Response of Buried Pipelines Subject to Earthquake Effects

8/11/2019 Response of Buried Pipelines Subject to Earthquake Effects

1/276

3

Response of Buried PipelinesSubject to Earthquake Effects

M.J. ORourkeX. Liu

Monograph Series

Multidisciplinary Center for Earthquake Engineering ResearchA National Center of Excellence in Advanced Technology Applications


2/276

The Multidisciplinary Center for Earthquake Engineering Research

The Multidisciplinary Center for Earthquake Engineering Research (MCEER)is a national center of excellence in advanced technology applications that is

dedicated to the reduction of earthquake losses nationwide. Headquartered atthe State University of New York at Buffalo, the Center was originally establishedby the National Science Foundation (NSF) in 1986, as the National Center forEarthquake Engineering Research (NCEER).

Comprising a consortium of researchers from numerous disciplines and in-stitutions throughout the United States, the Centers mission is to reduce earth-quake losses through research and the application of advanced technologies thatimprove engineering, pre-earthquake planning and post-earthquake recovery strat-egies. Toward this end, the Center coordinates a nationwide program ofmultidisciplinary team research, education and outreach activities.

Funded principally by NSF, the State of New York and the Federal HighwayAdministration (FHWA), the Center derives additional support from the FederalEmergency Management Agency (FEMA), other state governments, academicinstitutions, foreign governments and private industry.


3/276

i

RESPONSE OF BURIED P IPEL INES

SUBJECT TO EARTHQUAKE EFFECTS


4/276

i i


5/276

ii i

by Michael J. ORourkeXuejie Liu

RESPONSE OF BURIED P IPEL INES

SUBJECT TO EARTHQUAKE EFFECTS


6/276

iv

Copyright 1999 by the Research Foundation of the State University of NewYork and the Multidisciplinary Center for Earthquake Engineering Research. Allrights reserved.

This monograph was prepared by the Multidisciplinary Center for Earthquake

Engineering Research (MCEER) through grants from the National Science Foun-dation, the State of New York, the Federal Emergency Management Agency, andother sponsors. Neither MCEER, associates of MCEER, its sponsors, nor any per-son acting on their behalf:

a. makes any warranty, express or implied, with respect to the use of anyinformation, apparatus, method, or process disclosed in this report or thatsuch use may not infringe upon privately owned rights; or

b. assumes any liabilities of whatsoever kind with respect to the use of, orthe damage resulting from the use of, any information, apparatus, method,or process disclosed in this report.

Any opinions, findings, and conclusions or recommendations expressed in thispublication are those of the author(s) and do not necessarily reflect the views ofMCEER, the National Science Foundation, Federal Emergency ManagementAgency, or other sponsors.

Information pertaining to copyright ownership can be obtained from the authors.

Published by the Multidisciplinary Center for Earthquake Engineering Research

University at BuffaloRed Jacket QuadrangleBuffalo, NY 14261Phone: (716) 645-3391Fax: (716) 645-3399email: [email protected] wide web: http://mceer.eng.buffalo.edu

ISBN 0-9656682-3-1

Printed in the United States of America.

Jane Stoyle, Managing EditorHector Velasco, IllustrationJennifer Caruana, Layout and CompositionJenna Tyson, Layout and CompositionHeather Kabza, Cover DesignAnna J. Kolberg, Page Design

Cover photographs provided by M. ORourke and T. ORourke.

MCEER Monograph No. 3


7/276

v

F o r e w o r d

Earthquakes are potentially devastating natural events which threatenlives, destroy property, and disrupt life-sustaining services and societalfunctions. In 1986, the National Science Foundation established theNational Center for Earthquake Engineering Research to carry out sys-

tems integrated research to mitigate earthquake hazards in vulnerablecommunities and to enhance implementation efforts through technol-ogy transfer, outreach, and education. Since that time, our Center hasengaged in a wide variety of multidisciplinary studies to develop solu-tions to the complex array of problems associated with the developmentof earthquake-resistant communities.

Our series of monographs is a step toward meeting this formidablechallenge. Over the past 12 years, we have investigated how buildingsand their nonstructural components, lifelines, and highway structures

behave and are affected by earthquakes, how damage to these structuresimpacts society, and how these damages can be mitigated through inno-vative means. Our researchers have joined together to share their exper-tise in seismology, geotechnical engineering, structural engineering, riskand reliability, protective systems, and social and economic systems tobegin to define and delineate the best methods to mitigate the lossescaused by these natural events.

Each monograph describes these research efforts in detail. Each ismeant to be read by a wide variety of stakeholders, including academi-

cians, engineers, government officials, insurance and financial experts,and others who are involved in developing earthquake loss mitigationmeasures. They supplement the Centers technical report series by broad-ening the topics studied.

As we begin our next phase of research as the Multidisciplinary Cen-ter for Earthquake Engineering Research, we intend to focus our effortson applying advanced technologies to quantifying building and lifelineperformance through the estimation of expected losses; developing cost-effective, performance-based rehabilitation technologies; and improving

response and recovery through strategic planning and crisis management.These subjects are expected to result in a new monograph series in thefuture.

I would like to take this opportunity to thank the National ScienceFoundation, the State of New York, the State University of New York at


8/276

vi

George C. Lee

Director, Multidisciplinary Center

for Earthquake Engineering Research

Buffalo, and our institutional and industrial affiliates for their continuedsupport and involvement with the Center. I thank all the authors whocontributed their time and talents to conducting the research portrayedin the monograph series and for their commitment to furthering our com-

mon goals. I would also like to thank the peer reviewers of each mono-graph for their comments and constructive advice.

It is my hope that this monograph series will serve as an importanttool toward making research results more accessible to those who are ina position to implement them, thus furthering our goal to reduce loss oflife and protect property from the damage caused by earthquakes.


9/276

vii

Foreword ......................................................................... vPreface ........................................................................... xiAcknowledgments ......................................................... xvAbbreviations .............................................................. xviiNotations ...................................................................... xix

1 Seismic Hazards and Pipeline Performance inPast Earthquakes ..................................................................... 1

1.1 Seismic Hazards ........................................................................... 11.2 Performance in Past Earthquakes ................................................ 21.3 Empirical Damage Relations ....................................................... 3

1.3.1 Wave Propagation Damage ............................................... 3

1.3.2 PGD Damage.................................................................... 61.4 System Performance ................................................................... 11

2 Permanent Ground Deformation Hazards .............. 132.1 Fault ............................................................................................. 132.2 Landslide ..................................................................................... 162.3 Lateral Spreading ........................................................................ 20

2.3.1 Amount of PGD .............................................................. 212.3.2 Spatial Extent of Lateral Spread Zone............................... 252.3.3 PGD Pattern .................................................................... 27

2.4 Seismic Settlement ..................................................................... 30

3 Wave Propagation Hazards ........................................... 333.1 Wave Propagation Fundamentals ........................................ 333.2 Attentuation Relations ................................................................ 353.3 Effective Propagation Velocity .................................................. 38

3.3.1 Body Waves .................................................................... 383.3.2 Surface Waves................................................................. 39

3.4 Wavelength ................................................................................. 423.5 Ground Strain and Curvature Due to Wave Propagation ...... 443.6 Effects of Variable Subsurface Conditions ............................... 46

C o n t e n t s


10/276

viii

3.6.1 Numerical Models........................................................... 473.6.2 Simplified Model ............................................................. 543.6.3 Comparison..................................................................... 55

4 Pipe Failure Modes and Failure Criterion ................ 59

4.1 Continuous Pipeline ................................................................... 594.1.1 Tensile Failure ................................................................. 594.1.2 Local Buckling ................................................................ 614.1.3 Beam Buckling ................................................................ 624.1.4 Welded Slip Joints ........................................................... 67

4.2 Segmented Pipeline .................................................................... 694.2.1 Axial Pull-out .................................................................. 714.2.2 Crushing of Bell and Spigot Joints .................................... 724.2.3 Circumferential Flexural Failure and Joint Rotation.......... 73

5 Soil-Pipe Interaction .......................................................... 775.1 Competent Non-Liquefied Soil ................................................. 77

5.1.1 Longitudinal Movement .................................................. 795.1.2 Horizontal Transverse Movement .................................... 805.1.3 Vertical Transverse Movement, Upward Direction .......... 815.1.4 Vertical Transverse Movement, Downward Direction ..... 83

5.2 Equivalent Stiffness of Soil Springs ........................................... 845.2.1 Axial Movement .............................................................. 855.2.2 Lateral Movement in the Horizontal Plane ...................... 855.2.3 Vertical Movement .......................................................... 87

5.3 Liquefied Soil .............................................................................. 87

6 Response of Continuous Pipelines toLongitudinal PGD ............................................................... 91

6.1 Elastic Pipe Model ...................................................................... 926.2 Inelastic Pipe Model ................................................................... 97

6.2.1 Wrinkling ...................................................................... 1006.2.2 Tensile Failure ............................................................... 1016.3 Influence of Expansion Joints .................................................. 1026.4 Influence of an Elbow or Bend ............................................... 106

7 Response of Continuous Pipelines toTransverse PGD ................................................................ 113

7.1 Idealization of Spatially Distributed Transverse PGD .......... 1157.2 Pipeline Surrounded by Non-Liquefied Soil .......................... 116

7.2.1 Finite Element Methods ................................................ 1177.2.2 Analytical Methods ....................................................... 1307.2.3 Comparison Among Approaches ................................... 1377.2.4 Comparison with Case Histories .................................... 1397.2.5 Expected Response ........................................................ 140


11/276

ix

7.3 Pipelines in Liquefied Soil ....................................................... 1417.3.1 Horizontal Movement ................................................... 1427.3.2 Vertical Movement ........................................................ 143

7.4 Localized Abrupt PGD ............................................................. 147

8 Response of Continuous Pipelines to Faulting ........... 1498.1 Strike-slip Fault ....................................................................... 150

8.1.1 Analytical Models .........................................................1508.1.2 Finite Element Models ................................................... 1578.1.3 Comparison Among Approaches ................................... 1628.1.4 Comparison with Case Histories .................................... 163

8.2 Normal and Reverse Fault ....................................................... 166

9 Response of Segmented Pipelines to PGD ........... 1679.1 Longitudinal PGD ..................................................................... 168

9.1.1 Distributed Deformation................................................1689.1.2 Abrupt Deformation ...................................................... 170

9.2 Transverse PGD ....................................................................... 1719.2.1 Spatially Distributed PGD ............................................. 1719.2.2 Fault Offsets .................................................................. 174

10 Response of Buried Continuous Pipelinesto Wave Propagation ..................................................... 179

10.1 Straight Continuous Pipelines ................................................. 17910.1.1 Newmark Approach ......................................................18010.1.2 Sakurai and Takahashi Approach ..................................18110.1.3 Shinozuka and Koike Approach .................................... 18210.1.4 M. ORourke and El Hmadi Approach ...........................18310.1.5 Comparison Among Approaches ................................... 18810.1.6 Comparison with Case Histories .................................... 189

10.2 Bends and Tees ......................................................................... 19110.2.1 Shah and Chu Approach ............................................... 19110.2.2 Shinozuka and Koike Approach .................................... 19410.2.3 Finite Element Approach ............................................... 19510.2.4 Comparison Among Approaches ................................... 196

11 Response of Segmented Pipelinesto Wave Propagation ..................................................... 199

11.1 Straight Pipelines/Tension ....................................................... 199

11.2 Straight Pipelines/Compression ............................................... 20411.3 Elbows and Connections ......................................................... 20711.4 Comparison Among Approaches ............................................ 20911.5 Effects of Liquefied Soil ............................................................ 211


12/276

x

12 Countermeasures to MitigateSeismic Damage ............................................................... 215

12.1 Routing and Relocation ........................................................... 21512.2 Isolation from Damaging Ground Movement ....................... 21612.3 Reduction of Ground Movements .......................................... 21712.4 High Strength Materials ........................................................... 21812.5 Flexible Materials and Joints ................................................... 219

References ...................................................................223Author Index ...............................................................237Subject Index ...............................................................241Contributors ................................................................249


13/276

xi

Buried pipeline systems are commonly used to transport wa-ter, sewage, oil, natural gas and other materials. In the contermi-nous United States, there are about 77,109 km (47,924 miles) of

crude oil pipelines, 85,461 km (53,114 miles) of refined oil pipe-lines and 67,898 km (42,199 miles) of natural gas pipelines (FEMA,1991). The total length of water and sewage pipelines is not readilyavailable. These pipelines are often referred to as lifelines sincethey carry materials essential to the support of life and mainte-nance of property. Pipelines can be categorized as either continu-ous or segmented. Steel pipelines with welded joints are consid-ered to be continuous while segmented pipelines include cast ironpipe with caulked or rubber gasketed joints, ductile iron pipe withrubber gasketed joints, concrete pipe, asbestos cement pipe, etc.

The earthquake safety of buried pipelines has attracted a greatdeal of attention in recent years. Important characteristics of bur-ied pipelines are that they generally cover large areas and are sub-

ject to a variety of geotectonic hazards. Another characteristic ofburied pipelines, which distinguishes them from above-groundstructures and facilities, is that the relative movement of the pipeswith respect to the surrounding soil is generally small and the in-

ertia forces due to the weight of the pipeline and its contents arerelatively unimportant. Buried pipelines can be damaged eitherby permanent movements of ground (i.e. PGD) or by transient seis-mic wave propagation.

Permanent ground movements include surface faulting, lat-eral spreading due to liquefaction, and landsliding. Although PGDhazards are usually limited to small regions within the pipelinenetwork, their potential for damage is very high since they impose

large deformation on pipelines. On the other hand, the wave propa-gation hazards typically affect the whole pipeline network, butwith lower damage rates (i.e., lower pipe breaks and leaks per unitlength of pipe). For example, during the 1906 San Francisco earth-

P r e f a c e

By Michael ORourke


14/276

xii

quake, the zones of lateral spreading accounted for only 5% ofthe built-up area affected by strong ground shaking. However,approximately 52% of all pipeline breaks occurred within onecity block of these zones, according to T. ORourke et al., (1985).Presumably the remaining 48% of pipeline damage was attributedto wave propagation. Hence, although the total amount of dam-age due to PGD and wave propagation was roughly equal, thedamage rate in the small isolated areas subject to PGD was about20 times higher than that due to wave propagation.

Continuous pipelines may rupture in tension or buckle in com-pression. Observed seismic failure for segmented pipelines, par-ticularly large diameters and relatively thick walls, is mainly due

to distress at the pipeline joints (axial pull-out in tension, crushingof bell and spigot in compression). For smaller diameter segmentedpipes, circumferential flexural failures (round cracks) have alsobeen observed in areas of ground curvature.

This monograph reviews the behavior of buried pipeline com-ponents subject to permanent ground deformation and wave propa-gation hazards, as well as existing methods to quantify the re-sponse. To the extent possible and where appropriate, the review

focuses on simplified procedures which can be directly used inthe seismic analysis and design of buried pipeline components.System behavior of a buried pipeline network is not discussed inany great detail. Where alternate approaches for analysis or de-sign are available, attempts are made to compare the results fromthe different procedures. Finally, we attempt to benchmark theusefulness and relative accuracy of various approaches throughcomparison with available case histories.

This monograph is divided into twelve chapters. Chapter 1 re-views seismic hazards and the performance of buried pipelines inpast earthquakes. Chapter 2 describes the different forms of per-manent ground deformation (surface faulting, lateral spreading,landsliding), and presents procedures to quantify and model boththe amount of PGD as well as the spatial extent of the PGD zone.Chapter 3 reviews seismic wave propagation and presents proce-dures for estimating ground strain and curvature due to travelling

wave effects. Chapter 4 presents the failure modes and correspond-ing failure criteria for buried pipelines subject to seismic effects.Chapter 5 reviews commonly used techniques to model the soil-pipe interaction in both the longitudinal and transverse directions.


15/276

xiii

Chapters 6 and 7 present the response of continuous pipelinessubject to longitudinal PGD and transverse PGD respectively, whileChapter 8 discusses pipe response due to faulting. Chapter 9 pre-sents the response of segmented pipelines subject to permanentground deformation. Chapters 10 and 11 discuss the behavior ofcontinuous and segmented pipeline components subject to seis-mic wave propagation. Chapter 12 presents current countermea-sures to reduce damage to pipelines during earthquakes.


16/276

xiv


17/276

xv

A c k n o w l e d g m e n t s

This state of the art monograph is one of the products resultingfrom the Multidisciplinary Center for Earthquake Engineering Re-search (MCEER), formerly the National Center for Earthquake En-gineering Research (NCEER), research projects 94-3301A and 95-

3301A at Rensselaer Polytechnic Institute. These projects providedpartial financial support for the second authors doctoral studies.Both authors gratefully acknowledge this support.

Much of the U.S. research reviewed in this monograph was anoutgrowth of NCEER projects in the lifeline area. The NCEER life-line activity was lead by Professor M. Shinozuka, and the authorswould like to thank Professor Shinozuka for his tireless leadershipof that effort.

The monograph attempts to also include key results from over-seas, particularly Japan. Much of the Japaneses research was pre-sented at a series of six U.S. Japan workshops. This workshopseries was originally organized by Professor Shinozuka of the U.S.and the late Professor K. Kubo of Japan. More recently, the work-shop series was organized and lead by Professors M. Hamada (Ja-pan) and T. ORourke (U.S.). Hence, in addition to their signifi-cant individual technical contributions, the authors would like to

acknowledge the admirable international cooperation and profes-sional leadership of Professors Hamada, Kubo, T. ORourke andShinozuka.


18/276

xvi


19/276

xvii

AC Asbestos CementASCE American Society of Civil EngineersATC Applied Technology CouncilAWSS Auxiliary Water Supply System

CI Cast IronConc Concrete Pipe

DI Ductile Iron

EBMUD East Bay Municipal Utility DistrictECP Prestressed Embedded Cylinder Pipe

FE Finite ElementFS Factor of Safety

L-waves Love WavesLCP Prestressed Lined Cylinder PipeLSI Liquefaction Severity Index

MMI Modified Mercalli Intensity

NIBS National Institute of Building Sciences

P-waves Compressional Waves

PE PolyethylenePGD Permanent Ground DeformationPVC Polyvinyl Chloride

R-waves Rayleigh WavesRCC Reinforced Concrete Cylinder Pipe

S-waves Shear Waves

TCLEE Technical Council on Lifeline Earthquake Engineering

WSAWJ Welded Steel Arc-Welded JointsWSCJ Welded Steel Caulked JointsWSGWJ Welded Steel Gas-Welded Joints

A b b r e v i a t i o n s


20/276


21/276

xix

N o t a t i o n s

A cross-section area of pipe

a(t) ground acceleration as afunction of time

ac

critical acceleration

Acore area of the concrete coreA

mmaximum ground accel-eration

amax

maximum acceleration atground surface

ax

horizontal acceleration atground surface

az

vertical acceleration atground surface

C apparent propagationvelocity of seismic wave

CH

shear wave velocity of ahalf space

CL

shear wave velocity ofuniform soil layer

Cph

phase velocity of seismicwave

Cs

shear wave velocity ofsurface soil

d closest distance to surfaceprojection of fault plane

D pipe diameter

DA, D

Bpeak ground displacements

D5015 mean grain size in T15(mm)

da

pull-out capacity of joint(axial deformation)

dl

lateral deformationcapacity of joint

Dm

peak ground displacement

DN

Newmark displacement(cm)

Dr

relative density of soil

e pore ratio of soil

E modulus of elasticity

Ei

initial Youngs modulus

Ep

modulus of pipe after yield

Es soil modulus, Es= 2(1+s)GsF axial force in pipe

f frequency in Hz

Fi

axial force at the ithjoint

Fcr

compressive force at joint

FL

liquefaction intensity factor

F15

average fines contents inT

15(%)

FR

restraint strength against

axial tensiong acceleration due to Earths

gravity

G, Gs

shear modulus of soil

h, HA, H

Bthickness of layer (m)

H depth to center-line ofpipeline

H1

thickness of saturated sandlayer (m)

H2 height of embankment (m)H

cdepth to top of pipe

Hs

thickness of uniform soillayer (m)

Ia

Arias intensity in gs

Ip, I moment of inertia

k reduction factor dependingon outer-surface character-istics and hardness of pipe

ko coefficient of lateral soilpressure at rest

K1

equivalent soil spring fordisturbed soil


22/276

xx

K2

equivalent soil spring forundisturbed soil

Kc

bearing capacity factor forundrained soil

Kg soil stiffness per unit lengthK

Lsoil spring constant formovement in horizontalplane

Kv

soil spring constant fordownward movement

Kv1

soil spring coefficient forsmall relative displace-ment

Kv2

soil spring coefficient formoderate relative dis-placement

Kv3

soil spring coefficient forrelative displacementequal to or larger than y

u

L length of PGD zone

L effective slippage length atbend

L0

pipe segment length or

distanceL

aeffective unanchoredlength

LAB

horizontal projection ofinclined rock surface

Lc

length of curved portion

Lcr

critical length of PGD zoneL

epipe length in whichelastic strain develops

Lem embedment length definedas the length over whichthe constant slippage forcetumust

act to induce a

pipe strain equal toequivalent ground strain

Lp

pipe length in whichplastic strain develops

Ls

separation distancebetween two stations

LSI Liquefaction SeverityIndex, LSI is arbitrarilytruncated at 100

M bending moment at pipebent

Mw

earthquake magnitude

n number of joints withinPGD zone, number ofsand layers, or Ramberg

Osgood parameterNc

bearing capacity factors forhorizontal strip footingsfor clay

Nch

horizontal bearing capacityfactor for clay

Ncv

vertical uplift factor forclay

(Nl)60

corrected SPT N-value

Nq

bearing capacity factorsforhorizontal strip footingsfor sand

Nqh

horizontal bearing capacityfactor for sand

Nqv

vertical uplift factor forsand

Ny

bearing capacity factors fordownward loading forsand

p internal pressure (operatingpressure) in pipe

pu

maximum resistance inhorizontal transversedirection

Pw

excess pore water pressure

qu

maximum resistance invertical transverse direc-tion

Q 316

KAEg

R source distance (km) orpipe radius

r Ramberg Osgood param-eter

r parameter of PGD distribu-tion

Rc

radius of curvature of pipe

rd stress reduction factorvarying from a value of 1at the ground surface to avalue of 0.9 at a depth ofabout 30 ft (10 m)


23/276

xxi

Rd

distance from the epicenterto site (km)

Re

Reynolds number ( VD / )

Rs

closest distance to

seismogenic rupture orhypocentral distance

s distance between twomargins of PGD zonenormalized by width W

S ground slopes (%) or shearin pipe

S1

axial force acted on bentfor Element 1

sm

normalized distance frommargin of PGD zone tothe location correspondingto peak transverse grounddisplacement

Su

undrained shear strength ofsurrounding soil

t pipe wall thickness

T shaking period, predomi-nant period of soil (s) or

axial tension in pipeT

15thickness of saturatedcohesionless soils withcorrected SPT value lessthan 15 (m)

tu

maximum resistance inhorizontal axial direction

uj

joint displacement thresh-old

Ug, ug ground displacement inlongitudinal direction

Up, u

pdisplacement of pipeline inlongitudinal direction

V velocity for pipe moving inliquefied soil

Vm

maximum horizontalground velocity

W width of PGD zone

Wa

length from center of PGDzone to anchor point

Warc

arc length of pipe withinPGD zone (m)

Wcr

critical width of liquifiedzone

Wmedia

weight of medium

Wpipe

self-weight of pipe

Ws

distance between pipesupports

x non-normalized distancefrom the margin of thePGD zone

xu

maximum elastic deforma-tion in horizontal axialdirection

Y free face ratio (%)

y lateral displacement of soil

y1

transverse pipe displace-ment in PGD zone

y2

transverse pipe displace-ment outside PGD zone

yu

maximum elastic deforma-tion in horizontaltransverse direction

zu

maximum elastic deforma-tion in vertical transversedirection

inclined angle of slope,

adhesion coefficient forclay or equivalent groundstrain

o

empirical coefficientvarying with S

u

intersection angle betweenpipe and fault trace

c,

oconversion factors

optimal

optimal orientation of

pipeline

ppipe burial parameter

total unit weight

cr

critical shear strain

o

maximum shear strain atpipe-soil interface

effective unit weight of soil

sactual incidence angle ofS-wave

permanent displacement ofground or pipe

cr

critical displacement ofground movement


24/276

xxii

f

average fault displacement

1

pipe displacement at bent

L total elongation of pipe

u relative displacement at

jointu

c deformation capacity of ajoint in compression

uT

deformation capacity of ajoint in tension

uult

relative displacement forjoint closure

xr

joint opening due to jointrotation

xt

joint opening due to tension

x total maximum opening atone side of a joint due totransverse PGD

relative rotation at pipejoint

engineering strain

average pipe strain

o maximum ground strain atx=L/4

a maximum axial strain dueto the elongation of pipe

b pipe bending strain

ap upper bound for pipe axialstrain

g ground strain

p pipe axial strain

v volumetric strain forsaturated sandy soil layer

y yield strain

K EIg / ( )44

coefficient of viscosity ofliquefied soil

g

slope of lower boundary ofliquefied layer or groundsurface

g

maximum ground curva-ture

wavelength or beam-on-elastic foundation

parameter friction coefficient

s

Poisson ratio of soil

factor which depends onwidth of PGD, 0 5 1.

density of liquefied soil

uniaxial tensile stress

o

total overburden pressureon sand layer underconsideration

initial effective overburdenpressure on sand layerunder consideration

ap

axial stress in pipe result-ing from internal pressure

comp

compressive strength ofconcrete

cr

ultimate compressive stressof segments

hp hoop stress in pipe due tointernal pressure

v

total overburden pressure

v effective overburdenpressure

y

apparent yield stress

parameter of PGD distribu-tion

ave

average shear stress

s shear force at pipe-soilinterface

angle of shear resistance ofsand

principle direction ofground motion


25/276

1

1Seismic Hazards and

Pipeline Performance

in Past Earthquakes

Seismic damage to buried pipelines has been observed anddocumented by many individuals during post-earthquake recon-naissance. These surveys provide useful information about thefailure modes and failure mechanisms of buried pipelines. More-

over, empirical damage relationships, based on statisticalinformation from a number of past events, can be used to estimateexpected damage and system performance in future earthquakes.This chapter reviews seismic hazards due to both permanent grounddeformation and wave propagation effects, and presents informa-tion on the performance of both segmented and continuouspipelines in past earthquakes.

1.1

S e i s m i c H a z a r d s

For buried pipelines, seismic hazards can be classified as be-ing either wave propagation hazards or permanent grounddeformation hazards. There have been some events where pipe

damage has been due only to wave propagation. An example isdamage in Mexico City occasioned by the 1985 Michoacan earth-quake. More typically, pipeline damage is due to a combinationof hazards. As mentioned previously, T. ORourke et al. (1985)noted that roughly half the pipe breaks in the 1906 San Franciscoevent occurred within liquefaction-induced lateral spreading zoneswhile the other half occurred over a somewhat larger area wherewave propagation was apparently the prominent hazard. That is,

permanent ground deformation (PGD) damage typically occurs inisolated areas of ground failure with high damage rates while wavepropagation damage occurs over much larger areas, but with lowerdamage rates.


26/276

2 C H A P T E R 1

Wave propagation hazards are characterized by the transientstrain and curvature in the ground due to travelling wave effects.PGD (such as landslide, liquefaction induced lateral spread andseismic settlement) hazards are characterized by the amount, ge-ometry, and spatial extent of the PGD zone. The fault-crossingPGD hazard is characterized by the permanent horizontal andvertical offset at the fault and the pipe-fault intersectional angle.More detailed information on characterization of these PGD haz-ards is presented in Chapter 2.

1.2

P e r f o r m a n c e i nP a s t E a r t h q u a k e s

The vulnerability of buried pipelines to seismic hazards hasbeen demonstrated by the extensive damage observed during pre-vious earthquakes. Examples of documented pipeline damage

include: the 1906 San Francisco, Manson (1908); 1933 Long Beach,Wood (1933); 1952 Kern County, Steinbrugge and Moran (1954);1964 Alaska, Hansen (1971); 1964 Niigata, Hamada and T.ORourke (1992); 1971 San Fernando, McCaffrey and T. ORourke(1983); 1976 Guatemala, Dieckgrafe (1976); 1976 Tangshan, Sunand Shien (1983); 1979 Imperial Valley, Waller and Ramanathan(1980); 1983 Coalinga, Isenberg and Escalante (1984); 1983Nihonkai-Chubu, Hamada and T. ORourke (1992); 1987 Whittier

Narrows, Wang (1990); 1989 Loma Prieta, T. ORourke and Pease(1992); 1991 Costa Rica, M. ORourke and Ballantyne (1992); 1993Kushiro-Oki, Wakamatsu and Yoshida (1994); 1994 Northridge, T.ORourke and Palmer (1994). Presented below is a brief descrip-tion of the amount and types of pipeline damage which have beenobserved (T. ORourke et al., 1985).

In 1964, the Anchorage, Alaska earthquake caused over 200

breaks in gas pipelines and 100 breaks in water distributionpipelines at Anchorage. Gas lines within fault zones were rup-tured. Most of the pipeline damage was due to landslides andground cracking.


27/276

3SEISMIC HAZARDS AND PIPELINE PERFORMANCE IN PAST EARTHQUAKES

The 1971 San Fernando earthquake resulted in 1,400 breaksin various piping systems. The city of San Fernando temporallylost water, gas and sewage services. Liquefaction-induced lat-eral spreading along the eastern and western shores of theUpper Van Norman Reservoir damaged water, gas and petro-leum transmission lines.

The 1987 Ecuador earthquake destroyed the trans-Ecuadorianpipeline (660 mm in diameter), which represented the largestsingle pipeline loss in history. It cost roughly $850 million inlost sales and reconstruction.

The damage briefly discussed above was due to some combi-

nation of wave propagation and PGD effects. In the followingsubsection, empirical relations for pipe damage due to these haz-ards will be discussed separately.

1.3

Empirical Damage Relations

Often the first step in the seismic upgrade of a pipeline systemis an evaluation of the likely amounts of damage in the existingsystem due to potential earthquakes. For buried pipelines, empiri-cal correlations between observed seismic damage and somemeasure of ground motion are typically used. In 1975, Katayamaet al. developed one of the first relations, primarily for segmentedcast iron pipelines, in which damage rate is plotted as a function

of peak ground acceleration. This relation, shown in Figure 1.1,includes both wave propagation and PGD damage data. As shownin Figure 1.1, the damage ratio increases by a factor of 100 for adoubling of the peak ground acceleration.

1 . 3 . 1 W av e P r o p a g a t i o n D a m a g e

It appears that Eguchi was the first to separate wave propaga-

tion damage and PGD damage. For wave propagation, Eguchi(1983) summarized pipe break rate versus Modified Mercalli In-tensity (MMI) for several earthquakes in the United States, anddeveloped fragility relations for six different pipeline materials sub-


28/276

4 C H A P T E R 1

ject to wave propagation only. Eguchi (1991) modified his rela-tionship and obtained a bilinear curve shown in Figure 1.2, whereAC = Asbestos Cement, CONC = Concrete, CI = Cast Iron, PVC =Polyvinyl Chloride, WSCJ = Welded Steel Caulked Joints, WSGWJ= Welded Steel Gas-Welded Joints, WSAWJ (A, B) = Welded Steel

Arc-Welded Joints (Grade A, B), WSAWJ (X) = Welded Steel Arc-Welded Joints (X Grade), DI = Ductile Iron, PE = Polyethylene.Note that the damage rates increase fairly rapidly for MMI 8,and then more slowly for MMI > 8.

Based on data from three U.S. earthquakes, Barenberg (1988)established an empirical relation between seismic wave propaga-tion damage to cast iron pipe and peak horizontal ground or particlevelocity. Note that one would expect that pipe damage to corre-late fairly well with peak ground velocity since as will be shownlater ground strain, and hence pipe strain, is a function of V

max.

Including additional data from three other earthquakes, M.ORourke and Ayala (1993) prepared a plot of wave propagation

After Katayama et al., 1975

Figure 1.1 Pipe Damage in Repairs per Kilometer vs. Peak Ground Acceleration

5.04.0

2.0

1.00.8

0.6

0.4

0.2

0.1

0.08

0.06

0.0250.05 0.1 0.25 0.5 1.0 2.5

Peak Ground Acceleration (g)

Repairs/km

A

B

1

1

1

1

12

1 1

Poor

con

ditio

ns

Goo

dc

on

ditio

ns

Averagecon

ditio

ns

Managua, 1972N. Los Angeles, 1971Tokachi-Oki, 1968Niigata, 1964Fukai, 1948Tokyo, 1923Asbestos cement pipeincludedSome liquefaction

AB1

20.04


29/276


damage rate versus peak ground velocity, which includes cast-iron pipe, concrete pipe, prestressed concrete pipe and asbestoscement pipe. Both the relations are shown in Figure 1.3, wherethe M. ORourke and Ayala best straight line (point A through K)

gives higher damage rates than Barenbergs. The somewhat higherestimated damage rates using the M. ORourke and Ayala relationwere not due to the inclusion of pipe materials other than castiron. They are thought to be due at least in part to the effects ofcorrosion and variable subsurface conditions.

More recently, various researchers have developed empiricalwave propagation damage relations for different pipe materials(Eidinger et al., 1995) or for different diameter ranges (Honegger,

1995). Although the individual mean regression curves differ some-what from that in Figure 1.3, the data points still fall in the generalband shown in Figure 1.3 and the scatter of data points is similar.

After Eguchi, 1991

Figure 1.2 Wave Propagation Pipe Damage vs. Modified Mercalli Intensity

1

1

0.01

0.001

0.00016 7 8 9 10

Modified Mercalli Intensity

Rep

airsPer1000Feet

Ground Shaking WSGWJ

AC/CONC

PVC

CI/WSCJWSAWJ (A,B)/PE

DI

WSAWJ (X)


30/276

6 C H A P T E R 1

1 . 3 . 2 P G D D a m a g e

There are a variety of patterns of PGD depending on local soilconditions and the presence of faults. One type of PGD is local-

ized abrupt relative displacement such as at the surface expressionof a fault, or at the margins of a landslide. The second type of PGDis spatially distributed permanent displacement which could re-sult, for example, from liquefaction-induced lateral spreads, orground settlement due to soil consolidation. For localized abruptPGD, pipeline damage mainly occurs around the ground rupturetrace. On the other hand, breaks for spatially distributed PGD mayoccur everywhere within the PGD zone. Empirical damage rela-

tions for both types of PGD (spatially distributed and abrupt) arepresented in the following discussion.

After M. ORourke and Ayala, 1993

Figure 1.3 Wave Propagation Damage to Common Water System Pipe vs. Peak HorizontalParticle Velocity

1.0

0.50

0.20

0.10

0.05

0.02

0.01

0.005

0.002

0.00011 2 5 10 20 50 100Peak Horizontal Particle Velocity (cm/sec)

PipeDamageRatio(repairsperkilometer)

Best Straight Line(Pts. A thru K)

Barrenberg Line(Pts. A thru D)

I

E

K

J

G

D

H

C

B

A

F


31/276


Spatially Distributed PGD

Porter et al. (1991) developed an empirical relation for bell

and spigot cast iron water pipe with lead and oakum joints asshown in Figure 1.4.As shown in Figure 1.4, the damage rate is a function of per-

manent ground displacement. A bilinear curve is fitted to the datafrom the 1906 San Francisco and 1989 Loma Prieta earthquakes.The initial portion of the curve (PGD < 5 inches (13cm)) is basedon damage information for the Marina District in San Franciscoduring the 1989 Loma Prieta event (vertical settlements) while thelater portion (PGD > 5 inches) is based upon the 1906 San Fran-cisco event (lateral spreads).

As shown in Figure 1.4, the normalized pipe break rate is anonlinear function of PGD. Relatively small ground displacementproduces initial pipe breakage. At larger ground displacement,break rates increase, but at a smaller rate. To explain thisnonlinearity, Porter et al. postulate that damage initiates at lowmagnitudes of PGD, breaking the original pipe network into shortersegments that are relatively free to move with the surrounding soil.

Relatively larger displacements are then required to cause furtherbreaks in the remaining intact segments.

After Porter et al., 1991

Figure 1.4 Empirical Damage to Cast Iron Water Pipe vs. Spatially Distribution PGD

10

8

6

4

2

00 20 40 60 80 100 120

Permanent Ground Displacement (inch)

PipeBreaksPer1

000feet


32/276


33/276


Porter et al. assume that the rate of damage to gas welded steelpipes was half of that for cast iron pipes. Similarly, the vulnerabil-ity of arc welded steel pipes was taken as 12.5% of cast iron pipes.Porters pseudo-empirical relations for steel pipes and various othermaterials are shown in Figure 1.5.

More recent studies have suggested other empirically basedrelations between damage and the amount of ground movement.For example, Heubach (1995) suggests that expected damage tocast iron pipe with rigid joints is roughly a factor of four largerthan that for modern welded steel pipelines. Along similar lines,Eidinger et al. (1995) propose PGD damage relations which arefunctions of pipe material and joint type. Nevertheless, a com-parison of Figures 1.1, 1.3 and 1.4 suggests that the scatter of data

points about the mean regression curve is reduced when one con-siders wave propagation and PGD damage separately.

After Porter et al., 1991

Figure 1.5 Pipe Breaks vs. Permanent Ground Displacement

7

6

5

4

3

2

1

0

0 10 20 30 40 50 60

Permanent Ground Displacement (inches)

PipelineDamage(breaks/1000ft) Pre-1960 CI

Pre-1940 RS & WS

AWSS

Pre-1989 DI

Post-1940 WS & Post 1989 DI


34/276

10 C H A P T E R 1

Localized Abrupt PGD

Porter et al.s relations were obtained for spatially distributedPGD, specifically ground settlement and lateral spread. As such, itprobably should not be used to predict damage to pipelines sub-

ject to localized abrupt PGD such as at a fault offset. This is becauseone expects higher pipe strain, for a given amount of PGD move-ment where the PGD is abrupt as opposed to distributed.

For fault rupture, Eguchi (1983) presented a relationship be-tween the damage rate and the amount of fault offset, shown inFigure 1.6. It is based on damage from the 1971 San Fernandoearthquake and applies to pipelines within 300 feet (91 m) eachside of the predominate line of rupture. Note that the WSAWJ1

curve presents pre-modern (i.e., prior to late 1950s) welded steel,

modern (e.g. X-grade) pipe would be expected to vary by a factorof 70% lower.

As shown in Figure 1.6, the break rate per 1,000 feet for castiron (CI) pipes is about 1.5 for abrupt PGD equal to 10 inch (25

After Eguchi, 1993

Figure 1.6 Vulnerability Relationships for Buried Pipelines in Fault Rupture Areas

10

1

0.1

0.011 10 100

Maximum Permanent Displacement (in)

BreaksPer1000Feet

Fault Rupture

AC

CI

WSCJ

WSGWJ

WSAWJ1


35/276


cm) and 4.0 for abrupt PGD equal to 100 inch (2.5 m). However,for the same cast iron pipes, the break rate due to spatially distrib-uted PGD given in Figure 1.4 is 3.2 for the displacement equal to10 inches and 7.4 for the displacement equal to 100 inches. Thisresult is counter-intuitive since one expects higher pipe strain forabrupt PGD. The authors understand that Eguchi currently consid-ers the relations in Figure 1.6 to be lower bounds, with expecteddamage being possibly a factor of three times the lower boundvalue.

Even when the correct empirical relation is used, caution shouldbe exercised in its application. As discussed in Chapter 8, the vul-nerability of buried pipes to fault offset is strongly influenced by

the pipe-fault intersectional angle. Since the intersectional angleis not a parameter in Eguchis relation, the modified relation likelyapplies in an average sense to a wide range of intersectional angles.

The above discussion is not intended to suggest that theseempirical relations are useless. They are, arguably, the best cur-rently available. They are appropriate for evaluating overall systemperformance. However, by themselves, they are probably not ap-propriate for vulnerability analysis of an individual component.

1.4

S y s t e m P e r f o r m a n c e

There has been a large amount of research work over the pastdozen of years or so on pipeline system performance. Notable

contributions have been made by Isoyama and Katayama (1982),Liu and Hou (1991), Sato and Shinozuka (1991), Honegger andEguchi (1992), and Markov et al. (1994). A detailed discussion ofoverall system modeling and performance is beyond the scope ofthis state-of-the-art review, which focuses primarily on compo-nent performance, behavior and design. However, a summary ofthe results of system performance evaluations as a function of bur-ied pipeline component performance (specifically breaks per unit

length) will be discussed briefly.Isoyama and Katayama (1982) evaluated water system perfor-

mance following an earthquake for two supply strategies: supplypriority to nodes with larger demands, and supply priority to nodes


36/276

12 C H A P T E R 1

After NIBS, 1996

Figure 1.7 Serviceability Index vs. Average Break Rate for Post-Earthquake System Performance

Evaluation

with lowest demands. These two strategies correspond to the bestand worst system performance, which is shown in Figure 1.7. Re-cently, Markov et al. (1994) evaluated the performance of the SanFrancisco auxiliary water supply system (AWSS), while G&E (1994)did a similar study for the water supply system in the East BayMunicipal Utility District (EBMUD). Their results are also shownin Figure 1.7. Based on these results, NIBS (National Institute ofBuilding Sciences) (1996) proposed a damage algorithm, in whichthe system serviceability index is a lognormal function of the aver-age break rate. Note that in this figure, the serviceability index isconsidered as a measure of reduced flow.


37/276

13

2

Permanent Ground

Deformation Hazards

The principal forms of permanent ground deformation (PGD)are surface faulting, landsliding, seismic settlement and lateralspreading due to soil liquefaction. Whether the buried pipelinefails when subjected to PGD depends, in part, on the amount and

spatial extent of the PGD, which are introduced here.The aim of this chapter is to provide a general overview of

permanent ground deformation. First, we discuss the types of faultsand the expected amount of fault offset, which is empirically cor-related with earthquake magnitude. Second, we describe the typesof landslides, empirical relations for the occurrence of landslides,and analytical relations for the amount of earth flow movement.Third, two approaches to evaluate ground settlement induced by

soil liquefaction are introduced. Finally, we present the character-istics of lateral spreads induced by soil liquefaction.

2.1

F a u l t

An active fault is a discontinuity between two portions of theearth crust along which relative movements can occur. The move-ment is concentrated in relatively narrow fault zones. Principaltypes of fault movement include strike-slip, normal-slip and re-verse slip as shown in Figure 2.1. In a strike-slip fault thepredominant motion is horizontal, which deforms a continuouspipe primarily in tension or compression depending on the pipe-fault intersectional angle. In normal and reverse faults thepredominant ground displacement is vertical. When the overhang-ing side of the fault moves downwards, the fault is normal, whichdeforms a horizontal pipe primarily in tension. When the over-hanging side of the fault moves upwards, the fault is reverse, whichdeforms a horizontal pipe primarily in compression.


38/276

14 C H A P T E R 2

As mentioned previously in Section 1.2, the strain in a con-tinuous pipe subject to fault offset depends on the amount of thefault offset and the pipe-fault intersectional angle. Here we only

discuss the amount of fault offset. The effects of the intersectionalangle will be discussed in Chapter 8.Various empirical relations between fault displacement and

moment magnitude have been proposed. They all have a similarlogarithmic form. Here we only introduce the relationship by Wellsand Coppersmith (1994) because it extends previous studies byincluding data from recent earthquakes and from new investiga-tions of older earthquakes. Based on a worldwide data base of

421 historical earthquakes, Wells and Coppersmith selected 244earthquakes, and developed the following empirical relationships:

logf= -6.32 + 0.90Mfor Strike-Slip Fault (2.1)

logf= -4.45 + 0.63Mfor Normal Fault (2.2)

logf= -0.74 + 0.08Mfor Reverse Fault (2.3)

After Meyersohn, 1991

Figure 2.1 Block Diagrams of Surface Faulting

Footwall Side

Overhanging Side

Block Before Faulting Normal Fault Reverse Fault

Strike-Slip Fault Oblique-Slip Fault


39/276


40/276

16 C H A P T E R 2

2.2

L a n d s l i d e

Landslides are mass movements of the ground which may betriggered by seismic shaking. A large number of systems have beendeveloped to classify landslides. The most widely used classifica-tion system in the United States was devised by Varnes (1978).Varnes identified five principal categories based on soil movements,geometry of the slide, and the types of material involved. Varnesscategories are: falls, topples, slides, spreads, and flows. Herein,

lateral spreading is considered to be a liquefaction-induced phe-nomenon, and is discussed in Section 2.3.

Based on the different effects on pipelines, Meyersohn (1991)established three types of landslides as shown in Figure 2.3.

As shown in Figure 2.3, Type I includes rock fall and rock topple,which can cause damage to above-ground pipelines by direct im-pact of falling rocks. This type of landslide has relatively little effecton buried pipelines and will not be discussed in detail. Type II

includes earth flow and debris flow, in which the transported ma-terial behaves as a viscous fluid. Large movements (several metersor more) are often associated with this type of landslide but theexpected amount of movement is difficult to predict. Type II land-slides will not be discussed herein. Type III includes earth slumpand earth slide, in which the earth moves, more or less as a block.They usually develop along natural slopes, river channels, andembankments. Because pipelines often cross such zones, the fol-

lowing will focus on this type of landslide.Empirical methods have been used to determine upper bounds

for the occurrence of landslides. Figure 2.4 shows one such rela-tion (Applied Technology Council, 1985), in which the maximumdistance of observed landslides to the fault rupture zone is plottedas a function of earthquake magnitude.

Recent work by Jibson and Keefer (1993) resulted in analyticalestimation of the expected amount of landslide movement. They

used the computer program STABL (Siegel, 1978) to search for thecritical failure surface by randomly generating slip surfaces andcalculating the factor of safety (FS). The factor of safety is the ratio


41/276

17PERMANENT GROUND DEFORMATION HAZARDS

of the sum of the resisting forces and sum of the driving forces thattend to cause movement. That is, the critical failure surface is theslip surface with the lowest factor of safety.

Based on Newmarks Block model (Newmark, 1965), the criti-cal acceleration, a

c, can then be defined as:

ac

= g(FS - 1)sin (2.5)

wheregis the acceleration due to gravity and

is the inclinedangle of the slope.The displacement of the block is then calculated by double

integration of the ground accelerations larger than the critical ac-celeration a

c.

After Meyersohn, 1991

Figure 2.3 Selected Ground Failure Associated with Landsliding

Rock Fall RockTopple

Source Area

Main Track

DepositionalArea

Bedrock

Debris Avalanche (Very Rapidto Extremely Rapid)

Earth Flow (Very Slow to Rapid)

Main Scarp Bluff Line

Toe

Earth Slump Earth Block Slide

(a) Type I

(b) Type II

(c) Type III

Grab

en

SlipsurfacePressu

reRidg

e

Weathered BedrockSoil, etc.


42/276

18 C H A P T E R 2

Jibson and Keefer selected 11 strong-motion records to esti-

mate the Newmark displacement. For each of the strong-motionrecords, they calculated the Newmark displacement for severalcritical accelerations between 0.02 and 0.4 g, which is consid-ered to be the practical range of interest for mostearthquake-induced landslides. The resulting data are plotted inFigure 2.5, for which the best regression function is:

logDN= 1.460logI

a- 6.642a

c+ 1.546 (2.6)

After ATC-13, 1985

Figure 2.4 Occurrence of Landslide vs. Magnitude of Earthquake

1000

500

200

100

50

20

10

5

2

1

0.5

0.2

0.14.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5

Disrupted Slides & Falls

Coherent Slides

Lateral Spreads & Flows

Magnitude (M)

Maximum

DistanceofLandslideFrom

Fault-Rup

tureZone(km)


43/276


where DN

is the Newmark displacement in centimeters and Iais

the Arias intensity in gs, defined as:

I2g

a t dt a2=

[ ( ) ] (2.7)

where a(t)is the ground acceleration time history.In this regard, Wilson and Keefer (1983) developed a simplerelationship between Arias intensity, earthquake magnitude, M,and source distance,R, in kilometers:

logIa= M- 2logR- 4.1 (2.8)

Note that Equation 2.8 is developed from California earth-

quakes and may slightly underestimate shaking intensity in thecentral United States.

1000

100

10

1

0.1

0.1 1 10Arias Intensity (m/s)

NewmarkDisplacement(cm

)

0.02 g

0.05 g

0.10 g

0.20 g

0.30 g

0.40 g

After Jibson and Keefer, 1993

Figure 2.5 Newmark Displacement vs. Arias Intensity for Critical Accelerations of 0.02-0.40g


44/276

20 C H A P T E R 2

2.3

L a t e r a l S p r e a d i n g

Lateral spreads develop when a loose saturated sandy soil de-posit is liquefied due to seismic shaking. Liquefaction causes thesoil to lose its shear strength, which in turn results in the flow orlateral movement of liquefied soil. Although the ground move-ment is primarily horizontal, Towhata et al. (1991) observed thatvertical soil movement often accompanies liquefaction-induced

lateral spreading. However, the vertical component is typicallysmall and will be disregarded herein.In terms of pipeline response, two situations are possible. In

the first case such as at the Ogata Primary School site during the1964 Niigata event, the top surface of the liquefied layer is essen-tially at the ground surface. For this first case, a pipeline is subjectto horizontal force due to liquefied soil flow over and around thepipeline, as well as uplift or buoyancy forces. In the second casesuch as at the Mission Creek site during the 1906 San Franciscoevent, the top surface of the liquefied layer is located below thebottom of a typical pipeline. That is, the pipeline is contained in anon-liquefied surface soil layer which rides over the liquefied layer.For this second case, the pipeline is subject to horizontal forcesdue to non-liquefied soil-structure interaction but not subject tobuoyancy effects. Pipeline response to such horizontal loading isdiscussed in Chapter 6 and 7. Pipeline response to buoyancy forcesis discussed in Chapter 7.

The direction of movement for the lateral spread is controlledby geometry. When the lateral spread occurs at or near a free face,the movement is generally towards the free face. When the lateralspread occurs away from a free face, the movement is down theslope of the ground surface or down the slope of the bottom of theliquefied layer. For the PGD towards a free face data, the ob-served distance is from 10 to 300 m (33 to 984 ft) away from thefree face with average value of 100 m (Bartlett and Youd, 1992).

For the PGD away from a free face data, the observed slope isfrom 0.1% to 6% with an average value of 0.55%.

There are four geometric characteristics of a lateral spreadwhich influence pipeline response in a horizontal plane. With ref-


45/276


erence to Figure 2.6, these are the amount of PGD movement,the transverse width of the PGD zone W, the longitudinal lengthof the PGD zone L, and the pattern or distribution of ground move-ment across and along the zone.

In the following subsection, available information which canbe used to quantity each of these characteristics is presented.

2 . 3 . 1 A m o u n t o f P G D

In general, the potential for PGD to induce pipe damage isrelated to the amount of ground movement, the length and widthof the PGD zone as well as the pattern of deformation. Predictingthe amount of ground displacement due to soil liquefaction is a

Length L

WidthW

(a) Plan View

(c) Longitudinal Pattern(b) Transverse Pattern

W

L

A A

B

B

Figure 2.6 Characteristics of a Lateral Spread


46/276

22 C H A P T E R 2

challenging problem. Nevertheless there have been a number ofstudies, both analytical and empirical, which have addressed thisissue. These studies are reviewed below.

Analytical and Numerical Models

Dobry and Baziar (1990), and Mabey (1992) estimate lique-faction-induced displacement using a Newmark sliding blockanalysis. In this analysis a 1-D, rigid soil block model is allowed todisplace along a planar failure surface during time intervals whenthe sum of the inertial (i.e., earthquake) and gravity (i.e., self weight)components along the slide surface exceeds the shear strength of

the soil.Hamada et al. (1987), Towhata et al. (1991), and Yasuda et al.

(1991) used 2-D, static elastic models to estimate the amount oflateral spreading displacement. They model the non-liquefied sur-face layer as a 2-D, elastic beam, floating on the liquefied layerbelow, and subject to lateral loading (the component of the grav-ity load parallel to the inclined ground surface). An analytical,closed form solution is used to calculate the ground displacement

by minimizing the potential energy of the system.Orense and Towhata (1992) used a variational principle to

develop a 3-D analytical relation for the amount of liquefaction-induced ground displacement. The method is based on the principleof minimum potential energy. The lateral displacements are calcu-lated based on the assumption on a half sinusoidal distribution oflateral displacement along a vertical section (i.e., zero at the bot-tom and maximum at the top) and vertical displacements are

calculated based on constant volume assumption. The Rayleigh-Ritz method is employed to obtain the solution.

However, as pointed out by Bartlett and Youd (1992), theseanalytical and numerical models have not been applied to a widerange of earthquake and site conditions. More validation and cali-bration studies are likely needed before these analytical andnumerical techniques can be used directly by practicing engineers.

Empirical Model

Several empirical models have been proposed to predict lat-eral spread displacements. The following brief review describes


47/276


these existing empirical relations, their underlaying assumptionsand expected range of applicability.

Work by Hamada et al. (1986) suggests that the amount ofPGD induced by liquefaction is closely related to the geometricconfiguration of the estimated liquefied layer. They proposed thefollowing regression formula for the magnitude of horizontal PGD,, in meters:

= 0.75 3h g (2.9)

where his the thickness of the liquefied layers, in meters, and g

is

the slope of the lower boundary of the liquefied layer or the groundsurface (%), whichever is larger.Note that the Hamada et al. relation does not distinguish be-

tween the amount of expected PGD at a free face as opposed tothat for gently sloping ground. In addition, the thickness of theliquefied layer is in a sense a pseudo parameter which accountsfor the amount of ground shaking (related to earthquake magni-tude and distance) as well as the soil characteristics at the site.

According to Bartlett and Youd (1992), it produces reasonable es-timate for earthquakes with magnitude around 7.5 and epicenterdistance in the 20 to 30 km range.

Youd and Perkins (1987) introduced the concept of a Lique-faction Severity Index (LSI) which is defined as the amount of PGD,in inches, associated with lateral spreading on gently slopingground and poor soil conditions. LSI is arbitrarily truncated at 100.Youd and Perkins established a correlation between LSI, earth-quake magnitude and distance for the western U.S. as follows:

logLSI = -3.49 - 1.86logRd+ 0.98M

w(2.10)

where Rdis the distance from the epicenter to the site, in kilome-

ters for western U.S. earthquakes, and Mw is the earthquake

magnitude.Using that correlation as a starting point and data available for

the 1811-12 New Madrid earthquakes, Turner and Youd (1987)

proposed the following relation for the New Madrid area:

logLSI = 4.252 - 1.276logRd

(2.11)


48/276


49/276


which most strongly influence the amount of PGD are the averagefines contents, followed by the mean grain size and the groundslope/free face ratio. The accuracy of the Bartlett and Youd (1992)empirical relations are relatively good, in that predicted valuesare generally within a factor of two of the observed values. Assuch, they are arguably the best currently available relations foruse in the western U.S.

2 . 3 . 2 S p a t i a l E x t e n t o f L a t e r a l

S p r e a d Z o n e

As will be seen later, the width and the length of the PGDzone have a strong influence on pipe response to PGD. Unfortu-nately the currently available information on the spatial extent oflateral spread zone is somewhat limited. Although one expectsthat the spatial extent of the lateral spread zone strongly correlateswith the plan dimensions at the area which liquefied, analytical orempirical relations are not currently available. In the following,both the width Wand length Las shown in Figure 2.6 will bediscussed.

Information on observed values for the spatial extent of thelateral spread zone has been developed by Suzuki and Masuda(1991). Using data from the 1964 Niigata and 1983 Nihonkai-Chubu earthquakes, they presented scattergrams in Figure 2.8 ofthe amount of ground movement and spatial extent of the lateralspread zone for PGD away from a free face. Note that most all theobserved widths are distributed in the range of about 80 to 600 m(262 to 1968 ft) and the lateral displacement tends to increase

with increasing width.In terms of the length of the lateral spread zone at a free face,

the study by Bartlett and Youd (1992) provides useful information.

B B

AA

Ground surface

Slip Surface

(a) Ground Slope, S = 100A/B (b) Free Face Ratio,Y = 100A/B

Figure 2.7 Elevation View Showing Ground Slope and Free Face Ratio


50/276

26 C H A P T E R 2

6

4

2

0

0 200 400 600W (m)

(m)

(R=0.62)

N=221

0.00287

Y

Z

XW

After Suzuki and Masuda, 1991

Figure 2.8 Observed Data on the Amount of PGD and the Width of the Lateral Spread Zone AwayFrom a Free Face

12

10

8

6

4

2

00 50 100 150 200 250 300

Length of PGD Zone (m)

G

roundMovement(m)

After Bartlett and Youd, 1992

Figure 2.9 Observed Data on the Amount of PGD and the Length of the Lateral Spread Zone at aFree Face


51/276


Figure 2.9 shows observed data on the amount of PGD and thelength of the lateral spread zone at a free face. As with the ob-served PGD zone width shown in Figure 2.8, the observed lengthsin Figure 2.9 are less than about 400 m (1312 ft), with most of thevalues below 200 m (656 ft). Although there is a large amount ofscatter, the ground displacement appears to be a decreasing func-tion of the length of the lateral spread zone for this free facesituation. On the other hand, as discussed in relation to Figure2.8, the ground displacement appears to be an increasing func-tion of the width of the lateral spread zone for gently sloping groundsituations.

Nevertheless, due to the large amount of scatter in these fig-

ures, it seems that the expected length and width of a lateral spreadzone, particularly for site specific studies, should be based uponthe expected plan area of liquefaction as opposed to the estimatedground movement.

2 . 3 . 3 P G D P a t t e r n

As noted previously, the response of buried pipelines to PGD

is influenced by the pattern of deformation, that is the variation ofpermanent ground displacement across the width (Figure 2.6(b))or along the length (Figure 2.6(c)) of the lateral spread zone. Thestudy by Hamada et al. (1986) of liquefaction in the 1964 Niigataearthquake and 1983 Nihonkai-Chubu earthquake provide awealth of information on observed longitudinal PGD patterns. Fig-ure 2.10 shows longitudinal PGD observed along five of 27 linesin Noshiro City resulting from the 1983 Nihonkai-Chubu earth-

quake. In this figure the height of the vertical line is proportionalto the observed horizontal PGD at the point.

Note that about 20% of the observed patterns (6 out of 27)have the same general shape as Figure 2.10(a). That is, they showrelatively uniform PGD movement over the whole length of thelateral spread zone. The response of continuous buried pipeline toidealizations of these longitudinal patterns of PGD is discussed inChapter 6.

Information on transverse patterns of PGD, as shown in Figure2.6(b) is more limited. Figure 2.11 shows five transverse PGD pat-terns observed in the 1971 San Fernando earthquake and 1964Niigata earthquake.


52/276

28 C H A P T E R 2

0

0

0

0

0

50

50

50

50

50

100 (m)

100 (m)

100 (m)

100 (m)

100 (m)

5

5

5

5

5

4

4

4

4

4

3

3

3

3

3

2

2

2

2

2

1

1

1

1

1

Pe

rmanentGroundDisplace

ment(m)

(a) Section Line N-2

(b) Section Line S-15

(c) Section Line S-16

(d) Section Line S-4

(e) Section Line S-13

After Hamada et al., 1986

Figure 2.10 Observed Ground Deformation


53/276


a., b. T. ORourke and Tawfik, 1983; c. Hamada and T. ORourke, 1992

Figure 2.11 Observed Transverse PGD Patterns


54/276

30 C H A P T E R 2

2.4

S e i s m i c S e t t l e m e n t

Earthquake induced subsidence may be caused by densifica-tion of dry sand, consolidation of clay or consolidation of liquefiedsoil. Among these three types, the liquefaction-induced groundsettlement is somewhat more important in that it can lead to largerground movement and hence higher potential for damage to bur-ied pipeline system. Ground settlement induced by soil liquefactionis discussed below.

An example of observed seismic settlement due to liquefac-tion is shown in Figure 2.12. This figure presents contours of groundsettlement in the Marina District occasioned by the 1989 LomaPrieta earthquake (T. ORourke et al., 1991).

Note that the maximum ground settlement is about 140 mm(5.5 in). In comparison to the expected amount of lateral spreaddeformation discussed previously, the expected amount of groundsettlement, for the same general level of ground shaking, are typi-cally smaller.

After T. ORourke et al., 1991

Figure 2.12 Contour of Ground Settlement in Marina District


55/276


For saturated sands without lateral spread movement, Tokimatsuand Seed (1987) developed an analytical procedure to evaluateground settlement. The fundamental relation is:

= = ( ) , , ...,v i ih i n1 2 (2.14)

where vis the volumetric strain for a saturated sandy soil layer, his the layer thickness and nis the number of sand layers with dif-ferent SPT N-values.

The volumetric strain in each layer depends on the SPT N-value and the cyclic stress ratio as shown in Figure 2.13, where

(Nl)60is the corrected SPT N-value.The cyclic stress ratio can be computed by:

ave

o

max o

od

a

gr

=

0 65. (2.15)

where amax

is the maximum acceleration at the ground surface, o

and oare the total overburden pressure and the initial effective

overburden pressure on the sand layer under consideration and rdis the stress reduction factor varying from a value of 1 at the groundsurface to a value of 0.9 at a depth of about 10 m (30 ft) (Seed etal., 1987).

Note that T. ORourke et al. (1991) used a similar approach toestimate liquefaction induced settlement in the Marina District.As noted by T. ORourke et al., there is good agreement betweenthe estimated and measured settlements for the natural soils and

land-tipped fill, but the estimated settlements of the hydraulic fillare almost twice as much as those observed in the field.

Seed et al.s analytical approach provides a relatively accurateestimate of ground settlement. However, it is somewhat complexand requires detailed information on site condition and soil prop-erties.

Takada and Tanabe (1988) developed two empirical regres-sion equations for liquefaction-induced ground settlement at

embankments and plain (level) sites based on 404 observationsduring five Japanese earthquakes.


56/276

32 C H A P T E R 2

After Tokimatsu and Seed, 1987

Figure 2.13 Relation between Cyclic Stress Ratio, (Nl) 60, and Volumetric Strain for SaturatedSands

For embankment:

= 0.11H1H

2a

max/N+20.0 (2.16)

For plain site:

= 0.30H1a

max/N+2.0 (2.17)

where is the ground settlement in centimeters, H1

is the thick-ness of saturated sand layer (in meters), H

2 is the height of

embankment (in meters), Nis the SPT N-value in the sandy layer,and a

maxis the ground acceleration in gals. Takada and Tanabes

empirical approach is simple but somewhat less accurate than the

Seed et al.s approach.


57/276

33

3

Wave P ro paga t i on

H a z a r d s

The wave propagation hazard for a particular site is character-ized by the peak ground motion parameters (acceleration andvelocity) as well as the appropriate propagation velocity. This chap-ter briefly reviews attenuation relations for peak ground parameters,

as well as simplified procedures for determining the apparent propa-gation velocities for both body and surface waves. The ground strainand curvature due to wave propagation are then presented. Fi-nally, the influence of variable subsurface conditions on groundstrain is discussed.

3.1

W a v e P r o p a g a t i o n

F u n d a m e n t a l s

There are two types of seismic waves: body waves and surfacewaves. The body waves propagate through earth, while the surfacewaves travel along the ground surface. The body waves are gener-

ated by seismic faulting, while for the simplest case surface wavesare generated by the reflection and refraction of body waves at theground surface. Body waves include compressional waves (P-waves) and shear waves (S-waves). In compressional waves, theground moves parallel to the direction of propagation, which gen-erates alternating compressional and tensile strain. For S-waves,the ground moves perpendicular to the direction of propagation.

The situation for surface waves is somewhat more complex.

Rayleigh and Love waves are two main types of surface wavesgenerated by earthquakes. For the Love waves (L-waves), the par-ticle motion is along a horizontal line perpendicular to the directionof propagation, while for R-waves the particle motion traces a ret-rograde ellipse in a vertical plane with the horizontal component


58/276

34 C H A P T E R 3

of motion being parallel to the direction of propagation. For bothL- and R-waves, the amplitude of motions decreases with depthbelow the ground surface.

Figure 3.1 shows the East-West ground velocity histories inthe hill and lake zones of Mexico City during the 1985 Michoacanearthquake. Notice in the hill zone record, the peak ground ve-locity is about 10 cm /s (Figure 3.1(a)) and the ground motion diesout about 60 s after initial triggering. In the lake zone record (Fig-ure 3.1(b)), the ground velocity during the first 30 to 40 s afterinitial triggering is roughly about 10 cm/s. However the peak groundvelocity of 30 or 40 cm/s occurs roughly a minute or two afterinitial triggering. This suggests that Rayleigh waves could well have

been present in the lake zone. Note that if R-waves are present,they occur after the arrival of the direct body waves. That is, P-waves arrive at a site first, followed by S-waves. If surface wavesare present, they typically arrive after the body waves.

After M. ORourke and Ayala, 1990

Figure 3.1 Ground Velocity Time History in Hill and Lake Zones During the 1985 MichoacanEarthquake

40

0

-400 10 20 30 40 50 60 70 80

60

0

-600 20 40 60 80 100 120 140

a) Ciudad Universitaria - Lab. (Hill Zone)Time (sec)

Time (sec)b) Central de Abastos - Oficinas (Lake Zone)

Veloci

ty(sec)

Velocity(cm/sec)


59/276

35WAVE PROPAGATION HAZARDS

3.2

At t e n tu a t i on R e l a t i on s

Many authors have developed empirical attenuation relationsfor peak ground acceleration, velocity and displacement. T.ORourke et al. (1985) presents a review and summary of many ofthe existing relations. Those relations typically are functions ofearthquake magnitude, site-to-source distance and local site con-ditions. More recently, relations accounting for fault mechanism

have been developed. This section introduces some of these at-tenuation relations which were based on relatively large amountsof data.

One of the most recent relations for peak ground accelerationis that proposed by Campbell and Bozorgnia (1994). They used645 near-source accelerograms from 47 worldwide earthquakesfrom 1957 to 1993 to update the strong-motion attenuation rela-tion. They found that reverse and thrust earthquakes generate largeracceleration than strike-slip earthquakes, and that soft rock hashigher acceleration than hard rock. Their relation is given by:

ln . . . . exp( . )

. . ln ( ) .

. . ln ( ) . . ln ( )

A M R M

+ R M F

R S R S

m w s w

s w

s sr s hr

= + +[ ] [ ]

+ [ ] + [ ]

3 512 0 904 1 328 0 149 0 647

1 125 0 112 0 0957

0 440 0 171 0 405 0 222

2 2ln

(3.1)

whereAm

is the horizontal component of peak ground accelera-tion (g), M

Wis the moment magnitude, R

Sis the closest distance to

the seismogenic rupture of the fault (km), F=0 for strike-slip andnormal faulting earthquakes and 1 for reverse, reverse-oblique andthrust faulting earthquakes, S

sr=1 for soft-rock sites, S

hr=1 for hard-

rock sites and Ssr= S

hr= 0 for alluvial sites.

Joyner and Boore (1981) developed a relation for peak groundvelocity. This empirical attenuation relation was obtained by a re-gression analysis based on 38 data points from the 1979 Imperial


60/276

36 C H A P T E R 3

Valley earthquake and 68 data points from other earthquakes. Thatis:

log Vm= -0.67 + 0.489M

w- logR

s- 0.00256R

s+ 0.17S + 0.22P

(3.2)

where R ds = +2

16 , Vmis the peak ground velocity in cm/sec,dis the closest distance to the surface projection of the fault rup-ture (in km), S=0 and 1 for rock and soil sites respectively, and P=0and 1 for the 50th percentile and the 84th percentile, respectively.

More recently, Kamiyama et al. (1992) developed semi-em-pirical relations for peak ground responses, which account for

effects of local site condition. They found that the ground responsedepends on the earthquake magnitude, the hypocentral distance(R

s)and the amplification factor of the site (AMP(V)). For example,

the peak ground velocity, Vm, can be estimated by:

VR

R Rm

Ms

M

Ms s

M=

>

+

+

2 879 10 10

3 036 10 10

0 153 0 014 0 218

0 511 1 64 0 014 0 218

.

. /

. . .

. . . .

AMP(V)

AMP(V)(3.3)

After Kamiyama et al., 1992

Figure 3.2 Comparison of Peak Ground Velocity at Rock Sites in the 1989 Loma Prieta Event withKamiyama et al. Relation

100

10

1

0.11 10 100 1000

Hypocentral Distance (km)

PeakVelocity(cm/sec)

Observation

Mean

One StandardDeviation


61/276

37WAVE PROPAGATION HAZARDS

For rock sites (AMP(V)=1), Figure 3.2 presents the semi-em-pirical relation by Kamiyama et al. and observed data from the1989 Loma Prieta earthquake. Note that the amplification factorsare available from response spectra of local sites, and vary withfrequency content.

The most recent empirical relations for peak ground displace-ment were developed by Gregor (1995). He considers the effectsof different fault mechanisms and wave types. For example, thepeak rock displacement for shear waves (i.e., SH waves) due tostrike-slip fault can be estimated by:

logDm

= -5.0 + 1.02Mw

- 0.87logRs

(3.4)

Figure 3.3 shows data points for the 1989 Loma Prieta earth-quake (M

w=7.0) and the mean attenuation curve (solid line) from

Equation 3.4. The dashed lines are the mean curve one stan-dard deviation.

As shown in Figures 3.2 and 3.3, even with the best attenu-ation relation, observed values are only within

Documents

Response of Buried Pipelines Subject to Earthquake Effects