Modeling and identification of centrifugal compressor dynamics with approximate realizations

This paper deals with the parameter identification of a model for the dynamic behavior of a large industrial centrifugal compression system. Experimental results are presented to evaluate a new approach for determining the parameters of a modified version of the well-known Greitzer model. This approach is based on an approximate realization algorithm that constructs an LTI model from step response data. It is shown that the method is capable of providing estimates for a limited set of model parameters. In a subsequent sensitivity analysis it is proven that the identifiable parameter are all related to the dominant dynamics of the system. The main conclusion is that the suggested approach is capable of providing estimates for the various model parameters. However, identification is limited to those parameters that are associated with dynamics that can be excited sufficiently during forced response experiments.