Download as pdf
Download as pdf
You are on page 1of 85
REPORT 103 Design of laterally-loaded piles W.K. ELSON PhD CEng MICE “The project leading to this Report formed part ofthe joint Building Research Establishment (BRE) and CIRIA programme of research on pil tract to CIRIA at Brighton Polytechnic whe Padfield and Mr ling. The work was carried out under con- ‘ere Dr Elson isa Senior Lecturer. DrC. J. ren fock, both of Seatt Wilson Kirkpatrick & Partners, contributed 10 the Report on editorial and technical aspects, respectively Following CIRIA’s usual practice, a Steerin project, advised on the content of the Report and ‘comprise P. A. Green BSc DIC ACGI CEng FICE FGS (Chairman) |W. G. K, Fleming PhD BSe CEng MICE R.M.C, Driscoll BSe CEng MICE B. A. Leach BSc(Eng) CEng FICE Fistruet MASCE, W.1.J, Price MSe(Eng) CEng MICE FIHE D. Tobutt CEng MIStructE, J. M. Head MSc CEng MIMM FGS M. F. Randolph PhD 1g Group was established whi the validity ofits inform: Scott Wilson. spatrick & Partners ‘Cementation Piling and Foundations ‘Limited Building Research Establishment Allott & Lomax Gitford & Partners ‘Transport and Road Research Laboratory caRIA University of Cambridge {ARIA Report 03 Contents CURIA Report 03 LIST OF ILLUSTRATIONS LIST OF TABLES NOTATION ‘SUMMARY 4. INTRODUCTION 2, LATERALLY-LOADED PILE DESIGN 2.1 General comments 2.2. Design of pile groups 3. RECOMMENDED ANALYTICAL METHOD FOR THE DESIGN OF LATERALLY-LOADED PILES 3.1 Ultimate lateral resistance of piles 3.2. Deformation of single pl 3.3 Deformation of pile groups 4, ASSESSMENT OF SOIL PROPERTIES 4.1 Modulus of elasticity 42. Cosfficient of subgrade reaction 43. Discussion of stiffness parameters 5, RESULTS AND CONCLUSIONS 5.1 Single piles 5.2 Pile groups REFERENCES APPENDIXA Ultimate fateral resistance of single piles and pile groups APPENDIXB Analysis of lateral displacement of single piles APPENDIXC Analysis of pile groups APPENDIXD Case histories APPENDIXE Computer program references Page 10 10 " W 2 6 5 15 16 18 8 19 23 24 24 24 List of illustrations Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8. Figured Figure 10 Figure 11 Figure 15 Figure 16 Figure 17 Figure 18 Figure 19 Figure 20 Figure 21 Figure 22 Figure 23 Figure 24 ‘Common occurrences of incidental lateral loading for bridge foundations and substructures Flow chart for design of pile group Coefficient of subgrade reaction for sands as a function of the relative density Earth pressure coefficients for overburden pressure and cohesion Deflection, soil reaction and bending moment distribution for a long free-headed pile in cohesive soil Ultimate lateral resistance of short piles in cohesive soil Ultimate lateral resistance of long piles in cohesive soil Deflection, soil reaction and bending moment distribution for @ long restrained pile in cohesionless soil lateral resistance of short piles in cohesiontess soil lateral resistance of long piles in cohesionless soil Foundation response models for elastic continuum and Winkler spring medium Soil/pile interaction model used in Winkler spring idealisation Lateral deflection at ground surface of laterally-loaded pile in cohesive soil Effect of a superficial crust, definition of parameters Influence of crust thickness on deflection at ground surface of free-headed pile subjected to lateral load Influence of crust thickness on deflection at ground surface of free-headed pile subjected to moment loading Influence of layer coefficient on deflection at ground surface of fixed-headed pile subjected to lateral load Deflection coefficient for # range of fixity conditions for a pile subjected to moment and lateral loading in na1--homogeneous soil with modulus increasing linearly with depti: ‘Schematic representation of p-y curves and their relation to the geometry of pile displacement Values of deformation factor, ly, for free-headed pile in homo- geneous soil Values of deformation factor, huy, for free-headed pile in homo- geneous soil Values of deformation factor, Iw, for fr geneous soil Values of deformation factor, 1 headed pile in homo- for tree-headed pile in non- non-homogeneous soil with modulus increasing linearly with depth ‘CIRIA Report 13 (ARIA Report Figure 25 Figure 26 Figure 27 Figure 28 Figure 29 Figure 30 Figure31 Figure 32 Figure 33, Figure 34 Figure 35 Figure 36 Figure 37 Figure 38 Figure 39 Figure 40, Figure 41 Figure 42 Figure 43 Figure 44 Figure 45 Figure 45 homogeneous soil with modulus increasing linearly with depth Volues of deformation factor, Ir, for fixed-headed pile in homo- geneous soil Values of deformation factor, I'rw, for fixed-headed pile in non-homogeneous soil with modulus increasing linearly with depth Generalised curves giving deflected pile shape and bending ‘moment profile for lateral force loading Generalised curves giving deflected pile shape and bending moment profile for moment loading Yield displacement factor for free-hee soil with constant yield resistance Yield rotation factor for free-headed pile in homogeneous soil with constant yield resistance Yield displacement factor for free-headed pile in non-homo- geneous soil with modulus and yield resistance increasing Enearly with depth Yield rotation factor for free-headed pile in non-homogeneous soil with modulus and yield resistance increasing linearly with depth Ultimate lateral resistance of unrestrained free-headed rigid piles Graphical determination of pile loads within laterally-loaded pile group ‘Schematic representation of static load distribution in pile group Definition of parameters for stiffness method of pile group analysis Extract from Navfac DM7 concerning pile group analysis Reduction factors for the lateral deflection of square pile groups in homogeneous soils Interaction factor, te, for groups of free-headed piles subjected to lateral load in homogeneous soils Interaction factors, ays and eeu, for free-headed piles subjected to ‘moment in homogeneous soils Interaction factor, ay, for free-headed piles subjected to moment in homogeneous soils Interaction factor, ays, for fixed-headed piles in homogeneous soils led pile in homogeneous Discretisation of pile-soil interfara used in PGROUP program ‘Newhaven overbridge: abutment details and borehole fog Newhaven overbridge: bridge dead loading List of tables Tablet Table2 Tables Table 4 Table5 Tables Table 7 Te Table 9 Table 10 Table 11 Table 12 Table 13, Table 14 Table 15 Table 16 Table 17 Table 18 Table 19 Table 20 Table 21 Table 22 ‘Summary of the output of method of analysis: Typical values of Young's modulus and Poisson's ratio for various soil types considered as homogeneous elastic solids Values of the coefficient of subgrade reaction for clays Values of the rate of increase of coefficient of subgrade reaction with depth for sands Values at the elastic limit of ground-line deflection Single pile: summary of calculated and measured deformation and moment Computed and measured axiel loads in piles at the end of construction, Newhaven overbridge Computed and measured deformation of pile cap, Newhaven overbridge Expressions for deflection coefficients A and B for a constant subgrade reaction Ground-line values of deflection coefficients A and B fora single pile in Winkler medium (tong piles, uniform soil) Deflection coefficients A and B for a pile in a Winkler medium (Zroax = 1.0) Deitection coefficients A and B fora pile in a Winkler medium (Zyrax = 1-5) Deflection coefficients A and B for a pile in a Winkler medium (Zmax = 2.0) Deflection coefficients A and B for a pile in a Winkler medium (Znax = 2.5) Deflection coefficients A and B for a pile in a Winkler medium Zrax = 3.0) Deflection coefficients A and B for a pile in a Winkler medium (Zax = 3.5) Deflection coefficients A and B for a pile in a Winkler medium (rex = 4.0) Deflection coefficients A and B for a pile in ® Winkler medium Zmax = 6.0) Deflection coefficients A and B for a pile in @ Winkler medium (Zax = 10.0) Reduction factors for group action Unit displacement of a laterally-loaded pile Group efficiency factors for lateral loading CURIA Report 03 Notation CRIA Report non-dimensional coefficients relating an applied lateral force to deflection, ‘slope, moment, shear and soil reaction, respectively ‘cross-sectional area of pile ‘geometric matrix breadth ofthe loaded area ‘non-dimensional coefficients relating an applied moment to deflection, slope, moment, shear and sol reaction, respectively non-dimensional coefficient giving the deflection of a pile, depending on the degree of fixity ‘undrained shear strength of soil ‘effective cohesion of sil pile diameter relative density of sand ‘Young's modulus of elasticity of sil initial Young's modulus of elasticity of the oil ‘Young's modulus of elasticity ofthe j* pile ‘Young's modulus of elasticity ofthe soil a base of pile ‘mean value of pressuremeter modulus of elasticity over characteristic length of pte ‘Young's modulus of elasticity at the surface of soil ‘Young's modulus of elasticity ofthe pile shat equivalent Young's modulus of elasticity of the pile ‘undrained Young's modulus of soil ‘Young's modulus of elasticity of silat depth 2 protrusion of pile above mudline in “flagstaff problem, ‘eccentricity within pile group force matrix deflection and rotation factors which allow for soil yield (&, Fi refer to ‘non-homogeneous soils) ‘non-dimensional pile group flexibility matrix components depth to plastic hinge over which there is a soil reaction from a laterally-loaded pile flexibility coefficient ‘hear modulus of soil modified shear modulus of sol (=G(1+3v/4)) ‘modified shear modulus of soil ata depth l/2 horizontal load ‘orizontal load on pile group hortzontal oad at ground line ultimate lateral load capacity ofa pile ‘ultimate lateral load capacity cf single pile acting within group moment of inertia of apile influence factor in vertical plas test dimensionless deformation factors (refers to non-homogeneous sols) ‘subscript denoting inital value subscripts denoting typical members of group shape factor coefficient of subgrade reaction related to pile (K=hyB) coefficient of active soll pressure POR ree RRRAIERETO GOS b 8 = pes po Re coefficient of earth pressure at rest ‘coefficient of passive sol pressure (Brinch Hansen's earth pressure coefficients) pile flexi secant value of the coefficient of subgrade reaction undrained value of K coefficient of subgrade reaction applied to a beam ‘Terzaghi's coefficient of horizontal subgrade reaction “Terzaghs coefficient oforizntl subgrade reaeon rated to 305mm square plate length of pile characteristic length of pile (1/8) equivalent pie length in bending and compression depth below ground level to rigid base layer Randolph's critica pile length bending moment ‘g7ound-line moment loading on pile yield moment of pile section coefficicat of increase of Young's modulus with depth (=dE/é2) coefficient of increase of G* with depth ‘Standard Penetration Test result (blows/300 mm) bearing capacity factor lateral bearing capacity factor fora pile 0 exponent, number of piles ina group rate of increase of coefficient of horizontal subgrade reaction with depth value of my at smal strain axial load ina pile loading matrix ‘vertical component of load on j* pile ‘ultimate lateral resistance of sol per unit length of pile soil reaction per anit length of pi vertical bearing pressure pressure on vertical plate test cone resistance from static (or Dutch) cone test reference radius ‘group reduction factor rats Rautas Raw: 8f0Up reduction factors with respect to lateral load and moment i T radius radius of pile or hole stiffness matrix pile-to-pile spacing characteristic length of pile for a homogeneous case - Es (nearly increasing subgrade modulus) [EE ceaumnipne wie Isteral deflection of a pile deflection ofa single pile under action of unit lateral load horizontal displacement of ple group ‘ARIA Report 3 CRIA Report 8 ground-line deflection of rile or pile group lateral displacement of pile at ground-line from unit horizental force limiting deformation of ple for which my applies total vertical load applied to pile group, shear acting on pile vertical deflection of a ple (group) or of vertical test plate non-dimensional depth parameter = 2/T (Zax = L/T) depth below ground level a rheological factor interaction factors for deformation and rotation caused by lateral load and ‘moment relative stiffness of apile = IL, = { pile Me ‘orientation of pile in plan with respect to direction of lateral load bulk unit weight of soil submerged unit weight of soil pile group displacement deflection matrix ‘group efficiency factor for lateral load at ground-tine Poisson's ratio homogeneity factor (Randolph) ‘major principal stress minor principal stresses vertical stress rotation ofthe pile group effective angle of shearing resistance ‘undrained angle of shearing resistance angle of rake ofa pile to vertical Summary “This Report reviews currently available methods for the analysis of laterallytoaded single piles and pile groups. The analytical methods discussed generally include lateral loading as 8 particular case, together with vertical and moment loading. Major cyclic loads represent a Special case. and they are not covered. The Report highlights the limitations imposed by the available methods, and it gives guidance on the practieal problem of assigning realistic values to the necessary soil parameters, particularly concentrating on the value of soil stifines. Recommendations are made concerning methods of analysis which may be satisfact adopted in specific circumstances. Vertical deformation ofthe piles is not com: Report. The recommended methods of analysis ae applied to selected case hi illustrate their several strengths and weaknesses. 1. Introduction 10 Lateral loads on piles urise from a variety of causes, and they range in importance from the major load componeit in such structures as transmission towers or mooring dolphins, toa relatively insignificant force in the foundations of buildings. Inthe latter case i isnot ienerally necessary to design the pile to resist Iateralloeds in a formal manner. because they Fave a large reserve of resistance, Some ofthe analytical methods reviewed inthis Report. ‘while being primarily concerned with laterally-loaded ples, may be applied to the general ‘case of apile group subject. 10 vertical, horizontal and moment loading. ‘Allowance for the considerable lateral restraint afforded even by soft alluvial deposits in the desiga ofa piled structure produces an economic design Itis also possible to assess the {interaction of the foundation with the sol o thatthe resulting deformations may be accommodated by the design of the superstructure, Piles constructed by any ofthe commonly available techniques” may be designed to resist tateral loads, provided they have the capacity to resist the applied bending moments, Instal- lation, whether by driving or boring, disturbs the soil around the pile and affects the lateral Toad:carrying characteristics toan extent whichis difficult o quantity. Iti, therefore. desi able, particularly on important projects, to undertake load tests on a new site wherever this is pracicable. The greatest need for such testing is in soft alluvial deposit and fill materials Of the available techniques forthe analysis of single pites and pile groups, among the more advanced and generally applicable are those based on elastic continuum mechanics. Although the assumption of an clastic sol is an idealisation, the predictions based on it are reasonably eaurate for working loads. Elastic continuum analyses have the further advantage that they prediet the interaction of ples within a group with reasonable success, and they enable the [ieigner to attain a better understanding ofthe lond/deformation behaviour of pile group. In spite of various theoretical reservations, the use of other, simpler methods may be justified some cases. ‘An excellent general reference covering virtually all aspects of pile foundation design and si available in the form of the recent textbook by Poulos and Davis CIRIA Regent 2. Laterally-loaded pile design Figure? Common occurrences of incidental fataral loading for bridge foundations ‘and substructures (Hambly") CIRIA Report 9 2.1 GENERAL COMMENTS: ‘Structures may be subjected to a range of lateral loads, which are more or less easily quan- tifed. The lateral loads ean be conveniently grouped into the following classes: ‘stacie —_structutal reactions, earth pressure, bollard pulls cyclic wave loading, reaction from rotating machinery, earthquake transient wind, braking, berthing, impact, earthquake ‘other _thermaleffects, creep and shrinkage, consolidation of surrounding soil. Static and transient lateral loads may be treated reasonably satisfactorily by the methods ‘outlined in this review. However, major cyeic loads presenta special problem, whichis tatside the scope ofthis Report, because this form of loading modifies the stress/strain ‘behaviour of the soil. Cyelie loading leads to deformations in excess of those resulting from the equivalent stati values, and it can lead to the progressive failure ofthe foundation. ‘Various forms of construction are capable of resisting horizontal loading: gravity foundations ret ing and diaphragm walls ‘ground anchors piles. “These forms may be used in combination to optimum effect to resist and, where possible, to minimise lateral forces. “Whichever foundation type is selected, whether on technical or economic grounds, the {global behaviour ofthe structure should be earefully considered. fn particular. the effect (see Figure 1) of vertical and lateral deformations ofthe soil induced by adjacent earthworks should be evaluated, nen abide moped i ‘Sitting ery ok leone Tonsartng dovmag anata iaShean ci ponent ent ean amen Ki apy tno atcpremcs onthe is n 22.1 Stability Figure2 Flow chart for design of pile group 2.2 DESIGN OF PILE GROUPS sv of a pile group isillustrated in Figure 2. Its suggested that three levels of ap- praisal should be adopted: 1. consideration of the ultimate failure mechanism of the foundation and incorporation of an overall reserve of strength for safety 2. computation ofthe lateral translation and rotation of the foundation at working loads, sand consideration ofthe effect of this deformation on the whole structure 3. bending resistance ofthe piles. Load factors applicable to the design ofa piled foundation subjected to substantial lateral toad are not well established, Selection of the appropriate factors depends on the type of loading, the reliability ofthe ground investigation data, and the response of the completed structure tothe deformation of the foundation. “The ultimate load is not normally a controlling factor in the design of laterally loaded pile ‘groups, and elementary methods of assessing the ultimate resistance may be adopted. ‘OUTLINE PILE GROUP Shelonespretngenemnent ot iret oing STATIC ANALYSIS TELECT METHOD OF DETAL"DANALYSIS Lgl ANALYSESIDATA Goer seeasie ste Siectdenen age of trata, uronetcn “ietpey operon Leaner ots “Tels coinaum mcibods TERETE ans mere Te SE ema "fg PGROUP-LAWFICE). foto hay mateo "APPRAISE PERFORMANCE OF FOUNDATION eee efanere omic sat era “pmetene MINOR STRUCTURES ‘Arpaepertormance. CURIA Repo 0 Inherent uncertainties in assessing the loads and stresses ina laterally-londed pile group equite that reasonably conservative overall safety factors should be adopted, combined with, Timits on deformation, The following tentative guide-lines are suggested forthe design of individual 1, an overall factor of safety for lateral load of not less than 3 mits on lateral deformation at the ground surface of not greater than 2% of the pile diameter for sands and stiff clays, and not greater than 5% of the pile diameter for soft clays, subject tothe tolerances unposed by the structure itselt “These criteria are conservative, and they should rest zones oa shallow depth. the development of plastic failure 22.2 Deformation Three conceptual models are currently in use for the design of pile groups: 1. Structural frame modet- Inthe static and stiffness methods, the pile are implicitly assumed to be end bearing on a competent layer, and the contribution of overlying soft ‘material to load capacity is entirely discounted. Computation of the forces inthe frame js earried out by conventional structural analysis”. Atthough this is not a realistic model of actual ste conditions, the method has given satistactory results, and it is still extensively used. The method is now principally useful fora preliminary appraisal ofthe Tayout of a pile group and for the design of lightly loaded groups. For an economic design of pile groups subjected to large lateral loads or moments, other forms of analysis, are preferred. 2, Spring idealisation Tn this method, the soils modelled hy an infinite number of discrete springs (Winkler medium). Transfer or shear stresses within the soil mass is not modelled. The method has been extended (by Matlock and Reese) to include non-linear springs, and itis generally referred to as the ‘p~ y"or the ‘subgrade reaction’ method. “The model is well developed, giving satisfactory predictions ofthe behaviour of single piles, although t should not generally be used to analyse the behaviour of large pile groups, because the interaction of one pile with another cannot be calculated. 43. Blastic continuum model~ An elastic continuum models useful forthe analysis of both single piles and pile groups when the soil can reasonably be assumed to be linearly ‘clastic: In practice, provided an appropriate secant modulus is selected, the method tives satistactory results for piles at working loadin most soil types. Complex elasto- plastic sil models for pile group analysis are not generally either available or necessary ‘Continuum analyses (e-g. by Poulos” " and by Bannerjee and Driscoll) incorporate pile/soi/pile interaction. The method usually involves the use of acom- puter, and the designer should be fully aware of the limitations of the particular program tsed, While the method is currently best suited tothe final design ofthe foundation. the publication of parametric studies" makes the method more generally applicable. Several programs are already available commercially, and itis anticipated that other comprehensive programs will be produced, to cater for most common practical problems. slisation ofan elastie continuum allows calculations to be performed which insight into the behaviout of the ple group, and which help to establish the Sensitivity of the group to changing loading conditions and material parameters “The application ofthese concepts is summarised in Table 1, together withthe output of each method. For special circumstances, more involved procedures may be used (e.g, the fi Clement technique), with correspondingly greater costs in data preparation and equation solution. However, finite element modelling of realistic sll stress-strain behaviour in three ‘dimensional situations i, as yet, better suited to research than design. “The designer should select the method appropriate tothe problem in hand, bearing in mind the complenity ofthe problem and the resources available. For the design of large piled foundations, an analysis based on the elastic continuum approach is consi" red to be the most satisfactory method available at present. ‘Analysis of the foundation at wor the computed deformations and t design ofthe structure. king load enables the designer to assess the significance of, include the stiffness matrix of the foundation in the overall CRIA Report 03 B Table 1 Summary of the output of method of analysis Model Limitations Application Output Structural frame ‘Unrealistic model (the soilis | End-bearing pile groups, | Axial ond on pites isthe ignored) with a small lateral load ‘only reasonable output ‘component (say up to 10% of the vertical load) Winkler medium ‘A reasonable model for “Any lateraliy-oaded single | Deflection. slope. moment or single piles. However pile or widely-spaced pile | and hear of he pile at any peyanalysis inappropriate for pile groups | group. The analysis can depth withs/D <8, because the | provide reasonable Continuity ofthe soil isnot | predictions for eyclic loading modelled ‘or account forthe development of plastic zones if suitable p-y data are selected Elastic continuum ‘A reasonable model for Single pilesor pile groups | Output dependson the ingle piles or pile groups at | under working loads particular program adopted. ‘working load. Yield of the ‘but typically includes soil cannot be included deflection. slope. moment exactly. The limitations shear and axial load depend on the mathematics Uistribution for each pile in of the particular computer the group, and the overall solution chosen. Available stiffness andior Nexiility programs are limited to matrix ofthe pile group constant or linearly increasing soil modulus with depth | 223. Practical Practical constraints on positioning and aligning piles within a group (particularly raking considerations piles) should also be considered athe design stage. Heavily-raked piles are dificul wo eietol_ and they may depart from the intended line by a considerable amount. Mis- Siignment of piles within a group usually resus in higher bending stresses. An analyse of» edi pile group by O'Neill, Ghazzaly and Hat™ indicated a 30% increase in ver aling moment ca inat computed forthe designed geometry, together with substantial moments on the minor axis ofthe piles. Individual experience and preferences play a patin choosing the overall size and arrange- sent ofthe pile group. The following comments may be made. 1, The piles selected should be ofa readily a possible, the pile spacing qochd not be les than three diameters, particularly ifthe group is large. A group of very Sioaly spaced piles tends to behave as a large single pile. and it may be designed as such 2. Raking ples to resist lateral loads should not be battered excessively. Practical limits ave Reese ee dered to be 1 in $ for driven piles and 1 in 8 for bored piles. Stceper batters may desniven, but usualy only atthe expense of a loss of efficiency ofthe piling plant. While it is posible to design vertial piles to resist large lateral forces, it seems structurally sensible ares some raking piles to absorb part ofthe lateral load and to appreciably stiffen the roup. where this cn be done with tte or no extra cos. In appropriate eitenmstanees so reduce the maximum bending moments to which the piles are subjected there i significant settlement of the surrounding soil, a heavy batter may Hhnroduce problems of unquantifiable secondary stresses and i walidate the estimation of 3. Inprinciple, the patern of each group of ples shouldbe simple and identical tthe others Tre imtations ofthe analytical techniques are such that complicated ping patterns can rarely be justified on theoretical grounds. 44. Asa first approximation, the piles in the group should be arranged to resist the applied {Sading from a structural point of view (le. the centre of action of the pile group should lic ‘Roar the resultant thrusts ofthe various load cases). Having outlined a tral pile group, ‘etailed analysis may be carried out to refine the design 14 CURIA Repo 8 3. Recommended analytical methods for the analysis of laterally-loaded piles ‘Accritical review of the three methods outlined below is given in the Appendices, and worked ‘examples based on published case histories are presented in Section 5, page 24. 3.1 ULTIMATE LATERAL RESISTANCE OF PILES ‘The ultimate lateral resistance ofa laterally-loaded pile may be estimated from the approxi- mate solutions proposed by Broms"™, whose basic assumption was thatthe passive soil fesstance at depth for a pile was 3K,o, and Se, in granular and cohesive soils respectively. ‘These solutions are simple to apply. and they are presented as non-dimensional charts in “Appendix A for various commonly-oceurring pile geometries Broms's solutions, though known to be conservative, are recommended for routine design, because the ultimate lateral resistance ofa pile is not usually the governing criterion. Other futhors have presented more refined solutions to the problems. These are discussed in “Appendix A. ‘A long, verti redundant structure, and it doesnot fail unless one or more plastic ‘hinges develop within its length. Thus. provided the pile has a sufficiently large section modu- luscit transfers any overload deeper into the soil mas, albeit deformation. tha greatly increased “The ultimate laterar resistance ofa pile group containing n piles may be assessed by con- ring the two extreme limiting cases: 1. group capacity ofm times the resistance ofa single pile 2, the lateral resistance ofthe block of soil containing the pile group which represent a group of widely spaced piles (s/D > 8) and a group of close! (iD <3). respectively. The lateral resistance ofa block may be calculated ci application of the principles of soll mechar ameter pile. Inthe latter cate, itis unreasonable to adopt Broms's assumption that the lateral resistance in cohesive soils is zero toa depth of 1.5. The use of efficiency factorsto calculate the contribution of each pile is not recommended, because very little data are avail fable, either from model tests of fullscale experience, to establish reliable factors. spaced er by direct by considering the block as a single large- 3.2 DEFORMATION OF SINGLE PILES “The behaviour of single piles hasbeen studied in considerable det proven analytical techniques have been developed (see Appendix B) In selecting values of sil parameters for design it is necessary to carefully wr2ss the properties ofthe important Upper sol layers. Where lateral loads are substantial ot soi rroperties uncertain tis desirable to also carry outa lateral-oading test on a tra ite. The Norking pile relies onthe stiffness ofthe upper sol layers to mobilise lateral resistance. In tonsa profiles the surface layers are weak, and they are subject to seasonal moisture ‘Content fluctuations. [tis therefore suggested that conservative soll parameters should be Shoigned to the etical sol layer, and thatthe esstane of any zone influenced by drying ‘hrkikage should be ignored. Particular attention shouldbe paid tothe possibilty of scour or Temoval ofthe upper sol layers, “The performance of lateraly-loaded piles in service may usually be predicted with remsonable confidence, ising one of to simple idealisations of sil stiffness: constant (homo- tencous sol oF linearly proportional to depth (non-homogeneous soil). Of these, the for-homogeneous mode is generally the more realise. For design, the elastic continuum model or one ofthe alternative versions ofthe spring analogy may be adopted. Its suggested thatthe latter affords the best practical approach fr toutins dengn, Although the concept lt resticied by basic theoretical short-comings, these do fot significantly affect the results of ealeulations for single ples. |. and relatively well 32.1, Winkler medium- Based on the theory of subgrade reaction, solutions a Mredulus of subgrade moment, shear, and load distribution along a pile for stiffness constant or proportional to reaction method ‘Sept or shor or long piles, various pile Head constraint conditions, and layered sols, “The principal practical dificult attending this method i that of establishing a reasonable svalucof the modulus of subgrade rection, K. Its not clear whether K is dependent onthe Ulameter ofthe pile, and various authors make diferent assumptions (sce Appendix B.1). However. it should be stressed that Kis nota fundamental soil constant. This causes prob- tema when applying K, evaluated from the results of pile loading test, to prototype piles ith different stiffness characteristics. The method is jusiied on the bass that it allows a Tranval calculation tobe made, Wit suitable parameters the accuracy achieved is ‘Mfc dev However. the tho oe rt purport o model he mechan of a continu sil behaviour. ilable for deflection, slope, 32.2 Thep-yformof _ Incircumstances where heavily-loaded ples cause yielding of the soil or where appreciable ‘the Winkler sbilmodel cyclic loading occurs, the only method which gives reasonable predictions is a p-y analysis (Gee Appendix B.2). The accuracy of the output is strongly dependent on the selection of suitable empirical p-y curves, and reference should be made tothe original papers """. A ‘computer program has been published by Reese™, ‘The p-y method has an insubstantial theoretical background. However, the numerical technique can be extremely powerful. P-y curves may be chosen to closely model experi- mental data, such that designing other piles in similar soil conditions is relatively straight Forward. Without field test data, the establishment of reliable p-y curves s based on subjective judgement which greatly detracts from the accuracy ofthe method". ‘The method is readily applicable to the idealisations of homogeneity and linear non- ‘uum analysis homogeneity, and some recent work on layered soils has been carried out™. The elastic ‘continuum model, which is an idealisation, is suitable or the evaluation of deformations the stress range where sol can be considered to be reasonably elastic, up to working lond. ‘Many soils ae not linearly elastic, and they may exhibit a high stiffness at very low loads, ‘although this effect may be modified by disturbance from pile driving. Its therefore desirable tobe able to assign values tothe sil stiffness which are everywhere compatible with the stress Tevel, but at present this isnot feasible. The elastic continuum model is theoretically internally consistent, and development work refining the model and extending the range of its pplication i in progress. Poulos!”*" has published a considerable number of solutions for the deformation and rotation of the head ofthe pile, based on the idealisation of an elastic continuum. including ‘Constant and nearly inereasing stiffness with depth. fixed and free-headed piles, inclined piles and yielding soil. 3.3 DEFORMATION OF PILE GROUPS None of the existing methods of analysing pile groups which curry substantial lateral load may be considered to be well established. The main restraint has been the lack of reiable dats from instrumented fllscale trials with which to compare the results.of the proposed methods, ‘Some such comparisons are made in Section 5. Itis axiomatic that the method of analysis should suit the problem in hand. In general, simple methods should be preferred to numerically more complex methods unless the de- “Signer has high confidence in the accuracy of input data, thus justifying the expense of refined design. “The methods discussed in this Section range from those of simple statics, to comprehensive analytical methods which take account of sll-sructure interaction. See Appendix C forthe ‘analysis of pile groups. 3.3.1 Static and other For simple cases of identical piles and minor structures, where lateral loads ar small a ‘simplified models Straightforward resolution of forces may give a practical design, despite the lack of realism of, the conceptual model, This method of design is normally limited to cases where the lateral Joad is less than 10% of the vertical load, “The piles are assumed tobe pin-jointed atthe pile cap, and to act as axiallytonded columns. The forces in individual piles are calculated by resolving forces or by constructing a polygon of forces. Simple statics cannot be used ifthe group contains more than three rows of vicewth different rakes. 1 inpict inthe analysis thatthe deformation ofthe group is 3.3.2 Winkler soit model 2.3.3. Elastic contin- ‘uum analysis within acceptable limits, provided thatthe allowable load on each pile is not exceeded. In ‘oceasional particular eases, this may not be s0. Generally, the use of the var procedures may lead to impra more realistic than a simple static anal effort. ‘These simple methods do not take into account the lateral restraint offered by the soil. Consequently, calculated deformations ofthe pile group cannot be expected to be accurate, us stiffness methods is not recommended, because such fcomplicated geometries, These methods are n0 is, and they require considerably more computational Methods of analysis based on the spring analogy are not generally applicable to pile groups. Discrete springs are not capable of predicting interaction effects, because they cannot transmit shear forces. The use of factors related tothe pile group size and pile spacing t0 modulate the coefficient of subgrade reaction is an empirical device which makes some ‘llowance for the group effect. The reduction factors are based on a limited laboratory test programme on model pile groups without extensive field correladions. The method is not Enitable for development 1o cover general pile groups. However, for straightforward groups ‘of widely spaced, identical vertical ples, the procedure outlined by the NAVFAC DM7 Manual™ may be adopted. “The most accurate method of analysing the deflection and rotation ofa pile group, together ‘vith the axial, shear, and bending loads within individual ples, isa method based on an “Clastic continuum model, which allows the interaction of one pile with another to be Computed. However the use of the model is limited by the computational effort involved, Gnd no general computer program to cover all commonly-occurting situations is avalable at present “The mathematics incorporated into the computer programs vary, depending on the assumed shape ofthe pile andthe sophistication ofthe numerica integration techniques used Poulos’ analysis” is possibly the least accurate, although the result are presented in a form ‘shich is easly used. This analysis leads to an overestimate, by up to 20%, ofthe pile group deflection Poulos‘ describes an analysis of pile groups comprising identical vertical piles embedded in an elastic continuum of constant stiffness which may be used in manual calculations or simple computer programs to compute the group reduction factor, depending on the lo conditions and fixity of the piles. ‘A number of computer programs using elastic continuum analyses have been published. -The two programs outlined below are realy available in Britain, and they are known to give reasonable results, (See Appendix E and Appendix C3. 1. PGROUP 3.0-available from Highways Computing Division of the Department of ‘Transport. This program is capable of analysing arrays of piles raked in two dimensions ‘embedded in an elastic continuum and loaded in one plane. Pile caps contai piles may be accommodated, end the maximum number of pile segments which may be {sedis 11. This later restriction limits reliable calculations to piles with L/D ratios of less than $0."The pile cap may be in contact with the sol, or, ifthe piles freestanding. the free ‘height may be varied. “The present version (3.0) allows modelling of piles in homogenous soi, ina soi with ‘modulus linearly increasing with depth, or ina two-layer soil, 2. PIGLET- published ty Randolph. This program is based on power curves fitted to the normalise resilts ofa large number of finite element analyses. The program is suitable for the analysis of three-dimensional groups comprising long flexible piles (i.e. ples longer than 10 1020 diameters typically). The piles are assumed to be of equal length, and they nay be raked in any direction, while subject to vertical, lateral, moment, of torsional foading. The soil is assumed to have astiffness which increases linearly with depth, but no provision is made fora ground-contacting ple cap. Because this program is based on the Fesults of finite element analyses, it is not necessary to solve the full elastic continuum problem each time. The computer time requited is about 10% of that used by PGROUP, Pith a consequent saving in cost, A version for micro computers wll be available in the fear future from the Highways Computing Division, 4. Assessment 4.1.1 Assessment from correlation with other ‘oll properties 4.1.2. Assessment from laboratory tests: 4.1.3. Assessment from in-situ tests of soil properties “There are no clear guidelines forthe assessment of soil property values to be adopted for the design of laterally loaded ples, Many faciors influence the actual mobilised values. In par~ ticular, the disturbing effet of pile installation is difficult to quantity. Most soll exhibit ‘considerable loss of stiffness under the action of cylic loading, whichis virtually impossible to felate to laboratory test data, Generally, itis desirable o select upper and lower limits tothe ‘rifeal soll properties, and o check the sensitivity ofthe design to variation within the chosen range For the successful design of laterally oaded piles or pile groups, the upper layers of the soi should be thoroughly investigated, because these tend tobe particularly variable, In the case ‘of single pies, the stffnes ofthe tol toa depth of only afew pile diameters has a dominant influence on the behaviour ofthe pile. “The most obvious and satisfactory procedure for establishing the response of laterally Joaded piles isthe load testing ofa pile. If lateral loads are high, of the structure is subjected to significant cyclic loading, its strongly recommended that pile tests should be undertaken Tocantirm the design assumptions. However, for design purposes, the enginecris often faced ‘with th tsk of selecting suitable parameters from limited site investigation data. 4.1. MODULUS OF ELASTICITY ‘Young's modulus of elasticity, E, of the shear modulus, G, ofthe soil may be assessed by rmpitical correlations with the undrained shear strength, Such relationships are based on the imyted pile tet data published. For an initial approximate value of undrained Young's ‘modulus, the following relationship may be considered appropriate: in the general range 100t0 400 ff over-consolidated clays, the higher values should be adopted, perhaps even up toa tuliple of 800, but for long-term loading, including an allowance for creep strains, values of half the above may be ennsidered suitable. The above factors include some allowance for the disturbance of the soll structure caused by piling operations. However, there are insufficient data to differentiate between the influence of the various methods of installing piles. ‘Typical values of Young's modulus and Poisson's ratio for various soil types, consh homogeneous elastic solids, are given in Table 2, redas ‘Young's modulus may be conservatively assessed from carefully conducted triaxial tests on good quality samples. Either drained or undrained test conditions may be adopted as appro {the assurmption that shear modulus is approximately independent of Srainage conditions is often used to connect drained and undrained moduli via Poisson's io) "To obtain realistic values, and to minimise the effect of stress relief, the most {alistactory procedure is to reconsolidate the samples to the estimated in-situ stress conditions prior to starting the modulus determination. The modulus should be measured over 8 part tunload-reload cycle. “The stfnes ofthe soil may be assessed from various in-situ tests. Attempts have been made toestablish an empirical relationship between Young's modulus E, andthe Standard Pen- tration Test (SPT) N valve. Proposed relationships range from E= 600M (kN/m') (Yoshida and Yoshinaks™), to E= SOQON (KN) (Parry), Both are based on the back analysis of case histories of buildings. For preliminary design of piles, itis euggested that a value of modulus should be taken in the range 1200N < E< 2500N (kN/n?) Driving piles in granular strata can modify the modulus, effecting an increase for loose soils and a decrease for dense soils 4.1.4 Assessment from pile tests, Table 2. Typical values of Young's modulus and Poisson's ratio for various soil types considered as homogeneous elastic solids* ‘Solltype ‘Young's modulus, E Poleson's rato, ¥ (MN!) Loose sand Sto 20 031004 ‘Medium dense sand 16t0 20 0.210035, Densesand 30 t0 100 0.151003 Undrained rained. Softelay 200 6 Ito 4 0.5 forundrained Firmelay Stolz 300 8 Stifctay 10t020 Sto1s tvieetecibl, tat maynot be snvelet or poet toes andgnd ca ele tea wih mds ei si ebSmnpmeou es reed er ae “The results of static cone penetrometer tests may be used inthe following relationship to assess Young's modulus™. Eee o Pressuremeter tests may be used to measure the modulus n-sts, the Ménard type being for tundrained testing and the Camkometer type for either drained or undrained test conditions. For interpretation ofthe results of these tess, the reader is referred to Ménard et al.™ and ‘Windle and Wroth?™. Direct (plate) loading tess can be used to determine the in-situ measurement of modulus, put no rigorous investigation of the applicability ofthe results to the design of laterally-loaded piles has been undertaken, Ground disturbance and stress relief can result in a conservative {assessment of the modulus of elasticity. ‘Young's modulus may be calculated from the theoretical relationship (17) 2) a set”) @ where fis an influence factor (J, = 7/4 for w righ circular plate and lg = 0.88 fora square plate). ‘Marsland and Randolph describe a satisfactory procedure fora vertical plate test. {deally, horizontal tests should be conducted, but, in view of the difficulties involved, a vertical test may prove equally satisfactory. The Transport and Road Research Laboratory thas designed portable test equipment for use ina trench. ‘Because the designer is interested in the deformation characteristics of the top few metres of sil, itis suggested thatthe application of soll data derived from horizontal pate-bearing tests in a tril pit to the analysis of laterally-loaded piles merits further investigation. “The most realistic values of Young's modulus for use in prediction of pile behaviour may be established from back analysis of pile test data (see Section 3.2.3, page 16). However, the test Should be carefully conducted, and f possible the tes pile should be similar to the prototype piles. As aminimum, the translation and rotation of the pile head should be recorded. Such Enalyses are considered tobe valid only inthe range of working load, where any plastic zone ‘which may develop atthe head of the pile is restricted in size. It may be possible to test “Smaller piles, using appropriate dimensionless parameters to scale the tes pile correctly. 42. COEFFICIENT OF SUBGRADE REACTION “The computed bending moments in a laterally-loaded ple are not very sensitive to the valve of the coe ficient of subprade reaction, K, adopted, although deflections are roughly pro- portional to it. Quite a crude estimate of the modulus of subgrade reaction may thus suffice for a preliminary design. 0 4.2.1. Asseasment from Values of K may be derived from indirect indicators of sol stiffness published data and correlation with other soil properties SPT N values) or unconfined compression test results, The values quoted by Terzaghi™ may be considered reasonably conservative, Typical values of the horizontal coeficient of subgrade reaction for clays and sand are given in Tables 3 and 4. For clay, fyi the coefficient of sub-grade reaction for a 305-mm square plate in MN/m? as defined by Terzaghi™, which is related to K for piles using the relationship hae e where Fy isthe Terzaghi coefcient of horizontal subgrade reaction, 8 the breadth ofthe loaded area isin mettes, and then Ls KakBa (MN/m’) a In sand, the soil stiffness i usually found to increase with depth quite markedly, and the appropriate parameters the rate of increase ofthe coefficient of horizontal subgrade reaction, ny. This is related to K by the equation Kame © and the Terzaghi coefficient of horizontal subgrade reaction, ky. by ne ne © “The values suggested by Terzaghi™ are valid for stresses up to one half the ultimate soit pressure, and they include an allowance for long-term movements. Back analysis of field test Gata by various authors gives values ofthe constant of subgrade reaction upto five times larger than Terzaghi's values (see Figure 3). These calculated values of my are initial values at the origin ofthe p-y curve for short-term ioad tests, and are thus not strictly comparable with ‘Terzaghi’s recommendations. For the design of pilesin sands, itis suggested that Terzaghi's data should be used as a lower limit, and that te following relationship" should be con~ sidered for the upper limit: m= 0.190," (MNim') a Table 3 Vatues of the coefficient of subgrade reaction for clays” Consistency ofetay | Firmtostit Stiettoverystitt | Hard cx (kN/m') 50 to 100, 100 to 200 >200 (Nim) 15 10 30 30 t0 60 >60 (for 305mm plate) K (MINIM) 3106 61012 p12 Table 4 Values of the rate of increase of coefficient of subgrade reaction with depth for sands*™ Relative density Loose | Medium | Dense 1N (blows/300 mm) 41010 | 10t030 | 30 my (rysand) .MN/m’) (ie20smmpiaey |? 6 7 ‘ny (gubmerged sand) (MNim') 1 4 10 CIRIA Reon 103 CRIA Repo 13 octal sate econ. (MO) Figure3 Coefficient of subgrede nds os 2 density (Gacessino etal) For soft normally-consolidated clays, values of the coefficient of subgrade reaction, K, have ‘been related to the undrained shear strength of the material, and a wide range of values has been recorded”, These range from the general recommendation by Terzaghi™ of K = 67 RS escent work by Jamiolkowski and Lancellota™, showing Ki = 500 to 700c, at Porto Tolle in italy. Again, the values quoted by Terzaghi are for long-term loading, For normal design ‘urposed, easonably conservative values should be adopted, For short-term loading, the following values are proposed: For long-term loading. and stresses up to 50% of ultimate load, one third ofthe above values Fray be tonsidered appropriate, The suggested values assume thatthe undrained cohesive rrength of the deposit was determined using an in-situ method, or from high quality un- ‘inturbed samples recovered with a thin-walled sampling device. “To account fot the non-linear behaviour of piles at higher loads, Garassino eta.” suggest the following relationship: wen( 5) ° uje_\ = ron (35) © where the exponent, b, isin the range ~0.5 to ~0.7 for normally consolidated clay and sand. duets the limiting deformation ofthe pile to which my oF K applies (ie. the limit of elastic behaviour). ‘Typical values of us are given in Table 5. Equations (8) and (9) involve the use of an iterative procedure, and tis suitable for manual caleulations.. 2 422 Assessment of the costficlent of subgrade reaction from In-situ testa 42:3. Assessment from pile tests, Table § Values at the elastic limit of ground-line deflection Soll type glB () Sands 0.21010 Normally-consolidatedclays | 0.2t00.5 Over-consolidatedclays 0,04100.1, “The stress condition produced by a pressuremeter, particularly the sell-boring type, is some ‘what similar to the oil reaction of alaterally-loaded pile, Attempts have been made to relate the pressuremeter modulus tothe coetfcient of subgrade reaction applicable toa Interally-londed pile. ‘Ménard, Bourdon and Gambin™ propose the formula given in Equation (10) to calculate the value of the coefficient of subgrade reaction. eel (2) +e (wan) 10) where R = 0.3m (a reference radius), ois a rheological factor varying from 1.0 t0 0.50 for ‘laye, 0.87 t0 0.33 for silts and 0.50 100.33 for sands, and Ex isthe mean value of the pressuremeter modulus of elasticity over the characteristic length of the pile, ‘Typically, K = 1.6102.5E, using this relationship, Baguelin and Jezequel™ suggest that the initial soil modulus of subgrade reaction may be related tothe self-boring pressuremeter modulus by the empirical factors K,=1.6102.0Eq. ‘and Jamiotkowski and Lancellotta ® found similar values for a self-boring pressuremeter. Based on a comparison of analyses usin the Winkler soil model andthe elastic continuum model, various authors 29 compared the coefficient of subgrade reaction with Young's ‘modulus. Such studies show a relationship ofthe form K=081018E depending on the method of comparison (deflection or moment) and the pile ané conditions sdopted, However, nosuch straightforward relation between E and K cam be expected to holdin general when consideration is extended to parameters for the behaviour of piles in pile groups. ‘Vertical (plate) bearing tests may also be used to assess the coefficient of subgrade re action, following Terzaghi™. Fully-instrumented test piles may be used to assess values of the modulus of subgrade. feaction of to determine a comprehensive set of p-y curves. A satisfactory procedure is 10 ‘eatrument a steel pile with strain gauges at closely-spaced intervals™. The measured bending moments may then be integrated twice to calculate the deflection, and differentiated twice to Joteranine the load per unit length, These calculations are usualy carried out using numerical techniques, together with acurve-fiting procedure to minimise the effect of experimental error. ‘Such tests are expensive, and they are not considered suitable for evaluating design data Instrumentation of concrete piles for such purposes is extremely difficult, because the sentctial is not fully elastic, and because at high loads the flexural rigidity of the pile changes. Py curves or reasonable values of K may be established from test piles, for which only the round line deflection and rotation are measured. Reese and Cox™ descr asoltion tothe problem in which itis necessary to assume a variation of K with depth. CRIA Report 8 “The coefficient of subgrade reaction may also be determined from tests in which only the ground-line deflection is weacured. The values of the coefficient of subgrade reaction ‘Setermined from uniustrumenced test piles are not as accurate as those derived from fully- instrumented test ples. Any experimental errors in the tess are amplified, because the characteristic length, T, is a powsr function of nq, and the results are forced to fit an assumed ‘variation with depth, However, provided that the tes piles are similar in all respects tothe ‘design piles and are installed by similar equipment, these relatively simple lateral-load tests provide adequate design data, 4.3. DISCUSSION OF STIFFNESS PARAMETERS ‘As suggested atthe beginning of Section 4 the selection of appropriate soll parameters for design of laterally loaded piles isin practice a difficult process. Individual eases should be carefully considered on their merits. Incommon with many geotechnical calculations, because of the grea simplifications made inthe available theoretical models, appropriate parameters are strongly dependent on the sctual calculation method employed, rather than being true fundamental soi properties. In principle this should be less the ease for elastic methods than for subgrade reaction methods of calculation, “The following points summarise considerations which may be applied when evaluating parameters in the light of representative values presented in this section. 1. For single piles, and ‘constant’ materials, Young's modulus, E, and subgrade reaction parameter, K, were found to be roughly comparable. For single piles and ‘proportional’ fraterials, rate of increase of Young's modulus with depth, m (= 4/62), and rate of {nerease of subgrade reaction with depth, m, may similarly be regarded as roughly synonymous. 2. For given soil conditions, E and dE/dz should be relatively independent of pile group size. However, elastic continuum models over predict interaction effects, particularly in the case of proportional soils, Values of K and m effective in pile groups are generally ‘considerably less than single pile values, and dependent on position within the group, 3. Assumption of constant E or of constant K is seldom the most appropriate idealisation. ‘Thus, for soft clays and for granular soils a linear increase of sol stiffness with depth is generally recommended as the best approximation. Even for stiff over-consolidated clays, {may well be that although sol stiffness can be considered more uniform with depth, the ‘effects of local yield at the pile head and surface weathering may make the proportional Stiffness model more suitable, However, currently available methods of analysis for groups ‘can only be applied to relatively simple soil profiles, and a compromise mode! has usualiy ‘tobe accepted in practice. 5. Results and conclusions ‘Several analytical niodels are recommended so that the designer may select the technique ost suited toa particular problem, bearing in mind the limitations ofthe methods, “The methods of analysis discusted range from simple manual calculations to complex computer programs. The cost ofthe design ofa pile group can range from a few hundred to Several thousand pounds. Itis thus important to balance the degree of sophistication of the analysis with the accuracy of the available data ‘Some case histories taken from the literature were analysed using the design methods described in tis Report to indicate the validity ofthe calculations. The detailed calculations are presented in Appendix D, and the output of the various methods of analysis is sum- imarised below to assist the designer in the selection of an appropriate technique. 5.1. SINGLE PILES “The analysis of two case histories is presented in Table 6. The soil test data foreach of these tests are somewhat limited, and it was necessary to estimate the stiffness of the soil indirectly. ‘Generally, the computed deformations compare reasonably well withthe observed values the elastic continuum solutions giving lower values. All the methods of analysis under est- tate the maximum bending moment in the piles by up to 20%. However, the accuracy of the ‘calculations is largely dependent on the correct selection of the soil stiffness 5.2 PILE GROUPS “The results of analyses of the piled foundations tothe east abutment of Newhaven ‘Overbridge are presented in Tables 7 and 8, CIRIA undertook a compreliensve instru mentation programme to monitor the behaviour ofthis structure”. “The computed end measured axial loads in the piles are given in Table 7. The stifiness method over estimates the load on the middle piles ofthe group. Applied loads were not ‘measured, and so there fs some uncertainty a8 to the magnitude ofthe horizontal earth pressure and to the volume of backfill which acts with the abutment. ‘The stiiness method not unexpectedly over predicted the lateral movements. Its likely that the Poulos analysis™ gives a high computed translation, because of: effect ofthe raking piles not being fully accounted for 2, the over prediction of pile interaction inherent in this method 3, theeffect ofthe desiccated surface crust of the alluvium and passive resistance of the pile cap. “The PGROUP (2.0) program used could only accommodate one soil layer, and a check of the sensitivity of the foundation to changing the modulus of elasticity was made. Doubling & Secreased the computed lateral translation by 1 mm, illustrating thatthe pile layout chosen. svas quite insensitive to the assumed sol properties. The output of the PGROUP program Included bending moments and shear force distribution for each pile. “These case histories illustrate some of the difficulties facing the designer ofa piled foun- dation, Establishing reliable soil properties isthe most pressing problem. Currently available echniques are discussed in detail in this Report. However, further case studies of real Sructures are necessary to give confidence to the values measured. ’) CCIRIA Repo 18 ngle pile: Summary of calculated and measured deformation ‘moments Method of analysis Computed values Pile test data Lateral deformation | bending (om) moment (kN m) Maximum | Lateral deformation (mm) Maximum bending, moment (kN m) [Pitetest at Mustang Island, Reese, Coxand Koop") Elastic analysis (Poulos) (Gee Appendix B, Section 3.1) 40 107 5 Ehstic analysis (Randolph) (Gee Appendix, Section 3.2) 35 101 Subgrade reaction method (Gee Appendix B, Section 1.2) 40 13 136 Subgrade reaction methodat high oa (seepage 21) 19 328 374 McClelland and Foch) Elastic analysis (Poulos) 4 355 0 1.8 m above mud line Elastic analysis (Randolph) Subgrade reaction ‘method 31 490 Canta Regt 08 2s Table 7 Computed and measured axial loads in piles at end of construction, Newhaven overbridge ‘Method of calculation Front iles~ Middlepiles- | — Backpiles~ load per pile load per pie load per pile (en) (en) (kN) Staticmethod 360 360 330 (Gee Appendix, Section 1.1) Stiffness method 130 600 210 (Gawko™— see Appendix, Section 1.2) PGROUP 290 400 300 (E=S000kNim', ‘constant with depth) ‘Measured 320 350 310 Table 8 Computed and measured deformation of pile cap, Newhaven overbridge Method of analysis Lateral Vertical Rotation translation deformation (mm) (mm) (radian) Statiemethod Not applicable | Not applicable | Not applicable Stitfness method a 4 = 24x10" Poulos? 101020 Not calculated | Not calculated (E varyinglinearly ‘with depth) m= l.14MNin? PGROUP 6 3 = 5.2.x 10" (E= S000kNim, ‘constant with depth) Measured <10.mm (assessed)| © 15,Py= 9D. ‘This approximation may be used in design, From statics, the following formulae may be derived, depending on the fixity conditions of the pile head and the failure mechanism involved. 1. For short free-headed piles (less then twice the characteristic length), which fil by rotation of ultimate lateral resistance with depth was proposed by oe ee A e Hy L AY? Mh [fesse stip] 4L,) M, [res Fe] 4.long restrained piles Mo Me He ——S (22) [-on ft] “These equations are evaluated in dimensionless form, and are presented in the form of design charts in Figures 9 and 10, page 37. He 4, PILE GROUPS For widely-spaced piles in a group, the ultimate lateral resistance ofthe group may be calcu- lated asthe sum of the lateral resistances ofthe individual ples. For groups of closely-spaced piles (W/D<3), the piles and soil fail asa rigid block. Hence the group can be considered as a single large pile and Equations (11) or (15) 10.(18) used to EEtoulate the ultimate resistance, Broms's simplification of discounting the lateral resistance ‘of the top 1.5D of soll may, in this case, be unrealistic. Alternatively, the stability against block failure of the soil containing the piles and pile cap may be calculated using a wedge analysis. ‘Usually, a consideration of the’ above two limiting cases is sufficient for design purposes. [No solutions for intermediate cases are available, and the use of efficiency factors is not recommended because very little published field data exist to establish reasonable values. {ARIA Repent 1 imate ea eta Het Figur Uhimete lateral resistance of short piles in cohesive soll (Brome) Uninat nteraesiane. eD? Figura? Unimate lateral resistance ‘of long piles in cohesive soit (Broms“*) RIA Report 8 35 Figures Deflection, soll reaction and ‘bending moment distribution for along restrained plein cohesiontess soll (Broms'") 36 (8) Deflection 40K, (b) Soil reaction (© Bending moment (ARIA Repo 3 Uninet eitace, IK DY Figure Ultimate lateral resistance of short pites in ‘cohesioniess soll (Broms") nite eaves, MIKDY ral resistance incohesionioss ARIA Report 3 AppendixB Analysis of lateral displacement of single piles Estimation of the lateral deformation of a pile is particularly difficult unless many ideal~ isations are made regarding the behaviour of the soil, Two methods of calculation a commonly adopted, based on ether the assumption thatthe sol isan elastic continuum, oF that it may be replaced by a Winkler spring medium (perhaps in the form of a p-y analysis) ‘The methods are Uceueced here in the order in which they appeared in the text, which isto some extent the historial order in which they came to have an application to laterally-loaded piles. 1. WINKLER MEDIUM. “The concept of a beam supported by a series of discrete springs (see Figure 11) to representa soil mass was introduced by Winkler. This model has been widely used to analyse the beha jour of foundations, and it may be applied toa pile if itis assumed to be a thin vertical elastic strip. The stiffness of a Winkler foundation was defined by Terzaghi"™ as, ka cen 3) where kis the coefficient of subgrade reaction applied toa supported beam, wis the ‘eflection and p is the vertical bearing pressure. The Winkler model does not represent a continuum, because there is no transfer of shear stresses. In general, the application of this, ‘concept leads to an over estimate of structural deformations. Figure 12 illustrates the application of a Winkler medium to ples. For piles, itis usual practice to express the horizontal coefficient of subgrade reaction, ky. slightly differently: KakB= £ (Nim) as where p has units of FL“! and K (coefficient of subgrade reaction) has units of stress. and ky Units of FL In American literature™, the coefficient is called a ‘sil modulus’ and is denoted by te symbol Ey This defiton lends to conuson with Young's modules of Stacy anich sno has nis of FL, However, K isnot materia constant, but varies with ‘onttgensand breath ofthe ple, and withthe lading conditions. In geneal, K varies with tepthandthe forms Keke tme @ on ee ae (as a Kame ( F J (26) may be adopted, where mys the rate of increase of subgrade reaction with depth. The ‘exponent n commonly vaties from zero for over-consolidated clays to 2 for some sands. (a) Elastic (Winkler) springs (b) Elastic continuum Figure 11. Foundetion response models for elestic continuum and Winkler spring medium 8 ARIA Repo 8 ‘The differential equation governing the deflection of a pile subject to axial, transverse and moment loadings For piles of practical dimensions, buckling isnot usualy negligible effect on the flexural behaviour. The term conts lem, and the ig P may be omitted, leaving tu Elyse + Ku=0 (28) Ess 0-0 @) Bega tMno 0) the moment of inertia of the pile. ‘Closed-form solutions to these differential equations are available only when Kris constant ‘and for simple boundary conditions, More general solutions are available in inte difference form, for which variations of K with depth and with deflection, and layered soils have been considered 4.1 Constant coefficient The simplest solutions require K tobe considered constant with depth. The solution", t0 Of subgrade reaction Equations (28) to (30) then takes the form tu exp(fz) (A cospz + BsinBz) + exp(—Br) (Coosbz + D sinfz) “where the constants of integration, A, B, Cand D may be determined for various boundary ‘conditions and p= S @2) {isa measure of the stiffness ofthe sol relative to that ofthe pile. Because has units of L"* its inverse may be thought of as a characteristic Tength of the pile, Le. Figure 12 ‘Soil/pile interaction model used in Winkler spring idealisation CRA Report 09 ‘The boundary conditions usually considered are for fixed and free-headed piles subject to laterat loads or moments, The piles may be long or short, and they may also be stiff. These various dealisations lead to a multitude of ‘standard! formulae. ‘Matlock and Reese carried out a dimensional analysis ofthe problem and developed a series of equations containing similar groups of parameters: HP, Mo Interal deftection, w= =A, + =° By (3) : ot” Ele” Ele ‘HOT | MTB, slope, 0x ——A, + oH Ey ' aly bending moment, M= HT Aa + MeB 9 36) on G8) Tis a characteristic length for the homogeneous case (K = constant). Long, free and fixed-headed pites ‘Analytical expressions for the coefficients A and B are given in Table 9, and numerical values atthe ground surface are given in Table 10. ‘The case of along fixed-headed pile (1. > 1.5L) may be considered by setting the slope ‘equal to zero at the ground surface, which is equivalent to applying a moment of ~ HT toatree-heatd pls, Equations (33) and (34) may be combined for piles of known end restraint to give HoT? us cy @) En” were the defestion coefficient MOB, GA ae (40) Son or sire and ficed-headed piles For short or stiff piles (L< 1.5 Leand L<0.5 L, for free and fixed piles, respectively), ihe ‘deformation may be calculated from simple statics. Piles of intermediate length are treated as beams of finite length. ARIA Report Table 9 Expressions for deflection coefficients A and B for a constant subgrade reaction Parameter Applied loading Horizontal oad, H Moment, M Deflection, u Aya Ve Peosft B= €™ (cone ~sinB2) Slope, ® M(cospe + sinB2) By= ~ VEeM(cos82) Moment, M An = V Be Ping By = e-P(cosps + sinfs) Shear, ¥ Aya — © (cosBe sind) By —V Te sins Soil reaction, p At Ay Table 10 Ground-line values of deflection coefficients A and B for a single pile in Winkler medium (long piles, uniform soil) Parameter Lateralond Moment Defletionu | Ay = 141 | 8, = 100 Slope,® A = -t00 | Bm -1at Moment,M | dn = 000 | & = 1.00 shear,¥ A = 10 | B= 000 12 Coefficient of subgrade reaction Increasing with depth Discussion For the particular ease of ground-tine deflection, to, substi efficients A and B given in Table 10 into Equation (33) gives jon of the values forthe co- HB | 2M a worK TK which is identical with that of Broms'®. A design chart besed on this expression, together ‘with those for fixed and rigid piles, is shown in Figure 12, in which Broms has sketched a relationship for ples of intermediate length. Davisson and Gill® studied the problem of a two-layer profile and found that the upper layer of soilhas a dominant influence on the behaviour of a laterlly-luaded pile. The depth Was which such a superficial layer has a particularly strong influence is only 0.27, or a few pile diameters. This zone i likely tobe subject to seasonal variations in stiffness. Values for The coefficients for deflection and moment, calculated by Davisson and Gill, are reproduced In Figures 1410 17. These coefficients may be used in conjunction with Equations (31) and (G5)Atternatvely or preliminary conservative desig, the contribution of softened soil may simply be neglected over the upper few pile diameters, For the successful design of a lterally-loaded pie, the soil should be thoroughly invest- igated toa depth of about 0.47 to assess a realistic vue of the coefficient of subgrade weiction, Over-consolidated clays may be represented by a constant value of KX with depth. However, the assumption of a constant value of K may lead to a serious under estimate of the ‘deflection and moment inthe pile, in the event that the surface layers are softened by weathering, Conversely, the lateral resistance of a pile may be augmented by the presence of 1 superficial crust or by the placement of a thin layer of stiff fill around the pile. For normally-consolidated clays and sands, where the coefficient of subgrade reaction may be ‘considered to vary linearly with depth Kame a Closed-form solutions to Equations (28) to (30) are not available, but useful solutions in a ‘non-dimensional form have been obtained using finite difference techniques. These solu tre identical in form to Equations (33) and (37). Values fr the various coefficients are given ‘Tables 11 to 19. The values are expressed in terms ofa dimensionless depth parameter {Z= 217) and a similar pile length parameter (Zax = LT), where, for this ease, the characteristic length El 3) ™ ‘Tables 11 10 19 show that the behaviour of piles of length exceeding 47 is effectively indis- tinguishable from that of infinitely long members. Conversely, piles with a length of less than dT behave as short rigid members. Variation of C, withthe degree of fixity, Mo/Ho7. 1s ittustrated in Figure 18. “The deflection of short piles may be calculated from statics. For free-headed piles. roms obtained: rey thee (2) Tr tg 18H «) and for fined-headed piles: eee a 0" Tim 4s) ‘Other solutions are available for particular profiles of K with depth, the most useful being & ‘power function developed by Matlock and Reese". However the use of uch solutions i not encrally justified, because of the uncertainties of assigning realistic values to K throughout ‘he oll profile. Provided appropriate valucs are selected, the simple approaches outlined above give an estimate of the limits of behaviour of aleterally-londed pile whichis satisfactory for most practical problems. : rercominueonpee RIA Repo 0" meson nel Seition KL. Figure 13 ‘deflection at ground ‘surface of latoraly-loaded plein cohosive soil (Bros) Dinersionles eng fl, leer oie son dines yer hicks Figure 14 Effect of @ superficial crust, definition of parameters (Davisson and Gill) CRIA Repo 8 2B “4 efecto ceticea ate Ay Figure 15 Influence of crust thickness ‘on deflection at ground surface of free-headed pile ‘subjected to lateral load (Davieson and Gilt) Detection onticet te 8, Figure 16 Influence of rust thickness ‘on deflection at ground surface of ree-headed pile wubjected to moment loading (Davisson end curs) a0 as ry oar car oer On “Tekno ere get 19 oar car oer aT “Tks ova eyet Figure 17 Influence of layer ‘coefficient on deflection at {ground surface of fixed- hhoaded pile subjected to lateral load (Davisson and in) Figure 18 Deflection costicient for a range of fixity conaltions fora pile eubjacted to moment and lateral iogding Innon-homogeneous soll with modulus increasing linearly with depth (Matlock and Reese “) ona Beg 4s Tablet 1 Deflection coefficients A and B fora pile in a Winkler medium (Zmax = 1.0) za [a | 4m | 4 | «a | % | & Bn | oe ao | 1 | a7 | oo | 10 | om | mm | asa | 1m | om is | -a7 | on | os | -rs | tyes | -a5a8 | 100 | -o10 30 | Tare | oie | os | <250 | tote | -aszn | ome | -039 tos | Tam | oas | oa | hie | tee | -asis | oss | -ors eis | can | o2r | oo | “326° | 3m | “0 | om | -2i5 Soe | Tam | om | ae | cae | Sse | Hole (oy ee “The maximum moment occurs at a depth of /4ra (homogeneous) and l/r (proportionally varying stiffness) © Vacs rom Bnei an Davies ™ wipe Detonation tr Fad Foe) g 1 Figure 24 * Values of deformation factor, aw. for free-heedod » pile in non-homogeneous soil with modulus 0 Increasing linearly with ae ee depth (Poulos™) Peete Ke soon veos ‘ton an Den ey UD =a 10 * factor, fa for free-headed » pile In non-homogeneous ‘oll with modulus increas ” ing inearly with depth wee (Poulos™) Phe ety ar 7 Figure 26 Values of deformation factor, Ir, for fixed-headed pile in homogeneous soil (Poulos™) Figure 27 Values of deformation factor, for fxed-headed pile in non-homogeneous ‘oil with modulus increasing linearly with depth (Poulos ") 38 » 0s © Vote tom Baneree and Dives ™ iea>upss yao aS le ex aot. © Voter Banerec and Devs ™ ieupew wo wt 0 we eet ator Ke 1 CURIA Regt 08 Randolph presents generalised curves showing the deflected shape, together with the -variation of bertding moment down the length ofa pile, reproduced in Figures 28 and 29. (hese gre are not bed on Eauations (4) ‘and (53), which relate to ground-level feformations only). 3.3 Yielding soit Poulos ™ extended the elastic analysis to account for plastic yielding ofthe soil near the ground line, by introducing a yield stress. The yield streses are those proposed by Broms'*.*” {for cohesionless and cohesive sols. The deflections and rotations calculated using Equations (G0) and (51) are divided by a yield factor (Fy, Fa, F'y and Fe). Values ofthe yild factors expressed in terms of H/H, are presented Figures 30 to 34, This analysis should be used with ‘care, because the model employed is only a rather crude approximation to the soll behaviour. termine on page 4 mos, Gc 2 Soo (a) Deflected pile shape (@) Bending moment profile Figure2® Generalised curves giving deflected pile shape and bending moment profile for lateral force loading (Rendolph“) (a) Me (a) Defected pite shape (©) Bending moment profile Figure 29 Generalised curves giving deflected pile shape and bending moment profitefor moment loading (Randolph) CURIA Rept 08 9 ‘Yield pement ot Fe Vaverol tor Si —— ee re Figure30 Yield eisplacement factor for free-headed plain homogeneous soll wth constant yield resistance (Poulos*"*) {RIA Report 10 os a a “Veteran or Fe os] a 02 os as 4M, Figure31 Yield otation factor for freesheaded plan homogeneous soll with constant yild resistance (Poulos) Canta Report 03 os ‘ied placement ate Vater fo: ° 02 oe oe on rm mn, Figure32 Yield displacement factor for free-heeded pile n non-homogeneous soil with modulus and yield resistence increesing linearly with depth (Poulos™"*) e ARIA Rep 8 i Vato oe wm, Figure33 Yield rotation factor fo fee-headed pile in non-homogensous sol with m joduius and yield resistence inereasing linearly with depth (Poulos ™) DAF werge aac Por inh 0 Liormyitt estas Snbaton Figure 34 ° Ukimate resistance of ose ese unrestrained free-headed © L rigid plies (Poulos™) T ? ARIA Repo AppendixC Analysis of pile groups 41.1 Staticresolution of forces ‘A number of models exist for analysing ile groups which carry axial, lateral and moment foading. These methods range from simple structural procedures, which ignore the restraint offered by the soil to those which consider the pies tobe embedded in an elastic continuum ‘The lateral resistance ofa single ple is considerable even in soft soils, and for economic reasons should not be neglected, Unfortunately, the effect of group action has not been well documented, and very litle field data exist to correlate with design procedures, For this reason, some of the recommendations made inthis Appemlix are probably rather conser vative. 1. STRUCTURAL METHODS OF ANALYSIS Various structural methods of analysing indeterminate frames are widely used for designing, pile groups In all these methods, the piles are assumed to be end bearing, and the restraint Df any soft overlying soll is neglected. Thus the model is not a good representation of practi behaviour. The success ofthese methods, in particular the stiffness method, relies on the fact that designers select piles which are each adequate to carry the maximum calculated pile load ‘within the group. Designs using these methods are thus inherently conservative. = In this very simple method, the piles are assumed to be identical and to act as slender pin jointed columns which carry only axial loads, ;, The lateral restraint of the sol is ignored, ‘except insofar asthe piles are presumed not to ‘and the restraining effect ofthe pile '35) isnot considered, Simple resolution of forces yields a solution to the forees in the piles. This method is not appropriate for groups of piles containing more than three different directions of rake, because the group then becomes statically indeterming For groups of vertical piles (see Figure 36), the load in each pile may be calculated from the formula @) Loads in groups of piles with up to three planes of rake may be determined by graphical resolution of forces (see Figure 35). ‘and ar rent vee. 4 1 t 5 ' ' Figure 35 Figure 36 Graphical determination of pile loads within ‘Schematic representation of static load distribution lateraily-loaded plia group (CP2™) Inplte group (CP2") a CRIA Repo 19 1.2 Stiffness method (or elastic centre method) Figure 37 Definition of parameters for stiffness method ofp ‘group analysis (Sawko' CHRIA Rept 102 Inthe static method of analysis, no reference is made to the deformation of individual piles corto the pile group as a whole. It is implicitly assumed that, provided the ple size is selected ‘on the basis ofthe calculated working loads, after taking account of group action, the be~ haviour of the group is satisfactory. In effect, the assumption is made that ifthe settlement of ‘an individual pile from the group (considered in isolation at working load) is satisfactory, the settlement of the pile group is also acceptable. This may not always be the case. “The stiffness method of analysis for indeterminate systems has also been applied to the analy- tisof pile groups ©, The following simplifying assumptions are made: the lateral restraint ofthe soils negligible 2, the piles are pin jointed, although the analysis may be extended to account for bending stifiness 3. the piles deform elastically 4.the pile cap is rigid and horizontal. “Arelationship between the displacements of the pile cap (u,v, 6) and the axial distortion (A) ‘of the piles may be established by applying ont values of rigid body displacements tothe pile cap (see Figure 37) For atypical pile (the j® ple), it may be shown thatthe transformation matrix for deter- ‘mining the axial distortion is: ie [ines og) nots] «© ‘Thus the geomet matric forthe ple group may be obtained wl=[% (64) qh Ty “The mia sttnes ofan individual piles AoE: cry ™ " aon % nem Londs acting st 0 i = fe sever Elevation on eae new Displacements at O 65 ‘and the stiffness matrix ofthe unassembled structure, neglecting bending stiffness is wie 0 :] 5) oS, 0 00 Ss ‘The loading matrixis = fH (l= [H 6) M ‘The three matrices {¢, [5], and [P] are the only ones which have to be formulated. The various matrix operations may now be carried out to give tal (8) (67) where [8] is column vector of the unknown displacements ‘The foreesin the pile are given by (A= (5) 14} = (5 (a) 18) (68) ‘The extemal loads may be related tothe internal forces by (el @ therefore (rl = (51. (8) (70) where [S] isthe stiffness matrix of the assembled pile group with respect to the chosen axes. ‘Solution of Equation (70) enables the forces, F, in the individual piles to be determined using Equation (68). “The distribution of axial pile loads and the displacement of the pile cap are not greatly fected by neglecting the bending stiffness of the piles using this method. In many pile groups, bending stresses ae of secondary importance, and they need not be evaluated, provided the piles are nominally reinforced (e.g. to resist handling and driving stresses) ‘Computer programs fr structural analysis based on the stiffness method may be adapted to solve ple group problems (e.g. “Minipont’ and ‘SW Pile’ based on Sawko's analysis). While being ble to cope with complex geometry and boundary conditions, these programs suffer the same inherent disadvantages as the desk method described above, in thatthe restraint afforded by the soil isneglected. 2, WINKLER SOIL MODEL “The application of the Winkler soll model to the unalyssof single plesin well developed. However, asa consequence of the properties of the medium. the theory is not applicable to the accurate modelling ofthe interaction of piles in groups. To overcome this defect inthe re Jat reduction factors are applied tothe moduli of subgrade reaction to account for group ‘ition, Such factors are proposed by Prakash and Saran", based on tests of model piles in send. "Theses values ae given in Table 29, It should be appreciated that few other published field data exist to corroborate these salves. Table 20 Reduction factors for group action Pie scion os [3 Ihara Redonton for for m y a Figura 38 Extract from Naviec OM7"" group CRIA Repo 3 While deformations are approximately proportional to ny, bending moments depend on Tima) and they are thus relatively insensitive to the coefficient of subgrade reaction adopt- ed. Vertical, identical ples with pinned or iced heads For simple cases shown in Figure 38 with widely-spaced pile, a straightforward analysis may ‘be made ®, From Figure 38, the lateral loading a the ground line of each pile may be taken Ho Hoo, Mo= om He where Hg is the horizontal load on the pile group. “The piles may then be analysed as isolated members using the relationships outlined in “Appendix B, pages 38 1082. Groups with raking piles For this case, the concept of equivalent! ‘member in free air which gives both the: line. ngth may be introduced, defined as the length of the ime exial and lateral displacement atthe ground ‘Conon Tondo Dosen roses 1. Comput rela sities ator ere = i 3, Obi sctiensAAn- As at ep ares i 4. compute deton, moment ther ees acpne ung Egentions 291019) 4 = number les Defeged E roo ‘Case Piles wth flexible eap or hinged end condition 1, ProeedasinSep Coe 2. Compute eection and ment ted ‘poeunngeocen ye Ay Eaentomn avo). 2 Maximum shea oan at pt i and qunh = Hgninesh le ‘Case2 Piles with fixed head ov 3.1 Poulos method ‘tis not usually possible to satisfy both of these conditions at once, and the equivalent length may be assigned tothe controling deformation. Equivalent length may be calculated from the following formulae: 1. Axial loading, ‘Consider the compression of a column set Pec | Poa am AEe | Aab In practice, the equivalent pile length, Lec, falls within the range 0.5L to 1.0L, depending on the vertical distribution of shear stress. 2. Lateral loading Consider the deformation of acantilever @) 3h ‘Values of Ay and Bye for various sil conditions are given in Tables 100 19, pages 41 and 461050. Having defined the equivalent pile structure, the group is analysed using the stiffness method. For planar ple groups, Sawko's solution may be used (see page 65), Nair eral” present the elements fr the flexibility matrix of more general pile groups, wih different end Faity conditions. A computer program (STRAP II) based on this approach is also available. ‘The axial and shear forces, moments and deflections calculated using this method are only Tid forthe pile length above the ground line. The embedded portion should be analysed as “Appendix B, page 40. Using this typeof analysis, the calculation effort involved in dealing ‘with groups containing batter pilesis large. Ingeneral use of Winkler medium methods not recommended forthe analysis of pile groups, except forthe straightforward cases outlined above. The concept cannot be se yeloped further to model group action in a realistic manner, and litle further development of the method islikely. 3, ELASTIC CONTINUUM SOLUTIONS Tiable to over-predict the effects of soi/pile inte i tehavioar from linear elasticity, Poulos *” produced design charts which may be used to, txtrapolate the behaviour ofa singe pile to that of a group, and Bannerjee and Driscoll ‘s comprehensive computer program to predict pile group behaviour under general Groups of vertical piles Poulos *" calculated reduction factors, for groups comprising vertical piles only, and for soil Thich may be characterised by the two simple idealisations: E constant and E varying linearly swith depth. ‘The method of solution forthe case is to integrate Mindlin's expression ® for the stresses caused by a point load acting within an elastic continuum, and to predic the inter- Aifon between two piles. The interaction of any number of piles may then be computed by Tepeated superposition, Sotutions for square pile groups are presented (see Figure 39), ‘where the group reduction factor, Rp. is glven by ig _ defection of group Rn ™ Figg ~ defieston af singe pile carrying the sae load where de isthe displacement ofa single pile subjected to a unit horizontal load for the ‘appropriate finty conditions, AIA Repo 1s Itis known that for E increasingly linearly with depth, the effects of interaction are less marked. “The deflection of the k* pile may be calculated for a group of » free-headed piles from fs y= 09 (pean + HOD 3) it i where jy = the interaction factor between the jand the K* piles Seapeed common cre for srw argc an 33 Group redetion tr Figure 39 Reduction factors for the lateral deflection of square pile groups (Poulos™) in homogeneous soils ‘ta Repo ‘ ‘The total load on the grep may be calculated from the equation, Ho= EH m™ “The unit displacement ofa pile subjected to various loads and constraints may be calculated from the formulae given in Table 21. Table 21 Unit displacement of a lateraily-loaded pile Unit horizontal E = constant E=m displacement (isotropic case) | (non-homogencous case) h it ‘Ny (lateral force) he i EL mt h It Ny (moment loading) a a ze mo fy (pile head fixed) as an iE mo For ple groups, one ofthe following assumptions is generally made, depending on the degree of restraint afforded by the superstructure: 1. that of equal displacement of all piles. This requires the solution of n simultaneous ‘equations arising from Equations (73) and (74) 2. that of equal load inal piles. In this ease 11, = Higln and Equation (73) may be solved directly. Pile groups subject to moment loading and fixed-head pile groups may be analysed in as for teanner using the appropriate interaction factors (Figures 42 and 43). Inthe latter case, it ls reasonable to assume equal displacement of all piles. Values ofthe interaction factors, for ie ease, are given in Figures 40 1043. presents a simplification tothe above procedure. Ifthe loads in the piles are assumed to be equal, calculations may be carried out for a representative pile, usually one in {he centre ofan outer row, using the following expression forthe group reduction factor 1 Rant (ant) 5) ‘ “Thus, various group reduction factors may be calculated for the applied loading conditions: Raat Rast Rao» Ree ser contincs om pone 78 CCIRIA Report 108 Figure 40 Interaction factor, ‘groups of free: ‘subjected to (Poulos) in homogeneous soils ies so Figure 4t Interaction factors, yyy 20d Gime for free-headed piles 1d to moment inhomogeneous n ericson aor tt at ‘CARLA Report 10 Figure 42 Interaction factor, 0, for free-headed piles subjected tomoment (Poulos) in homogeneous soils {ARIA Report id B Figure 43 Interaction factor, ayy, for fixed-headed piles (Poulos™) in homogeneous soils ” 9 0 os 03 os 2 om oe CHRIA Repo 08 CrRtA Repo 08 “The deflection of afree-headed pile group may be calculated by combining Equations (50) and (51) with Equation (75). ‘o Ho. (ryan + £Rrosin) 6) Hs (nme wo, (manda Ra) o where He = total horizontal load on the group Rani = group reduction factor for deflection resulting from lateral load ‘Rune ™ group reduction factor for deflection resulting from moment Thi figs a= singe pile influence factors (see Figures 20025) Raw = group reduction factor for rotation from moment Equations (76) and (77) may be extended to calculate the defection of ples in group which projects above the mudline: Hg ‘ e é defiection of] (78) “G* Te (Raul +> Ral) + 5 Ramla + Raul) | + | free-standing oot ath asf rl, Expressions similar to Equations (76 to (78) may be used for nonhomogeneous sil of for eaeeretr loading. by inserting the appropriate influence and reduction factors. Poulos" Truodiuces yield displacement factors into these Equations to account for local yielding of ae tnexe factors are similar to those used for single piles, but are estimated (see Figures 30 to 33, pages 60 0.63) using the following equat Ho the es treatin capac, Hy, o single pil stn within pile group maybe esimated fom Hee = Ae (80) were iste proupetieny factor forlatea ond ‘Values of group efficiency factors for laterally loaded groups are not well established, and athe values presented in Table 22 are suggested a reasonable. These values are based on model tests by Prakash and Saran ™ and Oteo™. Table 22 Group efficiency factors for lateral loading Tie g ano) eye +1] cra ney or orig SS oom [os | oe | os | ve “Alternatively, to account for non-linear behaviour of the piles, a modification to Poulos's {interaction method is MI by Fockt and Koch?™. In this modification, itis assumed that aa reermation of the Ke pile, 1m may be calculated from a p-y analysis, and thatthe effect of {Re surrounding piles on the behaviour ofthe k® pile may be calculated from elastic theory. It Pr avatore tact assumed that local yielding ofthe soll lose tothe ple does not affect the displacements ata distance. Equation (73) may thus be written as : uy = RoE (Hy ary + RH) ; a en) where R= 1 % ‘The analysis is then carried out as before by solving the set of s generated by Equations (74) and (79). roma series of finite element analyses of flexible piles, Randolph ® shows that the pattem of lateral movement around pile maybe related tothe interaction factors proposed By Poulos. Randolph fis approximate expresions to ths normalised results of the finite clement analyses, which results in the following interaction factors wltaneous equations, E_\""r0 our s».(2) SA + cos") @) oy = 0.8 OnE (83) gy = ou OH es) ays on? 5) “These approximate expressions are sufficiently accurate for most practical purposes, unless the piles are very closely spaced, and they may realy be incorporated in a computer program. General pe groups Poulos shoves that the deformations of a battered pile may be calculated with sufficient Toniagy from the normal and axial components of the load, using solutions fore vertical ple aeeprgcd in an elastic half space. The interaction of two battered piles may be estimated by Considering the battered piles to be equivalent to two vertical piles atthe same spacing a Septh equal to one third cf tei vertical penetration. In addition, Poulos assumes that normal {oading on one batter pile causes deformation normal othe second pile, and similarly for axial deformation. Hence the deformation and loading of a general pile group of n ples may be related by the following expressions, (28) Hl -E - “The sub-matrices A, Band Care of ordern x nand he vectors are of order n. ‘Equation (86) may be solved, together withthe three equations for equilibrium forthe desired boundary conditions. “The coefficients ofthe sub-matrices are ‘Any = 4 089; 0891+ By sin COR Boj = Beyjeosy, sing; ~ By aun inv sing Gy = Masi ‘Argj =F ej sin Cony) ~ Ha cn CON HOW Biuj = Bey siny sing + By sun COFY) COMA iy = Ba sun S08% By cn SY, Boy = Byer COV Coy = By cen Tm axial deflection of asingle pile caused by unit axial load Ty = lteral deflection of singe pile caused by unit lateral load yy = lateral deflection of single pile caused by unit moment Bq = rotation of single pile caused by unit lateral load By _ = rotation of single pile caused by unit moment uy, = axial displacement iteration factor for piles Land j taj uaa = lateral displacement interaction factors for piles andj cen, ya = rotation interaction factors for piles andj vey Sateen ‘of piles | andj fom the vertical (positive ifin the direction of the applied loa EIRIA Report 13, 3.3 Computer programs Figure 44 Diseratisation of pte-soil interface used in PGROUP program (Bennerjee and Drisco!!™) CHRIA Report 108 PGROUP PGROUP is based on the boundary element method (Banné are modelled as cylindrical members with a uniform dis Figure 44), Mindlin's solution for a point force acting within a semi calculate stresses and displacements at any point within the medium. jee and Driscoll). The piles jon of stress on the elements (see finite solids used to ‘The program allows forthe estimation of forces and moments acting on piles within a planar group, embedded within an ewetic continuum, The continuum i essentially a homo- enous half space. However, by utilising approximstions based on analysis of a layered {taste system, the continaum may be treated as though the modulus increased linearly with epth (Gibson model), or as ifit were a two-layer system. The latter case may be employed forthe situation where there isa rigid base stratum. Limiting values are assigned to the shear and normal stresses acting on the piles, and al- though not used inthe analysis, the program prints a warning when caleulated values exceed those assigned. The ples are divided into segments (maximum of 11) and the analysis is valid {or piles witha length/diameter ratio not exceeding 50. The output ofthe program comprises the axial loads, shear forces and moments acting on the sections ofthe individual piles, together with the global stiffness of the pile group. ‘Aseries of ease histories was analysed using the program, and reasonable agreement was obtained in most cases. The results are broadly similar to those from a Poulos type analysis of tingle piles and square groups. However, the mathematics ofthe PGROUP method is con- sidered to have greater precision. The assignment of a value to Young's modulus of the soil Discrete ves a8, Seecbuton tends to large uncertainties. In many cases, the stessstrain behavior of soils markedly op linear’The assumption of linearity inbezeat in GROUP may lead to significant over ‘Prediction of interactive elects within the soil between different pars of the foundation. Care Thould be taken only to use PGROUP when the other assumptions inherent inthe program sre reasonably fathol representations of the ste under consideration. The program does not {ke account of external influences (e.g. consolidation) which may cause sol deformation. ‘A parametric study using PGROUP of a range of vertical pile group geometries was undertaken by Butterfield and Douglas. In this study, a range of square pile groups, from Toco 8 8, was analysed fr varying lengh to diameter ratio, LID modular ratio, EG, ‘and depth of stratum topile length, Ly “The results are presented in the form of non-dimensional flexibility coefficients, F*y Pueophan ey) Por ooien « D 9) (9) oy “The values fy are the flexibility coefficients which may be inserted inthe exibilty mstix for vertical pile group [ [ be A | ti] ul -10 fe fl |e (2) ® 0 fz: fa] 4. where fas = fas ‘Values ofthe flexibility coefficients may be read off graphs in Reference 12, enabling the slobal behaviour of vertical pile groups tobe rapid assessed with sufficient accuracy for many designs. “Asnoted above, the PGROUP program has been extended to accommodate Gibson soil and a two-layer soll system (PGROUP 3). “Accomplementary program which models yielding ofthe soil (PLLYLD) is under develop- ‘ment by the Department of Transport, Highways Computing Division. Other programs Several other programs ae available (see Appendix C) based on the assumption of aerpedded in an elastic continuum. The list of programs discussed is not exhaustive, an not within the scope of this Report to assess their reliability. Poulos produced a program forthe design of general pile groups based on Equation (86). “Tha program (DEFFIG) can ascommodate vertical, horizontal and moment loads for pile caer botalnng battered piles in notrope or nonkiomogencous,yeiding soi. The program Hare rdally svalabe in Australia, but notin the United Kingdom. Wood published 8 ‘gram forthe analysis of vertical elastic pile groups embeded in an elastic or an lati roding medium A program based onthe normalisation of lage numberof ine element sina has been published by Randolph ©. This program (PIGLET) i simple and quick 19 aaa se it does not involve the soltion of large matrices resulting from the discretisation rin races embedded in an elastic continuum. The program can cope with general loading of {he pile group (including torsion), piles raked in any aiection, and soils where the stiffness woe ecarly wth depth. The analysis of pile group using PIGLET is dequate for most {esign purpotes, butt is restricted to groups of long flenible piles. ARIA Report Appendix D Case histories 12 Pile in den: CRIA Report 03 “The case histories presented here, with the results of back-analyses using the design methods described in this Report, are both to ilustrate the methods recommended and to indicate the validity ofthe calculations. 1. SINGLE PILES ‘Two case histories are taken from the literature to illustrate the analysis of single piles. The pile test results analysed a Dense sand: Mustang Island (Reese, Cox and Koop) oft clay: Mississippi Delta (McClelland and Focht™) ‘The sol data from these ple tests are somewhat limited, because the stiffness ofthe soil was ‘assessed from unconsolidated undrained triaxial tests, unconfined compression tests, oF it~ Jiu SPT N values, The empirical relationships given inthis Report were used to assess the saiadulus or evefficient of subgrade reaction of the soil. The methods of analysis discussed are {generally restricted to linear load/deformation response. Hence the calculations surm, Errrised in Table 6 (page 25) are compared with the observed deformation at small loads. ‘The calculated moments are also summarised in Table 6. Generally, calculated defor- ‘compare reasonably well with observed values, The elastic continuum solutions ily give lower computed deformations. However, the accuracy of all the calculations is Eirgely dependent on the correct selection ofthe soll stiffness. Allthe methods of analysis ‘under estimate the maximum measured bending moment by up to 20%. Pile test data obtained by Reese, Cox and Koop" at Mustang'Island are used in this example. ‘The assumption of an elastic continuum isnot valid at high lateral load, hence solutions for the initial load only have been presented. Data Geometry ‘Loning and deflection Soll properties For Embedded length of H, = 894N se fine sand with alitle =2.0m M, = 23KNm silt Diameter = 0.61 m Measured deflection D, = 8010 1W% Eglp SNSTMN | gS moment N= 40 blows/300 mm at =125kNm oe ‘and for High groundwater level Hy = 220KN M, = 56kNm Measured deflection =3em and maximum moment = 374kNm n Dense sand elastic analysis Randolph analysis i OMNI! i. A+ yy es im? =7.7MNim? (Equation (60), page 57) Pe 0S 659), pages 56 and 57) G. = 18.4 MNim* J, = 4.77 (Equations (57) and Soll modulus Deformation Bending moment For initial loading, N= 40blows/300 mmat 6.1 m From Section 4.1.3, (page 18), E. = 2500 x 40 = 100 MNimt m= 16.4MNim* Poulos analysis Ky 32.5 x 10" Equation (50), pageS3 | Poulos and Davis (Equation (49), page $3) from Figures?" and 24, | Myuc™ 0.05 He (Fig 8.40) Lid=35 pages 56.and 57 +06 Mo (Fig8.26) ea = 107kKNm Te 2000 «24mm fp = 24,6 NI? uy 93S mm (Equation ($8), page $6) | page $9 Eom Equation (8), pa | Mon” 10188 From Figures 28 and 29, Dense sand ~ subgrade reaction method ‘Corfficlent of subgrade reaction ‘Deformation Bending moment Foriniial loading From Reese eta" (Equation (7), puge 20) my = 30MNIm? T =14m (Equation (43), page 42) Zac 15 For high loading From Terzaghi™ (Table 4, page 20) m= 10MNim? cor fron. Equation (8) on my 930% E 5 =9MNim* T =17m Zan ™ 12 Ayo 22.435 By = 1.623 4mm Pm (Equation (33), page 40) (Equation (33) page 40) For long piles, the maximum value of Ag = 0.75, and the corresponding value of By, = 0.66 Mag = 13KNm (Equation (35) page 40) Max 3284N Comment on dense sand case history ‘The top 0.75 mf sand was newly deposited to replace an excavated clay layer. Itis unlikely that ths sand was as dense as the underlying deposit. This may partly explain the under- prediction of wand M. ‘CRIA Report 13, 1.3 Pilein soft clay CURIA Rept 103 Pile text data obtained by McClelland and Focht™ in the Mississippi delta are used in this example. Data [Geometry Loading and deftection Soll properties Hy = 267KN ‘Normally-consolidated clay| with average cy = 37 KNim? at6.1 m, Below 6.1 m, «cil Embedded length Mo= ~325KNm ere eiatnitier =29m Diameter = 0.61 m Deflection at 1.83. above mud-line = 30mm Eqlp = 466MNen? | Maximum moment = 490kN/m at$.§ m below mud-line Soft clay elastic analysis ‘Soll modulus ‘Deformation Bending moment m =25MNic? Poulos analysis (Equation (49), page 53) Ky 3.0% 10° ud=37 Randolph analysis G =_E m+n 49MM mt = 1.1 MN? (Equation (60), page 57) be =05 i =93m ‘(Equation (57), page 56) G. = 5.0TMNIn! From Section 4.1.1, page 18 E = 400x37=14.8MNimt From Equations (80) ard (51), page 33 ig 120 aya = 800 I= 6000 uy = 15.7mm and Oya 43x 10rd Deflection at 1.83 m above mud-line = 24 mm E'y = 2.04 * 10°GNimt (rom Equation (58), page 57) From Equations (54) and (55), page 56, uy = 174mm 0, 3% 10° rad Deflection at 1.83 m_ above mud-line = 23 mm Poulos and Davis ‘Myx = 0.09% Hol (Fig 8.40) +0.6 Mo (Fig 8:26) = 355kNam From Figures 28 and29, page 59 Mya = 300 KN at adepth of 3.7m at 2 Soft ta Subgrade reaction method [Cocmctent of subgrade Deformation Bending moment reaction ‘Ka =400%37 | Seciion | | From Equations (33) ad (34), | Equation (35), =14gMNim? | 42.1, } | page 40 page 40 page 207 | 4,,= 2.435, B= 1.623 148 uy = B3mm zr M gy A MN Aig 1.62, Bry = 1.5 0, = 4.10% 10% rad 0s 2 T =2.86m / Equations 10 78 (and (45) } | Defections at 1.89 m ts 380 page a2 above medline = 31mm |2.0 350 Zou = 8 Meag = 400 KN en ata depth of 4.8m Comment on softclay case history thas been noted by Randolph thatthe test data reported are not completely consistent ‘with the reported shear force and moment diagrams, A lateral load of 300 KN, rather than 1267 KN, was probably applied. 2, PILE GROUPS ‘The methods of analysis, discussed in this Report, are applied in this ection tothe piled foundations of abridge abutment. The computations illustrate the importance of obtaining & {ealistic messure of the lateral stiffness of the sol, andthe rather sweeping simplifying, ‘assumptions which usually have to be made in des “Anoverbridge for the A259 was constructed on the fload plain of the River Ouse. The bridge foundations were piled to the underlying chalk, using West's Hardrive piles. CIRiA ‘undertook a comprehensive instrumentation programme to monitor the behaviour of the east abutment, igure 45 shows the general arrangement and a section through the abutment, together Bictails ofthe soil profile. The loading on the abutment was assumed to be: 1, dead load of the bridge 2 aprism of PFA back-fill at the abutment 3. lateral load equivalent to that of a fluid of density S kN/m’. ‘These loadings are shown in Figure 46. Calculations were carried out using four methods of unalysis: 1. static resolution of forces 2, the stiffness method 3, Poulos method 4. PGROUP program 2.0. ‘The bridge foundations were designed by the Highways and Transportation Department of East Sussex County Council, using the stiffness method, and later checked using PGROUP. “The calculations using Poulos's analysis were carried out assuming the stiffness ofthe soll increased linearly with depth, No fn-ttu measurements of stiffness were made during the site investigation, and a value was assessed as follows: ‘Average ¢, at Sm depth (in-situ vane test results) was 28.6 KN/m* Taking Es = 200 x 28.6 = ST20KN/mn* ‘Thus m= 1140KNin? CIRIA Rept 108 lo. $+ os| 10 20 ce Mile From oo wm Plan Section ‘MADE GROUND om 200. son sit shy SAND Figure 45 Nowhaven overbridge: com abutment details and nom 0”. borehole log (Reddaway ‘and Elson™) CRIA Report 03 In the PGROUP program used, only one soll layer could be modelled. An average ‘modulus of S MN/m' was used (ie. a value corresponding to the stiffness of the upper layers Of the alluvium), The sensitivity of the analyses was checked by repeating the computation with a value of £ = 10 MN/m?, Doubling the soil modulus made no significant difference to the computed forces, and it only marginally decreased the displacements of the pile ag ‘The computed axial loads at end of construction and pile cap deformations for the various methods of analysis are presented as Tables 7 and 8, respectively (see page 26). eck trata Abwiment Figure 48 Newhaven overbridge: bridge dead loading atm (Reproduced by fount permission of the County ‘Surveyor, East Sussex County Counc 105m0. CCIRIA Report | Appendix E Computer program references ‘Single Piles Souree pry analysis REESE™ Pile Groups, SW Pile Midland Road Construction Ur ‘Warwickshire County Council (Developed from SAWKO") Minipont Department of Transport Highways Computing Division PGROUP3.0 (Earlier versions no Tonger supported) PILYLD Department of the Environment (Notyet released) Highways Computing Division PIGLET RANDOLPH® LAWPILE. woop" DEFPIG POULOS ‘Australian program Availability Not currently available in the UK (Ove Arup and Partners, Warwick Various Bureaux ‘Various Bureaux University of Cambridge London United Computing, Systems Engineering Laboratory, University of Sydney (Not currently available in Britain)

You might also like