more than a fail/no fail "proof test" of a design and should be arranged to 

 provide the maximum information of use for similar future designs. 



48. A procedure used for many years to verify the stability of proposed 

 cross-section designs involves subjecting a model breakwater to a short series 

 of monochromatic waves at the design (stability) wave height at various wave 

 periods above and below the design period. The water level is also varied 

 within the range of possible levels predicted for the prototype site. This 

 sensitivity analysis approach is intended to reveal the breakwater's response 

 to the severe condition when plunging breakers are directly impacting the sea- 

 ward face, as seen in Figure 8. Displacement of some fraction of the armor 

 layer is measured by before and after soundings of the model structure. This 

 procedure is relatively economical and provides an indication of the design's 

 resistance to armor unit displacement by a group of waves with a "worst case" 

 combination of period and water level. Some statistical confidence is lost 

 since the design criteria for wave period and water level are not held con- 

 stant in the modeling procedure. Subsequent changes to the cross section in 

 response to unacceptable damages in the model contribute to further departure 

 from initial design criteria and any associated risk analysis. Design cri- 

 teria must then be reformulated and associated analyses repeated with the new 

 criteria. 



49. Tests of cross-section designs with irregular waves typically 

 involve a longer series of waves, since a significant number of waves (100 or 

 more) are necessary to adequately resolve a specified energy spectrum. The 

 added test condition parameters related to reconstructing a specific spectral 

 shape in a wave flume discourage the sensitivity analysis method described 

 above. Hydraulic laboratories differ in their approach to tests for the 

 effect of wave groups with irregular waves, however. Some favor manipulation 

 of spectral shape parameters to enhance wave groupiness, while others prefer 

 spectra that are as natural as possible. Recorded spectra are reproduced in 

 some instances to assure a completely natural incident wave condition in sta- 

 bility tests. Durations of individual tests also vary from relatively short 

 tests of around 100 waves (30 to 45 min) to tests of thousands of waves and 

 many hours simulating the growth and decline of a storm, as illustrated in 

 Figure 5. Further discussion of model tests with irregular waves is available 

 in Jensen (1984) and Bruun (1985). 



50. Evaluation of damages after a test is a critical step which 



31 



