7 items to compare power generation forecasts and actuals when generation is low

When you feel that power generation is low, it is not uncommon to immediately suspect equipment failure or insulation faults. However, the actual performance of solar power generation fluctuates due to multiple factors such as weather, solar irradiance, temperature, shading, soiling, downtime, output curtailment, measurement conditions, and the assumptions behind forecast values. The important thing is not to judge “low” intuitively, but to compare the generation forecast and actual performance under as similar conditions as possible and then check, step by step, where the differences arise.

This article outlines seven items to consider when comparing predicted and actual power generation during periods of low generation. Use them not to definitively identify equipment faults, but as a way to distinguish natural factors, aggregation conditions, site environment, and equipment condition. Because work involving internal inspection or measurement of electrical equipment carries risks of electric shock and equipment shutdown, it is important to consult a specialist or a qualified professional as needed.

Table of Contents

• Align the assumptions for generation forecasts with the actual performance period

• Check solar irradiance and weather differences to isolate natural factors

• Check for output reductions due to temperature and seasonal factors

• Check for local environmental changes such as shading and soiling

• Reflect downtime and curtailment history in the performance discrepancies

• Check for generation imbalances by system and by equipment

• Review the forecast generation method and the timing of updates

• Summary: When generation is low, record and explain the reasons for the discrepancy

Align power generation forecast assumptions with the actual performance period

The first thing to check when determining whether power generation is low is whether the generation forecast and the actual generation being compared are under the same conditions. Generation forecasts are produced at various intervals—annual, monthly, daily, or by time of day. Actual data, meanwhile, is often obtained from records of monitoring devices or energy meters, and the aggregation period or cutoff time may be different from that used for the forecast. If you compare them without confirming this discrepancy, differences that are not actually abnormal can appear larger.

For example, if the forecast is prepared on calendar days from the beginning to the end of the month, while the actuals are aggregated based on meter-reading dates or arbitrary periods used by monitoring equipment, a difference of just a few days can change the results. Especially in seasons when the weather varies greatly, the evaluation of monthly generation can change if the number of sunny or rainy days shifts by only a few days. Before concluding that generation is low, it is necessary to align the comparison period’s start date, end date, times, time zone, and aggregation units.

Also confirm which generation capacity the forecast assumes. If the equipment's rated capacity, number of modules, power conditioner capacity, the presence or absence of additions or partial removals, assumptions about output curtailment, etc., differ between when the forecast was created and now, discrepancies with actual results will occur. For generation facilities that have been in operation for a long time, past design values or the generation forecasts from the initial planning are sometimes used unchanged as the comparison target. In that case, they may not match the current equipment configuration.

Power generation forecasts include those based on standard solar irradiance and past weather data, simulation values used at the design stage, and management forecasts that reflect actual performance after operations begin. The meaning of a comparison changes depending on which forecast is being used. Design forecasts are convenient as a long-term benchmark, but they can be difficult to use as-is for short-term anomaly detection. For daily management, comparing against forecasts that incorporate recent weather conditions and past performance makes it easier to isolate the causes of low power generation.

In practice, it is important to first clarify "which forecast the current power generation is low against." Whether it is lower on a monthly basis compared to the annual forecast, lower on a daily basis compared to the monthly forecast, or lower compared to past performance in the same month will change what needs to be examined. If you proceed with a root-cause investigation while the comparison baseline is unclear, the inspection scope can become too broad, making it difficult to prioritize on-site checks and equipment investigations.

When receiving a consultation that power generation is low, first organize the predicted values, actual values, comparison period, installed capacity, and aggregation method. If a mismatch in conditions is found at this stage, it may be explainable as an artifact of the aggregation rather than an equipment fault. Conversely, if a difference remains even after aligning conditions, proceed to the next items and check, in order, the weather, temperature, shading, shutdown history, and equipment-specific deviations.

Isolate natural factors by checking solar radiation and weather variations

When comparing the actual performance of photovoltaic power generation with forecasts, the basic task is to check solar irradiance. Because power output is strongly influenced by irradiance, generation that is lower than forecast may be a natural result if the irradiance during the same period was lower. Instead of judging a result as low based solely on generation, it is important to confirm how the irradiance during the same period compared with the assumptions used in the forecast or with the long-term average.

When assessing solar irradiance, check not only the monthly totals but also daily and time-of-day trends. Even if monthly figures appear similar, generation performance can decline due to time-of-day biases — for example, many cloudy mornings, cloud cover in the afternoons, or a string of rainy days. Because the times of day when generation is most favorable vary with the installation orientation, when periods of poor weather coincide with the equipment’s characteristics the impact can appear large.

When power generation forecasts assume standard weather conditions, actual months with unstable weather tend to show larger deviations from the forecasts. During the rainy season, periods with many typhoons, or in regions where cloudy weather persists in winter, month-to-month variability occurs. If such natural variability is not taken into account, the cause of low generation can be attributed too much to equipment. Equipment inspections are important, but first confirming whether the discrepancy can be explained by weather factors is effective in reducing wasted investigation.

In practice, whenever possible, compare on-site or nearby meteorological data with actual power generation. If there is a pyranometer at the site, using its readings makes assessment easier. Even if there is no pyranometer on site, you can grasp general trends by referring to data from nearby observation points and weather information. However, local clouds, mountain shadows, coastal fog, and the effects of surrounding terrain can cause nearby data not to match the site exactly. Therefore, when assessing weather-related differences, it is safer to check not only the numerical data but also site condition records, monitoring images, and work reports.

If there was a day with low power generation, check not only that day’s weather but also compare the generation curve on clear days. If generation on clear days is close to the forecast, it is likely a temporary drop caused by cloudy or rainy weather. On the other hand, if the generation curve is low even on clear days, or if it only dips during specific time periods, you should suspect factors other than solar irradiance.

When looking at the relationship between solar irradiance and power generation, it is helpful not just to consider whether generation is high or low, but to check how much power is being produced relative to the irradiance. It is natural for power generation to be low on days with little irradiance, but if irradiance is sufficient and generation does not increase, possible causes include soiling, shading, equipment shutdowns, output curtailment, temperature effects, and grid-related faults. In other words, checking solar irradiance is the entry point for separating declines that can be explained by natural factors from declines that require investigation of the equipment side.

When power output is low, operators are more eager to identify the cause quickly. However, if they skip weather factors and move straight to equipment inspections, they may find no clear abnormalities even after checks, which can prolong decision-making. It is important to first confirm how much of the gap between forecast and actual output can be explained by differences in solar irradiance, and to establish a process that proceeds to further investigation only when an unexplained gap remains.

Confirm output reduction due to temperature and seasonal factors

An often-overlooked factor when generation is low is the effect of temperature and season. Solar photovoltaic (PV) systems do not always produce near-maximum output simply because sunlight is abundant. In general, photovoltaic modules tend to lose output as temperature increases, so even on clear summer days with ample sunlight, actual generation can fall short of forecasts.

Especially when power generation forecasts are based on standard temperature conditions, days when the actual air temperature or module temperature is higher may show discrepancies. Module temperature is affected not only by ambient air temperature but also by the roof surface, the area around the mounting structure, ground surface reflectance, and ventilation conditions. In poorly ventilated locations or environments where heat tends to accumulate, power output can more easily decrease even under the same solar irradiance.

On the other hand, in winter the lower temperatures can be advantageous from a temperature standpoint for the equipment, but daylight hours are shorter and the sun's altitude is lower. Shadows tend to lengthen in the morning and evening, and effects from snow, frost, and fallen leaves are more likely to occur. In other words, in summer output tends to decline due to high temperatures, while in winter declines are more likely due to shorter daylight, shading, and snow, so the points to check differ by season.

When comparing power generation forecasts with actual performance, check not only the temperature during the target period but also the expected shape of the generation curve for the season. In summer, output tends to be higher around midday, but when high temperatures and output curtailment coincide, the peak may be suppressed. In winter, total generation hours are shorter, and the morning and evening ramp‑ups and drop‑offs may appear more pronounced. If these seasonal characteristics are not taken into account, there is a risk of misinterpreting normal variations as anomalies.

When investigating low power generation, comparing the same month of the previous year or the same period over recent years is also useful. However, you should not simply judge it as abnormal because it is lower than the previous year; you need to consider differences in conditions such as solar irradiance, temperature, snow accumulation, typhoons, and prolonged rainfall. If the previous year had exceptionally good weather, this year’s results may appear low even if they are around the long-term average. Conversely, if this year’s weather has not been significantly worse but output is substantially lower than the previous year, you need to examine equipment-related factors in detail.

When examining seasonal factors, not only monthly power generation but also time-of-day data for clear-sky days is useful. Temperature effects tend to appear around late morning to early afternoon when solar irradiance is strong, so check how pronounced the peak is in the generation curve for clear-sky days. If irradiance is sufficient and shading effects are minimal yet output is lower than expected during those hours, you need to further assess whether the deviation can be explained solely by temperature effects or whether equipment limitations or partial system malfunctions are present.

For practitioners searching with the intent of "low power output," understanding seasonal factors is indispensable for improving the accuracy of fault isolation. Power output is not constant throughout the year; even the same equipment produces different results month to month. If you view the gap between forecasts and actuals with those variations in mind, you can carry out necessary inspections appropriately while avoiding excessive concern.

Check for changes in on-site conditions such as shadows and dirt

If the difference between predicted and actual power generation cannot be explained solely by solar irradiance or temperature, the next thing to check is changes in the on-site environment. Even if a photovoltaic system had no problems at installation, the surrounding environment can change over time. Power output may be reduced by tree growth, buildings or structures on adjacent land, weeds, deposits around the mounting racks, bird droppings, fallen leaves, dust, yellow sand, or snowfall.

Shading is one of the common causes of reduced power generation. Even if a shadow only covers part of an installation visually, it can affect generation at the string level or circuit level. Shadows from nearby trees, utility poles, fences, buildings, or forests extend particularly far when the sun is low in the morning and evening. Even locations that were unshaded at the time of installation can become shaded a few years later due to tree growth or surrounding construction.

When comparing predicted and actual power generation, you can sometimes infer the possibility of shading from the shape of the generation curve. If generation ramps up slowly only in the morning, drops sharply during certain hours in the afternoon, or exhibits unnatural fluctuations even on clear days, suspect shading or local obstructions. When solar irradiance is stable but actual output falls only at specific times, it is more likely due to on-site factors rather than weather.

Soiling can also cause reduced power output. When the module surface is covered with dust, bird droppings, fallen leaves, pollen, volcanic ash, or salt-containing deposits, sunlight has more difficulty reaching the cells. The effects of soiling may be washed away by rain, but they tend to persist on low-tilt installations or in environments where deposits readily adhere. Also, partial soiling can sometimes appear to have a large impact on energy production depending on the circuit configuration, so on-site inspection and photographic documentation are important to assess the condition.

Weed and vegetation management must not be overlooked. For ground-mounted systems, weeds growing in front of the modules can cast shadows on the bottom edge. Growth is rapid in summer, so even if there was no problem at the previous inspection, they can cause shading in a short time. If periods of low power generation coincide with times of weed overgrowth, check the weed removal history and site photos and examine their relationship with the generation curve.

When checking changes in the on-site environment, simply recording "dirty" or "there is a shadow" is insufficient. It is important to record when, over what area, during which time periods, and to what extent an impact is likely. To explain the difference between predicted and actual power generation, you need to verify not only qualitative impressions but also whether the time periods and affected systems experiencing the generation decline correspond to the on-site conditions.

During on-site inspections, and with attention to safety, record overall photos, the condition of each row, the direction of shadows, the distribution of soiling, and changes in surrounding structures. If possible, recording regularly from the same position, the same orientation, and the same time of day will make it easier to compare changes. Causes of low power generation may become clearer not from a single inspection but by comparison with past records.

Power generation forecasts are often based on the conditions at the time of installation, so if the on-site environment changes, it may become difficult to achieve the predicted output. When evaluating differences between forecasts and actual performance, it is important in practice to check not only for equipment failures but also how changes in the surrounding environment affect generation.

Reflect downtime and suppression history in actual variances

When generation is low, it is also important to check how much time the power generation equipment was actually in a state capable of generating. Generation forecasts often assume that the equipment is operational. In reality, however, there can be periods when generation is impossible or output is limited due to inspections, repairs, communication failures, protection trips, unexpected equipment shutdowns, grid-side constraints, output curtailment, and so on. If these are not reflected in the comparison between forecasts and actuals, you cannot correctly determine the reason for low actual generation.

First, you should confirm whether there was any downtime. Review the shutdown history of the power conditioners and related equipment, alarm logs, recovery logs, and on-site work records to determine if power generation was interrupted during the period in question. If an outage occurred, document the equipment that stopped, the outage start time, the recovery time, the reason for the outage, and the scope of the impact. Even a short outage can have a large effect on power generation if it occurs during periods of strong sunlight.

Next, check for generation limits caused by output curtailment or control. Even when generation equipment is functioning normally, output may be curtailed for certain periods due to grid conditions or operational controls. In this case, the generated energy will be lower, but this is different from an equipment fault. Looking at the generation curve, you may see it plateau at a fixed output despite sunny conditions and sufficient solar irradiance, or show output being restricted during specific time periods.

When checking curtailment history, verify not only whether curtailments occurred but also whether those curtailments were incorporated into the forecasts. If the forecasts did not include curtailment, the difference between actuals and forecasts may be attributable to the curtailment. On the other hand, if the forecasts already anticipated some curtailment but the discrepancy is larger than that, other factors need to be considered.

Also, attention must be paid to communication errors and data loss. Even if generation appears low, it may actually have been generating but the monitoring data is missing. Cross-checking the monitoring-screen data with the electricity meter values, records of electricity sold, and logs from on-site equipment makes it easier to distinguish whether the drop is in the data or an actual reduction in generation. Verifying the reliability of the data used to conclude that generation is low is fundamental to investigating the cause.

To reflect downtime in the difference between actual results and forecasts, it is helpful to roughly estimate the expected generation during the period when operations were stopped. A stoppage in the early morning or evening has a different impact than one around noon, even if both last the same hour. By referring to generation forecasts or the output curves of nearby days and organizing how much generation would likely have occurred had the system not been stopped, you can identify the portion of the forecast deviation that can be explained by the downtime.

In practice, simply reporting "there was a stoppage" is insufficient when explaining the cause of low power generation. By recording the downtime, the equipment involved, the affected capacity, and the expected impact on generation together, it becomes easier to explain to stakeholders. This is especially true when managing multiple facilities or multiple systems/grids; if stoppage histories are not well organized, it takes time to reconfirm the cause when similar drops recur.

When comparing power generation forecasts with actual performance, it is necessary to exclude periods when the system was not operating or was constrained. If equipment was stopped or was being controlled, the actual output will be correspondingly lower. After confirming outages and curtailments, if unexplained discrepancies remain, proceeding to additional investigations such as shading, soiling, equipment malfunction, or wiring faults will make it easier to isolate the cause.

Examine generation imbalances by system and by equipment

When generation is low, looking only at the total for the entire facility can make it hard to see the cause. Even if the overall figure is lower than forecast, the possible causes change greatly depending on whether every system is equally low or only some equipment or strings are underperforming. Therefore, when comparing generation forecasts and actuals, it is important to break down and check them by system, by power conditioner, by string, and by area wherever possible.

If all strings are similarly low, candidate factors are those that affect the entire installation, such as weather, solar irradiance, temperature, overall soiling, output curtailment, and deviations in forecasting assumptions. On the other hand, if only a particular string is significantly low, you should focus on checking the equipment, wiring, connections, shading, soiling, fuses, breakers, and input circuits related to that string.

In comparisons by equipment, it is useful to view multiple pieces of equipment under the same conditions side by side. When several systems on the same site have the same orientation, the same tilt, and similar capacities, their power generation trends tend to be similar. If only one is lower, the likelihood of a local issue increases. Conversely, if all are similarly low, prioritize checking the weather and overall site conditions.

The shape of the power generation curve is also important. Features of the curve—such as only certain equipment showing low output from the morning, a sudden drop at midday, not rising above a certain output, intermittent shutdowns, or large day-to-day variations—can narrow down the possible causes. For example, if output is low only during a specific time period, shading or control issues may be possible; if output is low all day, possibilities include dirt, circuit faults, equipment malfunction, or connection abnormalities.

With string-level monitoring, you can observe differences in greater detail. However, as the data becomes more granular, it is also necessary to distinguish between normal variability and abnormalities. When module orientation and tilt, the way shading is received, or wiring configuration differ, you cannot judge based on simple numerical comparisons alone. The basic principle is to compare systems that are under the same conditions.

When examining system-level imbalances, cross-checking equipment drawings and system diagrams is essential. The system name on the monitoring screen alone can make it difficult to tell which physical area it corresponds to on site. If you find a system with low generation, verify which row, which area, which module group, and which wiring route on site that system corresponds to. This allows you to narrow down the targets for on-site inspection.

Also, even if a string shows low power output, avoid immediately concluding that the equipment has failed. Environmental factors—shaded locations, areas with concentrated soiling, places where weeds have grown, or rows where snow tends to remain—can appear as variations between strings. It is necessary to make a determination by combining equipment logs with on-site conditions.

When the facility's overall power output is low, considering only the aggregate value makes it difficult to pinpoint the cause. By examining the data by system and by equipment, it becomes easier to determine whether the issue is system-wide or localized. With that distinction, you can limit the scope of on-site investigations and more easily prioritize inspections.

Review the method for creating forecast values and the timing of updates

When comparing generation forecasts with actual performance, people tend to focus only on the actuals, but it is also necessary to verify whether the forecast values themselves are appropriate as the current basis for judgment. Sometimes the reason you feel the generation output is low is not actual equipment malfunction but that the forecast values are too high or do not reflect current conditions.

Predicted values are heavily influenced by the assumptions made when they are produced. Results change depending on how factors such as solar irradiance, ambient temperature, installation tilt, orientation, installed capacity, loss rates, shadow assumptions, expected degradation over time, the treatment of output curtailment, and assumptions about maintenance outages are set. The power generation forecasts in design documents are based on conditions at the planning stage and therefore may differ from the actual performance after operation begins.

For example, if the surrounding environment changes after installation, part of the equipment is replaced, the grid configuration is altered, curtailment increases, or the site is more prone to soiling than anticipated, the gap between the initial forecast and actual results may widen. In such cases, rather than indicating low power generation, it may be that the forecast itself needs to be revised.

Also, for long-term operation, consider output changes due to aging. Photovoltaic modules and related equipment gradually experience performance changes with prolonged use. The degree of change varies depending on the equipment and environment, but if initial forecasts are used as the same standard for many years, discrepancies with reality tend to arise. It is important to confirm whether the forecast values reflect aging-related changes and whether they have been updated to match the current condition of the equipment.

When reviewing how forecast values were produced, organize which weather data were used, how the loss rate was set, how much shading and soiling were anticipated, and whether curtailment and downtime were included. Forecasts whose assumptions are unclear can be difficult to use as a basis for root-cause analysis. To explain differences between forecasts and actual results, you also need to verify the rationale behind the forecast values.

In operations management, it is easier to stay organized if you distinguish and use the design-stage power generation forecast, the annual plan, the forecast for monthly management, and the daily expected generation. The design-stage forecast is used as a guideline for long-term profitability and planning, while for daily anomaly assessment you use management criteria that reflect weather conditions and past performance, applying different forecasts according to their purpose. Judging everything by a single forecast value can lead to excessive anomaly determinations or oversights.

Even when revising forecast values, avoid attributing low performance solely to forecasting errors. It is important to check both deviations in the forecast assumptions and the actual condition of the equipment. For example, if the forecast is somewhat optimistic while there are also issues such as weed growth or a history of shutdowns, the causes of the discrepancy are not singular. If you record separately how much each factor contributed, it becomes easier to devise corrective measures.

When comparing during periods of low power generation, it is important to treat forecasts not as absolute truths but as management benchmarks based on assumptions. By checking how the forecast was created and when it was updated, you can calmly distinguish whether the issue lies with actual performance or with the comparison baseline.

Summary: When power generation is low, explain the reasons for the discrepancy in the records

The purpose of comparing predicted and actual generation when output is low is not simply to find numerical differences. It is to determine whether those differences can be explained by the weather, seasonal factors, shading or soiling, shutdowns or output curtailment, equipment-specific biases, or differences in the assumptions behind the forecasts. Rather than attributing the cause to a single factor, build up verifiable information step by step to improve the accuracy of your judgment.

The first thing to check is the comparison conditions. If the forecast and actual periods, aggregation units, installed capacity, and data acquisition methods are not aligned, the evaluation of differences will be unstable. Next, check solar irradiance, weather, temperature, and seasonal factors to see what can be explained by natural variability. On that basis, checking shading, soiling, weeds, snow accumulation, outage history, output curtailment, and system-specific biases makes it easier to narrow down the areas of the equipment that require investigation.

When power generation is low, there is a tendency to want to identify the cause quickly, but if you make an assessment without keeping records, it becomes difficult to explain later. It is important to record when and over what period, compared with which forecast, how large the difference was, and which factors can account for that difference. Combining the power generation curve, solar irradiance, weather information, outage logs, on-site photos, work records, and the single-line diagram makes it easier to present an explanation that stakeholders can understand.

Also, to determine whether a drop in power output is due to a temporary natural factor or a persistent equipment issue, it is essential to check not just single data points but time-series data. By checking whether the same decrease continues on sunny days, whether it only drops during specific time periods, or whether only a particular system is low, you can narrow down the likely cause. If necessary, consult a specialist or a qualified professional, and proceed with inspections while ensuring safety.

Compare power generation forecasts and actual output under the same conditions and document the reasons for any differences; doing so makes it easier to take action when generation is low. Before determining the cause, confirming each of the comparison conditions, natural factors, site environment, outage history, system-specific biases, and the assumptions behind the forecasts one by one is a safe approach that also makes the results easy to use in public materials and internal reports.