How much power does an electric truck depot need? Sizing from kWh per day and dwell time

Fleet depots get sized the wrong way round. Someone counts the stalls, multiplies by a dispenser rating, and lands on a service either unbuildable or three times larger than the trucks will draw. A depot is an energy problem first: the trucks need a fixed number of kilowatt-hours back before dispatch, and the question is what power moves that energy inside the window the schedule allows. Thirty Class 8 tractors running 180 miles a day at 2.0 kWh/mi need 10,800 kWh at the batteries, 11,612.9 kWh at the meter, and 1,451.6 kW averaged across an 8-hour window — a 2,000 kVA transformer. The sum of their dispenser nameplates says 2,500 kVA, and NEC 625.42 makes that larger number the code load unless a controller caps it. These are the rules behind the EV charging service capacity calculator.

Why is a depot sized from energy and not from charger nameplates?

A plaza is a power problem: arrivals are random, so every dispenser must meet its rating on demand. At a depot, vehicles, mileage and parked hours are known. Energy binds, power follows.

Daily energy comes from the duty cycle. The National Center for Sustainable Transportation’s review of battery-electric truck studies puts near-term long-haul consumption between 1.3 and 3.8 kWh/mi, narrowing to 1.3–2.5 long term, with NREL’s scenarios at 2.46 and 2.51 kWh/mi for 2018 tractors. A design value of 2.0 kWh/mi sits inside that band.

Two companion articles share the relation P = N × E ÷ (η × t). Sizing a site without a service upgrade takes E as a given (18 kWh in a 200-space Level 2 garage); here E is derived from miles and kWh/mi, which is what changes when the vehicles are tractors. What a DC fast charging site needs converts DC fast charging output kW to AC input kVA for a plaza, where no dwell-time argument applies.

What do NEC 625 and CEC Section 86 require?

NEC 625.42 makes every dispenser continuous, so the unmanaged load is the sum of nameplates with no diversity. 625.42(A) is the only escape.

Worked example — 30 Class 8 tractors on an 8-hour overnight window

A regional-haul depot outside Toronto. Thirty tractors, 180 mi/day, 2.0 kWh/mi, on the yard at 20:00, dispatched at 04:00. Utility primary 13.8 kV, secondary 480Y/277 V, 160 kW DC dispensers rated at DC output. Assume an AC-input-to-DC-output efficiency of 0.93 session-average (the Rocket DC ultra-fast charger is rated to 95.5 % at its peak, so 0.93 is a deliberate derate) and power factor 0.99, both to be confirmed on the datasheet along with the dispenser’s AC input voltage. The DC plaza article uses η = 0.95, which gives a 1,421.1 kW floor and 1,701.2 kVA for ten dispensers — a 2.1 % spread landing on the same ratings.

1. Energy delivered to the trucks. 30 × 180 × 2.0 = 10,800 kWh of DC energy per night.
2. Energy at the meter. 10,800 ÷ 0.93 = 11,612.9 kWh. The 812.9 kWh of losses are not optional.
3. Average AC power across the window. 11,612.9 ÷ 8 h = 1,451.6 kW — the energy floor; no controller or tariff goes below it while the trucks run 180 mi.
4. Apparent power at 80 % loading. 1,451.6 ÷ (0.99 × 0.80) = 1,832.8 kVA.
5. Round up the ANSI/IEEE ladder. The next standard three-phase rating is 2,000 kVA.
6. Dispenser count. Each tractor needs 360 kWh, or 360 ÷ 160 = 2.25 h. Eight hours fits floor(8 ÷ 2.25) = 3 whole sessions per dispenser, so 30 ÷ 3 = 10 dispensers.
7. The unmanaged code load. Ten 160 kW dispensers are 1,600 kW DC, or 1,600 ÷ 0.93 ÷ 0.99 = 1,737.8 kVA at the AC input. Under 625.42 that whole figure is continuous: 1,737.8 ÷ 0.80 = 2,172.3 kVA → a 2,500 kVA transformer, one step above the energy floor.

P = N × d × c ÷ (η × t)

S = P ÷ (PF × L)

h = d × c ÷ Pc

I = S × 1000 ÷ (√3 × V)

At 480 V that is 1,737,800 ÷ (√3 × 480) = 2,090.2 A. NEC 625.41 and 210.20(A) put the main at 1.25 × 2,090.2 = 2,612.8 A, forcing a 3,000 A rating off the NEC 240.6(A) ladder. Add two spare bays and the unmanaged load becomes 2,085.4 kVA, 2,606.7 kVA after the derate and a 3,000 kVA transformer — and 2,508.3 A, which after the 125 % multiplier is 3,135.4 A and a 4,000 A main. Spare bays are cheap; the pad, main and feeder they drag along are not.

A 2,000 kVA unit at 80 % loading and 0.99 power factor delivers 2,000 × 0.80 × 0.99 = 1,584 kW. Against the 1,451.6 kW floor that is 1.091, about 9 % headroom — a cold night or one late truck, not a route extension. Setting the EVEMS ceiling at 1,584 kW is what makes the 2,000 kVA choice legal under 625.42(A), and what makes it tight.

What does the dwell window do to the transformer?

The window is the most leveraged variable here: energy is fixed, so power scales inversely with hours.

Moving dispatch from 04:00 to 06:00 takes this depot from 2,000 kVA to 1,500 kVA. NREL’s workshop report notes utilities commonly — though not consistently across all service territories — place a peak limit of 2.5 MW on an interconnection point, owing to a common maximum size of pad-mounted transformers for secondary service. The window is often the difference between a standard service and a utility study; the same report offers 50–500 kW as the span depot charging could cover.

Depot reference table

At 2.0 kWh/mi, η = 0.93, PF = 0.99, an 8-hour window and 80 % loading the divisor is 0.93 × 8 × 0.99 × 0.80 = 5.89248, so kVA = trucks × miles × 2.0 ÷ 5.89248.

Every row to 3,000 kVA is one pad-mounted unit; the last is a two-transformer lineup. Note that 3,000 kVA is a common manufacturer rating, not a step on the preferred series, where 2,545.6 kVA rounds to 3,750 kVA — see how to size a transformer.

What does the same depot look like in Canada at 600 V?

Voltage and continuous load. At 600 V, 1,737,800 ÷ (√3 × 600) = 1,672.2 A instead of 2,090.2 A, and CEC 8-104(6)‘s ÷ 0.80 gives 2,090.2 A and a 2,500 A main against 3,000 A at 480 V — see choosing a facility distribution voltage. 8-104(5) allows 100 % where the device is marked; the NEC gets there via the 100 %-rated-assembly exceptions to 210.20(A) and 215.3.

Load management is the real divergence. NEC 625.42(A) removes the controlled load from feeder and service sizing. CEC 8-106(10) instead sets the demand load equal to the EVEMS maximum, and 8-106(11) removes it from the calculated load altogether where the EVEMS both monitors the service, feeders and branch circuits and controls the EVSE loads per Rule 8-500 — but only for the calculations enumerated in Rules 8-200 through 8-210, and IAEI advises relying on it only with the local authority’s consent until an EVEMS standard is published. Declare the ceiling as AC input: 1,584 ÷ 0.99 = 1,600.0 kVA, or 1,600,000 ÷ (√3 × 600) = 1,539.6 A, and 1,924.5 A after the 80 % rule — a 2,000 A main. Declared as DC output it would also have to be divided by η. NEC 220.57’s 7,200 VA floor has no CEC counterpart, so Level 2 bays in Canada are calculated at nameplate.

Can a battery buffer hold the service down?

When the utility offers 1,000 kW and the depot needs 1,451.6 kW, storage covers the gap — the peak-shaving duty of cutting demand charges with battery storage, sized by the kW-versus-kWh method in how to size a battery energy storage system. Cap the site flat at 1,451.6 kW, take 1,000 kW from the utility, and the buffer supplies 451.6 kW to the bus for 8 hours.

E = (P − Pu) × t ÷ (DoD × eff)

tr = E × DoD ÷ (eff × Pb)

The shortfall is 451.6 × 8 = 3,612.8 kWh at the bus. At a one-way conversion efficiency of 0.95 the cells give up 3,612.8 ÷ 0.95 = 3,803.0 kWh, and at 0.90 depth of discharge that is 4,225.5 kWh installed with about 500 kW of conversion. Replacing those 3,803.0 kWh costs 3,803.0 ÷ 0.95 = 4,003.1 kWh at the meter — 5.00 hours at 800 kW spare, inside the 16-hour idle period. The 0.95 is paid both ways, so round-trip efficiency is about 0.90, and it is measured, not assumed.

A buffer deferring a service upgrade at a 30-truck depot is therefore a multi-megawatt-hour asset, not a 500 kWh cabinet. The storage is specified against UL 9540 and sited to NFPA 855 — see battery storage fire safety — and its step-up follows battery storage interconnection sizing. A DC-coupled storage and charging system keeps buffer and dispensers on one DC bus.

What to specify

• Daily energy per vehicle, in kWh, and the charging window in hours. Energy from a telematics export or route study, not a brochure range; the window from the dispatch schedule as a guaranteed worst case. Eight hours versus ten is 500 kVA here.
• Charger efficiency, power factor and AC input voltage. From the datasheet at session-average conditions, not the efficiency peak. Assuming 0.95 instead of 0.93 shifts this example by 31 kW.
• The EVEMS ceiling in kW, as AC input, and its fallback state. NEC 750.30 bars the controller from overloading any branch circuit, feeder or service and CEC 8-500(2) from exceeding Rule 8-104(5) or (6); that ceiling becomes the code load under 625.42(A) or 8-106(10).
• Transformer kVA, secondary voltage and %Z. From the transformer sizing calculator and the nameplate — %Z sets the fault current the switchgear clears.
• Main, feeder and interrupting ratings. NEC 240.6(A) after the 125 % of 625.41, or CEC 8-104(6), with the transformer side to 450.3(B) or 26-254 (overcurrent protection) and SCCR to 110.9 from the utility available fault current letter.
• Buffer kW, kWh and depth of discharge. Installed and usable energy quoted separately, under NEC Article 706.

Common mistakes

• Sizing from the nameplate sum and stopping there. Ten 160 kW dispensers read as 1,737.8 kVA and a 2,500 kVA transformer, but the fleet moves only 11,612.9 kWh in 8 hours — 1,451.6 kW average, a 2,000 kVA unit. That sum is the correct code load under NEC 625.42, so the fix is a controller, not a bigger pad.
• Dividing daily energy by 24 hours. 11,612.9 ÷ 24 = 483.9 kW under-sizes by a factor of 3.0 against the 1,451.6 kW the 8-hour window demands. NEC 220.87 allows measured maximum demand on an existing service, never a future load averaged over a day.
• Buying a 500 kWh buffer for a megawatt-hour problem. Holding a 1,451.6 kW depot at a 1,000 kW service needs 4,225.5 kWh at 0.90 depth of discharge — 8.5 times a 500 kWh cabinet.
• Assuming diversity on charger load. NEC 220.57 fixes the EVSE load at the greater of 7,200 VA or nameplate, so the escape is a controller under 625.42(A), not a 0.80 spreadsheet factor. Canada is not symmetrical: CE Code Table 38 does allow demand factors above four EVSE loads, though the ESA subcommittee has recommended deleting it — confirm with the authority first.

Where Entogo fits

Entogo manufactures the depot’s power train in its own roughly 20,000 m² vertically integrated factory — three-phase pad-mounted transformers from 75 to 2500 kVA and up to 3000 kVA at 15/25/35 kV class, low-voltage switchboards for the 1,600–4,000 A mains these calculations land on, and DC ultra-fast chargers at 160/240/300 kW. Equipment is designed and built to ANSI/IEEE C57 and C37, NEC Article 625 and CEC Section 86 as the installation requires; UL (cULus)/CSA certification is available on request. Every product carries a 36-month minimum warranty, with the exact term set per product and confirmed in the contract.

Because transformers, switchgear and chargers are built in the same vertically integrated factory, a depot package can be released against one coordinated submittal rather than chased across three suppliers — which matters when the utility interconnection, not the equipment, should be the critical path. Start from the EV charging service capacity calculator or the transformer configurator; see also EV charging infrastructure.

Depot sizing comes down to four numbers in order: energy delivered to the trucks, energy at the meter, the window it must move through, and the ceiling a controller will enforce. The transformer, main and buffer then follow arithmetically — 10,800 kWh, 11,612.9 kWh, 8 hours, 1,584 kW, and a 2,000 kVA pad instead of a 3,000 kVA one. Take the first number from a brochure instead of the trucks and everything after it is wrong the same way.