PV simulators comparison: Overview

In this document

This article introduces the comparison of Solargis Evaluate against five other PV yield simulators across the full PV simulation chain. It summarizes the comparison setup, the main results, and the methodology differences across the simulators, and points to dedicated articles for each stage of the simulation chain.

Overview

PV yield simulation software differ in the data they consume, the models they implement, the assumptions they make, and the way they report intermediate results. For an investor, an independent engineer, or a developer choosing between simulation software, these differences affect the simulated energy yield and impact design decisions.

The Comparison of PV simulators series compares Solargis Evaluate against four widely used PV simulators: PVsyst, the National Laboratory of the Rockies (NLR) System Advisor Model (SAM), the open-source pvlib Python library, and DNV's SolarFarmer. Solargis Prospect, the simplified site assessment tool from Solargis, is included as a sixth comparator to show how it relates to Solargis Evaluate.

The comparison is methodology-driven: each simulator runs the same six test sites and the same four PV system configurations, and the results are compared topic by topic across the full simulation chain. The goal is to characterize where simulators differ in approach and in output, not to validate any simulator against measured plant data. The methodology references industry-standard, peer-reviewed models.

PV simulators compared

Where possible, each compared simulator is configured to use the same input data, the same model selections, and the same system parameters to observe differences in methodology rather than configuration choices.

PV simulator	Version	Type
Solargis Evaluate	v2.5	Commercial, cloud-based
Solargis Prospect	v2.5	Commercial, cloud-based
PVsyst	v8.1.4	Commercial, desktop
System Advisor Model (SAM)	v2025.4.16	Free, desktop (NLR)
pvlib	v0.13.1	Open-source Python library
SolarFarmer	v1.6	Commercial, desktop and cloud (DNV)

Table 1: Compared PV simulators.

What is compared

The comparison covers the three stages of the PV simulation chain:

Inputs: Solar resource, meteorological and environmental data, and scene definition.
Optical simulation: Plane-of-array irradiance, including transposition, rear-side irradiance for bifacial PV modules, horizon shading, near shading, and PV module losses (soiling, snow, angular, spectral).
Electrical simulation: DC/AC conversion, DC cabling, inverter, auxiliary, AC cabling, transformer, grid connection, and unavailability.

These stages are unpacked across eight dedicated articles in this series:

Setup and test methodology: test sites, test system configurations, simulator versions, and statistical metrics.
Solar data and system configuration: data inputs, scene modelling, and simulation chain variables.
Irradiance modeling: GTI transposition, rear-side irradiance for bifacial PV modules, horizon shading, near shading.
Optical losses: soiling, snow, angular, spectral.
Electrical modeling of the DC side: electrical chain from PV cell to inverter AC output.
Electrical modeling of the AC side: electrical chain from inverter AC output to grid delivery.
Conclusion: overall results of the comparison with interpretation
Detailed results and analysis: a deep dive into the most salient results with a detailed analysis

Results summary

Across the full PV simulation chain, from solar radiation to power at the grid connection point, pvlib, PVsyst, and SAM all agree relatively closely with Solargis Evaluate. The median bias stays inside ±3.2% for every simulator and mounting configuration, most values (44/72 test cases) fall inside ±2%, and the largest single deviation stays under 8%. All three simulators tend to calculate slightly less energy delivered to the grid than Solargis Evaluate, but none of them sits consistently below it: each one crosses to the other side at some sites and mounting configurations. The direction and size of the difference depend on the mounting configuration and the site climate, and the differences are systematic rather than random scatter, traceable to specific steps in the simulation chain. These effects in individual simulation steps are investigated in the next articles in this series.

For the full breakdown of the differences in the final output of the simulation (energy delivered to the grid) and the underlying analysis of why these differences occur see the Conclusion article. A sample result is shown in Figure 1 below – the total energy delivered to the grid as a bias of the three compared simulators against Solargis Evaluate.

Note: Solargis Evaluate is used as the mathematical reference series for bias and RMSE calculations across this comparison series. These statistics require a reference, and Solargis Evaluate is one suitable choice. This convention does not imply Solargis Evaluate is more accurate than the other simulators.

Figure 1: Total energy delivered to the grid bias for pvlib, PVsyst, and SAM against Solargis Evaluate, as a percentage of the Solargis Evaluate value, by mounting configuration across the six test sites.

How the simulators compare on key capabilities

Beyond the numerical results, the simulators differ structurally in the models they implement, the resolution they support, and the workflow they offer. Table 2 summarizes capabilities that users of PV simulation software commonly look at when choosing the right solution.

The table shows:

Where the simulators are functionally equivalent (for example, all simulators support sub-hourly simulations in some form).
Where they differ in approach (ray tracing versus view factor, cell-level versus submodule IV curves).
Where each tool has its own characteristic strengths.

Users should choose the simulator whose strengths align with their project requirements and workflow.

Capability	Solargis Evaluate	Solargis Prospect	pvlib	PVsyst	SAM	SolarFarmer
Built-in solar and meteorological database	Yes	Yes	No	No	No	No
Native temporal resolution	15-minute, 1-minute	15-minute input, hourly output, 12 daily profiles	Hourly or sub-hourly	Hourly or sub-hourly	Hourly or sub-hourly	Hourly or sub-hourly
Built-in soiling model	Yes	No	Yes	No	No	No
Built-in snow model	Yes	No	Yes	No	Yes	No
Visual scene editor	Yes	No	No	Yes	Yes	Yes
Shading computation method	3D ray tracing	Advanced view factor	View factor	View factor	View factor	View factor
IV curve modeling	Cell level	Cell level	Configurable	Submodule level	Submodule level	Submodule and string level
Transient thermal correction for PV cell	Yes	No	Available	Available	Available	No
Inverter model	SANDIA	SANDIA presets, or Euro-efficiency	SANDIA, Anton Driesse grid connected, PVWatts	PVsyst model	SANDIA, datasheet, NREL part-load	Efficiency curves
Transformer model	Iron and copper losses, two stages	Simple efficiency	Iron and copper losses	Iron and copper losses, two stages	Iron and copper losses (as factors)	No-load and full-load efficiency
Integrated reporting and analysis	Yes	Yes	No	Yes	Yes	Yes

Table 2: Summarized capabilities of compared PV simulators.