Techno-Economic and Simulation-Based Assessment of Onshore Flare Gas Utilization for Rural Electrification and National Grid Integration in Nigeria

Abstract Nigeria remains one of the leading gas flaring countries globally, with onshore oil operations contributing about half of the routine flaring despite persistent electricity deficits and low rural electrification levels. In 2024, Nigeria experienced a 12% increase in gas flaring, totalling approximately 6.5 billion cubic metres (BCM) that year, an increase that outpaced oil production growth and exceeded twice the global average flaring intensity. In the first half of 2024 alone, an estimated average of 148.7 million standard cubic feet of gas per day was flared, corresponding to a power generation potential exceeding 3.4 GW, sufficient to supply electricity to approximately 3 million households. Despite these losses, reliable electricity access remains constrained, with rural electrification rates estimated at 25 to 35%, underscoring a significant energy access gap. This study presents a novel AI-enhanced techno-economic and simulation-based framework for assessing onshore flare gas utilization for rural electrification and national grid integration in Nigeria, aligned with national flare commercialization and energy transition objectives. Historical production and flaring datasets are analysed using tree-based ensemble learning models, specifically Random Forest and Extreme Gradient Boosting, to forecast flare gas availability under varying operational conditions. Long Short-Term Memory (LSTM) neural networks are employed to capture temporal dynamics in both flare gas supply and electricity demand for rural mini-grid and grid-connected power systems. To enable efficient large-scale scenario evaluation, ML-based surrogate models are developed to approximate gas-to-power system performance and associated economic outputs. AI and ML predictions are integrated into a techno-economic simulation framework to evaluate net present value, internal rate of return, payback period, and levelized cost of electricity, with project uncertainty quantified through Monte Carlo simulation. Results demonstrate that LSTM reduces forecasting RMSE by 65.3% relative to ARIMA baselines, both utilization scenarios yield IRR values exceeding 12% WACC, and the probability of positive NPV is 87.3% and 96.1% for mini-grid and grid-connected scenarios respectively. The framework provides a scalable decision-support tool aligned with Nigeria's flare gas commercialization and energy transition objectives.