Simple & Standard Contact Modelling
The simplest option for modelling rollerbox contact is a single-point elastic contact scheme. With this approach, the support from each rollerbox is considered to act at a single node of the pipe model only, and the node at which the support acts remains the same for a given rollerbox (that is, if the pipe moves axially relative to the rollerbox, the support is still applied at the same location on pipe – in other words, the support point moves with the pipe). Friction effects are also ignored.
Recent work undertaken indicates that this type of model is very robust and has little or no implication for runtimes when compared to a model that, for example, uses fixed boundary conditions to model pipeline/stinger contact (that is, a model where intermittent contact is not considered). This approach is extremely robust and permits analyses to be run with timesteps of the order of 0.1 seconds, thereby permitting analyses to be completed very quickly.
It should be noted that, although this approach considers only contact at a single point along the pipeline (for each rollerbox), it is nevertheless fully capable of modelling contact with rollerboxes with multiple surfaces in the cross-section (for example, U-shaped, V-shaped or O-shaped rollerboxes).
A further advantage of this approach is that a relatively coarse FE mesh can be used as the support point does not move relative to the pipeline. This further increases the speed of analyses carried out with this scheme.
This approach is an extension of the single-point elastic contact algorithm, and provides an option with intermediate complexity and realism. The approach is basically similar to the single-point elastic contact algorithm with the exception that the support point is permitted to move along the pipeline – effectively moving from one node to the next. This allows relative axial motion between the pipe and the rollerbox to be considered.
Friction effects are also accounted for, and these are included by means of a non-linear spring with a force-deflection relationship such as that shown in Figure TN.5.3 below. Here, the approach to modelling friction is closely based on the seabed friction algorithm, whereby friction forces are generated by the spring, rather than by the application of boundary conditions.
An important point here is that the friction forces are computed on a contact surface-by-contact surface basis, rather than on a node-by-node basis. This scheme is more appropriate for contact with relatively short surfaces, whereas the alternative node-by-node approach is more suitable for seabed contact where relatively large sections of pipe are in contact with seabed.
Figure TN.5.3: Force-Deflection Characteristic for Non-Linear Spring used to Model Friction
This approach requires a finer FE mesh density than the “single point elastic contact” approach described above. This is because closer node spacing is required to avoid significant changes in configuration occurring when the support point moves from one node to an adjacent node.