Tóm tắt. Các kết quả từ một loạt các bài kiểm tra trực tiếp giao diện cắt trên một loạt các loại đất cohesionless được trình bày. Các xét nghiệm được sử dụng giao diện thép với tính chất so sánh với những cọc công nghiệp và điều tra sự ảnh hưởng trên sức đề kháng cắt của mật độ tương đối, có nghĩa là kích thước hạt và mức độ căng thẳng. Những kết quả này được so sánh (và nhìn thấy được trong thỏa thuận tốt) với các phép đo thu được bằng cách sử dụng một đống chuyển nhiều instrumented cài đặt trong một hang trung> cát. một số những yếu tố quan trọng hơn ảnh hưởng đến hệ số ma sát được phát triển bởi cọc trong đất cohesionless được xác định và tác động của chúng đối với thiết kế được thảo luận.
Trang 1FRICTION COEFFICIENTS FOR PILES IN
SANDS AND SILTS
R 1 JARDINE
Senior Lecturer, Imperial College, London
B.M.LEHANE
Senior Engineer, Arup Geotechnics, London (formerly Research Assistant Imperial College)
and
S J EVERTON
Sir Alexander Gibb & Partners (formerly Post-graduate student Imperial College)
Abstract The results from a series of direct shear interface tests on a range of cohesionless soils are presented The tests used steel interfaces with properties comparable to those of industrial piles and investigated the influence on the shearing resistance of relative density, mean particle size and stress level These results are compared (and seen to be in good agreement) with measurements obtained using a heavily instrumented displacement pile installed in a medium dens!> sand Some of the more important factors affecting the friction coefficients developed by piles in cohesionless soils are identified and their implications for design are discussed
1 Introduction
Recent reviews describe how the shaft capacity of piles in sand is usually assessed
by one of three routes; Briaud and Tucker (1988), Poulos (1989) or Jardine and Christoulas (1991) One popular approach is to assume a direct relationship between the local skin friction (1"1) and an in-situ test parameter such as SPT N, CPT qc
of pressuremeter Plim' Alternatively, 1'1 is assumed to be a simple multiple (;3) of the initial free-field vertical effective stress (a~o)' The third route is to assume a Coulomb failure criterion and to assess the normal effective stresses and frictional coefficients acting at failure separately:
Although the third route is the most attractive analytically, it is not clear how a~ 1
should be evaluated and 8' is rarely measured in appropriate laboratory tests As
a result, approximate rules of thumb (such as taken 8' = 0.81', after Potyondy, 1961) are adopted for design calculations
The API RP2A recommendations for offshore piles in sand suggest that Equa-tion (1) should be evaluated assuming that a~ 1 / a~o' or J( 1, is equal to 0.8 and 1.0 for open and closed-ended piles respectively and that 8' varies with soil type and relative density (Dr); a graphical interpretation of the implied variations of
Volume 28: Offshore Site Investigation and Foundation Behaviour, 661-677,1993
© 1993 Society for Underwater Technology
Trang 2662 R J JARDINE ET AL
6' with Dr and mean grain size (Dso) is given in Figure 1 Upper limits to T are also specified, but we should note that the API recommendations were derived by back-analyzing tests on un-instrumented piles and the individual parameters may
be more notional than realistic
For advances to be made, it is necessary to establish real values for a~ f and 6' This paper explores the factors affecting the values of 6' using data from instrumented pile tests and laboratory direct shear interface tests Factors affecting
a~ f are also discussed
2 Previous Studies of Soil-Interface Shearing Resistance
Potyondy (1961) presented some of the best known work on soil-interface shearing resistance His study involved direct shear tests on five soils (ranging from sand to clay) using concrete, steel and timber interfaces Broadly, he showed that the peak resistance (6~ealJ depended on soil grading, density, stress level, interface material and surface roughness For very rough surfaces, 6~eak approached the 4>~eak value seen in equivalent soil-to-soil shear tests, but, in most cases, lower 6' values were
Trang 3FRICTION COEFFICIENTS FOR PILES IN SANDS AND SILTS 663 obtained Tests on silt gave higher 6' values than equivalent experiments on a medium sized sand, indicating the opposite trend to that projected in Figure 1 The trend for 6' to increase as the grain size falls breaks down if the soil contains
a significant proportion of platy clay particles Comprehensive research using the ring-shear apparatus (e.g see Lupini et ai, 1981, Lemos, 1986, Tika-Vassilikos,
1991) has since shown that the interface shear behaviour of such clays can be dominated by the development of thin zones of oriented (residual) fabric which can show very low 6' values
Leaving aside the question of residual clay fabric, Potyondy's results for sands and silts cannot easily be generalised as the roughness and hardness of his inter-faces were not quantified, just three granular soils were tested and only the peak strength data are reported However, more recently, several research projects on the interface shearing resistance of sans have been reported, including those by Yoshimi and Kishida (981), Lemos (1986), Boulon and Foray (1986), Kishidaand Uesugi (1987), Everton (1991) and Lehane (1992) Most of these studies will be referred to later, but we will consider first the results obtained by Kishida and his co-workers using a specially designed simple shear apparatus
Kishida's apparatus made it possible to distinguish the point at which slippage starts between the soil and interface in direct shear It turns out that, with dense sands, this yielding condition coincides with the attainment of 6~eak' and therefore with the peak in measured shear resistance For samples denser than their critical density, the first slippage also coincides with the start of measurable dilation; the interface resistance falls as shearing continues until, after a few millimetres
of relative displacement, an ultimate condition is reached where dilation ceases and the resistance remains constant With loose sands, slippage appears to start significantly before the maximum 6' value is achieved
Kishida and Uesugi (1987) report that the soil parameters which affect 6~eak in sand-to-steel tests include the shape, size (Dso) and hardness of the sand grainsl
They showed that the steel interface's roughness was an important factor, but considered that normal effective stress had little overall influence Uesugi and Kishida (1986) propose straight line relationships for each sand between tan 6~eak
and roughness as shown in Figure 2 This relationship is truncated by an upper limit, where 6' = ¢/ When their data were re-plotted using a normalised Roughness Index (Rmax/ Dso) with account being taken of the angularity of the particles, a
general bi-linear relationship was found to hold for all sands composed of the same minerals
3 Instrumented Field Pile Tests in Labenne Sand
Field research with instrumented piles is required to assess the relevance of labo-ratory interface tests to the design of offshore foundations
1 Kishida and Uesugi defined roughness as the maximum asperity measured over a gauge length equal to Dso
Trang 4664
c 0.8 .Q
- Q
It 0.6
-0
-c
• !ll 0.4 .2 :e: Q)
0
0 0.2
R J JARDINE ET AL
Maximum Shear Stress Ratio
in Siml2le Shear Tests Critical ROUghes~ .~' ')::'-·0.90 r
/ / ~ •• D _.-0.85
;t 0 •
; ' / • ,D / /"0 -·0.58
/ D , /
.(" 0,1-' 0
, / 0
A Fujlgawa Sand
~"'D
D o6 • FukushIma Sand
V'" o Gloss Beads
o Toyoure Send
Roughness of Steel (pm)
Fig 2 Effects of steel interface roughness and sand type on tan 6' (coefficient of friction) after Uesugi and Kishida (1986a)
A programme of pile tests was performed in 1989 at the Laboratoires des Ponts
et Chaussees sand research site at Labenne, S.W France; full details of these
experiments are given by Lehane (1992), Jardine et at (1992) and Lehane et at
(1993) The tests used the Imperial College instrumented Pile (ICP)2, described
by Bond et at (1992), which allowed the local values of a~ and Trz developed at
a series of points on the pile shaft to be monitored with unprecedented accuracy Data were obtained throughout the processes of installation, equalisation and load testing
One of the key results obtained was that the equalised radial effective stress (a~c) was not a constant multiple of a~o' but was strongly and systematically dependent
on the relative density of the sand (which varied with depth at Labenne) and also
on the relative position of the pile tip Lehane (1992) explores the many practical implications of this finding We will return to this point later, but the principal point
of interest here is the way in which a~ varied with Trz during pile loading
Figure 3 presents a typical plot from one of the Labenne ICP (compression) pile tests, using data from an instrument cluster located in medium dense quartz sand The load cells show that shaft loading causes relatively small reductions in a~ at first, but that the response becomes more 'dilatant' as Trz/ a~ approaches a limiting ratio The stress path then deviates sharply to the right, mobilising its peak 0' angle
before the maximum shaft resistance is attained; the ultimate 0' is ~ 3° less than o~eal(' Further pile data from Labenne showed that the ultimate 0' value (i) did not
2 The ICP is shot blasted before use to have the typical surface roughness of an offshore pile, i.e
~ 8/Lffi Centre-Line Average (CLA)
Trang 5FRICTION COEFFICIENTS FOR PILES IN SANDS AND SILTS
60
40
t rz
(kPa) 20
TEST LB2C
O~ ~ ~ ~ ~-L ~~ -(Ir' (kPa ) -20
Fig 3 Variations of 1T~ with T during pile loading at Labenne, after Lehane et al (1993)
665
vary with the initial undisturbed sand density, (ii) was the same for wet and dry sand and (iii) was independent of the pile displacement rate (Lehane, 1992) Uesugi and Kishida's laboratory studies suggest that the increases in CT~ seen
in the Labenne tests were associated with sand grains being displaced radially outwards (by an amount 6r) as local slippage started at the pile shaft The ultimate failure (or critical state) was reached when the grains became 'unlocked' from the rough steel and large scale relative displacements become possible without any further dilation or change in shear stress The operational 8' value is the critical state angle (c5~s) and not the peak angle - which has been the focus of most previous research
Unlike the conventional shear box test, dilation at the pile-soil interface is resisted by the stiffness of the surrounding ground, causing CT~ to increase once slippage starts As a result pe3k obliquity (c5~eak) and maximum shaft resistance
(Tj) do not coincide Following the analysis given by Boulon and Foray (1986), Johnston et al (1987) and others it is surmised that the change in CT~ can be related
to the pile radius (ro), the average radial movement of the soil grains (6r) and the secant 'pressuremeter' shear modulus (G p ) (which falls as 6r Iro increases) by the expression:
The radial movement depends on roughness, Dso, stress level and Dr, but is
independent of pile radius per se (Boulon and Nova, 1990) We can therefore expect 6CT~ to vary inversely with pile radius and infer (from the Labenne experiments)
Trang 6666 R J JARDINE ET AL that dilation effects will dominate the capacity of small model piles, but make no significant contribution to the shaft resistance of large offshore piles Laboratory model pile tests reported by Lebegue (1964), Hettler (1982) and others are collated
by Lehane (1992) and are shown to be consistent with this trend
The controlled normal stiffness interface shear tests advocated by Boulon and his co-workers provide one way of investigating by interactions between dilation,
G p , pile radius and shaft resistance
4 Laboratory Studies on Labenne Sand
Laboratory studies were performed in parallel with the Labenne field pile research These included index tests, triaxial stress path experiments, shear box tests and soil-interface tests using the shear box and ring~shear apparatus (Lehane, 1992) Labenne sand has the uniform and consistent particle size distribution illustrated
in Figure 4a (soil 10); the principal results from soil-soil and soil- interface direct shear tests are presented in Figures 5, 6 and 7 The tests were conducted over a relatively narrow range of vertical effective stress (70±45 kPa)
Figure 5 shows the marked influence of the sand's initial voids ratio (and Dr)
on its peak ¢/ angle, with the densest samples giving angles 10° higher than the loosest However, in all cases ¢/ tended towards similar ultimate (or critical state) values (~ 33°) as dilation ceased and the shear stresses stabilised Figure 6 shows the individual tests data from one of the three suites of interface tests A range of initial void ratios and normal stresses were investigated for a steel interface which had a similar roughness to that of the Labenne pile The main features seen were: Dense samples showed reductions in shear stress (T) after achieving their peak values at relatively small displacements, but loose samples showed no clear peak in resistance
The maximum rate of dilation for dense samples occurred near the point at which T was greatest The loose samples showed either a net contraction, or
no overall dilation
All samples tended towards approximately constant volume conditions after relative displacements of 2 to 4mm
Figure 7 combines the data from Figure 6 with those from other tests on Labenne sand involving (i) a slightly smoother steel interface and (ii) a Teflon interface which was both smoother and softer The picture that emerges is similar to that from the sand-sand tests: for each particular interface, the initial relative density affects the amount of dilation and o~eak' but not o~s' For the two slightly different steel interface roughnesses considered the o~s values were comparable; peak angles were more sensitive to the changes in R Substantially lower angles were mobilised against
Trang 7FRICTION COEFFICIENTS FOR PILES IN SANDS AND SILTS
100
BO
60
40
20
o
(}()()()1
100
80
60
40
20
o
/
/12
0001
nAY
<HlOOl (}001
CLAY
7 'r ' I .'
I~ I
II' !
V / f!l
't
I •
y/
VJ~~
2 3 4 5 6
/((( (
1/))))
FUrticle size mm
FINE MEDIlt1 DJARSE FINE MEDIUM COARSE FINE SILT SAND
667
100
on
~
<n
40
20
0
1 00 Ion
20
0
MEOUM GRAVEL
Fig 4 Particle size distribution curves for soils in parametric study; (a) ungraded soils, (b) graded Cretaceous sand
Teflon interfaces
As the direct shear box has imperfect boundary conditions, check tests were run
(against an 8JLm steel interface) in the ring-shear apparatus described by Bishop
et al (1971); o~s was found to be within 1 o'of the average equivalent shear box value Stress-path triaxial tests gave the same rp~s as the soil-soil shear box tests However, the most interesting result was that the laboratory 0' values were virtually identical to those measured in the field instrumented pile tests (see Figure 3) The same agreement between laboratory and field was found in a later programme of instrumented pile tests in the Bothkennar clay-silt (Lehane and Jardine, 1992a)
Trang 8668
100
I
50 <1>1
46
42
36
34
30
R J JARDINE ET AL
1
"y/<t>p trend
• Peak
" ., ,
' " , " ,
~ .'
"-o
o Constant volume
o o o -. ======-<Pey :::: 33
26~ ~~ ~~ ~ ~=_ ~~ ~-0·46 (}SO 0·54 0·58 0~2 0·66 0-70
Initial voids mho ej
Fig 5 Variation of ¢/ with Dr for Labenne sand (1T~ = 75 =/- 45 kPa)
5 Parametric Study of Sand-Interface Shear
A parametric laboratory study was performed by Everton (1991) to supplement the Labenne research This involved direct shear tests on eight other granular soils in which the initial relative density, interface roughness and normal stress level were varied Control soil- soil tests were also performed Five of Everton's sands were fractions graded from the same deposit of Cretaceous Greensand3 The remaining three soils were: an industrial rock flour silica silt (HPF4), Ham River sand and a calcareous sand from the Middle East (AI Shattie)
Everton's data are collated here with: the Labenne data, results for Leighton Buzzard sand· (.given by Lemos, 1986), ring-shear interface data for a North Sea glaciomarine sand (Ridley and Jardine, 1992) and the Bothkennar clay-silt ring
3 A predominantly silica sand produced by the David Ball Company of Cambridge, UK
Trang 9FRICTION COEFFICIENTS FOR PILES IN SANDS AND SILTS
70 "t
(kPo)
60
50
40
30
20
10
51 (mm)
o~ ~ ~~ ~ ~ -~
+0·2
5h
brm)
+0·1
o ·0
- 0·1
5·0
o~ (kPo)
0 53
0 53
/:; 115 + 53
)( 53
<> 28
Fig 6 Interface shear tests on Labenne sand (R = 9.5 /LID, (J'~ = 75 =/- 45 kPa)
669
ej 0-55
0·54
(}55
0·64
0-64 0-55
Trang 10670 R J JARDINE ET AL
40 0'
rees)
0,.%
• R = 9·SjJm pedI
• R=5·Spm.ped<
36
32
28
24
20
• R=2·~.peak R=9·51J.m
Regression lines open S)1IboIs rmst voIune
~
R=5·51J.m • ~
~
~ 0 28° JJ_ 8-o-_o~ ""'_.-IJ _ cv
o - : _ _ (steel nterfoce R=2·01J.m
i;- 0 19°
- - 4 _- cv 6 (teflon interface 16~ ~ -L ~ L -~ ~
0·46 050 054 0-58 0-62 0·66 07
lnitinl voids ratio e j
Fig 7 Variation of 6' with Dr for Labenne sand (1T~ = 75 =1-45 kPa)
TABLE 1
Soils considered in the study
1 HPF4 - Crushed industrial rock flour; angular 0.04
3 5211 00 graded sand; sub-rounded 0.22
7 Ham River Sand; sub-rounded to sub-angular 0.32
8 Al-Shattie calcareous sand; rounded to angular 0.44
10 Labenne dune sand; sub-rounded to sub-angular 0.32
shear tests described by Lehane and Jardine (1992b) The twelve soils are identified
in Table 1, the grading curves are given in Figure 4 and Figure 8 presents the minimum and maximum void ratios, as determined by British Standard procedures The limiting void ratios appear to be controlled principally by the grain size,