M&V tutorial · maximum base load
Maximum base load: paying a site to drop to a firm level
Most wholesale capacity demand response, including commercial and industrial sites in virtual power plants (VPPs), is measured without a baseline. The site promises a level, and the grid checks that it got there. Here is how “drop to” works, why it dominates capacity markets, and why PJM’s market monitor disputes the results.
VPP Economics has no affiliation with, sponsorship from, or endorsement by any utility, aggregator, platform, or device maker named here.
The short answer
- Drop to, not drop by. The resource commits to cut its load to a set level, whatever it would otherwise have used. NAESB calls it maximum base load; PJM and MISO call the level a firm service level (NAESB; ERCOT).
- It is the backbone of wholesale capacity demand response. Firm service level covered 99.9% of PJM’s committed load management in 2023/24 (PJM).
- No baseline, so no baseline error, and no baseline to estimate what the site actually did. PJM credits the reference peak minus metered load. A store that is closed on Sundays is an overperformer in a Sunday event (Monitoring Analytics).
- The gap is live. For PJM’s winter 2026 events, PJM scored demand response at about 100%; its market monitor, measuring against a baseline, at about 50% (Monitoring Analytics).
From reference peak to firm level
Two numbers set before the season
A worked example for one commercial site in a PJM-style capacity program.
Swipe to see the full chart →
Event day
The site drops below its firm service level within the lead time and holds it.
Swipe to see the full chart →
Compare this with the other methods. There is no lookback window, no weather, no adjustment. The site sold 3.5 MW because its reference peak is 5.0 MW and it can always get down to 1.5 MW. On event day it metered 1.4 MW, so it is credited 3.6 MW: slightly more than it sold. A test simply checks that the site can reach the firm level.
Drop to versus drop by
What the credit measures
The same site meters 1.4 MW in both events. Only what it would have used differs.
Swipe to see the full chart →
On a hot peak day the two measures agree closely: 3.6 MW credited against 3.8 MW actually cut. On a mild or quiet day they split: still 3.6 MW credited, but only 0.6 MW of real reduction. That is the method’s core trade-off.
It measures what the grid planned for
Capacity is bought against peak load. A site that is below its firm level is not adding to the peak, whatever the reason. MISO stakeholders argue it delivers the same system benefit regardless of why load is reduced (MISO stakeholder feedback).
It pays for load that was already off
A site that is closed, slow or idle gets full credit without acting, and negative performers are zeroed out. The market monitor prefers measuring against a baseline (Monitoring Analytics).
Where the debate stands
Two ways to score the same events
PJM pre-emergency and emergency load management, winter 2026.
Swipe to see the full chart →
The market monitor’s example makes the mechanics plain. Two sites each sold 15 MW. One cuts 5 MW; the other runs 10 MW above its reference peak. PJM’s metric zeroes the second site and scores the pair at +17%; a baseline metric nets them to −17% (Monitoring Analytics). The monitor also reports that many resources could not perform in extreme cold, exactly when they are called. FERC approved a new PJM penalty for underperformance outside emergency intervals in June 2026, at half the emergency penalty rate (Monitoring Analytics).
Who uses it
Firm service level
Capacity sold = peak load contribution − firm service level. Compliance = peak load contribution − metered load, floored at zero, no export credit (PJM Manual 18).
Firm service level for load modifying resources
Must reach the firm level within the lesser of 1 MW or 10%; reduction below it is not counted. Only load modifying resources may use it (MISO).
Average coincident load
The reference is the average of a site’s 20 highest loads in the top 40 zone peak hours of the prior like season (NYISO ICAP Manual). A proposal to measure by a baseline instead stalled with stakeholders (industry summary).
Emergency Response Service, “drop to”
Assigned when an accurate default baseline cannot be built. Availability = average 15-minute load in the committed period − declared maximum base load (ERCOT).
NAESB does not allow it for reserves, where the grid needs a measured change, not a level (NAESB). It also fits poorly with home fleets: a household has no meaningful firm level, so residential virtual power plants use device data, sampling or baselines instead. The device tutorial and sampling tutorial cover those.
Where it fits and where it does not
Large, steady commercial and industrial loads
Plants and buildings that can always get down to a known level, such as process lines or chillers that can be shut off.
Loads with no good baseline
Irregular schedules defeat typical-day baselines. ERCOT assigns “drop to” for exactly this reason (ERCOT).
Loads that are often already low
Weekend, seasonal or idle sites get credit without acting.
Reserves and fast products
They need a measured response, so NAESB points to meter before / meter after or device metering (NAESB).
How it shows up in the VPP Program Designer
The VPP Program Designer does not offer maximum base load in Step 5. It is built around residential and mixed fleets, where a firm level per site is rarely meaningful. A C&I fleet settled this way would see credited capacity at or above the true reduction, depending on how often sites are already below their firm level when called.
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
Measures what the device itself delivers.
See the tutorial →For firm loadsMaximum base load
You are here. Pays for staying below a pre-agreed level.
This pageFor 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 settlement rules. The worked example uses illustrative numbers. Loss factors and winter adjustments are omitted. Checked October 23, 2026.
Performance figures for winter 2026 are the market monitor’s published summary; PJM’s own reporting may differ in scope and rounding.