7 Ways to Improve the Accuracy of PVSyst Generation Simulations

In planning, proposing, internal review, and financial evaluation of solar power generation systems, the accuracy of generation simulations greatly affects decision quality. Especially when using PVSyst, producing numbers is not the hard part; whether those numbers reflect reality is what determines their value. Results that look plausible can nevertheless diverge from reality if the assumptions are coarse, increasing the need for design revisions and explanatory effort after operation begins.

Improving simulation accuracy requires more than just familiarity with the software. You must consider the whole workflow: how to choose meteorological conditions, how to input orientation and tilt, how far to reproduce shading, how to break down losses, how to match equipment configuration to the real system, how to validate results, and how to handle simulations in day-to-day operations. In other words, only when input accuracy, setting accuracy, and interpretation accuracy are combined do PVSyst results become numbers useful on site.

This article organizes and explains seven methods to improve the accuracy of PVSyst generation simulations. It is structured for those starting accuracy improvements for the first time as well as for those already using PVSyst but struggling with prediction variance, presented in a way that is easy to review in actual practice.

Contents

• Make meteorological conditions closer to site reality

• Don’t stop at drawing-based values for orientation and tilt

• Reproduce shading with sufficient detail and repeatability

• Don’t handle loss settings as a single lump sum; review element by element

• Align equipment configuration and wiring conditions with the actual installation

• Validate simulation results against actual performance and alternative conditions

• Embed the process into operational workflows to continuously improve accuracy

Make meteorological conditions closer to site reality

The first foundation that determines simulation accuracy is meteorological conditions. No matter how meticulously you set equipment and loss parameters, if the irradiance, temperature, and wind assumptions you input are far from site reality, the final results will be off. Therefore, the first step to improve accuracy is not to adopt the most convenient meteorological data as-is, but to choose conditions that match the location’s specific characteristics for each project.

PVSyst reads meteorological data to run simulations, but it is important to recognize that the nearest station data is not necessarily the best choice. Coastal and inland areas exhibit different temperature and wind tendencies, and plains versus mountainous areas differ in irradiance patterns, morning/evening shadow effects, and susceptibility to snow or fog. Differences can appear even within the same prefecture. Therefore, selection should be based not on distance alone but on terrain, elevation, surrounding environment, and climate zone.

Focusing only on annual total irradiance is also risky. Even if annual totals are similar, differences in monthly distribution or seasonal variability will affect power sales plans, self-consumption planning, and how surplus energy appears. Whether a site is weak in winter or suffers output drop due to high summer temperatures will change how you consider system capacity and losses. To improve accuracy, check not only annual totals but also monthly trends when assessing data validity.

If on-site observations or performance data from existing systems are available, do not ignore them. If past generation performance, local temperature trends, presence or absence of snow, or seasonal weather variability are known, review meteorological data selection and correction approaches based on those facts. Standard data are convenient for initial assessments, but during project scrutiny, cross-checking with site information is indispensable.

Be careful that improving meteorological assumptions does not mean simply making them more conservative. While it may be tempting to choose conservative values for safety, excessively low assumptions can lead to incorrect system designs. What’s needed is realism, not conservatism or optimism. Rather than adjusting everything with a uniform coefficient for ease of explanation, ensure you can justify why you adopted a particular meteorological condition—this is effective both for accuracy and for practical use.

When comparing multiple projects, avoid situations where one project uses high-resolution meteorological data and another uses coarse data. Such inconsistency undermines fairness in comparison, because differences in inputs, not system conditions, may drive the outcome. When comparing projects, make sure you apply consistent rules for meteorological data selection.

Meteorological data selection sets much of the simulation’s direction. Therefore, don’t treat the initial settings screen as mere routine; approach it as part of understanding the project. You cannot improve accuracy without knowing the site. Making meteorological conditions close to site reality is the starting point for turning PVSyst outputs from desktop estimates into results usable in practice.

Don’t stop at drawing-based values for orientation and tilt

The second way to improve simulation accuracy is to carefully refine orientation and tilt settings. In solar PV, the module’s facing direction and tilt angle determine how much irradiance it receives. If orientation and tilt are set coarsely, not only the annual generation but also hourly and monthly generation patterns will be off.

In practice, orientation and tilt are often entered based on preliminary drawings at the initial stage. That approach is fine for early studies, but if those values are used unchanged downstream, they won’t reflect design changes or on-site conditions. For example, racking layout may be fine-tuned based on post-earthworks ground conditions, and on rooftop projects the actual roof pitch, ridge direction, and installation area constraints may change the assumed angle or orientation. Ignoring these differences as minor can accumulate into a source of prediction error.

First determine whether the project can be treated as a single-condition case. If all faces are the same orientation and tilt, it is relatively simple, but in reality many projects consist of multiple surfaces. When systems include east–west splits, a mix of south- and west-facing surfaces, roof steps, or continuous small faces following terrain, averaging is suddenly dangerous. A single average orientation or tilt cannot accurately reproduce the generation behavior of the real system.

Be especially wary of apparent consistency from averaging. For example, if east- and west-facing surfaces are averaged into a near-south condition, annual totals might look similar, but morning/evening generation balance and peak timing will differ significantly. For high-voltage or extra-high-voltage studies, reverse power flow control, or self-consumption peak measures, these differences matter. To improve accuracy, treat each surface’s conditions separately wherever possible.

Also, check orientation and tilt against site surveys and design documentation. Don’t simply copy drawing values; confirm the final layout considering earthworks, racking specifications, installation constraints, and maintenance access. For ground-mounted systems, subtle slope differences on embankments or constructed surfaces can have an impact, and for rooftop systems, roof material and waterproofing details can change racking height and effective tilt. Setting an ideal angle in the office is pointless if the site cannot be installed as drawn.

Furthermore, orientation and tilt affect temperature conditions and shading. Changes in angle alter ventilation and thus module temperature. Row spacing and height changes affect self-shading. In other words, orientation and tilt are not independent input items but fundamental conditions linked to other settings. Vagueness here will cause cascading uncertainty in shading and loss settings.

From an accuracy perspective, a three-stage review—initial values, design-confirmed values, and construction-assumed values—is useful. Use representative values for early studies, split by surface during design, and re-adjust against construction assumptions at the final stage. Don’t treat PVSyst outputs as a one-off; when orientation or tilt information is updated, rerun the results.

Orientation and tilt may look like simple numbers, but they are the foundation of generation. Don’t be satisfied with copying drawing values; confirm whether modules will actually be installed at that orientation and angle and whether those settings express the multi-surface nature of the project. Doing so will raise simulation accuracy by one level.

Reproduce shading with sufficient detail and repeatability

The third method is to treat shading carefully. Although shading has a large impact on PVSyst simulation accuracy, it is often simplified in practice. Shading factors include distant terrain and surrounding buildings, utility poles, fences, neighboring equipment, and self-shading between arrays. If these factors are treated roughly, not only annual generation but seasonal losses and morning/evening output reductions will be poorly reproduced.

Shading can be broadly divided into the influence of surrounding terrain and distant horizon, and the influence of nearby objects around the system. The former relates to horizon visibility and terrain-based irradiance blocking; the latter involves obstruction by buildings, trees, electrical cubicles, racking rows, parapets, etc. To improve accuracy, identify which shading factors dominate and reproduce them to the necessary extent for the project. There’s no need to over-model everything, but do not ignore shading that matters.

Beware of assuming shading only affects short periods like mornings and evenings and therefore is negligible. During winter, when the sun is low, even short periods of shading can accumulate losses, and in east–west or low-tilt projects shading can have a much larger effect than expected. Shading impacts are not simply a reduction in illuminated surface area. Depending on string configuration and connection scheme, partial shading can propagate output reductions across the system. Therefore, assess shading not only by area but also by its electrical impact.

Insufficient on-site verification when setting shading can fail to reflect tree growth, removal of temporary structures, or potential future buildings. The present-looking landscape is not guaranteed to persist, so at a minimum separate current impacts from future concerns. For example, if existing trees change leaf coverage seasonally or if neighboring development is possible, don’t lock in optimistic assumptions as final values.

Also, do not overlook self-shading. For ground-mounted systems, inter-row spacing, height, tilt, and ground slope change the times when front-row shading affects back rows. If you fix row spacing based on a rough layout, you may miss differences arising after design changes. Inter-row spacing is often revised due to earthworks or maintenance access, so when layout plans are updated, update shading settings accordingly.

On rooftops, parapets, penthouses, equipment platforms, and lightning protection can produce localized shading. These are easy to miss but can repeatedly affect specific strings, causing performance discrepancies. If you cannot explain why output drops only in the morning or only widens in winter during performance validation, suspect insufficient shading settings.

To improve shading accuracy, combine drawings with site photos, drone imagery, survey results, and height information of surrounding obstacles. The larger the project, the more labor shading inputs may seem, but capturing major contributors in advance is more efficient than carrying unexplained errors later. The goal is not to complicate shading modeling per se, but to manage shading as a controllable source of error.

Trust in PVSyst outputs requires not treating shading as if it does not exist. By avoiding excessive simplification and reproducing necessary shading at the required precision, simulated generation moves closer to on-site behavior.

Don’t handle loss settings as a single lump sum; review element by element

The fourth method is to stop treating losses as a single aggregate and instead review them element by element. Many discrepancies in simulation results come from coarse loss assumptions. In practice, people often reuse past project values or apply company-standard uniform loss rates, but that fails to capture project-specific conditions.

Losses include temperature-related loss, wiring loss, mismatch, soiling, degradation over time, conversion efficiency, utilization decline, downtime for maintenance, and more. Each has different causes and different variability. Treating them all as one loss rate obscures what is actually influential and makes root-cause analysis difficult when actual performance differs from simulation.

To improve accuracy, first decompose losses. Take wiring loss as an example: conditions differ between DC and AC sides, and losses change with cable length, cross-sectional area, and current. Values change with equipment layout, so fixing initial estimates through to the final stage is inappropriate. For temperature losses, factors include not just module specs but installation form, ventilation, separation from the roof, and duration of high-temperature conditions. Setting uniform values without considering site specifics is risky.

Soiling losses are another example that varies greatly by location. Coastal sites face salt and wind effects, farmland-adjacent sites suffer dust, sites along major roads see deposition from exhaust, and industrial sites face dust from operations. Whether the slope receives rain, and whether a cleaning regime is in place, also affects results. Even when using a standard value, be prepared to explain why that value is appropriate given site and operational conditions.

Also avoid mixing assumptions for the first year with long-term operational assumptions. Whether the simulation is for first-year evaluation or for multi-year financial analysis changes how you treat degradation and availability. Comparing numbers without clarifying this leads to mismatched discussions. Improving accuracy is not merely inputting more detailed numbers; it is clarifying what evaluation the numbers represent.

Watch out for both over- and under-estimation when reviewing losses. Inflating losses under the guise of safety can underestimate legitimate system performance, while understating losses to appeal to sales can create headaches after operation begins. The important thing is to document the basis for each loss element and link them to project conditions.

Loss settings are not a one-time input. Equipment choices change efficiencies, wiring plans alter losses, and maintenance regimes change availability. Therefore, establish a process to review loss settings whenever design or construction plans are updated. Using unchanged loss assumptions while reusing results makes the numbers look coherent while their substance grows outdated.

If you want to improve PVSyst simulation accuracy, do not use losses as a convenient tuning knob. Losses are not a box to absorb errors but important settings that represent system behavior. By decomposing losses and reviewing them against each project’s conditions, you significantly increase the explanatory power and reproducibility of simulation results.

Align equipment configuration and wiring conditions with the actual installation

The fifth method is to align equipment configuration and wiring conditions with the actual installation. Simulation outcomes depend not only on irradiance and losses but also on how the system is configured. If module count, string configuration, equipment capacity combinations, connection methods, and input circuit assignments differ from reality, the calculated generation may not match the behavior of the actual system.

A common problem is proceeding with detailed analysis while retaining only a conceptual configuration. For example, at an early stage you might roughly size module-to-equipment ratios, but in detailed design, roof surface divisions, wiring routes, input quantity constraints, or maintenance considerations can change stringing. This can lead some inputs to be overloaded or others to have excessive margin, changing assumed conversion efficiency and output control behavior.

Improving accuracy requires more than checking module and equipment counts. Confirm which modules on which surface feed which inputs, whether surfaces with different shading are grouped into the same input, whether voltage conditions are acceptable, and whether operational ranges under high and low temperatures are covered. Coarse configurations translate directly into coarse simulation accuracy.

Be cautious with assumptions about oversizing. Judging by capacity ratios alone overlooks differences caused by regional conditions, orientation and tilt, and temperature. Some projects can rationally tolerate large oversizing while others experience clipping at levels higher than anticipated even with slight oversizing. To improve PVSyst accuracy, evaluate how much curtailment will occur by time of day and by season, not just the capacity ratio.

Wiring conditions are another area that tends to remain at desk-assumed values. In practice, cable routes change for obstacle avoidance, equipment placement, constructability, and safety, often becoming longer than originally assumed. Deviations in wiring distance directly affect losses, so if not updated as design progresses, they become a source of error. Projects with multiple collection routes or changes in panel/switchgear locations are especially susceptible to differences from initial simplifications.

Treatment of equipment efficiency is also important. Judging by nameplate efficiency under rated conditions does not guarantee performance in the system’s actual operating range. Whether a project has many low-output periods, severe high-temperature exposure, or a large share of morning/evening generation changes effective efficiency. Therefore, improving equipment-configuration accuracy means aligning system operating characteristics with equipment characteristics, not merely filling a bill of materials.

In practice, configuration information is updated at sales, design, and construction stages. If PVSyst configurations are not updated each time, the final simulation may rest on outdated assumptions. Presenting such outdated numbers to clients or for internal approvals leads to inconsistencies. That is why you need a mechanism to update simulations when configuration information changes.

Generation simulation is not determined by irradiance and losses alone. How equipment is configured and connected shapes the results. By refining equipment configuration and wiring conditions to match the real system, PVSyst outputs become closer to on-site behavior and differences from actual performance are easier to reduce.

Validate simulation results against actual performance and alternative conditions

The sixth method is to validate results rather than accepting them at face value. PVSyst is a convenient tool, but because it returns outputs based on the assumptions you set, it does not automatically guarantee that the results are correct. To improve accuracy, don’t trust numbers that simply look good; verify them from multiple perspectives.

First, avoid looking only at annual total generation. Even if annual values are close, monthly distributions can be skewed. For example, if the model overestimates summer and underestimates winter, temperature, shading, or meteorological assumptions may be at fault. Conversely, if monthly trends match reality even when annual totals differ slightly, the direction of assumptions may be correct. Therefore, validation should include monthly, and where appropriate daily or hourly, trends— not just annual totals.

When existing system performance data is available, comparison with that data is the most powerful validation method. However, simple comparisons are insufficient. You must separate factors such as curtailment, faults, downtime, cleaning frequency, measurement errors, and anomalous weather years, or else you will mix simulation issues with operational issues. When comparing with actual performance, check not only the magnitude of differences but also when, at what times, and in which direction the deviations occur.

Comparisons with similar projects are also valuable. Seeing the differences between projects with similar site and equipment conditions helps you check the reasonableness of your settings. If your simulation is extremely high or low compared to comparable projects, some bias likely exists in your meteorological, loss, or configuration settings. When improving accuracy, do not work in isolation; adopt a comparative perspective.

Sensitivity checks are also an effective validation method. For example, see how much the result changes when you slightly alter orientation, or how far it moves when you change a loss rate, or whether shading changes primarily affect winter. By perturbing inputs and observing responses, you can identify which settings dominate. If the result is overly sensitive to a particular input, prioritize improving that input’s accuracy. Conversely, if a finely tuned item has minimal overall effect, reprioritize your work accordingly.

Also, make validation explainable. When sharing results with clients, supervisors, or internal stakeholders, simply presenting a number may not be convincing. If you can explain why you chose a particular meteorological condition, loss rate, and configuration, trust in the numbers increases. Validation is necessary not only for accuracy but also for accountability.

Validation in PVSyst is not the last step; it is a learning process to improve inputs for future projects. Capture biases revealed by differences with past performance and apply them to subsequent projects to raise organizational simulation accuracy. Treat results as a step in a continuous improvement cycle rather than an endpoint—this is how practical operations become robust.

Embed the process into operational workflows to continuously improve accuracy

The seventh method is to embed accuracy improvement into operational workflows rather than leaving it as an individual task. PVSyst simulation accuracy will not stabilize just because a single person set inputs carefully once. Project conditions differ, designs are updated, and construction and operational realities change. Therefore, improving accuracy requires integrating input, review, update, and validation processes into daily operations.

First, clarify which level of accuracy is required at each stage. The information available and the required precision differ between an early-stage sales proposal, pre-design detailed study, and final pre-construction check. Demanding the same granularity at every stage increases workload and undermines sustainability. Use representative values in the early stage, project-specific values in the design stage, and construction-reflective values at the final stage—having stage-specific rules is realistic.

Next, keep records of input conditions. Record why you chose a particular meteorological dataset, the basis for loss rates, which materials supported shading settings, and which drawing revision you used for configuration. These records prevent confusion when revisiting the project. If only results remain without the underlying assumptions, updates and validation become difficult. Managing the assumptions leading to numbers is more important than the numbers themselves.

Interdepartmental information flow is also essential. If sales hold site intelligence, design holds configuration details, construction knows actual wiring conditions, and operations hold performance data, but this information is fragmented, PVSyst inputs will inevitably be incomplete. Improving accuracy requires processes that deliver necessary information to the person setting up the simulation. Don’t rely solely on individual effort; build organizational processes for information collection.

Having rules for reconciliation with actual performance is effective. For example, check divergence between predicted and actual generation after a set period following commissioning, and run a cause-analysis workflow for projects with large differences. Recording differences without capturing causes does not create knowledge, but retaining causes and countermeasures raises input quality for future projects. Accuracy improvement is not a one-off gimmick; it perpetuates only when knowledge is reusable.

Also, do not seek perfection in practice. As accuracy improves, the information and effort required increase, and it is not realistic to treat every project to the same depth. The key is to identify items that strongly affect results and prioritize improvements. By reviewing the seven areas—meteorological conditions, orientation and tilt, shading, losses, configuration, validation, and operations—in sequence, you can steadily improve accuracy for many projects.

Improving PVSyst simulation accuracy is not merely about producing favorable numbers. It enhances proposal credibility, stabilizes design decisions, makes internal and client explanations easier, and facilitates response to post-commissioning discrepancies. In short, accuracy improvement is both a generation-estimate issue and a matter of operational quality.

Accurately understanding site conditions, thoroughly documenting input assumptions, validating results, and applying lessons to the next project—when this flow is established, PVSyst becomes more than a calculation tool and functions as a practical decision support tool. If you are struggling with simulation accuracy, start by reviewing your company’s operations against the seven viewpoints introduced here.

Finally, an unexpectedly important starting point for accuracy improvement is how accurately you can capture site conditions. Misreading meteorological conditions, misidentifying orientation or tilt, overlooking shading factors, and misrecognizing equipment layout are more likely when site information is vague. If you want to make site verification more reliable, consider tools that let you acquire high-precision positional information while checking the site. For those who want to improve on-site position confirmation, equipment placement mapping, and recording of shading factors, consider using LRTK (iPhone-mounted GNSS high-precision positioning device). Simulation accuracy will not rise from the desk alone. Only by linking precise site understanding to input assumptions will you approach accuracy that is useful in practice.