Jump to content

Simon Križnik

  • Content Count

  • Joined

  • Last visited

  • Days Won


Posts posted by Simon Križnik

  1. You need to define local skew and Def_orth=1 only for those plies that have an invalid orientation assigned by /STACK (in your case Vx=1, Vy=0, Vz=0 means only plies whose normal points in global x-direction has to be corrected).


    All elements with PCOMPP prop_ID type must belong to a ply (/PLY) inside a stack (/STACK). 


  2. The first anisotropic direction of material could be defined with angle ϕ and global vector (Vx ,Vy ,Vz ). It is also possible to use angle ϕ and skew. In this case, x-axis of skew replaced the global vector V.


    The orthotropy direction is set with Vx, Vy and Vz parameters in /STACK card (laminate). To override laminate direction set up ply with local skew and Def_orth=1. 


  3. Hi,


    when assigning material orientation, make sure the direction is not perpendicular to shell elements. You have probably assigned material orientation in the height direction for the whole component, which is invalid for the elements on top (check material orientation of element 1600).  The best practice is to assign material orientation for each surface one by one.

  4. Upon looking into the model again I found the reason why the buckling initiates at the base of upper crashbox. Looking at the cross-section (deformation scaled 5x) it is revealed the clamping region at the base of upper crashbox slightly rotates inwards, which initiates the buckling. The reason for this behavior is the primary load paths are not aligned (crashbox diameters are different) which induces bending moment. shem.gif


  5. Hi,


    the geometry has to be properly partitioned before using automatic  hexa mesher (one volume, multisolids). In case of complex shapes, it is challenging to properly partition the solid in order to be mappable in one or more directions- it is even more challenging to strategically plan how these mappable solids come together to produce a structured mesh. Unfortnately, there is no easy fix for hexa meshing as it is the most difficult type of meshing.


    Refer to the following for details:





  6. Hi,


    there are modeling errors:
    -3 beam elements have collinear nodes (invalid orientation node). Use the 1D>beam>update panel to resolve this issue.
    -remove intersections and penetrations with tools>penetration check
    -the timestep imposed is a bit too high


    The deformation starts at the base of the upper crashbox due to random mesh flow lines that behave as a geometric imperfection.

    2-D quad mesh should be systematic (ruled or mapped), avoid 2-D auto mesh: Flow lines should be maintained with minimum number of trias and diamond or rotating quads should be avoided. Use of the auto mesher on surfaces sometimes results in a zig-zag or random mesh which might lead to unexpected problems later. 
    Better Mesh Flow: For crash or non-linear analysis, systematic mesh flow lines where all the elements satisfy the required quality parameters is very important. Using a mix-mode element type instead of pure quad element type helps to achieve better flow lines and convergence of solution.

    Refer to the page 




  7. Hi,


    it is common practice to create a rigid (RBE2) spider and apply the load to the independent node. 

    In general, it is best not to apply loads and displacements directly to design spaces, as this often leads to incorrect results. Instead, you should split the part into design and non-design spaces, and apply loads and displacements to the non-design spaces.

  8. This error is due to non-convergence (check the out file for details).


    Increase the number of cutbacks allowed (NCUTS) in the load collector with NLADAPT card image and refer this load collector under NLADAPT subcase definition (loadstep).


    However, there is a limit when implicit methods become computationally inefficient (buckling, wrinkling, unconstrained rigid body motion, large deformation, rupture, contact with friction,...). These cases can be solved by the explicit method.


    enriqueng likes this

  9. Glad to help.


    I suggest you go through free Altair e-books and start with  Practical Aspects of Finite Element Simulation.


    You can also learn from learning and Certification program:

    https://certification.altairuniversity.com/  > (Learn Modeling and Visualisation)


    Check the following youtube channels:


    Altair India Student Contest

    ELEATION By Apoorv Bapat


    If you are new to FEA I recommend Finite Element Analysis For Design Engineers by Paul M. Kurowski:


    enriqueng likes this

  10. Hi,


    1. Use large displacement non-linear static (LGDISP NLSTAT) or transient analysis. Only explicit analysis (Radioss) can converge in case of wrinkling.

    2. To get membrane behavior set MID2 and MID3 as BLANK. Membrane has no bending stiffness which can cause convergence difficulties. As a workaround, assign another material of negligible stiffness under MID2 & MID3.

    3. Linear static cannot be used because of geometric non-linearity inherent to inflatable structures.

    4. It is necessary to assign thickness in any type of analysis.

    5. Material MATHE cannot be referenced by properties other than PSOLID/PLSOLID. Therefore it can not be used with shell elements.

    6. For computational efficiency this model could employ quarter symmetry as load, BC and geometry are all symmetric. 


    enriqueng likes this

  11. Hi,


    Emax and EPS_max are not mandatory, if the loading curve is defined.


    From Radioss help:




    When εmaxεmax is reached, Emax is used whatever the curve definition is.

    E0 and Emax used to calculate the current time step. According to current value of strain, Radioss interpolates Young's modulus between E0 and Emax linearly, where E0 is also used to calculate contact stiffness. Radioss automatically modifies E0 if it is less than the initial value according to the input stress/strain curves tangents.
    • If E0 is not specified, use maximal initial slope of all stress strain loading curves as E0.
    • If Emax is not specified (or set default), use Emax as E0. Specified value of Emax should be greater than E0, otherwise also take = E0 as Emax.
    • If εmaxεmax is not specified (or set default), take the strain where, Emax is reached for the first time on one of the loading curves.
    • If both εmaxεmax and Emax are specified, take εmaxεmax where, Emax is reached for the first time on one of the loading curves.



    Giroud likes this

  12. Yes, according to Learn Thermal Analysis with Altair OptiStruct ebook:



    When there is only one TLOADi present in the model setup, it can be directly referred by the DLOAD subcase entry in Heat Transfer (Transient) load step. In case of multiple TLOADi definitions, user needs to create another load collector with card image DLOAD, in which all TLOADi IDs must be referred with proper scaling factors and global scale factor. The DLOAD load collector is then referred in DLOAD subcase entry in load step definition


    Attached is an example from the ebook which I modified to include multiple heat sources (ambient & local).

    Note: The ambient temperature is applied with convection on the outer surface. The local temperature is imposed at one end of the tube. This approach is valid for 3D elements, but not for 1D beams.


  • Create New...