Model reference adaptive control specification for a steam heated finned tube heat exchanger

Abstract

Model reference adaptive control is applied to a steam heated finned tube heat exchanger to improve the dynamic performance as the system undergoes changes in its describing transfer function. Outlet coolant temperature is regulated by steam flow in a direct digital control feedback loop. Process identification is carried out in closed loop. A prespecified dynamical behavior is imposed upon the system by adjustment of the control algorithm parameters. The process output is subsequently forced to track a model unit step set point response. Simulation studies are in initially conducted with a realistic nonlinear mathematical model for the heat exchanger system. The nonlinear process is then modeled by a second-order transfer function with deadtime (linear model). Process identification based upon quasilinearization, maximum likelihood, generalized least squares, and Kalman's pulse transfer function method is achieved to obtain linear discrete process models at three different operating conditions in each of three conventional feedback control modes. Comparison of adapted and unadapted system model reference performance indices reveals that quasi linearization is most accurate for modeling and implementation of the adaptive algorithm. In addition, statistical correlation is used with open-loop data to estimate the impulse response of the exchanger about a linearized operating condition. Two techniques are proposed for calculating process response information i f the impulse response is available. Equations describing the digital control loop, process, and adaptive algorithm are presented. The simulation studies show theoretically that closed-loop process modeling is least accurate for proportional-integral control as evidenced by wider 95 percent confidence intervals on the estimated linear model coefficients. This conclusion is attributed to the "tighter" control exhibited about the control set point.