Model Predictive Control of a Rod Pump

Situation: Develop a control system to control the fluid level height of a sucker rod system that takes into account rod pump, well, and reservoir dynamics.

Task: Design a model predictive controller (MPC) and moving horizon estimator (MHE) to automatically control fluid height in a sucker rod well.

Action: Designed a moving horizon estimator to determine uncertain well parameters using available surface measurements. Developed an MHE/MPC control algorithm to adjust the stroking speed of a sucker rod pump to maintain fluid level above the pump.

Result: MHE/MPC provided the best control when compared against PI and pump-off control. Results from this study were accepted to be published by Computers and Chemical Engineering Journal.

Summary of rod pump, well, and reservoir dynamics. ×

Moving horizon estimation of porosity from surface loading measurements and stroking speed. The estimator is able to closely approximate the true value of porosity as more data becomes available. ×

Moving horizon estimation of multiple reservoir parameters from surface loading measurements and stroking speed. The estimator converges to almost identical values of the true porosity and skin. ×

Shows the controller performance of pump off control. ×

Shows the controller performance of PI control. ×

Shows the MHE and MPC controller performance. ×