Str^m-Tejsen and Chislett 



GENERAL CONSmERATIONS ON TESTING PROCEDURES 



It is seen that the mathematical model, Eqs. (5), is comprised of numerous 

 coefficients such as Y^^^, n^, and Xf,,. These coefficients in general depend on 

 the particular geometry and design of a ship, and they must be known with rea- 

 sonable accuracy before manoeuvres can be simulated by solving the mathemati- 

 cal model with the aid of a computer. 



Ideally, numerical values for the coefficients would be evaluated by theo- 

 retical means, but although some of the hydrodynamic coefficients can be cal- 

 culated approximately, the only reliable way at this time of obtaining values with 

 the accuracy needed for simulations is to conduct captive model tests. 



In a captive model test, the model is forced to perform precisely controlled 

 movements, one or two of the different motion and rudder parameters being 

 assigned values simultaneously. The resulting hydrodynamic forces and mo- 

 ments acting on the model are measured as functions of the parameters, and the 

 coefficients are subsequently obtained from these measurements. The expres- 

 sions for the hydrodynamic forces and moments, Eqs. (5), may be considered 

 having been developed on the basis of a superposition process, the accuracy of 

 which is progressively improved by successive inclusion of "crosscoupling" 

 terms expressing deviations from simple superposition. Similarly, in the exe- 

 cution of the captive model tests, parameters are first explored one at a time, 

 all other parameters being zero. The resulting forces and moments can then 

 be expressed by coefficients which are functions of the single parameter. The 

 next step in the captive model tests is to vary two parameters simultaneously, 

 and if the resulting forces and moments differ from the superimposed results 

 of the individually measured values, then the difference is expressed as a two- 

 variable function of the parameters, and the coefficient representing the "cross- 

 coupling" effect can be determined. Whereas, it is possible to generate three 

 or more parameters simultaneously and so obtain crosscoupling terms in more 

 than two variables, such terms have been found to be less than the accuracy of 

 measurement. It is perhaps interesting in this connection to consider a free- 

 sailing ship as a special case of captive model testing in which the ultimate 

 stage of superposition has been reached. The totals of forces and moments 

 acting can be inferred from the accelerations, and compared with model data 

 expressed as one- and two-variable functions. This is done in effect when full- 

 scale trajectories are compared with simulations based on model results. 



The range of motion and rudder parameters explored during testing should, 

 in principle, cover the range of subsequent simulation. Surge, sway- and yaw- 

 accelerations, speed loss, drift angle, yaw velocity, and rudder angle should 

 therefore be varied systematically up to the values corresponding to maximum- 

 rudder manoeuvres for the free-sailing ship. 



Captive model tests in which measurements are made of hydrodynamic 

 forces and moments resulting from drift angle, from rudder angle, and from 

 combinations of drift and rudder angles, can be conducted with relatively sim- 

 ple equipment in a conventional towing tank. 



Many methods have been used to measure forces and moments due to angu- 

 lar velocity, among them being such devices as curved models, curve-flow 



322 



