Methodology & trust

Transparent assumptions, defensible results

The product is the validation, not an afterthought. Here is exactly where our data comes from, what we assume, and what we deliberately do not yet claim.

Public data foundation

Official first, open second, user-supplied always allowed — every run fingerprinted and versioned.

Honest status: the local engine's default resource is a synthetic, seeded clear-sky model (physically-based, fully reproducible). Live NREL data is now wired in for several flows: the REopt V3 engine pulls real PVWatts/NSRDB solar by coordinate, the modeler's load generator builds an 8,760-hour profile from DOE reference buildings (REopt simulated-load), and the standalone calculators query PVWatts, Solar Resource Data, Utility Rates, and Alternative Fuel Stations directly. Your own 8,760-hour load CSV is supported too.
DomainPrimary sourceWhat it provides
Solar / weatherNSRDBTime-series solar & meteorological data, TMY and time series
Solar / weatherNASA POWERFree global solar & meteorological parameters, API access
Solar / weatherPVGISOpen global PV resource/performance, no registration
Solar / weatherERA5Global hourly reanalysis 1940–present for backfills & uncertainty
Load profilesNREL End-Use Load ProfilesCalibrated 15-minute residential & commercial archetypes
TariffsOpenEI URDBUtility rate structures via API (CC0 where noted)
IncentivesDSIREU.S. incentives & policy database
Cost baselinesNREL ATBTechnology cost/performance defaults (override per project)
Fuel pricesEIADiesel & petroleum price series for scenario sensitivities

Documented assumptions

Year-1 representative dispatch. The 8,760-hour dispatch is run for a representative year; fuel and O&M are escalated over the horizon. Headline LCOE/NPC therefore reflect a year-1 resource — per-year re-dispatch with PV degradation is an explicit roadmap item, not silently assumed.
Synthetic, seeded resource. Demo weather is synthesized from a physics-based clear-sky model plus a seeded clearness index. Production runs ingest measured NSRDB/PVGIS/ERA5 time series for the exact site.
Constant-dollar (real) economics. All cash flows are discounted at the real rate i = (1+nominal)/(1+inflation) − 1. Fuel escalates at a real rate. This keeps the assumption sheet transparent.
Single-bus energy balance. The wedge models energy balance, not AC power flow. Voltage/thermal feasibility on the physical network is on the roadmap, not assumed here.
Renewable-fraction definition. Renewable fraction is the PV share of served load — direct PV plus battery discharge apportioned by the energy origin that charged it (PV vs genset). COE excludes the value-of-lost-load penalty (reported separately); NPC includes it.

Clean-room & licensing

Clean-room build

Built entirely from public documentation, open literature, and original implementations. No proprietary code, no result scraping, no UI cloning. Models are derived from first principles and published methods.

Open-source references

REopt.jl · Apache-2.0
Cross-check reference engine for PV/storage/generator optimization.
PySAM · BSD-3-Clause
Physics & finance cross-checks against NREL’s SAM core.
oemof.solph · MIT
Alternative LP/MILP modeling layer for custom formulations.
What we don’t yet claim. This wedge does not perform AC power-flow, protection, or cable-sizing studies, and is not a controller-grade real-time dispatcher. Those are deliberately on the roadmap — we ship the rigorous techno-economic core first rather than overstate scope.