Optimal Control Architecture for Balancing Performance and Cost in Oil and Gas Production Systems

In the domain of process design, stakeholders pursue two interrelated yet potentially conflicting objectives: maximization of system performance and reduction of plant cost. The control architecture of a process not only determines the cost of the system, but also significantly influences its potential performance. Nevertheless, conventional processes for designing control architectures prioritize economic objectives while overlooking system performance. This paper introduces a systematic approach that integrates both these objectives simultaneously into the design of control architectures for oil and gas production systems. The method involves quantifying the trade-off between controllability, observability, and the cost associated with the control architecture. This quantification is posed as a multi-objective integer nonlinear programming problem, which is specified as a Pareto optimization problem. Solving this optimization problem yields a set of Pareto-optimal control architectures, enabling design engineers to explore optimal tradeoffs between cost and performance. The efficacy of the proposed procedure is demonstrated through a real-world oil field example. Pareto-optimal architectures for the oil field are found using the developed framework. Subsequent analysis of the results reveals the indispensability of physical sensors for certain variables and the importance of well-balanced sensor distributions among the different wells in the oil field. To assess the impact of different architectures on closed-loop control performance, linear quadratic Gaussian (LQG) controllers are designed. Comparisons are made between the performance of LQG control systems instantiated on the identified Pareto-optimal architectures and non-optimal alternatives. This comparison highlights the pivotal role of optimal architectures in simultaneously enhancing performance and minimizing costs.