Study On Economic Performance Assessment For Process Control

With the widespread implementation of advanced process control strategies in chemical and petrochemical plants in the last two decades, there have been an increasing need for a tool to reliably assess the economic performance of advanced process control applications. Economic performance assessment of process control has been recognized to be one of the best ways to identify potential benefits resulting from control upgrade projects. It can provide control engineers with a quantitative metric to justify the potential benefits due to control system improvement and can priorize the best control upgrading opportunity. It also provides useful information on performance monitoring or better management of abnormal situations in order to maintain or improve the economic performance of modern process control systems.Quantifying the economic benefit resulting from improved control is often based on the reduction of variability. Once the probability density function of key process variables and the economic performance function are both identified, the statistical-based or optimization-based method is utilized to calculated the potential benefits. Economic performance assessment of process control has been an area of active research in process control community since it involves the technology and knowledge of system identification and parameter estimation, performance assessment and monitoring, numerical analysis and optimization. Taking the new trends of the technology into accounts, this work will investigate the following issues from the theoretic development and engineering applications.(1) An approach to economic performance assessment of advanced control system is presented. The method builds on steady-state economic optimization techniques and uses the linear quadratic gaussian (LQG) benchmark other than conventional minimum variance control (MVC) to estimate the potential of reduction in variance. The LQG control is a more practical performance benchmark compared to MVC for performance assessment since it considers input variance and output variance, and it thus provides a desired basis for determining the theoretical maximum economic benefit potential arising from variability reduction. Combining the LQG benchmark directly with benefit potential of MPC control system, both the economic benefit and the optimal operation condition can be obtained by solving the economic optimization problem. The proposed algorithm is illustrated by a simulated example of Shell standard problem.(2) Uncertainty is an inherent characteristic in most industrial processes. Process uncertainties may lead to significant disturbances to the processes, thereby degrading the operation performance. It is therefore necessary and desirable to incorporate the effects of uncertainty dynamics into the assessment problem to make sure that the estimated process performance is relevant and practically realizable. An optimization-based approach for economic performance assessment of the constrained process control is integrated with the LQG benchmark as the variance benchmark. By explicitly incorporating uncertainty into the performance assessment problem, the performance evaluation can be formulated as a stochastic optimization problem, which helps to identify the opportunity to improve the profitability of the process by taking appropriate risk levels. Using the LQG benchmark to estimate the achievable variability reduction through the control system upgrades, the proposed method provides an estimate of both the performance that can be expected from the control system and the operating condition that delivers the improved performance. The results obtained can serve as a tool for control engineers to make decisions on control system tuning and/or upgrading. The proposed algorithm is illustrated via a simulation example of a model predictive control system for distillation process model.(3) Based on the results of estimated economic performance calculation, the economic performance indices under different control upgrade strategies are defined. Then a decision is made on whether a control system upgrade can improve the process performance, and which proposed control system should be implemented. The proposed approach is illustrated by the application to economic performance assessment of an industrial model predictive control system for xylene distillation unit.(4) Economic performance of advanced process control should pay attention to the control performance since the improvement of control performance does improve economic performance of process control. Performance assessment and monitoring of model predictive control (MPC) systems has been a great interest for both academia and industry. The presence of constraints renders MPC controller nonlinear and, thus, makes the use of traditional linear techniques problematic. When in consideration of the constraint and optimization prosperities of MPC, the controlled variables (CVs) are expected to be constrained within a certain region but not at a certain set point. Performance assessment and monitoring of such type of problem is meaningful and practical. In this study, a new approach based on the weighted points statistics is developed for the performance assessment of the above problem. The important advantage of the proposed approach is that just the routine closed-loop operation data of the system and constrained region of each CV are required, which is convenient for the industrial applications. Simulation example and industrial case study illustrate the applicability of the proposed approach.(5) Parameter estimation is a key problem in the development of process models, and thus is an important issue in both economic performance assessment and control performance monitoring of process control. A novel three stage computation framework is developed, which is based on a quasi-sequential dynamic optimization method. By dividing the variables space into a dependent and an independent space, only independent variables are treated by " the SQP solver. The lower stage is also named simulation layer, where the process model in terms of DAEs are discretized with orthogonal collocation and solved using Newton method to compute the dependent variables. Since the degree of freedom is limited to the number of parameters in the upper stage and the number of independent variables in the middle stage, any standard NLP solver can be used to solve the problem. Thus, the computation expensive is significantly reduced. An example of parameter estimation for a CSTR model is employed to demonstrate the effectiveness of the proposed approach.