You are hereExtreme response of nonlinear ocean structures: Identification of minimal stochastic wave input for time-domain simulation

Extreme response of nonlinear ocean structures: Identification of minimal stochastic wave input for time-domain simulation

Rune Torhaug
Principal Advisor: 
Steven R. Winterstein
Year Published: 
Mon, 01/01/1996 (All day)
Torhaug_Rune.pdf3.66 MB

The wave loads on complex ocean structures typically vary nonlinearly with the wave elevation. For these nonlinear ocean structures time-domain simulations remains one of the few general techniques for estimating response statistics under random wave loads. We here critically evaluate strategies to identify minimal portions of stochastic wave input to form reliable extreme response estimates of nonlinear ocean structures.

We consider first the nonlinear dynamic response of a jackup structure under random wave loads. For a simplified jackup model, average behavior and variability in extreme forces and responses are found from simulation over many 6-hour seastates. Weibull and Hermite analytical models of response extremes are also presented and evaluated. Necessary simulation lengths are established both for direct simulation of extremes, and for analytical extreme models.

Next, we consider how the use of "design seastate histories" reduces the cost of time-domain response analysis. This is motivated by the fact that the computationally expensive part of time-domain simulation lies not in the simulation of multiple wave histories, but rather in the propagation of each through the hydrodynamical and structural model to obtain forces and responses. Excluding wave histories when their wave characteristics differ too much from their theoretical ensemble average values and only simulate response in a few most representative "design seastate histories", is shown to be a promising means of reducing the variability in the response estimates without need for an increase in the number of simulations.

Finally, we study the nonlinear extreme response of a flared container ship through short time-domain analyses. We focus on the use of "critical wave episodes", which are short wave segments which are likely candidates to produce extreme response in the hour-long history. We concentrate on propagating through the structure only a few critical wave episodes (each critical wave episode is only a few wave cycles long) per hourly simulation. We discuss how we can identify the location of critical wave histories within a longer wave history. Once a location is identified we discuss how long (how many cycles) a critical wave episode must be to ensure accurate results.