The use of fast-time models to implement minimum time control policies is considered for two distinct modes of operation. The first case considered is that in which the fast model operates from current process state, or output, towards desired process state, or input. This forward searching model enables estimates to be made of future process behaviour based upon currently applied control signals.