Installation Stages Tab Best Practice
1.Restart stages allow you to take the solution from a previous stage and apply configuration changes (e.g. cable length, active line length, vessel position) or new loadings (e.g. current, wave) in a follow on analysis. This follow on analysis has memory of previous solution variables, which is important when trying to assess frictional effects. It also allows you to apply changes/loadings incrementally over a series of solution steps and this can be more stable than using non-restart stages which apply all changes/loadings immediately in a single step and are also dependent on the simplified line profile from the Model component. A good example of this is vessel yaw/roll offsets as mentioned in the Model and Vessel Offset components best practice. However, it is worth noting that no criteria analysis is performed during restart stages and also in the case of dynamic restarts the tensioner (if present) supports the pipe (rather than static boundary conditions) and its behaviour is dictated by the characteristic curve/damping inputs specified by on the Tensioner component.
2.The following points represent best practice for time related inputs:
a.Static restarts generally require more than one time increment with the greater the change in configuration/loading the greater the number of required increments. Too little increments will result in convergence difficulties. 10 increments would be viewed as a minimum, but 100s of increments can be required for extreme cases.
b.For dynamic analysis try use fixed time steps as they make it easier to understand how long the dynamic analysis is going to take. The size of the time step is dependent on the level of support contact. If there is significant intermittent support contact (e.g. during S-lay) then the time step should be between 0.025 to 0.05 seconds. If there is no support contact then a larger step (0.1s to 0.2s) should be possible. Using too large a time step for a given scenario can lead to convergence difficulties.
c.If you find a fixed time step is running into convergence difficulties at a particular time and you want to avoid having to use a smaller step throughout then consider using a variable time step with a smaller minimum step. The aim would be that the time step would reduce around the problem time, but then increase back to a normal size afterwards.
d.The dynamic analysis duration for regular waves should correspond to a number of wave periods (e.g. 5 periods is typical) whereas the duration for a random sea should be a number of hours (e.g. 3 hours). If you are interested in assessing just the governing motions from a random sea then use the Vessel Motion Timetrace History analysis to pin point when the governing motions occur and then you can adapt the dynamic analysis start/end times accordingly to fit just around the governing time, thus avoiding long analysis durations.
e.Try to use a non-linear ramp so as to ensure a smooth transition at the end of the ramp time. The ramp time itself should ideally correspond to 2 wave periods for regular waves and 50s to 100s for random seas.
3.In terms of wave loadings, remember to assign a Vessel Motion component to the Vessel component otherwise no wave generated vessel motion will occur during the dynamic analysis. It is also recommended to perform equivalent regular wave analysis (short duration) first before proceeding to a full random sea analysis (much longer duration). This allows for quickly assessing the dynamic analysis convergence and predictions for the maximum wave height, without waiting hours for a random sea completion. Equivalent regular waves can be created by assuming the following:
a.Regular Wave Amplitude = Hmax/2, where Hmax = 1.75 to 2 times Hs
b.Regular Wave Period = Tp, although it may be best to assess a number of regular wave periods around Tp (e.g. +/- 2s)
4.By default PipeLay performs post-processing on automated plots and tabular output for each installation stage. This does not take long for static analyses, but for dynamic analysis, particularly for random seas, this can be time consuming and so you may wish to turn off some/all of the default post-processing. If you are only interested in governing values then summary post-processing across all stages would be much more efficient than individual stage post-processing.
5.The Update button appears on many tabs of the Analysis component including the Installation Stages tab. If it is enabled then you must remember to press it if you wish to store changes made to a stage/input on the tab.
6.The Stage Preview is a useful tool for checking multi-stage operations where different aspects of the model are being changed on the Parameters tab. More details on this are provided in the Parameters tab subsection. Note, the Stage Preview can take time to fully solve/display depending on the complexity of the model and so this may slow down the navigation around the Installation Stages tab. If you find this happening then you can disable the Stage Preview altogether using the provided tick box.
7.The Automate Stages function can save you time when defining multi-stage analyses, but you must remember to define the desired parameter changes and criteria first. This often means manually setting up stage 2 first to demonstrate the required changes from stage 1 and then using the Automate Stages function to create stages 3, 4, etc.