(344i) A Framework for Steady-State Process Operability Analysis Using Kriging-Based Surrogate Models

The objective in this work is to develop a supervised machine learning-based framework for Process Operability [1] using responses based on Kriging surrogate models. Currently, the available operability approaches for nonlinear systems are limited by problem dimensionality, not being computationally tractable for high-dimensional systems [2], [3], [4]. Due to this high-dimensionality challenge, a parallel programming approach needed to be employed [4], which is not always readily available in all programming platforms and user software infrastructures. As an alternative, the proposed approach in this work will use Kriging (also known as Gaussian Process Regression) [5], [6], [7] to substitute the developed first-principles or process simulation-based models by surrogate models. The built surrogate models can generate results that are comparable to the first-principles nonlinear models in terms of accuracy, while reducing the computational effort.

To achieve this goal, a steady-state framework for the systematic analysis of highly nonlinear, large-dimensional systems is proposed. In this framework, the following tasks are performed: model selection, training, validation, and testing of the produced surrogate models when employed in existing process operability algorithms [2], [3], [8] that are based on first-principles models. The proposed framework is benchmarked against the current operability tools to provide a new direction in the Process Operability field employing surrogate models.

Two case studies associated with natural/shale gas conversion are addressed to illustrate the effectiveness of the proposed methods, namely a membrane reactor for direct methane conversion to hydrogen fuel and benzene and a natural gas combined cycle (NGCC) plant. The proposed approach generates results with an error margin less than 1% and with computational effort reduction by up to four orders of magnitude, when compared to the results obtained with the first-principles-based process operability case studies from the literature [2], [9], indicating the trend of having reduced computational effort for high-dimensional problems. Therefore, this research is expected to contribute with supervised machine learning formulations and algorithms/software tools for Process Operability to enable the improved design, operations and manufacturing of complex chemical and energy systems.

References

[1] C. Georgakis, D. Uztürk, S. Subramanian and D. R. Vinson, “On the operability of continuous processes,” Control Engineering Practice, vol. 11, pp. 859-869, 2003.

[2] J. C. Carrasco and F. V. Lima, “An optimization-based operability framework for process design and intensification of modular natural gas utilization systems,” Computers & Chemical Engineering, vol. 105, pp. 246-258, 2017.

[3] J. C. Carrasco and F. V. Lima, “Novel operability-based approach for process design and intensification: Application to a membrane reactor for direct methane aromatization,” AIChE Journal, vol. 63, pp. 975-983, 2017.

[4] J. C. Carrasco and F. V. Lima, “Bilevel and parallel programing-based operability approaches for process intensification and modularity,” AIChE Journal, vol. 64, pp. 3042-3054, 2018.

[5] S. N. Lophaven, H. B. Nielsen and J. Søndergaard, DACE - A Matlab Kriging Toolbox, Version 2.0, 2002.

[6] A. Forrester, A. Sobester and A. Keane, Engineering design via surrogate modelling: a practical guide, John Wiley & Sons, 2008.

[7] C. E. Rasmussen and H. Nickisch, “Gaussian processes for machine learning (GPML) toolbox,” Journal of machine learning research, vol. 11, p. 3011–3015, 2010.

[8] V. Gazzaneo and F. V. Lima, “Multilayer Operability Framework for Process Design, Intensification, and Modularization of Nonlinear Energy Systems,” Industrial & Engineering Chemistry Research, vol. 58, pp. 6069-6079, 2019.

[9] J. C. Carrasco and F. V. Lima, “Nonlinear Operability of a Membrane Reactor for Direct Methane Aromatization,” IFAC-PapersOnLine, vol. 48, pp. 728-733, 2015.