Power Master Load Balancing
Every charging platform says it supports load balancing. The question that matters is what the allocation optimises for, because a scheme that maximises total energy delivered and one that guarantees everyone leaves with enough are different algorithms and they produce different winners.
Where the balancing decision runs and why the edge wins, the four allocation algorithms and what each optimises for, which one suits which kind of site, and what happens when controllers cannot talk to each other.
Tell us what the site needsFor An engineer or operator who wants to know how the allocation is actually decided rather than that load balancing is supported.
Where the decision runs, and why the edge wins
Load balancing can sit in three places, and on a site with more than one charger only one of them actually works.
Swipe to compare
| In the charger | In the cloud | At the edge, on Power Master | |
|---|---|---|---|
| Sees other chargers | No. Each unit decides alone | Yes, if all report in | Yes, directly |
| Sees site load | No | Only what is fed to it | Yes, measured locally |
| Survives loss of connectivity | Yes | No | Yes |
| Reacts in | Immediately | Seconds, network permitting | Milliseconds to seconds |
| Suits | A single charger | Policy and scheduling | Any site with more than one charger |
The first column is the one that catches people out. A charger deciding its own allocation works perfectly until there is a second charger, because each unit assumes it has the whole supply and together they exceed it. That is not a tuning problem, it is the architecture.
Power Master at the edge is the arrangement we recommend for any real site. It sees the building's own load and every charger at once, it keeps deciding when the connection to the platform is gone, and it is where the algorithms below actually run. The cloud sets policy; the edge enforces it.
The four allocation algorithms
Charger360 Edge carries four, because sites want measurably different outcomes from the same supply. Choosing between them is a commissioning decision rather than a default.
- 1
First come, first served
Whoever plugs in first gets their full rate, and later arrivals take what is left. Simple and predictable, and it produces the resident who plugs in at nine in the evening and gets very little. Suits sites where arrival order genuinely reflects priority.
- 2
Equal share
The available capacity is divided evenly across active sessions and redivided as vehicles arrive and leave. Feels fair and is easy to explain, and on a busy site it can leave everybody short rather than most people satisfied. The default choice for apartments, where twelve hours absorbs the reduction.
- 3
Need based
More capacity goes to the vehicle with the furthest to go. The most efficient use of a constrained supply, and it needs to know state of charge, which DC exposes and AC does not.
- 4
Deadline based
Allocation worked backwards from when each vehicle has to leave. The strongest outcome where departure times are known, which in practice means depots and workplaces rather than public sites.
There is also a rotational arrangement that can sit on top of these: a vehicle that has already received a large share is stepped down so one that arrived later can be stepped up, and the allocation cycles between them rather than settling. It suits sites where every driver needs a useful amount and none needs a full charge.
Choosing between them
Swipe to compare
| Site | Usually | Because |
|---|---|---|
| Apartment, overnight | Equal share | Twelve hours absorbs a reduced rate and no driver is waiting |
| Workplace, all day | Equal share or rotational | Maximises how many staff get a useful charge in one day |
| Fleet depot | Deadline based | Departure times are known, and a vehicle must leave ready |
| Public, short dwell | First come, first served | Arrival order reflects who is waiting |
| Mixed, constrained supply | Need based | Where state of charge is visible, it uses a tight supply best |
The honest way to choose is to decide who the site is worst for. Every scheme has a driver it treats least well, and picking that deliberately is better than discovering it in a complaint.
Why there is more than one algorithm
A single allocation rule cannot serve every site, because sites want different outcomes and the differences are real rather than a matter of tuning.
- An apartment overnight wants everyone to leave with enough. Nobody cares who charged first
- A workplace wants as many staff as possible to get a useful charge in a fixed window
- A depot wants every vehicle ready by its departure time, which is a scheduling problem rather than a sharing one
- A public site wants energy delivered and sessions completed, because that is the revenue
- A site on a demand-charge tariff wants the peak held below a threshold, even at the cost of slower charging for everyone
Those five want measurably different allocations from the same supply. That is why Charger360 Edge carries several algorithms rather than one, and why choosing between them is a commissioning decision rather than a default.
When controllers cannot talk to each other
This is the failure that takes a site off supply, and it is the question worth asking any vendor.
- Two controllers that each assume they have the whole supply will together exceed it. This is the specific failure to design against
- The safe behaviour on losing contact is to reduce allocation, not to hold it and not to assume capacity
- Each unit needs a standalone floor: a rate it may offer with no coordination at all, chosen so that every unit at that floor is still within the connection limit
- Loss has to be alarmed, because a site quietly running at its floor for weeks is a fault that went unnoticed
- Recovery has to be tested as well as failure
The floor is the important number and it is rarely asked about. It has to be safe with every point active simultaneously, which on a dense site is a small figure.
Want this applied to your own site?
Tell us what the site needsTechnically reviewed by Deepu Joy, Director of Products and Delivery. Last reviewed 2026-09-10.
Frequently asked questions
Should load balancing run in the cloud or at the edge?
At the edge. The sites that most need load management are frequently the sites with the worst connectivity. The cloud sets policy, the edge enforces it, and the charger holds a safe floor if both are unreachable.
Why does the choice of algorithm matter?
Because sites want different outcomes. Maximising energy delivered and guaranteeing everyone leaves with enough are different objectives and produce different allocations from the same supply.
What are the four algorithms?
First come first served, equal share, need based and deadline based. A rotational arrangement can sit on top of any of them, stepping a vehicle that has already had a large share down so a later arrival can be stepped up.
What happens if the controllers lose contact?
Each unit falls back to its default current limit, a standalone floor that is safe with every point active at once. Holding the last allocation, or assuming full capacity, is the failure that trips the site.
Is fairness a technical decision?
No. It is a business decision about who the site is for, implemented technically. Every scheme has a driver it treats worst, and choosing that deliberately is better than discovering it.
More bays than supply?
Tell us what the site needs to achieve, not how much capacity it has. The objective decides the algorithm, and the algorithm decides how much capacity you actually need.
Tell us what the site needs