What is PVSyst? An introductory article to understand the approach to power generation forecasting

PVSyst is software for the study, capacity design, simulation, and data analysis of solar photovoltaic (PV) power generation systems. It handles not only grid-connected systems but also stand-alone and pumped-storage applications, and includes various features for meteorological data, equipment databases, solar irradiance, loss modeling, and performance evaluation. The official documentation also presents it as a tool that supports the entire workflow from preliminary-stage design through detailed time-step simulations, result comparison, and report generation.

What practitioners searching "What is PVSyst" really want to know is not simply the name or its main features, but the way of thinking: how an energy yield prediction is built on what assumptions, where sources of error hide, and which figures to look at to judge plausibility. Energy yield prediction is the process of sequentially incorporating irradiance conditions, installation conditions, temperature, shading, electrical losses, and operational constraints. The shortcut to understanding PVSyst is not to be overwhelmed by the number of screens, but to methodically understand at which stages the forecast is pared down.

Table of Contents

• What is PVSyst?

• Why is forecasting power generation difficult?

• Input conditions that form the basis of the forecast

• Approach to deriving effective irradiance from solar irradiance

• How to account for temperature and electrical losses

• Why shading and layout can significantly affect forecasts

• Key items to always check in the report

• Practical steps to improve forecasting accuracy

• Summary

What is PVSyst?

Put simply, PVSyst is not a tool for estimating solar power generation by "back‑calculating from system capacity," but a tool for "reproducing how energy is produced from the site's meteorological and system conditions and at which stages it is lost." Even the official description is structured so that, at the conceptual design stage, it produces rough monthly estimates with few inputs, while at the detailed design stage it moves on to hourly simulations using orientation and tilt, array configuration, shading, and meteorological data. In other words, PVSyst is easier to understand if you regard it not as a box that only outputs a generation figure, but as a design‑support framework for progressively refining assumptions.

The reason this way of thinking is important in practice is that, in the early stages it is realistic to compare candidate options with sufficient accuracy, and when a project becomes concrete you refine the detailed conditions and update the predictions. If you try to input everything in detail from the start, input accuracy cannot keep up unless site conditions are settled. Conversely, if you proceed to contract or profitability decisions based on rough estimates, later shading conditions, temperature conditions, and wiring and control conditions will take effect and result in large deviations from the assumptions. The essence of PVSyst is not to fix the generation forecast in a single pass, but to be able to raise the quality of the forecast as the maturity of the assumptions increases.

Furthermore, the official documentation shows that it can handle a large number of simulation variables as a result, allowing results to be reviewed by month, day, and hour and design proposals to be compared. What this implies is that PVSyst’s value lies not in the mere annual generation figure, but in being able to explain why that figure was obtained. What is questioned in design meetings and internal approvals is not only the final energy output but also the validity of the assumptions, the breakdown of losses, and where there is room for improvement. The ease with which materials can be organized to withstand that accountability is why PVSyst is valued in practical work.

Why is power generation forecasting difficult?

The difficulty in predicting solar power generation lies in the fact that the amount generated is not determined by a single factor. A common misconception is that multiplying annual irradiance by the installed capacity will yield an approximate energy output, but in reality many corrections are applied before and after that. First, horizontal-plane irradiance does not directly reach the generation surface. The irradiance received changes with azimuth and tilt, and horizon masking by terrain and surrounding objects and near-field shading also come into play. Furthermore, light incident on the module surface does not simply convert into electrical power: angle-dependent reflection, temperature rise, low-irradiance characteristics, mismatch, wiring losses, conversion losses, and output limiting by control all take effect in sequence. The reason PVSyst’s loss diagram is emphasized is precisely to visualize these multi-stage attenuations.

Also, meteorological data itself is not perfect. Official documentation also states that weather data can be distorted by sensor accuracy, logging, and processing issues, and that when importing proprietary data you should carefully check for time shifts, units, outliers, and consistency with clear-sky models. In other words, power generation forecasting is not only about choosing the correct formula but also about inspecting whether the input data itself can be trusted. In the field, even if simulation conditions are meticulously prepared, a mere time shift in the weather data can change how morning and evening shading and peak output appear, and as a result can distort the interpretation of annual values.

Another important point is that a forecast is not magic that predicts future measured values; it is a model that, under a set of assumptions, shows "under these conditions, it should be about this much." If the meteorological year differs, solar irradiance and temperature will change, and the way soiling develops after installation, how the system is operated, and the frequency of output curtailment also vary from project to project. Therefore, when reading PVSyst's numbers, you should not compare them one-to-one with a single year's actual results and talk about hits or misses; you must always check what assumptions the forecast was based on. Simply adopting this stance will make your interpretation of simulation results considerably healthier.

Input conditions that form the starting point for predictions

The starting point for power generation forecasting is defining where, in what orientation, and what kind of system will be installed. Official documentation likewise outlines a workflow in which, during detailed design, azimuth and tilt are decided, the mounting configuration—such as tracking or row layouts—is specified, and the system configuration is entered before running time-based simulations. What is important here is that the geometric conditions of the installation surface are as significant as the equipment ratings. Conditions such as whether the array faces south or east–west, what the tilt is, whether row spacing is sufficient, and whether the horizon is obstructed greatly affect the time series of incident solar irradiance. In other words, the entry point for generation forecasting should be defining "how sunlight will reach the installation" before equipment selection.

The next most important factor is meteorological data. PVSyst provides features for creating geographic locations, importing meteorological data, comparing and visualizing it, and performing quality checks, and it assumes meteorological data are prepared before simulation. The official documentation also states that hourly data may be representative-year data or synthetic data generated from monthly averages. In practice this difference is often overlooked, but whether a time series was made from representative monthly values or from more detailed observational series changes how peaks and fluctuations are handled. While that may be sufficient for rough estimates of annual energy production, if you want to examine peak clipping or short-term output behavior, you need to ascertain the nature of the input data.

Furthermore, in practice, inputting a system configuration does not end with simply registering nominal values. Because factors such as the number of series and parallel connections, the range of input voltages, combinations with power conversion equipment, conditions for energy storage and self-consumption, and grid-side constraints are involved, the outcome can vary even for the same installed capacity if the design approach differs. The official guidance also shows that detailed design results include a large number of simulation variables, and that for grid interconnection it can handle constraints related to self-consumption, storage, and even the injection point. Therefore, when reading power generation forecasts, you should not compare based only on installed capacity but also check which constraint conditions were applied.

Approach for Converting Solar Radiation into Effective Insolation

One of the first major hurdles in understanding energy yield forecasting is the concept of not using irradiance as-is but converting it into effective irradiance. PVSyst’s loss diagram takes effective irradiance and module efficiency at standard test conditions as its starting point, and then builds up behavior according to environmental conditions. Put another way, the core of energy yield forecasting is “how much of the irradiance effectively reached the power-generating cells.” Even if the global horizontal irradiance on the horizontal plane is large, the energy the cells actually receive is diminished by conversion to the plane of array, horizon shading, nearby shading, reflection due to the angle of incidence, and so on. To understand PVSyst, you need to get into the habit of not viewing irradiance as a single number, but of separating the light before it is reduced and the light after it has been reduced.

The effect of the angle of incidence is a point that beginners particularly tend to overlook. The official documentation explains that the incidence-angle correction represents the phenomenon in which light is reflected at glass surfaces and the like, reducing the amount that reaches the cell, and that it models the angle dependence of this effect. In other words, even with the same solar irradiance, whether sunlight strikes the surface head-on or at a shallow angle changes the effective component the cell receives. Ignoring this obscures the reason why there may appear to be irradiance in the morning and evening but generation is less than expected. In practice, when discussing azimuth and tilt, it is important not to confuse the irradiance incident on the receiving surface with the irradiance that effectively reaches the cell.

Also, PVSyst's predictions do not exist in a simple world where "the same illuminated area yields the same result." This is because the distribution of effective irradiance changes by time of day and season due to row layout and obstacles. Solar irradiance is important not only in its annual total but in when it arrives, at what angle, and at what intensity. Therefore, azimuth, tilt, and row-spacing settings should be regarded not as cosmetic layout inputs but as core inputs that shape the generation time series. If these are handled carelessly, no matter how precisely equipment specifications are refined downstream, the fundamental prediction will remain unstable.

How to consider temperature and electrical losses

Once you have grasped effective irradiance, the next step is to look at how much of the incoming light is converted into electricity. In PVSyst's simulation variables, losses at low irradiance, temperature losses, module quality losses, initial degradation, mismatch, DC-side wiring losses, and so on are organized. What is important here is that generated energy is not determined solely by the amount of light; it fluctuates according to the operating conditions of the modules and the system. For example, during periods of strong irradiance, temperature tends to rise, and even if conditions appear favorable, conversion efficiency can drop due to temperature. Conversely, although low temperatures are favorable for conversion, irradiance may be weak to begin with. Energy production forecasting is the task of balancing these opposing factors over time.

The mindset practitioners should have here is that "losses are not bad; they are a verbalization of assumptions." You can make losses look small and produce attractive numbers, but if those figures don't match reality they are meaningless. What makes PVSyst's loss diagram useful is that it lets you see at a glance which stages of attenuation were expected and to what extent. Moreover, the official documentation explicitly states that each loss rate in the loss diagram is a percentage relative to the immediately preceding energy quantity and should not be simply added together. This is very important: beginners are especially prone to want to "add up all the loss rates," but in fact the structure is compounded—the values are sequentially reduced. Once you adopt this perspective, your understanding when looking at a loss diagram deepens.

Additionally, the constraints on the conversion equipment cannot be overlooked. The official documentation explains that because short-term fluctuations — which are not captured by time averages — cause additional losses near the upper limit, a correction model using finer meteorological data is provided. This helps to understand the common practical phenomenon of “annual solar irradiance is sufficient, but the near-peak output does not extend as much as expected.” Especially in projects considering higher output or adjusting the DC-side ratio, it is necessary to pay attention not only to annual totals but also to behavior near the peak. Forecasting generation is not just about the annual sum; it is also about examining how instantaneous constraints affect accumulated results.

Why Shadows and Layout Greatly Affect Predictions

Shading is one of the factors most likely to be underestimated in energy yield forecasting for solar power generation. Official documentation also notes that nearby shading is caused by surrounding buildings, trees, mounting structures, and so on, and that treating it in detail requires an accurate three-dimensional reproduction of the PV installation and its surrounding environment. In other words, shading assessment is not a matter of a subjective impression like “there’s a little shade in the morning,” but a task of geometrically defining when, where, and how much obstruction will occur. If this is treated too lightly and only annual energy production is estimated, losses in specific seasons or at particular times of day can be overlooked, yielding results that are harsher than expected.

What's more troublesome is that shadows do not simply act in proportion to their area. PVSyst's module layout feature is described as defining each module's position and string connections based on a three-dimensional scene and the electrical configuration, for the purpose of calculating electrical mismatch losses in detail. This means that even if a shadow only covers part of a module, its effect can propagate through the wiring configuration to an entire string or input. In practice, it is dangerous to simplify by assuming "if the shaded area is a few percent, the loss will also be a few percent." The same shadow can produce different results depending on the shadow's position, the cell orientation, how the strings are cut, and the system configuration.

Shading assessment is not a minor supplementary item to be briefly touched on at the end. Official guidance also states that a detailed electrical shading assessment requires both 3D modeling and a fully defined system, and that module layout is normally a process reviewed at the final stage of study. This is because if the layout changes later, the assumptions about electrical impacts also change. Therefore, if you want to improve the accuracy of power generation forecasts, shading should be central to the design conditions, not an auxiliary diagram for appearance. This stance significantly affects the reliability of results, especially for projects around buildings, on sloping ground, in close proximity to trees, or with high equipment density.

Items to Always Check in a Report

When reading PVSyst results, many people stop after looking only at the annual energy yield. However, in practice you should make a habit of looking at the loss diagram first. The official documentation also positions the loss diagram as a tool to quickly assess design quality and identify the main sources of loss. Furthermore, because it can be reviewed not only on an annual basis but also month by month, its structure makes it easy to track seasonal differences and when specific losses become pronounced. For example, if you can determine whether temperature losses are strong in summer, whether shading is significant on winter mornings and evenings, or whether wiring or control constraints stand out during particular periods, your priorities for improvement become clear.

Next, the performance ratio is important. In official documentation, the performance ratio is defined as the ratio of the energy actually produced to the energy that could ideally be produced under rated conditions, and is described as a comprehensive indicator that includes optical losses, shading, soiling, mismatch, wiring, conversion losses, and so on. In practice, it is important not to evaluate this performance ratio as simply high or low on its own, but to interpret it in light of the project conditions. A project with little shading in cold conditions cannot be directly compared with a project in high temperatures that includes self-consumption or energy storage using the same yardstick. The performance ratio is a useful indicator, but comparing numbers alone without knowing the background conditions can lead to incorrect judgments.

Moreover, hourly and monthly results should not be overlooked. Even if the annual values look good, if the monthly downturns appear unnatural, there may be something off in the input conditions, the weather data, or the shading settings. Official documentation shows that results can be displayed by month, day, and hour, and can even be compared with measured data. Therefore, in practical report checks, you should not stop at “viewing the annual energy production,” “viewing the loss diagram,” and “viewing the performance ratio”; only after confirming the plausibility of the seasonal distribution and the time series is the review complete. Being able to do this elevates the process from mere numerical verification to a design review.

Practical Steps to Improve Prediction Accuracy

If you want to improve the accuracy of power generation forecasts, it's more important to eliminate assumptions that cause large errors first than to painstakingly refine the numbers from the outset. In practice, the five measures that are especially effective are: validating meteorological data, verifying installation azimuth and tilt, assessing shading conditions, checking row spacing and obstacle heights, and verifying system configuration consistency. However, the important thing here is not the number of items but the order. First, confirm the conditions affecting how light enters, then tighten up assumptions about temperature and electrical losses, and finally review controls and detailed shading evaluations — this sequence is the most reasonable. I understand that the fact official documentation provides weather data quality checks, three-dimensional shading evaluation, and detailed module-layout calculations as separate functions also aligns with the philosophy of eliminating sources of error in stages.

Also, whenever possible you should reconcile simulations with measured data. Official documentation states that you can import measured data, compare it with simulation variables, and analyze even small irregularities. This is highly effective for improving prediction accuracy. Even for new projects, referring to trends from similar sites or existing installations allows you to reassess whether assumptions about soiling, temperature, and control losses are overly optimistic. Skilled PVSyst users treat simulations not as tools to produce a "correct answer" but as a way to test and update hypotheses.

Furthermore, in recent practical work it has become increasingly difficult to ignore short-term fluctuations and constraints near peaks. Official documentation also presents the idea that using finer-grained meteorological data can correct additional losses that are hard to capture with time averages alone. Even if differences in annual totals appear small, in projects with grid constraints or curtailment control those differences can affect project viability. Therefore, deciding up front what prediction granularity a project requires is also part of improving accuracy. In practice, it is important not to treat an analysis that is sufficient at the annual-total level and one that requires time-of-day behavior at the same resolution.

In terms of understanding on-site conditions, there are limits to merely adjusting input values at a desk. If slope gradients, separations, obstacle locations and heights, or row spacing allowances are ambiguous, assumptions about shading and installation conditions will shift. In such situations, whether you have a means to quickly and accurately capture position and height directly affects the quality of the simulation. Energy yield forecasting is not a task that can be completed within the software alone; differences arise depending on how faithfully you can reflect the site’s geometric conditions in the inputs. If you want to strengthen initial on-site checks, leverage measures such as LRTK, an iPhone-mounted high-precision GNSS positioning device, and by refining the identification of installation locations and obstacle positions at an early stage, you will make it easier to tighten prediction conditions in later stages.

Summary

Summarized from a practical perspective, PVSyst is not simply software for calculating photovoltaic generation; it is a design and verification tool that, by sequentially incorporating irradiance, installation conditions, shading, temperature, electrical losses, and control conditions, explains the final energy output and the reasons for it. In the official documentation, features such as preliminary design, detailed time-step simulation, loss diagrams, performance ratio, and comparison with measured data are organized into a single framework. For that reason, when learning PVSyst it is important not just to memorize the sequence of screen operations, but to understand at which stage what is reduced and which assumptions drive the results.

Understanding the approach to generation forecasting greatly changes how you view simulation results. Rather than reacting only to the annual generation figure, you should be able to question whether the meteorological data are reasonable, whether shading assumptions are realistic, whether expected temperature and mismatch losses are plausible, and whether the interpretation of the performance ratio fits the project conditions. That perspective is the first step to using PVSyst not merely as a calculation tool but as a foundation for design decisions. To further improve the accuracy of desk-based predictions, it is essential to lock down the site location, elevation, and separation conditions early to firm up the input assumptions. Considering those practical links, an approach that combines high-precision site-survey methods such as LRTK with an iterative exchange between simulations and field information leads to the most reproducible generation forecasts.