Fig. 3-1(b) show the contents of hydrodynamic force for each derivatives at the time of turning angle 180 degree that is steady turning situation. At this moment, forces and moments caused by nonlinear derivatives give a big affect on the estimation of YH and NH. Although the difference between estimated force concerning Yrrr from limited range data and one from whole range data is big, the affect on YH is small because Yrrr is minor factor in total force YH. On the other hand, with regarding to the contents of total moment acting on hull NH, hydrodynamic derivatives obtained by difference experimental range show big difference. Moment estimated by limited range is half of one estimated by whole range data. Fig. 3-2 shows the ratio of the difference of each moment estimated by limited range from one by whole range to difference of total moment estimated by limited range from one by whole range data. Non-linear and coupling terms show big difference between the both experiment range. As a result, in steady turning situation, non-linear and coupling term's moment give a big influence. In Figure 2, estimated rate of turn by whole range data is smaller than one by limited range data. The reason of this difference is to estimate big turning resistance because the difference of estimated moment caused by Nvvv and Nvrr.

 Consequently, the difference of experimental ranges gives a big influence for the big motion, such as, big yaw motion and big swaying motion. The influences of experimental range correspond to the non-linearity of hydrodynamic force and moments. As the non-linearity may relate to the hull form, the rational experimental range will be decided after the relation between hull form and non-linearity of hydrodynamic force and moment is clarified.

　

　

　

3.2 The analyzing procedure on estimation of hydrodynamic derivatives

　

 The influence of the analyzing procedure on hydrodynamic derivatives is discussed in this section. Hydrodynamic derivatives, which are identified by applying the following two kinds of analyzing procedures are compared and discussed. The result of captive model tests shown in Figure 1 is applied to this investigation.

　

　

 Identified derivatives according to each procedure are shown in table 2.

　

 On identified linear derivatives, both variation of Nv and Yv are about 10%, and Nr and Yr are about 5% or less. The variation of drifting derivatives is larger than that of turning derivatives. On identified non-linear derivatives, the variation of drifting derivatives is also larger than that of turning derivatives. Variations of coupling derivatives on turning and drifting are 30%〜40%. It is clear that the analysis procedure affects on accuracy of identification for linear and non-liner derivatives on drifiing.

 A simulation of the turning test with rudder angle 35 degree was carried out using each identified derivatives. Simulated trajectories are shown in figure 4. The dotted line in the figure shows estimated results using derivatives, identified by the sequential identification, and the chain line shows results using those by the package identification. The solid line shows a free running test result of the 2.5m model.

　

　

　

 On time histories of turning, the difference of both simulations is small through turning. Simulated turning trajectory based on the sequential identification is quite similar to that on the package identification. This investigation shows that the influence of the analysis procedure on estimation of maneuverability is small in case of Esso Osaka.

　

　

　

 The influence of the propeller loading condition, on hydrodynamic derivative is described in this section. Three kinds of captive tests for the 2.5m model were carried out. First one was captive model test without a propeller. Second one is with a propeller at the self propulsion point of the model, where the propeller advance ratio Js = 0.472. The last one is with a propeller at the self propulsion point of the ship, where the propeller advance ratio Js = 0.691. Identified derivatives based on the result of the captive model test with each propeller loading are shown in table 3.

　

 Variations of identified linear derivatives except Yr are about 3%. The value of Yr becomes larger as JS increase. Variations of non-linear derivatives and coupling derivatives also become larger as JS increase.

　

 A simulation of the turning test with rudder angle 35degree was carried out using each identified derivatives. Simulated trajectories are shown in figure 5. The dotted line shows estimated results using derivatives identified based on the result of the captive test without a propeller. The chain line show that with a propeller at the self propulsion point of the model, and the broken line show that with a propeller at the self propulsion point of the ship. The solid line shows a free running test result of the 2.5m model.

　

 On trajectories and time histories of turning, these three kinds of simulations agree closely in the beginning of the motion. But the variation of simulated drift angles becomes significant as the turning proceeds. The difference of those between with a propeller and without a propeller is not small. But the difference of those between at the ship propulsion point and at the model propulsion point is small. It is considered that, for the Esso Osaka, the influence of the propeller loading on estimation of maneuverability is small.