1. Hello,

i hope you can help me. I thinking about how to constrain my frame/Chassis model in the fea program to simulate different load cases like brake in corner, brake, accel. and so on. The forces on the Tire/upright for each loadcase is not the problem. I have the loads for each situation.

The torsional analysis is also not that difficult I constrain the rear upright in all dof. And apply the loads at the front upright.

What are your steps to simulate the frame?

My idea is to fix the center of gravity and apply the loads on the tire contact patch. But to what points of the Chassis I havt to make my ridgids? The engine mounts? The mounting points for the seat?

I have no idea what is the right way to not over and under constrain the chassis and to get the right stress value on the tubes and welding.

I hope we get a good conversation.

2. Heyho,

I have never done such a simulation myself, but I think the correct way of doing one is to use inertia relief. To quote this PDF:
"Typical applications of inertia relief include modeling an aircraft in flight, an automobile on a test track, or a satellite in space."

As far as I understood it, displacements are constrained on one point or some points and then loads are applied. Then the solver updates the acceleration of the model until it has reached an equilibrium between the applied loads and the inertia forces. At this equilibrium, the constraints do not see any forces at all, meaning that they only define the position of your model in 3d space but nothing else.

Consult your FEA software manual on how to use inertia relief.

3. Inertia Relief.

4. Inertia relief is for sure the right way to do it. I don't know the gritty technical details of it, but as far as I know it basically smears the necessary reaction forces to keep your model from accelerating away across the model based on the mass and inertia distribution of the model. You can run it with no constraints at all or with up to 6, I believe. All adding these constraints do is change the frame of reference you see the body deforming in when you review the results. I'm not 100% certain that 6 is the max you can have, but it would make sense seeing as you've got the 6 rigid body DoF to play with.

The difficulty is that, at least the way I see it, you'd need to model every single part of the car to get it right.

Lets say you just modeled your chassis and suspension with beam elements, slapped on some cornering loads and ran it with inertia relief. The forces are going to all have to be reacted by the mass and inertia that's there in the model, and so will only be balanced by inertial forces acting on the suspension and chassis. But in reality, the chassis and suspension only make up a portion of the vehicle mass and inertia. A large part of the forces that the tires generate have to go into pushing the driver and engine around, and if these aren't in the model, then the forces you apply at the contact patches will be taking a different path to the inertial forces(all going to the chassis/suspension instead of the driver/engine) and hence any forces or stresses you calculate will not be what your car sees on track.

And since the driver is a pretty significant portion of the car's mass/inertia in FSAE, now you need a finite element model of the driver! Or at least some way to add the driver's equivalent mass and inertia to the model without artificially stiffening/strengthening the chassis. Most solvers have point mass and inertia elements, although I've never used them myself so I can't say for sure that they'd work in inertia relief.

Another difficulty in trying to put a driver in is that he needs to be connected properly to the chassis so that the loads that have to go to the driver take the proper path from the contact patches. That is, the car applies most of the forces to the driver through the seat and harness, so you need some way of connecting the driver to the chassis that models this.

What fun! Obviously you wouldn't be able to model everything, but the bigger(relatively) the mass and inertia contribution of something you leave out is, the bigger the error will be.

While this is fun to talk and think about, it'd be a hell of a lot of work to do it right, and I don't think there's much benefit to be had given that torsional stiffness already drives so much. The only simulation we've ever done on our chassis, aside from torsional stiffness, is strength of mounts and mounting tubes.

The other implication in this(at least the way I see it) is that K&C testing is wrong, at least slightly. By clamping down the chassis(or whatever they decide to clamp down) instead of using inertia relief like the car sees on track, the path of the loads from the tires to the constraint changes and hence the deflections should change, at least a bit.

Unfortunately, big arms to swing cars around on to generate the loads don't seem very practical. And the whole danger that the car might fly off if you screw up the test may also be a deterrent.

The error is probably not big enough to be important, and it's probably better the heavier the chassis gets, but this was something that came to mind when I was thinking about a full car compliance finite element model.

Also, you might want to rethink your constraints for your torsional stiffness analysis.

I can really write a lot when I'm supposed to be studying for exams

5. Wow,

Thank you for your answers. It sound very difficult, as you said, to define the mass and connect it to the frame correctly.

I´ve read that the intera Relief is possible in Hyperworks and Nastran.

The main goal is at first to look at the mounting parts and the Tube in the near what stress is there an that parts during conering and braking.

@ Moop
Who do you have analys this? Do you make a sub Model of these Tubes? It would be glad if you could tell me a few details.

What do you think about the contrains for torsional stiffness test, is it wrong to contrain the rear uprights or say contact patch in all dof.?

Thank you

6. What do you think about the contrains for torsional stiffness test, is it wrong to contrain the rear uprights or say contact patch in all dof.?
Constraining both rear uprights in all 6 DOF's will over-constrain the system. You are introducing artificial stiffness into the chassis that away. Look for a way to constrain the chassis such that all 6 DOF are constrained, but nothing is over-constrained. Also, make sure you're loading the chassis through the actual load-paths, i.e. through the suspension linkage, push-rod, rocker, shock, etc.

7. BTW: I wrote an SAE paper in 1979 on how to do this. There is an inertia relief demo in the paper showing the distribution of body and frame strain (flexing) during a cornering maneuver. The major masses are the important parts to include, not necessarily all the parts. The masses are the load reaction generating pieces.

Paper Number: 790376

Cobb, W., "Suspension Parameter Prediction Using Finite Element Analysis," SAE Technical Paper 790376, 1979, doi:10.4271/790376.

Abstract:

A technique has been developed for applying finite element computer models, assembled for vibration work, to analysis and prediction of vehicle handling performance. The method facilitates the prediction of chassis stiffnesses important to handling before a vehicle is built and available for laboratory testing by more traditional methods. The technique incorporates the general purpose finite element computer program (NASTRAN), total vehicle system models developed for structural analysis, and applied loads intended to simulate existing laboratory and road tests. A post-processor program summarizes chassis parameters related to handling from the general NASTRAN displacements in familiar units. Example results from actual and simulated testing are compared.

There you go...

8. But back to one of the interesting questions...

A industry test called a "Torsional Racking Test" is used to evaluate a frame assembly. It is a naked frame without any additional subsystems. The test involves locking two rear suspension attachment points to T-posts on a bedplate. There are only displacement constraints. (Free rotations). One front attachment point is similarly anchored. A series of weights is then used to pull down the unconstrained attachment point. Measurments are made of the eleements all along the 'frame' from front to rear on both sides.

This test may not have all the face validity you might insist on, but has a direct connection to the ability to properly model it in a FEA program. Yes, motor and suspension pieces alter the normal modes displayed during the test and analysis, but, when you perform other solutions (statics or inertia relief for example) these modes are constrained by the additional rigid or non rigid pieces and cause flexure in the system model, just like the car does when its all put together.

The racking test keeps you from getting off the beaten path (so to speak) and is a simulation of a simple test, not a very complicated dynamics event.