(189h) Chance-Constrained MINLP Optimization for the Process Synthesis of the Oxidative Coupling of Methane

Uncertainty in process models is of vital importance for most process synthesis tasks. This is especially the case when a close interrelation between process unit outputs, such as required product purities, and parametric uncertainty exists. In recent contributions, such as (Quaglia et al., 2013) and (Steimel et al., 2014), frameworks have been presented, which incorporate uncertainty in mixed-integer (non-)linear programming problems associated with superstructure optimization in chemical engineering. Most approaches published so far follow along the lines of a two stage technique with an outer stage in charge of making the superstructure decisions on the design and the inner stage reacting to a set of different scenarios generated from a description of the uncertainty. Consequently, these frameworks require a great number of simulations and the structural optimization reacts only indirectly to the effect of the uncertainty. As an alternative approach to tackle the aforementioned class of optimization problems, Esche et al. (2016) introduced a chance-constrained MINLP approach. The framework incorporates uncertainty into process synthesis tasks by requiring the adherence to probabilistic process constraints, namely chance constraints.

In this contribution, a comparison of the results of the chance-constrained MINLP framework to deterministic, robust, and other sampling-based methods (e.g. Monte Carlo sampling) for the introduction of uncertainty into the optimal process synthesis task is carried out. The oxidative coupling of methane process is employed as an exemplary process synthesis task. The process concept considers choices on different reactor feeding policies and various downstreaming options. The latter consist of combined pressure and temperature swing adsorption, combinations of gas permeation membranes, and absorption desorption processes using different scrubbing liquids for the removal of carbon dioxide from the product stream.

In addition to the comparison of the results, options for the extension of the chance-constrained optimization framework for larger scale systems, greater numbers of uncertain variables and chance constraints, as well as the interfacing with additional solvers is discussed.

References 

Quaglia, A., Sarup, B., Sin, G., Gani, R. (2013). A systematic framework for enterprise-wide optimization: Synthesis and design of process networks under uncertainty. Computers & Chemical Engineering, 59:47â??62.

Steimel, J., Harrmann, M., Schembecker, G., Engell, S. (2014). A framework for the modeling and optimization of process structures under uncertainty. Chemical Engineering Science, 115:225-237

Esche, E., Müller, D., Werk, S., Grossmann, I.E., Wozny, G. (2016). Solution of Chance-Constrained Mixed-Integer Nonlinear Programming Problems. In: Proceedings of the 26^th European Symposium on Computer Aided Process Engineering, June 12^th-15^th, 2016