Model-Order Reduction (MOR) is an important technique that allows Reduced-Order Models (ROMs) of physical systems to be generated that can capture the dominant dynamics, but at lower cost than the full order system. One approach to MOR that has been successfully implemented in fluid dynamics is the Eigensystem Realization Algorithm (ERA). This method requires only minimal changes to the inputs and outputs of a CFD code so that the linear responses of the system to unit impulses on each input channel can be extracted. One of the challenges with the method is to specify the size of the input pulse. An inappropriate size may cause a failure of the code to converge due to non-physical behaviour arising during the solution process. This paper addresses this issue by using piston theory to estimate the appropriate input pulse size.
