All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Related Documents

Share

Description

Simple method for dynamic stiffness and damping of floating pile groups

Transcript

Makris, N. & Gazetas, G. (1993).
Lotechnique
43, No. 1, 135-150
Displacement phase differences in a harmonically oscillating pile
N. MAKRIS* and G. GAZETAS*
Analytical solutions are developed for harmonic wave propagation in an axially or laterally oscil- lating pile embedded in homogeneous soil and excited at the top. Pilesoil interaction is realisti- cally represented through a dynamic Winkler model, the springs and dashpots of which are given values based on results of finite element analyses with the soil treated as a linear hysteretic contin- uum. Closed form expressions are derived for the phase velocities of the generated waves; these are compared with characteristic phase velocities in rods and beams subjected to compression- extension (axial) and flexural (lateral) vibrations. The role of radiation and material damping is elu- cidated; it is shown that the presence of such damping radically changes the nature of wave pro- pagation, especially in lateral oscillations where an upward propagating (reflected) wave is generated even in a semi-infinite head-loaded pile. Solutions are then developed for the phase differences between pile displacements at various depths. For most piles such differences are not significant and waves emanate nearly simultaneously from the periphery of an oscillating pile. This conclusion is useful in analysing dynamic pile to pile interaction, the consequences of which are
shown
in this Paper.
KEYWORDS: deformation; dynamics; piles; vibration;
WBWS
Des solutions analytiques ont ctb d&velopp&s afin d’Ctudier la propagation harmonique des ondes dans un pieu ancri dans un sol homogbe, excite zi son sommet, et oscillant IaGralement et axi- alement. L’interaction sol-pieu est bien reprbent&e par le modkle dynamique de Winkler dont les ressorts et ‘pistons’ sont affect(ts de valeurs calcu- l&s P partir d’analyses par Blbments finis, le sol &tant supposit g hyst&sis linbaire. Des expressions de forme bquivalente sont d&iv&s pour calculer les vitesses de phase des ondes induites. Elles sont compari?es aux vitesses en phase caracti?ristiques obtenues dans des barres et poutres soumises I des vibrations de type compression-extension (axiales) ou de type flexion (IatCrales). Le ri31e de la radi- ation et celui du ‘damping’ du matkriau sont expli- qub; l’on
montr
que l’existence d’un ‘damping’ modifie totalement la nature de la propagation des ondes, tout particuli&ement lors d’oscillations late- rales oi
un
onde se propageant vers le haut apparait, m@me dans un pieu semi-h&i charge P son sommet. Des solutions permettant de calculer les differences de phase entre les deplacements des pieux P differentes profondeurs sont alors develop- pi?es. Pour la plupart des pieux, ces differences ne sont pas signiticatives et les ondes Cmergent $ peu pr&s simultanCment de la p&iphi?rie du pieu oscil- lant. Cette conclusion est trb utile pour I’analyse de I’interaction dynamique pieu-pieu dont I’article d&rite les consCquences.
INTRODUCTION
This work was prompted by the need to develop a deeper understanding of some of the wave pro- pagation phenomena associated with the dynamic response of piles and pile groups. For example, it is well known (Kaynia & Kausel, 1982; Nogami, 1983; Novak, 1985; Roesset, 1984) that two neighbouring piles in a group may affect each other so substantially that the overall dynamic behaviour of the group is vastly different from that of each individual pile. This pile to pile inter- action is frequency-dependent and is a conse- quence of waves that are emitted from the Discussion on
this
Paper closes 1 July 1993; for further details see p. ii. * State University of New York at Buffalo and Nation- al Technical University of Athens. periphery of each pile and propagate until they ‘strike’ the other pile. As an example, for a square group of 2 x 2 rigidly-capped piles embedded in a deep homoge- neous stratum Fig. 1 shows the variation with fre- quency of the vertical and horizontal dynamic group stiffness and damping factors, defined as the ratios of the group dynamic stiffness and dashpot coefficients, respectively, to the sum of the static stiffnesses of the individual solitary piles. At zero frequency the stiffness group factors reduce to the respective static group factors (also called ‘efficiency factors’ by geotechnical engineers) which are invariably smaller than unity. The continuous curves in Fig. 1, adopted from the rigorous solution of Kaynia & Kausel (1982), reveal that, as a result of dynamic pile to pile 135
36
MAKRIS AND GAZETAS
“r
Sd =
5
or
Sd =
10
Sd = 5 6-
4- 2- 0 0 I I I I , 0.2 0.4 0.6 0.6 1.0 ao =
wdlV, 4
Sd =
5
0 Sd=O 0 0 0 o Sd = 5 ALa 0 I I I I I 0.2 0.4 0.6 0.6
1.0 a,, =
cudV (b)
Fig. 1. Normal&d vertical and lateral impedances of a 2 x 2 pile group (E,/E, = 1000,
L/d =
15, v = O-4, l = O-05): solid curves = rigorous solution of Kaynia & Kausel (1982); points = simplified solution of
:
(a) Dobry & Gazetas (1988); (b) Makris & Gazetas (1992) (impedances are expressed as t + iu, Q; subscripts z and x refer to vertical and horizontal mode* KC’) and QC’) of the single (solitary) pile) are the total dynamic stiffness and damping of the Qpile group; ZP) is the static stitTuess
interaction, the dynamic stiffness group factors achieve values that may far exceed the static efh- ciency factors, and may even exceed unity. Both stiffness and damping factors are not observed in the single pile response. Specifically, the peaks of the curves occur whenever waves srcinating with a certain phase from one pile arrive at the adjac- ent pile in exactly opposite phase, thereby indu- cing an upwards displacement at a moment when the displacement due to this pile’s own load is downwards. Thus, a larger force must be applied to this pile to enforce a certain displacement amplitude, resulting in a larger overall stiffness of the group as compared to the sum of the individ- ual pile stiffnesses. Also shown in Fig. 1 as points are the results of a very simple analytical method of solution pro- posed by Dobry & Gazetas (1988) and further developed by Makris & Gazetas (1992), Makris, Gazetas & Fan (1992) and Gazetas & Makris (1991). The method introduces a number of physi- cally motivated approximations, and was src- inally intended merely to provide a simple engineering explanation of the causes of the numerically observed peaks and troughs in the dynamic impedances of pile groups. Yet, as is evident from the comparisons shown in Fig. 1, the results of the method plot remarkably close to the rigorous curves for all three pile separation distances considered (two, five and ten pile diameters). Even some detailed trends in the group response seem to be adequately captured by the simple solution. Further successful com- parisons are given in the above-mentioned refer- ences. The fundamental idea of this method is that the
PHASE DIFFERENCES IN FIXED HEAD PILE
137 displacement field created along the sidewall of an oscillating pile (in any mode of vibration) pro- pagates and affects the response of neighbouring piles. It is assumed that cylindrical waves are emitted from the perimeter of an oscillating pile, and propagate horizontally in the r direction only. This hypothesis is reminiscent of the shear- ing concentric cylinders around statically loaded pile and pile groups assumed by Randolph & Wroth (1978, 1979), and is also similar to the dynamic Winkler assumption introduced by Novak (1974) and extensively used with success in dynamic analyses of pile groups. It is further assumed that these cylindrical waves emanate simultaneously from all points along the pile length; hence for a homogeneous deposit they spread out in phase and form a cylindrical wave- front, concentric with the generating pile (unless the pile is rigid, the amplitude of oscillation along the wavefront will be a (usually decreasing) func- tion of depth). The resulting dynamic complex- valued pile to pile interaction factor for vertical oscillation takes the simple form (Dobry 8z Gaze&s, 1988) ~V=~)“‘exp(-@{)exp(-io~) (1) where r,, =
d/2
is the radius of the pile, S is the axis to axis distance of the piles, and V, and /l are the S wave velocity and hysteretic damping ratio of the soil respectively. The most crucial of the introduced simplifying assumptions is that the waves created by an oscil- lating pile emanate simultaneously from all peri- metric points along the pile length, and hence, for a homogeneous stratum, form cylindrically expanding waves that would ‘strike’ an adjacent pile simultaneously at various points along its length, i.e. the arriving waves are all in phase, although their amplitudes decrease with depth. The question arises as to whether the satisfac- tory performance of such a simple method is merely a coincidence (e.g. due to cancellation of errors), or a consequence of fundamentally sound physical approximations. Answering this question was one of the motives for the work reported in this Paper. Hence, the first objective was to inves- tigate whether or not this key assumption of syn- chronous wave emission from an oscillating pile is indeed a reasonable engineering approximation and, if it is, for what ranges of problem param- eters. A second, broader, objective of the Paper is to obtain a deeper physical insight into the nature of wave propagation in a single harmonically oscil- lating pile embedded in homogeneous soil. To this end, realistic dynamic Winkler-type models for vertically and horizontally oscillating single piles are developed, from which analytical solu- tions are derived for the apparent phase velocities of the waves propagating along the pile and for the variation with depth of pile displacements and phase angle differences. A limited number of rigorous finite element results are also obtained to substantiate the findings of the Winkler model. It is shown that the apparent phase velocities for typical piles are indeed quite large, and the dis- placement phase differences correspondingly small, especially within the upper, most active part of the pile. It is also found that at very high frequencies the phase velocities in a pile embed- ded in homogeneous soil become asymptotically equal to the wave velocities of an unsupported bar or beam in longitudinal and flexural oscil- lations.
PROBLEM DEFINITION
The problem studied involves a single floating pile embedded in a uniform halfspace and sub- jected at its head to a harmonic loading of circu- lar frequency w. The pile is a linearly elastic flexural beam of Young’s modulus
E, ,
diameter
d,
cross-sectional area A,, bending moment of inertia I, and mass per unit length m. The soil is modelled as dynamic Winkler medium, resisting pile displacements through continuously distrib- uted linear springs (k, or k,) and dashpots (c, or c,), as shown in Fig. 2 for horizontal (x) and verti- cal (z) motion. For the problem of lateral vibra- tion (horizontal motion), the pile is considered to be fixed-head (zero rotation at the top). The force to displacement ratio of the Winkler medium at every depth defines the complex-valued imped- ances
k, iwc,
(vertical motion) or
k, iwc,
(horizontal motion), i = J( - l), where c, and c, would, in general, reflect both radiation and material damping in the soil.
k,
and
k,
are in units of stiffness per unit length of the pile (i.e. [F] CL]-‘); they correspond to the traditional subgrade modulus (in units [FJ CL]-“) multiplied by the width (diameter)
d
of the pile. Frequency-dependent values are assigned to these uniformly-distributed spring and dashpot coefficients, using the following algebraic expres- sions developed by matching the dynamic pile- head displacement from Winkler and from dynamic finite-element analyses (Roesset & Angelides, 1979; Blaney, Kausel & Roesset, 1976; Dobry et al., 1982; Gazetas & Dobry, 1984a, 1984b)
k E
0.6E,(l + - ,/a,,) (2a) c~ x
Cr)radiation + Cz)hysteresis
(2b)
138
MAKRIS AND GAZETAS
Fig. 2. Dynamic Winkler model for axially and laterally oscillating pile k, z 1.2E, ‘.X E
(CxLliation + (Cxhystcresis
(24 z 2dp+ + ( >“4]~;1,4 +2/I b
(2d) where B is hysteretic damping, ps is mass density,
E,
is Young’s modulus, V, is S-wave velocity of the soil, a, =
ad/l/
and
V,,
is an apparent veloc- ity of the compressionextension waves, called ‘Lysmer’s analogue’ velocity (Gazetas & Dobry, 1984a, 1984b) 3.4 v,, = n(l - v)v, (3) where v is the Poisson’s ratio of the soil. For an average typical value v = 0.4, equation (3) gives
V
La z
1.8 V, and equation (2d) simplifies to Similar springs and dashpots can be obtained using Novak’s plane-strain elastodynamic solu- tion for a rod oscillating in a continuum (Novak, 1974, 1977, 1985; Novak et al., 1978). Novak’s results would be exact for an infinitely long, infi- nitely rigid rod fully embedded in a continuum space. In contrast, equations (2b) and (2d) for radiation damping are derived in two steps
(4 (4
their form is determined from a simple one- dimensional ‘cone’ model (Gazetas & Dobry, 1984a; Gazetas, 1987; Wolf, 1992) which resembles Novak’s model but does allow for some non-zero vertical deformation of the soil during lateral motion, as is appropriate due to the presence of the stress-free surface and to the non-uniformity with depth of pile deflex- ions the numerical coefficients of the two expres- sions are then calibrated by essentially curve- fitting rigorous finite element results for a variety of pilesoil geometries and properties, as well as for different loading conditions (Gazetas & Dobry, 1984a; Gazetas, 1987; Wolf, 1992). The spring constants, however, are derived solely through curve-fitting, i.e. by matching pile- head stiffnesses of the Winkler and the finite

We Need Your Support

Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks