Advances in MVC Technology

The Multi-Variable Controller (MVC) is ubiquitous in refining and petrochemicals industries. The MVC technology in recent years has changed a lot from its start three decades ago. The underlying architecture of MVC is still the same but there have been changes in the modeling methodology. Typical MVC in each cycle execution performs three steps: i) Open loop prediction ii) Steady state optimization and iii) Generates dynamic move plan. A linear programming problem is formulated in steady state optimizer routine to compute optimized targets. A dynamic move plan is generated from the optimized targets and open loop prediction. The move plan from the dynamic optimization is downloaded to the lower level regulatory control. The set points of regulatory control is used as Manipulated Variables (MV) in MVC. The latest in MVC technology is MVC directly manipulating the control valve output (OP) to control the process. The model used for this purpose is called PV (Process Variable) model, where MVs are PVs. The MVC control execution is same as the traditional MVC but the dynamic move plan is computed in terms of PV change and transformed to OP change via OP-PV transformations. This OP change is downloaded to DCS to implement the move. The advantage of this method is that a valve can operate to its saturation limit without losing model fidelity. PV Models also increase a Plant's Capacity by increasing the feasible operating region of the process.