tensile axial stresses that increased the principal static stress by 17 per- 

 cent and 34 percent. From results of large-scale tests, Burcharth and Liu 

 (1990) have shown that dolos axial stresses are relatively small for extreme 

 stress states. It is also likely that the axial stresses will not signifi- 

 cantly affect the static stress probability distributions that will be used to 

 determine the design stress level. The maximum prototype stresses without 

 axial contributions are therefore reasonable estimates of likely maximum 

 static stresses on the breakwater. 

 Static response 



45. In general, the static dolos stress can be computed as the mean of 

 a given stress time series. The stress time series is computed as shown in 

 the previous section. Of course, the mean of the time series may not be the 

 static stress if the oscillations in the time series are not symmetric about 

 the mean or if there is drifting or shifts in the data set. These peculiari- 

 ties in the data set must be accounted for in the computation of the mean. 



For the Crescent City prototype data, there were few shifts and little drift- 

 ing in the 30-min time series and nearly all of the data were symmetric about 

 the mean so the static response was computed as the mean of the time series in 

 all cases . 



46. Although the static data set for the Crescent City prototype 

 dolosse is a very small sample consisting of the 14 maximum stress values cor- 

 responding to the 14 working instrumented prototype dolosse, a conservative 

 probability density function (PDF) can be fit to this data set. For the 

 histogram, the mean is m cc = 416 psi and the standard deviation is s cc = 



188 psi , where the subscript cc indicates that the moments were determined 

 using Crescent City prototype dolosse data. The best-fit PDF to the static 

 stress histogram is of the log-normal type and can be characterized as 



P(°scc) = 





Hi 



Pec 



**cc 1 



o gc £ cc s/2* 



(3) 



where the mean of lna scc is given by a cc = 5.93 , and the standard deviation 

 is given by /3 CC = 0.45. Due to the extremely small data set, the fitting of 

 the density function was done visually. The log-normal distribution repre- 

 sents the skewness and the tail or extremal values of the data well. Also, 

 Burcharth and Liu (1990) support this distribution for static dolos response 

 with large-scale test results. The static stress can be nondimensionalized by 



24 



