Model A. Cumulative present-value cost of staying on LPG vs converting one household to solar-induction over 25 years. The two lines cross at the break-even year — if they cross at all under the chosen assumptions.
Cumulative present-value cost per household
Model D. 4,000 simulations drawing oil drift, volatility, discount rate, FX and battery learning from distributions centred on your sliders. ΔNPV > 0 means solar wins for that run.
Distribution of ΔNPV per household
LPG present-value cost minus solar present-value cost. Mass to the right of zero is the viable region.
Sensitivity scenarios. Expected net benefit and viability probability under the four named scenarios from §16.2, each a fixed bundle of assumptions. Tap a scenario in the rail to load its sliders.
Expected ΔNPV by scenario
Outer islands, social view · 2,500 runs each · positive = solar-induction is the better deal
Probability solar-induction is viable
Share of runs with ΔNPV > 0
§16.1 Tornado. Each bar swings one assumption across its low–high range while holding the rest at your current settings. The widest bars are where your data-collection effort matters most.
What moves ΔNPV the most
Model B. Forecast household LPG bill under two stochastic processes. Geometric Brownian motion lets prices wander with fat upper tails; Ornstein–Uhlenbeck pulls them back toward a long-run mean.
Geometric Brownian motion — wide tails
Median with 25–75% and 5–95% bands · drift follows your oil-growth slider
Ornstein–Uhlenbeck — mean-reverting
Same starting bill, constrained around a long-run equilibrium
Model C. A 2 kW array generates far more than a household cooks — so the solar resource is never the binding constraint. The constraint is timing: evening cooking needs storage, and storage is what breaks the budget.
Annual generation vs cooking demand
Generation degrades ~0.5%/yr; surplus above the demand line is exportable or stored
Self-consumption by cooking pattern
How much solar is used directly (η_match) before storage — and the battery each pattern needs
§17 Subsidy design. Government outlay over a 10-year rollout under three financing structures. Battery learning is baked into later cohorts, so phasing genuinely lowers the bill.
Annual government outlay by option
A — full subsidy, all 94,424 HH · B — outer-island priority, 52,000 HH first · C — blended, government funds panels only