How Delfos diagnosed physical damage to photovoltaic modules through data analysis

Context

During the continuous monitoring of a large-scale solar power plant (operated by a major client in the renewable energy sector), Delfos’ Performance Engineering team noticed abnormal behaviour in one of the units. The inverter 2 was consistently underperforming compared to the other inverters at the same plant.

Upon evaluating the indicators, it was found that the specific output (kWh/kWp) of this inverter was approximately 4% below the average for the rest of the plant. Although a 4% deviation does not trigger shutdown or critical failure alarms, as the equipment continues to operate and feed energy into the grid, this ongoing underperformance represents a silent and cumulative financial loss that requires investigation.

‍

Implemented Solution

To understand what was dragging down the inverter’s performance, the Delfos team utilised the platform’s granular data to carry out a multi-layered diagnosis, ruling out obvious faults:

• String Box Mapping: an in-depth analysis of the data enabled the team to pinpoint the problem. The platform identified that the deviation was not widespread across the inverter, but concentrated in a few specific string boxes connected to it, which showed clear signs of underperformance.
• Ruling Out Common Electrical Faults: the next step was to check availability. The data confirmed that none of the strings were out of operation (disconnected). This finding made the diagnosis non-trivial: power was flowing, but in smaller quantities.
• Prescriptive Guidance: with this evidence, the Delfos team presented the case to the client during the Performance Engineering meeting. The anomalous behaviour was explained and the highly probable hypothesis raised that the root cause was physically damaged photovoltaic modules in those specific string boxes.

‍

Results Achieved

Armed with precise location data (affected inverters and string boxes) provided by Delfos, the client’s team was able to carry out a highly targeted intervention, without wasting time on general inspections:

1. Root Cause Validation: the on-site visual inspection confirmed exactly the hypothesis raised by the platform. Several damaged photovoltaic panels were found in the indicated strings.
2. Targeted Recovery: confirmation of the physical damage directly explained the observed underperformance. The collaborative effort enabled the client to swiftly replace or repair the affected modules.
3. Performance Restoration: the inverter’s performance was restored to normal, returning to the plant’s average specific production levels and halting the ongoing 4% loss of capacity.

Conclusion

This case perfectly illustrates the limitations of basic monitoring and the value of data intelligence. Broken or damaged solar panels do not trigger alarms, and the inverter to which they are connected will continue to operate, albeit inefficiently. The Delfos platform’s ability to drill down from the inverter level to the string box level, combined with the Performance team’s expertise in ruling out common electrical faults, transformed a suspected ‘4% loss’ into a precise physical target for the O&M team to resolve.

Key Data

• A 4% persistent shortfall in the specific output of the Inverter 2 compared to the plant average, detected proactively.
• High-precision diagnosis: an isolated issue in specific string boxes, saving the O&M team time on-site.
• 0 SCADA alarms: the fault (broken panels) kept the electrical connection active, rendering it invisible to traditional alert systems.
• Root cause confirmed: hypothesis of damaged photovoltaic modules 100% validated during the first field inspection.

‍