Thanks for all your replies. After some trial and error I've learned a few more things about Mechanica. Using contact analysis was a given for our problem but options such as mesh refinement and multi-pass had differing effects which could prove useful in some situations. Using mesh refinement can increase the accuracy in the immediate area of the contact and high stress gradients but where stress gradients aren't so steep a normal run without mesh refinement seems to give acceptable results. Mesh refinement takes much longer to run so if you're just looking for rough numbers and general areas of high stress not using mesh refinement will save time. Multi-pass also takes longer but once again depending on your purpose you may get away with single-pass. I also found that using a plotting grid of 6 or 8 gives a much smoother appearance in the output although at the expense of time. I would say running with the default of 4 gives pretty good results but if a high resolution image is required (say for a report) you may want to spend the extra time and use 6 or 8. Plotting grid doesn't change the actual stress levels but it does change the way the output looks on the screen. High stress gradients also look better with a higher plotting grid number. Another thing that I learned was that when doing contact analysis it is important to set the legend properly to see all the stress areas that may affect the design. The default legend gives you an indication of the top stress levels but tends to wash out the lower level gradients by making them all one colour. By lowering the maximum stress value shown in the legend you can get more insight into the stress distribution. Lastly, using the "contour" option when viewing results can point you to stress areas that you can't see easily from a shaded image, even with cutting or capping surfaces. With contour mode the model is transparent but allows you to see the stress iso lines so you can more easily see inside the part and zero in on areas on interest.
Anyway that's what I gleaned from this exercise. If anyone disagrees with any of my statements please let me know.
My only comments are related to the plotting grid value.
1) The run time should not be affected very much by changing the plotting grid number, just the results file sizes. I am surprised that this is the case for your model. And yes, your output resolution is absolutely better with a higher plotting grid value.
2) Also, the maximum stress reported in results can increase if you increase the plotting grid value. I have proven this to myself in the past with a test model.
Randy Speed President and CEO Speed Consulting, LLC (972) 938-0490 ph (972) 937-2319 fax www.speedconsulting.com
Thanks for clarifying this. I was under the impression that the plotting grid was just for visuals but I checked the documentation and you are right it actually affects the results that are reported. In high gradient areas you could probably miss some high values if you used too coarse a plotting grid.
Thanks for the feedback.
----- Forwarded by Mark Dolata/CenterLine/CA on 02/02/2006 12:06 PM -----
"Randy Speed" <email@example.com> 02/02/2006 11:50 AM Please respond to "Randy Speed" <firstname.lastname@example.org>
To <email@example.com>, "PTC/USER Simulation and Behavioral Modeling Discussion" <-> cc