As an assignment for my Automotive Engineering study, I have to practice manual FEM calculations. I'm free to choose whichever existing construction, and with that I have to make FEM analyses of both the original and the improved construction. I chose the chassis of the '57 MGA i'm restoring, which is shown below:

mga chassis

I suppose there are several ways to improve the driveability, but I expect increased torsional stiffness to have the most impact. The rear bridge seems rather wobbly to me, but I can't really tell; I haven't driven an MGA before.
(If you have better ideas how to improve the chassis please do suggest.)

What would be a good approach to improve the torsional stiffness of the chassis?
With that, I mean what kind of construction/structure would be effective; a triangular structure between the two side beams for instance.

Please note: it doesn't have to be practically feasible or cost effective. It's just for practicing.

  • $\begingroup$ Wait, what does your actual question have to do with FEM? You mention you're doing this for a FEM class, but as I see it your question itself has nothing to do with FEM. Or am I misunderstanding something here? $\endgroup$
    – Wasabi
    Commented Feb 7, 2017 at 19:03
  • $\begingroup$ @Wasabi I have to do FEM calculations on the original frame, and on the improved frame. I have to make an improvement to the frame, so hence my question; 'How could i best improve the frame?' $\endgroup$
    – Bart
    Commented Feb 7, 2017 at 19:42
  • $\begingroup$ Start by modelling the original design, and compare the results with some test measurements on your frame. If you don't do that comparison, the entire exercise is likely to be a work of fiction, given the amount of FE experience you have (I assume, not much). By the time you have got the FE model to match the real world, you will probably have some useful ideas about what is wrong with the original design. Then change your model to evaluate those ideas. $\endgroup$
    – alephzero
    Commented Feb 7, 2017 at 21:39
  • $\begingroup$ @alephzero So you say that i'd get ideas about how to improve the chassis, once i've found the weak points after an analysis of the original chassis? That'd be a good strategy, but I have to do the proposal of my idea before I can start the assignment, so that won't be possible.. I was hoping for a simple idea to further investigate, like a triangluar bridge or something. Regarding the assignment, costs and feasability are no issue, because it's only an excercise. It'd an issue if i'd decide to really implement the improvement in the frame, but i think i'll edit that out of the question. $\endgroup$
    – Bart
    Commented Feb 7, 2017 at 22:21

2 Answers 2


The overall stiffness of the frame is determined by the stiffness of the side beams augmented primarily by the three tubular cross members.

An interesting academic exercise would be to evaluate how changing the diameter of the cross members affects the overall stiffness.

  • $\begingroup$ So to simplify, create larger cross members. That's a good approach for me I think. The shape of the construction isn't changed then, only the corss members' diameter. It's a relatively small and simple change it seems, so it's fine for a first excercise. I'll do that. $\endgroup$
    – Bart
    Commented Feb 11, 2017 at 10:57

There are really two aspects to this:

  1. Will increasing roll stiffness improve handling to an extent which offsets any increase in weight by adding extra chassis members ?
  2. If yes, what is the best strategy for improving the stiffness of the chassis?

The issue here is that this is a fairly simple ladder chassis so we are really looking at creating some sort of ladder/spaceframe hybrid. However we also need to consider the fact that the rear suspension is pretty basic in the form of longitudinal leaf springs so there is no guarantee that stiffer is better without identifying specific characteristics which you need to improve. Eg increasing rear roll stiffness isn't necessarily better unless this is identified as a limiting factor on performance.

In some ways it might be a more productive exercise to investigate where you can lose weight by removing material from the existing chassis by identifying areas where it is under-stressed. Apart from anything else this should definitely give some easily quantifiable benefit.

In terms of an assignment it may also make more sense to demonstrate how a space-frame chassis could give the same torsional stiffness for less mass. Again this gives your project a much more well defined objective.

For example if you manage to find a way to make the existing chassis 10% stiffer it is then very hard for you to show what real world performance benefits this gives without actually carrying out the modifications and testing them. On the other hand if you can make it 10% lighter for the same stiffness that is a very obvious benefit.

I would also add that having a project with a well defined strategy for improving a design is fundamentally better than just changing some arbitrary design parameter without any clear idea of how it will improve performance.

So project titles like 'A worked example of the weight advantages of a space-frame vs ladder chassis' or 'Reducing the weight of an MGA chassis' will make much more sense to an engineering professor than 'Showing that adding more mass to a structure can make it stiffer'.

Eg say you make it 10% stiffer and 10% heavier... is that better? Unless you can prove it is better the whole exercise is a bit pointless. If it was an experimental exercise things would be a bit different.

The point here is that in any scientific and engineering experiment you need to be able to reach some conclusion. If your intended approach didn't work that is fine as long as you can say that it didn't work and why what is bad is when there is no way to tell within the scope of the project whether it was worth doing or not.

Eg 'I investigated whether it was practically possible to lighten the chassis of an MGA using the following strategies but none of the approaches I used were effective' is useful information.

However 'I made the chassis of an MGA 10% stiffer by increasing its mass by 10%' doesn't really tell you anything useful, unless you can add further information how this improves its performance.

As an aside there is also the fact that for classic cars (in the UK at least) any marginal increase in performance has much less value attached to it than preserving its authenticity and originality.

  • $\begingroup$ You'd be 100% right if I were to actually modify the chassis with the construction, but it's just a FEM excercise, not even a serious experiment. I agree my approach isn't very scientific, but my goal was to come up with some kind of construction that one could expect would make the chassis more torsion resistant. Maybe that wouldn't even improve driveability, but that's not relevant. It'd be nice if it showed to be, so that i could use that information while rebuilding my chassis, but it's not necessary. $\endgroup$
    – Bart
    Commented Feb 8, 2017 at 21:23
  • $\begingroup$ I agree that keeping it original is better, I wouldn't introduce dramatic modifications in the chassis even if they'd have good potential. But the chassis itself is rotten and rusted through, so it needs rebuilding anyway. I might as well add some minor mods to increase stiffness here and there while i'm at it. That is, if that'd show to improve performance. If it doesn't, or if i'm not sure I'll just leave the chassis completely original, and the excercise is still a success. $\endgroup$
    – Bart
    Commented Feb 8, 2017 at 21:30
  • $\begingroup$ I agree, may main point is that, as an academic exercise, something like weight reduction or illustrating the difference between two different approaches to chassis design will probably give you more payoff than designing for an objective which is hard to prove in practice. $\endgroup$ Commented Feb 8, 2017 at 21:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.