A machine learning approach to determine the elastic properties of printed fiber-reinforced polymers

This work focuses on the simultaneous determination of elastic constants and fiber orientation state for a short fiber-reinforced polymer composite by performing minimum experimental tests. We introduce methodology that enables inverse in-situ properties tensile tests at coupon level. demonstrate approach extrusion deposition additive manufacturing (EDAM) process to illustrate one application methodology, but development is such it can be applied (SFRP) systems processed via other methods. Currently, developing composites digital twins require extensive material characterization. In particular, mechanical characterization orthotropic involves sample preparation testing, therefore elasticity tensor generally populated using micromechanics model. This, however, requires measuring in addition knowing constituent properties. Experimentally tedious time consuming. Further, optical methods are limited resolving cylindrical fibers or cluster non-cylindrical fibers, computed tomography (CT) scan regions volume much smaller than full printed bead. Therefore, we propose accelerated machine learning, identify anisotropic same time. Early results show inference possible with as few three Our combination choice model reliable set experiments yield nine constants, well as, state.