(483b) Parallel Solution of Optimal Gas Network Control Under Uncertainty

We present the parallel solution of optimal gas network control problems on distributed memory machines while considering uncertainty, obtaining solution times two orders of magnitude faster than state-of-the-art NLP solvers. Optimal control of gas networks is critical both for reliable operation of the electric grid and residential heating. Furthermore, consideration of uncertainty is becoming increasingly important as renewable penetration increases [1]. However, the addition of uncertainty adds a great deal of complexity to an already complex problem. The combination of spatial discretization, time discretization, and uncertainty lead to extremely large nonlinear programming problems (NLPs). However, these problems have inherent structure that can be exploited for parallel solution.

We present a framework for both parallel modeling and solution of optimal gas network control under uncertainty. Our pipeline models are distributed with IDEAS [2], making the development of gas network models quick, intuitive, and less error prone. Parallel solution on distributed memory machines is enabled with Parapint [3]. Parapint is an open-source Python package that utilizes the Message Passing Interface (MPI) and Schur-Complement decomposition within an interior-point algorithm for parallel solution of structured NLPs [4]. Our numerical results demonstrate over two orders of magnitude speedup compared to state-of-the-art solver Ipopt [5].

Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. This work was funded by the Institute for the Design of Advanced Energy Systems (IDAES) with funding from the Office of Fossil Energy, Cross-Cutting Research, U.S. Department of Energy.

Works Cited

1. [1] V. Zavala, "Stochastic optimal control model for natural gas networks," Computers & Chemical Engineering, vol. 64, pp. 103-113, 2014.

2. [2] A. Lee, J. Ghouse, J. Eslick, C. Laird, J. Siirola, M. Zamarripa, D. Gunter, J. Shinn, A. Dowling, D. Bhattacharyya, L. Biegler, A. Burgard and D. Miller, "The IDAES process modeling framework and model library - Flexibility for process simulation and optimization," Journal of Advanced Manufacturing and Processing, vol. 3, no. 3, p. e10095, 2021.

3. [3] J. Rodriguez, R. Parker, C. Laird, B. Nicholson, J. Siirola and M. Bynum, "Scalable Parallel Nonlinear Optimization with PyNumero and Parapint," Optimization Online, 22 09 2021.