A careful examination of these data indicates that the best predictions 

 for all except two tests are obtained with prediction schemes using 

 remolded strengths. In terms of the importance of the viscosity 

 parameter, \i, the results are inconclusive. Table 11, which presents 

 the average percentage error produced with each scheme, illustrates 

 some of the problems associated with determining the importance of 

 soil viscosity. In terms of the overall average error, the scheme 

 which uses remolded strengths and a soil viscosity of 3.0 appears 

 to be most accurate. However, if the results listed in Table 10 

 are consulted again, it is seen that the schemes using lower 

 viscosities (0.0 to 1.0 Ib-sec/ft-^) are the most accurate for 

 most of the tests. However they are greatly in error for tests PT-3 

 and PT-5. Since these tests were performed with the cone penetrometer, 

 and since, owing to the small base area of this device, the soil entry 

 point is difficult to determine, there is reason to believe that the 

 original accelerometer records may have been misinterpreted. In 

 any case the results of these tests conflict so strongly with the 

 results and analyses of the other tests that there is a strong 

 probability that an error or misinterpretation has been made somewhere 

 in the data reduction process. Considering the average error resulting 

 from the tests other than PT-3 and PT-5 as also presented in Table 10, 

 it is seen that the scheme which uses remolded strength and the 

 relatively small soil viscosity coefficient, y, of 1.0 Ib-sec/ft-' 

 yields the best results. 



It should be noted, however, that all of these schemes yield 

 relatively good penetration depth predictions. The scheme with the 

 largest error is still within 50 percent of the correct solution. It 

 may be concluded, therefore, that a penetration prediction scheme 

 founded on "pseudo-static" concepts apparently yields good results 

 for low velocity penetration. The use of remolded strength in the 

 force prediction equations improves the prediction accuracy. The use 

 of at least some modification for soil viscosity apparently produces 

 greater accuracy. This is not a dominant factor, however. 



DISCUSSION 



The various results of the analysis and testing conflict somewhat. 

 This is best illustrated by considering the following points: 



1. The soil tests indicated a strong variation of shearing 

 strength with shearing velocity. 



2. The regression analysis indicated that velocity-dependent 

 terms are the most significant in influencing penetration response. 



3. The physical analysis indicated that static soil resistance 

 altered only slightly to account for velocity dependency was most 

 effective in predicting penetration response. The predictions obtained 

 were rather good. 



18 



