Proceedings of the 19 th International Conference on Soil Mechanics and Geotechnical Engineering, Seoul 2017 1 Predicting the behaviour of non-circular, curved-in-plan retaining walls using the trial load method Prédire le comportement des murs de soutènement non-circulaires incurvés en utilisant la méthode de la charge d’essai (trial load method) Daniel Gilmore* & Raul Fuentes** *Sellafield LTD, United Kingdom, Daniel.Gilmore@sellafieldsites.com ** University of Leeds, United Kingdom. ABSTRACT: The trial load method is a long established method to predict the behaviour, and design of arch dams. This paper presents the first implementation of this method to predict the behaviour of non-circular, curved-in-plan, deep excavations. The results of the method were compared to a curved excavation, observed in literature and results of models in PLAXIS 3D. The results show that the trial load method can be successfully used to predict the behaviour of curved walls, and provides insights on further research to validate its use more widely for design purposes and its potential to be applied to the design of typical (i.e. straight) retaining walls. RÉSUMÉ: La méthode de charge d'essai (trial load method) est une méthode établie depuis longtemps pour prédire le comportement et la conception des barrages d'arc. Cet article présente la première mise en œuvre de cette méthode pour prédire le comportement des excavations profondes non circulaires, courbées dans le plan. Les résultats de la méthode ont été comparés à une excavation courbe, observée dans la littérature et les résultats des modèles dans PLAXIS 3D. Les résultats montrent que la méthode de la charge expérimentale peut être utilisée avec succès pour prédire le comportement des murs incurvés et fournit des informations sur des recherches plus poussées pour valider son utilisation plus largement à des fins de conception et son potentiel à appliquer à la conception d ' Murs de soutènement. KEYWORDS: Retaining wall, PLAXIS 3D, Numerical method 1 INTRODUCTION A designer can use a variety of different methods to design retaining walls. These mainly fall under two categories: numerical methods using computer programmes and analytical methods such as Terzaghi or Peck design charts and other limit equilibrium solutions as described for instance in CIRIA C580 (Anderson, 2012). Both methods have their advantages and disadvantages. Analytical methods are easy to analyse but give conservative designs (evidenced by the excessively high stiffness in cantilever retaining walls resulting in very small displacements) (Long, 2001). Numerical methods are able to model complex loading and excavation sequences but the inputs and analysis options in the software need to be carefully and correctly applied or the consequences can either lead to very conservative designs (O’Brien, 2010) or, in extreme situations, even collapse (Whittle and Davies, 2006). The design of retaining walls using analytical methods mainly relies on either 2D section designs (e.g. Coulomb’s or Rankine’s theory) or on empirical evidence (e.g. Clough and O’Rourke (1990) design charts). These design methods whilst having some advantages as stated above also have a few disadvantages such as not incorporating the effect of the horizontal stiffness of the retaining wall. This disadvantage can lead to very conservative designs as shown by Long (2001) resulting in higher material usage, higher economic and environmental cost and wasted personnel hours. Its use for more innovative designs of the plan of the retaining wall such as the “peanut” shaped retaining walls is impossible. For such shapes, designers need to resort to using numerical methods (Puller et al, 2015). Having an analytical method or process which includes the additional stiffness of the horizontal dimension of the retaining wall has the potential to allow for more economic design whilst still using simple calculations. A non-circular, curved-in-plan retaining wall (referred to as an arched retaining wall) is a proposed variation on the use of straight retaining walls (as shown in Figure 1) for deep excavations. Such shape helps to avoid/reduce the amount of props or excavation supports required by using the theory of arches to distribute the load from the centre of the wall to the corners of the excavations: hence reducing the displacements. This reduction in displacements and props has a series of benefits, including; economic, production rates, material use and health and safety (Gilmore, 2015). Research on this topic has been focused on diaphragm walls as done by the author (Gilmore, 2015 and 2016), and on arched secant pile walls by Yi-ping and Tu-qiao (2000). This paper will present a new analytical method for designing curved-in-plan cantilever retaining walls using a method which has been used extensively to design arch dams (The trial load method). Case study data are used to show its effectiveness. It is thought that this method can be used as both a stand-alone design method and also as a checking tool for numerical models. 2 TRIAL LOAD METHODOLOGY The trial load method is to estimate the behaviour of an arched dam by using the relative stiffness of the vertical and horizontal sections to predict its behaviour (Fanelli, 1999). The method divides the arch dam into vertical (or inclined members) and horizontal members as shown in Figure 2 (which for an arched retaining wall are a cantilever and arch beam respectively). The deflections of each of the members are then calculated in isolation, adjusting the applied loading (shown in Figures 3 and 4) in iterative manner until the displacement at points where the vertical and horizontal sections cross are the same. Within the trial load method there are also different forms of analysis each with their own levels of accuracy and complexity. These are: the Crown-Cantilever Analysis, Radial Deflection Analysis and Complete Adjustment Analysis. These methods increase in accuracy and also in complexity (i.e. the Crown-Cantilever analysis