Physical modelling of RC structures through digitally manufactured small-scale specimens for centrifuge testing

Blind prediction contests of shake table tests of RC structures show that most contesting models fail to predict the seismic response of a structure. Given that models behave better at the component level, we conclude that a large part of the error/uncertainty sources from the “global level assumptions” (e.g. damping formulation, interaction of components etc.). In parallel, attempting to predict the maximum of the response to an individual ground motion is too strict of a validation test. In Earthquake Engineering, the hazard is defined by a set of ground motions and the task of the engineer is to predict some statistic of the maxima of the time history responses. There are cases where current models fail to predict the response to individual ground motions, but can excellently predict the CDF of the maxima of the responses to multiple ground motions. Therefore, there is a need to statistically validate the global assumptions of numerical models. Such a procedure requires a virgin specimen (e.g. a full building) for each ground motion. This is not feasible in full scale, but it can be done at small scale in a centrifuge. Then, building reinforcement with submillimeter diameters and manually placing it becomes a major issue. This can be solved by a metal 3D printer. This paper presents some first experimental results on compression tests on micro-concrete that uses sand and gypsum plaster, 4 point bending tests on 10x10x80mm unreinforced micro-concrete beams, tension tests on 3D printed steel rebars having a 0.5mm diameter, and 4 point bending tests on a 12.5x12.5x70mm micro-RC beam. The results are encouraging: the compression and tension behavior of concrete, the tensile behavior of steel, the beam ductility observed, the cracking patterns, and the bonding behavior between the concrete and the rebars is similar to full scale models.