behavioral parameters. In many foundation designs, the validity of the 

 sample with respect to actual conditions was never established. A single core 

 was often assumed to represent the material surrounding the site. Neither 

 sample disturbance nor areal variability was considered in analysis. Conse- 

 quently, structures located near the site did not always perform as expected. 

 Some laboratory analyses consisted solely of classifying the soil according to 

 grain size and mineral constituents. Design was based strictly upon the 

 expected performance of the soil type. Since the range in behavior for a 

 soil type was large, a conservative design technique was employed. More 

 sophisticated laboratory testing techniques often failed to consider the low 

 effective strengths of the soil. Early attempts at performing consolidation 

 tests missed the behavior of the soil in the low pressure ranges. 



Even when relatively undisturbed representative samples were 

 evaluated for strength and consolidation characteristics, in-situ behavior 

 often deviated from analytical predictions. I n most cases, the difficulties 

 were attributed to the inability of analytical techniques to predict perfor- 

 mance. For example, the bearing capacity of cohesive soils has been found 

 to be lower than often anticipated. Keller's model footings penetrated to a 

 depth greater than predicted by calculations based upon the undisturbed 

 strength of the soil. However, calculations employing the remolded strength 

 of the soil predicted the depth of penetration rather closely. Apparently the 

 penetrating blocks progressively remolded the soil. Jack-up rigs designed to 

 apply a bearing pressure of less than 200 psf have failed in the underconsoli- 

 dated soils of the Mississippi Delta area. The factors of safety against bearing 

 capacity failure (undisturbed strength) for these soils were thought to be signi- 

 ficantly greater than one. However, at other sites traditional bearing capacity 

 estimating techniques are sometimes conservative. Results from the NCEL 

 plate bearing device and the ESSA plate load device indicate that the bearing 

 capacity for cohesionless materials is larger than predicted by methods sug- 

 gested by Terzaghi and Peck (1964). Foundation designs based upon these 

 latter calculations are conservative from a soils standpoint. 



Techniques for predicting the settlement of a structure have also 

 been found to differ from in-situ performance. The LOBSTER tests con- 

 ducted by NCEL suggested that a large amount of secondary compression 

 occurs in seafloor soils. A settlement analysis based on laboratory consoli- 

 dation tests therefore underestimated settlement. 



In other case histories, no reliable analytical technique was found to 

 be applicable to the particular condition in question. The Tektite project, 

 for example, abandoned the use of embedment anchors as a foundation for 

 the Tektite habitat when performance data were found to be nonexistent. An 

 acoustic array in the St. Croix Range slid nearly 1 ,000 feet down a gentle slope 



69 



