14262 JOURNAL OF THE GEOTECHNICAL ENGINEERING DIVISION ANALYSIS OF DEFORMATION OF VERTICALLY Loapep PILEs By Mark F. Randolph’ and C. Peter Wroth? lnmnooucTION ‘A large proportion of research on piles has been concemed with ultimate bearing capacity, and relatively litte attention has been paid to the deformation characteristics of piles at working loads. Currently, the maximum permissible settlement of a foundation is often a more stringent design criterion than the overall stability of the construction and, as such, the settlement of single piles techniques of analysis, such as the integral equation method adopted by Poulos and Davis (1968), Poulos (1968), and Butterfield and Banerjee (1970, 1971), and the finite clement method, have been used with some success. However, both methods have their limitations and it is not clear how to modify the solutions obtained to take account of inhomogencities or nonlinearity in the behavior of the ground that may occur al any particular site. ‘An attempt to clarify the manner in which a pile transfers load to the surrounding soil is presented in this paper. An analytical approach has led to the development of an approximate closed-form solution for the settlement of a pile under a given load, which has been checked using the numerical techniques mentioned previously, The resulting equations represent a flexible design aid whereby rapid estimates of pile deformation may be made without recourse to a computer. Moreover, the analysis indicates the relevant soil parameters that need to be measured and demonstrates the manner in which local softening near the pile - or the effect of a.weak stratum of soil-at some depth down the pile may be taken into account. Note.—Discussion open until May 1, 1979. To extend the closing date one month, a written request must be filed with the Editor of Technical Publications, ASCE. This - paper is. part of the copyrighted Jouraal of the Geotechnical Engiocering Division, .. Proceedings of the American Society.of Civil Engineers, Vol. 104, No. GT12, December, 1978. Manuscript was submitted for review for possible publication on August 2, 1977. "Univ. Asst. Lect., Cambridgé Univ., Cambridge, England. 2 Reader in Soil Mechanica, Cambridge University, Cambridge, England.” 14651450 DECEMBER T3878 ari? Dimensionay Anatysis Any analytical solution for the deformation of a pile must be applicable over a wide range of pile geometries and soil stiffnesses. It is helpful to form dimensionless groups of the relevant parameters, rather than investigale how the solution is affected by variations of each individual pile or soil parameter. The basic variables in the pile-soil system considered to be relevant are as follows: w = settlement of the pile; P = applied load; / = length of the pile; r, = radius of the pile shaft, preferred to the diameter since other horizontal distances will be measured from the center line of the pile; E, = Young’s modulus of the pile; G = shear modulus of the soil, preferred to Young’s modulus £, since the soil deforms primarily in shear and because the shear modulus is usually assumed to be unaffected by whether the loz.ting is drained or undrained; and v = Poisson’s ratio of the soil. Note that the Poisson’s ratio of the pile is ignored, as it has an insignificant effect on the load-seitlement behavior. In general, the settlement of the pile may be written as a function of the other six variables: WHf\(Prhty,EysGv) scree eeens Se inas © SN eee ws 1) ‘The dimensions of t.:¢ various terms in this equation are: MT, Lb Me eT, MES 0! ais seg ack i Q Following the technique of dimensional analysis, there are two independent dimensions, L and MT~?, which will lead to five dimensionless groups. Using r, to eliminate L and G to eliminate MT ™ gives: w a(Z 1 £, ) ® r \Gri'y, 6° 8h = ee ee ee cae Since the settlement will be proportional to the load, the groups containing w and P may be combined to give: Grew qd - : = ~1( a.) en ee Pete nena acsearle Ore sw & oe ee AD To in which k = E,/G. ; In the subsequent analysis an attempt is made to find the form of the function in this equation. For convenience, th. results will be expressed in terms of the inverse of the left-hand side, i.e., P/(G r, w), which will be referred to as the load-seitlement ratio. The dimensionless groups //r, and \ will be referred to as the slenderness ratio and-the stiffness ratio, respectively. . Numerica Tecunraues a gee, -Finite element and integral equation analyses have been used to establish and check the analytical model outlined later in this paper. Both these methods" have their limitations; the former is restricted, in. practical and economic terms, to the analysis of single piles, while the latter method, until recently, has beea restricted to a homogeneous linearly elastic soil. [The integral equation method has now been extended to deal with a nonhomogeneous soil, by treating theGT VENTRALLY eurieee 2 == soil as a number of homogeneous layers (Banerjee and Davies, 1977}.] However, the methods may be used together to explore most facets of pile-soil interaction. Finite Element Analysis——The finite element program used by the writers was based on six-noded isoparametric triangular clements in-an axisymmetric mode, which allow a linear variation of strain across each element. The method of solution was Gaussian elimination of a banded matrix, the bandwidth being restricted to 100. The pile and surrounding soil were divided into 267 elements (580 nodes) with a rigid horizontal boundary imposed et a depth, A, where h = 2.5 1 (Fig. 1). This is close enough to affect the solution for a pile in an elastic half space, but only by less than 5% (see Banerjee, 1970). The outer vertical boundary was ata radius of 50 r, and care had to be taken to ensure that this did not affect the solution. For piles longer than / = 40 r,, it was found that the method of fixing these boundary nodes affected the computed N \ | wv Vi a A ni JM 4“ VV AW Bi Ss L | i ‘Fares boundary FIG. 1.—Finite‘ Element Grid for Pile “FIG. 2-—Schematic Disgram of Method Analysis i of Integral Equations - pile settlement. Fixing the nodes in the radial direction only produced a good * compromise, but for very long piles (! > 80 7,) the mean of the settlements for totally fixed and for totally free vertical boundary nodes gave the best agreement: with “equivalent analyses. by -integral equations. Ideally, the outer vertical boundary ‘should be at a radius of at least i the length of the £ pile. oa ’ “+ Tntegral Equation Analysis.—The integral equation method of pile analysis— --’ based on the solution by Mindlin (1936)'for a point load ‘acting in an elastic -. half space—has been described by Poulos and Davis (1968), Butterfield and Banerjee (1971) and, in more detail, in the thesis of Banerjee (1970). . | The method is shown in Fig. 2. ‘The surfaces of the pile shaft and base » are divided into a number of segments; an underreamed pile may be representedby a disk of the appropriate radius at the base of the pile. Stress resultants are assumed to act ia the soil mass around the eventual pile surface, These Stress resultants will be shear stresses down the shaft of the pile and normal stresses at the base [Fig. 2(a)]. Equal and opposite stress resultants are assumed to act on the free standing pile [Fig. 2(4)]. The stresses are assumed to be uniform over ¢ach pile segment, and overall equilibrium is provided by balancing the applied load on the pile, P., against the sum of the stress resultants. The resulting displacements at the midpoint of each segment are calculated (Figs. 2(c) and 2(d)] for the soil mass from the Mindlin solution and down the pile from simple elastic theory. For n segments, Matching these two displacement fields provides n ~ 1 equations that, together with the equilibrium equation, enable the m unknown stress resultants to be calculated. Since equilibrium and compatibility of displacements are both satisfied, the pile may now be “‘installed’’ in the soil mass as shown in Fig. 2(). Although this description of the integral equation method is oversimplified, the important feawure is that the displacement field in the soil mass due to the Stress resultants is calculated from the Mindlin solution before the pile is actually “installed.” The writers’ computer Program was based on the technique described by Poulos and Davis (1968) where, following the conclusions of Mattes (1969) and Banerjee (1970) no account has been taken of radial compatibility of displacement between the pile and soil. The method of allowing for a rigid layer at some depth below the pile (essential for comparison with finite element analyses) described by Butterfield and Banerjee (1971) was adopted whereby fictitious stress intensities are introduced over the layer to give zero overall vertical displacements at that depth. A facility was also included to allow relative movement of the soil and pile next to the pile shaft when some Specified threshold value of shear stress was reached, The results quoted in this paper were obtained by dividing the pile into 15 shaft and five base segments. This gave values. of the load settlement ratio some 2-1/2% greater than a less accurate solution using seven shaft and three base segments. It was also found that the solution became unstable for very long compressible piles. The method does not appear suitable for piles where the quantity (J/7,)7/ > 20. Before applying the results of either type of analysis, checks were made that the two methods gave agreement with other published results and that they were consistent with each other when analyzing the simple case of a pile embedded in a homogencous linear elastic layer. Anacysis oF Pusey Seraranon oF Loans Carmen ey SHAFT AND’ Base The approach to the analysis is shown in Fig. 3(a), in. which the soil has been divided inio an upper and a lower layer by a horizontal plane AB, at the level of the base of the pile. It is assumed, initially, that the upper layer of soil will be deformed exclusively by the load shed by, the pile:shaft and that the lower layer of soil will be deformed exclusively by the pile base load. Fig. 3(6) shows the. separate deformation patterns anticipated; the plane AB has been “texploded” to A, B, and A,B, . The deformation patterns along A,B, and A,B, will not be compatible and this will lead to some interaction between the upper and lower soil layers. ‘It is emphasized that this simplified approach to pile anelysis an exact solution to the problem. However, the acceptability of the prece: assumption and any subsequent assumption may be checked with the aid of finite element and integral equation analyses. By cnsuring that the essential features of the problem are modeled correctly, @ solution satisfactory for engineering purposes may be obtained. Thus, in the present instance, the separate deformation patterns along A,B, and A,B, will be matched at the pile base and at large radii. : Interaction of Pile Shaft with Upper Layer of Soll.—Since the analysis will be developed for a pile in a linear elastic soil, the effect of the installation of the pile may be ignored at this point. (It is assumed that the soil parameters are not affected by the installation of the pile.) Also, only increments of stress due to loading of the pile will be of concern. For piles with slenderness ratio th) Separate deformation pettarns of unper and lower Ivers - Ct) FIG. 3.—Uncoupling of Effects Due to FIG. 4—{a) Mode of Deformation of Pile Shatt and Base Shate; (b} Stresses on Soll Element I/r, of 20 or more, the stress changes in the soil caused by load transfer st the pile base may be uncoupled from those caused by load transfer down the pile shaft in the manner described previously. ‘ The deformation of the soil around the pile shaft may be idealized as shearing : of concentric cylinders [seo Fig. 4(a)|. This approach has been considered by previous workers: Cooke (1974) has shown,good agreement between the resulting deformation patterns and measurements from pile tests; Frank (1974, 1975) and Bagueiin, et al. (1975) have used finite clement analysis to demonstrate the validity of the assumption. For completeness in the present paper, the analysis resulting from the idealized mode of deformation will be briefly outlined ss : follows. ‘ i : ‘6712 Consideration of Vertical equilibrium of an element of soi] [see Fig, tb) yields z = 0 5 ——(r+tr SE MAAS 12s Bean oo a ? az : ® in which + = the shear stress increment; and o, = the vertical total stress increment (taking compressive stresses as Positive). When the pile is loaded, the increase in shear Stress, 1, in the Vicinity of the pile shaft will be much greater than the increase in Vertical stress, a, and thus Eq. 5 may be approximated to become 2 0 6 = RTO et 18-5 tt tty co aang acy (6) Writing the shear stress at the pile shaft, r = Tor a8 T,, Eq. 6 may be integrated to give: assumption is that there is Some magical radius, +, at which the shear stress es negligible [Cooke (1974), Frank (1975), and, thus mam (2) ape BEER yn Sei amcta tx. « (9) G ry G in which { = In (7, /r,), . ve Effect of Lower Layer of Soil on Upper Layer.—The lower layer of soil sore eee Tne, ane F the pile acting tn 2 rgd ence following)." At;GT12 VERTICALLY LOADED PILES 1471 latter inequality implies that the shear stress will decrease more rapidly with r than given by Eq. 7. The magaitude of the term 4/dr(rt) will decrease with vertical distance above A,B, (Fig. 3), so that the value of +, (at which the Shear stress becomes negligible) will vary with depth and will follow the pattern shown in Fig. 5, Since, for a rigid pile, w, must be independent of depth, the value of +, must also vary with depth in such a fashion that (ir. /G) In (7, /r,) is constant, This variation has been found [Butterfield and Banerjee (1970, 1971), Frank (1975) Ottaviani (1975), Poulos (1968)] and typical results from the writers’ own finite element computations are shown in Fig. 6, for a pile of slenderness ratio 40, and soil having v = 0.4, SEE eat veo: —— impret equstion enatyaie + SEB Hen) FIG. 6.—Distribution of Shear Stross FIG. 7.—Settlement Profile at Mid- down Pile Surface (Sudden increase near Depth of Pile ; Free Surtace Is irregularity of Finite Ele- 2 ment Method) shear modulus are ‘Proportional to I-— vy: Thus, the: stresses set-up by the in compatibility of the displacement patterns: along AB (Fig. 3) will vary in. some fashion inversely with 1 — vy, At low values of Poisson's ratio v, the - ‘term 30, /2z will be smaller at a given radius than for higher values of Poisson tatio. This will lead to the effects of the pile loading being felt"at greater472 + (DECEMBER -1978 eri wean value of r_.will-siso be proportional to the length of the pile; thus, :r, igen ee ; a eament ot 2 edi, r, may be expressed Grom Eas. 9 and 10) 2 a iadnenmtes ay 9 aeGe Med “eS 7 . ‘This settlement profile is compared to that calcellated from’ integral equation” “analysis at the pile mid-depth in Fix. 7, for the case of a pile having i/r, = ee for two values Of v. The valve of © affects the value of rand thus ve, a. 10). ‘The logarithmic variation of settlement given by Eq. Ll is a good approximation for small r. Similar agreement as been found by Baguelin, ct © al. (1975) and Frank (1975). The magaitade of the settlement, w, computed from Ea. ant ends to zero for r = 52 7, for the case of v = 0.5 and for Oe eg, the verte of, vadivs ist, = 2510 — v)- In the analysis following, the variation of , rating wil be igncrod and = single averkgs vies 07s Viper Layer = De cen CF aa i ee ores ockormamice ear a pits 1s solic to reid punch, amv se ore, i Ine Geformation oad behavior is described by [Timoshenko and Goodier (1970): wb PH) 2. an La Wane ane sane Oe ee ae wee eee OD in which a factor 7 has been introduced to allow for the depth of the pile” tas below the surface, ie., the interaction of the upper layer of soil with er layer. i ie oe much debate and confusion over the. coefficient y. This ficient, Of depth factor, was introduced to modify the original solution that applies to 8 punch at the surface of an elastic half space, in order to take secount of the stiffening effect of the soil above the level of the loaded area. Since @ closed-form solution is no longer possible for the general case, correction factors must be used, and the original figures were due to Fox (1948) who considered & Joaded area embedded in a soil mass at a depth, h. He showed, and his work has been substantiated by Banerjee (1970), that for h/d > 6, in which dis the diameter of the Joaded ared, a factor of 1/2 should be oil Clearly, at large depths this is a limiting value—from the symmetry of the problem. | For plate loading tests conducted at the bottom of an open borehole, this | factor is markedly different. Burland (1969) showed by_fi inite element analyses. that the limiting value was nearer 0.85 [used by Marsland (1971) in London clay with v = 0.4], though it depended on Poisson’s ratio. This figure of 0.85, instead of 0.5, accounts for much of the discrepancy between the dimensionless number K (the ratio of settlement as a fraction of plate diameter, to fraction \ of ultimate load) evaluated earlier by Burland, et al. (1966) from plate tests | and from data obtained from triaxial tests. When the loaded area is the base of a pile, it has been customary to revert to Fox's figure of 4 = 0.5 {Burland, et al. (1965), Burland and Cooke (1974] since the joaded area is NOW buricd—in the sense that there is no longer an open borehole above it. However, this cannot be correct, because the pile isGT12 -. VERTICALLY LOADED PILES -- --+ 4473 not capable of absorbing any of the load applied at its base (as would be a column of soil) since it is the means by which the load is transmitted to the base. Furthermore, the upper layer of soil is already being deformed by the action of the shear stresses along the pile shaft. Thus, at any rate, for a straight (ie., not underreamed) pile, 1 > 0.85 and is probably close to unity.’ Most finite element and other elastic numerical analyses give ratios of base settlement to load consistent with a factor much greater than unity, together with high computed shear stresses on the pile shaft near the base. If these shear stresses -are included in the load acting on the base (in a real soil, plastic strains-would reduce the peak in shear stress), then values of y of about unity are obtained. Frank (1975) gives typical values of this factor, taking P,as all the load transferred * to the soil within half a diameter of the pile base. Even with this broad definition of P,, 7 is about 0.75. s The factor 1 may need to be modified for dealing with load increments after the shaft friction has been fully mobilized and further load is being taken solely by the pile base, The base then acts like a plate at the bottom of a borehole, and so y should be reduced to approx 0.85 r,/r, for increments in this range of loading. Compa Bast ano Suart Transren of Loan By combining the results of the last two sections, it is possible to produce a reasonable estimate for the deformation of a rigid pile in an homogeneous linearly elastic continuum, which has been checked against numerical solutions. ‘The purpose of such an estimate is to provide a simple basis that may then be modified to account for nonhomogeneity of a particular soil deposit. Since only the overall load settlement ratio of the pile shaft is important, it is sufficient, as a first approximation, to take r,, and thus, for a rigid pile, 7,, a8 constant with depth. Eq. 10 may thus be written as tora OP, tees oD COE? Scien & 3 13 G 2nIG on w,= For a rigid pile, w, = w, = w, and P, = P, + P,; thus, making use of Eqs. 12 and 13: P, P, P, 4 Wl + ee Re ai Be oe O14 Grew, Grew, Grew, ali-v) Fr an This may be compared with the results of integral equation analyses. From the analysis in the previous section, y = 1 and { may be calculated from ( = In (r,/r,). AS considered previously, an empirical value of r,, has been determined as r,, = 2.5 1 (1 — v). The variation of { with //r, and v is shown in Fig. 8. Using these values, the load settlement ratio has been computed for a range of slendemess ratio, !/r,, and two different values of Poisson’s ratio, v = O and v = 0.5. The results are plotted in Fig. 9 and compare well with the results of integral equation analyses. The fraction of the load taken by the base is20 arayaes Os © Reuter (ition emo, © Eeuatns Hoe 0 5 FIG. 8.—Variation of { = In (r_ /r,) with FIG. $.—Compsrison of Load Settle- Slenderness Ratio //r, and Poisson's ment Ratios for Different Slenderness Ratio v Ratios of Piles * For example, for a pile of slenderness ratio //r, = 40, P,/P, varies from 7% ( = 0+-11% (v = 0.5). Pus Compnessmary It is necessary to modify the preceding analysis in order to estimate the effect of pile compressibility on the settlement of the pile head. Eq. 10 may be written as L TAZ) 1, . et ms eee = oie Shey RGSS Gale 4 Howece @ saws 16) w(z) = = as) in which w and +, now vary with depth z, The compressive strain of the pile (assumed elastic) is written in terms of the load Fe) transmitted by the pile at any depth z: aw @) _ =P@) » ~P@) ‘ =. ae 7 — SeGag SHS epee ape i a Ses E in which E, is the Young’s modulus st the pies ‘ted = “E, /G,. This’ will : be related t0'the shear stress on thé shaft surface, 4), bY : aP(z) oz mm dn ner (B)ee Seu < bi5ik Bd Coterfeunlete Bau xe | tice Differentiating Eq. 17 and substituting Eq. 18 gives:a wiz) 1 < aet2KG, az rAG, Finally, the relationship between 7, (2) and w(z) (Eq. 16) may be used to form the governing differential equation: a? w(z) 2 rth | The solution of this equation is WEE) = Ae Ben gee ke a Hee He es (21) in which (ul)? = [2/0] @/r.)- The constants A and B may be found by substituting the boundary conditions at the base of the pile, in which WEY seasqnGd RSd PORES sate ee Teg (20) wi) = PY) BAD pense LE RE OER (22) 47,G, and (=) = a (omg: 1) sac wea eeees (23) az ar, G, _ Thus Ae Beta Eg Nee RM eee e Roe ceerae need (24) and p(Ae“ - Be) = aa coosugtatangante eiwii-d HERE SEN 5) ar,iG, > Solving for A and B and substituting back into Eq. 21 yields ¢ 1 i i wipe te ([8to ene 27,6, 4 "Tr kB ai - v) 1 wal shes ol CaS a TS Faw} wines Ree 8 (26) | Since the term (wr, Ap)7* will be very small (<0.02) for relevant values of ty, Eq. 26 may be simplified to give . : P, n(l-» ee ee Bd-2)ps ene. QD e Ve In particular, the settlement of the pile top may be expressed! in forms of the settlement of the pile base by ye wemhQn sue (aice shia acanDa. (eae oa ube oe. Q8) :7he toial load supported by the. pile maybe obtiined by integrating Bq. It 18, Se ae era a. 2aP, po tr. 4p PG) = sink [n (- 2 + cos ean} a 29)1476 DECEMBER 1978 arg Thus, the expression for P,/G,r,w, becomes: P, [ 4 = G,r,w, nl —v) ; i lee li. 4 1 7 tanhQ@s) ]-' a3 tr pl nQl-v)akr, pl Se el ag This expression is simpler to apply than it appears at first sight since the group] @/r,) [tanh (4.)/ 1) occurs in both the numerator and the denominator. - This result has been used to investigate the manner in which the compressibility J of. a pile affects its load deformation behavior. Some results are presented in Table 1, in which values of the load settlement ratio and the ratio of tip 3 settlement to base settlement, w, /w,, are tabulated for two different slenderness ratios and a range of stiffness ratios, FE, /G,. Corresponding results from integral i % TABLE 1.—Comparison of Effect of Pile Compressibllity as Computed by Integral Equation Analysis and Eqs. 28 and 30 it Eq. 28 P/(G,r.,) | Integral equation analysis Eq. 30 Note: Poisson's ratio for the soil is y = 0.4, equation analyses are also given for comparison. Note that hoth methods become: _ unstable for long compressible piles, and the values given for //r, = 80, E,/G, = 300 are suspect. ConswwerATION OF Soit INHOMOGENEITY The analysis so far has dealt with a pile embedded in a homogeneous soil. In practice, the stiffness of the undisturbed soil usually increases with depth and, particularly for bored piles, there may be a softened annular zone of material around the pile shaft. Radial Inhomogeneity.—For the case of radial inhomogeneity, Eq. 9 may be rewritten as [see Banguelin, et al. (1975) and Frank (1975)] Lae W,= 7,7, Sree Sistine, asia Feel lee, viesueiecey War de=sn Senor GiGT12 VERTICALLY LOADED PILES 1477 in which G = G(r). If an estimate of the variation of G with radius is possible, thea the foregoing analysis may be applied but with a different value of [ given by: Fannsglpibe ds 4 ¢ aa.| gaia, GigE Snel Ae Seer ee ee ae ee .~.. G2) Gr jn which G, is the undisturbed value of G(r) at some distance from the pile. In order to check ‘the validity of this approach, a group of finite element. analyses was undertaken in'which the geometry was kept constant and Poisson’s ratio was taken as 0.4, but the soil modulus was varied as follows: (1) Ge G_ (constant) throughout the soil mass; (2) G = G,/4 for 1 1.25; and G) G = G,/4 for 1 < r/r, = 1.25, G = G, for r/r, > 2.0, and G varied linearly from G, /4 to G_, 1.25 < r/r, = 2.0. Expressions for the shaft settlement, calculated from Eq. 3} are: fa Sofa Q3a w, a, See am a wae nce ecemeieie ue ) w= (sia125 +n“) cee wdew, acme nomen HERES PERS (33) Toho fl w= (stots +02) Sees . ow ge ntise di RR 3c) - Urs The value of r,, was taken as r,, = 2/(1 — v). This is smaller than the value of r,, previously used because these results were compared with finite element analyses, in which there is a rigid boundary at a depth, A, where A = 2.51. This rigid layer has the effect of increasing the vertical stress changes, ¢,, thereby decreasing the radius at which the shear stresses become negligible. The load settlement ratios computed by the two methods are given’ in Table 2. This limited number of comparisons is sufficient to show that the simple analysis outlined previously is capable of predicting the effect of a softened zone around a pile. Vertical Inhomogeneity.—Almost all soils shown some variation of stiffness with depth and, although it may sometimes be pcssible to assign a single value of shear modulus to the whole deposit, it is essential that an analytic model is capable of dealing with vertical inhomogeneity. It is possible to account for simple cases of inhomogeneity, where there are two or three distinct layers of soil, each one being homogeneous, by assuming that the shear strain distribution is unaffected. The value of r, at any depth may then be established by multiplying by the appropriate shear modulus [see Frank (1975)}. A more important type of vertical inhomogeneity is where the stiffness of a single soil type increases gradually with depth. It will be assumed here that the stiffness increases proportionally as in a “‘Gibson”’ soil (Gibson (1967, 1974)] or linearly with depth. Thus the shear modulus may be written as G=m(b +2)1478 DECEMBER 1978 gives rise to stress increments o, along AB (Fig. 3). The size of these will depend on the value of the soil stiffness in that region. If the case 8 Gibson soil is considered, whose modulus at this depth is the same as thaf® of a previously analyzed but homogeneous deposit, then the terms o, and aq, .f02) will be uachanged from the homogeneous case. However, for uniform settlement} wiidoie Ta kcdiown ik tias 10 for the case of a pile with sleademens ratio of 40, in a soil with v = 0.4. The effect of the variation in shear Note: Predictions are given by P,/(G_r,w,) = [4/(1 — vy] + (2=/O/r,), in which {is calculated from"Eqs. 334, 335, and 33c, written in the form: w, = {7,7,/G.- nx Sent Gent Bane Redo 2 ao n04 “ys stew mocks Be deo 8 Oy 7 vomoganacus voi — FS. enabysst ’ oe [a 66)] FIG. 10.—Distribution of Shear down FIG. 11.—Comparison of Settlement Pile Shaft, in “Gibson” Soil from Finite Patterns at Mid-Depth of Pile of Slen- Element Analysis derness Ratio 40, for Homogensous Soil and “Gibson” Soil, Both with v = 0.4 is to decrease the relative size of the shear stresses compared to da, /dz. _. From Eq. § it may be seen that this will entail a more rapid decrease of shear stress with r than the corresponding homogeneous case and so the value” of r,, (at which the shear stresses become negligible) will decrease. At the midpoint of the pile, where the shear stress is of the order of half that for the homogeneous case, it is reasonable to expect that r,, will be half its previous value. This reduction in 7,, between a homogeneous soil deposit and a Gibson soil is shown in Fig. 11. The logarithmic variation of settlement- with tadius is a good approximation for small r in both cases. Applying Eq.. GTI2 > VERTICALLY LOADED PILES wee u in the same manner as for Fig. 7 (sce section on “* ysis of Piles by . Separation of Loads Carried by Shaft and Base’’) estimates for r,, of 48 Fr. and 24 r, are obtained. This confirms the supposition that r,, is reduced by a factor of two for a pile in a Gibson soil. ” In general, introducing an inhomogeneity factor, p, which is the ratio of the shear modulus at the pile mid-depth to that at the base, ie., p = G(I/2)/GO)s the new expression for r,, is pee PSECU Ww eager ens EME ES we ie TE npg ao daaix (35) = for a pile in an infmite half space, or . ADIT og ASN Tews ene AED IN ors (36) for comparing with finite clement analyses where there is a rigid layer at a depth of 2.5 1, Note that, for a given soil, p will vary with the length of the pile except for the extreme cases of a homogeneous soil (p = 1) and a Gibson soil (p = 0.5). ‘ ‘The finite élemeat method was used to analyze a group of problems featuring vertical inhomogeneity. All analyses had the same (homogeneous) value of shear modulus below the level of the pile base, in order that the base behavior would not be affected..In fact, the effect of G continuing to increase with depth below the base of the pile is very small as was shown by: two finite element analyses—one with G = mz throughout and the other with G = mz for 0 i, where the overall settlements for a given load differed by less than 5%. This constant value of G at z >.1 will be written as G,and the results expressed in terms of the nondimensional parameter P,/G, T. W,): Following the arguments previously outlined, if G is of the form G = m(b + 2), in which m(6 + 1) = G,, then the distribution of shear stress will be similar, i.€.: aGVREO 42) co cee errs es rere reste ee ee (7) k rz k from which. w,=—r in =—rb ieee semen wal FANE Soke ee (38) m Te 7 The total load taken by the shaft is P, given by t I a o={ aersnden derbi (64>) ee ltyese QS SRA When eet (39) o P, Qa lt Shes, eB ae RG SH ew tee 40) Grew, 0 Fe This expression may be substituted into Eq. 14 to give the overall load-settlement ratio, P,/(Gr. ™,)- Results for different values of p for 9 pile where Ur, = 40, are compared with those from the finite element analyses in Table 3 and show good agreement. Tn addition, these results show that, ifasingle homogeneous shear modulus is to be chosen for a soil where G increase with depth, then the value of G at ~2/3 of the depth of the pile base gives a settlement equivalent to the correct analysis. Although the analysis for a compressible pile in @ homogeneous medium is1480 DECEMBER 1878 GTIZ i no longer applicable to the nonhomogeneous case, it has been found that the n variation of settlement down the pile may still be approximated by Eq. 27.. Thus, Eq. 30 will still hold except that the shaft load will be reduced by the .. factor p, as shown by Eq. 40. The general expression for the load settlement '®: ratio P,/(G,r, w,) is now P, [ 4 Grew, La(l-») 20 iT ame | 4 1 f tanh(id I 1 eee tor ul al —¥) war, : pl in which { = In ¢, mf?) = In (2.5(0/r,) o(1 — vhs (iy = BAGEL r, ; e = G(I/2)/G(; and y= 1. TABLE 3.—Comparison of Computed Settlement Ratios for Vertical Soil inhomo- yi geneity. Degree of vertical inhomogeneity, 0 = 1Gd/2/G) joe geneous) P,/(G,r,w,) | Finite element analysis = imple analyses (see Note) i Note: Predictions are given by P,/(G,r,w,) = [4/(1 — ¥)] + (P,/(G,r,w,)]. given by Eq..40. .. Additional radial softening close to the pile may be accounted for by suitable modification to { as indicated by Eq. 32. Summary of Anacynicat Move: ron Sinate Pies The first point to emerge from the analysis concerns the settlement of a ‘pile that is necessary to develop full shaft friction. Early research into pile behavior assumed that the shear strains around a pile shaft were confined to a narrow softened zone close to the pile, of the order of 5 cm-10 cm (2 in.-4 in.) wide. Full cohesion was assumed to be developed at shear strains of ~10% [Burland, et al. (1966)) giving corresponding settlements of the pile of 5 mm-10 mm (0.2 in.—0.4 in.), independent of the pile shaft radius. For bored piles, where a softened zone independent of the pile radius may occur, this may be a reasonable approach. For most piles, however, the soil may be taken, in the first instance, as radially homogeneous and dimensional analysis demonstrates the dependence of this critical settlement on the pile radius, Eq. 3 shows the proportional dependence of w/r, on P/(Gr?2). For 4 particular pile geometry, (//r,), the later term (ignoring the contribution to P from the pile base} is a measure of the shear strain in the soil adjacent to the pile shaft. Thus, the settlement needed to produce a particular value of this shear strain is proportional to the pile radius. As shaft failure is approached, the high stress level near theGTI2 VERTICALLY LOADED PILES __ 7 1481 shaft reduces the secant shear modulus and thus, in effect, a softened zone is formed. This is, superficially similar to the bored pile case except for the important difference that the width of this softened zone is a function of the shear stress level and so (from Eq. 7) is proportional to the pile shaft radius. Thus the settlement of a pile needed to develop full, shaft cohesion will, in general, be approximately proportional to the pile radius. Some experimental | evidence to support this has been given by Whitaker and Cooke (1966). It is not immediately obvious which are the important parameters in Eq. ‘Al for determining the settlement of a pile at a given load. In order to investigate th, the expression will be simplified by ignoring the second term in the denominator (usually <0.1) and to compensate by ignoring the contribution of the pile-base represented by the first term, 4/[q( — v)], in the numerator. This gives: P, 2Qe- 1 tanh(pl) Grow & % wl which may be rewritten for design purposes to give the settlement explicitly as: . This simplified expression tends to overestimate the settlement of a pile by up to 10%. The soil properties that need to be measured are the value of G and its variation with depth and, for the long-term settlement of a pile, v. Most soil tests give a quantitative estimate of the variation of soil strength and stiffness with depth from which, for a given pile length, p may be calculated with reasonable accuracy. The difference between undrained and drained pile settlement may be obtained by allowing Poisson’s ratio, v, to vary inthe expression for [, assuming that the shear modulus is unaffected. For most piles, approx 85% of the settlement may be expected to take place immediately. There is no simple approximation for tanh (LD)/pl, but Fig. 12 shows the variation with pd. Inaccuracies in estimating u! will arise mainly from the ratio . = E,/G,. The value of E, will be known, but G, must be estimated from soil data. For most practical values of ul, in which 0.5 < pl < 2.0, the variation in G, tanh (w/)/p! will be approx two-thirds the variation in G,. Thus, an underestimate in G, of 30% will lead to an overestimate of the pile settlement of about 20%. It is clear that accurate measurement of the soil shear modulus is of prime importance in estimating the settlement of a piled foundation. AppucaTion of Anavymicat Moet to Pit Tests The proposed analytical model may be used to back analyze values of the soil shear modulus from pile tests. If this parameter is known with reasonable accuracy from independent soil tests, then any discrepancy will give 2 measure of the effects of pile installation that may be used later in the design or the actual foundation. For the purposes of illustration, a pile test in London clay will be studied and the values of the shear modulus of the soil obtained by148200~«C DECEMBER 1878 et back analysis will be compared with the results from plate loading tests pressuremeter tests. sn The test to be considered is one of the tests conducted on mini piles, Cooke, et al. (1973, 1974) at Hendon, North London. The pile [f= 3.5 (11.5 £1), r, = 0.084 m (3.3 in.), I/r, = 41.7] was fabricated out of tu steel with a wall thickness of 6.4 mm (0.25 in.) [see Fig. 13(a)}. It was jacked into the ground and subsequently loaded to just over half the maximum ‘load 02 0408 08 10 +7 Se ¥8 20 22 2426 2630 ¥! FIG. 12.—Variation of tanh (j:2)/(u2) with pi ie Shae 2 2 een Les > Set bering pra nnwremter tenis 2 toeoe 0 “ pe (1 won Ateon FIG. 13.—Details of Pile Test by Cooke and Price (1973) (1 KN = 0.112 tons: 1 m = 3.28 ft: 1 MN/m? = 145 psi) measured during the jacking process. The eee Young's modulus of the pile is E, = Eyoa—> = 30,800 MN/m? (4.5% 10% psi)... ee (44) ° in which A, = the net cross-sectional area of the pile. The results of the loading test have been reported by Cooke and Price (1973) and the relevant details are, P, = 36.4KN (4.09 tons); w, = 0.516 mm (0.0263 in.);GT12 VERTICALLY LOADED PILES =e - 1483 w, = 0.383 mm (0.0151 in.); - . Se se 3 EN/mm (201 tons/im) 0.0606 ee eee (45) wy, i . : Little drainage will have taken place during loading of the pile, so Poisson's ratio for the soil may be assumed to be close to 0.5. From the ratio of w,/w,, Eq. 28 gives a value of »J of 0.81, from which tanh (ul)/p! = 0.83. ‘The simplified expression given by, Eq. 42 may now be used, in the form _ FC, ROD ah de welewwrmera Been wf al : The product G,p is the average value of the soil shear modulus down the length of the pile. The value of { = In (r,,/r,) = In [2.5 (t/r,) p(l — v)] may vary from 3.26 (for p =. 0.5) up to 3.95 (for p = 1.0) with an average of 3.61. Thus, for this site the shear modulus at mid-depih is o(5)- 13.9 MN/m? (2.01 x 10? psi) £108... eee eee «47 From, the value for pi of 0.81 (A = 5,300. For { = 3.61, p = 0.71, and G, = 13.9/0.71 = 19.6 MN/m? (2.84 x 10° psi). This gives A = E,/G, = 1,570, which, in turn, gives a mew estimate for { of 3.38. This iteration may be repeated. until the most consistent set of values is reached. These are G, = 20.6 MN/m7@2.99 x 10? psi), p = 0.66, A =1,500......... (48) If the more accurate expression of Eq. 41 is used, then the new set of soil parametersis = G, = 19.8 MN /m? (2.87 %10' psi); p= 0.58; A=1555.....6- .. (49) Marsland (1971) has conducted large diameter plate tests at this site and, although the effects of anisotropy have still to be investigated, has deduced values of the shear modulus of Lendon clay assuming isotropic conditions. The shallowest test was at 6 m (20 ft) where shear moduli of the order of 25 MN/m’? (3.6 x 10° psi) were measured. Below this depth, an increase in stiffness with depth is recorded and thus, it is reasonable to expect the shear modulus at 3-m (10-ft) depth to be ~ 20 MN/m? (2.9 x 10° psi). The value of p = 0.58 indicates that the shear modulus decreases rapidly near the ground level. Windle and the second writer (1977) report the results of self-boring pres- suremeter tests on the same site. Values of the shear modulus in the horizontal plane were measured, which varied from 14 MN/m? (2 x 10° psi) at a 2-m (6.5-ft) depth up to an average of 25 MN jm? (3.6 x 10° psi) at 4 m (13 ft). These figures agree well with those deduced from the pile test results [Fig. 13(5)}. Cooke and Price (1973) take a value of p = 0.75, but assume that the pile installation process has softened the soil next to the pile. At mid-depth, they take a radial distribution of shear modulus varying from 3.75 MN/m? (0.54 x 10° psi} to 37 MN/m? (5.4 x 10? psi) as the radius increases from r, to 18 r,. This compares with a figure for G{!/2) of 19.8 x 0.58 = 11.5 MN/m? (1.67 x 10° psi) estimated previously, assuming radial homogeneity. It is not14840 DECEMBER 1978 stig clear how pile driving affects the stiffness of the surrounding soil, but it seems 4g: possible that any softening duc to remolding may be balanced, after consolidation, by the resulting increase in effective stress level. ‘The pile test analyzed previously is, to some extent, unique in that the settlement = of the pile base was measured as well as the settlement at the top of the ; pile. This additional information enabled the variation of the soil shear modulus with depth to be estimated. To analyze a pile test where-only the load and = settlement at the pile head are measured, Eq. 46 may be used to provide an‘ estimate of the average shéar modulus, G,p, within approx 10% since the variation of (1/%) [tanh (n1)/(u1)] with p is small. Arrucanion of Anatyncat Moos. ro Pu Desran Design of a piled foundation is facilitated if design charts covering a wide - range of pile geometries are prepared. Burland and Cooke (1974) illustrated . the preparation of such charts for underreamed piles im London Clay. Each” © chart related to a’ single pile shaft diameter and showed the load settlement variation for a renge of pile base diameters. The analysis presented in this paper may be used to advantage in the preparation of similar design charts, To illustrate the procedure, charts have been prepared for straight shafted piles in two soft clay deposits whose stiffness profiles have been measured using a self-boring pressuremeter. For any given pile stiffness and profile of soil stiffness, Eq. 41 may be used to calculate the ratio P,/w, for a range of possible pile lengths / and radii r,. Contours of equal values of P,/w, may then be drawn for the chosen range of pile geometries. = Fig. 14(a) shows a design chart for piles in the soft clay of Canvey Island. Values of the soil shear modulus have becn measured by Pender (1974) and” the stiffness profile may be approximated as G = 0.9 z MN/m? (p = 0.5), in which z is in meters (G = 40z psi, where z is in feet), and v = 0.5. The pile stiffness has been taken as E, = 20,000 MN/m? (2.9 x 10° psi); this figure is typical for a concrete pile, or a tubular steel pile where the ratio < of net pile area to gross pile area is 10%. The chart shows curves of equal values of P,/w, for a range of pile geometries. In most design cases, the pile load and maximum pernissible settlement is known and thus the chart may be used to provide suitable values of Jand r,. Fig. 14(b) shows a similar design chart for Boston Blue Clay, where following self-boring pressuremeter tests conducted by Hughes (1974), the soil stiffness has been taken as G = 9.0 + 0.05 : MN/m? and z is measured in meters (v = 0.5) (G = 1.3 x 10’ + 2.2 z psi, in which z is measured in feet). Note that checks musi be made to ensure the overall stability of the final design. Eq. 42 may be rewritten as w, zz 1 tpl r, wir, G,p tanh (wl) The first two terms on the right-hand side represent the average shear stress and average shear contpliance down the pile and thus (ignoring the load taken by the pile base) an estimate of the settlement as a fraction of the pile radius may be obtained fromGT12 VERTICALLY LOADED PILES 1485 wfc) tl , : (@) =. Si 4 sam Pie eur EES & HAS Bory 2 on Typical values of G/c, in which c, is the undrained shear strength range from 100-200 and, for a factor of safety of two on the shaft stress, typical values for w, to mobilize full shaft friction are such that 0.0] < w,/r, < 0.04, substituting appropriate values for { and wi. An interesting feature of Fig. 14 is that the curves of constant P,/w, tend to be parallel to the “‘length”’ axis for small pile radii. This implies an inefficient area of design whereby little decrease in working settlement is gained by increasing = om) fa) Design chert for clay with G = 092 MNjin? Ep = 20,000 MNjm? Curves represent constant vata ot 8 ia kfm dle) {b) Design chart for clay with G = 90 + 0-052 MN/m? FIG. 14—Example of Design Charts for Piles in Soft Clay (1 MN/m? = 1% pai; 1m = 3.28 ft; 1 kN/mm = 2.86 tons/in.) the pile length. A dashed line has been drawn that defines the lower limit - of efficient design for these particular pile and soil stiffnesses. Conc.usions The analysis outlined in this paper has been developed in an attempt to model the manner in which a pile transfers load to the soil. Agreement between the results of the analysis and solutions obtained from numerical methods that need & computer is very good, being within the accuracy of these latter methods themselves. Two points that emerge from the analysis need to be stressed again since they have often been misunderstood in the past: