Renewable Energy Simulation Best Practices: Accuracy, Validation, and Bankability

A renewable energy simulation is only as good as the assumptions that go into it. A 2% error in loss factor assumptions can change project IRR by 1-2 percentage points. An incorrect weather data set can shift P50 yield estimates by 5-10%. For a $200M project, that translates into millions in valuation uncertainty.

This guide covers the best practices for producing bankable renewable energy simulations — covering weather data selection, simulation resolution, loss factor benchmarking, model validation, sensitivity analysis, and documentation standards.

Weather Data: TMY, P50, P90, and Site-Specific Measurements

The single biggest driver of simulation accuracy is the weather data input. Using the wrong data set — or failing to account for interannual variability — is the most common cause of post-COD performance shortfalls.

TMY (Typical Meteorological Year) data represents a composite of "typical" months selected from a multi-year database (typically 10-30 years). TMY data is appropriate for:

• Technology comparison (which inverter, which module, which tracker)
• General project feasibility screening
• P50 baseline energy estimates

P50 and P90 analysis uses the full multi-year time series to calculate the probability distribution of annual energy production. The P50 value is the median expected generation; the P90 value is the generation that will be exceeded 90% of the time. Lenders typically require P50 for valuation and P90 for debt service coverage ratios (DSCR).

Site-specific measurement campaigns are the gold standard. A minimum of 12 months of on-site pyranometer and anemometer data, correlated to long-term satellite records, provides the most bankable weather input. The measured data corrects for local microclimate effects that satellite databases (PVGIS, NSRDB, Solargis) may miss.

Simulation Resolution: Sub-Hourly vs Hourly vs Monthly

Simulation resolution directly impacts accuracy, especially for hybrid systems with battery storage. The choice depends on the application:

• Hourly (8760 timesteps/year): Minimum acceptable resolution for any bankable analysis. Adequate for PV-only projects where clipping losses and temperature effects dominate. For BESS projects, hourly simulation captures diurnal charge/discharge patterns and energy arbitrage but may miss sub-hourly frequency regulation events.
• Sub-hourly (8760-52560 timesteps): Required for projects with frequency regulation, fast-ramping constraints, or detailed battery dispatch. Sub-hourly simulation captures intra-hour SOC swings, instantaneous C-rate effects on degradation, and ramp-rate compliance.
• Monthly (12 timesteps): Only appropriate for preliminary feasibility screening. Monthly simulation cannot capture battery cycling dynamics, clipping effects, or time-of-use optimization.

Energy Optima runs its core simulation at hourly resolution (8,760 timesteps per year) and supports sub-hourly dispatch for BESS applications requiring detailed EMS modeling. For more on dispatch strategies, see our EMS dispatch guide.

Loss Factor Benchmarking and Validation

Loss factors are frequently the weakest link in renewable energy simulations. A 1% discrepancy in assumed soiling loss or wiring loss can have a material impact on energy projections. Best practice is to benchmark each loss category against industry data and project-specific conditions.

Our PV loss waterfall analysis details the 10 categories of energy loss. Key benchmarking values for well-designed utility-scale projects:

• Soiling loss: 1-3% in arid climates with regular cleaning; 3-7% in high-dust environments without cleaning
• Shading loss: 0.5-2% for well-designed layouts; >5% for constrained sites
• IAM (incidence angle modifier): 1-3% depending on module glass and anti-reflective coating
• Spectral loss: 0.5-1.5% depending on climate (higher in tropical regions)
• Temperature loss: 3-8% depending on climate, mounting, and module temp coefficient
• Module quality/mismatch: 1-2% for nameplate tolerance and LID/LeTID
• Wiring loss (DC): 1-2% (design target < 1.5%)
• Inverter/MPPT loss: 1-2% including MPPT efficiency and conversion
• Clipping/curtailment: 0.5-5% depending on ILR (inverter loading ratio)
• Transformer/auxiliary: 1-2% for station transformer and parasitic loads

Any simulation where the sum of loss factors is outside the typical range of 12-25% total loss should trigger a validation review. An experienced reviewer should be able to identify which loss category to question first.

Degradation Modeling and Long-Term Trends

Linear degradation assumptions (e.g., 0.5% per year for PV modules, 2% per year for batteries) are the default in many tools, but they introduce significant error. Our BESS degradation guide shows how a linear 2%/yr assumption underestimates energy throughput by 18% over 25 years compared to manufacturer-specific 3D interpolation.

For PV modules, degradation is also non-linear. LID (light-induced degradation) causes a 1-3% drop in the first 100-200 hours of operation. LeTID (light- and elevated temperature-induced degradation) can cause an additional 1-5% in PERC modules over the first 2-3 years. After these initial effects, PV degradation typically stabilizes at 0.3-0.7%/year.

Best practice: use manufacturer-specific degradation curves for both PV and BESS, not generic annual degradation assumptions. For modules, request the PID and LeTID test data from the manufacturer. For batteries, use the cycle life and calendar aging tables.

Energy Balance Verification

A fundamental sanity check for any simulation: the energy balance must close.

Energy Conservation: Generation + Grid Import = Load + Grid Export + Losses + Storage Change

For a hybrid PV+BESS system, verify that every MWh of generation is accounted for — either consumed by the load, exported to the grid, stored in the battery, or lost. A gap of more than 1-2% in the energy balance indicates a modeling error that must be resolved before the results are usable for financial analysis.

Energy Optima includes an energy balance verification tool that reports the closing error for each simulation year. Closing errors below 0.5% are achievable for well-configured models.

Sensitivity Analysis and Uncertainty Quantification

Bankable energy projections require more than a single point estimate. A sensitivity analysis shows which variables have the most impact on project economics and where uncertainty is concentrated.

Key variables to include in any sensitivity analysis:

• Irradiance (GHI/DNI): ±5-10% variation depending on weather data confidence
• Temperature: ±2°C variation in ambient temperature
• Degradation: ±0.2%/yr variation in PV or BESS degradation
• Soiling: ±1% absolute variation in soiling loss
• Availability: ±1-3% variation in plant uptime
• Energy pricing: ±$10-20/MWh in tariff or PPA price
• Discount rate: ±0.5-1% in WACC
• CAPEX: ±10% variation in equipment and installation costs

Tornado charts showing the impact of each variable on IRR or LCOE are standard deliverables in independent engineer (IE) reports. A well-structured sensitivity analysis can identify which risks need mitigation and which are within acceptable tolerance.

Documentation Standards for Bankability

Independent engineers (IEs) and lenders require thorough documentation of all simulation assumptions. A bankable simulation report should include:

• Weather data source and methodology — TMY version, satellite provider, measurement correlation, P50/P90 methodology
• Component datasheets — Manufacturer and model for each component, with key parameters listed
• Loss factor justification — Each category with source data or engineering judgment basis
• Degradation curves — Source of degradation data (manufacturer tables, published studies)
• Dispatch assumptions — EMS strategy, charging/discharging logic, reserve requirements
• Financial assumptions — Discount rate, inflation, tariff escalation, tax incentives
• Sensitivity results — Tornado chart or spider diagram for key variables

Energy Optima's reporting engine generates three audience-specific report formats: Engineer (full technical detail), Investor (financial summary with charts), and Owner (executive summary with KPIs).

Model Validation Against Operational Data

The gold standard for simulation credibility is validation against actual operational data. For projects under development, this may not be possible — but for a platform like Energy Optima, validation against real projects builds trust in the modeling engine.

Validation techniques include:

• Back-casting — Simulating an existing operational project and comparing results to SCADA data
• Cross-validation — Comparing results from Energy Optima to results from PVsyst or SAM for the same inputs
• First-principles checks — Manual calculation of specific yield, capacity factor, and loss budget
• Banking context — Comparing P50/P90 estimates to actual production for similarly configured projects

For more on how Energy Optima's simulation accuracy compares to other tools, see our comparison guide.

How Energy Optima Supports Best Practices

Energy Optima was designed from the ground up around these best practices:

• Multiple weather data sources: PVGIS TMY and NSRDB with 15+ year historical time series for P50/P90
• Hourly simulation with sub-hourly dispatch: 8,760-timestep energy simulation with configurable EMS resolution
• Manufacturer-specific degradation: 147+ battery models with 3D interpolation, not linear assumptions
• 10-category loss waterfall: Granular loss modeling with industry benchmarks for validation
• Energy balance verification: Automatic closing error reporting for each simulation year
• Sensitivity analysis: Built-in tornado charts and scenario comparison
• Audience-specific reporting: Engineer, Investor, and Owner report formats with full assumption documentation

For financial projections, see our BESS financial modeling guide. For loss modeling details, see the PV loss waterfall analysis.