The complexity of this expression arose essentially from a desire to match every 

 observed peak in the experimental frequency-response characteristics, as shown in Figure 9. 

 In retrospect, it would have been better to use a much simpler approximation and to reserve 

 a finer frequency correction for later application. 



The overall analog computer configuration is pictured in Figure 10. It was excited by 

 white noise generating a random voltage which was recorded on magnetic tape in order to 

 control the wavemaker system. 



DIFFICULTIES ENCOUNTERED IN WAVE GENERATION 



Having fixed the desired relative distribution of wave height for each increment of 

 frequency, there are two basic goals in optimizing the controlling program: 



1. To raise the upper limit of average wave height which can be created in the basin. 

 This is constrained by the allowable range of travel of the flapper valve in the wavemaker 

 and by the maximum rpm (1500) at which the air-blower motors can be driven. 



2. To obtain the specified distribution of wave height at each frequency. 



Several deficiencies were obvious in the first set of programs developed in June 1962» 

 The measured wave heights were much smaller than desired, the spectral distribution was 

 greatly distorted, and the hydraulic servo systems were rate-saturated. It was immediately 

 obvious from an examination of the spectrum of the taped excitation signal that a very consid- 

 erable excess of power had been concentrated in the frequency range above 1 cps in order to 

 compensate for the inefficient generation of waves at these frequencies. This excess energy 

 was undoubtedly the reason for the observed rate saturation, and because of the nonlinear 

 behavior, it contributed to the spectral deformation. Also, since the total power of a command 

 signal determines the root- me an -square (rms) motion of the flapper valve, which cannot exceed 

 a certain amount because of physical limitations on travel, this additional inefficient energy 

 caused a large reduction in observed wave-height levels in the basin. 



To lessen the effect of this undesirable high-frequency power, the programs were later 

 operated on by a low-pass filter, which had a flat frequency response to 1 cps and a 

 160-db/decade attenuation thereafter, synthesized on the analog computer. 



Still another correction was introduced to increase the maximum wave heights which 

 could be generated in the basin. In theoriginal programs, the average amplitude of the flapper 

 valve motion, which was continuously variable, was adjusted so that the largest peaks in 

 20 min of recording were a little less than the mechanical stops on the valve. To increase 

 this average valve motion, an electronic clipping of the subsequent taped program was made 

 which sliced off occasional peaks at the same voltage. This permitted adjusting the running 

 level so that the clipped peaks caused motion just within the mechanical stops and resulted 

 in a considerable increase in wave height with a slight cost in nonlinearity. 



12 



