Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Neural Network-enabled Domain-consistent Robust Optimisation for Global CO$_2$ Reduction Potential of Gas Power Plants

Waqar Muhammad Ashraf, Talha Ansar, Abdulelah S. Alshehri, Peipei Chen, Ramit Debnath, Vivek Dua

TL;DR

The paper tackles domain-inconsistency that arises when neural surrogates interface with optimisation solvers in gas power plant operation. It introduces a domain-consistent robust optimisation framework that uses a data-derived Mahalanobis constraint and multi-level ANN surrogates embedded in a nonlinear program, validated on a 1180 MW CCGPP. The approach yields a mean energy-efficiency gain of about $0.76$ pp at the plant level and projects a global CO2 reduction potential of roughly $26$ Mt per year, with notable regional distribution. This work demonstrates a scalable, near-term decarbonisation pathway enabled by machine learning-guided robust optimisation, applicable to the global gas power plant fleet.

Abstract

We introduce a neural network-driven robust optimisation framework that integrates data-driven domain as a constraint into the nonlinear programming technique, addressing the overlooked issue of domain-inconsistent solutions arising from the interaction of parametrised neural network models with optimisation solvers. Applied to a 1180 MW capacity combined cycle gas power plant, our framework delivers domain-consistent robust optimal solutions that achieve a verified 0.76 percentage point mean improvement in energy efficiency. For the first time, scaling this efficiency gain to the global fleet of gas power plants, we estimate an annual 26 Mt reduction potential in CO$_2$ (with 10.6 Mt in Asia, 9.0 Mt in the Americas, and 4.5 Mt in Europe). These results underscore the synergetic role of machine learning in delivering near-term, scalable decarbonisation pathways for global climate action.

Neural Network-enabled Domain-consistent Robust Optimisation for Global CO$_2$ Reduction Potential of Gas Power Plants

TL;DR

The paper tackles domain-inconsistency that arises when neural surrogates interface with optimisation solvers in gas power plant operation. It introduces a domain-consistent robust optimisation framework that uses a data-derived Mahalanobis constraint and multi-level ANN surrogates embedded in a nonlinear program, validated on a 1180 MW CCGPP. The approach yields a mean energy-efficiency gain of about pp at the plant level and projects a global CO2 reduction potential of roughly Mt per year, with notable regional distribution. This work demonstrates a scalable, near-term decarbonisation pathway enabled by machine learning-guided robust optimisation, applicable to the global gas power plant fleet.

Abstract

We introduce a neural network-driven robust optimisation framework that integrates data-driven domain as a constraint into the nonlinear programming technique, addressing the overlooked issue of domain-inconsistent solutions arising from the interaction of parametrised neural network models with optimisation solvers. Applied to a 1180 MW capacity combined cycle gas power plant, our framework delivers domain-consistent robust optimal solutions that achieve a verified 0.76 percentage point mean improvement in energy efficiency. For the first time, scaling this efficiency gain to the global fleet of gas power plants, we estimate an annual 26 Mt reduction potential in CO (with 10.6 Mt in Asia, 9.0 Mt in the Americas, and 4.5 Mt in Europe). These results underscore the synergetic role of machine learning in delivering near-term, scalable decarbonisation pathways for global climate action.
Paper Structure (11 sections, 5 equations, 3 figures, 3 tables)

This paper contains 11 sections, 5 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Prior work
  3. Method
  4. Results
  5. Plant-level energy efficiency gain
  6. Emissions reduction from global fleet of CCGPP
  7. Brief description of combined cycle gas power plant
  8. Training of ANN models for CCGPP
  9. Optimisation problem for subsystems of CCGPP
  10. Robustness evaluation of optimal solutions
  11. Emission reduction

Figures (3)

  • Figure 1: Multi-level optimisation of CCGPP. Solver convergence for optimisation problem solved (a)(i) without and (b)(i) with Mahalanobis constraint for $Power_{\text{ Set Point}}$ of 950 MW (red) and 1090 MW (blue). (a)(ii-iii) and (b)(ii-iii) Mapping the optimal solutions, and (c) comparing the optimal solutions with the actual data of power plant.
  • Figure 2: Global CO2 reduction potential from gas power plants.
  • Figure E1: Distribution of global gas power plants.