Summary Chapter 13

Training Manual March 15, 2001 Inventory # Chapter Objectives In this chapter, a review will be given of the basic concepts learned in this course. 1. Describe what makes the difference between implicit and explicit 2. Review of Practical guidelines a. Modeling b. Materials c. Contact d. Loading e. General

Training Manual March 15, 2001 Inventory # RESULTS SOLUTION PROCESS “black box” SOLUTION PROCESS “black box” INPUT DATA Demands What Makes the Difference ? This slide is about the different aspects of explicit finite element programs compared to implicit finite element programs. The main request of any user is to get a solution with a minimum of effort. What will be the expenditure to reach the required results?

Training Manual March 15, 2001 Inventory # Computational efforts with explicit programs –Model size –Critical time step Element edge length Sonic wave velocity: –Young’s modulus –Density –Termination time What Makes the Difference: Computational Efforts Computational efforts with implicit programs –Model size –Degree of Nonlinearity –Number of time steps

Training Manual March 15, 2001 Inventory # Estimation of the Computation Time with ANSYS/LS-DYNA with T CPU = total CPU usage in CPU-sec k= system factor SGI PowerIndigo CPU-sec/element SGI Crimson 100 MHz CPU-sec/element HP CPU-sec/element N elem = number of elements t= simulation timee.g. 160 msec c= sonic velocitye.g mm/msec in steel l min = smallest length of elemente.g. 9 mm Deviations in both directions depend on material laws, element types, and contacts..

Training Manual March 15, 2001 Inventory # System Request for the FE-Analysis with Different Solvers Example: Square plate meshed with SHELL63 vs. SHELL163 Table 1 : Model size for system comparison

Training Manual March 15, 2001 Inventory # (continued) System Request for the FE-Analysis with Different Solvers

Training Manual March 15, 2001 Inventory # What Makes the Difference: User Efforts User efforts with implicit programs –Try to find convergence criteria for the solution. –SOLVE –If the calculation has not converged, repeat the first task until “solution is done”. User efforts with explicit programs –Tune up the model to reach solution as fast as possible. Avoid small elements Use rigid bodies Mass scaling Use higher velocities –Only dynamic analysis is possible; if a static solution is asked, try to load as fast as possible, but without overshoot. –In metal forming simulation, the punch velocity can be higher than in real life without lack of accuracy. –Check the results to validate the kinetic energy.

Training Manual March 15, 2001 Inventory # Practical Guidelines - Modeling Avoid small elements whenever possible, as they will significantly reduce the time step size. If small elements are required, use mass scaling to increase the critical time step. Do not use triangular/tetrahedron/prism elements whenever possible. Although these elements are supported, they are not recommended. For best results, use cube-shape bricks. If the overall hourglass energy is more than 5% of the internal (strain) energy, use some form of hourglass control in the model. The energies can be monitored in the GLSTAT and MATSUM files. Alternatively, fully integrated elements can be used to combat hourglassing problems. However, these elements can give poor results in cases involving large deformation and/or bending. Use rigid bodies in any part of a model where the deformation results are not important. Rigid bodies save significant amounts of CPU time.

Training Manual March 15, 2001 Inventory # Practical Guidelines - Materials Make sure to use consistent units when defining your material properties. Incorrect units will not only determine the material response, but will also effect the model’s contact stiffness. Make sure that the material data being used in a model is accurate. The precision of most nonlinear dynamic problems hinges upon the quality of the material data input. Spend extra time and effort to obtain accurate material data. Select the best possible material model for a given application. If it is not certain whether a part’s physical response should include a particular characteristic (e.g., strain rate effects), it is always best to define a material model that includes all possible features.

Training Manual March 15, 2001 Inventory # Practical Guidelines - Contact Initial penetrations between contact surfaces are not allowed. Make sure that the model is defined without any overlapping surfaces where contact is defined. Always use realistic material property and shell thickness values. The material properties and geometry of contacting surfaces are used to determine the penalty stiffness. Do not make multiple contact definitions between the same parts. Use automatic contact for shell elements unless contact forces are required. Use automatic general contact (AG) whenever possible. It is the easiest type of contact to define and doesn’t cost much in CPU time. List the defined contact surfaces prior to solution to ensure that contact has been properly defined.

Training Manual March 15, 2001 Inventory # Practical Guidelines - Loading Avoid single point loads; they are known to excite hourglass modes. Since one excited element transfers the mode to its neighbors, point loads should not be applied, if possible. After a load curve has been defined, use the EDLDPLOT command to plot it in order to ensure its accuracy. Because LS-DYNA may ‘overshoot’ a solution by a few microseconds, it is often useful to extend loads for longer time period than the final solution (termination) time. For ‘quasi-static’ problems, it is often advantageous to apply a higher velocity than that in the real application. This can significantly shorten a problem’s computation time. Constraints on the nodes of a rigid body are not allowed. All constraints must be applied to the center of mass of the rigid body (via the EDMP,RIGID command).

Training Manual March 15, 2001 Inventory # Practical Guidelines - General Use mass scaling in most analyses to increase the critical time step and reduce the solution time. Check the LS-DYNA output window to ensure that the percentage of mass increase is small. Always output all energy data. Make sure that hourglass, damping, and interface energies are calculated. This information is useful for monitoring solution accuracy and debugging problems. Use consistent units for material properties, lengths, and time. If units are incorrect, a model will typically diverge. Use LS-DYNA control switches to monitor the solution. Periodically use sense switch sw2 to check the progress of a run. If the model begins to diverge, use sw1 to terminate the analysis. Learn how to use LS-TAURUS. Use the EDOPT command to obtain d3plot and d3thdt files. LS-TAURUS has better support for certain types of postprocessing (e.g., failed elements).

Training Manual March 15, 2001 Inventory # * 1 slug = kg; 1 ft = m = 12*2.54 cm; 1 N = 10 5 dyne = 1 lbf/4.4482; 1 Mbar = dynes/cm 2 ; 1 bar = 14.7 psi = 1.0 atm = 105 Pa; 1 kg/m 3 = gm/cm 3 = slug/ft 3 Consistent Units