Rotating bending on a crankshaft

AndreaDR

Hello everyone, I'm newbie with Mecway simulation. I would ask if it s possible to simulate the rotating bending (Wöhler) on a crankshaft. What I mean if it s possible to made a support that act like a bearing and allow the rotation. I also would ask if It s possible to see the rotation in the animation.
Thanks in advance and best regards.

Victor

Yes, that should be OK if you're just looking for stress over one cycle. Use:

Analysis settings:
- CCX
- Nonlinear Static 3D
- Quasi-static

Meshing parameters:
- Maximum element size 30 mm works but the default doesn't.

Support with Node-surface coupling (rigid, RBE2), frictionless support if the mesh is fine enough, or elastic support.

Use node-surface coupling (rigid, RBE2) to apply a prescribed displacement and rotation as a function of time. These need to be on two separate nodes.

The solution animation will show the rotation.

Sebastianmaklary

I've done a bunch of analysis of gears where I've used the method that Victor proposes with a time-based rotation constraint on a node. Works fine.

Sergio

Would be possible to make it dynamic and also include inertial forces due to rotation (at least at constant speed)? The ideal model would include also connecting rods and piston, and an external force in the piston head.

There is a recent video on youtube of a cranckshaft assembly, but guess that they are using MBDyn for the dynamic stuff

AndreaDR

Hi Victor, firstly I would thank you for your advice. I've tried to set the case as you told me but I still don t figure it out. I am not sure about the constraints I set and also about time step. I ve seen from another example but it doesn t work.
I have a question: does the node on which you need to impose the rotation conditions have to belong to the rotation surface in question or can they be assigned to an auxiliary node that is on the rotation axis?
I finally ask if you can check my work.
Thank you again.

disla

-One boundary conditions is repeated.
-You can not apply BC on the Surface of the rigid body
-Your Young modulus is very small. Don't you mean 100 GPa ?¿.
-If you are willing to see some bending on your shaft you can't make it Rigid.
-You probably will try a sort of diferent configurations so try to reduce your model as much as possible just untill you find the right configuration.



Seems very interesting problem. I'm not experience enought with this kind of analisys so I do not dare to propose anything.

I would only say that I see some problem with the nolinear analysis. Some loads are not followers so you will have to look carefully where and how to apply them. Sergio's idea seems good to me. The piston only oscillates vertically and the pressure at its base does not change direction, only intensity. For me that must be the point of entrance of loads in the system.

JohnM

I did this work for 20 years, there are many ways to approach this problem depending on the fidelity and design goals - an ambitious starter project!

For a general approach if the goal is a fatigue assessment, you will want a good looking mesh that runs efficiently. To this end, you can develop loads as an initial step and apply them to a static model. Comsider rotating the loads, not the crankshaft. You can do this if you are clever with your loadings.

Mesh: pin and main fillets should be meshed with very good detail. Skip this refinement step until you've got a working model.

Loads: For constant speed operation, you will need to consider the centrifugal load, and the load at the crankpins of your crankshaft. These days I would use Open Modelica to create a crank slider system that allows you to develop the crankpin loads over 720°. You will need the crankpin stroke, connecting rod pin to pin length, mass and moment of inertia of the connecting rod, mass of the piston, operating speed and a combustion pressure curve. There is a small offset built into the conrod-piston wrist pin that you can probably neglect for a first cut. Ideally you will want to output these developed loads in a cylindrical system, radial and tangential, to simplify application to the FEA model.

The loads will need to be applied to the four pins based on the firing order of the engine. For a typical 4 stroke 4 cylinder, this is 1-3-4-2. This means that pin 3 will be 180° behind pin 1, pin 4 will be 360° behind, and pin 2 will be 540° behind.

How these loads get applied in the finite element model will be a function of your design goals, but a simple way to start would be to create a node to surface coupling at each pin with a limited width surface with the masternode centered in the crankpin. These nodes can be rotated into a cylindrical system using custom model definition. You can add levels of sophistication that include squeeze film stiffnesses but this is a good first cut.

Reaction forces will be taken out as bearing loads and crank torque. Include the flywheel in this model because if you take the next step to applying dynamic loads you will need it. You can use the same approach as the crank pins for the mains but this is probably extreme given the size of the bearings. A more sophisticated approach would be to react through a set of springs that represent a squeeze film. Crankshaft mains bearings will experience edge loading that might be significant depending on your design goals.

This approach allows you to run many cases quickly to develop stresses for calculating fatigue. You can also easily change operating speed and combustion profile depending on the load point, usually peak torque and wide open throttle.

Sergio

I have recovered an old model of a simple compressor that works similar, I will create another version as soon as possible with even more simplified components, including the con rod bolts, piston rings and cylinder liners, and adding a load on a top of the piston and bolt preloads. Also the contact pairs must be reversed to get the contact stress at the cranckshaft and piston pin sides.

My idea is to have a very (extreme) simplified assembly model, and then replace only one component at time with a more realistic representation to be able to get more accurate stress components.

But still is a static simulation, I´m very interested in running this in a dynamic way to include accelerations and be able to run at different rpms. I left this model if someone want to take the post, now takes 2 hours to solve with 16 cores. I will include the next simplified model also in the next days

disla

I have applied different combinations of the recommendations and hints contained on the attached files, but I found a big issue solving with NLGEOM Quasiestatic.
Shaft bends with centerline offset properly but the system lacks from rotational inertia to sort out the critical turn around and keeps oscillating no matter what.




The vertical position is an equilibrium position, and I don’t think I can inspect the whole 360 rotation without enforcing the rotation by hand with the rigid bodies.

The load driven has many points in common with the Spur gears and needs a brake to remove the rigid body rotation.
I have found something curious. Centrifugal force helps on the convergence for some reason. Similar to that dummy load on contacts.
I’m using nonlinear springs to remove the contribution of the piston once it is returning to the starting position before a new expansion.

Good looking but not satisfied yet.

Thanks both,

JohnM

Make sure you are considering the flywheel inertia in the system, if you are actually going to run a full dynamic you will need this for sure. You will find that at normal operating conditions, the inertia effects dominate the system. Also, the piston conrod loadings and their contribution to the dynamics of the system are essential. The Piston at TDC during the exhaust stroke still imparts significant load on the crank pin in the opposite direction for example. Look for references with light end / heavy end or rotating/reciprocating discussions about how to simplify the piston/ conrod system in this type of analysis. There are papers written about this.

Back in the 1990s when we built these models with stone knives while wearing bear skins, the idea of running a full-up model like this would have been impossible, so I'm very interested to see how far you take this. We would use stick models that were tuned based on the modal response to the solid or use a forgotten technology called super elements. It's fun to see the potential unlocked with the run times we get these days!

disla

Thanks JohnM,

I have moved to Dynamic.
I think I have dirty spark plugs but I‘m progressing.