Grid-oriented control under §14a EnWG

Two rulings, two chambers

On 23 November 2023 Ruling Chamber 8 of the Federal Network Agency decided what controllable consumer devices save on grid charges¹, and four days later Ruling Chamber 6 decided when the grid operator may dim them in exchange². Both have applied since 1 January 2024.

In the statute it's one deal. Section 14a(1) of the Energy Industry Act (EnWG) lets the agency require agreements on grid-oriented control "in return for grid charge reductions"³, and yet the agency spread the exchange over two chambers: control, including the guaranteed minimum power and documentation, in BK6-22-300, and the three modules of the charge reduction in BK8-22/010-A.

So a question about the control box goes to BK6, one about the price sheet to BK8. Even the penalty proceedings over module 3 carry BK8 case numbers.

Who has to take part

Four groups, one threshold. Heat pumps and cooling systems behind one grid connection are added up per group, so three small heat pumps in a cascade count as one device once their combined rating exceeds 4.2 kW⁴.

Every device commissioned after 31 December 2023 has to take part. The ruling has exactly two exceptions. One covers charging points run by bodies with special road traffic rights, such as police and fire services; the other covers heat pumps and cooling systems that don't heat or cool homes, offices or other rooms people spend time in, in particular units serving essential commercial purposes or critical infrastructure. A company fleet's wallbox is still in, and so is a business's battery. A grid without congestion is no excuse⁴, and commissioning doesn't have to wait until the iMSys and control box are in².

Money isn't all the owner gets. A grid operator may not delay or refuse the connection of a controllable consumer device by pointing to a lack of grid capacity⁴. The discount is only half the deal.

The 4.2 kW come from charging technology. Chamber 6 had consulted on 3.7 kW, but a charging point needs at least 6 amperes per phase, around 4.1 kW on three phases, and a technical safety margin made it 4.2 kW². The same number does two jobs on purpose, as threshold and as minimum power.

Curative: when dimming is allowed

One sentence in the ruling carries this whole chapter.

Annex 1 turns this into a procedure. The trigger is a current- or voltage-related threat or disruption in a network section, which the grid operator establishes through the grid status assessment; it then passes the reduction to the metering point operator without undue delay, and under the annex's presumption that means five minutes at most. Every controllable device in the section is treated alike. Dimming goes only as far as needed, and the way back is in steps⁴. Upstream grid operators have no role under §14a².

VDE FNN, the forum within the electrical engineering association VDE that writes the technical rules for grid operation, confirmed the five minutes in April 2025. It also reads them as the maximum validity of a grid status assessment, so an older one no longer counts⁹.

Does curative mean waiting for the fuse to blow? No. Under the FNN recommendation a grid operator may set its power limit at 80 percent of the permissible continuous current rating or of the permissible voltage drop, which leaves the rest as room to check and react¹⁰.

There's an upward exception. For directly controlled heat pumps and cooling systems rated above 11 kW, the minimum is the connected load times a scaling factor, presumed appropriate at 0.4 until a different recommendation exists, which is roughly 4.2 divided by 11^4,2. In April 2025 VDE FNN recommended sticking with it¹¹. Storage units and wallboxes keep their 4.2 kW, even where that's less than 40 percent of their rating².

So the 40 percent rule covers only two of the four groups: per device under direct control, and as the base of the overall calculation under an EMS⁴.

"There is no preventive dimming" therefore only holds without a footnote from 2029. Until then two regimes run side by side. Since 1 March 2025 grid operators have had to report both each month, per network section, on the VNB-Digital platform^4,12.

The control chain

Once a metering point has an iMSys, control has to run through the smart meter gateway. That's in the statute, with a single exception for the agile rollout that ends with the application update³. In the mandatory rollout the default metering point operator installs the control unit, and the data traffic through it is one of its standard services under the Metering Point Operation Act (MsbG)^13,14.

The ruling cuts responsibility into three pieces. The grid operator assesses the grid and hands over the order. The metering point operator gets it into the metering system. Making the device follow is the owner's job².

Inside the gateway only one party has write access, the gateway administrator. It configures which connections the CLS channel allows, and through that channel the active external market participant (aEMT in German shorthand) talks to the control box; the gateway secures the link without reading the traffic¹⁵. What the control box has to do is set out in a VDE FNN requirements document, currently version 1.5 of August 2025¹⁶. Since 2025 manufacturers have, as a rule, proved through a BSI certification that it meets BSI TR-03109-5^17,18,19. From the box, the setpoint goes to the device or to an energy management system.

Whether a device is controlled directly or through an energy management system is the owner's call, device by device⁴. Under direct control each device gets its own setpoint and keeps at least 4.2 kW. An EMS instead gets one total value that the owner allocates freely, and the minimum is then 4.2 kW plus (number of devices minus one) times a simultaneity factor times 4.2 kW; the factor falls from 0.8 for two devices to 0.45 from nine. If a heat pump or cooling system above 11 kW is part of it, 0.4 times the combined rating of all heat pumps replaces the 4.2 kW base, and likewise for cooling systems. With both groups present, the larger of the two values applies⁴.

To check the maths: a wallbox, a heat pump and a storage unit, all below 11 kW. Direct control leaves 12.6 kW. An EMS gets 4.2 + 2 × 0.75 × 4.2 = 10.5 kW, so less guaranteed power, but a free hand in how it's spread.

The digital interface on the last stretch to the device now has a minimum standard. Back in 2023 the chamber considered EEBUS fit to carry both the command and the proof of execution, without ruling out other interfaces². The FNN recommendation describes EEBUS as the "industry-wide favoured solution for a minimum standard". The agency, though, still considers other variants possible in principle, to be settled between the parties involved²⁰.

And the box itself? The 2023 ruling had already asked that the recommendations anticipate the control function moving elsewhere, into the gateway for example². The box would then disappear as a separate device and the rest of the chain would stay. Chapter 4 covers where protection profile and field tests stand, along with the open question of which certification basis some competitive metering point operators rely on for the gateways with integrated control they already use.

The supplier gets no balancing compensation for the energy that dimming removes. It gets a message through electronic market communication, for every grid-oriented control action and at the start of preventive control⁴. Under notice no. 58 the API web services will probably stay limited, until further notice, to market location identification and control²¹. Chapter 7 covers the processes behind them.

From readings to grid status

Between a reading and a command sits a calculation, which the ruling calls the grid status assessment: the loading of a network section derived from current measurements and grid models⁴.

How much metering is enough? Until a recommendation arrived, the presumption was minute-level real-time values from at least 15 percent of all connections in a network section, or 7 percent plus measurements at the transformer feeders⁴. The recommendation came in April 2025, and on 14 April the agency recommended it for use²². The FNN guidance works out, per grid topology, what share of suitable connections needs an iMSys delivering TAF 10 data, and it assumes communication never fails²³.

The table really answers an investment question. In a radial grid, measuring each feeder in the substation cuts the share of gateways that must deliver grid status data from 70 to 15 percent. Then there's the cap in section 34 MsbG, under which minute-level grid status data is an additional service, limited to a quarter of the connections the metering point operator has equipped with an iMSys in the grid area¹⁴. The two percentages use different bases, yet for planning they point the same way, to the substation first.

Where the calculation runs, the ruling doesn't say. On the market the layer is called a low-voltage control system or DERMS, for distributed energy resource management system. A pilot that GWAdriga announced on 5 February 2026 shows what it can look like: at Westfalen Weser Netz a control system from envelio is coupled directly to CLS management, TAF 10 data feeds a digital twin, and when congestion appears, control commands go out automatically through CLS management. Twenty low-voltage grids are part of the pilot²⁴.

Three modules, two metering worlds

The reward sits in BK8-22/010-A. If the owner doesn't choose, the grid operator bills module 1²⁵.

Nobody needs an iMSys for modules 1 and 2. Once the owner has asked the metering point operator or the grid operator to make the device controllable, the owner's duty is met, and the reduction doesn't wait for the hardware^3,2. If a grid operator grants module 1 only once a gateway is installed, it's holding back money the customer is owed.

Module 3 works differently. It rests on the economic incentives that section 14a(1) sentence 2 names explicitly as a form of grid-oriented control^1,3, and it has nothing to do with section 14c EnWG, which governs market-based procurement of flexibility services and expressly leaves §14a untouched²⁷. (Dynamic tariffs and the grid charge reform follow in chapter 6.)

The usual order route runs through the supplier. The customer orders from the supplier, the supplier orders the configuration from the grid operator, the grid operator orders the time windows in the metering system from the metering point operator, and only after a positive confirmation do the billing data follow²⁵. Four roles in a row, and if one stalls, the customer waits.

That's exactly where it jams. On 28 May 2026 the Federal Network Agency threatened two grid operators with penalty payments, after complaints from market participants and consumers that "especially" concerned module 3. They had until 30 September 2026 to fix the shortcomings²⁸. As of 13 October 2026 the enforcement list shows four proceedings against distribution system operators, all at the stage of a threatened penalty, none imposed²⁹. The introduction on the same page still speaks of two companies. Count the table.

Counterpoint: how much transition is enough?

The fiercest dispute was over clause 10.5, and it came late: the 24-month clock only appeared in the draft of the second consultation, which closed on 27 July 2023^5,2. Grid operators rejected the limits (24 months, two hours a day, the 4.2 kW floor) outright or wanted them relaxed considerably. One distribution system operator even saw grid stability at risk. The energy industry association BDEW feared that operators might skip their first control actions just to keep the clock from starting, and proposed case-by-case reviews with possible extensions².

The chamber held its ground. The call to carry on "as before" could "in no way" be reconciled with the growth in heat pumps and electric vehicles, it wrote on 27 November 2023; the "lamentably inadequate state of observability of the low-voltage grid" had long been known, and the BDEW proposal was "not suitable"². Each grid operator had to act on its own responsibility. With service providers, if need be.

Pressure from the other side came from building technology, and it worked. The air conditioning association FGK, the building services federation BTGA and the heat pump association BWP called for a minimum power proportional to the rating of large heat pumps. The chamber agreed, and that's where the 0.4 factor comes from².

A third group came away empty-handed. Under slogans such as "no §14a without §14c" and "introduce preventive congestion management", numerous consultation participants wanted the devices' flexibility to be used through markets first. The chamber refused: §14a governs last-resort control in an emergency and doesn't pre-empt the competition over market-based flexibility².

Preventive, by the way, means two different things in this dispute. The consultation participants wanted flexibility bought on a market before congestion builds up; clause 10.5 allows dimming based on grid planning, without real-time data and only for a while. Whether the chamber's bet on the 24-month clock pays off will show on 1 January 2029.

Deadlines and legacy devices up to 2029

For legacy devices one question counts: was a reduced grid charge actually billed before 1 January 2024¹? Night storage heaters keep their old arrangement until they stop operating. Controllable legacy devices without an earlier discount aren't covered at all, though their owners may switch voluntarily at any time, just not back⁴. The exemption for devices that can't be controlled has its price, since they lose the guaranteed connection under clause 5².

There's no amendment. The list of notices on BK6-22-300 ends with no. 6 of 14 April 2025, and Ruling Chamber 6's current notices up to 2 October 2026 contain nothing on §14a^5,8. The only built-in review is a duty on grid operators to revisit their recommendations at least every three years. The recommendations were due by 1 October 2024 and 1 January 2025², and the final versions date from March and April 2025^30,9. When the first review falls due depends on where the clock starts, somewhere between autumn 2027 and spring 2028, just before or already in the last year of preventive control.

Four checks for grid and metering point operators:

1. Know the clock in each network section

   Record for every network section whether, and since when, it has been under preventive control. The 24 months run from the first use; on 31 December 2028 they end everywhere, however much of them would be left.

2. Test your metering concept against the FNN table

   For every network section with foreseeable control needs, record the topology, any measurement in the substation and the share of gateways delivering TAF 10. Only then can you tell whether substation measurement or minute-level gateway data is cheaper. The cap from section 34 MsbG belongs in that sum.

3. Decouple modules 1 and 2 from the rollout

   No gateway, still a discount.

4. Test the chain up to the acknowledgement

   Run one command from the control system to the device, with timestamps at every hand-over. The five minutes run from the grid status assessment to the trigger at the metering point operator; after that the device has to follow without undue delay⁴. Market communication counts the hand-over to the metering point operator in seconds: the instruction has to reach it at least 60 seconds before the effective period starts, it replies within 30 seconds, and it confirms or rejects the set-up at the latest one second before the start^31,32. That reply covers the chain up to the control box. Whether the device followed only shows in its acknowledgement, which exists only over a digital interface. If it has justified doubts, the grid operator can request the metering point operator's logs³⁰. Keep the documentation for at least two years⁴.