Optimal Filtering Applied to the Vacuum Arc Remelting Process

Optimal estimation theory has been applied to the problem of estimating process variables during vacuum arc remelting (V AR), a process widely used in the specialty metals industry to cast large ingots of segregation sensitive and/or reactive metal alloys. Four state variables were used to develop a simple state-space model of the V AR process: electrode gap (G), electrode mass (M), electrode position (X) and electrode melting rate (R). The optimal estimator consists of a Kalman filter that incorporates the model and uses electrode feed rate and measurement based estimates of G, M and X to produce optimal estimates of all four state variables. Simulations show that the filter provides estimates that have error variances between one and three orders-of-magnitude less than estimates based solely on measurements. Examples are presented that verify this for electrode gap, an extremely important control parameter for the process.