Solar Plant Design Basics: What PVsyst Actually Models, and What It Cannot

The short answer

PVsyst simulates the energy a photovoltaic plant will produce over a year by applying a chain of losses to irradiation data. The headline outputs are specific yield in kWh per kWp per year and performance ratio, and the credibility of both depends almost entirely on three inputs: the weather data source, the shading model, and the loss assumptions the engineer chose. Two engineers can model the same plant and differ by ten per cent, honestly.

The loss chain

Nameplate capacity is measured at standard test conditions: 1000 W/m², 25 degrees cell temperature, AM1.5 spectrum. No plant in India ever operates there. What arrives at the meter is the nameplate reduced by, roughly in order:

• Irradiation on the tilted plane rather than horizontal, which is a gain for fixed tilt at Indian latitudes.
• Shading, both far horizon and near, from rows, structures and trees.
• Incidence angle modifier, because glass reflects more at shallow angles.
• Soiling. This is the big one in India, and it is site-specific. A plant near a cement works or an unpaved road may lose several times what a coastal plant loses, and monsoon timing decides how much is washed off.
• Temperature. Module output falls with cell temperature, typically about 0.35 per cent per degree above 25. In a Rajasthan summer cell temperature can exceed 65 degrees, which is a double-digit percentage loss at the time of day when irradiation is highest.
• Module quality, light-induced degradation and mismatch.
• DC cabling losses, then inverter conversion, then transformer and AC cabling, then availability.

Performance ratio is the delivered energy divided by the energy that the irradiation in the plane of array would have produced at nameplate efficiency. A well-built Indian plant lands somewhere around 0.75 to 0.82 in year one. A report showing 0.88 in a hot dusty location is describing assumptions, not a plant.

The three inputs that decide honesty

Weather data. Meteonorm, PVGIS, NASA and purchased satellite datasets disagree, sometimes by several per cent for the same coordinates, and the difference goes straight into the yield. State the source and the years covered. A bankable report uses a dataset with an uncertainty figure attached.

Shading. A 3D near-shading scene takes effort and changes the answer, particularly for row spacing decisions and for anything near a building. A model with no near-shading scene is a spreadsheet with a nice interface.

Soiling and availability. These are the two most commonly set to a default and forgotten. Soiling should come from the site, and availability should reflect the O&M contract you actually have.

Where an electrical engineer fits

The simulation is one deliverable among several, and the rest is electrical engineering: string sizing against inverter voltage windows at record low and high temperatures, DC and AC cable sizing and voltage drop, earthing and lightning protection, protection coordination, and the SCADA and metering that proves what the plant produced.

That last one is where automation engineers come in. A utility-scale plant has a plant controller doing active and reactive power control to grid code, inverters on Modbus TCP, weather stations, and a SCADA layer feeding both the O&M team and the offtaker's settlement. It is a control system with panels on it.

Courses: PVsyst solar power plant design for the simulation and design side, ETAP power system analysis for protection and load flow, and electrical design engineering for the wider design discipline.

Frequently asked questions

Is PVsyst the industry standard? For yield assessment, yes, and lenders expect it. Helioscope and others are used for faster layout work.

What specific yield should I expect in India? Broadly 1,400 to 1,700 kWh/kWp per year for fixed tilt depending on location, and higher with tracking. Treat any single number quoted without a location and a dataset as marketing.

Does the automation side need solar-specific knowledge? Some. Grid code compliance, plant controller behaviour and inverter communications are specific enough that experience is valued, and the underlying skills are ordinary industrial control skills.