Simulating ice accretion and subsequent ice shedding on a rotating propeller is a challenging multiphysics problem. A particularly frustrating situation can occur when the ice accretes correctly in FENSAP-ICE / ICE3D, but the ice never detaches from the blade — even when the adhesion shear strength is reduced to extremely low values.
This article examines the possible reasons behind this behavior in a typical CFX → Airflow Import → DROP3D → ICE3D workflow.
🧊 The Typical Setup
Consider a rotating propeller with parameters such as:
- Propeller radius: approximately 0.6 m
- Rotational speed: approximately 3000 RPM
- Glaze icing conditions: approximately -5 to -10°C
- Ice thickness: approximately 1–3 mm
- Two CFX domains: stationary + rotating
- DROP3D configured with rotational body forces
- ICE3D ice shedding model enabled
- Very low ice adhesion shear strength
- Glaze Advanced ice model
At first glance, this appears to be a reasonable setup. However, several different pieces of physics and data transfer have to work correctly before ICE3D can actually predict shedding.
🔄 First Question: Is the Rotational Information Actually Reaching ICE3D?
The most important thing to verify is that the rotational velocity information used by the shedding model is actually being transferred through the workflow.
It is not enough for the CFX rotating domain to use the correct angular velocity.
There is a difference between:
- CFX solving a rotating flow field,
- DROP3D using rotational effects,
- ICE3D receiving the correct rotational information for its shedding calculation.
A correct rotating CFX solution therefore does not automatically guarantee that ICE3D is applying the expected centrifugal loading to the accumulated ice.
⚙️ Check the Rotation Axis and Angular Velocity
For a propeller rotating at 3000 RPM, the angular velocity is approximately:
ω = 3000 × 2π / 60 ≈ 314.16 rad/s
For a point at radius r, the centrifugal acceleration is:
ac = ω²r
At the propeller tip, r = 0.6 m:
ac ≈ 59,200 m/s²
That corresponds to roughly 6,000 g.
Therefore, if the ice is genuinely being subjected to the expected rotational acceleration, the tip region should experience a very large centrifugal loading.
If no shedding occurs even with extremely low adhesion strength, this is a strong reason to investigate whether the rotational information is being used by the ICE3D shedding calculation as expected.
🧊 Low Adhesion Does Not Automatically Mean Ice Must Shed
One important misconception is that simply reducing the adhesion shear strength should force the ice to detach.
The shedding model generally depends on a comparison between the loads acting on the ice and the resistance associated with ice attachment.
Conceptually, the condition can be viewed as:
Applied load > Adhesion resistance
However, the relevant loading is not necessarily just the centrifugal force.
Depending on the model implementation, additional effects can influence the detachment criterion, including:
- ice density,
- ice geometry and thickness,
- surface orientation,
- local acceleration,
- aerodynamic loading,
- rotational velocity,
- adhesion parameters,
- local stress distribution.
Therefore, changing one parameter by several orders of magnitude is not necessarily sufficient to trigger shedding if another required input is missing or incorrectly interpreted.
🌀 CFX Rotating Domain vs. ICE3D Rotational Physics
This is one of the most important areas to investigate in a two-domain workflow.
CFX can correctly solve a rotating reference frame, while ICE3D may require its own rotational information for the ice-shedding calculation.
These are not necessarily the same thing.
A useful troubleshooting approach is therefore to verify independently:
- CFX rotating domain angular velocity.
- DROP3D rotational body-force settings.
- ICE3D wall rotational velocity.
- Rotation axis direction.
- Rotation origin.
- Units of angular velocity.
- Whether the rotational data is actually included in the imported solution.
A mismatch in any one of these can produce a perfectly reasonable-looking flow solution while preventing the expected shedding behavior.
📐 Pay Special Attention to the Rotation Origin
The rotation axis and origin are especially important for a propeller.
The centrifugal acceleration depends on the distance from the rotation axis:
a = ω²r
If ICE3D interprets the rotation origin differently from CFX, the effective radius used in the shedding calculation may be incorrect.
This could result in a major difference between the centrifugal force you expect physically and the force actually applied in the shedding calculation.
🔬 Check the Imported Data — Not Just the CFX Results
Another useful diagnostic is to inspect the actual data passed from the airflow solution into FENSAP-ICE.
Do not assume that because the CFX solution contains rotation, every required rotational quantity is automatically available to ICE3D.
Look for evidence that the imported solution contains the required quantities at the blade surface.
For example:
- velocity components,
- wall velocity,
- surface orientation,
- temperature,
- pressure,
- rotation parameters.
If possible, compare a known stationary case with the rotating case and determine exactly which quantities change after the Airflow Import step.
🧪 Perform a Simplified Test Case
Before trying to debug the complete propeller simulation, create a simplified verification case.
For example:
- Use a simple rotating cylindrical surface.
- Apply a known angular velocity.
- Generate a controlled ice layer.
- Use an extremely low adhesion value.
- Run the shedding calculation.
If shedding still does not occur, the problem is probably related to the rotational/shedding workflow rather than the propeller geometry.
This type of reduced test case is often much easier to diagnose than a complete 3D propeller simulation.
⚠️ Check Whether the Shedding Model Is Actually Active
Another surprisingly common problem is that the shedding model is selected in the interface but is not actually being applied to the relevant surface or calculation stage.
Verify:
- shedding is enabled globally,
- the correct blade wall is included,
- the correct ice model is selected,
- the shedding calculation is executed after accretion,
- the correct ice state is used as input,
- the solver output contains shedding-related quantities.
Simply seeing ice accumulation in the visualization does not prove that the shedding stage is being executed correctly.
📊 Why the Tip Is the Best Diagnostic Location
For a rotating propeller, the tip is an excellent location for troubleshooting because centrifugal acceleration increases linearly with radius:
ac = ω²r
At the hub, r is small.
At the tip, r is maximum.
Consequently, if the rotational contribution is correctly implemented, the tip region should be one of the locations where the centrifugal contribution is strongest.
If the tip behaves exactly like the hub despite a large rotational speed, investigate the rotational input and shedding model before changing the adhesion value again.
🛠️ Recommended Troubleshooting Sequence
Instead of changing many parameters simultaneously, test the workflow systematically:
- Verify RPM → rad/s conversion.
- Verify the rotation axis.
- Verify the rotation origin.
- Verify CFX rotating-domain settings.
- Verify DROP3D rotational body forces.
- Verify ICE3D wall rotation settings.
- Verify the Airflow Import data.
- Check whether shedding is enabled on the actual ice-covered surface.
- Run a simplified rotating test case.
- Only then modify adhesion strength and ice parameters.
🚀 Final Takeaway
If ice accretion works correctly but no shedding occurs at all, even with extremely low adhesion shear strength, repeatedly lowering the adhesion value is probably not the best next step.
The first thing to investigate is the data path between CFX, Airflow Import, DROP3D and ICE3D.
A rotating CFX domain can be physically correct while the ice-shedding calculation may not be receiving or interpreting the rotational information in the way you expect.
For a propeller operating at 3000 RPM and a radius of 0.6 m, the theoretical centrifugal acceleration at the tip is enormous — approximately 59,200 m/s². If the predicted behavior shows no sensitivity to rotation, that makes the rotational data transfer and ICE3D shedding implementation the first areas worth checking.
The key lesson: don't assume that a correct rotating flow solution automatically means that every downstream icing model is using the same rotational physics.
If you are troubleshooting a similar CFX + FENSAP-ICE + DROP3D + ICE3D workflow, checking each stage independently can save a huge amount of time.
💬 Discussion
Have you successfully simulated ice shedding from a rotating propeller using CFX and FENSAP-ICE?
If you have experience with this workflow, sharing the exact Airflow Import, DROP3D and ICE3D rotational settings could help identify whether this is a setup issue, data-transfer limitation or model compatibility problem.
0 Comments