Solar plant monitoring: what to measure and why

A solar plant produces what the sun delivers, minus losses. Monitoring exists to tell one from the other: how much of the shortfall is clouds and how much is a string that tripped, a failed inverter, soiling or a loose cable. Without measuring irradiance, every bad day looks like a cloudy one.

This article lists what to measure on a distributed-generation plant, from an industrial rooftop to a ground-mounted array, and explains what each measurement answers. We work as integrators: the inverter, the pyranometer and the sensors are standard market products; what we design is the gateway that reads them and the platform that brings it all together.

Generation: per inverter and per string

The inverter already measures and exposes, over Modbus, instantaneous power, cumulative energy, DC and AC voltage and current, frequency and internal temperature. It is the first data to collect, and on small plants it may be the only one.

On larger plants, the interesting data sits one level down: the current of each string (a set of modules in series connected to one inverter input). Inverters with multiple MPP trackers expose current per input; string combiner boxes with metering expose each string. Comparing strings is what reveals a broken module, a burnt connector or shading, because irradiance is the same for all of them.

Irradiance

A pyranometer measures solar radiation in W/m². It must sit in the plane of the modules, with the same tilt and orientation, because that is what the modules receive. A reference cell does the same job at lower cost, with a spectral response closer to the module’s.

Without irradiance there is no way to know how much the plant should have generated. With it, every day has an expected value and an actual one.

Module temperature

Silicon modules lose power as they heat up: the typical coefficient is around −0.3 to −0.4% per °C above 25 °C, and the exact value is on the module datasheet. A sensor bonded to the back of a representative module, plus ambient temperature and wind from the weather station, let you correct the expected value and explain why a 40 °C afternoon produces less than a 28 °C morning under the same sun.

Performance ratio

The indicator that sums up the plant’s health is the performance ratio (PR), defined in IEC 61724-1. It is the ratio of the final yield to the reference yield:

PR = Yf / Yr

Where Yf is the AC energy delivered over the period (kWh) divided by the plant’s nominal DC power (kWp), and Yr is the plane-of-array irradiation over the period (kWh/m²) divided by the reference irradiance of 1 kW/m². In expanded form:

PR = E_AC / (P_nominal × H_POA / G_ref), with G_ref = 1 kW/m²

Example: a 1,000 kWp plant, a day with 5.8 kWh/m² of plane-of-array irradiation, 4,640 kWh generated. PR = 4,640 / (1,000 × 5.8) = 0.80.

PR cancels out the clouds, because irradiance is in the denominator. What remains is loss: temperature, soiling, shading, cabling, inverter efficiency, downtime. That is why PR drops in summer (heat) even on a healthy plant, and the right comparison is with the same month of the previous year or with the temperature-corrected PR that the same standard defines.

Inverter alarms

The inverter logs events with codes: grid voltage or frequency out of range, DC insulation fault, arc fault, overtemperature, fan failure, low DC input voltage, anti-islanding trip. The platform reads the status register and the event vector and turns each code into an alert with a name and a timestamp. The inverter manual has the code table, which varies from manufacturer to manufacturer.

Two alarms deserve their own rule: inverter not communicating and inverter at zero with irradiance above a threshold. The first is a data-network failure; the second is a plant standing still on a sunny day.

What the utility requires, and what it doesn’t

It is worth separating regulation from commercial practice. Grid connection of a distributed-generation plant is governed by the grid-connection and net-metering rules of your country and your utility. In general terms, they cover:

• how the plant owner applies for connection, how the utility reviews the request and inspects the installation before energizing it;
• the revenue metering, usually a bidirectional meter that records what is imported and what is exported, and who pays for it;
• how exported energy is credited on the bill;
• the technical requirements for connection and protection (anti-islanding, voltage and frequency ranges), set by the grid code and the utility’s technical standards.

Check what yours actually require before assuming an obligation. In Brazil, for example, the distributed-generation regulation does not require the plant to send generation data to the utility continuously: the utility’s own meter already records what flows in and out. The monitoring described here is for the plant owner, for the operation and maintenance contract, for the investor and, in shared or community generation schemes, for the participants.

What to do now

1. Check whether your inverter has a Modbus port (RS-485 or Ethernet) and whether the manufacturer publishes the register map.
2. Install a pyranometer or reference cell in the plane of the modules. Without it there is no PR.
3. Define the alerts that matter: inverter stopped in sunshine, a string below its siblings, PR below the monthly target.

If you’d like to discuss a project, talk to one of our engineers.

---

Sources: IEC 61724-1, Photovoltaic system performance, Part 1: Monitoring.