Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


Posts posted by Hyperman

  1. Hi,


    The engine file is not included (again). Use Engine File Assistant to easily define control cards. Engine file can be merged with starter on export: check the "Merge starter and engine file" option so the engine control cards defined will be in the starter file (_0000.rad). Open the _0000.rad file in Textpad or Notepad and ensure that engine control cards are present. 



    It is not obvious how the model is not plausible- do you have experimental tests or previous experience? I can only think of the axial crushing might be too stiff. A geometric imperfection can be introduced with a /RANDOM nodal noise.


    There are intersections and penetrations in the contact interface which should be corrected. There are also 10 bad quality elements (negative volume).

  2. Hi,


    if the geometry changes significantly it is possible the crash dynamic also varies from the original in:

    • mass, affecting the kinetic energy 
    • stiffness, because of different load paths/crushing behavior


    It is difficult to estimate how much submodel affects the global model. This is why the reanalysis of the optimized global model is recommended.


    RD-E: 5400 Cut Methodology Conclusion (from Radioss help):



    The cut methodology allows you to reduce the computation time by running smaller models once a first full model has been run. Its main limitation lies from the fact that the interactions between the selected parts (for sub-model) and the unselected parts must be small enough so that the behavior of the unselected parts is not significantly affected by unselected parts. Therefore, the interactions between the unselected parts and the selected parts (for sub-model) could be well defined by the section (no extra contact for instance).

    To assess those hypothesis, it is also important to compare the section force (or moment) between full model and sub-model.

    In industrial applications, like in dummy analysis, it is advised to keep only the dummy and its direct environment as sub-model. Then about 90% of the CPU cost could be saved compared to the full car model. In low speed impact or beginning of impact with high speed, it is advised to keep only the deformed area as sub-model. Then about 30% of the CPU cost could be saved compared to the full car model.



  3. Hi,


    The long calculation issue is not due to the unloading curve; the time step will drop due to the element compressive deformation. This happens when elements are very deformed and their characteristic length goes to 0. In the case of large strain formulation, the time step of an element goes to 0 when the element is compressed. In a mathematical way, the element can not reverse its orientation since its stiffness increases to an infinite value; but due to numerical accuracy, the element may go to reverse its orientation.
    In order to solve the problem of both the drop in cycle time step and subsequent termination of the run due to a negative volume, you might first check that the material used is well-suited to the physics which is represented. Then switch the elements to a small strain formulation. This is done as follows:
    In the RADIOSS Starter input file (Runname_0000.rad), use Ismstr =2 in the solid property or in the option /DEF_SOLID; in RADIOSS Engine file (Runname_0001.rad) use the option /DT/BRICK/CST which will set the time step value symbol_tritmin at which the solid elements will switch to small strain.
    This means that the solid elements using Ismstr =2 will use large strain formulation while their time step remains greater than symbol_tritmin, and will then switch to small strain formulation. Their volume will then remain constant and the element can even reverse its orientation. The drop of their time step normally stops
    except for some materials, especially viscous materials.


    Check the attached video for more information (relevant from 6:00)


    V_h_Solid Elements in RADIOSS - Altair University

  4. Hi,


    Hyperview can be used to post-processing animations files. Hyperview can be used to plot time history files. 
    Energy balance curves are output by default, only need to define TFILE output frequency.

    energy balance.jpg

    Check the following video (relevant from 1:38:30):


    2016 Directed Learning Webinar Series - Crash Analysis with RADIOSS - Altair University

    TP005_editT01 TP005_edit_0001.rad TP005_edit_0000.rad

    JKL likes this

  5. Hi,


    Radioss solver is capable of highly nonlinear transient dynamic and quasi-static analyses. To answer your general question would require an extensive reply. Therefore I suggest you first check the free Radioss ebook and some videos:

    Introduction to Radioss 


    Radioss for Non-linear Crash using Altair HW SE 









    Impact Attenuator Crash Analysis for SUPRA SAEINDIA 2017 - 3D Meshing & Analysis using HW SE 14.0


    Impact AOC Webinar: Introduction to Radioss for Explicit Nonlinear


    Impact AOC Webinar: Nonlinear Analysis using Radioss by HW SE 14.0



    Hypermesh Crash Simulation


    HyperWorks 2019 - Radioss - Steel Ball and Plate Impact Simulation 


    BAJA SAEINDIA Online Training-Dynamic Crash Analysis using Altair RADIOSS


    AOC India 2019 RADIOSS



    Crash simulation


    2016 Directed Learning Webinar Series – Crash Analysis with RADIOSS


    SUPRA SAEINDIA 2015 - Webinar Series - Day - 5


    SAEINDIA BAJA 2016 Online Training Session Day 3 


    AOC2015: Linear / Non-Linear Contest Models Solution recorded webinar


    Dynamic Crash Analysis of Roll cage



    Radioss for Dynamic Analysis



    Dynamic Crash Analysis using Radioss




  6. Hi,


    Tperiod is period to be damped (less than or equal to the highest period of the system). Please refer to RD-E: 1600 Dummy Positioning about how to use various damping options. 


    /ADYREL Dynamic relaxation with auto-defined adaptive damping is supported since 2019.1 version.

  7. Glad to help.


    Error ID 44 is a general error and can be due to many modeling issues. Share the model for debugging if possible. 


    With total tensile failure defined, the element will be deleted when the specified strain is reached. To output damage evolution use /ANIM/SHELL/DAMA which generates animation files containing a damage value for each layer of a shell element. The damage value is the maximum of damage over time of all failure criteria associated with the material.


    You may also find this presentation useful (especially p18-19):


  8. Glad to help.

    1. The load collector 'displacement_spc' constrains the moving plate in all degrees of freedom except z translation. Without this constraint the plate could rotate and/or displace in unwanted directions.

    2. The master segments are both plates. Usually, the stiffer component is chosen as the master. In case both components use the same material, the coarser mesh is chosen as the master.

    3. I haven't used this function yet.

    4. Yes, Type 7, 11, 19 need to have initial clearance (non-zero gap, Physical gap). 

  9. Hi,


    non-linear penalty-based contact interfaces require a non-zero initial gap. Type 24 linear contact interface can be used with a zero gap. Type 19 is computationally inefficient and since there is no edge to edge issue in your model a type 7 will be more efficient. 

    A rigid body was put on plates so they will not affect the timestep. The model and boundary conditions are symmetric so a quarter symmetry could be used for efficiency (this will also properly constrain the cylinder in the symmetry planes- currently it is held in place only by friction).


  10. The model you shared had several issues:

    • no card image and material parameters were assigned to the "rubber" material 
    • rigid elements should be placed in a separate component
    • the pressure load collector did not have pressure versus time function referenced (The cause of error id 120)
    • only one linear (first order) tetrahedral element through thickness is not recommended (too stiff in bending). In order to calculate the bending stresses accurately in thin-walled structures, a minimum of 3 solid elements through the thickness should be used.

    You can take advantage of geometry and boundary condition symmetry by modeling only a quarter of the model. 


    Use Quality> ModelChecker in HyperCrash or Tools>model checker>radiossblock in Hypermesh


    In order to get different buckling modes and Buckling load factor a geometric imperfection should be introduced.  Either use a /RANDOM nodal noise or the procedure described in thread:
  • Create New...