I am trying to simulate transitional flow in a smooth circular pipe using Ansys FLUENT. Is LES is suitable for transition flow modeling? if yes, which SGS model can capture the transition behaviour?
Hi,
Well, knowing if LES is "suitable" actually depends on what you want to analyze in your simulation results. Indeed, LES will allow you to observe smaller structures than URANS. It also allows a better representation of cycle to cycle variations, but it is also more computationally expensive especially with wall bounded flows... So, first, you must define your need.
Concerning the SGS model, it depends a lot on what is available in the CFD softwre you are using. Maybe you could give us more details about it. Everyday, I am using STARCCM+ and I would personnally give a shot at the WALE model as it is especially suited for wall bounded flows.
Article Subgrid-Scale Stress Modelling Based on the Square of the Ve...
Thanks a lot Jonathan. This is indeed very helpful.
I am trying to figure out the physics of a flow through a pipe with an insert. I am not sure about the suitability of RANS models in the transitional zone. So I am trying one configuration with LES.
Are you not sure about URANS because of some previous work (your own or literature)? If you have some, could you join them to the conversation please?
If it is not the case, I would first give a shot at URANS especially if you are not familiar with CFD. Indeed, doing a first simulation with URANS (even if it is not perfect) may allow you understanding partially what is hapenning in your system and have a better understanding of the physics without the cost involved by LES, which can be much more than URANS. After this, you can eventually transfer to LES knowing better what you expect from it...
Also, be aware that a complete validation of an LES simulation may be much more demanding and challenging.
Run LES with various spatial resolutions and check whether the results you are interested in alter. Typically they will, unless the resolution is fine enough, close to DNS resolution in wall-normal direction, in axial and circumferential direction it can be coarser. Note that pipe transition is not trivial at all as there is no exponential growth of small-amplitude disturbances, rather you need so-called transient (algebraic) growth of streak-like structures with a non-small amplitude, or large-amplitude disturbances of any kind.
yes, you can use LES formulation using the dynamic SGS model (an option in ANSYS/FLUENT). However, use second order discretization (time and space) and refine the grid near the wall to ensure to have at least 3-4 nodes within y+=1
however your task is not simple and higher order own-made code would be more suitable...
The orginal question is about a specific issue: simulation of transition in a pipe.
You cannot hope to do such a simulation using RANS/URANS. This is a case where only LES or DNS can do a physically meaningful simulation.
- Badges
- Science topic