Fleet managers evaluating charging scheduling software often encounter phrases like "constrained optimization," "mixed-integer programming," or "predictive energy modeling." These are real technical terms describing real processes, but they do not always help a fleet ops team understand what the software actually does or why it matters for their operations.
This is a plain-language primer. If you run a commercial EV fleet and are trying to understand what charging schedule optimization actually solves, this is the explanation we wish existed when we started building Fleetvolts.
The Problem Manual Scheduling Cannot Solve Well
Imagine a depot with twenty-four electric delivery vans. Each van returns at a different time in the evening, depending on route completion. Each van has a different state of charge depending on route distance, payload, and conditions that day. Each van has a departure time tomorrow morning, and some routes require more charge than others.
Your depot has a 200 kW service connection. You have twelve Level 2 charging ports at 7.2 kW each. Your utility rate charges $16/kW per month for peak demand measured in 15-minute intervals, and your off-peak electricity rate starts at 11 PM.
A dispatcher managing this manually will do something reasonable: plug vehicles in as they arrive, hope the peak demand does not get too high, and check in the morning that everything is fully charged. This works. It is not optimal. The simultaneous plug-in ramp that typically happens between 6 PM and 8 PM creates a demand spike that costs money every month. And the vehicles that least needed an overnight top-off used charger capacity that could have been reserved for the ones that did.
Manual scheduling fails not because dispatchers are bad at their jobs. It fails because the problem has too many interacting variables to track by hand, and the cost consequences of suboptimal decisions accumulate quietly on the electricity bill rather than showing up as visible operational failures.
What Constrained Optimization Means in Practice
Optimization in this context means finding the charging schedule that minimizes total electricity cost while satisfying a set of constraints. The constraints are the non-negotiable operational requirements: every vehicle must reach its required state of charge by its departure time. The objective is to achieve that at the lowest possible cost given the tariff structure and available charging capacity.
The constraints interact. You cannot charge all vehicles simultaneously because depot power capacity is finite. You cannot delay every vehicle's charge until cheap off-peak hours because some vehicles depart before those hours arrive. You cannot always perfectly spread the load because vehicles arrive at different times and the charger ports are a shared resource.
A scheduling algorithm navigates these tradeoffs systematically. For each fifteen-minute interval through the overnight window, it decides which vehicles should be charging at what rate, subject to the constraints above. It does this across dozens of vehicles simultaneously, which is the part that is genuinely not tractable by hand.
The result is a schedule where vehicles that can wait do wait, vehicles that need early charges get them, the simultaneous draw on the depot's power service is smoothed, and demand-charge-generating spikes are reduced or eliminated.
The Role of Energy Prediction
Scheduling optimization requires knowing, at the start of the overnight window, how much charge each vehicle will need. That sounds simple, but it is where most basic scheduling tools fall short.
A naive approach uses the current state of charge and targets 100% for every vehicle. This wastes charging capacity on vehicles that did not deplete much today, and it does not account for tomorrow's route energy requirement. A vehicle that will run a 90-mile urban route tomorrow needs a different charge target than one running a 40-mile suburban loop.
Accurate per-vehicle energy prediction changes the problem. Instead of "charge to 100% by departure," the constraint becomes "charge to the state of charge required for tomorrow's route plus a reasonable buffer." This unlocks scheduling flexibility. Vehicles with lighter tomorrow routes can charge later and more slowly, preserving early high-rate charging capacity for vehicles that genuinely need it.
This is why we built energy prediction into Fleetvolts before building the scheduling layer. Without per-vehicle, per-route energy forecasts, the scheduler is working with a blunt target that leads to systematic over-charging of some vehicles and under-optimized load distribution.
Why the Outcome Is Better Than Manual Scheduling
The measurable difference between optimized and manual charging schedules appears primarily in two places: the monthly demand charge line and total kWh consumed.
Demand charge reduction comes from flattening the simultaneous load peak. When vehicles charge in a staggered sequence instead of all at once, the 15-minute peak demand interval is lower. At $16/kW, reducing peak demand by 30 kW is $480 per month, every month, regardless of route activity.
The kWh savings are smaller but real. Avoiding systematic over-charging to 100% when vehicles only needed 75% reduces total energy purchased. Over a fleet of twenty-four vehicles, consistently avoiding 10 kWh of unnecessary charging per vehicle per night is 240 kWh per night, which at even $0.09/kWh off-peak accumulates over a month's operations.
We are not claiming optimization eliminates all electricity cost variability. Weather, route changes, and vehicle health events create genuine unpredictability that no scheduling system can fully absorb. The goal is to remove the structural inefficiency of uncoordinated simultaneous charging, which is predictable and addressable.
When Manual Scheduling Is Actually Fine
It is worth being direct here. A depot with four or five vehicles, an abundance of charger capacity relative to load, and a flat-rate electricity tariff with no demand charges gains very little from automated scheduling optimization. The problem is simple enough to manage by hand, and the tariff structure does not create cost incentives that optimization can capture.
The value of scheduling optimization scales with fleet size, charger-to-vehicle ratio tightness, demand charge exposure, and route variability. Once a depot hits roughly fifteen to twenty vehicles and is on a commercial tariff with meaningful demand charge rates, the math on optimization starts to work clearly.
The fleets that benefit most are the ones operating at a charger-to-vehicle ratio below 1:1, meaning more vehicles than charging ports, with heterogeneous routes, on commercial tariffs that include both demand charges and time-of-use energy pricing. That describes most growing delivery fleet depots. If that describes your operation, manual scheduling has a ceiling that automated optimization does not.