Summary of Both Laboratory and Field Test Results . The final analy- 

 sis of the ultrasonic inspection technique after the laboratory and field 

 tests is based on the following criteria: (1) accuracy, (2) range of 

 damage detection, (3) repeatability of readings taken, and (4) operational 

 simplicity. 



Accuracy is defined as the extent to which a given measurement agrees 

 with the standard value for that measurement. One method for estimating 

 accuracy uses the regression curves plotted in Figure 16. The R-value 

 is shown for each curve. The test piles, standard, and 8335D show 

 R-values of at least 0.6, indicating some relationship between the stan- 

 dard deviation and percent cross-sectional area loss. Complete depen- 

 dence, however, requires that R = 1. Piles 4x1 and 5x have R-values 

 equal to 0.02 and 0.06, respectively. Since R is close to 0, little 

 relationship exists between the standard deviation and cross-sectional 

 area loss for these piles. According to these results, the only correla- 

 tion is for the standard pile which has only simulated damage with a 

 maximum cross-sectional area loss of 26% and for pile 8335D which has 

 extensive damage with a minimum cross-sectional area loss of 45%. Con- 

 sistent correlation between standard deviation and percent cross-sectional 

 area loss has not been found. 



The range of damage detection can be evaluated by examining the 

 regression curves (Figure 16) . The curve for the standard pile 

 (Figure 16a) shows that for a standard deviation value of 0.25 mV, the 

 percent of cross-sectional loss could be between 0.0 and 20%. The re- 

 gression curve for pile 8335D (Figure 16b) shows that a standard devi- 

 ation value between 0.1 and 0.2 mV could indicate a cross-sectional loss 

 in the range of 55 to 85%. Piles 5x and 4x1 show no decrease in standard 

 deviation values with an increase in cross-sectional area loss. The 

 variability of readings makes it impossible to discriminate the range of 

 cross-sectional loss that can be detected. 



The repeated ultrasonic readings varied in the same location. The 

 variance between standard deviation readings was less that 0.05 mV. 

 However, according to the regression curves, a small difference in stan- 

 dard deviation values (0.05 to 0.1 mV) can seriously affect the value of 

 cross-sectional area loss. 



The deployment of the ultrasonic equipment in an operational envi- 

 ronment and subjected to actual use as in field testing showed the sim- 

 plicity of using the ultrasonic device. The only function required of 

 the diver inspecting the pile is to hold the transducer in the proper 

 location. Both the distance between the transducers and the angle of 

 orientation of the transducers in the transducer holder is pre-determined 

 and pre-set. 



IMPACT TESTING 



Theory 



Impact testing at NCEL was based on the simple hammer-sounding 

 inspection method currently used by Navy divers to detect internal damage 

 of timber piles. When a timber pile is struck with a hammer, a sharp 

 ringing sound indicates a solid pile; a hollow or dull thud indicates a 



35 



