RD-E: 0903 Interfaces Study

Comparison of results obtained using different interfaces.

Options and Keywords Used

To reduce the dynamic effect, dynamic relaxation can be used (/DYREL in the Engine file). A diagonal damping matrix proportional to the mass matrix is introduced into the dynamic equation:(1) [ M ] { u ¨ } + [ C ] { u ˙ } + [ K ] { u } = F MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbwvMCKf MBHbqefqvATv2CG4uz3bIuV1wyUbqedmvETj2BSbqefm0B1jxALjhi ov2DaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaacaGacmGadaWaaiqacaabaiaafaaake aadaWadaqaaiaad2eaaiaawUfacaGLDbaadaGadaqaaiqadwhagaWa aaGaay5Eaiaaw2haaiabgUcaRmaadmaabaGaam4qaaGaay5waiaaw2 faamaacmaabaGabmyDayaacaaacaGL7bGaayzFaaGaey4kaSYaamWa aeaacaWGlbaacaGLBbGaayzxaaWaaiWaaeaacaWG1baacaGL7bGaay zFaaGaeyypa0JaamOraaaa@4E40@ (2) [ C ] = 2 β T [ M ] MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaWaamWaaeaaca WGdbaacaGLBbGaayzxaaGaeyypa0ZaaSaaaeaacaaIYaGaeqOSdiga baGaamivaaaadaWadaqaaiaad2eaaiaawUfacaGLDbaaaaa@3FC0@
With,
β
The relaxation value by default, equal to 1
T MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbwvMCKf MBHbqefqvATv2CG4uz3bIuV1wyUbqedmvETj2BSbqefm0B1jxALjhi ov2DaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaacaGacmGadaWaaiqacaabaiaafaaake aacaWGubaaaa@39B0@
The period to be damped (less than or equal to the largest period of the system)
Thus, a viscous stress tensor is added to the stress tensor:(3) σ ij viscous =( λ+2μ ) ε ˙ kk δ ij +2μ ε ˙ ij MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaeq4Wdm3aa0 baaSqaaiaadMgacaWGQbaabaGaamODaiaadMgacaWGZbGaam4yaiaa d+gacaWG1bGaam4Caaaakiabg2da9maabmaabaGaeq4UdWMaey4kaS IaaGOmaiabeY7aTbGaayjkaiaawMcaaiqbew7aLzaacaWaaSbaaSqa aiaadUgacaWGRbaabeaakiabes7aKnaaBaaaleaacaWGPbGaamOAaa qabaGccqGHRaWkcaaIYaGaeqiVd0MafqyTduMbaiaadaWgaaWcbaGa amyAaiaadQgaaeqaaaaa@569E@
In an explicit code, the application of the dashpot force modifies the velocity equation:
V t + Δ t / 2 = V t Δ t / 2 + γ t Δ t MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaamOvamaaBa aaleaacaWG0bGaey4kaSIaeuiLdqKaamiDaiaac+cacaaIYaaabeaa kiabg2da9iaadAfadaWgaaWcbaGaamiDaiabgkHiTiabfs5aejaads hacaGGVaGaaGOmaaqabaGccqGHRaWkcqaHZoWzdaWgaaWcbaGaamiD aaqabaGccqqHuoarcaWG0baaaa@4A92@
without relaxation
V t + Δ t / 2 = ( 1 2 ω ) V t Δ t / 2 + ( 1 ω ) γ t Δ t MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaamOvamaaBa aaleaacaWG0bGaey4kaSIaeuiLdqKaamiDaiaac+cacaaIYaaabeaa kiabg2da9maabmaabaGaaGymaiabgkHiTiaaikdacqaHjpWDaiaawI cacaGLPaaacaWGwbWaaSbaaSqaaiaadshacqGHsislcqqHuoarcaWG 0bGaai4laiaaikdaaeqaaOGaey4kaSYaaeWaaeaacaaIXaGaeyOeI0 IaeqyYdChacaGLOaGaayzkaaGaeq4SdC2aaSbaaSqaaiaadshaaeqa aOGaeuiLdqKaamiDaaaa@554A@
with relaxation

with ω = β Δ t t MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbwvMCKf MBHbqefqvATv2CG4uz3bIuV1wyUbqedmvETj2BSbqefm0B1jxALjhi ov2DaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaacaGacmGadaWaaiqacaabaiaafaaake aacqaHjpWDcqGH9aqpcqaHYoGydaWcaaqaaiabfs5aejaadshaaeaa caWG0baaaaaa@40B3@

This option is activated in the Engine file (*_0001.rad) using /DYREL.
Inputs
β
= 1
T MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbwvMCKf MBHbqefqvATv2CG4uz3bIuV1wyUbqedmvETj2BSbqefm0B1jxALjhi ov2DaebbnrfifHhDYfgasaacH8srps0lbbf9q8WrFfeuY=Hhbbf9v8 qqaqFr0xc9pk0xbba9q8WqFfea0=yr0RYxir=Jbba9q8aq0=yq=He9 q8qqQ8frFve9Fve9Ff0dmeaacaGacmGadaWaaiqacaabaiaafaaake aacaWGubaaaa@39B0@
= 0.2

The dynamic problem (impact between balls) is considered in a second run managed by the second Engine file (*_0002.rad) with a time running from 30 ms to 130 ms.

Input Files

The input files used in this example include:
Inter_7_Penalty
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_7_Penalty/TEST7P*
Inter_7_Lagrangian
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_7_Lagrangian/TEST7L*
Inter_16_tied
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_16_tied/TEST16T*
Inter_16_sliding
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_16_sliding/TEST16S*
Inter_17_tied
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_17_tied/TEST17ST*
Inter_17_sliding
<install_directory>/hwsolvers/demos/radioss/example/09_Billiards/Contact_modelling/Inter_17_sliding/TEST17S*

Model Description

The balls and the table have the same properties as previously defined. The dimensions of the table are 900 mm x 450 mm x 25 mm and the balls' diameter is 50.8 mm.

Six interfaces are used to model the contacts (ball/ball and balls/table):
Interfaces Used in the Problems
TYPE16 (Lagrange Multipliers) tied or sliding
Secondary nodes/main solids contact
TYPE17 (Lagrange Multipliers) tied or sliding
Secondary 16-node shells/main 16-node shells contact
TYPE7 (Lagrange Multipliers)
Secondary nodes/main surface contact
TYPE7 (Penalty) sliding
Secondary nodes/main surface contact

Model Method

The TYPE16 interface defines contact between a group of nodes (secondary) and a curved surface of quadratic elements (main part). The TYPE17 interface is used for modeling a surface-to-surface contact. For both interfaces, the Lagrange Multipliers method is used to apply the contact conditions; gaps are not required. Contact between the balls and the table is set as tied or sliding. Contact between the balls themselves is always considered as sliding. The TYPE7 interface enables the simulation of the most general contact types occurring between a main surface and a set of secondary nodes. The Coulomb friction between surfaces is not modeled here (sliding contact) and the gap is fixed at 0.1 mm. The other parameters are set to default values.

The TYPE7 interface with the Penalty method is not available with 16-node thick shell elements. Thus, brick elements replace the 16-nodes shells in this case (check in the input file).

Contact modeling between balls (always sliding).

rad_ex_fig_9-19
Figure 1. Definition of Secondary and Main Sides for Contact
The symmetrical interface definition is not recommended when using the Lagrange Multipliers method (TYPE16, TYPE17 and TYPE7-Lag). The problem using the interface with the Penalty method uses two interfaces to model the symmetrical impact.

rad_ex_fig_9-20
Figure 2. Symmetrical Configuration of the TYPE7 Interface using the Penalty Method
Interface
Secondary (red) and Main (blue) Objects
TYPE16 - tied
Secondary: nodes
Main: solids (16-node shell)
TYPE16 - sliding
Secondary: nodes
Main: solids (16-node shell)
TYPE17 - tied
Secondary: 16-node shell
Main: 16-node shell
TYPE17 - sliding
Secondary: 16-node shell
Main: 16-node shell
TYPE7 - Lagrange Multipliers
Secondary: nodes
Main: surface (segments)
TYPE7 - Penalty method
Secondary: nodes
Main: surface (segments)
Contact between the balls and the table (sliding or tied depending on the problem):

rad_ex_fig_9-21
Figure 3. Definition of Secondary and Main Objects for Balls/Table Contacts
Interface
Secondary (red) and Main (blue) Objects
TYPE16 - tied
Secondary: nodes
Main: solids (16-node shell)
TYPE16 - sliding
Secondary: nodes
Main: solids (16-node shell)
TYPE17 - tied
Secondary: 16-node shell
Main: 16-node shell
TYPE17 - sliding
Secondary: 16-node shell
Main: 16-node shell
TYPE7 - Lagrange Multipliers
Secondary: nodes
Main: surface (segments)
TYPE7 - Penalty method
Secondary: nodes
Main: surface (segments)

Pre-loading: quasi-static gravity loading to reach static equilibrium.

The explicit time integration scheme starts with nodal acceleration computation. It is efficient for the simulation of dynamic loadings. However, a quasi-static simulation via a dynamic resolution method needs to minimize the dynamic effects for converging towards static equilibrium and describes the pre-loading case before the dynamic analysis. Thus, the quasi-static solution of gravity loading on the model shows a steady state in the transient response.

Results

Kinetic Energy Transmission between Balls during Collision

clip0755
ex_9_type17_interface_zoom35 Figure 4. TYPE17 Interface. Contact between quadratic surfaces Balls/table contact: tied / Ball/ball contact: sliding
ex_9_type17_interface2_zoom37 Figure 5. TYPE17 Interface. Contact between quadratic surfaces Balls/table contact: sliding / Ball/ball contact: sliding

ex_9_type16_interface_zoom40
Figure 6. TYPE16 Interface. Contact nodes/quadratic surface Balls/table contact: tied / Ball/ball contact: sliding
ex_9_type16_interface2_zoom36 Figure 7. TYPE16 Interface. Contact nodes/quadratic surface Balls/table contact: sliding / Ball/ball contact: sliding
ex_9_type7_interface_zoom38 Figure 8. TYPE7 Interface: Lagrange Multipliers Method. Contact nodes/linear surface (sliding contact)
ex_9_type7_interface2_zoom42 Figure 9. TYPE7 Interface: Penalty Method. Contact nodes/linear surface Balls/table contact: sliding / Ball/ball contact: sliding

Conclusion

  Interface 16 Tied Interface 16 Sliding Interface 17

Tied

Interface 17 Sliding Interface 7

Lagrange Multipliers

Interface 7

Penalty

Cycles 241392 241385 241387 241385 241385 773099
Error on Energy -30.8% -1.4% -55.5% -10.8% -1.2% -46.1%
Rolling yes no yes no no no
Momentum Transmission partial quasi-perfect partial good good partial
Quadratic surface main side main side main and secondary sides main and secondary sides no no

A non-elastic collision appears using the TYPE7 interface Penalty method. After impact, each ball has about half of the initial velocity. The momentum transmission is partial and can be improved by increasing the stiffness of the interface despite the hourglass energy and degradation of the energy assessment.

Error on energy is more noticeable for interfaces using the Tied option, due to taking into account the rolling simulation.

This study shows the high sensitivity of the numerical algorithms for the modeling impact on elastic balls. Regarding the interface type, the kinematics of the problem and the transmission of momentum are more or less satisfactory. TYPE16 interface allows good results to be obtained.