40.2 MWh/month Loss Reduction with Disconnected String Box Identification

Context

During scheduled maintenance on a Sungrow SG3125HV central inverter, six string boxes were temporarily disconnected to carry out the activity. After the maintenance was completed, only five of them were correctly reconnected. One of the string boxes inadvertently remained out of operation, without the field team or staff noticing the fault, since, although the SCADA system recorded the data, detection would have required manual analysis. In the Delfos system, automatic alerts and specific indicators made this identification faster and more assertive.

The fault was only identified thanks to analysis carried out by the Delfos Performance Engineering team, which observed an anomalous reduction in the inverter's generation, despite the absence of alarms or local operational indications.

Each string box represents 16 strings of photovoltaic modules, and their disconnection implied an estimated daily loss of 1.34 MWh - which, if not quickly identified, could result in monthly losses of over 40 MWh.

‍

Implemented Solution

Delfos used different analytical modules of its platform to quickly identify the fault:

1. Comparative Power Analysis
Initially, the team detected a significant reduction in the active power of the ITS11_INV2 inverter when compared to the neighboring ITS11_INV1 inverter, both with the same specifications.

‍

2. Visual detection with Performance Heatmap
With the Performance Heatmap module (part of Performance Analyses), it was possible to clearly observe, day by day, the lack of performance of one of the string boxes.
The analyses were made taking into account the KPI Performance Ratio per string box over the course of the days.

‍

3. Even after reconnection, the Heatmap indicated underperformance, showing that the string box, although connected, still required quality inspection to recover its ideal performance - an action that could have gone unnoticed without this type of visualization.

‍

Achieved Results

- Preventive identification of operational failure: The Delfos platform detected the problem before it generated significant losses and reported it to the client via Performance Engineering.
- Fast response time: The fault was reported on 16/10 and corrected the following day (17/10).
- Losses avoided: It is estimated that losses of up to 40.2 MWh/month were avoided, equivalent to €1,272.66 in non-generated revenue.

‍

Conclusion

This case reinforces the fundamental role of the Delfos APM platform in the early detection of subtle operational faults that are difficult for conventional systems, such as SCADA and routine inspections, to detect. With analytical resources such as Performance Heatmap and automated comparison between similar assets, Delfos provides a level of technical supervision that anticipates failures, avoids substantial financial losses and improves asset performance.
In addition, the continuous monitoring carried out by the Performance Engineering service enables proactive action to be taken, strengthening the operational reliability of the monitored assets.

‍

- 1 disconnected string box not automatically identified by SCADA - required manual analysis
- 1.34 MWh/day of potential loss
- 40.2 MWh/month of avoided generation
- 100% of the fault identified and resolved in less than 24 hours after the alert from Delfos

‍

‍