Temperature control of hot isostatic pressing (HIP) furnace using model predictive and risk sensitive optimal controllers

Fabrication of metal matrix composites (MMC) requires the use of sophisticated manufacturing processes. One such process is hot isostatic pressing (HIP) which is becoming a key technology for manufacturing MMC airframe structural panels for advanced aircrafts, superalloys and titanium alloys. We present in this paper the results of applying modern control techniques such as model predictive control (MPC) to HIP furnaces to achieve optimal temperature and pressure profiles-this would minimize the ultimate cost per unit of production. A HIP furnace is primarily a nonlinear distributed parameter process. The problem was solved in three steps. First, a nonlinear physical model was developed and validated from experimental data from a real 3-zone HIP furnace located at Industrial Materials Technology (IMT), Andover, Mass. The model was linearized in the second step to approximate the steady state behavior of a HIP furnace. Finally, in the third step, the linear model was used to design an MPC controller. Another control technique similar to H-infinity known as risk sensitive optimal (RSO) control was used in combination with MPC to increase robustness and improve steady state regulation. The controllers were tested on the nonlinear simulator. The response time of the combined MPC-RSO controlled system was less than one-third of the closed loop time response of the system under current PID control. The combined controller was also found to attenuate low frequency disturbances to a satisfactory level.