Dynamic Stall Characteristics and Modelling of Time-Varying Pitching Kinematics

This paper presents an experimental investigation into how complex time-varying pitching kinematics influence dynamic stall characteristics, demonstrating the limitations of static pitch rate definitions for stall prediction and proposing modifications to the Goman-Khrabrov model to better capture nonlinear aerodynamic responses.

The Gist

Imagine you are riding a bicycle up a steep hill. If you pedal at a steady, slow pace, you might start to feel the tires slipping (stalling) at a certain point. But what if you suddenly pedal much faster? You might find you can climb even steeper before the tires slip. This is the basic idea behind Dynamic Stall: when an object (like a helicopter blade or a wind turbine) moves its angle quickly, it can handle much steeper angles before "stalling" (losing lift) compared to when it moves slowly.

This paper investigates a tricky question: Does the speed of the movement matter, or does the change in speed (acceleration) matter just as much?

Here is a breakdown of the research using simple analogies.

1. The Setup: The "Swinging Door" Experiment

The researchers used a NACA0018 airfoil (a wing shape) in a water channel. Think of this wing as a door on a hinge.

• The Goal: They swung the door open (pitched it up) at different speeds and patterns.
• The Patterns:
  • Linear: Swinging the door open at a perfectly constant speed.
  • Accelerating: Starting slow and swinging faster and faster.
  • Decelerating: Swinging fast at first and then slowing down.
• The Measurement: They watched exactly when the "slip" (stall) happened and how much "lift" (force) the door generated before slipping.

2. The Big Discovery: The "Traffic Light" Rule

For a long time, scientists believed that to predict when a wing would stall, you only needed to know how fast it was moving at the exact moment it hit the "danger zone" (the static stall angle).

Think of the static stall angle as a Red Traffic Light.

• Old Theory: If you are driving at 60 mph when you see the red light, you will skid a certain distance before stopping. If you are at 30 mph, you skid less. The speed at the light determines the skid.
• The New Finding: The researchers found that for the timing of the stall (how long it takes to stop after seeing the red light), the old theory is actually correct! Whether you were speeding up or slowing down before hitting the light, the "skid distance" (stall delay) was almost the same if your speed at the light was the same.

The Analogy: It's like a runner crossing a finish line. Whether they were sprinting faster or slowing down just before the line, the time it takes for their heart rate to peak after crossing the line depends mostly on how fast they were going at the moment they crossed.

3. The Twist: Acceleration Changes the "Peak"

While the timing of the stall didn't change much based on acceleration, the result did.

• Accelerating Motion (Speeding Up): If the wing is speeding up as it approaches the stall angle, it generates more lift (a bigger "push") before it finally slips. It's like a runner who is still sprinting as they cross the finish line; they carry more momentum.
• Decelerating Motion (Slowing Down): If the wing is slowing down, it generates less lift and slips earlier. It's like a runner who is tired and slowing down as they cross the line; they have less momentum.

So, while the "clock" for the stall starts ticking at the same time regardless of acceleration, the height of the wave (maximum lift) depends on whether the wing is speeding up or slowing down.

4. Fixing the Computer Model: The "Smart Lag"

The researchers tested a popular computer model (the Goman-Khrabrov model) that tries to predict these forces.

• The Problem: The original model treated the wing's "memory" as a simple delay. It assumed that if the wing is moving fast, the air "lags behind" by a fixed amount. This worked fine for constant speeds (like a car on cruise control).
• The Glitch: When the wing was speeding up or slowing down, the model got confused. It predicted the stall would happen too early for slowing wings and too late for speeding wings.
• The Fix: The researchers realized the "lag" has two parts:
  1. Reaction Time: How fast the air reacts to the current speed (this changes if you accelerate).
  2. Vortex Formation: A physical process (like a whirlpool forming) that takes a fixed amount of time to happen once it starts, regardless of what you do next.

The Metaphor: Imagine you are trying to stop a heavy shopping cart.

• Reaction: How hard you push the brake depends on how fast you are going right now.
• Inertia: Once the wheels lock, the cart takes a fixed distance to stop because of its weight, no matter if you were speeding up or slowing down before you hit the brakes.

The original model mixed these two up. The new "Modified Model" separates them: it calculates the reaction based on the current speed, but the "vortex formation" (the fixed stop distance) is based on the speed the wing had when it first hit the danger zone.

5. Why Does This Matter?

This research is crucial for designing:

• Helicopters: Their blades flap and twist wildly.
• Wind Turbines: Especially vertical ones that spin and change angles constantly.
• Micro-drones: Which need to be agile and quick.

By understanding that the timing of the stall is predictable (based on speed at the critical moment) but the force depends on acceleration, engineers can build better models. This means they can design machines that are safer, more efficient, and less likely to break under the stress of sudden, jerky movements.

In a nutshell: The paper tells us that while the moment a wing stalls is predictable by its speed at that exact second, the power it generates depends on whether it's speeding up or slowing down. And to predict this correctly, our computer models need to stop treating the air like a simple delay and start treating it like a complex dance between reaction and momentum.

Technical Summary

1. Problem Statement

Dynamic stall is a critical phenomenon in unsteady aerodynamics, affecting the performance and structural integrity of systems like helicopter rotors, vertical-axis wind turbines, and micro air vehicles. While the delay in stall onset (the time between exceeding the static stall angle and actual flow separation) is well-understood for linear, constant-rate pitch motions, its behavior under time-varying (nonlinear) kinematics remains less explored.

Previous studies often characterize time-varying motions using the instantaneous pitch rate at the moment the static stall angle is exceeded ($\dot{\alpha}_{ss}$). While this parameter successfully predicts stall delay for linear motions, it is unclear if it adequately characterizes the complex force responses (lift overshoot and post-stall fluctuations) of motions with varying acceleration (e.g., accelerating or decelerating pitch rates). Furthermore, existing semi-empirical models, such as the Goman-Khrabrov model, rely on an "effective angle of attack" formulation that assumes a constant lag based on instantaneous pitch rate. This study investigates whether this formulation holds for nonlinear kinematics and proposes necessary modifications.

2. Methodology

The study combines experimental fluid dynamics with semi-empirical modeling.

Experimental Setup:

• Facility: SHARX recirculating water channel at EPFL.
• Model: NACA0018 airfoil ($c=0.15$ m, span=0.58 m) at $Re = 60,000$.
• Flow Control: Zigzag tripping tapes were used to force turbulent boundary layer transition, preventing laminar separation bubbles that could complicate force data.
• Kinematics:
  • Quadratic Motions: Prescribed to enforce constant second derivatives (constant acceleration/deceleration). These included accelerating ($\ddot{\alpha} > 0$), decelerating ($\ddot{\alpha} < 0$), and linear ($\ddot{\alpha} = 0$) motions.
  • Range: Pitch angles from $0^\circ$ to $30^\circ$. Non-dimensional pitch rates ($\dot{\alpha}c/2U_\infty$) ranged from 0.0001 to 0.04.
  • Control: Motions were executed with high precision (error $<0.1^\circ$).
• Measurements: Six-degree-of-freedom load cells measured aerodynamic forces. Dynamic stall onset was identified by the peak lift coefficient ($C_{L,max}$).

Modeling Approach:

• Generalised Goman-Khrabrov (GK) Model: A first-order state-space model predicting flow separation based on a separation point variable $X$.
• Original Formulation: Uses an effective angle of attack $\alpha_{eff} = \alpha(t) - \tau_2 \dot{\alpha}(t)$, where the lag is determined by the total stall delay $\tau_2$ and the instantaneous pitch rate.
• Proposed Modification: The authors hypothesized that the total stall delay consists of two distinct physical processes: a reaction delay (kinematic-dependent) and a vortex formation delay (invariant physical process). They proposed separating these in the effective angle definition.

3. Key Contributions

1. Validation of $\dot{\alpha}_{ss}$ for Stall Delay: The study confirms that for nonlinear motions, the stall onset delay (in convective time) follows the same universal power-law relationship as linear motions when characterized by the pitch rate at the static stall angle ($\dot{\alpha}_{ss}$).
2. Identification of Acceleration Effects on Force: While the timing of stall delay is insensitive to acceleration, the angle of attack at which stall occurs and the resulting maximum lift are significantly influenced by pitch acceleration.
3. Model Limitation Diagnosis: The paper identifies that the standard Goman-Khrabrov model fails for nonlinear motions because it treats the lag from reaction and vortex formation as a single term dependent on instantaneous pitch rate, which varies during the motion.
4. Novel Model Modification: A new formulation for the effective angle of attack is proposed that separates the reaction delay (dependent on instantaneous $\dot{\alpha}(t)$) from the vortex formation delay (dependent on characteristic $\dot{\alpha}_{ss}$).

4. Results and Discussion

Stall Characteristics:

• Stall Delay: Across all 126 test cases (linear and quadratic), the stall onset delay $(t_{ds} - t_{ss})U_\infty/c$ collapsed onto a single power-law curve defined by linear motions. The difference between accelerating and decelerating motions was statistically negligible ($<5.7\%$), confirming that the delay is primarily a function of $\dot{\alpha}_{ss}$.
• Stall Angle ($\alpha_{ds}$) and Lift ($C_{L,max}$):
  • Accelerating motions ($\ddot{\alpha} > 0$) delayed stall to higher angles of attack and produced higher peak lift coefficients.
  • Decelerating motions ($\ddot{\alpha} < 0$) caused stall to occur at lower angles of attack with reduced peak lift.
  • This indicates that while the "clock" for stall is set at the moment of crossing $\alpha_{ss}$, the subsequent acceleration profile dictates the magnitude of the lift overshoot.

Model Performance:

• Original GK Model: Accurately predicted the attached flow region but failed to predict the correct stall onset timing and post-stall lift for nonlinear motions.
  • For decelerating motions, the model predicted stall too early (underestimating delay).
  • For accelerating motions, the model predicted stall too late (overestimating delay).
  • Reason: The original model uses the instantaneous pitch rate $\dot{\alpha}(t)$ for the entire lag. In decelerating motions, $\dot{\alpha}(t)$ drops below $\dot{\alpha}_{ss}$, reducing the calculated lag and triggering premature stall prediction.
• Modified GK Model: By redefining the effective angle of attack as:
  $$\alpha_{eff} = \alpha(t) - [(\tau_2 - \tau_1)\dot{\alpha}(t) - \tau_1\dot{\alpha}_{ss}]$$
  The model successfully separated the kinematic reaction (using instantaneous rate) from the invariant vortex formation time (using the characteristic rate at stall onset).
  • Outcome: The modified model significantly reduced prediction errors in stall onset timing and improved the correlation ($R^2$) for post-stall lift, particularly for high-acceleration cases.

5. Significance

This research bridges a critical gap in dynamic stall modeling for complex, time-varying kinematics found in real-world applications (e.g., wind turbine blades in gusts or helicopter rotors in maneuvering).

• Practical Utility: It validates that engineers can use simple linear-motion stall delay correlations to predict the timing of stall for complex motions, provided they use the pitch rate at the static stall angle.
• Modeling Advancement: The proposed modification to the Goman-Khrabrov model allows for accurate prediction of aerodynamic loads in nonlinear scenarios without requiring computationally expensive CFD simulations. This is crucial for real-time control systems and structural load analysis in aerospace and renewable energy sectors.
• Physical Insight: The study clarifies that the stall development process has two distinct phases: an initial kinematic reaction phase sensitive to instantaneous acceleration, and a subsequent vortex formation phase that proceeds at a fixed physical timescale once initiated.