Residents will open the window anyway! What TM59:2026 gets right, and what we would challenge…

CIBSE’s second edition of TM59 brings welcome changes to how overheating is assessed in UK homes. We summarise what has changed, then set out where we would challenge the guidance, which, alongside Part O, can still lead designs towards more mechanical ventilation and cooling than homes need.

Overheating design for new homes in England currently sits between two documents. Part O compliance still relies on the 2017 edition of TM59, while CIBSE recommends its new TM59:2026, published in July 2026, as best practice.

The new edition is a real step forward. Here we set out the key changes, then challenge, constructively, the parts of the guidance we think could be more pragmatic, particularly where they combine with Part O.

What has changed in TM59:2026…

The adaptive test for living rooms is unchanged, but much of what sits around it is new.

A passive-first, three-stage approach. Every home must first pass using passive measures, with windows assumed usable and site noise, security and air quality constraints set aside. Those constraints are added at Stage 2, where enhanced mechanical ventilation may be introduced; cooling comes in only at Stage 3, in specific spaces, if Stage 2 fails.
A new bedroom test. Based on sleep research in 591 English homes, it limits the number of nights, not hours, when the mean bedroom temperature between 11pm and 8am exceeds 27 °C (26 °C for vulnerable occupants) to four between May and September. It now applies to every bedroom, including mechanically ventilated and cooled ones.
A clear line between natural and mechanical. Where constraints still let windows open for at least 50% of occupied hours, rooms are judged on the adaptive criterion; below that, on a fixed 26 °C.
Credit for ceiling fans. Fans producing 0.6 m/s or more can raise the pass mark by up to 1.2 K in naturally ventilated rooms and 2.1 K in mechanically ventilated ones, but not in bedrooms at night or in communal areas.
Communal areas can fail. Corridors, stairwells and lobbies must stay at or below 28 °C for all but 3% of occupied hours, at every stage.
Tighter modelling inputs. Windows now open according to operative rather than air temperature, infiltration is set to zero for new build, pipework heat losses follow Approved Document L (2026), trees are excluded as shading, and home offices get their own heat-gain profile.
New weather files. CIBSE now specifies the weather data in a separate document, based on the latest UK climate projections.

Mohamad’s technical briefing sets out each change in detail; ask us for a copy.

These changes are a step forward. In practice, though, a few areas, some new and some carried over from Part O, can still lead designs towards machines rather than better buildings. We design for people, not gadgets, and offer the challenges below in that spirit.

Windows shut on paper, open in real life…

Where Part O expects windows to stay shut at night because of noise, and where pollution or security concerns close them too, the usual answer is a larger MVHR unit sized to run on boost. We are seeing these installed, and in occupation residents who are too hot often open the window anyway.

MVHR is not a cooling system. It brings in outdoor air at outdoor temperature, plus fan and duct heat, and rarely gets close to the air change rate of an open window; Approved Document F asks for purge ventilation of 4 air changes an hour.

Bigger units also mean bigger ducts and more fan noise, and both cost developers money: larger plant, more acoustic attenuation, and deeper ceiling voids that eat into the same floor-to-ceiling height fans would need. With TM59 pointing to bedroom limits of about 30 ± 5 dBA, the extra cost can buy limited benefit, particularly if residents open the window anyway.

The bill does not stop at handover. An oversized MVHR needs filter changes, servicing and electricity for as long as it runs, paid by residents through their bills and service charges, or by the housing association or build-to-rent landlord that holds the building.

The voids add up. On one of our current schemes, a six-storey student residence, a smaller heat recovery unit would have met Part L; it was the overheating strategy that pushed the MVHR, and its ducts, up a size. Leaving MVHR out altogether would take 150 mm out of every ceiling void, 900 mm over the building, and a right-sized unit would recover part of that.

On taller schemes, a saving like that can decide whether a building stays below a fire-safety height threshold such as 11 m or 18 m, and avoids the stricter requirements that apply above it. None of this is an argument against heat recovery, which earns its place in airtight, low-energy homes; it is an argument for sizing MVHR for ventilation, and solving overheating by other means.

TM59:2026 already offers a more balanced route, and we would encourage its use. Section 3.8 lets the acoustician set how many hours or nights a window can be open, aiming for “a reasonable balance between thermal and acoustic comfort”, for example opening “only during the hottest days of the year”. In practice, three things can still lead projects back to sealed windows:

Acoustic reports that give a simple open-or-closed answer instead of hours.
Part O in England, which expects bedroom windows to be closed at night where noise inside would exceed 40 dB LAeq over 11pm to 7am, or 55 dB LAFmax more than 10 times a night.
Air quality constraints, for which TM59 does not yet set criteria, which can lead to cautious, blanket window closure near busy roads.

What we would do instead: ask the acoustician for hours, not a yes or no. Put bedrooms on the quiet side of dual-aspect layouts, use attenuated openings (TM59 accepts their equivalent areas), and read the Stage 1 result as what the home will do when residents open the windows anyway. Someone opening a window on a noisy street is trading sleep for heat; good design makes that trade less painful rather than assuming it will not happen.

Bedrooms that must stay sealed at night…

Criterion B now applies to mechanically ventilated bedrooms, with no fan credit and boost rates capped by noise. Where bedroom windows must stay shut at night, cooling can become the only route left, and we think this is the most likely way air conditioning becomes common in UK flats.

What we would do instead: solve it in the layout and façade at concept stage, with quiet-side bedrooms, acoustic balconies or winter gardens and attenuated night openings, before the model is asked to.

Ceiling fans: a good idea that needs clearer rules…

TM59:2026 credits ceiling fans. A fan producing 0.6 m/s or more can raise the pass mark by up to 1.2 K in naturally ventilated rooms and 2.1 K in mechanically ventilated ones. Fans are cheap and low-energy, and we support their use.

“My concern is not whether ceiling fans are a good solution. They can be an excellent low energy solution. The question is how we define, model and verify their benefit when the result may become part of a mandatory pass or fail assessment.” Mohamad Tabatabaee, D-For, on MEP Today

Where is 0.6 m/s measured? TM59 does not say. Air speed at seated height (0.6 to 1.1 m) differs from standing height, and both differ from the air speed close to the fan. People in a living room sit, stand and cook; at the bed, the air speed is likely to be below 0.6 m/s, and two fans leave weaker zones between them.

The evidence comes from a different room. As Mohamad points out, the CBE research behind the design guide TM59 cites tested a 1.5 m diameter fan in a 5.6 × 4.3 m chamber, 2.6 m high, with the blades at 2.3 m. That is not a typical UK new-build bedroom.

Results depend on the software. Modelling tools calculate mean radiant and operative temperature in different ways, results shift with room geometry, and TM59 does not say how a fan’s own heat should be counted. A credit that may one day decide a Part O pass or fail needs one agreed method.

And the fan has to fit. A typical ceiling fan hangs about 300 mm below the ceiling, so a 2.4 m ceiling, the lowest TM59 considers feasible, leaves the blades at around 2.1 m: within reach of a raised arm, and below the 2.3 m many manufacturers specify. Mohamad suggests 2.6 to 2.7 m floor-to-ceiling heights, before any service void; the London Plan minimum is 2.5 m and the national space standard 2.3 m.

Without the fan, a heavily noise-constrained living room is judged against a fixed 26 °C, and the method’s next steps are more mechanical ventilation, then cooling. The fan credit also stops at the bedroom door at night.

What we would do instead: before TM59:2026 feeds into Part O, we would welcome clear guidance on the height and position at which air speed is assessed for each room type, how fan heat is counted, how software must calculate operative temperature, and a safe minimum ceiling height. On projects, settle ceiling heights with the architect at concept stage, and credit occupant-controlled air movement by its performance where people actually sit and sleep, whatever the device.

The 50% threshold…

If noise or security constraints still let windows open for at least 50% of occupied hours, a room is judged against the adaptive comfort criterion. At 49%, it switches to a fixed 26 °C. In a warm spell, the adaptive limit for a typical home sits at around 27 to 28 °C.

So one percentage point of window availability can move the pass mark by around 2 K, with nothing about the room changing. People in a room whose windows open 45% of the time still open them, change clothes and adapt; a fixed threshold assumes they cannot.

What we would do instead: replace the switch with a sliding transition in proportion to window availability, or keep the adaptive criterion wherever residents can still open a window.

Trees and green infrastructure…

TM59 excludes trees and vegetation as “not a fixed element”, while admitting this is conservative and that plants could improve summer comfort. Yet the GLA cooling hierarchy that TM59 itself cites starts with green infrastructure.

What we would do instead: allow protected or retained mature trees as a reported sensitivity, with seasonal leaf cover, so landscape-led shading is rewarded rather than ignored.

Corridors: fix the heat, not the air…

Communal corridors can now fail TM59 outright. That is right, but the cause is usually heat-network pipework, and the risk is that corridor cooling becomes the standard fix.

What we would do instead: shorter pipework runs, lower flow temperatures and properly installed insulation, as TM59 itself recommends, before any corridor plant.

What we would like to see…

A defined, verifiable method for fan credits (measurement height, fan heat, software) before Part O adopts it
A gradual transition at the 50% threshold
Time-limited window opening, agreed with the acoustician, as the norm, with Part O aligned to it
Evidence-based criteria for air quality constraints, as there are for noise
Protected trees allowed as a reported sensitivity
Heat removed at source before corridors are cooled

TM59:2026 is a better method than the one it replaces. With a few refinements, and Part O brought into line, it could steer far fewer homes towards sealed windows, oversized MVHR and air conditioning. Our approach is simple: design the building to work with the windows open, make opening them bearable, and use machines only where they genuinely earn their place.

If overheating, noise or air quality is already shaping your residential scheme, talk to us at concept stage: enquiries@d-for.com.

Listen: Mohamad Tabatabaee unpacks the ceiling fan question, including operative and mean radiant temperature and how TM59 compares with ASHRAE 55:2023 and ISO 7730:2025, on MEP Today, in “TM59:2026. Operative Temperature, MRT and the Ceiling Fan Challenge”.