Ask a fleet manager what they are trying to reduce with smart charging, and the answer is almost always "electricity costs" or "kWh rates." That framing is understandable. Per-kWh rate is the number most familiar from home electricity bills, and it is the one most commonly cited in conversations about time-of-use tariffs and off-peak charging. But for a depot with multiple simultaneous charging sessions, the demand charge component of the electricity bill is often larger than the energy charge component, and it responds to a completely different lever.
Understanding that distinction changes what smart charging actually needs to do, and how you evaluate whether your charging management system is working.
Why the kWh rate gets the attention and the demand charge doesn't
The per-kWh rate is easy to understand and easy to communicate. It shows up clearly on the bill, it is comparable to consumer electricity prices, and the math for computing expected monthly charging cost is straightforward. If each van uses 40 kWh per day and you have 15 vans, you need 600 kWh per day. At $0.12/kWh that is $72 per day or roughly $2,200 per month. Understandable, plannable.
Demand charges are harder to explain and harder to model. They depend on the maximum 15-minute or 30-minute power draw in the billing period, which is a function of how many vehicles charge simultaneously and at what rate, not of how much total energy you consume. Two fleets consuming the same total kWh per month can have very different demand charges depending entirely on whether their charging overlaps or is spread out.
A depot drawing 7.2 kW per charger across 14 simultaneous sessions is pulling 100.8 kW. At a demand charge rate of $14 per kW per month, that single 15-minute peak costs $1,411 for the month. Spread those 14 sessions across a 10-hour window so no more than 5 are running simultaneously at full rate and the peak drops to around 36 kW, bringing the demand charge to roughly $504. The total energy consumed is identical. The bill difference is $907.
Energy cost savings from shifting to off-peak windows on a TOU tariff might be $80 to $200 per month for the same fleet. Demand charge management is the larger prize.
What "staggering charge start times" actually means in practice
Staggering charge start times sounds simple. It is more constrained than it appears. The scheduler is not freely choosing when vehicles charge; it is choosing within a set of constraints that includes: each vehicle must be ready by its next dispatch time, the depot power service connection has a hard ceiling, each charger has a maximum output rate, and some vehicles have more urgent charging needs than others based on current SOC and tomorrow's route.
A naive stagger that delays all vehicles by a fixed offset (start vehicle 1 at 5:00 PM, vehicle 2 at 5:15 PM, vehicle 3 at 5:30 PM, etc.) reduces the simultaneous draw compared to all vehicles starting at 5:00 PM. But it may not minimize the monthly demand peak, because the peak might now occur at the point where the first several vehicles are still charging at full rate and the next few have just started. The actual peak power moment depends on the charge rate and session duration of each vehicle, which varies by SOC.
A better stagger is rate-aware: it considers each vehicle's current SOC, maximum charge rate, and required SOC at departure. Vehicles that need a lot of energy and have a late departure can start slow or start late. Vehicles with high urgency (low SOC, early dispatch) get prioritized. The scheduler constructs a timeline where the sum of simultaneous draw at any 15-minute interval stays below a target ceiling, usually set at 60 to 80 percent of the depot's service capacity to leave headroom for other facility loads.
Rate limiting as a complement to start-time staggering
Start-time staggering is not the only tool. Per-charger rate limiting, which is possible on chargers that support OCPP Smart Charging profiles, lets the scheduler reduce the output of a charger below its maximum rather than delaying the session entirely. This matters when a vehicle needs to charge but cannot wait for a later start time.
A vehicle returning at 4:30 PM with 18 percent SOC and a 6:00 AM dispatch the next day has plenty of time available. If that vehicle starts charging immediately at 7.2 kW, it finishes in about 5.5 hours, completing at 10:00 PM. If instead it charges at 3.6 kW from 4:30 PM to midnight, it finishes at the same SOC target and the depot draws half the power for that vehicle during the busy return window.
Rate limiting is not appropriate for every vehicle. A vehicle with a later return and a morning dispatch that has genuinely tight timing should charge at full rate. The scheduler needs to distinguish between vehicles where reduced rate is safe given the time available and vehicles where reduced rate risks a missed dispatch target. That requires knowing the latest-acceptable-start charging time for each vehicle, which requires knowing the route schedule for tomorrow.
The interaction between demand management and TOU optimization
These two goals interact, and the interaction has a trap. The natural outcome of TOU optimization on its own is to delay all charging sessions until after the on-peak window closes, typically 8:00 PM to 10:00 PM depending on the tariff. If 14 vehicles all start charging at 9:00 PM, you have eliminated on-peak energy charges and created a 100 kW demand spike at 9:00 PM. The on-peak kWh savings are real but smaller than the demand charge increase.
The correct approach is to treat demand management as a constraint that the TOU optimizer operates within, not a separate objective that comes second. The scheduler should produce a solution that keeps the demand ceiling below a target throughout the billing period while minimizing the energy cost under that constraint. In most cases, this produces a solution that naturally uses some off-peak window without creating a simultaneous start cluster when the window opens.
We are not claiming this is technically difficult to reason about. It is standard constrained optimization. What makes it practically difficult is that the input data (per-vehicle SOC, tomorrow's dispatch schedule, today's facility load baseline) needs to be reasonably accurate for the output to be useful. A schedule built on stale SOC data or a dispatch plan that changes after the schedule was computed is a schedule that may not reflect actual conditions on the floor.
The metric that tells you if demand management is working
The clearest signal that your smart charging system is managing demand effectively is your monthly demand charge value trending down and stabilizing at a predictable level, rather than varying month to month based on whatever happened to coincide on the worst day.
A demand charge that swings between $800 and $2,600 month to month without clear explanation indicates that demand peaks are being driven by uncontrolled simultaneous charging events, probably on days when a lot of vehicles return within a short window and all plug in at once. A demand charge that stays within a narrow range suggests the scheduler is successfully maintaining a power ceiling across varying fleet conditions.
Fleet managers who have implemented effective demand management describe the change in similar terms: the electricity bill stops being a surprise. The demand charge line item becomes a number in a range you expect, rather than a number you have to explain to accounting after the fact. The energy cost per kWh is still there and still worth optimizing, but it becomes the smaller part of a bill that is now substantially predictable.