1. The most accurate and complete time-domain models have only one 

 restriction — that the buoyancy of the vessel be linear with immersion. 

 This does not result in any significant errors for moderate vessel 

 motions. However, this restriction does introduce errors for severe 

 vessel motion when bow/stern submergence occurs. This restriction is a 

 consequence of the mathematics required to transform frequency-domain 

 vessel motions to the time domain. Since no available vessel motion 

 models can handle extreme vessel motions, this restriction is unimportant. 

 However, recent efforts in the OTEC project towards the development of 



an extreme vessel motion model may spur research aimed at the development 

 of a corresponding mooring model. 



2. At the present time, there is no accepted technique for pre- 

 dicting and simulating the slowly varying drift forces on a floating 

 vessel. These forces can be significant in comparison to the other 

 environmental loads. Approximate techniques of unknown accuracy are 

 available for estimating this load. 



3. Another limitation in reducing the errors associated with 

 mooring simulation comes from the uncertainty in defining the environ- 

 mental loads, particularly the wind and wind-driven surface waves. 

 Errors associated with the use of a wind wave model (Bretschneider, 

 Pierson-Moskowitz, etc.) have been shown to approach 100% for spectral 

 components as compared to actual measurements. These errors can seriously 

 affect the simulation due to the frequency sensitivities of the vessel 

 response and the use of the wave spectrum in determining the mean and 

 slowly varying drift forces. 



The development of spectral families for wind wave models by Dr. 

 Ochi is an important development reported at the seminar. By identifying 

 the error bounds (admittedly a statistically averaged value) in these 

 wind wave spectral models, much of the uncertainty in the final results 

 can be reduced. For many mooring models, the error introduced by using 

 a single spectral model was significant compared to the error due to 

 approximations in the mooring model itself. 



4. It was also pointed out that developing and using a very accurate 

 mooring model may not be cost effective if the criteria by which the 

 results are evaluated are not well-defined. This is illustrated in 

 Figure 2, which shows the uncertainty (also probability) in the simulated 

 results, p(s), and the uncertainty in the criteria, p(c); the bandwidth 



of either curve is analogous to the standard deviation of the error. 

 The area of overlap gives an indication of the probability of system 

 failure. For example, in long-term applications, p(c) for failure loads 

 may be large due to uncertainty in the corrosion, wear, etc. of system 

 components. This has important implications because the mooring designer 

 could simulate such a system with an inexpensive, simplified model and 

 save computation costs from a more refined model. A more detailed 

 illustration of model errors versus evaluation criteria is shown in 

 Figure 3, using CEL's mooring models as an example. Definition of p(c) 

 is dependent on each application, so generalizations would be difficult. 

 Recognizing that the evaluation criteria play a role in the choice of 

 analysis models is the first step. 



