Solar self-consumption: how to calculate it from meter data

Solar self-consumption is the share of the energy a PV system generates that the site uses itself instead of exporting to the grid. Calculate it from two meter readings for the same period: self-consumption rate = (PV generation − grid export) ÷ PV generation. Its partner figure, the self-sufficiency or autarky rate, divides the same solar energy used on site by the site's total demand instead: self-sufficiency rate = (PV generation − grid export) ÷ site demand.

Both figures are only as good as the measurements behind them. An inverter reports generation, the grid connection meter reports import and export, and neither alone gives the site's demand. Mark the measurement points on the single-line diagram first, then name each data series after the flow it measures.

How to calculate solar self-consumption

These equations assume a common AC boundary, no other generator and no storage. Distribution losses inside that boundary become part of the inferred site demand. Meter tolerances and clock differences can leave a small residual when you compare this balance with separately measured loads.

If generation is zero, self-consumption is undefined for that period. If demand is zero, self-sufficiency is undefined. Display an explanation rather than dividing by zero or reporting an invented percentage.

Worked example: one hour

During one hour, the solar system produces 4 kWh. The site imports 1 kWh and exports 2 kWh. Its demand is 4 + 1 − 2 = 3 kWh, of which 2 kWh came directly from solar. Import and export can both occur during the hour as clouds and loads change.

Self-consumption vs self-sufficiency over a day

The two rates answer different questions. Self-consumption tells you whether the array is sized and timed to the load: a low value means much of the solar is exported. Self-sufficiency tells you how far solar reduces what the site buys: a low value means the site still imports most of its energy. A large array on a small load has low self-consumption and high self-sufficiency; a small array on a large load has the reverse.

This day is measured in four six-hour blocks at the grid meter and the inverter, with no battery:

Self-consumption for the day is 32 ÷ 46 = 69.6% and self-sufficiency is 32 ÷ 62 = 51.6%. The daily totals alone could suggest that 46 kWh of solar covered 74% of a 62 kWh demand. It covered 51.6%, because 14 kWh arrived in the afternoon when the load could not use it, and the night and evening demand had no solar to use.

Interval length matters when export is not metered. If you estimate solar used on site as the smaller of generation and demand in each block, the morning block gives min(18, 20) = 18 kWh instead of the measured 16 kWh, because within those six hours the site exported 2 kWh while also importing 4 kWh. The longer the interval, the more this estimate overstates self-consumption. With a separately metered export register the result does not depend on interval length, which is why the export measurement comes first.

The calculator sums aligned interval results in the same way. Its worked inputs are illustrative and exclude batteries.

Where to measure: generation, grid connection and battery

Document whether each value is measured or calculated. A hybrid inverter may combine PV and battery output at one AC terminal. That total is not a separate PV generation measurement. Use its documented internal channels if you need the split, and record whether those channels are DC or AC.

Sign conventions

A meter or inverter reports power with a sign, and manufacturers do not agree on which direction is positive. A grid meter set up in the consumer convention shows import as positive and export as negative. An inverter's own meter may show generation as positive. A battery may report charge as positive or negative depending on the vendor and the register.

Store each flow as a positive directional total (import, export, generation, charge, discharge) and record the source register and its sign convention beside it. Commission every sign with a known condition: an importing load with the PV isolated, then generation on a sunny period, then a battery charge and discharge. A reversed current sensor or a wrong sign setting turns export into import, and the self-consumption rate becomes meaningless. The negative power and meter reconciliation guide works through the checks.

Preserve import and export separately

A signed net-energy total discards useful information. Importing 2 kWh and exporting 2 kWh produces a net of zero, but both flows still matter for billing and solar use.

Prefer the meter's separate cumulative import and export registers. Calculate interval energy from counter differences, handling resets and rollovers explicitly. If you integrate power instead, establish its sign convention, sampling interval and treatment of gaps. Short periods of export between samples can otherwise disappear.

Keep timestamps in one time base, retain the local billing time zone and check daylight-saving transitions. Do not subtract a daily inverter counter that resets at midnight from a lifetime meter counter. Commission the signs with a known importing load and, when conditions permit, a known exporting state.

Export limitation and zero export

Where the network operator limits export, an export limitation scheme measures power at the connection point and reduces inverter output so that export stays below the agreed value. A zero-export site sets that value to zero, or close to it, so the inverters follow the site load.

That changes what the meters show. Export stays near the limit, self-consumption rises towards 100%, and the generation meter records only the energy the inverters were allowed to produce. The energy the limiter withheld is not measured anywhere. If you need it, estimate the available generation separately, for example from irradiance and a model, and keep it in its own series labelled as an estimate. Do not add it to measured generation.

Check the limiter's own measurement against the grid meter at commissioning: a sensor on the wrong side of a load or a slow control loop lets short export peaks through. Record the agreed limit, the measurement point and the response the network operator requires. In Germany, the VDE-AR-N 4105 guide covers the low-voltage connection rules; the zero-export solar monitoring application lays out the meters and signals.

Add battery flows explicitly

With a separately metered AC-coupled battery, the corresponding balance is:

Site demand = PV generation + grid import + battery discharge − grid export − battery charge

All five terms must refer to the same interval and AC boundary. Do not mix battery DC charge energy with inverter AC discharge energy in that equation.

Solar sent into a battery can later serve loads, be exported, or be lost during storage and conversion. If the battery can charge from the grid, discharge alone cannot identify the original source. State the attribution method before publishing a solar self-consumption percentage. Comparing stored energy at the beginning and end also matters when assessing a short reporting period.

Worked example: a battery ledger over three intervals

Three one-hour intervals, all AC energy at the same boundary, each flow a positive directional total:

The ledger balances in every interval: in A, the 4 kWh of solar went 1 kWh to export, 2 kWh to the battery and 1 kWh to the load. Site demand of 5 kWh is established by measurement, even though no load was metered directly. The demand is inferred, so label it that way.

What the ledger cannot establish is how much of the 3 kWh discharged in C was solar. The battery took in 2 kWh from PV in A and 2 kWh from the grid in B. A proportional rule would call 1.5 kWh of the discharge solar. A rule that assumes solar leaves first would call 2 kWh solar; one that assumes grid energy leaves first would call 1 kWh solar. If the battery already held energy before interval A, none of these rules is right without knowing what that energy was.

The 1 kWh difference between charge (4 kWh) and discharge (3 kWh) is not a measured loss either. It may still be in the battery, or it may have been lost in conversion. Only the battery's stored energy at the start and end of the period, read from the battery management system, separates the two.

The useful results are therefore:

• site demand: 5 kWh, measured by balance;
• solar used directly by loads: 1 kWh, in interval A;
• solar sent to storage: 2 kWh;
• self-consumption, if the stated definition counts stored solar as consumed: (4 − 1) ÷ 4 = 75%;
• self-sufficiency: not determined until the attribution rule, the opening and closing stored energy and the treatment of losses are stated.

Publish a battery-inclusive percentage only with those three things written beside it. The energy monitoring project worksheet has room to record them per site.

For control and commissioning, read the roles of BMS, EMS and PCS and BESS command verification. A successful command response does not prove the requested power was delivered.

Before estimating annual yield, check that the proposed module strings fit the inverter's voltage and current limits. Use the module's temperature coefficients and the site's design temperatures; this screening does not replace the manufacturer's design rules.

Keep energy estimates separate from measured performance

A PV model such as PVGIS estimates generation using location and system assumptions. Keep the model version, weather dataset, array capacity, orientation and loss assumptions with the result. Compare like periods and boundaries when checking the installation against that estimate. The Solar PV generation estimate tool requests a PVGIS estimate for a location and array.

Annual modelled generation does not establish self-consumption or battery size. Those depend on when generation and demand coincide. Use aligned interval data for that comparison. Export limits and curtailment also mean the energy actually generated can differ from the available solar resource.

Value imported and exported energy separately

For a simple flat-rate estimate without storage, value solar used directly on site at the avoided import unit price. Value exported energy at the export unit price. Keep both prices editable.

For example, 2 kWh used directly at an illustrative €0.30/kWh avoids €0.60 of energy charges. Exporting another 2 kWh at an illustrative €0.10/kWh earns €0.20. The combined value is €0.80, not 4 × €0.30. These are arithmetic examples, not offered tariffs.

Time-of-use prices require matching each flow to its price period. Standing charges, demand charges and battery costs need their own treatment. Start with measured energy and the actual tariff rather than promising savings from the array's nameplate power. Shared schemes change the accounting again: in an Italian renewable energy community the incentive follows energy shared within the community in each hour, and Ireland's SRESS pays for exported energy under its own metering rules.

Compare those two energy values in a fixed-tariff cash-flow model. Enter installed cost, on-site use, export payment and degradation explicitly; the example rates are editable assumptions, not current tariff quotations.

Measure a solar site with EpiSensor

The ZEM electricity monitor reports voltage, current, active power, power factor and energy for each phase, so one ZEM on the inverter output and one on the site incomer give generation and the grid flow on the same Gateway and the same time base. Inverters and grid meters with a documented Modbus register map can be read directly through the Gateway instead. Edge stores the readings on site, so the interval balances in this guide can be calculated from one record. The solar PV monitoring application maps the measurement points for a commercial site.