EPBD 2026: New Building Automation (BACS) Rules Explained

If you run a non-residential building with a big heating or cooling plant, the EPBD 2026 building automation requirement now applies to you, and the deadline to act has already passed in most of Europe. The recast Energy Performance of Buildings Directive set 29 May 2026 as the date every member state had to have it written into national law. From that point, the obligation to fit a building automation and control system stops being a Brussels directive and becomes something your national regulator can enforce.

Most coverage of the recast has focused on the headline targets: zero-emission new builds, solar mandates, the phase-out of fossil boilers. The building automation clause got less attention, which is exactly why it catches portfolios off guard. It is narrow, it is technical, and it lands on a specific slice of the commercial stock right now.

Here is what changed, who it applies to, what a compliant system actually has to do, and the gap most portfolios discover when they look closely.

What changed, in one paragraph

The recast EPBD requires non-residential buildings to be fitted with a building automation and control system (BACS) where the rated output of the heating, air-conditioning, or combined ventilation system is above 290 kW. That building automation control system 290 kW threshold is not new in itself; it carried over from the 2018 revision, which set the end of 2024 as the original cut-off. What the 2024 recast did was reaffirm it, fold it into the wider directive, and lock in the next step: the threshold drops to 70 kW by 31 December 2029. The EPBD transposition 2026 deadline of 29 May is the moment all of this became national law across the EU. So if your building sits above 290 kW and has no compliant BACS, you are not approaching a deadline. You are past one.

Does this apply to your building?

The requirement is specific, which is good news, because it means you can rule yourself in or out fairly quickly. Three questions decide it.

Is it non-residential? Offices, retail, hotels, hospitals, schools, logistics, manufacturing, mixed-use above the residential portion. The BACS obligation does not reach single-family homes or the residential floors of apartment blocks. If you manage commercial or institutional space, you are in scope by building type.

Is the rated output above the threshold? This is the part people read wrong. The 290 kW figure refers to the effective rated output of the technical building system, the heating plant, the air-conditioning plant, or a combined heating-and-ventilation or cooling-and-ventilation system. It is not the floor area, not the connected electrical load, not the peak demand you saw last January. A single large building usually clears 290 kW on its central plant alone. A mid-sized building can clear it once you count a combined system correctly. Pull the nameplate ratings on your central plant before you assume you are under the line, and remember the line moves to 70 kW in 2029, which pulls in a much larger share of the stock.

Is it technically and economically feasible? The directive includes a feasibility clause, and it is genuine, not a loophole. Where retrofitting a BACS is not technically possible or cannot be done at reasonable cost, the obligation eases. But feasibility is assessed and documented, not assumed. “We didn’t think it was worth it” is not a feasibility assessment. If you intend to rely on the clause, you need the analysis on file.

If you answered yes, yes, and feasible, the EPBD compliance non-residential buildings requirement is yours to meet, and the practical question becomes what a compliant system has to do.

What a compliant BACS must actually do

A BACS requirement EPBD-style is defined by function, not by a brand of hardware. The directive lists what the system has to be capable of, and each capability maps onto something a sensor-intelligence layer already does.

Continuously monitor, log, and analyse energy use. Not a monthly meter read. The system has to track energy consumption across the building’s technical systems on an ongoing basis and keep the record. This is the continuous-data backbone the rest of the requirement sits on.

Benchmark efficiency and detect losses. The system must be able to establish where the building’s efficiency sits and flag when it slips, identifying losses in the technical systems and telling whoever runs the building where the waste is. This is the clause that trips up the most portfolios, because it asks for analysis, not just data capture.

Detect faults and flag them to the operator. A compliant BACS has to surface faults in the technical building systems and notify the facility manager or technical operator so they can act. Detection plus a notification someone receives, not a value sitting in a historian that nobody opens.

Communicate with connected systems and across interfaces. The system has to be able to talk to the building’s other technical systems and to interoperate across different proprietary technologies, devices, and manufacturers. In a real estate of mixed-vintage plant and three different BMS vendors, interoperability is the difference between a system that meets the bar and one that meets it on paper.

Read that list back and a pattern emerges: the directive is not asking for more control hardware. It is asking for monitoring, analysis, fault detection, and interoperability, which is the work an analytics layer does on top of the controls you already run. That distinction is where the gap lives.

The gap most portfolios have

Here is the trap. Most non-residential buildings above 290 kW already have a building management system. The instinct is to assume a BMS satisfies the BACS requirement. Often it does not, and the difference is in the functional clauses, not the box on the wall.

A typical BMS is strong at control. It runs schedules, holds setpoints, sequences plant, and shows you live values on a graphic. What it tends to be weak at is the three things the directive actually names. Continuous logging is frequently partial: trends are kept for some points and not others, retention is short, and submeters drop offline without anyone noticing because the aggregate bill still looks normal. Loss and inefficiency detection is usually absent altogether, because spotting that a chiller is drifting 8% off its expected efficiency curve is an analytics task, not a control task, and the BMS was never built to do it. Fault detection, where it exists, is a wall of threshold alarms that the operations team muted months ago because the false-positive rate made them useless.

So the building has a BMS, the directive asks for monitoring, analysis, loss detection, and fault flagging, and the honest answer is that the controls layer covers maybe half of it. The missing half is exactly the analytics work: continuous logging that does not lose points, benchmarking that knows what good looks like, and fault detection an operator actually trusts enough to act on. Closing that gap does not mean ripping out the BMS. It means reading it, which is what a BMS analytics layer is for.

From BACS box-ticking to operational value

The temptation with any compliance deadline is to do the minimum, file the certificate, and move on. With this one, the minimum is a waste, because the data you have to collect to satisfy the directive is the same data that runs everything worth having.

Treat the BACS requirement as the floor, not the ceiling. The continuous energy logging the directive mandates is the input to demand forecasting, so the same feed that proves compliance also tells you next week’s load and where your peaks will fall. It is also the feed that turns the annual audit scramble into continuous compliance tracking and feeds ESG reporting straight from sensor data instead of estimates. The loss-detection clause is anomaly detection by another name, so the system that flags inefficiency for the regulator is the system that catches a fan running 24/7 against an override or a heat exchanger losing delta-T. The fault-flagging requirement, done with analytics rather than raw thresholds, becomes a prioritised queue of the handful of issues actually worth a technician’s time this week, instead of an alarm list nobody reads.

This is the reframe that matters. A BACS bought to pass an audit is a cost. The same capability, read as a decision layer over your existing controls, pays for itself in energy and avoided downtime regardless of what the directive said. The dashboard tells you the temperature on Floor 4. The decision layer tells you which of four hundred buildings needs you today, and why. One of those is compliance theatre. The other is the reason to bother.

A short compliance-readiness path

You do not need a transformation programme to close the gap. The path is short and it works with the plant you already have.

Audit what your systems actually do. Go building by building above the threshold and check the four functional clauses honestly: is energy use logged continuously and completely, is efficiency benchmarked, are losses detected, are faults flagged to a person who acts? Most portfolios find the control layer is fine and the analysis layer is missing.

Identify the functional gaps. Separate “we don’t have the hardware” from “we have the data and don’t analyse it.” The second is far more common and far cheaper to fix, because the sensors are already in the building and the BMS is already reading them.

Layer analytics on top, read-only. A sensor-intelligence layer ingests from the BMS you already run, without touching control logic and without a rip-and-replace. It supplies the continuous logging, benchmarking, loss detection, and fault flagging the directive asks for, and because it reads any BMS it handles the interoperability clause across a mixed estate. The same layer maps to the wider compliance picture too, since BACS, CSRD, and BREEAM all draw on the same measured data. Compliance gets satisfied as a byproduct of running the building better.

The EPBD 2026 building automation requirement is, in the end, a forcing function. It asks every large non-residential building to do continuous monitoring, analysis, and fault detection on its energy systems. That is worth doing on its own terms. The directive just set a date.

Frequently asked questions

What is the EPBD 2026 building automation requirement?
It is the recast Energy Performance of Buildings Directive’s obligation for non-residential buildings to be fitted with a building automation and control system (BACS) where the heating, cooling, or combined ventilation system has an effective rated output above 290 kW. Member states had to transpose it into national law by 29 May 2026.

Does the EPBD BACS requirement apply to my building?
It applies if the building is non-residential, the rated output of its heating, air-conditioning, or combined ventilation system exceeds 290 kW, and fitting a BACS is technically and economically feasible. The threshold drops to 70 kW by 31 December 2029, bringing many more buildings into scope.

Isn’t 290 kW an old threshold?
The 290 kW figure originated in the 2018 EPBD revision, with an end-of-2024 deadline. The 2024 recast reaffirmed it and added the step down to 70 kW in 2029. The 29 May 2026 transposition deadline is when the obligation became enforceable national law across the EU.

Does my existing BMS already meet the requirement?
Not necessarily. The directive defines a compliant BACS by function: continuous energy monitoring and logging, efficiency benchmarking, loss and fault detection, and interoperability across systems. Many building management systems are strong on control but weak on continuous logging, loss detection, and trustworthy fault flagging, which is whe