Highlights
Main enhancements relate to TYPE24 contact, SPH and Multi-Domain.
General contact TYPE24 has been enhanced concerning the treatment of initial penetrations; formerly with solids coated with shells, the contact defined both on shells and solids, with penetration larger than half the shell thickness, was causing nodes anchorages on the wrong side. This is no longer the case in 14.0.
SPH: major improvements regarding scalability on high number of cores; performance measured on several tests shows a solution acceleration factor of 10x about.
Multi-Domain compatibility has been extended to SPH and ALE; for FSI problems, due to their time-step and amount of elements differences between the fluid and lagrangian domains, the solution acceleration factor is much higher than in regular crash applications, measurable in something between 6x and 15x, depending on the model. Solution theoretical acceleration factor is also estimated at the Starter level and printed in output file.
| • | Node-to-element contact based on penalty formulation with constant stiffness; this formulation allows large time-steps, as opposed to contact TYPE7, which may severely affect time-steps. |
| • | Initial penetrations treatment (INACTI=5), with possible release of nodes with initial penetrations, and INACTI=1000 to handle tiny initial penetrations. |
| • | Automatic treatment of Solids coated with shells. The contact stiffness is the highest among the shell and the solid; the contact segment position is always determined by the shell. |
| • | Scalability improved in H-MPP mode with high number of cores. |
|
| • | Single input file setup extended to FSI involving ALE and/or SPH: significant speed-up compared to regular Mono-Domain. Speed-up factors from 5x to 15 x with industrial models. |
| • | Single Time History file output. |
| • | /GAUGE card made compatible with ALE and SPH. |
| • | Starter printout of a theoretical speed-up estimate when Multi-Domain is used. |
| • | Starter now checks and errors out if more than 50% nodes are coupled with cross domain contacts (was just a warning in earlier versions). |
|
| • | Failed elements converted into SPH: instead of being deleted, solid elements can be converted into SPH particles. All failure and deletion strategies can be used, including minimum time-step. |
| • | Parameter to reduce tensile instability. |
| • | Heat transfer is enabled between SPH particles and finite elements. |
| • | Thermal Expansion is also available for SPH. |
| • | H-MPP performance improvement: an optimized domain decomposition improves greatly the scalability on high number of cores. |
|
| • | Design Response via Equations for Design Optimization (/DRESP2) and Equation Definition (/DEQATN) entries are supported in RAD-OPT Format. |
| • | Support Free-size Manufacturing Constraints/Size Optimization for composites in RADIOSS-Optimization (RADOPT). |
| • | Support Free-size Optimization for Composites in ANALYSIS=NLGEOM/IMPDYN/EXPDYN subcase. |
| • | Support of 3D Topology Optimization in RAD-OPT Format. |
| • | Support of Velocity response in RAD-OPT and ESL-B2B. |
| • | Handling of “Soft” (Low-density) elements from OS ESLM Topology Optimization in RAD-OPT and ESL-B2B. "Soft" (Low-density) elements from OS Topology Optimization are removed from the next RADIOSS nonlinear analysis. They are assigned to a /DEL entry in the RADIOSS Engine file. These elements are reintroduced at the next ESLM linear static optimization run. |
| • | Handling of failed elements from RADIOS Nonlinear Analysis in RAD-OPT and ESL-B2B. Stiffness of Failed elements from RADIOSS nonlinear analysis run, is not considered when generating the Equivalent Static Loads for the subsequent linear static optimization problem. |
| • | Support more animation options (/ANIM/VECT, /ANIM/BRIC/TENS/STRAIN, /ANIM/SHELL/TENS/STRAIN, /ANIM/Eltyp/EPSP, /ANIM/Eltyp/ENER, /ANIM/Eltyp/HOURG and /ANIM/Eltyp/THIC). |
|
| • | Output parameter value used in which card which field in output file. |
| • | Weld beads, glue lines: /CLUSTER defines a cluster of elements for modeling spotweld/connection lines; failure can be applied to the cluster and/or to the single elements There are also specific cluster outputs for post-processing /ANIM/VECT/CLUST, /TH/CLUST. |
| • | New keyword /INISHE/FAIL and /STATE/SHELL/FAIL: To improve predictivity of metal forming simulation, it is necessary to consider damage, together with stress, strain and thinning. Need damage output for shell in .sta file. (already have damage output for brick). |
| • | New keyword /ANIM/ELEM/SSP (ALE only): To output sound speed. |
| • | New keyword /ANIM/ELEM/SCHLIEREN: To output schlieren image (optical method widely used in CFD field). |
| • | New keyword /ANIM/SHELL/TDEL and /ANIM/BRICK/TDEL: Time of deletion of solid and shell element. |
| • | Outputs global plastic strain (equivalent to total mean plastic strain in all integration points) for beam (with /PROP/TYPE18). |
| • | Animation file is generated when Engine fails with ERROR 166. |
|
| • | AMS compatibility extended to RBE2 and RBE3. |
| • | Added warning if master node of a rigid body is part of the mesh now errors out if running /AMS. |
| • | Compatibility checks extended to /MAT/LAW13: if it is used with AMS, Starter errors out. |
|
| • | Read/write multiple sections into multiple SC01 files to prevent global SC01 file huge sizes resulting from the amount of stored nodal information. Now can be split into multiple independent SC01 files, enhancing usability, range of applications, as well as disk space and data transfer reductions. However, backward compatibility with a global SC01 option gathering multiple sections is maintained. |
|
| • | Compatibility extended to /SENSOR/TYPE11, and 12. |
| • | Local parameters: parameters can be restricted to a SUBMODEL (do not apply to any other part of the model). |
|
| • | Improved scalability on high number of cores for User Material laws. |
| • | Possibility to use free Fortran compilers for Windows (MinGW) and Linux (GCC); warning: performance might not be at the same level as Intel compiler. |
| • | User material laws compatible with solid spotweld property. |
| • | Add User Animation Outputs /FAIL/SNCONNECT. |
| • | Possibility to specify in Starter (or execution) a specific dll. Users may need to use different dll or even no dll at all. |
| • | Add deformation gradient tensor in Users' laws when Ismstr=10. |
|
| • | Implicit nonlinear compatibility extended to /RBE2, /RBE3 and /PROP/SPR_PRE. |
| • | Improve the convergence for implicit using /IMPL/SINIT. |
|
| • | Starter messages have been reformatted to include option titles. |
| • | Warning and Error messages have been compacted to improve readability and avoid huge out files. For example, Incompatible kinematic conditions checks for rigid walls and tied interfaces. |
| • | Warning message is issued in case subset, surface or group are empty (instead of Error). |
| • | Mass and inertia of each part are printed anytime (from the output level Ipri=0). |
| • | Added DOF output for RBE3. |
| - | submodel version is not supported (<= 44). |
| - | only one node is defined for rotation axis in /TRANSFORM/ROT. |
| - | a wrong value (besides 0 or 1) is used for /BCS. |
| - | no defined node in input deck, which leads to NUMNOD=0 in .sta file. |
| - | Engine errors out if invalid engine keywords are used (for example: /DT/NODE/CST is not valid option but difficult to debug: was leading to deletion of all elements). |
| • | Output complete error message in case of RHO0 <= 0.0 in LAW6 for user to better understand the error message. |
| • | Improved the warning message in case a 4-node degenerated triangle shell element. |
|
Contact TYPE24:
| • | Large sliding for SH3N and at corners. |
| • | In case of T-connections and X-connections, possible wrong direction of contact forces. |
| • | Single precision: possible loss of contact. |
| • | Solids coated with shells: possible node anchorage in case of initial intersection. Improve coating shells treatment in /INTER/TYPE24 which may leads to too small stiffness or gap in contact. |
| • | Wrong forces output if shell sliding in Contact TYPE24. |
| • | TYPE24 + INACTI=5 Starter infinite loop. |
| • | No contact treated in type 24 when master surface (shell) is defined by /SURF/SEG. |
| • | Lost contact in int24 due to master surface using triangular shell elements. |
| • | Possible Starter segmentation violation error with int24 in win64 and messages repeated too many times. |
| • | Improvement of computational efficiency of interface type24 if using gravity in implicit simulation. |
| • | Possible lose contact in interface type 24 if run with single precision. |
Multi-Domain:
| • | Contact across domains: Contact segments were not deleted upon element deletion (flags Idel and Idel2 information was not passed between domains). |
| • | Possible Engine segmentation violation, if /ATFILE in 1.rad file exists without /ATH in 0.rad. |
| • | Segmentation violation in Engine phase when /TH/SUBS is present in the Starter file. |
| • | Use of TYPE2 with penalty formulation may lead to very low time-step in the domain with small time-step. |
| • | SPH with Multi-Domain TH combined T01/T02 output plot interruption between restart phases. May affect curve processing (for example a derivative operation). |
| • | /GAUGE for ALE and /GAUGE/SPH compatibility with Multi-Domain. |
| • | An infinite loop in the reading of .sta file at Starter level was fixed. |
| • | SPH/INOUT issues were fixed. |
| • | Starter issue with large ALE models. |
| • | /GAUGE/SPH failing /PARITH/ON issue was fixed. |
SPH:
| • | Possible Starter failure in H-MPP mode with large number of SPH particles. |
| • | Damage tensor (LAW24) output to animation files does not work for SPH. |
| • | Possible segmentation violation if mass is not defined; error message (ID=138) is introduced for an easy localization of the issue. |
| • | Possible Starter segmentation violation if MAT_ID=0 is assigned to SPH. Starter will issue an error if MAT_ID=0 is used. |
| • | Filtering in /GAUGE/SPH/ not working due to bad initialization of gauges which cause F_cut frequency to be equal to 0 in all cases. |
Optimization:
| • | When a /ESLPART entry was defined in RAD-OPT, the RADIOSS Contact definitions (/INTER/ entries) were not considered in the linear static optimization phase. |
| • | Some of the generated equivalent static load subcases had zero compliance. The ESLM Time Step Screening Process in RAD-OPT and ESL-B2B now filters out zero compliance (equivalent static) subcases. |
Output:
| • | Possible engine segmentation violation error with ANIM/BRICK/USRi. |
| • | Output for USER4 and USER5 in LAW58 was eingineer strain not true strain. |
| • | Output Null (with Isolid=17) or wrong output (with Isolid=24) in /ANIM/BRICK/TENS/DAMA for /MAT/LAW24. |
| • | T02 file generated after implicit run followed by explicit run might be empty. |
| • | Possible Engine segmentation violation error when /OUTP/SHEL/USER5 or /OUTP/ELEM/USER5 used. |
| • | Animation files might be corrupted when /ANIM/SHELL/EPSP/ii and /ANIM/SHELL/EPSP/ALL are used in same Engine input. |
| • | TSHELL property reports incorrect plastic strain (extremely large values). |
| • | Incorrect damage initialization with TETRA4. |
| • | The equivalent plastic strain and damage values for LAW72+Ishell=12 is wrong in TH file. |
| • | TH output does not take into account the contact viscous energy from /INTER/TYPE24. |
| • | No Plastic Strain show in contour for solid element using /MAT/LAW44 Cowper Symonds. |
| • | KE and TE from /TH/PART did not give the right results. |
| • | HyperGraph can not read results (from /THPART) in the T01 file. |
| • | Different results from abf output and T01 output. |
| • | Wrong stress results in composite layers. |
| • | Engine error out if /MAT/LAW01 with /STATE/SHELL/AUX/FULL. |
| • | Possible Engine segmentation violation error when using /TH/QUAD. |
| • | Wrong Output values in the abf file, if compare to the T01 Output. |
| • | /TH/BRIC/EPSXX output the incorrect value for part with LAW4. |
| • | Correct Isolid value of /INIBRI in .sta file in case Isolid=24 + TETRA4 used in model. |
| • | .sta file is not compatible with TETRA10. |
| • | Use /INIBRI/ST..._F the initial stress and strain are not correctly mapped on the model. |
| • | Strain tensor output in animation file for LAW25, solids and thick shells is not correct. |
| • | /PROP/TYPE11+LAW25: no strain tensor output in animation file. |
| • | /FVMBAG1 with tetra mesh and internal surfaces. Result of pressure in time history and animation does not match. |
| • | Improve of the spring element output in Starter listing file *_0000.out (aligned the comment and value). |
| • | Possible segmentation error in Engine, if using /OUTP/VECT/VROT for solid. |
| • | Starter error when using /TH/FRAME. |
| • | Wrong Internal Energy in case restart the job after it previously stopped by "STOP" in control file. |
| • | Wrong animation results (DAMA variables and Von-Mises stress) if uses Sol2SPH +/ANIM/BRICK/DAMA. |
| • | /STATE/BRICK/STRESS/FULL is not working for thick shell 16 node and for Brick 20 node. |
| • | Possible segmentation error in Starter due to Initialize deformation with NPT=0 in /INISHE. |
| • | Wrong strain rates (always Zero) in Animation files (using /ANIM/ELEM/EPSD) for material LAW2. |
| • | Inertia info in warning message #542 (rigid body inertia) is wrong (mixed each other). |
| • | Starter blocked, if using /INIBRI/AUX. |
AMS:
| • | Possible instability due to large time-step and solids with full integration (Isolid=12). |
| • | Suppress all Warnings relating to TYPE7 v. TYPE2 interfaces slave nodes incompatibilities when switching /INTER/TYPE2 from kinematic to penalty formulation (Spotflag 25 or 26). |
| • | AMS divergence with m=8 in /MAT/BARLAT3. |
| • | Time step higher than DTMIN for /DT/AMS and elements may cause instability in interface elements. |
Cut-section Method:
| • | Cut Model cannot find SC01 file. |
| • | Restarts were not functioning. |
| • | Checks and auto-correction if some sections refer to a specific root-name and others do not. |
| • | Wrong results (zero section force) if cut plane(defined with /SECT/CIRCLE, /SECT/PARAL or /SECT with frame) is just on a row of nodes. |
Submodels:
| • | Units of the /RWALL parameters are not transformed in a submodel. |
| • | /UNIT is not applied on the submodel. |
| • | /FAIL/CONNECT is not supported in submodel. |
| • | Model crashed due to SUBMODEL, in case of more than two times transformations are requested. |
| • | /MAT/LAW65 is not supported inside submodel. |
| • | Using submodel, M (Maxwell order flag) in LAW42 is ignored. |
| • | Starter errors out if more than 101 //SUBMODEL and /TRANSFORM are defined (out of bound). |
User Subroutines:
| • | Advanced user laws are not working in multithread. |
| • | Wrong user values in animation output if user law has high number of uvar (like number 99). |
| • | Results repeatability (/PARITH/ON) is not ensured for solid with user defined property (igtyp > 29). |
| • | Possible segmentation error in Engine, if interface type 2 with failure + using user dynamic library. |
Implicit:
| • | Possible Engine crash (ISTOP=4) of BCS solver on Windows (temporary files management on Windows). |
| • | Implicit simulation does not end properly on several CPUs. |
| • | /IMPACC and /IMPVEL provide incorrect displacement for both implicit static and implicit dynamic. |
| • | Possible Run freeze for implicit nonlinear with RBE3. |
| • | Possible Engine segmentation violation error due to Linear implicit + /INTER/TYPE24. |
| • | Inconsistent results between multi-proc and mono-proc by implicit nonlinear using quasi-newton. |
| • | Possible segmentation violation error due to big memory usage in case large slave node number of RBE2 used in implicit. |
| • | Possible segmentation violation error in implicit nonlinear due to wrong hierarchy RBE2/RBE3 definition. |
| • | In implicit linear+ explicit in SPMD, computed external work for implicit part incorrectly (doubled). |
| • | Node displacement do not respect imposed displacement value in local skew for nonlinear implicit. |
| • | Wrong displacement result in implicit analysis when /IMPDISP and /BSC (using local skew) used on the same node. |
| • | No convergence in implicit if solid element with orthotropic laws (like LAW28). |
| • | Run in SMP but failed in SPMD shortly after start due to masters of RBE3 are slaves in Rigid body. |
Messages:
| • | Starter does not printout Material ID and material title in *.out message or does not print info for number of airbag element in case automatic meshing of airbags failed. |
| • | Starter reports incompatibility for orthotropic material laws with sandwich property. |
| • | Starter prints out wrong default values for shells numerical damping (dn). |
| • | Starter writes wrong pressure values to 0000.out file. |
| • | Starter does not printout all hardening options for Springs. |
| • | Possible Starter segmentation violation error if user sets incompatible value for Ish3n (Ish3n=24) in /PROP/SHELL. |
| • | Possible Engine segmentation violet error if use /ANIM/BRICK/TENS/STRAIN/010 or ANIM/BRICK/TENS/STRESS/010, which those integration points are not existing. |
| • | Message for brick activation is not correct in case of big identifier number (more than 8 digits). |
| • | Correct error message which with ASCII characters. |
| • | Possible segment violet Starter error due to use "TEMP" in /TH/NODA in non-thermal analysis. |
| • | Remove Engine warning message, if using /UPWM/SUPG + /MAT/LAW51 for /MAT/LAW51 now is compatible with SUPG. |
| • | Get unreadable error message, if define non exist material ID in part. |
| • | Rupture message repeated in Engine output file many times, if use /LAW58 + /FAIL/TENSSTRAIN. |
|
