5 Steps to Investigate Low Power Generation Caused by Solar Irradiance Sensor Errors

When generated power seems low, many field teams first check the weather, panel dirt, shading, PCS shutdown history, wiring abnormalities, and so on. However, one factor that is easily overlooked is error in the irradiance sensor. If you make a judgment without distinguishing whether the power generation itself is truly low or whether the irradiance data used as the basis for evaluating power generation is offset, you may steer the investigation in the wrong direction. This article explains 5 steps to check the causes of low power generation from the perspective of irradiance sensor errors, aimed at practitioners of photovoltaic power generation equipment, presented along a workflow that is easy to use on site.

Table of Contents

• Why suspect a solar irradiance sensor error when power output is low

• Step 1 Check the relationship between solar irradiance and power generation over time

• Step 2 Find anomalous trends by comparing with neighboring and past data

• Step 3: Check the sensor's installation angle, orientation, dirt, and shadows

• Step 4 Check for apparent errors due to temperature, measurement interval, and missing data

• Step 5: Prevent recurrence by reviewing corrections, replacements, and operational rules

• To improve the accuracy of assessments of declines in power generation

Reasons to suspect solar irradiance sensor error when power generation is low

The power output of a solar power generation system is closely related to solar irradiance. On clear days with high irradiance, output tends to increase, while on cloudy or rainy days output tends to decrease. Therefore, when judging whether output is low, it is important not simply to look at the numerical value of the output, but to check whether the output was reasonable given the weather and solar irradiance on that day.

A problem here arises when the sensor that measures solar irradiance itself has an error. If the irradiance sensor’s reading is higher than the actual value, the power generation equipment may appear to be generating less than it actually is. Conversely, if the irradiance sensor’s reading is lower than the actual value, a decrease in power output may be overlooked. In other words, the irradiance sensor is the yardstick for evaluating power output, and if that yardstick is off, the assessment of the power generation equipment’s condition will also be off.

In practice, when consulted about low power generation, people often immediately suspect panel failures or PCS malfunctions. Of course, those are important items to check. However, if the solar irradiance data used to evaluate power generation is incorrect, you cannot distinguish between equipment-side abnormalities and measurement-side abnormalities. Especially when the remote monitoring screen shows that “generation is low relative to irradiance,” you need to check not only the power generation equipment itself but also the condition of the solar irradiance sensor and the data acquisition system.

Errors in solar radiation sensors can occur suddenly and significantly or develop gradually. Causes vary and include dirt on the sensor surface, bird droppings, fallen leaves, shadows from surrounding structures, misalignment of the installation angle, loosened mounting hardware, poor cable connections, and incorrect data logger settings. Also, even if the sensor’s measured values themselves show no major abnormalities, a mere time offset between power generation data and irradiance data can disrupt apparent evaluations.

To correctly identify the cause of low power output, it is important not to look at the power generation system and the solar irradiance sensor separately, but to check the relationship between them. By confirming whether power output naturally increases when irradiance rises, whether it likewise decreases when irradiance falls, whether the curve on clear days is smooth, and whether any unnatural discrepancies occur only during specific time periods, you can determine the likely direction of the cause.

The five steps covered in this article are not intended solely for specialized analysis. They are organized in an order that makes them easy for on-site personnel to check during routine inspections and remote monitoring. First, check the time series of solar irradiance and power generation, then compare them with nearby and historical data. Next, inspect the sensor installation conditions and the surrounding environment, and check for apparent discrepancies caused by temperature or data processing. Finally, prevent recurrence through calibration, replacement, and review of operational rules.

The phenomenon of low power generation is often not explainable by a single cause. There are cases where, alongside errors in the irradiance sensor, panel soiling or malfunctions in some strings occur simultaneously. Therefore, checking the irradiance sensor should be regarded not as a way to conclude "there is no equipment malfunction," but as a preliminary check to correctly assess a drop in power generation.

Step 1 Confirm the time-series relationship between solar irradiance and power generation

The first thing to do is to align solar irradiance and power generation on the same time axis and examine them. When assessing whether generation is low, looking only at daily or monthly totals makes it difficult to isolate the cause. If you suspect an irradiance sensor error, it is important to at least look at how irradiance and power output vary over the course of a single day.

If conditions are normal on a clear day, solar irradiance increases from morning, becomes high around midday, and decreases toward the evening. Power generation output generally follows a similar pattern. Of course, it will not match exactly due to factors such as system capacity, panel orientation, tilt angle, temperature, output control, and PCS capacity. Even so, when irradiance changes significantly, it is natural for power output to change correspondingly.

If the solar radiation sensor has errors or improper installation, this correlation can break down. For example, if solar irradiance suddenly rises but power output hardly changes, there may be restrictions or abnormalities on the power generation side, but it may also be that the solar sensor is picking up localized reflected light. Conversely, if power output transitions smoothly while only the irradiance is fluctuating finely, you need to check for dirt on the sensor surface, passing shadows, the mounting condition, wiring, and instability in data acquisition.

When checking time-series data, be careful not to judge based only on instantaneous values. On days when clouds pass, both solar irradiance and power output can change dramatically over short periods. Therefore, if you look at only a single instant, the power output may appear low relative to the irradiance. In practice, it is easier to make a judgment by examining trends over multiple time windows, such as intervals of a few minutes, tens of minutes, and one hour. If short-term data are noisy but the one-hour average returns to a natural relationship, it can be considered that weather variability is the dominant influence.

On the other hand, if on sunny days the solar irradiance alone drops unnaturally at the same time every day, suspect shading of the sensor. Surrounding mounting structures, fences, utility poles, trees, buildings, adjacent equipment, etc., can cast shadows on the sensor during certain seasons or times of day. If the irradiance sensor is shaded while the panel surface is not, the power generation system may be operating normally but the solar irradiance will be recorded as low. In this case, the calculated generation efficiency can appear better than it actually is, which can cause an actual drop in generation to be overlooked.

Conversely, if the sensor is placed where it receives more solar irradiance than the actual panel surface, the power output tends to appear lower. For example, if part of the panel surface is shaded but the irradiance sensor is installed higher where there is no shade, the recorded irradiance may look good while the power generation drops. In this case, it is less a sensor error than a situation where the sensor does not represent the light-receiving conditions of the entire installation. When investigating the causes of low power generation, confirming this representativeness is also indispensable.

Also, the time offset between solar irradiance and power output should not be overlooked. If the timestamps of the irradiance sensor data and the power generation data are misaligned, their relationship can appear greatly distorted on cloudy days. If the irradiance peak is displayed several minutes to ten or so minutes offset from the power output peak, the cause may be differences in measurement intervals or clock settings rather than a fault in the equipment. When a remote monitoring system collects data from multiple devices, it is important to check each device’s clock settings and the timing of data aggregation.

In Step 1, examine the relationship between solar irradiance and power generation not only in terms of "high or low" but also from the perspectives of "do they follow the same pattern?", "at what times do they diverge?", and "do trends differ between clear and cloudy days?" If you can narrow down the time periods or conditions under which anomalies occur at this stage, the subsequent comparison work will be easier.

Step 2 Detect abnormal trends by comparing with nearby and past data

After checking the time series of solar irradiance and power generation, next prepare comparison references. It is difficult to determine whether a solar radiation sensor’s readings are correct from a single value alone. Therefore, by comparing them with past data from the same installation, weather trends at nearby locations, other sections within the same power plant, or data from installations under similar conditions, it becomes easier to detect abnormal trends in the solar radiation sensor.

A useful first step is to compare with past sunny days. If the configuration of the power generation equipment has not changed significantly, the relationship between irradiance and power output on sunny days in the same season will tend to show a similar pattern. Of course, it will not match exactly because of ambient temperature, panel temperature, aging degradation, soiling, and changes in the surrounding environment. Even so, if the irradiance alone is unnaturally higher or lower compared with sunny days from last year or a few months ago, that is a reason to suspect a change on the irradiance sensor side.

When comparing, look not only at the monthly totals but also at the daily curve. Even if the monthly solar irradiance hasn't changed much, anomalies can appear only during specific time periods. For example, changes such as low irradiance only in the morning, high only in the afternoon, or a sudden dip around noon may indicate shadows near the sensor, dirt, misalignment of the mounting angle, or the effects of reflections. Anomalies concentrated in particular time periods should be considered not only as equipment failures but also as changes in the field environment.

Comparing with nearby data is also useful. If your measurements deviate substantially from local weather trends, there may be an issue with the solar radiation sensor or with data acquisition. However, because solar radiation is easily affected by passing clouds and localized weather, it is dangerous to immediately conclude a fault just because it does not exactly match nearby data. The purpose of the comparison is not to make the numbers match perfectly, but to see whether an obvious, hard-to-explain bias persists.

When multiple solar irradiance sensors are installed at the same power plant, comparing the sensors is particularly useful. Even within the same site, differences in azimuth, tilt, shading, or mounting height can lead to variations in readings. However, if sensors are installed under similar conditions yet one sensor is consistently higher or lower, you should check the sensor unit itself, the installation condition, the wiring, and the data acquisition settings. If you can identify when the differences between multiple sensors suddenly widened, that point in time can provide a clue as to what happened on site.

When looking at past data, also check the equipment change history. If there have been panel additions, PCS replacements, monitoring-device updates, sensor relocations, wiring work, mowing or tree removal, or construction of new nearby buildings, the relationship between solar irradiance and power generation can change. Even if it looks like a solar sensor error, the actual cause may be changes in equipment configuration or the surrounding environment. When choosing comparison periods, it is important to use periods with conditions as similar as possible.

To determine that power generation is low, power performance indicators are sometimes used. By looking at the ratio of generated energy to solar irradiance, you can smooth out weather variability to some extent when evaluating. However, this indicator depends on the accuracy of the solar irradiance data. If the irradiance sensor reads high, the indicator will look worse; if it reads low, it will look better. Therefore, when the indicator suddenly worsens, you should consider not only faults in the generation equipment but also anomalies in the irradiance data.

When performing comparisons, it is important to determine when an anomaly began. If the relationship between irradiance and power output changes around a specific day, check the work history, inspection history, cleaning history, outage history, and any changes to monitoring settings before and after that day. If the change is gradual, consider sensor soiling, aging, growth of surrounding vegetation, loosening of fixings, and similar factors. Different causes should be suspected for sudden changes versus gradual changes.

The purpose of Step 2 is not to declare the solar irradiance sensor "absolutely correct" or "definitely wrong." It is to judge how much confidence can be placed in the solar irradiance data when investigating low power generation. If comparisons increase suspicion of the irradiance sensor, proceed to an on-site inspection. Conversely, if the irradiance data are stable and only a clear decline in power generation is evident, it becomes easier to prioritize investigation of the equipment.

Step 3 Inspect the sensor's installation angle, orientation, dirt, and shadows

If an anomaly in the solar radiation sensor is suspected from the data, check the installation condition on site. Because the solar radiation sensor is installed outdoors, it is affected by rain and wind, dust and sand, bird droppings, fallen leaves, pollen, yellow sand (Asian dust), salt, snow accumulation, frost, and so on. Even if the power generation equipment itself shows no abnormalities, if the sensor’s light-receiving surface is dirty the irradiance may be recorded lower than the actual value. Since this undermines the premise for judging that power generation is low, checking the condition of the sensor’s light-receiving surface is fundamental.

When checking for dirt, look to see whether the surface is uniformly dulled or whether there are localized deposits. Uniform dulling can make the irradiance appear lower overall. Localized deposits can change the effect of shadows depending on the sun’s position, producing unnatural variations at different times of day. Checking whether the relationship between irradiance and power output improves after cleaning makes it easier to assess the impact of dirt. However, cleaning must be carried out in accordance with sensor specifications and site rules, and in a manner that does not damage the surface.

Installation angle and orientation are also important. Whether the solar irradiance sensor is installed with the same tilt and azimuth as the panel surface or is installed horizontally changes the meaning of the data. If you want to evaluate power generation using the irradiance on the panel surface, the relationship with power output can appear shifted if the sensor’s mounting surface does not match the panel surface. For example, if the panel is tilted but the sensor is installed close to horizontal, the correspondence with power generation will vary by season and time of day.

On site, angles that were correct at installation can gradually change due to wind, vibration, contact during work, or loosening of mounting hardware. Even if there are no obvious visual abnormalities, changes in tilt or orientation will affect measurements. It is important to check the sensor mount, mounting hardware, support posts, cable tension, and loose bolts, and to restore the installation to the installation standards as needed. In particular, checks should be prioritized after strong winds, after snowfall, and after nearby construction.

When checking for shading, it is important not to judge based only on the inspection time. Even if there are no shadows at the time you visit, shadows can occur in the morning or evening, or vary by season. Check in advance the times when anomalies appear in the solar irradiance data, and search on-site for shadow sources during those times to be more efficient. Inspect the surroundings—mounting racks, utility poles, power lines, fences, surveillance camera posts, communication boxes, weeds, trees, etc.—to confirm that nothing casts shadows solely on the sensors.

Also check whether the sensor installation location is representative of the entire power plant. In large plants, terrain and shading conditions can vary across the site. If a solar irradiance sensor reflects conditions only in a specific area, using it to evaluate overall power generation can lead to misinterpretation. For example, even if the area around the sensor is well sunlit, if another section experiences shading from vegetation or terrain, the power output may appear low compared with the measured irradiance. In such cases, the cause is not a failure of the irradiance sensor but a mismatch between the evaluation target and the measurement location.

Check the cables and connections as well. Outdoor wiring can deteriorate or loosen due to ultraviolet light, rainwater, temperature changes, small animals, or contact during work. Poor contact can cause solar radiation data to fluctuate intermittently or suddenly show values near zero. If fine noise or missing data are observed in the data, check not only the sensor itself but also the junction box, converter, data acquisition device, and communication path.

Also, pay attention to reflections around the sensor. Reflections from white walls, metal surfaces, puddles, bright sheets, or nearby equipment can cause the sensor to detect higher-than-normal solar irradiance. The effects of reflections tend to appear in certain seasons or times of day and may be noticeable only on clear days. If solar irradiance is recorded as high while power generation does not increase correspondingly, overestimation due to reflections should also be considered.

In Step 3, link anomalies in the data to the physical conditions in the field. When investigating the causes of low power generation, there are many things you cannot determine from the figures on the screen alone. The irradiance sensor is a small device, but its installation condition affects the overall assessment of generation performance. During on-site inspection, it is important to check and record each of the following: soiling, angle, orientation, shading, representativeness, wiring, and reflections.

Step 4 Verify apparent errors due to temperature, measurement interval, and missing data

Even when there are no obvious abnormalities in the installation of the irradiance sensor, the relationship between power output and irradiance can appear distorted. Typical examples include apparent errors caused by temperature, measurement interval, missing data, and aggregation methods. Before concluding that power output is low, you need to check not only the measured values themselves but also how the data are acquired and how they are aggregated.

Solar power output is influenced not only by irradiance but also by panel temperature. In general, on days with strong sunlight and high ambient temperatures, panel temperatures tend to rise, and the power output may not increase as much even at the same irradiance. Therefore, when it looks like "irradiance is high but power generation is not as high as expected" on a clear summer day, consider output reduction due to temperature as well as possible irradiance sensor errors. If you judge solely by irradiance without checking temperature, you may mistakenly conclude that the generation equipment has a fault.

On the other hand, if temperature correction or the handling of evaluation formulas is not appropriate, the assessment of power generation can become either overly strict or overly lenient. It is important to check whether the performance metrics shown on the monitoring screen assume particular temperature conditions, use temperature data, or simply compare based only on solar irradiance. Before suspecting an irradiance sensor error, you should understand how the performance metrics are calculated.

Differences in measurement intervals are also a major factor. Because solar irradiance can change rapidly over short periods, if the measurement interval for power output differs from that for irradiance, the data may appear to be from the same time but in reality the comparison conditions are not aligned. For example, if irradiance is recorded every minute while power output is logged as 30-minute averages, short-lived changes in irradiance can appear not to match the power output. This mismatch becomes more pronounced on cloudier days.

Data gaps are another point to check. If there are missing values in the solar irradiance data and those gaps are filled with zeros or the previous value, the daily totals or averages can deviate from reality. The same applies when there are missing values in the power generation data. Depending on the method used to handle missing data, generation may appear low, or conversely the irradiance may appear low. When investigating a drop in power generation, check not only the graphs but also whether missing data exist, the imputation method used, and the rules for excluding outliers.

A mismatch in time settings is also a common oversight. If the irradiance sensor, data acquisition device, PCS, and monitoring system each record at different times, data may appear to be for the same time period but actually be offset. Especially on days when clouds move quickly, even a few minutes' offset can make the relationship between irradiance and power generation look unnatural. Time offsets are more noticeable on days with greater variability than on clear days, so be careful not to conclude equipment failure based solely on data from cloudy days.

Also pay attention to the units and aggregation of solar irradiance. Instantaneous solar irradiance and accumulated irradiance over a given period have different meanings. Power output and generated energy are not the same either. Confusing instantaneous power with accumulated energy, or instantaneous irradiance with accumulated irradiance, leads to inaccurate conclusions that the generated energy is low. In practice, it is fundamental to match the types of data being compared and to view them over the same time span.

Even when there are output curtailments or operating limits on the PCS, the relationship between solar irradiance and power output breaks down. Even if solar irradiance is high, if the plant’s output is being curtailed by the equipment, power output will not increase. In this case, even if the irradiance sensor is correct, power output will not be proportional to irradiance. Before suspecting an irradiance sensor error, you can avoid misjudgment by checking PCS shutdown logs, output curtailment logs, grid-side constraints, protection actions, and the timing of maintenance work.

In Step 4, check for apparent errors caused by how the data are interpreted rather than by a fault in the sensor itself. The phenomenon of low power generation can sometimes appear to occur simply because the assumptions for measurement and aggregation are not aligned. When on-site inspections do not reveal any abnormalities, it is especially important to carefully check temperature, measurement interval, missing data, timestamps, units, aggregation methods, and output limits.

Step 5 Prevent recurrence by reviewing corrections, replacements, and operational rules

When the causes of errors in the solar irradiance sensor become apparent, the next step is to implement countermeasures and prevent recurrence. In investigations into low power generation, it is important not to stop at merely detecting anomalies but to stabilize the accuracy of subsequent evaluations. Because the solar irradiance sensor serves as the basis for judging generation performance, continuing to operate with residual errors will affect future anomaly detection and reporting.

The first step is to implement corrective measures appropriate to the cause. If soiling of the light-receiving surface is the cause, perform proper cleaning and verify how the relationship between solar irradiance and power output changed before and after cleaning. If shading is the cause, consider changing the installation location, managing nearby obstructions, or maintaining vegetation. If deviation in tilt or azimuth is the cause, restore the correct installation condition and prevent loosening of the mounting hardware. If there are problems with wiring or connections, check the connection condition and waterproofing, and carry out necessary repairs.

If degradation or failure of the sensor itself is suspected, consider inspection or replacement. Sensors that have been used for a long time may have shifted measurements even if there are no obvious external abnormalities. When performing replacement or calibration, keep the data from before and after the change and check how much the evaluation has changed, as this makes later explanations easier. It is important not simply to replace with a new unit, but to understand how the basis for power generation evaluation has changed as a result of the replacement.

When applying corrections, careful handling is required. Applying a correction factor to the readings of a solar irradiance sensor can temporarily adjust the assessment, but if the root cause is soiling or shading, correction alone will not prevent recurrence. Also, if the basis for the correction is unclear, it will be difficult to judge when reviewing the data later. When applying corrections, it is important to record the date the correction was started, the reason for the correction, the correction method, and the data used for verification.

Reviewing operational rules is also essential. A solar radiation sensor is not something you install once and forget; it should be checked regularly. During routine inspections, check the sensor’s receiving surface for dirt, nearby shading, mounting condition, and cable condition. In remote monitoring, check whether the relationship between solar radiation and power generation has suddenly changed, whether there are unnatural discrepancies at specific times, and whether missing data or abnormal values are increasing. Combining on-site inspections with data checks enables early detection of anomalies.

Standardizing the verification procedure for inquiries about low power generation is also effective. If the order of checks differs by person in charge, the accuracy of root-cause investigations will vary. First confirm the drop in power generation, then examine its relationship with solar irradiance, compare with past data and nearby trends, check the on-site condition of the irradiance sensor, and, if necessary, proceed to a detailed investigation of the equipment side; deciding on this flow in advance reduces the likelihood of overlooking issues during the investigation.

How records are kept is also important. The dates on which the solar irradiance sensor was cleaned, the mounting angle was adjusted, surrounding vegetation was trimmed, wiring was repaired, or monitoring settings were changed all affect how power generation data are interpreted. If these histories are not recorded, it becomes impossible later to understand “why the relationship between solar irradiance and power generation changed from that date.” Linking inspection records with measurement data and managing them together streamlines analysis of the causes of power output declines.

Also, when reviewing the placement of a solar irradiance sensor, simply choosing a sunny spot is not enough. You need to consider the representativeness for the entire power generation installation, ease of maintenance, low susceptibility to shading, minimal surrounding reflections, wiring safety, and the potential for future vegetation growth or equipment additions. Since the sensor is used for assessing power generation, it is important to choose a location suitable for evaluation rather than one that is merely easy to measure.

The purpose of Step 5 is not only to confirm the cause of the current drop in power generation but also to improve the accuracy of future judgments. At sites where a solar irradiance sensor error has occurred once, the same conditions may cause it to recur. It is important to review cleaning frequency, inspection items, data monitoring items, and contact rules for abnormalities, and to establish a system that can detect signs of low power generation quickly and accurately.

To improve the accuracy of determining declines in power output

When investigating the causes of low power output, errors in solar irradiance sensors are easy to overlook. Even if a power generation system's output appears to have declined, the measured solar irradiance may simply be incorrect. Conversely, low recorded solar irradiance can cause an actual drop in the system's power output to be missed. For that reason, solar irradiance sensors should not be treated as mere accessories but regarded as an important benchmark for evaluating power generation performance.

In the workflow introduced here, we first check the relationship between solar irradiance and power generation as a time series, then compare it with past data and nearby trends. After that, we inspect on-site the sensor installation angle, orientation, soiling, shading, representativeness, and wiring condition, and further check for apparent errors caused by temperature, measurement interval, missing data, time offset, and aggregation methods. Finally, we prevent recurrence through correction, replacement, cleaning, review of installation, and the establishment of operational rules.

It is important not to immediately conclude equipment failure based solely on the result of low power generation. If the benchmark of solar irradiance is not accurate, the assessment of power generation cannot be correct. By checking the condition of the solar irradiance sensor and the data acquisition system as thoroughly as inspections of the power generation equipment, the accuracy of root-cause investigations is likely to improve.

Also, errors in irradiance sensors often arise from small on-site changes. A bit of dirt, a slight angular misalignment, shading only at dawn or dusk, a few minutes' time offset, and differences in how missing data are handled can all affect energy-yield assessment. Each of these may seem like a small issue when viewed individually. However, when explaining the causes of low energy yield and deciding which countermeasures to prioritize, they can lead to significant differences in judgment.

By incorporating inspection of irradiance sensors and verification of irradiance data into routine operations, you can not only detect decreases in power generation early but also reduce the risk of unnecessary on-site inspections and incorrect component replacements. When you feel that power output is low, before looking at the generation equipment itself, first confirm whether the irradiance data used as the basis for evaluation can be trusted. Adopting this mindset will make on-site responses more reliable.

If you have an environment that allows solar irradiance, power output, on-site photos, and inspection records to be reviewed together, it becomes much easier to isolate the cause. Establishing a system that detects signs of reduced power output from both data and on-site conditions, and that can make judgments including irradiance sensor errors, leads to stable operation of solar power generation facilities.