6585845002

213

South Africa

strength determined from Eąuations (9A) and (9B) and the following approximate relations suggested by Skempton (1951).

l/ro_v^cu = 70 to 120

(overconsolidated)    (15A)

l/m_v^cu = 25 to 80

(normally Consolidated)    (15B)

From observed values of settlement, indi-cations are that although there is a wide scatter, results are generally useful for preliminary estimates.

A further method which has recently been employed, by the author, for computing settlement of shallow rigid foundations over alluvial and aeolian fine to medium sands and clayey sands is that of Schmert-mann (1970). The value of q_c from the Dutch static cone penetrometer test is used to estimate insitu compressibility from the relation E_s = 2 q_Q, and the strains and resulting settlement calcu-lated using a single triangular distribu-tion of vertical strain, in the form of strain influence factors, for all cases. The intermediate parćimeter of change in vertical stress is not required in this method.

Application of Schmertmann's theory to prediction of settlement of three large oil tanks founded on sand in South Africa has been discussed by Webb and Melville (1971). Ratios of calculated to measu-red values were 1,22 and 0,81 for the tanks underlain by deep sands but up to 2,53 in the case of a tank underlain by rock at a depth equal to a third of its diameter. The strain distribution is clearly important in shallow sands and, as anticipated by Schmertmann, is apparently affected by density, compressibility, layering, depth to bedrock and flexibility of the applied load. From recent expe-rience with column bases in estuarine sands, greater in depth than twice the least width of the base, encouraging results have been obtained.

Piled Foundations.

Piles in estuarine and alluvial sediments are usually designed as end bearing piles, and shaft friction neglected unless negative shaft friction is expected. The interpretation of Dutch static cone pene-tration test results is based on the prin-ciple that in a homogeneous soil the re-sistance to penetration of a cone divided by its cross-sectional area is a constant, and that the surface of failure, initia-ted beIow the cone in the form of a loga-rithmic spiral, extends for an increasing distance upwards as the cone is pushed downwards.

In most cases the soil is heterogeneous, with erratic variation in penetration re-sistance. Pile point resistance is then calculated by averaging the cone resistance over a depth a D above the pile founding level and 8 D beIow it, as re-commended by Van Der Veen and Boersma (1957), where D is the pile diameter.

For loose sands a is taken as 3,75 and 8 as 1,0. With increasing density of sand a and 8 are increased to as much as 8,0 and 3,5 respectively for medium dense sand as recommended by Begemann (1963). Permissible end bearing loads of bored piles are often taken as about ²/j of those for a driven pile of the same size founded in the same sand stratum.

There is an increasing tendency to allow for shaft friction, but with the Dutch static cone penetrometers generally avai-lable in South Africa only accumulated side friction is measured. Because of possible disturbance of the soil as the outer casing penetrates it the recorded variation of accumulated friction with depth is evaluated qualitatively and employed to indicate whether the various subsoil strata consist of clays or sands. In clay soils shaft adhesion, f_g# is com-puted from q_ę, using Equations (9A) and (9B) to obtain c_u, and ratios of f_§/c_u ranging from 1,0 for soft clay to 0,3 for stiff clay, as given, for exaraple, by Tomlinson (1963). Use is also madę of the relationship q_c = 50 f_s for loose sands and clays given by Mohan, Jain and Kumar (1963). Direct relationships bet-ween q_c and f_s are employed for sands such as those of J.H. Schmertmann refer-red to by Sanglerat (1972). For loose, medium dense and dense sands ratios of f_s/q_c are ^l/_60t V₁₂₀ and ²/n« respectively. Corresponding q_c values are 20, 80 and 220 bar.

In three recent controlled tests on pre-cast concrete piles driven through loose sands containing clay layers to a stratum of dense sand at 25 m the ratio between ultimate load and predicted load calculated from q_c values ranged between 1,10 and 1,25. A factor of safety of 2,5 is usually applied to computed end bearing and shaft frictional loads.

Attention must be drawn to a possible source of serious error in the interpretation of the static cone penetrometer test when pebble or coarse gravel layers occur in alluvial sediments. While such layers may yield very high values of q~ and refusal to further penetration of the cone, they may be relatively thin and offer little resistance to penetration of piles.

The Dutch static cone penetrometer test is also used extensively in South Africa for control of the vibroflotation process for densifying loose sands as described by Webb and Hall (1969). Crushed stone of about 25 mm in size is usually employed for backfill. Provided the test positions are morę than about 450 mm from