[VladCFD]

Good afternoon,

This forum has been often useful to me in the past, hopefully someone will be able to help today

I am working on a very long pipe (~ 1 Km) with a small diameter (< 60 mm). Water is flowing outside the pipe, in the direction of the pipe (x direction, no radial component).
I am interested in the drag of this underwater pipe, in particular the drag resulting from friction. Therefore, I consider a very small Y+.

My question : Can I use 2D axisymmetric (axisymmetric swirl ?) to extract a reliable drag, or are there 3D features that should be expected and modelled, significant for the drag ? The question is of importance, as the involved amount of cells is huge I have been searching the litterature, but I cannot find works with such long geometries.

Thanks a lot for your time.
Kind regards,

Vlad

[vinerm]

If the pipe is straight, then you don't really need CFD. A simple correlation would be way faster and equally accurate. If there are some other effects, such as, cross-flow or pipe oscillations due to flow or pipe is not straight, then CFD is the only option.

[VladCFD]

Thank you, but let's assume my boss wants me to run CFD anyway
► Would 2D axisymmetric be accurate with respect to 3D, for a drag study ?
► Should I use opt for 2D axisymmetric or 2D axisymmetric swirl ? I don't really understand the nuance.
Thank you,
Vlad

[vinerm]

Yes, you can use either 2D axisymmetric or 2D axisymmetric swirl. Both will have the assumption of straight pipe built-in.

Axisymmetric implies flow can only be in axial and radial direction, while with swirl, flow is also allowed in circumferential direction. However, for a 1 km long pipe, swirl cannot maintain itself until and unless there is a mechanism to maintain it. So, for your case, use 2D axisymmetric. And you don't need to run it for 1 km long duct. Just run for a few cm with translational periodic condition. This will give you pressure drop and drag per unit length. Then, you can just multiply by the ratio of 1000 m and length of your domain to get final pressure drop and drag value.

Do note though that translational periodic assumes that the flow is always fully developed. If that assumption is not applicable to your case, then you just use a long domain and 2D axisymmetric. You may not require a 1 km long pipe because after a certain length, drag (or essentially shear stress) will become almost independent of axial position.

[VladCFD]

Thank you very much !