Difficulty Resolving Surface Quantities Using IBM in Incompact3d (Re = 3900 Cylinder Flow)

[GSP-021]

Hi,

I'm currently using Incompact3d to simulate flow over a cylinder at Re = 3900. The global flow parameters velocity, turbulent fluctuation etc. align well with benchmark data from the literature, which suggests that the overall flow field is being captured reasonably well.

However, I'm encountering significant challenges in accurately resolving surface quantities such as:

Heat flux

Wall shear stress

Pressure coefficient

I've experimented with both Lagrangian reconstruction and the cubic spline-based immersed boundary method (IBM) implementations available in Incompact3d, but neither approach seems to yield satisfactory results for these surface-level quantities.

I'd really appreciate any insights or suggestions regarding the following questions:

Is this limitation in resolving surface quantities inherent to the IBM implementation in Incompact3d?
Are there known limitations with IBM (especially spline-based or Lagrangian interpolation) when it comes to near-wall gradients or surface scalar transport?

Would it be advisable to implement a dedicated routine for scalar transport near the surface (e.g., for temperature or passive scalars) to improve accuracy in surface heat flux estimation?

Are there any specific strategies, parameter tweaks, or code-level modifications you would recommend based on your experience with IBM in Incompact3d, particularly for better resolving near-wall quantities?

Any feedback, pointers to relevant references, or practical experiences would be hugely appreciated. Thanks in advance for your time.

[mathrack]

Hello,

Surface quantities can be estimated locally or using a control volume. Currently, some routines are available in forces.f90 when using a control volume.

If you want to investigate or develop some advanced IBM capabilities for the fluxes on the surface, I would suggest to look at the PhD from R. Vicente Cruz https://theses.fr/2021POIT2302.

Cheers,
Mathrack

[GSP-021]

Hi,
Thank you again for your message.
No, my geometry is not a pipe — it's a small cylinder placed inside a larger domain. I'm using this setup to [25D X 20D X 3D domain size with cylinder diameter as D].

[mathrack]

Have you tried to obtain some θ plot on a simpler 2D cylinder case ?

[GSP-021]

No,So far I have only tested the 3D case. For the 3D case at Re = 300 using trilinear interpolation, the surface flux calculation was somewhat acceptable with fewer fluctuations. However, for the Re = 3900 case, the method is not working properly. Additionally, I encountered another issue: I am unable to accurately capture the temperature gradients near the wall. Specifically, the gradient near the stagnation point is lower than in the surrounding areas, which is incorrect. This problem persists even with a very high-resolution grid.
I’ve attached a figure for reference, where the temperature gradient is calculated using Para-view for the entire domain, but the view is zoomed in near the cylinder.

[mathrack]

Hello,
The IBM reconstruction strategy depends on the pencil orientation of the dataset. It is not possible to calculate gradients with Paraview in case of IBM.
If you want to visualize the temperature gradient in paraview, you should compute grad(T) in Incompact3d and export it.

[GSP-021]

Hi,
I calculated grad(T) similar to the vorticity post-processing and its comes out similar to what I have got in paraview. The values of grad(T) comes out to be lower at font stagnation zone.
Thank You!