Adaptive Multi-Objective Optimization of PV-Diesel Hybrid Systems for Climate Resilience in Semi-Arid Regions

Abstract

The electrification of semi-arid regions confronts unprecedented reliability challenges stemming from extreme climate variability challenges that climate change continues to intensify at an accelerating pace. Conventional optimization methods for PV-diesel hybrid systems often fail to account for these deep uncertainties, leading to designs that are either unreliable or economically unviable. This paper introduces a novel Adaptive Multi-Objective Genetic Algorithm with an Evolving Fitness Function (AMOGA-EFF) to optimize system design for long-term resilience. The framework is uniquely built on a comprehensive uncertainty analysis, integrating high-resolution stochastic modeling from 20-year historical weather data with an ensemble of CMIP6 climate projections (SSP2-4.5 and SSP5-8.5). To quantify performance under these stresses, we introduce two new metrics: the Climate Robustness Index (CRI) and the Economic Vulnerability Factor (EVF).

The AMOGA-EFF approach was rigorously tested through extensive simulations for three distinct sites: Ouagadougou (Burkina Faso), Jodhpur (India), and Petrolina (Brazil); where it consistently demonstrated superior performance. It yields a 35-42% improvement in system resilience with only a 6-11% increase in the levelized cost of energy (LCOE) compared to the standard NSGA-II algorithm. Under the high-emission SSP5-8.5 scenario, optimized systems maintain 89% availability during extreme weather events, in stark contrast to the 61% achieved by conventional deterministic designs. Furthermore, the resulting configurations achieve a 13.3% reduction in total life-cycle cost, primarily through a 22.1% decrease in fuel consumption. This work provides a robust methodological blueprint for designing resilient and cost-effective energy infrastructures in climate-vulnerable regions worldwide.