SolarDoc Pro · Operating plants
Plant Measurement & Performance Report
Example 50 MWac / 65 MWp plant, Rajasthan · 4 in-plane irradiance stations · 2024 · 10-minute data
Summary
2 of 4 stations had problems during the year. MET-1, the sensor the plant uses for PR, had a dirty dome in May, a clock running an hour ahead from June to August and a 4-day outage in July, so the reported PR was 1.2 points too high. The plant itself had 2 real events: a trip and a week of curtailment.
What this could be costing you
Estimate- Your PR is reported 1.3 points too high. About 1,611 MWh a year of real underperformance (₹48.3 lakh) is hidden, so it can't be claimed under your PR guarantee.
- Trips and curtailment found in your data lose about 244 MWh a year (₹7.3 lakh) at the rate seen.
Estimate: 98,171 MWh a year from your energy data × the PR error, plus energy lost in plant events, × ₹3/kWh. Use your own PPA tariff for a better figure.
Do the stations agree?
Every station is compared with the plant consensus (the median of all stations) day by day. When one walks away from the others under a clear sky, it's the sensor. Short disagreements are passing clouds and are left alone.
- MET-3 · 22 Mar – 17 AprReads 5.9% low against the other stations under clear skies: soiling or calibration
- MET-1 · 1 May – 31 MayReads 2.7% low against the other stations under clear skies: soiling or calibration
- MET-1 · 10 Jun – 31 AugClock about 1 hour off: mornings and afternoons disagree with the other stations
- MET-1 · 20 Jul – 24 JulNo data while the other stations were reporting
- Not a fault · 29 daysStations differed briefly as clouds passed across the site. These readings are correct and were kept.
Monitoring class
The station meets Class B. 4 changes reach Class A.
Checked from the data (recording interval, availability, station agreement) and the station details supplied (sensors, cleaning, calibration).
- RequirementThis plantClass A needsTo close the gap
- ✗Recording interval10 min5 min or less (1 min best)Set the logger to store 1-minute averages
- ✗Global horizontal sensors13Add 2 horizontal pyranometersSMP10 · Get a quote
- ✗Recalibrationlast 3 years agoevery 2 yearsRecalibrate the sensors (swap with calibrated units to avoid downtime)calibration service · Get a quote
- ✗Module temperature sensors69Add 3 module temperature sensors
- ✓Sampling interval3 s5 s or less–
- ✓In-plane irradiance sensors (50 MWac plant)43–
- ✓Pyranometer classClass A, spectrally flatClass A, spectrally flat–
- ✓Calibration uncertainty1.8%2% or less–
- ✓Sensor cleaningweeklyweekly, or a soiling monitor–
- ✓Sensor inspectionweeklyweekly–
- ✓Air temperature and windmeasuredmeasured on site–
Product suggestions are independent recommendations; product names are trademarks of their owners.
PR Check: is it the plant, or the sensor?
Monthly PR
May: MET-1's dome was dirty and read low, so PR looked better. July: MET-1 stopped logging for 4 days while the plant kept producing, so the reported PR jumped.
Power vs irradiance
With the clock an hour off, mornings and afternoons split into two loops. After cleaning, the points fall on one line, so whatever stays off it is the plant: clipping at the inverter limit (flat top), curtailment, and a trip.
Real plant events
- 1 May, 11:40–13:50Trip: no power while the sun was strong
- 28 Oct – 3 Nov, afternoonsCurtailment: output held at 30 MW
Statement for EPC and O&M performance reviews
The irradiance data used for PR was checked by SolarDoc against all 4 plant stations. After removing 3 sensor faults at MET-1, the 2024 PR is 78.2%, not 79.4%. The 2 events listed above are plant-side; the rest of the difference comes from the measurement.
Method
- Station agreement: each station's daily irradiance is divided by the plant consensus (median of all stations). Differences above 2.5% for three days or more are flagged; a morning/afternoon split points to the clock.
- Cleaned irradiance: the consensus of the stations that are healthy at that moment.
- PR: AC energy divided by in-plane insolation times DC capacity, by month. Intervals without valid irradiance are left out rather than counted as zero.
- Plant events: zero output with strong sun (trip), output held flat below the rating while the sun would allow more (curtailment), flat top at the inverter limit (clipping).
- Monitoring class: recording and sampling intervals, sensor count for the plant size, sensor class and calibration, cleaning and inspection.
This sample uses a simulated plant. Automated analysis complements, not replaces, an independent engineer's review.