Simulation Length |
 
|
Simulation Length |
|
Selection of an appropriate simulation length depends on whether you are running a regular wave or random sea dynamic analysis. In regular wave analysis, the simulation length needs only to be long enough for the effects of initial transience to dissipate fully and for steady state response to be established. Generally speaking, a simulation length of 4-6 times the period of the regular wave is sufficient. However, there may be some exceptional cases when it takes considerably longer for steady state conditions to be fully established. For example, a catenary riser subjected to cross current loading may experience “riser walking” – a phenomenon which involves lateral displacement of the touchdown zone increasing over successive wave periods. For this reason, it generally advisable to check some sample time histories of response to ensure that steady state conditions have been attained, rather than simply assuming this is the case, and proceeding directly to envelopes of response over the last couple of wave periods.
In the case of random sea dynamic analysis, the simulation time needs to be much longer than that of a regular wave analysis. The rationale behind this is as follows. The input to a random sea analysis is in the form or a wave spectrum (or spectra). Flexcom discretises the spectrum into discrete component harmonics, and uses these to generate a time history of random surface elevation. Each harmonic is essentially a regular wave, with its own individual amplitude, period, direction and random phase angle. As the assignment of phase angles is a completely random process, the resulting time history of water surface elevation represents only one possible realisation of a frequency domain excitation in the time domain. For this reason, the results of interest are really the statistics of response (as opposed to the actual time histories themselves). So the analysis must be run for long enough to provide confidence in the statistical measures – in order words until the response statistics reach steady state. A simulation time of 1800s (0.5 hours) would be considered a lower bound, but random sea analyses are often run for 10,800s (3 hours). Such lengthy simulation times obviously have implications in terms of run time and storage capacity.