1Click 🧱 Build-up assistant (top-left of the canvas). Pick a standard junction from the list — the full SAP E-series, party-wall P-series and roof windows — or list your own wall / floor / roof layers from outside to inside.
2The junction opens as editable layers: set each material and thickness in mm (swap the insulation, change a block, adjust the slab), then press 🧱 Draw junction onto grid.
3ΨMonkey picks a grid resolution and paints the whole section — both elements plus the internal & external surfaces — with 1 m flanking arms. Fine-tune features (lintel, DPC, frame) on the grid with Paint / Erase if you need to.
4Hit ▶ Solve to get Ψ and fRsi. Then Export PDF for a Building-Control-ready report.
▸ Full reference & theory (ISO 10211, boundary conditions, f_Rsi, validation…)
Standard junctions (SAP Appendix K)
The Build-up assistant ships ready-made templates for the SAP 10 junction library — the E-series (E1–E25), party-wall P-series (P1–P8) and roof windows / ridge (R1–R4). Each drops a complete, representative 2D bridge onto the grid with 1 m flanking arms and surfaces pre-assigned. Pick one, edit its build-ups (materials and thicknesses) in the layer view, then Draw junction onto grid. The R4 ridge has a live roof-pitch slider (15–50°). Party-wall junctions model a full-fill, edge-sealed party cavity (no convective bypass, as in nearly all modern construction) — set the cavity layer to air in the editor to model an unfilled wall. The default build-ups follow a published accredited full-fill construction (U≈0.176 wall) but line the inner leaf with dot-and-dab (12.5 mm plasterboard on 15 mm dabs, R 0.22) as most UK housing is built. Each standard junction subtracts its flanking Σ(U·l) analytically on SAP internal dimensions (BR 497 convention), computed live from the layers you set. Most junctions land within ≈±0.02 W/m·K of the accredited figures; because that set is modelled with a wet-plaster lining, set the inner layer to 20 mm gypsum plaster if you want to reproduce its published values exactly. Edit the layers and both the 2D model and the convention flanking re-derive automatically.
1. Draw your cross-section
Select a material, then paint cells on the grid. Each cell is a square of the size set in "Cell size mm". Empty cells are outside the model — no heat flows through them. Paint air voids with an air/cavity material.
2. Assign boundary conditions
Paint boundary cells on the empty cells just outside each surface — BC cells are environment (air) nodes, so any material painted underneath them is ignored (such cells are flagged with a diagonal cross and a warning at solve time). Keep the full construction thickness painted, then add the BC line beside it, or use the edge dropdowns. Surface resistances per BS EN ISO 6946 are applied automatically: External Rse = 0.04; Internal Rsi = 0.13 (walls / horizontal heat flow), 0.10 (ceilings / heat flow upward), 0.17 (floors / heat flow downward). Use Adiabatic at cut-off planes.
3. Model extent (ISO 10211)
Cut-off planes should be at least 1 m from the central junction, or 3× the element thickness, whichever is greater. Use a larger grid / smaller cells if your flanking elements are short.
4. Solve
The 2D finite-difference solver (SOR-accelerated) iterates to convergence, including surface films. The heat-balance row shows the mismatch between heat entering and leaving the model — it should be well below 0.1%.
5. Read Ψ and f_Rsi
Ψ = L_2D − ΣU·l per ISO 10211, where L_2D = Q_2D/ΔT and ΣU·l is the 1D flanking heat loss. The auto mode derives U·l from clear 1D paths in the model; switch to Manual to enter your own U-values and lengths (e.g. from your U-value calculations, using the same dimension convention as your SAP/HEM assessment — internal dimensions in the UK). f_Rsi is the temperature factor at the worst internal surface point; ≥ 0.75 is required for dwellings (BRE IP 1/06). For thin-feature details — openings (lintel / sill / jamb), roof windows and party-wall-to-roof junctions — refine the cell size and use Manual U·l (enter your flanking U-values) for a certified Ψ; the auto flanking is approximate there.
6. SAP 10.x / HEM
Enter the Ψ value for the relevant junction type (E1–E25, P1–P8, R1–R11 etc.) and multiply by junction length for the ΨL contribution (W/K). HEM applies punitive defaults where Ψ values are not provided, so calculated values will usually improve the result.
Tools & shortcuts
Paint (P), Erase (E), Rectangular Marquee (R), Line (L) for pitched roofs and diagonals, Fill (F) bucket, Boundary (B). The Erase tool has a shape selector — Brush, Rectangle, Line or Fill — so you can clear areas the same way you draw them (it wipes both material and boundary). Brush sets stroke width for Paint/Erase/Line. Alt+click picks up the material or boundary under the cursor. Ctrl+Z / Ctrl+Y undo & redo. T / X / I toggle temperature, flux and isotherm overlays. Right-click erases with any drawing tool.
Isotherms & mould risk
After solving, enable Isotherms to overlay temperature contours. The dashed red contour is the critical fRsi = 0.75 line — an internal surface on the cold side of it indicates condensation/mould risk.
Library & batch reports
★ Save to Library stores the junction in this browser; ▤ Library reloads junctions and exports/imports the library as JSON.
Validation
✓ Validate runs the built-in suite: exact analytical wall checks, a Laplace field benchmark, and BS EN ISO 10211:2017 test reference case 2 (±0.1 K / ±0.1 W/m acceptance criteria). The whole junction library has also been stress-tested for energy balance, mesh-independence and mirror symmetry.
Ground junctions
Soil conductivities (BS EN ISO 13370) are under the "Ground" group. For ground-floor junctions, model a soil block extending well beyond the slab edge with adiabatic cut-offs; a full ISO 13370 treatment needs a much larger domain, so treat results as indicative.
⚠ Air cavities — always use 1 cell
The "Air cavity (ISO 6946)" and similar Rc-based materials apply a fixed total resistance per cell regardless of cell size. Always draw these exactly one cell thick, then set the cell size to match the cavity thickness. Using multiple cells will multiply the resistance incorrectly.