Truck cfd analysis
 

Hi all,

I am given a project on cfd analysis of truck trailer configuration the model is shown in image below. I have to find drag coefficient at different Reynolds number and validate data with NASA experimental value.

http://www.freeimagehosting.net/t/ebdfb.png

I would like to know how to start, as I don't have any idea about cfd analysis. I can learn software that's not an issue. I just need step by step guideline on how to approach this problem.

I have basic theoretical knowledge but this is my first time on working on cfd analysis. SO I really need your guidance.

Thanks in advance.

I would start by doing the following:
1)making the geometry first
2)Then mesh
3)Set up the critical parameters in FLUENT
4)Make sure to output the drag on whatever walls your looking at and in the correct direction
5) Post-Process
6) Analyze your results

Hope this help =)

First step is to read the CFD literature in regards to truck trailer configurations. Find out what other people did. A big issue with this configuration is the base, i.e. back end. Also, since you are new to this, I would suggest that you try to replicate someones solution as a test case.

Before you start in on any geometry you should also work out exactly what you want to find, what situation you're modelling and all boundary data, properties and expected flow characteristics you think might be occuring. As Martin said read up on similar cases first to give you a better idea. Only once you know in basic terms what you need and what you expect to happen should you start doing anything. You can then determine what your geometry should look like (e.g. what you need and don't need) and what type of mesh and where to concentrate the cells. I sugest you look for the introduction to CFD introcutory fluent training power point for a good start to the process you should follow.

Hope this helps

Andy

@ albedo

Thanks for reply.

I have made geometry in catia in .igs file now I am confused what software to use. I have ansys cfx, ansys fluent and ansys icem cfd.

In what software do i add mesh. I have tried icem cfd, but there is query what is blocking in it? and other question is that how should i define inflow & outflow (Is it supposed to be defined while meshing)


@ martin thanks I would look for such other literature in the internet.


@andrew thanks..

have u tried Phoenics or OpenCFD
 

hello Seth,
Do u have Phoenics package ? This can help you to solve fast and easy to put initial inlet flow conditions .

In addition to the comments of Martin and Andrew I think it might be easier to start with a 2-d configuration. This will enable you to run more solutions since 2-d grids are less computationally expensive! Additionally you will gain knowledge on CFD and hopefully the physics behind it. I think icem is a good meshing tool but you might need some time to learn it. You can use Velocity inlet, Pressure outlet for the front and back of your domain, wall for the ground (that you can specify as moving in fluent), wall for your configuration (non-slip) and pressure far field for the top part of your domain.

Best of luck,

A.D.E

Can I create a geometry in CATIA and then do meshing.

and What software in ANSYS should I use to easily mesh the 3d geometry that I have attached in my first post.

Thanks in advance.

What do you have access to in regards to software? If you have work bench with designmodeller and ansys meshing then I would use those for such a simple geometry. They aren't too hard to get the hang of. If you insist on doing the modeling in CATIA then if possible export is as a parasolid file for import into ansys geometry modeller. If you don't have access to workbench then either gambit or ICEM are generally the meshing packages of choice for CFX and Fluent.

Andy

Thanks Andrew for detailed explanation.

I have normal version of Catia without ansys meshing tools.

So basically what's difference in cfx and fluent? I am bit confused what software to use.

Cheers..:)

The choice you make on which software to use depends on what you want to do. As you appear to have never done any CFD analysis before i'd say the differences are likely to mean very little to you. Some questions you'll need to ask to determine in a basic sense which will work best for you are:

Is it a Transient or Steadystate problem?
Is it compressible or incompressible flow?
What kind of turbulence modelling do i need?

I'm assuming the object you're modelling is a wind tunnel model judgeing by the pressure taps?

If so then i would treat it in a similar way to an aerofoil problem, there is plenty of info on running CFD on such things.

In this case you might as well use Fluent, there is a tutorial on modelling an aerofoil which you can probably adapt to your model (at least in 2D, though 3D wouldn't be much harder) Fluent also has the k-w SST and S-A models which are widely used for external flows which you might find useful.

So:
Geometry - CATIA
Mesh - probably Gambit though maybe ICEM
Solver - Fluent

You should be able to do everything you want with those.

Andy

@ Andy

Thanks a tons buddy. You gave a very clear idea on how should I work.

BTW lets say for eg. I make a flat plate in CATIA, Then do I need to define domain and all stuff in mesh generator software right? I have made only flat plate not the domain around in CATIA.

cfd flow over truck
 

hi


you do one thing this type of problem is easy to solve in fluent 6.326 v
step -1
first creat the model in gambit
step - 2 using fluent
or you try to use on tutorial problems ok


velliangiri
velli_giri@yahoo.co.in

Yes in CFD you only need to create the fluid domain, so although you have your "plate" you would create a big box around it representing the fluid and subtract the "plate" from it leaving a hole in the middle. You'll just give the edges of the hole representing the "plate" wall boundary conditions so the flow will go round them. I sugest you do a few tutorials like the aerofoil one suggested earlier as this will give you a better idea. The image shows an example of an external flow mesh for an aerofoil, you can see there is a hole in the middle.

Andy

P.S. Yes you should be able to do the domain deffinition in the meshing software if you import you geometry.

http://lsec.cc.ac.cn/%7Ecjxu/image/nc15mo2/mesh.gif

So lets say if I want to analyze flat plate then firstly I need to make one solid flat plate in catia. Then Import to meshing software. Now my question is how do I define boundary in such case do I need to create boundary with structure itself?

it will depend on the software as to how you do it, i use Ansys DesignModeller which is built in to Ansys Workbench. With that is it easy as it has been designed to do it. You just create and enclosure around your geometry and it will create a solid body. What i sugest you do as it seems you're more used to using CATIA, is to create a large solid box that encloses your model and then remove your model leaving a hole in the middle of the new solid. you can then import the solid fluid domain without the actual geometry you're interested in. So you'll just have the fluid with a hole, you mesh the fluid and ignore the hole. If its 3D make sure that you have fluid on all sides. You will apply boundary conditions to this fluid domain. The picture shows what a mesh for a flat plate might look like (this is cut through to show the plate but the plate can't be seen normally). In Ansys Meshing you can then apply lables to the boundaries in this case the outer sides of the cube and the surfaces of the hole in the middle. These lables can then be used to locate the boundary data. This was done useing Ansys DesignModeller and AMP, i'm not certain how to do this using Gambit of ICEM so you may need to search through their documentation on how to.

Andy

So I can also do like , make a model in catia import in ansys modeler and then enclose it? In that way I wont have to care about defining fluid around geometry.

Yes that would work too, it would probably be the easiest option but it depends on the software you have available to you. The enclose feature in geometry modeller makes everything much easier. Import>Enclose>supress the body of your import. you can then just move on to meshing.

@andrew Can you tell me How can I find drag co-efficient in CFX ?

These links explain it.

http://www.cfd-online.com/Forums/cfx...efficient.html

http://www.cfd-online.com/Forums/cfx...efficient.html

Thanks for all your help. I will be updating my results here. Rep added to u.

Do one need a licence copy of the ansys cfx to publish the paper of analysis?

You always need either a Research License or a Commercial license if you're going to be doing anything that will be published. An academic teaching license should only be used for learning the software.

Ok. So I have done one experiment but it has about 25% error than the data i have to validate.

I have attached pic of mesh and analysis. I have question that how do I refine mesh? As you can see in image there is no vortex seen behind the truck so I think I need to do refinement near that region.

Can you post a pic of your mesh as well? Also what setting are you using in CFX, what turbulence model, boundary conditions etc?

Boundary conditions
At Inlet:
Velo x dir = 23.55 m/s
Velo y & z dir = 0 m/s
Turbulance: Intensity and length scale ; fractional int 0.05 m & eddy length scale 0.1 m
Temp : Static temp. (288 K)
At outlet:
Static pr. = Relative pr. : 0 pa
At the walls of domain (Including ground): Free slip wall
At the surface of truck & support: No slip wall

First two image show the meshing over the enclosure. Second shows meshing over truck body (Enclosure suppressed)

I sugest you try to produce a structured mesh with more refinement behind the truck where the features you're interested in are. you'll also need to do a mesh independence study. You should look up wake turbulence modelling and find out which turbulence models are most popular, you'll probably need to do a comparison between a few and your validation data.

But there must be something there, regardless of the mesh sizing. It can not be a vacuum and I doubt the flow would wrap around the corner at the back. Did you place stream lines at the back?

Or do you mean there is no unsteady vortex shedding?

So how exactly do I do structured meshing.

Here is how I do analysis.

1. In ansys workbench select CFX in analysis system table..
2. Import my truck design. & Made enclosure around it.
3. Did auto meshing.. (I think this is where I have to do something to make results accurate)

So in ansys meshing how do I create a proper structured grid? All I do is just do auto meshing.

4. Then i give input values.

I want to know which turbulence model did u chose?

5. Post processing.

I wanted to mean that flow doesnt curl around the rear bottom part.

Not sure that I understand. Since there are sharp edges around the parameter of the base, the flow will separate there. Depending on the turbulence model, you may get steady results. And this seems to be what you are getting. If the results are steady a ring vortex will exists at the back end. Are you seeing the ring vortex?

Or, are you expecting unsteady vortex shedding? Of course in real life that is what you will get. But, a turbulence model might not give you that. For example, in general, an ogive-cylinder body will give you a steady base flow using SA or SST turbulence model.

Can anyone please tell me how do I do structured meshing?