Urban Wind Vector Engine HT (Beta)

Heuristic urban wind screening: terrain-parallel wind deflected by building wakes, corner channelling and gaps, coloured as a heatmap draped on the terrain (DrapeHeatmap, default) or flat at a custom elevation, with vectors and streamlines.

Site Analysis Nickname: WindEngineHTVer

Overview

Simulates urban wind fields using terrain-parallel raycasting. Outputs a perfectly flat, crisp XY pixel-screen heatmap at a custom elevation.

Inputs (19)

  • Run (Boolean): Global execution toggle switch
  • TerrainMesh (Mesh): The underlying site topography
  • ContextBuildings (Mesh): Lightweight mesh context structures
  • WindDirection (Vector): Travel vector of incoming air
  • WindSpeed (Number): Baseline velocity metric
  • AnalysisHeight (Number): Human pedestrian offset
  • GridSpacing (Number): Resolution size of pixel elements
  • CustomColors (Colour): Custom color spectrum override
  • HeatmapHeight (Number): Z offset applied to the heatmap: added to the terrain bounding box top in Flat mode (DrapeHeatmap=false), or added to each point's own local terrain height in Drape mode (DrapeHeatmap=true)
  • WakeLengthRatio (Number): Wake length as a multiple of the obstructing building's height (typical near-wake range 6-8x)
  • SlopeSpeedup (Number): Coefficient controlling how much upward terrain slope accelerates wind (0 = no slope effect)
  • CornerTangency (Number): How tangent (0-1) wind must be to a wall before corner-channeling kicks in; lower = stricter
  • CornerBoost (Number): Speed boost coefficient applied when wind channels around a building corner
  • GapBoost (Number): Extra speed boost applied when a point is squeezed between two facing building walls (venturi/tunneling), scaled by how narrow the gap is
  • Iterations (Integer): Relaxation passes per point: each pass re-evaluates wake/corner/gap using the previous pass's own direction, letting a point settle through multiple sequential building interactions instead of a single snapshot. 1 = original single-pass behavior
  • Damping (Number): Blend factor (0-1) between a point's previous iteration and its newly computed one. Lower = slower, more stable convergence; 1.0 = no damping (replace outright each pass)
  • DrapeHeatmap (Boolean): true = heatmap follows terrain relief, offset from each point's own local ground height. false = a single flat plane, offset from the terrain's bounding-box top
  • SeedBox (Box): Optional bounding box - every valid analysis grid point inside it is used as a streamline seed instead of the default sparse grid sampling
  • StreamlineMaxSteps (Integer): Safety ceiling on streamline length. A streamline normally stops on its own once it exits the mesh, goes solid, or its speed dies out - this only guards against a pathological case looping forever

Outputs (9)

  • VelocityHeatmap (Mesh): Flat, unwelded crisp horizontal pixel-tile matrix
  • WindVectors (Line): Spatial direction markers
  • VectorColors (Colour): Velocity color map matching lines
  • Streamlines (Curve): Continuous particle flow paths
  • TagPoints (Point): Anchor coordinates for Text Tag
  • PlainMesh (Mesh): Original topography mesh without vertex colors
  • VelocityValues (Number): Raw unformatted velocity values, aligned with WindVectors
  • Info (Text): Component information and interpretation
  • Dashboard Data (Text): JSON legend payload (min/max speed + the actual color ramp in use) - feed directly into the Legend Geometry component's Color Legend input

Method, assumptions and simplifications

The same notes appear in the component's Info output.

Method

  1. A grid of GridSpacing is laid over the terrain bounding box; each cell is ray-cast down to the terrain and lifted by AnalysisHeight. The wind direction is projected onto the local face plane and speed is scaled by (1 + dirZ × SlopeSpeedup), so uphill flow speeds up and downhill flow slows down.
  2. Points inside a closed building mesh get speed 0. Otherwise three factors multiply the baseline speed: wake, corner and gap.
  3. Wake: a ray is cast upwind; if it hits a building at distance d below L = min(building height × WakeLengthRatio, 0.35 × site diagonal), the factor is max(0.12, (d / L)²).
  4. Corner: for every building wall within 2.5 × GridSpacing whose normal is nearly parallel to the flow (|n · dir| < CornerTangency), the direction is bent towards the wall tangent with weight b = 1 − dist / (2.5 × GridSpacing) and speed is multiplied by (1 + CornerBoost × b). When several buildings qualify, the bend direction is taken from the nearest qualifying wall; their speed boosts still multiply.
  5. Gap: if another building within the same radius has a wall facing the nearest one (normals dot product below −0.2), speed is multiplied by 1 + GapBoost × (1 − corridor width / (5 × GridSpacing)). GapBoost 0.45 caps the gap speed-up near the 1.45 to 1.5 times seen in wind-tunnel and CFD studies.
  6. With Iterations above 1 each point re-evaluates the factors from its own previous direction and blends the result with Damping; no information is exchanged between points.
  7. Outputs: tiles coloured by speed normalised between the run minimum and maximum (CustomColors or a default blue to pink ramp), vectors of length 0.5 × speed, and streamlines traced from seed cells by stepping 0.2 units along the nearest cell direction until they leave the grid, hit a solid cell or reach MaxSteps. Comfort % is the share of open points at or below 5 m/s.

Assumptions

  • WindSpeed is the reference speed at pedestrian height in m/s; model units are metres.
  • ContextBuildings are closed meshes; their height is the bounding-box height.
  • The terrain is a height field seen from above; one wind direction per run.

Simplifications and limits

  • This is a rule-based screening model, not CFD: no pressure or continuity solve, no turbulence, no vertical wind profile, and no interaction between neighbouring points.
  • Wake, corner and gap factors are empirical rules of thumb with tunable coefficients; results show patterns, not reliable absolute speeds or comfort classes.
  • Colours are relative to the run speed range, and the comfort share uses a single 5 m/s threshold, not a Lawson or similar criterion with frequency of exceedance.
  • Streamlines use the nearest-cell direction (no interpolation) with a fixed 0.2 step; without SeedBox the default seeds follow an irregular stride over the grid, not a regular upwind row.
  • The gap rule uses one closest point per building, so corridors inside one U-shaped or courtyard building are not detected.

Workflow

  • Terrain mesh -> TerrainMesh; context massing (closed meshes) -> ContextBuildings; prevailing wind direction and speed from a wind rose.
  • Dashboard Data -> Legend Geometry; for more reliable results use Fast CFD and Streamlines.

References

  • Lawson T.V. (1978), The wind content of the built environment: pedestrian wind comfort criteria (for context on the 5 m/s comfort threshold).