different requirements in each country. One of the final objectives of the project is also to develop testing and design guidelines for vibration design of lightweight steel joist floors. Based on the new research results and earlier test results and design guidelines it is very possible that the semi-empirical guidelines can be improved. Vibration testing is already widely used in Canada and also in some countries in Europe (FI, SE, UK). Based on earlier and new experience, the harmonisation of the testing procedures is also possible.

2 COMPARISON OF INITIALLY PLANNED ACTIVITIES AND WORK ACCOMPLISHED

The work has been accomplished mainly according to original plan described in Technical Annex. Some deviations have been recorded in testing program. Standardized tests on full-size panels cannot be performed at the UPC facilities; for this reason, it was decided in a project meeting to redefine the initially planned UPC-testing campaign. In general terms, the number of specimens was reduced, but much more detailed experiments were carried out; as a result, the total amount of work of the modified test series was at least the same than that of the original work plan. Emphasis was put on understanding the role of connections between panels in airborne sound transmission.

The tasks described in Technical Annex are described in details in this report. In order to improve the readability of the document, the content of the final report is not following directly the order of the work packages in the project plan in Technical Annex. Acoustic and vibration issues are presented here

15

separately as a whole including analysis, testing and comparisons. The table below shows the corresponding chapter to each work package or task.

Work Package Work described in WP 1 Chapter 3 WP 2, Tasks 2.1 & 2.3 Chapter 4 WP 2, Tasks 2.2 & 2.4 Chapter 5 WP 3 Chapter 6 WP 4 Chapter 4 WP 5 Chapter 5 WP 6 Chapter 7

16

3 REQUIREMENT AND TESTING PROCEDURES FOR ACOUSTIC AND VIBRATION PERFORMANCE OF STRUCTURES (WP1)

3.1 Introduction and objectives The general objectives of the work described in this chapter relate to WP1 (Tasks 1.1 to 1.4). They were to carry out a comprehensive review of current and future sound insulation requirements, especially at low frequencies, where sensitivity of lightweight structures is critical. A study was to be carried out of the current acoustic requirements and design principles adopted for the design of façade structures, and to consider the suitability of ISO 12354 for lightweight steel structures (especially the flanking parameters). A further objective was to review how these acoustic requirements are met with current solutions and to consider the improvement possibilities. It was intended to harmonise the testing procedure for vibration performance of lightweight floors and to introduce a standard way of making vibration measurements on the basis of previous experience. The final objective in WP1 was to harmonise the rating of the annoyance of vibrations.

The information gained in WP1 has been incorporated in the design guide, and it formed the basis for modelling, design and product development.

3.2 Current and future requirements for acoustic performance of structures (Task 1.1)

A study of the acoustic requirements for dwellings in Europe found that they vary both in terms of measurement used, and in terms of absolute values. Table 3.1 compares the requirements for the UK, France, The Netherlands, Spain, Finland, Sweden, Norway, Denmark and Iceland. It covers airborne sound between separating walls and floors, impact sound (for separating floors), and limits for the attenuation of external noise. Acceptability is expressed sometimes in laboratory values, in field values, and in other cases by calculation. Spectrum adaptation terms are sometimes applied to cater for traffic or low frequency noise. Limits for the attenuation of external noise are expressed as a facade noise reduction in some countries, but in others as an absolute level in the dwelling.

A review of current relevant EN-Standards (EN 12354 series) concluded that they define how to calculate the acoustic performance of buildings (airborne and impact sound insulation between rooms, as well as airborne sound insulation of façades) from the performance of building elements (floor, wall, lining, window etc), including junctions between elements, but they are only valid for rather heavy construction. EN 12354 is now under revision, mainly in order to adapt/modify it for its use with lightweight construction. [A member of the ACOUSVIBRA team is also a member of the group revising the Standard.]

17

Table 3.1 Summary of requirements for acoustic performance in the UK, France, Spain, Iceland and Scandinavia

Country Airborne sound for separating walls and floors (dB)

Impact sound for separating floors (dB)

Façade insulation/Indoor noise from traffic or other outdoor sources (dB)

Remarks

UK ≥45 (DnT,w + Ctr,100–3150)

≤62 (L´nT,w)

–

Building in 'noisy' locations is restricted according to planning regulations

France ≥53 (DnTA)

≤58 (L´nT,w)

DnTA ≥ 30 Façade noise reduction

Spain Current code (NBE-CA 88)

R ≥ 45 dBA LN ≤ 80 dBA R ≥ 30 dBA

Values can vary depending on the use. The isolation of façades is global

Spain Future code (CTE DB-HR, draft June 2006)

DnT,A ≥ 45-55 dBA (different unit) RA ≥ 35 dBA (same unit)

L'nT,w ≤ 60-65 dB

D2m,nT,Atr ≥ 30-51 dBA

Requirements for façades depend on external noise and use of room

Finland ≥55 (R´w)

≤53 (L´n,w)

LAeq, 7-22 ≤ 35 (day) LAeq, 22-7 ≤ 30 (night),

Indoor noise level from traffic

Sweden * ≥52 (R´w with C50-3150)

≤58 (L´n,w with Ci,50-2500)

LAeq ≤ 30 Indoor noise level from traffic

Norway ** ≥55 (1)

(R´w with C50-5000) ≤53 (L´n,w with Ci,50-2500)

LAeq, 24h ≤ 30, LAmax, 22-06 ≤ 45,

Indoor noise level from traffic

Denmark Code Standard class C

≥52/53 ≥55 (R´w)

≤58 ≤53 (L´n,w)

LAeq, 24h ≤ 30 LAeq, 24h ≤ 30

Indoor noise level from traffic Code requirements not linked to Standard

Iceland ≥55 (1)

(R´w) ≤53 (L´n,w)

LAeq, 24h ≤ 30 Perhaps not the Code requirements

Netherlands ≥51 ≤59

* The standard SS 02 52 67 will be probably revised in the near future. ** The standard NS 75 will be probably revised in the near future (No great changes are expected) *** CTE Code being revised, proposed values in brackets [ ]. In the UK, there is a unique approach to the determination of acoustic acceptability for dwellings. This involves either, compliance with the acceptance criteria above through site tests, or, compliance by the use of accredited details throughout the construction. These accredited details are known as ‘Robust Details’ (RDs)1 and they have been established for a wide range of wide range of floor and wall types through rigorous site testing carried out by UK government approved agencies. RDs have to perform at about 5DB better than the minimum acceptable acoustic limits given above. A SCI publication2 illustrates details having ‘RD’ status which are relevant to steel framed construction, and it includes those which are expected to achieve satisfactory acoustic performance, but have not yet been given full approval. The number of details is increasing, as companies obtain approval for their products and forms of construction.

18

3.2.2 Special consideration of traffic noise The most harmful noise source through facades in cities is from traffic. Noise problems arise mostly in residential buildings and other overnight accommodation, such as hotels and hostels, and also in schools, day-care centres and business apartments. The effects of traffic noise can be reduced through town planning by placing such buildings far enough from roads, railways and airports, and by shielding them from noise. Also, acoustic barriers can be used along the noisiest roads. However, it is normal practice in cities to take account of external noise when new buildings are planned. When the daily average noise level exceeds the sound acceptance level at the facade, the sound insulation of the facade needs to be specified. The sound level acceptance criteria for building façades for several countries are given in Table 3.1. In large cities, buildings can be located very close to the streets, so that there is no yard between the street and the building. In such cases, the sound level in the front of a facade can be as high as 75 dB, or more. This demands a very high performance from the sound insulation of the facade.

Basic facade structures can reduce the traffic noise by 30 to 40 dB. For example, when the sound level is 60 dBA outside the building, the sound level inside the building can reduce down to between 20 and 30 dBA. It is possible to reduce noise by 45 to 50 dB using standard products. However, higher sound insulation requires special solutions for the structures, windows and window areas. The use of fresh air ventilators can be impossible because of the noise penetration. Such requirements can exist, for example, close to airports.

3.2.3 Relevant International Standards for acoustics Site measurements should be carried out in accordance with EN ISO 140. For airborne sound measurements, a steady sound of a particular frequency is generated in the source room, and the sound pressure level in the source and receiving rooms are compared to ascertain the reduction. For impact sound measurements, a standard impact sound source is used to strike the floor, and the impact sound pressure level is measured in the room below (or beside). For both airborne and impact measurements, the receiving room levels must be corrected to take into account of the absorption of the receiving room (for example standardised to 0.5s reverberation time) before comparison with performance standards. Measurements are taken at 16 one third-octave frequency bands across the hearing spectrum from 100 Hz to 3150 Hz.

To convert the site measurements into a single figure rating, the method set out in ISO 717 compares the set of 16 measured results with a reference curve. The rating is made by considering only those measured values which fall short of the reference curve, and by choosing a reference curve where the sum of the deviations is as large as possible, but not greater than 32 dB. The value of the reference curve at 500 Hz gives the single figure rating.

A summary of the relevant International Codes is given below:

Laboratory: Airborne sound insulation – EN ISO 140 – 3: Laboratory measurements of airborne sound insulation of building and of building elements.

Impact sound insulation – EN ISO 140 – 6: Laboratory measurements of impact sound insulation of floors.

Impact sound insulation – EN ISO 140 – 8: Laboratory measurements of the reduction transmitted impact noise by floor covering on heavyweight reference floors.

Impact sound insulation – EN ISO 140 – 11: Laboratory measurements of the reduction transmitted impact noise by floor covering on lightweight reference floors.

pr EN 10848 Acoustics – Laboratory measurement of the flanking transmission of airborne and impact sound between adjoining rooms:

Part 1 (frame document),

Part 2 (application to light elements when the junction has a small influence),

19

Part 3 (application to light elements when the junction has a substantial influence) and

Part 4 (all other cases)

Field: Airborne sound insulation – EN ISO 140 – 4: Field measurements of airborne sound insulation between rooms.

Airborne sound insulation – EN ISO 140 – 5: Field measurements of airborne sound insulation of façade elements or façades.

Airborne sound insulation – EN ISO 140 – 7: Field measurements of impact sound insulation of floors.

Standards for rating sound insulation of building and of building element: Airborne sound insulation – EN ISO 717 – 1

Impact sound insulation – EN ISO 717 – 2

Standards for predicting building performance from performance of elements: EN 12354 – 1: Building Acoustics – Estimation of acoustic performance of buildings from the performance of elements – Airborne sound insulation between rooms.

EN 12354 – 2: Building Acoustics – Estimation of acoustic performance of buildings from the performance of elements – Impact sound insulation between rooms.

EN 12354 – 3: Building Acoustics – Estimation of acoustic performance of buildings from the performance of elements – Airborne sound insulation against outdoor sound.

20

3.2.4 Glossary of acoustic terms The relevant acoustic terms used in this report are shown in Table 3.2, below, are defined in ISO 140 and ISO 717 as follows:

Table 3.2 Acoustic terms used in this guide

Type of sound

Term for laboratory

measurement (dB)

Term for field measurement

(dB) Description

Ln L′n Normalised impact sound pressure level (measured at a given frequency, corrected for absorption)

Ln,T L′n,T Standardised impact-sound pressure level (measured at a given frequency, standardised to a reverberation time of 0.5 seconds)

Ln,w L′n,w Weighted normalised impact sound pressure level (single figure value, taken at 500 Hz after frequency weighting)

LnT,w L′nT,w Weighted standardised impact sound pressure level (single figure value, taken at 500 Hz after frequency weighting)

Impact Sound Insulation

Ci Ci

Spectrum adaptation term for impact sound (value added to single number quantities to take account of the unweighted impact sound level over a defined frequency range)

R R′ Sound reduction index (apparent sound reduction index) (value for a building element for a given frequency band, adjusted for panel size and receiving room absorption)

Rw R’w

Weighted (or weighted apparent) sound reduction index (single figure value, taken at 500 Hz after frequency weighting)

Dn Dn Normalised airborne sound pressure level difference (measured at a given frequency, corrected for absorption)

Dnw Dnw Weighted normalised airborne sound pressure level difference (single figure value, taken at 500 Hz after frequency weighting))

DnT DnT the standardised airborne sound pressure level difference (measured at a given frequency, standardised to a reverberation time of 0.5 seconds)

Airborne Sound Insulation

DnT,w DnT,w Standardised weighted airborne sound pressure level difference (single figure value, taken at 500 Hz after frequency weighting)

C C Spectrum adaptation term for airborne sound (value to be added to a given single number rating, weighted for A-weighted pink noise over a specified frequency range)

Ctr Ctr Spectrum adaptation term for airborne sound (value to be added to a given single number rating, weighted for A-weighted urban traffic noise over a specified frequency range)

Leq Equivalent continuous sound pressure level (measurement for fluctuating sounds over a specified time period)

LAeq (LAeq,T)

A-weighted equivalent continuous sound pressure level (measured for fluctuating sound over a specified time period, but weighted for frequency)

Airborne Sound

LAmax (LAmax,T)

Maximum sound pressure level (measured for fluctuating sound over a specified time period)

Numerically Ln,w = Ln,T (L´n,w = L’n,T), if the volume of the receiving room, V, is 31 m3. If the volume of the receiving room, V, is for example 62 m3, L'n,w is 3 dB greater than the respective standardised value L'nT,w. If the proportion of the surface area of the structure, S, and the volume of the receiving room, V, is 0.32 (for example S=10 m2, and the volume of the receiving room V= 31 m3), the value of Rw (or R´w) coincides with value of DnT,w. If the proportion is twice that, the value of Rw is 3 dB smaller than the value of DnT,w. If the surface area S is 10 m2, the value of Rw (or R´w) coincides with value of Dn,w.

21

3.3 How the requirements are met with current solutions (Task 1.2)

Current design practice to meet acoustic requirements often involves the application of standard solutions which have been developed and approved by testing. Typical methods for façade, roof, floor and wall construction are presented below in Table 3.3 to Table 3.6. The tables show, an image and description of the make up, and their acoustic performance. Further construction details and their performance are presented in the design guide3 associated with this project. Note that the sound insulation product data of building components are declared using the Standards ISO 717-14 and ISO 717-25.

Table 3.3 Typical details for Façade construction

Detail Description

Façade panels:

Façade panel 175 -gypsum board 13 mm -vapour barrier -Thermal stud + mineral wool 175 mm -gypsum board 9mm +steel cladding Rw = 50 dB Rw + Ctr,100–3150 = 43 dB

Table 3.4 Typical details for roof construction

Roof Construction:

1) PVC membrane 2) plywood 3) mineral wool 300 4) thermal purlin 5) steel sheathing Rw = 47 dB

22

Table 3.5 Typical details for floor construction

Detail Description

Floor Construction:

Suitable for use as a separating floor • 18 mm chipboard or similar • 19 mm gypsum board • 25 mm to 30 mm of resilient layer

(mineral wool 120 to 200 kg/m3) • Chipboard or OSB base • Light steel joists (min 150 mm deep) • 100 mm of unfaced mineral wool quilt

(10 to 30 kg/m3) between the joists • Proprietary resilient bars • 2 layers of gypsum board (≥23 kg/m2) DnT,w + Ctr,100–3150 = 48 to 52 dB L’nT,W = 54 to 57 dB Fire rating = 60 minutes

Table 3.6 Typical details for separating walls and internal partitions

Detail Description

Partition Walls:

Partition wall width 66mm -gypsum board 13 mm -light stud 66 mm+ mineral wool 50 mm -gypsum board 13mm Rw > 45dB -double gypsums on both sides Rw > 49dB

Separating wall width 125 mm -Stud thickness 0,7 mm -2 × Gypsum EK (12 kg/m2) -Rw > 59dB -Wall width +25 mm => +1 dB

The work on the ACOUSVIBRA project has identified a number of recommendations for good practice in floor and wall construction, and these have been incorporated into the design guide (see Section 7).

23

3.4 Harmonisation of the testing procedure for vibration performance (Task 1.3)

Since the early 1990s, rapid advances in the instrumentation, digital data acquisition and processing have enabled a transfer of the advanced vibration measurement and testing technologies from mechanical and aerospace engineering disciplines to civil structural engineering applications. These technologies are nowadays available for the testing of floors. In this context, the purpose of dynamic testing is to:

• measure the actual dynamic responses of floors at the dynamic loading specified in the design

• post-process these responses in accordance with the specified national (or international) Standard to obtain a relevant response parameter, and

• to rate this parameter against an acceptance criterion specified for the floor design.

3.4.1 Modal testing Modal testing or experimental modal analysis6 is a complex technology whose aim is to establish modal properties of test structures experimentally. It is based on measuring and post-processing dynamic responses of a test structure at one or more locations. Two types of floor modal testing exist; when the excitation force creating these responses is not measured, and when it is measured. In all cases the response of the floor is recorded at a number of points on the floor to form a “test grid” in order to give an impression of the overall behaviour of the floor.

Modal testing of floors without measuring the excitation force Ambient vibration survey (AVS) The AVS is based on the assumption that the floor dynamic excitation is provided by the environment in which it resides and has energy more or less evenly distributed within the frequency range of interest. The acceleration of each point on the test grid over a period of time (usually about 15 minutes) is recorded and the acquired time-histories are processed to give the natural frequencies, operational deflection shapes corresponding to those natural frequencies (which are very close to mode shapes), and modal damping ratios (note that modal mass cannot be estimated from AVS).

Heel-drop Excitation in heel-drop tests is provided by a single person raising themselves on the balls of their feet and dropping onto their heels, so providing an impact. The multi-modal decaying response to this impulsive broadband excitation can be measured and the response data used to estimate modal properties in a manner similar to AVS. Some researchers7 have recently developed an instrumented heel-drop test, where the heel-drop is executed on top of a slim, purpose-built force plate containing a load cell which measures the excitation force. This set-up amounts to a modal testing, whereby the force is measured.

Rotating mass shaker The rotating mass shaker applies a sinusoidally varying vertical force at particular frequencies to excite floor harmonic responses at the same frequencies. These are then measured and recorded at a range of frequencies so the recorded amplitudes can be plotted against the frequencies for each test point. From this, it is possible to estimate likely natural frequencies which correspond to the increase in amplitudes of harmonic responses, and a single-degree-of-freedom (SDOF) half-power method is typically used to estimate damping.

A common feature in all these modal testing methods is that modal properties tend to be less complete and reliable. This is because a number of assumptions have to be made to enable extraction of modal properties, and some assumptions may not be correct.

24

Modal testing of floors with measurement of the excitation force The following testing methods yield a frequency-response-function (FRF), which relates the input signal to the output signal at each frequency. The FRF will give information about natural frequency, damping, modal mass and, if an FRF is determined for a grid of points, the mode shapes of the floors.

Instrumented impact testing The frequency content of a force produced either by heel-drop or hammer is measured and compared to the resulting response to produce the FRF at a range of positions. To minimise the seriously detrimental effects of unmeasured extraneous excitation of floors, the FRF-based modal testing of floors should be performed on an unoccupied floor structure, preferably in an unoccupied building. Advice on impact testing of floors is given by Reynolds and Pavic8,9, and Blakeborough and Williams7.

Shaker testing Low signal-to-noise ratios in the instrumented impact testing can be resolved by employing an instrumented shaker excitation. The excitation is generated by a moving shaker armature of known mass. This mass is driven by a signal generated by a spectrum analyser that is also used to acquire all of the force and response data.

3.4.2 Dynamic performance assessment Although a floor’s modal properties are very important, its vibration performance depends most directly on the actual vibration response due to given dynamic excitation. Therefore it is necessary to create relevant dynamic excitation, then, after measuring and post-processing floor vibration responses due to the excitation, rate the obtained response parameter. For office floors, it is considered that floor responses due to a single person walking should be measured, as this kind of excitation occurs frequently in floors and is difficult to isolate.

Human sensitivity to vibrations is frequency dependent, and so a method of frequency weighting is employed to account for this. The weighting leaves vibration levels unchanged where the contour is low (say, between 4 and 8 Hz) and attenuates the levels at frequencies to which humans are more ‘resistant’. In this way, vibration response is ‘normalised’ to the same sensation level, irrespective of the excitation frequency. There are two parameters which are typically used in modern codes of practice for assessing the amount of vibration and its effects: these are the RMS acceleration, and the recently established vibration dose value (VDV).

RMS acceleration is used because it is a measure of the total vibration causing distress to the human body. Greater RMS accelerations correspond to higher vibration magnitudes, which cause more annoyance. However, an assessment of the human distress using the RMS relationship is appropriate for “well behaved” (as Griffin10 defines them) vibrations which are steady-state-long-lasting-periodic or stationary-random. If the vibrations are short-lived transients, the RMS acceleration is no longer a reliable effective value (Griffin, 199610).

A method that addresses this problem, and is gaining acceptance internationally, is the vibration dose value (VDV) method. The VDV is a cumulative measure of the vibration transmitted to a human receiver during a certain period of interest.

3.4.3 Subjective evaluation The human body is a very sensitive vibration instrument, and many objects and some furniture are also very sensitive to floor vibrations. Human perceptions are often more authentic, and also more accurate, than many theoretical calculations.

During subjective tests, observers are asked to give their opinion of the intensity and acceptability of the vibrations induced by a walker. At least two people are needed when evaluating the properties of a floor. A walking person may not himself feel the floor vibrations, whereas someone sitting or standing nearby can feel them more clearly. The rating of intensity and acceptability is highly subjective. Even when most of the observers classify the vibrations as perceptible, there may be one who classifies the same vibrations as barely or strongly perceptible. Also, when the majority classify the floor as acceptable, one

25

observer may find it wholly unacceptable. Opinions of several observers are therefore needed; the more the better.

The observations are made both from body perception in a sitting position from vibrations induced by a walker weighing about 80 kg and from objects - clinking of a coffee cup with a spoon in the cup and with the cup on a saucer; leaf movements of a 30-40 cm high plant; rippling of water in a glass bowl; and chinking of a glass pane.

The observers are asked to form an opinion on the intensity and acceptability of the vibrations after the walker has passed the observation point three times. If more than one floor is rated in succession, the order of testing may affect the results. If the worst performing floors are tested first, the properties of better floors may be over-estimated. Conversely, if the better performing floors are tested first, the properties of worse floors may be under-estimated.

Ratings for intensity and acceptability are asked from the observers according to test form shown in Figure 3.1. More detailed guidance for subjective testing is given in the Design Guide3 prepared for Acousvibra project.

INTENSITY OF VIBRATIONSThe vibrations are- imperceptible (No)- barely perceptible (B)- clearly perceptible (C)- strongly perceptible (S)

ACCEPTABILITY OF VIBRATIONS

Is the floor acceptable in a newly built livingroom ?+ yes ++ absolutely acceptable- no - - absolutely unacceptable

Test 1 Intensity AcceptabilityNo B C S ++ + - - -

Body perceptionClinking of a coffee cupLeaf movements of a pot plantWater rippling in a glass bowlChinking of a glass pane

Test 2 Intensity AcceptabilityNo B C S ++ + - - -

Body perceptionClinking of a coffee cupLeaf movements of a pot plantWater rippling in a glass bowlChinking of a glass pane

Figure 3.1 Assessment form for rating the vibrations

3.5 Harmonisation of the rating of the annoyance of vibrations (Task 1.4)

3.5.1 Review on current practice Guidance relating to the dynamic response of floors to walking activities and their acceptability has been reviewed. There is a significant lack of specific information regarding lightweight floors. Much of the guidance reviewed in this section focuses on floors which are subject to resonant response. The very nature of lightweight construction dictates that the response of the floor is likely to be transient in nature, not resonant. In these cases, the response is dominated by a track of impulses corresponding to the heel impacts. The effect of these impulses is that they set the mass of the floor in motion; the floor vibrates at its natural frequency and decays rapidly as energy is dispersed over the floor as a whole. As a consequence, successive peaks and decays typify the overall dynamic response of a floor of this type.

26

The classic approximation, as shown below, is commonly used to determine the natural frequency of a system

max0

18

δ=f

where:

δmax = total deflection (in mm), based on the gross second moment of area of the components with a load corresponding to the self weight, plus permanent loads, plus 10% of imposed load.

Despite a general agreement on the method of determining fundamental frequency, there is not the same consistency when determining floor response, and various methods are available11,12,13,14,15&16.

Eurocode 5 BS EN 1995-1-1:200417 for timber structures includes two requirements for floor acceptability. The first requirement limits static deflection under a unit load: this is described in Table 3.7. The second requirement limits the unit impulse velocity, as follows:

( )11 −≤ ζfbv

Where:

v = Unit impulse velocity maximum value of velocity under 1N/s idealised impulse (m/Ns2)

b = Floor width (m)

f1 = Fundamental frequency of floor (Hz)

ζ = Modal damping ratio

Additional Stiffness/frequency methods In addition to the methods of calculating an explicit value for the response of the floor, a conservative method exists for limiting the response by specification of a minimum stiffness/frequency criterion. Guidance documents have been produced by SCI, AISC,the board of design for Sweden and Finnish Constructional Steelwork Association18. Table 3.7, below, contains a summary of the requirements from each of these guidance documents.

27

Table 3.7 Summary of Stiffness/frequency requirements Body Requirement AISC Frequency ≥ 9-10 Hz

Minimum stiffness under a force of 1kN per mm (it is intended, however, that the peak acceleration should also be checked - such that apeak/g ≤ 0.5%)

BKR (Sweden) Under a 1 kN force, the deflection of a single beam in a timber floor structure should not exceed 1.5 mm

SCI Frequency ≥ 8 Hz Under a 1 kN force the deflection of a single beam in a timber floor structure should not exceed the following:

Span (m) 3.5 3.8 4.2 4.6 5.3 6.2 Deflection (mm) 1.7 1.6 1.5 1.4 1.3 1.2

.

Finland Frequency ≥ 10 Hz Under a 1 KN static load, deflection should be limited by classes of acceptance

Acceptance Class A B C D E Deflection (mm) ≤0.12 ≤0.25 ≤0.5 ≤1.0 >1.0

. prEN1995-1-1 Frequency > 8Hz

Under a 1 KN static load, deflection should be limited for a given width of floor as follows:

Floor width (m) 50 70 90 110 130 150 Deflection (mm) 4.0 2.7 1.8 1.3 0.8 0.5

.

ISO 2631 + British Standards Human perception of vibration depends on the frequency of the vibration, the type of activity being undertaken and the direction of the vibration relative to the body (i.e., whether the subject is standing, sitting or lying). To allow for this, the vibration is weighted according to the frequency, so that the equivalent vibration can be determined. Values of frequency weighting are given in ISO 263119,20. Various weighting curves are given in the Standard, depending on the direction of vibration and the activity. The three most common weighting curves are shown graphically in Figure 3.3 and Figure 3.4, along with suitable linear approximations. Table 3.8 shows when each curve applies, and the axes of vibration are given in Figure 3.2.

28

In most cases, the aim of vibration analysis is to reduce or remove discomfort, but, in special circumstances - such as operating theatres, the vibration will need to at a level that cannot be perceived and does not affect the steadiness of hand or vision. Perception and discomfort use the same weightings, but perception will tend to have a lower allowable threshold (i.e., a subject can detect vibration without being discomforted by it).

Table 3.8 Weighting factors appropriate for floor design

Room Type Axis of vibration Category BS 6841 weighting curve

z-axis Vision/Hand control Wg Critical working areas (e.g. hospital operating theatres, precision laboratories) x-, y-axis Perception Wd

z-axis Discomfort Wb Residential, offices, wards, general laboratories, consulting rooms x-, y-axis Discomfort Wd

z-axis Discomfort Wb Workshop and circulation spaces

x-, y-axis Discomfort Wd

Supportingsurface

y

z

xSupportingsurface

y

x

Supportingsurface

x

z

y

z

Figure 3.2 Directions for Vibration defined in ISO 2631

29

Acceptance magnitudes are no longer given in the current version of ISO-2631-220 (from 2003). BS 6472: 199221 covers many vibration environments in buildings. Rather than specifying specific levels of acceptable vibration, limits of satisfactory magnitude are expressed in relation to a ‘base curve’ and a series of multiplying factors ranging from 1 to 128. The base curves for vibrations in the z-, x- and y-axis, together with a typical range of factored curves, are shown in Figure 3.5. Each line shown on the figure represents a constant level of human reaction, known as an isoperceptibility line: the area above a line corresponds to an unacceptable human reaction; the area below represents acceptable levels of vibration.

1 10 100

1

Wei

ghtin

g fa

ctor

0.1Frequency (Hz)Wd Weighting

1 10 100

Wei

ghtin

g fa

ctor

1

0.1Frequency (Hz)Wg Weighting

Figure 3.3 Wd and Wg frequency weighting curves

1 10 100

1

Wei

ghtin

g fa

ctor

Frequency (Hz)Wb Weighting

0.1

Figure 3.4 Wb frequency weighting curve

30

To ensure that satisfactory levels of vibrations are maintained for different environments encountered within health and hospital buildings, ISO 2631-2: 198922 specifies the multiplying factors (response factors) shown in Table 3.9 that should be applied to the z-, x- and y-axis ‘Base curve’ shown in Figure 3.5.

Table 3.9 Response factors used in ISO 2631-2:1989 to specify satisfactory magnitudes of building vibration

Place Time Response factor for continuous vibration

Operating theatre, precision laboratories 1 Wards, residential Day

Night 2 to 4 1.4

General laboratories, offices 4 Workshops 8

The values given within Table 3.9 are similar to those recommended by BS6472: 199221, with the exception that no relaxation is permitted on the response factor of 1 when an operating theatre is not in use. Furthermore, no guidance is given on the length of the exposure periods that should be considered for day- and night-time use (BS 6472 specifies a 16-hour day and an 8-hour night).

3.5.2 Harmonisation of criteria for the acceptability of vibrations

Two different methods have been selected in the Acousvibra-project for the assessment of vibration acceptance of light-weight floors. The first method is based on deflection criteria and is valid for high-frequency floors with natural frequency over 9 to10 Hz. The second method is based on acceleration criteria.

11045 & 11046

1 10 1000.001

rms

acce

lera

tion

(m/s

²)

Frequency (Hz)

0.010

0.100

1.000

12 x base curve

Base curve

4 x base curve

1 10 1000.001

rms

acce

lera

tion

(m/s

²)

Frequency (Hz)

1.000

0.100

0.010

12 x

base

curve

Base c

urve

4 x ba

se cu

rve

(a) (b)

Figure 3.5 Building vibration (a) z-axis curves; and (b) x- and y-axis

curves for rms acceleration according to BS6472

31

Deflection-based model This model proposes a five-class classification of floors in residential and office buildings (Table 3.10). The classification has already been adopted in Finland (for example) by the Certified product Declaration procedure18. The class represents the sense-perception of a person in a seated or standing position, and the sense-perception from vibrations of objects. The classification is material independent and presumes that walking-induced vibrations are the basis of the design. Because the perception of vibrations is a subjective feature, the descriptions given in Table 3.10 are initial suggestions. Also, the vibration of objects depends on the properties, and on the position of the object. The vibration class should always be agreed with the customer. A lower or higher class may be agreed for special products, dimensions or floor usage.

Table 3.10 Class recommendations for floors of residential and office buildings.

Vibration class Scope of application

A Normal class for vibrations transferred from another apartment/room. Special class for vibrations inside one apartment/room.

B Low class for vibrations transferred from another apartment/room. High class for vibrations inside one apartment/room.

C Normal class for vibrations inside one apartment/room.

D Low class for vibrations inside one apartment/room. For example attics and holiday cottages.

E Class without restrictions.

The design criteria of high-frequency floors may be based on floor deformations due to a 1 kN point load. Both the global deflection δ0 of the floor beams and local deflection of the floor surface should be studied. The local deflection δ1 comprises the deflection of the floor slab, deformation of the floating floor and deformation of raised floors. Table 3.11 gives tentative limiting values for parameters for the corresponding vibration classes to be used in either design or in the testing of floors. Verifications of acceptance limits can be found, e.g., from the article of Toratti and Talja (2006)23. Natural frequencies and deflections can be determined by testing, or by simple calculation methods. Equations for natural frequency and deflection are presented in Chapter 7.

Table 3.11 Vibration classification of high-frequency floors.

Vibration class

Criterion for the global deflection δ0 of the floor beams

Criterion for the local deflection δ1 of the floor surface

A δ0 < 0,12 mm δ1 < 0,12 mm

B δ0 < 0,25 mm δ1 < 0,25 mm

C δ0 < 0,50 mm δ1 < 0,50 mm

D δ0 < 1,0 mm δ1 < 1,0 mm

E δ0 > 1,0 mm δ1 > 1,0 mm

The classification mentioned above is based on a comprehensive test series for light-gauge steel and timber floors carried out in VTT, Finland23. Figure 3.6 shows the correlation between subjective ratings and vibration measurements. It can be seen that static deflection highlights the difference between acceptable and unacceptable high-frequency floors, when f0 > 9.

32

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

0 5 10 15 20 25 30 35

Fundamental frequency [Hz]

Def

lect

ion

due

to 1

kN

[mm

]acceptednot accepted

Class C, δ < 0,5mm

Figure 3.6 Deflection of 1 kN point load compared to acceptability.

Acceleration-based model The acceleration-based model is based mainly on the methods described in Chapter 3.5.1, sub-section “ISO 2631 + British Standards”. The frequency-weighted acceleration ar.m.s and the response factor can be determined by testing, or it can be calculated according to the method shown in Chapter 7. However, it has been recognised that the response factors contained within Table 3.9 are based on continuous vibrations, and are therefore appropriate for floors that are continuously very heavily trafficked. For less heavily trafficked floors, walking activities will produce intermittent vibrations. For that reason, higher limits are needed for light-weight floors - especially for residential buildings. Some preliminary proposals for acceptability limits can be derived from the same Finnish study as described in previous chapter23. The test data is presented in the graph of response factor against natural frequency in Figure 3.7. It can be seen that the response factor of 16 may be appropriate, but, on the other hand, many floors above that limit were also judged to be acceptable.

0

20

40

60

80

100

120

8 13 18 23 28 33

Fundamental frequency [Hz]

Res

pons

e fa

ctor

AcceptedNot accepted

Figure 3.7 Response factor compared to acceptability

33

If the calculated response values are within the acceptability limit, there is no need to consider the intermittent nature of dynamic forces further. However, for the situation where the floor has a higher response than would be acceptable under the conservative limits for continuous vibration, the acceptability should be assessed by considering the intermittent nature of the dynamic forces. In these circumstances a cumulative measure of the floor response may be made through the use of vibration dose values (VDV’s); this method is described in detail within Appendix B of BS 647221 and is also described in Chapter 7. According to this method, the total time of an acceptance intermittent activity can be determined as follows:

4

)(68.0 ⎥⎦

⎤⎢⎣

⎡×

=rmsa

VDVt

where t is the total time duration of activity and VDV is the appropriate limit from the BS 647221. VDV valued of 1.6 has been proposed in this project as discussed in Chapters 5 and 7.

34

4 ACOUSTIC MODELLING AND TESTING

One of the objectives of WP2 in the Acousvibra project was to develop/improve tools for modeling the acoustic performances of lightweight building structures in order to get a better understanding of sound transmission through such structures and optimize the different components. These tools include models for the acoustic performance of building components such as walls or floors, but also models for the acoustic performance of the whole building, taking into account both direct and flanking transmissions. As a result, this chapter is divided into two main parts: one dealing with building components (section 4.1) and one dealing with the whole building (section 4.2). Moreover, one objective of WP4 of the project was to perform (i) laboratory tests of cost effective and acoustically optimized structures and (ii) field tests to measure the building acoustic performance and identify the importance of direct and flanking transmissions; but another objective was to use these tests to validate the models developed in WP2. It seems therefore more convenient and clear to group the activities and results concerning both modeling and testing in the same chapter.

4.1 Component level

4.1.1 Introduction In buildings, separating elements (inner walls and floors or outer walls) transmit airborne sound or impact sound; this transmission is called direct transmission (as opposed to flanking transmission, see section 4.2) and is characterized by the following two standardized acoustic quantities:

• the quantity expressing the direct airborne transmission through external walls (against outdoor sound) or through separating walls or floors between dwellings is the sound reduction index R of the wall or floor considered

• the quantity expressing the direct impact noise transmission through floors is the normalized impact sound level Ln of the floor.

These two quantities are frequency dependant and can be either measured in laboratory or estimated using models. In the Acousvibra project, existing models were used and new models were developed in order to better understand the role of the different components (boards, cavity, studs, rails…) in sound transmission through lightweight steel frame building elements; oppositely, not much work was done on modeling impact sound transmission.

In the first sub-section below, the different modeling methods used by the Acousvibra partners are described. In the second section (testing) all the laboratory tests of walls, floors or elementary components (such as studs or rails) performed during the project are given. Finally, all the important results related to the models (calibration /validation, comparisons between models, as well as parametric studies) are given in the last two sections.

4.1.2 Modeling Most of the models developed or improved during the Acousvibra project concern the airborne sound transmission through walls. Lightweight steel walls are made of double leaf elements with single frame or two separate frames and one or more gypsum boards on each side (see examples given in Figure 4.1).

35

Figure 4.1 Examples of double leaf lightweight walls a) with single frame; b) with separate sets of studs; c) façade wall with single frame and external finish (plastering)

Sound propagates through a double leaf wall mainly through 2 paths: (i) airborne sound transmission from one leaf to the other through the air cavity; this transmission can be attenuated by inserting absorbent (porous) material in the cavity; (ii) structure borne sound transmission through any mechanical contact between the two leaves: through the studs in the case of single frame double wall, and through the boundary rails and boundary studs in all cases. When modelling the acoustic performance of such building elements, all these paths must be taken into account.

Mainly three types of approach have been used in the project to model lightweight walls: Statistical Energy Analysis (SEA) developed and used by VTT, wave approach and hybrid wave approach / SEA developed and used by CSTB and numerical methods (finite elements, boundary elements) developed and used by UPC.

Statistical energy analysis (SEA) • The SEA approach is an energy based model. The model includes airborne transmission through

cavity (case of double walls) and structure borne transmission through studs (case of single frame double wall). Using SEA, the sound transmission is expressed on a power basis with the assumption of many resonant modes present in rooms and cavities (acoustic modes) and in plates (structural modes) in the frequency bands considered. Sound power flows between sub-systems (typically room → plate → cavity → plate → room) are calculated from (i) the energy stored in each sub-system expressed from the average sound level in rooms and cavities, and from the average vibration level in plates; (ii) parameters called coupling loss factors. All the SEA assumptions and parameters can be found in existing literature (see for example [24]); many parameters (coupling loss factors for example) are based on theoretical equations also found in the literature. All the airborne and structure borne sound paths mentioned above are taken into account using SEA as seen in Figure 4.3; in the case of single frame double wall for example, path 7 represent the structural path through the studs, which is modeled as springs point connected between plates at screw locations (see Figure 4.2).

Figure 4.2 SEA spring connection between 2 infinite plates

a) b) c)

Plate

Plate 2

Point connecting tie36

• In the Acousvibra project, (i) some original work on modeling studs as springs line connected between plates has been done; the line connections are modeled with the same logic as for the classical point connections, the point impedance of plates being replaced by the line impedance of plates and the sound bridges modeled as line springs; (ii) laboratory measurements have been performed (see 4.1.3 testing) to check SEA assumptions (checking natural frequencies of plates and cavities) and experimentally estimate damping properties (loss factors) of boards in the case of lightweight structures; (iii) comparisons between SEA calculation results and measured results (see 4.1.4 calibration/validation) have been made, showing in particular the limitation of SEA in modeling the path through the air.

Figure 4.3 SEA transmission path diagram for a double wall

Wave approach and hybrid models (wave approach combined with SEA)

• The wave approach is well known for its application to sound transmission through infinite thin plates in flexure, or through infinite double plate systems with air cavity in between (see for example [25]). The sound power transmitted is calculated in 3D, wave by wave, using the classical wave equation for thin plates in flexure. Figure 4.4 shows this diagrammatically, whereby the wall is being excited by a plane wave at different angles of incidence; a diffuse field transmission loss is then obtained by adding the sound powers incident or transmitted at different angles. This approach is valid over the whole frequency range and the computation time is quite short (compared to FEM for example). However, real lightweight steel frame walls or floors are more complicated than infinite thin plates for the obvious following reasons:

- plates are stiffened by studs or joists - the air cavity is not the only path; structural paths through inner frame and/or boundary frame must be taken into account

- the size of real walls is not infinite - in the case of floors, both airborne and impact sound insulations must be considered.

As a consequence of this, more sophisticated tools (but still based on the wave approach) have been developed to treat lightweight structures; details on these tools are given further in this section.

37

Figure 4.4 Wave approach basic configuration

• Hybrid models (wave approach combined with SEA): In the following models, both wave approach and SEA are combined. The wave approach is applied to infinite plates and the effect of finite size plates is taken into account using a spatial filtering technique (ref. [26]). Five models have been developed and used in the project :

(i) Sound transmission through stiffened single plate (see Figure 4.5): the model is based on the wave approach, the stiffeners being taken into account as reaction line forces and moments (bending and torsion); the model was developed and results were published (ref. [27]) before Acousvibra

Figure 4.5 Examples of single leaf and double leaf lightweight walls studied using hybrid models

(ii) Sound transmission through double walls with separate sets of studs (see Figure 4.5); an hybrid wave/ SEA model is used: the wave approach is used for the airborne path through cavity (if present, the absorbent material is modeled as an equivalent fluid) and SEA is used for the transmission path at boundaries (see Figure 4.6). What is really new and developed during the project is (i) the use of a semi infinite plate input mobility for estimating the SEA transmission paths through boundary rails and boundary studs; (ii) the laboratory characterization method developed for estimating the corresponding SEA parameter (spring values); the coupling loss factor is determined experimentally from vibration level differences measured between plates when one is mechanically excited (see 4.1.3 testing).

z

Infinite wall

Incident plane wave

θ

x

Periodically Spaced

Stiffeners

Lightweight panelDiffuse

Acoustic Field

Lightweight panelDiffuse

Acoustic Field

Lightweight panels

Absorbing cavity

Periodically Spaced

Stiffeners

Diffuse Acoustic Field

Lightweight panels

Absorbing cavity

Periodically Spaced

Stiffeners

Diffuse Acoustic Field

(a) (b)

38

Figure 4.6 Examples of structural path at the boundaries of a double wall with separate sets of studs

(iii) Sound transmission through double walls with common set of studs modeled as point connections; an hybrid wave / SEA model is used: the wave approach is used for the transmission path through cavity over the whole frequency range and for the transmission path through studs, modeled as line spring connections between plates at low frequencies. SEA is used at higher frequencies where studs and boundary rails are modeled as springs point connected between plates at screw locations. Here again, the spring values are determined using the laboratory characterization method developed during the project. This model, developed and validated during the Acousvibra project (see 4.1.4 calibration/validation) is the most advanced model developed so far for this type of double walls (see 4.1.5 parametric studies and conclusions). This model has also been used to validate/develop new types of walls (see chapter 6 new products)

(iv) Sound transmission through double walls with common set of studs modeled as line spring connections; this pure wave approach model has been developed during the Acousvibra project and applied to both the airborne transmission through cavity and the structure borne transmission through studs; studs are modeled as line connections between plates with both translational and rotational stiffness. This model has been used in a parametric study to show particularly the relative importance of translational and rotational stiffness in the sound reduction index of the wall (see 4.1.5 parametric studies and conclusions).

Figure 4.7 Example of single frame double wall studied using models (iii) and (iv)

(v) Impact sound transmission through stiffened lightweight floor; this pure wave approach model has been developed during the Acousvibra project; the excitation force is decomposed into propagating normal stress waves using the 2D spatial Fourier technique and the wave approach is then used the same way as for airborne excitation to calculate the sound power radiated. Comparisons between calculated and measured results show that the model is good at low frequencies but over estimate the impact sound level at mid and high frequencies as shown in Figure 4.8. It seems that the floor input mobilities are

Lightweightpanels

Periodicallyspaced studs Periodically

spaced screwsalong studs

Lightweightpanels

Periodicallyspaced studs Periodically

spaced screwsalong studs

39

not correctly modeled at mid and high frequencies. No further work has been done to improve the model.

Figure 4.8 Impact noise level of a single board with joists

Finite element method Deterministic numerical models have been developed and used during the Acousvibra project. A combination of numerical techniques —the standard finite element method (FEM); the spectral finite element method (SFEM); the boundary element method (BEM) — has been used. These numerical models are especially indicated in the low- and medium-frequency range, and allow a finer level in the geometrical description of the building components than the other two modeling approaches described above. Three types of model have been developed:

(i) 2D FEM/BEM model of sound transmission through double walls

The model calculates the sound reduction index of finite size double walls, taking into account the airborne transmission path through the cavity and the structure borne path through the studs, line connected to plates (because of the 2D geometry). The propagation of sound in the sending and receiving rooms and in the cavity is modeled with the Helmholtz equation. If present, the absorbent material is modeled as an equivalent fluid. The structure borne path is described with the equation of structural dynamics. These equations are solved in a coupled fashion by means of FEM and/or BEM for the acoustic domains and FEM and/or SFEM for the structure. A simplified version of the model for single walls has also been developed.

Studs can be modeled either as real stud profiles, taking into account the actual cross-section shape, or as translational and/or rotational stiffness (see Figure 4.9). Comparisons between calculated and measured sound reduction index have been made and the model has been used in a parametric study (see 4.1.5 parametric studies and conclusions) to show the relative importance of translational and rotational stiffness in the sound reduction index of single frame double walls.

Figure 4.9 Studs are modelled as real profile (left) or as translational and rotational springs (right))

Tapping machine

Lightweight panel

Periodically spaced stiffeners

(joists)

50

60

70

80

90

100

100

160

250

400

630

1000

1600

2500

4000

Frequency (Hz)

Ln (d

B)

Calculated - without stiffener

Calculated - with stiffeners

Measured

40

(ii) 2D FEM model of vibration transmission in double walls

The model calculates the vibration level difference between finite size plates in the case of a double wall mechanically excited. The model takes only into account the structure borne path through the studs, line connected to the plates. Like in the model of sound transmission just discussed, studs can be modeled as real stud profiles or as translational and/or rotational stiffness, see Figure 4.9. The model with real stud profiles has been used to compute the vibration transmission loss Dij of various stud profiles. Figure 4.10 shows how “acoustic” studs (i.e. open profiles with folded webs) have a larger transmission loss (and hence, a superior acoustic performance) than standard studs.

Figure 4.10 Influence of the stud shape on the vibration transmission loss

The model with springs has been used to identify the translational and/or rotational stiffness by matching the vibration level differences obtained with real stud profiles and with ideal spring connections. The ultimate goal would be to use these stiffness values as input for a SEA model. The identification procedure leads to a frequency-dependent stiffness. The fact that the spring stiffness depends on the frequency can be explained by noting that different sectional vibration modes of the actual stud profile, with different resonance frequencies, correspond to different apparent stiffness. Although SEA models can in principle deal with frequency-dependent input, this is not common practice, so the practical use of these results is not straightforward.

(iii) 3D FEM model of studs

A 3D model of a length of stud, either isolated or connected to two boards and representing the laboratory experimental setup (see section 4.1.3) for measuring the translational stiffness of studs has been developed (see Figure 4.11). The applied force can be either static or dynamic (frequency range below the first section mode of the stud); the stud translational stiffness is estimated from the ratio between stud response and force applied. This model has been developed during a cooperation project between UPC and CSTB associated to the Acousvibra project: with a mobility grant from the RFCS, a PhD student from UPC stayed at CSTB for five months.

Figure 4.11 3D finite element model of laboratory setup for stud testing

41

Stiffness calculated by the model and measured using the experimental setup have been compared: differences were observed but the ranking of the studs with respect to their stiffness was the same.

The model has then been used to rank the stiffness of various stud designs (see Figure 4.12). Stiffness is the most relevant parameter for the acoustic performance of studs used in double walls: as already pointed out above, a lower stiffness means less vibration transmission between the two leaves connected by the studs and, hence, better acoustic performance.

Starting from the usual C profile, the different types of folds in the web shown in Figure 4.12 were analyzed. The main conclusions of the study are that (1) acoustic profiles indeed have a considerably lower stiffness (55% to 75% of profile C), so a significant improvement in acoustic performance is to be expected; (2) outward folds are more effective that inward folds in providing extra flexibility to the stud and (3) fold width is a parameter much more relevant than fold height.

Figure 4.12 Stud profiles analyzed

4.1.3 Testing Acoustic testing corresponds to tasks in WP4 which has been divided into 3 parts: laboratory tests, field tests and laboratory characterization of building components (such as studs or gypsum boards); all the tests are summarized in a table given in section 4.4.

Laboratory tests Laboratory standardized tests: Several laboratory measurements of building elements according to standards ISO 140 part 3 (sound reduction index), part 6 (impact sound for heavy floors) and pr ISO 140 part 11 (draft standard for lightweight floors) have been performed during the Acousvibra project.

Here is a list of the building element tested, indicating why they were tested and giving reference to the sections of this report where the measurement results are used, and where all the details of interest can be found:

- sound reduction index of outer walls measured in order (i) to test new performing outer walls (see chapter 6 for details) and (ii) to compare with field measurements (see 4.2.3 testing); two types of outer walls have been tested:

- outer wall with thermal studs TC175 (see VTT report RTE 149-05)

- multilayer thermal stud wall (see FIOH report 5210-2004-25974e)

42

50

175 80

15

80

Figure 4.13 Typical outer wall tested (with thermal studs TC175)

- sound reduction index of several separating single frame double walls using different sets of studs and with one or two boards on each side, measured in order (i) to test new performing separating walls (see chapter 6 for details), (ii) to validate models (see 4.1.4 calibration/validation) and (iii) to compare with field measurements (see 4.2.3 testing); several types of walls were measured:

- partition walls with embossed 66mm studs (see FIOH report TY02-2006-31771E)

- partition walls with different stud and rail types (see FIOH report TY02-2006-34973E)

Figure 4.14 New studs and rails used in walls tested in laboratory

Laboratory non-standardized tests: Non-standardized laboratory measurements of double leaf panels have been performed at the Acoustics Lab of the Universitat Politècnica de Catalunya (LEAM/UPC). Small size panels (1.455 m × 1.615 m) were tested between a semi-anechoic chamber and a reverberant room (see Figure 4.15).

The goal of the testing campaign was to assess the influence of connections between panels. To this end, five specimens were designed and manufactured (see example in Figure 4.16). The first specimen is a single-panel double-wall. The second specimen consists of two half-panels with a junction with no

43

special treatment between them. Then, this connection is progressively improved by sealing the junction between plasterboards, placing a C profile between the two half-panels, and filling the cavity between half-panels with foam.

No important differences were found between these five specimens, even if the degree of filling of the cavity with absorbent material was varied. The most probable cause for this result is that the boundary rails are so stiff that they are the predominant path of sound propagation, so acting in other paths (junction, cavity) is of little relevance. These results cannot be generalized, because of the too small size of the panels.

Figure 4.15 View of the LEAM/UPC laboratory aperture

Figure 4.16 Example of double leaf panels tested

Field tests Field tests consist in measuring the building performances and therefore are reported in section 4.2 of this chapter which deals with the performance of the whole building.

Laboratory characterization of building components Buildings components such as boards, studs or rails were experimentally characterized in order to estimate input parameters of the models described in 4.1.2:

(i) Detailed laboratory tests (see Figure 4.17) were performed on single leaf wall and double leaf wall with and without mineral wool in cavity; plate loss factors and natural frequencies of plates and cavity were measured in order to check the validity of SEA and to get the loss factors.

The total loss factor was estimated by measuring the structural reverberation time in 1/3 octave bands using two accelerometers (one on stud and one in the mid span between studs)

44

and two to four excitation points (using hard hammer, more massive rubber hammer or even fist excitation), in order to get data over the whole frequency range. The loss factors were also estimated from narrow band spectrum analysis (bandwidth at 3 dB) of plate vibration modes.

The vibration modes of three structures were also analyzed in order to get the plate modal densities and check the validity of SEA; for the same reason, the cavity modes of a double leaf structure without wool insulation were measured with a microphone installed in cavity using pink noise excitation.

Figure 4.17 VTT experimental setup for detailed SEA tests on lightweight partition walls

(ii) a laboratory characterization method for studs has been developed in order to estimate the SEA input parameter for studs (translational section stiffness) ; this stiffness is obtained from the input mobility of a length of stud connected to two boards, measured at frequencies below first modal behavior of studs. The stud element tested (see Figure 4.18) is positioned between two 120 x 65 cm2 gypsum boards; three screws, spaced every 30 cm, connect each gypsum board to the stud. Two pieces of wood are placed a both ends of the experimental setup to minimize rotation. A force transducer and an accelerometer are used to measure the input mobility from which the stiffness per unit length of stud is deduced in the low frequency range (around 10 Hz), below modal behavior. An example of stiffness spectra is given in Figure 4.19.

Figure 4.18 CSTB experimental setup for stud characterization

Force Excitation(hammer)

Rigid Floor

45

0.0

100000.0

200000.0

300000.0

400000.0

500000.0

600000.0

700000.0

800000.0

900000.0

1000000.0

1.000 10.000 100.000Frequency (Hz)

Stiff

ness

(N/m

)

TC125TC175AWS125

Figure 4.19 Example of experimental results for stud characterization

(iii) a laboratory characterization method for boundary rails has been developed in order to estimate SEA input parameter for rails (translational section stiffness); the vibration level difference (VLD) between two boards only connected through top and bottom rails is measured. Two rail elements are positioned at the top and bottom between two 2.4 x 1.2 m2

gypsum boards (see Figure 4.20). The two rail elements are screwed to the laboratory concrete ceiling and floor with the required (used in practice) screw spacing. A hammer is used to apply a “rain on the roof” type excitation (distributed excitation over the board surface). In order to deduce the rail stiffness characteristic, the VLD between the two gypsum boards is measured and compared in the mid-high frequency range to a predicted VLD, calculated using SEA model of rails represented by springs point connected to plates at screw locations (an example of result is shown in Figure 4.21).

Figure 4.20 CSTB experimental setup for boundary rail characterization

Force Excitation(hammer & Rain on the roof)

Rigid floor

Rigid ceiling

46

-10

0

10

20

30

40

50

100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

VLD

(dB

)

Predicted with SEA - Kt=0.25 MN/m

Measured TC125 rail (with glass wool in cavity)

Figure 4.21 Example of experimental / predicted velocity level difference for boundary rail characterization

4.1.4 Calibration/validation and comparison between models Calibration/validation The work was focused on single frame double walls.

VTT SEA model Comparison between calculated and measured results shows that the VTT SEA model can estimate the sound reduction index Rw (single number value) of double walls quite exactly. However, 3 important limitations should be mentioned: (i) there is a problem to make accurate estimations of the index R at the resonance frequency of double walls; this shows the limit of SEA for estimating the transmission through cavity (see also comparison between models down below); (ii) at low frequencies, there are not enough modes for a proper SEA modeling, and results should be used carefully; (iii) at critical frequency of plates (equality between sound wave length and plate vibration wave length, leading to high radiation, usually at high frequencies), the SEA model seems to overestimate the sound radiation in some cases. In the future, the model must be improved by introducing more realistic models.

CSTB hybrid model The CSTB model for double walls with common set of studs combines wave approach and SEA as described in detail in section 4.1.2. The stiffness values for studs and rails used in the predictions correspond to those measured in laboratory as discussed above. The different paths (studs, boundary studs, and boundary rails) can be studied independently (as shown in Figure 4.23) and the whole wall system can be optimized. In general, when the cavity is filled with absorbing material, the frame is the dominant path in sound transmission except in the low frequency range where the transmission path through the cavity is dominant. For relatively stiff studs (TC175 or TC125 type), the paths associated to the boundary rails, boundary studs and the studs contribute equivalently to the total sound transmission in the mid-high frequency range. For more resilient studs (AWS125 type, see Figure 4.22), the path associated to the boundary rails is mostly responsible for the sound transmission in the mid-high frequency range. The predictions provide results rather close to those measured, within 1 dB in term of index Rw, for the different systems studied.

Figure 4.22 Example of Acoustic Wall Stud (AWS)

47

Figure 4.23 Predicted and measured index R for a double wall mounted on TC175 studs and rails; detailed results for the different transmission paths are given

Comparison between models Three models (VTT SEA model, CSTB hybrid model and UPC finite element model) have been compared in 2 different studies:

SEA/hybrid models The CSTB hybrid wave/SEA model and the VTT SEA model have been compared on the performance of single leaf and double leaf single frame walls using Ruukki TC 175, 125 or AWS 125 studs. The main results are the following:

- both models give similar results at mid and high frequencies (see example Figure 4.24 for a double wall with one gypsum board on each side, TC 125 studs and mineral wool in cavity); however, studs have only small effects on sound transmission according to the VTT model, and bigger effects according to the CSTB model - sound transmission at the resonance frequency of double walls is over estimated using SEA; as before, studs have no effect on sound transmission; sound transmission is lower using CSTB wave approach, and decreased even more with the presence of studs - single values Rw obtained with the two models are very close to each other (within 1 dB) and close to measured results for walls with standard studs, but not as close for walls with acoustic studs (differences higher than 2 dB between models and between calculated and measured results).

Boundary rails Boundary studs Other wall Studs

Boundary rails Boundary studs Other wall Studs

58 (-2)Calculated – Structure borne path trough boundary rails only

58 (-2)Calculated – Structure borne path trough boundary studs only

56 (-2)Calculated – Structure borne path trough studs only

51 (-3)Calculated – All paths

52 (-3)Measured

Rw in dB (C in dB)

58 (-2)Calculated – Structure borne path trough boundary rails only

58 (-2)Calculated – Structure borne path trough boundary studs only

56 (-2)Calculated – Structure borne path trough studs only

51 (-3)Calculated – All paths

52 (-3)Measured

Rw in dB (C in dB)

10

20

30

40

50

60

70

50 63 80 100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

TL (d

B)

Measured

Calculated - All paths

Calculated - Structure borne path through studs only

Calculated - Structure borne path through boundary studs only

Calculated - Structure borne path through boundary rails only

48

0

10

20

30

40

50

60

70

80

90

50 63 80 100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

TL (d

B)

CSTB - Without studsVTT - Without studsCSTB - Predicted with CSTB stiffnessVTT - Predicted with CSTB stiffnessMeasured

Figure 4.24 Measured and predicted R index for a single frame double wall; comparison between SEA and hybrid models

Finite element/wave approach models The CSTB wave approach model (with studs modeled as line connections) and the UPC 2D FEM/BEM model have been compared on the performance of single leaf and double leaf single frame walls. The main results are the following (see results in Figure 4.25 in the case of a double wall with one gypsum board on each side, TC125 studs and absorbent material in the cavity):

- UPC model under estimate sound transmission at mid and high frequencies, both in case of single leaf and double leaf walls without absorbent material in the cavity - both models agree on the small influence of studs when there is no absorbent material in the cavity - they give similar results at mid and high frequencies in cases of double walls with absorbent material in the cavity, but different results at low frequencies (the CSTB R index shows a deep drop at resonance frequency as the UPC R curve is much smoother) - the usefulness of these two models are limited since they don’t take into account the transmission through boundary rails, dominant at mid and high frequencies

10

20

30

40

50

60

70

80

90

16 20 25 32 40 50 63 80 100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

TL (d

B)

CSTB - Without studsUPC - Without studsCSTB - Predicted with line connectionsUPC - Predicted with line connectionsMeasured

Figure 4.25 Measured and predicted R index for a single frame double wall; comparison between finite element and wave approach models

49

4.1.5 Parametric studies and conclusions Two parametric studies have been performed in order to show the relative importance of translational and rotational stiffness in the acoustic performance of single frame double walls:

Parametric study using the wave approach model In this model, studs are represented as line connections between plates over the whole frequency range with both translational and rotational stiffness. Examples of index R spectra for different values of translational and rotational stiffness are given in Figure 4.26. It can be observed that a spring rotational stiffness below 1kNm/rad/m (which seems to be the case of most existing studs) has little influence on the sound reduction index; above this value, it mainly affects the sound reduction index in the mid-high frequency range. The translational spring stiffness modifies the sound reduction index in the mid-frequency range but the slope remains the same (the translational stiffness values shown in the results are common values measured on existing studs).

0

10

20

30

40

50

60

70

80

90

50 63 80 100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

TL (d

B)

Predicted - Kt = 0.7 MN/m2 and Kr = 10 Nm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 340 Nm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 1 kNm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 10 kNm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 0.1 MNm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 1 MNm/rad/m0

10

20

30

40

50

60

70

80

90

50 63 80 100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

TL (d

B)

Predicted - Kt = 0.7 MN/m2 and Kr = 1 kNm/rad/m

Predicted - Kt = 3.5 MN/m2 and Kr = 1 kNm/rad/m

Predicted - Kt = 0.7 MN/m2 and Kr = 1 MNm/rad/m

Predicted - Kt = 3.5 MN/m2 and Kr = 1 MNm/rad/m

Figure 4.26 Effect of stud translational (Kt) and rotational (Kr) stiffness values on the index R of single frame double walls

Parametric study using the 2D finite element model In this model, studs are represented as line connections between plates with both translational and rotational stiffness.

Figure 4.27 shows the sound reduction index vs. frequency for different values of the translational stiffness Kt (denoted by colours) and the rotational stiffness (denoted by symbols). Note that, for the sake of clarity, the values Kt = 105 N/m2 and 107 N/m2 are shown in the left figure, while the values Kt = 106 N/m2 and 108 N/m2 are shown in the right figure. The large range of variation in the sound reduction index R shows that stud stiffness is indeed a relevant parameter. Three different parts can be identified in the figures: (1) for very low frequencies (below 200 Hz approximately), the cavity transmission path is dominant and there is little influence of the stud stiffness; (2) in the central part, the response is dominated by the translational stiffness and the rotational stiffness is not relevant (all curves of the same colour are grouped together); (3) for high frequencies, there is an effect of the rotational stiffness.

50

Figure 4.27 Effect of translational and rotational stiffness values of studs on the index R of single frame double walls

Both models (wave approach and 2D finite elements) show the same relative importance of translational and rotational stud stiffness over the frequency range of interest.

Parametric study using the CSTB hybrid model The CSTB hybrid model has been used in a parametric study in order to optimize the composition of single frame double walls having a given acoustic performance; part of the results are given in chapter 6 new products.

51

4.2 Whole building level, including junctions 4.2.1 Introduction Airborne and impact sound transmissions in buildings include both direct transmission through the separating element (wall or floor) and flanking transmissions resulting from vibration transmissions through junctions between elements in the sending room and elements in the receiving room. Understanding better flanking transmissions in the case of steel frame lightweight constructions was one of the goal of the Acousvibra project.

The performance of buildings is expressed as airborne and impact sound insulation between rooms and airborne sound insulation from outdoor sound. A European standard (EN 12354) exists for predicting these quantities: parts 1 and 2 of the standard deal with airborne and impact sound insulation between rooms and part 3 deals with airborne sound insulation from outdoor sound. These quantities are calculated from the performance of building elements (index R of floors, walls, windows, doors… and impact level Ln of floors) and the performance of the junctions between these elements, taking into account both direct and flanking transmissions. However EN 12354 does not apply to steel (or wood) frame lightweight constructions because the acoustic behavior of these elements is different from heavy concrete or masonry elements and corrections have to be made on both the method for predicting the performance of building and also the methods for measuring some of the acoustic properties of lightweight building elements; work has just started, at European level, to estimate and later standardize these corrections. CSTB is active in these European groups, has proposed a modified method (see 4.2.2 modeling) and has tested it in the frame of the Acousvibra project by thoroughly analyzing one building (see 4.2.3 testing).

Other field tests have been performed during the project, without any flanking path analysis, in order to compare the performance of different building elements in laboratory and in real buildings; these results are also reported in section 4.2.3 testing.

4.2.2 Modeling Two types of models are considered in this section: a prediction model for calculating the building performances and models for calculating some of the input parameters of the prediction model.

Models for calculating the building performances Two models are considered: one model for calculating the airborne and impact sound insulation between rooms taking into account direct and flanking transmissions and one model for calculating the airborne sound insulation from outdoor sound taking into account all the façade elements (wall, window, door, air inlet, rolling shutter…)

Model for flanking transmissions Figure 4.28 gives an example of vertical acoustic transmissions in a lightweight building and schematically shows the direct path and one of the flanking paths through a junction between floor and wall.

52

Figure 4.28 Typical example of flanking transmission path

Direct transmissions are known since they are very close to the performance of the separating element tested in laboratory and expressed in term of sound reduction index R or impact sound level Ln; in lightweight constructions, separating elements are usually double leaf elements (double walls and floors with suspended ceiling).

The CSTB model used for estimating flanking transmissions in this project is adapted from standard EN 12354-1 and -2. Flanking transmissions from element i in the source room to element j in the receiving room is expressed in term of a flanking sound reduction index Rij in the case of airborne sound transmission and a flanking impact sound level Ln,ij in the case of impact sound transmission as follows:

ji

sijv

jiij SS

SDRR

R log.102

*,

**

+++

= (A1)

j

iijv

ijiinijn S

SDRR

LL log.102

*,

**

,, −−−

−= (A2)

In the case of airborne sound transmission, equation (A1) shows that flanking transmission depends on the sound reduction indices ji RR , of elements i and j and the vibration attenuation at the junction

ijvD , between these elements; high indices R or high junction attenuation will lead to small flanking. The symbol * indicates that the corresponding term has been adapted (or measured in a special way) in the case of lightweight constructions; S represents the surface area of the building elements i, j and index s stands for separating element.

In the case of impact sound transmission, equation (A2) shows that flanking transmission depends on the direct impact sound transmission iinL , of the floor (without suspended ceiling as shown in Figure

4.1) and the junction vibration attenuation ijvD , .

Flanking transmissions in lightweight buildings are more difficult to predict than for heavy constructions for the following reasons:

- indices R must be corrected for flanking because lightweight elements radiate sound differently when excited mechanically; CSTB has proposed correction terms, which are calculated from the radiation efficiency of the element considered. The radiation efficiency is proportional to the ratio between the acoustic power radiated by a wall and the vibrational energy stored in the wall. The radiation efficiency of several steel frame lightweight elements have therefore been measured in laboratory (see 4.2.3 testing) in order to estimate these corrections

1/2 separating wall participating to flanking path ij

Floor without suspended ceiling participating to flanking path ij

j Direct path

i

Flanking path

53

- single leaf elements (bare floor or half separating wall) are often involved in flanking transmission through lightweight buildings (as shown in Figure 4.28) and their acoustic performance (R or Ln) is usually not known and has to be measured or estimated. During the project, index R and impact level Ln of bare floors and half separating walls have been measured in laboratory (see 4.2.3 testing) in order to get typical values which can be used in the prediction method.

- equations A1 and A2 show that the vibration attenuation Dvij at junctions is the key parameter to estimate flanking transmissions. There is a new standard (EN 10848 series) showing how to test junctions between building elements, which can be used to evaluate and optimize junctions in laboratory; however this standard is not directly applicable to lightweight constructions, where vibration fields are not at all homogeneous (they are in concrete buildings). CSTB has adapted and used the EN 10848 standard in the project for testing junctions between lightweight elements (see 4.2.3 testing) in order to get typical values which can be used in the prediction method.

Models and measuring methods for façade insulation (1) General

In the different European countries the regulations concerning the facade insulation greatly differ in philosophy. In many countries, the requirement is expressed as an indoor sound level which, due to varying outdoor noise levels and spectrums, makes the situation quite complicated - i.e. what is the objective for the façade sound insulation?

In some countries the requirement is directly expressed as a façade sound insulation value which can be measured easily or predicted quite accurately using existing European standards, if sufficient safety margins are used; the same concepts of sound insulation are used in prediction and in evaluation, and concepts or quantities correlate well which each other. For example in France requirements are expressed as façade sound level differences, making the situation quite easy to handle.

(2) Modeling and predicting:

The most common method to model and predict sound insulation of a façade is presented in the European standard EN 12354-3. The method is based on calculating the façade performance from the performance of the different parts (measured in laboratory) and their dimensions. The method is general and several sound insulation quantities can be used and predicted: R’ (apparent sound reduction index), Dn (normalized sound level difference) or DnT (standardized sound level difference); single number values can also be calculated with or without spectrum adaptation terms.

According to standard EN 12354-3, and if for example the quantity Dn is used, the building façade insulation can be estimated from the laboratory performance of the elements composing the façade (wall, windows, doors, air inlets or rolling shutters) using the following equation:

).1010.10log(.10 10/10/

0

10/∑ ∑∑ −−− ++−=i k k

kD

j

DiRn L

lASD nekneji

where iR is the sound reduction index of the wall, window or door considered (of surface area

iS ), nejD the insulation of the air inlet and nekD the insulation of the rolling shutters (of length

kl ).

This type of equation can be used for lightweight façades to predict their performance in practice (with safety limits).

54

(3) Measuring:

Standard EN ISO 140-5 proposes different methods (using different concepts and quantities) for evaluating the sound insulation of a building façade in the field. The sound source can be either real traffic noise (its spectrum may vary a lot from case to case) or a loud speaker. The outdoor microphone may be located close to the structure considered or 2 m away from it, leading to different sound levels which must be used in different equations in order to calculate the façade sound insulation. Practice has shown that in most cases the use of traffic noise as sound source is impossible because (i) the sound level is too low for accurate measurements and (ii) the spectrum may vary a lot; the loudspeaker method is more repeatable and often recommended. Finally, two different approaches are also proposed in the standard to measure either the index R or the level difference D of the façade considered.

If for example the normalized sound level difference is used, the building façade insulation can be measured as:

)/log(.10 022,1,2 AALLD mnm −−=

where mL 2,1 is the outdoor sound level at 2m in front of the façade, 2L the sound level in the

receiving room and 0A a reference equivalent absorption area (10m2). A can be estimated for the measured reverberation time of the receiving room. Figure 4.29 schematically shows how these sound levels are measured.

Figure 4.29 Schematic of façade sound insulation measurement

As a conclusion, it should be mentioned that even if there are some problems in predicting the sound insulation of a real façade or in evaluating this quantity by measurement, the greatest uncertainties lie presumably in predicting the outdoor noise levels from which requirements are often given.

Models for calculating the input parameters for the prediction method Two input parameters play an important role in predicting flanking transmissions (see above) and are not easy to measure: the vibration attenuation at junctions between elements and the radiation efficiency of building elements. An attempt at calculating these two parameters (or at least getting orders of magnitude) using FEM models has been done during a cooperation project between UPC and CSTB and two models have been developed:

2D FEM/BEM model for radiation efficiency A 2D model for estimating the radiation efficiency of finite stiffened panels has been developed during the project. The model takes into account the geometrical shape of the stiffener sections (see Figure 4.30) and compares the radiation efficiency on both sides of the panel (stiffener side or board side). Panel and stiffeners are modeled with the finite element method (FEM) and the acoustic domains with the FEM or the boundary element method (BEM). The radiation efficiency has been evaluated in two different ways: (1) calculating the acoustic power flow through a control surface or (2) assuming that

L1,2 m

2m

L2 (receiving room)

55

the pressure field in the interior acoustic domains is a diffuse field. The results obtained (see Figure 4.30) clearly show the extra radiation provided by the stiffeners: the radiation efficiency on the beam side of the panel is larger than on the board side, which is similar to the case of a board without stiffeners. The model also reproduces the coincidence phenomenon (frequency for which the wave length of the excited modes of the structure is similar to the wave length of the acoustic waves in the air and thereby the radiation is maximum).

Figure 4.30 Radiation efficiency of a stiffened single plate: 4 cm thick wooden board with T shape 14 cm high steel joists

2D SFEM model for junction vibration level difference A 2D model for estimating the vibration level difference of junctions between building elements (walls and floors) has been developed during the project. The building elements (walls and floors) are modeled as beams with proper mass per length and bending stiffness and junctions between elements are similar to the section profile of the real junctions (see Figure 4.31). The spectral finite element method (SFEM) has been applied to compute the vibration level difference of the various building elements connected at the central junction.

Figure 4.31 Two dimensional model of part of a full size building tested during the project

The main difficulty lies in correctly modeling the complex connections between components (beams and boards) at the junction. The results are shown in Figure 4.32. Note that, in spite of the 2D nature of the numerical model and the complex connection behavior, the experimental and calculated curves have the same ordering. These results are encouraging, because they show that the numerical model is a useful tool for determining the dominant flanking transmission paths in a junction.

56

(a) (b)

Figure 4.32 Vibration level difference Dvij of the central cross junction: a) experimental results; b) numerical results

4.2.3 Testing Laboratory tests Laboratory tests have been carried out at CSTB in order to get all the input parameters needed for estimating flanking transmissions in a typical two storey constructions.

These laboratory tests include the following measurements (building elements shown in Figure 4.33) :

Standardized measurements (1) The sound reduction index and impact sound level of a bare lightweight floor have been measured according to ISO 140 part 3 and pr ISO 140 part 11 respectively

(2) the sound reduction index of a stiffened panel (half a separating wall) have been measured according to ISO 140 part 3

The results (see Figure 4.34) show the poor acoustic performance of these single leaf lightweight elements, since only half of the systems (top floor without ceiling and ½ separating wall) participate to flanking; fortunately, as explained further, this weakness is compensated by poor acoustic radiation and high vibration attenuation at junctions.

Figure 4.33 Bare floor without suspended ceiling (22 mm thick wooden chipboard) and ½ separating wall tested (two gypsum boards screwed on studs)

140 mm

150

mm

Floor CTBH22

Double Ω joists spaced every 60 cm

2x 13mm gypsum boards

60 cm

70 mm

C shaped studs

57

Figure 4.34 R index of the bare floor tested (without suspended ceiling)

Radiation efficiency measurements The radiation efficiency of the two elements above (bare floor and stiffened panel) have been measured; the measurement method is not standardized and has been adapted from ISO 140 part 3 and EN 10848 (standard for measuring junctions performance); the radiation efficiency is estimated from the ratio between the acoustic power radiated by a building element and the vibrational energy stored in the element; the measurement method consists in measuring the space averaged sound pressure level in the room where the element radiates sound (using microphones) and in measuring the space averaged vibration level of the element (using accelerometers).

The examples of results given in Figures 4.34 and 4.35 show that (i) single leaf elements involved in flanking have a very low index R, leading to high flanking; (ii) however their radiation efficiency is very low, particularly at low frequencies: about 10 dB less than heavy elements in the case of airborne excitation and between 15 and 20 dB less than heavy elements in the case of mechanical excitation; results also show that lightweight elements do radiate more on stiffener side than on board side, particularly in the mid frequency range. This result confirms/validates the results obtained numerically by UPC using their 2D FEM/BEM model (see section 4.2.2).

-30

-20

-10

0

10

20

30

100

160

250

400

630

1000

1600

2500

4000

Frequency (Hz)

σ (d

B)

Structural Excitation - Radiation onboards sideStructural Excitation - Radiation onstuds side

-30

-20

-10

0

10

20

30

100

160

250

400

630

1000

1600

2500

4000

Frequency (Hz)

σ (d

B)

Acoustic Excitation - Radiation onboards side

Acoustic Excitation - Radiation onstuds side

Figure 4.35 Radiation efficiency of the ½ wall tested; with structural excitation (left); with acoustic excitation (right)

28 (-2)27Emission roomon boards side

28 (-1)27Emission roomon studs side

Rw in dB (C in dB)R in dB(A)

28 (-2)27Emission roomon boards side

28 (-1)27Emission roomon studs side

Rw in dB (C in dB)R in dB(A)

0

10

20

30

40

50

60

100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

R (d

B)

Emission room on boards side

Emission room on studs side

58

Field tests

CSTB tests in Colmar, France The building tested in Colmar is the only lightweight constructions thoroughly studied on site in the Acousvibra project, where flanking transmissions have been measured and analyzed and where prediction models have been validated.

(1) Description of the building

The building has a separate column-beam load bearing frame (Arcelor construction system) as shown in the different figures below; details of the connections between joists and load bearing frame are also given. Three types of junctions can be identified, as shown in Figure 4.39: junctions through load bearing beams in horizontal transmission (type 1), junctions through load bearing beams in vertical transmission (type 2) and junctions through load bearing columns (type 3)

Figure 4.36 Primary separate load bearing frame; connection between beam and column

Figure 4.37 Part of the separate load bearing frame of the building tested

Figure 4.38 Connections between joists an primary load bearing beam

59

Figure 4.39 Schematic of the 4 rooms tested (load bearing frame in red; joist orientation in doted lines)

The cross junction between the four rooms studied is given in Figure 4.40 (the T junction between floor and façades would be similar); the floor consists of a 22mm wood board with or without floating floor.

Figure 4.40 Cross junction between the 4 rooms studied

Junction types:

Type 1

Type 2

Type 3

60

(2) Measurements and calculations performed

The two storey four room building has been studied in 4 steps:

Step1: measurement of airborne and impact sound insulation between rooms (vertical, horizontal and diagonal transmissions) in the 2 cases of building with bare floor and with floating floor

Step2: measurement of vibration attenuation at junctions (key parameter for estimating flanking transmissions); the measurement method developed during the project, is based on standard pr EN 10848 and adapted to lightweight elements

Step3: calculation of the different transmission paths using EN 12354 modified for lightweight constructions (see model described in 4.1.2) and ranking of these paths. Direct transmissions were estimated from existing laboratory tests of all the separating walls and floors of the building. Ranking shows the influence of separate load bearing frame in flanking transmissions (see mains results and conclusions below).

Step4: re-composition of total airborne and impact sound insulation between rooms from all the calculated paths and comparison with measurement. The results compare quite well and thus validate the method (see main results and conclusions below).

(3) Main results and conclusions

The results of the tests for the 3 types of junctions identified and measured (step2) are the following:

(i) the floor-floor junctions with structural continuity through both floor boards and joists (see Figure 4.40) corresponds to the junction type 1 in Figure 4.39. These junctions have a poor vibration insulation of the order of 15-20 dB; this insulation is close to the values that would be obtained with junctions between monolithic elements;

(ii) the floor-wall or wall-wall junctions (vertical flanking paths) through the load bearing beams correspond to the junction type 2 in Figure 4.39. These junctions have rather good vibration insulation of the order of 30 dB or more; this shows the advantage of using a primary load bearing frame, which allows loose connections between walls and load bearing beams.

(iii) the wall-wall junctions (inner leaves) through the load bearing columns (horizontal flanking path) correspond to the junction type 3 in Figure 4.39. These junctions are not line connected (see point connections between column and beam in Figure 4.36) and these transmission paths can be neglected

The different paths were analyzed and ranked (step3) in terms of single value airborne sound insulation DnTw+C and single value impact sound level LnTw ; both cases of vertical and horizontal insulation between rooms were considered. Results particularly show that for buildings with primary load bearing frame, flanking is as important as direct transmission for impact noise in vertical transmission and that separating walls and floors equally participate in the horizontal transmission of impact noise. Figure 4.41 shows an example of sound insulation spectra either directly measured in the building or re-composed from the different flanking paths estimated using the CSTB prediction model; the shapes of the two spectra are very similar and the calculated paths show that in the case presented (horizontal transmission with floor without floor covering) the direct transmission is dominant at low frequencies and that flanking is dominant at mid and high frequencies.

61

20

30

40

50

60

70

80

100

125

160

200

250

315

400

500

630

800

1000

1250

1600

2000

2500

3150

4000

5000

Frequency (Hz)

DnT

(dB

)

Direct Path (measured)Flanking floor-wall (estimated)Flanking floor-floor (estimated)Flanking wall-floor (estimated)Total 4 paths (estimated)Measured

Figure 4.41 Airborne sound insulation spectra measured and calculated in the case of the horizontal transmission Cross junction between the 4 rooms studied

Ruukki field tests Ruukki have performed several field tests, mainly to compare laboratory performance of separating elements with in situ performance.

(1) Field Airborne sound insulation of outer walls with thermal studs

Several façades with the structure shown in Figure 4.42 have been tested on site. Details of the results can be found in test report “Promethor report 6405” and summary report “Promethor report PR-R-1149-1”. Comparisons between measured sound insulations and sound insulations re-composed from the laboratory performances of all the façade elements (façade wall, window, door and air inlet) according to EN 12354-3 (see 4.2.2 modeling) show that the predicted results of Dn,w are higher than the measured ones and that a safety margin of at least of 4 dB should be used.

In Finland where the acoustic requirement for facades is defined from the difference ΔL between the sound levels outside and inside the building, a safety margin of at least 7 dB in predicting Dnw should be used. It can be also roughly approximated that if measured with the loudspeaker method, the façade insulation D2m,n,w should be about 3 dB greater than the required ΔL .

Figure 4.42 Typical outer wall structure with thermal studs

62

Figure 4.43 Example of façade tested (from report PR-R-1149-1)

(2) Role of outer walls in flanking transmission

Ruukki have made field measurements of sound insulation between dwellings in order to determine the importance of flanking transmissions along façade walls in cases of heavy concrete buildings with lightweight outer walls as shown in Figure 4.44. The outer wall structure is the basic lightweight steel wall based on thermal studs carefully studied during the Acousvibra project.

Figure 4.44 Example of floor-outer wall junction measured on site

Measured results expressed in terms of apparent sound reduction index R’w between dwellings were compared to results measured in other buildings of the same type, but with heavy concrete façades: both results show that flanking transmissions are of the same order, but leading to apparent index R’w sometimes 5 or 6 dB lower than the index Rw of the separating element - concrete wall or floor -. This result shows that (i) lightweight façades have a similar effect on sound insulation between dwellings as heavy concrete façades and (ii) in both cases, flanking can be important in such building structures.

(3) Field performance of partition walls

Partition walls with acoustic wall studs AWS 150 (see Figure 4.45) have been tested on a construction site where junctions between flanking elements were carefully studied in order to reduce flanking transmissions to a minimum. The one storey building tested had a heavy floor lined by a lightweight access floor and all the other elements (walls and roofs) were lightweight with frame discontinuities at junctions (see Figure 4.46 showing a roof/partition junction).

Figure 4.45 Acoustic wall stud used in partitions tested on site

63

Figure 4.46 Detail of the partition/roof junction (building tested on site)

Horizontal sound insulations between rooms separated by the AWS partitions were measured and compared to the laboratory performance of the wall; comparisons showed that flanking was practically negligible in this case.

(4) Field tests of buildings with Kvantti floors

A new type of floor (the semi heavy Kvantti floor, see Figure 4.47) was both measured in laboratory and tested in the field. Unfortunately, the floor coverings were not the same in laboratory and in the field; so the impact sound insulation obtained could not be compared. However, in one case the floor coverings had some similarities (28mm of wood jointed floor in the lab and 2x15mm gypsum boards + carpet in the field, both mounted on rubber strips) so that very closed values of the R index could be assumed. The measured field performances were in this case very close to lab performance, showing that flanking was not significant for airborne noise vertical transmissions (see details on floor/façade junctions given in Figure 4.48).

Figure 4.47 Kvantti floor: case with plain jointed floor installed on timber battens and damping rubber strips

64

Figure 4.48 Detail on Kvantti floor/façade junction

4.3 Conclusion Acoustic modelling and testing correspond to tasks spread over two work-packages of the project: WP2 on modelling and WP4 on testing with the aim of better understanding and mastering both the acoustic performance of lightweight building elements, such as floors or walls, considered separately as well as the performance of the whole building.

4.3.1 Acoustic performance of lightweight building elements The acoustic performance of building elements considered separately can be either measured in laboratory or estimated using models. In the project, existing models were used but also new models were developed in order to better understand the role of the different components (boards, cavity, studs, rails…) in sound transmission through lightweight building elements. Three different modelling methods were used and compared: an energy based method (SEA), a simple wave based analytical method and different numerical methods (finite elements – FEM; spectral finite elements – SFEM; and boundary elements - BEM). A rather accurate and fast model has been obtained by combining wave approach and SEA. The numerical methods allow a finer level in the geometrical description, but are more time consuming. Most of the modelling work has been focused on understanding the role of metal frames (studs and rails) in airborne sound transmission, particularly in the case of single frame double walls where the frame plays a dominant role. The models show that the key parameter is the section stiffness of studs and rails; therefore, efforts were put into developing laboratory methods for estimating this key parameter; the characterization method works and rail or stud section stiffness can be measured and then used as input data in the model for calculating the R index of the wall considered. Moreover, a 3D numerical model has been developed to reproduce the laboratory method for characterizing studs; the model can numerically estimate the section stiffness of new stud profiles, existing only on paper.

All the models developed or improved during the project have been validated and compared to each other, particularly in order to find out their limits: (i) it seems that in the case of a double wall, the wave approach is more accurate than SEA for modelling the transmission path through the wall air cavity, dominant at low frequencies; SEA is however well adapted for modelling the structural transmission paths through all the mechanical contacts between the two leaves of the wall at mid and high frequencies; the best model seems therefore to be an hybrid model combining both wave approach and SEA; (ii) when a finer level in the geometrical description is needed (optimizing stud profile for example), numerical models can then be used; in this case, 2D models seem to be a good compromise between computing time and accuracy; the absolute results given can be inaccurate, but at least, 2D models are capable of giving the right ranking between different solutions.

65

Two parametric studies have also been performed using two different models; the results are comparable and clearly show the dominant role of the stud translational stiffness (compare to rotational stiffness) in the sound transmission through the studs.

Many building elements (mainly outer walls and separating walls) have been tested in laboratory during the project, for different reasons: (i) because the building element was new and its acoustic performance unknown, (ii) in order to validate models, (iii) in order to compare the performance of the element tested separately in laboratory to its performance in the building on site.

Not much work has been done on modelling impact noise, mainly because of the difficulty in modelling impact excitation which requires a good knowledge on lightweight structure input mobility. The same difficulties would also arise in studying structure borne noise from building technical equipment (such as waste water pipes, heating devices…), which has not been considered in the Acousvibra project. It should be mentioned that work on this difficult subject has started at European level in the frame of two of the TC126 standardization working groups.

4.3.2 Acoustic performance of lightweight buildings Two problems have been considered: (i) the problem of airborne and impact sound transmissions between dwellings, which include both direct transmission through the separating element (wall or floor) and flanking transmissions through other elements and their junctions; the main goal of the project was to understand, quantify better and if possible reduce flanking transmissions in the case of steel frame lightweight constructions; (ii) the problem of airborne sound transmission through façades, which includes the transmission through all the elements composing the façade (wall, windows, doors, air inlets…).

For the first problem, a European standardized prediction model for calculating the building performances, only valid for heavy concrete structures, has been adapted to lightweight structures. The model input data are mainly the performance of the building elements (index R and impact level Ln) and two less known parameters: (i) the performance of the junctions between elements, which can be measured either in laboratory or on site, and (ii) the radiation efficiency of the building elements, which can be measured in laboratory. During the project, this prediction model was validated on an existing small building thoroughly tested on site; moreover, the performance of all the building elements was tested separately in laboratory. The prediction model was able to estimate rather accurately the sound insulation corresponding to the different transmission paths in the building. It was particularly shown than for the lightweight building tested, which had a separate load bearing column-beam frame (which allows loose connections between walls and load bearing frame), flanking transmissions were still present, particularly in the case of impact noise and for both vertical and horizontal transmissions.

A attempt has also been successively made in estimating two of the rather numerous input parameters of the prediction model using the finite element method; (i) a 2D FEM/BEM model has been developed for estimating the radiation efficiency of lightweight stiffened panels; the results are rather good, showing the extra radiation provided by the stiffeners; (ii) a 2D SFEM model has been developed for estimating the vibration level difference at junctions between building elements; this 2D model is capable of giving the right ranking (compared to measured results) between the different flanking paths.

Other tests have been performed on site during the project: (i) several buildings carefully studied in order to reduce flanking (avoiding any structural continuity at junctions between elements) has been successfully controlled, showing that flanking can be almost suppressed; (ii) heavy concrete buildings with lightweight façade have been tested in order to evaluate the importance of flanking through the façade in the airborne sound insulation between apartments; the results show that overall flanking is rather important, but of the same order as for heavy concrete façades.

For the second problem of airborne sound transmission through façades, a European standardized model for calculating the façade performance from the performance of all the façade elements (wall, windows…) has been used and validated on site; it seems that the European model is too optimistic for lightweight structures and that a safety margin has to be taken into account.

66

4.4 Summary table of the acoustic tests performed Type of test

Element tested aim report

Outer walls: multilayered outer wall with thermal studs Number of elements tested: 4

- new type of walls - comparison with field tests

- VTT report RTE 149-05 - FIOH report 5210-2004-25974e

Standardized Laboratory measurements of index R

Separating walls: single frame double wall with new types of studs Number of elements tested: 4

- new type of studs - validation of models

- FIOH report TY02-2006-31771E - FIOH report TY02-2006-34973E

- Standardized Laboratory measurements of R and Ln- Non standardized laboratory tests of radiation efficiency

Single leaf wall and floor Number of elements tested: 2

- performances required for estimating flanking

- Acousvibra technical report CSTB CR02025 D002

Non standardized laboratory tests of index R

Double leaf panels Number of elements tested: 2

- influence of connections between panels

- Acousvibra technical report UPC CR03025 E008

Studs and rails Number of elements tested: 12

- measurement of stud and rail section stiffness

- Acousvibra technical report CSTB CR02025 D003

Laboratory characterization of building components (input data for models)

Single and double leaf panels Number of elements tested: 3

- measurement of loss factor of plates - measurement of plate and cavity natural frequencies

- Acousvibra technical report VTT CR02025 F002

- Field measurements of airborne and impact sound insulation - measurement of vibration reduction index of junctions

2 storey building in Colmar France 1 building with separate load bearing frame

- complete analysis of transmissions paths between rooms

- Acousvibra technical report CSTB CR02025 D002

67

Outer walls with thermal studs 4 facades tested

- Comparison with results predicted using European model EN 12354-3

-Promethor report PR-R-1149-2.

Lightweight Outer walls mounted in heavy concrete buildings 3 buildings tested

- check the importance of flanking (through outer wall)- comparison with heavy façades

Promethor PR-R1116-1 Promethor PR-R1116-2 Promethor PR-R1193-1

Partition walls with acoustic studs (AWS 150) 1 building tested

- check the importance of flanking

H. Helimäki report 3133

Field measurements of airborne sound insulation

Kvantti floors 1 building tested

- check the importance of flanking

VTT memo (7.4.2004)

68

5 VIBRATION MODELLING AND TESTING

5.1 Introduction and objectives This section covers modeling, simulation, dynamic testing and subjective evaluation of vibrations in lightweight floor structures.

The main objective of modeling and simulation of vibration performance of structures –chapter 5.2 (Task 2.2) - is to obtain an understanding of which structural parameters have the most significant influence on vibration properties of floors. The modeling is performed with finite element in which also connections and support has been incorporated.

A group of field- and laboratory tests – chapter 5.3-5.4 (Task 5.1) - have been carried out with, to high extent, a unified testing procedure. The dynamic properties of the floors have been measured together with subjective evaluations where the latter concerns both body perception and vibration induced noise and trembling of articles. The test results have then been compared to vibration models – chapter 5.5 (Task 5.1) - which gives possibilities to update and refine the models afterwards.

The role of connections in vibration performance of structures – chapter 5.6 (Task 5.2) – have been studied through a number of laboratory test in order to increase the knowledge about how connected structures influence the vibration properties of lightweight floors.

Laboratory test with a, within the project, developed motion simulator to subjectively evaluate some special phenomena related to floor vibrations has been performed – chapter 5.7 (Task 5.3). Mainly artificial vibration signals with specific characteristics have been used but also some vibration signals- slightly modified though – measured in field have been reproduced thorough the simulator for assessment.

5.2 Modeling and simulation of vibration performance of structures (Task 2.2)

Finite element modelling Finite element modelling (FEM) is an excellent tool in order to better understand the dynamic behaviour of a floor and it is well suited for parametric studies of construction changes of the floor. As a first example, the floor described below was used for a three dimensional FE model. Later on, the same floor was used for dynamic and subjective tests in laboratory environment.

The floor The model floor is shown in Figure 5.1. The span was 6.0m and the width 3.6m. It consisted of load bearing C-profiled beams joined together by two rim beams. A profiled metal sheet added stiffness transverse the main beams. The top-layer consisted of either two 15mm plasterboards or 50mm concrete. The ceiling consisted of a double layer of plasterboard resiliently connected to the C-beams by flexible ceiling joists. The ceiling joists behaved like a spring and the purpose of the design was to reduce vertical sound and vibration transmission. The floor was connected to a tubular beam which in turn rested upon rigid concrete supports. The support was chosen both to allow access to the ceiling and to approximate the grounded condition, i.e. to avoid any interaction between floor and support in the vertical direction in order to measure the true vibration response of the floor. The floor was categorized as “lightweight” when plasterboard was used as covering and as “semi-heavyweight” in the case of concrete covering. The weight of the floor was about 70 and 170 kg/m2 for the plasterboard and concrete case respectively.

69

Figure 5.1 Section of the modelled floor with its support.

Eight test setups involving three variables, each of which could be varied in two ways, were considered The floor covering is either plasterboard or concrete, the connection between the tubular beam and the concrete support was either continuous or by four corner points, and finally the two long sides were either free or supported by vertical floor braces, three at each side. The configurations are numbered 1-8 according to Table 5.1 For all configurations, the floor was assumed to be bare, without any significant effect from the surroundings.

Table 5.1 The eight different setups of the floor.

Setup No Covering Main support Long sides 1 Plasterboard Continuous Supported 2 Plasterboard Continuous Free 3 Plasterboard Corners Free 4 Plasterboard Corners Supported 5 Concrete Corners Free 6 Concrete Corners Supported 7 Concrete Continuous Free 8 Concrete Continuous Supported

5.2.1 FE - Model A The model was built up as follows.

Top layer, plasterboards / concrete - homogenous isotropic plate, shell elements. Sheet metal – slender beam elements in transverse direction. C-beam – beam elements. Rim beam – beam elements. Ceiling joist – spring elements combined with beam elements Ceiling plasterboards – homogenous isotropic plate, shell elements. Tubular beam – beam elements. Concrete supports – grounded. All elements used isotropic material characteristics where the related cross sections and densities corresponded to either measured or calculated data from the floor parts. An isotropic behaviour was assumed for the floor properties besides the sheet metal which indeed is orthotropic. The stiffness in the transverse direction is considerable higher than in the main direction. In the model, this behaviour is approximated by beam elements, which yield the correct stiffness in the transverse direction with only a minor contribution to the stiffness in the main direction. As an alternative, the sheet metal could be

1 2 3a

4 5

6 7

3b

1) 2x15mm plasterboard (or 50mm concrete) 2) 31mm profiled sheet metal (18mm in case of concrete) 3a) 350mm C-beam, t=2.0, c600 3b) 354mm rim beam, t=2.0 4) 25mm ceiling joist 5) 2x13mm plasterboard 6) 8x250x150mm tubular beam 7) 80mm wide concrete support

70

modelled by orthotropic shell elements. The nodes of the tubular beams controlled the main boundary condition. Depending on the type of support, either all nodes or the four corner nodes were clamped. Some additional restraints were used. The sheet metal and top layer were restrained to the C-beams for both translations and rotations and the same applied to the rim beams in relation to the tubular beams.

Results The first two natural frequencies in the vertical direction are shown in Table 5.2. The first frequency f1 refers to the fundamental frequency, a bending mode of first order, and the second frequency f2 refers to the first torsion mode. The true mode shapes can be seen in Figure 5.5. Table 5.2 The first two natural frequencies obtained from FE model A for the eight setups.

Setup # Freq.

1

2

3

4

5

6

7

8

f1 (Hz) 10.3 10.2 10.0 10.1 8.1 10.1 8.5 10.3 f2 (Hz) 12.8 10.6 10.6 12.7 9.5 18.7 9.6 19.8

5.2.2 FE - Model B - including orthotropic elements The same floor is here modelled with a different approach in which FE with orthotropic characteristics have been used. The model can be divided into four layers; 1) the to tubular beams, 2) the C-beams and rim beams, 3) the profiled sheet metal in conjunction with floor plasterboard or concrete and 4) the ceiling joists in conjunction with the ceiling plasterboard. 1) and 2) are represented by beam elements while 3) and 4) are modelled by using orthotropic shell elements, i.e. elements with different properties depending on direction. The composite floors are oriented in such a way that they possess higher stiffness in direction perpendicular main beams.

This effect will be modelled by means of orthotropic shell elements. The parameters required are then: E1 (Young’s modulus in the stiffest direction), E2 (Young’s modulus in perpendicular direction), h (thickness of the shell element, is the same for both directions), ρ (density) and ν (the poisson ratio). These parameters have been chosen in order to reproduce the actual mechanical and geometrical properties of the composite floors. The first hypothesis assumed here is the uniqueness of poisson ratio for the composite material νxy= νyx= ν. The other parameters will be determined by studying states of pure axial stress and pure bending stress in every orthogonal direction. The pure axial stress states are characterised by the imposed deformation fields:

),0,,( === xyyyxx γεεεε )0,0,0( === xyyx γεε . (5.1)

The pure bending states are characterised by imposed deformation fields of the type; ( ) ( )0,,0,0,0,

2

2

2

2

22

2

2

2

2

====== ∂∂∂

∂

∂

∂

∂∂∂

∂

∂

∂

∂

∂yx

uyy

uxu

yxu

yu

xxu zzzzzz κκ . (5.2)

The described constant deformation fields will be utilised in order to impose the resultant force, due to axial stress in the section of orthotropic plate, to be the same as in the composite floor section for both orthogonal directions;

)1(

2

)1(//2

/12 1

11

11

1ss

s

gcgcgc

AEAEhEννν −

+−

=−

(5.3)

)2(

2

)2(//2

/22 1

11

11

1ss

s

gcgcgc

AEAEhEννν −

+−

=−

(5.4)

The same thing can be imposed for the bending moments;

)1(

2

)1(//2

/

312 1

11

1)1(12

1ss

s

gcgcgc

IEIEhEννν −

+−

=−

(5.5)

71

)2(

2

)2(//2

/

322 1

11

1)1(12

1ss

s

gcgcgc

IEIEhEννν −

+−

=−

(5.6)

Finally, it is imposed that that the total mass of the FE shell must be equal to the mass of the composite floor;

ssgcgc AAh ρρρ)1(

/)1(/ += (5.7)

The following notation is used in the equations: E (Young’s modulus), A (Area per unit length of the cross section), I (moment of inertia per unit length of the cross section), c/g (to distinguish between the concrete and plasterboard case), s (steel), (1) (to indicate the direction with highest stiffness, direction Y), and (2) (to indicate the orthogonal direction with lower stiffness, direction X). The numerical value of these parameters can be found in [1]. Five unknown parameters (E1, E2,, h, ρ, ν) remain and as a result a system of five non-linear equations ((5.3)-(5.7)) is to be solved. However, the system is almost unaffected by all possible poisons ration ν (0.01-1.0). This means that in practise four parameters are to be determined by the five equations since ν here is set to be the mean poisson ratio of two materials;

2

/ sgc ννν

+= (5.8)

The criteria in order to determine E1, E2 and h will be: - Equations (5.5) and (5.6) will be exactly satisfied. That means that the FE shell will have the

same bending stiffness as the composite floor for both directions. - E1, E2 and h will then be adjusted to minimise the error in the axial stiffness of the FE shell. - Once E1 and E2 are determined, ρ can be obtained from equation (5.7).

Results The first two natural frequencies in the vertical direction for some setups are shown in Table 5.3. Table 5.3 The first two natural frequencies obtained from FE model B for some setups.

Setup # Freq.

3

4

5

6

f1 (Hz) 13.9 17.0 8.5 12.9 f2 (Hz) 21.1 30.2 14.2 25.8

5.2.3 Prediction of walking response The same floor now serves as an example how the response from walking can be simulated by FEM. The model B described above has been used. A walking person is supposed to follow a straight line according to Figure 5.2. The walking frequency is set to 5/3 Hz and the step length to 0.75m.

Figure 5.2 Walking line along the floor, FEM.

72

In Figure 5.3 the walking force in time domain is presented [29]. For the model, nine point-like time dependent forces have been considered. The numerical value of the force is 1N multiplied with the time dependent function in Figure 5.3 which is the relation between static and dynamic force during the period of time when the foot is in contact with the floor. The results can easily be scaled to fit the weight of the walking person. Calculated response displacements as a result of a 100kg person should then be multiplied by 981.

Figure 5.3 Dynamic action of a human step.

An important aspect in time step analysis is the consideration of damping. A dynamic system might often be considered as being 1) undamped, 2) Rayleigh damped or 3) modal damped. As an example; if a modal damping ratio of 5% for the first five modes of vibration is assumed, the resulting vertical displacement of the floor - in the walking path and in the mid span - is shown in Figure 5.4.

Step positions

Along the floor: A 0.0m B 1.2m C 2.4m D 3.6m E 4.8m F 6.0m

Across the floor: 0.9m

a)

73

Figure 5.4 Vertical displacement due to walking; a) along the floor and b) across the floor.

5.3 Laboratory tests on the vibration performance of lightweight floors (Task 5.1)

The modeled floor described in chapter 5.2 and Figure 5.1 was here used for dynamic testing and subjective evaluation, all in laboratory environment. Both coverings, plaster board and concrete, were tested in total eight different combinations according to Table 5.1. Main properties and descriptions of the tested floors are shown in Annex A.

5.3.1 Dynamic test Modal test The modal parameters of the floor were tested for eight different configurations as given in Table 5.1. Complementary subjective evaluations also were performed. In the dynamic measurement, the input force was generated by a LDS v406 electromagnetic shaker in conjunction with an amplifier LDS PA100E. The force was measured by a Brüel and Kjær (B&K) transducer type 8001 combined with a B&K charge amplifier type 2635 and the response by accelerometers B&K type 4508. B&K type 2827 served as front-end and B&K’s Pulse® was used as software. The input force was applied to one of the C-beams from underneath. The response was measured at 154 points, 77 on the floor surface and 77 on the ceiling, equally distributed on each surface respectively to represent the averaged spatial response. The response from walking traffic was measured in defined observation points described below. The acceleration was measured in both vertical and horizontal directions.

Results The first two natural frequencies and related damping ratios are presented in Table 5.4. It can there be seen that the fundamental frequency for the plasterboard set-ups were in the interval 10.1-10.4 Hz, while for the concrete cases, were 10.0-11.7 Hz. For the support at the long sides, the fundamental frequency, f1, was about 1 Hz higher (1.3 Hz for No 1 vs. No 8 and 0.9 Hz for No 4 vs. No 6) when using concrete as a top layer, compared to the plasterboard. With the long sides left free, the fundamental frequency was not affected by the kind of top layer (0.0 Hz for No 2 vs. 7 and 0.1 Hz for No 3 vs. No 5). In a similar way, the long side supports, compared to free sides, only affected the fundamental frequency when concrete was used. It was 1.4 Hz higher, (for both No 5-6 and for No 7-8) but almost nothing (0.2 Hz for No 1 vs. No 2 and 0.1 Hz for No 3 vs. No 4) in the case of plasterboards.

The type of main support, continuous vs. corners, shows minor changes, of 0.1-0.3 Hz for the fundamental frequency for the continuous support (0.1 Hz for No 1 vs. No 4, 0.2 Hz for No 2 vs. No 3, 0.3 Hz for No 5 vs. No 7 and 0.3 Hz for No 6 vs. No 8). The second natural frequency, f2, corresponds to the first torsion mode and is very much affected by the support at the long sides. In the plasterboard cases, the frequency increases from about 13.5 Hz to about 15.3 Hz after installing the support.

Step positions

Along the floor: 3.0m

Across the floor A 0.3m B 0.9m C 1.5m D 2.1m E 2.7m F 3.3m

b)

74

The mode shapes are presented in Figure 5.5. At around 14 Hz, there is a mode with high similarity to the fundamental when plasterboard is used and at around 14-16 Hz when concrete is used. For this mode, the shape of the floor surface is the same as for the fundamental mode while the ceiling oscillates in a second order bending mode in the main direction, i.e. with two bellies having opposite phase to each other.

Table 5.4 Frequency (f) and damping ratio (ζ) for the fundamental mode (f1) and the first torsion mode (f2).

f1 f2 Setup No f (Hz) ζ(%) f (Hz) ζ(%)

Covering Main support

Side support

1 10.4 1.9 15.2 0.8 Plasterboard Continuous Supported2 10.3 1.3 13.5 0.6 Plasterboard Continuous Free 3 10.1 1.3 13.6 0.6 Plasterboard Corners Free 4 10.3 1.5 15.3 0.8 Plasterboard Corners Supported5 10.0 1.2 12.0 1.5 Concrete Corners Free 6 11.4 1.5 16.4 1.1 Concrete Corners Supported7 10.3 1.4 10.8 1.5 Concrete Continuous Free 8 11.7 2.0 16.4 0.9 Concrete Continuous Supported

Figure 5.5 Mode shapes of the a) fundamental frequency and b) first torsion mode.

5.3.2 Subjective evaluations The subjective evaluations were performed in accordance to a procedure described in chapter 3 and in [30] but an overview is given here. The evaluation consisted of two parts. In the first, which is about body perception, the test subject sat on a chair in the “observation point” while the experiment leader walked along the floor in one of two predefined routes. If the fundamental floor frequency is found to be above 10 Hz, Talja stipulates a walking pace of 2 Hz, while the pace in the case of a fundamental frequency below 10 Hz should be adjusted so that a complete number of harmonics exactly match the actual frequency. The walking pace should be set as high as possible without exceeding 2 Hz, e.g. if the fundamental frequency is 9.0 Hz, the corresponding walking pace should be 1.8 Hz. The leader walked back and forth over the floor, passing the subject three times. The subject’s observation point is 60 cm from the walking line, (see Figure 5.6 for walking lines and observation points). The test subject marked in a protocol the intensity of the body perception and the degree of acceptance.

The second evaluation was about vibrating articles. Four different articles, each with a special vibration/sound pattern were used: clinking of a coffee cup (a cup together with a saucer and a spoon), leaf movements of a 30-40 cm high pot plant, rippling of water in a glass bowl and chinking of a glass pane. The objects were put, one at a time, on a 1.2 m. high tripod besides the glass pane which was mounted vertically on the side of the tripod by mirror hooks. The tripod was placed at observation point 1 and the test subject stood at observation point 3 while the experimental leader walked along the path according to Figure 5.6.

a b Floor surface

Ceiling surface

75

Figure 5.6 Walking lines and observation points used for subjective evaluations of vibration intensity (point 1), and for subjective evaluations of vibrating articles (tripod in point 1, test subject in point 2).

When judging, the subjects were instructed to assume that the floor was a part of a living room in a new building. A four-point scale, ranging from zero to three was used in accordance to Talja et al. The vibration intensity was assessed as imperceptible (3), barely perceptible (2), clearly perceptible (1) or strongly perceptible (0) and the vibration acceptability was assessed as absolutely acceptable (3) , yes – acceptable (2) , no – unacceptable (1) or absolutely unacceptable (0).

Results Ten test subjects, all university personnel, took part in this study. In Figure 5.7, the outcome of the body perception tests is shown for the two walking paths, along the centre and along the edge of the floor. (Note that the latter test was not performed for setup No1) The means, together with 95% LSD-interval, are shown. The subjects are, in general, critical of the floor and none of the setups is judged as acceptable. The best floor – but still not good – is setup No 8 where the mean intensity/acceptance is within the interval 0.7-1.0. The actual setup which is concrete, continuous and supported long sides, also represents the stiffest setup.

Figure 5.7 Rating of vibration from walking along the centre of floor; a) intensity and b) acceptance and from walking along the edge of the floor; c) intensity and d) acceptance.

The outcome from the vibrating article tests is more positive and the results vary more between the floor setups. The acceptance for the four articles is presented in Figure 5.8. The leaf movements of the pot plant shows about the same acceptance for all test setup but for the remaining three articles, there is a difference depending on the top layer. A majority of the mean acceptance values are significantly higher for the concrete cases compared to plasterboard cases. The acceptance ranges for the concrete and plasterboard layer respectively are; 2.2-3.0 and 1.4-1.8 for the coffee cup, 1.7-2.1 and 1.0-1.2 for the glass bowl and 1.2-2.0 and 0.2-0.6 for glass pane.

1 2

a)

Centre walking

Test No.

Bod

y pe

rcep

tion,

Inte

nsity

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Centre walking

Test No.

Bod

y pe

rcep

tion,

acc

epta

nce

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Edge walking

Test No.

Bod

y pe

rcep

tion,

inte

nsity

2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Edge walking

Test No.

Bod

y pe

rcep

tion,

acc

epta

nce

2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3a) b) c) d)

76

Figure 5.8 Vibration acceptance from articles, a) clinking of a coffee cup, b) leaf movements of a pot plant, c) water rippling in a glass bowl and d) chinking of a glass pane.

5.3.3 Analysis The negative results of the body perception tests initially seemed somewhat perplexing. All the floor setups were assessed as unacceptable which was not expected, at least not for the concrete layer setups. During the experiments, the walking response was measured both horizontally and vertically and the total horizontal acceleration was found to be considerably lower than the vertical. However, a frequency analysis in the horizontal direction indicated a major response at around 5 Hz and a further dynamic test revealed that the experimental setup suffers from a strong natural frequency of 5.1 Hz. The procedure of the test was that one tubular beam was excited horizontally by an impulse force while the response was measured in three points; on the rim joist, on the tubular beam and on the concrete support. Since the response on the concrete supports is significantly lower that the others it is suggested that the natural frequency is a result of vibration at the junction between the concrete support and tubular beam. This particular vibration is a result of the experimental setup. For installations in buildings, the beam connection to the other building frame components normally prevents the horizontal vibration more effectively.

To obtain subjective responses without the horizontal vibration, the tubular beams were removed from the setup, i.e. the rim joist rested directly on the concrete supports. A new test, referred to as test No 9, regarding body perception, was then performed with identical conditions as setup No 7: concrete layer, continuously supported and free long sides. A dramatic change in perception occurred. The mean acceptance increased to 2.0 units and the floor setup is therefore considered as acceptable. See Figure 5.9 for details.

Figure 5.9 Rating of vibration from walking along the centre of floor; a) intensity and b). Test No 9 is identical to test No 5 besides that the tubular support beams have been removed

Coffee cup

Test No.

Acc

epta

nce

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Pot plant

Test No.

Acc

epta

nce

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Glass pane

Test No.

Acc

epta

nce

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3

Glass bowl

Test No.

Acc

epta

nce

1 2 3 4 5 6 7 80

0,5

1

1,5

2

2,5

3a) b) c) d)

Centre walking

Test No.

Bod

y pe

rcep

tion,

Inte

nsity

1 2 3 4 5 6 7 8 90

0,5

1

1,5

2

2,5

3

Centre walking

Test No.

Bod

y pe

rcep

tion,

Acc

epta

nce

1 2 3 4 5 6 7 8 90

0,5

1

1,5

2

2,5

3

a) b)

77

5.4 Field tests on the vibration performance of lightweight floors (Task 5.1)

The results from four different field tests referred to as 1) Brixton, 2) Derby, 3) Daventry and 4) Kauklahti are here summarised. Further information of the test floor, test procedures and results are found in the reports [31] [32] [33] and [34], respectively. Summary of floor descriptions are given in Annex A.

5.4.1 Brixton A lightweight floor in the Loughborough Community Centre in Brixton, London, was tested. The brief of the testing was to estimate experimentally modal properties and to measure dynamic response from walking. During the tests, most of the partitions were erected.

Test procedure A plan of the tested floor is shown in Figure 5.10 together with the 16 test points (TPs) and three walking paths (WPs). The test points were selected so that a good representation of potential mode shapes and the effect of partitions could be established but also so that representative levels of vibration could be recorded in the pedestrian tests where test persons with a weight of about 85 kg were used. Pacing rates were set to 1.4 to 2.0 Hz. For the modal test, two electrodynamic shakers excited the floor simultaneously by uncorrelated random forces.

Figure 5.10 Plan of the tested floor. 16 test points and three walking paths.

Results Results in form of the modal parameters natural frequency and damping are presented in Table 5.5 for the first ten modes of vibration.

78

Table 5.5 Natural frequencies and damping ratios for the first ten modes.

Mode number 1 2 3 4 5 6 7 8 9 10 Frequency, Hz 14.4 14.6 15.0 17.4 17.9 19.2 19.8 20.6 22.9 23.5 Damping, % 2.8 1.0 4.7 4.2 4.8 8.7 4.3 8.4 3.8 11.9

The walking response for some selected cases is presented in Table 5.6 after that each measured time response was weighted using the wg weighting function [35]. The following parameters have been calculated from the experimental data; vibration dose value (VDV), maximum transient vibration value (MTVV) which is the maximum value of 1s running RMS [36] and R factor [37].

Table 5.6 Pedestrian response data.

Walking Path 1 1 1 2 2 2 3 3 3 Pacing Rate 1.4 1.7 2.0 1.4 1.7 2.0 1.4 1.7 2.0 MTVV [m/s2] 0.112 0.137 0.161 0.195 0.189 0.0735 0.0873 0.0791 0.116 R factor 22.4 27.3 32.2 39.0 37.7 14.7 17.5 15.8 23.2 VDV [m/s1.75] 0.104 0.133 0.160 0.189 0.176 0.068 0.0848 0.0632 0.128

5.4.2 Derby A lightweight dwelling floor in the “Rosebury” house, Figure 5.11, Royal County Park, Chellaston, Derby, was tested. The brief of the testing was to estimate experimentally modal properties and to measure dynamic response from walking.

Figure 5.11 Test house, Derby.

Test procedure A plan of the tested floor is shown in Figure 5.12 together with the 16 test points and the walking path. The test points were selected so that a good representation of potential mode shapes and the effect of partitions could be established but also so that representative levels of vibration could be recorded in the pedestrian tests where test persons with a weight of about 85 kg were used. Pacing rates were set to 1.4 to 2.0 Hz. For the modal test, two electrodynamic shakers excited the floor simultaneously by uncorrelated random forces.

79

Figure 5.12 Plan of the tested floor. 16 test points and the walking path.

Results Results in form of the modal parameters natural frequency and damping are presented in Table 5.7 for the first five modes of vibration. Besides these “global” modes, some local modes were found, the lowest one at 14.3 Hz.

Table 5.7 Natural frequencies and damping ratios for the first five modes.

Mode number 1 2 3 4 5 Frequency, Hz 5.7 7.2 9.2 9.6 10.5 Damping, % 11.3 8.4 3.2 2.1 2.1

The walking response for some selected cases is presented in Table 5.8 after that each measured time response was weighted using the wg weighting function.

Table 5.8 Pedestrian response data.

Pacing Rate 1.4 1.6 1.8 2.0 MTVV [m/s2] 0.0637 0.0580 0.0984 0.1054 R factor 12.7 11.6 19.7 21.1 VDV [m/s1.75] 0.0512 0.0568 0.0920 0.111

5.4.3 Daventry Dwelling floors in a house developed by Advance Housing Ltd, Figure 5.13, Heartlands Business Park, Daventry, was tested. The brief of the testing was to estimate experimentally modal properties and to measure dynamic response from walking.

80

Figure 5.13 Test house, Daventry.

Test procedure A plan of the two tested floors is shown in Figure 5.14 together with the 14 test points and four walking paths. The test points were selected so that a good representation of potential mode shapes and the effect of partitions could be established but also so that representative levels of vibration could be recorded in the pedestrian tests where test persons with a weight of about 85 kg were used. Pacing rates were set to 1.4 to 2.0 Hz. For the modal test, an electrodynamic shaker excited the floor by a random force.

Figure 5.14 Plan of the tested floosr. 14 test points and four walking paths.

81

Results Results in form of the modal parameters natural frequency and damping are presented in Table 5.9 for the first five modes of vibration. Besides these “global” modes, some local modes were found, the lowest one at 14.0 Hz.

Table 5.9 Natural frequencies and damping ratios for the first five modes.

Mode number 1 2 3 4 5 Frequency, Hz 5.9 8.0 11.1 12.5 15.3 Damping, % 3.4 2.4 2.1 2.8 3.5

The walking response for some selected cases is presented in Table 5.10 after that each measured time response was weighted using the wg weighting function.

Table 5.10 Pedestrian response data.

Walking Path 1 1 1 1 2 2 2 2 3 4 Pacing Rate 1.4 1.6 1.8 2.0 1.4 1.6 1.8 2.0 N/A N/A MTVV[m/s2] 0.005 0.006 0.010 0.010 0.044 0.051 0.065 0.093 0.011 0.006 R factor 0.97 1.27 1.97 1.98 8.72 10.2 12.9 18.5 2.24 1.12 VDV [m/s1.75] 0.004 0.006 0.010 0.012 0.038 0.045 0.058 0.101 0.010 0.005

5.4.4 Kauklahti The test object is steel framed floors with concrete deck inside two one-family houses located in Kauklahti, Espoo, Finland. The brief of the testing was to measure modal parameters, walking induced vibrations, static deflection and subjective rating.

Floor construction The construction of the floors is presented in Figure 5.15 and Figure 5.16. The actual dimensions are as follows: House 1 - floor 1: span 4.2m and width 8.5m, house 1 – floor 2: span 4.35m and width 7.4m, house 2: two span structure, span 3.5m and width 4.1m.

Figure 5.15 Floors in house 1. To the left: floor 1, (1) and (6) were not installed when tested. To the right: floor 2

1 Mortar layer + tile 10mm 2 Concrete 55-85mm 3 Corrugated steel sheet RAN 35-0.7 4 Floor beam C200-2.0, c/c600mm 5 Hat profile c/c400mmm 6 15mm plaster board

1 Mortar layer + tile 10mm 2 Concrete 55-85mm 3 Corrugated steel sheet RAN 35-0.7 4 Thermo Floor beam TC200-2.0, c/c600mm 5 12mm Wind shield

82

Figure 5.16 Floor in house 2.

Test procedure Plans of the tested floors of the two houses are shown in Figure 5.17 and Figure 5.18 together with the test points and walking paths for each floor. The subjective rating should be performed with at least five observers which give their opinion both in terms of body perception and in form of sense perception from vibrating articles standing on a tripod when another person walking by at a pace of 2 Hz. The used articles are clinking of a coffee cup with a spoon in the cup and on a saucer, leaf movements of a plant, rippling of water in a glass bowl and chinking of a glass pane. A full description of the test procedure can be found in [30] and [38]. In this study only three to four observers were available and all the mentioned articles were not used.

Figure 5.17 Test points and walking paths of floors in house 1.

1 Surface structure 2 Concrete 52-70mm 3 Corrugated steel sheet RAN 20-0.7 4 Floor beam C200-2.0, c/c250mm 5 Hat profile c/c400mmm 6 Primary beams CFRHS 200x100x6, centre restraint CFRHS 150x100x6, edge restraints 7 Suspended ceiling: Hat profile + 15mm plaster board

83

Figure 5.18 Test points and walking paths of the floor in house 2

Results Results in from of the modal parameters fundamental frequency (the lowest natural frequency) and damping are presented in Table 5.11.

Table 5.11 Fundamental frequency and damping.

Fundamental frequency [Hz]

Damping [%]

House 1, floor 1 18.5 4-5 House 1, floor 2 25.0 4.5-6 House 2, point 1 11.1 5-7 House 2, point 2 11.1 4.5-5.5

Static deflection as well as the walking induced based parameters maximum peak displacement and response factors (R) are shown in Table 5.12. The static deflection, δ, is the result of a 1 kN point load, the maximum peak displacement, |umax|, and the so called ISO-factor – based upon frequency weighted acceleration [30] - are determined from the measurements.

Table 5.12 Measured maximum peak deflection (|umax|), ISO-factor and static deflection (δ).

δ [mm] |umax| [mm] ISO-factor House 1, floor 1 0.03 0.055 14 House 1, floor 2 - 0.045 16 House 2, point 1 (Horizontal x) (Horizontal y)

0.025 0.080 (0.02) (0.02)

10 (0.5) (0.9)

House 2, point 2 - 0.08 10

The result of subjective evaluation is given in Table 5.13. The result shows that for body perception, all floors were accepted by all observers. For vibrations of articles, in the case of one point (house 1, point 1), two out of four persons found it acceptable..

Usually the criterion is that the floor is found acceptable by the majority of observers [30]. Although the criterion may appear too loose, it should be kept in mind that the test look at the worst points of the floor, that quite high walking velocity is used and that in a test situation the observers are probably more critical. Additionally, in a real residential situation the threshold for complaints is considerably higher than the acceptance level in a test situation.

84

Table 5.13 Subjective rating of vibrations

INTENSITY OF VIBRATIONSThe vibrations are - imperceptible (No) - barely perceptible (B) - clearly perceptible (C) - strongly perceptible (S)

ACCEPTABILITY OF VIBRATIONS Is the floor acceptable in a newly built living room? + yes ++ absolutely acceptable - no - - absolutely unacceptable

House 1, point 1 (room1) Intensity Acceptability

No B C S ++ + - - - Body perception ••• • •• •• Clinking of a coffee cup • ••• •• •• Leaf movements of a pot plant •• •• Water rippling in a glass bowl ••• • Chinking of a glass pane ••• • • •••

House 2, point 1 Intensity Acceptability

No B C S ++ + - - - Body perception ••• • •• Clinking of a coffee cup •• • ••• Leaf movements of a pot plant Water rippling in a glass bowl Chinking of a glass pane ••• • ••

5.5 Vibration models in comparison with experimental results (Task 2.4)

Comparison with FE model The floor construction described in chapter 5.2 and Figure 5.1 has been the object for two FE models, with different approach, and since it also has undergone extensive laboratory tests, chapter 5.3 it is well suited as representative when it comes to validation of the models and comparison with measured data.

The dynamic results obtained from laboratory measurements are given in Table 5.14, compared with the results from the FE models. It can be seen that for FE model A, the first mode is predicted accurately for setup No 1-4 (plaster board) while for setup No 5-8 (concrete) the mode is predicted with lower frequencies (1.3-1.9 Hz) compared with measurements. The second mode is underestimated in frequency (0.8-3.0 Hz) which indicates that the model for some reason proves to have a lack of stiffness in the transverse direction. Overall, the frequency shift from one setup to another is better predicted than the frequencies as absolute values. The FE model B does in general deviate more from the measured data. It seems that the have taken into account too much stiffness since most natural frequencies are overestimated.

Table 5.14 Comparison of FEM results vs. measured data in laboratory tests.

FE model A FE model B Measured Setup No f1 (Hz) f2 (Hz) f1 (Hz) f2 (Hz) f1 (Hz) f2 (Hz)

1 10.3 12.8 - - 10.4 15.2 2 10.2 10.6 - - 10.3 13.5 3 10.0 10.6 13.9 21.1 10.1 13.6 4 10.1 12.7 17.0 30.2 10.3 15.3 5 8.1 9.5 8.5 14.2 10.0 12.0 6 10.1 18.7 12.9 25.8 11.4 16.4 7 8.5 9.6 - - 10.3 10.8 8 10.3 19.8 - - 11.7 16.4

85

In addition, finite element models were created for two of the buildings described in Section 5.3: Brixton and Daventry. The models were created using isotropic shell elements to represent the caberboard or chipboard, and beam elements to represent the supporting joists. The ceilings, services, acoustic flooring and occupation were represented by an additional imposed load, and the analysis was conducted using design values and layouts for all inputs except the damping, for which the damping value from the test results for the first mode was used.

The fundamental frequencies from these FE models and the corresponding response factors are given in Table 5.15. As the results show, the FE models suggest similar frequencies but much larger response factors than the tests. There are several possibilities for these differences (e.g. assumptions about the fixity of the floor material to the steel joists or assumptions about the mass), but a further study using the finite element software shows that the major effect on both frequencies and response factors is the provision of partitions, which were not included in the original model.

Table 5.15 Comparison of FEM results vs. measured data for field tests.

FE model Measured Field Test f1 (Hz)* R f1 (Hz)* R Brixton 13.9 66.9 14.4 35.5

Daventry 16.3 53.9 14.0 16.5 * Only local modes of vibration are considered, as these have the lowest modal masses and hence higher response factors By including a thin steel plate along the lines of the partitions, and gradually increasing the depth of the plate, the natural frequencies of the floor can be matched in the FE model. At a depth of 50mm and width of 1.5mm, the frequencies correlate, which suggests that the partitions are not fully effective, but are significant. Performing a response analysis on the output from the Daventry tests with the partitions as described gives a response factor, R = 16.2, which is approximately equal to that obtained in the test.

Comparison with design criteria

Deflection based method Subjective rating of intensity and acceptance were made using the procedure described in Chapter 3 for all laboratory test floors tested in LTU and Kauklahti field test floors tested in Finland. For the British field test cases (Brixton, Derby and Daventry) subjective ratings were not carried out. For that reason those results were not included in the comparison. Only general observations by the personnel that carried out the measurements indicate that there should be no vibration problems in these buildings. Some results from laboratory tests carried out prior to the Acousvibra project are included in the comparison. These tests were performed by VTT [44] and are marked as VP1…VP3 in Table 5.16. Floor properties are given in Annex A.

Summary of the comparison is given in Table 5.16. Calculated and measured fundamental frequency static deflection due to 1kN load and vibration class is shown. Last columns of the table gives results from subjective rating. The floor was judged as acceptable if at least 50% of the observers evaluated the floor as “acceptable” for new built residential (Class C minimum).

86

Table 5.16 Comparison of calculated and measured floor performance Calculated values Measured values Acceptance based

on sense perception

Floor f0 [Hz]

δ 1 kN [mm]

Vibration class

f0 [Hz]

δ 1 kN [mm]

ζ [%]

Vibration class

Body perception acceptance

”Observation of articles” acceptance

Kauklahti 1 17,5 0,11 A 18,5 0,03 4-5 A YES YES Kauklahti 2 7,5 0,39 C3 11,1 0,03 5-7 A YES YES Lab 1 14,1 0,41 C 10,4 1,9 NO2 NO2

Lab 2 12,9 0,41 C 10,3 1,3 NO2 NO2 Lab 3 11,4 0,46 C 10,1 1,3 NO2 NO2 Lab 4 12,3 0,46 C 10,3 1,5 NO2 NO2 Lab 5 11,6 0,17 B 10,0 1,2 NO2 NO2 Lab 6 12,5 0,17 B 11,4 1,5 NO2 YES2 Lab 7 14,2 0,14 B 10,3 0,11 1,4 A NO2 NO2 Lab 8 15,7 0,14 B 11,7 0,12 2,0 A NO2 YES2 Lab 9 14,2 0,14 B 0,11 1,4 A YES VP1 -a 10,9 0,98 D 11,8 3,01 3,3 E NO NO -b 8,8 1,22 E 10,2 2,91 4,0 E NO NO -c 6,9 1,55 E 8,1 3,31 3,3 E NO NO VP2 -a 9,5 0,55 D3 12,1 0,6 1,7 D YES YES -b 7,7 0,67 D3 10,7 0,7 2,1 D YES YES -c 6,0 0,88 D3 8,8 0,7 1,3 D3 YES NO VP3 -a 10,8 0,18 B3 12,0 0,15 7,7 B YES YES -b 8,7 0,22 B3 10,3 0,20 9,9 B YES YES -c 6,9 0,28 C3 8,1 0,21 11,9 B3 YES YES 1Measured value included local deflection of the floating floor 2Horizontal vibration effects on the results 3The criteria is based on deflection criteria, although the fundamental frequency < 10Hz.

In some cases, the calculated fundamental frequency was below 10Hz which is the limit for high-frequency floors. Anyhow, the same criterion was used for all cases, because the measured values in these cases were usually higher fulfilling the frequency limit. Both field test floors (Kauklahti 1 and 2) were acceptable based on calculations, measurements and subjective tests. In Kauklahti 2 the calculated fundamental frequency was lower and calculated deflection higher than the measured values. In this case, the floor was supported by beams and the fundamental frequency and deflection of the beam was also considered. But in the actual building, it seems that the influence of the supporting beam is not so high.

Unfortunately all laboratory test series are not relevant for comparison due to horizontal vibration of the test floors which dramatically effects the sense perception as described earlier. For test floor 9, the horizontal vibration was eliminated and then the calculated values agree very well with measured values and with subjective tests.

For test floors VP1…VP3, the measured values are also well in accordance with predictions. In the second group of floors (VP2) the performance was subjectively judged to be somewhat better than predicted.

Acceleration based method For the acceleration based method, the weighted acceleration aw.rms has been determined by calculation and by experiments. Then response factor and total time of acceptable intermittent were determined from aw.rms (See equation in Chapter 3.5.2). Adjusted value of DVD = 1,6 has been used in the calculation. This value should correspond to a “low probability of adverse comment” in day time and is based on a response factor of 16. Calculated and measured values are shown in Table 5.17. From laboratory measurements only setups 1 and 9 were selected for comparison representing two different floor types.

87

Table 5.17 Comparison of calculated and measured values for test floors Calculated values Measured values Floor f0

[Hz]

aw.rms [m/s2]

Response factor R

Acceptable time t [h]

f0 [Hz]

aw.rms [m/s2]

Response factor R

Acceptable time t [h]

Kauklahti 1 18,3 0,022 4,4 cont. 18,5 0,067 13,4 cont. Kauklahti 2 31,8 0,011 2,2 cont. 11,1 0,051 10,2 cont. Brixton 8,4 0,179 34,2 9,9 14,4 0,178 35,5 8,5 Derby 14,1 0,296 59,2 1,1 14,3 0,147 29,4 cont. Daventry 13,7 0,140 28,1 cont. 14,0 0,083 16,5 cont. Lab 1 12,9 0,144 28,8 cont. 10,4 0,139 27,8 cont. Lab 9 14,2 0,045 9,0 cont. 10,3 0,035 7,0 cont. VP1 -a 10,9 0,048 9,6 cont. 11,8 0,256 51,2 2,0 -b 8,8 0,057 11,4 cont. 10,2 0,186 37,2 7,1 -c 6,9 0,060 11,9 cont. 8,1 0,193 38,6 6,1 VP2 -a 9,5 0,044 8,7 cont. 12,1 0,216 43,2 3,9 -b 7,6 0,049 9,9 cont. 10,7 0,134 26,8 cont. -c 6,0 0,047 9,4 cont. 8,8 0,121 24,2 cont. VP3 -a 10,6 0,024 4,8 cont. 12,0 0,093 18,6 cont. -b 8,6 0,029 5,8 cont. 10,3 0,070 14,0 cont. -c 6,7 0,029 5,9 cont. 8,1 0,109 21,8 cont.

The Table 5.17 shows that most of the floors are below the response factor limit R = 16 for continuous vibration when aw.rms value is determined theoretically. When measured aw.rms is used the R value is in most cases above 16, but still the total acceptable time is so high that practically it leads to acceptance of continuous vibration (t > 16 h). The table also shows that the floors that were subjectively judged as “unacceptable”, such as VP-1 series, are acceptable based on this method. VP-1 series floors were floating floors (see details from Annex A). It seems that the prediction of aw.rms is quite challenging because the predictions seems to be concidrably lower than the measured values thus leading the results to the “unsafe” side. In Lab 1 and Lab 2 floors the predicted aw.rms are in congruent with measured values.

5.6 Role of connections and boundaries of a lightweight floor (Task 5.2)

A series of dynamic measurements was performed as a step by step procedure gives an insight into how different changes of the surroundings, corresponding to different installation steps of the floor into a real building, affect the dynamic properties. The experimental setup and measurement procedure was the same as described in chapter 5.1. Setup No 3 – plasterboards, corners, free long sides - was used for all tests described here. The experiment was performed according to the following three approaches; 1) Line loads in two different directions in combination with different type of mountings, 2) Mass loaded supports and 3) Reinforced supports.

5.6.1 Line loads with different mountings The use of line loads is a way to represent lightweight partition walls in a real building, although the representation is a substantial simplification of a true partition. The line load consists of two glulam beams joined together to form one complete timber beam of length 3.76m and with an approximate weight of 215 kg. One or two complete beams were used at the same time. The beams were put to the floor along the short sides (i.e. at the support), in the centre in both main – and transverse direction, in ¼-point in transverse direction and a combination with beams in the centre and ¼-point (both transverse) were used simultaneously, Figure 5.19.

88

Figure 5.19 Placement of the line loads in terms of timber beams. 1a) and 1b) over the supports, 2) centre transverse 3) ¼-point transverse and 4) centre main.

Also, different types of mountings of the beams to the floor were tested. The simplest was with the beams lying freely, with no mounting to the floor. In the next step angle bars were attached to the glulam beam in seven positions, directly over the load bearing C-beams, each bar on successive opposite side of the timber beam in transverse direction. The angle bars were connected to the floor by screws of different length. A short screw fixed to the sheet metal while a longer screw was used for connection to the C-beams. See Figure 5.20 for details.

Figure 5.20 Mounting of the complete timber beam to the floor.

A comparison between the different setups of transverse line loads connected to the sheet metal is shown in Figure 5.21. The fundamental frequency varied a few tenths of a hertz indicating that the effect of increasing stiffness and mass cancel. The first torsion mode was more affected. It dropped from originally 13.6 Hz to 13.0 Hz with the line load at ¼-point, to 12.7 Hz at ½-point and to 12.1 Hz when the two line loads were combined. It is clear that the added mass has a significant effect for this mode.

Figure 5.22 shows the corresponding case where the line load is applied to the centre line in the main direction. Here there is a stiffness gain for the fundamental mode and the frequency increases by 0.3 Hz to 10.4 Hz. The torsion mode is almost unaffected (+0.1 Hz).

1a 2 3 1b 4

Screw to sheet metal

Screw to C-beam Timber beam

89

-70.0

-30.0

-60.0

-50.0

-40.0

8.000 20.00 12.00 16.00Frequency (Hz)

Figure 5.21 FRFs that show the effect of adding line loads to the floor in transverse direction.

-70.0

-30.0

-60.0

-50.0

-40.0

8.000 20.00 12.00 16.00Frequency (Hz)

Figure 5.22 FRFs showing the effect of adding a line load to the centre of floor in main direction.

The importance of mounting of the line loads is presented in Figure 5.23 for the transverse ½-point case. As expected, the stiffness increases as the degree of mechanical connection of the timber beam to the floor construction increases. This is clearly seen for the second natural frequency, the torsion mode. Originally the frequency was 13.6 Hz and as the timber beam is placed freely on the floor it dropped to 12.4 Hz. When the beams were connected to the structure, first by screws to the sheet metal and then by screws to the C-beams, the natural frequency increased to 12.7 Hz and 12.8 Hz respectively.

Figure 5.24 shows the situation when the line load was placed in the main direction of the floor and attached by screws either to the sheet metal or to the C-beams. The fundamental frequency increases from 10.1 Hz in the reference case to 10.4 Hz and 10.5 Hz when screws are mounted to the sheet metal and to the C-beams, respectively. The stiffness gain, due to attachment to the C-beams instead of the sheet metal, is minor in both of the two shown cases. It can be explained by the fact that, although its inherent stiffness is significantly lower, compared to the C-beams, the combined package of sheet metal and plasterboard is closely mounted to C-beams by screw connections. Then, composite action occur, and the C-beams’ stiffness are utilised even though the glulam beam is not mounted directly to the C-beams.

Gypsum, Corners, Free edges Line load ½, across, mounted to sheet metal Line load ¼, across, mounted to sheet metal Line loads ½ and ¼, across, mounted to sheet metal

Gypsum, Corners, Free edges Line load ½, across, mounted to sheet metal Line load ½ along, mounted to sheet metal

90

-70.0

-30.0

-60.0

-50.0

-40.0

8.000 20.00 12.00 16.00Frequency (Hz)

Figure 5.23 FRFs that show the effect of different degree of mechanical connection of the timber beam to the floor structure in the case of a transverse centre line load.

-70.0

-30.0

-60.0

-50.0

-40.0

8.000 20.00 12.00 16.00Frequency (Hz)

Figure 5.24 FRFs that show the effect of different degree of mechanical connection of the timber beam to the floor structure in the case of a centre line load along the floor’s main direction.

5.6.2 Mass loaded support The same timber beams were now placed with no mechanical attachment on the short sides of the floor, above the supports, one beam to each support. The fundamental frequency increased by 0.6 Hz to 10.7 Hz and the torsion mode by 0.2 Hz to 13.8 Hz, see Figure 5.25. The added timber acts like a motion resistor and the support takes one step further on the scale from a pinned towards a clamped condition.

Gypsum, Corners, Free edges Line load ½, across, no mechanical connection Line load ½, across, mounted to sheet metal Line load ½, across, mounted to C-beams

Gypsum, Corners, Free edges Line load ½, along, mounted to sheet metal Line load ½, along, mounted to C-beams

91

-70.0

-30.0

-60.0

-50.0

-40.0

8.000 20.00 12.00 16.00Frequency (Hz)

Figure 5.25 FRFs showing the effect of adding mass on the floor above the supports and the effect of stiffening the supports by U-profiled beams connected to each C-beam.

5.6.3 Reinforced support In an attempt to further increase the stiffness of the system the supports were modified by in total 14 U-profiled beams of 600 mm length, each connected to the tubular beam and to the rim beam with proper distances in between. The U-beams were mounted with a centre distance of 600mm, i.e. the same as the C-beams. The details are shown in Figure 5.26 and the resulting FRF in Figure 5.25. The FRF is almost the same as for the reference case except for the fundamental frequency. The original frequency peak has been split into two – one of higher and one of lower frequency than original. The two mode shapes are similar and an explanation is that the stiffness, for some practical mounting reason, did not become identical for the two supports after the reinforcement.

Figure 5.26 U-profiled beams (cross section to the right) was added to the support (left) Five screws positioned in the rib centre together with spacers where appropriate were used for mounting.

5.6.4 The connections’ relation to a real building To fully understand and predict the response from a floor in a real building is difficult. Parameters, like the exact boundary condition, the exact dynamic properties of the surroundings, the effect of partitions etc. must be known. A few of these parameters were treated in the foregoing experiments where among

Gypsum, Corners, Free edges Line load to the supports Reinforced supports

U-beam Rim beam

C- beam

92

other things, gluelam beams were used to simulate the effect of partitions as well as walls standing on the support. The additional weight and stiffness shift the natural frequencies of the floor, typical by 1.5-2.0 Hz, and the damping should increase due to friction between floor and partition and inside the partition. The influence greatly depends upon the partition’s construction and on the direction of the partition. The various techniques of mechanical connection tested were not found to be of any major importance. The experiment with the mass loaded support shows a relatively large increment in fundamental frequency. Even though the obtained difference was only 0.6 Hz, it must be noted that a corresponding wall in a real building could, due to expected higher stiffness and mass, affect the result even more. The stiffer and heavier a wall, the more clamp-like the support obtained.

5.7 Subjective assessment of floor vibrations with a motion simulator (Task 5.3)

A major concern for the serviceability of lightweight floors is the low frequency vibrations induced by normal human activities, primary walking. Walking possesses a pace frequency of about 1.6 – 2.4 Hz, and its harmonics might well excite a floor at its fundamental frequency leading to severe response amplification. From a structural dynamic point of view, the human-floor system forms a dynamic system that can be modelled as:

VIBRATION RESPONSE = DYNAMIC PROPERTIES × INPUT FORCE.

Humans play double roles in this system; both as the source and as the sensor. The activities exert forces on the floor and at the same time, occupants receive floor vibration through their body, by visual impression and/or by sound. The dynamic properties of the floor system can be calculated theoretically or measured experimentally by modal testing. In most cases, the vibration in the vertical direction is, naturally, dominant. Even though floor constructions exist in which also horizontal vibrations play a crucial role they are omitted here.

A number of researchers have come up with different suggestions for floor design criteria over the years. A common method is to use a deflection criterion for high-frequency floors and an acceleration limit for low-frequency floors. The definition of a high/low frequency floor depends on the fundamental frequency, where suggestions of 7, 8 or 10 Hz are commonly found for the frequency of transition.The argument for this choice of fundamental frequency is that no higher harmonics than the fourth should be considered when it comes to walking-induced vibration. Then, making sure that the fundamental frequency of the floor is above this limit, typically 8-10 Hz, no annoying vibrations are assumed to occur. The statement that the fourth harmonic is the limit for consideration could be questioned and Ellis [39] proclaims that harmonics up to the eighth order should be taken into account. Light weight floors typically fall into the category high frequency, whereas heavy floors normally relate to low frequency.

Floor serviceability is in general indeed a challenging task; the nature of floor vibration is a multimodal topic of multiple frequencies, and the human perception is difficult to predict even though the exact vibration is known. Having a clear understanding of the human response to this kind of well defined and well controlled vibrations, should result in the development of proper design criteria. Within the Acousvibra project, four experimental studies of human perception of vibration were conducted; I) Threshold values from single frequency vibrations, II) Threshold values for a second frequency component, III) Annoyance of dual sinusoidal vibrations and IV) Annoyance and acceptance of multi-frequency vibrations. Study I will, providing the results agree with those from other similar studies, serve as a kind of assurance that the experimental set-up and methods used are reliable. Study II can be seen as an indication of the importance in perception of dual sinusoidal vs. single. Study I and II use a reduced number of test persons while Study III - with focus on the annoyance of dual sinusoidal vibration - uses more test persons for better consistency in the results. Study IV is extended from Study III in the way that it includes acceptance, i.e. whether or not the vibration should have been accepted if it occurred in a real floor. Study IV considers signals of up to five discrete frequency components.

93

5.7.1 Experimental set-up A new motion simulator, designed to simulate floor vibration, has been developed within the project. It consists of a forced air cooled electromagnetic shaker (Derritron VP30) installed in a specially made steel frame. A light, but proportionately stiff wooden plate (in order to avoid resonance frequencies within the typical range of floor vibration), rests on the foundation’s four corners, with cellular polyetherurethane dampers (Sylomer® R25) in between. The shaker is connected to the centre of the wooden plate by a steel rod. A wooden chair for the test person to sit on completes the arrangement which can be seen in Figure 5.27. The chair’s position is controlled before every experiment, to assure an exact position, but has no fixed mechanical connection to the plate. The test subjects were always asked to sit on the chair in a comfortable upright position. The back was then in natural contact with the backrest of the chair, even though it was not given as a specific instruction. The subjects wore ear cups which were used in order not to be influenced by ambient noise. Normal indoor shoes and clothes were worn.

Figure 5.27 The motion simulator used in the experiments.

5.7.2 Study I: Threshold values from single frequency vibrations Objective The objective was to determine the human threshold values when exposed to vertical vibrations of single frequencies. The threshold value is here defined as the lowest possible vibration magnitude that can be detected from a sinusoidal signal with constant amplitude.

Method Seven subjects took part in the study and nine single sinusoidal signals were tested: 5, 6.3, 8, 10, 12.5, 16, 20, 25 and 31.5 Hz. The signals correspond to the centre frequencies of one third octave band. A low weight control amplifier put in the test person’s lap, was used for amplitude adjustment by a discrete turn knob without cue marks. For each test signal, the amplitude was initially set to zero after which the test person slowly increased the level until he just could feel the vibration. The threshold was measured once for each signal and subject.

Results The results clearly show that the sensitivity is higher, i.e. the threshold value is lower, for lower frequencies compared to higher. At 5 Hz, the mean threshold was about 8mm/s2 while it was 23mm/s2 at 31.5 Hz. The obtained results are also more consistent for the lower frequencies. This is shown by the confidence intervals (95%) of the threshold values’ mean, Figure 5.28a. In Figure 5.28b the mean threshold values are compared with the ISO base curve [40]. Values from the present study are higher for each of the tested frequencies, approximately a factor 1.6 higher. Some other similar investigations [41] and [42] have also obtained greater threshold value than the ISO base curve. Since the results are in agreement with other studies, it is assumed that the experimental equipment and testing procedure are reliable and that the results from the following studies II and III can be treated with confidence.

94

5 6.3 8 10 12.5 16 20 25 31.50

10

20

30

Frequency (Hz)

Am

plitu

de (m

m/s2 )

a)

5 6 7 8 9 10 20 30

5

6789

10

20

Frequency (Hz)

RM

S a

ccel

erat

ion

(mm

/s2 )

ISO base curvePresent study

b)

Figure 5.28 a)Means with confidence intervals of the threshold values from sinusoidal vibration. b)Means of threshold values from the present study ——— compared with ISO base curve ----.

5.7.3 Study II: Threshold values for a second frequency component Objective The objective was now to find the perception threshold of a sinusoidal vertical vibration signal in the presence of another sinusoidal signal having fixed amplitude.

Method Six persons from Study I participated. A base frequency of 8 Hz with fixed amplitude of 35, 50 or 70 mm/s2 rms was used together with a second signal, the test frequency of 10, 12.5, 16, 20 or 25 Hz. In total, 15 combined signals, each built up of two frequency components, were tested (three base amplitudes times five test frequencies). During the experiment, the base frequency was first set to one of the fixed amplitude levels. Then, the test subject controlled the amplitude of one test frequency at the time by using the same control amplifier as in Study I. Starting from zero level, the amplitude of the test frequency was slowly increased until it just was felt that the signal was changed, i.e. the presence of the second frequency component could be noticed, and finally, the response level was measured. The test was performed in three steps related to the three amplitudes of the base frequency.

Results Figure 5.29a shows the average threshold value and corresponding confidence interval (95%) as a function of test frequency at the three different amplitudes of the base frequency. For comparison, the threshold values for single sinusoidal vibrations from Study I are incorporated in Figure 5.29b.

10 12.5 16 20 250

10

20

30

40

50

60

70

Frequency (Hz)

Acc

eler

atio

n (m

m/s2 R

MS

)

8 Hz: 35mm/s2

8 Hz: 50mm/s2

8 Hz: 70mm/s2

a)

8 9 10 20 30

5

6

789

10

20

30

40

Frequency (Hz)

Acc

eler

atio

n (m

m/s2 R

MS

)

8 Hz: 35 mm/s2

8 Hz: 50 mm/s2

8 Hz: 70 mm/s2

Threshold from study I

b)

Figure 5.29 Mean threshold values in the presence of a base frequency of 8 Hz at three different amplitudes, 35mm/s2———, 50mm/s2---, and 70mm/s2 ···; a) together with confidence intervals and b) together with threshold values ·-·- from single frequencies as comparison.

95

Subjects are more sensitive, i.e. lower threshold level, as the test frequency gets closer to the base frequency. The result at 10 Hz is significantly different from the other test frequencies. For example, when the base frequency level is set to 35 mm/s2 the threshold value is 6 mm/s2 at 10 Hz while it is 22 mm/s2 at 25 Hz. The 10 Hz test frequency was actually even more easy to detect in the presence of the base signal compared to the case when it was played alone (Study I).Threshold values at all test frequencies tended to be lower when the base frequency amplitude was set at 35 mm/s2 as compared with the higher settings of 50 and 70 mm/s2 though this difference was not statistically significant.

5.7.4 Study III: Annoyance of dual sinusoidal vibrations Objective The objective was to subjectively evaluate the annoyance of vibration signals consisting of dual frequencies. The signals were composed to be representative of those that might occur in a floor structure.

Method Fifteen subjects took part in this study, 13 male and two female. A base component, with constant frequency and amplitude throughout the experiment, was set to 8 Hz, 35 mm/s2 rms. In addition, a second frequency component was added to the base one. The added component consisted of one of the five frequencies: 10, 12.5, 17, 20 or 25 Hz. The reason to choose 17 Hz instead of 16 Hz as in previous experiments, is that severe phase dependence would occur using a 16 Hz component since it is a harmonic of 8 Hz. By using 17 Hz, the phase relation to the base component is by far less sensitive. Five different amplitudes were used for the added component: 7,14,21,28 and 35 mm/s2 rms that correspond to 20,40,60,80 and 100% respectively of the base component’s amplitude. Thus the test subjects were in total exposed to 26 signal combinations 1 signal comprised one frequency component only, 8 Hz at 35 mm/s2 for reference, while the remaining 25 signals comprised two components.

The choice of the base component is motivated by the fact that 8 Hz is a typical fundamental floor frequency. It also constitutes the borderline between low- and high frequency floors and should consequently be the most annoying frequency that is allowed to occur in a high frequency floor (typical lightweight constructions). The amplitude of 35 mm/s2 rms is the upper acceptance limit in home and office environments according to AISC [43]. The subject was exposed to the 26 test signals in a random order. After a short settle time when starting up each signal, typically a few seconds, the subject was informed that the correct amplitude was achieved and the signal then lasted for 10 seconds. Then the test person was asked to rate the experienced annoyance due to the vibration on a 11-point (0-10) numeric scale where “0” is defined as “not at all annoying” and “10” as “extremely annoying”. When judging, they were instructed to think of them selves as being at home or in office environment. The whole test lasted about 20-25 minutes for each subject.

Results Figure 5.30 shows the mean annoyance of the 26 test signals in which it can be seen that the perceived annoyance is frequency and magnitude dependent. The general tendency is that the annoyance increases as the amplitude of the second frequency component increases while the annoyance decreases as the frequency of the second component increases. As an example of the former; when the second frequency component is 10 Hz, the average annoyance is 5.2 at 7 mm/s2 but 6.8 at 35 mm/s2 and as an example of the latter; when the amplitude of second frequency component is 21 mm/s2 the average annoyance is 6.8 at 10 Hz but 5.4 at 25 Hz. It was also found that the lowest annoyance, 4.4 in average, occurred in the case where the second frequency component was omitted.

96

8 +10 +12.5 +17 +20 +254

4.5

5

5.5

6

6.5

7A

nnoy

ance

Frequency (Hz)

Figure 5.30 Averaged annoyance rating. The single point to the left is the base frequency only, 8Hz (signal No. 1). The next group consists of the signals 8 + 10 Hz (signals No. 2-6) and after that follows 8 + 12.5 Hz (signals No. 7-11), 8 + 17 Hz etc. The amplitudes of the second frequency component are 0 (*), 7(x), 14(o), 21(◊), 28(□) and 35(∆) mm/s2.

5.7.5 Study IV: Annoyance and acceptance of multi-frequency vibrations Objective The objective was to subjectively evaluate the annoyance and acceptance of vibration signals comprising one to five frequencies. The signals were composed to be representative of those that might occur in a floor structure.

Method Fifteen subjects took part in the study, twelve male and three female. The signals contain one or several of the frequency components 8, 10, 12, 14 and 16 Hz. Assuming a walking pace of two Hz, the chosen frequencies represents its harmonics. Frequencies below 8 Hz have been omitted since no such low natural frequencies are allowed to occur according to design criteria. The frequency components’ amplitude was always set to one out of two possible fixed levels; High level – 28 mm/s2 rms and Low level – 24 mm/s2 rms. In total, 44 signals were used; 10 contained one frequency component only, 14 contained two components, 18 contained three components and 2 contained five frequency components.

The experiment was performed in the following way; After a short settle time when starting up each signal, typically a few seconds, the subject was informed that the correct amplitude was achieved. Once at correct level, the signal lasted for 10 seconds and then stopped. Then the test person was asked to rate the vibration perception in two ways, independently of each other; a) The annoyance was rated on a 11-point (0-10) numeric scale where “0” is defined as “not at all annoying” and “10” as “extremely annoying” and b) whether or not they should accept that actual floor vibration. When judging, they were instructed to think of them selves as being at home or in office environment. The whole test lasted about 30-35 minutes for each subject.

Results The results are generally presented in two ways. Annoyance refers to the 11-point annoyance scale and Acceptance says to what degree a floor vibration is acceptable. The acceptance ranges from 0-1 where 1 means that 100% of the test subjects accepted the vibration and 0 means that none found the floor vibration acceptable. High and Low level are related to vibration amplitudes of 28 and 14 mm/s2 rms respectively. The presented parameter is the averaged value from the 15 test subjects.

Amplitudes of the second frequency component:

* 0 mm/s2 x 7 mm/s2 o 14 mm/s2 ◊ 21 mm/s2 □ 28 mm/s2 ∆ 35 mm/s2

97

One frequency, high vs. low level In Figure 5.31 the results are shown from vibration of single frequencies at high and low level, signals 1-10. It is clearly seen that the annoyance is higher at high level compared to low and that the acceptance is lower at high level compared to low. It is also noticed that the annoyance/acceptance difference between high and low level decreases as the frequency increase. An explanation could be that at low level, the vibration is so small that it barely can be felt regardless of the frequency (recall the absolute threshold value of about 10 mm/s2) while at high level, the reduced sensitivity for higher frequencies is seen.

a) b)

Figure 5.31 a) annoyance and b) acceptance from single frequencies at two different amplitudes. Signals 1-10.

One vs. two frequency components Figure 5.32 and Figure 5.33 show how the test subjects react when exposed to dual frequency components.

a) b)

Figure 5.32 a) annoyance and b) acceptance from a pure 8 Hz signal of either low or high level vs. a combined signal of 8 Hz + a second low level frequency component of 10-17 Hz. Signals 1-2 and 11-18.

Low 14 mm/s2

High 28 mm/s2

8 Hz Low 14 mm/s2

8 Hz High 28 mm/s2

98

a) b)

Figure 5.33 a) annoyance and b) acceptance from a pure 10, 12 or 14 Hz signal at high level vs. a combined signal where a second frequency component of 12-16 Hz at low level has been added. Signals 3, 5, 7 and 19-24.

In Figure 5.32, the response from the pure 8 Hz signals is compared to the case where a second component of 10-17 Hz is added, signals 1-2 and 11-18. With the 8 Hz at low level there is an immediate difference in annoyance/acceptance for 8 Hz vs. 8 + 10 Hz. The annoyance increased from 1.5 to 3.7 and the acceptance dropped from 87% to 53%. When the second component’s frequency increases, test subjects are gradually less affected by the added component. With the 8 Hz component at high level there are only small differences in terms of annoyance while the acceptance is lower when the second component is present. As an example, the acceptance dropped from 40% to 13% for 8 Hz vs. 8 + 10 Hz.

In Figure 5.33, pure 10, 12 and 14 Hz signals at high level are compared to signals where a second frequency component of 12-16 Hz at low level is added. Looking at the 10 Hz case, the change in annoyance is small but for the 12 and 14 Hz case the change is somewhat higher. The annoyance of 14 Hz is 3.3 while it is 5.4 for 14 + 16 Hz. The same tendency appears clearly in the acceptance. The acceptance for 10 Hz and 10 + 12 Hz is 53% and 33% respectively while it for 14 Hz and 14 + 16 Hz is 73% and 7% respectively.

One and two vs. three frequency components Figure 5.34-Figure 5.35 show how the test subjects react when exposed to signals containing up to three (up to five in one case) frequency components. In Figure 5.34 a pure 8 Hz signal at high level and an 8 + 10 Hz signal at high+low level are compared to signals where a third frequency component of 12-17 Hz at low or high level is added. Going from one to two components yield an annoyance increment of about one unit. Going from two to three components yield an annoyance increment of approximately half a unit if the third component is at low level and about one unit if it is at high level, this regardless of the frequency 12, 14 or 17 Hz. In terms of acceptance, the pure 8 Hz signal was accepted by 40% of the subjects while the rest of the signal combinations showed very low acceptance, 13% or below.

Single frequency

Dual frequencies

99

a) b)

Figure 5.34 a) annoyance and b) acceptance from a pure 8 Hz at high level, an 8+10Hz signal at high+low level and combined signals where a third frequency component of 12-17 Hz of low or high level is added. Signals 1, 11 and 25-30.

a) b)

Figure 5.35 a) annoyance and b) acceptance from a pure 8 Hz at low level, an 8+10Hz signal at low+low level and combined signals where a third frequency component of 12-17 Hz of low or high level is added. Signals 2, 15 and 31-36.

5.7.6 Further analysis and prediction model Some patterns and conclusions can immediately be drawn from the results presented in the previous section. Floor vibration acceptance is primarily a matter of vibration amplitude and frequency characteristics according to the following three implications:

1) As the vibration amplitude increases the annoyance increases and the acceptance decreases. 2) In general, when the vibration frequencies increase but the amplitude remains constant, the annoyance decreases and the acceptance increases. 3) The perception depends on the vibration’s composition in terms of number of frequency components and their frequency separation.

Beating effect The second of the implications mentioned above enacts that the acceptance increases as the frequency increases. But this is only in general. For several cases reported here it is not valid. A signal that comprises two frequencies, say 14 and 16 Hz, should then, if it is assumed that the frequency components have equal amplitude, be judged with lower annoyance than a similar signal that comprises the frequencies 8 and 10 Hz. In Figure 5.36, selected signals from Figure 5.32 and Figure 5.33 are extracted for comparison. Each of them consists of two components with two Hz separation where the

8 Hz (high) / 8+10 Hz (high+low)

8+10+12/14/17 Hz (high+low+low)

8 Hz (low) / 8+10 Hz (low+low)

8+10+12/14/17 Hz (low+low+low)

100

amplitude of the lower frequency (high level) is twice the amplitude of the higher one (low level). It can be seen that they do not agree with implication 2, i.e. the annoyance will decrease as the frequency increases but instead the annoyance remains at about the same level independently of the frequency. The main reason why implication 2 is not valid in this case is probably the beating. Due to the frequency separation of 2 Hz there is in the summed signal a strong contribution of this 2 Hz beating component for all the actual signals and the perception of this beating component appears to dominate the perception from the original frequencies.

8+10 10+12 12+14 14+160

1

2

3

4

5

6

7

8

9

10

Ann

oyan

ce

Frequency (Hz)

Figure 5.36 Annoyance from signals built up of two frequencies (high+low level), all with 2 Hz separation.

Prediction model In order to find out to what degree the total energy (un-weighted total amplitude) and the frequency weighted acceleration (weighted total amplitude) correlates with the perceived averaged annoyance, a series of multiple linear regression analyses were performed. For each model, the statistical significance (95% confidence level) of the following potential parameters was tested: total amplitude (weighted and un-weighted), fundamental frequency (the lowest of the participating discrete frequencies, only in the case of two or more frequency components) and frequency separation (only in the case of two frequency components). In Table 5.18-Table 5.19, grouped with respect of the number of frequencies, the best fitted models in terms of R-squared adjusted and mean absolute error is shown.

Table 5.18 Regression models in the form of: Annoyance = a + b · Amplitude + c · Fundamental freq. + d · Frequency separation.

Coefficients No. freq.

comp. a)

Constant b)

Amplitude c)

Fund freq. d)

Freq. sep.

Mean absolute

error

R2 adj. (%)

1 -1.68 0.19 - - 0.43 80.7 2 -1.63 0.27 0.15 0 0.51 81.1

3 or more 0.99 0.15 0 - 0.47 68.0 2 or more -0.72 0.19 0 - 0.58 78.8

Table 5.19 Regression models in the form of: Annoyance = a + b · Weighted amplitude + c · Fundamental freq. + d · Frequency separation.

Coefficients No. freq.

comp. a)

Constant b)

Weighted amplitude

c) Fund freq.

d) Freq. sep.

Mean

absolute error

R2 adj. (%)

1 -1.26 0.39 - - 0.28 94.2 2 -4.21 0.45 0.28 0 0.37 89.0

3 or more -1.14 0.33 0.21 - 0.31 82.8 2 or more -3.17 0.43 0.24 - 0.48 84.7

101

As expected, the regression analysis gave better mathematical models in the cases where weighted amplitude was used in stead of un-weighted. The degree of explanation is higher and the mean absolute error is less. The frequency separation was not found to be statistically significant while the fundamental frequency indeed was. If the fundamental frequency was omitted from Table 5.19, i.e. the annoyance is a function of the weighted amplitude only, the R-squared adjusted is 74.4 and 75.7 % for the cases of 2 and 3 or more frequency components respectively. So, when the frequency weighted total amplitude is used as the only parameter to describe the annoyance, the R2-adjusted is lower for multi frequency cases compared to sinusoidal. This fact leads to an important indication, namely that ISO’s frequency weighting Wm works well as long as a single sinusoidal is concerned but is less accurate in terms of a limited set of discrete frequency components. It is also shown in Table 5.19 that a joint model for all signals containing two or more frequency components only has a minor effect on the degree of explanation. Regression models of higher polynomial order were tested but no evidence for invoking such parameters could be found.

For the prediction model to be extra valuable when it comes to floor design, the averaged annoyance must be related to the averaged acceptance. A linear regression analysis gives the relation according to Eq.3.1 where R-squared adjusted is 94.8% and mean absolute error 0.054 units.

AnnoyanceAcceptance ×−= 16.013.1 (5.9)

Then a delicate issue follows. What level of Acceptance is acceptable? A simple answer could be that a floor is not acceptable unless 100% of the occupants could accept its vibration properties. However, such an attitude is somewhat naive and it is normally accepted that not all can be pleased. An acceptance level of 50% for the actual type of laboratory test might be appropriate [44] meaning that if the parameter Acceptance is 50% or higher, the floor is generally judged as acceptable. The argument for this relatively low level is that during these special tests, the subjects tend to be significantly more critical than they should have been during more realistic circumstances. It follows from Eq. 5.9 that the Annoyance is equal to 4 (3.94 exactly) when the Acceptance is 0.50.

The following prediction models are proposed for calculating the annoyance of floor vibration in home- and office environment on an 11-point scale ranging from zero to ten.

Sinusoidal case: amplitudetotalWeightedAnnoyance 39.026.1 ⋅+−=

Multiple frequency case: frequencylFundamentaamplitudetotalWeightedAnnoyance 24.0 43.017.3 ⋅+⋅+−=

Where amplitude is given in mm/s2 rms and frequency in Hz. Frequency weighting, Wm, according to ISO 2631-2 2003.

Interpretation: If Annoyance ≤ 4, the floor is acceptable If Annoyance > 4, the floor is unacceptable

In Figure 5.37, the predicted annoyance is compared with the observed data. It is seen that the predicted annoyance agrees well to the observed. The annoyance difference is less than ±1.00 units for all of the 44 signals whereas 26 of them lie within ±0.50 units.

102

1-10 11-24 25-440

1

2

3

4

5

6

7

8

9

10

Ann

oyan

ce

Signal No.

ObservedPredicted

Figure 5.37 Observed vs. predicted annoyance.

Application for parameter study The suggested prediction model can also be used in parameter studies. If, for instance, the fundamental frequency of a floor is 8 Hz and its response amplitude is 14 mm/s2, what is then the maximum allowable amplitude of a second natural frequency at 10 Hz if the floor still should be acceptable? Figure 5.38 presents a number of predictions from such questions at issued. For cases when the fundamental frequency is 8 or 10 Hz of varied amplitudes, the maximal amplitude for a second component at different frequencies is predicted. The tendency that successively higher amplitudes are tolerated as the frequency increases stands out.

10 12 14 175

10

15

20

25

30

35

40

Frequency (Hz)

Am

plitu

de (m

m/s2 )

8Hz, 7mm/s2

8 Hz, 14mm/s2

8 Hz, 20 mm/s2

a)

12 14 17

5

10

15

20

25

30

35

40

Frequency (Hz)

Am

plitu

de (m

m/s2 )

10Hz, 7mm/s2

10Hz, 14mm/s2

10Hz, 20mm/s2

b)

Figure 5.38 Predicted maximum amplitudes of a second component for acceptance in the presence of a fundamental frequency of a) 8 Hz and b) 10 Hz.

Example to analysing floor test data Eight setups for a floor test in laboratory was described in Table 5.1 For setup No 5, the frequency spectrum due to walking was found to be dominated by one single frequency, 10.0 Hz, where the corresponding weighted acceleration was 12 mm/s2. The annoyance might then be calculated according to the “sinusoidal case”.

4.31239.026.1 =⋅+−=Annoyance

For setup No 4, the frequency spectrum revealed two dominating frequency components. The frequencies were 8.0 and 10.0 Hz with corresponding weighted acceleration 16 and 6.5 mm/s2 respectively. The annoyance might then be calculated according to the “multiple frequency case”.

103

2.6824.05.51643.017.3 22 =⋅++⋅+−=Annoyance

The interpretation of these results is then that floor setup No 4 is not acceptable (annoyance > 4) while setup No 5 indeed is (annoyance < 4).

5.8 Discussion and conclusions

5.8.1 Horizontal vibrations The results from the laboratory tests in chapter 5.3 showed the importance of considering not only vertical but also horizontal vibration in a floor construction. Although the effect might not be as dramatic as here reported, the horizontal contribution from walking might still significantly affect the perception inside a real building, all dependent on how the floor is installed. Even though the response in the horizontal direction should be low in comparison to the vertical, it should not be neglected since interaction effects, e.g. beating, could occur in conjunction with the vertical response. The perception from multiple frequency vibration was discussed further by Ljunggren et al. [45].

In addition, in the presence of horizontal vibration, although the body perception was unacceptable, the vibration of articles was accepted (for the concrete layer). Thus, the vibrating articles seem not to be affected by the low frequent horizontal vibration and the explanation is probably that the articles do not respond greatly to that limited input frequency range. In order to get the spoon to clink against the coffee cup, or to get the pot plant to swing, another type of force input - more broad-banded or with higher frequencies- is required to get an adverse perception. It also could be the case that vibrating articles in general should be of considerable less concern compared to body perception, although this suggestion is rejected by Talja et al [46]. It was there shown that besides the coffee cup that indeed was judged with higher acceptance, the vibrating articles showed a tendency towards slightly higher acceptance than body perception. However, the discrepancy was not at all on such a level that it could explain the large deviation in the actual test.

5.8.2 Parameters affecting the floor response Several parameters affecting a floor’s dynamic were highlighted in chapter 5.6. Although hardly mentioned, it must be stated that the amount of damping affects the floor performance significantly. The damping ratio can for a laboratory floor be about one percent while a floor inside a building can possess a damping ratio of more than ten percent [47]. The acceptance, especially from impulse like sources, is higher for damped vibrations. A specific problem concerning damping is that there exist no engineering methods to predict the damping within a construction. Instead, data from measurements must be put in to a mathematical model. The data might come from a direct measurement of the actual floor but it can also be approximated with known damping values from similar floor constructions.

Another aspect of floor vibration is that the vibration response depends on the location in the room. A person standing along the main centre line of the floor would not get any sensation at all from the first torsion mode since the mode shape possesses a nodal line with zero response. A person standing in the exact centre point of the floor will perceive a maximum response amplitude at the fundamental frequency. At the centre point along one of the floor’s long sides, the contribution is high from both the fundamental mode and the first torsional mode. A “worst case scenario”, i.e. the spot at which the highest summed vibration amplitude is obtained, is therefore often referred to when the vibration properties are described.

5.8.3 Vibration design criteria There has been a lack of specific guidance regarding vibration design of lightweight floors. Two different methods have been partly developed and evaluated in the Acousvibra project. A common design criteria for both methods is concerning fundamental frequency of lightweight floors. Due to the

104

special vibration nature of lightweight floors, they should be designed in such a way that the fundamental frequency exceeds 10 Hz (in some cases above 8 Hz can be accepted).

The first method studied in the project was based on deflection limits. The floors can be classified to different vibration classes with specified deflection limits. For typical residential buildings, the deflection limit due to 1 kN load is 0,5 mm. Based upon earlier tests together with tests carried out within the project, this design criterion has proved to be working well for different kind of floors and is therefore recommended for wider use. Advantages of this method is the simplicity and that it is user-friendly.

The second studied and developed design criterion is based upon weighted floor acceleration. The method originates from the ISO-2631 standards, in which however no specific limits are given. For heavy weight floors, the acceleration and the so called response factor usually predicts the floor performance quite well but for lightweight floors it is harder to find limits that distinguish acceptable floors from un-accepted. Limits have been introduced in the project concerning response factors of lightweight floors, but it seems that the method does not predict e.g. the acceptance of the floating floors very accurate. Furthermore, preliminary recommendations for response factor limits are given for total time of acceptance intermittent activity. This method described in British standards has been modified for lightweight floors, but more validations may be needed for final design purposes.

5.8.4 Multiple frequency vibrations One important reason that makes it so hard to find a robust, but at the same time simple, design criterion is the fact that depending on the floor’s nature, different properties are in focus. For floors with resonant behaviour, it should be adequate just to look at the fundamental frequency if, and only if, higher natural frequencies can be neglected. The next question to arise is then where to set the limit of “higher frequencies”? The answer, as long as human activities are concerned, could be a limit set to 20 Hz which is about the frequency for the eight harmonic from fast walking.

However, many floors can not be treated that simple. The general recommendation must therefore be to 1) include natural frequencies other than the fundamental and that 2) parameters other than the total energy is used if highest possible accuracy is wanted. The beating effect is according to the presented study a main key but whether the corresponding frequency separation is best treated as a design parameter of its own or if it should be complemented/combined by others is a task that hopefully future research will answer.

An important conclusion is that the suggested prediction model for annoyance based upon the ISO Wm weighting, constitutes a valuable evaluation method that seems to handle the difficulties with single vs. multiple frequency response in a decent way. However, it seems not to be adequate to use the ISO weighting directly for prediction – or even for comparison between signals - since the Wm weighting underestimate the importance of multiple frequency components. The ISO weighting handles single frequencies well and even though it has not been tested here, it could be the case that it also handles more broadband like signals well, but for signals that comprises two to three well defined discrete frequencies – a situation where the beating effect is significant but also a situation that might occur for real floors – it is not accurate enough. An objection to use this one-third-octave band frequency weighting in general design guides is that it might be experienced as a too extensive procedure by potential users.

105

6 PRODUCT DEVELOPMENT

6.1 General objectives Objectives of this work package are to develop high quality acoustic and vibration performance solutions for lightweight outer walls/facades, partitions and floors. Especially external noise of urban locations is taken into consideration in the development of the envelope of the building. New solutions are developed in order to reduce the vibration sensitivity of light weight floors. The models developed in WP2 were used to perform parametric studies in order to identify the important parameters and the corresponding constituent properties and to propose improved systems. Specimen of improved products were tested in laboratory in WP4 (acoustic performance) and WP5 (vibration performance).

6.2 Façade and outer wall structures (Task 3.1) 6.2.1 Requirements for facades The noise insulation requirements for the new façade development were based on the environmental noise acceptance criteria in Finland presented in Table 6.1. Criteria are given for the average noise level, for both day and night time inside the building. The noise requirements apply to the whole day or night period, not to momentary noise peaks. Temporary sound levels can exceed the average sound levels, even by 20 dB. This can occur, for example, when an individual noisy car passes the building in the middle of the night.

Table 6.1 The Finnish requirements for the A-weighted average sound level caused by environmental noise (Finnish Government, 993, 1992)

day time 7 a.m. to 10 p.m.

night time 10 p.m to 7 a.m.

Noise inside the building

LA,eq,s (dB) LA,eq,s (dB) Residence, health care buildings, accomadations 35 30 School buildings, meeting rooms 35 35 Offices and business 45 45

The sound insulation requirement of the facade is often determined by the town planning architect, who specifies the sound level difference ΔL. It can be calculated by equation

ΔL = LA,eq,u - LA,eq,s

This requires that both the traffic noise level outside the facade LA,eq,u (dBA) and the requirement for the highest allowed indoor noise level LA,eq,s (dBA) are known. LA,eq,s depends on the use of the room as shown in Table 1.

The requirement in urban area can be even ΔL ≥ 35 dB. That value was chosen as target value that a façade should fulfil.

Another task in this work package was to develop outer wall structure to the industrial building with noise source inside the building. In this case, the requirement for the sound insulation value was Rw ≥ 56 dB.

6.2.2 Development process Facades against traffic noise In order to fulfil the sound level difference requirement ΔL ≥ 35 dB of the whole façade, the laboratory measured value of the prefabricated wall element (without windows) should fulfil Rw+Ctr,100-3150 ≥ 45 dB. This is valid for typical size of office or residential room and gives already a safety margin of about 4 dB (Finnish guidance for dimensioning façade components[48]).

107

The first studied façade type is light-weight wall panel based on thermal studs and used as infill walls in office or residential buildings with primary steel or concrete frame. Figure 6.1 shows the structure and the installation of the element. The element is composed of inner layer gypsum board (13 mm), vapour barrier, perforated C-stud +mineral wool (175 mm) and outer layer gypsum board (9mm). The outer façade material can vary but in office buildings it is typically steel cassette as shown left in Figure 6.1.

Figure 6.1 Structure of the light steel element and installation.

Some previous tests results for the pure wall element without outer steel cassette skin were available. Those results gave Rw+Ctr,100-3150 = 43 dB. During this project, some modifications were made to the wall stud and new test series were carried out in VTT in WP4. Two versions were tested: First test wall was similar than shown in Figure 6.1 without steel facade. In the second wall there was 0,6 mm steel sheet under the inner gypsum board. The purpose of the steel sheet is not only to improve acoustic behaviour, but it is used as weather guard during transportation and installation and the inner gypsum board will be assembled later at the same time as partition walls are finished. The influence of the extra steel sheet was initially estimated according to methods developed in WP2. Following test results were achieved:

• Wall 1: Rw+Ctr,100-3150 = 51 - 9 = 42 dB

• Wall 2: Rw+Ctr,100-3150 = 54 - 9 = 45 dB

Thus it was noticed that the test result for wall 1 was quite similar than received in the previous tests, although modifications of stud. The improvement of the sound insulation behaviour due to extra steel sheet was quite expected and the result was also realised to be adequate to the facades in urban areas with the requirement of ΔL ≥ 35 dB.

The sound insulation of the built façade with windows against traffic noise was validated by measuring in four different field tests. In all cases the laboratory measured sound insulation value for the windows were Rw+Ctr,100-3150 = 40 dB. The measured sound level differences ΔL were 38…41 dB for the facades with external steel cladding or without cladding and 41…44 dB for facades with brick covers. Measured and calculated exact sound level difference values (EN12354-3) are shown in Table 6.2.

Table 6.2 Measured and calculated sound level differences in field tests Test case

EN 12354-3 D2m,n,w +Ctr

Measured Dn,w + Ctr

Field test 1: Wall 1 + 135mm brick 46 43 Field test 2: Wall 1 + 135mm brick 47 44 Field test 3: Wall 1 42 41 Field test 4: Wall 1 + steel cladding 42 38

108

As shown in Table 6.2 measured values were quite good and the difference to the calculated values was 4 dB in maximum. The difference is acceptable due to installation errors and accuracy of measurement. All measured valued fulfilled the requirement of ΔL ≥ 35 dB even with wall structure 1.

Flanking transmissions In order to avoid flanking between two rooms inside the building the junctions of the building elements presented in the previous chapter must be carefully designed. Two different wall- concrete floor connection types are introduced in Figure 6.2. In both cases the sound transmission through the wall stud was tried to minimize by cutting the element between two storeys. Both versions were also tested in site.

Figure 6.2 Two different floor-wall connection types

In both cases in Figure 6.2, the laboratory test results for pure floor structure have been measured Rw = 64 dB. Field test results for the structure shown in left were R’

w = 59…64 dB in test building 1 and R’w

= 58…62 dB in test building 2. In test building 1, measurements were made also in the case of concrete outer wall and the measured results were R’

w = 61…63 dB. Field test results for structure shown in right were R’

w = 61…63 dB and corresponding measurement with concrete outer wall from the same building gave R’

w = 60 dB. In the field measurements the results are usually R’w = 58…62 dB when the

outer wall is made of concrete. Thus it can be concluded that in the buildings with lightweight steel facades the sound insulation between dwellings is quite similar compared to the concrete outer wall system. Generally the measured sound insulation indexes were good and fulfilled clearly the Finnish sound insulation requirement R’

w ≤ 55 dB.

Outer walls for industrial building against internal noise Because the noise insulation requirement Rw ≥ 56 dB in this case was quite high, it was obvious that more mass and/or multilayered structures are needed compared to structures presented in previous chapter. Modified element type was developed for industrial buildings where the thermal studs (purlins) are in horizontal position from the column to the column whereas in the conventional wall stud element the studs are in the vertical position from the floor to floor. In the modified element, the internal surface of the element was slightly profiled thin steel sheet with thickness of 0,8 mm…1,5 mm. Different multilayered structures were developed in order to improve sound insulation performance. The use of different kind of boards, such as cement boards (further or instead of gypsum boards), was investigated. Before testing, calculations were made in order to estimate the influence of the different layers on the airborne sound insulation. Finally, the structure shown in the Figure 6.3 was chosen to the tests. The calculated airborne sound insulation index for the structure with 0,8 mm steel plate was Rw = 58 ±2 dB and thus the structure should fulfil the Rw ≥ 56 dB requirement.

109

Figure 6.3 Wall element for industrial buildings

Airborne sound insulation tests were carried out for structure type shown in Figure 6.3. Following test results were achieved:

• Wall 1 (with steel plate t= 0,8 mm): Rw = 56 dB

• Wall 2 (with steel plate t= 1,5 mm): Rw = 59 dB

Thus both versions fulfilled the requirement as was expected. By introducing more layers to the structure shown above, it is possible to get very high acoustic performance wall structures for special needs. For example the structure shown in Figure 6.4 can achieve airborne sound insulation index Rw = 70 dB. In this case, further to previous case, there is also cement board inside the element and one normal gypsum board on the outer side of the element.

Figure 6.4 High acoustic performance wall element

110

6.3 Floors (Task 3.2) 6.3.1 Requirements for floor structures Floor structures shall fulfil airborne sound insulation and impact sound insulation requirements. In Finland, those limits are R’

w ≥ 55 dB and L’n,w ≤ 53 dB. Furthermore a vibration performance of the

floor should fulfil vibration class C requirements in the case or typical residential and office building (See Chapter 3.5.2). In practise this means that the fundamental frequency of the floor should be ≥ 10 Hz and the deflection of the floor due to 1 kN load should be less than 0,5 mm.

6.3.2 Lightweight floors Introduction Light-weight steel floors are typically used in residential buildings, but some experiences are also from office buildings. In two storey residential buildings, the floors are fixed on the load-bearing light-weight exterior walls while in multi-storey buildings, the floor elements are supported by separated steel frame.

Light steel framing has several advantages, for instance: • Dry construction, the risk of moisture problems can significantly be reduced.

• A high level of prefabrication is possible. This leads to reduced construction time and reduced costs.

• Reduce the need for ground stabilization due to the light weight building

• Easy to place water pipes, ventilating piping, electricity wires etc.

Typical light-weight floor structure and installation of partly prefabricated element are shown in Figure 6.5. Typically the load bearing profiles are cold-formed C-sections. Trapezoidal sheeting is installed perpendicular to the steel joists and two layers of gypsum boards are screwed to the sheeting. Gypsum boards can be replaced also with thin concrete layer (about 50 mm) in order to increase stiffness of the floor. Ceiling is composed of one or two layer of gypsum boards.

Figure 6.5 Example of structure and installation of light-weight floor

Prefabricated floor elements normally span across a single bay, while floors built on site may span two or more bays. The advantages with continuous spanning are increased stiffness and damping, the disadvantages are that vibrations may be felt in adjacent rooms and that flank transmission may increase.

111

Typical span areas for single span structures are from 4 m to 6 m. The challenge is to avoid unpleasant vibrations for dynamic loads and the fact is that usually the vibration criteria determine the maximum span length of the light-weight floor structure.

The first problem is that a light construction is much easier to excite than a heavy construction. As we do not want to make the construction heavy, the solution is to make it as stiff as possible for point loads. The other problem is that steel is very elastic and it does not absorb vibration energy well, which means that vibrations after an impact may be felt for a long time. The solution to that is to make the composite construction as energy consuming as possible.

The floor development in this project was focusing on increasing the perpendicular stiffness of the floor in order to divide the load to the wider area of the floor. Especially the composite action between the floor cover (gypsum boards or concrete layer) and steel sheeting was utilised. In longitudinal direction, also the composite action between steel joist and concrete layer was considered.

Vibration problems in light-weight floors are usually caused by walking. Thus the loading levels are usually very low and the floor behaves fully elastically. For that reason, it can be assumed that full composite action can be assumed between different structural layers even with small amount of connectors.

Developed cover structures with fastening details are shown in Figure 6.6 for gypsum board covered floor and in Figure 6.7 for concrete covered floor. In the last-mentioned case the composite action between steel joist and concrete (and trapezoidal sheeting and concrete) is ensured with uplifted screws (every second screw). In concrete covered version the perpendicular stiffness is considerable high (about 1/50 of main direction stiffness) and the deflection due to walking should be effectively distributed to other main joists thus decreasing the total deflection.

Figure 6.6 Cover structure fastening details in gypsum board light-weight floor.

112

Figure 6.7 Cover structure fastening details in concrete topped light-weight floor.

Tests and conclusions Light weight floors shown in Figure 6.6 and Figure 6.7 were tested in laboratory at LTU (See more detailed description of the tests in Chapter 5). Concrete covered floor was tested also in detached house in Kauklahti, Espoo, Finland. Dynamic properties and static deflections of the floors were measured and sense perception tests were carried out. The laboratory tests were partly unsuccessful because of horizontal vibrations due to experimental set-up. Anyway, the concrete covered floor was tested also in more controlled support conditions and the floor dynamic properties and static deflections were as expected and the floor was also judged as accepted in subjective tests. The field test results were also accordance with predictions, although the measured static deflection 0,03 mm was even considerable lower than expected value 0,11 mm. That floor was also accepted in subjective tests.

Although some adversities in laboratory tests, a lot of valuable information for product development was possible to extract from the results. First, the importance of the support conditions to the floor performance was realised. The horizontal vibration of the floor due to walking activities should be effectively prevented. In practice this is easier to fulfil than in laboratory conditions because the supporting members belongs to the frame of the building and the frame itself is quite well stabilised.

The second important observation from the tests was that more effective connectors are needed between steel joists and concrete slab. A fast drying of the concrete slab caused warping of the concrete slab and thus the adhesion between trapezoidal sheeting and concrete slab was lost at the edges of the floor. Stronger fasteners are needed with more dense spacing at least at the edge areas of the floor.

Finally, at least the results of concrete covered light weight floors showed that the performance of the floors were acceptable and the test results were accordance with design formulas and criteria given in Chapter 3.5.2 and in Finnish design guide for vibrations[49]. Design guide includes different vibration classes where class C can be used in normal residential and office buildings. As a simple tool for designer, design charts for the light-weight floors were prepared. Figure 6.8 shows example of design chart for a concrete covered light-weight floor. The solid lines represent different steel joist heights in different live load values when the criterion is either ultimate limit state or serviceability limit of L/600. Dashed lines show limits for different vibration classes for each steel joist heights.

113

Light-weight floor, concrete covering: C200... 350-2.5

1

2

3

4

5

6

7

8

3.0 4.0 5.0 6.0 7.0 8.0

Span [m]

Live

load

[kN

/m²]

C350-2.5

C300-2.5

C250-2.5

C200-2.5

Class A

Class B

Class C

Class E

Figure 6.8 Example of design chart for light-weight floor

6.3.3 Semi-heavy floors Development of semi-heavy floors (so called Kvantti-floors) continued in the framework of Acousvibra project. The structure of the floor is described in Figure 6.9. The steel beams (c/c 1200 mm) are either I- or C-section with height about 300…400 mm. The lower flange of the beam is casted into the concrete slab with thickness of 80 mm. The floor covering may be constructed e.g. by wooden planks, by trapezoidal sheeting + gypsum boards or by composite slab. The space between the concrete slab and covering layer can be utilized to house technique. The floor is delivered to the building site as an element with lower concrete slab and beams. The element width is typically 2400 mm. The acoustic and vibration performance of the floor had already been measured in laboratory conditions formerly. During the project, the floor system was piloted in two residential buildings, where acoustic field tests were carried out.

Figure 6.9 Kvantti-floor with wooden plank covering

114

Airborne sound insulation results were acceptable and the results were good in accordance with the laboratory measurements fulfilling the Finnish sound insulation requirement of R’w ≥ 55 dB. The weighted impact sound level L’n,w in the field was not as good as expected (from 56 to 59 dB). In the laboratory measurements it was found that besides the damping properties, the position of the damping material was an important factor in achieving the impact sound insulation requirement. If the damping material was situated on the flange of the steel beam (See example in Figure 6.10) the results were somewhat worse than when it was positioned right over the web. It might be that this was the main reason for poor results in field tests also. It should be also noted that the whole upper part of the floor should be supported through damping rubbers and the real workmanship was not checked on the building site before the floor covering.

Figure 6.10 Kvantti floor structure measured in field in the case of gypsum board + trapezoidal sheeting covering.

For further developing work it may be concluded that:

- the structural connection (its damping properties and the way how the impacts are transferred to steel beam ) between top floor and steel beams is of a great importance

- it is also of importance to know the modes of the steel beams section, especially the bending waves in the flange and in the web, and to study if there is possibilities to change or modify the beam section so that the basic modes of the beam and the concrete slabs are not awakened

- also the path from the steel beams to the concrete slab should be studied more precisely (the coincidence frequency of the concrete slab is about 200 – 250 Hz and over these frequencies the slab is radiating sound due to bending waves very effectively).

6.4 Partition walls (Task 3.3) 6.4.1 Requirements for partition wall structures Two targets were set in this task. The first aim was to make improvements to the current partition wall structures in order to achieve more economical solution with at least the same acoustic performance as with current products. The second task was to make a preliminary design for totally new partition wall system with new type of studs and rails. The target values set for the sound insulation indexes in different cases in the second task are given in Table 6.3.

Table 6.3 Target values set for sound insulation indexes Stud height

Gypsum boards/side

Rw [dB] Max. wall height [mm]

70 mm 1 44 3000 70 mm 2 52 4000 95 mm 1 46 4250 95 mm 2 54 5000

115

6.4.2 Development process Improvement of the current partition wall system Main aim in the improvement of current partition wall structures was that the acoustic performance would be at least as high as with current products, but steel material consumption in studs and rails could be reduced. Also the roll forming of the studs and rails should be possible without major modifications to the production line. Beside the acoustic performance of the partition wall, the wall should also meet the requirements for stiffness, i.e. a deflection of the wall under horizontal static load should not exceed some specific limit that is usually 10 mm or L/300. Furthermore the installation requirements should be fulfilled e.g. the stud can not be too flexible. The material saving is possible by introducing thinner studs and rails as used previously. Fulfilling the stiffness requirements the studs were embossed in order to increase the section stiffness of the stud profile. Different embossed types were studied (See examples in Figure 6.11).

Figure 6.11 Different embossing types

Simple bending tests were carried out to gypsum sheathed wall structures with studs of different embossing types and with studs without embossing. The purpose of the test was to find out relative change of the stiffness of the wall by using embossed studs manufactured from 0,47 mm steel steel coil and non-embossed profiles manufactured from 0,56 mm steel coils as standard products. The main goal was to find out if the standard studs could be replaced with embossed thinner studs. The test set-up is shown in Figure 6.12.

Figure 6.12 Bending test for partition wall structures

The bending tests showed that e.g. the embossing type shown on centre and right in Figure 6.11 worked very well and the stiffness of the structure was adequate and was only slightly lower than with solid stud with thicker material. The tests also showed that ultimate load bearing capacities were improved by using embossing. Practical installation tests were also carried out in order to ensure the mountable of the system.

After final embossing type was chosen, standardised airborne sound insulation test was carried out in laboratory conditions. Basic single frame structure with 66 mm studs, rails and one layer 13 mm gypsum boards on both sides and mineral wool infilling gave sound insulation index Rw = 43 dB. The achieved value was the same as received for the structure with normal studs.

116

After confirmation of acoustic performance, a comprehensive set of full-scale standardised horizontal static load tests were carried out to the wall structures on vertical position. The test program consisted of 22 tests including different wall heights and wall thicknesses. Furthermore specimens had the single or double boarding on the both sides. As a final result, the maximum wall heights were determined for each wall type corresponding to the line loads of 0,5 kN/m and 1,0 kN/m in the middle of the span. Figure 6.13 shows example of design chart for maximum wall heights with different stud heights and other configurations.

Figure 6.13 Example of design chart for maximum wall heights

Novel partition wall stud and rail development As discussed in Chapter 4, an influence of the stud and rail characteristics on acoustic behaviour is considerable. The stiffness of the profiles is the most important factor. In the beginning of the project, it was shown that the dynamic stiffness of the profile can be used in the estimating airborne sound insulation indexes. Thus, some preliminary dynamic stiffness tests were carried out in order to determine the dynamic stiffness values and make comparison between different profiles. The dynamic stiffness tests were carried out according to test method described in Figure 6.14. Seven different stud profiles were tested.

117

Figure 6.14 Test set-up for measuring dynamic stiffness

The results achieved from the tests were utilised in the ranking of different profiles, but absolute numerical values were difficult to implement in calculation models. Later research on CSTB showed that actually the punctual static stiffness of the studs and tracks can be used as an input parameter in the prediction model (See Chapter 4.1.2). The stiffness values can be determined either by laboratory tests or by FEM-calculations. More detailed description is given in Chapters 4.1.3. In order to calibrate the calculation method and also to get information to the product development process, airborne sound insulation laboratory tests were carried out to three different types of partition wall types shown in Figure 6.15. In these tests it was also paid attention to stiffness of the rails, what is also recognised as significant factor. The stud height was 66mm in all cases. The stud type a) was embossed as described in earlier chapter. The test walls were composed of studs, rails and one layer or gypsum sheathing on both sides with mineral wool filling.

a) b) c)

Figure 6.15 Three different tested partition wall stud and rail combinations

Tested sound insulation index values were Rw = 42 dB in case a), Rw = 41 dB in case b) and Rw = 43 dB in case c). The case a) but with normal rail gave Rw = 43 dB. Differences between different configurations were relatively small. Laboratory characterization was made to the type a) studs and type 3) stud and rail and the test results were good in accordance with predictions in these cases.

After evaluation process it was possible to estimate the airborne sound insulation index reliably if the punctual stiffness of the stud and rail is known. Thus it was also possible to calculate out the maximum punctual stiffness values for the wall studs and rails in partition wall structures fulfilling different Rw -target value cases shown in Table 6.3. As an example of the calculations made by CSTB are shown in Table 6.4. In this case the target value for the Rw value was 44 dB and it can be see that the maximum punctual stiffness is then 0,60 MN/m.

118

Table 6.4 Predicted acoustic index for the 66 mm wall with single gypsum board layers on both sides.

Studs/Rails stiffness (MN/m)

0,05 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 0,90 1,00

Rw [dB] 49 49 47 46 45 44 44 43 43 42 42 C [dB] -7 -7 -5 -4 -4 -3 -4 -3 -4 -4 -4 Rw+C [db] 42 42 42 42 41 42 40 40 39 38 38

Final task in this work package task was to try to find out wall stud shapes that fulfill the punctual stiffness requirements found for different sound insulation classes. This analysis was carried out by FEM by UPC for the some initial profile types shown in Figure 6.16. The reference profile was typical C-section. Some profile dimensions such as were varied and optimized during the analysis.

Figure 6.16 Different stud profiles analysed

The main conclusions of the parametric study was that (1) acoustic profiles such as types 2,3,4 and 5 indeed have a considerably lower stiffness (55% to 75% of profile C), so a significant improvement in acoustic performance is to be expected; (2) outward folds are more effective that inward folds in providing extra flexibility to the stud and (3) fold width is a much more relevant parameter than fold height. In fact, a practical guideline for the design of acoustic studs is to explore fold shapes that two (or more) widths.

6.5 Conclusions Different high performance products were developed in the framework of the project. Two types of light-weight façade elements were studied and developed: 1) for residential and office building fulfilling high traffic noise insulation requirements and 2) for industrial building fulfilling high airborne sound insulation value. Both applications are based on perforated light-gauge steel thermal studs fulfilling also high thermal insulation requirements. In both cases, prediction models developed in the WP2 were utilized in the design of walls and finally products were also tested in the laboratory. In the first case, also field tests were carried in order to validate the acoustic performance in actual buildings and in order to validate calculation methods. Connection details were also verified by field measurements.

Two types of floor structures were studied and more developed in the project. Light-weight floors were composed of light steel joists, gypsum board ceiling and the covering that was built from trapezoidal sheeting and either two layers of gypsum boards or thin concrete slab (~50 mm). Special attention was paid on composite action between different components in order to improve floor stiffness properties in both directions to steel joists. Vibration tests were carried out in the laboratory as well as in situ. Laboratory tests indicated the importance of connection details on horizontal vibrations of the floors due to walking activities. Finally, subjective tests with proper floor connections showed acceptable vibration behavior of the light-weight floors. Similar results were received also from field tests. The subjective tests also indicated that the deflection based design criterion predict well the floor performance. Furthermore, some field acoustic tests were carried out to semi-heavy type of floor, that was earlier developed and tested in laboratory. The acoustic field tests showed good airborne sound insulation properties, but some minor development needs are still concerning impact sound insulation.

119

Third group of developed products composed of partition walls. First, some modifications were made to the typical partition wall. The material saving was possible by introducing thinner studs and rails as used previously. Fulfilling the stiffness requirements the studs were embossed in order to increase the section stiffness of the stud profile. Different embossed types were studied and finally bending tests and acoustic tests were carried out to the modified partition wall type. Furthermore, some preliminary design of totally new type of partition wall was started. The study for the optimum shape of the wall stud and rail was made in close cooperation with other partners and research on work packages 2 and 4. It was possible to define some simple parameters to the partition wall studs in order to be able to predict the airborne sound insulation analytically. Those parameters for wall studs can be determined by simple laboratory tests or by fe-modelling. Some preliminary proposals for new section types were given.

120

7 DESIGN GUIDE FOR ACOUSTIC AND VIBRATION PERFORMANCE OF LIGHTWEIGHT CONSTRUCTION

7.1 Introduction and objectives

One of the main objectives of the ACOUSVIBRA project was to compile a European design guide for practical engineers and architects. Amongst general guidance on the design of lightweight construction for good acoustic and vibration performance, the guide was to introduce how to utilise modelling as a tool in the development of new structures. This guide1 was produced during WP6, and relied on work carried out in the other work packages. It involved the collaboration of all the participants in the project.

The design guide provides practical information for the design of lightweight steel structures so that they may have an adequate acoustic and vibration performance. A set of design rules for acoustic performance (Task 6.1) and vibration performance (Task 6.2) have been developed, and are included in the guide. It illustrates many examples of buildings using lightweight construction and their components and connection details. It demonstrates that lightweight steel construction is perfectly capable of meeting modern acoustic and vibration serviceability requirements.

The guide outlines the current requirements for acoustic performance in many European countries for floor, separating wall and façade design. It illustrates many examples of floor and wall construction, together with their insulating properties, and makes recommendations for best practice (Task 6.3). Major parameters that affect the performance at the whole building level are discussed, as well as for component design and junctions. Sound insulation against traffic noise is also covered. The basic principles of the modelling methods used to predict sound transmission are provided, including the ‘wave’, energy (SEA) and finite element approaches, and guidance is given on their specific use for lightweight structures.

Specific proposals for the dynamic assessment of lightweight floors are presented in the guide (Task 6.4): these include a deflection-based model and an acceleration-based model. Advice on finite element modelling of lightweight floors is also given. Case studies are presented, whereby field measurements of dynamic behaviour of floors are compared with predictions using these methods. The key parameters influencing floor vibrations are also discussed, and current methods of on-site testing and measurement, data processing and subjective evaluation are given. The guide also presents proposals for harmonization of annoyance criteria.

The guide’s contents include 5 sections, plus references and Appendices, as follows:

1 INTRODUCTION 2 TYPICAL LIGHTWEIGHT steel CONSTRUCTION

2.1 Lightweight steel construction systems 2.2 Building parts

3 Acoustics 3.1 Current requirements and Standards 3.2 Major parameters influencing acoustic performance 3.3 Modelling methods

4 Vibrations 4.1 Current requirements 4.2 Major parameters influencing floor vibrations 4.3 Testing procedures 4.4 Design methods 4.5 Case studies

5 RECOMMENDATIONS FOR CONSTRUCTION – DESIGN AND DETAILING 5.1 Designing for good acoustic performance 5.2 Designing for good dynamic performance

121

6 REFERENCES Appendix A Further examples of typical construction and their acoustic performance Appendix B Standard Testing Procedures

A précis of the main sections in the design guide is given below in Section 7.2 to Section 7.6.

7.2 Design Guide Chapter 2: Typical Lightweight Steel Construction

Lightweight steel construction is a collective concept for construction systems that primarily contains light cold-formed steel profiles, plasterboards and mineral wool insulation. Steel studs have been used in walls for more than 30 years, but it was not until competitive light floor structures were developed that the entire construction system with structural steel studs was established. Typical light steel construction systems are:

Structural framing for walls, floors and roofs in low-rise residential buildings.

Walls and flooring in multi-storey office and residential buildings, complemented by a primary structural steel and concrete framework.

Interior, infill and separating walls in almost any building enveloping people, such as schools, hospitals, offices etc.

Cladding and roofing for almost any building type, from residential to industrial.

The technology and materials enable the dry and efficient prefabrication of structural elements, or the manufacture of fully furnished modular units, which are easily connected on site. With its low weight relative to the floor surface, (typically 150 kg/m2), it is also suitable for the extension of existing buildings, seismic areas, and where the foundation conditions are poor. Typical lightweight steel construction is also cost competitive, quality assuring, time efficient and environmentally sustainable. Lightweight steel technologies are now used worldwide.

7.2.1 Lightweight steel construction systems Multi-storey structures Lightweight steel multi-storey buildings often use a technology based on a structural steel frame with beams and columns, complemented with steel stud interior walls, light or semi-light flooring of different types, and light steel infill walls at the periphery of the building. This enables a rapid dry envelope to be created in the temporary condition, and provides insulation and lateral support to the cladding in the permanent condition. Multi-storey buildings tend to be heavier when more than six storeys high, but the lightweight steel technology can be used beneficially up to about eight storeys, see Figure 7.1. The acoustic and vibration performance is not related to the number of floors, but is dependent on the principal design and workmanship.

122

Figure 7.1 Two eight-storey lightweight steel residential buildings with a gross weight of 480 kg/m2, including all materials. (Sweden)

Housing Light steel framing can be used for single family housing, terraced housing, small apartments and multi-occupancy buildings. The various steel components may be used in almost any part of the building. The occupants often require a better performance than is stated in existing building regulations, which can be met by lightweight steel construction. Lightweight steel housing is now used almost worldwide. The structure can be given a traditional appearance, an ultra-modern functional design, or its design can reflect the need for temporary living accommodation in countries needing disaster relief. An example of modern cold climate light steel housing is shown in Figure 7.2. It can be beneficial to prefabricate housing units because of the advantages of speed of construction and quality control.

Figure 7.2 Single-family housing using lightweight steel (Loiste, Finland)

Element and modular systems The traditional method of construction using beam and column elements for steel framed buildings, known as “beam and stick”, has not proved popular for lightweight steel, and the efficient industrialised prefabrication of 2D elements (walls, floors and roofs) has been regarded as the only competitive

123

method for the system. The next step in this development to improve efficiency was reached by introducing the 3D modular system, including industrialised production of fully equipped volume elements ready for assembly. The choice of production method mainly depends on the size of the project and the degree of standard design.

Light framed elements The lightweight steel framed elements for walls, floors and roofs are produced in a safe and dry off-site industrial facility, see Figure 7.3, or in an established weather protected on-site assembly hall. The prefabrication facilitates accuracy, time efficiency and high quality. The parts can be fully assembled before final erection, including windows, services and surface finishes, or they can be partly assembled so that surfacing works can be completed on site. The low weight of the final construction elements make them easy to handle and mount, and involves few workers. The prefabrication also ensures optimised acoustic performance of details in the construction, which may otherwise be difficult to achieve with on-site construction. For large projects and repetitive production, some manual processes can be automated.

Figure 7.3 Off-site assembly of a wall element.

Composite elements Thin plate steel sandwich panels are widely used in industrial and storage buildings, and in other large buildings where efficient, economical fire resistance and thermal insulation is needed. The composite action between the steel and the stiff mineral wool insulation makes the system self-bearing, and also usable as roof panels for, say, offices and residential buildings. A typical sandwich panel is shown in Figure 7.4. The filling can also be expanded polystyrene (EPS), which is not fire proof. For increased structural strength, the panels are joined by steel columns and beams.

124

Figure 7.4 Schematic example of a "sandwich panel" wall element with sheet steel and high density mineral wool.

Modular construction Modular construction is primarily a system based on a combination of a load-bearing steel framework of thin-walled sections and prefabricated room modules, based on the lightweight steel technology. A building can comprise a single module, or many modules, such as with a multi-storey apartment block. This construction technology has been used efficiently in large residential projects, and one example is shown in Figure 7.5. The modules can be fully furnished and include all surface finishes, services and equipment when the module leaves the area of assembly for transportation to the construction site. The philosophy with the modular system is to industrialize the production of the structure and the interior in order to reduce time and cost, and to increase quality and sustainability The acoustical properties can differ some from stick-built and 2D element construction, mainly owing to short spans, joints between modules, and cavities related to the frame.

Figure 7.5 Lightweight steel module being elevated into its final position. (Sweden).

7.2.2 Building parts Frame The combination of light steel framing and plasterboard provides a construction system with the highest load-bearing capacity and rigidity in comparison to its weight. However, higher structures need a stabilising core or frame in order to withstand increased vertical and horizontal loads, such as wind or seismic forces. All loads must be transferred to the foundation of the building, and typical stabilisation methods for lightweight steel buildings are:

125

Trusses, or a combination of wallboards and trusses through stressed skin diaphragm action

Rigid steel frame, fully or partly integrated with the light system.

Rigid concrete core, normally including stairs and elevator shaft.

The choice of method depends mainly on the design of the building, the layout and the number of floors. Typically, a lightweight structure can be built with up to three floors without an additional rigid frame. The frame itself can affect the acoustic and vibration properties of the building, i.e., in terms of structural rigidity and sound transmission through the frame.

Wall and façade Lightweight steel walls are commonly used in offices or residential buildings with steel studs as framing, complemented with plasterboards or plywood and mineral wool insulation. Steel walls can be used as sandwich panels in public and industrial buildings, where they comprise steel sheets with a core of insulation (mineral wool or EPS). The lightweight load-bearing walls are most cost-effective in buildings of up to three storeys. The most common use of steel framed walls is the “drywall” for internal walls, owing to its flexibility and easy construction principles. The requirements concerning sound insulation, fire prevention and thermal insulation are met by the use of plasterboard and rockwool insulation. External steel framed walls can be used in almost any building type, and are often prefabricated complete with finishes.

The steel profiles are normally hot dip galvanised with 20 μm of Zinc on each side for corrosion protection during storage and transportation. Light steel walls generally comprise C-shaped studs at 600 mm spacing, but occasionally at 400 mm, and are joined to C-shaped rails fastened to the floor and ceiling. External wall sections range in size from 100 to 225 mm, and for internal walls normally 45-70 mm. Slotted thermo-profiles have been developed in order to reduce heat losses through the steel section (thermal bridging), see Figure 7.6. It is then normally sufficient to place the insulation between the studs. These “slots” have almost no effect on the acoustic properties.

Gypsum plasterboards are the most commonly used as sheeting boards. These serve the normal wall functions, including: structural load resistance, stabilising the studs against buckling from vertical loads; resistance to horizontal loads due to diaphragm action; fire resistance. There are different kinds of plasterboards available for different applications, e.g., reinforced fireboard, impregnated paper for bathroom usage, and also outdoor board. In load bearing walls plywood or particleboards with sheeting can be used for increased load bearing capacity. Fibre cement boards can also be used in combination with lightweight steel frames.

For thermal and sound insulation, mineral wool is used in the wall. In those cases where warm frame construction is used, EPS can be used, but it is not fire proof or preferred for acoustic insulation. The insulating capacity can be varied by using a different insulation density and/or thickness.

126

Figure 7.6 Lightweight floor and wall construction showing the assembly of an external wall element with “thermo profiles”, external plasterboard and a rendered EPS façade

The façade design possibilities are almost unlimited when using light steel walls, and some examples are shown in Figure 7.7. In certain cases the wall and façade is integrated with composite interaction, but the wall normally acts as a support for the façade. Common façade types are brick walls, EPS with rendering, thin steel sheet or cassettes, Minerite fibre plates and wooden panels. All fundamental requirements can be met by each of these solutions.

(a) (b)

Figure 7.7 (a): Combination of rendered facades and Minerite fibreboard panels in a light steel residential project using eight different facade systems (Malmö, Sweden). (b): Brick wall façade and a lower level curved roof using lightweight steel construction (Oxford, UK).

127

Lightweight steel partition walls are used primarily for room separation, and enable flexible, dimensionally stable, functional and easily-built solutions for many building types, see Figure 7.8. The system is also used for bathrooms, but is not used for structural load support. In some countries almost no other framing material is used for partition walls. Separating walls between units of accommodation provide both acoustic insulation and fire compartmentation functions. The thickness of these is usually twice that of a partition wall.

Figure 7.8 Schematic design of an insulated partition wall. (a) Ceiling rail. (b) 70 mm light steel stud, c/c 600 mm. (c) Hole for hidden services. (d/e) Single or double plasterboard layers. (f) Mineral wool for acoustic insulation. (g/h) Surface treatment. (i) Rubber strips for improved acoustic performance.

Flooring, ceiling and roof Lightweight steel floors are used most commonly in residential buildings, but some use is made in office buildings. In low-rise residential buildings the floor is connected to the load-bearing lightweight walls, while in multi-storey buildings the floor elements are generally supported by a steel framework. Prefabricated floor elements normally span across one single bay, while floors built on site may span two or more bays. The advantage with continuous spanning is that it provides an improved vibrational performance from the increased stiffness and damping. A disadvantage can be that vibrations may be felt in adjacent rooms, and flanking of sound may increase. Typical single spans are up to 6 m using a very light floor structure, and much more for semi-light structures.

These double-layer flooring systems consist of a top floor system and a bottom system, which are separated. The bottom system acts as a suspended ceiling. The cavity in between these layers includes the load bearing steel profiles, services and mineral wool for acoustic and fire insulation. With regard to impact sound and vibrations, this light double layer system has very different properties compared to heavy single layer flooring systems. These properties are given in the following chapters.

128

Most lightweight steel floors are based on load bearing C-, Z- or sigma shaped purlins, 150-350 mm profile height, made from cold-rolled hot dip galvanised steel. Typical purlins have a thickness range of 1.5-3 mm. Standard sections are normally supplied from purlin producers. An example of a C section used in a floor is shown in Figure 7.9(a). Semi-light structures also include standard H or I section hot rolled steel profiles for increased strength and stiffness, as shown in Figure 7.9(b). Stiffeners are usually needed at the end supports of the lightweight steel purlins in order to avoid web crippling. End stiffeners can be U- or L-shaped sections.

Where trapezoidal sheeting is used, it is screwed on the top of the floor joists. The sheet profile is often 35 to 45 mm high and has a material thickness of 0.7 to 0.8 mm. Trapezoidal sheeting distributes vertical loads and gives perpendicular stiffness to the floor, but also functions as a horizontal diaphragm. The top of the floor can consist of up to three plasterboard layers, or alternatively of a rather thin layer (50-70 mm) of concrete or anhydrite. Boards are screwed to the purlins and glued together. Normally, 15 mm thick extra hard plasterboard or plywood is used in board-based floor construction. Composite action between the steel purlins and the top layer may be utilised in stiffness calculations.

The suspended ceiling normally consists of one or two layers of plasterboard mounted with resilient connections to the joists, in order to reduce sound transmission. This flexible connection can consist of an “acoustic stirrup” to which secondary “acoustic profiles” are attached. The plasterboard also adds fire resistance to the structure.

(a) (b)

Figure 7.9 (a): A light steel flooring system, comprising a double gypsum layer on a profiled steel sheet, load-bearing C-profiles and a resilient ceiling connected to "acoustic rails". (b): A semi-light steel and concrete composite flooring system with a parquet floor solution and provision for hidden services.

Roof structures in modern buildings are often constructed using light steel. They can be made using profiled sheet steel on a light steel framework, sandwich panels, light steel roof trusses carrying any top layers, or ‘warm’ light steel elements integrated with the top floor construction. With regard to acoustic performance, the roof structure mainly affects sound insulation from exterior noise, but might also affect sound propagation from machinery on the top storey and service installations. Light steel roof structures are flexible and cost-efficient, and are therefore commonly used in many building types, covering small and large areas. An example of a light steel roof structure is shown in Figure 7.10.

129

Figure 7.10 A roof structure using light steel.

Details Much of a building’s acoustic and vibration performance depends on the parts often regarded as minor details, such as the junctions between the many components of a building structure. Continuous development leads to optimised design of details. It is of great importance that the building contractor follows the specification set by the designing engineer for the best overall performance of the building.

Junctions Wall-to-wall and floor-to-wall junctions can take different forms, depending on the structure type, span and requirements. The key issue it to prevent sound and vibration transmission between neighbouring rooms, both vertically and horizontally, and from propagating throughout the structure. Flanking paths are reduced by structural breaks and soft insulation materials. Figure 7.11 shows two junction types between lightweight walls and semi-light steel floors. In example (a), the free end floor support is placed outside the wall, owing to efficient design of the supporting steel beam. The junction and flooring solution in image (b) includes several details for improved acoustic and vibration performance.

130

(a) (b)

Figure 7.11 Junction types between lightweight walls and semi-light steel floors. (a): a light steel external wall and semi-light concrete flooring. (b): Similar junction but semi-light steel flooring and fixed end support.

Fasteners and connections Structural capacity and acoustic properties also depend on the connections between the different components. Screws or rivets are used as fasteners for the light steel; screws for the boards, and screws or bolts for other connections. Fasteners are normally standardised and traceable components. The number and type of fasteners, torque, pad and/or the gap between the connected components can affect sound propagation in the building.

7.3 Design Guide Chapter 3: Acoustics The main information given on current requirements and Standards in this chapter in the guide has been presented in Section 3 of this document, so it is not repeated here.

The guide provides information on the major parameters influencing acoustic performance at a component level, a whole building level, and includes the effects of junctions. Some case studies of buildings which have been modelled in detail are presented. Methods of modelling sound transmission through buildings, including wave, SEA and finite element approaches are covered.

7.3.1 Component level Walls and façades Lightweight steel walls are made of double leaf elements mounted either between apartments (separating walls) or between rooms inside the same apartment (internal partitions). Depending on their level of acoustic performance, they can have a single frame, or two separate frames, and one or more gypsum boards on each side (see examples given in Figure 4.1).

Façades are usually composed of different elements, such as the wall itself, windows, doors and air inlets etc. The façade insulation depends on the acoustic performance of all the different elements; the way of estimating this insulation is given in Section 0 in the guide. Lightweight façade walls are, like lightweight walls, made of double leaf elements, but have additional layers such as profiled sheeting or external finish.

131

Figure 7.12 Examples of double leaf lightweight walls a) with single frame; b) with separate sets of studs; c) façade wall with single frame and external finish (plastering)

The acoustic behaviour of walls and façade walls is evaluated in terms of airborne sound and is characterized by the Sound Reduction Index R of the wall; this quantity can be measured in the laboratory according to standard ISO 140, Part 3. Note that, in buildings, the direct transmission through separating walls is not the only path from one room to another; surrounding walls also participate in flanking transmission; parameters influencing flanking transmissions are considered in Section 7.3.2.

Sound propagates through a double leaf wall mainly through 2 paths: (i) airborne sound transmission through the air cavity, and (ii) structure borne sound transmission through the frame.

Airborne sound transmission through the air cavity Airborne sound transmission through the air cavity is known; Figure 7.13 shows a typical transmission loss spectrum (TL) of a perfect double wall with no mechanical ties between the leaves. The frequency range can be divided into 3 domains, as shown in the figure:

• domain (1): the wall has a single wall behaviour where both leaves vibrate in phase with the same amplitude, which follows the mass law (doted line in the figure);

• domain (2): the wall has a mass-spring-mass resonance behaviour where both leaves vibrate out of phase and with high amplitude, which leads to a low TL, lower than the mass law; the resonance frequency is usually located between 50 and 100 Hz;

• domain (3): the wall has a double wall behaviour where the two leaves are decoupled, leading to a TL much higher than the mass law. In this frequency range, the TL is often lowered by the ‘coincidence phenomenon’, which corresponds to an equality between the sound wave length and the leaf vibration wave length, and leads to a high radiation efficiency of the leaves.

a) b) c)

132

Figure 7.13 Typical transmission loss frequency spectrum for a double wall

The corresponding major parameters influencing the acoustic performance are therefore the mass per unit area of each leaf, the distance between leaves and the properties of absorption material inside the cavity (thickness, density and flow resistance). The resonance frequency should be as low as possible, which requires sufficient mass for each leaf and sufficient distance between leaves. Absorption material inside the cavity decreases the airborne sound transmission between the leaves.

Structure borne sound transmission through the frame Structure borne sound transmission through the frame is more complicated and not as well understood; the research work performed during the Acousvibra project has helped clarify and evaluate these structural paths.

Lightweight double walls with two separate frames - the only structural path between the leaves is at the wall boundaries; through boundary rails or boundary studs connected to any surrounding structures. This path becomes dominant at high frequencies, which reduces the index R of the wall. Studies of double walls with separate sets of studs and one gypsum board on each side show that:

• studs increase the transmission through the cavity at mid and high frequencies (compared to boards with no studs); the key parameter is then the bending stiffness of the studs (the lower the bending stiffness, the better the insulation);

• the flanking paths through the wall boundaries become important at high frequencies. A dip is experienced typically at around 2000 Hz, and corresponds to the coincidence frequency mentioned above.

For single frame double walls, there are structural paths through studs and, as before, through boundary rails and boundary studs. This case has been well studied during the Acousvibra project, and laboratory characterization methods for studs and rails have been developed, as well as models to evaluate the corresponding structural transmission. The key parameter is the stud and rail section profile stiffness: the lower the stiffness, the better the insulation. Studies of typical single frame double walls show that:

• the structural path through the wall frame is dominant (compared the path through cavity) at the middle and high frequencies and

• paths through studs and through wall boundaries are balanced (similar transmission loss).

Acoustic wall studs (AWS) can be developed with low section profile stiffness in order to decrease structural transmission, as discussed above, but, when acoustic wall studs are used, models indicate that the structural paths through boundary rails and boundary studs (which are stiffer because of connections to surrounding structures) become dominant.

Frequency

(1) (2)

(3)

Resonance frequency

Coincidence frequency

(1) single wall behaviour

(2) mass-spring-mass resonance behaviour

(3) double wall behaviour (leaves decoupled)

Transmission Loss

133

Figure 7.14, below, summarizes the major parameters influencing sound insulation for lightweight steel walls.

Figure 7.14 Major parameters influencing sound insulation for lightweight steel walls

The major parameters influencing sound insulation properties of lightweight steel walls are the:

• material properties of the leaves, particularly their mass per unit area and number of layers

• distance between the leaves

• thickness, density and flow resistance of absorbent material in the cavity

• bending stiffness of the studs in the case of double walls with separate sets of studs

• section stiffness of studs in the case of single frame double walls

• section stiffness of boundary rails or studs

Floors The low weight, structure type and relatively short span of these flooring types, categorise the floor as a high-frequency floor, with a fundamental frequency above 10 Hz. For economy and flexibility in use, it is often important to be able to create as long a floor span as possible. It is not normally a problem to make the double-leaf lightweight steel floor strong enough to carry the loads for typical spans of 4 to 6 m, and deformations can be limited. It is possible to span longer distances, e.g., by utilising composite action between the steel joist and top of the flooring (concrete or boards), or by using a slightly stronger and heavier system like any semi-light system, as described earlier. One challenge is to avoid unpleasant vibrations from dynamic loads. Usually, the vibration criteria determine the maximum span of a lightweight floor structure.

The acoustic behaviour of lightweight floors is evaluated in terms of both airborne and impact sound. The floor is then characterized by two quantities - the Sound Reduction Index R and the normalized impact level Ln - which can be measured in laboratory according to ISO 140, Part 3 and Part 11 respectively. Note that, in buildings, the direct transmission through floors is not the only path from one room to another, surrounding walls also participate in flanking transmission; parameters influencing flanking transmissions are considered in section 0 in the guide.

Sound propagates through a double-leaf floor construction in two main ways: (i) airborne sound transmission through the air cavity inside the element and (ii) structure-borne sound transmission through the bridges (connected components) between the top floor surface and the ceiling. Moreover, both airborne and impact sound insulation must be considered.

Section stiffness of studs (single frame)

Distance (number of studs)

Bending stiffness of studs (separate frames)

Distance of studs (same leaf)

Mass, number of layers

Thickness, density, flow resistance

Distance

Section stiffness of rails

134

Airborne sound insulation Any basic lightweight floor is made of the floor structure and coverings, and a ceiling, and therefore has the acoustic behaviour of a double wall; the main parameters are therefore the same as those presented above for walls. In particular, the resonance effects between the top and bottom layers are present at frequencies of about 60-100 Hz. To achieve good properties, the main cavity between the flooring layers has to be insulated with mineral wool, where absorption takes place. It is also important that the structure-borne transmissions through the connections between floor and ceiling can be attenuated; one efficient solution consists of using a resilient ceiling attached by flexible fixings, which stop vibrations from spreading to the ceiling. For frequencies down to about 100 Hz, lightweight floors normally perform well.

Impact sound insulation It is very efficient to reduce impact sound by applying an elastic layer between the surface board and the floor structure. This layer is independent of the floor structure and is much softer than the surface board. The impact sound insulation is more of a problem in the low-frequency area, and invariably needs to be improved. Experience shows that with sufficient impact sound insulation an adequate airborne sound insulation is reached automatically. When designing light-gauge floors for adequate acoustic performance, the improvement to the impact sound insulation should be a prime consideration.

Figure 7.15 Major parameters influencing airborne and impact sound insulation for lightweight steel floors

The major parameters influencing airborne and impact sound insulation properties of lightweight steel floors are shown in Figure 7.15, and can be summarised as the:

• material properties of the floor, the top surface and the suspended ceiling, in particular, the mass per unit area of the individual layers

• material properties and thickness of the impact sound insulation layer, in particular, its dynamic stiffness

• distance between top and bottom layers, i.e., the major part of the construction depth

• stiffness of the connection between the suspended ceiling and the supporting beams of the floor

• method chosen for providing sound absorption in the flooring cavity

• sound insulation of the heavy top layers, such as concrete and plasterboards.

7.3.2 Whole building level (including junctions) Three types of building acoustic performances are considered in this section in the design guide: airborne and impact sound insulation between rooms, as well as the airborne sound insulation against outdoor noise.

135

Airborne sound insulation against outdoor noise The airborne sound insulation against outdoor noise can be measured from the sound level outside the building (at 2 m in front of the façade) and the sound level inside the receiving room, according to ISO 140-3; results are expressed in single number values defined in ISO 717-1. The façade insulation mainly depends on the acoustic performance of the different elements comprising the building façade: façade wall, window, air inlet, rolling shutters etc. These elements are characterized by their laboratory performance, i.e., the Sound Reduction Index R for elements such as walls, windows etc., and acoustic insulation Dne for smaller elements such as air inlet, rolling shutters etc. High façade insulation is obtained by choosing elements having either a high index R or a high insulation Dne .

European Standard EN 12354-3 defines how to estimate the building façade insulation from the laboratory performance of the different elements comprising the façade and their geometrical size. This can be used at the design stage. Optimal and cost-effective solutions are obtained by choosing balanced façade elements, whereby each element contributes equally to the sound insulation. Each contribution can be estimated the total normalized façade insulation calculated. Two examples of this calculation are given in the guide.

Experience shows that predicted results obtained using EN 12354-3 are often too optimistic (compared to measurements) in the case of lightweight construction and a safety margin has to be defined. This should take into account the uncertainty in the prediction, the uncertainty in the measurements, and the decreasing performance of building elements and their connections with age.

Airborne and impact sound insulation between rooms Airborne and impact sound transmissions Airborne and impact sound transmissions in buildings include both direct transmission through the separating element (wall or floor) and flanking transmissions resulting from vibration transmissions through junctions between elements in the emission room and elements in the receiving room.

Figure 7.16 shows an example of vertical acoustic transmissions, and illustrates schematically the direct path through the floor and ceiling, and one of the flanking paths through the junction between the floor and wall. Note that the whole separating elements (floor + ceiling or double wall) participate in the direct transmission, and only parts of the separating elements (top floor or ½ separating walls) participate in flanking. As a result, when a given airborne or impact sound insulation between rooms has to be reached, flanking transmissions must be taken into account and, if possible, reduced to a minimum so that the resulting insulation is as close as possible to the insulation of the separating element alone.

1/2 separating wall participating to flanking path ij

Top floor participating to flanking path ij

j direct path

i

Figure 7.16 An example of vertical acoustic transmissions

Figure 7.16 shows that one way of reducing flanking transmissions consists of attenuating vibration transmission at junctions between elements. This can be done by mechanically disconnecting elements as much as possible and/or inserting resilient layers, in order to break any structural continuity. 3D

136

modular buildings, where 3D modules can be placed on top of one another with no line connections between elements (walls and floors), can be a good solution; flanking is then reduced to a minimum, but still exists through point connections between modules.

Multi-storey buildings with a primary separate beam-column load bearing frame have some inherent freedom to improve floor/wall junctions, since the walls themselves are not load bearing; some examples are given in one of the case studies presented in the guide, where measured values of junction attenuation are also quoted.

However, one or two storey buildings with load bearing walls where floor/wall junctions participate in the building structural stability cannot be treated easily at the junction; in this case, the only way to reduce flanking consists of fixing linings on the inner leaves of the wall involved in the flanking path.

Junctions between lightweight elements Some examples of junctions between lightweight elements are given in the guide, and recommended and undesirable types are illustrated. Further guidance is given on the application of EN 12354, which is the European standard for predicting the airborne and impact sound insulations in buildings from the performances of building elements.

7.3.3 Case studies Two case studies are presented in the guide; a lightweight building and a heavy building with a lightweight façade. The buildings have been modelled in detail, and the transmission of sound predicted.

7.4 Modelling methods Sound transmission through double leaf walls or floors can be separated into different transmission paths: the airborne path through the air cavity and structural paths through the inner frame and boundary frame. Three different approaches can be used for modelling these different paths: the wave approach, the energy approach (SEA) and the finite element approach, including the Finite Element Method (FEM) and the Boundary Element Method (BEM). The following sections briefly present the principles of these approaches, as well as their modelling capabilities and limitations when applied to lightweight construction, and further information is presented in the design guide.

7.4.1 Wave approach combined with SEA Basic principles Wave approach The wave approach is well known for its application to sound transmission through infinitely thin plates in flexure, or through infinite double plate systems with an air cavity in between54. The sound power transmitted is calculated in 3D, wave by wave, using the classical wave equation for thin plates in flexure. Figure 4.4 shows this diagrammatically, whereby the wall is being excited by a plane wave at different angles of incidence; a diffuse field transmission loss is then obtained by adding the sound powers incident or transmitted at different angles. This approach is valid over the whole frequency range and the computation time is quite short (compared to FEM for example). However, real lightweight steel frame walls or floors are more complicated than infinite thin plates for the following reasons:

• plates are stiffened by studs or joists

• the air cavity is not the only path; structural paths through inner frame and/or boundary frame must be taken into account

• the size of real walls is not infinite, and

• in the case of floors, both airborne and impact sound insulations must be considered.

As a consequence of this, more sophisticated tools (but still based on the wave approach) have been developed to treat lightweight structures; details on these tools are given further in this section.

137

Figure 7.17 Wave approach basic configuration

SEA approach SEA (Statistical Energy Analysis) is an energy based model55 where sound transmission is expressed on a power basis. Each sub-system (i.e., volumes, plates etc) is supposed to have many resonant modes in each frequency band considered (i.e., acoustic modes in volumes and air cavities, and vibration modes in components such as plates or beams); this assumption limits the practical frequency range where SEA can be used to mid and high frequencies. Power flows between sub-systems (such as cavity or plate) are then calculated from the energy stored in each sub-system and from parameters known as coupling loss factors. Figure 7.18 gives an example of a SEA power flow diagram in the case of a double wall, and indicates all the sub-systems considered. All the SEA assumptions, as well as the way to estimate the different SEA parameters from the dynamic properties of the sub-systems, can be found in existing literature. The application of SEA to lightweight structures is discussed later in this section.

Figure 7.18 SEA transmission paths between sub-systems; the case of a double wall

Application of the wave approach to lightweight structures In the CSTB wave models of single wall or double wall with separate frame, studs or joists are modelled as periodically-spaced beams line-connected to thin plates. The presence of beams is expressed as reaction forces and moments applied to the plates, and these forces and moments are calculated from the flexural and torsional line impedance of the beams. The air cavity, with or without absorbent material, is modelled as an equivalent fluid. The finite size of the system is taken into account using a spatial windowing technique applied to the infinite system56; this technique introduces the diffraction effect associated with the finite size of a structure using spatial filtering. Comparisons

z

Infinite wall

Incident plane wave

θ

x

138

between calculated and measured results show that the wave approach is a good way for modelling the path through the air cavity57.

The structural paths through the inner frame and/or the boundary frame are usually modelled using SEA and not the wave approach, since these paths are important at mid and high frequencies, where SEA is valid and very simple to apply. However, in the case of single frame lightweight double walls, the structural paths through the inner studs have an influence on the sound transmission at low frequencies, where SEA is not valid, and must then be modelled using the wave approach. In this case, the connection through the studs is modelled as periodically-spaced line springs coupling the two plates. The spring stiffness can be estimated experimentally using the characterization method for studs given below.

Application of SEA to lightweight structures It is important to note that comparisons between calculated and measured results show that SEA does not give accurate estimations of the double wall transmission loss at resonance frequency. The transmission path through the cavity is overestimated, mainly because:

• SEA does not model thin air cavities well and

• double wall resonances occur at low frequencies - where SEA is theoretically not valid.

However, in the case of single frame lightweight walls, SEA is well adapted to model the structural paths between the two leaves through inner studs at mid and high frequencies: studs are then modeled as springs point-connected between plates at screw locations. The corresponding SEA coupling parameter mainly depends on (i) the stud section profile stiffness, which can be experimentally estimated using the set-up shown in Figure 7.19, and (ii) the input point impedance of the plates, which can be estimated from theoretical impedances of infinite plates. In the experimental setup, the stud stiffness is obtained from input mobility measurements (ratio between velocity and force applied), using a force sensor and an accelerometer, in the low frequency range (around 10 Hz) before modal behaviour.

Figure 7.19 Experimental set-up for stud characterization

SEA is also well adapted to model the structural paths through the wall boundaries, and this path exists in both cases of double walls with single frame, and double walls with separate frames. A SEA coupling parameter between the leaves can be estimated as before, but differently: (i) the input point-impedance of the plates is estimated from theoretical impedances of semi-infinite plates (instead of infinite plates), and (ii) rails and boundary studs are connected to surrounding structures and their apparent stiffness can not be characterized the same way as for inner studs; another experimental set-up, shown in Figure 7.20, is then used. In the test, one of the plates is mechanically excited and the velocity level difference (VLD) between plates is measured and compared in the mid –high frequency range to a SEA predicted VLD in order to deduce the correct boundary frame stiffness.

Force Excitation(hammer)

Rigid Floor

139

Figure 7.20 Experimental setup for boundary frame characterization

Modelling of impact sound transmission In this case, the light weight floor is excited by the tapping machine, as shown in Figure 7.21. The excitation force can be represented by normal stress plane waves using spatial Fourier transforms, and the wave approach can then be used in the same way as for airborne excitation to calculate the sound power radiated and the impact sound level. The difficulty lies in the estimation of the force applied by the tapping machine, taking into account not only the hammer characteristics (mass and impact velocity) but also the floor characteristics (global thin plate point mobility, which depends on the location and local point mobility associated with local deformations due to the hammer impact). Impact sound transmission models are currently being researched and are under development.

Figure 7.21 Impact noise of stiffened lightweight floors

7.4.2 Finite element method Introduction The finite element method (FEM) was devised in the 1950s as an advanced method of structural analysis, originally for aeronautical engineering applications and soon extended to civil and mechanical engineering58.

The scope and range of applicability of FEM has grown dramatically since then. Today, FEM is the most widely used numerical method to solve partial differential equations (PDE) of any type, which are ubiquitous in virtually all branches of engineering59. Heat transfer, compressible or incompressible flow, pollutant dispersion and wave propagation, to name just a representative few, can be modelled with FEM. The finite difference method has a similar broad scope60.However, it is generally accepted that FEM is more powerful than finite differences, especially in handling complex geometries: finite differences require a regular grid, whereas unstructured meshes are handled very naturally by FEM.

FEM can also be used in building vibro-acoustics. The main application of interest is the sound transmission through (typically multi-layered) walls and floors. This is a coupled problem that involves, firstly, sound propagation through air (sending and receiving room, air cavities inside the structure) or

Force Excitation(hammer & Rain on the roof)

Rigid floor

Rigid ceiling

Tapping machine

Lightweight panel

Periodically spaced stiffeners

(joists)

140

through acoustically absorbent materials, and, secondly, vibration transmission through structural parts (boards and studs), as shown in Figure 7.22.

]

Figure 7.22 The coupled vibro-acoustic problem

The propagation of sound through air is modelled by the wave equation60,61, a PDE that involves space and time derivatives. The unknown is the field of acoustic pressure p(x,t), which is a function of space and time.

Like any other evolutive PDE, the wave equation can be solved by combining finite elements for spatial discretization and a numerical marching scheme to advance the solution in time. However, the usual approach is to assume a steady-harmonic solution (i.e., a harmonic variation in time with respect to a steady reference state). With this assumption, the wave equation (in the time domain) is transformed into the Helmholtz equation (in the frequency domain) 61,62.

The vibration transmission through the structural parts is modelled by the usual equation of structural dynamics, involving the mass, stiffness and the damping of the structure63. The steady-harmonic assumption is also made for the displacements of the structure. Hence, the coupled vibro-acoustic problem is solved in the frequency domain, to obtain the acoustic pressure in the air and absorbent material, and the displacements in the structure62,64. These fields can then be post-processed into the relevant output, such as the sound reduction index R. Apart from the coupled vibro-acoustic problem, the purely mechanical problem is also relevant. Solving the equation of structural dynamics for a double wall, for instance, yields the vibration transmission loss Dij between the two boards.

Specific issues of FEM modelling in building vibro-acoustics From the viewpoint of the user, a finite element analysis consists of four major phases: input of data, finite element mesh generation, computation and post-process. Figure 7.23 details these four steps in the case of sound propagation through a wall.

141

Figure 7.23 Phases in finite element analysis of sound transmission

Further information is given in the guide to emphasize the main principals of FE modelling in building vibro-acoustics (as compared, for instance, with FE structural analysis). Guidelines are given on mesh generation and computation, and the capabilities and limitations of FEM explained and illustrated with some examples.

7.5 Design Guide Chapter 4: Vibrations 7.5.1 Introduction

Design methodology is presented here for analysing and modelling floors. Information on harmonisation of the testing procedure and the rating of the annoyance of vibrations is given in Chapter 3.4 and 3.5 of this report.

Design recommendations are given below for lightweight floors in residential (or office) buildings when the fundamental frequency of the floor is more than 10 Hz. They are based on walking-induced vibrations caused by one person. The recommendations do not apply to other applications which have different loading conditions or demands, or to rooms in which the vibrations are machine-induced. Low fundamental frequencies are typical for heavy floors, while high frequencies are typical for lightweight floors. The design of floors with a low or high fundamental frequency is different.

Humans are very sensitive to vibrations, particularly of low frequencies. Floor vibrations transmitted to furniture, glassware and pot plants can also cause a disturbing noise. Ordinary walking consists of both harmonic and impact components. The lowest harmonic component is the step frequency, which is, typically, 1.6-2.2 Hz, but multiple frequencies of 3.2-8.8 Hz are also present. Annoying vibrations often result if the cyclic components are magnified by resonance (i.e. when the load frequency matches the natural frequency of the floor) or a heel impact causes unduly large floor response. It may also be annoying if the floor beams deflect too much, or the floor surface bends too much, under the footsteps.

Resonance governs if the fundamental frequency of the floor is less than 10 Hz. If the fundamental frequency is higher than this, the deflection of the floor beams governs. The vibrations of objects on

Input of geometrical and material data

Dimensions, mechanical (mass, stiffness, damping) and acoustic (absorption, resistivity) properties

Geometrical model and mesh generation

Geometrical model of acoustic domains and structure discretized in finite elements

Computation

Solve system of linear equations for each frequency, to obtain acoustic pressure and displacements

Post-process

Compute output of engineering interest (e.g. Sound Reduction Index) from FE solution

142

shelves are mainly induced by slanting of the shelf due to the local bending of the floor surface. It is well-established design rule that light-weight floors having a fundamental natural frequency lower than about 8-10 Hz should be avoided. The most efficient way to increase the fundamental natural frequency is to increase the stiffness of the floor. An effective way is to increase the stiffness of the floor in the transverse direction, which is normally much lower than the stiffness in the main direction. However, the increased stiffness in the transverse direction will not have a significant affect on the fundamental natural frequency, but it will distribute loads to adjacent beams, so reducing the deflection under the footstep.

The design procedures for assessing the dynamic performance of floors comprise:

• Determination of natural frequency

• Definition of modal mass for the floor

• Evaluation of response

• Checking of responses against acceptance criteria

Detailed guidance on the procedures for lightweight floors is given below. Two alternative dynamic design models are presented below; one based on the calculation and assessment of deflections, and the other on the calculation and assessment of accelerations. Guidance is also given on finite element modelling.

7.5.2 Deflection-Based Model Determination of fundamental frequency The fundamental frequency of floor joists, when the floor has simply supported edges, is calculated from

l

blEIEI

bl

bl

mEI

lf

)()(

21)(

2

42

20⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎠⎞

⎜⎝⎛+⎟

⎠⎞

⎜⎝⎛+⋅=

π (7.1)

where l is the span of joists, (EI)l is the stiffness of floor joists and slab per unit width, (EI)b is the stiffness of floor slab per unit width in a perpendicular direction to the joists, b is the floor width and m is mass density by area of floor, increased by service load 30 kg/m2.

In many cases the edge condition parallel to the floor joist may be neglected, and the frequency may be approximated by

mEI

lf l)(

2 20π

= , (7.2)

The underestimation in frequency is less than 5% when b/l >1.0 and (EI)l /(EI)b>30, but if b/l = 0.5, the same accuracy is achieved only when (EI)l /(EI)b>200.

Determination of global deflection The global deflection due to a point load for a floor may be based on the deflection of an orthotropic plate with all edges simply supported. The maximum deflection due to the point load F = 1 kN is in the middle of the floor, and may be approximated by

143

lEIFl

)(

20 ⋅= γδ , where (7.3)

l

b

i j EIEI

lb

ji)()( ja ;

12 )12(

144

44 ==

⎟⎠⎞

⎜⎝⎛ −⋅

+−⋅⋅

= ∑∑ βα

αβ

παγ (7.4)

In many cases, the edge condition parallel to the floor joists may be neglected, and γ may be approximated by

4/1

)()(

42

1

⎥⎦

⎤⎢⎣

⎡⋅

=

l

bEIEI

γ (7.5)

The error in estimation by comparison with theoretical deflections is less than 2.5 %, when b/l >1.0 and (EI)l /(EI)b>20, but, if b/l = 0.5, the same accuracy is achieved only when (EI)l /(EI)b>300. If the deflection δ0 approximated by the equation for orthotropic plate gives higher values than

lEIsFl

)(48

3max ⋅⋅

=δ , (7.6)

which is the deflection of a single floor beam with (spacing s) due to point load F = 1 kN, δ0 is taken as δmax.

The calculated deflection is then compared with the acceptance criteria, as explained in Chapter 5.6.

7.5.3 Level 2: Acceleration-based method

Natural Frequency For free elastic vibration of a beam, or uniform section, the natural frequency is given by:

42 mLEIk

f nn π

= (7.7)

Where EI is dynamic flexural rigidity of the member (Nm²), m is the effective mass (kg/m), L is the span of the member (m), and kn is a constant representing the beam support and/or loading conditions.

Some standard values of kn for the first mode of vibration of elements with different boundary conditions are as follows:

pinned/pinned (‘simply-supported’)

fixed/pinned (propped cantilever)

fixed both ends (encastré)

fixed/free (cantilever)

π²

15.4

22.4

3.52

144

A convenient method of determining the natural frequency of a beam f, is given by first finding the maximum deflection δ (in millimetres) caused by the weight of a mass m. For a simply-supported element subjected to a uniformly distributed load (kn = π²), this is of course:

EImgL

3845 4

=δ (7.8)

where g is the acceleration due to gravity (i.e., 9.81 m/s2).

Rearranging this equation, and substituting in the value of m and EI/L4 into the equation for fo, te flowing simple formula is established:

δ≈

δ=

18817.f (7.9)

For beams with different loading types and/or boundary conditions, similar results can be found with the numerator varying between 16 and 20. However, for practical design, a value of 18 will normally produce results of sufficient accuracy.

Pace frequency Vibration assessment should be carried out for a continuum of pace frequencies within the range appropriate to the walking activities for the relevant location. For design, the assessment of vibration should determine the worst response within the following pace frequency ranges:

Walking activities in a corridor/walkway area = 1.8 – 2.2 Hz

Walking activities in a room (L ≤ 10m) = 1.8 – 2.0 Hz

Critical working areas, although a special case for vibration response, are classified for pacing frequency as a room. In the above ranges L refers to the dimension of the walking path for the activity. Where it is possible to show that for a given corridor the dimension of the free walking path will never be more than 10 m, the second frequency range may be used.

The above design pace frequencies have been derived from measurements made on an office building in Delft51, which showed that the distribution of walking frequencies were lognormal, with a mean frequency of 2.0 Hz and a coefficient of variation of 8.5%. The design values for a corridor/walkway area given above have been evaluated according to EN 1990, Annex C53 with a target reliability index β = 1.5 and a First Order Reliability Method (FORM) sensitivity factor of αE = 0.7.

The above pacing frequencies also correspond to the categories described by Wheeler52 as a ‘normal-walk’ for up to 2.2 Hz and a ‘slow-walk’ for up to 2.0 Hz. The length chosen for the transition from slow walk to normal walking in rooms of 10 m represents the number of steps required for the floor response to reach a steady state. Generally at 2.0 Hz this takes approximately 4-5 steps.

Floor Response If the fundamental frequency is greater than 10Hz, the floor will not be subject to resonant response due to the first four harmonics of walking activity. The floor will therefore exhibit a transient response to the impulsive loading of footfalls. The following expression should be used to estimate the rms acceleration:

2154

2 3.1oeffeff

43.1p

reorms fWmL

ffa μμπ= (7.10)

Where:

fo is the fundamental frequency of the floor (Hz)

μe is the mode shape at the point of excitation, normalised to the anti-node

145

μr is the mode shape at the point of response, normalised to the anti-node

fp is the pacing frequency of the walking activity considered (Hz)

m is the distributed mass (kg/m2) comprising the self weight, the superimposed dead load and, a proportion of the imposed live load (considered as permanent) or less if a more accurate estimate can be made of the semi-permanent live load. It is recommended that an upper limit of 10% of imposed loading should be considered for dynamic design.

Leff is the effective floor length (m), calculated as below.

Weff is the effective floor width (m), calculated as below.

Modal mass The modal mass is calculated from the effective floor area that participates in the response of the floor, which is determined from the calculation of the effective floor length and width. The true floor area that participates in the motion will commonly not be rectilinear due to the complex interaction of stiffness in the orthogonal directions. However, a representative floor area may be estimated from these effective perpendicular values.

Effective floor length: The effective floor length is defined in the direction of span of the main floor support members. It is partly dependant on the stiffness of the support members and the number of consecutive spans. The flowing equation may be used to estimate the effective length:

[ ] 6xx2

eff 103.55.71.22.0 −×

×+−=I

LLnL y (7.11)

Where:

ny is the number of consecutive floor spans (ny<5)

L is the supporting member span (m)

Ixx is the composite moment of inertia of the main support member (m4/m)

Effective floor width: The effective floor width is limited to the actual floor width but may be greater than the width of a single bay where multiple bays are used. It is further dependant on the composite inertia of the main support member and the transverse stiffness (defined by the spacing of members). The effective floor width may be estimated from the following equation:

Weff = nWSI

W xx ≤−+×

×+ − )6.0(8.503.5

)1(75.0 6 (7.12)

Where:

W is the overall width of the floor bay considered (m)

Ixx is the composite moment of inertia of the main support member (m4/m)

S is the spacing of floor support members (m)

n is the number of consecutive floor widths considered (m)

Notes: 1. The effective floor width is normalised to 3m, this assumes that bays will always be 3 m or

greater in width. Where the bay width that is under consideration is less than this, the Weff is limited to 3m; this is independent of the number of bays considered overall.

2. The effective floor length is limited to the total length of the floor. The effective floor length is further limited to 4 consecutive spans due to inability to transmit the impulsive force over a greater length because of the stiffness of the main support members.

146

3. Increasing the width of the bay by 1 m will increase the modal mass by approximately 25% overall.

4. A conservative estimate for the ue and ur factors in the calculation of Leff is unity. This will effectively place the response at the antinode of the floor with the excitation force co-incidental.

5. These formulas are appropriate for floors with regular bays. Where the widths of bays are variable a reduction in the Weff may be made by a linear adjustment of (1 – Ws/Wm) to the calculated value of Weff, where: Ws is the width of the smaller bay; and Wm is the width of the main bay considered.

Mode shape factors For a single floor panel the fundamental mode shape will give the most onerous response. An example of a typical fundamental mode shape profile for a simple floor with uniformly spaced joists is shown in Figure 7.24. The shape is affected by the flexibility of the joists, and may be generalised as a single half-sine wave as shown in Figure 7.24. Using this generalisation it is possible to estimate conservative mode shape factors for the points of excitation and response. If the maximum point of displacement is normalised to unity (i.e. the anti-node) the relative displacement and, hence, the mode shape factor, μe or μr, can be calculated from:

⎟⎠⎞

⎜⎝⎛=

Wxor re

180sinμμ (7.13)

Where:

x is the distance of the point considered from the centre of the individual floor panel

w is the total width of the individual floor panel

A

A

A

A

Mainbeams

Floorbeams

(a)

(b)

Floor & beaminteraction

Idealisedinteraction

Figure 7.24 Possible floor and beam mode interactions

The calculated acceleration is then compared with the acceptance criteria, as explained in Chapter 5.6.

147

7.5.4 Finite Element Modelling of Floor Vibrations The design guide presents a method of response analysis developed by the SCI65 that is appropriate for any floor, but assumes the use of finite element analysis to determine the natural frequencies, modal masses and mode shapes of the floor under consideration. These techniques are particularly useful for structures which are complex or have very stringent requirements with regard to vibration.

The use of finite element (FE) methods is described along with suitable design methods which may be used to assess floor response. Finite element modelling is useful to establish a reasonably accurate prediction of the nature of the floor or the whole-building structure and, in all but the simplest of structure, will give a better prediction than that given by hand calculation methods.

The dynamic performance can be established through finite element modelling of the floor or the whole structure, or similar numerical modelling techniques. FE is an approximation: it takes a continuous structure and breaks each part of the structure into a number of parts, or a finite number of elements. The relationships between these elements are then determined using methods for multi-degree-of-freedom discrete systems. The accuracy of the solution is always dependant on the number of elements into which the system is split, but with increased accuracy comes increased complexity and hence higher computation times.

Implementation suggestions From comparisons with measurements made on a wide variety of composite floor types, it is recommended that the following parameters and modelling details should be adopted. Any improvements over these recommendations will obviously lead to a greater accuracy:

• All connections should be assumed to be rigid (although joints are designed at ULS to be pinned, in vibration the strains are not large enough to overcome the friction and so pinned joints may be treated as fixed)

• Column sections should be provided and pinned at their theoretical inflexion points (typically located at mid-height between floors for multi-storey construction).

• Continuous cladding provided around façades may be assumed to provide full vertical restraint to perimeter beams. The edges of clad buildings should therefore be modelled as free in rotation but restrained in direction for all three directions of freedom (i.e. pinned).

• The mass of the floor should be equivalent to the self-weight and other permanent loads, plus all imposed loads which might be reasonably expected to be permanent.

• Movement joints may be considered to be rotationally free, though fixed in location. For greater accuracy, the exact transfer of stiffness through the joint may be allowed for by consideration of the deflected form. However, as the stiffness transfer is small, it is often inefficient to allow for such detail.

One of the most difficult properties to estimate is the level of damping that is present on the floor, owing to the fact that it is strongly influenced by finishes and non-structural components. Unless better information is available, it is recommended that the values for damping given in Section 4.2.2 of the guide should be adopted.

There are no hard and fast rules for the size of the elements (or mesh), but, in general, if the number of elements can be doubled without significantly changing the frequencies, then there are sufficient elements.

Modal mass The required outputs from the finite element analysis are the modal frequencies, the mode shapes and the modal masses. The mode shape can generally be output in two forms; mass-normalised and unity-normalised.

With mass-normalised mode shapes, the output displacements are determined so that the modal mass, Mm, is 1kg. This combination of mode shape and modal mass can be used in the equations given below, but it does not give any indication about the effect of each mode on the overall response.

148

With unity-normalised mode shapes, the maximum displacement is set to unity for each mode, but as a non-dimensional value; other displacements for a given mode shape are then relative (values) to this. To calculate the modal mass corresponding to a unity-normalised mode shape, the designer will need to determine the maximum kinetic energy in each mode shape (which can generally be determined by the FE software), and the relationship between this and the modal mass is given by:

222 m

mm f

KEM

π= (7.14)

where:

Mm is the modal mass for mode m (kg)

KEm is the maximum kinetic energy in mode m that corresponds to the unity normalised mode shape (kg/s², or J/m²)

fm is the frequency of mode m (Hz)

Note that some finite element packages have an output of mass participation or effective mass, which is generally not the modal mass. To check that the calculation for the modal mass is correct it is suggested that a model of a simply supported beam is considered, for which all the modal masses are theoretically half of the total mass. Note that at higher frequencies the modelling assumptions will mean that this value is not achieved.

Response analysis To find the peak response, the following analyses will need to be performed for a range of floor frequencies within the range of walking frequencies, and the maximum response taken. It should be noted that these methods assume that the force is applied at the most responsive location on the floor even though the walking path will only pass across this point briefly. However, this is a conservative assumption and an analysis based on walking paths rather than individual points may be taken; but this leads to an increased level of complexity. For low frequency floors (where the fundamental frequency is lower than the values given in Table 7.1), both the steady-state response and the transient response need to be checked, as the higher frequencies of the floor may result in the transient response being greater than the steady-state. However, as lightweight floors are high frequency floors, only the transient response needs to be checked.

Table 7.1 Low frequency floor to high frequency floor cut-off

Floor type Low to high frequency cut-off

General floors, open plan offices etc. 10Hz

Enclosed spaces, e.g. operating theatre, residential

8Hz

Transient response of floors For transient analysis, the response is dominated by a train of impulses, which correspond to the heel impacts made by the walker. In these circumstances, it is recommended that all modes with natural frequencies up to twice the fundamental (first mode) frequency should be taken into account, as above this the effects of the frequency weighting will make the results insignificant. The weighted peak acceleration response at a position j in a single mode m of frequency fm may be obtained from the following equation:

mmj,mi,

2mmj, 1π2

MIfa μμζ−= (7.15)

where:

μi,m is the mode shape amplitude, from the unity or mass normalised FE output, at the point on the floor where the impulse force I is applied

149

μi,m is the mode shape amplitude, from the unity or mass normalised FE output, at the point where the response is to be calculated

I is the excitation force (as given by Equation (7.16)) [Ns]

Mm is the modal mass of mode m (equal to 1kg if the mode shapes are mass normalised) [kg]

The equivalent impulsive force I (representing a single footfall) in Newton-seconds (Ns) that may be used in transient design is given by Equation (7.16). This equation was developed by comparing the performance of the theoretical model proposed by Young65 according to the requirements given in EN 1990 Annex C53.

70060 0

3..1

43.1 Pf

fI

m

p= (7.16)

where:

fp is the pace frequency

fm is the frequency of the mode under consideration and

P0 is the static force exerted by an ‘average person’ (normally taken as 76 kg × 9.81 m/s² = 746 N).

The acceleration to each impulse is found by summing the acceleration responses of each mode using the following superposition formula. The rms acceleration can then be determined using a time period, T, of 1/fp.

( ) tfm

m

m

m

etfM

If

tata

mπ22m

m1mj,mi,

2m

1mj,j

1π2sin1π2

)()(

ζζμμζ −

=

=

⋅−−=

=

∑

∑ (7.17)

Excitation and response positions The excitation point, i, and the response point, j, should be chosen to produce the maximum response of the floor. In most cases i and j will represent the same point (as the maximum response to any excitation is coincident with the excitation point), and so should be checked for every point defined in the finite element analysis. In practice, however, only the locations of the maximum amplitudes for each mode need to be checked, but the displacements of each point are required for every mode.

In the case of a response in a room from excitation in a corridor, i and j should be chosen to be the maximum amplitude of displacement in the respective areas for each mode. In all cases, the mode shape amplitudes can be taken conservatively as 1 when using the modal mass corresponding to unity- normalised mode shapes. Examples of selecting the mode shape factors are given in Figure 7.25, where the case corresponds to the modes for which the maximum amplitudes are chosen.

150

Table 7.2 Mode shape factors for corridor to room response as shown in Figure 7.25(b)

Amplitudes of mode 1

Amplitudes of mode 2

Amplitudes of mode 3

Case μi,1 μj,1 μi,2 μj,2 μi,3 μj,3

1 0.59 1.00 -0.95 0.00 0.95 -1.00

2 0.59 0.71 -0.95 1.00 0.95 0.71

3 0.50 0.50 -0.87 0.87 1.0 1.00

Assessment of vibration levels Once the rms acceleration has been determined, it can be weighted and compared to the acceptability criteria described in Chapter 3 in this report.

7.6 Design Guide Chapter 5: Recommendations for Design and Detailing

This section in the design guide presents a summary of the key information and the recommendations made in the main contents of the guide.

7.6.1 Acoustic performance of the structure and componenets Types of acoustic performance:

• performance of building elements such as walls and floors which contribute to the performance of the whole building, expressed in terms of sound reduction index and impact noise level, and

• performance of buildings, which depends on the performance of all the building elements (walls, floors, linings, floor covering etc.) and also the junctions between elements, expressed in terms of airborne and impact sound insulation between rooms.

Walls and façades Types of light steel walls:

• double leaf elements mounted either between apartments (separating walls) or between rooms inside the same apartment (internal partitions) with a single frame, or two separate frames, and one or more gypsum boards on each side.

i,j

i,j

i,j

j

j

j i

i

i

Room Corridor

Figure 7.25 Mode shape factors

151

Types of light steel façades:

• double leaf elements, as lightweight walls, but with additional layers such as profile sheeting or external finish.

Sound is characterized by the Sound Reduction Index R of the wall, measured in the laboratory according to standard ISO 140. Sound propagation is by:

• airborne sound transmission through the air cavity, and

• structure borne sound transmission through the frame

A reduction in the airborne sound transmission through lightweight walls can be achieved by increasing the following:

• mass per unit area and number of layers of the leaves

• distance between the leaves

• thickness, density and flow resistance of absorbent material in the cavity

and by reducing the:

• bending stiffness of the studs in the case of double walls with separate sets of studs

• section stiffness of studs in the case of single frame double walls

Acoustic wall studs, with perforated webs, are recommended.

Considerations for minimising discomfort from sound transmission of traffic noise through façades:

• location of building through planning and architectural considerations

• location of fresh air ventilators – can be impossible in extreme cases, such as airports

• window areas and careful location of windows

• special (expensive) building elements are required if sound insulation required > 45dB

• balancing the performance of the window element with the structural element making up the façade unit

Examples of lightweight steel wall systems and their measured acoustic performance are given in Table 3.3 and Table 3.6. Examples of light steel roofs are presented in Table 3.4.

Floors Type of floor:

• double-layer flooring system consisting of a top floor system and a bottom system, which are separated. The top system has an elastic layer between the surface board and the floor structure, and the bottom system acts as a suspended ceiling. The cavity in between these layers includes the load bearing steel profiles, services and mineral wool for acoustic and fire insulation.

Types of acoustic sound transmission:

• airborne, characterized by the Sound Reduction Index R, and

• impact, characterized by the normalized impact level Ln

Note that the direct transmission through floors is not the only path from one room to another; surrounding walls also participate in flanking transmission.

The major parameters influencing airborne and impact sound insulation properties of lightweight steel floors are:

• material properties of the floor, the top surface and the suspended ceiling, in particular, the mass per unit area of the individual layers

152

• material properties and thickness of the impact sound insulation layer, in particular, its dynamic stiffness

• distance between top and bottom layers, i.e., the major part of the construction depth

• stiffness of the connection between the suspended ceiling and the supporting beams of the floor

• method chosen for providing sound absorption in the flooring cavity

• sound insulation of the heavy top layers, such as concrete and plasterboards

Good sound insulation for light steel floors is achieved mainly by the reduction of impact sound level, and the following measures relating to the various components are recommended:

Suspended ceiling: • attach the suspended ceiling by transverse elements, and use multi-folded profiles or spring elements

• use gypsum boards or chipboard with a thickness up to 20 mm

• use boards with a high mass

• use a suspended ceiling with two layers (double planking) for better performance

• avoid acoustic losses through joints in the boards by careful detailing and construction supervision

• use an elastic connection of the suspended ceiling to adjoining walls

For a ceiling which is attached elastically by spring elements, the sound is transmitted mainly via the floor cavity, and the influence of the beams is small.

Floor cavity: • use fibrous insulation material with a high flow resistance (r ≥ 5 kPa×s/m²), or a high density

• use an insulation layer ≥ 100 mm thick, with a depth of up to 80-90 % of the height of the floor cavity

Upper floor layer: • use gypsum board or chip boards up to 20mm thick, with a thickness of steel ≤ 2mm

• use planking boards with high mass, and an additional covering

• use an elastic intermediate layer between the planking and the floor beams (unless there is an additional floor covering)

Structure of floor covering: • use insulation material with a low dynamic stiffness (s' < 30 MN/m³)

• use insulation material with a high dynamic stiffness (s' ≥ 16 MN/m³) in dry plaster floors

• consider improving the impact sound reduction in dry plaster floors by additional loading of the floor or by increasing the floor weight

Floor covering: Carpets improve the impact sound insulation of floors in the upper frequencies. At low frequencies, the effect is relatively small. Ceramics and tiles are rigidly fixed to the floor and lead to a slight improvement of the sound insulation at low frequencies due to the mass increase, but at high frequencies the sound insulation worsens severely by the increased stiffness and the higher sound excitation.

Examples of lightweight steel floor systems and their measured acoustic performance are given in Table 3.5.

153

Flanking sound transmission between rooms and junctions: Attenuate vibration transmission at junctions between elements by.

• mechanically disconnecting elements as much as possible and/or

• inserting resilient layers, in order to break any structural continuity.

• avoiding line connections between elements (walls and floors).

Multi-storey buildings with a primary separate beam-column load bearing frame have some inherent freedom to improve floor/wall junctions, since the walls themselves are not load bearing, but one or two storey buildings with load bearing walls where floor/wall junctions participate in the building structural stability cannot be treated easily at the junction. In this case, the only way to reduce flanking consists of fixing linings on the inner leaves of the wall involved in the flanking path. Some examples of junctions between lightweight elements are given in Figure 7.26. Junction types L1 and L2 should be avoided because of purlings crossing the junction: this provides structural continuity, but allows undesirable flanking transmissions). Junction type L3 is acceptable because of the linings used to reduce the flanking path. Junctions L4 to L6 are recommanded because studs are parallel to the junction (no structural continuity) and there is no direct connection between boards.

JUNCTION L1

Boards cutted

VS 1

US 1

VS 1

US 1

JUNCTION L2

VS 1

US 2

VS 1

US 3

JUNCTION L3 JUNCTION L4

VS 1

US 4

JUNCTION L5 JUNCTION L6

VS 1

US 4

Figure 7.26 Example of junctions between lightweight elements

154

7.6.2 Designing for good floor dynamic performance

The most significant parts of a floor system are:

• mass - self weight of structure and occupancy

• stiffness - floor stiffness, and

• damping - friction at joints in the supporting frame and energy losses by vibration of non-structural components, fittings and furnishings.

The following damping ratios, ζ,, should be used in design for estimating the response of floor systems:

ζ = 1.1% completely bare floors, or floors where only a small amount of furnishings are present

ζ = 3.0% fully fitted out and furnished floors in normal use

ζ = 4.5% when the designer is certain that the partition lines are perpendicular to the main vibrating elements of the critical mode shape and where evidence from previous experience justifies the use of this value.

For the design of lightweight floors, the most practical ways of improving vibration performance are:

• increase the stiffness of the floor in the transverse direction by a stiffer/deeper floor structure

• increase the stiffness of floor beams – make them deeper/heavier

• provide more main beams (at closer spacing)

• reduce the beam spans, if possible.

Increasing mass is not efficient, and increasing damping is normally impractical.

155

8 CONCLUSIONS

The main objectives of the project were: 1) improved modelling of acoustical and vibration behaviour of lightweight structures, 2) better understanding of acoustical behaviour of lightweight structures in respect of flanking, 3) standardisation of testing methods for vibration of lightweight floors, 4) introduction of new construction methods and solutions, and 5) introduction of a guide for designers.

In the project, existing models were used but also new models were developed in order to better understand the role of different components (boards, cavity, studs, rails etc.) in sound transmission through lightweight building elements. Three different modelling methods were used and compared: an energy based method (SEA), a simple wave based analytical method and different numerical methods. An accurate and fast model has been obtained by combining the wave approach and SEA. With the developed method, it is possible to predict the sound insulation index R of double wall by an accuracy of about 2 dB compared with laboratory measurements. In the case of single frame double walls, metal frames (studs and rails) play a dominant role in airborne sound transmission. The models show that the key parameter is the section stiffness of studs and rails; therefore, effort was put into developing laboratory methods for estimating this key parameter. The modelling method works and rail or stud section stiffness can be measured (or calculated by fem) and then used as input data in the model for calculating the R index of the wall considered. The developed methods make it possible to use advanced calculations instead of expensive testing in the product development processes, in order to find out relevant parameters affecting the sound insulation performance and optimizing the structure.

One goal of the project was to understand, quantify better and if possible reduce flanking transmission in the case of steel frame lightweight constructions. A European standardized prediction model for calculating the building performance, which isonly valid for heavy concrete structures, has been adapted to lightweight structures. The model input data are mainly 1) the performance of the building elements, 2) the performance of the junctions between elements, which can be measured either in laboratory or on site, and 3) the radiation efficiency of the building elements, which can be measured in laboratory. During the project, this prediction model was validated on an existing small building thoroughly tested on site; moreover, the performance of all the building elements was tested separately in the laboratory. The prediction model was able to estimate the sound insulation corresponding to the different transmission paths in the building accurately. In particular, it was shown that for the lightweight building tested, which had a separate load bearing column-beam frame (which allows loose connections between walls and load bearing frame), flanking transmissions were still present, particularly in the case of impact noise and for both vertical and horizontal transmissions.

Other tests have been performed on site during the project: (i) several buildings carefully studied, in order to reduce flanking (avoiding any structural continuity at junctions between elements), were successfully controlled, and showed that flanking can be almost suppressed; (ii) heavy concrete buildings with a lightweight façade have been tested in order to evaluate the importance of flanking through the façade in the airborne sound insulation between apartments; the results show that overall flanking is important, but of the same order as for heavy concrete façades.

For the second problem of airborne sound transmission through façades, a European standardized model for calculating the façade performance from the performance of all the façade elements (wall, windows etc.) has been used and validated on site; it seems that the European model is too optimistic for lightweight structures and that a safety margin has to be taken into account.

The main objective of modelling and simulation of vibration performance of floor structures was to obtain an understanding of which structural parameters have the most significant influence on the vibration properties of floors. The modelling was performed with finite elements, where the connections and supports were incorporated.

A group of field- and laboratory tests have been carried out with (to a high degree), a unified testing procedure that was developed in the project. The dynamic properties of the floors were measured, together with subjective evaluations, where the latter concerns both body perception and vibration induced noise and shaking of articles. The results from one specific series of laboratory tests showed the

157

importance of considering not only vertical but also horizontal vibration in a floor construction. Horizontal vibration can be eliminated in real buildings by careful design and installation of lateral fixing of the floors.

Two different vibration design methods have been partly developed and evaluated in the Acousvibra project. A common design criterion for both methods concerns the floor’s fundamental frequency. Owing to the special vibration nature of lightweight floors, they should be designed in such a way that the fundamental frequency exceeds 10 Hz. The first method is based on deflection limits. The floors can be classified into different vibration classes with specified deflection limits. For typical residential buildings, the deflection limit due to 1 kN load 0,5 mm is proposed. Based upon earlier tests, together with laboratory and field tests carried out within the project, this design criterion has been shown to work well for different kinds of lightweight floor and is therefore recommended for wider use. An additional advantage of this method is its simplicity. The second method studied and developed uses a design criterion based upon weighted floor acceleration, but it seems that for lightweight floors it is hard to find limits that distinguish acceptable from unacceptable floors. Preliminary limits have been introduced in the project concerning response factors of lightweight floors, but more research is needed in this area.

A new motion simulator, designed to simulate floor vibration, was also developed within the project. The tests carried out showed interesting results concerning vibration response from multiple frequencies. It also seems that natural frequencies other than the fundamental should be included in the design. The beating effect – an interaction effect of two adjacent natural frequencies – is according to the presented study a main parameter, but, whether the corresponding frequency separation is best treated as a design parameter of its own, or if it should be complemented/combined by others, is a task that hopefully future research will answer. A model that predicts the human annoyance from floor vibration based upon the ISO Wm weighting handles single frequencies well, but, for signals that comprises two to three well defined discrete frequencies a situation where the beating effect is significant but also a situation that might occur for real floors, it is not accurate enough.

Various high performance products were developed in the framework of the project. Two types of light-weight façade elements were studied and developed: 1) for a residential and office building fulfilling high traffic noise insulation requirements, and 2) for an industrial building fulfilling a high airborne sound insulation value. Both applications were based on perforated light-gauge steel thermal studs which also fulfilled high thermal insulation requirements. Connection details were also verified by field measurements.

Floor development focused on light-weight floors consisting of light steel joists, gypsum board ceiling and the covering that was built from trapezoidal sheeting, and either two layers of gypsum boards or thin concrete slab (~50 mm). Finally, subjective tests with proper floor connections showed acceptable vibration behavior of the lightweight floors. Similar results were also received from field tests.

The third group of products developed consisted of partition walls. Firstly, some modifications were made to the typical partition wall. A material saving was possible by introducing thinner studs and rails to those used previously. Fulfilling the stiffness requirements, the studs were embossed in order to increase the section stiffness of the stud profile. Furthermore, some preliminary design of a totally new type of partition wall was started. The study for the optimum shape of the wall stud and rail was made in close cooperation with other partners and research on work packages 2 and 4. Some preliminary proposals for new section types were given.

Finally, the design guide was prepared in the project, providing practical information for the design of lightweight steel structures so that they may have an adequate acoustic and vibration performance. It illustrates many examples of buildings using lightweight construction and their components and connection details. It demonstrates that lightweight steel construction is perfectly capable of meeting modern acoustic and vibration serviceability requirements. Comprehensive recommendations for design and detailing for good acoustic and vibrational performance are given. It illustrates many examples of floor and wall construction, together with their insulating properties. Major parameters that affect the performance at the whole building level are discussed, as well as for component design and junctions. Sound insulation against traffic noise is also covered. The basic principles of the modelling methods

158

used to predict sound transmission are provided. The key parameters influencing floor vibrations are also discussed, and current methods of on-site testing and measurement, data processing and subjective evaluation are given. The guide also presents proposals for harmonization of annoyance criteria.

159

9 EXPLOITATION AND IMPACT OF THE RESEARCH RESULTS

9.1 Actual applications Calculation methods for acoustic performance of lightweight structures were developed. New type of façade elements, lightweight floors and partition walls were developed in the project.

9.2 Technical and economic potential for the use of the results The developed modelling methods for prediction of sound insulation properties of the lightweight structures can be utilised in product development processes. Usually, many expensive laboratory tests are needed in order to find optimum acoustic solution. Now, the influence of different construction parts and layers on acoustic performance can be estimated by calculations in adequate accuracy and only few validation tests are needed. Vibration problems of floors are always very difficult to solve. Once constructed, it is very expensive to modify an existing floor to reduce its susceptibility to vibration, since only major changes to the mass, stiffness or damping of the floor system will produce any perceptible reduction human-induced vibrations. The proposed classification of the floors to vibration classes will assist designers and clients in making decisions regarding floor vibrations in order to avoid complaints by the occupants. The proposed vibration design method can be used 1) for the prediction of the floor vibration properties and 2) for determining the vibration class. Furthermore, it is recommended that the properties of a new floor type and its maximum span are always verified by prototype tests. Easiest way is to check the properties by sense perceptions, which is more authentic and accurate than most of the theoretical calculations. Standardised method for sense perceptions is proposed in this project.

Different analyses and performance tests have proved that it is possible to fulfil high acoustic and vibration requirements by lightweight steel products such as outer wall elements, partition walls and different flooring types.

9.3 Patent filings Within the framework of the project Rautaruukki Oyj has filed one patent concerning embossed partition wall stud.

9.4 Publications resulting from the project Following publications have been resulted from the project:

Articles in scientific journals

Ljunggren F. and Ågren A. Perception from simulated multiple frequency floor vibration. Submitted to Journal of Sound and Vibration, 2006.

Ljunggren F., Wang Y. and Ågren A. (2007) Human vibration perception from single and dual frequency components. Journal of Sound and Vibration, Volume 300, No 1-2, pp. 13-24.

Ljunggren F. and Ågren A. Dynamic and subjective analysis of a lightweight/semi-heavyweight floor in laboratory. Building Acoustics, Volume 13, No 4, pp. 255-272.

Conference papers

Sirari P. et al (2006) Experience of modelling sound insulation of lightweight structures with modified SEA. Proceedings Euronoise 2006, Tampere, Finland

161

Poblet-Puig J. and Rodriguez-Ferran P. (2006) Numerical modelling of sound transmission in double walls. Proceedings Euronoise 2006, Tampere, Finland

Guigou-Carter C. and Villot M. (2006) Analytical and experimental study of single frame lightweight double wall. Proceedings Euronoise 2006, Tampere, Finland

Poblet J. et al (2006) Experimental and numerical characterization of metallic studs. Proceedings Euronoise 2006, Tampere, Finland.

Technical reports

CR-03025/B001, SCI (2005) Work Package 1: Requirement and testing for acoustic and vibration performances of lightweight structures

CR-03025/E002, UPC (2004) Planning of numerical vibroacoustic simulations

CR-03025/E004, UPC (2005) Planning of WP”: Modelling and simulation

CR-03025/E005, UPC(2006) Finite element modelling of orthotropic floors

CR-03025/E006, UPC(2006) Calculation of elastic stiffness of metallic beams as input data for SEA

CR-03025/E007, UPC(2006) Stiffness of different stud profiles

CR-03025/E008, UPC(2007) UPC acoustic tests

CR-03025/D001, CSTB(2005) CSTB calculation models and related experimental validations

CR-03025/D002, CSTB(2006) CSTB calculation models and related experimental characterizations

CR-03025/D003, CSTB(2006) CSTB calculation models improvement, related experimental characterizations and sound transmission predictions

CR-03025/D004, CSTB(2006) Sound transmission prediction of two double walls tested at FIOH

CR-03025/D005, CSTB(2006) Sound transmission prediction - Double wall models comparison

CR-03025/F001, VTT(2004) SEA in sound insulation - Brief state of art

CR-03025/F002, VTT(2005) Verification of SEA parameters - Results obtained from laboratory tests

CR-03025/F003), VTT(2006) Theoretical analysis, validation and sensitivity analysis of a modified SEA based airborne sound insulation programme

CR-03025/F004, VTT(2006) Vibration tests on floors in Kauklahti

CR-03025/G001, LTU(2005) Harmonisation of testing procedure and annoyance rating regarding floor vibrations

CR-03025/G002, LTU(2006) Human vibration perception from single and dual frequency components

CR-03025/G003, LTU(2006) Oerception from simulated multiple frequency floor vibration

CR-03025/G004, LTU(2006) Dynamic and subjective analysis of a lightweight/ semi-heavyweight floor in laboratory

9.5 Dissemination of results Further to scientific articles and conference papers resulting from the project, project results can be disseminated through Design Guide. The Design Guide can be delivered to designers by providing practical information for the design of lightweight steel structures so that they may have an adequate acoustic and vibration performance. Each partner can also utilize the Design Guide for their own purposes such as for producing product guidance, lecture material etc.

162

10 LIST OF FIGURES AND TABLES

10.1 List of figures Figure 3.1 Assessment form for rating the vibrations 26

Figure 3.2 Directions for Vibration defined in ISO 2631 29

Figure 3.3 Wd and Wg frequency weighting curves 30

Figure 3.4 Wb frequency weighting curve 30

Figure 3.5 Building vibration (a) z-axis curves; and (b) x- and y-axis curves for rms acceleration according to BS6472 31

Figure 3.6 Deflection of 1 kN point load compared to acceptability. 33

Figure 3.7 Response factor compared to acceptability 33

Figure 4.1 Examples of double leaf lightweight walls a) with single frame; b) with separate sets of studs; c) façade wall with single frame and external finish (plastering) 36

Figure 4.2 SEA spring connection between 2 infinite plates 36

Figure 4.3 SEA transmission path diagram for a double wall 37

Figure 4.4 Wave approach basic configuration 38

Figure 4.5 Examples of single leaf and double leaf lightweight walls studied using hybrid models 38

Figure 4.6 Examples of structural path at the boundaries of a double wall with separate sets of studs 39

Figure 4.7 Example of single frame double wall studied using models (iii) and (iv) 39

Figure 4.8 Impact noise level of a single board with joists 40

Figure 4.9 Studs are modelled as real profile (left) or as translational and rotational springs (right)) 40

Figure 4.10 Influence of the stud shape on the vibration transmission loss 41

Figure 4.11 3D finite element model of laboratory setup for stud testing 41

Figure 4.12 Stud profiles analyzed 42

Figure 4.13 Typical outer wall tested (with thermal studs TC175) 43

Figure 4.14 New studs and rails used in walls tested in laboratory 43

Figure 4.15 View of the LEAM/UPC laboratory aperture 44

Figure 4.16 Example of double leaf panels tested 44

Figure 4.17 VTT experimental setup for detailed SEA tests on lightweight partition walls 45

Figure 4.18 CSTB experimental setup for stud characterization 45

Figure 4.19 Example of experimental results for stud characterization 46

163

Figure 4.20 CSTB experimental setup for boundary rail characterization 46

Figure 4.21 Example of experimental / predicted velocity level difference for boundary rail characterization 47

Figure 4.22 Example of Acoustic Wall Stud (AWS) 47

Figure 4.23 Predicted and measured index R for a double wall mounted on TC175 studs and rails; detailed results for the different transmission paths are given 48

Figure 4.24 Measured and predicted R index for a single frame double wall; comparison between SEA and hybrid models 49

Figure 4.25 Measured and predicted R index for a single frame double wall; comparison between finite element and wave approach models 49

Figure 4.26 Effect of stud translational (Kt) and rotational (Kr) stiffness values on the index R of single frame double walls 50

Figure 4.27 Effect of translational and rotational stiffness values of studs on the index R of single frame double walls 51

Figure 4.28 Typical example of flanking transmission path 53

Figure 4.29 Schematic of façade sound insulation measurement 55

Figure 4.30 Radiation efficiency of a stiffened single plate: 4 cm thick wooden board with T shape 14 cm high steel joists 56

Figure 4.31 Two dimensional model of part of a full size building tested during the project 56

Figure 4.32 Vibration level difference Dvij of the central cross junction: a) experimental results; b) numerical results 57

Figure 4.33 Bare floor without suspended ceiling (22 mm thick wooden chipboard) and ½ separating wall tested (two gypsum boards screwed on studs) 57

Figure 4.34 R index of the bare floor tested (without suspended ceiling) 58

Figure 4.35 Radiation efficiency of the ½ wall tested; with structural excitation (left); with acoustic excitation (right) 58

Figure 4.36 Primary separate load bearing frame; connection between beam and column 59

Figure 4.37 Part of the separate load bearing frame of the building tested 59

Figure 4.38 Connections between joists an primary load bearing beam 59

Figure 4.39 Schematic of the 4 rooms tested (load bearing frame in red; joist orientation in doted lines) 60

Figure 4.40 Cross junction between the 4 rooms studied 60

Figure 4.41 Airborne sound insulation spectra measured and calculated in the case of the horizontal transmission Cross junction between the 4 rooms studied 62

Figure 4.42 Typical outer wall structure with thermal studs 62

Figure 4.43 Example of façade tested (from report PR-R-1149-1) 63

Figure 4.44 Example of floor-outer wall junction measured on site 63

164

Figure 4.45 Acoustic wall stud used in partitions tested on site 63

Figure 4.46 Detail of the partition/roof junction (building tested on site) 64

Figure 4.47 Kvantti floor: case with plain jointed floor installed on timber battens and damping rubber strips 64

Figure 4.48 Detail on Kvantti floor/façade junction 65

Figure 5.1 Section of the modelled floor with its support. 70

Figure 5.2 Walking line along the floor, FEM. 72

Figure 5.3 Dynamic action of a human step. 73

Figure 5.4 Vertical displacement due to walking; a) along the floor and b) across the floor. 74

Figure 5.5 Mode shapes of the a) fundamental frequency and b) first torsion mode. 75

Figure 5.6 Walking lines and observation points used for subjective evaluations of vibration intensity (point 1), and for subjective evaluations of vibrating articles (tripod in point 1, test subject in point 2). 76

Figure 5.7 Rating of vibration from walking along the centre of floor; a) intensity and b) acceptance and from walking along the edge of the floor; c) intensity and d) acceptance. 76

Figure 5.8 Vibration acceptance from articles, a) clinking of a coffee cup, b) leaf movements of a pot plant, c) water rippling in a glass bowl and d) chinking of a glass pane. 77

Figure 5.9 Rating of vibration from walking along the centre of floor; a) intensity and b). Test No 9 is identical to test No 5 besides that the tubular support beams have been removed 77

Figure 5.10 Plan of the tested floor. 16 test points and three walking paths. 78

Figure 5.11 Test house, Derby. 79

Figure 5.12 Plan of the tested floor. 16 test points and the walking path. 80

Figure 5.13 Test house, Daventry. 81

Figure 5.14 Plan of the tested floosr. 14 test points and four walking paths. 81

Figure 5.15 Floors in house 1. To the left: floor 1, (1) and (6) were not installed when tested. To the right: floor 2 82

Figure 5.16 Floor in house 2. 83

Figure 5.17 Test points and walking paths of floors in house 1. 83

Figure 5.18 Test points and walking paths of the floor in house 2 84

Figure 5.19 Placement of the line loads in terms of timber beams. 1a) and 1b) over the supports, 2) centre transverse 3) ¼-point transverse and 4) centre main. 89

Figure 5.20 Mounting of the complete timber beam to the floor. 89

Figure 5.21 FRFs that show the effect of adding line loads to the floor in transverse direction. 90

Figure 5.22 FRFs showing the effect of adding a line load to the centre of floor in main direction. 90

165

Figure 5.23 FRFs that show the effect of different degree of mechanical connection of the timber beam to the floor structure in the case of a transverse centre line load. 91

Figure 5.24 FRFs that show the effect of different degree of mechanical connection of the timber beam to the floor structure in the case of a centre line load along the floor’s main direction. 91

Figure 5.25 FRFs showing the effect of adding mass on the floor above the supports and the effect of stiffening the supports by U-profiled beams connected to each C-beam. 92

Figure 5.26 U-profiled beams (cross section to the right) was added to the support (left) Five screws positioned in the rib centre together with spacers where appropriate were used for mounting. 92

Figure 5.27 The motion simulator used in the experiments. 94

Figure 5.28 a)Means with confidence intervals of the threshold values from sinusoidal vibration. b)Means of threshold values from the present study ——— compared with ISO base curve ----. 95

Figure 5.29 Mean threshold values in the presence of a base frequency of 8 Hz at three different amplitudes, 35mm/s2———, 50mm/s2---, and 70mm/s2 ···; a) together with confidence intervals and b) together with threshold values ·-·- from single frequencies as comparison. 95

Figure 5.30 Averaged annoyance rating. The single point to the left is the base frequency only, 8Hz (signal No. 1). The next group consists of the signals 8 + 10 Hz (signals No. 2-6) and after that follows 8 + 12.5 Hz (signals No. 7-11), 8 + 17 Hz etc. The amplitudes of the second frequency component are 0 (*), 7(x), 14(o), 21(◊), 28(□) and 35(∆) mm/s2. 97

Figure 5.31 a) annoyance and b) acceptance from single frequencies at two different amplitudes. Signals 1-10. 98

Figure 5.32 a) annoyance and b) acceptance from a pure 8 Hz signal of either low or high level vs. a combined signal of 8 Hz + a second low level frequency component of 10-17 Hz. Signals 1-2 and 11-18. 98

Figure 5.33 a) annoyance and b) acceptance from a pure 10, 12 or 14 Hz signal at high level vs. a combined signal where a second frequency component of 12-16 Hz at low level has been added. Signals 3, 5, 7 and 19-24. 99

Figure 5.34 a) annoyance and b) acceptance from a pure 8 Hz at high level, an 8+10Hz signal at high+low level and combined signals where a third frequency component of 12-17 Hz of low or high level is added. Signals 1, 11 and 25-30. 100

Figure 5.35 a) annoyance and b) acceptance from a pure 8 Hz at low level, an 8+10Hz signal at low+low level and combined signals where a third frequency component of 12-17 Hz of low or high level is added. Signals 2, 15 and 31-36. 100

Figure 5.36 Annoyance from signals built up of two frequencies (high+low level), all with 2 Hz separation. 101

Figure 5.37 Observed vs. predicted annoyance. 103

Figure 5.38 Predicted maximum amplitudes of a second component for acceptance in the presence of a fundamental frequency of a) 8 Hz and b) 10 Hz. 103

Figure 6.1 Structure of the light steel element and installation. 108

Figure 6.2 Two different floor-wall connection types 109

166

Figure 6.3 Wall element for industrial buildings 110

Figure 6.4 High acoustic performance wall element 110

Figure 6.5 Example of structure and installation of light-weight floor 111

Figure 6.6 Cover structure fastening details in gypsum board light-weight floor. 112

Figure 6.7 Cover structure fastening details in concrete topped light-weight floor. 113

Figure 6.8 Example of design chart for light-weight floor 114

Figure 6.9 Kvantti-floor with wooden plank covering 114

Figure 6.10 Kvantti floor structure measured in field in the case of gypsum board + trapezoidal sheeting covering. 115

Figure 6.11 Different embossing types 116

Figure 6.12 Bending test for partition wall structures 116

Figure 6.13 Example of design chart for maximum wall heights 117

Figure 6.14 Test set-up for measuring dynamic stiffness 118

Figure 6.15 Three different tested partition wall stud and rail combinations 118

Figure 6.16 Different stud profiles analysed 119

Figure 7.1 Two eight-storey lightweight steel residential buildings with a gross weight of 480 kg/m2, including all materials. (Sweden) 123

Figure 7.2 Single-family housing using lightweight steel (Loiste, Finland) 123

Figure 7.3 Off-site assembly of a wall element. 124

Figure 7.4 Schematic example of a "sandwich panel" wall element with sheet steel and high density mineral wool. 125

Figure 7.5 Lightweight steel module being elevated into its final position. (Sweden). 125

Figure 7.6 Lightweight floor and wall construction showing the assembly of an external wall element with “thermo profiles”, external plasterboard and a rendered EPS façade 127

Figure 7.7 (a): Combination of rendered facades and Minerite fibreboard panels in a light steel residential project using eight different facade systems (Malmö, Sweden). (b): Brick wall façade and a lower level curved roof using lightweight steel construction (Oxford, UK). 127

Figure 7.8 Schematic design of an insulated partition wall. (a) Ceiling rail. (b) 70 mm light steel stud, c/c 600 mm. (c) Hole for hidden services. (d/e) Single or double plasterboard layers. (f) Mineral wool for acoustic insulation. (g/h) Surface treatment. (i) Rubber strips for improved acoustic performance. 128

Figure 7.9 (a): A light steel flooring system, comprising a double gypsum layer on a profiled steel sheet, load-bearing C-profiles and a resilient ceiling connected to "acoustic rails". (b): A semi-light steel and concrete composite flooring system with a parquet floor solution and provision for hidden services. 129

Figure 7.10 A roof structure using light steel. 130

167

Figure 7.11 Junction types between lightweight walls and semi-light steel floors. (a): a light steel external wall and semi-light concrete flooring. (b): Similar junction but semi-light steel flooring and fixed end support. 131

Figure 7.12 Examples of double leaf lightweight walls a) with single frame; b) with separate sets of studs; c) façade wall with single frame and external finish (plastering) 132

Figure 7.13 Typical transmission loss frequency spectrum for a double wall 133

Figure 7.14 Major parameters influencing sound insulation for lightweight steel walls 134

Figure 7.15 Major parameters influencing airborne and impact sound insulation for lightweight steel floors 135

Figure 7.16 An example of vertical acoustic transmissions 136

Figure 7.17 Wave approach basic configuration 138

Figure 7.18 SEA transmission paths between sub-systems; the case of a double wall 138

Figure 7.19 Experimental set-up for stud characterization 139

Figure 7.20 Experimental setup for boundary frame characterization 140

Figure 7.21 Impact noise of stiffened lightweight floors 140

Figure 7.22 The coupled vibro-acoustic problem 141

Figure 7.23 Phases in finite element analysis of sound transmission 142

Figure 7.24 Possible floor and beam mode interactions 147

Figure 7.25 Mode shape factors 151

Figure 7.26 Example of junctions between lightweight elements 154

10.2 List of tables

Table 3.1 Summary of requirements for acoustic performance in the UK, France, Spain, Iceland and Scandinavia 18

Table 3.2 Acoustic terms used in this guide 21

Table 3.3 Typical details for Façade construction 22

Table 3.4 Typical details for roof construction 22

Table 3.5 Typical details for floor construction 23

Table 3.6 Typical details for separating walls and internal partitions 23

Table 3.7 Summary of Stiffness/frequency requirements 28

Table 3.8 Weighting factors appropriate for floor design 29

Table 3.9 Response factors used in ISO 2631-2:1989 to specify satisfactory magnitudes of building vibration 31

Table 3.10 Class recommendations for floors of residential and office buildings. 32

Table 3.11 Vibration classification of high-frequency floors. 32

168

Table 5.1 The eight different setups of the floor. 70

Table 5.2 The first two natural frequencies obtained from FE model A for the eight setups. 71

Table 5.3 The first two natural frequencies obtained from FE model B for some setups. 72

Table 5.4 Frequency (f) and damping ratio (ζ) for the fundamental mode (f1) and the first torsion mode (f2). 75

Table 5.5 Natural frequencies and damping ratios for the first ten modes. 79

Table 5.6 Pedestrian response data. 79

Table 5.7 Natural frequencies and damping ratios for the first five modes. 80

Table 5.8 Pedestrian response data. 80

Table 5.9 Natural frequencies and damping ratios for the first five modes. 82

Table 5.10 Pedestrian response data. 82

Table 5.11 Fundamental frequency and damping. 84

Table 5.12 Measured maximum peak deflection (|umax|), ISO-factor and static deflection (δ). 84

Table 5.13 Subjective rating of vibrations 85

Table 5.14 Comparison of FEM results vs. measured data in laboratory tests. 85

Table 5.15 Comparison of FEM results vs. measured data for field tests. 86

Table 5.16 Comparison of calculated and measured floor performance 87

Table 5.17 Comparison of calculated and measured values for test floors 88

Table 5.18 Regression models in the form of: Annoyance = a + b · Amplitude + c · Fundamental freq. + d · Frequency separation. 101

Table 5.19 Regression models in the form of: Annoyance = a + b · Weighted amplitude + c · Fundamental freq. + d · Frequency separation. 101

Table 6.1 The Finnish requirements for the A-weighted average sound level caused by environmental noise (Finnish Government, 993, 1992) 107

Table 6.2 Measured and calculated sound level differences in field tests 108

Table 6.3 Target values set for sound insulation indexes 115

Table 6.4 Predicted acoustic index for the 66 mm wall with single gypsum board layers on both sides. 119

Table 7.1 Low frequency floor to high frequency floor cut-off 149

Table 7.2 Mode shape factors for corridor to room response as shown in Figure 7.25(b) 151

169

11 REFERENCES

1. Robust Details Handbook, Robust Details Ltd, 2004

2. Acoustic Detailing for Multi-Storey Residential Buildings, SCI Publication P336, 2004

3. Design Guide for the High Quality Acoustic and Vibrational Performance of Lightweight Steel Constructions (ACOUSVIBRA), RFCS 2007

4. BS EN ISO 717-1:1997 Acoustics - Rating of sound insulation in buildings and of building elements. Part 1. Airborne sound insulation. British Standards Institution

5. BS EN ISO 717-2:1997 Acoustics - Rating of sound insulation in buildings and of building elements. Part 2. Impact sound insulation. British Standards Institution

6. Maia, N. M. M., Silva, J. M. M., He, J., Lieven, N. A. J., Lin, R. M., Skingle, G. W., To, W-M and Urgueira, A. P. V. (1997). Theoretical and Experimental Modal Analysis. Taunton, UK: Research Studies Press Ltd. and John Wiley and Sons.

7. Blakeborough, A. and Williams, M. S. (2003) Measurement of floor vibrations using a heel drop test. Structures and Buildings, Proceedings of the ICE, No. 4 Vol. 156, 367-371.

8. Reynolds, P. and Pavic, A. (2000a). Impulse Hammer versus Shaker Excitation for the Modal Testing of Building Floors, Experimental Techniques, Vol. 24, No. 3. 39-44. Society for Experimental Mechanics.

9. Reynolds, P. and Pavic, A. (2000b). Quality Assurance Procedures for the Modal Testing of Building Floor Structures, Experimental Techniques, Vol. 24, No. 4. 36-41. Society for Experimental Mechanics.

10. Griffin, M. J. (1996). Handbook of Human Vibration. London, UK: Academic Press

11. YOUNG, P. Improved floor vibration prediction methodologies Arup Vibration Seminar, Institution of Mechanical Engineers, 2001

12. MURRAY, T., ALLEN, D. E., UNGAR, E. E. ATC Floor Vibrations Due to Human Activity. American institute of Steel Construction, One East Wacker Drive, Suite 3100, Chicago, Illinois, 1997

13. HICKS, S. J., KING, C. M. and DEVINE, P. J. Design guide on the vibration of floors in hospitals. The Steel Construction Institute

14. OHLSSON, S. Springiness and human-induced floor vibrations. Swedish Council for Building Research

15. AS3623. Australian Domestic Metal Framing Code. Standards Australia

16. ONYSKO, D. M., et al. ATC Design guide 1. Minimising Floor Vibration. Applied Technology Council, Redwood, California

17. BS EN 1995-1-1:2004 Eurocode 5, Design of timber structures: General rules and rules for buildings.

18. FCSA (2005). Walking-induced floor vibrations. Certified Product Declaration TRY-93-2005. 9 p.

19. ISO 2631-1: Mechanical vibration and shock – Evaluation of human exposure to whole-body vibration: Part 1: General requirements. International Organisation for Standardization, Geneva. 1997

170

20. ISO 2631-2: Evaluation of human exposure to whole-body vibration: Part 2: Continuous and shock-induced vibration in buildings (1 to 80 Hz), Switzland, International Organisation for Standardization, 2003

21. BS 6472: 1992 Evaluation of human exposure to vibration in buildings (1 Hz to 80 Hz). British Standards Institution

22. ISO 2631-2: 1989 Evaluation of human exposure to whole-body vibration: Part 2: Continuous and shock-induced vibration in buildings (1 to 80 Hz), Switzerland, International Organisation for Standardization, 1989

23. Toratti, T. & Talja, A. (2006) Classification of human Induced Floor Vibrations, Building Acoustics, Vol. 13, No: 3, pp. 211-221.

24. R.M. Craik, 1996, Sound transmission through buildings using Statistical Energy Analysis, Gower Publishers, Aldershot, England

25. Frank Fahy, 1985, Sound and structural vibration, Academic Press, England

26. M.Villot and C. Guigou-Carter, “Predicting the Acoustical radiation of finite size multi-layered structures by applying spatial windowing on infinite structures”, Journal of Sound and Vibration, (2001) 245(3)

27. C. Guigou-Carter and M. Villot, “Modelling of sound transmission through lightweight elements with stiffeners”, Journal of Building Acoustics, 10(3), 2003

28. Finite element modelling of orthotropic floors. Report to the project “High quality acoustic and vibration performance of lightweight steel constructions”, Document CR03025/E005. Universitat Politécnica de Catalunya 2006

29. Ellingwood B, Talin A. Structural serviceability: floor vibrations. Journal of structural engineering. ASCE, 10(2), 401-418, 1984

30. Talja A, Toratti T. Testing of human induced vibrations in residential buildings. VTT. Background paper for RFSC project “Acousvibra” 2005.

31. Pavic A, Reynolds P. Dynamic testing of lightweight floor structure, Loughborough Community Centre, Brixton. The Steel Construction Institute, VES/05/07 2005

32. Pavic A, Reynolds P. Dynamic testing of ‘Rosebury’ house first floor, Miller Homes (East Midlands) Ltd, Derby. The Steel Construction Institute, VES/05/11 2005

33. Pavic A, Reynolds P. Dynamic testing of ‘Ferndale’ house, Advance Housing Ltd, Daventry. The Steel Construction Institute, VES/05/13 2005

34. Talja, A. Vibration tests on floors in Kauklahti. Report to the project “High quality acoustic and vibration performance of lightweight steel constructions”, Document CR03025/F004. VTT Technical Research Centre of Finland 2006.

35. BS6841Guide to measurement and evaluation of human exposure to whole-body mechanical vibration and repeated shock. British Standards Institution (BSI) 1987.

36. ISO 2631 Mechanical vibration and shock – Evaluation of human exposure to whole-body vibration – Part 1: General requirements. International Standards Organisation, 1997.

37. Wyatt, T A. Design Guide on the Vibration of Floors (SCI Publication 076). Ascot, UK: SCI.

38. Talja A, Toratti T, Järvinen E. Vibration of floors. Design and testing procedures. VTT Research Notes 2124 2002.

39. Ellis B R. On the response of long-span floors to walking loads generated by individuals and crowds. The structural engineer 78(10) 2000.

171

40. ISO 2631 Evaluation of human exposure to whole-body vibration – Part 2: Continuous and shock-induced vibrations in buildings. International Standards Organisation, 1989.

41. Parsons K C, Griffin M J. Whole-body vibration perception thresholds. Journal of Sound and Vibration 121(2) 1988.

42. McKay J R. Human response to vibration: Some studies of perception and startle. Ph.D. Thesis, University of Southampton 1972.

43. Murray T M, Allen D E, Ungar E E. Floor vibrations due to human activity. American Institute os Steel Constructions, 1997.

44. Kullaa J, Talja A. Vibration performance test for light-weight steel-joist floors. Proceedings of Light-weight steel and aluminium structures, Helsinki, 1999.

45. Ljunggren F, Ågren A. Perception from Simulated Multiple Frequency Floor Vibration. Submitted to Journal of Sound and Vibration 2006.

46. Talja A, Toratti T. Effect on floating floors on the vibration performance of wood-concrete composite floors. Proceedings of World Conference on Timber Engineering WCTE200, Canada, 2000.

47. Al-Foqaha’a A A, Cofer W F, Fridely K J. Vibration Design Criteria for Wood Floors Exposed to Normal Human Activities. Journal of Structural Engineering 125(12) 1999.

48. Sipari P, Saarinen A, Dimensioning of the sound insulation of building façade; Ministry of the Environment, Environment guide 108, October 2003, Finland.

49. FCSA (2005). Walking-induced floor vibrations. Certified Product Declaration TRY-93-2005.

50. Design Guide for the High Quality Acoustic and Vibrational Performance of Lightweight Steel Constructions (ACOUSVIBRA), RFCS 2007

51. Generalisation of criteria for floor vibrations for industrial, office, residential and public building and gymnastic halls, Final report RFCS project 7210-PR/314, Unit G..5 – RFCS, European Commission, 2003

52. Wheeler, J. E., prediction and control of pedestrian-induced vibration in footbridges, Journal of Structural Engineering, ASCE, Vol.108, No.9, pp.2045-2065, 1982.

53. BS EN 1990:2002 Eurocode Basis of structural design: Brussels, CEN, 2002

54. F. Fahy, Sound and Structural Vibration, Academic Press, London, 1985

55. R.J.M. Craik, Sound transmission through buildings using Statistical Energy Analysis, Gower, London, 1996

56. M. Villot, C. Guigou-Carter and L. Gagliardini, “Predicting the acoustical radiation of finite size multi-layered structures by applying spatial windowing on infinite structures”, Journal of Sound and Vibration, 145(3), 2001.

57. C.Guigou-Carter and M.Villot, “Modeling of sound transmission through lightweight elements with stiffeners”, Journal of Building Acoustics, 10(3), 2003

58. Zienkiewicz, O.C. and Taylor, R.L. The Finite Element Method. Butterworth-Heinemann, 5th edition (2000).

59. Johnson. C. Numerical Solution of Partial Differential Equations by the Finite Element Method. Cambridge University Press (1987).

60. Hoffman, J.D. Numerical methods for engineers and scientists. McGraw-Hill, (1992).

61. Kinsler, L.E., Frey, A.R. and Coppens, A.B. Fundamentals of Acoustics. Wiley (1990).

172

62. Ihlenburg, F. Finite Element Analysis of Acoustic Scattering Springer-Verlag (1998).

63. Clough R.W. and Penzien, J. Dynamics of Structures. McGraw Hill (1993).

64. Beranek, L.L. and Vér I.L.. Noise and Vibration Control Engineering. Wiley (1992).

65. Devine, P.J., Smith, A.L., and Hicks, S.J. Design Guide on the Vibration of Floors (2nd Edition). The Steel Construction Institute, 2007.

66. YOUNG, P. Improved floor vibration prediction methodologies Arup Vibration Seminar, Institution of Mechanical Engineers, 2001.

173

APPENDIX A TEST FLOOR DESCRIPTIONS

A.1 Laboratory tests

Test identification Lab 1 and Lab 2

Structure:

1) 2x Gypsum board 15+15mm 2) Trapezoidal sheeting 35/0,8 3) a ) C-350-80-30x2.0

b) U-354-65x2.0 4) Ceiling joist 25mm 5) Gypsum boards 13+13mm 6) 250x150x8 tubular beam 7) Concrete wall, b= 80mm

Floor span length [m] 6,0 Main joist support condition1 RIGID Floor width [m] 3,6 Support condition of longit. edges2 Lab 1 YES

Lab 2 NO Spacing of joists [m] 0,6 Number of spans 1 Total mass [kg/m2] 72 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

6.28E+06

Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

9.78E+04

LABORATORY TEST LTU

Test identification Lab 3 and Lab 4

Structure:

1) 2x Gypsum board 15+15mm 2) Trapezoidal sheeting 35/0,8 3) a ) C-350-80-30x2.0

b) U-354-65x2.0 4) Ceiling joist 25mm 5) Gypsum boards 13+13mm 6) 250x150x8 tubular beam 7) Supports at the end of beam

Floor span length [m] 6,0 Main joist support condition1 ON BEAM Floor width [m] 3,6 Support condition of longit. edges2 Lab 3 NO

Lab i 4 YES Spacing of joists [m] 0,6 Number of spans 1 Total mass [kg/m2] 72 Supporting beam3 length [m] 3,6 Longitudinal stiffness [Nm2/m]

6.28E+06

Supporting beam stiffness [Nm2] 1,026E+07

Perpendicular stiffness [Nm2/m]

9.78E+04

LABORATORY TEST LTU

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

174

Test identification Lab 5 and Lab 6

Structure:

1) Concrete 15…65mm 2) Trapezoidal sheeting 20/0,8 3) a ) C-350-80-30x2.0

c) U-354-65x2.0 4) Ceiling joist 25mm 5) Gypsum boards 13+13mm 6) 250x150x8 tubular beam 7) Supports at the end of beam

Floor span length [m] 6,0 Main joist support condition1 ON BEAM Floor width [m] 3,6 Support condition of longit. edges2 Lab 6 YES

Lab 5 NO Spacing of joists [m] 0,6 Number of spans 1 Total mass [kg/m2] 170 Supporting beam3 length [m] 3,6 Longitudinal stiffness [Nm2/m]

1.79E+07

Supporting beam stiffness [Nm2] 1,026E+07

Perpendicular stiffness [Nm2/m]

3.19E+05

LABORATORY TEST LTU

Test identification Lab 7, Lab 8 and Lab 9

Structure:

1) Concrete 15…65mm 2) Trapezoidal sheeting 20/0,8 3) a ) C-350-80-30x2.0

d) U-354-65x2.0 4) Ceiling joist 25mm 5) Gypsum boards 13+13mm 6) 250x150x8 tubular beam 7) Concrete wall, b= 80mm

Floor span length [m] 6,0 Main joist support condition1 RIGID Floor width [m] 3,6 Support condition of longit. edges2 Lab 7, 9 NO

Lab 8 YES Spacing of joists [m] 0,6 Number of spans 1 Total mass [kg/m2] 170 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

1.79E+07

Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

3.19E+05

LABORATORY TEST LTU

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

175

A.2 Field tests

Test identification Kauklahti, House 1 (Ruukki)

Structure:

-Ceramic tiles (not installed at tests) -Concrete 50mm /85mm -Trapez. sheeting 35/0,7 -Steel joists C200-70-22x2.0 -Mineral wool 100mm -Vapour barrier -Hat profile 20mm -Gypsum board 13mm (not installed)

Floor span length [m] 4,2 Main joist support condition1 RIGID Floor width [m] 8,5 Support condition of longit. edges2 YES Spacing of joists [m] 0,6 Number of spans 1 Total mass [kg/m2] 150 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

6,36E06 Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

5,44E05 FIELD TEST VTT

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

Test identification Kauklahti, House 2 (Ruukki)

Structure:

1. Surface structure 2. Concrete 52…70 mm 3. Trapez. sheeting 20 / 0.7 4. Steel joist C200 – 2.0, c/c 250

mm, 5. Hat profile c/c 400 mm 6. Primary beams:

a. CFRHS 200x100x6 in the centre restraint,

b. CFRHS 150x100x6 in the edge restraints,

7. Suspended ceiling: hat profile c/c 400 mm + gypsum board GF 15

Floor span length [m] 3,5 Main joist support condition1 on beam Floor width [m] 4,17 Support condition of longit. edges2 YES Spacing of joists [m] 0,25 Number of spans 2 Total mass [kg/m2] 181 Supporting beam3 length [m] 4,17 Longitudinal stiffness [Nm2/m]

1,11E07 Supporting beam stiffness [Nm2] 2,33E06

Perpendicular stiffness [Nm2/m]

4,11E05 FIELD TEST VTT

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

176

Test identification Brixton (SCI)

Structure: -18mm chipboard -15mm db board -50mm Rockfloor -22mm Caberdeck -200 ‘C’ Section -15mm plasterboard -15mm Plasterboard

Floor span length [m] 6.68 Main joist support condition1 Into wall Floor width [m] 7.2 Support condition of longit. edges2 Simple supports Spacing of joists [m] 0.6 Number of spans 4 Total mass [kg/m2] 27.1 Supporting beam length [m] 4.38 Longitudinal stiffness [Nm2/m]

1.528E+06 Supporting beam stiffness [Nm2] 2.56E+06

Perpendicular stiffness [Nm2/m]

2840 Field Test SCI

Test identification Daventry (SCI)

Structure:

1) 22mm Caberdeck 2) 220 ‘C’ Terrapin section 3) 15mm Fermacell

Floor span length [m] 4.875 Main joist support condition1 Into wall Floor width [m] 3.145 Support condition of longit. edges2 Simple supports Spacing of joists [m] 0.59 Number of spans 2 Total mass [kg/m2] 68.1 Supporting beam3 length [m] 3.145 Longitudinal stiffness [Nm2/m]

2.94E+06 Supporting beam stiffness [Nm2] 7.45E+06

Perpendicular stiffness [Nm2/m]

3549 Field test SCI

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

22mm Caberdeck

220 ‘C’ Section

15mm Fermacell

18mm Chipboard15mm db board

50mm Rockfloor22mm Caberdeck

200 ‘C’ Section

15mm plasterboard15mm plasterboard

177

Test identification Derby

Structure:

1) 22mm chipboard 2) 200mm timber beam 3) Ceiling

Floor span length [m] 3.9 Main joist support condition1 On to beam Floor width [m] 5.7 Support condition of longit. edges2 Simple supports Spacing of joists [m] 0.6 Number of spans 4 Total mass [kg/m2] 28.5 Supporting beam3 length [m] 6.6 Longitudinal stiffness [Nm2/m]

0.53E06 Supporting beam stiffness [Nm2] 2.17E+07

Perpendicular stiffness [Nm2/m]

1564 Field test SCI

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

178

A.3 Old laboratory tests used in verifications

Test identification VP1

Structure:

1) Chipboard 22mm 2) Hard mineral wool 30mm 3) Plywood 15mm 4) C-sections 250mm 5) gypsum board GTS9 9mm 6) Acoustic profile 25mm 7) Gypsum board 13mm

Floor length [m] 7,0; 7,8; 8,8 Main joist support condition1 RIGID Floor width [m] 8,3 Support condition of longit. edges2 YES Spacing of joists [m] 0,4 (2x) Number of spans 1 Total mass [kg/m2] 83 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

9,608x106 Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

2,30x103 LABORATORY TEST VTT

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

Test identification VP2

Structure:

1) Two layer gypsum 15+15mm 2) Trapez. sheeting 35/0,7 3) Plywood 15mm 4) C-sections 250mm 5) gypsum board GTS9 9mm 6) Acoustic profile c/c400 25mm 7) Gypsum board 13mm

Floor length [m] 7,0; 7,8; 8,8 Main joist support condition1 RIGID Floor width [m] 8,3 Support condition of longit. edges2 YES Spacing of joists [m] 0,4 (2x) Number of spans 1 Total mass [kg/m2] 110 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

9,608x106 Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

2,275x104 LABORATORY TEST VTT

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

179

Test identification VP3

Structure:

1) Concrete 25…60mm 2) Trapez. sheeting 35/0,7, 2

additional 5.5 screws in alternate corrugation, screw heads 15mm uplifted

3) Plywood 15mm 4) C-sections 250mm 5) gypsum board GTS9 9mm 6) Acoustic profile c/c400 25mm 7) Gypsum board 13mm

Floor length [m] 7,0; 7,8; 8,8 Main joist support condition1 RIGID Floor width [m] 8,3 Support condition of longit. edges2 YES Spacing of joists [m] 0,4 (2x) Number of spans 1 Total mass [kg/m2] 170 Supporting beam3 length [m] - Longitudinal stiffness [Nm2/m]

1,873x107 Supporting beam stiffness [Nm2] -

Perpendicular stiffness [Nm2/m]

3,028x105 LABORATORY TEST VTT

1 on the rigid wall / on the beam 2 YES = Supported / No = Free edges 3 If the joists are supported on the beams

180

European Commission

EUR 23319 — High quality acoustic and vibration performance of lightweight steel constructions

J. Kesti, S. Hicks, J. Rackham, J. Widman, M. Villot, C. Guigou, A. Rodríguez-Ferran, J. Poblet-Puig, P. Sipari, A. Talja, F. Ljunggren, A. Ågren,

Luxembourg: Office for Official Publications of the European Communities

2008 — 180 pp. — 21 × 29.7 cm

Research Fund for Coal and Steel series

ISBN 978-92-79-08304-4

ISSN 1018-5593

Price (excluding VAT) in Luxembourg: EUR 20