Beyond a single solution: Liquefied natural gas process optimization using niching-enhanced meta-heuristics

Abstract

The escalating global energy demand and the imperative to mitigate climate change necessitate optimizing the natural gas liquefaction process to enhance energy efficiency, reduce costs, and improve sustainability. Traditional approaches typically focus on identifying a single high-quality solution. However, such methods often overlook other potential solutions that are equally good or comparable to the global optimum. These solutions provide alternative, economically viable options for engineers, enabling more insightful decisions. Motivated by this research gap, we propose the application of multi-solution meta-heuristics for the dual-effect single-mixed-refrigerant (DSMR) process, a novel approach in the field of liquefied natural gas (LNG) process optimization. To comprehensively evaluate algorithm performance, we introduce a Quantity-Quality comparative evaluation method, which ranks algorithms based on both the number and quality of solutions identified without requiring prior knowledge of the global optimum. We highlight the effectiveness of multi-solution meta-heuristics in identifying multiple feasible high-quality solutions outperforming commonly used single-solution approaches. Our assessment encompasses exergy efficiency and economic viability, with results showing that 15 out of 17 multi-solution variants surpass the popular genetic algorithm (GA) in global optimization, achieving up to 3% reduction in energy consumption compared to the state-of-the-art solution. Additionally, all niching-enhanced meta-heuristics consistently identified more high-quality solutions than single-solution meta-heuristics. An exergo-economic analysis of the top 15 solutions revealed exergy efficiency improvement of up to 37.73%, with capital cost reductions ranging from 0.27% to 5.22%, and operating cost reductions from 0.24% to 3.1%, compared to the base case. The code is available on GitHub.