M&V tutorial · metering device output
Metering device output: measuring what the battery delivers
When a virtual power plant (VPP) can read each battery or device directly, the measurement gets precise. What the device would have done anyway still has to be estimated. Here is how device metering works and why the rules matter more than the meter.
VPP Economics has no affiliation with, sponsorship from, or endorsement by any utility, aggregator, platform, or device maker named here.
The short answer
- The meter goes on the device, not the home. NAESB calls it metering generator output: the reduction is based on the output of a generator behind the customer’s revenue meter. For batteries, that is inverter or sub-meter data (NAESB).
- A precise meter still needs a baseline. CAISO subtracts what the device normally delivers in the last 10 similar non-event hours; National Grid compares event discharge with non-event days (CAISO; National Grid).
- The rules move the number more than the meter does. In the example below, one battery delivering 5 kW is credited anywhere from 0.9 to 5.0 kW depending on the baseline and export rules.
- It is the default for batteries. Massachusetts and Rhode Island pay for battery performance as average kW discharged, from 15-minute or finer inverter data (National Grid).
Two meters, one home
Where the device meter sits
A home battery behind the utility’s revenue meter, mid-event.
Swipe to see the full chart →
The revenue meter sees only the net: -2.6 kW, an export. It cannot tell whether the home cut back or the battery pushed harder. A sub-meter on the battery removes that guesswork. It measures 5.0 kW directly, to the accuracy of the meter itself. NAESB’s framework lets grid operators set that accuracy; its own example is no worse than 3% of full scale (Federal Register).
The device still needs a baseline
What the battery would have done anyway
The same battery on an event day and on a typical non-event day.
Swipe to see the full chart →
This is where device accuracy meets the baseline problem. Many home batteries already discharge in the evening to cover the home’s own use, especially on time-of-use rates. If the event pays for all 5.0 kW, it pays for 1.5 kW the battery would have delivered anyway. So National Grid credits only the increase over non-event days, and sets performance to zero when the battery discharges less than usual (National Grid). The meter is exact; the counterfactual still has to be estimated.
For thermostats, water heaters and EV chargers, device data reports what the device did: runtime, setpoint or charging power. It still needs a model of what the device would have done without the event, which brings back the same baseline methods as the interval-meter tutorial, applied to the device instead of the home.
Same battery, five rules
Credit for one battery under five rules
A 5 kW discharge, a 2.4 kW home and 1.5 kW of usual evening discharge.
Swipe to see the full chart →
The battery did the same thing in every row. The credit runs from 5.0 kW to 0.9 kW. The biggest swing is exports. CAISO counts device output only up to the home’s own load, so a battery that could push 2.6 kW more onto the grid gets nothing for it. Sunrun and Tesla argue that this undercounts batteries that can serve the grid, not just their own home (CAISO stakeholder materials). Massachusetts takes a middle path: export credit is capped at 150% of the site’s annual peak less the battery’s nameplate (Mass Save).
What device data shows
Enrolled capacity is not delivered capacity
Share of available battery capacity delivered, averaged across events.
Swipe to see the full chart →
Rhode Island’s 2024 season shows the method at scale: about 900 batteries per event, averaging 5.29 kW each (Rhode Island Energy). Missed events count as zero, and a battery that starts late is averaged over the whole event: 0, 8, 8 and 8 kW over four hours is credited 6 kW (MassCEC). That honesty is the point. A deemed value would have assumed every enrolled kilowatt showed up; device data shows about two-thirds did.
Where it is used
ConnectedSolutions (MA, RI, NH, NY)
Batteries paid on average kW discharged from inverter data at 15 minutes or finer; other curtailment at the site still uses a meter baseline (Mass Save).
Metering generator output
Sub-metered device output minus its own 10-in-10 baseline, capped at site load. Can be combined with a load baseline for the rest of the site (CAISO BPM).
Distributed generation
Emergency generation metered at the generator is credited its full output; other generation-based demand response subtracts a baseline (ISO-NE).
Aggregated DER pilot
Fleet telemetry is checked against premise meters: the telemetered response must land within 10% of the 15-minute meter data (ERCOT).
NAESB allows this method, alongside meter before / meter after, for regulation, the fastest grid service (NIST). The two often work together: a fleet’s device telemetry is the data, and a meter before / meter after rule turns it into a reduction.
Where it goes wrong
Paying for normal operation
Without a device baseline, routine self-consumption gets paid as a grid service.
The rest of the home moves
A battery discharges while the home starts an EV charge or pre-cools. The device meter misses it; the revenue meter does not.
Gaps and vendor data
Inverter data comes through the manufacturer’s cloud. Programs require approved vendors and full-season 15-minute data, and treat missing data as zero (National Grid).
What the grid is buying
Whether exports count is a market rule, not a measurement one. It can halve or double the credit.
How it shows up in the VPP Program Designer
The VPP Program Designer lists direct device measurement in Step 5 at about ±3% with no bias. That assumes the program subtracts a device baseline, as National Grid and CAISO do. A program that pays full device output with no baseline would overstate dependable capacity by whatever the devices deliver on ordinary days; in the example above, that is 1.5 of 5.0 kW.
All six ways performance is measured
Baseline from interval meter
Past meter data, sometimes with weather, predicts what the site would have used.
See the tutorial →For short eventsMeter before / meter after
Usage just before an event against usage during it.
See the tutorial →For fleetsStatistical sampling
A metered sample stands in for the whole fleet.
See the tutorial →For batteriesMetering device output
You are here. Measures what the device itself delivers.
This pageFor firm loadsMaximum base load
Pays for staying below a pre-agreed level.
See the tutorial →For residential programsControl group
Compares dispatched homes with homes held back.
See the tutorial →Method and caveats
A tutorial for program designers, utilities, aggregators and other readers. It is not a substitute for a tariff, manual or program rules. The worked example uses illustrative numbers, and the five rules are simplified for comparison. Checked October 23, 2026.
Rule labels describe how each program handles baselines and exports as published; programs update annually, and some allow either device or meter data.