Validate an updated equation-of-state formulation against 30+ years of well and production history before it was used for new development scenarios.

Steep topography and a shallow magmatic heat source produce strong lateral temperature gradients that simple models under-predict.

Multi-scenario natural-state and history-matched runs across the revised EOS, cross-checked against a combined-field simulation case.

Confirmed EOS fix carried through to production forecasts; resource risk range tightened for the client's development committee.

Build a natural-state model from sparse exploration data to support a well-siting decision ahead of the first deep well.

Limited well control meant permeability structure had to be inferred largely from surface geochemistry and resistivity surveys.

Ensemble of plausible permeability structures run in parallel, ranked against surface manifestation temperatures and chemistry.

Narrowed the candidate well-pad locations from five to two, with an explicit confidence range on target reservoir temperature.

Test whether developing Tecuamburro and Infiernitos as one combined project changes either resource's own forecast, using a framework where every combined-scenario sample shares the identical natural-state realisation, injection strategy and pump-sizing logic as its single-resource counterpart.

Determine whether joint development causes genuine resource-scale interaction — shared pressure support, competition for reinjection capacity — or whether independent forecasts simply sum, and prove it at the level of reservoir pressure itself rather than inferring it from power output alone.

293-sample stochastic ensemble; automated well-siting growth algorithm; source-network-coupled reinjection in Waiwera; matched sample-by-sample against each resource's independent forecast; interference tested directly via per-sample reservoir pressure at Infiernitos' own feedzones.

Combined development changes each resource's own forecast by under 1% (Wasserstein-1 distance 0.13–0.38 MWe), and the distributions are not statistically distinguishable (Kolmogorov–Smirnov, p>0.05). A real but small hydraulic interference exists — 0.18 bar median pressure reduction at Infiernitos by year 25 — an order of magnitude below natural-state uncertainty.