🐋Bio-Inspired Wings with Tubercles–Wind Tunnel Study 

 📌 Overview 

This project investigates the aerodynamic benefits of leading-edge tubercles, inspired by humpback whale flippers, using wind tunnel experiments on 14 3D-printed wing designs. The goal was to quantify how variations in tubercle geometry (peak shape, spacing, amplitude, sweep) affect stall delay, drag, and lift-to-drag ratio (L/D) — especially in post-stall conditions.

The experiments revealed new performance-enhancing geometries and led to the discovery of an additional design parameter (peak width), offering insights applicable to MAVs, UAVs, and control surfaces.

🛠 Technical Highlights

• Designed and tested 14 different wings with sinusoidal, elliptic, triangular, and swept tubercle geometries.
  
  

• Used low-turbulence wind tunnel testing with Reynolds numbers 77,000 and 174,000.
  
  

• Developed a modular two-axis load cell and a precision worm gear pitch mechanism.
  
  

• Applied full aerodynamic corrections and uncertainty analysis to validate measurements.
  
  

• Verified results against established experiments (e.g., Hansen et al.).

🛠 Tools & Equipment

• Low-turbulence wind tunnel (Cairo University)
  
  

• PLA 3D-printed wings with internal spars
  
  

• HX711 + Arduino for digital force acquisition
  
  

• MATLAB for post-processing

🔧Experimental Setup

• All tests were conducted in a low-speed, open-return wind tunnel at Cairo University. The 3D-printed wings were mounted vertically on a custom-designed rig equipped with an inhouse modular two-axis load cell for force measurements and a worm-gear pitch controller to adjust angle of attack with a resolution of 0.2° per full turn. 

• The force data was collected using an HX711 digital amplifier and Arduino-based system, sampling at 10 Hz. Each test case was measured at angle-of-attack increments of 1°, with three repeat runs per point and 400 data samples per run. Freestream velocity was measured using a high-resolution digital manometer, and Reynolds number was controlled by adjusting tunnel speed.

• All wing designs were printed in PLA with embedded spars to ensure rigidity. Aerodynamic corrections and full uncertainty quantification were applied during post-processing using MATLAB.

 📊 Uncertainty & Statistical Methodology 

• To ensure the reliability and credibility of the experimental results, a comprehensive uncertainty analysis was performed following the guidelines of Bentley (2005) and Barlow et al. (1999). This was especially critical due to the comparative nature of the study, which involved evaluating small differences in aerodynamic performance across multiple wing geometries and Reynolds numbers. 

• Each test point was measured in triplicate, with 400 data samples per run to capture variability in lift and drag forces.

• Uncertainties were propagated from:

  1. Load cell force readings

  2. Angle of attack measurements

  3. Freestream velocity (from digital manometer)

• Key contributors to total uncertainty included:

  1. Load cell resolution

  2. Pitch gear mechanism precision

  3. Manometer accuracy

• Type A (random) uncertainties were calculated based on repeated measurements and standard deviation.
  
  

• Type B (systematic) uncertainties included sensor resolution, aerodynamic correction errors, angle misalignment, and fixture tolerances.
  
  

• The combined standard uncertainty was computed using the root-sum-square method, in accordance with AIAA and NIST guidelines.
  
  

• Expanded uncertainties (95% confidence) were reported for key parameters (CL, CD, CL/CD).

• Final uncertainty bounds confirmed that performance differences between wings were statistically meaningful and not due to measurement error.

• This validation was essential for confidently comparing and ranking wing configurations.

⚙️ Wind Tunnel Corrections

• Corrections were applied based on procedures from Barlow et al. (1999), after using the method of images to adapt for the half-model testing approach.
  
  

• Solid blockage: Accounts for the model's frontal area relative to the tunnel cross-section. This was used to adjust dynamic pressure and surface stresses.
  
  

• Wake blockage: Adjusts for the added drag due to the wake’s restriction of airflow in a closed test section.
  
  

• Streamline curvature: Corrects for the difference in flow curvature caused by tunnel walls compared to free-stream conditions.
  
  

• Downwash effects:
  
  

  1. Normal downwash change: Alters the apparent lift and drag due to induced flow from the test section boundaries.
     
     

  2. Spanwise downwash distortion: Affects the local angle of attack along the span — mitigated by keeping the wing span <80% of tunnel width.
     
     

• Horizontal buoyancy: Adjusts for pressure gradients in the tunnel unrelated to the model itself.
  
  

• Additional geometric constraints were applied:
  
  

  1. The gap between wing root and tunnel floor was kept below 0.005 times the span to avoid flow leakage without introducing frictional error.
     
     

These corrections were applied to all aerodynamic coefficients (CL, CD, and α), and their effects were quantified during verification experiments. For example, at α = 14°, the corrections led to a 0.4% drop in CL, a 7.4% increase in CD, and a 4.5% increase in α for the baseline straight wing

🧩 Key Findings

• Wings with leading-edge tubercles showed significant post-stall performance improvements over the straight (baseline) wing.
  
  

• The sinusoidal peak-only design (P7) outperformed full sine tubercle geometries in both lift-to-drag ratio and stall angle.
  
  

• Lift-to-drag ratio (CL/CD) improved by up to 83% for P7 compared to the straight wing at Reynolds number 77,000.
  
  

• Stall angle increased by up to 36% for the best-performing tubercled wings.
  
  

• Wings with elliptical and triangular peaks performed well but were generally outperformed by the sinusoidal peak-only designs.
  
  

• Introducing a new geometric variable — peak width — revealed its strong influence on post-stall behavior and overall aerodynamic performance.
  
  

• Partial trailing-edge tubercles, mimicking natural flipper morphology (e.g., SLPT7), outperformed fully tubercled trailing-edge wings in certain cases.
  
  

• Drag was reduced by over 30% in some tubercled configurations compared to the straight wing.
  
  

• The positive effects of tubercles were more pronounced at lower Reynolds numbers (Re = 77,000).
  
  

• Sweep angle had nuanced effects, with forward-swept configurations showing improved stall delay but not always better efficiency.
  
  

• Some designs with larger amplitudes and short wavelengths suffered performance degradation due to excessive flow disruption.

❌Key Findings on AR=2 Wings 

• Wings with AR = 2 were tested at three Reynolds numbers: Re = 77,000, 120,000, and 174,000 to evaluate how performance scales with freestream velocity and viscous effects.
  
  

• The study observed the same general trends as with AR7 wings:
  
  

  • Sinusoidal and peak-only tubercles delayed stall
    
    

  • Post-stall lift was increased
    
    

  • CL/CD improved compared to straight wings
    
    

• However, high uncertainty at certain angles of attack — especially at higher angles and lower Reynolds numbers — limited the conclusiveness of some comparisons.
  
  

• In particular, comparisons between SLPT2, S2, and SLT2 were noted as unreliable due to:
  
  

  • Large uncertainty bars overlapping across multiple data points
    
    

  • Inability to confidently attribute aerodynamic gains to geometry rather than measurement noise
    
    

• For this reason, those comparisons were excluded from the performance summary, and were explicitly considered as inconclusive.

🔭 Future Directions

• Extending tubercle analysis to unsteady flapping and rotary wings
  
  

• Embedding vortex tracking using PIV
  
  

• Integration with computational optimization workflows