A local energy brain in a shipping container

The scenario

Picture a shipping-container shop at a trailhead or a festival site — no fixed grid connection, or only a weak one. On the roof: a PV array. Inside: a battery, a fridge and lighting, an AC charging station for e-cars, and a rack of switchable sockets for an e-scooter fleet. The job: charge as many vehicles as the sun and battery allow, never trip the (optional) grid fuse, and keep the fridge running — automatically, with nobody on site.

How the Edge Controller optimises

The controller runs three loops on the box itself, every few seconds:

• Measure — PV yield, battery state-of-charge and every meter (Modbus) plus each charge point's live power (OCPP) land in one model.
• Optimise — a constraint solver shares the truly available power (PV + battery headroom − base load) across all loads under every limit at once. Strategies: fair share, priority, first-come-first-served, minimum-guarantee or departure-time.
• Control — it pushes a charging profile to each EV charger (OCPP) and switches the scooter sockets and shop loads (Modbus / smart plugs) — shedding the lowest-priority load first when the sun fades.

The building blocks

The Edge Controller is a small fan-less box (ARM or x86). It speaks OCPP to the chargers, Modbus TCP/RTU to meters, inverters and sockets, and MQTT internally. Everything below runs in containers on that one device.

Works when the line is down

Because the brain is on-site, a dropped internet link changes nothing: the controller keeps measuring, optimising and charging, and buffers everything the cloud needs (sessions, meter data) until the link returns — then drains it in order. The same mechanism makes it §14a-EnWG ready: a grid operator's dimming signal becomes a top-priority cap that overrides everything, instantly and verifiably.

Build your own — in five steps

1. Mount the Edge Controller in the container and wire it to the LAN with the chargers, meters, inverter and sockets.
2. Run the commissioning wizard: it discovers the chargers, scans the Modbus meters, checks the phases and sends a safe, reversible test-limit.
3. Draw the topology: grid/PV/battery at the top, the charge points and sockets as leaves, with the connection and sub-circuit limits.
4. Pick a strategy (e.g. minimum-guarantee so every scooter gets a base charge, then fair share of the surplus).
5. Go live — and manage it locally on the box or remotely from the CPMS portal.