14th International Benchmark Workshop on Numerical Analysis of Dams Stockholm, Sweden, September 6-8, 2017 A systematic assessment of the influence of geometry and materials properties on the performance of arch dams Salazar F. 1 , Toledo, M.Á. 2 , Vicente D.J. 1 1 International Center for Numerical Methods in Engineering (CIMNE), Barcelona, SPAIN 2 Civil Engineering Department: Hydraulics, Energy and Environment, Technical University of Madrid (UPM), Madrid, SPAIN E-mail: fsalazar@cimne.upc.edu ABSTRACT: Arch dams have different properties that play a relevant role in their behavior, although it is not clear to what degree or in what sense. There is some consensus regarding the relevance of certain factors such as length at crown, height, base and crest thickness, or Young modulus of dam and foundation. However, others such as the shape of arcs and cantilevers, which are correlated and whose effect is more difficult to consider, can also be influential. In this work, a systematic study of the response of arch dams in front of the common loading scenarios has been carried out, taking into account the usual range of variation of their properties. In total, 39 input variables related to geometry, material strength and thermal load were considered. Ranges of variation for each of these parameters have been defined according to the usual design criteria and 3,000 different geometries – together with the corresponding FEM models - have been generated with random values of these parameters. The resulting displacements and stresses have been used to fit prediction models based on a machine learning technique named ‘random forests’ that give an estimate of the dam response. The interpretation of these models can be associated with the relative importance of the characteristics of arch dams on each of the behavior variables. 1 Introduction The first filling and the initial stage of dam operation are critical periods in terms of dam safety; the new loads applied by the presence of the reservoir frequently induce a transient behavior [1]. This is more acute for arch dams, since they transmit higher loads to the foundation and abutments. Moreover, the joints between cantilevers are grouted at that time, hence the structure becomes monolithic and hyperstatic. Data-based models cannot be applied properly during the first years of operation due to the lack of monitoring data for model fitting [2]. Numerical models are sometimes available, though they need to be calibrated, for which behavior data are also necessary. As a result, dam safety assessment during that period is mostly based on engineering knowledge and experience from similar cases. This approach is intrinsically subjective and biased by the particular know-how of each practitioner. Moreover, each dam features different properties whose influence on the relevant outcome indicators is not fully understood. It is generally acknowledged that several factors are relevant on dam behavior, such as the crest length/height ratio [3], or that between the elastic modulus of dam and foundation [4]. According to the USBR [5], if the ratio base thickness/height is lower than 0.2, the thermal load is preponderant. However, other variables, whose effect is harder to consider, can also be influential. In addition, some of them are obviously correlated. It can be concluded that the effect of each parameter of the dam on each of the response variables is difficult to determine a priori.