Articles Amir M. Alani* Morteza Aboutalebi DOI: 10.1002/suco.201100043 Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs This paper confirms that the structural behaviour of groundsupported slabs is a non-linear function of the structural properties of slabs as well as the supporting soil. The findings reported emphasize that the suggested equations used in design codes pay insufficient attention to the effect of the supporting ground stiffness within the context of the mechanical behaviour of slabs as far as ductility is concerned. The results presented demonstrate that ground stiffness has a significant effect on the ductility of ground-supported slabs. It also pays particular attention to the possibility of determining the ductility limit of slabs. Keywords: stiffness, ground slab, failure, ductility, behaviour 1 Introduction The increase in the capacity of a structure is not essentially equivalent to improved structural behaviour. For around two decades, the leading engineering codes have guided designers towards more ductile design procedures. These procedures do not always end in structures with a greater loadbearing capacity, but may well end up with the structure’s energy-dissipating ability being increased. Almost all industries use ground slabs as a vital base for production lines or storage areas. The vast areas covered by ground-supported slabs are usually subjected to both predicted and unpredicted stationary and moving loads. These mostly transverse loads on slabs may cause failure in the form of settlement, punching shear, shear fracture, bending failure or bearing crush. In the UK, the Concrete Society has published the 3rd edition of the TR34 report [1], which is currently under review. It has been noted both in this report and in the National Cooperative Highway Research Program (NCHRP) Report 372 [2] “Support under Portland cement concrete pavements” that the subgrade stiffness has a minor effect on the required slab thickness and the flexural stresses built up by the transverse loads. This conclusion in both references is based on the yield-line method analysis in which the effect of the support reaction has not been considered. Although in most construction cases engineering the subgrade layer is not considered an economically feasible * Corresponding author: m.alani@gre.ac.uk, am50@gre.ac.uk Submitted for review: 01 September 2011 Revised: 10 February 2012 Accepted for publication: 21 February 2012 102 option, the sub-base layer is often compacted to a certain degree. Potentially, the subgrade stiffness can be altered to the desired value by some compaction efforts. In this, the modulus of subgrade reaction, as an independent factor, can be considered as a design parameter in most practical cases. The issue of bearing stress distribution in the soil supporting the ground slab was introduced by Westergaard more than 80 years ago [3]. Westergaard assumed the slab to be a homogenous, isotropic elastic solid sustained by the reactions from an elastic subgrade which are only vertical and proportional to the deflections of the slab. He introduced the term “modulus of subgrade reaction k”, which represents the behaviour of the subgrade based on the load per unit area causing unit deflection. This behaviour is based in principle on the Winkler model, which is traditionally preferred for slab-on-ground design. In this research, the authors have based the soil representing Winkler springs on the elastic isotropic constitutive model of behaviour in the same way as Westergaard proposed. Researchers worldwide have reported different results based on both experimental and theoretical studies into the effect of the k factor on the first crack and final bearing capacity of ground-supported slabs [4, 5]. Although the design codes of practice commonly used, e.g. the 3rd edition of the TR34 report [1] and the 1995 NCHRP Report 372, disregard the effect of the supporting soil stiffness variation, some researchers do consider it to be a significant influencing factor. Moreover, there is no wide-reaching agreement about the trend of the effects of changing the modulus of resilience of the soil layer. Irving, as part of his work, reported that “increasing the compaction effort gives a slab an increased first cracking load and an increased failure load” [4]. He also concluded from his experimental studies that “an increase in k factor results in improved post-cracking ground slab behaviour”. Opposing that, Kearsley has concluded that the increase in ductility of fibre-reinforced concrete ground slabs is more remarkable if the supporting soil is not so stiff [5]. Regarding these rather contradictory reports and similar beliefs within the community, the authors of this paper recognized that there is ample scope to investigate the state of the effect of subgrade and sub-base layers on the behaviour of ground slabs from the ductility point of view. © 2012 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Structural Concrete 13 (2012), No. 2 A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs The subject of “distribution of the bearing stresses in supporting soil media beneath a slab under vertical load” has been extensively investigated both theoretically and experimentally [6–13]. The most important shortcoming of the investigations was the presumed elastic behaviour of the slab prior to Meyerhof’s reported work. He studied the phenomenon of the concrete slabs and their ability to experience plastic behaviour under higher loading conditions. Meyerhof concluded that the theory developed would lead to considerably more economic slab design. The distinctive feature of this work was that the equations drawn up were based on equating the reduction in potential energy as the load was imposed to the amount of strain energy absorbed by the formation of cracks. The strength and deformation behaviour of plain and steel fibre-reinforced concrete ground slabs has also been investigated by Falkner et. al [8]. In that study, experimental results for two plain concrete and four steel fibre-reinforced concrete ground slabs under central loading were used to calibrate and validate the finite element (FE) models presented. Based on those results, a rationalized design principle applied to plain and steel fibre-reinforced concrete ground slabs was introduced. A study by Meda et. al [9] on the fracture behaviour of steel fibre-reinforced concrete ground slabs was performed by using non-linear fracture mechanics FE analyses. In this study, the non-linear fracture mechanics FE analyses compliantly predicted experimental results for the crack development, ultimate load and collapse mechanism of the ground-supported slabs. Chou applied the finite element method (FEM) to demonstrate that the slab panel size has a significant effect on the build-up of stresses and also the settlement of the slab [7]. Further, Shentu demonstrated in another study that the subgrade reaction under a circular-loaded slab has a non-linear inverse relationship with the extent of the slab [10]. Tests of slabs have also been conducted by Hadi in which the use of synthetic fibres has once again been compared to steel fibres using slab specimens of 820 × 820 mm supported at four corners and subjected to a central point load [11]. In addition, numerous experimental investigations, particularly in the UK, have led to a much clearer understanding of the mechanical behaviour of slabs. In the late 1980s, tests were carried out by Beckett establishing the effect of steel fibre type and size within this domain [12, 13]. It was established by Beckett that the aspect ratio of the fibres played an integral part in the load at which first cracking occurred in slabs. Subsequently, the most recent investigations have been performed (and are being performed) at the University of Greenwich by Alani et al, in which tests on larger slabs (6 × 6 m) have led to a better understanding of the mechanical behaviour of slabs. Early results have revealed that the effect of corner uplift can be resolved by considering large-size slabs. As emphasized before, this paper is a part of wider studies that Alani and Beckett are currently carrying out at the University of Greenwich. These long-term, substantial studies of large ground-supported slabs are being carried out in specially developed, state-of-the-art facilities dedicated to this field. These studies have concentrated on size and types of reinforcement used (steel, synthetic and fabric fibres as well as Fig. 1. Ground slab test facilities at the University of Greenwich plain concrete) for ground-supported slabs for industrial applications. Fig. 1 illustrates certain aspects of a steel fibre-reinforced concrete slab under investigation. The subject of ground-supported slabs can be considered within the context of three interlinked elements: the supporting soil, the slab and their mutual effects on the failure phenomenon. A recent paper by Tchrakian highlights the inadequacies of investigations into the structural performance of ground-supported slabs when the interaction between the slab and the subgrade is disregarded or oversimplified [14]. This finding supports the idea of the inclusion of the supporting soil characteristics as well as the slab properties in any in-depth study of the subject in the future. The equations by Meyerhof [6] are empirical in nature and since the innovation of FEM many researchers have set upon the task of predicting the plastic response of slabs more accurately. The development of versatile and powerful FE packages such as ANSYS and improvements in computing hardware such as computers with multiple CPUs has made it possible to investigate such phenomena in more detail and with more precision. The following section presents the assumptions and the techniques used to develop the FE models with the help of ANSYS 12.1. 2 Theoretical background A concrete slab resting on an elastic subgrade can be represented by an elastic slab bearing on a dense liquid or elastic solid by the conventional theory of elasticity [15]. The theory of elasticity predicts the behaviour of these slabs for loads at or below initial yield. The theory of elasticity becomes less accurate for higher loads [6]. For that reason, a plastic approach might be a feasible option for analytical purposes. The yield-line theory can also be adopted in order to evaluate the ultimate bearing capacity of reinforced concrete slabs, which is based on assuming a rigid-plastic slab resting on an elastic subgrade. The case studied in this paper concerns a load applied at the centre of a flat slab resting on the ground. The mode of failure considered is a circular negative failure on the top surface and positive radial failure lines from the Structural Concrete 13 (2012), No. 2 103 A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs load application point to the perimeter of that circle on the bottom surface. Fig. 2 illustrates this case more clearly. The governing equation for the vertical deflection w(x, y) of a plate subjected to a point load P at the centre of the slab and a spread support reaction q(x, y) is given by Timoshenko [16]: ( ) ( ) ( ) D∇ 2∇ 2w x, y = Pδ 0 , 0 − q x, y (1) In this equation, δ is the Dirac function, which is equal to zero apart from in the vicinity of the origin of the system of coordinates where the load is applied. Based on this differential equation, the displacement of a fixed circular plate of radius r with a central concentrated load P is calculated by Timoshenko [16] as: ∆= Pr 2 16π D (2) where: D= Eh3 12 1 − v 2 ( ) (3) and h stands for the slab thickness and v is Poisson’s ratio. The elastic stiffness is therefore equal to: KS = 16π D r2 (4) In this case the collapse load without soil support PCo is found to be: PCo = ( ) 2π M + M′ (5) 2a 1− 3r where M and M' are the positive and negative moment capacities respectively, and a is the radius of the load application area [17]. It is also possible to prove that the collapse load of the soil support PP with soil stiffness K is equal to [15]: P= ( ) 2π M + M′ 4 2a γ Kπ r 1− 1 + 2a 3r 3D 1 − 3r (6) This in turn demonstrates the considerable effect of the soil on the ultimate capacity of the ground-supported slab. As mentioned before, based on this equation, the loadbearing capacity is a non-linear function of the slab and the supporting soil properties. It is shown later that the ductility of the slab is not only a function of the slab properties but also depends on the supporting soil condition. It has been proved that as slab plan area increases, settlement decreases due to increased bending action, and the built-up stresses rise as well [7]. It is interesting to note that in Chou’s investigation, a limit has been found for the 104 Fig. 2. Cone-shaped failure mechanism Structural Concrete 13 (2012), No. 2 slab plan size at which this relationship breaks down. Two distinct types of behaviour were observed by Chou in his study: a) “large slab effect”, and b) “small slab effect”. In his studies of large slabs, slab plan size was noted to have limited effect on the variation of resulting stresses and deflections under loading conditions. A thorough study of how different factors such as slab size, modulus of subgrade resilience and rebar ratio affect the behaviour of the slab under different types of failure modes, e.g. ductile flexural and/or brittle punching shear failure, is under investigation by the authors of this paper. However, the focus here is on a better understanding of how modulus of subgrade reaction affects the failure characteristics of ground-supported slabs. Although when studying the behaviour of ground slabs it is traditionally accepted that the first yield line mechanism demonstrates the ultimate load of the system [17], if the failure mode tends to become a more ductile one, then there is a strong possibility that the first crack line becomes tolerable when the load is not permanent. Moreover, the idea of rehabilitating the slabs under a ductile failure condition before they are permanently damaged will save considerable effort and resources, whereas the brittle modes of failure do not allow such attributes to be exercised. 3 Finite element models It was considered that the use of finite element (FE) methods is a viable way of studying how subgrade stiffness affects the ductility of ground-supported slabs. No doubt this approach supersedes any form of practical and experimental study which will require a relatively substantial amount of resources and time. The non-linear capabilities of ANSYS for modelling crack formation and patterns of concrete structures/elements make it particularly suitable for investigating the ductility of ground-supported slabs. It also caters for possible elimination of tension causing upward deflection of the edges. The FE models in this study consisted of 3 × 3 m square panels 20 cm deep with 5 cm long Link10 elements representing the elastic soil based on the Westergaard theory and the Winkler spring method [3]. In order to achieve A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs Fig. 4. Loading pattern on top surface of slab Fig. 3. Solid and link elements in FE model consistent results, the no-tension property of the Link10 element was used. Fig. 3 shows the slab and the link elements representing the supporting soil. Solid65 concrete solid elements capable of modelling the reinforced concrete slabs have been used, for which the concrete material has been modelled as precisely as possible. The Solid65 element is capable of modelling cracking in tension and crushing in compression. In the ANSYS 12.1 models, the failure criteria of the concrete were described via the Willam and Warnke model [18] given the following constants: – the shear transfer coefficient was set for an open and a closed crack with values of 0.2 and 0.8 respectively – uniaxial crushing stress ( ƒc) was set to 20 MPa – uniaxial tensile cracking stress was set to 2 MPa The above constitutive model is capable of considering the tensile cracking of concrete. The slab modelled in this study consisted of several layers of elements in the FE models. This will allow cracks to develop in the most vulnerable layer with respect to the applied load which in turn can spread to the other layers. The programme developed was written such that the effect of cracks on the stiffness of the slab is taken to account according to each load increment. This was achieved by modifying the stiffness matrix in each previous loading condition within the yielded solution. It is also useful to note that the reduced ability of concrete to transfer shear across cracks is considered in ANSYS. The reinforcement was defined in three perpendicular directions for which a 0.001 rebar ratio was assumed for all directions based on the random reinforcement alignment for steel fibres. The reinforcing material is a multi-linear representation of ordinary rebar material. The load was applied to a set of nine central nodes (the load applied on the central node was twice that on the other eight nodes) at the upper slab surface as shown in Fig. 4. Spreading the load over the nine central nodes is understood to increase the chance of convergence of the FE models. By preventing a crush failure in the centre of the slab, this would allow a better understanding of the to- tal effects within the slab rather than restricting the final step to the effects of local crushing under the applied load. The amount of load applied to the slab was far greater than the anticipated capacity of the slab, so cracking and/or crushing is guaranteed before the analysis finishes. This way, the number of results saved was increased to 20 000 prior to the termination of the analyses. The supports of the system were placed at the bottom of each link element separately to confine the movement in all three perpendicular main directions. It should be noted that the stiffness produced by each of the Link10 elements is the function of three factors: the modulus of elasticity of the material, the cross-sectional area and the inverse of their length. In order to calculate the resulting stiffness of the subgrade, this should be multiplied by the number of link elements in each unit area. In this particular work, the variation in the modulus of subgrade reaction was controlled by the area of the link elements and the other factors were kept constant. The soil was assumed to range from a very soft non-impact layer to a theoretically almost infinitely stiff one. A total of 25 different FE models were analysed, and the most important numerical results are presented. Loss of contact between the slab and the soil is accounted for by using the gap property of the Link10 element of ANSYS. If the no-tension contact between soil and structure is not modelled, the unreal tension between the soil and the slab in the vast areas where the deflection of the slab is upward will cause considerable moments and shears in the slab and consequently the structural results will be unreliable. 4 Numerical results The structural response factors monitored in the FE models consist of the deflections, stresses and cracking pattern representative of the plastic energy dissipation. To validate the FE results, as implied before, the experimental data for 6 × 6 × 0.15 m slabs as the result of a separate study was used. To that effect, a similar FE model with the same dimensions as in the test was developed and the deflections of the slab in the test were compared with those of the FE model. The comparison between results is shown in Fig. 5. The results demonstrate a satisfac- Structural Concrete 13 (2012), No. 2 105 A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs Table 1. Effect of the support stiffness on the structural response of the slabs Fig. 5. Comparative experimental and FE modelling results for the deflection of a 6 × 6 m slab under central load k (N/mm3) Bearing capacity (kN) Extent of cracked zone (m) 0.1 111 0.6 0.08 102 0.9 0.06 88 1.8 0.0575 85 1.8 0.055 83 2.1 0.05 77 2.1 0.0475 74 2.1 0.045 71 2.1 0.04 66 2.4 0.035 63 2.4 0.03 61 2.4 0.025 55 2.4 0.02 50 2.7 0.0196 50 2.7 0.0185 49 2.7 0.0175 47 3 tory degree of compatibility between the experimental and the FE results. The short-term settlement of the slab is inversely proportional to the modulus of subgrade reaction. As shown in Fig. 6, the maximum deflection of the slab under a specific amount of load at the centre of the slab increases almost linearly as the amount of subgrade resilience modulus decreases. It is an obvious and predictable result, but what is very important and must be emphasized is the rate of change of deflection, especially in the central region of the slab. As seen here, the gradient of deflection is directly related to the stiffness of the subgrade. This leads to the belief that as the stiffness of the soil increases, the main resisting part of the slab is more and more confined to a smaller central region of the slab. These results could lead to two major consequences. Firstly, increasing the stiffness of the sub-base material causes a more locally activated flexural region in the slab which will decrease the slab’s energy dissipation capability. Secondly, it will increase the crack width in the central part of the slab, which can subsequently increase the possibility of rebar deterioration and excessive crack opening. The concentration of the flexural behaviour of the slab in the central region is also evident tracing the cracking zone of the slab as subgrade resilience modulus is increased. Fig. 7 illustrates the cracked zones of the slab at the ultimate loading condition for selected k values. It should be noted that the amount of the ultimate load is not the same for these cases. This will be discussed later. The confinement of the cracked zone as the k value increases is an endorsement of the disadvantage of a higher value of subgrade stiffness claimed previously. (a) (b) 0.015 46 3 0.0125 44 3 0.01 0.005 41 37 3 3 Fig. 6. Deflection of a 3 × 3 m slab under similar loads: (a) for k = 0.020 N/mm3, (b) for k = 0.080 N/mm3 106 Structural Concrete 13 (2012), No. 2 A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs (a) for k = 0.020 N/mm3 (b) for k = 0.040 N/mm3 (c) for k = 0.0475 N/mm3 (d) for k = 0.060 N/mm3 Fig. 7. Crack formation in a 3 × 3 m slab under similar loads with different k values Table 1 indicates the results of the finite element models. As shown in Fig. 8, the extent of the plastic zone, where most of the input energy is dissipated by plastic deformation of the cracks, decreases as the modulus of resilience of the supporting soil increases. This very important result has never been experimentally investigated because setting up experimental models for different soil types with various controlled sub-base stiffness conditions is not only very expensive but also very time-consuming. Furthermore, Fig. 9 demonstrates that the loadbearing capacity of the slab is a non-linear function of the soil stiffness. The ultimate slab capacity increases as the k factor increases, but it is not linearly proportional, and the increase in capacity is not considerable with respect to the increase in k. It has also been observed that the capacity of the slab increases to some extent as the stiffness of the supporting soil layer increases. Basically, structural engineers are required to provide the minimum essential value of resilience modulus of the subgrade and sub-base layers within the context of ground-supported slabs. They are also required to provide the predicted short- and long-term settlements and bearing capacity as well as practical prerequisites such as the integrity of the sub-base under construction traffic. To that effect, the findings of this investi- Fig. 8. Extent of the cracked zone plotted against soil stiffness Fig. 9. Ultimate plastic bearing capacity of slab plotted against soil stiffness Structural Concrete 13 (2012), No. 2 107 A. M. Alani/M. Aboutalebi · Analysis of the subgrade stiffness effect on the behaviour of ground-supported concrete slabs gation suggest that it would be useful to consider that using stiffer soils under the ground-supported slabs can reduce the area of the slab in which the maximum stresses are built up and the energy is dissipated by cracking and crushing. Hence, the increase in the modulus of subgrade stiffness does not essentially improve the behaviour of the slab. 5 Conclusions No particular specifications are presented in the widely available design codes for the modulus of resilience of the subgrade reaction and hence ground slabs could be designed regardless of the subgrade soil characteristics. This paper sets out to investigate how subsoils with different stiffness values affect the mechanical behaviour of ground-supported slabs. This aim was fulfilled by a set of finite element (FE) analyses carried out by applying the ANSYS 12.1 package to 3 × 3 × 0.2 m slabs made up of Solid65 elements with a 0.001 volumetric rebar ratio and Link10 supporting elements which represented the supporting soil layer with various stiffness values. This work demonstrated that the increase in the modulus of resilience will increase the rate of deflection of the slab in a smaller area under the applied point load with an overall decrease in the maximum deflection of the whole slab. This subsequently leads to the area in which concrete cracking occurs being confined to a smaller zone, which in turn can be interpreted as restricting the area affected by the energy dissipation. This will cause both a decrease in the ductile behaviour and an increase in the crack width and depth of the slab. It can be concluded that although a minimum value of ground stiffness is necessary to provide the required capacity, greater stiffness will avoid a desirable ductile behaviour which besieges a considerable portion of the slab in the load-conveyance process. Therefore, it is of paramount importance to stress the fact that the effect of the modulus of resilience on the behaviour of ground slabs cannot be ignored from a ductility point of view, which contrasts with the predictions of the design codes commonly used. It is considered that the findings of this paper could potentially be used in the design process of any groundsupported concrete slabs in order to provide the minimum soil stiffness required to avoid excessive deflections. It can be recommended that in designing ground-supported concrete slabs, designers establish a minimum value for the subgrade modulus of resilience, based on the deflection and bearing capacity requirements, and then design the slab. This will no doubt provide sufficient information to engineer the soil accordingly. 3. Westergaard, H. M.: Computation of stresses in concrete roads. Proceedings of 5th annual meeting of Highway Research Board, vol. 5, part 1, 1925. pp. 90–112. 4. Irving, J.: Soil-structure interaction of fibre-reinforced concrete floor slabs on grade. MSc dissertation, University of New Brunswick, Canada, 1999. pp. 108. 5. Kearsley, E. P., Elsaigh, W.: Effect of ductility on load-carrying capacity of steel fibre-reinforced concrete ground slabs. Journal of the South African Institute of Civil Engineering, vol. 45, No. 1, 2003. pp. 25–30. 6. Meyerhof, G. G.: Load-carrying capacity of concrete pavements. Journal of Soil Mechanics and Foundations, vol. 88, No. 3. 1962. 7. Chou, Y. T.: Stress analysis of small concrete slabs on grade. Journal of Transportation Engineering, vol. 110, No. 5. 1984. 8. Falkner, H., Huang, Z., Teutsch, M.: Comparative study of plain and steel fiber-reinforced concrete ground slabs, Concrete International, vol. 17, No. 1. pp. 45–51. 1995. 9. Meda, A., Plizzari, G. A.: New design approach for steel fiberreinforced concrete slabs-on-ground based on fracture mechanics, ACI Structural Journal, vol. 101, No. 3, 2004, pp. 298–303. 10. Shentu, L., Jiang, D., Hsu, C. T.: Load-carrying capacity for concrete slabs on grade. Journal of Structural Engineering, vol. 102, No. 1. 1997. 11. Hadi, M.: An investigation into the behaviour of steel and polypropylene fibre-reinforced concrete slabs. University of Dundee, 2008. 12. Beckett, D.: Comparative tests on plain, fabric reinforced and steel fibre-reinforced concrete ground slabs. Concrete, vol. 24, No. 3. 1989. 13. Beckett, D.: Thickness design of concrete industrial ground floors. Concrete, vol. 29, No. 4. 1995. 14. Tchrakian, T., O’Dwyer, D., West, R. P.: Load spreading and moment distribution in fibre-reinforced slabs on grade. Trinity College Dublin, Ireland, 2008. 15. Baumann, R. A., Weisgerber, F. E.: Yield-line analysis of slabs-on-grade. Journal of Structural Engineering, vol. 109, No. 7, Jul 1983, pp. 1553–1568. 16. Timoshenko, S., Woinowsky-Krieger, S.: Theory of Plates and Shells, 2nd ed., McGraw-Hill Book Co., New York, N.Y., 1959. 17. Johansen, K. W.: Yield-Line Theory, William Clowes & Sons Ltd., London, 1962. 18. Willam, K. J., Warnke, E. P.: Constitutive models for the triaxial behaviour of concrete. Proceedings of the International Association for Bridge and Structural Engineering, vol. 19, 1975. References 1. The Concrete Society, Technical Report 34, Concrete industrial ground floor slabs – a guide to their design and construction, 3rd ed., 2003. 2. Darter, M. I., Hall, K. T., Kuo, C. M.: Support under Portland cement concrete pavements. Transportation Research Board, Washington D.C., 1995. NCHRP Report 372, p. 50. 108 Structural Concrete 13 (2012), No. 2 Prof. Amir M. Alani Head of Dep. of Civil Engineering and the Bridge Wardens’ Chair in Bridge & Tunnel Engineering, University of Greenwich, Chatham, UK m.alani@gre.ac.uk Morteza Aboutalebi Post-doctoral Fellow Department of Civil Engineering University of Greenwich, Chatham, UK aboutalebi@greenwich.ac.uk