Depot charging planning: how to share power so every bus makes its first run

Every evening, the dispatcher at an electric bus depot has the same question: will everything get out in the morning? With a strong grid connection, the answer is easy. With a weak one, it depends on how you share power between the buses overnight, and that split often isn't designed by anyone. Power gets divided equally, the chargers report no faults and everyone's relaxed until the first block pulls out short of energy.

SoC alone won't give you a charging schedule

When we talk about readiness here, we only mean energy. A bus needs to leave with enough kilowatt-hours for its block, plus whatever reserve the operator sets. Brakes, doors and the rest are a different conversation.

How much is enough? It depends on the route. A short city-centre route and a long suburban one burn very different amounts, even with identical buses, and in winter heating widens the gap. NREL's guidebook on deploying battery electric buses lists what to account for: route profile, passenger load, temperature, heating and air conditioning. [1]

That's why we don't set the charging order by SoC. 60% in a bus doing two short trips tomorrow is more than it needs. The same 60% before an all-day block on a long route is trouble. What we calculate is the missing energy: how much has to go in so the bus can finish its block with the reserve intact.

That's only half the sum, though. The other half is time. You need to know when the bus actually gets to a charger (not when the timetable says it should be back) and when it has to be unplugged. Charging stops before departure, because the driver still has to pull the plug and get the bus off the bay.

Then there's the hardware: how many charge points you have, whether a charger splits power across two connectors, how much the vehicle will actually take at its current state of charge. And there's the depot's power budget, meaning what's left for charging after the bus wash, the workshop and the yard lighting have taken their share.

The data has to be fresh. A plan built on SoC readings from three hours ago looks great on screen, but the yard has moved on.

A charging schedule example: six buses, 120 kW

The depot is made up, and the numbers are chosen so you can check them in your head. At 23:00, six buses come in and each needs 150 kWh from the grid. A and B leave at 5:00, C, D, E and F at 7:00.

Option one: split it equally. 120 kW across six buses is 20 kW each. There are six hours until five o'clock, so A and B end up with 120 kWh each. They're 30 short. The chargers won't flag anything all night, because they did exactly what they were told.

Option two: A and B get 30 kW each from the start, the rest get 15. Still 120 kW in total. At four, A and B have their 150 kWh and stop, so C to F can move up to 30 kW. Each of them has 75 kWh by four and picks up the other 75 by 6:30.

Everyone makes it. The first two with an hour to spare, the others with half an hour.

In option one the totals added up too: the depot's needs fit within the available power. The order was the problem. So a depot-wide energy balance proves nothing on its own. You have to check each bus against its own departure time.

The other thing is the margin. Between 23:00 and the last departure at 7:00, the depot can deliver 960 kWh at most. It needs 900. That's 60 kWh of slack, or half an hour at full power. Lose 30 kW of budget for two hours and the slack is gone. Faced with a result like that, the first thing we'd ask is what happens if one bus comes back an hour late.

A real depot also has the charging curve to deal with. Above a certain SoC, a bus takes less than you offer it, so the last stretch takes longer than simple division suggests. We left it out here to keep the maths readable.

Late returns, charger faults and lost comms

The 23:00 plan holds until the first phone call from the road. It has to be recalculated every time something changes: a late return, an interrupted session, less power, two blocks swapped.

A late return is easy to work out. A gets back at 3:00 and needs 150 kWh. It has two hours at a maximum of 60 kW, so the best it can do is 120 kWh. You can give it top priority, you can unplug everyone else. It makes no difference. It'll be 30 kWh short.

What matters is when the dispatcher finds out. At 3:05, they have two hours to put out a different bus or shuffle blocks. At 4:55, they have a problem. The system should flag the shortfall as soon as the bus is back, with an actual number: this many kWh missing, this much time left.

A charger fault looks easier, since power is freed up. Except the bus is still parked at a dead charger. Someone has to move it to a free bay with a compatible connector, and at night there isn't always anyone in the yard to do it. The FTA's best-practice report draws attention to exactly this: infrastructure interoperability and operational readiness. [2] A new power allocation that assumes the bus will hop to another bay by itself is useless.

Lost comms are the trickiest. The charger goes quiet and you can't tell whether it's charging or not. We don't assume either. We flag the data as stale, switch to the fallback mode agreed in advance, and someone goes to check.

Smart charging at the depot: what to build it from

You don't have to replace anything. The fleet system knows which block leaves when. The CSMS, the charging station management system, knows what's happening on the chargers. The EMS, the site's energy management system, knows how much power you're allowed to draw. What's usually missing is something that puts those three together and says: this bus gets 30 kW, that one 15. That's the planning layer.

Limits reach the chargers through OCPP's smart charging mechanisms. [3] Just check that your chargers and your CSMS really support them. Then watch the meter, not the acknowledgement. A charger that has accepted a "30 kW" command doesn't necessarily deliver 30 kW.

At i4B we treat this as building blocks. Each piece can be plugged in on its own: into your own product, a system built for an operator, or something bought off the shelf. You can also start small. At first, planning just calculates and advises the dispatcher. We switch on power control only once the integrations are tested and everyone knows what the system will do when a charger goes quiet.

How to tell depot charging planning is working

One metric matters more than the rest: the percentage of departures where the bus had hit its energy target. For every miss, we log the reason: late return, fault, not enough power, bad data. After a month you can see what's really spoiling your mornings.

Beyond that, we look at the time buffer before departure, the number of overnight interventions, and how many minutes pass between spotting a risk and calling the dispatcher.

Energy cost? That too, but last. Cheap charging that leaves a block unable to run isn't a saving.

Sources and assumptions

1. NREL (2021). Electrifying Transit: A Guidebook for Implementing Battery Electric Buses. A. Aamodt, K. Cory, K. Coney. Chapter 8: route, infrastructure and operations planning.
2. Federal Transit Administration (2023). Procuring and Maintaining Battery Electric Buses and Charging Systems: Best Practices. Report No. 0253, Chapter 5.
3. Open Charge Alliance. What is new in OCPP 2.0.1, version 1.0. Smart Charging section and Functional Block K.