|
[Sponsors] |
Question of modeling for Reynolds Stress Model |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
February 14, 2022, 20:37 |
Question of modeling for Reynolds Stress Model
|
#1 |
Senior Member
Sangho Ko
Join Date: Oct 2021
Location: South Korea
Posts: 135
Rep Power: 4 |
Hello everyone.
I have very many questions of Reynolds Stress Model. I've studied k- SST Model before but I can't understand review paper of Reynolds Stress model. First I'm going to write down link of this paper. https://ntrs.nasa.gov/api/citations/...9950018160.pdf Cause of the many questions, it can be irritating. Sorry. But I need help. 1. Equation (37) in page 8 A. Why laplacian of fluctuating pressure is emerged? I guess it is emergied to derive Equation (39) and Equation (40). B. Why it is derived like that? Why fluctuating pressure is related to fluctuating velocity gradient and mean velocity gradient? 2. Equation (25) in page 7 A. Why we call this 'pressure-strain correlation'? I think equation (25) is related to fluctuating pressure and fluctuating velocity gradient. 3. Equation (39) in page 8 and Equation (40) in page 9 A. Meaning of 'slow pressure-strain' and 'rapid pressure-strain' In this paper author said these mean 'slow pressure-strain' and 'rapid pressure-strain'. 4. Equation (43) in Page 9 A. Meaning of I guess dot means (Material differentiation). Is it right? 5. Equation (44) in page 9 A. Meaning of 6. In page 9, "The fundamental assumptions underlying two-equation models are that the turbulence is locally homogeneous and an equilibrium state is reached where attain constant values that are largely independent of the initial conditions."I guess this also have similar meaing with upper. A. Indepedent Constant What it the exact meaing of independent constant about initial condition? Thanks for reading this long questions. I'm not good at English so please excuse my bad English. Also I've studied k- SST Model before but I'm not enough. I'm not professional about turbulence. But I'm interested in that field. So help me please Thanks Last edited by FluidKo; February 15, 2022 at 00:33. Reason: Question |
|
February 15, 2022, 01:42 |
|
#2 |
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,683
Rep Power: 66 |
1. That's the Poisson equation and for an incompressible flow, the pressure satisfies the Poisson equation. The gradient of pressure appears in the momentum equation. You can take the divergence of the momentum equation (and then substitute the continuity equation into it) to derive this Poisson equation.
2. It doesn't matter whether or not you call it fluctuating pressure and fluctuating strain because the correlation between a constant (i.e. a mean) and anything is zero. There's no ambiguity. 3. Pressure can be decomposed into two parts (one slow and one rapid). Both satisfy the Poisson equation. The rapid correlation has a source term that contains the mean velocity gradient. The slow correlation has a source term that does not and the slow correlation is primarily non-linearities that indirectly interact with the mean flow. Rapid/slow is with regard to the rate at which they interact and distort one another (pressure and strain). You don't have to call it rapid and slow if you don't want to because rapid/slow comes from distortion theory. 4&5. Yes (almost always) but (sometimes) no. Overdot is time rate of change which for the purposes of this conversation means material derivative (without going in-depth into why the material and total derivative would not be the same). 6. Take for example a turbulent viscosity which depends only on the mean velocity and mean velocity gradient (in other words, every Boussinesq-like turbulence model). The turbulent viscosity (i.e. the turbulence) depends only on the current state of the flow. It has no hysterisis and doesn't know where it has been. This is equivalent to assuming that the constants are independent of the initial conditions. The turbulence is treated as a state variable, which is a big conceptual assumption. If your turbulent viscosity doesn't contain a path integral in it, your turbulence model falls into this category. If paths are involved then we need to be really careful about that material derivative in 4&5. |
|
February 15, 2022, 06:17 |
I have questions to your answer.
|
#3 | |
Senior Member
Sangho Ko
Join Date: Oct 2021
Location: South Korea
Posts: 135
Rep Power: 4 |
Quote:
First, in third question, you've told me 'rapid pressure-strain ' interacts with mean velocity gradient directly and 'slow pressure-strain ' interacts with mean velocity gradient indirectly. But I don't know about that. I think this is related with distortion theory that you've told me. Right? I want to get recommendation of paper to learn what is distortion theory. Edit: Oh I've found content related to this from the text book by S.B.Pope. Second, in sixth question, you've told me path. But I can't understand the meaning of the path. In my opinion, I guess path that you're meaning is pathline(trajectory) because you've talked 'There is no hysterisis in turbulence. There is no function(Doesn't matter with initial condition) to describe accurate turbulent state that varies over each time. It is random.', If we know all of informations about pathlines for whole particles, then it is possible to consider every situation at every each times. It means we can find change of turbulence viscosity and any other change about turbulence at every each time along the pathline.(We know history and where turbulence has been) Do I understand it well? Thanks Last edited by FluidKo; February 17, 2022 at 05:20. |
||
February 18, 2022, 11:18 |
|
#4 |
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,683
Rep Power: 66 |
Real eddies and turbulence have some memory of where they've been. If you know the history, you can maybe build models to take that into account. But that is not the way we do this today and that is the fundamental assumption being mentioned. We don't need a complete history of every fluid particle dating back to the Big Bang to solve this. We know that that temporal correlation time-scale of turbulence is small. At the same-time, we know that time-scale is not zero. We only need the history up to this time-scale and maybe a little bit more.
Let's say you are doing RANS/URANS using your favorite turbulence model today). Your mut is at most a function of x,y,z, and u,v,w at the current time t. That is, given that u=u(x,y,z,t), v=v(x,y,z,t), and w=w(x,y,z,t) you have mut=mut(x,y,z,u(x,y,z,t),v(x,y,z,t),w(x,y,z,t)). Only the current time t exists in this dependency (and not any previous time t'). And if only the current t is allowed, then you have assumed that all your turbulence constants depend only on the current t and not the initial conditions. This issue also applies to LES but in a limited scope in the SGS model. A generalized turbulence model would need to know where every eddy has been at last up to t' going back the temporal time-scale. That's what you would need to do to no longer make this fundamental assumption. |
|
February 18, 2022, 11:46 |
|
#5 | |
Senior Member
Sangho Ko
Join Date: Oct 2021
Location: South Korea
Posts: 135
Rep Power: 4 |
Quote:
|
||
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Question about matching of solver and turbulence model | louistse | OpenFOAM Running, Solving & CFD | 1 | February 1, 2017 21:36 |
Modeling a thermomechanical bending process | Oueg18 | ANSYS | 0 | December 14, 2016 16:07 |
Overflow Error in Multiphase Modelling with Two Continuous Fluids | ashtonJ | CFX | 6 | August 11, 2014 14:32 |
Reynolds Stress model in CFX vs Fluent | Tim | CFX | 1 | October 7, 2009 06:19 |
Questions about the Reynolds stress model | empirer2002 | Main CFD Forum | 1 | January 5, 2006 07:37 |