An efficient method for optimal design of large-scale integrated chemical production sites with endogenous uncertainty

Abstract

Integrated sites are tightly interconnected networks of large-scale chemical processes.Given the large-scale network structure of these sites, disruptions in any of its nodes, or individual chemical processes, can propagate and disrupt the operation of the whole network. Random process failures that reduce or shut down production capacity are among the most common disruptions. The impact of such disruptive events can be mitigated by adding parallel units and/or intermediate storage. In this paper, we address the design of large-scale, integrated sites considering random process failures. In a previous work (Terrazas-Moreno et al., 2010), we proposed a novel mixed integer linear programming (MILP) model to maximize the average production capacity of an integrated site while minimizing the required capital investment. The present work deals with the solution of large-scale problem instances for which a strategy is proposed that consists of two elements. On one hand, we use Benders decomposition to overcome the combinatorial complexity of the MILP model. On the other hand, we exploit discrete-rate simulation tools to obtain a relevant reduced sample of failure scenarios or states. We first illustrate this strategy in a small example. Next, we address an industrial case study where we use a detailed simulation model to assess the quality of the design obtained from the MILP model.

Asset maintenance optimisation approaches in the chemical and process industries – A review

“… Terrazas-Moreno et al (2011) combined simulation and optimisation to design the integrated chemical processing networks, which are subjected to random failures of units. The superstructure optimisation model is formulated to add parallel units or additional buffers to the integrated sites.…”

Multistage Robust Mixed-Integer Optimization Under Endogenous Uncertainty

“…This results in a model that encodes a very large conditional scenario tree, which dramatically increases the computational complexity compared to the case with only exogenous uncertainty. Significant advances have been made in solving such stochastic programs, with approaches that focus on two general strategies: identifying redundant NACs that can be removed (Goel & Grossmann, 2006;Colvin & Maravelias, 2008;Gupta & Grossmann, 2011;Boland et al, 2016;Apap & Grossmann, 2017), and applying tailored solution algorithms based on Lagrangean decomposition (Goel & Grossmann, 2006;Gupta & Grossmann, 2014), branch-and-cut (Colvin & Maravelias, 2010), Benders decomposition (Terrazas-Moreno et al, 2012), sequential scenario decomposition (Apap & Grossmann, 2017), or heuristic knapsack decomposition (Christian & Cremaschi, 2015). Vayanos et al (2011) apply decision rules to obtain tractable conservative approximations for multistage stochastic programs with type-2 endogenous uncertain parameters that are continuously distributed.…”

Models and computational strategies for multistage stochastic programming under endogenous and exogenous uncertainties

“…Additionally, Boland et al (2016) and Hooshmand Khaligh and MirHassani (2016b) explored non-anticipativity constraint reduction for multistage stochastic programs with arbitrary scenario sets, which will be of particular interest later in this thesis. Other publications on stochastic programming under endogenous uncertainty which we will not discuss here, but may be of interest to the reader, include: multistage stochastic network interdiction (Held and Woodruff, 2005); the decision-rule approach to multistage stochastic programming (Vayanos et al, 2011); the optimal design of integrated chemical-production sites (Terrazas-Moreno et al, 2012); computational strategies for nonconvex, multistage mixed-integer nonlinear programs (Tarhan et al, 2013); and the dynamic single-vehicle routing problem with uncertain demands (Hooshmand Khaligh and MirHassani, 2016a).…”