was documented; the wave height was then increased approximately 10 percent, 

 and wave bursts were generated again. This procedure was continued until a 

 wave height was reached which caused failure of the riprap. Failure was 

 defined as the riprap being shifted enough to expose part of the filter layer 

 which was removed by the wave action. 



The incident wave height was calculated using strip-chart recordings from 

 two resistance-type wave gages spaced one-quarter of a wavelength apart and 

 placed as close to the wave generator as feasible such that the waveform sta- 

 bilizes. The gages were placed as close to the blade as convenient to increase 

 the recording time of the gages before the waves were reflected from the 

 structure and returned to the gages. The average wave height, which was the 

 average of the wave heights for each gage, was measured by visual inspection of 

 the strip-chart recordings. 



In the LWT tests, a correction was applied to the wave height because of 

 the last wave effect (Madsen, 1970) , The term "last wave effect" refers to the 

 occurrence of one to three waves noticeably higher than the modal wave height; 

 the number of higher waves is related to the water depth-to-wavelength ratio, 

 d/L. This term was used because the highest wave usually occurred near the end 

 of the burst. A high wave also occurred near the beginning of the burst, causing 

 the highest waves to bracket the smaller waves of almost uniform height. These 

 smaller waves were considered the modal waves. 



The highest waves in the burst caused more stone movement than the modal 

 waves; thus, a correction was made to the modal wave height. The use of the 

 modal wave height to characterize the height of a wave burst would result in 

 an invalid comparison of riprap stability for tests with different wave periods. 

 The correction to the modal wave height was determined by the depth-to-wavelength 

 ratio, d/L: a decrease from 1.11 for a wave period of 2.8 seconds to 1.04 for a 

 wave period of 11.3 seconds. 



The last wave effect was not apparent in the small-scale tests because of 

 the initial and final position of the wave generator blade. The initial and 

 final position of the blade in the LWT tests was in the center of its total 

 stroke where the water particle velocities were at a maximum, creating some 

 irregularities in the initial and final waves. In the small-scale tests the 

 blade started and stopped in a rear position where the water particle veloci- 

 ties were zero, causing no apparent irregularities in the wave burst. 



The apparatus used to survey the filter layer and riprap armor layer con- 

 sisted of six vertical sounding rods mounted on a rack that moved along rails 

 mounted on the tank walls. Attached to the end of each survey rod by a ball 

 and socket joint was a foot measuring 1.8 centimeters in diameter which was 

 approximately one-tenth of that used in the LWT tests. The model surface was 

 surveyed in the same manner as that of the prototype. Surface elevations were 

 measured over square grid points 61 by 61 centimeters (prototype, one-tenth of 

 that for the model) apart on a horizontal plane. 



The following procedure for the small-scale tests was the same as for the 

 prototype tests (Ahrens, 1975) : 



(a) Place and compact core material. 



(b) Place and smooth filter layer material. 



13 



