CFD Analysis: High Residuals + Inconstant Values

AndreP · October 30, 2017, 17:34

Hey guys!

My name is André, im a MSc Motorsport Engineering student and i'm having some problems with my Star CCM+ Analysis. Although I have some basic and solid knowledges of CFD Simulations and Analysis, i don't consideer myself a Pro user, so I think I could really use some of your input.

What I want to do, is study the aerodynamics influences by varying the distance on 2 similar bodies. So, to start it all, i'm starting with one body only and trying to have a benchmark. The flow is 50 m/s and has a 5º entry angle (to simulate more real world aspects).

The problem: Although my mesh is quite good refined (7 milions elements) and my prism layer beeing sufficient good done (Y+ values alongside the body between 30-100). I just can't seem to have some good residuals (bellow 0.0001) or even some real values, as they flutuate a LOT. I'll try to put as many screenshots I think that may help you to help me! Any info you need, just ask me! Thanks in advance!

AndreP · October 31, 2017, 08:03

AndreP · October 31, 2017, 08:05

Miguel Parente · November 1, 2017, 22:07

Have you already tried with out angle of attack? The problems seems simple, so you shouldn't had any problem solving that with steady state. If it stills does not work try use unsteady model. Then you my have to situations, if for instance for drag result of drag is periodic, your problem is definitely unsteady. If not, run the simulation in unsteady, after a while, when you see that you velocity field and pressure are already defined, change for steady state.

AndreP · November 2, 2017, 15:35

Quote:

Originally Posted by Miguel Parente

Have you already tried with out angle of attack? The problems seems simple, so you shouldn't had any problem solving that with steady state. If it stills does not work try use unsteady model. Then you my have to situations, if for instance for drag result of drag is periodic, your problem is definitely unsteady. If not, run the simulation in unsteady, after a while, when you see that you velocity field and pressure are already defined, change for steady state.

Hey Miguel! let me start by thanking for your time!

It seem strange, but I never suspected the problem could be in the Velocity vector. I tried, as you said, with zero angle of attack and the Residuals immediatly went perfect just like the values!

Could this be a matter of bad designation of the flow direction?

How to I setup correctly the flow at 5º inlet (Simulating lateral winds)?

Miguel Parente · November 3, 2017, 18:19

Hey André. Well it could be the case, that you have a or a stationary problem with small perturbations that lead to convergence problems, or you can really have a unsteady problem. There several ways of doing it, on way, at least for me seems to be "easier" is to rotate the volume control. I mean, when you design the control volume don't design it with the axles parallels to the vehicle axles. Give them the 5º degree rotation. Then you make the normal simulation setup. I don´t know if I make myself clear.

AndreP · November 3, 2017, 19:34

Quote:

Originally Posted by Miguel Parente

Hey André. Well it could be the case, that you have a or a stationary problem with small perturbations that lead to convergence problems, or you can really have a unsteady problem. There several ways of doing it, on way, at least for me seems to be "easier" is to rotate the volume control. I mean, when you design the control volume don't design it with the axles parallels to the vehicle axles. Give them the 5º degree rotation. Then you make the normal simulation setup. I don´t know if I make myself clear.

Yes you made yourself clear, and that would be a perfect case, if my study was only for 1 car.

But as I'm going to put another car behind it, in Slipstream. It will not be the same situation if i rotate both cars 5º, you see what I mean? Thats why I was trying to solve this with a true 5º flow angle!

Miguel Parente · November 3, 2017, 20:04

Hum yes, I see your point. In that case be aware with the distance of the side wall, to be sure that the side wall do not interfere with flow around both cars. Try start the simulation in steady state, it even can be with first order discretization of the convective term. If do not converge don't worry, after some time, change the to unsteady. Be aware of the time step you choose and the number of inner iterations. You have to play with that, and to change the discretization again for second order. Look for the residuals and the variables of interest, I belive that you're looking for drag and moments. If in unsteady your residuals are good and values of drag and the other ones have converge is because the problem is steady, but you my have some perturbation that difficult converge in steady.

October 30, 2017, 17:34	CFD Analysis: High Residuals + Inconstant Values	#1
AndreP Member André Pinto Join Date: Oct 2017 Location: Brussels, Belgium Posts: 84 Rep Power: 9	Hey guys! My name is André, im a MSc Motorsport Engineering student and i'm having some problems with my Star CCM+ Analysis. Although I have some basic and solid knowledges of CFD Simulations and Analysis, i don't consideer myself a Pro user, so I think I could really use some of your input. What I want to do, is study the aerodynamics influences by varying the distance on 2 similar bodies. So, to start it all, i'm starting with one body only and trying to have a benchmark. The flow is 50 m/s and has a 5º entry angle (to simulate more real world aspects). The problem: Although my mesh is quite good refined (7 milions elements) and my prism layer beeing sufficient good done (Y+ values alongside the body between 30-100). I just can't seem to have some good residuals (bellow 0.0001) or even some real values, as they flutuate a LOT. I'll try to put as many screenshots I think that may help you to help me! Any info you need, just ask me! Thanks in advance!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
High Cm values in 2D analysis of VAWT	Pranav96	FLUENT	8	November 7, 2018 17:44
Cfd to ansys thermal to ansys structural interface	ssixr	ANSYS	17	July 31, 2015 16:18
Configuration for CFD analysis?	lolcocks	Hardware	0	June 9, 2015 12:25
CFD Online Celebrates 20 Years Online	jola	Site News & Announcements	22	January 31, 2015 01:30
Truck cfd analysis	sheth	Main CFD Forum	29	August 4, 2011 11:15

October 31, 2017, 08:03		#2
AndreP Member André Pinto Join Date: Oct 2017 Location: Brussels, Belgium Posts: 84 Rep Power: 9

October 31, 2017, 08:05		#3
AndreP Member André Pinto Join Date: Oct 2017 Location: Brussels, Belgium Posts: 84 Rep Power: 9

November 1, 2017, 22:07		#4
Miguel Parente New Member Miguel Join Date: May 2017 Posts: 10 Rep Power: 9	Have you already tried with out angle of attack? The problems seems simple, so you shouldn't had any problem solving that with steady state. If it stills does not work try use unsteady model. Then you my have to situations, if for instance for drag result of drag is periodic, your problem is definitely unsteady. If not, run the simulation in unsteady, after a while, when you see that you velocity field and pressure are already defined, change for steady state.

November 3, 2017, 18:19		#6
Miguel Parente New Member Miguel Join Date: May 2017 Posts: 10 Rep Power: 9	Hey André. Well it could be the case, that you have a or a stationary problem with small perturbations that lead to convergence problems, or you can really have a unsteady problem. There several ways of doing it, on way, at least for me seems to be "easier" is to rotate the volume control. I mean, when you design the control volume don't design it with the axles parallels to the vehicle axles. Give them the 5º degree rotation. Then you make the normal simulation setup. I don´t know if I make myself clear.

November 3, 2017, 20:04		#8
Miguel Parente New Member Miguel Join Date: May 2017 Posts: 10 Rep Power: 9	Hum yes, I see your point. In that case be aware with the distance of the side wall, to be sure that the side wall do not interfere with flow around both cars. Try start the simulation in steady state, it even can be with first order discretization of the convective term. If do not converge don't worry, after some time, change the to unsteady. Be aware of the time step you choose and the number of inner iterations. You have to play with that, and to change the discretization again for second order. Look for the residuals and the variables of interest, I belive that you're looking for drag and moments. If in unsteady your residuals are good and values of drag and the other ones have converge is because the problem is steady, but you my have some perturbation that difficult converge in steady.