Guidance documents that all claim to describe a UK dwelling's heat loss can return design loads that differ by a large margin, and the differences are not random: they come from a small number of identifiable convention choices.
MIS 3005-D Issue 3.0, the MCS heat pump design standard, requires the design heat load to comply with BS EN 12831-1:2017. In practice UK designers work from a mix of sources: the CIBSE Domestic Heating Design Guide 2026 (DHDG), which publishes a reduced implementation of the European method for UK dwellings; SAP 10.3 and RdSAP 10, built for energy rating rather than plant sizing; and the MCS Heat Load Calculator's documentation. These are often treated as interchangeable. They are not.
This post compares them. It sets out which source governs what, where they genuinely disagree, where a plain reading of the generic European standard gives the wrong UK answer, and which modelling choices move a result most. We illustrate with a single semi-detached 1960s house, using assumed geometry, so the numbers demonstrate the conventions rather than prescribe a design.
The guidance landscape
The sources sit at different levels of authority and were written for different purposes, which is the root of most disagreements.
| Rank | Source | Purpose | Role in our method |
|---|---|---|---|
| 1 | MIS 3005-D Issue 3.0 (05/12/2025) | MCS compliance envelope for heat pump design, up to 70 kW thermal | Sets minimum internal temperatures, design external temperatures, ground floor rule, 100% sizing |
| 2 | BS EN 12831-1:2017 | Generic European method for design heat load | The calculation the standard requires |
| 3 | CIBSE DHDG 2026 + corrigendum DHDG-2026-1-COR-001 | UK reduced implementation with worksheets and national data | Primary implementation source for domestic UK dwellings |
| 4 | Approved Document L Vol 1 | Regulatory targets | Target U-values and air permeability only |
| 5 | RdSAP 10 / SAP 10.3 | Energy rating and annual energy modelling | Separate mode; fallback U-values for EPC-derived data |
| 6 | MCS Heat Load Calculator and vendor documentation | Tooling | Cross-check only |
Two scope gates apply before any of this can be used. MIS 3005-D covers heat pump systems with a heat load not exceeding 70 kW thermal. DHDG's reduced method covers UK domestic properties without large openings, with more than one linked heated space, no whole-house or multi-room ventilation system, balanced continuous ventilation only, and room-sealed combustion appliances. Flats are outside DHDG scope entirely, so a designer working on one needs a fully compliant implementation of the European standard instead.
The key conceptual split is that SAP and the design method answer different questions. SAP estimates annual energy use, so it uses seasonal climate and average conditions. EN 12831 sizes plant for a design day, so it uses an extreme external temperature and treats different parts of the building differently. Numbers from one should never be compared directly with numbers from the other.
The two calculations answer different questions, and most cross-source confusion comes from mixing their inputs:
How the method fits together
DHDG organises the calculation as a sequence of worksheets (A1 to A3) with numbered cells, so any result can be checked by hand. Read alongside MIS 3005-D, the chain looks like this:
A few features of this chain are easy to miss:
- Scope comes first. If a property fails a DHDG or MIS condition, the reduced method does not apply, and a fully compliant implementation of the European standard is needed instead.
- Each element uses its own boundary temperature. External air, loft, ground (annual mean or design, depending on floor type), another heated room, a neighbouring dwelling at 15 °C, or an unheated space at 5 °C. Negative differences are legitimate, because a living room can gain heat from a warmer bathroom.
- Intermittent heating is a calculated allowance, not a factor. MIS 3005-D says no uplift is required when Table 2 column B is selected. DHDG gives the allowance in W/m² of floor area from setback drop or period of disuse, thermal storage and reheat time, and zero with start/stop optimisation.
- Room results are outputs in their own right. MIS 3005-D 5.9.2 requires specific room heat losses in W/m² and the design emitter temperature of the worst performing room to be given to customers in writing, and air-to-air systems must size each indoor unit against its own room's load.
- Heat pump selection is separate. At least 100% of the calculated load at the design flow and external temperatures, without supplementary heat.
DHDG also carries a continuous chain of worked examples through its method, including a ventilation example (436.1 W, 239.3 W and 129.3 W from its factors) and an MVHR supply temperature example. That is the closest thing to a published reference case available without the licensed standard.
The compliance position is worth stating precisely. MIS 3005-D requires the calculation to comply with BS EN 12831-1:2017. DHDG states that it will produce similar results to the full general method when used within its scope, and MCS's own calculator cites it for internal-element data and design temperatures. Following DHDG faithfully is therefore a defensible route, but it is not independent verification against the licensed text, and wording such as "compliant" or "certified" should wait for that.
Emitter vs generator basis: two totals, two jobs
One calculation produces two different whole-dwelling totals because emitters and the heat generator face different worst cases. On the example house they are 5.60 kW and 4.93 kW, a 14% difference that comes entirely from ventilation.
| Emitter basis | Generator basis | Typical basis | |
|---|---|---|---|
| What it sizes | Radiators or underfloor loops and pipework, room by room | Nominal rating of the heat generator | Minimum rating and modulation check |
| How it is built | Each room's own design load: fabric, ventilation with the orientation factor of 2, intermittent allowance | Sum of fabric losses, plus sum of ventilation components without the factor of 2, plus intermittent allowance only if most rooms start heating together | Sum of room losses at typical (reference) temperature differences |
| Worst case assumed | Any single room could be the windward one | Wind can only hit one side of the house at a time | Typical rather than extreme conditions |
| Example house | 5.60 kW | 4.93 kW | 2.32 kW |
Because the leakage component is doubled only on the emitter basis, the gap widens as ventilation becomes a bigger share of the loss, so it is largest in leaky houses.
The practical consequences are these:
- Sizing the heat pump from the room sum overstates it. Here that is about 14% too large, and DHDG is explicit that oversizing a low-temperature heat generator degrades efficiency.
- Sizing emitters from the generator total understates them. A windward room can see its full design ventilation loss, so it would be undersized.
- Room-level results come from the emitter basis. The W/m² figures and the worst-room design emitter temperature that MIS 3005-D 5.9.2 requires are room quantities.
- Published figures need labelling. A quoted heat load is only comparable if it says which basis it is, so two reports of the same house can both be right and still differ.
Where the documents disagree
The table lists the conflicts most likely to change a calculated result or a comparison between published figures. Where the rank order above conflicts, the higher source wins.
| Topic | Source A | Source B | Practical effect |
|---|---|---|---|
| Ground floor temperature difference | MIS 3005-D: solid floor uses internal temperature minus local annual mean; suspended floor uses design external temperature | DHDG worksheet: annual mean difference for any ground floor, no solid/suspended split | On the example, 7.6 K rather than 21.1 K across a solid slab (typical 18 °C room) |
| Thermal bridging | DHDG Table 2-9: +0.10 W/m²K to every element, including internal, party and ground | MCS calculator (from RdSAP 10): +0.15 for age bands A to I, +0.11 for J, +0.08 for K onward | About 5% on the example (4.93 vs 5.17 kW) |
| Minimum ventilation rate | Generic EN 12831 practice as described by vendors: design airflow is the larger of infiltration and a minimum rate | DHDG Table 2-14: minimum used only as a check that warns when leakage is below it | Raising the airflow gives a materially different, non-DHDG answer |
| Ventilation outputs | A single ventilation figure per room | DHDG: three figures, with an orientation factor of 2 for emitter sizing only | Whole-dwelling load is not the sum of room loads (4.93 vs 5.60 kW) |
| Design external temperature | MIS Table 2: nearest CIBSE Guide A Table 2.5 station, column A (99%) or B (99.6%) | DHDG Table 2-3 zone values | 0.1 K difference on the example (−3.1 vs −3.0 °C) — tight now that the nearest-station rule correctly selects Cardiff; an earlier draft that used the wrong station diverged by 1.5 K, exactly the gap this cross-check exists to catch |
| Party walls | SAP 10.3 Table 3.10: party wall U-value, zero temperature difference | DHDG / EN 12831: adjoining dwelling at 15 °C (5 °C if typically unheated) | About 244 W across the party wall on a 31 m² semi-detached party wall area |
| Air permeability defaults | EN 12831 Table B.6 (as used by a vendor implementation) | DHDG Table 2-11, from a DESNZ study of 5,125 dwellings, by built form, age and wall type | Different defaults for the same house |
| Internal design temperatures | MIS Table 1: minimum values by room type | DHDG Table 2-2: richer, higher for some bedrooms (21 °C vs 18 °C) | DHDG values are used, floored at the MIS minimum |
| U-values | RdSAP 10 age-band defaults | DHDG construction tables 2-19 to 2-54 for surveyed constructions | 0.57 (filled cavity) and about 1.08 (as-built timber frame) in DHDG for the example's two wall types |
Two of these are worth dwelling on because they change how a researcher should read a published heat load.
First, the ground floor. Ground temperature beneath a slab tracks the annual mean rather than the design day, so applying the design-day external temperature across a solid floor overstates the loss. It is one of the few places where a design-day approach overstates a load in a calculation that is otherwise easy to understate. DHDG's own worksheet applies the annual mean to suspended floors as well, so MIS 3005-D and DHDG disagree on that case, and MIS takes precedence.
Second, party walls. The DHDG instruction is to assume the adjoining property is unheated at the time of calculation, even if it is known to be occupied and heated. That produces a real heat flow through the wall, whereas SAP treats the same wall as adiabatic. Neither is wrong; they answer different questions.
The picture shows why the ground floor and party wall matter so much: they are the elements where the conventions assign completely different temperature differences, while the exposed elements are the same under every method.
The biggest sources of error
Most of the divergence between calculations comes from inputs and conventions rather than from the formulas. These are the choices that move a result most, roughly in order of effect on a typical older house.
- Envelope geometry. Real element areas, measured internally with storey height taken floor to floor (DHDG Figure 2-4), matter far more than any coefficient. Windows, doors and rooflights are measured including frames and subtracted from the wall. Areas estimated as fixed fractions of floor area hide the largest errors.
- Missing elements. Rooms-in-roof, roofs over lower storeys, ceilings under a cold loft on a lower floor, and party walls on semi-detached and terraced houses are all easy to leave out. A model can be internally consistent and still omit large parts of the envelope.
- U-value provenance. DHDG stresses that the U-value chosen has a large effect on system size and cost, and expects the construction to be identified by survey: client history, visual inspection, knocking to identify dot-and-dab, loft inspection, cavity fill injection points, and wall thickness compared against its tables. A free-text EPC description is a weaker basis, and a glazing description with no installation year cannot be resolved to a single U-value.
- Thermal bridging base. The DHDG addition applies to every element, so the area it multiplies matters as much as the value. On the example it adds a small but measurable amount across the whole envelope. Where junction-level psi and chi values are available they replace the blanket addition; using both double counts.
- Ground floor. DHDG gives U-values by construction, soil type, wind shielding, perimeter-to-area ratio and insulation (Tables 2-43 to 2-54). The perimeter counts only exposed edges, so party walls and walls to heated extensions are excluded. Feeding a tabulated whole-floor U-value into an ISO 13370 calculation as if it were a construction U-value double counts the ground.
- Ventilation structure. Leakage is built from air permeability, exposed envelope, additions for vents, fans and flues, and a shielding factor. The example comes to about 0.33 air changes per hour equivalent. Using a flat air change rate is not far off numerically here, but it removes the checks and the three-output structure below.
- Margins inside the calculation. MIS 3005-D asks for a heat pump providing at least 100% of the calculated load, and DHDG is explicit that oversizing a low-temperature generator degrades efficiency. Any margin belongs at equipment selection, labelled, not inside the heat loss.
The three ventilation numbers
The least intuitive feature of the DHDG method is that ventilation yields three results per room, because emitters and generators are sized against different physical risks. A windward room can see its full design ventilation loss, but not every room does at once. DHDG applies an orientation factor of 2 to the leakage component, for emitter sizing only:
where f_leak and f_cont are temperature-weighted leakage and continuous ventilation factors. A typical-conditions variant sets the lower bound of the generator modulation range. The consequence for anyone comparing results is that the whole-dwelling generator load is not the sum of the room design loads. The example gives 4.93 kW for generator sizing against 5.60 kW as the room sum for emitters and pipework.
The chain that produces those factors runs from air permeability to the room leakage rate, with the minimum check as a side branch rather than a floor:
Carrying SAP conventions into a design calculation
The most tempting shortcut is to reuse SAP inputs for a design heat load. SAP puts the full design temperature difference across the ground floor, applies a thermal bridging factor of 0.20 to total exposed area when details are unknown, and derives ventilation from structural infiltration, chimneys, fans and shelter. Each is reasonable for annual energy and different from the design method. On our example house, a rough estimate built on SAP conventions came out around 6 kW, against 4.6 to 5.2 kW under the design method's own convention sensitivities.
Worked example: a 1960s semi-detached house
Applying the DHDG method with MIS 3005-D's temperature rules to an assumed-geometry 1960s semi-detached house gives 4.93 kW for generator sizing. The geometry is assumed throughout, because the property had no measured envelope, so treat this as a demonstration of the conventions, not a design.
Assumptions: 109 m² total floor area over three levels (68.5 m² ground floor, 26.9 m² first floor, 13.6 m² room-in-roof), with 30.9 m² of party wall across the ground and first floors, storey height 2.4 m, 22.5 m² of double glazing, 3.7 m² of doors, walls 90% filled-cavity masonry and 10% timber frame. Design external temperature is −3.1 °C (MIS Table 2, Cardiff station, column B), annual mean 10.4 °C (MGD 007 Southern region), and area-weighted internal temperature 19.22 °C.
| Element type | Loss W |
|---|---|
| External wall | 1,362 |
| Window | 1,345 |
| Roof (ceiling under cold loft + room-in-roof) | 631 |
| Ground floor | 426 |
| Party wall | 244 |
| Door | 242 |
| Fabric design heat loss | 4,250 |
Note the temperature differences: the ground floor and party wall carry roughly 7.6 K and 3 K in a typical 18 °C room, not the 21.1 K of the exposed elements. That is where the design method departs most visibly from a SAP-style calculation. The ground floor U-value also varies room by room here — DHDG scales it by each room's own exposed-perimeter-to-area ratio, so the small rooms (the rear lobby, the larder) get a noticeably worse U-value than the larger reception rooms for the same construction.
Ventilation adds 1,350 W on the emitter basis and 680 W on the generator basis. The minimum-ventilation check (DHDG Table 2-14) is applied room by room, not to the whole dwelling, and it catches two of the eleven rooms here: the Sitting Room's calculated leakage of 10.1 m³/h falls well short of its 24.6 m³/h minimum (0.5 air changes per hour, since it is a living room with a window), and the Dining Room's 15.2 m³/h falls short of its 21.5 m³/h minimum by a smaller margin. The method raises a warning that ventilation may be insufficient in those rooms and that specialist advice is needed. The design airflow itself is not increased.
| Result | Value |
|---|---|
| Heat generator sizing | 4.93 kW |
| Sum of room design loads (emitters, pipework) | 5.60 kW |
| Specific heat loss, generator basis | 45.2 W/m² |
Sensitivity to the convention choices above is modest: column A rather than B (a less extreme design temperature) gives 4.62 kW; the DHDG zone reference temperature (−3.0 °C) is now barely different from the Cardiff station value used here (−3.1 °C), giving 4.91 kW; and the MCS calculator's 0.15 thermal bridging value gives 5.17 kW.
A caution on comparing quantities. The EPC for this house implies a heat transfer coefficient of 300 to 370 W/K, but that is a SAP annual-energy figure and a different quantity from a design heat load. Recomputing the same envelope on SAP conventions gives a notably higher heat transfer coefficient than the design-method figure, for the reasons set out above, and the two should not be compared directly even where the SAP-convention figure happens to sit inside the EPC's own range.
Same house, different methods
Applying each main convention to the same house gives design heat loads from about 4.6 to 6.0 kW, a spread of about 29% between the lowest and highest, before any real-world uncertainty is considered. The reference is DHDG with MIS 3005-D rules on the generator sizing basis (4.93 kW).
Single-convention choices move the result by 0.4% to 6%. The larger figure at the bottom is different in kind, and only partly engine-verified: it stacks two conventions we could reproduce directly by changing the engine's own inputs (a full design-day temperature difference across the ground floor, and no heat flow through the party wall) with two we estimated by hand on top of that real output (a flat 0.7 air changes per hour in place of the DHDG ventilation model, and a thermal bridging factor of 0.20 applied to total exposed area instead of the DHDG per-element addition). It is our own rough illustration of what happens when SAP inputs are carried into a design calculation, not a SAP or EPC calculation and not a fully engine-verified figure, and the 0.7 air changes per hour is an assumed mid-range value.
The bridge shows that no single convention explains the gap. Swapping in the SAP ground floor and party wall conventions together adds a little over a third of a kilowatt, on real engine output — within that step, the ground floor's much larger design-day temperature difference and the party wall's smaller one partly offset, a reminder that a plausible-looking result can still rest on effects that partly cancel. Replacing the DHDG ventilation model with a flat air change rate then adds roughly two-thirds of a kilowatt, our own hand estimate on top of that engine output, and the flat thermal bridging convention adds only a little more once that flat ventilation rate is already in place.
What none of these methods capture
Everything above compares calculation conventions on one fixed set of inputs, so it is the smallest part of the uncertainty that a real retrofit carries. None of the methods models the following, and none can:
Uncertainty enters at every step between a survey and a delivered result, and calculation conventions are only one link:
| Source of uncertainty | Why it matters | Retrofit example |
|---|---|---|
| Measurement and survey limits | Inputs are inferred, not observed | Cavity wall insulation cannot be visually checked; room-in-roof and floor constructions are hidden; glazing installation year is often unknown |
| Assumptions and user error | Every default is a choice, and every input can be mis-entered | Wrong built form or age band, misclassified rooms, transcription errors, EPC descriptions taken at face value |
| Air permeability | A single default stands in for a leaky, variable envelope | DHDG's GB mean is 8.6 with a spread of ±3.3 m³/(h·m²) at 50 Pa, before any measured test |
| Installation quality | The installed system differs from the designed one | Emitters, pipework and insulation as fitted; airtightness of works; workmanship on thermal bridges |
| Controls and commissioning | Efficiency depends on how the system is set up | Flow temperature, weather compensation curve, setback, cylinder settings, defrost behaviour |
| User behaviour | Occupants set the real conditions | Set-points, heating hours, interior doors, hot water use, window opening |
| Real weather | The design temperature is an extreme percentile | Actual winters, wind exposure and local microclimate |
The consequence is that the conventions gap we have shown (about 4.6 to 6.0 kW here) sits on top of all of these. A design load can be out by tens of percent from calculation choices alone, an emitter or flow temperature can then be wrong for the room that limits it, and commissioning and behaviour are left to fill the gap. We have not quantified how these combine, and we would not claim any single figure, but it is a plausible contributor to the wide variation in heat pump experience and performance seen in practice, particularly in retrofit, where the fabric is old, partly hidden and hard to survey.
Implications for researchers and open questions
Published heat load figures are only comparable when their conventions are stated. We suggest reporting at least the following alongside any number: the standard and issue, design external temperature source and percentile, ground floor temperature rule, thermal bridging basis, air permeability method, and whether the figure is a generator sizing load or a sum of room loads.
A few observations that may be useful for anyone comparing datasets:
- SAP-derived heat transfer coefficients and design heat loads are different quantities, and back-calculating one from the other from an EPC can be off by a wide margin (roughly 300 to 370 W/K on our example, notably higher than the design-method figure).
- Thermal bridging conventions differ by source and by which elements carry the addition, so a shared value does not mean a shared result.
- Air permeability defaults derived from the 2025 DESNZ study of GB dwellings (Roberts, Li, Allinson and Lomas) differ from older generic defaults, and DHDG's corrigendum withdrew an earlier SAP 10.2 estimation pathway, so results from the first edition of the guide need checking.
- The minimum-ventilation warning fired on our example. Whether a low-leakage retrofit design method should routinely flag under-ventilation is a question for indoor air quality researchers as much as for heating engineers.
Some questions remain open. MIS 3005-D points to MCS MGD 007 for local annual mean temperatures, which we have not seen, and DHDG zone values are the interim substitute. It is not clear that MCS accepts interpolation to the nearest CIBSE Guide A station beyond the eight named in MIS Table 2, although its own calculator does exactly that. And the licensed BS EN 12831-1:2017 is still needed to test how closely the reduced DHDG method tracks the full one.
If you work on measured heat loss, co-heating or in-situ U-value studies, or on validation of design heat load tools, we would welcome comparison data and critique.
Note: the worked-example figures in this article were corrected on 21 September 2026 against the current calculation engine. The previous version of this page overstated several totals by roughly 10–15%; the conventions and conclusions described are unchanged.
Sources
- MIS 3005-D: The Heat Pump Design Standard, Issue 3.0, 05/12/2025 (MCS)
- CIBSE Domestic Heating Design Guide 2026-02 and corrigendum DHDG-2026-1-COR-001, July 2026
- BS EN 12831-1:2017 (referenced by MIS 3005-D; not independently consulted)
- Approved Document L Volume 1: Dwellings
- SAP 10.3 (12-06-2026)
- Roberts, Li, Allinson and Lomas (2025), Baseline airtightness of the GB housing stock, DESNZ
- MCS Heat Load Calculator documentation: U-values, thermal bridging, design conditions
