ABSTRACT
Abstract This paper adresses the interest and relevance of large model tests, according to the LCPC experience at the Structures Laboratory, in improving the design of concrete structures or helping to solve field structural problems. Large scale tests should neither be a substitute to numerical modelling, nor only a simple validation of material research. They should be used at three steps of the innovation process. First, to identify the major mechanism which limits the structural capacity, then to validate numerical modelling (and/or simplified design method) and the adapted material characterization, and finally, to assess the improved global solution. Especially for new materials or new structural concepts, the first point must be remembered, since it helps to take account of major features of concrete structures : size effects and physical origins of nonlinearities. Keywords: Concrete, numerical modelling.
