Smart Charging for Fleet Depots: Sequencing 40 Vehicles Through 12 Chargers Overnight

Smart Charging for Fleet Depots

Smart charging for fleet depots is a constrained-optimisation problem wearing a charging-infrastructure costume. If you are a CPO bidding for a bus-depot or logistics-yard contract, or an enterprise fleet operator running your own yard, the question that decides your margin is not “how many chargers do we install?” — it is “what is the smallest grid connection that still gets every vehicle out of the gate on time?” Those two questions have very different answers.

This is written for the people who own that number: CPOs tendering for depot infrastructure, and the fleet and facilities teams behind EV fleet charging management at bus depots, LCV yards and warehouse sites. Public charging optimises for throughput and willingness to pay. A depot optimises for one binary outcome — every vehicle at its required state of charge at its scheduled departure — at the lowest sanctioned load you can get away with. Everything below is the mechanics of doing that with 12 chargers instead of 40.

Why a Depot Breaks Every Public-Charging Assumption

A public charging site is a queueing problem with unknown arrivals. A depot is a scheduling problem with known arrivals. That single difference inverts almost every design rule you brought with you from public infrastructure.

The vehicles are captive, so utilisation is a cost line, not a revenue line

On a public site, an idle charger is lost revenue and you fight for utilisation. In a depot, the fleet charges on your asset regardless — so every charger above the minimum viable count is pure capital you did not need to deploy.

Departure times are known hours in advance

A bus depot knows the 05:40 timetable. A logistics yard knows the first wave leaves at 06:00 and the second at 11:00. That foreknowledge is the asset: it converts “charge everything as fast as possible” into “finish each vehicle just before it is needed,” which is what flattens the power curve without missing a departure.

There is exactly one grid connection, and it is the real constraint

Public sites are usually power-constrained per socket. A depot is power-constrained in aggregate: chargers, workshop, compressors, lighting and HVAC all sit behind one meter and one sanctioned load. The Smart Freight Centre’s depot guidance is blunt — grid connection capacity must cover both the charging cluster and the site’s base load, and upgrades “commonly entail significant investment costs for equipment and installation, as well as higher grid connection fees.” For the mechanics of site-level power sharing, start with load management basics for non-engineers.

The charger-to-vehicle ratio is deliberately below 1:1

A fleet with a homogeneous schedule — every vehicle arriving and leaving together — genuinely needs a 1:1 charge-point ratio. But the same guidance notes that “for fleets where the trucks have different operating schedules the number of charge points may be chosen to be less than the number of trucks, presuming that their schedules allow to use the same charge points consecutively.” Heterogeneity is the resource you are monetising. Sequencing is how you spend it.

Departure-Based Sequencing: How 12 Chargers Serve 40 Vehicles

Plugging vehicles in as they arrive is the default behaviour of every uncontrolled depot, and it is the worst possible order. Arrival order correlates with nothing that matters. What matters is laxity: the slack between how long a vehicle has until departure and how long it needs on a charger.

The three inputs the platform actually needs

Sequencing needs three things per vehicle, from three different systems. Departure time comes from the duty roster or timetable. Required state of charge at departure comes from tomorrow’s route energy — distance × observed kWh/km, plus a reserve. Current state of charge comes from telematics, or from the charger via ISO 15118. Energy still needed divided by deliverable power gives charge duration; departure time minus charge duration minus now gives laxity. Sort ascending, and the queue writes itself.

Least-laxity-first sequencing at 22:00. LCV-21 is the emptiest vehicle in the yard and charges last, because it has nearly seven hours of slack. LCV-12 charges first on 48 kWh, because it leaves in 90 minutes.

Why one 60 kW charger clears three to four vehicles a night

Replenishing 60 kWh at 60 kW takes 60 minutes of energy transfer, or about 67 minutes at a realistic 90% end-to-end efficiency. Add eight minutes for unplug, shunt and re-plug and a slot costs about 75 minutes — up to eight slots per charger in a 22:00–06:00 window, far more than the 3.3 the arithmetic demands. The binding constraint at 12 chargers is therefore never charger count. It is the site power cap, which is why sequencing logic and power-sharing logic have to be the same logic.

Rebuild the queue continuously, not once at 22:00

A vehicle returns 40 minutes late. A route is reassigned. A charger faults. Each invalidates the plan, so the queue is recomputed on every event, not solved once at lock-up — laxity re-evaluated on each transaction start, stop, meter-value and roster change, with revised limits pushed to chargers already mid-session.

The Grid-Connection Ceiling: Size to Energy, Not to Peak

Here is the number that pays for the whole software layer. An unmanaged depot has to be electrically sized for the worst case: every charger drawing rated power at the same instant. A managed depot is sized for energy delivered across the window. Work the same 40-vehicle fleet both ways.

Same fleet, same 2,400 kWh, same departures — an 80% smaller grid connection. The managed column only works if a control layer enforces the 300 kW setpoint every minute of the night.

The honest version matters. No competent electrical designer sanctions the full 2,000 kW; they apply a diversity factor and land near 1,400 kW. But a diversity factor is an assumption, and an assumption cannot be enforced. A dynamic load management setpoint can be — it is a number the chargers are physically held to. That is the difference between hoping the peak stays down and contracting for it.

The saving lands in two places. Capital: 28 fewer chargers plus the civil work, cabling, switchgear and transformer behind them. Operating: demand charges. Tamil Nadu’s HT public EV charging category carries a demand charge of ₹304/kVA/month before the state’s 50% EV rebate; on 1,600 kVA of avoided sanctioned load that is roughly ₹4.9 lakh a month, or ₹58 lakh a year. In Gujarat, where HT EV charging demand charges sit near ₹25/kVA/month, the same 1,600 kVA is worth about ₹4.8 lakh a year and the case shifts onto capital and connection charges instead. The mechanism is universal; the magnitude is state-specific, so run it against your own tariff schedule before quoting a saving.

The Standards Floor: OCPP Profiles That Make Smart Charging for Fleet Depots Work

Sequencing is only as good as the control channel underneath it. In OCPP, that channel is the charging profile — a schedule of power or current limits the platform pushes to a charger, which the charger is obliged to respect. The Open Charge Alliance has shipped three relevant generations: OCPP 1.6 in 2015, 2.0.1 in 2020, and 2.1 in 2025.

OCPP smart charging capability by version. The row that decides depot architecture is the third one: without an EV-declared departure time, the platform has to infer it from the roster.

What OCPP 1.6 hardware still lets you do

Plenty — and this matters, because most installed depot hardware in India is 1.6J. You can enforce a site cap, throttle individual connectors, and run time-of-day schedules. What you cannot do is let the vehicle tell you when it leaves; the platform infers departure from the duty roster and SoC from telematics, which works well for a fixed bus timetable and poorly for an ad-hoc logistics yard. Profiles are also static, so a late arrival needs the platform to push a replacement profile rather than the charger adapting on its own.

Verify the profile is honoured, not just accepted

The most common depot failure has nothing to do with protocol version: a charger returns Accepted to a SetChargingProfile and then ignores it, or applies the limit per connector when you meant it per station. Many chargers implement TxProfile and TxDefaultProfile but not the station-level cap at all. Test every SKU against measured meter values before you size a connection around a setpoint, and keep demand-responsive charging policies keyed to what each model actually obeys rather than what its datasheet claims.

Smart Charging for Fleet Depots Looks Different in Each Segment

E-bus depots: high simultaneity, unforgiving contracts

Bus depots are the hardest case because the schedule is homogeneous — buses return in a cluster and leave in a cluster, exactly the condition under which the 1:1 ratio genuinely applies. The energy is large too: Delhi Transport Corporation’s Rajghat-II electric depot consumed about 615,000 kWh charging 72 buses over 440,000 km in July 2024 — roughly 285 kWh per bus per day at 1.38 kWh/km. Compress 20.5 MWh into a seven-hour window and you are drawing over 3 MW. Two levers move it: shifting a share of energy into the midday layover, which is where the IEA’s up-to-60% figure comes from, and staggering by timetable block so the 05:40 departures finish before the 08:00 block starts. Under PM-eBus Sewa — 10,000 buses across 169 cities on a ₹57,613 crore outlay — depot upgrades and behind-the-meter power infrastructure are explicitly in scope, so this sanctioned-load decision is being made now, at scheme scale, and once per depot.

Logistics yards: staggered shifts and opportunity top-ups

Warehouse and last-mile yards are the easiest segment to run below 1:1 because schedules are heterogeneous by design: a first wave out at 06:00, a second at 11:00, returns trickling in from 14:00. That produces real slack to trade, plus a mid-shift window where a 20-minute opportunity top-up is worth more than an hour overnight. The catch is variance — routes change daily, so required SoC must be recomputed from tomorrow’s plan, not a standing assumption. That is where integration with the fleet’s own scheduling system earns its keep; YoMobility’s fleet management platform is the sister product we most often pair with a depot deployment for exactly that roster-to-charger handshake.

Mixed LCV fleets: heterogeneous charge acceptance

A yard running three-wheelers, small LCVs and a couple of medium trucks has a subtler problem: the vehicles do not all accept the same power. Depot guidance is precise on this — if a vehicle accepts only 22 kW on AC, a 43 kW AC charge point buys you nothing, and you need DC to hit the required average power. Mixed depots therefore need a mixed AC/DC estate with per-connector limits and a sequencing engine that knows each vehicle’s charge-acceptance curve rather than assuming the charger’s rating is the delivered rate. The commercial layer on top — who pays what for which kWh when a CPO owns the chargers — is covered in our guide to B2B fleet charging tariffs.

Six Metrics That Tell You Sequencing Is Actually Working

Depot dashboards default to public-charging metrics — sessions, kWh, revenue — none of which answer whether the depot did its job. Instrument these instead.

1. SoC-at-departure success rate. Share of departures that left at or above required state of charge. The depot’s only real SLA; below 99% is a scheduling failure, not a charging one.
2. Peak kW against contracted kVA. Monthly maximum demand as a percentage of sanctioned load. Persistently under 70% means you over-sanctioned; touching 100% means your setpoint is not enforced.
3. Throughput and idle-plugged time. Sessions per charger per night, and the gap between connected hours and energy-delivering hours. A finished vehicle nobody moved is a charger you did not have.
4. Setpoint breach events. Intervals where measured site draw exceeded the commanded ceiling. Every breach is either a charger ignoring a profile or a base load you failed to model.
5. Profile acceptance versus compliance. Share of profiles accepted and observably honoured at the meter, per charger model — the number that tells you a firmware update broke your depot.
6. kWh per km by vehicle. Drift is an early warning of battery degradation, load changes or a longer route — all of which quietly raise tomorrow’s required SoC and eat your buffer.

All six come from data the chargers and telematics already produce; what is usually missing is one place that joins them. That is the practical argument for running the depot on a charging management system like YoCharge rather than a spreadsheet and the charger vendor’s own portal: sequencing decisions, power setpoints and departure outcomes have to be reconciled against the same clock. Get that right and the depot’s grid connection becomes a design choice rather than a bill you inherit.

Frequently Asked Questions

Sources: IEA — Global EV Outlook 2026, Electric Vehicle Charging | Open Charge Alliance — OCPP | Open Charge Alliance — What’s New in OCPP 2.0.1 | Smart Freight Centre — Charging Infrastructure for Truck Depots | PIB / MoHUA — PM-eBus Sewa | Down To Earth — DTC depot energy data | Mercom India — Tamil Nadu FY2026 tariff order