(273e) Resilient Design and Operations of Chemical Process Systems Using Robust Optimization

A major goal of risk management is to avoid the occurrences of undesired events by implementing effective prevention and protection strategies [1]. However, many disruptive events suggest that not all unexpected events can be avoided. Disruptive events usually strike a process system and cause critical failures in vulnerable processes [2]. The undesired consequences of disruptive events highlight the need for enhancing the resilience of process systems. Although resilience may be interpreted by different terminologies in various contexts, a resilient system is always capable of absorbing a portion of the impacts from disruptive events and recovering to the original state rapidly. There are several research challenges to develop a general framework for resilience optimization. The first challenge is to propose a novel quantitative measure of resilience for process systems. The quantitative measure should be able to account for both performance degradation and system recovery. Additionally, resilience is an intrinsic property of a system, indicating that the quantitative measure should be independent of external systems or volatile markets. The second challenge is how to model resilience enhancement strategies, and how to integrate the resilience enhancement strategies and process models into an integrated systems analysis and optimization framework. Because the resilience of a process system is relevant to both safety and operability, the system performance under the worst-case realization of disruptive events is of paramount importance [3]. Adaptive robust optimization is an emerging method to handle uncertainty in sequential decision-making processes and to hedge against the worst-case realization of uncertainty. Although adaptive robust optimization could be an effective tool for addressing problems on resilient design and operations, there are research challenges on developing a two-stage adaptive robust optimization model for resilience optimization and on effective solution of the resulting optimization problem that cannot be tackled directly by any off-the-shelf optimization solvers due to its multilevel structure.

In this work, we address the resilient design and operations of process systems in response to disruption events. A novel quantitative measure of resilience is proposed as the ratio of the quantity of products manufactured with disruptive events to that without disruptive events. Five resilience enhancement strategies are then introduced, including selecting the most resilient technology/process alternatives, increasing the capacities of operating processes, employing parallel operating processes, building backup processes, and optimizing the operating levels after the occurrence of disruptive events. A general framework for resilience optimization is further proposed to incorporate the quantitative measure of resilience and the resilience enhancement strategies into process design and operations. In the first step of the proposed resilience optimization framework, a preliminary risk assessment is performed for a given system to identify the disruptive events that are worth considering in process design and operations. The numbers of failed processes for the identified disruptive events and the recovery time of each process are used as input parameters for resilient design and operations. In the second step, a multiobjective two-stage adaptive robust mixed-integer fractional programming model is formulated. There are two objective functions: the first objective function is to maximize the resilience under the worst-case realization of disruptive events, and the second objective function is to minimize the total capital cost. Both objective functions are independent of external processes and volatile markets, thus reflecting the intrinsic properties of the given system. The resulting optimization model has a three-level structure: the first level determines the optimal network configuration, equipment capacities, and capital costs; the second level determines the worst-case realization of disruptive events; the third level determines the optimal number of available processes and operating levels in each time period. To tackle the computational challenges stemming from the multilevel structure and the nonlinear objective function, a tailored global optimization method is proposed by integrating the inexact parametric algorithm and the column-and-constraint generation algorithm. The applicability of the proposed resilience optimization framework is illustrated through two applications on process design and planning of a chemical process network and superstructure optimization of shale gas processing and natural gas liquids recovery processes, respectively.

References

• Hosseini, K. Barker, and J. E. Ramirez-Marquez, "A review of definitions and measures of system resilience," Reliab. Eng. Syst. Safe., vol. 145, pp. 47-61, 2016.
• Bhamra, S. Dani, and K. Burnard, "Resilience: the concept, a literature review and future directions," Int. J. Prod. Res., vol. 49, pp. 5375-5393, 2011.
• T. T. Dinh, H. Pasman, X. D. Gao, and M. S. Mannan, "Resilience engineering of industrial processes: Principles and contributing factors," J. Loss Prevent. Proc., vol. 25, pp. 233-241, 2012.