Computational Fluid Dynamic Analysis of an Impreza CAD model
#31
Scooby Regular
Thread Starter
Join Date: Jan 2006
Location: North Wales
Posts: 931
Likes: 0
Received 0 Likes
on
0 Posts
AeroAus... yes the aerofoil does have pretty extreme camber towards the rear but this is a copy of the actual aerofoil used on the car so I cant change that. However, the lower surface curve is gradual and I believe I could angle the wing better to keep the airflow attached on the underside for a bit longer than it is currently showing. The extreme camber on the upper surface will no doubt cause a fair bit of drag as well as desired high pressure but not in a very efficient way but I am not too bothered about that. I only tested this to see what sort of downforce gains the wing I actually have on the car is giving.
The mesh is fairly broad for a full scale car simulation and consists of around 2.7M polyhedral elements. I was going to change the mesh to a brick/trimmer mesh but haven't got around to it. The polyhedral mesh utilises a volume control in the form of a rectangular block around the car that limits the cell size to 5cm. Closer to the car it reduces down to a few mm's and away from the volume control it expands relatively quickly. I would say its not the most accurate mesh but still gives a fair idea of the direction of air flow and rough force data without taking too long to run. The computer I'm using couldn't handle much more than 8.5M cell polyhedral mesh.
I have used both k-epsilon and k-omega and the results were surprisingly similar. Again I was only after a rough estimate of lift and drag forces as well as an overview of flow directions. If it were to be an industry standard simulation a lot more refinement would need to be done from the mesh quality to the model accuracy but so far I am happy with the results and am taking them with a pinch of salt.
Its nice to see a few people taking an interest in this, it is quite time consuming work but I will try to continue it when I can.
Gaz
The mesh is fairly broad for a full scale car simulation and consists of around 2.7M polyhedral elements. I was going to change the mesh to a brick/trimmer mesh but haven't got around to it. The polyhedral mesh utilises a volume control in the form of a rectangular block around the car that limits the cell size to 5cm. Closer to the car it reduces down to a few mm's and away from the volume control it expands relatively quickly. I would say its not the most accurate mesh but still gives a fair idea of the direction of air flow and rough force data without taking too long to run. The computer I'm using couldn't handle much more than 8.5M cell polyhedral mesh.
I have used both k-epsilon and k-omega and the results were surprisingly similar. Again I was only after a rough estimate of lift and drag forces as well as an overview of flow directions. If it were to be an industry standard simulation a lot more refinement would need to be done from the mesh quality to the model accuracy but so far I am happy with the results and am taking them with a pinch of salt.
Its nice to see a few people taking an interest in this, it is quite time consuming work but I will try to continue it when I can.
Gaz
#32
Scooby Regular
iTrader: (2)
Join Date: Jul 2007
Location: Essex
Posts: 552
Likes: 0
Received 0 Likes
on
0 Posts
Very interesting mate, there's very little information like this available and I've always been interested to see what all these splitters and spoilers are actually doing! I'm sure a lot of people/us are not gaining the full potential of aftermarket products we're putting on our cars
#33
Scooby Newbie
Join Date: Jan 2013
Location: Sydney, Australia
Posts: 3
Likes: 0
Received 0 Likes
on
0 Posts
AeroAus... yes the aerofoil does have pretty extreme camber towards the rear but this is a copy of the actual aerofoil used on the car so I cant change that. However, the lower surface curve is gradual and I believe I could angle the wing better to keep the airflow attached on the underside for a bit longer than it is currently showing. The extreme camber on the upper surface will no doubt cause a fair bit of drag as well as desired high pressure but not in a very efficient way but I am not too bothered about that. I only tested this to see what sort of downforce gains the wing I actually have on the car is giving.
The mesh is fairly broad for a full scale car simulation and consists of around 2.7M polyhedral elements. I was going to change the mesh to a brick/trimmer mesh but haven't got around to it. The polyhedral mesh utilises a volume control in the form of a rectangular block around the car that limits the cell size to 5cm. Closer to the car it reduces down to a few mm's and away from the volume control it expands relatively quickly. I would say its not the most accurate mesh but still gives a fair idea of the direction of air flow and rough force data without taking too long to run. The computer I'm using couldn't handle much more than 8.5M cell polyhedral mesh.
I have used both k-epsilon and k-omega and the results were surprisingly similar. Again I was only after a rough estimate of lift and drag forces as well as an overview of flow directions. If it were to be an industry standard simulation a lot more refinement would need to be done from the mesh quality to the model accuracy but so far I am happy with the results and am taking them with a pinch of salt.
Its nice to see a few people taking an interest in this, it is quite time consuming work but I will try to continue it when I can.
Gaz
The mesh is fairly broad for a full scale car simulation and consists of around 2.7M polyhedral elements. I was going to change the mesh to a brick/trimmer mesh but haven't got around to it. The polyhedral mesh utilises a volume control in the form of a rectangular block around the car that limits the cell size to 5cm. Closer to the car it reduces down to a few mm's and away from the volume control it expands relatively quickly. I would say its not the most accurate mesh but still gives a fair idea of the direction of air flow and rough force data without taking too long to run. The computer I'm using couldn't handle much more than 8.5M cell polyhedral mesh.
I have used both k-epsilon and k-omega and the results were surprisingly similar. Again I was only after a rough estimate of lift and drag forces as well as an overview of flow directions. If it were to be an industry standard simulation a lot more refinement would need to be done from the mesh quality to the model accuracy but so far I am happy with the results and am taking them with a pinch of salt.
Its nice to see a few people taking an interest in this, it is quite time consuming work but I will try to continue it when I can.
Gaz
Interesting to hear that k-e and k-w returned similar results, perhaps a finer boundary layer is required to really show the differences. Speaking of which, was it difficult to obtain decent convergence? I would imagine that the flow would be extremely transient due to vortex shedding effects and such.
May I ask where you got the CAD model from? I'd like to try putting it through ANSYS CFX, I've got a workstation capable of 20M + element meshes at my disposal.
Anyway, love your work
#37
Scooby Regular
Thread Starter
Join Date: Jan 2006
Location: North Wales
Posts: 931
Likes: 0
Received 0 Likes
on
0 Posts
Yea... I ended up getting a job and have hardly any time to play with this stuff anymore shouldn't complain really but would have liked to continue the work but due to the very messy CAD file of the car it's not simple to test configuration changes quickly as the model has to be 'patched' and 'wrapped' within the CFD software each time which takes ages. Was nice to get a few positive comments on it all though. I'm also still working on the real car when I can which used to be updated on a project here called rex racer, not updated that either but ended up trying to learn mapping and done a lot of painting on it, so just have too much on my plate. Sorry all!
Gaz
Gaz
Thread
Thread Starter
Forum
Replies
Last Post
Wingnuttzz
Member's Gallery
30
26 April 2022 11:15 PM
Scott@ScoobySpares
Full Cars Breaking For Spares
55
05 August 2018 07:02 AM
Danny0608
Subaru Parts
0
12 September 2015 02:59 PM
airflow, automobile, cad, centre, computational, drag, dynamics, fluid, impreza, low, modeling, pressure, starccm, supercomputer, wing