Jump to content

Simon Križnik

Members
  • Content Count

    1090
  • Joined

  • Last visited

  • Days Won

    99

Everything posted by Simon Križnik

  1. Inertial effects are due to acceleration in dynamic events: we must consider mass (inertia), i.e. sum of all forces = m*a
  2. @Samorit -solid elements (hexa, penta, tetra, pyramid) do not have rotational DOF (only translational DOF) -use law 38 which works- I do not know why material law 70 does not -grnodnode can be used to reference a set of nodes in for example BC, loads, contacts, etc. -check the unit consistency again- Units: Kg, MPa, sec, mm are not a set of consistent units. You can still use this unit system, but you have to set up analysis>control cards>HeaderCard according to the input and output units required and let the solver convert units appropriately -you can define output requests to plot force vs. displacement in HG and do model calibration in Hyperstudy 3. Versuch_edit4.hm You can give back and show appreciation by hitting the like button on comments you find helpful
  3. The reason why it is not running the analysis for creating flexible bodies could be that there are already .h3D flexbody files created in the previous run.
  4. Thanks for liking my help and subscribing to my content Success at last, the simulation is perfect in terms of energy (Check the T01 files in HG). Material law 38 offers more parameters to characterize ABS. Advanced Mass Scaling is used to speed up computation. For additional speedup when running type: -nthread (number of cores) -sp (single precision) Second and tonne units are assumed so interpret accordingly. 3. Versuch_edit1.hm
  5. Hi, if you are new to FEA I recommend you go through the following free ebooks: https://altairuniversity.com/free-ebooks-2/free-ebook-practical-aspects-of-finite-element-simulation-a-study-guide/ https://altairuniversity.com/free-ebooks-2/free-ebook-crash-analysis-with-radioss-a-study-guide/ Refer to the following Youtube channels for practical demonstrations: Altair University Altair India Student Contest ELEATION By Apoorv Bapat Here is tensile test tutorial: Refer to RD-T: 3500 Tensile Test Setup and RD-E: 1100 Tensile Test for Radioss tensile test. OS-E: 0110 Elastic-Plastic Large Displacement Analysis is Optistruct tensile test. I have made some modifications to your model: -define runtime, animation and plot output frequency, output requests. All can be created simultaneously in Radioss user profile by utility>engine file assistant -remesh with hexahedral elements which are more computationally efficient and accurate than tetrahedral mesh -simplified the model exploiting two plane symmetry in boundary condition, loads and geometry The model does run, but there is a lot of hourglassing that results in huge energy error. Using fully integrated element formulation or under integrated with physical stabilization (Isolid=14,17,24) results in elements switching to small strain and run failure. Currently, I do not know how to solve this problem. What unit system is used in this model? My guess is MPa, kg, but I am not sure about time units. The loading stress-strain curve is referencing the same as unloading but should be different due to hysteresis. 3. Versuch_edit.hm
  6. Hi, force vs displacement curve can be done using non-linear quasi-static analysis (NLSTAT) in Optistruct. There are a few examples: Radioss should be used if there are inertial effects due to high loading rate.
  7. Hi, please go through Hyperform videos: https://www.google.si/search?q=hyperform&tbm=vid&ei=Bzv4W5WdGOeDrwTV0o6gBg&start=0&sa=N&ved=0ahUKEwjVvNaVh-veAhXnwYsKHVWpA2Q4ChDy0wMIXA&biw=1689&bih=845&dpr=1.1 Although TWBs is not demonstrated you can get an overview of how forming simulations are done in general. Attached is tutorial HF-1040: Laser Weld Manufacturing_Solutions_Tutorials.pdf 1Step_laser_weld.hf 1Step_Laser_weld_FLC_curve.csv It might be that some guy named Taylor welded sheets, but they are more commonly known as tailor welded blanks
  8. Hi, models can be exported as solver decks. File>Export>Solver deck or using the icon in the picture below
  9. Tabulated means function can be referenced to describe material behavior. Material laws 38 and 70 provides this functionality. About law 38 from Radioss help: So create the engineering stress versus engineering strain function in a tabulated format using XYplots>Curve Editor then reference this curve in the fct_IDiL field of material law and set the Itota=1. In compression, the nominal stress-strain curves for different strain rates are defined by you. Up to 5 curves may be input. The curves represent nominal stresses versus engineering strains. You can find some examples of law 70 in the following topic:
  10. Hi, you can find details about various viscoelastic material models in Radioss help (see files attached). Tabulated Strain Rate Dependent Law for Viscoelastic Materials (LAW38) could be suitable for your application. Viscous Materials.pdf Tabulated Strain Rate Dependent Law for Viscoelastic Materials (LAW38).pdf _MAT_LAW38 (VISC_TAB).pdf
  11. Hi, sorry for the late reply- I was hoping someone with spring-back expertise would respond. Anyway, here are my suggestions: 1. In Example 24 – Spring-back it is noted in table 1 that dynamic relaxation using explicit also during spring-back improves correlation with experimental results. 2. Looking at the comparison it can be observed that the experimental specimen is not as flat (or is slightly rotated upwards) in the left portion (where the sheet is constrained). Modeling appropriate boundary condition in this region would improve the correlation (there should be some examples in literature) To reduce dynamic effects three approaches are available in RADIOSS: Dynamic Relaxation (/DYREL), Energy Discrete Relaxation (/KEREL) and Rayleigh Damping (/DAMP). Refer to Example Guide for application examples. I suggest using Rayleigh Damping in this case.
  12. I found the problem: the composite material density you defined was not consistent with the model unit (mm, newton, second, tonne)- the mass of b-pillars was 1300 tons Now you owe me some likes taurus_forum2_0000.rad taurus_forum2_0001.rad
  13. The run is failing because of deep penetrations therefore node deactivation in type 7 interface. I have tried with symmetric contact and type 11 line contact without success. The issue is probably due to weak b-pillar shattering. I will get back to this model later, because it frustrated me too much
  14. Connecting parts of crash-box assembly with mesh connectivity instead of using type 2 or 7 contacts would greatly reduce the computation time. Traditional mass scaling /DT/NODA/CST) can be used. AMS can also be used for crash simulations, but it should be compared against CST run for validity. AMS is strictly not recommended only for high-speed events like blast and ballistics. You can find elements/nodes/interfaces contributing to low timestep during simulation in the engine out file (runname_001.rad) or in the starter out file (runname_000.rad) search for NODAL TIME STEP (estimation). Element check can also be performed in HM using tools>check elem>time (or F10) where minimum timestep criteria can be defined and failing elements highlighted. Mesh should be finer where more deformation is expected (close to impactor).
  15. Hi, the solver deck should be in: <install_directory>/demos/hwsolvers/optistruct/examples/snap-fit.fem snap-fit.fem
  16. Hi, a couple of basic modeling errors: -pole (rigid wall) was referencing node group for frontal crash -initial velocity was moving the car away from the pole -there was a self-impacting type 7 contact interface on the whole car, which is problematic as there are many unresolved intersections and penetrations. So I defined self-impact contact only on components near the impact area and removed penetrations and intersections using tools>penetration check in HM (in Hypercrash use quality>check all solver contact interfaces). It is computationally efficient to include only parts of the model where contact influences deformation mode. The main problem was a negative shell and numerical damping in a composite property. Also, the composite b-pillar is too weak so it is rupturing immediately after impact, which is also causing trouble in type 7 contact. With proper b-pillar structural integrity the contact interface would probably work. taurus_forum2_0000.rad taurus_forum2_0001.rad
  17. Hi, please refer to Radioss help (help>hyperworks help home>radioss or F1). Tstop is the simulation end time. Tfreq is frequency to write either animation or history plot file.
  18. Hi, a negative value should reverse the rotational direction. Maybe there is some other modeling issue. Try reversing the joints or markers on which motion is defined (flip the axis used for rotation).
  19. Hi @Farper I have modified only one model (attached below): -SPC should be referenced in load-step -nodes which are given the enforced displacement (SPCD) must also be fixed in same DOF using SPC -changed the SPCD value to -50 so compressive displacement is applied Compression_edit.hm
  20. Hi @guardianzm local coordinate system is called marker in MV. Please refer to attached document or MV help for further details. I would also recommend a free ebook: Practical Aspects of Multi-Body Simulation with HyperWorks MotionView User's Guide.pdf
  21. I do not remember all the modifications I made, but it will be a good practice for you to find and correct mistakes- the error is probably due to initial penetration in type 7 interface or due to nodes not found in type 2 interface. Use the above-mentioned (model checker, out files) methods to debug models. It is an important skill to master so you will be able to correct the models yourself. Composite orientation and the drape estimator tools are used to define element orientation. Each element has its own orientation and if anisotropic materials are used (like UniDirectional or woven composites) it is essential to establish proper element orientation. Note that angle defined for plies are in relation to element orientation i.e. plies are rotated by that angle.
  22. As suggested by @Prakash Pagadala, use NLSTAT (quasi static simulation). you have excessive deformation in you analysis, so Loads and BCs are not appropriate To get accurate loads and boundary conditions full scale crash simulations should be performed first and then loads extracted using Equivalent Static Load Method.
  23. @Amasker Plasticity describes non-linear material behaviour (stress-strain curve is non-linear) where the material deforms permanently. This type of problem can not be solved in linear static analysis. Radioss explicit solver should be used for crashworthiness, beacause it is a dynamic event with material, contact and geometric non-linearity.
  24. Hi, review and debug models using tools>Modelchecker>Radiossblock and starter out file. select user profile 2017, because for an unknown reason 2018 is not supported. There are problems with contacts- type 7 and 2 require initial gap so some parts ware offset, and some other parameters were defined. I suggest connecting parts of crash-box assembly with mesh connectivity (matching) instead of using type 2 or 7 contacts. There were also issues with composite orientation. here is the manual option: and here is Drape Estimator: Drape Estimator - Altair University You should also reduce the model, as my computer would run this simulation for the next 2 months nonstop- imagine the desperation IA_3.0-edit.hm
  25. Yes, we control the number of modes with either NMODES or UB-FREQ. NMODES explicitly defines the number of modes, while UB_FREQ defines frequency upper bound, i.e. all modes below the specified frequency will be included (note that we usually do not know in advance the number of nodes below certain frequency).
×
×
  • Create New...