NL Transient - Contact
Hello, i am still working on my problem with a NL Transient Contact Analysis. I've searched for Tutorials and help, but i still have problems with the relation between contact of two solids and the applied load.
The gravity is supposed to push the wheel against a plate for a running time of 1s. Therefore i created gravity in negative y-axis with tabled1 of 0 to 1s and a tload collector. Between 0.6 and 0.8s a load is applied on the bottom of the wheel. As a result the wheel should move in positive y-axis against the gravity. The whole wheel carrier, is constrained in all dofs except rotation of x-axis (dof4).
As contact between the wheel and the plate i chose stick: While gravity is pushing the wheel down it has contact with the plate, as soon as the load is applied (between 0.6 and 0.8s) the contact should be separated. For NLPARM i have N=10 and DT=0.1 ==> 1s running time (TTERM=1).
The problem is, that the model does not converge, even after 16 hours i have no results. Does somebody know what exactly the problem is? Maybe the gravity oder the contact is wrong, but i cant see the problem.
I've attached the hm/fem file, i hope somebody can have a look.
Thanks!
Answers
-
Hi User1357
You can start to constrain your model a bit more, just to make sure that convergence is not due to any SPC mistake, try to fix upper bar 1-6 for example!To save computational time you could also consider to use shell elements instead of solids, also use beam elements!
To stretch it even further you can remove most elements (just keeping those close to rbe2/loads/spc and then stitch the model together with RBE2 element. This is just for making debugging runs quick!Also, I am not sure how you want your loads, seems like the wheel will be sticked to the ground and that you apply a force at the contact interface. Is this as intended?
Hope this can help!
/johan0 -
Johan Dahlberg_20306 said:
Hi User1357
You can start to constrain your model a bit more, just to make sure that convergence is not due to any SPC mistake, try to fix upper bar 1-6 for example!To save computational time you could also consider to use shell elements instead of solids, also use beam elements!
To stretch it even further you can remove most elements (just keeping those close to rbe2/loads/spc and then stitch the model together with RBE2 element. This is just for making debugging runs quick!Also, I am not sure how you want your loads, seems like the wheel will be sticked to the ground and that you apply a force at the contact interface. Is this as intended?
Hope this can help!
/johanHello Johan,
thanks for your quick response.
In reality, I have a wheel test rig. The wheel carrier is loaded with weight on the upper bar and pressed against a drum. The drum rotates, and a mounted striker bar generates an impact on the wheel. As a result, the wheel carrier (which can rotate along the X-axis) is slightly "thrown" upwards. This is equivalent to riding a bike over an obstacle, causing an impact. This impact exerts a high force for a very short period.
For simplicity, I assumed a plate against which the wheel is pressed. I calculated the force of the impact to be 10,000N, acting within a very short time. If I only consider gravity and the contact between the wheel and the plate, I get results after 20 minutes (for this, I could also use linear-static analysis). However, as soon as I add the force, the solver no longer finds a solution.
When looking at the results of the non-converged model, it seems as though gravity is not slowing down the impact, and the wheel carrier is rotating and deforming infinitely. If i choose all dofs=0 for the upper bar, then i receive a result after 20 minutes, so i guess i have a SPC problem. But restricting all dofs at the upper bar is not like in reality .
I hope you could understand what i mean.
Thank you for helping!
0 -
User1357 said:
Hello Johan,
thanks for your quick response.
In reality, I have a wheel test rig. The wheel carrier is loaded with weight on the upper bar and pressed against a drum. The drum rotates, and a mounted striker bar generates an impact on the wheel. As a result, the wheel carrier (which can rotate along the X-axis) is slightly "thrown" upwards. This is equivalent to riding a bike over an obstacle, causing an impact. This impact exerts a high force for a very short period.
For simplicity, I assumed a plate against which the wheel is pressed. I calculated the force of the impact to be 10,000N, acting within a very short time. If I only consider gravity and the contact between the wheel and the plate, I get results after 20 minutes (for this, I could also use linear-static analysis). However, as soon as I add the force, the solver no longer finds a solution.
When looking at the results of the non-converged model, it seems as though gravity is not slowing down the impact, and the wheel carrier is rotating and deforming infinitely. If i choose all dofs=0 for the upper bar, then i receive a result after 20 minutes, so i guess i have a SPC problem. But restricting all dofs at the upper bar is not like in reality .
I hope you could understand what i mean.
Thank you for helping!
Ok, good that it runs fairly quick with fully constrained model!
So you need to take some more parts of the carrier into consideration. I guess you will have some support from the rear axle and that the rear somehow is also connected to the front axle (otherwise your carrier will collapse at the pivoting point!)
If you still only wants to have your model as it is you can consider to test:
1)remove contact and just apply an acceleration (you can use your force curve just change type and adjust amplitude to some proper level)2) keep contact but add striker bump on road and move the road/plate in neg z. So "real contact" then. Let the gravity set first and then start to slide plane
In any case, it would be good if you could add some kind of spring/damper in your hinge!You can also consider to model this with Inspire motion (or with Motionsolve) as well!
/johan0 -
Johan Dahlberg_20306 said:
Ok, good that it runs fairly quick with fully constrained model!
So you need to take some more parts of the carrier into consideration. I guess you will have some support from the rear axle and that the rear somehow is also connected to the front axle (otherwise your carrier will collapse at the pivoting point!)
If you still only wants to have your model as it is you can consider to test:
1)remove contact and just apply an acceleration (you can use your force curve just change type and adjust amplitude to some proper level)2) keep contact but add striker bump on road and move the road/plate in neg z. So "real contact" then. Let the gravity set first and then start to slide plane
In any case, it would be good if you could add some kind of spring/damper in your hinge!You can also consider to model this with Inspire motion (or with Motionsolve) as well!
/johanFirst of all, thank you for the prompt response.
What do you mean by additional parts of the wheel carrier? I have connected all components with Boolean; in reality, these are securely bolted. At the point where the two rigids (all dofs=0 except dof4) are located, there is usually a bolt around which the wheel carrier can rotate. How can I add dampers to these rigids in Hypermesh? Do you mean I should additionally connect the mass that is only on the front rod with the remaining parts of the wheel carrier?
Regarding 2), I attempted this by adding a sort of impact bar to the end of the plate. The plate is still firmly fixed underneath with an SPC. Additionally, I created a velocity of 25 km/h (6944 mm/s) for the plate using SPCDs. Unfortunately, Optistruct does not seem to recognize the impact of the impact bar, leading to convergence problems again. I have attached the model here in case I made a mistake.
Regarding 1), If i use SPC with all dofs=0 and additionally SPCD with a proper acceleration at the same nodes of the rim of the wheel, i also get convergence problems.
Thank you for your assistance!
0
Altair Employee