Skip to main content

Availability Forecasts Methodology

Power availability forecasts built from outages (UMMs) and Kpler modeling adjustments methodology

Written by Raphael Khan

Power availability represents the generation capacity available on the grid after accounting for outages. It is a key indicator of supply conditions, system tightness, and power price formation. Availability is live and reflects the latest outage data at query time.

Kpler provides two availability metrics:

  • Raw Availability (by fuel type & by unit) — derived from reported outages

  • Kpler Availability — an enhanced forecast that applies proprietary modeling to account for underreporting and outage behavior

Raw Availability

Raw availability represents the available capacity, computed as:

Availability = Installed Capacity – Unavailability

We provide the availability at both fuel-type and unit levels.

Fuel-Type Availability

Fuel-type availability follows a top-down approach:

  • Installed capacity per fuel type is sourced from TSOs, ENTSO-E, and other validated reference datasets.

  • Valid reported outages are aggregated to construct an unavailability curve.

  • Availability is derived at the fuel level

Unit-Level Availability

  • Availability is similarly computed for each generation unit reporting outages.

  • The availability of a production unit (plant) is calculated as the sum of the availability of its generation units.

  • This unit-level data can then be aggregated by zone and fuel type to derive a bottom-up view of fuel-type availability.

Note : the sum of unit-level (bottom-up) availability may not always match top-down fuel-type availability exactly.

Differences can arise from smaller generation units (typically below 100 MW) that can't be individually tracked due to missing generation data, outages, or EIC codes - these units are still accounted for in installed capacity figures used in top-down calculations, but are absent from the bottom-up sum.

Bottom-up availability is queryable directly via the Fuel-Type Availability endpoint (see Accessing the data) with aggregation=bottomup.

Coverage

Outages are collected from multiple providers - see coverage here. For each unit reporting outages, we build an unavailability curve, from which availability is derived.

Raw availability therefore covers all zones, fuels, and units available in these sources.

Kpler Availability

Kpler Availability is an enhanced availability forecast, available at the fuel-type level. It builds on raw availability by applying proprietary modeling adjustments to account for underreporting - not all outages are reported in advance, which can lead to overestimated availability.

To address this, derate raw availability using an unavailability factor based on historical patterns across different horizons, producing:

  • a central estimate

  • upper and lower bounds reflecting the minimum and maximum scenarios historically observed by the model for each horizon - the central curve will not necessarily sit at the midpoint of this interval

Kpler Availability = Raw Availability + Model Adjustments

Coverage

Country

Coverage

🇫🇷 France

Nuclear, Fossil gas

🇩🇪 Germany

Fossil hard coal, Fossil brown coal/Lignite, Fossil gas

🇮🇹 Italy

Fossil gas

🇳🇱 Netherlands

Fossil hard coal

🇵🇱 Poland

Fossil hard coal, Fossil gas

🇪🇸 Spain

Fossil gas, Nuclear

🇬🇧 United Kingdom

Fossil gas

Note : when a model update is released, the latest revision applies by default to all dates - past and future, so historical estimates stay consistent with the current model. Previous revisions remain accessible via API with the modelVersion parameter (see Accessing the data), letting you reproduce our exact output as it stood at a given point in time - in combination with asOf, which controls which outages were known at that time. Model updates are communicated in advance.

Kpler Availability vs Raw Availability

The examples below compare raw availability derived from reported outages and Kpler availability forecast (both as of a given date) with the actual availability.

Legend:

  • Black line solid - Actual availability

  • Dark blue dashed - Raw availability derived from outages

  • Light blue band dashed - Kpler availability forecast (central curve + confidence interval)

FR Nuclear - forecasts as of January 1st 2025 (link here)

DE Fossil Hard Coal - forecasts as of June 1st 2025 (link here)

Across both examples, raw availability consistently overestimates capacity, while Kpler availability provides a more accurate forecast.

The confidence interval provides a realistic range of outcomes, especially in forward periods where availability may be over- or under-estimated.

Accessing the data

You can access our availability forecasts through :

Kpler Terminal

  • Fuel-type availability

  • Unit availability

Kpler Power API

Using provider=kpler (our default provider mapping):

  1. With methodology=kpler (default): if Kpler Availability is available for the zone/fuel (see coverage above), returns low, central and high curves; otherwise falls back to raw availability (central curve only).

  2. With methodology=raw: always returns raw availability (central curve only), still benefiting from our provider mapping - even where Kpler Availability coverage exists. Useful if you want our best outage-provider selection without the model correction applied on top.

This endpoint returns all generation and production units for a selected zone and fuel type, along with their availability.

Related Articles
Did this answer your question?