HeatLens is a UK platform for designing domestic heat pump systems and assessing retrofit options. At its core is a room-by-room design heat load engine, built on the CIBSE Domestic Heating Design Guide 2026, that records every input, default and assumption alongside each result.
HeatLens welcomes collaboration with universities, colleges, training providers and research groups. This page sets out the opportunities for joint research, teaching and data sharing, the developments coming next, a worked example, the current state of validation and a summary of the method.
Features for research and teaching
The short video below walks through the HeatLens features most relevant to researchers and educators. They fall into four groups: transparency, teaching, modelling, and data and API access.
Transparency
- Traceable to the source. Every result can be followed from the headline figure back to its inputs, defaults and the source table row behind each one.
- Evidence levels and provenance. Each input is tagged as measured, from a plan, from a table or estimated, so the effect of evidence quality on a result can be seen and taught.
- A documented method. Room-by-room heat loss following the CIBSE DHDG, which is based on BS EN 12831-1. Every element's area, U-value and temperature difference is listed, so each figure can be checked by hand.
- A tested engine, honestly scoped. More than 180 automated tests with pinned results, alongside a clear statement of what has not yet been tested.
Teaching
- Design and rating conventions. Compare design heat load with SAP rating conventions, and emitter and pipework sizing with heat generator sizing, on the same home.
- Worked examples for coursework. Documented worked examples, including the one on this page, and a free demo project that needs no account.
Modelling
- Solar, battery and heat pump together. Planned modelling of the heat pump with solar PV and battery storage, for studies of how these technologies interact. Not yet available.
- Design against measured performance. A weather-adjusted comparison of design and measured energy use, with seasonal performance and findings, for research into the performance gap.
Data and API
- Research API (planned). Batch calculations with inputs varied systematically, each result returned with its audit trail.
- A research dataset (being explored). Design-stage data with provenance, shared only on an opt-in basis with personal details removed, subject to legal advice.
Why HeatLens suits research
- A documented, traceable method. The engine follows the CIBSE Domestic Heating Design Guide 2026 (DHDG), which is based on BS EN 12831-1:2017, for MCS's heat pump design standards, MIS 3005-D:2025 Issue 2.0 and MIS 3005-D Issue 3.0. Source tables and modelling decisions are documented and versioned alongside the code.
- Provenance on every input. Each wall, window, roof, floor and party wall records whether its area and U-value were measured, taken from a plan, looked up from a construction table or estimated, together with the table row used.
- Results that state their own confidence. Each result carries an evidence level, from blocked through indicative to fully evidenced, so any result resting on open assumptions is identified as such.
- Room-level outputs. Results are reported room by room in W/m², with separate whole-dwelling totals for heat generator sizing and for emitter and pipework sizing.
- A self-contained, versioned calculation core. The engine runs independently of the web application and database. It takes a plain description of a dwelling and returns a result with a full audit trail.
- Results that can be checked by hand. Each engine result lists every element's area, U-value, temperature difference and ventilation inputs, so students can check a room's heat loss by hand.
Essential inputs missing. No figure is produced. Calculated with stated assumptions. Calculated from evidence, with points still to resolve. Every input recorded with an evidence source.
Collaboration opportunities
HeatLens is particularly interested in working with partners in six areas:
Measured heat loss Compare design loads with coheating, smart meter or in-situ measurements on the same homes. Monitored performance Link design-stage inputs to monitored heat pump data to study the performance gap. Independent method review Review the DHDG implementation against BS EN 12831-1:2017 and other design tools. Sensitivity and uncertainty Quantify which inputs matter most when defaults are replaced with measured values. Dataset design Shape the structure and governance of any future research dataset. Student projects Dissertations and projects built around the method, examples or engine outputs.
Upcoming developments, including a research API and solar PV, battery and thermal storage modelling, are set out under What's coming next below.
Teaching and education
HeatLens is designed to make the reasoning behind a heat loss calculation visible, which suits teaching well. Every engine result can be traced back to its inputs, defaults and source tables.
Step 1 Result Heat generator and emitter sizing, with an evidence level Step 2 Room Design heat load and W/m² for each room Step 3 Element Area, U-value and temperature difference for each wall, window, roof and floor Step 4 Source Measured, plan, construction table or estimate, with the table row used
Educators can use HeatLens to:
- demonstrate room-by-room heat loss using the UK DHDG implementation of BS EN 12831-1;
- show how evidence quality, defaults and assumptions change a result;
- compare design heat load conventions with the conventions used for energy rating under SAP (the Standard Assessment Procedure);
- explore the difference between emitter and pipework sizing and heat generator sizing; and
- set coursework around documented worked examples.
Universities, colleges and training providers can enquire about free access to HeatLens for teaching and research.
Data sharing
HeatLens records design-stage information that is rarely captured elsewhere: what was specified for a home, from what evidence and with what level of confidence, before anything was installed. This could support research into heat loss methods, retrofit decision-making and the performance gap.
A research dataset could describe dwelling type, build era, banded floor area, region, design temperatures, fabric inputs with their provenance, room-level heat loss components, proposed heat pump capacity and flow temperature. Researchers should note its likely limits: the data would come from HeatLens users, who are mostly heat pump professionals working on retrofit assessments, and it would describe design intent with no monitored performance after installation.
No HeatLens user data is currently shared with researchers. Any future sharing would be opt-in, limited to design data with personal details removed and location generalised to region, and governed by a written data sharing agreement, subject to legal advice. As a first step, HeatLens is exploring a synthetic reference dataset that research teams could use to build and test analysis pipelines. Views from researchers on what it should contain are very welcome.
Design projects Design-stage records in HeatLens Opt-in consent Separate agreement from the account holder De-identified Personal details removed, location generalised, small groups suppressed Data sharing agreement Purpose, handling and publication agreed in writing Research dataset With provenance flags and documentation First step being explored: a synthetic reference dataset, generated by the engine, for building and testing analysis pipelines before any real data is involved. Proposed approach, subject to legal advice. No user data is currently shared.
What's coming next
One development of direct relevance to research is now in beta, a research API is planned, and solar PV and battery modelling is in development.
Research API. Scripted access to the design heat load engine is planned. It is intended to let research teams run calculations in batches, vary inputs systematically and retrieve results with their audit trail, without using the web interface.
Solar PV and battery modelling. A solar PV and battery module is in development and is not yet available to customers. Its settings and figures are draft and have not yet completed HeatLens's evidence review. It is intended to support studies of how these technologies work together in a home. Thermal storage modelling is also planned.
Performance monitoring (beta). A performance module compares a home's measured energy use with its design, adjusted for the weather that actually occurred. It reports seasonal performance, measured heat loss and balance point, and flags findings such as low efficiency or immersion heater use. It is being tested on a pilot home before wider use, and could provide a route to the design and measured comparisons described under Validation and open questions.
Now in beta Performance monitoring Measured energy use compared with the design, adjusted for the weather. Planned Solar PV and battery modelling In development: solar PV and battery storage alongside the heat pump. Research API Scripted access to the design heat load engine for batch runs and systematic input variation. Thermal storage modelling Thermal storage and its interaction with heat pump operation. Under consideration Sensitivity analysis Varying uncertain inputs together to examine the spread of results. Citable engine release A tagged release with a DOI.
Built-in sensitivity analysis and a citable engine release with a DOI are also under consideration. Research partners are invited to help set priorities for all of these developments.
Worked example
To illustrate the method, this example uses a property from the HeatLens beta with all identifying details removed: a semi-detached house built in the 1960s in southern England, with two storeys plus a room in the roof and a total floor area of about 110 m².
No measured survey was available for the property. Its geometry was therefore established in two different ways from the same evidence on record: once from the energy performance certificate (EPC) alone, and once by inferring a floor plan consistent with the recorded rooms, built form and certificate floor area. Comparing the two shows how sensitive a design load is to the way missing geometry is filled in.
The example also shows why complete inputs matter. The property's original records omitted the party wall and parts of the envelope. An early run of the engine on those records alone returned about 2.15 kW, together with validation errors identifying the missing geometry.
With complete geometry in place, the two approaches gave:
| Result | EPC-inferred geometry | Inferred floor plan |
|---|---|---|
| Heat generator sizing | 4.61 kW | 4.93 kW |
| Emitter and pipework sizing | 5.17 kW | 5.61 kW |
| Typical load (modulation check) | 2.21 kW | 2.33 kW |
| Specific heat loss | 48.4 W/m² over about 95 m² of recorded room area | 45.2 W/m² over the certificate floor area |
| Open assumptions flagged by the engine | Yes | Yes (the reasoned plan) |
The EPC-inferred geometry records less heated floor area than the certificate states, which is why its specific heat loss is higher despite its lower total. The inferred floor plan is itself an assumption, so the engine flags it as an open assumption and reports the result as not usable as MCS evidence. In the product, a reading taken from a floor plan drawing counts as evidence only after a person has reviewed and accepted it. Neither result should be relied on as MCS evidence without a measured survey.
The two approaches agree to within about 7% on heat generator sizing and about 8% on emitter and pipework sizing. This is useful evidence of internal consistency. Because neither approach used a measured survey, neither figure has yet been compared with the property's actual heat loss.
The inferred floor plan also produces a full breakdown by room and by building element:
The article Where UK heat loss guidance diverges uses a house of this type to compare how different UK guidance documents, including SAP conventions, treat its heat loss.
Validation and open questions
Research partners can add most value where evidence is still missing, so this section sets out what has and has not been established.
Tested so far Open for collaboration
What has been tested. The engine has more than 180 automated tests covering its climate, ground floor, ventilation, thermal bridging and construction logic, with expected results pinned so that code changes cannot silently move a figure. It reproduces the DHDG's published worked examples for ventilation, air permeability conversion, mechanical ventilation with heat recovery (MVHR) supply temperature, intermittent heating and a selection of construction and ground floor U-values. The transcribed source tables have also been checked for physical consistency.
What remains to be tested. The engine's predictions have yet to be compared with measured whole-house heat loss, with monitored heat pump performance, or with other design tools on a shared set of homes. The performance monitoring module, now in beta, is intended to make the comparison with monitored performance possible. Results are described as calculated following the DHDG method. Independent verification, both against the standard and by engineers outside HeatLens, is a priority and an area where academic partners would be particularly welcome.
Why this matters. Studies have found substantial gaps between calculated and measured heat loss. One UK trial of 56 homes by Build Test Solutions, Veritherm and Elmhurst Energy compared measured heat loss with room-by-room calculations to BS EN 12831 conventions. It reported an average absolute difference of 35%, with 59% of homes overestimated and 13% underestimated (source). The worked example above shows incomplete input data alone moving a design load by a factor of about two.
Method in brief
Standards and sources
MIS 3005-D:2025 Issue 2.0 and MIS 3005-D Issue 3.0 require design heat loads to be calculated to BS EN 12831-1:2017. The DHDG provides a reduced method based on it for dwellings within its scope, and HeatLens follows the DHDG. Each result carries a short methodology statement naming the guide, the corrigendum applied and the engine version, so any figure can be traced to the method that produced it. Where sources differ, HeatLens applies them in a fixed order of precedence:
The DHDG method applies to houses with more than one heated space, all linked; no large openings; no whole-house or multi-room mechanical ventilation; any continuous ventilation balanced; and room-sealed combustion appliances. Flats and heat loads above 70 kW are out of scope. The engine checks these conditions before a result can be treated as fully evidenced.
How a calculation is built
Climate Nearest CIBSE station, elevation corrected Geometry and U-values Element by element, with sources Boundary temperatures Temperature beyond each element Ground floor DHDG tables, whole-floor ratio Ventilation Three figures per room Room design heat loads and W/m² Heat generator sizing Whole dwelling, sizes the heat pump Emitter and pipework sizing Sum of room loads, sizes radiators and pipes Reported separately and never added together. No safety margin is folded into either total.
Climate. The nearest of the eight CIBSE Guide A reference stations is selected by distance from the site, with a correction for elevation. The design external temperature is cross-checked against the DHDG's climate zone value, and a difference of more than 1 K is flagged. The annual mean external temperature used for solid ground floors comes from MCS guidance document MGD 007.
Geometry and U-values. Each room carries an explicit list of building elements, so each area is recorded directly for that element. U-values follow a set order of preference: measured or manufacturer-declared values; values calculated from an identified construction; the DHDG's construction tables; and finally RdSAP 10 default values where there is no survey evidence. A thermal bridging allowance is added to each element individually.
Boundary temperatures. Each element uses the temperature on its far side. Solid ground floors use the annual mean external temperature, as MIS 3005-D requires. For a wall shared with a neighbouring home, 15°C is assumed beyond it (5°C if that home is unheated), and heat gained from a warmer adjoining room is counted.
Ground floors. U-values come from the DHDG's ground floor tables by construction, soil type, wind shielding and the ratio of exposed perimeter to area, calculated once for the whole floor.
Ventilation. Air leakage is built up from air permeability, the exposed envelope and shielding. The method produces three ventilation figures: one for emitter and pipework sizing, one for heat generator sizing, and one for checking minimum modulation. A separate check flags any room whose calculated air supply falls below the recommended minimum, and reports it as a warning without changing the design load.
Totals and sizing. Room loads and specific heat losses are reported for every room, and heat generator sizing and emitter and pipework sizing are reported as two separate totals. The heat loss result contains no added safety margin. MIS 3005-D requires the heat pump to meet at least 100% of the calculated load, and any margin belongs at equipment selection as a labelled figure.
Source tables. The DHDG's 179 constructions (over 9,000 tabulated values) were transcribed and checked for physical consistency. Where this check identified published values that appear to be in error, the engine flags each affected lookup and keeps the printed value, so every result can still be checked against the guide.
A selection of the defaults applied when evidence is missing:
| Parameter | Default | Source |
|---|---|---|
| Thermal bridging, typical existing dwelling | 0.10 W/m²K on every element | DHDG Table 2-9 |
| Air permeability | By built form and age band, e.g. semi-detached 9.2 m³/(h·m²) at 50 Pa | DHDG Table 2-11 |
| Minimum ventilation check | 0.5 air changes per hour for living rooms, kitchens, bathrooms and WCs with windows | DHDG Table 2-14 |
| Adjoining dwelling | 15°C heated, 5°C unheated | DHDG 2.5.2.3 |
| External design temperature | Nearest CIBSE Guide A station, 99% or 99.6% column | MIS 3005-D:2025 Table 2 (Issue 3.0 Table 2) |
| Annual mean external temperature | 18 regional values | MCS MGD 007 |
| Room height where not recorded | 2.4 m, assumed and flagged on the result | RdSAP 10 section 3.9.1(b) 'normal room height' |
| Building height for the air leakage pressure factor, where not recorded | 2.7 m per storey, used only to choose the DHDG Table 2-13 band | Engine assumption, recorded |
| Safety factor | None | MIS 3005-D:2025 §3.4.1(e) (Issue 3.0 §5.5.1(e)) |
Citing HeatLens
Until a tagged public release is available, please cite the HeatLens design heat load engine by name and version (currently 0.1.1), with the date of access. Each calculation result records the engine version that produced it. A formal citation is intended to accompany the first tagged release.
Get in touch
To discuss a collaboration, a validation study or data sharing, email hello@heatlens.co.uk. To enquire about access for teaching or research, use the Apply for education access button below.
Work with HeatLens
Interested in a collaboration, a validation study or using HeatLens in teaching?
Email hello@heatlens.co.uk to start a conversation, or send an education enquiry for your institution.