EXPLOSIVE SPALLING OF CONCRETE IN FIRE: NOVEL TESTING TO MITIGATE DESIGN RISK

Ieuan Rickard and Luke Bisby, The University of Edinburgh, UKSusan Deeny, Arup, UK

SYNOPSIS

Heat-induced explosive spalling in fire poses a credible risk to concrete structures, and has received considerable research attention in recent decades. However, no validated guidance to enable the design of concrete mixes to prevent spalling, nor any established, widely verified, repeatable test methods are yet available to confidently quantify or demonstrate spalling resistance for a particular mix in a given application. As a result no models yet exist that can predict spalling with sufficient confidence to be used in design. This paper summarises contemporary research on heat-induced concrete spalling, with particular emphasis on design for fire of concrete-lined tunnels. The topic is also highly relevant for modern concrete buildings. A novel, repeatable, and economical testing method to reduce project risk by quantifying the propensity of concrete mixes for spalling under a range of different thermal and mechanical conditions is described. The intent of this paper is to present the limitations of knowledge to enable design for heat induced spalling, and to highlight research currently underway to overcome some of the issues faced in practice.

EXPLOSIVE SPALLING OF CONCRETE

Concrete structures have historically performed very well in building fires1,2. Concrete is non-combustible and has a relatively low thermal conductivity and diffusivity, so that in a fire, provided the concrete cover to the internal reinforcement remains in place, heat flow to the reinforcement and the inner core occurs slowly so as to yield the necessary fire resistance. Traditional ‘fire resistance’ design of concrete structural elements is usually accomplished by prescribing minimum overall member dimensions and minimum concrete cover to the reinforcement. The available evidence from real fires suggests that these simple, prescriptive approaches have historically yielded an acceptable level of fire performance in concrete buildings2,3 and that heat-induced concrete spalling has not been a serious concern. However, contemporary concrete mixes, particularly those with higher strength and/or reduced permeability, appear more prone to heat-induced spalling, and this has been shown by both experiences from real fires and recent research on heat-induced spalling. This raises concerns that heat-induced spalling may become a more prominent issue in future fires and therefore that the potential consequences and the use of assessment and mitigation methods ought to be considered by structural engineers when undertaking structural fire design of modern concrete structures.

Spalling refers to the breaking away of pieces from the surface of concrete elements when exposed to heat1,2; it can take several different forms and may significantly affect the load carrying capacity of a concrete structure. This is due to reductions in cross section, changes in the load distribution, and loss of thermal protection to internal reinforcement. It is an extremely complex phenomenon involving time and temperature dependent mechanical stresses, temperature gradients, differential thermal stresses, moisture movement, and microstructural and chemical changes with increasing temperature4. Figure 1 below shows two spalled samples after testing.

Figure 1: A) Extreme case of explosive spalling of ultra high performance concrete, and B) Spalling of a concrete slab surface exposing internal reinforcement during a standard furnace test

The available research suggests that heat-induced spalling exhibits a stochastic nature, and experimental results are sometimes contradictory. Whether this is due to genuine randomness or to insufficiently controlled or instrumented testing environments is a matter of debate2. This uncertainty hinders design and often results in conservative and/or semi-arbitrary mitigation measures being used. However, general trends in factors increasing the risk of spalling can be highlighted. Spalling risk tends to increase with increases in compressive strength, rate of heating, moisture content, mechanical restraint, and imposed compressive load1,2,5. An exhaustive list of factors known to influence spalling is given by Maluk6, Bailey and Khoury1 provide a highly accessible summary of the available knowledge on heat-induced spalling of concrete, and a detailed and comprehensive summary of the available literature is given by Jansson5.

BACKGROUND

The current paper is interested in a particular manifestation of heat-induced spalling often referred to as ‘explosive spalling’; this is violent and typically (but not always) occurs when concrete elements experience steep in-depth thermal gradients, as in the early stages of a severe fire. A number of notable examples of real fires in which explosive concrete spalling has occurred are available in the literature, including a number of tunnel fires4,7, fires on or under bridges8,9, and building fires5.

Two key mechanisms are now acknowledged as contributing to explosive spalling. The first mechanism is related to the development of differential thermal stresses in the concrete induced by thermal gradients, differential thermal expansion, and induced restrained deformations that develop during heating10; the second mechanism attributes spalling to a build-up of pore pressure due to evaporation, along with transport of moisture within the concrete microstructure11,12. Harmathy11 originally hypothesized that the pore pressure in heated concrete is increased due to the formation of a liquid water ‘moisture clog’ that results from moisture transport and condensation in the cooler zones within the concrete core. More recently other mechanisms have also been proposed5.

Regardless of the mechanisms responsible, fire induced explosive spalling erodes cross-sectional dimensions and removes the concrete cover, thus exposing internal reinforcement to more severe heating and reducing its strength and stiffness. This could ultimately lead to structural damage, loss of structural stability, or loss of water-tightness. The potential costs of spalling in buildings are hard to quantify, however in tunnels the costs associated with repairs and loss of revenue during rehabilitation are likely to be considerable. For example, the direct repair costs following the 2008 Eurotunnel fire were estimated at £46 million, and given that the tunnel facilitates £91.4bn of trade annually, downtime as a result of fire could be as high as £250m per day, not accounting for indirect losses13. There are no obvious documented cases of large-scale structural failures or collapses that can be attributed directly to explosive spalling, and the consequences of spalling for structural performance of concrete structures admittedly remains a topic of on-going research14.

The importance of preventing, or properly accounting for, explosive spalling during design of concrete structures is related to at least three design considerations:

1. Life safety – Explosive spalling reduces structural capacity, and hence ‘fire resistance’, and could result in life-threatening structural collapses, exacerbated fire spread, or loss of water-tightness (particularly important in submerged tunnels).

2. Asset protection and financial losses – Explosive spalling causes damage to concrete elements, raising concerns for both direct and indirect economic losses, increased post-fire downtime and repair, asset protection, business continuity, and so on.

3. Project risk – The predominant approach to actually assess spalling risk for concrete mixes used in real projects (at least in tunnelling projects where spalling mitigation often governs concrete mix design) is by full-scale furnace testing of representative structural elements. This approach presents serious technical challenges, and generates project risk as a concrete may fail at a late stage of design. It is generally not feasible to run multiple tests, and in tunnel design and construction testing is often performed late in the design stage after the concrete mix and tunnel lining thickness have been fixed. Test failure would result in redesign and project delays, thus reducing design confidence in the early stages. Reliance on small sets of tests also places restrictions on the Contractor, limiting their ability to change concrete mix designs in response to supply, constructability, or economic factors occurring after testing4. Project risk due to heat induced spalling of concrete is usually less critical in other areas of construction, where spalling is rarely explicitly considered and ‘fire resistance’ is assured by relying on historical evidence for furnace tests and real building fires. Whether structural engineers ought to be more reflexive as regards the potential consequences of concrete spalling in modern building projects is open to debate.

Addressing the above design issues for modern concrete mixes requires the ability to quantitatively predict explosive spalling – such that the depth of spalling to be expected during a credible design fire can be approximated and accounted for during the structural design process – or prevent spalling – such that design can performed confidently neglecting it’s influence. Whilst considerable progress has been made on both fronts in recent decades, both remain difficult challenges and are the focus of a great deal of research internationally15.

AVAILABLE RESEARCH ON SPALLING

The available research on explosive spalling of concrete can be classified into three broad categories, as outlined and discussed in the following sections.

(1) Standard Furnace TestingThe ‘fire resistance’ of structural elements (of all material types) is assessed in practice using large-scale ‘fire resistance’ tests (Figure 2). These tests subject loaded, representative, and typically full-scale (or as close to full-scale as possible) structural elements to a standard gas temperature versus time curve within a fire testing furnace. Explosive spalling has been directly observed for a large number concrete elements tested in this manner, for more than a century16, however this spalling has historically been mild and not thought to be of critical importance for the overall fire resistance of concrete structures. In recent decades, however, with the advent of engineered high-strength, high-performance, and self-consolidating concrete mixes with increased propensity for explosive spalling, spalling has become a more common failure mode for concrete elements tested in furnaces17.

Since explosive spalling depends not only on material parameters but also on structural parameters, it is necessary to repeatably and accurately reproduce realistic conditions (e.g. geometry, boundary conditions, applied mechanical and thermal loads, etc) when experimentally studying spalling of different concrete mixes. Such control over testing parameters is challenging when testing in furnaces, particularly during the early stages of heating which are critical for influencing explosive spalling. Furnace tests therefore have limited applicability for quantifying spalling risk, although in some cases the time to spalling, final depth of spalling, and volume of spalled material are noted in furnace testing reports. Such testing is typically used to support design assumptions of: (1) no spalling under a ‘standard fire’ heating scenario (i.e. where none is observed in furnace testing); or (2) a certain depth of spalling (sometimes assumed as the final spalling depth observed during standard furnace testing). However, given the known variability of spalling for a single mix under slightly different thermal and mechanical conditions, neither assumption can usually be confidently defended based on the available experimental evidence.

Figure 2: A large number of samples positioned on a standard fire testing furnace ready for spalling testing by standard fire exposure from below

The reliance on full scale furnace testing is particularly problematic for assessing the spalling risk of concrete tunnels or tunnel lining segments. A number of furnace test procedures have been applied to assess spalling of concrete for tunnelling applications; however there has been little harmonization of these from either a thermal or a mechanical perspective. The majority of these test large-scale tunnel structure or lining samples exposed to a ‘tunnel fire’ heating regime18,19 based on data from tunnel fire tests that are assumed to adequately represent the thermal environment during a severe fire within a tunnel.

Furnace testing of structural elements other than tunnel segments is also challenging as it is often only possible to test single elements, and loading/restraint cases may be hard to replicate. Demonstration by testing that a concrete mix will perform adequately is not as commonplace elsewhere in structural engineering as in tunnel engineering, but may be advisable where modern, high-strength concrete mixes are being used or where spalling could be a critical to meet functional performance objectives. This is the case, for example when applying slender ultra high performance concrete slabs, pre-stressed with carbon fibre reinforced polymers, in building applications20.

(2) Spalling ExperimentsOwing to the numerous challenges associated with large-scale furnace testing of concrete elements to assess their propensity for spalling, a wide range of scientifically-based test methods and research projects are presented in the technical literature5,21. A detailed summary is avoided here, but it is noteworthy that researchers have used a range of sample sizes (from small cubes of only a few centimetres up to full-scale structural elements on the scale of metres), heating conditions (including furnace environments, electrical ovens, open pool and bonfires, direct flame impingement, cone heaters, and electrical, optical, and propane-fired radiant panels), and mechanical restraint and loading conditions (including unloaded, passively restrained, and actively restrained samples under uniaxial or biaxial, steady-state or time varying stress states).

A shortcoming of the available data from such experiments is that often insufficient care is taken when controlling the relevant material, thermal, and mechanical parameters that may affect spalling. For instance, samples have been exposed to heating scenarios which are poorly controlled or potentially non-uniform. Other testing has paid little attention to either mechanical or differential thermal stress development in the samples during heating6. As a result, there is currently no agreement or harmonization within the spalling research community regarding the appropriate small-scale experimental methods that should be used to quantify spalling risk or the

respective influences of parameters influencing spalling. It is therefore difficult to compare results from different authors. The RILEM Technical Committee on Spalling, 256-SPF, is currently seeking to address this by developing an internationally harmonised testing method to assess spalling risk.

(3) Predictive Numerical ModellingA number of related coupled thermo-hygro-chemo-mechanical codes of varying complexity are available in the literature; these generally attempt to predict spalling time and depth by simulating the stress state in heated concrete resulting from the relevant thermo-hygral and thermo-mechanical processes that take place during heating10,22. However, even the most developed of these advanced models are, at present, unable to accurately predict spalling for a given concrete mix in a particular application and subjected to either standard or project-specific design fires. A number of simplified spalling models (or spalling criteria) are also available in the literature – for instance assuming that only pore pressures are relevant23 or that spalling will occur at a particular temperature in the concrete – however neither of these approaches is easily defended based on the available experimental evidence.

STATE-OF-PLAY

The situation described above has resulted in a state-of-play where:1. the mechanisms of explosive spalling, and the respective influences of the various factors that

are thought to exacerbate explosive spalling, are not fully understood and cannot be quantified;2. confident prediction and/or prevention of explosive spalling, under the full range of relevant

thermal or mechanical actions for a given mix in a given application, is not currently possible;3. a lack of harmonization of testing methodologies has resulted in an ad-hoc, project-specific

approach to testing and experimentation, thus somewhat muddying the waters of scientific advancement on this topic; and

4. the potential consequences of spalling for the response of concrete structures in fire have not been quantified in most cases and in many applications (aside from tunnelling and nuclear engineering) spalling is rarely explicitly considered by designers (although again RILEM Committee 256-SPF is studying this issue).

DESIGN APPROACHES TO ACCOUNT FOR SPALLING

Given the above state-of-play, structural designers currently apply one of four approaches to addressing the issue of spalling in the structural design process:1. Assume no spalling – In most cases (aside from in tunnel design, see below) designers

assume that spalling is sufficiently unlikely to occur, or that its consequences will be sufficiently benign, that they need not explicitly consider it in design; this is the approach suggested in Eurocode 224 for concrete with compressive strength less than 80 MPa and moisture content less than about 3%. Such an approach is only partly supported by the available experimental data.

2. Mitigate spalling – Eurocode 224 also provides a range of means to reduce the risk of explosive spalling. For concrete with a strength greater than 55 MPa and with more than 6% silica fume (by weight of cement), or for concrete with a strength greater than 80 MPa, four specific options are given: (1) adding steel reinforcement mesh within the concrete cover at a depth of 15 mm, (2) demonstration ‘by experience or testing’ that no spalling occurs under fire exposure, (3) addition of supplemental fire protection (see Point 3 below), or (4) addition of 2 kg/m3 of polypropylene (PP) anti-spalling fibres. Polypropylene fibres have been shown1,2,5 to reduce the risk of explosive spalling, although the mechanism by which this is achieved remains a matter of debate. Eurocode 2 fails to make clear, however, that the current understanding of the phenomena of spalling is insufficient to enable a genuine risk-based approach to design for spalling. It is not currently possible to understand the risk of spalling for untested concrete mixes, nor to generate specific guidance on what would constitute acceptable spalling for the wide range of elements and load cases that are possible.

3. Apply supplemental thermal insulation – Addition of supplemental thermal insulation (e.g. rigid board systems or spray applied coatings) is a common approach to protect concrete from increased temperatures and avoid damage altogether, as well as mitigate spalling, particularly

in concrete-lined tunnels25. Such solutions are indeed supported by the available experimental evidence, but are likely to be conservative, to result in significant cost increases, and will require ongoing inspection and maintenance for the lifetime of the structure. This approach is rarely used in normal buildings, aside from parking garages.

4. Assume a sacrificial spalling depth – In some concrete tunnelling projects designers have assumed a sacrificial spalling depth, which is then accounted for in subsequent heat transfer and structural design calculations. The assumed spalling depth is typically ‘verified’ using large-scale furnace tests on representative structural elements. Given the stochastic nature of spalling and its demonstrated sensitivity to thermal and mechanical testing conditions, this approach must be considered as having considerable uncertainty, and also being potentially over-conservative.

It is noteworthy that if designers are sufficiently confident that explosive spalling will not occur in a credible design fire (using approaches 1, 2, or 3 above, for example), then fire engineering design of concrete structures becomes possible, since performing accurate heat transfer calculations within non-spalling concrete is relatively straightforward with available models. If, however, designers cannot confidently prevent or predict spalling, then structural design calculations for concrete structures in fire are both uncertain and rather difficult to defend. The alternative is to take conservative and/or arbitrary actions that may result in additional costs (i.e. the costs of supplementary fire insulation or sacrificial concrete cover).

NOVEL TESTING TO MITIGATE DESIGN RISK

Current approaches to heat-induced explosive spalling design can be described as ‘design by experiment’4 and/or use uncertain measures to reduce risk. ‘Design by experiment’ currently relies on large scale furnace tests which are costly, and variable. A reliable and verifiable design tool is not feasible at present, so this cannot be an immediate solution.

The spalling process is driven by a range of influencing parameters and complex thermal, mechanical, physical, and chemical processes. Some researchers6 have suggested that what is needed is a means of experimentally characterising and quantifying the propensity for spalling of different concrete mixes under various credible conditions of heating and mechanical stresses, rather than a detailed understanding of (and ability to computationally predict) spalling. Through this approach it may be possible to effectively guarantee (by experiment) that explosive spalling will not occur for a given mix in a given application. This is not just an alternative to understanding the underlying mechanisms, but a necessary precursor to properly test hypotheses and models.

Fire engineering has frequently sought to address knowledge gaps through the establishment of test methods and sets of empirical data. The success of this approach is dependent upon the ability of those test methods to characterise the key parameters affecting behaviour. The University of Edinburgh is working with Arup to address some of the key issues facing design for fire induced spalling by adapting a novel fire testing method and apparatus9,21,26

which is shown in its current form in Figure 3. The approach is to undertake tests on concrete under a range of carefully controlled conditions of thermal exposure and sample loading/restraint.

Figure 3: The current H-TRIS spalling test setup at The University of Edinburgh

Thermal ExposureThe temperature-time curves used in current structural and tunnel fire testing are notionally based on experimental data; however furnace testing is known to provide variable thermal exposures due to its approach of matching temperatures inside the furnace to a prescribed ‘standard fire’ temperature-time curve (within acceptable limits). Spalling can be particularly sensitive to this variation20. Applying the thermal load as a prescribed received thermal exposure (or incident heat flux) is potentially more consistent in that amount of energy received by each sample is directly controlled. Maluk26 developed a novel test method and apparatus named H-TRIS (Heat-Transfer Rate Inducing System), which heats prismatic concrete samples using a computer controlled array of radiant panels, and has outstanding repeatability and economy when compared to conventional furnace testing, particularly as regards spalling testing.

The H-TRIS thermal test method can produce exposures that are equivalent to the temperature-time curves currently used in furnace testing and design. Alternatively, this method is also capable of replicating realistic design fire exposures; for example determined on the basis of CFD or zone fire modelling, and thus offering greater flexibility in support of functionally based fire engineering design.

Loading and Restraint Differential thermal stress is one of the key factors known to contribute to spalling in some cases. To impose representative mechanical stress conditions during testing, the new test method replicates the pre-fire stress state by externally pre-stressing the sample using a bespoke 3 MN uniaxial loading frame. This loading frame has been developed with tunnel testing applications in mind, but different loading frames have been used to provide general insights into the spalling phenomenon – applicable to essentially all areas of construction.

As the sample is heated, the internal sample self-restraint changes but the total external stress is maintained constant in the H-TRIS method. While this may not represent the true time-dependent loading and restraint conditions in a given concrete structure, it generates a known external mechanical stress condition and allows multiple tests to be quickly and inexpensively performed under a range of mechanical conditions, to highlight the potential influence of loading on spalling risk for a given concrete mix. Some typical images from recent H-TRIS testing are shown in Figure 4. Both samples shown have the same geometry and were tested to the same thermal

exposure. A uniaxial compressive load of 10 MPa (Sample B) was observed to be sufficient to cause severe spalling after 7 minutes 49 seconds, compared to no spalling occurring during a 30 minute exposure when no load was applied (Sample A).

Figure 4: Before (top) and after (bottom) images of the heated surface of two identical samples tested using H-TRIS - (A) Unloaded, and (B) loaded to a sustained compressive stress of 10 MPa uniaxial compression

The H-TRIS test method is currently based on medium scale samples21 (500 x 500 mm in plan). At this scale, tests with H-TRIS address many of the design confidence and risk issues associated with spalling prevention. Tests are less expensive and faster, allowing more candidate mixes to be tested repeatedly, testing earlier in the project programme, and rapidly retesting to allow mix optimisation. The H-TRIS test method has been validated in terms of the thermal exposures it applies by comparison with in-depth temperatures measured in full scale furnace tests of identical samples6,27. The authors are actively seeking research partners interested in undertaking a programme of validation testing in parallel with a conventional spalling assessment involving large scale furnace tests on loaded samples (most likely involving tests on precast concrete tunnel lining segments).

THE STRUCTURAL ENGINEERS’ RESPONSIBILTY

Reinforced concrete has a good record of performance in real fires, and the information presented in this short paper should not be construed as a criticism of concrete as compared with other candidate construction materials. All construction materials present risks in fire; how the structural engineering community manages these risks will differ between materials and must necessarily change as construction materials continue to evolve, presenting new and different fire risks.

However, the available research suggests5,6 that many modern concrete mixes, used across all areas of construction, are more prone to spalling than has historically been the case. The structural engineering community therefore ought (1) to consider whether spalling is likely for a given concrete mix and credible design fire conditions for a particular design scenario, and (2) account for the potential impact of spalling on the structure’s ability to meet the agreed functional performance objectives for the structure in fire. Mitigating actions, which will depend on the particular circumstances of a given application, should be taken if necessary.

The available guidance24 on concrete spalling in fire is based in research but is not fully supported by the available experimental evidence, posing a challenge for designers. Additional research on this issue is therefore needed, particularly as advances in concrete technology continue to generate novel high performance concrete mixes for use in construction. Designers need to ensure that potential for spalling is not overlooked.

A novel experimental method and apparatus such as H-TRIS presents an opportunity to advance current approaches. It can address a number of the necessary issues, both technical and financial, in the assessment and mitigation of explosive concrete spalling. Its basis on heating

specimens by directly controlling the thermal exposure received also offers flexibility in its capability to replicate thermal exposures predicted using e.g. CFD analysis in performance-based structural design for fire.

REFERENCES

1. Bailey, C. & Khoury, G. A. Performance of Concrete Structures in Fire. (MPA - The Concrete Centre, 2011).

2. Bisby, L., Mostafaei, H. & Pimienta, P. White paper on fire resistance of concrete structures. (2014). doi:10.6028/NIST.GCR.15-983

3. Bisby, L., Gales, J. & Maluk, C. A contemporary review of large-scale non-standard structural fire testing. Fire Sci. Rev. 2, 1–27 (2013).

4. Deeny, S., Gildersleeve, C., Bisby, L. & Rickard, I. A Novel Design. World Tunneling 9–12 (2014).

5. Jansson, R. Fire Spalling of Concrete : Theoretical and Experimental Studies. (KTH Royal Institute of Technology, 2013).

6. Maluk, C. Development and Application of a Novel Test Method for Studying The Fire Baheviour of CFRP Prestressed Concrete Structural Elements. (University of Edinburgh, 2014).

7. Channel Tunnel Safety Authority. Inquiry into the fire on heavy goods vehicle shuttle 7539 on 18 November 1996. (HMSO, 1997).

8. Giuliani, L., Crosti, C. & Gentili, F. Vulnerability of bridges to fire. in Bridge Maintenance, Safety, Management, Resilience and Sustainability (eds. Biondini, F. & Frangopol, D. M.) 1565–1572 (Taylor & Francis, 2012).

9. Rickard, I., Maluk, C., Bisby, L. & Deeny, S. Predictive Testing for Heat-Induced Spalling of Concrete. in IABSE Conference Geneva, Structural Engineering: Providing Solutions to Global Challenges (ed. Brühwiler, E.) 929–936 (IABSE, 2015).

10. Zhang, H. L. & Davie, C. T. A numerical investigation of the influence of pore pressures and thermally induced stresses for spalling of concrete exposed to elevated temperatures. Fire Saf. J. 59, 102–110 (2013).

11. Harmathy, T. Z. Effect of moisture on fire endurance of building elements. ASTM STP385 (1965). doi:10.1520/STP48429S

12. Kalifa, P., Chéné, G. & Gallé, C. High-temperature behaviour of HPC with polypropylene fibres - From spalling to microstructure. Cem. Concr. Res. 31, 1487–1499 (2001).

13. Ernst and Young LLP. Economic footprint of the Channel Tunnel fixed link. (2016).14. Kelly, F. & Purkiss, J. Reinforced concrete structures in fire: a review of current rules. Struct.

Eng. 86, 33–39 (2008).15. 4th International Workshop on Concrete Spalling due to Fire Exposure. (MFPA Leipzig

GmbH, 2015).16. Hull, W. A. Fire Tests of Concrete Columns. in Proceedings of the 16th annual Convention

of the American Concrete Institute 20–45 (ACI, 1920).17. Maluk, C., Bisby, L. & Terrasi, G. P. Effects of polypropylene fibre type and dose on the

propensity for heat-induced concrete spalling. Eng. Struct. 141, 584–595 (2017).18. Haack, A. Fire protection in traffic tunnels: General aspects and results of the EUREKA

project. Tunn. Undergr. Sp. Technol. 13, 377–381 (1998).19. Ministère de l’Equipement. Circulaire interministérielle no 2000-63 du 25 Aout 2000 relative

à la sécurité dans les tunnels du réseau routier français. (2000).20. Maluk, C., Terrasi, G. Pietro, Bisby, L., Stutz, A. & Hugi, E. Fire resistance tests on thin

CFRP prestressed concrete slabs. Constr. Build. Mater. 101, 558–571 (2015).21. Rickard, I., Bisby, L., Deeny, S. & Maluk, C. Predictive Testing for Heat Induced Spalling of

Concrete Tunnels - The Influence of Mechanical Loading. in Structures in Fire - Proceedings of the Ninth International Conference (eds. Garlock, M. & Kodur, V. K. R.) 217–224 (DEStech Publications, Inc., 2016).

22. Gawin, D., Pesavento, F. & Schrefler, B. A. What physical phenomena can be neglected when modelling concrete at high temperature? A comparative study. Part 1: Physical phenomena and mathematical model. (2011). doi:10.1016/j.ijsolstr.2011.03.004

23. Dwaikat, M. B. & Kodur, V. K. R. Fire Induced Spalling in High Strength Concrete Beams.

Fire Technol. 46, 251 (2009).24. British Standards Institution. BS EN 1992-1-2 : 2004 Eurocode 2 : Design of concrete

structures. (2010).25. NFPA. NFPA 502, Standard for Road Tunnels, Bridges, and Other Limited Access

Highways. (NFPA, 2011).26. Maluk, C., Bisby, L., Krajcovic, M. & Torero, J. L. A Heat-Transfer Rate Inducing System (H-

TRIS) Test Method. Fire Saf. J. 1–13 (2016). doi:10.1016/j.firesaf.2016.05.00127. Rickard, I. et al. Development of a Novel Small-Scale Test method to Investigate Heat-

Induced Spalling of Concrete Tunnel Linings. in 4th International Workshop on Concrete Spalling Due to Fire Exposure (ed. Dehn, F.) 195–205 (MFPA Leipzig GmbH, 2015).