Additional modules for forming simulation

Simufact Forming Die Analysis

This module enables you to simulate die strains and stresses in order to predict the stress distribution in the forming dies. This allows for a much more detailed view of the die’s behavior than is possible when only using Simufact Forming Hub. For every die, whatever the level of complexity, whether they are pre-stressed dies or die reinforcements with an unlimited number of components, this module can offer a greater insight into the die's behavior. Knowledge of stress distribution, especially the location and amplitude of tension maxima within the dies, is necessary to plan effective counter-measures at the time of design in order to extend the die life, analyze die failures and to evaluate alternative concepts.

Reinforcements (for example, through shrinkage assembly, even multi-tiered ones) can of course be simulated and optimized in regard to their position and overlap.

The die stresses can either be simulated coupled directly with the forming simulation, or can be simulated in a decoupled analysis as a separate step during the post-processing of the underlying forming simulation, for the moments of greatest die loads. The latter procedure is especially user-friendly and can easily be integrated into the workflow because of its short computation time. First, the forming process is simulated and the material flow is optimized. Once a good process design has been achieved, a decoupled die analysis re-using the already conducted forming simulation is used for a review of the dies loads and die stresses. If needed simulation can also be used to optimize the die concept in regard to the optimum die life.

Simufact Forming Parallel Core

This application module offers the parallelization of simulations by using the effective Domain Decomposition Method (DDM) for the FE-Solver, and Shared Memory Parallelization (SMP) for the FE-Solver and the FV-Solver. Due to these methods computation time is dramatically reduced.

Through parallelization by DDM, the simulated model is disassembled automatically into different domains that are linked by additional numerical boundary conditions. For every domain a distinct stiffness matrix, which contains all elements, nodes and boundary conditions is generated. The displacements and forces are then calculated for every time step when solving the stiffness matrix.

When compared to non-parallelized simulations, this method allows for several smaller stiffness matrices to replace one big stiffness matrix. The computational effort required, due to the increasing size of the stiffness matrix, increases more than linearly in a FEM simulation. DDM generates smaller stiffness matrices, which therefore allow for the computational effort of each domain to be reduced disproportionately, ultimately resulting in an increase in computational speed.

Additional speed-up is also gained by using different cores or processors within the work station or a cluster to solve the stiffness matrices of the single domains simultaneously and independently from each other.

Remeshing operations are achieved by joining the domains, remeshing the mesh and then re-disassembling the model into new separate domains. DDM parallelization can be combined with shared memory parallelization. The domain computation is then further distributed to additional cores / processors for selected computing operations related to the stiffness matrix assembly and the equation solving. Shared memory parallelization can be used without DDM parallelization, too.

Tips:

• The number of domains that can be utilized for parallelization depends on the number of licensed Simufact.forming Parallel Core modules. For example, if a simulation should be parallelized on twelve cores, then eleven Simufact.forming Parallel Cores must be acquired in addition to Simufact.forming Hub.
• DDM cannot be used for FV simulations since the computational effort of FV simulations increases linearly with the number of nodes / elements. Shared memory parallelization is the most effective method for FV simulations.