WISH LIST

edited July 2014
My First Suggestion. I would like to be able to use ISO REGIONS not only by selecting value and displaying required span of field values but would love to use scroll on the mouse to dynamicly scroll through the model certain span of eg.stress from MIN to MAX.
It's possible in Creo Simulate and is very useful and FAST.
«13

Comments

  • Hello apadzak.

    That would be an interesting feature. Are you talking about displaying internal iso surfaces in a transparent solid, or just contours on the outside surface?

  • Yes Victor, transparent solid and just iso surfaces coloured according to the scale.
    It is really fast and shows propagation of certain values.
  • Iso surfaces can be scrolled in MIDAS as well. It doesn't have to work with scroll but could be used with the dragger for cutting plane in Mecway.
  • edited July 2014
    I have another tip. I believe that modern engineer/scientist/manufacturer can't work just on one type of software and then there is requirement for translations. CAD has got several formats available for data interchange but FEM isn't that straight forward. I think Mecway would benefit from being able to transfer pre processed meshes to other formats. Some people would want it focus on commerciall products but I think that going GNU (free licence) way would be much more beneficial. ANSYS, NASTRAN, ABAQUS have their tools to create geometries, or they work as CAD attachment. GNU License software lacks of nice and easy pre-processing and sometimes user friendly post-processing. Post-processing is very useful but pre-processing would be so much more beneficial. Z88 Aurora can deliver formats for Abaqus, Ansys, Nastran but how easy would be to extend Mecway to do really powerful things by using ELMER, OpenFOAM, CalculiX, OOFem... Verification of fluid flow done in MecWay but then transfered to OpenFOAM or multiphysisc with Elmer would make it so much more extended tool.
    Is it a massive job to make Mecway being able to export universal files for mentioned GNU programs? If not I would be over the moon with that option available.
  • Good suggestions. Not many people have asked for that which is why it's not there but I'll consider more export formats for future versions. You can already export the mesh to Gmsh's format which is fairly convertible, but it doesn't include any boundary conditions, etc.
  • edited July 2014
    I know about GMSH in MECWAY but going straight away from an analysis in MECWAY to OpenFOAM for example with material specified, boundary conditions set or just regions of surfaces, elements, nodes gathered and named to be easy used. I think that mesh generators such as GMSH, SALOME and NETGEN and their capabilities in this matter could be quite helpful. I am not entirely sure but Creo Simulate exports remeshed p-element mesh as h-element mesh to Nastran format for sure but my doubt is about selecting boundaries. Having results from OpenFOAM I can perform stress analysis in MECWAY etc.
  • I'll consider importing results from a fluid mesh to a structural mesh too but can't promise anything. Especially since this is perhaps a more narrow application than simply converting to other FEA formats.
  • Hi Victor, I have another suggestion, wish... Is there any way to enable functionality of Mecway to select verstices and edges for loas or constraints in solids? I have seen test case of Abaqus which I tried to duplicate but instead of edges I had to choose nodes to be constrained in one of the directions. It was Modal analysis and I got quite good results but probably using edges they would have been better. Highest difference on one of the modes in frequency was arround 6% in comparison to the test case. CalculiX and Abaqus werent different more than 3% in all modes and sometimes exact. Requirement was to keep mesh density the same and shape as well as the test case. The best results in Mecway were obtained using hex20 elemnts as expected but probably could be better. On of the shape modes didn't match the test case but I don't know if it wasn't a mistake on their website (Abaqus) as frequency was within tolerance.
  • Hello apadzak

    I recognize the limitation of not being able to select edges. However using nodes will give exactly the same results as edges because edge constraints would be internally converted to node constraints.

    I'd be interested to see the model. Some error with the same mesh size is inevitable, but different mode shapes is a little concerning.
  • Please have a look at this CASE: http://fobos-project.ru/abqdoc611/Documentation/docs/v6.11/books/bmk/default.htm?startat=ch04s04anf25.html
    Mode 9 should be in plane and it isn't on the attached file.
    It looks better when hex8 is being used with regards to mode shapes.
    The only difference between the models is that they have different order elements and in hex20 midside nodes have been constrained. In fluid analysis midside nodes have different degrees of freedom so I maybe in this case it creates a problem.
    I think it would be a good proof for Mecway to go through NAFEMS tests to provide more assurance of complying with high FEM standards. It is possible that I coud be the cause of the problem but it's worth to check it anyway.
  • Hello apadzak

    I think this is not a problem, just a difference. By refining the mesh several times, it looks like the "true" mode 9 should be out-of-plane as you found with Mecway. In that case, the hex8 elements in both Mecway and Abaqus are incorrectly giving it too high a frequency so it doesn't appear below mode 11. This is to be expected because linear elements are usually over-stiff in bending and those out-of-plane modes have very high bending curvatures.

    Not having access to the NAFEMS documentation, it's not clear what the correct solution should be. If the reference solution was obtained from plate theory, it can be less accurate than a solid model.

    Another possible issue is whether the meshes are the same. I'm just trying to understand the .inp format. It looks like the C3D20 file's *ELGEN command

    *ELGEN,ELSET=ALLE
    1, 4,2,1, 4,200,10

    is only generating a 4x4x1 grid of elements, not 8x8x3 as in your liml file. If that's the case, then mode 9 in Mecway matches Abaqus exactly (frequency to 5 digits) as shown in the picture here.

  • Thank you Victor.
  • edited August 2014
    There is another test case with Reference Values from NAFEMS, but it's not NAFEMS documentation so there could be an argue if this is true. http://www.compassis.com/downloads/Manuals/Validation/RamSeries_ValidTest_18_OK.pdf
    Mecway got very close to the reference values in Frequencies
  • Hi Victor,
    Temperature dependent properties in thermal analysis is veru usefull but as we know the temperature changes stress/shock resistance of the material as well. This could be useful. Composites get brittle within quite low amplitude of temperatures which has got strong effect on the matterial behaviour. I Z88 Aurora and Creo SImulate you have to choose which solver/option to use if you deal with glass, concrete, steel, rubber... How would that be possible in Mecway to make thermal analysis with thermal stress depending on the gradient of the temperature and material behaviour (steel structure in huge fire).

    Thanks
  • Thanks for the suggestion. I can't promise temperature dependent mechanical properties though. For now you would have to do it manually using components to define regions of uniform properties.
  • Victor,

    Can we add element to element friction-less surface contacts? Similar to your compression only or friction-less support but instead of normal load transfer to support, normal transfer to another element support.

    This can be very beneficial for analysis that has gap between elements and much more. For example, a cantilever beam that has a column below the free end. Once the beam is loaded and deflect to a certain point due to its stiffness (stressed to a certain point), the bottom surface of the beam come in contact with the top surface of the column. Then the cantilever beam becomes a fixed/simple support beam for the delta loading after coming in contact with the column while the stress in the cantilever beam has already captured.

    A more advance application is user-defined coefficient of friction for the contact surfaces. If zero is specified for the two tangential directions, then the contact become a friction-less contact as mentioned above.
  • Thanks for the suggestion. This is a widely requested feature, so I'm aiming to get it into Mecway in a near future version as long as there are no major obstacles.
  • Victor, for non-linear static analysis, only geometric nonlinearity is included in version 2. Would material nonlinearity be a new feature in the near future?

    It also seems there is no option between following/follower load and nonfollowing load for the current version. If yes, would this feature be available in the near future?

    Thanks in advance!
  • Material non-linearity is certainly on the radar but it's not a very popular feature so may be a while. However version 3 will have Abaqus/Calculix file support allowing you to use the Calculix solver which has plastic and hyperelastic materials.

    Can you describe a use case for controlling follower loads? In version 2, loads are in principal all follower loads except that some of them aren't subject to change because they're fully defined by the input. For example:
    - Pressure is a follower load because the orientation of the surface might change.
    - Force on faces is somewhat a follower load because the relative sizes of the faces it's applied to might change. However the total force is constant because you specify that explicitly.
    - Force on nodes is not a follower load because the number of nodes cannot change during the solution.
    - Gravity is a follower load because the distribution of the mass can change. But again the total force is constant because the total mass is constant.
  • VMHVMH
    edited September 2014
    Victor,

    Thanks for the prompt response. Please see attachment. I'll use applied force on nodes so the force remains in the same direction.

    Can Mecway output the amount of strain for at a given node?

    Thank you
  • In this case, force on either faces or nodes will be OK because the total magnitude and direction is determined by what you specify and won't change during the solution.

    There's an undocumented way to output strains but it won't be accurate for large deformations so not useful here. Email me if it's still of interest.
  • VMHVMH
    edited September 2014
    Thanks Victor,

    Just another comment on another feature that I think will be very beneficial for engineers who work with existing structures: I have encountered a few cases where client asked to perform a FEA on an existing loaded girder to column connection (fully welded together). FEA showed that the connection had some local material yielding under gravity and lateral load therefore reinforcement was required.

    Due to the fact that the connection was already preloaded with dead load (without live and wind), any reinforcement added will not take the preloaded dead load unless the connection was fully unloaded which was not possible. Therefore a two steps loading FEA was performed: dead load without reinforcement and live load plus wind load with reinforcement. The solutions were then combined to check the stresses in the existing material and the new material. This gave us a better understanding of the effectiveness of the reinforcement and the stresses in the existing material.

    The analysis was performed using Abaqus using the same model with the reinforcement element modeled. The first step ran the analysis with dead load only with the reinforcement elements turned off (I don't know the details on this). The second step ran the live load plus wind load with the reinforcement elements turned back on.

    Thanks
  • Mecway isn't well suited to the kind of multi-step analysis because it doesn't have the concept of arbitrary steps. If it can do this in future, it'll probably be manually controlled - do one solve, then another, then combine the results.
  • Victor, that will be great too. As long as we can combine the solution to see the final results, that meets the purpose. Thank you and am looking forward for future upgrades!
  • I have a couple suggestions.

    add the other orthogonal views xz, yx, etc

    this one might be more difficult. on the imported cad define certain edges to have a set number of nodes before meshing. I attached a picture
  • edited October 2014
    Hello tcarson

    All 6 orthogonal views are available by left or right clicking on the triad arrowheads. If it ends up upside-down, use alt+left/right arrows to rotate it.

    Regarding the edge refinement. A less convenient way is to add a string of local (spherical) refinements along the location of the edge. Just make each refinement's radius at least as big as half their spacing. Are you really looking to refine along the edge, or just use the edge as a convenient way to locate something else?
  • Thanks for the triad tip that helps a lot for screen shots.

    Regarding edge refinement. I like to manually control how many elements around a hole or some other feature?

    Thanks
    Timm
  • Thanks for the suggestion about edge refinement. I'll keep it in mind for possible future updates, but can't promise it. For now there's only the spherical local refinement, so it might take several of them to cover a linear region. Also, you can refine all curves in the object together using "Min. number of elements per curve" in the meshing parameters.
  • Hi,

    This is a VERY nice FEM code. Its very useful for me. I have previous experience with ANSYS, Catia, NISA, MARC and Nastran, 25 years in all.

    I would like the following in the prioritised order:
    -Shear force pr unit length or shear stress on shell elements in the U and V directions.
    -Definition of element U and V directions in a local cartesian, cylindrical or spherical coordinate system.
    -Definition of boundary conditions and loads in a local cartesian, cylindrical or spherical coordinate system.
    -Definition of failure criteria for different materials.
    -Plotting of margin of safety.
    -Plotting of failure criteria, max stress, Tsai-Wu and others.
    -Contact elements.
  • Hello Sture. Thanks for the suggestions. I've added all those to my own list.

    There is an undocumented way to obtain shear force per unit length for shell elements. However, the results are sometimes misleading and need to be interpreted appropriately. This is why it's undocumented. I'll get back to you about this.

    I see the usefulness of cylindrical coordinates, but can I ask what kinds of applications you have in mind for spherical coordinates?

    You can currently get some of the effects of cylindrical and spherical element local axis coordinates by generating the elements with Mesh tools-> Rotate/copy. That rotates the coordinates along with the elements.
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!