German Grid Congestion Monitor (Redispatch)

Structured, diff-based tracking of Germany's official redispatch measures — the only country-wide, day-level, machine-readable measure of real grid congestion. Source is the redispatch publication of the German Federal Network Agency (Bundesnetzagentur, SG HoBA) on netztransparenz.de.

Field names and values are in German (grundDerMassnahme, congestieStatus: "hoch", anweisenderUENB, …) because that is what the source publishes and what a German grid audience expects. Listing text and this page are in English for reach; the local search terms are in the Keywords section and in the Häufige Fragen (deutsch) section at the bottom.

When should an AI agent use this?

• "Where were the worst grid congestion hotspots in the German transmission grid last month?"
• "Which plants was Amprion ordered to curtail, and how often?"
• "How much congestion-relevant redispatch energy did each transmission operator shift, and did it escalate?"
• "Which redispatch measures exceeded 1000 MW, and what triggered them?"
• "Were output cuts or offsetting increases dominant in September 2026?"
• "Is transmission operator X's congestion level worse than last month?"

What this Actor does

• Downloads the official redispatch CSV through the publication page's download export (an ASP.NET WebForms postback). No API key, no login, no guessing a rotating filename.
• Normalises every measure: start/end as ISO timestamps, duration, MW (average and maximum), MWh, reason, direction, the ordering and requesting transmission operators, the affected plant, the primary energy carrier and whether the row is a genuine congestion measure.
• Separates congestion causes from commissioning tests: rows with reason Probefahrt (probe starts and function tests of reserve plants) are flagged istEngpassrelevant: false and excluded from every congestion metric, while remaining visible in the totals.
• Additionally delivers one monthly profile per transmission operator and calendar month, with the total and congestion-relevant energy, the energy per covered calendar day, measure and plant counts, the direction split, the energy-carrier mix, the main reason and two derived congestion statuses.
• Detects new and changed measures between runs via a durable named key-value store.
• Fires a dedicated pay-per-event alert when a monthly profile drops into a stricter status band than the previous run.
• Caps delivery (maxMassnahmen, default 500) and orders by significance, so a country-wide overview is cheap; a free summary always covers the full filtered set, and records outside the cap are re-detected on the next run instead of being dropped.

Scope and limitations

This is a transmission-level indicator (high voltage, 220/380/400 kV) covering the four German transmission system operators. It says nothing about connection-level capacity at distribution level: Germany has no country-wide machine-readable Netzengpassgebiet or connection queue per region. The Apify Store searches for Netzengpass, Netzausbaugebiet and netzkonflikt returned zero direct competitors on 2026-09-29.

Congestion status

The statuses are documented threshold classifications of the published energy — not a forecast, not a queue statement, and not a statement about any individual plant's connectability. Both are defined in one place (classifyCongestionStatus and classifyCongestionStatusTagessatz) so they stay auditable.

congestieStatus — on the congestion-relevant energy of the calendar month (excluding commissioning tests):

Status	Congestion-relevant energy per month
keine_engpassmassnahmen	0 MWh (only tests, or no measures)
gering	below 150,000 MWh
erhoeht	150,000 – 500,000 MWh
hoch	500,000 – 1,000,000 MWh
kritisch	above 1,000,000 MWh

congestieStatusTagessatz — on the energy per covered calendar day, which is the fair comparison for a current, still incomplete month:

Status	Congestion-relevant energy per day
keine_engpassmassnahmen	0 MWh
gering	below 5,000 MWh
erhoeht	5,000 – 15,000 MWh
hoch	15,000 – 30,000 MWh
kritisch	above 30,000 MWh

abgedeckteKalendertage is the denominator: for the running month it is the calendar days elapsed so far (the source is still incomplete), and for a completed month it is the month length. That way a partial month stays comparable without a month that simply ended quietly looking artificially intense. Both scales are calibrated on the 2026-01 to 2026-09 data set (observed range roughly 19,000 to 1,587,000 MWh per operator-month, and 620 to 52,900 MWh per day). The monthly escalation alert uses the absolute status.

Other source limitations the Actor does not hide

• Day-level granularity only. The agency deliberately publishes daily averages and maxima ("Mittelwerte und Maximalwerte beziehen sich auf die jeweiligen Tagesanteile"), not the second-by-second curve.
• Probefahrt is not a congestion measure. The source explains these are probe starts, probe calls and function tests of domestic reserve plants, performed to maintain operational readiness and explicitly not triggered by a congestion. They made up about 5.9% of published 2026 energy. Every congestion figure excludes them; gesamtArbeitMWh, massnahmenJeGrund and arbeitMWhJeGrund keep them visible, and nurEngpassursachen can drop them from the delivered rows too.
• Measures crossing a month boundary (104 in the 2026 set) are attributed to their start month.
• dauerStunden is a naive local-time difference. Timestamps carry no offset, so a measure spanning a daylight-saving switch can be off by one hour. With typical durations of half an hour to a few hours this affects at most a couple of days per year.
• No sub-plant geography. Grid assignment is carried via the requesting operator in anforderndeBetreiber; anweisenderUENB is the control-zone assigner. The publication contains no coordinates per plant.
• No simultaneous peak. summierteMittlereLeistungMW sums daily averages over the month and therefore adds up across time. maximaleLeistungMW is the peak; durchschnittlicheMittlereLeistungMW is the average per measure.
• The source has no period filter. The page's download export always returns the year-to-date set, whatever the date fields in its form say. zeitraum therefore filters inside the Actor, on day level, and that is stated here rather than passed off as a source feature.
• No foreign exchange volumes. For cross-border redispatch and countertrade, the source publishes only the German part.

Input

Field	Type	Description
anweisenderUENB	array of strings	Optional: 50Hertz, Amprion, TenneT DE, TransnetBW (case-insensitive). Empty = all.
grundDerMassnahme	array of strings	Optional: official reason, e.g. Strombedingter Redispatch, Probefahrt, Spannungsbedingter Redispatch, Strombedingter Countertrade DE-DK1. Empty = all.
richtung	array of strings	Optional: Wirkleistungseinspeisung reduzieren or Wirkleistungseinspeisung erhöhen. Empty = both.
primaerenergieart	array of strings	Optional: Konventionell, Erneuerbar, Sonstiges. Empty = all.
betroffeneAnlage	string	Optional: substring of the official plant name, e.g. Mehrum, Börse, Dörpen. Case-insensitive with German umlaut folding: Doerpen and dörpen both find Dörpen-West.
nurEngpassursachen	boolean	Default false. If true, drop rows that are not congestion measures (in practice Probefahrt). The statuses and the summary exclude them regardless.
minLeistungMW	number	Optional: minimum for the maximum capacity in MW.
minArbeitMWh	number	Optional: minimum for the total energy in MWh.
maxMassnahmen	number	Default 500, minimum 1. Cap on delivered measure records; the most significant come first (highest energy, then highest peak, then most recent). Records outside the cap stay out of the change-detection baseline, so the next run detects them again and delivers the next batch — nothing is silently lost, you just receive large sets in batches. Set a high value for everything at once.
zeitraum	string	aktuelles_jahr (default), letzte_30_tage, letzte_7_tage.
liefermonatsprofil	boolean	Default true: also deliver monthly profiles per operator.
nurNeueOderGeaenderte	boolean	Default true: only deliver and charge new/changed measures. The baseline is shared by all runs of this Actor, so it is not a per-user "first run" switch.

Output

Every dataset item carries a typ field to tell the three item types apart.

redispatch_massnahme — one individual measure:

{
  "typ": "redispatch_massnahme",
  "massnahmeId": "15.06.2026|02:00|15.06.2026|06:00|TenneT DE|Wirkleistungseinspeisung reduzieren|Strombedingter Redispatch|Mehrum|Erneuerbar@1500.5~1955.85~5999.4",
  "massnahmeBasis": "15.06.2026|02:00|15.06.2026|06:00|TenneT DE|Wirkleistungseinspeisung reduzieren|Strombedingter Redispatch|Mehrum|Erneuerbar",
  "land": "DE",
  "beginn": "2026-06-15T02:00:00",
  "ende": "2026-06-15T06:00:00",
  "zeitzone": "CEST",
  "dauerStunden": 4,
  "monat": "2026-06",
  "grundDerMassnahme": "Strombedingter Redispatch",
  "richtung": "Wirkleistungseinspeisung reduzieren",
  "mittlereLeistungMW": 1500.5,
  "maximaleLeistungMW": 1955.85,
  "gesamtArbeitMWh": 5999.4,
  "anweisenderUENB": "TenneT DE",
  "anforderndeBetreiber": ["TenneT DE", "TransnetBW"],
  "betroffeneAnlage": "Mehrum",
  "primaerenergieart": "Erneuerbar",
  "istEngpassrelevant": true,
  "isNeuSeitLetztemLauf": true,
  "isGeaendertSeitLetztemLauf": false,
  "vorherigeGesamtArbeitMWh": null,
  "quellenUrl": "https://www.netztransparenz.de/de-de/Systemdienstleistungen/Betriebsfuehrung/Redispatch",
  "abgerufenAm": "2026-09-29T12:00:00.000Z"
}

massnahmeBasis is the event slot — time window, operator, direction, reason, plant and energy carrier — taken from the source fields, never from the CSV row number, so a record stays identifiable when the source reorders or inserts rows. massnahmeId adds the value tuple (<basis>@<average>~<peak>~<energy>, a missing value as n), with #2, #3, … only when the source publishes exact duplicates. Because the identity is the slot plus the values and not a position within the group, a record keeps the same ID when the source adds or removes a neighbouring row in the same window, and a corrected value shows up as isGeaendertSeitLetztemLauf on the same slot instead of as a brand-new record.

monatsprofil — aggregate per operator and month:

{
  "typ": "monatsprofil",
  "profilId": "profil-TenneT DE|2026-09",
  "land": "DE",
  "uebertragungsnetzbetreiber": "TenneT DE",
  "monat": "2026-09",
  "anzahlMassnahmen": 834,
  "anzahlMassnahmenEngpassrelevant": 822,
  "anzahlBetroffeneAnlagen": 89,
  "abgedeckteKalendertage": 30,
  "gesamtArbeitMWh": 1647382,
  "arbeitMWhEngpassrelevant": 1586788,
  "tagessatzArbeitMWh": 52893,
  "summierteMittlereLeistungMW": 205166.61,
  "durchschnittlicheMittlereLeistungMW": 1930.05,
  "maximaleLeistungMW": 2368,
  "anzahlReduzierungen": 653,
  "anzahlErhoehungen": 181,
  "richtungVerteilung": { "Wirkleistungseinspeisung reduzieren": 653, "Wirkleistungseinspeisung erhöhen": 181 },
  "congestieStatus": "kritisch",
  "congestieStatusTagessatz": "kritisch",
  "congestieStatusVorherigemLauf": "hoch",
  "istEskaliert": true,
  "hauptGrund": "Strombedingter Redispatch",
  "energietraeger": { "Erneuerbar": 1119131.91, "Konventionell": 257091.79, "Sonstiges": 201383.5 },
  "quellenUrl": "https://www.netztransparenz.de/de-de/Systemdienstleistungen/Betriebsfuehrung/Redispatch",
  "abgerufenAm": "2026-09-29T12:00:00.000Z"
}

zusammenfassung — one free item per run covering the full filtered set (also when delivery was capped): reporting period, measure and congestion-measure counts, total and congestion-relevant energy, maximum capacity, overall status, per-operator counts and energy (both variants) with statuses, and counts and energy per reason and per energy carrier. It reports anzahlVorAbschneidung, anzahlGeliefert and abgeschnitten so a truncated delivery can never look complete.

Use cases

• Project developers and energy consultants who need to know whether their region is under network pressure, and how hard.
• Transmission operator and regulatory watchers tracking the distribution of redispatch load per control zone, trends and escalations.
• Energy traders and analysts comparing curtailments against offsetting measures.
• Journalists and analysts who need a citable monthly figure for German grid congestion.
• Data pipelines that need stable JSON instead of parsing the source CSV themselves.

Pricing

This Actor uses Apify Pay-Per-Event pricing.

• Actor Start: Apify default.
• redispatch-massnahme: $0.001 per delivered measure, once the shared baseline has been seeded.
• monatsprofil: $0.01 per delivered monthly profile.
• congestie-meldung: $0.002 extra per measure that is new or changed since the previous run.
• congestie-eskalation: $0.05 per monthly profile that dropped into a stricter status band than the previous run.
• Summary: free, no separate event.

How the billing actually works. The change-detection baseline lives in a named store shared by every run of this Actor, and it is seeded once before publication. It is therefore not a per-user "first run is free" switch: your runs are compared against that shared baseline from the start. This is deliberate and cheap in practice:

• With the default nurNeueOderGeaenderte: true, a run delivers only measures that are new or changed since the shared baseline. Once the baseline covers the year set, a run on unchanged data delivers 0 measure records, so you pay only monatsprofil (about $0.08 for the eight operator-months). That is the normal monitoring cost.
• With nurNeueOderGeaenderte: false, a run delivers the full filtered set subject to maxMassnahmen and bills redispatch-massnahme per delivered record — at most $0.50 with the default cap of 500.
• A one-off complete export of the year set (~16,900 records) therefore costs about $16.90; set maxMassnahmen: 20000 for that. Raising maxMassnahmen later also works, because records outside the cap were never added to the baseline.
• The free summary always covers the full filtered set, so the country-wide picture never costs more than the delivered records.

All events are charged once per run with an aggregate count.

Legal and source

Source: redispatch publication of the HoBA special unit (transmission grid) of the Bundesnetzagentur, published on netztransparenz.de and produced by the four German transmission system operators.

https://www.netztransparenz.de/de-de/Systemdienstleistungen/Betriebsfuehrung/Redispatch

Only official, publicly accessible data is processed. No credentials are bypassed, no private areas are scraped, and no personal data is collected — the source names industrial plants, exchanges and operators, never natural persons. The Bundesnetzagentur licence governs reuse of the raw data; this Actor delivers normalised aggregate and metric objects.

The Actor gives no project, investment or legal advice. The current official Bundesnetzagentur publication is always authoritative.

FAQ

How current is the source? The publication is updated day by day and grows continuously. The export returns the year-to-date set at the moment of retrieval. Two runs on the same day normally return the same set.

Why is there a maxMassnahmen cap? Because a full year set is around 17,000 records. Without a cap, one default run would deliver and later bill for all of them. The cap keeps a first look cheap, the summary still covers everything, and abgeschnitten tells you when you are seeing a subset. Records outside the cap are deliberately left out of the change-detection baseline, so the next run picks them up again: you receive a large set in batches rather than losing the tail.

Is the first run free? No, not for you personally — and the listing never promised otherwise once you account for how the baseline works. The change-detection baseline is shared by every run of this Actor and is seeded before publication, so your runs compare against it immediately. The good news is that the default mode makes this almost free anyway: once the baseline covers the data, a run on unchanged data delivers 0 measure records and bills only the monthly profiles.

Why doesn't the Actor report a queue or connection capacity? Because Germany has no country-wide machine-readable source for it. The Netzausbaugebiet/Netzengpassgebiet disclosures under § 13 EnWG are published per distribution operator, mostly as a map or PDF, and the § 13a EnWG curtailment volumes sit behind a login. Redispatch is the only bundesweit day-current, officially quantified congestion signal.

Why are commissioning tests excluded from the status? Probefahrt covers probe starts and function tests of reserve plants. The source states they are not triggered by a congestion, yet they were about 5.9% of 2026 energy. Leaving them in would inflate the congestion picture. They remain in gesamtArbeitMWh and in the per-reason breakdown, so nothing is hidden.

Why two congestion statuses? The absolute monthly status answers "how much congestion happened this month". It is misleading for the current, still incomplete month. The per-day status answers "how intense is it here, right now" and stays comparable across months of different lengths. abgedeckteKalendertage shows the denominator.

What is the difference between anweisenderUENB and anforderndeBetreiber? The operators apply the Anschluss-ÜNB principle: publication is the responsibility of the operator in whose control zone the plant sits — that is anweisenderUENB. anforderndeBetreiber lists who requested the measure, possibly several operators or foreign partners (RTE, Statnett, APG, Swissgrid, Elia, CEPS, EnDK). Multi-value entries are &-separated in the source and split into an array here.

What does Börse in betroffeneAnlage mean? Since 2019 the source also publishes the TSOs' exchange trades. Such rows have no local plant reference — they occur for voltage-driven redispatch, where there is no local countermeasure. They count in the aggregates.

Why does capacity or energy sometimes come back as null? Because the source leaves the value out for some rows (e.g. n.v.). The Actor writes null rather than 0 so a missing value stays distinguishable from a real zero. minLeistungMW and minArbeitMWh exclude such rows instead of letting them through.

How does the change detection survive a growing source? It stores a compact snapshot of every measure ever seen, in a named key-value store shared with scheduled runs. Entries older than 400 days are pruned, because the source only ever publishes the running year — without pruning the snapshot would eventually exceed the key-value record limit.

Häufige Fragen (deutsch)

Wann wird die Quelle aktualisiert? Die Publikation ist tagesscharf und wächst laufend weiter. Der Export liefert den Bestand seit Jahresbeginn bis zum Abrufzeitpunkt. Zwei Läufe am selben Tag liefern in der Regel denselben Stand.

Warum gibt es die Grenze maxMassnahmen? Weil der Jahresbestand rund 17.000 Datensätze umfasst. Ohne Grenze würde ein Standardlauf alle liefern und später auch alle abrechnen. Die Grenze hält den ersten Blick günstig, die Zusammenfassung deckt weiterhin alles ab, und abgeschnitten zeigt an, wenn nur ein Teil geliefert wurde. Datensätze außerhalb der Grenze kommen bewusst NICHT in die Änderungs-Baseline: der nächste Lauf erkennt sie wieder und liefert die nächste Charge. Große Mengen kommen also in Portionen, statt dass der Rest verloren geht.

Ist der erste Lauf gratis? Nein, nicht für Sie persönlich. Die Änderungs-Baseline liegt in einem Store, den alle Läufe dieses Actors teilen, und wird schon vor der Veröffentlichung befüllt; Ihre Läufe werden also sofort damit verglichen. Mit der Standardeinstellung ist das trotzdem fast kostenlos: sobald die Baseline die Daten abdeckt, liefert ein Lauf ohne Neuigkeiten 0 Datensätze und berechnet nur die Monatsprofile.

Warum werden Probefahrten nicht als Engpass gezählt? Probefahrt sind Probeabrufe, Funktionstests und Testfahrten inländischer Reservekraftwerke. Die Quelle stellt ausdrücklich klar, dass sie nicht durch einen Engpass ausgelöst werden — 2026 machten sie trotzdem rund 5,9 % der veröffentlichten Arbeit aus. Sie mitzuzählen würde das Engpassbild aufblähen. Sie bleiben in gesamtArbeitMWh und in der Aufschlüsselung nach Grund sichtbar.

Warum zwei Congestion-Status? Der absolute Monatsstatus beantwortet „wie viel Engpass war in diesem Monat". Für den laufenden, noch unvollständigen Monat ist er irreführend. Der Tagesstatus beantwortet „wie intensiv ist es hier gerade" und bleibt über Monate unterschiedlicher Länge vergleichbar. abgedeckteKalendertage nennt den Nenner.

Worin unterscheidet sich anweisenderUENB von anforderndeBetreiber? Die ÜNB wenden das Prinzip des Anschluss-ÜNB an: Für die Veröffentlichung ist der Betreiber zuständig, in dessen Regelzone die Anlage liegt — das ist anweisenderUENB. anforderndeBetreiber nennt, wer die Maßnahme angefordert hat, gegebenenfalls mehrere Betreiber oder ausländische Partner (RTE, Statnett, APG, Swissgrid, Elia, CEPS, EnDK).

Was bedeutet Börse bei betroffeneAnlage? Die Quelle veröffentlicht seit 2019 auch Börsengeschäfte der deutschen ÜNB. Solche Zeilen haben keinen örtlichen Anlagenbezug, etwa bei spannungsbedingtem Redispatch, wo es keine räumliche Gegenmaßnahme gibt. Sie zählen in den Aggregaten mit.

Wie finde ich eine Anlage mit Umlaut? Die Namenssuche faltet Umlaute auf die deutsche ASCII-Konvention: Doerpen findet Dörpen-West, und dörpen ebenfalls. Eine Suche nach Dorpen findet bewusst nichts — einfaches o und oe bleiben getrennt.

Kostet ein Lauf über das ganze Jahr pauschal 16,75 $? Nein. Das ist der Preis für die vollständige Auslieferung von rund 16.900 Einzeldatensätzen ($0.001 je Datensatz), also ein Einmalextrakt mit maxMassnahmen: 20000. Mit den Standardeinstellungen (maxMassnahmen: 500, nur neue/geänderte Datensätze) liegt ein Lauf bei höchstens $0.50 plus Monatsprofile, und die kostenlose Zusammenfassung liefert trotzdem das Gesamtbild. Ein reiner Überwachungslauf ohne Neuigkeiten kostet nur die Monatsprofile (rund $0.08).

Related Actors

• ../nl-netcongestie-monitor — same netcongestie pattern for the Netherlands (live, same portfolio author).
• ../uk-netcongestie-grid-capacity-tracker — transmission connection queue for Great Britain (live, same portfolio author).

Keywords

Netzengpass, Redispatch, Stromnetz, Übertragungsnetz, Übertragungsnetzbetreiber, Engpass, netztransparenz, Bundesnetzagentur, 50Hertz, Amprion, TenneT DE, TransnetBW, Regelleistung, Stromdaten, Kraftwerk, Einspeisemanagement, Gegenmaßnahmen, Probefahrt, Netzausbaugebiet, netzkonflikt, Germany electricity grid, grid congestion, redispatch, transmission system operator, curtailment, countertrade

Changelog

0.1.0

• Initial release.
• Live-verified download of the official redispatch CSV through the WebForms export endpoint, with retry, an HTML body check against error pages, and automatic discovery of the download module so a layout change cannot silently select the wrong dataset.
• Monthly aggregation per transmission operator with two documented congestion statuses (absolute and per day).
• Commissioning tests (Probefahrt, ~5.9% of 2026 energy) are flagged and excluded from every congestion metric while staying visible in the totals; nurEngpassursachen can filter them from the delivered rows.
• Filters for operator, reason, direction, energy carrier, plant (with German umlaut folding), capacity, energy and period.
• maxMassnahmen cap with significance ordering and transparent truncation reporting. Records outside the cap are deliberately not added to the change-detection baseline, so a later run detects and delivers them too: large sets arrive in batches instead of losing the tail.
• Change detection is keyed on the event slot (window, operator, direction, reason, plant, energy carrier) rather than on a full row ID, so a source row added to or removed from a window no longer makes a neighbouring row look new, and a corrected value is reported as changed on the same slot.
• Durable, size-bounded baseline in a named key-value store (compact slot form, pruned to 400 days, tolerant of older stored shapes), plus an escalation alert for monthly profiles.
• Free per-run summary over the full filtered set, with abgeschnitten, anzahlVorAbschneidung and anzahlGeliefert from a single source so the log, the summary and the billing cannot diverge.