Abstract: A control of a multi-variable process throughout a plurality of successive, distinct operational phases of the process involves accumulation of sets of values of the process and quality variables (a-j) from previous multiple operations of the process through all its phases, as respective datasets of a historical record. Each dataset includes an identifier (p) of the phase to which it relates. The current values of the applicable process variables (Qa-Qh) during each phase and the phase identifier (p) are indicated on respective axes (Xp,Xa-Xh) of a multi-dimensional display in relation to an operational envelope which defines bounds (UL,LL) for the values of the individual process and quality variables (a-j) relevant to that phase and which is derived from the datasets of the historical record identified with that phase. Changes in the current process variables (Qa-Qh) result in changes of the bounds (UL,LL), and alarm is given when a current value departs from within them.

Abstract: Controlling a multi-variable process involves multi-dimensional representation of the values (Qa-Qh) of the process-variables (a-h) according to individual coordinate axes (Xa-Xh), and response based on historical values for the process-variables accumulated from multiple, earlier processes. An envelope (UL-LL) showing the best operating zone (‘BOZ’) for each process variable based on current values of the other variables is calculated from the accumulated historical values, and alarm conditions in which the current value of a variable lies outside the BOZ is rectified by changing the values (Qa-Qc) of manipulatable variables (a-c). Variable targets are achieved, alarms rectified and value optimisation realised using an inner envelope (UI-LI) derived from a subset of the BOZ-defining set of historical values. Where the alarm rate is low, operation is improved by narrowing the BOZ set to tighten the BOZ envelope (UL-LL) reducing an inner envelope where alarm rate remains acceptable, as a new BOZ.