The effect of model fidelity on real-time optimization performance

Abstract

The ability of a real-time optimization (RTO) system to track the changing optimum closely relies on an accurate model for representing the plant behavior. This paper investigates the effect of model fidelity on RTO performance using a simulated industrial boiler network case study. Three optimization approaches with very different modelling fidelity are investigated; 1) model-free direct search method, 2) modelbased method using a simplified efficiency curve model, and 3) model-based method using a fundamental model. The model-free direct search method builds a locally linear model using empirical data. It takes many steps to reach the optimum, which causes a significant profit loss during tracking. This tracking loss can be reduced by using the model-based RTO system. The RTO system with an updated, detailed fundamental model is able to track fast and large disturbances because the model is accurate in a large range of operation. An RTO system with a simplified efficiency model requires periodic experimentation to correct for the disturbances, which can cause significant profit loss during experimentation and tracking. This study demonstrates how quantitative performance measures improve as higher fidelity models are used in real-time operations optimization.

Optimización en Tiempo Real con Disyunciones Lógicas: Aplicación a Sistemas de Calor y Potencia

“…La variabilidad horaria de los precios de la electricidad, los cambios en la demanda de vapor en distintos niveles de presión, los cambios en la eficiencia de las turbinas de gas con la temperatura ambiente, y las restricciones ambientales y operativas, hacen que la operación óptima sea diferente en distintos horarios del día, y difícil de prever o planificar. Algunos autores han abordado el problema sin tener en cuenta la adaptación del modelo (VELASCO-GARCIA, 2011); otros han propuesto la optimización en tiempo real de una caldera aislada (YIP; MARLIN, 2004).…”

Performance Evaluation of Real Industrial RTO Systems

“…Great advantages are attributed to the use of a priori information in the form of a process model, and model-based techniques may present superior performance among others; generally, the more accurate the model, the better will be the RTO system performance [19,20]. Thus, such RTO applications are typically based on rigorous steady-state models of processes.…”

Model-free optimization based feedforward control for an inkjet printhead

“…[1] and [2]). The model-fitting task is further complicated by the fact that process data are usually noisy and signals often do not carry sufficient information for efficient process identification [3]. Therefore, optimization using an inaccurate model might result in suboptimal operation or, worse, infeasible operation when constraints are present.…”