Depot Charging for EV Fleets

By Michael Samuel Naeem · January 27, 2026 · 5 min read

Fifty electric delivery vans return between 6 PM and 9 PM to a depot fed by a 400 kW transformer — and someone ordered fifty 19 kW Level 2 chargers because "one per van is simple." Peak demand blows the connection fee, blows the budget, and still leaves van #37 at 62% SoC when drivers leave at 5 AM. Depot charging for fleets is an operations research problem dressed as electrical installation: right-size power, schedule intelligently, and tie charging to when vehicles actually need to leave, not when they physically plug in.

Site power and charger mix

Steps:

1. Utility capacity — firm kW, demand charge structure, upgrade timeline
2. Dwell window — hours available × vehicle count
3. Energy per vehicle — kWh/day from telematics or route model
4. Peak simultaneous — histogram of return times

Example calculation sketch:

Fleet: 40 vans, avg 45 kWh/day replenishment
Dwell: 10 hours (19:00–05:00)
Minimum average power: 40 × 45 / 10 = 180 kW
With 20% scheduling inefficiency → 216 kW plan

Mix DCFC for quick-turn vehicles and Level 2 for overnight dwell — not uniform hardware.

Dynamic load management (DLM) shares unused capacity:

Site limit 300 kW
  ├── Bus A: 80 kW (departs 05:00, SoC 40%)
  ├── Bus B: 60 kW (departs 06:00, SoC 55%)
  └── Vans C–N: 7 kW each capped dynamically

CSMS scheduling architecture

def schedule_depot(vehicles, site_limit_kw, horizon_hours):
    # sort by departure urgency / energy deficit
    ranked = sorted(vehicles, key=lambda v: v.energy_deficit_kwh / v.hours_until_departure, reverse=True)
    allocations = {}
    for v in ranked:
        need_kw = min(v.max_charge_kw, v.energy_deficit_kwh / max(v.hours_until_departure, 0.5))
        allocations[v.id] = allocate_with_cap(need_kw, site_limit_kw, allocations)
    return allocations  # push via OCPP SetChargingProfile

Inputs from telematics webhook:

{
  "vehicle_id": "van-104",
  "eta_depot": "2026-01-27T19:22:00Z",
  "soc": 0.31,
  "next_shift_miles": 112,
  "departure": "2026-01-28T05:00:00Z"
}

Recompute every 15 minutes as arrivals shift.

OCPP smart charging integration

OCPP 2.0.1 ChargingProfile with TxProfile stack:

{
  "chargingProfile": {
    "chargingProfilePurpose": "TxProfile",
    "stackLevel": 0,
    "chargingSchedule": {
      "duration": 36000,
      "chargingRateUnit": "W",
      "chargingSchedulePeriod": [
        { "startPeriod": 0, "limit": 11000 },
        { "startPeriod": 7200, "limit": 0 }
      ]
    }
  }
}

Validate charger supports profile units (W vs A). Fallback: contactor-level DLM hardware if legacy AC chargers ignore software limits.

Depot layout and operations

• Pull-through vs back-in affects cable reach and session start latency
• Pre-conditioning — cabin/battery heat while plugged reduces route energy; schedule before departure spike
• Maintenance bays — isolated circuits so repair lifts do not steal schedule capacity
• RFID/driver login — assign session cost center; prevent personal vehicle plug-in abuse

TCO and reliability

Model:

• Energy cost vs diesel equivalent
• Demand charge avoidance value of smart scheduling
• Charger downtime SLA — redundant dispensers for critical morning wave
• Labor — automated start on plug (no manual app) saves shift change minutes

Track ready-at-departure SoC KPI — missed departures cost more than kWh savings.

Transit vs last-mile differences

Transit may need in-route opportunity charging — depot becomes partial node in larger network.

Load management and demand charges

Depot electricity bills have two cost components — energy (kWh) and demand (peak kW):

Monthly bill = (kWh × energy_rate) + (peak_kW × demand_charge)

Example:
  50,000 kWh × $0.12/kWh = $6,000 energy
  800 kW peak × $15/kW    = $12,000 demand charge
  Total: $18,000/month

Smart charging reduces peak kW without reducing total kWh:

def optimize_depot_schedule(vehicles, chargers, departure_times, max_site_kw):
    # Sort by departure time (most urgent first)
    sorted_vehicles = sorted(vehicles, key=lambda v: departure_times[v.id])
    schedule = []
    for vehicle in sorted_vehicles:
        # Assign charger slot that doesn't exceed site limit
        slot = find_slot(chargers, vehicle, schedule, max_site_kw)
        schedule.append(slot)
    return schedule

Target: reduce peak kW by 30–40% vs uncontrolled charging. Demand charge savings often exceed energy cost optimization.

Vehicle-to-depot (V2D) and grid services

Fleet batteries can provide grid services when not charging:

Overnight: vehicles charge (low grid demand, cheap rates)
Morning peak: fleet departs
Midday: parked vehicles export to grid (demand response revenue)
Evening: vehicles return and charge

Requires ISO 15118-20 BPT-capable chargers and fleet OEM agreement on battery degradation compensation. Start with unmanaged charging; add V2D after baseline operations stable.

Charger redundancy and uptime

Morning departure failure costs more than overnight energy savings:

For 20-vehicle depot with 6 AM departure deadline: N+2 redundancy on DC fast chargers. AC overnight chargers tolerate N+1 — vehicles have 8+ hours to charge.

Failure modes

• Uncontrolled charging — demand charge exceeds energy cost; bill shock
• No departure SoC monitoring — vehicle leaves undercharged; route failure
• Single charger failure with no redundancy — morning departure missed
• Schedule ignores driver overtime — vehicle returns late; not in charge queue
• RFID not assigned — personal vehicle charges on fleet account

Production checklist

• Load management algorithm caps site peak kW
• Departure SoC guarantee enforced per vehicle schedule
• N+1 redundancy on DC fast chargers for morning departure
• RFID/driver login assigns session to cost center
• Ready-at-departure SoC tracked as primary KPI
• Demand charge vs energy cost modeled in TCO analysis

Resources

• OCPP 2.0.1 smart charging use cases
• NREL fleet electrification analysis tools
• CharIN fleet and depot charging guides
• ISO 15118 scheduled charging modes
• California HTF depot charging case studies (CARB)