Simulating small features on large models

siw · February 16, 2012, 13:06

Hi,

Just wanting got get some thoughts from fellow CFDers.

For those who do external aerodynamics (esp. transonic) how do you go about your CFD workflow when assessing aerodynamics of a small feature on a large vehicle esp. when trying to capture the correct (or suitable) boundary conditions and resolving small geometrcial features whose aerodynamics can be influenced by the large vehicle geometry around it?

For instance those who work with cars (clearly not transonic

): lets say you want to model a car with and without a rear-spoiler (suppose the spoiler has some small details on it that are to be assessed as well) to determine the change in the flowfield and forces due to it being there.

So, in order to get the correct upstream flowfield would mean that you must first simulate the whole car and a large farfield domain and probably with a decent near wall resolution both with out without the simplified spoiler (no small details). Would you then conduct a more localised simulation of the spoiler (with the small details) with the nearby car geomtery (roof, rear sides and boot) and use some upstream results over the entire car to set up inlet conditions and boundary layer profiles for the smaller locilased domain? Otherwise, it would require simulating the whole car with the spoiler (including the small details) which would require very large meshes and computational resources. Maybe this last option is how Formula 1 teams work with their massive resources.

Just curious to know how others go about such work.

Thanks.

ghorrocks · February 16, 2012, 18:10

You are correct that the best approach is to simply model everything, and this is what the F1 teams do because they have access to some pretty big resources.

But for the rest of us you generally need to make some compromises. If you are looking for the relative effect of a small detail on a very large structure, then will a small model which is focussed on the small detail give good enough relative results to be useful? I mean is the optimum design in a small model close enough to the optimum design in the full model? The absolute drag/lift/whatever will not be right, but if the optimum is the same then it is a useful design tool.

Alternately you have two other approaches:
1) Use a simplified model of the upstream effects with a coarse mesh. Obviously this is a compromise between fine enough to catch the important detail and coarse enough to run on your hardware.
2) Use a full model simulation to define an inlet boundary profile for a small run focussed on the detail you are looking at. This assumes the small detail does not affect the overall flow field much.

And then of course you can always hire zillions of computers on cloud computing which is becoming easier and cheaper.

February 16, 2012, 13:06	Simulating small features on large models	#1
siw Senior Member Stuart Join Date: Jul 2009 Location: Portsmouth, England Posts: 742 Rep Power: 26	Hi, Just wanting got get some thoughts from fellow CFDers. For those who do external aerodynamics (esp. transonic) how do you go about your CFD workflow when assessing aerodynamics of a small feature on a large vehicle esp. when trying to capture the correct (or suitable) boundary conditions and resolving small geometrcial features whose aerodynamics can be influenced by the large vehicle geometry around it? For instance those who work with cars (clearly not transonic): lets say you want to model a car with and without a rear-spoiler (suppose the spoiler has some small details on it that are to be assessed as well) to determine the change in the flowfield and forces due to it being there. So, in order to get the correct upstream flowfield would mean that you must first simulate the whole car and a large farfield domain and probably with a decent near wall resolution both with out without the simplified spoiler (no small details). Would you then conduct a more localised simulation of the spoiler (with the small details) with the nearby car geomtery (roof, rear sides and boot) and use some upstream results over the entire car to set up inlet conditions and boundary layer profiles for the smaller locilased domain? Otherwise, it would require simulating the whole car with the spoiler (including the small details) which would require very large meshes and computational resources. Maybe this last option is how Formula 1 teams work with their massive resources. Just curious to know how others go about such work. Thanks.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to model many small holes in a large domain?	Bryan Woods	FLUENT	3	February 5, 2009 20:34
About simulating laminar flow through small orific	ncj	Main CFD Forum	1	August 14, 2002 05:46
Large volume and small flows	Astrid	CFX	5	April 26, 2002 06:47
CFX 5.5	Roued	CFX	1	October 2, 2001 17:49
Is turbulence disripable by chaos?	Matthias	Main CFD Forum	9	March 27, 2001 14:04

February 16, 2012, 18:10		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	You are correct that the best approach is to simply model everything, and this is what the F1 teams do because they have access to some pretty big resources. But for the rest of us you generally need to make some compromises. If you are looking for the relative effect of a small detail on a very large structure, then will a small model which is focussed on the small detail give good enough relative results to be useful? I mean is the optimum design in a small model close enough to the optimum design in the full model? The absolute drag/lift/whatever will not be right, but if the optimum is the same then it is a useful design tool. Alternately you have two other approaches: 1) Use a simplified model of the upstream effects with a coarse mesh. Obviously this is a compromise between fine enough to catch the important detail and coarse enough to run on your hardware. 2) Use a full model simulation to define an inlet boundary profile for a small run focussed on the detail you are looking at. This assumes the small detail does not affect the overall flow field much. And then of course you can always hire zillions of computers on cloud computing which is becoming easier and cheaper.