f97.beERIK SCHMIDT’S WORKSHOP
← All terms

Glossary · Energy industry & energy transition

Feed-in forecast

Feed-in Forecast wind energy photovoltaics forecast error electricity market Balancing group grid operation

A feed-in forecast is the prediction of how much electrical power a generation plant, a group of plants or an entire generation park will feed into the power grid at specific points in time. It therefore does not refer to the installed capacity, but to the feed-in actually expected. For wind and photovoltaic plants, it depends mainly on the weather. For conventional power plants, biomass plants, hydropower plants or storage units, schedules, fuel availability, technical restrictions, market prices and operating decisions also play a role.

The forecast is stated as power in kilowatts, megawatts or gigawatts for individual time intervals, today predominantly for quarter-hours in the European electricity market. If it is summed up over a period, the result is a quantity of energy in kilowatt-hours, megawatt-hours or gigawatt-hours. An example: a photovoltaic plant can briefly deliver a lot of power at midday and still generate a different quantity of energy over the day than a wind plant that feeds in less, but for longer. A feed-in forecast therefore always requires a temporal resolution. For grid operation and trading, what matters is not only how much electricity is generated in a day, but above all when.

A forecast is neither availability nor controllability

The feed-in forecast is often confused with the technical availability of a plant. Availability only means that a plant is in principle ready for operation. A wind turbine can be available and still feed in hardly anything because the wind is weak. A fully functional photovoltaic plant delivers nothing at night. Conversely, a plant can feed in less because of maintenance, grid congestion or a fault, even though the weather would lead one to expect more.

The forecast should also not be equated with controllability. A gas power plant can keep to a schedule as long as fuel, technical minimum load and grid conditions allow it. Wind and solar plants can be curtailed, but they cannot replace missing wind or missing irradiance. For weather-dependent generation, the forecast therefore describes an expected feed-in under certain natural conditions and operational assumptions. It is not firm capacity and does not directly state how much power is reliably available in times of scarcity.

The schedule must also be distinguished. It is a binding, or at least operational, notification of what power is to be generated or consumed at what times. The feed-in forecast is a basis for schedules, but not the same thing. Between the two lie market decisions, balancing group responsibility, plant control and, where applicable, interventions by the grid operator.

Data, models and forecast horizons

Feed-in forecasts are built from several data sources. For wind energy, these include wind speed at hub height, wind direction, air density, temperature, turbulence, icing risk and site characteristics. For photovoltaics, they include global irradiance, cloud cover, temperature, orientation, tilt, shading, snow or soiling, and the technical condition of the modules. In addition there are master data such as installed capacity, location, plant type and grid connection point, as well as historical feed-in values. Modern forecasting systems combine numerical weather models with measurement data, physical plant models and statistical or learning-based methods. The transmission system operators, for example, calculate wind and solar feed-in with their own forecast models based on meteorological input data.

The purpose of the forecast also changes with the forecast horizon. Assessments over several days serve operational planning, forward trading and the estimation of reserve needs. Day-ahead forecasts are decisive for the day-ahead market. There, large quantities of electricity are traded for the following day. Since the delivery day of 1 October 2025, this has been done in European market coupling in quarter-hours instead of hours. Intraday forecasts are updated continuously as soon as new weather model runs, measured values or plant information become available. Very short-term forecasts in the range of minutes to hours, often based on live measurements or satellite and cloud images, help system operation to detect deviations early.

How good a forecast is does not depend only on the model. It also matters whether the master data are correct, whether measured values arrive promptly, whether shutdowns are reported, and whether grid-related curtailments are cleanly separated from weather-related underproduction. If a plant is recorded in the database at the wrong location or with outdated capacity, it permanently creates false expectations. The deviation between forecast and actual feed-in is measured as the forecast error, often as a mean absolute or root-mean-square error, relative to installed capacity. Because independent individual errors partly offset one another, the relative error for a large area is generally significantly smaller than for a single plant.

Publication obligations in the EU

In the EU, feed-in forecasts are also subject to transparency obligations. Under Article 14 of Regulation (EU) No 543/2013, the transmission system operators determine for their control areas a forecast of wind and solar power generation in megawatts per bidding zone and market time unit of the following day. This forecast must be published no later than 18:00 Brussels time on the previous day. During intraday trading it must be updated regularly, at least once at 08:00 on the delivery day. The obligation applies only to Member States with more than 1 percent, or bidding zones with more than 5 percent, wind or solar feed-in per year. The data appear on the ENTSO-E transparency platform. They are aggregated system forecasts and do not replace the plant-specific forecasts of operators or direct marketers.

Significance for the market and the balancing group

In the electricity market, feed-in and withdrawal must be balanced in accounting terms in every quarter-hour. Each feed-in and withdrawal point is assigned to a balancing group. If a wind or solar plant feeds in more or less than forecast and traded, an imbalance arises. Physically, the transmission system operators offset it with balancing energy; economically, it is settled via imbalance energy. Good forecasts therefore reduce not only uncertainty but also the cost risks for direct marketers, suppliers and balancing group managers.

In day-ahead and intraday trading, forecasts determine which quantities are offered or bought back. If a lot of solar power is expected for the afternoon, this affects prices and the dispatch of other plants. If cloud cover develops differently than expected, market participants have to correct their positions. The closer the time of delivery comes, the more accurate the forecast becomes, but the less time remains to react. This is a practical trade-off between forecast quality, trading liquidity and operational feasibility.

Significance for the grid and redispatch

Grid operators have to assess where in the grid power is fed in and whether lines, transformers or voltage maintenance reach their limits. A Germany-wide wind forecast is not sufficient for this. Transmission system operators need spatially resolved forecasts for regions with a lot of wind feed-in. Distribution system operators need to know whether solar power raises the voltage too much in particular local grids. The same quantity of energy can be unproblematic or a burden for the grid, depending on location and time.

Redispatch is also based on forecasts. If expected feed-in at one location aggravates a grid congestion, grid operators curtail plants or ramp up others. In Germany, Redispatch 2.0 distinguishes two balancing models for generation plants and storage units. In the planned-value model (Planwertmodell), the operator responsible for dispatch supplies planning data and feed-in forecasts. In the forecast model (Prognosemodell), the grid operator estimates the feed-in itself and determines the lost energy (Ausfallarbeit). The balancing-related compensation is intended to place the affected balancing groups as if no redispatch measure had taken place. Under the determination procedure BK6-23-241 of the Bundesnetzagentur, more plants in the distribution grid are to switch to the planned-value model step by step. Forecast errors can result in interventions being planned too late, too strongly or at the wrong place. The cause then often lies not only in the weather, but in the interplay of forecast, grid model, reporting obligations and processes.

Typical misunderstandings

A common misunderstanding is that weather-dependent generation is unpredictable. That is not true. Wind and solar feed-in can be forecast fairly accurately, above all in aggregate across many plants and large regions. Uncertainty does not disappear entirely, however. Fog and high-fog situations, fast cloud movements, Saharan dust, snow on modules, storm fronts with storm shutdowns or a frontal passage shifted in time can trigger significant deviations. A forecast is therefore not a promise, but an expectation with an error range. Probabilistic forecasts make this range explicitly visible, for example through quantiles.

Equally simplistic is the assumption that better forecasts would solve all integration issues of renewable energy. Forecasts improve planning and reduce the need for balancing, but they do not generate electricity in times of little wind and sun. If there is hardly any wind for several days and photovoltaics are weak at the same time, a good forecast helps to recognise this situation early. It does not replace flexibility, storage, controllable power plants, load shifting or sufficient grid infrastructure.

A third mistake is to see forecast errors only as a technical modelling problem. Many deviations have organisational causes: faults are not reported in time, curtailments are documented inconsistently, master data are outdated, measured values arrive late or market actors have different levels of information. A forecast is never used in isolation, but is embedded in trading deadlines, balancing rules, data obligations and grid processes.

What the feed-in forecast makes visible

The feed-in forecast shows how precisely a power system has to be organised in time and space when large parts of generation depend on the weather. In the past, power plant dispatch was planned mainly according to expected demand, fuel costs and technical power plant data. With high shares of wind and photovoltaics, the supply side also becomes part of short-term coordination. Forecasts are thus no longer auxiliary information for individual operators, but a basis for price formation, congestion management, reserve planning and system operation.

The term also helps to keep energy quantity and system contribution apart. High annual solar power generation says little about how much photovoltaics contributes to covering load on a dark winter evening. A wind forecast for next week may lead one to expect a lot of energy and yet contain individual hours with low feed-in. For the residual load, that is, the load after deduction of weather-dependent feed-in, the temporal profile of the forecast is what matters.

The feed-in forecast thus links weather, technology, market and grid. It reduces uncertainty but does not eliminate it. It enables trading and operation, but does not remove the dependence on wind, sun and grid capacity. Used precisely, the term describes an expected feed-in at specific times and locations. It does not explain security of supply on its own, but it shows what information a power system with a lot of weather-dependent generation needs in order to remain plannable.

Sources

Related glossary entries

Edited by eCMS on 11.11.2026 19:42

© 2026 Erik Schmidt. All rights reserved. Quotations with a link to the source are welcome.