FEA on a tube frame chasis using COSMOS 3

[kikiturbo]

Hi, I posted this in the Solidworks section but was instructed to post here as this might be more appropriate section... here it goes..

Hi, I am just starting using SW 06 and am having some problems with modeling and analysis of a spaceframe construction for the car I am making.. as I want to model it in SW and analyse its torsional stiffness properties...

now, the easiest way for me is to use a 3D sketch and make my tube frame, then use the weld profiles tool and assign each section of the sketch with a tube.

now, when I do the analysis in COSMOS it doesn't understand that the tubes are actually connected togeather, so there are two ways of doing it, that I see:

1. use fillet welds and weld the tubes in SW
2. use the restraint tool in COSMOS and "tell" cosmos that the tubes are rigidly connected in the frame..

the first solution doesn't always work, in fact once I have a real 3D spaceframe, where I have joints with several tubes going in, the SW fillet weld feature just will not weld some tubes togeather...

the second solution is very time consuming, and makes COSMOS block and crash from time to time...

I have also tried exchanging the tubes in the model with solid bars of equal cross section area (as I presume that tension/compression of the tube will have much greater influence on the frame torsional stiffness than tubing bending) that does make it much faster but the problems persist...

does anyone use SW and COSMOS to model tube frames and analyse it's properties... care to share some tips? or should I look for other software...?

best regards

vlado

 

[ERT]

Vlado,

Are you using solid or shell mesh? We have seen issues with shell meshes as you have described. Even with solid meshes, it is important to fill in any gaps at the joints.

Your suggestion to use solid bars (obviously with a solid mesh) sounds interesting. Whether to use an equivalent area if deflection due to tension/compression is dominate, or use an equivalent moment of inertia if torsion/bending is dominant is the critical choice. I would guess this depends a lot on the triangulation of your frame.

Regards,

ERT

http://www.akeng.com

 

[corus]

If you're modelling a frame then you might be better just using beam type elements ang assigning the appropriate section properties of the tubes to each element. This will at least give you the right torsional, bending and direct stiffness properties to the frame.
Sometimes it's better not to use a constructional type drawing for a finite element model as you sometimes need to represent the details differently for the model. From the start it's better to construct the frame as you would want to model it.

corus

 

[kikiturbo]

>ERT - as far as I am concerned, tube framed spaceframe should be designed around propper triangulation and using tubing mainly in tension/compression is a major design objective for me... one of the main reasons for using FEA is to try a slightly different approach to the design of the floor in my car instead of starting with what is basically a ladder frame and adding a spaceframe cage around it..

>corus - great tip, and I have allready tried modeling the frame using solid bars instead of tubing (is this what you are reffering to?) My major problem with SW / cosmos is making it understand that my bunch of tubes/solid bars is connected togeather

I will have no problems modeling my car twice, once for FEA and then a "true" model for checking the assembly clearences.. but I need to find an elegant way of doing the FEA on the spaceframe in Cosmos

cheers

vlado

 

[corus]

Solid bars in 3D wouldn't be correct as you can't get the right stiffness for all types of loading (ie. bending torsion, etc.). A beam type element looks just like a wire, however you input the correct sectional properties (areas, inertias) so that the correct stiffness applies for all load types. The frame would therefore be drawn as a series of lines along the centrelines of the tubes. Each beam type element reprsenting a tube would connect with the next tube(s) at a single point. This type of model would be ok for generally sizing the tubes and getting overall stresses but wouldn't be good at looking in detail at the stresses at the intersections. For that you'd need a 3D model, either with solid elements or with shell elements. Solid elements would give you the more definitive answer to describe stresses everywhere but would be the most expensive (time wise) to set up and run.

As an aside, for the design I would have thought that a tube has better bending properties than a bar as in any section most of the stiffness is towards the outer fibres, hence you get lower stresses with a tube in bending. I don't see the point, therefore, why you'd want the tube to act mainly in tension/compression.

corus

 

[kikiturbo]

I agree with you 100% and I will try with beam elements...

as for my solid bar idea... that was primarily for speed, as it works much faster then using a tube in the analisys... and also, since the elements are much stiffer in tension/compression then bending, I thougt about dismissing the bending stiffness in the first instance and concentrate on getting the propper triangulation of the frame based on tension/compression alone...
well, I will do some experiments and will see...

cheers
vlado

 

[tnkayaker]

If your doing your analysis in Cosmos, the tubes must be connected to each other some how. From my limited experience with Cosmos and welded assemblies, it is easiest if you can mate common surfaces to each other, then Cosmos will treat the intersection as a "weld". This is not really praticle with a spaceframe. The beam element method would give you the most accurate results for torsional stiffness, but it would not be good for weld stresses.

 

[kikiturbo]

thanks..

weld stresses are not a concern really, as the frame will be severely overbuilt, as the target torsional stiffness is quite high..

I need to figure out how to do the beam element method in SW 2006... if anyone can share a few "for dummies" tips they would be greatly appretiated..

I just got some books on SW 2006 from amazon, so it is time to read..

cheers

vlado

 

[2design]

If you are using COSMOSWorks, it does not support 1D elements except for some rigid elements. The only elements you can use are tris and tets, both linear and second order.

 

[ERT]

CW2006 does not support beam elements.

The CW2007 "What's New" says that beam elements are now supported. It is not clear if mixed beam/solid meshes is supported; I think this is a key capability CW has been lacking. Has anyone seen this yet?

Regards,

ERT

 

[kikiturbo]

Hi,

I did some experiments with my tube frame,... I first put otgeather a "simplified" version of my frame, as I need to work on my suspension pickup points, but the general measurements are OK.. I used solid bars instead of tubing, as I could not get it to work with tubing... After I placed weld elements onto my 3d sketch, I used Insert-Feature-Combine function...

I had to fiddle around with meshing settings, as it would not mesh with default settings, going towards a smaller mesh size worked a treat... I also noticed that I had to pay attention to the placement of the tubes as if I left the corner so that the tubes would end in the centre and leave a corner gap (see pic.1) it would not mesh... (pic 2) worked...

pic 1

http://img.photobucket.com/albums/v723/turbolimac/FEA21.jpg

pic 2

http://img.photobucket.com/albums/v723/turbolimac/FEA2.jpg

a pic of the FEA displacement model... here I wanted to measure the torsionall stiffness, so I fixed the rear two spring turrets and applied force to the front one...

http://img.photobucket.com/albums/v723/turbolimac/FEA1.jpg

 

[corus]

Nice pictures, but you can see in the top roof of the structure how one of the bars is bending into a flattened S shape. This is why it'd be important to ensure you have the right bending stiffness of the bars as well as the right sectional area. Perhaps with such relatively few 'beams' you could use a free 3D beam program and get the co-ordinates of each beam from the solidworks model and so verify your results? The input for the model should be just a matter of the co-ordinates and the section properties, which for a tube are relatively easy to calculate. Look on this site

http://www.structural-engineering.fsnet.co.uk/free.htm

corus

 

[kikiturbo]

thanks for the tip, I have downloaded framework and will try it..

as for the co-ordinates that is not a problem as my model started with a set of points so I have all the co ordinates allready..

cheers
vlado

 

[badboggs]

Did you consider modeing the frame in surface shell elements?

Boggs

 

[dbh512]

I tried using Cosmos Express to model a tube frame chassis in SW2005. It didn't work. The wall thickness was about .040 or .060, so Cosmos tried to create a mesh with elements of about that size, so it created a huge number of nodes - too many to calculate.

I downloaded Cadre Lite (

http://www.structural-engineering.fsnet.co.uk/free.htm

) which does space frames using the typical beams/tubes and pinned joint solution, and I was able to gain some very valuable insight into chassis stiffness. It helped a lot. It isn't a CAD program so it was a pain to have to enter the data by hand then run an analysis, evaluate the Cadre data then consult the SolidWorks model to see where a tube could be added or how much it can be moved, then go back to Cadre and edit the input data or re-enter the whole lot again. It takes time, but it is faster than building a chassis and testing it.

I don't think I did it with tube and sheet metal. I recall finding that adding sheet steel to replace triangulations adds a lot of weight, and that if one uses aluminum sheet there is a problem with the rivets - regular pop rivets aren't strong enough and there is also a corrosion issue. -- things that don't come up in the FEA work.

Once the basic frame is developed in Cadre ( the size and placement of tubes is established ) you can use Cosmos Express to evaluate stresses/deflections on individual joints, pickup points or tubes if necessary.

Good luck in your design,

DBH512

 

[kikiturbo]

thanks for your tips...

SW2006 / Cosmos has some real trouble when using tubing... I agree..

I found that using solid bars with the same cros section area as the tube I want to use gives me a good idea of where I am heading with my design... and it is simple to do and very very fast..

I will try some simpler models, in SW or Framework with real tubes or beam elements to see how much my stiffness model that uses solid bars differs form the "real" one that is moddeled using tubing...