Appendix E: Comparison with Adams-Marc Co-simulation
In Adams 2014, the
Adams Co-Simulation Interface (ACSI) was introduced. The ACSI capability to perform a co-simulation between Adams (multibody dynamics) and Marc (implicit nonlinear finite element analysis) can be applied to similar engineering problems as the Adams nonlinear flexible body. Important characteristics of each solution in a number of categories are summarized below to highlight the similarities and differences between them:
■General Pre-Processing
■Nonlinear Flexible Body - The body is based on a valid MSC Nastran implicit nonlinear (SOL400) data deck. While MSC Nastran SOL400 is based, in-part, on Marc technology it is a separate and distinct implicit nonlinear finite element solver.
■Adams-Marc Co-simulation - The FEA model is a Marc dataset. A configuration file needs to be prepared to specify connections and data flow between the Marc model and the Adams model.
■General Post-Processing
■Nonlinear Flexible Body - Adams PostProcessor provides integrated post-processing in this solution including visualization (animation) and basic stress/strain plotting.
■Adams-Marc Co-simulation - With this solution there is no support for integrated post-processing in Adams. Each code generates its own results sets which must be reviewed either independently by Adams PostProcessor and Mentat or together via 3rd party tools (CEI Ensight).
■Solution Process
■Nonlinear Flexible Body - In this solution, all equations (Adams and FEA) are solved by Adams. It is an exact continuous closed-form approach with common single error control from Adams. There is no distributed memory parallel (DMP) processing support. There is shared memory parallel (SMP) support.
■Adams-Marc Co-simulation - Because this solution is a co-simulation each code (Adams and Marc) solves independently. The co-simulation communicates at the zero-order or, at best, 1st order. Marc does support DMP processing. Finding optimal run settings (results convergence vs. performance balance) is likely more difficult compared to the Adams nonlinear flexible body since there are more settings to adjust (error control in two solvers plus a communication interval)
■Features and Limitations
■Nonlinear Flexible Body - A few important points to consider here for this solution:
♦Solution time may be very large and difficult to predict
♦Direct contact with nonlinear flexible body unsupported
♦FE modeling techniques which require Nastran iterations to establish proper initial conditions (for example, glued contacts with penetration in FEM design position) are not supported in nonlinear flexible body deifintions
♦Users can locate Adams' Markers on the FEA model
♦Deployment options allow solving nonlinear flex bodies on separate machines or "fire and forget" cluster scenario
■Adams-Marc Co-simulation - A few important points to consider here for this solution:
♦Solution time may be estimated by user
♦Limitation in Adams model: no discrete GSE allowed
♦Limitations in Marc model: connection to Adams only through rigid surfaces
♦There are some Marc capabilities/elements not included in SOL400 (for example, certain elements for tire modeling among others).
♦A "glue" code controls the co-simulation. Adams and Marc installations can run on different machines.