Introduction
Every thermal anomaly detected by Sitemark has an estimated loss associated with it. It's visible in the information overview of a specific anomaly and in the chart view.
To learn more about loss estimation in charts view you can use this support article.
How was the Loss Estimation created?
The loss estimation system was developed with Imec* as part of the Analyst research project. A rule-based AI system was trained on monitoring system data from several sites to learn how the type of anomaly and the temperature information influence the induced losses.
Note that these losses are estimations based on the thermal signature of the anomaly. They're typically quite accurate, relatively speaking, allowing you to prioritize anomalies with respect to each other.
The loss depends on many factors (irradiance, air temperature, other panels in the string, etc.), so they should not be used to predict accurate energy gains.
How is the Loss calculated on a panel?
The table below shows the estimated loss (as a %, where applicable), the confidence level, and how each value is derived, with the five most common anomaly types listed first by frequency and the rest ordered from highest to lowest loss.
Anomaly Type | Estimated Loss % | Confidence | What this finding means |
Multi Hotspot | Calculated based on ∆T | Estimated | Several hot areas on one panel, each pointing to underperformance. The loss is estimated from how much hotter these areas are than the rest of the panel, adjusted to standard sunlight. |
Hotspot | Calculated based on ∆T | Estimated | A localised hot area usually means part of the panel is underperforming. We estimate the loss from how much hotter the spot is than the rest of the panel, adjusted to standard sunlight so results stay comparable between inspections. |
Single Bypassed | 33.30% | Standard value | One of the panel's three sections has been switched out of the circuit by its built-in protective diode, so the panel produces roughly one-third less power. This is a fixed, well-understood value. |
String (open circuit) | 100.00% | Maximum estimate | The whole string appears to be disconnected and not producing. ** |
Suspected PID | 10.00% | Estimated (indicative) | An early, unconfirmed sign of potential-induced degradation. A small loss is applied, and we recommend monitoring these panels over time to see how they develop. |
Inverter | 100.00% | Maximum estimate | When an inverter appears to be offline, all of the panels it serves can stop producing. ** |
Combiner Box | 100.00% | Maximum estimate | When a combiner box appears to be offline, all of the strings connected to it can stop producing. ** |
String (reversed polarity) | 100.00% | Maximum estimate | The string shows signs of a wiring or electrical fault that stops it delivering useful power, so the full string output is counted as lost (100%). We recommend an on-site electrical check to confirm the cause. |
Module (open circuit) | 100.00% | Standard value | The panel appears disconnected and is not contributing power, while the rest of its string keeps running. The full panel output is counted as lost (100%). |
Missing Module | 100.00% | Standard value | A panel expected from the site design is not present, so it produces nothing (100%). |
Double Bypassed | 66.70% | Standard value | Two of the panel's three sections have been switched out of the circuit by its built-in protective diodes, so the panel produces roughly two-thirds less power. This is a fixed, well-understood value. |
PID | 30.00% | Estimated | Potential-induced degradation is a gradual electrical ageing effect that reduces output and tends to worsen over time. We apply a representative average loss, as the precise figure varies from panel to panel. |
Multi Diode | 30.00% | Estimated | Two or more of the panel's protective diodes appear to be faulty, affecting several sections of the panel. We apply a conservative estimate, as the exact effect depends on the nature of the fault. |
Malfunctioning Tracker | 25.00% | Estimated | A tracker that is not following the sun correctly still produces power, but less than it should. We apply a typical average loss; the exact amount depends on the tracker's position and the time of day. |
Thin film multi cross cell | 20.00% | Estimated (indicative) | A defect affecting several cells across a thin-film panel. We apply a conservative estimate. Thin-film panels are harder to assess by thermography, so on-site confirmation is recommended for the more important cases. |
Single Diode | 15.00% | Estimated | One of the panel's protective diodes appears to be faulty, affecting one section of the panel. We apply a conservative estimate, as the exact effect depends on the nature of the fault. |
Thin film single cross cell | 10.00% | Estimated (indicative) | A localised defect on a thin-film panel. A modest estimate is applied. As with all thin-film findings, on-site confirmation is recommended where the impact matters. |
Misaligned Module | 10.00% | Estimated | A panel mounted at a slightly wrong angle loses a small amount of output compared with its neighbours. A modest average loss is applied. |
Heated Junction Box | 0.00% | Flag only | A junction box running warmer than its neighbours usually points to a connection or safety issue rather than a direct loss of power, so no production loss is assigned. It is flagged for a field check because of the potential connection and fire risk. |
Shaded Module | 0.00% | Flag only | The panel is cooler because something is shading it, rather than showing a fault. Shade moves through the day and the seasons, so no fixed loss is assigned. We recommend removing the source of shade where possible. |
Obstruction | 0.00% | Flag only | Something is blocking the view of the panel, so a reliable loss cannot be calculated. It is flagged so the obstruction can be removed and the panel reassessed. |
Miscellaneous issue | 0.00% | Flag only | A finding that does not fit a standard category. No production loss is assigned automatically; we recommend reviewing it individually. |
Visual issue | 0.00% | Flag only | A finding identified from visual imagery that does not directly indicate a loss of power. It is flagged for review and appropriate follow-up. |
Broken Glass | 0.00% | Flag only | Identified from high-resolution visual imagery. Cracked glass does not always reduce output straight away, so no production loss is assigned on its own. It is raised as a safety and warranty priority, because moisture can enter over time and cause further damage. Any measurable power loss is captured under the accompanying thermal finding, where one is present. |
Broken Glass - Probable | 0.00% | Flag only | The same as broken glass, with slightly lower detection confidence. Reported as a safety and warranty item rather than a production loss. |
Broken Glass - Suspected | 0.00% | Flag only | A possible glass break that could not be fully confirmed from the imagery. Flagged for a quick visual check on site; no production loss is assigned at this stage. |
Missing Data | 0.00% | Flag only | No usable data was available for this item, so it is not counted as a loss. It is left out of the totals rather than treated as a healthy panel. |
Losses are only calculated on the panel level and do not factor in the anomaly's effect on the rest of the string.
* IMEC: Interuniversity Micro-Electronics Center is an independent research center in the field of microelectronics, nanotechnology, artificial intelligence, design methods, and technologies for ICT systems.
** We report the maximum possible impact (100%). The actual figure may be lower if the box is only partly affected, which your monitoring data can help confirm.