When a fleet electrification project goes over budget in year one, the most common culprit isn't the vehicles, the chargers, or even the electrical service upgrade. It's a line item that most TCO models either bury or ignore entirely: the demand charge.
Demand charges are not exotic or obscure. They appear on virtually every commercial utility bill in the United States. But their mechanics are counterintuitive enough that fleet operators who are sophisticated about fuel costs, maintenance intervals, and depreciation schedules consistently get blindsided by them at the first billing cycle after EVSE installation.
What a Demand Charge Actually Is
A demand charge is a fee assessed on the highest 15-minute average power draw recorded during a billing period. It is billed in dollars per kilowatt, and it applies to peak demand — not to total energy consumed. This is the critical distinction that catches fleet operators off guard.
Under a typical commercial rate structure in the Pacific Northwest, demand charges might run between $8 and $22 per kW per month depending on the rate schedule. Under Portland General Electric's Schedule 32 (Large General Service), the demand component can constitute 35–50% of total monthly charges for a facility with a high peak-to-average power ratio. When you add a depot's worth of Level 2 EVSE — say, 30 stations at 7.2 kW each — and they all start charging simultaneously when drivers return to the yard at 4:00 PM, you've created a 216 kW demand spike. At $15/kW, that's $3,240 in demand charges from a single concurrent charging event. Monthly.
The situation gets worse under DC fast charging deployments. A depot with ten 50 kW DCFC units that charge concurrently adds 500 kW of demand. Even if that peak occurs exactly once in the billing period and lasts only 15 minutes, the charge applies to the entire month.
The Ratchet Clause Problem
Standard demand charge mechanics are painful enough. Ratchet clauses make them structurally hazardous for newly electrifying fleets.
A demand ratchet provision — found in many commercial rate schedules including Pacific Power's Schedule 47 in Oregon — sets a minimum billable demand based on a percentage (commonly 50–75%) of the highest peak recorded during the previous 11–12 months. In practice, this means a single uncontrolled charging event in January can set a demand floor that elevates your minimum bill through December, regardless of how well you manage load in subsequent months.
Fleet managers who pilot EVSE at one depot, observe the demand charge, and then attempt to address it through scheduling improvements often discover that the ratchet clause is eating their savings. The ratchet was set by the unmanaged pilot period. The only path to a clean reset is waiting out the ratchet window — typically 12 months — while maintaining disciplined demand management throughout.
This is not a reason to avoid electrification. It is a reason to get demand management right from day one, before the first charger goes live.
Structural Causes of Fleet Demand Spikes
Understanding demand charges requires understanding why EV fleet charging creates such adverse demand profiles compared to the facility baseline. Four structural factors compound the problem:
- Correlated return windows. Fleet vehicles — delivery vans, transit buses, utility trucks — return to the depot within a narrow window driven by route schedules and shift end times. Unlike employee vehicles that trickle into a workplace parking lot over 90 minutes, commercial fleets often dock within 20–30 minutes of each other. Every charger activates nearly simultaneously.
- Coincident utility peak hours. Most fleets return between 3:00 PM and 6:00 PM — precisely the period when utility TOU on-peak pricing is active and when system demand charges are highest under coincident peak billing structures. You're hitting peak demand at exactly the wrong time on the utility's clock.
- Charging speed requirements. Fleets with overnight return windows have time on their side for slower L2 charging. But fleets with mid-day layovers or split-shift operations may need DCFC to restore range within 90 minutes, and DCFC power levels create demand spikes that L2 simply doesn't.
- Large facility baseline demand. Interestingly, fleet operators with large warehouse or industrial facility loads sometimes find that their demand charge impact is lower in percentage terms because the facility baseline is already high. A small logistics company operating from a modest depot facility with a 50 kW baseline load will see a proportionally larger impact from EV charging than a large distribution center with a 2 MW baseline.
Demand Management Strategies That Work
Demand charge reduction for EV fleets is fundamentally a scheduling problem. The goal is to spread charging load across time to flatten the demand curve without violating the hard constraint that vehicles must reach their target state-of-charge (SOC) before their next departure window.
Several approaches are available, with meaningfully different cost-benefit profiles:
Sequential Charging Dispatch
The simplest intervention is staggered charging start times. Rather than activating all chargers when vehicles return, a scheduling system activates them sequentially — starting with vehicles that have the largest energy deficit, the earliest departure time, or both. Each charger activates only when the previous vehicle has reached a predefined SOC threshold and its charging rate has tapered.
Consider a 20-vehicle fleet returning between 3:30 PM and 4:00 PM with departures at 6:00 AM. If each vehicle requires an average of 40 kWh and charges at 7.2 kW (standard L2), each session takes approximately 5.5 hours. Staggering activations at 10-minute intervals across 20 vehicles spreads the charging window from 3:30 PM to roughly 7:20 PM, dramatically flattening the load curve while still completing all sessions well before 6:00 AM.
This works for homogeneous fleets with predictable return patterns. It becomes more complex as fleet composition and operational variability increase.
Load Cap with Priority Queuing
A more adaptive approach sets a hard facility-level demand cap — say, 150 kW for the charging system — and manages which vehicles charge at any given moment based on priority logic. Priority rules might weight departure time (vehicles leaving earliest get priority), energy deficit (vehicles with lowest SOC charge first), or a combination of both.
This approach requires EVSE hardware capable of dynamic load management — most modern stations running OCPP 1.6 or OCPP 2.0 support smart charging profiles through the SetChargingProfile command, which allows a central system to set per-connector current limits in real time. The practical question is whether your EVSE vendor's implementation actually honors these profiles reliably, which varies considerably across manufacturers.
Tariff-Aware Scheduling
Load cap management alone flattens the demand curve but doesn't necessarily time it to minimize cost. The most cost-effective approach combines demand cap management with TOU-aware scheduling: hold load below the demand cap target, and concentrate charging during the cheapest rate periods — typically the overnight off-peak window from 10 PM to 6 AM under most PNW utility schedules.
This requires live tariff data. Static schedules encoded once and never updated will drift out of alignment with seasonal TOU rate changes, holiday schedules, and special rate event windows. See our piece on building on live utility tariff feeds for a technical breakdown of what data is actually available from utility APIs.
What Demand Charge Management Won't Fix
It's worth being direct about the limits of software-based demand management. Scheduling optimization can reduce demand charges substantially — in well-controlled scenarios, demand peaks can be reduced by 50–70% compared to unmanaged charging. But it cannot overcome certain physical constraints.
If your electrical service capacity is genuinely insufficient for the fleet — if the transformer serving your facility cannot support the overnight charging load you need even when spread optimally across the off-peak window — scheduling software doesn't solve that problem. You need a service upgrade, additional transformer capacity, or on-site battery energy storage to buffer the grid draw. We're not saying software optimization is a substitute for right-sized infrastructure; we're saying software optimization is what makes right-sized infrastructure as cost-effective as possible, and what prevents over-sized infrastructure from generating unnecessarily large demand charges.
Similarly, demand management software cannot retroactively undo a ratchet clause that's already been triggered. The value of intelligent scheduling is highest when deployed before the first vehicle plugs in.
The Measurement Question
Fleet operators who've been managing demand charges manually — through spreadsheets, through simple timer-based outlet controls, or through their EVSE vendor's basic scheduling interface — sometimes ask whether dedicated fleet charging optimization software is actually worth it given what they're already doing.
The honest answer depends on fleet size and tariff complexity. For a 5-vehicle fleet on a simple TOU rate, basic timer controls may capture most of the available savings. For a 30-vehicle fleet on a rate schedule with coincident peak billing, seasonal TOU windows, and a ratchet clause, the scheduling problem has enough dimensions that manual management consistently leaves money on the table.
The way to find out is to pull 12 months of utility bills, identify the demand charge component, and ask how many of those peak demand events were avoidable. If the answer is "most of them happened in the first 90 minutes after vehicles returned to depot," the scheduling opportunity is real and material.
Demand charges reward attention. Fleet operators who build them into the budget from day one — and deploy management systems designed to minimize them — consistently find that the economics of electrification are better than the initial TCO model suggested, not worse.