Reduced order approximation of the lattice discrete particle model for the simulation of FRP confined concrete columns

It has been recently proved that the Lattice Discrete Particle Model (LDPM) is capable of accurately capturing the behavior of uniformly and non-uniformly confined concrete and, consequently, it can be considered a useful tool for the simulation of FRP confined concrete columns with different cross sections under compressive loading. The LDPM computational framework is implemented into a multi-purpose structural analysis program called MARS, which is based on an explicit dynamic algorithm scheme, advantageous in terms of convergence. However, decreasing the maximum stable time step with the highest natural frequency of the system, the computational time necessary to carry out simulations of quasi-static events, like the compression tests in the present study, might be highly demanding. In order to decrease the computational cost of the simulations, the Proper Orthogonal Decomposition (POD) as a model reduction technique has been explored for the application to the LDPM models of FRP confined columns and the relationship between efficiency gain and accuracy loss is discussed.