Ahmed Body mesh convergence

Costas · March 17, 2016, 13:41

Hello guys, im new to the topic so im trying to figure things out. I want to carry out a mesh convergence for the ahmed model. Im using a 25 deg. slant for 40m/s. I know i have to approximate double my mesh size to see whether me residuals have converged. I should be plotting a graph to show convergence of the number of elements against some other variable. What would be the most appropriate variable to take and how woould i take its value? Thank you

LuckyTran · March 17, 2016, 15:09

I highly recommended you look into Patrick Roache's paper/book on grid convergence index. It's a very good introduction and is also a minimum standard for many journals.

The variable you choose should be one of the parameters you are studying. What are you investigating in the Ahmed body? If it's local velocity then plot how the local velocity at a few characteristic points changes with mesh size.

Costas · March 18, 2016, 08:59

Well i'm trying simply to get the mesh to converge and later i will be comparing turbulence models, therefore i would assume that any variable chosen would be useful. So im using drag and 'k' in this case to see if they converge. Would that be enough?

LuckyTran · March 18, 2016, 09:48

grid convergence is very subjective so you have to pick variables to compare. I recommend using the same variables in your convergence monitors used to determine iterative convergence to compare across simulations for your grid convergence. I mean actual monitors and not residuals.

Global variables like drag are great (if that is what you actually want to compare). Suppose, you did run the turbulence models and you wanted to compare their "performance." How would you determine which model performs better? The one that gives the most accurate drag? If so, the sure, go ahead and use drag.

Local variables like x-velocity, y-velocity, z-velocity, k, etc. are primitive variables and can give you a lot of insight. But these are field variables, you must still select at which x,y,z locations you want to compare them. I would choose velocity over k for comparison since different models model k differently. Even worse would be to pick the dissipation rate for a k-epsilon model, which you cannot compare directly with the specific dissipation rate for a k-omega model.

Costas · March 18, 2016, 11:23

OK so then im doing a silly mistake of taking values of 'k'. However im also taking values for the lift and drag which should prove convergence from what i understand. Thank you.

March 17, 2016, 13:41	Ahmed Body mesh convergence	#1
Costas New Member Constantinos Join Date: Mar 2016 Posts: 19 Rep Power: 10	Hello guys, im new to the topic so im trying to figure things out. I want to carry out a mesh convergence for the ahmed model. Im using a 25 deg. slant for 40m/s. I know i have to approximate double my mesh size to see whether me residuals have converged. I should be plotting a graph to show convergence of the number of elements against some other variable. What would be the most appropriate variable to take and how woould i take its value? Thank you

March 18, 2016, 09:48		#4
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,676 Rep Power: 66	grid convergence is very subjective so you have to pick variables to compare. I recommend using the same variables in your convergence monitors used to determine iterative convergence to compare across simulations for your grid convergence. I mean actual monitors and not residuals. Global variables like drag are great (if that is what you actually want to compare). Suppose, you did run the turbulence models and you wanted to compare their "performance." How would you determine which model performs better? The one that gives the most accurate drag? If so, the sure, go ahead and use drag. Local variables like x-velocity, y-velocity, z-velocity, k, etc. are primitive variables and can give you a lot of insight. But these are field variables, you must still select at which x,y,z locations you want to compare them. I would choose velocity over k for comparison since different models model k differently. Even worse would be to pick the dissipation rate for a k-epsilon model, which you cannot compare directly with the specific dissipation rate for a k-omega model. Last edited by LuckyTran; March 18, 2016 at 11:34.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[snappyHexMesh] any mesh for Ahmed Body ?	adrieno	OpenFOAM Meshing & Mesh Conversion	5	August 6, 2019 05:56
[GAMBIT] Ahmed Body 3D mesh	Saima	ANSYS Meshing & Geometry	19	February 10, 2018 11:59
Gambit problems	Althea	FLUENT	22	January 4, 2017 03:19
Moving mesh	Niklas Wikstrom (Wikstrom)	OpenFOAM Running, Solving & CFD	122	June 15, 2014 06:20
[ICEM] Ahmed body Hexa mesh (with legs)	Far	ANSYS Meshing & Geometry	15	February 5, 2014 04:25

March 17, 2016, 15:09		#2
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,676 Rep Power: 66	I highly recommended you look into Patrick Roache's paper/book on grid convergence index. It's a very good introduction and is also a minimum standard for many journals. The variable you choose should be one of the parameters you are studying. What are you investigating in the Ahmed body? If it's local velocity then plot how the local velocity at a few characteristic points changes with mesh size.

March 18, 2016, 08:59		#3
Costas New Member Constantinos Join Date: Mar 2016 Posts: 19 Rep Power: 10	Well i'm trying simply to get the mesh to converge and later i will be comparing turbulence models, therefore i would assume that any variable chosen would be useful. So im using drag and 'k' in this case to see if they converge. Would that be enough?

March 18, 2016, 11:23		#5
Costas New Member Constantinos Join Date: Mar 2016 Posts: 19 Rep Power: 10	OK so then im doing a silly mistake of taking values of 'k'. However im also taking values for the lift and drag which should prove convergence from what i understand. Thank you.