July 18, 2006

• The explicit method is widely used for impact analyses. However, the time increment depends strongly on the mesh, therefore there are disadvantages that the mesh has to be taken uniformly as far as possible, and the time increment should be considerably small. Although we cannot take the time increment larger than the required time resolution of the problem, if the performance of the implicit method is sufficiently high, there is possibility that the implicit method overcome this disadvantage of the explicit method.

  Fig.1 Analysis model

  Here, we evaluate the drop impact analysis of ADVENTURECluster, comparing both an experiment and analysis result of LS-DYNA. We referred to the paper bellow for the experiment and analysis result of LS-DYNA.

  • T. Omori, H. Inoue, N. Kawamura, M. Mukai, K. Kishimoto, T. Kawakami, Evaluation of Drop Impact Load for Portable Electronic Components, 7th International Conference on Electronics Materials and Packaging, pp.262-267, 2005

  We owe the results here to Toshiba Corporation and Tokyo Institute of Technology.

  Since LS-DYNA uses the explicit method, the comparison will be much interested.

• A PMMA rod with a hemispherical end is dropped from 10 cm above and hits the top of the SUS rod with a flat end at 0, 30 and 60 degree angles. The dimension of the SUS rod is as follows:
  • Diameter: 10mm
  • Length: 900mm
  • Young's modulus: 193GPa
  • Poisson ratio: 0.31
  • Mass density: 8030kg/m³

  The dimension of PMMA rod is as follows:

  • Diameter: 10mm
  • Length: 200mm
  • Curvature of the hemispherical end: 5mm
  • Young's modulus: 5.7GPa
  • Poisson ratio: 0.3
  • Mass density: 1190 kg/m³

• 1/4 symmetricity has been assumed with the inclined angle 0 degree, and 1/2 symmetricity has been assumed with the inclined angle 30 and 60 degree. The element is linear hexahedral (some meshes are degenerated). The number of nodes and DOFs are 270,000 and DOFs is 810,000, respectively with the 0 degree case; The number of nodes and DOFs are 520,000 and 1,560,000 with 30 and 60 degree cases. As to the contact condition, the upper end of SUS rod is set to be the master surface, and the hemispherical end of PMMA rod is set to be the slave surface. The friction factors are assumed to be 0 (complete sliding), 0.2 and 0.4. In the paper noted above, 0.2 is concluded by the experiment. The selective reduced integration for volumetric strain for the linear hexahedral elements in ADVENTURECluster.

• The soft-constraint option has been chosen with the friction model of LS-DYNA rather than the penalty option. Geometric nonlinearity is assumed in ADVENTURECluster. The automatic time increment option has been set in the ADVENTURECluster calculation.

• The analysis results with the inclined angle 0, 30 and 60 degree are shown in Fig.2, Fig.3 and Fig.4, respectively. The vertical axes are contact reaction force, that is, the summation of the products of the contact pressure and the surface area.

• In the case with the inclined angle 0 degree, since no force appeared in the direction of sliding, the results of the three cases for the friction factor 0 (complete sliding), 0.2 and 0.4 have to coincide, as the three results of LS-DYNA, which can be seen in Fig.2. The analysis by ADVENTURECluster has been done only for the complete sliding. The result of LS-DYNA and ADVENTURECluster are almost equal, though a small difference at about 0.19ms is seen, which will be viewed as induced by the difference of the time increments. As to comparison of the experiment and analyses, the both can be concluded to coincide well; the same comment is given in the above paper.

  In the case with the inclined angle 30 degree, the results of LS-DYNA and ADVENTURECluster are almost identical with the friction factor 0 case. However, with the friction factor 0.2, the result of ADVENTURECluster is a little different from that of LS-DYNA, a little closer to the experiment (this cannot necessarily conclude that the friction analysis of ADVENTURECluster is superior to LS-DYNA). With the friction factor 0.4, according to the constraint of the sliding, larger reaction forces appeared both in LS-DYNA and ADVENTURECluster than the experiment and the results with the lower friction factors.

  In the case with the inclined angle 60 degree, since sliding appeared with the three friction factors, the reaction forces are smaller, and close results to the experiment have been obtained. The results of LS-DYNA and ADVENTURECluster almost identical with the friction factor 0 case, which is the same as in 30 degree case. From this problem setting, one could assume by intuition that the larger is the friction factor, the larger is the reaction force, but the result of LS-DYNA gave the opposite result that the larger is the friction factor, the smaller is the reaction force (this cannot necessarily conclude that the friction analysis of ADVENTURECluster is superior to LS-DYNA).

• Although we still have problems of the validity of the friction model or dependency of mesh precision on the analysis, the results here could be said to have shown the propriety of the drop impact analysis of ADVENTURECluster.

  The calculation of LS-DYNA was performed on a UNIX machine (IBM Power4 1.3GHz 1CPU). The calculation of 1/2 model (inclined angle 30 and 60 degree) was performed; simulation time was 0.6 milliseconds and the initial time increment were taken to be 2.10^-5milliseconds (2.10^-8s); the calculation time took about 35 hour. On the other hand, the calculation of ADVENTURECluster was performed on 8 node-machine of Pentium4 3.0GHz; the simulation time is 0.3 milliseconds, the total time steps obtained were 60 to 120, and the calculation time was about 4 hours. The number of time steps taken in the calculation of LS-DYNA is not known, but the total time step taken in the calculation of LS-DYNA is supposed to be much larger than that of ADVENTURECluster.

Fig. 2 Results with inclined angle 0 degree

Fig. 3 Results with inclined angle 30 degree

Fig. 4 Results with inclined angle 40 degree

if you have any questions.