Use day-ahead electricity prices

Electricity does not cost the same all day. On European wholesale markets a price is set for every hour of tomorrow, published the afternoon before — the day-ahead market. Between a cheap overnight hour and an expensive evening peak the difference is routinely several-fold, and occasionally the price goes negative.

For a fleet that charges overnight and is not in a hurry, moving load a few hours is one of the largest cost levers available.

Flowion collects these forecasts from ENTSO-E, the European transmission operators' platform, and serves them through the API.

Fetching prices#

Two ways to say which market you mean:

# By environment — uses its configured price zone
curl "https://api.flowion.io/v1/energy/prices?environment_id=$ENV" \
  -H "Authorization: Bearer $TOKEN"

# By raw bidding zone EIC code
curl "https://api.flowion.io/v1/energy/prices?zone=10YSE-1--------K" \
  -H "Authorization: Bearer $TOKEN"

Give exactly one of zone or environment_id. Optional from and to bound the window — from defaults to now, to to 48 hours after it.

There is also a per-site variant, which matters when one environment spans several markets:

curl "https://api.flowion.io/v1/sites/$SITE/energy/prices" \
  -H "Authorization: Bearer $TOKEN"

Reading a price point#

{
  "period_start": "2026-08-11T02:00:00Z",
  "resolution_minutes": 60,
  "price_amount": "18.44",
  "currency": "EUR"
}

Points come back chronological.

Two things to handle properly:

price_amount is a decimal string, not a number. Parse it as a decimal type. Reading it as a float and rounding to cents will drift once you are summing thousands of intervals.

Prices can be negative. When wind generation exceeds demand, producers pay to offload it. "-5.32" is a real value, and a comparison that assumes positive prices will pick the wrong hour precisely when the saving is biggest.

Do not assume resolution_minutes is always 60 either — European markets are moving toward 15-minute settlement, and the field exists to tell you.

Bidding zones#

A bidding zone is a market area with a single clearing price, identified by an EIC code. Zones are not countries: Sweden has four, Norway five, Germany and Luxembourg share one.

Set the zone on the environment, or override it per site. A fleet spread across zones and priced against one of them will be wrong for every site except that one — often by a lot, because zonal prices diverge most exactly when the grid is stressed.

What you can do with it today#

The forecast is available data, not an automatic behaviour. Nothing in the platform currently shifts charging toward cheap hours by itself.

What you can build on it:

Report on it. Combine transaction energy with the prices that applied to work out what a session or a site actually cost.

Schedule with it. Fetch tomorrow's prices, find the cheapest window long enough for the energy you need, and install a charging limit whose schedule opens up during it and throttles outside it.

Decide with it. Choose when to start a queued session, or whether to accept a discretionary one now.

Price-optimized allocation is designed but not built. Making the load balancer weight its allocation by price — rather than you scheduling around it — is planned work, not current behaviour. See OCPP feature coverage.

A worked pattern#

The common overnight case, in four steps:

1. Fetch prices from now to 48 hours out for the site's zone.
2. Work out how much energy the connected vehicles need, from their transactions' recent meter readings.
3. Find the cheapest contiguous window long enough to deliver it at the rate the circuit allows.
4. Install a default_policy charging limit whose schedule permits a full rate inside that window and a minimal rate outside it.

Re-run it when tomorrow's prices publish, in the afternoon.

Two things to build in from the start. Vehicles taper — a car near full draws far less than its allowance, so a window sized from the ceiling will be too short. And forecasts have an edge: past to, you have no prices, so a schedule must degrade to something safe rather than to nothing.

Availability#

Data comes from ENTSO-E, so it covers European markets. An environment with no price zone configured returns nothing — that is an unset setting, not an error.

Collection runs on a timer. A gap around publication time is normal; treat a missing point as "not yet known", not zero.

Endpoints#

Energy · Charging Limits · Sites & Circuits