How will the tank be held in place in real life?
Inertia relief is nice and easy, but you also give up some control on the model, and introduce a possibility for errors.
If you are going to use it, at least first do an analysis without it, so you know roughly what the results should look like.
Hey guys,
I have a question about the buckling results aquired from my analysis.
I am busy analyzing a component that has a global buckling mode the opposite direction of which it is being pushed as a result of an applied force.
Negative result filter settings are on.
When I repeat the analysis with the filter off it swings in both directions, the expected direction the loads pushes it and the opposite.
I am very skeptic if the results produced are reliable.
Does anyone have any experience which such results?
Does the visual deformation direction have to be the direction the forces push the component to?
Does anyone have an idea as to why this is happening? The mecbasiceng license is listed on the flex license server as being available and the feature.dat file on my computer lists both licenses. I am using Creo 3.0 Engineer IV, datecode M050.
I discovered this problem while taking the "Introduction to Creo Simulate 3.0" elearning course when I couldn't complete some of the exercises and got a message saying I needed an advanced license.
I reported my problem to the PTC tech support folks and they said they configured their machines just like mine and couldn't duplicate the problem. It's going on two weeks now and I haven't been able to get any further response from them.
The only thing I can think of is that my problem with the elearning course saying I need an advanced license for some of the exercises is due to the mecbasiceng license not being loaded.
I am attaching files I have submitted to PTC Tech Support. Note that the snapshot of the licenses on the flex server was taken when I wasn't using Simulate.
Fiona,
Have you watched this video from PTC on buckling?
http://learningexchange.ptc.com/tutorial/799/creating-a-buckling-analysis
The deformation results swings the column in both directions.
Oh ok. So the main point is that with default settings you may get better results with Creo 3. This is very good to know.
I haven't done any simulations with Creo 3... yet, but I have worked with Ansys for a couple of years and I agree that it tends to require a bit of experimenting with solver settings from time to time. We are only starting to introduce Creo simulations in our R&D and the aim is to get more and more engineers to do some simple simulations and maybe even more complex ones when they get to know the software. Your comparison shows that there are some good and easy capabilities in Creo Simulate, which may prove to be very useful for us.
Thanks again.
Nice work.
This is not posted as a question, but I just wanted to show that Creo does just as good as ANSYS when plotting stress linearizations, and I like the formatted output better than ANSYS. If you're attempting to determine the Kt value with Creo, you will get good results. The Kt from the FEA was compared to hand calculations and a macro in spreadsheet. All Kt values for all of these methods came out to approximately 1.32
Kt = (8794.7/6658.78) = 1.32
Denis,
Les deux graphiques sont faits avec Microsoft Excel. Dans Creo Simulate, on peut exporter les chiffres (x,y) d'un graphique à un fichier *.xls. Cliquez "File-Export Excel" (Creo 2.0), dans Creo 3 on clique "File-Save As". Pareil dans ANSYS ...
/M
Mats,
Merci pour le renseignement, j'avais pas pensez à Excel avec Creo, avec MathCad, Excel c'est bien.
Cordialement.
Denis.
Hi
nice and interesting initiative. Thanks for sharing.
I am wondering why having selected a mesh so different between Ansys and Creo.
Creo uses "P-method" which raises the degree of the polynomials that describe the displacement field. ANSYS uses the H-method where the polynomial degree is fixed,for all elements, and is normally 2, i.e. quadratic elements. This means that with the H-method and quadratic elements, stresses and strains can only vary linearly within an element. In creo, the P-level can be raised to 9 and this is done adaptively; in areas with higher stress gradient, the P-level is raised so that a more complex displacement field can be described with a single element. This allows a much coarser mesh than is required if the P-level is fixed to 2, as is the case for Ansys. So the meshing algorithm in Creo is adapted for the P-method, while the meshing functionality in ANSYS is adapted for the H-method. I guess the P-method makes it more difficult to implement nonlinear functionality, which is undoubtedly much better in ANSYS than in Creo Simulate. I think ANSYS has the P-method, but the mesher is not adapted for it, and the P-level is limited to 5 if I remember correctly. And I'm not sure of what nonlinear functionality is supported with the P-method in ANSYS. My guess is that very few Ansys users utilize the P-functionality in Ansys. Long story short: you can't compare Creo Simulate Mesh with ANSYS mesh.
My goal is to do a sensitivity analysis of a part where one dimension (x) is varied. What I want out, is the force that would cause the part to reach the yield point for each change in 'x'.
In other words, the user is allowed to reconfigure an assembly, and we want to plot the expected failure load for each configuration. Anyone know of a way to do this?
Just create a local parameter called Load. While applying load specify "Load" instead of value .then proceed static & Sensitivity study
You want to increase the load until some location in the model reaches yield stress, and then identify what that load is, correct?
If it is a linear problem, then create a computed measure M = Mtrl_yield_limit/Stress_model_max_von_Mises.
For example if you apply 1 kN, and M turns out to be 0.1, then you will need to scale down the load to 0.1 or else your max stress will be above yield. Conversely, if M turns out to be 10, then you must scale up your unit load by a factor of 10 to reach yield stress.
Apply a unit load, say 1 kN. Run sensitivity study. Plot yout measure M vs. x. Then M corresponds to the factor*(unit load) that will cause the material to reach the yield limit for each x.
In linear problems, everything is proportional- double the load you double the stress, displacement etc. For nonlinear problems you can't do that.
If it is a nonlinear model, then there are no shortcuts. Increase the load in steps, and see when max von Mises stress exceeds yield limit.
/Mats L
Nice.
My lunchtime meanderings in Ansys following on from my earlier frustration where I could not get it to work in Creo
Below took 2 mins to build and run 20mins to make pics and post.
Hi Mats
ok fair enough
Hopefully regardless of the meshing the results are similar and outcome the same
Best regards
Hello everyone,
We are about to make some key decision on software selection and one point of interest I have is the degree of standardisation I can implement in our product development process.
That is, when we develop new products (generally very similar one to another), we generally do not perform FEA or CFD because of the time it takes, or lack of skilled staff.
My intention is to standardize the way we model, so we can apply template for FEA or CFD and get the calculation running more quickly. The next step would then be to automate it.
That is, I have my model finished according to defined modelling policy, I then click a button and my FEA, CFD start.,
This could be pushed further, Upon check in CDF or FEA start automatically (as the CAD worker does to publish models), and if results are outside certain limits a warning (email) is sent, otherwise it is all good. Or simply the results are posted somewhere.
Obviously, as you can see, I am not expert in this area.
My questions to you are
Have you reached this level of standardisation and automatition
Do you think this is possible (technically speaking as I appreciate you do not know our products)
Thanks a lot
Both of these answers were useful. I was able to run a sensitivity analysis on the load parameter and determine that the response was indeed linear. Then I could extend the results of the dimensional sensitivity to find the failure load at each configuration.
I like ANSYS for CFD/FEA. The workbench interface is very intuitive and the translation in from Pro Engineer is excellent.
You can do parameter studies by naming parameterized dimensions with an DS_ prefix and Pro Engineer will automatically open and generate the next model for you.
Also if you implement EKM then you can use an automated workflow for future studies by even non-trained personnel. I think this is the answer that you are seeking.
Rule 1: Don't model anything using surfaces.
Rule 2: Make everything pass a Geometry Check before any simulation.
Rule 3: If the parts are assembled make sure that they are actually touching each other where you want them to touch.
Rule 4: You should model your parts so it is easy to change the dimension to worst case scenario (i.e. Minimum/Maximum conditions). Don't use created dimensions for important tolerances.
Hi Mike, thanks for your reply.
You got it right. The overall idea is to put FEA and CFD in our standard process but we do not have all our designers to be FEA/CFD expert.
We keep the very special cases for the expert.
Thanks for the info.
OK so even this level of automation will even work with Ansys, however will all the files (apart of the results files) be integrated into Windchill for future auditing and traceability ?
Best regards and have a good weekend