U-values, bridged constructions, the Glaser condensation check, annual moisture analysis, and 2D interstitial condensation.
An assembly is Psiclops's model of a real building element's cross-section — a wall, roof, floor, or ground floor — and it's the starting point for every U-value and Glaser check. There are two ways to build one, chosen to match how complex the real construction actually is. The simple route describes it as an ordered stack of layers — material, thickness, and whether a layer is bridged by a timber stud or joist — quick to set up and enough for most straightforward walls, roofs, and floors using the standard ISO 6946/ISO 13370 methods. The 2D route has you upload a DXF drawing of the real cross-section instead, which Psiclops meshes and solves with the same finite element solver used for ψ-value junctions — needed whenever the 1D layer-stack model can't represent the geometry at all (an I-joist, a metal web/posi-joist, an irregular material zone), or whenever getting an accurate result genuinely requires a full 2D solve rather than a 1D approximation. Either route feeds the same free U-value and Glaser checks. A repeating element like an I-joist or metal-web bay doesn't need redrawing from scratch each time either — build it once, the same way (1D layers, or a 2D drawing if its geometry needs that), then embed it directly as a single material inside a different assembly's own layer, or a junction's polygon, via the "Add from Assembly" picker. For the 1D layer-stack route specifically, pick "Internal" as its element type — the same option also covers a genuine internal partition (a stud wall between two rooms, say), since both share the same trait: no real external face. Either way, it still gets a real U-value for reference, but skips the Glaser condensation check entirely, since there's no genuine internal-to-external vapour pressure gradient to check against.
A U-value (thermal transmittance) measures how much heat passes through a building element per unit area per degree of temperature difference, in W/(m²·K). Lower is better. The standard method for simple, thermally uniform layers is the simple summation method set out in ISO 6946:2017 §6.2 (BR 443's own convention). Psiclops calculates U-values from the actual layer build-up you describe, and where a layer is bridged — a timber stud or joist running through insulation — it automatically switches to the ISO 6946:2017 §6.3 combined method rather than treating that layer as solid insulation. For bridging geometries the combined method itself isn't accurate for — I-joist panels, metal web joists (posi-joists), or other irregular cross-sections — a full 2D finite element analysis is needed instead; see the next two questions.
A timber stud or joist running through an insulation layer conducts heat far more readily than the insulation around it — softwood's thermal conductivity is roughly 3–4 times that of typical mineral wool. Treating that layer as 100% insulation, which is what the simple ISO 6946:2017 §6.2 summation method effectively does if bridging isn't accounted for, produces a U-value that flatters the real construction, sometimes substantially, depending on the bridged fraction. This isn't a minor rounding issue for building control or SAP/HEM purposes: BR 443 — the UK's own conventions for U-value calculations, itself built on ISO 6946:2017 — requires bridging to be accounted for using the ISO 6946:2017 §6.3 combined method for exactly this reason. An unbridged U-value submitted for a timber-frame wall or a joisted floor is not a compliant figure and can be challenged by a Building Control officer or SAP assessor. Psiclops applies the combined method automatically whenever a layer is marked as bridged, deriving the bridging fraction from the physical widths you enter rather than asking you to calculate a percentage yourself.
ISO 6946:2017's own combined method has a stated accuracy limit: if the upper and lower resistance limits it calculates for a construction diverge by more than 1.5:1, the standard itself says the method is not considered sufficiently accurate — and Psiclops flags this automatically whenever it happens. An I-joist (a timber or engineered flange either side of a thin web) or a metal web joist — a lattice of light-gauge steel struts between timber chords, often sold as a posi-joist — is exactly this kind of case: the bridging geometry is more complex than a simple repeating stud, typically with a large air fraction and a cross-section that genuinely varies with depth. For these constructions, a full 2D finite element analysis is the correct method, not the 1D combined method. The same EN ISO 10211 compliant FEM solver Psiclops already uses for ψ-value junctions can be applied to a drawn cross-section of the joist/insulation detail via the DXF-based 2D assembly route, giving a result the 1D method's own validity check would otherwise flag as unreliable.
Psiclops applies the surface resistances defined in ISO 6946:2017 Table 1, which vary by heat flow direction. For walls (horizontal heat flow): Rsi = 0.13, Rse = 0.04 m²K/W. For roofs (upward heat flow): Rsi = 0.10, Rse = 0.04 m²K/W. For exposed floors (downward heat flow): Rsi = 0.17, Rse = 0.04 m²K/W. Ground floors use ISO 13370:2017 which has a different approach — see the Ground Floors FAQ. An "Internal" element (used for a genuine internal partition, or any assembly meant to be embedded elsewhere) uses Rsi = Rse = 0.13 m²K/W on both faces when evaluated on its own — the same reasoning as a wall's internal side, since neither face is exposed to wind-enhanced external conditions.
They're excluded automatically. A standalone Internal assembly's own U-value correctly includes Rsi = Rse = 0.13, because its two faces genuinely meet room air. But once you embed it as a layer inside a different assembly (via the 🧱 "Add from Assembly" picker), only its material resistance carries over — its own surface resistances are stripped out, because its faces are no longer exposed to air directly; they're sandwiched between the parent assembly's other layers. The parent assembly's own Rsi/Rse are then applied once, at its own true outer faces. This means the same 1D construction — say, a timber-and-insulation stud zone — correctly reports a higher U-value when evaluated standalone (as a real internal partition) than when the identical layer stack is embedded inside a larger roof or wall build-up, and that difference is expected, not an inconsistency.
Psiclops calculates a real effective thermal conductivity for the cavity automatically, from its thickness, the direction of heat flow through it, and the emissivity of the surfaces bounding it — rather than asking you to look up or estimate a value yourself. It applies the equations in BS EN ISO 6946 Annex B directly, as BR 497 §2.4 requires, rather than interpolating the published summary table (that table is itself just those same equations tabulated for ordinary, non-reflective surfaces, so the two agree exactly wherever both apply — but only the equations can also handle a low-emissivity surface or an irregularly shaped air space). On the 1D layer-stack route — the common case, since a typical UK partial-fill cavity wall (masonry outer leaf, a residual unfilled cavity, insulation, inner leaf) is built this way — add the cavity as its own layer, give its thickness and heat-flow direction, and Psiclops derives the resistance and converts it to an effective λ for you. On the 2D DXF route, a cavity is drawn as its own region like any other material and gets the same treatment automatically, including BR 497 §2.4.3's method for an irregular or divided air space such as the voids inside a box lintel. Either way, you don't need to know the underlying air-gap convention yourself to get a correct result.
The air spaces Psiclops calculates are unventilated ones, which is why the picker names them that way. That covers more cases than it first appears, because BS EN ISO 6946 — and BR 497 §2.4.1.2 with it — splits ventilated cavities three ways by the size of the openings to outside, and two of the three are then treated exactly as unventilated anyway. A minimal cavity, with openings up to 500 mm² per metre of length for a vertical cavity (or per m² of surface for a horizontal one), is treated as unventilated. So is a slightly ventilated one, between 500 and 1500 mm². Only a well ventilated cavity, at 1500 mm² or more, is genuinely different: there the cavity's resistance and that of everything outboard of it are disregarded entirely, and an external surface resistance is applied at the inner leaf instead. Psiclops does not apply that treatment for you, but it is straightforward: omit the cavity and every layer outboard of it from the drawing, and tag the face this exposes with the “External sheltered” boundary condition. That is the one to use, not plain “External” — BR 497 asks for a surface resistance “corresponding to that for still air”, which is 0.13 m²K/W (or 0.10 for upward heat flow), where plain External is the wind-exposed 0.04 and would understate the construction. A typical UK partial-fill masonry cavity with normal weep holes is not well ventilated and needs none of this.
A Glaser analysis (also called interstitial condensation risk assessment) checks whether water vapour will condense within an assembly at the winter design condition. It works by tracking the vapour pressure and saturation pressure at each interface through the construction. If the vapour pressure meets or exceeds the saturation pressure at any interface, condensation risk is indicated at that point. Failing Glaser does not necessarily mean the construction is wrong — but it means the design needs to be reconsidered or evidence provided that the condensation will be managed. Psiclops runs Glaser automatically alongside every U-value calculation, free of charge. For a bridged layer — a stud or joist within insulation — Psiclops checks each material's own path separately ("at stud", "at insulation") rather than blending them into one figure, since moisture doesn't diffuse sideways across a bridge the way heat does, and a blended result could mask a real risk in whichever path actually performs worse.
Psiclops uses UK winter design conditions per BS EN ISO 13788: internal 20°C at 65% relative humidity, external 0°C at 80% relative humidity. These represent the steady-state worst-case winter condition. The Glaser method is a steady-state analysis — it does not model seasonal drying potential. For constructions that fail Glaser under these conditions but may be acceptable due to summer drying, a full hygrothermal assessment using WUFI or similar dynamic simulation would be required.
Vapour resistivity (measured in MNs/g·m) describes how resistant a material is to the passage of water vapour — the higher the value, the more it resists vapour movement. It is needed for the Glaser analysis. Values are published in material datasheets and in BS EN ISO 10456. Common values: mineral wool 5–10, PIR insulation 50–150, brick 25–50, dense concrete 60–100, vapour control layers 300,000+. If you do not have a value for a specific material, Psiclops will flag it and allow you to use a conservative default.
The Glaser method is a steady-state analysis using fixed winter design conditions. It does not account for: seasonal drying potential in summer, hygroscopic moisture storage in materials, the effect of rain-driven moisture loads, or dynamic temperature and humidity variation. It is a screening tool — a pass means the construction is unlikely to have condensation problems; a fail means further investigation is needed, not necessarily that the construction is unacceptable. Psiclops's annual condensation analysis (see next question) extends the check across a full year without needing a separate hygrothermal licence, though it remains a steady-state method. For a full transient hygrothermal assessment, dynamic simulation (WUFI, DELPHIN, or similar) is still the right tool.
Yes — an annual condensation analysis, available through Psiclops Assist for both 1D assemblies and 2D DXF-derived assemblies and junctions (the 2D version is covered in its own question below). Instead of one worst-case winter snapshot, it runs BS EN ISO 13788's own Annex C monthly moisture-accumulation procedure across a full year, showing whether moisture that condenses in the coldest months fully dries out over the rest of the year — validated to within ±0.00003 kg/m² against the standard's own published Table C.3 worked example (see the calculation verification page for the full published case set). Which climate it runs against depends on what your project tells it. If the project has a site address Psiclops can geocode, the analysis uses a real location-specific climate normal (a Typical Meteorological Year from the European Commission's PVGIS service), and that run is the compliance basis — a real climate for the real site is a better answer than a national average. The standard's fixed UK-wide Annex C.1 reference climate is then run a second time alongside it, clearly separated and labelled for comparison, so an assessor can still see the nationally comparable figure and check that the conclusion does not hinge on which climate was assumed. Where no location can be resolved, the fixed Annex C.1 reference climate is used on its own and is the compliance basis. Whichever applies, the report names the climate it used rather than leaving you to infer it. It's included in the cost of an Assist session, no separate charge. It is still a steady-state monthly method, not a full transient hygrothermal simulation modelling rain or real dynamic weather — for that, WUFI or similar remains the right tool.
Yes. The standard Glaser method only works by walking vapour pressure through a single ordered inside-to-outside stack of layers — it can't be applied directly to a 2D DXF-derived assembly or a junction, since neither has one layer path to walk. Psiclops instead runs a genuinely 2D check: a second finite element solve using each material's vapour resistivity (µ) or Sd value in place of thermal conductivity, producing a full 2D vapour-pressure field across the actual drawn cross-section. That field is compared against saturation pressure at every point in the mesh, not just at fixed layer interfaces — so it can catch a risk location a 1D method has no way to represent, such as where a thermal bridge locally concentrates both cold temperatures and vapour pressure. This applies to both 2D assemblies and junctions. The worst point found is marked directly on the vapour-pressure plot and identified by name — either as sitting on the interface between two named materials, or well within a single material — so the result points to somewhere specific in your actual construction, not just an unlabelled coordinate.
Every material assigned in the drawing needs a vapour resistivity (µ) or Sd value — the same information the standard 1D Glaser check needs. If even one material is missing this, the check is skipped entirely with a clear message identifying which material needs it, rather than guessing or silently using an incomplete result. Air cavities are handled automatically — still air's own vapour resistance factor is 1 by definition, so this is filled in and locked automatically rather than left for you to enter.
Effective Sd is the equivalent vapour resistance (in metres of still air) of the whole 2D cross-section, derived from the same solved field used for the risk check — the vapour-diffusion equivalent of an assembly's effective thermal conductivity. It lets a 2D construction be compared on the same basis as a simple 1D layer stack, or reused as a single composite material elsewhere.
Yes — available on demand through Psiclops Assist, the same way the 1D annual condensation analysis is (see the question above). Like the 1D version, it runs BS EN ISO 13788's own Annex C monthly moisture-accumulation procedure across a full year against the standard's fixed UK reference climate, showing whether moisture that condenses in the coldest months fully dries out over the rest of the year rather than accumulating indefinitely.
What's different is how it's done: the classic 1D annual method walks a handful of named layer interfaces in a fixed inside-to-outside order, which only exists for a simple layer stack. A 2D DXF-derived assembly or junction has no such order, so this method instead re-solves the full 2D vapour-pressure field every month and works node by node across the finite element mesh — but only at genuine material interfaces, exactly the same physical basis the 1D method itself uses: the boundary between two different materials, or the construction's own outermost external-facing surface. A point well within a single homogeneous material is never treated as a condensation plane, however locally supersaturated it might read, because BS EN ISO 13788's own model doesn't recognise one there either — this is the 2D method staying faithful to the standard's own theory, not a simplification of it.
In months where a given interface point's vapour pressure would meet or exceed saturation pressure — or where moisture from a previous month hasn't yet fully evaporated — that specific mesh node is held ("constrained") at exactly its saturation pressure using a Dirichlet boundary condition, the same standard finite element technique used to fix a known value at a specific point in a solved field, and the vapour field is re-solved with that constraint in place. The solver's own "reaction" at each constrained node — mathematically the same idea as a reaction force at a fixed support in a structural FEM model — gives the true rate of moisture condensing or evaporating at that exact point that month.
This is the precise 2D generalisation of how the 1D method already works (comparing flux in against flux out at a condensation plane), just solved directly for arbitrary geometry instead of a simple two-resistor layer stack, and it needs no assumption about which direction is "inside to outside" at all — including genuinely 2D cases the 1D method has no way to represent at all, such as condensation concentrating at a corner's own internal junction between materials. Each interface node's own moisture is accumulated month to month, clamped at zero, and carried forward exactly as the 1D method already does; the construction passes only if every point's accumulated moisture returns to zero by the end of the year.
This method has been cross-validated directly against the EN ISO 13788-validated 1D method on the standard's own Annex C.5 worked example, redrawn as an equivalent 2D cross-section, and matches it to within a fraction of a percent (see the calculation verification page for the published case set). One figure is reported differently from the 1D check: results are given in kg per metre of the drawing's own section depth, not kg per m² of wall face — a 2D cross-section, which may be a corner rather than a flat wall, has no single well-defined face area the way a 1D layer stack does, so this is the honest equivalent rather than a number dressed up to look directly comparable.
No, and this distinction matters. Psiclops's interstitial condensation checks — both the classic 1D Glaser method and the new 2D version — are steady-state calculations using fixed winter design conditions, the same family of method described in BS EN ISO 13788. Neither models real weather, moisture storage within materials, or drying over time the way a full transient hygrothermal simulation (WUFI, DELPHIN, or similar) does. This is the standard screening method SAP assessors expect and BR 497/Part L compliance is built around — genuinely useful for catching a real risk, but not a substitute for a full hygrothermal assessment on a higher-risk or non-standard construction.
U-value and Glaser checks are free regardless of which method actually computes them — the simple 1D layer-stack summation, or, when the geometry needs it, the same finite element solver used for ψ-value junctions (see the assembly question above). The FEM solver itself isn't what Psiclops charges for. What's paid is specifically a ψ-value junction calculation's own deliverable: a BR 497 compliant report, isotherm plots, SAP Table K1 classification, and a competent person declaration — a U-value or Glaser check produces none of that. It returns its result immediately on screen, with nothing generated or filed away, whichever method computed it.
U-value and Glaser calculations are synchronous — the result is returned immediately, typically in under one second. There is no queue, no waiting, no polling. You describe your assembly, click calculate, and see the result instantly.
ψ-value calculations use a finite element solver which takes longer than a simple U-value summation. Typical calculation time is 10–60 seconds depending on the complexity of the junction geometry and the mesh density required. The calculation runs asynchronously — you submit the job and Psiclops notifies you when it is complete. You do not need to keep the browser open while it runs.
Your first U-value, Glaser check, and ψ-value are free. No card required.
Get started free →