FE Model Requirements

You can use any finite element (FE) model of a component as the foundation for an Adams Flex flexible body. The following describes aspects of an FE model that you should consider when you transfer it for use in Adams. It starts by listing the vendors that support MNF translation.

FEA Programs and MNF Translation

The following table list the FEA vendors and their available Modal Neutral File (MNF) functionality.
FEA Vendors and MNF Translation
 
 
MSC Nastran
MSC Marc
ANSYS
ABAQUS
I-DEAS
PERMAS
CMS
Interface nodes
Orthogonalization
 
Modal Loads
Residual Vectors
 
 
 
Preloads
Stress Modes
Generalized damping
 
 
 
 
 

1 Solid elements only. PERMAS is a European finite element code.

Node Number Limitation

There is no limit on the number of nodes in the FE model. The number that you select, however, can affect data storage, transfer requirements, and the rendering performance of the graphics hardware. For example, an FE model with more than 20,000 nodes taxes the capabilities of video subsystems during animation and requires approximately 1 Mb of disk storage for each modal degree of freedom (DOF) you retain from the finite element analysis.
You should note, however, that during the Dynamic simulation:
Adams uses a modal description of the flexible body.
Computational speed is completely independent of the number of finite element nodes in the body.

Attachment Points

When you build a flexible body into an Adams model, you interface with the body using a variety of attachments, either joints or forces. In Adams Flex, you can model the variable boundary conditions at attachment points, which are nodes that have been idealized for attachment, by preserving all six Cartesian degrees of freedom (DOFs) of those points as you export the flexible body from your finite element analysis (FEA) program. An attachment point is equivalent to a superelement exterior grid point.
The FEA of the component is usually performed without detailed information about external constraints. These boundary conditions are often an unknown function of time. If the Adams analysis could only accommodate normal modes (eigenvectors), the number of modes that would be required to model the effect of attachments would be dramatically increased. With prior knowledge of nodes at which joint or force elements are applied to the body, Adams can achieve a high-fidelity solution with a minimal number of modes.
Attachment points are not without drawbacks, however. Each attachment point normally contributes six modal DOF. Corresponding to each attachment point DOF is a constraint mode, which is a static mode shape due to a unit displacement of that DOF while holding all other DOFs of all attachment points fixed. A large number of attachment points can result in unwieldy data files and can significantly impact CPU time, if the associated modes are enabled during an Adams dynamic simulation. You should note that you can always apply joints and forces to any node without it having been identified as an attachment point during the FEA.
The ability to capture structural interactions at attachment points is accomplished by applying a component mode synthesis (CMS) method similar to that proposed by Craig and Bampton ("Coupling of Substructures for Dynamics Analyses", Craig, R.R. and Bampton, M.C.C., 1968 AIAA Journal, Vol.6, No.7, pp.1313-1319). The Craig-Bampton modes are the constraint modes mentioned earlier, plus a set of fixed interface normal modes, the eigenvectors of the system, while holding all attachment point DOFs fixed. As one of the substructruing techniques, Craig-Bampton method is primarily developed for efficient model assembly and accurate coordinate reduction, however when used in Multibody system simulation, it provides one kind of modal basis that is equivalent to free-free modes if no reference condition is applied before carrying out the Craig-Bampton analysis.
Even though, the Craig-Bampton method is introduced here, Adams Flex does not limit the modal basis it uses. The modes can be obtained from any CMS such as reference condition, normal modes, Craig-Bampton and Craig-Chang etc. For a detailed technical description of the CMS modes, please see the Theory of Flexible Bodies.

Mode Selection

The rigid body modes (if exist) should be disabled since that will cause coordinates redundancy when integrated with multibody system. Any one of the other modes can be enabled or disabled during the dynamic simulation. You should disable a mode if it does not contribute to the response of the flexible component during a simulation. You should not, however, base the inclusion of a mode solely on its resonant frequency and the frequency of the excitation. For example, a static load can cause the deformation of the component in such a way that can only be captured by the shape that corresponds to a particular high frequency-mode. You should only disable modes because of unacceptable computational overhead or when you are certain that a particular mode does not contribute to overall response. You should never disable modes based on an arbitrary notion of a frequency range of interest. Because it is the mode shapes, not the associated frequencies that determine the solution accuracy.

Units in MNFs

When you generate a Modal Neutral File (MNF), you must specify which units were used in the FEA program. This units information is stored in the MNF. You should note that the finite element analysis does not need to have been performed in the same set of units as subsequent Adams simulations. As long as the MNF is labeled with the proper units, Adams can convert data to correctly represent the flexible body in the Adams model.
Some finite element programs do not accommodate inconsistent sets of units (for example, millimeter, Newton, kilogram, and seconds). You can still perform your finite element analysis in these units but you should be aware of the strange behavior that can occur, such as frequency being reported in units that are not Hz. If the data in the MNF is labeled with the correct units, Adams handles the data correctly.

Constraints and Rigid Body Modes

You should avoid using constraints in the FE model. Only in rare cases are constraints appropriate. For example, a constraint is necessary when a particular node on the component is fixed on ground or on a rigid body in the Adams model. In these cases, the displacement field of each deformable body that forms the multibody system must be uniquely defined by means of the set of reference condition ("Reference Conditions and Substructuring Techniques in Flexible Multibody System Dynamics", O’Shea et al., 2018, Journal of Computational and Nonlinear Dynamics, Vol.13, 041007). Proper constraints must be added in the FE model, before imposing any nonlinear set of algebraic constraints associated with the mechanical joints. An improperly constrained finite element model can seriously misrepresent the component. Consider, for example, an FE model in which two separate nodes are completely constrained. These two nodes never move relative to each other during the Adams simulation because none of the mode shapes feature any relative motion between the two nodes.
When you read modal information for an unconstrained component into Adams, you must ensure that all rigid body modes are disabled. Adams adds six nonlinear rigid body DOF to each flexible body and numerical singularities arise if the rigid body mode shapes are also included. Adams Flex attempts to disable rigid body modes by default, but Adams View can fail to detect some rigid body modes because rigid body modes sometimes have nonzero frequencies caused by numerical inaccuracies.