Capability · Techno-Economic Modeling · NREL SAM-based

Energy yield numbers that hold up at tax equity diligence.

Engine System Advisor Model (SAM), built and validated by NREL. Free for anyone to download. The differentiation is not the tool. It is who runs it, what assumptions go in, and whose interests the output serves.

A SAM model on its own is a starting point, not an answer. The default assumptions are reasonable averages, the resource files cover everywhere and nowhere in particular, and the loss stack is whatever the user typed in. We rebuild the model from resource data up, calibrate it against actual operating projects, and deliver a yield distribution your tax equity counsel and lender's IE will fund against.

vs vendor proformas

Modules and inverters get sold on optimism.

Manufacturer energy estimates assume best-case soiling, mismatch, and availability. We back-calculate the implied loss stack and tell you exactly where they stretched.

vs EPC pro forma models

The EPC's model exists to defend the contract.

An EPC's energy estimate sets the performance guarantee floor. Theirs runs low to make the LDs unreachable. Ours runs honest because we own the owner-side P&L.

vs in-house dev teams

Defaults are not assumptions.

Most dev teams run SAM with stock loss factors and call it a day. We tear the loss stack down to first principles, cite each input, and produce sensitivity bounds a lender will accept.

vs paid IE reports

The IE works for the developer.

Independent engineers are paid by the sponsor. We sit on the owner side of the table when the IE delivers, and we know what numbers they will push back on before they do.

Seven workstreams. From resource data to lender's IE briefing.

Each workstream produces a defensible artifact. Together they form the techno-economic file your capital stack will be priced against.
01
Resource and weather data

NSRDB, TMY3, Solargis, ground-station data. Multi-year, P50, and P90 weather files compared and reconciled. Climate adjustment for long-term trend, not just historical mean.

You get: resource memo, weather file selection rationale, multi-source variance analysis.
02
System design and topology

Module-inverter pairing, DC/AC ratio optimization, tracker type, ground coverage ratio, bifacial gain. Configuration trade-offs run in parametric mode, not picked from a vendor catalog.

You get: optimized configuration package, technology trade-off matrix, GCR sensitivity output.
03
Loss factor stack

Soiling, snow, shading, mismatch, wiring, transformer, availability. Each loss factor cited to a published source or calibrated to operating data. No round numbers, no defaults.

You get: loss factor assumption book with citations, defensible bounds for each input.
04
Distributional yield analysis

P50, P75, P90, P99 generated from Monte Carlo over weather years and degradation paths. Performance ratio carried out over the asset life. This is the output tax equity will price the deal against.

You get: yield distribution report, exceedance probability curves, year-by-year production forecast.
05
Battery dispatch and hybrid optimization

Behind-the-meter peak shaving, front-of-meter merchant arbitrage, capacity firming, standalone storage ITC modeling. Dispatch strategy modeled against actual rate structures and tariff data.

You get: dispatch strategy memo, revenue stream allocation, sizing optimization output.
06
Parametric and sensitivity analysis

Tornado diagrams on the inputs that actually move the financial outcome. Scenario sweeps on technology, financing terms, and incentive structures. Breakeven analysis on the variables that matter to the lender.

You get: sensitivity report, tornado diagram, breakeven analysis, scenario comparison matrix.
07
Validation and calibration

Modeled output checked against operating data from comparable projects in the same resource zone. Year-over-year actual vs forecast variance benchmarked. Calibration adjustments documented for the IE.

You get: validation memo, calibration adjustment log, lender's IE briefing pack.

Modeling decision gates · Sub-flow within DG1 Feasibility

DG1.1
Resource lock
Weather files selected, multi-source reconciled, climate adjustment applied.
DG1.2
Design lock
Topology and BOM frozen at optimized configuration, parametrics archived.
DG1.3
Loss stack lock
Loss factors cited and bounded, P50 through P99 distribution generated.
DG1.4
Bankability
Model handed to lender's IE, sensitivity package complete, validation closed.

What each exceedance number is actually for

P50
Expected

Median annual production. The number sponsor pro formas typically run. Useful for IRR projection, not for sizing debt.

P75
Conservative case

75 percent probability of exceedance. Often the figure tax equity asks for in base case sizing.

P90
Debt sizing

90 percent probability of exceedance. The number lenders typically use to size senior debt and DSCR.

P99
Stress case

One-year-in-a-hundred low production scenario. Used in stress testing, downside modeling, and DSCR floor analysis.

What sits on your desk when we're done.

A techno-economic file that holds up under tax equity diligence, lender's IE review, and IRS substantiation. Everything cited, nothing proprietary, all of it transferable to a successor advisor if you ever choose to bring modeling in-house.

  • Calibrated SAM project file Native .sam file, fully documented, ready to hand to your IE or successor advisor.
  • Resource and weather memo Source files, multi-source reconciliation, climate adjustment rationale.
  • Loss factor assumption book Each loss factor cited and bounded. Defensible to a tier 1 lender's IE.
  • P50 / P75 / P90 / P99 yield distribution report Exceedance curves, year-by-year production, long-term degradation path.
  • Parametric sensitivity report Tornado diagram, scenario sweeps, breakeven analysis on the variables that move financial outcome.
  • Battery dispatch strategy memo Where applicable. Dispatch logic, revenue stream allocation, sizing rationale.
  • Lender's IE briefing pack Validation against operating data, calibration log, anticipated IE pushback list with prepared responses.

Have a model that needs to clear an IE this quarter?

Initial scoping calls are complimentary. We work in NREL SAM, PVsyst when the project requires it, and PySAM for portfolio-scale automation.

Start a modeling scope →