AEROLAB – Manual (English edition)

Digital wind tunnel for printed parts Version 3.0 · 19 October 2026 · Thomas Schüller · Syntagma Forschung · neutral edition

1 Purpose

AEROLAB computes, in your browser, how air flows around a body and what force the body experiences. It lets you compare shapes before printing them: which housing shape produces less drag, which angle of attack gives the best lift-to-drag ratio, what a corner radius does. All computation runs on your device; no data is transmitted.

Not intended for: replacing measurement or proof. The computation is two-dimensional and valid for low Reynolds numbers (section 9). It is not suitable for certification, safety proof or load-bearing parts, and it is not provided for armaments, military, weapons or militarily usable aviation.

2 Quick start

Open aerolab.html (or aerolab-neutral.html) in a current browser.
Pick an example case, e.g. "Vortex street behind a cylinder (Re 100)". The simulation starts at once.
Choose a view (velocity, vorticity, pressure) and switch on the smoke lines.
Read the readings (Cd, Cl, Strouhal number) and watch for yellow warnings.
Change shape, angle or Reynolds number and press Restart.

3 Tabs

Tab	Content
Wind tunnel	Single run with live image, readings and force history
Study / polars	Series of runs with one varied quantity (e.g. angle of attack)
Optimisation	Automatic shape search with objective, bounds and minimum thickness
Part and export	STL, mass, image with readings, save and load design
Limits and help	Validity range, reference values, notes

Shapes: cylinder, ellipse, airfoil (NACA 4-digit), inclined plate, rounded box, streamlined body with tapered tail. Resolutions: Coarse (fast, comparison only), Standard, Fine (slow, for verification). Channel walls: slip walls (wind tunnel without wall friction) or periodic (row of identical bodies approximating a free stream).

4 Reading the results

Cd drag coefficient, Cl lift coefficient (positive upwards), Cl/Cd lift-to-drag ratio, St Strouhal number (vortex-shedding frequency), Re Reynolds number, τ relaxation time (close to 0.5 means risk of instability), blockage fraction of channel height blocked by the body.
The reference length is the length (chord, diameter) of the body, not its frontal height. Coefficients are per unit span (two-dimensional).
Above about Re 47 a cylinder sheds a vortex street; Cl and Cd then oscillate and the readings show the mean values.
Warnings appear if blockage exceeds 15 %, the body is thinner than 4 cells, the run becomes unstable, or Re is too high for the grid. They mean the coefficients are distorted.

5 Study / polars

Select a variable, range and number of points (2 to 15). Each point is a full run to convergence. The chart shows Cd, Cl and Cl/Cd; the table can be saved as CSV (semicolon-separated). Cancel stops the series and keeps the points obtained so far.

6 Optimisation

Set the initial shape and parameters in the Wind tunnel tab.
In Optimisation, select the free variables and their bounds, then the objective (minimise drag, maximise lift-to-drag, maximise lift, maximise downforce).
Enter chord/length and the minimum part thickness in millimetres; designs below it are penalised.
Set the maximum number of runs (30 is a good start) and use Coarse resolution for the search.
Start. The log shows every run; the chart shows the best objective so far.
Press Re-check fine: initial and best design are recomputed at fine resolution. Only this comparison should be reported.
Apply to wind tunnel loads the best design back.

The search is a local method; a different starting design can lead to a different result. With vortex shedding the objective fluctuates and the search becomes noisier.

7 Part and export

STL (prism of the profile; closed-shell edge check), optional embossed signature on the end face (can be switched off; omitted on profiles that are too thin), mass estimate (PLA, PETG, ASA, PA; infill), PNG with readings, animation (10-second video recording of the running simulation with readings, as WebM file; needs a browser with recording support such as Chrome, Edge or Firefox) and JSON to save or load a design. The angle of attack belongs to the installation and is not part of the STL geometry.

8 Printing test bodies

Print the body standing (span vertical) so layer lines follow the profile contour in the flow direction. Use fine layers (0.12 to 0.16 mm), smooth the surface, check trailing-edge thickness in the slicer, and measure on the prototype before relying on simulated values. Design, sizing and suitability remain with the user.

9 Validity and limits

Two-dimensional; real cylinder flow becomes three-dimensional above about Re 190.
Reliable for comparison in the laminar range: up to about Re 200 for compact bodies and about Re 1000 for slender profiles. Typical printed parts in air operate at Re 10,000 and above, where AEROLAB shows at most coarse tendencies, not coefficients.
Absolute coefficients of airfoils are not yet checked against an independent reference (see validation report). Comparisons between shapes are permissible.
Wall blockage distorts coefficients (cylinder at Re 40: Cd 1.81 versus 1.5 to 1.6 free stream, i.e. about one sixth to one fifth high at 10 % blockage). Compare bodies only at similar blockage.
Not included: surface roughness, layer lines, end plates, three-dimensional effects.

10 Troubleshooting

Observation	Cause	Remedy
"Computation unstable"	Re too high for the grid	Lower Re, raise resolution, switch on turbulence damping
Strongly fluctuating readings	Unsteady flow	Read mean values, run longer
Drag unexpectedly high	Body too thin for the grid, high blockage	Finer resolution, heed warnings
Optimisation stalls	Start at a bound, flat objective	Widen bounds, change start design
STL "not closed"	Self-intersection at extreme parameters	Stay inside the bounds

Disclosure: the author runs a commercial 3D printing and additive manufacturing business and has a commercial interest in printed parts. Developed with the help of an AI assistant.

Methods, validation and references: papers.