5 to 15 cm/sec; and U^ from 5 to 20 cm/sec). Resulting errors of 20 percent in Ti 

 values were typical. This effect is due to the filtering of the nonlinear output with 

 a nonzero average and is not a function of frequency. Since only the static cali- 

 bration is used, it is not an inertial effect. 



Steady-state linearity, threshold, and frequency response can be improved 

 by various servo-controlled systems,^ however, modifications of this kind would 

 be difficult under laboratory conditions and probably impossible in the ocean 

 environment. 



ROTOR ROTATION 



The impeller flowmeter usually cannot sense direction of rotor rotation and 

 is therefore unable to sense current direction. The outputs from the pickups are 

 usually the same regardless of the direction of rotor rotation. Several methods for 

 the solution of the problem exist; one is given by Koppe.^ 



ANGULAR RESPONSE 



Figure 7 shows the angulai- response for the Marine Advisers Inc. Model Q-8 

 ducted meter, which will be considered typical for this kind of meter. The response 

 is not a cosine function, and a single meter cannot be used to indicate the instan- 

 taneous component of the real flow parallel to the principal axis without some error.'' 



If more than one meter is used, the correct velocity components can still be 

 found for some cases. Specifically, one possible case is where two-dimensional 

 flow is present in the plane formed by the axes of two orthogonal meters. The de- 

 tails, however, will not be given here since the more important case is where 

 three-dimensional flow is present. Using three or more meters with noncosine 

 responses does not appear to be feasible. 



The present need is to determine the seriousness of a noncosine response 

 and to investigate the improvement of response characteristics. 



*Lanel'-' states, for a meter of similar design, "... the proper component of flow along the 

 axis of the instrument is recorded." (Result of tow tank tests at constant speed.) However, 

 no data were given and figure 7 will be assumed to be more typical for this kind of meter. 



14 



