What can PVSyst compare? Explaining how to interpret design studies

Table of Contents

• PVSyst is software for "finalizing a design through comparison"

• Prerequisites to prepare before starting comparisons

• Comparison target 1: Meteorological data and site conditions

• Comparison target 2: Azimuth, tilt, and mounting method

• Comparison target 3: Capacity allocation and electrical design

• Comparison target 4: Shading conditions and 3D model

• Comparison target 5: Assumptions for detailed losses

• Metrics to examine in comparison results

• How to read report comparisons

• How to conduct comparisons to avoid failures in design studies

• Summary

PVSyst is software for 'finalizing designs through comparison'

PVSyst is not a tool for calculating a photovoltaic system design just once and being done with it; rather, it is a design tool that, based on site information and time-series meteorological data, helps interpret performance differences by switching among multiple simulation conditions. The official documentation also explicitly states that projects are managed within a Project framework, and that optimization and parameter analysis within that framework are advanced through different simulation runs, i.e. variants. In the results screen you can call up other variants within the same project for quick comparison, and the report comparison lets you check differences between different projects or different variants. In other words, when asked what PVSyst is, it is more accurate in practice to view it not as “software that outputs generation figures” but as “software that supports design decisions by comparing differences in conditions,” which better explains how to use it.

This perspective is important because a variant saved in PVSyst is not merely a calculation result, but a bundle of the simulation’s specific assumptions—plane orientation, module layout, power converter configuration, loss coefficients, shading, horizon conditions, and so on. The official workspace description also states that a variant file contains the specific conditions and the calculation results, such as plane orientation, module layout, power converters, loss coefficients, shading, and horizon. In other words, creating a variant is not duplicating a proposal but rather isolating each design intent. Once you understand this, comparing in PVSyst appears not as “which of proposal A or B wins” but as the task of tracking “which differences in assumptions produce which differences in losses.”

Prerequisites to Prepare Before Starting Comparisons

When you start comparisons in PVSyst, the most important thing is not to include too many elements from the outset. The official tutorial likewise recommends first creating an initial system configuration with the minimum parameters, saving it, and then creating successive variants that add distant shading, nearby shading, individual losses, and so on in stages. This is to make the design differences clear for each comparison. If you limit changes to one or two items—for example, a variant that only changes orientation, one that only changes capacity allocation, or one that only adds shadow conditions—you are less likely to misread what caused the differences when looking at loss diagrams and monthly results.

Conversely, if you change meteorological data, orientation, capacity allocation, shading, and loss coefficients all at the same time, even when differences appear it becomes difficult to identify their causes. In particular, meteorological data are the starting point for project evaluation and are considered a major source of uncertainty. Therefore, at the stage when you want to see pure design differences, you should compare with the location and meteorological data fixed, and treat trying a different meteorological source or data from another year as a separate "sensitivity check of assumption uncertainty" rather than part of the design comparison. Simply enforcing this separation from the outset makes PVSyst comparison results much easier to interpret.

Comparison 1: Meteorological data and site conditions

Surprisingly, the thing you should compare first in practice is the meteorological data, before the equipment configuration. The official documentation states that meteorological data is the starting point for project evaluation and at the same time the main source of uncertainty. The meteorological data features also include comparing irradiance between different weather files and creating a representative year from multiple years of files. In other words, PVSyst treats not only the design conditions but also which meteorological assumptions the design is evaluated against as a subject of comparison. When a designer judges "this option produces higher energy," it is important to distinguish whether it really appears that way because of design differences or because of differences in the weather source.

What you should look for when comparing meteorological data is not just the annual totals. The official quality-checking function lets you inspect any variable in a weather file on an hourly, daily, and monthly basis, and by comparing it with a clear-sky model makes it easier to detect time shifts, amplitude anomalies on clear days, abnormal nighttime values, missing data, and so on. It is also explicitly stated that incomplete weather files are not recommended for use in simulations. In other words, comparing weather data is not simply about choosing the dataset with the largest annual insolation; it is about judging whether that data is of sufficient quality to be used for design decisions. Skipping this step and moving on to later comparisons will destabilize the discussion of orientation and capacity allocation itself.

Furthermore, comparing meteorological data can be used not only to determine which design option wins, but also to assess how resilient a design is to changes in assumptions. A proposal that is superior in a representative year may see its advantage shrink with data from other years or other sources. In such cases, whether the results remain stable against variations in assumptions is often more valuable in practice than the superficial differences in annual energy production. If you use PVSyst as a comparison tool, it is important to treat differences in meteorological data not as noise but as test conditions for measuring a design’s robustness.

Comparison Item 2: Orientation, Tilt, and Installation Method

The easiest parameters to compare in PVSyst are azimuth and tilt. For fixed installations, orientation is formally defined by the tilt angle and the azimuth angle, and these can be adjusted on-screen. In addition, for azimuth selection there is a simple optimization tool provided to check how appropriate the chosen tilt and azimuth are for that location, giving an overview of whether the current selection is near the optimum and how far it is from it. However, this simple evaluation is a rough estimate using monthly values and has been shown to differ from the final hourly simulation results. Therefore, for azimuth comparisons it is practical in professional work to use the simple optimization as an entry point and always make the final decision based on detailed simulations — a two-step approach.

A common mistake in orientation comparisons is to rush to conclusions based solely on energy production. In reality, when azimuth and tilt change, not only the incident radiation but also seasonal generation bias, morning-and-evening output distribution, how shading affects the system, and interactions with temperature conditions change. That is why, for orientation comparisons in PVSyst, it is more effective to judge by looking at monthly differences and the distribution of output during peak periods rather than a simple annual sum. In the design process, switching the mindset from “finding the angle that generates the most energy” to “finding the angle that offers the most manageable output characteristics under the installation and grid conditions” raises the value of the comparison.

Additionally, the current document presents a design philosophy that allows multiple independent orientations within a single project, enabling the simultaneous handling of fixed-mounted and tracking systems, or subsystems with different tilt and axis conditions. This is particularly effective for projects where only part of the site has a different slope or where multiple roof surfaces coexist. In practice, one may be tempted to crudely evaluate the project using a single average angle, but if you leverage the comparison function, it is far better for design studies to carefully separate per-surface condition differences as variants or subsystems and identify which surfaces are boosting overall performance and which are holding it back.

Comparison 3: Capacity allocation and electrical design

In PVSyst, a system is defined as being composed of modules, strings, power converters, and grid connections, and it can be assigned multiple sub-arrays. In other words, what can be compared is not just the total capacity. The electrical design itself is the subject of comparison: differences in the number of modules in series and in parallel, the allocation of converter units, how grid interconnections are grouped, and the electrical share per surface. The important point is that even with the same total capacity, different configurations can change voltage conditions, partial-load behavior, the way overloads present themselves, and how shading affects the system. PVSyst can treat these differences as variants, so its strength is that capacity comparisons do not end with “how many kW to add” but can delve into “how to configure it.”

In practice, the figure most often compared is the DC:AC ratio — that is, the ratio of the array’s rated output to the inverter’s rated output. Official documentation defines this ratio as the ratio of the array’s nominal output to the inverter’s nominal output and gives about 1.25–1.3 as a general guideline to avoid excessive overload losses. However, it also explains that a reliable assessment of overload losses can only be obtained through detailed time-resolved simulations. Furthermore, behavior under overload is not simply a matter of “dumping excess power as heat”: the inverter shifts its operating point to keep itself within a safe output range, which manifests as some of the power that could potentially have been extracted not being generated. For that reason, instead of judging solely by the ratio, it is essential to compare how much overload loss will actually occur annually and in which seasons and times of day those losses are concentrated.

Also, when comparing capacity allocations, focusing only on increased energy production can lead to incorrect conclusions. A proposal may be slightly advantageous in annual energy production, yet its output may plateau during certain daytime hours and show little difference during other periods. In such cases, the design benefit is not as large as it appears. Conversely, even when the difference in annual values is small, a proposal that substantially improves morning and evening or seasonal variability can be highly valuable when considering compatibility with the grid and demand conditions. PVSyst capacity comparisons should be read not as discussions of ratios but as comparisons of output characteristics over time.

Comparison item 4: Shadow conditions and 3D model

Shadow conditions are a comparison item that tends to produce differences in PVSyst and is also easy to misinterpret. In the official documentation, shadows are broadly classified into distant shadows and near-field shadows. Distant shadows are represented by the horizon and uniformly affect the entire power-generating surface in the sense of whether the sun is visible or not at a given time. In contrast, near-field shadows are the visible shadows cast onto the power-generating surface by nearby obstacles; they are much more complex to handle and require a detailed 3D description of the system and its surrounding environment. Moreover, near-field shadows affect not only the simple shaded-area fraction but also both optical losses and electrical partial-shading losses, thereby widening design differences more than one might expect.

Therefore, it is effective to compare shading conditions in stages. First create a baseline scenario with no shading, then a scenario with only the horizon, next a scenario including a 3D scene of near-field shading, and finally a scenario that also incorporates the module layout and refines the electrical shading losses. Comparing them in that order makes it clear at which stage losses increased. The official design flow is also organized so that the horizon, near-field shading, module layout, and detailed losses can be defined separately. What is important in shading comparisons is not "whether shading exists or not," but separating out "which types of shading, in which seasons, add how much loss."

Furthermore, when comparing shading, annual energy production alone is not enough. The official loss diagram is useful for quickly identifying weaknesses in system design, and can be checked not only on an annual basis but also month by month. In other words, it lets you distinguish whether the shading is concentrated only in winter mornings and evenings or whether it gradually affects performance throughout the year. In design reviews, the correct approach is not to dismiss a proposal as “no good” the moment shading conditions are introduced, but to examine which times of day and in what types of losses the shading appears, and, if necessary, use that information to guide layout adjustments or a re-evaluation of circuit splitting.

Comparison 5: Assumptions for Detailed Losses

One reason PVSyst is strong for design studies is that it does not lump detailed losses into a single black box, but treats them explicitly as items for comparison. The official documentation lists detailed losses such as soiling, incidence-angle correction, temperature parameters, wiring resistance, module quality, mismatch, and shutdowns or downtime. What designers often do is treat these as fixed values once entered, but in reality many are assumptions. Especially in early-stage studies, site conditions, construction quality, and operational assumptions are often not yet settled, so it is a more honest way to interpret results to separate and compare variants at three levels—optimistic, standard, and conservative.

For example, regarding soiling loss, even the official soiling loss description notes that it is an uncertainty that strongly depends on environmental and rainfall conditions, and it can be defined as monthly values. In other words, rather than roughly assigning a single annual average value for soiling, dividing into dry and rainy seasons and creating multiple scenarios improves the accuracy of comparisons. Also, for incidence angle correction, PVSyst allows model selection or comparison from the detailed losses screen, and user-defined profiles other than the defaults should be used with caution. For these items, in practice it is more useful to check whether the ranking of design options changes when values are varied, rather than trying to hit the "correct" value on the first try. If a design option maintains its advantage despite some variation in loss assumptions, that option can be considered a strong basis for further consideration.

A commonly overlooked point when comparing detailed losses is that they are not independent of one another. For example, a low-tilt proposal tends to accumulate more soiling, and changing the azimuth alters how the incidence angle correction behaves. If wiring length changes, resistive losses change, and if the layout changes, the electrical impact of mismatch and partial shading also changes. Therefore, detailed losses are not decorative items to be summed at the end but a comparison axis that determines the quality of the design proposal itself. The deeper you use PVSyst, the more comparisons at this layer will determine the persuasive power of the design.

Metrics to Look at in Comparison Results

So, what should you look at in the comparison results? PVSyst's outputs include numerous simulation variables and can be displayed by month, day, or hour. The official documentation also notes that the results contain many variables, and the loss diagram is particularly useful for identifying design weaknesses. Furthermore, you can use predefined tables, a custom monthly table where you select any eight variables, monthly graphs that can display up to four variables simultaneously, and hourly and daily plots. In other words, PVSyst comparisons are not a single showdown based on annual energy production, but are designed to be interpreted to include monthly trends, time-of-day biases, and the structure of losses.

Among indicators, annual energy production and specific yield are easy to understand and therefore tend to be looked at first, but drawing conclusions based on them alone is risky. In the official PR description, PR is an indicator that includes shading, incidence-angle correction, optical losses such as soiling, conversion losses and mismatch, array losses such as wiring, and even system losses; unlike specific yield, it is less directly dependent on meteorological inputs and installation surface orientation, so it is considered easy to use for comparing system quality between different sites and orientations. Therefore, as a basic practice, use annual energy production to see "how much is produced" and PR to see "how efficiently it is produced." Reading the two side by side makes it easier to see the difference between a proposal that simply produces more and one that is refined in quality.

Also, the loss diagram should be placed at the center of the comparison. Even if the difference in annual energy generation is small, if different losses were reduced to produce that result, the reproducibility of the improvement and the next steps will change significantly. For example, if PR has improved but the difference in energy generation is small, it may simply be masked by variability in solar irradiance conditions; and if energy generation has increased while PR remains flat, it may just be supported by favorable weather or orientation. In PVSyst comparisons, reading the loss structure rather than focusing on the result numbers raises the quality of design decisions.

How to Read Report Comparisons

The report comparison feature of PVSyst is highly effective for preventing misinterpretation of comparison results. The official documentation states that you can display the report currently shown side-by-side with the report of another project or variant and highlight the differences. This is useful not only when there are few changes, but also at stages when the study progresses and multiple parameters begin to interact. People tend to pick out only the convenient differences when they look at numerical results alone, but reading a report with highlighted differences lets you confirm at once which assumptions, configurations, loss settings, and result indicators have changed. This way of reading is also very powerful when explaining the rationale for comparisons in design meetings.

Furthermore, PVSyst can export hourly and daily data, allowing you to select the necessary variables and send them for external analysis. The official advanced simulation explains that output files can write monthly, daily, and hourly values; that special graphs allow predefining histograms and scatter plots; that batch simulation can perform parameter analyses; and that the optimization tool can investigate sensitivity to tilt, azimuth, pitch, and so on. In other words, PVSyst comparisons do not have to be confined to the report screen—you can increase the temporal resolution as needed and examine finer differences. When the study becomes more complex, it is efficient to narrow the scope to a few representative variants and move on to hourly comparisons.

PVSyst can also be used for comparisons after commissioning. In the official measured data analysis, on-site measured data and simulated values are compared hourly or daily, and it is said to be useful for detecting minor faults. In the design study phase there are no actual future measurements, but if there are results from existing similar projects, the persuasive power of design comparisons increases dramatically. Don’t leave comparisons as desk studies; by ultimately considering “how well the design assumptions align with measured data,” the quality of the comparisons will be further enhanced.

How to Proceed with Comparisons in Design Studies Without Failing

A comparison procedure in PVSyst that is less likely to fail is to first select a single reference variant, then add comparison axes one by one. As the official tutorial shows, the basic approach is to create the reference case with a minimal configuration at first, and then progressively save versions that add items such as distant shading, nearby shading, and individual losses. In practice, expand this idea slightly: after the reference case, perform an orientation (azimuth) comparison, then a capacity allocation comparison, next a shading-condition comparison, and finally a detailed-loss comparison — proceeding in that order stabilizes the interpretation of the differences. Then, by creating composite cases that combine only the promising options, you can deepen the study without increasing the number of comparisons too much.

The trick at this stage is to verbalize not only the magnitude of the results at each step but also the meaning of the differences. For example, in orientation comparisons, check not only how many percent the energy production increased, but also which months the difference is concentrated in, how the PR changed, and which elements moved on the loss diagram. For capacity allocation comparisons, rather than focusing on the DC:AC ratio number itself, look at whether overload losses are within a realistic range and whether the time-of-day distribution is biased. For shading comparisons, look at biases in electrical shading losses during winter mornings and evenings or on specific surfaces rather than the annual difference. Using PVSyst for comparisons is an extremely powerful tool when used not to produce a single conclusion but to sequentially confirm the issues needed for design decisions.

Common mistakes include arguing only about design differences while meteorological assumptions are unstable, making the shading model overly complex all at once so you can’t tell what’s actually effective, confusing the roles of PR and energy yield, and treating detailed loss terms as fixed truths. Official guidance also identifies meteorological data as the primary source of uncertainty and advises that quality checks should inspect time offsets, anomalous values, missing data, and the like. In other words, the accuracy of comparisons is determined more by how you partition and interpret the assumptions than by the software’s computational capability. Mastering PVSyst means not using every available feature, but organizing the axes of comparison and creating a state in which differences between proposals can be translated into design language.

Summary

If you had to sum up in one sentence what PVSyst can compare, it is a design-comparison tool that can organize and interpret, on a per-variant basis, meteorological data, orientation and tilt, mounting method, capacity allocation, shading conditions, detailed losses, and ultimately the resulting performance metrics. The important point is not the sheer number of comparable items, but the ability to put them on the same footing and explain the differences. In design studies, rather than simply choosing the option with the largest annual energy production, it is essential to adopt a perspective that selects proposals which are robust under different assumptions, have clear loss structures, and are easy to justify. Understanding PVSyst as software that provides that perspective makes its use much more practice-oriented.

Then, when putting desk-based design comparisons into practice on site, quickly verifying the site coordinates, elevation, orientation, and positional relationships with surrounding obstructions is the quickest way to improve the accuracy of those comparisons. If you want to streamline such on-site surveying, tools like LRTK, an iPhone-mounted GNSS high-precision positioning device, are well suited. Compare proposals using design simulation, and finalize the assumptions through on-site positioning. Establishing this workflow makes PVSyst’s comparison results a more practical basis for decision-making.