Jump to content

Simon Križnik

  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Simon Križnik

  1. Hi Nagaraj, there are a lot of optimization examples in the following Youtube channels: Altair University Altair India Student Contest ELEATION By Apoorv Bapat
  2. Hi Atul Yadav, it is the most convenient to set up boundary conditions using BCs Manager by clicking Tools > BCs Manager Attached is the RD-T: 3500 Tensile Test Setup (from Radioss help). RD-T_ 3500 Tensile Test Setup.pdf
  3. Hi Charles, most of the cards were generated from the Engine file assistant (Utility>engine file assistant). The tool generates typical output requests, such as stress, strain, velocity, etc. by checking generate default output requests. ANIM keywords generate data for element, vector and nodal animation output requests. Please check the radioss help for more information on output requests. ENG_RUN defines simulation duration. ANIM_DT defines the frequency of writing the animation files (A-files). ENG_TFILE defines the frequency of writing the time history file (T-file). ENG_DT_NODA is the mass scaling which I replaced with Advanced Mass Scaling (ENG_DT_AMS & AMS keywords).
  4. Hi Izzdham Shah, what you have described is the correct way to set up symmetry boundary condition. However, symmetry plane for topology optimization refers to a concept in optimization called Pattern Grouping where optimizier is forced to produce symmetric design even if loads and BCs are asymmetric.
  5. Hi Charles, in the model you shared the interface between cars is set to type1 Fluid/Structure Interaction (ALE/Lagrange Interface). So I defined type 7 general purpose interface instead and redifined self contact only on segments that actually experience contact for computational efficiency. Output requests should be set up- the most convenient way is by Utility>engine file assistant I also applied Advanced Mass Scaling to speed up the computation. Results as well as model files are shared bellow: crashtest.hm crashtest_0001.rad crashtest_0000.rad
  6. please refer to https://connect.altair.com/CP/SA/training/self_paced/optimization/content/opt6/pdf/Ex4a.pdf
  7. Hi dajeet, assuming the units are ton,mm,s,N,MPa it seems that the beams are too slender (0,5mm radius) for a cantilever with 125N total force with and a moment arm of 1000mm. So despite being made of stiff material (carbon fiber) this structure would deform excessively IMHO . I suggest running size optimization to determine proper dimensions.
  8. Hi fredriki, from Practical Aspects of Structural Optimization - A study Guide
  9. Hi r_ferriani, in the model you have shared there are only components, materials and properties. I have arbitrarily created the boundary conditions and loads in the file attached bellow. Is tis the result you are looking for? Also, the model will not run as is due to bad tetra mesh quality. You can disable element check with Analysis>control cards>PARAM>CHECKEL>NO, but this could lead to inaccurate results, so I suggest remeshing with hexahedral elements. 23-08 prova 5_edit.hm
  10. Thanks Prakash, I will try a different modeling approach for a better correlation.
  11. For even greater control over mesh element distribution you can 2D automesh the surfaces before using one volume/multi volume solid map, which then uses the 2D mesh as a reference. Care should be taken to ensure that the 2D mesh is mappable in 3D. It can be helpful to preliminarily solid map the component just to see in what order and directions HM would map.
  12. Hi, I want to perform a modal analysis of an inflatable tube. The analysis is split in two steps- inflation step and subsequent modal step. The pretension is carried over into modal through STATSUB (PRELOAD) from inflation step. Inertia relief is applied through PARAM>inrel -2 However the results are not matching the experiments sufficiently. Can the internal overpressure be directly included in modal analysis instead of indirectly through preloading?
  13. Hi AgataK, MASS parameter defines Mass per spring element or M/l0 depending on flag Ileng In case of linear stiffness you can use K parameter (again subject to Ileng flag). If nonlinear stiffness is required please use func_ID1 referenecing a load vs engineering strain curve. Definition from Radioss help is attached bellow _PROP_TYPE4 (SPRING).pdf
  14. Hi, there are so many modeling errors I suggest you rebuild it just for practice. Be careful when assigning element types (the 2D elements should be QUAD and edges should be plot elements). Use 1d>config edit and elem types panel to edit and set default elements respectively. Also, I am able to change by card edit from /line/seg to /surf/se in HM so it looks like this issue is solved in newer releases. Attached is tthe edited model which is able to run. Note that the concentrated load was not assigned to nodes so I picked those arbitrarily. machine axisymm.hm machine axisymm_0000.rad machine axisymm_0001.rad
  15. Hi, the video was posted in a similar topic, however it is in .swf format. The video demonstrates the steps I have alrady explained.
  16. Hi, if you want to filter the noise, then please refer to HG-1030: Referencing and Filtering Curves
  17. Hi SamuelIshak, the procedure to create type 5 contact interface can be found in the video from 0:30 to 1:10 type 5 contact In short: -create ^edges components by using find edges command on the tools>edges panel -create contact segments by selecting edge plot elements in the analysis>contactsurfs with card image set to SurfSeg -create contact type 5 in analysis>interfaces then add contactsurfs to master and nodes to slave
  18. By isolate I mean you should only have the elements to be connected visible by using mask panel. Alignment nodes are used to identify congruent features on both meshes, because currently Hypermesh can not automatically recognize mesh pattern to be connected. If for example there are two identical meshes but one is rotated then you should help HM map 1D elements between by specifing some reference nodes. There is analogy to pattern repetition optimization in Optistruct, wherby local coordinate systems are defined for each group of elements. You probably should edit/delete the link
  19. Simon Križnik


    it was exported from user guide which comes with above mentioned Hyperworks version. You have correctly observed that some energy was dissipated and the energy balance in the .out file is negative. There are two possible causes: -hourglass energy -contact energy, especially with friction Attached is documentation on energy balance; also check the same topic in Crash Analysis With RADIOSS- A Study Guide Plot both energies to see if they account for the difference (they are not included in total energy). Results Checking.pdf
  20. Hi Burner2k, this happens to me too sometimes, when it fails to connect. It helps to isolate only the elements/nodes to be connected and pick the alignment nodes away from the boundary. The meshes do not have to match exactly, as long as the elements are equal in type (tria, quad) and distribution (quantity in different directions). In other words, nodes can be perturbed. Alignment nodes can be seen as a frame of reference between meshes. You are right about density.
  21. Simon Križnik


    attached is documantation for type 2 interface (version is 2017.2.3) INTER_TYPE2.pdf
  22. Hi, this "^tempCompSurfSegment" could be a renamed "^faces" component (created by tool>faces panel) from the original file. If Radioss solver gives an interface error upon removal of such component then probably some of the elements in such component were directly (by element) or indirectly (by contact surface) referenced in some interface. The same applies to "^tempCompLineSegment" renamed from "^edges" component (created by tool>edges panel).
  23. Simon Križnik


    You could also use spotflag=27 or 28. I don't think there is a way to maintain the timestep with spotflag=25 as it is a penalty formulation- this means a spring is inserted which could lead to timestep drops.
  24. Hi, in the attached file the material model is duplicated verbatim (imported solver deck copied form Radioss help), but the elements still get deleted from the start. Therefore I suspect this material model is not functional with stress-strain curve input. Please verifiy at your end. law58_kg_mm_ms.hm
  25. Simon Križnik


    Using the Gapmin value of 1.5 will not produce contact from the start of simulation because there is no initial penetration at this distance. Try to increase this value and check the resulting penetrations with tool > penetration check. Igap=1000 is a constant gap equal to Gapmin. It is more computationally efficient, but does no take into account the thickness of parts in contact Igap=1,2,3 are variable gap which is computed for each impact as the sum of the master element gap (gm) and the slave node gap (gs). If Igap=1, variable gap is computed as: max[Gapmin,(gs+gm)]max[Gapmin,(gs+gm)] If Igap=2, variable gap is computed as: max{Gapmin,min[Fscalegap⋅(gs+gm),Gapmax]}max{Gapmin,min[Fscalegap⋅(gs+gm),Gapmax]} If Igap=3, variable gap is computed as: max{Gapmin,min[Fscalegap⋅(gs+gm),%mesh_size⋅(gs_l+gm_l),Gapmax]}max{Gapmin,min[Fscalegap⋅(gs+gm),%mesh_size⋅(gs_l+gm_l),Gapmax]} The problem with type 7 interface is it can't handle edge to edge contacts due to Edge to Edge Impact Locking (refer to User Guide). This is a possible reason why the computation crashes and it can be corrected with type 11 interface.
  • Create New...