Finite Element Discretisation |
 
|
Finite Element Discretisation |
|
Ideally, the aim is to create a mesh which is sufficiently dense to accurately capture structural behaviour, while not being unduly complex and resulting in lengthy simulation times. For example, if defining a steel catenary riser, you would typically require a relatively fine mesh in the touchdown region, while a more coarse mesh would suffice for the portion of riser lying flat on the seabed, or the portion extending upwards through the water column.
Intersections between different sections within the model, model boundaries (e.g. vessel connection) and the wave zone are generally regions of interest also and therefore should have a more refined mesh. In order to avoid over-meshing, you should use longer elements in the middle of long sections of continuous properties. You should also attempt to prevent large changes in relative element length across the finite element mesh by gradually stepping up and down element lengths along the structure. A suggested maximum aspect ratio between the lengths of adjacent elements would be approximately 1.5. These guidelines are particularly important if you are building a model manually using Nodes and Elements and Cables, but this approach has been largely superseded by the new Lines modelling feature. Since lines provide automatic mesh generation, meshing at intersections and boundaries, and meshing aspect ratios, are automatically adhered to when you use lines.
If you are modelling contact, it is important that the finite element mesh in sections of the model where contact is likely to occur is sufficiently fine to properly model contact. For example, in order to properly model contact between a riser and a guide surface, it is desirable that at least two nodes of the riser are in contact with the guide surface when contact occurs.
If your model is likely to experience significant compressive loads, you will need to use a relatively fine mesh in order to ensure that the critical Euler buckling load is not exceeded - refer to Compression and Buckling for further details.
It is generally recommended that you should endeavour to match the spatial and temporal discretisations. Specifically, if you are running a time domain dynamic analysis, the select time step (or range of steps in the case of a variable step analysis) should be broadly consistent with the finite element mesh density. In order words, if you have a relatively refined model, then you should be using relatively fine time steps. Conversely, if you have a relatively coarse model, then larger time steps will be more suitable.
Avoid the specification of zero length beam elements. The presence of any such elements will immediately cause a static indeterminacy, and the source of the indeterminacy may be difficult for you diagnose.
•*LINES is used to define a line (and sections within that line), by specifying relevant set names, lengths, start and end locations, and mesh generation settings.
•*GUIDE is used to define guide (contact) surfaces.
•*TIME is used to define time parameters for an analysis.