Multi-body simulation of a two-stage gear transmission in Rigid dynamics

  • 74 Views
  • Last Post 09 January 2019
  • Topic Is Solved
Akash91 posted this 19 December 2018

Hi,

I'm trying to simulate a two-stage gear transmission system in Rigid dynamics as part of my thesis work. To this effect, I've modelled a basic gear transmission in Creo, assembled it, created a step file and imported it into rigid dynamics module in Ansys Workbench. 

I've deleted auto-generated surface contacts and defined contacts between bodies via joints. I've defined revolute body-ground contacts at the ends of the three shafts so that they have only a rotational DOF along Z. Gears mounted onto the shafts as well as the supporting bearings were defined with fixed body-body contacts w.r.t shafts. The idea is that when rotational velocity is given at a revolute body-ground shaft joint, the gear and bearing fixed on the shaft rotate along with it. Finally, I've created named selections for the gear teeth and defined  contact between mating gears as frictionless contacts so that the rotational velocity from one gear can be transferred onto the second, eventually from the first stage onto the second stage.

However, when I run the simulation, the simulation ends prematurely with the warning message: Although the solution failed to solve completely at all time points, partial results at some points have been able to be solved. I've tried it with different settings but the simulation breaks midway every time. Does anyone have an idea as to what might be the problem here? Does it mean the joints are not defined properly? Or does it mean settings need to be better?

Thank you very much! 

PS: I've attached screenshots of the geometry I've used for reference

Attached Files

Order By: Standard | Newest | Votes
peteroznewman posted this 19 December 2018

Hi Akash,

Rigid Dynamics is not the tool to get bearing loads on a shaft. The reason is when you have a shaft in Rigid Dynamics, you only need one revolute joint to allow the shaft to rotate.  The Revolute joint could be anywhere along the shaft because the shaft is rigid. There is no need for a second revolute joint. It would be redundant.

The real shaft is flexible and has two bearings to support the side load of the gear forces. If you bring that shaft into Static Structural, you can apply forces to the faces of the gears, observe the shaft flex and show the bearing reaction forces at the two bearings, one at each end of the shaft.

Regards,
Peter

  • Liked by
  • SandeepMedikonda
peteroznewman posted this 21 December 2018

Hi Akash,

I think you mean you spent two weeks learning how to create geometry in Creo. That is very valuable knowledge. Engineers need to be good at CAD to do analysis.

You haven't spent that much time in ANSYS Rigid Dynamics have you?  In any case, you could apply a moment to one shaft and Rigid Dynamics could plot the angular velocity vs time curve as that moment accelerates the gear train, but you will get a lot of noise in the result and you can calculate that curve by hand from the known mass moment of inertia of all the shafts. There isn't much of an impact when a gear train starts moving. There is very little clearance or backlash in the gears.

Rigid Dynamics would be more useful to study the linkage of parts that steer the front wheels of a car, or the mechanism that latches the car door closed.  Here is a recent discussion on the front wheel suspension of a car.

Regards,
Peter

If one of the posts has answered your question, please click the Is Solution link to mark the discussion as Solved or reply with a follow-up question. You can always ask a new question in a new discussion.

  • Liked by
  • Akash91
peteroznewman posted this 19 December 2018

Hi,

May I ask what is the goal of the Rigid Dynamics simulation?

The gear ratio of each mesh is known, so if you have an input shaft rotational velocity, you also know the rotational velocity of the other two shafts.

If you want an animation to illustrate the gears, you can delete the contacts and apply the known rotational velocity to each revolute joint.

Regards,
Peter

Akash91 posted this 19 December 2018

Hi Peter,

 

Thanks for the reply.

I'm actually doing NVH analysis on the gear transmission but since I've designed it myself and don't have any experimental data/loads to go with, I'm trying to do an MBS in order to get loads and reaction forces on bearings and gears. These loads I can then use in harmonic analysis to analyse for vibrations.

I've applied rotational velocity on the driving gear and a moment on the driven gear in the second stage. I'd like to see how the reaction moment is on the bearing and driving gear in the first stage. I hope you get the gist.

Thanks!

Akash91 posted this 20 December 2018

Hi Peter,

Thanks a lot for your input. I'll try to do a static structural analysis on the system as you suggested and find out how the shafts, bearings and gears would react to applied loads.

But I'd like to ask, what exactly is rigid dynamics used for? Is it just to Analyse the motion within a system of bodies? If this is the case, the module practically does nothing as we define the motion of each body ourselves w.r.t connecting bodies. Could you perhaps let me know, with regards to my example of a gear transmission, what I can hope to find using rigid dynamic analysis?

I'm still a student taking baby steps into the world of FE simulation. So any information you can give would be greatly appreciated, thanks!

peteroznewman posted this 20 December 2018

Rigid Body dynamics is good to solve for unknown acceleration and velocity of mechanisms driven by a force or moment, such as an escapement mechanism in a clock.  If you RMB on the YouTube video, you can select Loop.

Akash91 posted this 21 December 2018

Hi Peter,

Thank you so much for the reply and for the link.

So since I know the gear ratio, I can directly calculate the speed of the driven gear and RBD simulation would be redundant. But since I've spent about two weeks on RBD, I'd still like to use it to find out the reaction on gear teeth, let's say the impact when rotation starts, is that possible?

Thank you for your time and help. Happy holidays!

Akash91 posted this 09 January 2019

Hi Peter,

A very happy new year to you.

I followed your advice and was able to simulate the working of gears albeit with the use of a revolute joint. So, there wasn't much reaction as you predicted on the gears.

I was talking to my supervisor yesterday and he pitched in with the idea of getting rid of revolute joints at the shafts and suspending the shafts using just bearings. So the weights of the shafts and gears would be bore directly by the bearings which would develop reaction forces when the gears are rotating. So I modelled bushing joints in Ansys with available bearing stiffness matrix values and tried to run the simulation by applying a rotational velocity on one bushing joint. I read a recent discussion on Spur gear analysis where you advised to check the contact status foremost. Turns out it was why the simulation was stopping midway.

To rectify this, I imported the model into Static structural, used the 'Adjust to touch' tool to close the contact between meshing gear teeth and transferred the model into a new rigid dynamics module. I've also read through Ansys help documents and as per suggested best practices, incorporated gravitational force and set the restitution factor to 0. It all seems well and the simulation runs but somehow the system is in a lock position. I'm not entirely sure if I've over constrained the model or if I'm unable to apply rotational velocity through bushings but there is no rotation whatsoever between the gears.

Do you have any advise for me regarding this? I'm attaching an archive file of my project. I'd really appreciate it if you could check and inform me of any errors or propose any ideas to solve it.

Thank you very much!

Attached Files

Close