Chassis FEA for Torsional Rigidity

[apoorv]

Hello Guys,
I am Apoorv from SRM University, India. We are participating for First time in 2015 FSAE Italy.
For our first car chassis I am having a some trouble in getting the appropriate torsional stiffness value from FEA. I have used ANSYS Workbench beam analysis method for the analysis by importing coordinates, drawing tube center lines and then defining cross section. I applied a couple at front 2 suspension nodes and fixed the rear bulkhead 4 nodes. The deflection I got gave me a TR value of 600Nm/degree. Which is less stiff I guess. Also when I tried by importing directly the solidworks file into ansys(as solid geometry) and did the same analysis it gave TR value of 1000Nm/deg (as the deflections were small). I also tried different forces from 500-2000N but the results were same(which is good).

Can anyone tell me which method is more appropriate beam element or solid geometry? Is there anything wrong I am doing in the process (fixtures or loads)?
Ansys images are attached below.
Solidworks.jpgBeam element Method - more deflection.jpgLoad and Fixtures.jpgSolid geometry Method - less deflection.jpg

Thanks

[Kevin Hayward]

Apoorv,

The chassis design you have is going to have very low torsional rigidity. The side pod areas will add next to nothing and you should look at integrating them into the car better. I would be using beam elements, but you need to make sure that all the connections are solving properly. Somewhere around the 600 Nm/deg looks reasonable accurate for what you have modelled.

Your fixtures are incorrect. Loads for the chassis originate from the tyres and are then transferred through links to the chassis. You have put vertical only loads at only a pair of front wishbone attachments. This will be far from accurate. The only are that you might be assessing with any reliability will be the centre section (i.e. far away from the fixture locations). Although I would doubt if that is accurate as well.

Before doing too much analysis it may be worth doing a bit of research into frames that other teams have developed, as well as maybe a few roughly made hand models. You need to do your design iterations from a decent starting point. By the time you get to the analysis stage the iterations are much slower than when you are conceptually designing. This means local rather than global improvements. With the design as is you need to stay looking globally first.

Kev

[apoorv]

@Kevin I absolutely understand your point that Our chassis design will not have high rigidity and need some iterations.
If I apply load to all the front suspension hard points and fix my rear suspension hard points, then which deflection value should I take??. Previously I used to take 2 points only and take average of their values. or should I model the suspension as well in ansys n apply load at uprights?

[DougMilliken]

...s well as maybe a few roughly made hand models. ...

Plastic straws and hot glue are a quick way to make small physical models to bend and twist. We included a note on this topic in "RCVD: Problems, Answers and Experiments", see page 212.

[Jay Lawrence]

apoorv,

Forget about what values you should get (although if you search through these forums there are many figures given). Kev is saying that you need to focus at a higher level on your chassis design. I would start by removing those side pod boxes and adding some floor cross-members.

[Kevin Hayward]

Apoorv,

You should be looking at a number of points when you are ready to. Once it is in the computer you can see every part you want. A good way is to model in some extension bars to the sides of the chassis that in no way contribute to the stiffness. Have them all extend to the same plane. These are along the car at different points. Once you have that you can assess the linearity of the twist. You shouldn't be looking at a single number, rather a graph of twist vs position along the chassis.

There are a number of really crappy chassis structures being made in FSAE, some by a few very good teams. One of the most common mistakes is to have rapid transitions between very stiff sections and very flexible areas of the chassis. Some of these teams are shocked that they have cracking problems. Even more disappointing when it is attributed to welding quality rather than design flaws.

However, just for arguments sake lets return to the higher level thinking. What are you looking for in a chassis? What number is okay? Is the chassis the biggest area contributing to twist deflection corner to corner that you are going to see?

Frankly once you have a decent approach to the chassis design it is almost a waste of resources to analyse the chassis in this way at all. If the chassis is sorted you will see much more compliance in other components, and your time might be better spent doing those tests. FEA is a verification tool, and wont help you much in idea generation. My advice to any FSAE team is to quickly get to the point where physical testing is your main form of validation. Plastic straws and glue is as much a simulation of the chassis structure as FEA, and validation is quick.

Design wise there is only one real area you need to work on for a spaceframe chassis in FSAE at the moment. That is the driver cell. The front is easy, the rear has to dodge a lot, but the only real calls you have to make are:

- High or low upper side member
- Integrated or no side frame sections

Side frames and high sides end up with much better stiffness, but more weight. So what do you want? Lower weight and more deflection, or higher weight and stiffer structure? Decent arguments exist for both cases and will have a bearing on your approach to vehicle dynamics.

Kev

[apoorv]

@kev Well i guess then we should consider actual physical testing methods and concentrate on better geometry.I tried to put 2 triangulated members, one in front bulkhead and one at the base. This significantly improved the stiffness in ANSYS model.
I think then FEA is good just for checking your deflection in geometry and have an indicated value. Also I Will soon start to design a new chassis and consider your advice to have some target values and then put some good triangulation as well. As u pointed out side pods. we already decided to remove them next year n place our batteries somewhere else.
Thanks

[Kevin Hayward]

Apoorv,

You don't necessarily need to remove them. Rather if they are there think of how you might design them to help with torsional rigidity.

From your comment about electrics I am assuming an EV car. If so your chassis can be awesome. Most of the problems of rear end packaging for the petrol cars is pretty easy to solve with an EV. That will only leave the driver cell compromised.

Kev

[apoorv]

Kevin,

You are right our vehicle is an EV.

We decided to remove them as to reduce the weight in the chassis and drag induced by them. If we remove them and and put batteries in rear significant weight can be reduced.