Comparison of finite element and finite difference modelling results with measured performance of a reinforced soil wall Damians I.P. 1 , Bathurst R.J. 2 , Josa A. 1 and Lloret A. 1 1 School of Civil Engineering, Universitat Politècnica de Catalunya - BarcelonaTech, Barcelona, Spain 2 Royal Military College of Canada, Kingston, Ontario, Canada ABSTRACT The performance of a metallic reinforced soil segmental retaining wall is examined in a systematic manner using two numerical modelling approaches. Finite element modelling was carried out using the commercial program PLAXIS and finite difference modelling using the commercial program FLAC. Numerical model results using both approaches were compared against measurements recorded for a well-instrumented full-scale 3.6-m high wall constructed with a sand backfill, modular-block facing, and steel reinforcement (welded wire mesh). This paper presents measured and predicted toe loads, facing displacements, and reinforcement connection loads at end of construction and during subsequent staged surcharge loading approaching failure. Both numerical models have been verified against recorded measurements. The sensitivity of the assigned backfill soil friction angle on the magnitude and distribution of reinforcement connection loads is also examined. The paper concludes with a summary of lessons learned to achieve satisfactory agreement between predicted performance and wall measurements using both modelling approaches. RÉSUMÉ La performance d’un mur de soutènement en sol renforcé par bandes métalliques est examinée de manière systématique en utilisant deux approches de modélisation numérique. La modélisation par éléments finis a été réalisée en utilisant le programme commercial (PLAXIS) et la modélisation par différences finies en utilisant le programme commercial (FLAC). Les résultats des modèles numériques, utilisant les deux approches, ont été comparées avec les mesures enregistrées pour un mur d’échelle réelle de hauteur de 3,6 m bien instrumenté construit avec un remblai de sable, une paroi de blocs modulaires et une armature en acier (treillis soudé). Cet article présente les charges mesurées et prévues au pied de la fondation, les déplacements de la paroi et les charges de connexion du renforcement à la fin de la construction et durant l’application par paliers des surcharges subséquentes approchant la rupture. Les deux modèles numériques ont été vérifiés par rapport aux mesures enregistrées. La sensibilité des propriétés du sol de remblai sur la magnitude et la distribution des charges de renforcement est également examinée. L’article se termine par un résumé des leçons apprises pour parvenir à un accord satisfaisant entre la performance prévue et les mesures prises sur le mur en utilisant les deux approches de modélisation. 1 INTRODUCTION Mechanically stabilized earth (MSE) walls (reinforced soil walls) are complex mechanical systems. They are comprised of a hard concrete facing in most cases, granular soil backfill and polymeric or steel reinforcement layers. Many walls today are constructed with a dry- stacked modular concrete block facing. Current design methods give recommendations for the internal stability design of these systems so that reinforcement layers do not fail or deform excessively within the soil backfill or at the connections (FHWA 2009). These design methods are based on closed-form solutions adapted from conventional earth pressure theory. For geosynthetic (polymeric) reinforced soil walls there is evidence that these design methods are very conservative (Bathurst et al. 2008). For metallic reinforced soil systems the conservatism is not as great because load equations have been empirically adjusted using measured loads from instrumented walls (Allen et al. 2004). Regardless of the reinforcement type, current design methods are most applicable for walls with simple configurations and competent foundation conditions. For walls falling outside of design guideline envelopes, numerical modelling may be the only alternative for design. However, there are a number of questions that must be addressed when using numerical models for design: 1. Has the numerical model been verified against one or more instrumented structures (Carter et al. 2000)? 2. Are project-specific material properties available? 3. Will different numerical modelling packages give different design outcomes and, if so, are the differences important? The current investigation is focused on the last question. Two widely used numerical modelling software packages that use different numerical solution schemes are used to predict performance features of a wire mesh (metallic) reinforced soil wall. One software package is the program FLAC which uses the finite difference method (FDM) solution scheme and the other is the PLAXIS software package that uses the finite element method (FEM). The 2D simulation results are compared to measurements taken from the well-instrumented full-scale experiment noted above and previously reported in the literature.