The highest 15-minute power draw in the period, with the time it occurred. On demand-billed accounts this peak drives the demand charge — often 30–40% of the bill — so flattening it with staggered start-ups and pre-cool-and-coast is the most direct way to cut cost.
The highest 15-minute average power draw the site reached during the period, together with the time of day it occurred. On accounts billed on demand or ratchet rates, this single 15-minute interval sets the demand charge for the whole period — a line item that often runs 30-40% of the bill. The metric reads the main meter and is reconciled to the utility bill; in V1 it is not weather- or occupancy-normalized, so an unexpected peak is flagged for a walkthrough rather than auto-attributed to a cause.
The calculation depends on the data and the tariff for each client and site. The base method takes the 15-minute interval kW from the main meter and reports the maximum interval in the period, with its timestamp. The interval length must match the utility's demand window — most tariffs use a 15-minute sliding or fixed window, but some use 30-minute or shorter ratchet windows, so the window is confirmed against the bill before the metric is activated.
The first step is always to confirm the account is actually demand-billed by reading the bill. On energy-only rates there is no demand charge, so the peak is informational context, not a cost lever, and no action is recommended. On demand or ratchet rates the peak is the billed quantity; on ratchet rates a single peak can set a floor that follows the account for up to twelve billing periods.
With the meter alone, Energy Intelligence reads the draw and recommends the shed — it does not command equipment. Where the customer also runs Keedian on HVAC, Lighting, or Refrigeration, the demand action (staggered start-up, pre-cool-and-coast, deferring a stageable load) is executed through those connected products. The exact threshold, window, and trigger logic vary by client tariff and the connected products in place.
| Range | Classification | Interpretation |
|---|---|---|
| Peak well below billed/ratchet level | Headroom | Demand sits comfortably under the billed peak — no action; informational context for the demand profile. |
| Peak within X% of the trailing-12-month peak | Approaching — monitor | A new peak is close to setting or matching the billed demand, especially inside the utility on-peak window; candidate for a managed shed. |
| New peak at or above the prior billed peak | Peak set — review | A higher demand charge is locked for the period; on ratchet rates a floor may now follow the account for up to twelve months. |
| Coincident peak inside the utility on-peak window | Highest cost exposure | Peak coincides with the tariff's most expensive window or a demand-ratchet trigger — the most valuable interval to flatten. |
Reference behavior only, not fixed thresholds. The demand window, the billed/ratchet level, and the on-peak hours come from the client's tariff and bill — calibrate per client before activating any demand alert.
There is no portfolio compliance percentage for peak demand the way there is for an HVAC compliance metric. The target is set per account against the tariff — typically a managed demand ceiling agreed with the client, with the trailing-12-month billed peak as the reference point. Confirm the account is demand-billed and agree the ceiling before reporting against it.
The table below shows how moving Peak demand impacts each customer value driver the product is designed to improve — the metric page explains the mechanism; the product pages express the magnitude.
| Value driver | Impact strength | How Peak demand moves this lever |
|---|---|---|
| Energy savings | Direct, primary | On a demand-billed account the 15-minute peak is the demand charge — often 30-40% of the bill. Flattening it with staggered start-ups, deferred load stages, and pre-cool-and-coast is the most direct lever on this metric. Anchored to the 7-Eleven energy outcome ($5M); the per-site demand saving is expected (pending validation) and is quantified directly from the client's own bill, not an invented benchmark. |
| Sustainability | Indirect | A coincident peak inside the utility's on-peak window is also the dirtiest, most carbon-intensive grid hour. Shifting the same load off that interval cuts the demand charge and the CO2e tied to it — the cost lever and the emissions lever move together, where the tariff carries time-of-use or coincident-demand pricing. |
The table below summarizes the alarms that fire directly from Peak demand. Each row links to the full operational detail (trigger, preconditions, action plan, human role, escalation, prevention) in the SOPs catalog.
| Alarm | Description | Severity | Tier | AI executes? | Value drivers | SOP |
|---|---|---|---|---|---|---|
| Demand peak approaching threshold | Rolling demand approaches the billed/ratchet peak or a demand-charge threshold, especially inside the utility on-peak window. | Medium | Essential | Hybrid | Energy savings | Open SOP → |
More alarms in development (single-metric): managed demand-limiting (active shed driven through connected products), ratchet-floor watch, and time-of-use cost optimization. Composite alarms in development combine the demand peak with per-circuit load and the tariff to attribute and pre-empt the peak at the asset level.
The action plan for each alarm lives on its own SOP page in the SOPs catalog — with the diagnostic steps, human role, value drivers, escalation, and prevention specific to that alarm. The list below maps each alarm to its SOP.
Every demand action depends on the tariff. Read the bill and confirm the account is on a demand or ratchet rate before treating any peak as a cost event. On energy-only rates the peak is informational context, not a charge — do not recommend a shed or report a demand exposure. Document the tariff terms (demand window, on-peak hours, ratchet period) per account so the alarm threshold is set correctly.
Energy Intelligence reads the meter — it sees the peak, it does not switch loads. Where the customer also runs Keedian on HVAC, Lighting, or Refrigeration, route the shed through those products (staggered start-up, pre-cool-and-coast, deferred load stage). On a meter-only site there is no connected product to act through, so the output is a recommendation to the operator, not a remote command. Confirm which connected products are in place per site so the action is the right one.
If the metered peak does not reconcile with the demand quantity on the bill, the difference can be a CT-ratio or meter-multiplier error, a misread demand window, or a billing error worth recovering. Escalate internally to the technical team before reporting a demand exposure. Do not present a demand-charge figure to the client until the meter and the bill agree, and document the period affected.
Per-alarm escalation criteria live in the Escalation block of each SOP in the SOPs catalog. The patterns below are metric-level — read from the portfolio view, not from any single alarm firing.