Publications

Model-based electroslag remelting control for simultaneous, consistent and responsive melt rate and immersion depth control

Melgaard, David K.; Beaman, Joseph J.; Shelmidine, G.J.

Electroslag Remelting (ESR) is a complex process used to produce high quality specialty alloy materials. The quality can be directly correlated to variances in melt rate and immersion depth. Conventional ESR furnaces control these quantities using two independent control loops using proportional changes in current for melt rate and driving the electrode up and down to match a voltage set point for immersion depth. However it is well known that the control loops are highly coupled, i.e. changing the current to account for melt rate deviations changes the voltage depth relationship and vice verse. In addition the noise in measurements of the ESR process can be considerable, forcing conventional controllers to use highly damped responses. A new model-based controller has been developed to embody the coupling and improve responsiveness by using estimates from a reduced-order linear ESR model and the typical process measurements to control melt rate and immersion depth simultaneously. Kalman filtering is used to optimally combine the model estimates of eight process states and the measurements of voltage, current, position and mass to estimate the instantaneous melt rate and immersion depth. Several ESR melts under steady state and transient conditions were conducted to evaluate the performance of the new controller. This paper will discuss the design of the new ESR model and controller and will present experimental results demonstrating its much improved control and responsiveness. While this controller was developed for the ESR process, the effectiveness of model-based control in managing such a complex process with relatively simple equations suggests the approach could be employed for many other processes as well.