Learning Forum

[benjrfletcher]

I am trying to re-create the physics a lower wishbone part in a suspension system would experience in Ansys so I can perform FEA analysis and perform topology optimization on it. In general, the system consist of a upward force applied by the wheel, a counteracting force by a damper, and some sort of support at the legs.

I have tried a simple method by using a fixed support at one end, and forces applied where the damper and wheel would apply forces. However, I don't think the physics is realistic enough. I want to used some sort of ground to body spring connection for the damper, and a cylindrical support at the legs allow tangential rotation, since in real life, it has a pin support. However, my model seems under constrained. Does anyone have any advice, or resources, or case studies I could look at.

Thank you very much

[peteroznewman]

The wheel applies force in many directions.  Upward when it is in uniform motion or standing still, but a large fore-aft force when the brakes are applied and a lateral force when cornering.

Even larger than braking forces are impact loads when the wheel hits a curb or the edge of a pothole.

[benjrfletcher]

 

 I’ve used fixed supports at D and B at the moment.

 

[benjrfletcher]

My main aim is to model uniform motion/still at the moment correctly, then I can try more complex loading situations.  I was looking for help on setting up the correct boundary conditions for this: i.e. if or how to use a spring ground to body connection and cylindrical supports to model pin joints at D and B; 

[peteroznewman]

It’s always best to have a Coordinate System in any image to reference directions in the descriptions.  Let’s create a Coordinate System where the X axis is along the D-B hinge line, Y axis is up and Z axis points toward A.  If that already matches the orientation of the Global Coordinates, then you can just use Global Coordiante System.

Make the hole at B a Remote Displacement and set X, Y and Z values to 0 while leaving rotations Free.

Make the hole at D be a Remote Displacement and set Y and Z values to 0 while leaving the other four Free.

Make the holes at C be a Remote Displacement and set Y to 0 while leaving the other five as Free.

Make the holes at A be a Remote Force and apply the force components as desired. Initially, weight of the car at that one tire in the Y component.

In the Solution branch, insert a Probe to get the Reaction Force of the Remote Displacement at C to know the force required in the suspension spring to support the force going into A.

These constraints will allow all the natural bending to occur in the lower wishbone arm.  You could replace the Remote Displacement at C with a Sping to Ground and Preload it with the Reaction Force measured with the Remote Displacement if you want.  Then the wishbone will rotate up or down when you apply more or less force at point A.

A more accurate model would be to add the upper wishbone arm with similar constraints except there is probably no point C on the upper wishbone, and a third body with the wheel axle that connects to the two A points on the two wishbones.  Move the tire force to the midpoint of the axle.  Use one Revolute Joint on the lower arm point A to connect to the matching holes on the third body and a Spherical Joint on the upper arm at point A to connect to the matching feature on the third body.  You can probe the forces going through these joints and find they will be higher than the Y direction force you are putting into the center of the axle because of the extra distance of the axle midpoint to the connection at the A points of the wishbones.

[benjrfletcher]

Hi thank you.  I followed your instructions, and put in all the necessary remote displacements.  The results are a lot better!  I get a very large deflection, so I would like to put the spring in: could you help me with the right settings for the spring. 

[peteroznewman]

How large is very large?  Please insert a screen snapshot of a Total Deformation plot into your reply.

Also reply with the value of Force for the Probe of the Reaction Force of the Remote displacement at C.

Create a Coordinate System using the two holes at C.

Insert a Spring under the Connections folder and change it to be Body-Ground.

Select the two holes at C as the Mobile Scope.

[benjrfletcher]

[benjrfletcher]

[benjrfletcher]

[benjrfletcher]

[benjrfletcher]

 

fSo now I set the preload to -10859, and set the others to what you told me, and it seems to be working, I get 18mm deflection, which is essentially the same as with the boundary conditions in the first case.  I have set pics of all the settings I've used.  

 

[benjrfletcher]

When I run the simulation using the spring, do I suppress the remote displacement at C?  (i tried it suppressed and it doesn't seem to work)...

[peteroznewman]

Yes, you want to suppress the remote displacement at C however the equilibrium of the part depends on a 10,859 N reaction force at point C with the Remote Force at point A at its full value.  What is the full value of that force?

Is the Remote Force “step applied” or Tabular that begins at 0 N at t=0 and ramps up to the full value at t=1s?  You want it step applied.

A different way to configure the spring is with a much higher spring rate and zero preload.  The spring rate you have of 800 N/mm means that the spring was compressed by 13.6 mm to get to 10,859 N.    If you use a spring rate of 10859 N/mm then the spring only needs to be compressed by 1 mm to get to that reaction force.  In this configuration, you can have the Remote Force at point A be Tabular and start at 0 at t=0 and ramp up to the full value at t=1.  This will be very easy to converge and result in 1 mm of deformation of the spring at point C.

[benjrfletcher]

At the moment it is ramped; when you say step applied, what steps should I set it too, i.e how many, starting at what force, etc .

[benjrfletcher]

You said, "A different way to configure the spring is with a much higher spring rate and zero preload.  The spring rate you have of 800 N/mm means that the spring was compressed by 13.6 mm to get to 10,859 N.    If you use a spring rate of 10859 N/mm then the spring only needs to be compressed by 1 mm to get to that reaction force.  In this configuration, you can have the Remote Force at point A be Tabular and start at 0 at t=0 and ramp up to the full value at t=1.  This will be very easy to converge and result in 1 mm of deformation of the spring at point C"

When I try to solve it as such, I've suppressed the remote force at C, I set spring stiffness to 10859 N/mm, but it says invalid boundary conditions.  Where have I gone wrong? 

[peteroznewman]

Ramped

Step Applied

[peteroznewman]

I don't know what made the Invalid Boundary Conditions error show up.  Right click on the error message, is there a pop-up that provides more information?  Under the Solution Information folder, look at the Solution output for more information.

[benjrfletcher]

Hi, nevermind, it was because I still had the reaction force probe in the solution.  I now have those two methods you have showed me and I get quite interesting results with both.  It seems to me, that the more realistic physics would be when I preload the spring with the correct reaction force, this gives me a very large deflection in the part at the wheel and at point C, however, it doesn't deform i.e. the wishbone isn't bending a lot.  Whereas, setting the spring stiffness to 10859 N/mm, this position of point C stays approximately the same, but the position at the wheel moves about 20mm causing a bending in the wishbone.  Do you think this would create more realistic stresses in the part.  Thanks for all the help so far!

I'll send a couple pics

[benjrfletcher]

[benjrfletcher]

[benjrfletcher]

 

 

I suppose the first models an unloaded suspension system, like a car on a car jacked with no wheels, which is then loaded i.e. by the wheels being put on, and the car being put on the ground.  Whereas, if I want to model acceleration or braking, where the spring is near max compression anyway, the other method with a remote displacement at C would work, since at max compression in the spring, point C will essentially be fixed.  Do you agree with this?  If I were to model acceleration or braking, do you know what forces or moments I would have to apply? Or suppose I want to model acceleration, then uniform motion, then braking, I wonder how I would do this.  Let me know if you have any advice or resources which I could read!  

 

Many Thanks

 

 

[peteroznewman]

In the image of Total Deformation, note that the time is set to 0.68 which means that only 68% of the full remote force has been applied.  You need to set the time to 1 to get results at the full remote force.

In the image of Equivalent Stress, note that the time is set to 0.95 which means that 95% of the full remote force has been applied. You need to set the time to 1 to get results at the full remote force.

To compare two models plot the same data in both images.

Since you have Show Undeformed Model, you can see where the original geometry started, but the display of the deformed shape can be abitrarily scaled using the value in the circled box.  When comparing two models, you want to make sure both images are using the same display scale factor.  If the scale factor of 1.0 results in a deformed shape that is too close to the undeformed model, you can type in a larger value like 10, 100 or 1000.

If you draw the orientation of the wishbone in the position it would be in when the car weight is supported by a lift, and you use the true spring rate on the spring with zero preload, you can ramp from 0 to the full force the ground applies to the tire and the wishbone will move up (relative to the frame) the amount it does when the car is lowered onto the ground.  There will be significant rotation of the wishbone as the force ramps on.  To get an accurate result, under Analysis Settings, you must turn on Large Deflection.  This will cause the solution to take longer to compute.

If you draw the orientation of the wishbone in the position it would be in when the car weight is supported by the tire, you can use the Remote Displacement at C to hold that point fixed.  Both models should end up with the same amount of bending in the wishbone at the same location in space. The bending of the wishbone creates insignificant rotation of the wishbone so there is no need to turn on Large Deflection and the solution is calculated more quickly.  This is what you want if you plan to use Topology Optimization.

The geometry you started with is a relatively thin solid plate.  This shape is not very good at resisting bending.  Wishbone arms have more thickness to resist bending, but they are generally not solid through the thickness as that would greatly increase the mass.  They may be 3D sheet metal forms, have pockets milled in to reduce mass or be made of pipes as shown in the images below. Once you increase thickness, the amount of bending will go down significantly.

[peteroznewman]

Some information on tire forces.

https://www.brachengineering.com/content/publications/SAE-2000-01-0357-Brach-Engineering.pdf

https://simpletire.com/learn/tire-news-information/circle-of-forces

https://www.racecar-engineering.com/tech-explained/tyre-dynamics/

 

[benjrfletcher]

When I preload the spring, if the reaction force is downwards, should the preload load be positive or negative load (for compression) ? I get two very different results depending if it is positive or negative.

[benjrfletcher]

[peteroznewman]

Spring elements are defined such that tension is a positive element force and compression is a negative element force.  You need the spring to be in compression so use a negative force which will push the lower wishbone down since the other end of the spring is above it.  The force on the tire will push the axle up and the solution will find the equilibrium.

[benjrfletcher]

My results don't seem correct.  Shouldn't there be a large amount of movement in the system as the spring compresses, until finallly it hits some equilibrium point.  Whereas, in the simulation, there is very little deflection, and hardly any movement!?

Ill attach an image

[benjrfletcher]

[peteroznewman]

It looks like it may be correct to me. The lower wishbone is bending that much because the thickness of the part is small.  As a quick verification, you can change the properties of that body from Flexible to Rigid and solve again.  That will show the equilibrium angle of the wishbone without bending. If the preload in the spring matches the measured force of 8549 N, the spring will not change length when it solves and the deformation plot will have very small values.

Where are you applying the Remote Force?

[benjrfletcher]

What is I want the spring to change length and to be like a real spring in a suspension system, what would I do in that case?

[peteroznewman]

What axial stiffness have you used for the spring?

You can run the solution with 8549 N of Preload and the Remote Force at the value you used at t=0 to start in equilibrium, enter a value of 0 force at t=1.  The solver will rotate the wishbone down as the spring gets longer until the load in the spring is zero and the force is zero. If you make the End Time 2 s, you can use the starting force at t=2 and see the spring compress back to its working length it had at the start of the simulation.  Before you run that simulation, under Analysis Settings, change Large Deflection to On.