Low-Speed Open-Circuit Wind Tunnel with Force Balance for Aerofoil Lift-Drag Measurement

Low-Speed Open-Circuit Wind Tunnel with Force Balance is a build-and-measure project our batch took on because a B.Tech aero programme should not end its aerodynamics lab with a rat trap and a paper glider. We designed and fabricated an open-circuit, suck-down tunnel: a 300 × 300 mm settling chamber with a drinking-straw honeycomb and two screens, a 9:1 contraction into a 100 × 100 mm acrylic test section, a gentle diffuser, and a hooded axial fan with speed control delivering about 4–8 m/s in the working section.

The measurement side is deliberately instrumented: the model sits on a two-component sting balance built from two 0.5 kg micro load cells read by HX711 modules through an Arduino Uno, so lift and drag are sampled and logged rather than eyeballed; dynamic pressure comes from a pitot-static probe and inclined liquid manometer. A 3D-printed NACA 0012 aerofoil (coordinates from the UIUC database) is tested from −4° to +14° angle of attack, with a smoke-rake and wool tufts for flow visualisation.

A Python pipeline converts raw counts into calibration-corrected forces and then into Cl, Cd and the drag polar, overlays thin-airfoil theory computed in MATLAB/Scilab, and states an uncertainty budget from calibration weights and repeat runs. The result is a department asset that costs roughly ₹10,500, runs in a corridor, and gives every future batch a rig to defend in the viva — plus our own honestly low-Reynolds-number data to argue about.

Aerospace students meet lift and drag as equations long before they ever see a measured force. In most Indian college laboratories there is no working, instrumented wind tunnel: department tunnels are scarce, shared and often out of service, commercial desktop units cost several lakhs, and the demonstrations that exist — string-and-smoke shows — produce no numbers an examiner can check. The consequence is that concepts like stall, induced drag and centre of pressure stay theoretical, and students graduate without the single most transferable experimental skill in the discipline: running a calibrated aerodynamic measurement and stating its uncertainty.

The problem is to design and fabricate a low-speed open-circuit wind tunnel with a calibrated two-component force balance that a four-member B.Tech team can build for about ₹10,000, that logs lift, drag and airspeed digitally, and that yields defensible Cl–alpha curves and drag-polar data on a standard aerofoil — together with a documented calibration, corrections and an uncertainty budget, so every plotted point can be traced to a measurement rather than a claim.

Abstract

This project designs, fabricates and calibrates a small open-circuit wind tunnel with a two-component force balance, then measures lift and drag on a 3D-printed NACA 0012 aerofoil. The tunnel is a suck-down configuration: settling chamber with honeycomb flow straightening and screens, 9:1 contraction, 100 × 100 mm test section, diffuser and variable-speed axial fan giving 4–8 m/s. Forces are logged through two load cells on a sting balance read by an Arduino Uno over USB serial; airspeed comes from a pitot-static probe and inclined manometer. Deliverables: the working tunnel, logged Cl–alpha and drag-polar results, flow-visualisation records of attached and separated flow, and a calibration and uncertainty report.

Introduction

Every aerodynamics textbook builds from lift and drag, yet few Indian college laboratories can show either with measured numbers. We sized the tunnel by the classical rules of Barlow, Rae and Pope: contraction for flow quality, diffuser angle against separation, honeycomb cell length-to-diameter about six, and speeds giving Reynolds numbers of roughly 15,000–35,000 for a 60 mm chord — low, but honestly measured and discussed.

Existing system and gap

• Department tunnels: few colleges own one; when present they are usually shared, uninstrumented or awaiting repair.
• Commercial desktop tunnels: cost several lakhs with instrumentation included.
• Student builds published online: usually stop at flow visualisation without a calibrated balance or logged forces.
• Gap: an affordable tunnel plus balance whose design decisions, calibration and corrections are documented well enough to defend in a viva.

Proposed system

• Fabricated tunnel: MDF settling chamber and diffuser, acrylic test section, straw honeycomb, two screens, hooded axial fan with speed controller.
• Sting-mounted model with a two-component balance: micro load cells, HX711 digitisers, Arduino firmware logging counts, airspeed and angle at 20 Hz to CSV.
• Calibration: dead-weight calibration of each load cell through a thread-and-pulley rig; manometer cross-check of dynamic pressure; tare runs with the model absent.
• Analysis in Python: calibration, blockage and wake corrections per standard practice, Cl and Cd computation, drag polar, and comparison with thin-airfoil theory and XFOIL at matched Reynolds number.
• Flow visualisation: smoke rake and wool tufts documenting attached flow, transition behaviour and stall.

Feasibility

• Technical: sheet-metal, carpentry and 3D printing skills from workshop courses; Arduino logging is a second-year skill; analysis is script level. Flow quality — the hardest item — is handled by proven contraction and honeycomb ratios rather than novelty.
• Economic: indicative cost ≈ ₹10,500 (fan ₹3,000, acrylic test section ₹2,500, MDF chambers and diffuser ₹1,700, balance load cells and HX711 ₹600, manometer and pitot ₹250, prints and hardware ₹1,400, fasteners and sealing ₹1,050) — a department asset, not a consumable.
• Operational: runs from a single 13 A socket, needs no compressed air, and a full Cl–alpha sweep takes under twenty minutes once calibrated.
• Schedule: fabrication-heavy first half, calibration and test campaigns in weeks 9–12, analysis and documentation in weeks 13–14 — matching the VTU Phase-I to Phase-II progression.

In scope

• One open-circuit tunnel, suck-down configuration, 4–8 m/s, powered from a single socket.
• Two-component force measurement (lift and drag); pitching-moment instrumentation is future scope.
• One 3D-printed NACA 0012, 60 mm chord, 96 mm span, plus a flat-plate comparison model.
• Angle-of-attack range −4° to +14° in 2° steps, at two fan speeds.
• Pitot-static + inclined-manometer airspeed measurement; optional differential-pressure sensor logging.
• Corrections per standard practice (blockage, wake) and an uncertainty budget from calibration weights and repeats.

Out of scope

• Three-component balances and strain-gauge moment measurement.
• Pressurised, cooled or closed-circuit tunnels; turbulence grids beyond the honeycomb.
• Reynolds numbers above roughly 35,000 — the report discusses low-Re effects explicitly instead of hiding them.
• Smoke generation beyond a hand-held rake; no heated or particle-laden seeding systems.

We follow a design–fabricate–calibrate–measure sequence mapped onto the VTU two-phase structure: Phase-I (BXX685) delivers the survey, sizing calculations, drawings and simulation-backed design; Phase-II (BXX786) delivers fabrication, calibration, the test campaign and the report.

Test plan: (1) calibration checks — five-point dead-weight linearity on each axis; (2) repeatability — three full sweeps on one day and one on a later day; (3) functional — stall detection on NACA 0012 and the flat plate; (4) sanity — zero-lift angle near 0° for the symmetric section, drag rising monotonically past stall. Every case is recorded with input, expected, observed and status, and each objective maps to at least one test.

The rig is a three-layer system: an aero layer (tunnel and model), an instrumentation layer (balance, probe, digitisers, microcontroller) and an analysis layer (host scripts).

1. Aero layer: air is drawn through the settling chamber (honeycomb removes swirl and large eddies; screens reduce turbulence intensity), accelerated by the 9:1 contraction into the 100 × 100 mm test section, past the sting-mounted model, and decelerates through the diffuser before the fan. The pitot-static probe sits ahead of the model.
2. Instrumentation layer: the sting passes through the test-section floor to a two-axis balance — one load cell aligned with lift, one with drag — each read by an HX711 24-bit digitiser. The Arduino Uno samples both channels at 20 Hz, reads the set angle from a rotary selector (with the printed protractor as the primary reference) and streams one CSV line per sample over USB.
3. Analysis layer: a Python logger stores runs; a second script applies zero-tares, dead-weight calibration factors, blockage and wake corrections, computes Cl and Cd from dynamic pressure and planform area, and produces the plots; MATLAB/Scilab overlays the thin-airfoil prediction.

flowchart TD
  INLET["Settling chamber: honeycomb + two screens"] --> CON["Contraction 9:1"]
  CON --> TS["Test section 100 x 100 mm acrylic"]
  TS --> DIF["Diffuser"]
  DIF --> FAN["Hooded axial fan + speed controller"]
  MODEL["3D-printed NACA 0012 on sting + protractor"] --> BAL["Two-component balance: lift and drag load cells"]
  BAL -->|"millivolt signals"| HX["HX711 digitisers x2"]
  HX -->|"I2C-like serial"| UNO["Arduino Uno: 20 Hz sampling, CSV stream"]
  PITOT["Pitot-static probe"] --> MAN["Inclined liquid manometer"]
  MAN -->|"dynamic pressure, manual entry"| PC["Laptop: Python logger and analysis"]
  UNO -->|"USB serial 115200"| PC
  PC --> OUT["Cl-alpha curves, drag polar, theory and XFOIL overlays"]

flowchart LR
  A["Raw counts + tare + angle"] --> B["Apply dead-weight calibration factors"]
  B --> C["Compute forces N and dynamic pressure q"]
  C --> D["Blockage and wake corrections"]
  D --> E["Cl = L / (q S), Cd = D / (q S)"]
  E --> F["Plot vs theory and XFOIL with uncertainty bars"]

Key design numbers we defend in the viva: contraction ratio 9:1 for acceptable flow quality at low cost; honeycomb cell length-to-diameter ≈ 6; diffuser equivalent half-angle ≈ 6° to avoid separation; model frontal-area-to-test-section ratio ≈ 1.6 per cent so blockage corrections stay small; sampling 20 Hz with a 5 s settle at each angle before averaging 100 samples per point.

• Three-component balance with a pitching-moment cell and an automated servo-driven angle sweep, removing the largest manual step from the campaign.
• Differential-pressure airspeed logging with a calibrated low-range sensor, so airspeed joins the CSV stream instead of the manometer notebook.
• Turbulence-grid experiments to measure the effect of inlet turbulence on transition and stall, turning the tunnel into a research teaching tool.
• A second model family — cambered sections, a finite wing with endplates, or a car body — to widen the department's use of the rig.
• Pressure-tap model with an Arduino multi-channel manometer bank to measure surface pressure distributions and integrate Cl directly.
• Repeatable lab experiment: a PCB for the balance electronics, a written three-hour procedure and a kit list so future batches can rebuild and extend the rig.