A comparison of proper orthogonal decomposition and Monte Carlo simulation of wind pressure data

A discussion on the efficiency of proper orthogonal decomposition (POD) and the Monte Carlo simulation (MCS) methods is presented. It is shown, by means of a theoretical and a practical example of application, that the efficiency of POD method strongly depends on whether a global or a local structural response to wind pressure loads is required. The POD method presents a fast convergence rate for global response, while for local response the effective correlation length of the pressure process should be considered. This conditional efficiency may be the cause of undue application of the method, if the user is not aware of its limitations. The MCS method is presented as an appealing alternative, since its efficiency does not depend neither on the response model nor on the statistics of the pressure data, but only on the number of simulations.