Streamlines

Generates smooth streamlines (particle traces) through any vector field. Regular grids (e.g. Fast CFD) are read directly with bilinear interpolation; scattered points use Inverse Distance Weighting. 4th-order Runge-Kutta integration, optional speed-coloured tubes.

Site Analysis Nickname: Streamlines

Overview

Generates smooth streamlines (particle traces) through any vector field using Inverse Distance Weighting.

Inputs (12)

  • FieldPoints (Point): The locations of the vector field data
  • FieldVectors (Vector): The vectors corresponding to each FieldPoint
  • SeedPoints (Point): The starting locations for the streamlines
  • StepSize (Number): Distance to move per integration step
  • MaxSteps (Integer): Maximum number of steps per streamline (per direction)
  • SearchRadius (Number): Scattered points only: radius to search for blending local vectors (ignored for regular grids)
  • MoveUp (Number): Visual Z-offset to prevent clashing with the analysis mesh
  • Direction (Integer): 1 = both ways from each seed (default), 0 = downstream only, -1 = upstream only
  • MinSpeed (Number): Lines stop where the field is slower than this (m/s)
  • TubeRadius (Number): Radius of the speed-coloured tube mesh (0 = no tubes)
  • Palette (Text): Tube colours by local speed (palette name, e.g. Wind, Turbo, Viridis)
  • Run (Boolean): Optional. False = skip the analysis and keep showing the last results (cached). Share one toggle with the Analysis Dashboard to switch everything on/off together.

Outputs (5)

  • Streamlines (Curve): The generated streamline curves
  • Info (Text): Component documentation and stats
  • Speeds (Number): Local field speed at every streamline vertex (one branch per streamline)
  • Tubes (Mesh): One mesh of tubes coloured by local speed (needs TubeRadius > 0)
  • Dashboard Data (Text): JSON legend payload for the tube colours

Method, assumptions and simplifications

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

Method

  1. Field: if FieldPoints lie on a regular XY lattice (holes allowed, spacing = most common gap between coordinates, 5 % tolerance), vectors are read with bilinear interpolation from the 4 surrounding cells, needing at least 3 of them. Otherwise inverse distance weighting with weights 1 / d² over points within SearchRadius in plan is used.
  2. Only the X and Y components of FieldVectors are used; the ground height Z is interpolated the same way so lines drape over the field points.
  3. From each SeedPoint a line is integrated with 4th-order Runge-Kutta on the unit direction field, so each step advances StepSize along the line; Direction 1 traces upstream and downstream and joins them, 0 downstream only, −1 upstream only.
  4. A line stops when it leaves the field (or reaches a missing grid cell such as a building), local speed drops below MinSpeed, it stalls, it returns within StepSize / 2 of its seed after more than 8 steps, or after MaxSteps.
  5. Points are lifted by MoveUp. Speeds are the local vector magnitudes at each vertex; with TubeRadius above 0 a 6-sided tube mesh is built, coloured by speed between the overall minimum and maximum using Palette (default Wind).

Assumptions

  • FieldVectors are velocities in m/s (MinSpeed and the legend use m/s).
  • The flow is essentially horizontal (2.5D).

Simplifications and limits

  • Steady field: lines are streamlines of a single snapshot, not particle paths through time-varying flow.
  • Inverse distance weighting smooths the field and does not respect obstacles; with scattered points lines stop at buildings only if no field point lies within SearchRadius.
  • Tube colours are relative to the speed range of this run.

Workflow

  • Fast CFD or a wind engine grid points and vectors -> FieldPoints and FieldVectors; seed points along the upwind edge -> SeedPoints.
  • Tubes -> preview; Dashboard Data -> Legend Geometry.

References

  • Press W.H. et al., Numerical Recipes: classical 4th-order Runge-Kutta integration.
  • Shepard D. (1968), A two-dimensional interpolation function for irregularly-spaced data (inverse distance weighting).