Accurate semi-empirical predictive modeling of an underdamped process

Abstract Rollins et al. (D.K. Rollins, J.M. Liang, P. Smith. Accurate simplistic predictive modeling of nonlinear dynamic processes, ISA Transactions 4 (1998) 293–303.) introduced a predictive modeling approach that uses a semi-empirical model in an algorithm that changes the model form when input changes occur. This approach was found to be very accurate under a variety of sampling conditions when processes follow first order dynamics. The purpose of this article is to demonstrate the ability of this approach to accurately predict output response for systems with complex dynamics. More specifically, this approach is evaluated on a mathematically simulated CSTR that approximately follows underdamped second order behavior for changes in coolant flow rate and inverse second order behavior for changes in feed rate. This work demonstrates creativity in obtaining accurate fitted models with coefficients that vary widely over the input space using a few number of experimental runs to obtain the model coefficients. The proposed method is evaluated under conditions of high input (coolant flow rate) change frequency, variable change rate of the input, and no sampling of the output.