(301a) Addressing Practical Considerations in Implementation of Optimization Models for the Chemical Process Industry

Model-based optimization tools using mathematical programming techniques have been researched and adopted extensively both in academia and industry [1-2]. In the chemical industry, they have been implemented in diverse scales and applications such as batch scheduling [3], process design and optimization [4-6], process control [3, 6-7], production planning [8], turnaround planning [9-10]. However, there are some practical considerations to note which influence the modeling and solution strategies and thereby ultimately the successful and sustainable implementation of mathematical models in an industrial setting. We will address some of these key points in this talk.

Important factors which need to be evaluated at the outset are the application-specific requirements and the mode of model implementation i.e., whether it is real-time or not, whether a quick and good feasible solution is acceptable or an optimal solution is preferred. Understanding in depth the decisions to be delivered based on the model and the requirements of the decision maker through multiple probative questions is critical. This helps in iterative need-based modeling with the introduction of complexities only when absolutely needed and valuable.

Another set of factors that are important in an industrial setting are the ease of usage for a wide range of operating scenarios and the maintainability of models developed. e.g., fewer requirements for intermittent case-specific tuning. A particular behavior may be easier to formulate through use of multiple additional terms and corresponding penalty weight parameters in the objective function. However, such an approach could lead to constant case-specific tuning attempts of the multiple weight parameters to achieve the desired behavior. In other cases, however, overly constraining a system may affect the ability to obtain a feasible solution quickly and hence it might be prudent to have additional terms in the objective function. In some cases, the use of precedence-based models (compared to other structures such as resource-task-networks (RTNs)) help in reducing the number of parameters to keep track of and thus enhance model transferability. Grouping together blocks of constraints which describe a specific operational behavior enables more flexibility along with code readability and modularity.

Based on our experience with developing optimization models for various industrial applications, we will also briefly describe other strategies related to initialization of nonlinear models, user friendly practices for models requiring longer solve times, solver-specific heuristics, good modeling practices. These influence both the short-term and long-term successes of projects deriving value from use of mathematical programming models in industry.

References:

[1] Georgiadis, G.P.; Elekidis, A.P.; Georgiadis, M.C. Optimization-Based Scheduling for the Process Industries: From Theory to Real-Life Industrial Applications. Processes 2019, 7, 438.

[2] Hansen, E.; Rodrigues, M. A. S.; Aragão, M. E.; de Aquim, P. M. Water and wastewater minimization in a petrochemical industry through mathematical programming, Journal of Cleaner Production, 2018, 172, pp. 1814-1822

[3] Wassick, J.M., Nie, Y., Lin, B., Matthews, B., Hoogerwerf, R. Integrating Scheduling and Control for a Mixed-Mode Process. 2015 AIChE Annual Meeting Proceedings, Nov 12, 2015.

[4] Wen, Y., Biegler, L. T., Ochoa, M. P., Weston, J., Nikbin, N., Ferrio, J., Continuous reactor network design for multiple rigid polyol productions Journal of Advanced Manufacturing and Processing. 2022, 4(1)

[5] Ekici, C.; Biegler, L.; Ho, C. R.; DeWilde, J; Kipp, D.; Witt, P. Optimization of Syngas to Olefin (STO) Reactors Under Model Uncertainty 2020 AIChE Annual Meeting Proceedings, Nov 19, 2020.

[6] Bindlish, R. Scheduling, optimization and control of power for industrial cogeneration plants, Computers & Chemical Engineering, 2018, 114, pp. 221-224.

[7] Bindlish, R. Nonlinear model predictive control of an industrial polymerization process, Computers & Chemical Engineering, 2015, 73, pp. 43-48

[8] Amaran, S.; Rajagopalan, S..; Joswiak, M.; Bury, S. J. Long-Term Maintenance and Production Planning for the Integrated Chemical Enterprise, 2018 AIChE Annual Meeting Proceedings, Nov 1, 2018.

[9] Amaran, S.; Sahinidis, N.V.; Sharda, B.; Morrison, M.; Bury, S. J.; Miller, S.; Wassick, J. M. Long-term turnaround planning for integrated chemical sites, Computers & Chemical Engineering, 2015, 72, pp. 145-158.

[10] Amaran, S.; Zhang, T; Sahinidis, N. V.; Sharda, B.; Bury, S. J. Medium-term maintenance turnaround planning under uncertainty for integrated chemical sites, Computers & Chemical Engineering, 2016, 84, pp. 422-433.