Understanding portfolio and plant data is a crucial pillar of creating an effective O&M strategy. Mere monitoring of a plant’s uptime is not adequate to address potential issues, which may arise. Furthermore, in order to minimize operating expenses, O&M providers should leverage both real-time and predictive data analysis. Rigorous examination of dynamic plant conditions, i.e. soiling and module degradation, with sophisticated data analytics enables the O&M providers to customize an optimal maintenance schedule. This leads to further security of a plant’s profitability in the long run. During this webinar, Ben Hansen & Zach Kreifels from SMA and Alexander Wolf from Meteocontrol discussed the role which effective data gathering, sorting and presenting play in bolstering O&M providers to construct an effective O&M strategy. The video recording of the webinar and the speakers’ slides are available for download.
1. Data Gathering & Visualization
1.1 Using Analytics to detect unknown issues
According to Ben Hansen, all plants have some hidden issues, most of which show up as an alarm or an error in the inverter. However, some of the more common issues are related to modules and cabling, and include glass breakage, bypass diode failures, module disconnects and short circuits. In contrast, inverters have a lower number of issues but when they do, they could have a greater impact. Nonetheless, inverters tend to serve as good connection points for the data on plants, making them a great starting point for evaluating overall plant performance data. Analytics can help in identifying the aforementioned unknown issues. Kreifels then further explored the key considerations when dealing with data and data analytics in this webinar.
1.2 Using Predictive Analytics to Minimize Operational Expenses
Suffice it to say that it is always better to repair, while it is cheap, than postpone to a later time, when it becomes more costly. Software can run an analysis throughout a period of time and, with automation and machine learning, a course of action can be recommended to preempt a bigger repair issue.
1.3 Optimizing Maintenance Schedules to Impact Profitability
-
Manage for least production impact and greatest revenue output: Any downtime of a plant has to be managed in order to minimize the revenue impact. Predictive analytics can immensely help to optimize maintenance scheduling, which will have the greatest influence on revenue.
-
Schedule preventative at optimal times: Preventive maintenance schedules tend to be carried out on a calendar schedule, which is typically based on the requirements of manufacturers, in order to sustain warranty coverage.
-
Maximize repairs when you will get the most return: Using data analytics, repairs can be scheduled when the most positive impact is maximized. For instance, the residential market are very good at maximizing the value of a truck roll, according to Hansen. He added that the aforementioned strategy may not work equally in utility-scale applications, nonetheless similar principles can be adopted.
2. Data Monitoring
2.1 Evaluating the power yield from “Irradiation” utilizing sensor data
Firstly, the irradiation needs to be measured using data collected through pyranometers or weather stations. Furthermore, according to Meteocontrol, this data should be measured in real-time and not based on an estimated value, for instance using an extrapolated track record. The actual irradiation value measured at the site is then utilized to calculate the expected power generation. The final step would be to compare the actual power generation against the expected power generation.
2.2 Portfolio Management
There is an abundance of big data from the PV system, which makes for an optimized overview of the system. According to Wolfi, the essential KPIs to measure performance are:
| For O&M providers | For asset managers |
|---|---|
| Performance ratio (PR) | Processing time of the maintenance |
| Open alarms, specific yield | Reaction times of the operator |
| Data availability | Invested time per “Asset” |
| System availability | Number of tickets per Portfolio |
| Fault Statistic |
