Solar KPIs: 12 Essential Metrics With Simple Formulas

Imagine a monitoring dashboard that says a solar plant produced 13,860 kWh last month. Is that good? You cannot answer that yet, because the answer depends on how big the plant is, how much sun it received, whether the inverters were running, how dirty the panels were, and how old the modules are.

This is where solar KPIs come in. They are the handful of numbers that turn a raw energy reading into a judgement about the plant. I come from an electrical and electronics background, and solar is one of the places where that interest and data analysis meet. So I wanted to understand which numbers solar teams actually watch and how each one is calculated. This article is that exploration, written for fellow learners.

If you want data to practise on while you read, the BloomInData dataset library is a good place to start.

I use one hypothetical plant, Plant A, for every formula, so you can see how the 12 metrics connect.

The example plant used throughout

These numbers are invented to make the arithmetic clear. They are not real plant data.

Other inputs appear where they are needed.

Group 1: Resource and output solar KPIs

1. Plane of array (POA) irradiation

What it is: The sunlight energy that actually lands on the tilted surface of the panels, measured in kWh/m².

Why it matters: Every quality metric compares what the plant produced with what the sun made available. Without the right sunlight number, the comparison is meaningless.

Decision it supports: Whether a weak month came from the weather or from the plant.

2. Energy yield (kWh)

What it is: The electrical energy the plant delivered in a period.

Decision it supports: Billing, revenue tracking, and the starting point for nearly every other KPI.

3. Specific yield (kWh per kWp)

What it is: Energy produced per kWp of installed DC capacity. It lets you compare plants of different sizes.

Decision it supports: Comparing a 10 kWp rooftop and a 5 MWp ground plant on a fair basis.

Group 2: Quality metrics

4. Performance ratio (PR)

What it is: The share of the theoretically possible output that the plant actually delivered, given the sunlight it received.

The IEC 61724 definition describes performance ratio as measured output divided by expected output for a reporting period, based on the system’s nameplate rating. PR is widely used in performance guarantees and in operations and maintenance.

Decision it supports: Is the plant healthy, whatever the weather was?

5. Temperature corrected performance ratio

What it is: PR adjusted so that hot weather does not make a healthy plant look weak. Temperature corrected performance ratios reduce seasonal variability.

Decision it supports: Tracking health over many months and years without summer dips fooling you.

Group 3: Utilization metrics

6. Capacity Utilization Factor (CUF) or capacity factor

What it is: Actual energy produced divided by the energy the plant would produce running at full rated capacity for every hour of the period.

Decision it supports: Financial modelling, tariff checks and lender reporting.

7. Availability

What it is: The share of time the plant was ready to produce when the sun was available.

Decision it supports: Judging O&M contractor performance and finding where lost energy came from.

Group 4: Loss metrics

8. Soiling ratio

What it is: How much dust and dirt reduce output. NREL describes the soiling ratio as the output of a soiled module divided by that of a clean module. A value of 1 means clean, and anything below 1 means soiling. The IEC 61724 monitoring standard includes a procedure for measuring it.

Decision it supports: Cleaning schedules. Compare the energy value of the lost kWh with the cost of a cleaning cycle.

9. Inverter efficiency

What it is: How much of the DC power arriving at the inverter leaves as AC power.

Decision it supports: Spotting a failing or underperforming inverter early.

10. DC to AC ratio

What it is: DC panel capacity compared with inverter AC capacity.

Decision it supports: Judging whether clipping losses are acceptable in a design review.

Group 5: Long term and money metrics

11. Degradation rate

What it is: The yearly loss in a module’s or system’s power output.

Decision it supports: Long term revenue forecasts and warranty claims.

12. Levelized Cost of Energy (LCOE)

What it is: The total lifetime cost of a system divided by its total lifetime energy output, with both discounted over time. It lets you compare solar with other sources, or one project with another.

Decision it supports: Go or no go on a project, and comparing technologies.

Calculating performance ratio in SQL

The question: What is each plant’s monthly PR?

The query assumes a hypothetical table called daily_plant_readings, with one row per plant per day. It adds up a month of energy and a month of irradiation first, then divides, following the PR formula above. A monthly PR column lets an analyst rank plants, spot a falling trend, and decide which site needs a visit.

SELECT
    plant_id,
    YEAR(reading_date)  AS yr,
    MONTH(reading_date) AS mth,
    SUM(ac_energy_kwh)  AS energy_kwh,
    SUM(poa_irradiation_kwh_m2) AS poa_kwh_m2,
    SUM(ac_energy_kwh) /
        (MAX(dc_capacity_kwp) * SUM(poa_irradiation_kwh_m2)) AS performance_ratio
FROM daily_plant_readings
GROUP BY plant_id, YEAR(reading_date), MONTH(reading_date);

Using solar KPIs together: a diagnostic path

Suppose Plant A’s PR drops from 84% to 79% next month. Here is how I would think about it, moving from question to data to decision.

1. Is it the weather or the plant? PR already adjusts for sunlight, so a real drop deserves attention. Check the temperature corrected PR to rule out heat.
2. Was it downtime? Check availability and inverter logs.
3. Was it dirt? Check the soiling ratio and the date of the last cleaning.
4. Was it one inverter? Compare inverter efficiency across units.
5. Was it a sensor problem? Check the POA sensor before blaming the plant. A faulty irradiance reading can fake a PR change.
6. Decide. Dispatch a cleaning crew, raise an O&M ticket, or recalibrate the sensor.

IEC lists the purposes of performance monitoring as spotting trends, locating faults, and comparing performance with design expectations and guarantees. The path above uses all three.

Common mistakes

• Comparing PR for different periods without adjusting for temperature.
• Using GHI where POA irradiation is required.
• Quoting CUF without stating whether the capacity is AC or DC.
• Averaging daily ratios instead of calculating from monthly sums.
• Treating one noisy year as a degradation trend.
• Reading LCOE as a selling price.

What to learn next

Try rebuilding Plant A in Excel or Python, then add a year of invented daily data and chart PR, temperature corrected PR and availability together. Because solar output follows a seasonal pattern, it also pairs well with time series forecasting. I explored forecasting on real energy data in my Global Energy Consumption Analysis & Forecast notebook on Kaggle, and you can find more practice data in the BloomInData dataset library.

Frequently asked questions about solar KPIs

Sources