Stochastic operational comparison of PV-integrated vanadium redox flow and lithium-ion battery systems for tropical remote microgrids: A GA-LP-MC framework applied to Thursday Island

Remote tropical island microgrids dependent on barged diesel face a contested storage-chemistry choice between lithium iron phosphate (LFP) and vanadium redox flow batteries (VRFB). This study asks whether VRFB's durability advantage yields a lower levelised cost of storage (LCOS) under contemporary LFP economics, using Thursday Island (Torres Strait; 24.6 GWh yr −1 ; landed diesel A$2.00 L −1 ; 27 °C annual mean) as a representative case. A genetic algorithm with hourly linear-programming dispatch sizes each technology independently; multi-year Arrhenius degradation enters the dispatch as year-on-year capacity constraints with sampled parameters; and a 10,000-run Monte Carlo propagates correlated uncertainty, with operations reported under realistic rolling-horizon control and bounded by an idealised perfect-foresight case. Under verified 2030-aligned parameters the central hypothesis — that VRFB would be cheaper — is not supported: PV–LFP–Diesel achieves a mean LCOS of 404 AUD/MWh against 693 for PV–VRFB–Diesel, with realistic control adding only ≈3% (the “optimism gap”). The advantage is decisive: all 10,000 paired draws favour LFP (mean difference − 289 AUD/MWh; 95% CI [−290, −287]), and the ranking holds across all factorial cells, carbon prices, cost trajectories and site temperatures. The advantage derives from round-trip efficiency (≈62% of the gap) and lower unit-energy capital cost, despite a heavier replacement burden. Absolute costs are bounded by the realism of a synthetic, statistically calibrated load; this conservatively bounded null result directly informs remote and First Nations clean-energy investment under the First Nations Clean Energy Strategy 2024–2030.