Swiss ZEV & vZEV · self-hosted · AGPL-3.0
OpenZEV runs the whole operating cycle of a Swiss energy community: import the load curves, split the solar between neighbours timestamp by timestamp, price it against real tariffs, and send a QR-bill invoice that reconciles.
Energy balance · demo community · Q3 2026
The same split the billing engine performs — min(production, consumption)
per 15-minute interval — is what every invoice line is derived from.
What it does
A ZEV is only simple until you have to bill one. OpenZEV covers the parts that actually consume the evenings: reconciling imports, keeping tariff history straight, and producing documents a participant will accept.
Import from the formats utilities actually send, and see what will land before it does.
Allocation runs per timestamp, not per month — so high/low bands and spot prices land correctly.
Invoices as PDF/A-3b with a QR-bill payment slip, plus annual statements and participation contracts — each issued version kept exactly as it was sent.
Admins, ZEV owners and participants see different products. Metering points carry validity-based assignments, so a participant moving out mid-period bills correctly.
A feasibility calculator estimates savings, payback, ROI and NPV — and can prefill from a real community's participants and measured self-consumption.
Invoice lifecycle
Every invoice moves through the same chain. Transitions lock the row and decide against committed state, so two people working the same billing run cannot overwrite each other.
Generated from the period's readings and tariffs. Re-generate freely.
Figures accepted. The PDF is rendered off-request and attached.
Emailed to the participant; delivery and failures both recorded.
Marked on receipt, closing the period for that participant.
Cancelled branches off any stage and is never a deletion — the document
and its number stay, because an invoice a participant already holds cannot be unsent.
Screenshots
From the demo community that ships with the project — seeded with one command.
Billing periods are listed oldest first with the action each one needs, so a session starts on the work rather than on a search. The period still collecting data says so and asks for nothing.
Production, consumption, self-consumption and grid exchange for the selected period, with a Sankey diagram tracing every kWh from producer to participant. Filter by participant, or change resolution down to the interval.
Every participant's invoice for the period with its metering completeness, status, total and PDF state — and batch actions for the steps you would otherwise repeat once per household.
Tariffs are versioned series, not editable rows. Each version keeps its own bands and validity, so an invoice raised last year still prices at last year's rate — and the chart shows exactly when each change took effect.
Quick start
Pick the path that matches where you want it to live. All three pull prebuilt images from GHCR.
Starts the stack and seeds a full community — participants, meters, tariffs and readings — so there is something to look at.
# clone, then: scripts/start-demo-environment.sh # → http://localhost:8080 # admin@openzev.local / admin1234
The full stack — frontend, API, Celery worker and beat, PostgreSQL and Redis — with no demo data.
docker compose up -d --build # frontend :8080 # api :8001 docker compose down
Deploys frontend, backend, worker and beat with an Ingress and a media PVC. Bring your own PostgreSQL and Redis.
helm repo add openzev \ https://splattner.github.io/openzev helm repo update helm install openzev openzev/openzev \ -n openzev --create-namespace
Under it
Nothing here is exotic. It is meant to still start in three years.
OpenZEV is built for self-hosters who enjoy running their own stack, and is shipped as-is with no warranty. Check your imports, your tariffs and your invoices before they reach a participant — responsibility for what the numbers say stays with you.
It is also built with substantial AI assistance, down to the specs, ADRs and user documentation. Some choices will look unconventional to an experienced team. That is the trade the project makes: it is optimised for learning, experimentation, and running one real ZEV — not for enterprise process.