Modular Multivariable Control Applied to the Shell Heavy Oil Fractionator

Modular multivariable controllers use interconnections of single process variable-multiple control effort controllers (i.e. Coordinated Controllers) plus logic to control multiple, possibly constrained, process variables with multiple, constrained, control efforts. The number of controls may be greater or fewer than the number of process variables. In the event that not all control objectives can be met, then process variables with higher user assigned priorities are kept at their targets in preference to process variables with lower priorities. When there are more controls than process variables which are to be maintained at set points, then the excess controls can be assigned nominal values and priorities. After all process variable objectives are met, the excess controls will achieve their nominal values, again in order of user assigned priority. We propose a modular multivariable control structure (i.e. process variable and control effort priorities) which should lead to a complete solution to the Shell Heavy Oil Fractionator Problem. The structure is justified based on performance objectives and robustness considerations. In addition, we compare the control achieved by a simple MMC (modular multivariable control) implementation to those achieved by Shell´s QDMC approach for setpoint changes for a two control effort, two process variable subset of the complete problem.