Modeling, estimation, and control of electroslag remelting process

Electroslag Remelting (ESR) is used widely throughout the specialty metals industry to produce superalloy and special steel cast ingots. High quality ESR casting requires that the electrode melting rate be controlled at all times during the process. This is especially difficult when process conditions are such that the temperature distribution in the electrode has not achieved, or has been driven away from, steady state. This condition is encountered during the beginning and closing stages of the ESR process and also during some process disturbances such as when the melt zone passes through a transverse crack. To address these transient melting situations, a new method of ESR melt rate control has been developed that incorporates an accurate, reduced-order melting model to continually estimate the temperature distribution in the electrode. The related state variables are estimated by the observer algorithms. Due to the highly nonlinear characteristics of the process, more sophisticated estimators than the Kalman filter are proposed. The unscented Kalman filter (UKF) based on the unscented transform and the particle filtering technique were chosen for possible candidates and applied in the controller design. During the highly transient periods during melting, the UKF showed the best performance for controlling the melt rate. Particle filtering can deal with non-Gaussian noises and the accuracy is totally based on the number of the Monte Carlo runs. Unfortunately, the particle filter is relatively slow in the real-time applications for controlling the ESR process with current computer technology.