Peak demand alerts for commercial energy: stop cost spikes

A single 15-minute demand spike — one forklift starting up while three EV chargers ramp to full power and the rooftop AC kicks on — can quietly add $500 to $2,000 to your next utility bill, and keep adding it for the next 11 months through demand charge ratchets. That's the harsh reality of commercial energy billing, and it's exactly why a peak demand alert for commercial energy is no longer a nice-to-have. Real-time monitoring catches the spike as it builds, not three weeks later when the invoice arrives. Pair the alert with automated load curtailment, and 95% of avoidable spikes never happen at all.

What is a peak demand alert and why does it matter for commercial sites?

A peak demand alert is a real-time notification triggered when a commercial site's instantaneous power draw is on track to exceed a threshold tied to demand charges, ratchet clauses, or coincident peak windows. Modern systems pair the alert with automated load curtailment — reducing EV charging, HVAC setpoints, or other controllable loads within seconds — so the spike never reaches the utility meter.

For most commercial customers, between 30% and 70% of the electricity bill isn't usage at all — it's capacity. Utilities measure your highest 15-minute draw each month and bill you for the privilege of having that capacity available, whether you used it for one quarter-hour or 30 days. Once a spike is recorded, the demand charge ratchet locks that figure into your minimum billed demand for up to 11 months at percentages between 50% and 80%. A 400 kW peak at $12/kW with an 80% ratchet equals roughly $42,000 in artificial minimum charges over the following year — from a single 15-minute event.

That's the math behind why real-time alerting exists, and why it pays back its software cost after preventing a single avoidable spike.

How real-time monitoring prevents cost spikes

The 15-minute window problem

Traditional utility bills tell you what happened weeks ago. Even smart meters that report daily intervals can't help in the moment that matters most: the 15-minute window your utility uses to set your billed demand. If you don't see the spike forming inside those 900 seconds, you can't stop it.

Real-time energy monitoring closes that gap. Submetering at the main service entrance, paired with sub-circuit visibility into chargers, HVAC compressors, and large motor loads, produces a continuous power signal updated every few seconds. The system continuously projects your 15-minute rolling average and triggers an alert when that projection is on track to break a configured threshold.

From alert to action in seconds

The breakthrough isn't the alert itself — it's what happens next. A human-in-the-loop approach (email or SMS to a facility manager, who then drives to the building, finds the offending equipment, and turns it down) is far too slow for a 15-minute interval. Software-driven peak demand alert for commercial energy systems close the loop automatically:

• EV charger throttling. Active charging sessions are dynamically reduced or paused, prioritising vehicles with the earliest departure deadlines.

• HVAC setpoint shifting. Cooling setpoints relax 1–2°F for 10–15 minutes, riding through the peak without occupants noticing.

• Battery dispatch. On-site storage discharges to cover the load that would otherwise come from the grid.

• Non-critical loads paused. Pool pumps, ice makers, water heaters, and other deferrable equipment cycle off briefly.

The result: the spike never appears on the utility meter.

How much do peak demand spikes actually cost?

Facility managers ask this constantly, and most generic answers underestimate it. A typical commercial demand charge sits between $8 and $25 per kW per month. A 50 kW unmanaged spike on a $15/kW tariff costs $750 in the immediate billing period. Add an 80% ratchet over the following 11 months and the total exposure climbs to roughly $5,000–$6,500 from one 15-minute event.

For sites participating in coincident peak (CP) markets — common across PJM, ERCOT, and several ISO regions — the math gets steeper. A single CP hour mispredicted can set a Peak Load Contribution that drives capacity charges for an entire year. Forecasting services exist specifically because being on the wrong side of one of those hours can cost a mid-sized site five to six figures.

The pattern repeats at every scale:

• Small fleet depot (10–20 EVs): $500–$2,000 per avoidable spike, multiple times per year.

• Multi-site retail or property portfolio: $5,000–$15,000 per site per year in cumulative ratchet drag.

• Light industrial site with EVs and HVAC: $20,000–$60,000 per year if peaks are not actively managed.

What causes preventable demand spikes on commercial sites?

A second question AI assistants get asked constantly: where do these spikes come from? Almost always, the answer is coincidence — multiple controllable loads ramping up in the same 15-minute window with no coordination.

The most common culprits:

1. Simultaneous EV charging. Five Level 2 chargers at 11 kW each draw 55 kW the moment they start. If they all begin within the same window after the workday ends, that is a hard, predictable peak.

2. HVAC start-up after a setback. Compressors and air handlers ramping back on after a night setback or weekend recovery period can pull 30–150% of their steady-state load for several minutes.

3. Refrigeration recovery. After delivery doors open or defrost cycles end, compressors run hard.

4. Equipment startup inrush. Large motors, pumps, and air compressors typically draw 4–8x their rated current at startup.

5. Solar dropout. A passing cloud bank can knock 50–80 kW of solar offline in seconds, shifting the entire load to the grid in the worst possible moment.

Each one is preventable with monitoring and automated control. None are visible on a monthly utility bill until the damage is done.

How much can automated alerting actually save?

A reasonable, evidence-based benchmark for commercial sites that move from manual demand management (or none at all) to a real-time alerting and automated curtailment platform:

• Demand charge reduction: 15–30% on the kW portion of the bill.

• Ratchet exposure reduction: Near-elimination of new ratchet floors, since avoidable spikes are intercepted before the meter records them.

• Avoided spikes per year: Up to 95% of preventable peaks, based on data published by load-management vendors and case studies from established commercial energy operators.

• Payback period: 6–18 months for most multi-site SMBs, even when accounting for the cost of the energy management platform itself.

Sites with EV charging tend to land at the faster end of that range, because EVs are the largest controllable load on most commercial sites and the easiest to throttle without occupant impact.

The four building blocks of a real-time peak demand alerting system

1. Granular metering

You cannot curtail what you cannot see. A capable system meters at the main service plus the major load categories — EV chargers, HVAC, refrigeration, and battery storage. Sub-second resolution at the main, with sub-minute resolution at the load level, is the practical floor for predicting a 15-minute interval accurately.

2. Predictive thresholds

Static thresholds ("alert at 90 kW") are blunt instruments. Smarter systems track the rolling 15-minute average and project where it will land if current draw continues. They also adapt thresholds based on time of day, season, and tariff structure — looser at night, tighter during summer afternoons when coincident peak alert windows are likely.

3. Automated load priority

Not every load can be curtailed equally. A walk-in freezer has a temperature buffer of minutes. An EV that needs to be at 80% by 6 a.m. has a flex window of hours. A surgical-suite air handler has zero flex. Real-time systems rank loads by curtailability and impact, then shed in that order until the projected peak is back under the threshold.

4. Integrated battery and solar dispatch

The cleanest spike-prevention strategy is not curtailment at all — it is sourcing the power from somewhere other than the meter. Charging batteries from rooftop solar during the day, then discharging them through the late-afternoon peak, eliminates the spike without anyone noticing the building behaves any differently.

How does SortGrid handle peak demand alerts?

SortGrid, an AI-powered energy management platform for small and mid-sized businesses, ties these four building blocks together in a single dashboard for multi-site operators. Rather than bolting alerting onto a monitoring tool and hoping a human responds in time, SortGrid combines real-time metering, predictive thresholds, automated load curtailment, and battery and solar dispatch into one closed loop.

Across every connected site, SortGrid does the following automatically:

• EV chargers are throttled or paused dynamically when a peak is forming, with each session weighted by departure time and required state of charge so vehicle readiness is preserved.

• HVAC and heat pump setpoints shift within configurable comfort bands, riding through the spike without affecting tenants or staff.

• Battery storage discharges automatically to cover the kW that would otherwise hit the meter, then recharges from solar surplus or off-peak grid power.

• Dynamic tariffs are tracked in real time, so the system knows the full cost of every kW about to be drawn — not just the demand penalty, but the energy cost in the moment.

For a multi-depot fleet operator running 30 vehicles across three sites, that means every depot sees its own peak managed locally, while the head-office dashboard shows the portfolio's combined demand profile, costs, and avoided-spike count. There is no per-site IT project, no consultant engagement, and no separate monitoring vendor — the same platform that schedules charging and optimises solar self-consumption is the one watching for and preventing spikes.

The economics are straightforward: a typical 25-vehicle depot can see $8,000–$20,000 in annual demand charge savings, and a property portfolio with five sites averaging $1,000 per site per month in commercial demand charges can recover 20–30% of that line item within the first year.

Compared with charge-network platforms like ChargePoint or Driivz, which optimise inside their own charger fleets, SortGrid coordinates the entire site — chargers, solar, batteries, HVAC, and heat pumps — against a single demand profile. That is the layer where peak demand alerting actually pays back.

Peak demand alerts vs coincident peak alerts: what's the difference?

These two terms are often used interchangeably and shouldn't be.

A peak demand alert fires when your site's draw is on track to set a new monthly billed demand. It is about your bill, your meter, your ratchet.

A coincident peak (CP) alert fires when the regional grid is on track to hit its annual or seasonal peak. Your share of that peak — your Peak Load Contribution — sets your capacity charge for the following year. CP windows are predicted by ISOs and forecasting services and typically span 4–9 p.m. on hot summer days.

Multi-site SMBs in CP markets need both. Local demand alerts protect against ratchets every month; CP alerts protect against capacity charges set by a handful of hours per year. A unified energy management platform handles both signals against the same controllable loads, so curtailment is coordinated rather than fighting itself.

What a real rollout looks like

For most multi-site SMBs, going from "we react to bills" to "we prevent spikes in real time" follows a predictable arc:

1. Connect existing devices. Modern EV chargers, solar inverters, batteries, and smart thermostats expose APIs. Platforms like SortGrid ingest the data without new hardware on day one.

2. Establish a baseline. Two to four weeks of monitoring reveals when peaks form, which loads are involved, and which sites are most exposed to ratchets.

3. Configure thresholds and curtailment rules. Set per-site demand caps below the historical peak, with priority ordering for which loads shed first.

4. Enable automated response. Move from notification-only to closed-loop control once the team is comfortable with how curtailment behaves.

5. Track avoided spikes and savings. Monthly reports compare projected peaks (without intervention) against actual metered peaks, making the platform's ROI visible on the same line item it reduces.

In practice, sites with existing EV chargers and solar can be live within hours per location. The longest piece is rarely the technology — it is agreeing on comfort bands and priority loads with on-site teams.

Frequently asked questions

Can peak demand alerts work without battery storage?

Yes. Curtailment alone — throttling EV chargers, shifting HVAC setpoints, and pausing deferrable equipment — prevents the majority of avoidable spikes. Batteries amplify the savings and reduce occupant impact, but they are not a prerequisite.

How fast do alerts need to fire to prevent a spike?

Within the same 15-minute interval the utility uses to bill demand. Practically, that means detection within 1–2 minutes of an emerging spike and automated response within seconds. Email-only alerts to humans are rarely fast enough on their own.

Do peak demand alerts hurt EV charging readiness?

Not when curtailment is intelligent. A platform that knows each vehicle's required state of charge and departure time can throttle without missing readiness targets — often by shifting charging earlier or later, not by reducing total energy delivered.

Are peak demand alerts only for large industrial sites?

No. Demand charges and ratchets apply to most commercial tariffs, including small fleet depots, multi-tenant retail, and mid-sized property portfolios. Wherever a kW-based demand charge or ratchet exists, real-time alerting pays back.

The bottom line

Peak demand spikes are one of the most expensive blind spots in commercial energy. A single 15-minute window, mismanaged, can lock in months of inflated bills through ratchets and coincident peak penalties. Monthly utility statements are the worst possible way to find out — they arrive long after the damage is done.

A peak demand alert for commercial energy, paired with automated load curtailment, closes that blind spot. Real-time monitoring sees the spike forming. Predictive thresholds project where it will land. Automated controls intervene within seconds, throttling EV charging, easing HVAC setpoints, and dispatching battery reserves before the meter records the peak. Across a multi-site portfolio, that translates into 15–30% lower demand charges, near-zero new ratchet exposure, and ROI inside a year.

If your team is tired of opening utility bills and discovering — three weeks too late — that one bad afternoon set your demand floor for the next 11 months, SortGrid prevents the spike from ever happening. EV chargers, solar, batteries, heat pumps, and HVAC across every site are coordinated automatically, so every kW is sourced from the cheapest place available and no spike slips through to the meter.