How to evaluate BESS revenue before building the battery

Begin with the investment decision—not the dataset

A revenue study may need to choose battery power, duration, location, grid capacity, co-location design, route to market or contract structure. State that decision first. A model built only to maximize historic revenue can select a configuration that is expensive, unfinanceable or impossible to connect.

Define the commercial boundary: asset-owning entity, market counterparty, BRP/aggregator relationship, settlement meter and who pays energy, fees, imbalance, degradation and auxiliary consumption. Revenue before these allocations is not project cash flow.

The minimum credible input set

Use an evidence hierarchy

1. Oracle or perfect-foresight simulation

This uses realized future prices to choose past dispatch. It estimates technical upper potential under the modeled constraints. It is valuable for sizing and diagnosing value, but not achievable revenue.

2. Forecast-driven historical backtest

At each historical decision time, use only the forecast and market data then available. Freeze the output, carry SOC and positions forward, apply realized outcomes, then make the next decision. This tests a repeatable policy out of sample if model selection did not use the evaluation period.

3. Forward scenarios

Historic volatility and spreads will not necessarily persist as storage, renewables, interconnection, fuel and regulation change. Use multiple price and market-access scenarios, including cannibalization. Preserve correlation between price, renewable output and activation rather than combining independent averages.

4. Shadow and live operation

Before autonomy, compare proposed actions with actual market and site conditions. Then introduce bounded live dispatch. Production evidence should report command acceptance, delivered energy, availability, exceptions and settlement—not only model revenue.

Battery sizing is a joint optimization

Power (MW) determines instantaneous capability. Usable energy (MWh) determines duration. The grid connection may cap both. A longer-duration battery can capture broader spreads or sustain services, but extra MWh create value only when opportunities, connection and cycling support them.

Compare configurations on identical scenarios. Include degradation-related loss of capacity and augmentation. For PV+BESS, size against the joint solar profile and point-of-connection limit. For balancing, include the SOC needed to sustain activation and recovery.

Sensitivities that change the decision

• Price spread and volatility compression as competing storage enters.
• Day-ahead and intraday forecast error, liquidity and transaction cost.
• Access date and qualification success for balancing services.
• Activation frequency, direction, energy and performance penalties.
• Round-trip efficiency, auxiliary consumption and temperature derating.
• Availability, connection delay and commissioning ramp.
• Cycle life, augmentation cost and residual capacity.
• PV capture-price and curtailment correlation for hybrid projects.
• Merchant, tolling, floor and revenue-share terms.

Report downside percentiles, break-even assumptions and the variables to which debt-service coverage is most sensitive. Do not hide a weak base case behind one high-volatility year.

What a decision-ready output should contain

A good report links assumptions to results: scenario definitions; source and review date; model version; dispatch constraints; revenue by stream; purchases, fees, losses, imbalance and degradation; annual cycles and throughput; availability; CAPEX/OPEX; and return metrics. It should show at least one dispatch week so reviewers can understand how revenue is created.

Keep a clear bridge from modeled schedule to executable system. Can the trader place the required orders? Can the EMS receive 15-minute targets? Does the BMS expose dynamic limits? Does the revenue meter reconcile? VoltMage’s simulation and BESS optimization workflow is designed to test commercial decisions alongside these physical questions.

Build an investment data room the model can defend

Market data should include licensing terms and revisions as well as prices. Forecast data should contain every vintage used for a decision. The technical room should include single-line diagrams, connection agreement, BESS/PCS/BMS data sheets, efficiency curves, auxiliary loads, warranty, fire and operating philosophy, point list and acceptance-test results. The commercial room should contain route-to-market terms, fee waterfall, collateral, imbalance allocation and availability obligations.

Create an assumption register with owner, source, date, unit, boundary, base/downside/upside values and approval. “Round-trip efficiency: 90%” is not sufficient without power, SOC, temperature and measurement boundary. “Two cycles per day” is not a strategy without defining equivalent-full-cycle accounting and why those cycles exist.

Red flags in a vendor revenue model

• One historic year annualized across the project life.
• Perfect future prices described as a trading strategy.
• Balancing capacity and arbitrage stacked without reserved MW or activation energy.
• Nameplate MWh used after degradation with no augmentation cost.
• 100% availability from commercial operation date.
• No bid-ask spread, market fee, imbalance or auxiliary consumption.
• Hourly simulation for a 15-minute decision and settlement process.
• No explanation of how a schedule reaches the PCS or how delivery is measured.

Use decision gates

At early development, a broad scenario screen can reject unsuitable power/duration combinations. Before FID, use current market rules, executed commercial terms and detailed site constraints. Before commissioning, replay schedules through the actual EMS interface. Before autonomous operation, run shadow mode and bounded dispatch. Each stage reduces a different uncertainty; one spreadsheet cannot substitute for all four.

A mature model is updated after construction. Commissioning reveals real efficiency, auxiliary load, ramp response and communications behavior. Those observations should replace development assumptions and become the baseline for operating forecasts.