467416 Health-Aware Operation of a Subsea Gas Compression Station Under Uncertain Operating Conditions
Traditionally, prognostics and health monitoring (PHM) systems are used by the operators to schedule maintenance. However, the information is seldom used to adjust the optimal operation strategy. Typically, the operators will choke back production until the estimated remaining useful life (RUL) is comfortably within the acceptable range. While this approach might work when there are only a few degrees of freedom that can be adjusted, it becomes difficult to ensure optimal for more complex processes.
Escobet et al. propose a control structure that combines PHM and process control, resulting in a health-aware control structure (Escobet, Puig, & Nejjari, 2012). The proposed control structure is able to ensure optimal operation without compromising the integrity of the system. Their approach involves increasing or decreasing the set-point of a PI controller by a fixed value until the predicted RUL is within an acceptable range, essentially making it an automated version of the current practice. While the algorithm is faster and more accurate than an operator, the set-point selection algorithm can result in cases where the proposed set-point is suboptimal or infeasible.
Pereira et al. propose a model predictive control (MPC) structure that utilizes prognosis and health monitoring of actuators to ensure optimal operation (Pereira, Galvão, & Yoneyama, 2010). By using an MPC, multivariate systems can also be treated. Reliability requirements are included by adding RUL constraints in the optimization routine. A similar approach is found in (Salazar, Weber, Nejjari, Theilliol, & Sarrate, 2016)
We extend the results of Escobet, Pereira, Salazar and coworkers by considering a stochastic degradation model and a nonlinear process model, with a nonlinear MPC formulation. We utilize the concept of health-aware control to derive a control structure for a subsea production system. A statistical model is used to predict the evolution of the remaining health of the system. Model uncertainty is treated by linearizing around worst case realizations, resulting in a robust control structure (Diehl, Bock, & Kostina, 2006). We also consider the case where the maintenance horizon is a decision variable, rather than optimizing over an a priori horizon. As an example, we consider a subsea gas compression station.
Diehl, M., Bock, H. G., & Kostina, E. (2006). An approximation technique for robust nonlinear optimization. Mathematical Programming, 213-230.
Escobet, T., Puig, V., & Nejjari, F. (2012). Health Aware Control and model-based Prognosis. Mediterranean Conference on Control & Automation (MED) (pags. 691-696). Barcelona: IEEE.
Pereira, E. B., Galvão, R. K., & Yoneyama, T. (2010). Model Predictive Control using Prognosis and Health Monitoring of actuators. International Symposium on Industrial Electronics (pags. 237-243). Bari: IEEE.
Salazar, J. C., Weber, P., Nejjari, F., Theilliol, D., & Sarrate, R. (2016). MPC framework for system reliability optimization. En Z. (. Kowalczuk, Advanced and Intelligent Computations in Diagnosis and Control (pags. 161-177). Springer.
See more of this Group/Topical: Computing and Systems Technology Division