CIBSE Domestic Heating Design Guide 2026: What Heating Engineers Need to Know

Published September 2026 · 10 min read

CIBSE published a new edition of the Domestic Heating Design Guide in January 2026. In July, version 2026-02 superseded the original 2026-01 release following publication of a corrigendum (CIBSE ref DHDG-2026-1-COR-001). The two printings number their tables differently — this article covers both.

If you design boilers, radiators or hot-water systems for UK homes, this is one of the principal reference guides for domestic heating design. One boundary to be clear about first: for an MCS heat-pump design, the compliance calculation follows MIS 3005-D and BS EN 12831-1 rather than the CIBSE DHDG calculation method. The two methods answer different paperwork and can produce different numbers for the same house — we cover the differences in CIBSE vs MCS — choosing a calculation method. Everything below is the CIBSE method.

A note on the guide's character: it is a calculation method, not a rule of thumb. In our line-by-line audit, we found no general safety uplift or arbitrary diversity factor applied across the heat-loss method — every watt is accounted for by a table, an equation or a documented designer choice. That puts more weight on your inputs. Here are the ones that move the numbers.

Airtightness drives infiltration

Infiltration heat loss starts from the building's air permeability at 50 Pa (q50), not from a per-room air-change guess. Section 2.5.4.1 gives three routes to a q50 value:

• • Measured — an airtightness test result, used directly. This is the route the guide recommends wherever possible. A low-pressure pulse test reporting at 4 Pa converts to 50 Pa via the guide's Equation 2.8.
• • Design target — for new builds and planned retrofits, the guide permits the design air-permeability target to be used; its referenced notional-dwelling value is 5 m³/(h·m²) at 50 Pa (as set out in Approved Document L, or the regional equivalent).
• • Estimated — where neither exists, a default. Since the July corrigendum this is a single table of empirical defaults (see the corrigendum section below). The all-dwellings mean is 8.6 m³/(h·m²).

From the whole-building q50, each room's leakage is derived from its share of the exposed envelope, background ventilation from vents and flues is added, and a conversion factor keyed on building shielding and building height brings it to design conditions. A doubling factor for wind orientation applies to the emitter (room) design figure only, not to boiler sizing.

On site: an airtightness test result feeds your sizing directly, and a leaky cottage and a taped-and-tested new build do not get the same allowance.

Room temperatures depend on build year

Section 2.5.2.2 splits the housing stock at 2006. Buildings built since 2006 are designed at 21 °C throughout. Buildings built before 2006 take room-by-room temperatures from Table 2-2 — bedrooms and kitchens at 18 °C, for example. That is roughly 3 K of design temperature difference in those rooms, which flows straight through every fabric and ventilation calculation for them. The guide also carries a 23 °C recommendation for rooms used by elderly or infirm occupants.

On site: the build year is a sizing input, not a survey footnote.

Your heating schedule sizes the plant

The guide allows for intermittent heating: Section 2.5.5.1 adds a reheat allowance in W/m² on top of the steady-state loss, from two tables (Tables 2-15 and 2-16 in the corrected reprint; 2-19 and 2-20 in the original printing) selected by heating strategy and by the room's thermal mass, low or high. The factor depends on the temperature drop the room is allowed and the reheat period. The guide's own example: a 14 m² high-thermal-mass room on constant heating with a 3 K setback and a 2-hour reheat gets 33 W/m² — 462 W of extra emitter capacity.

Two things zero the allowance: constant heating with no setback, and optimum start/stop control. And some combinations get no factor at all — the guide marks deep-drop, long-reheat cells as design-invalid, on the basis that a system sized to recover them would be unacceptably oversized.

The allowance also reaches boiler sizing, with a gate: Worksheet A3 only carries the intermittent column into the heat generator total when the majority of rooms use an intermittent or setback strategy and their reheat periods overlap (or may in future). Otherwise the generator is sized without it.

On site: the customer's heating habits are a design input. A house heated in two short bursts a day needs more from its radiators than one heated continuously.

Party walls are not free

The guide gives party walls a non-zero heat loss. The wall to a typically heated neighbour is calculated against an assumed 15 °C on the far side; a typically unheated neighbouring space is assumed at 5 °C. For a 21 °C living room that is a 6 K or 16 K temperature difference across the party wall. The method offers only those two cases — there is no "unknown" option to pick — and with either one, a normally heated room loses heat through its party wall.

This is a point where the CIBSE and MCS methods genuinely diverge: under the MCS framework, party walls to a normally heated neighbour are treated as lossless. Same house, different watts, both documented.

On site: ask what is on the other side of every party wall — a heated house, or a stairwell, garage or empty unit.

Unheated spaces and lofts

An unheated space adjoining a room is assumed to sit at 5 °C. Lofts are not unheated spaces: Worksheet A2 groups a loft with outside, so a ceiling below a loft is calculated at the full design temperature difference. Treating the loft as a 5 °C buffer understates the loss.

On site: a room under the loft takes the full outside difference through its ceiling. (Under the MCS method the same surfaces use BS EN 12831-1 Annex B temperature ratios instead — another documented method difference.)

Ground floors: location matters

Ground-floor U-values come from the guide's tabulated lookup, keyed on floor construction and insulation, on wind exposure, and on ground conditions — the guide's worked example runs an exposed, rural site on clay. The temperature on the far side of the floor is the location's reference ground temperature from the weather data, not a fixed constant; the guide's own examples use values like 8.9 °C and 10.6 °C.

On site: the same floor build-up gives different numbers in different places, and that is intentional. Heatworx implements the guide's ground-floor calculation and resolves the appropriate ground temperature from the property's weather zone.

Design outdoor temperature is a choice, with data to back it

The guide's weather data (Tables 2-3 and 2-4) covers UK weather zones with design temperatures at several percentiles — 98th, 99th and 99.6th among them — plus altitude corrections. The percentiles are a real choice: the guide publishes, alongside each one, the expected shortfall days and maximum expected underheat, and it explicitly cautions against picking the coldest (99.6th percentile) figure without further consideration. Colder design temperature, bigger plant, more oversizing on every mild day.

On site: the percentile is a conversation to have with the client, with the shortfall numbers on the table. Heatworx resolves the design temperature from the property's zone and altitude and lets you select the percentile; a manual override is available where a project specifies a temperature.

The minimum ventilation check

The guide carries a minimum design air-change table (Table 2-14 in the reprint; 2-18 in the original) — 0.5 air changes per hour for living areas and similar rooms, with other room types outside the check. In the guide this is an adequacy check: where the calculated room ventilation falls below the minimum, the worksheet's instruction is that there may be insufficient ventilation and specialist advice should be sought. Heatworx deviates here deliberately, in two ways: the minimum is applied as a floor on the calculated figure rather than as a flag, and it is applied to any room with outside walls rather than to living areas only — so a very airtight room is never sized below 0.5 air changes per hour. The result can only be equal to or higher than the worksheet's own arithmetic, and the deviation is documented.

The July 2026 corrigendum

In July 2026 CIBSE issued corrigendum DHDG-2026-1-COR-001, and a corrected reprint of the guide followed. It makes one substantive change: the SAP-based estimation route for air permeability (§2.5.4.1 "By estimation") is withdrawn in its entirety — its five tables, both worked examples, and the corresponding worksheet cells. CIBSE's stated reason is that the route's ACH-to-q50 conversion factors were incorrectly derived and, combined with a later table, cancelled themselves out, while double-counting exposure and height adjustments.

The replacement is simpler: a single new Table 2-11 of default air-permeability values from a DESNZ-commissioned study of 5,125 GB dwellings. You pick a default by built form, age band or main wall construction; where the categories disagree, the guide asks for judgement — the most representative value, or a conservative choice. The all-dwellings mean is 8.6 m³/(h·m²) at 50 Pa. In our comparison, the withdrawn route typically produced values around 13 against the new defaults' 8 to 9 — a system estimated with it would have been oversized relative to the corrected figures.

One practical consequence: deleting five tables renumbered every later table in section 2 down by four — the U-value tables among them — so a table citation now means different content depending on which printing you hold. Check §2.5.4.1: if it is headed "By estimation: default values", you have the corrected reprint. And one finding worth stating: in our page-by-page comparison, we found no other changes to the guide's table values.

The corrigendum deserves its own article, and a follow-up will cover it in full — the withdrawn SAP route, the new defaults table, and the renumbering map.

Heatworx implemented the corrigendum in August 2026: the withdrawn route was retired, the new default table was digitised in full, and estimation now works the corrected way.

What to do on Monday morning

1. Check which printing you hold, using the §2.5.4.1 heading above. If you have the original, stop using its SAP estimation tables — CIBSE has withdrawn them.
2. Prefer measured airtightness. A test result is the guide's recommended input, and a 4 Pa pulse test converts.
3. Record the build year on every survey. It decides the room temperatures.
4. Capture the heating schedule and setback, not just the fabric. They feed into emitter sizing and, when reheat periods overlap, the boiler.
5. Ask what is behind every party wall and above every top-floor ceiling.
6. Put the design-temperature percentile in front of the client, with the shortfall days that go with it.

Where Heatworx fits

Heatworx implements the DHDG 2026 method — the weather data, room temperatures, air-permeability routes, reheat tables and worksheets described above — and we audited the implementation line by line against the guide's text: every digitised table diffed against the printed appendix and the guide's worked examples reproduced through the calculation engine. Every result states which method and data pack produced it. The same survey also calculates to the MCS method (MIS 3005-D / BS EN 12831-1) — you switch per survey and nothing you measured changes.

Related guides