New to FEA & ANSYS Workbench - Help Please

[RayTrowsdale]

Hi Everyone.

I design filter housings which are made from die cast aluminium. They consist of a head and a bowl which screw together with a stub acme thread. Some models have a single turn of thread, some with one and half turns and some with two turns.

I am having problems analysing the housings. I keep getting the message that there are not enough supports. What I have done is to add a fixed support to the top of the filter head and a cylindrical support (fixed tangentially) to the bowl which stops it turning. The loads consist of pressure on all the internal surfaces.

My contact area is all the thread area which I have made frictionless.

At the moment, the CAD geometry I have sent to ANSYS is unchanged from the real geometry - what I mean by that is that I haven't simplified it at all and that there are small clearances all around the threads. Could that be the problem? Have I constranined it correctly?

I realise that the real solution to my problem is more training or the use of a mentor but training is something that is unlikely to happen quickly and a mentor just doesn't exist in my company!

many thanks for any help in tackling this kind of problem.

 

[cvan42]

dhray,

I believe what is going on is your contact pair isn't near or touching on the first substep and your bowl is taking off to the moon since it's only constrained not to rotate along the axis. Let it run to nonconvergence, look at Solution->Deformation->Total, right click and Evaluate All Results. If your max deformation is something ridiculous like 240 inches the bowl is heading down the street, and contact is your issue.

If this is the case, you're going to have an interesting time with the contact pair along a full detail thread pair.

1) If you're in SolidWorks, do an interference check, make sure your assembly part orientation is good. Do a Utilities -> Geometry Analysis to check for short edges/sliver faces in the model. Put a 0.001" -0.005" radius on your thread root and crest lines, no thread is perfectly machined and if there is no tool radius your stress might not be accurate.

2) If your model is good (and you have a really good computer), proceed with modeling the entire thread if thread stress is what you are concerned about. If the node count is too high, go back to CAD and cut the assembly to quarter model, put frictionless supports along the cut surfaces.

3) Select all thread surfaces, callout a surface mesh size of 0.001" and preview the results. If your node count is too high for your computer, increase the mesh size. For tetra elements, I'm guessing 0.005" mesh size could be adequate.

4) You should only have one contact pair for your threads. Double check your automatic contact pairs, if you're not paying attention, it could be adding pairs to surfaces not even close to each other. Make it Augmented Lagrange, stiffness factor manual at 0.1, update stiffness each iteration, pinball radius automatic detect. You could do frictionless, reality would be somewhere around 0.15-0.3. If your filter is huge industrial size, add a gravitational force component to the loading. If not no big deal.

5) On your Connections, right click ->Insert->Contact tool
On contact tool, insert ->Penetration, Gap, Status
In the Contact Tool, Contact Side, pick Contact. Run the results to check your pair is near, sliding or sticking. If it is far, you have a gap issue. The best fix is to back into your CAD, move the parts so they're around 0.0005" from interfering. Go to geometry-> Update, and rerun the contact tool.

6) Check your loads. If your pressure is 80 PSI, be sure at time 0 it is 0, and time 1 it is 80. If both 0 and 1 is 80, there would be issues and it would probably not converge. Start off with a time step around 0.05 s. If it takes more than ~14 iterations to converge on the first substep, change timestep to 0.01 s. If still not converging, probably there is a contact issue. Change the contact settings, increase the stiffness factor to 0.5, add a contact displacement of 0.00005, etc. etc.

You should be alright if all those are checked.