Sunteți pe pagina 1din 285

Ground Motion Intensity Factors and

Acceleration Response Spectra


Ninth Edition API RP 2 A
Earthquake Design Provisions
Eastern Gulf of Alaska Region
Prepared for

American Petroleum Institute


(APO)

November 1978
~ ~~

~~

S T D - A P I / P E T R O 1 1 ï - E N G L 1978 .
II 0 7 3 2 2 7 0 Ob02L99 517 m

TABLE OF CONTENTS

Page

List of Tables V

List of Figures vi

I. INlRODUCTION I- 1
I. 1 Perspective I- 1
1.2 API RP 2A 1-2
I. 3 Development 1-4
I. 4 Objectives I- 5
L5 Seismic Zoning I- 6
I. 6 Effective Ground Accelerations I- 7
1.7 Normalized Elastic Response Spectra I- 7
1.8 Strength (Elastic) Level Response Spectra I- 8
1.9 Strength (Elastic) Levei Analytical Considerations I- 9
I. 10 Ductility (Inelastic) Levei Provisions I- 9
1.11 Report Outline And Contents 1-1 1

II. RELIABILITY ASSESSMENT II- 1


II. 1 Basic Concepts II- 1
11.2 A ppii cable Probability Theory 11-2
11.2.1 Prob abiii ty Definitions 11-2
11.2.2 Key Concepts 11-3
11.2.3 Random Variables 11-8
II.2.4 Probability Papers 1r112
11.3 Reliabiiity Computations 11-13
II.4 Perspective 11-18

m. VALUE ASSESSMENT III- 1


III. 1 Tangible Costs III- 1
111.2 Risk Reduction 111-3
111.3 Intangible Costs 111-5
III. 4 Utility Evaluations 111-7
111.5 Cost Ratio Concepts 111-8
111.6 Perspective 111-1 1

i
~~

STD.API/PETRO 117-ENGL 1778 0732270 Ob02200 Ob7

IV. STUDY PLATFORM iv-1


IV. 1 Components of The Study Platform iv-1
IV. 2 Superstructure 1v-1
IV. 3 Substructure 1v-2
IV. 4 Design Procedure For Sizing Platform Elements 1v-3
IV. 4.1 Lateral Pile Behavior IV -3
IV.4.2 Axial Pile Behavior IV -3
IV. 4 . 3 Structural Elements and Member Joints IV-4
IV.5 Dynamic Modeling and Characteristics IV -4
V. OCEANOGRAPHIC LOADING V-1
v. 1 . Gulf of Alaska Waves, Winds, And Currents v-1
v.2 Predicted Imposed Loading On Study Platform v-4
v.3 Uncertainties In Predicted Loading v-5
v. Q Evaluation Of Actual Loading v-7
VI, PLATFORM RESISTANCE TO OCEANOGRAPHIC LOADINGS VI- 1
VI. 1 General VI- 1
V1.2 System Reliability v1-2
VI.3 Superstructure v1-3
VI.4 Substructure v1-6
VI.5 Platform System Resistance Evaluations VI- 7
VI.6 Calibrat ion v1-8
YIL EARTHQUAKE CûNûITIWS v11-1
VII. 1 Gulf Of Alaska Seismic Exposure VII- 1
V!I. I. 1 Seismic Modeling VII- 1
VU. 1.2 Seismicity v11-2
.
Vii i . 3 Attenuation v11-4
VU. 1 . 4 Random Quake Seismic Exposure v11-7
VII. 1.5 Non-Random Quake Seismic Exposure v11-8
vII.2 Firm Ground Motion Characteristics VII- 10
VII.2.1 Time Histories VII- 10
vII.3 Soft Ground Motion Characterization VII- 11

ii
~
~

STD.API/PETRO 117-ENGL 1978 0732290 Ob02201 T T 5

MI, PLATFORM RESPONSE TO EARTHQUAKES VIII- 1


VIII. 1 Induced Forces Due To Earthquakes Contrasted With
Imposed Forces Due To Oceanographic Conditions VIII- 1
VIII.2 Firm Ground Motion Earthquake Forces VIII-3
VIU. 3 Soft Ground Motion Earthquake Forces VIII-5
VIL4 Elastic Response Summary VIII- 7
VIII.5 inelastic Response And Damage States (Static Analyses) VIII-8
VI11.6 Uncertainty in Predicted Elastic Response VIII-11

m. RELiABiLm ASSESSMENï 1x-1


IX. 1 Oceanographic Loadings And Platform Resistance 1x-1
IX.2 Oceanographic Reliability Assessment IX -2
IX.3 Earthquake Induced Forces And Platform Resistances IX-2
IX. 4 Earthquake Reliability Assessment IX -4
IX.5 Combined Environment Reliability Assessment IX -4
X. VALUE ASSESSMENT x-1
x. 1 Initial Costs x-1
x.2 Damage S t a t e Costs x- 1
x. 3 Tangible Cost Analysis x-2
x. 4 Intangible Costs Analysis x-5
x.5 Utility Evaluations x-7
X. 6 Historical Target Reliability Analysis X-8

XL DESIGN STRATEGIES XI- 1


XI. 1 Analysis And Design Procedures XI- 1
x1.2 Element Sizing Procedure x1-2
XI.3 Design Environmental Conditions Required To
Achieve Target Reliabili ties x1-2
XI.4 Ground Motion Time Histories For Elastic And
Inelastic Analyses x1-3
XIS Response Spectra For Elastic And Inelastic Analyses x1-4
XI.6 Design Response Spectra and Effective
Ground Accelerations x1-4

i ii
-.
._- . c
XL CLOSURE XII-1
xn.1 Summary XII-1
xII.2 Perspective XII-3
xII.3 Limitations XII- 6
xII.4 Acknowledgement XII-8

REFERENCES .
APPENDEX A: Development And Application Of Response Spectra
Relationships For SPOF And MDOF Systems

-.
i- .c iv
-
=
~~ ~

S T D * A P I / P E T R O 117-ENGL 1778 0 7 3 2 2 9 0 Ob02203 8 7 8

LIST OF TABLES

IV. 1 Characteristics Of Study Platform


IV.2 Fundamental Bending & Extensional Mode Periods Of Study' Platform
IV.3 Damping Ratios
v. 1 Hindcast Historical Extreme Storm Oceanographic Conditions 1947-1969
VII. 1 Seismicity Parameters 1899-1974
VII.2 Earthquake Occurrence Rates
VII.3 Transmission Law Terms
VII.4 Random Earthquake Expected Ground Motion Parameters For Average Site
Return Period Of 100 Years
VII.5 Non-Random Earthquake Expected Firm Ground' Motion P a r m e t e r s For
The 50th Percentile
VIII.6 Platform Response For Soil-Pile System €3
VIII. 1 Platform Response Summary

V
-.
.i.c
-

. ..
~~ ~ ~~~~~

y--

STD.API/PETRO 117-ENGL 1978 0 7 3 2 2 9 0 Ob02204 70Li

LIST OF FIGURES

SECTION I
I. 1 Seismic Risk Map Of Alaskan Coastal Waters
1.2 API (1978) Design Response Spectra

SECTION II
11.1 Elements Of Reliability Assessment
11.2 Statistical Characterization Of Facility System Forces And Resistances
11.3 D a t a Fit On Normal Probability Plot
11.4 General Frequency Distribution Or Probability Density Function
11.5 Reliability Analysis
11.6 Overlap Of S And R Distributions
11.7 Cumulative Distribution Functions For S Or R
11.8 Relationship Between PF, Vs, And - - For VR = 30%
R/S
11.9 Relationship Between FF, Vs, And --
R/S For VR = 50%
11.10 Typical Results Of A Reliability Assessment - PF VS Design Load
SECTION III
111.1 Economics Of A Tangible Cost Value Analysis
111.2 Typical Tangible Cost Analysis
111.3 Effects Of Changing Parameters In Tangible Cost Analysis
111.4 Economics Of Risk Reduction
111.5 Intangible Cost (Human Life) Analysis (After Starr, 1971)
111.6 General Variation Of Cost With Probability Of Failure
IIL7 Cost Ratio Versus Annuai Optimum Probability Of Failure For Conventional
Template-Type Offshore Structures

SECTION Tv
IV.1 Broadside And End-On Elevations Of Study Platform
IV.2 Soil Shear Strength Profile
IV.3 Pile Configurations
IV.4 Pile Load-Moment-Deformation Characteristics -
IV.5 Pile-Soil Systems A And B Load-Deflection Characteristics A t Pile Head
IV.6 Axial Pile Capacities For Driven And Drilled And Grouted Piles
IV.7 Axial Pile Load-Deformation Curves

-.
i- .c vi
- ~ ~

STD.API/PETRO 1 1 7 - E N G L . 1978 0 7 3 2 2 7 0 Ob02205 b 4 0

SECTION V
V.l Statistical Characterization Of Wave Height Data
V.2 Annual Expected Maximum Wave Height For Study Area Based On Extrapolation
Of Measured Data
V.3 Annual Expected Maximum Wave Heights For Study Area Based On Marine
Advisors Hindcas t Study
V.4 Expected Maximum Wave Heights A t Study Sites
V.5 Wave And Current Forces On Study Platform
V.6 Probability Of Platform Storm Loadings
V.7 Characterization Of Additional Uncertainties In Predicted Lateral Wave
Forces

SECIION V i
VI. 1 Foundation And Structurai Systems For Study Platform
VI.2 Risk Factor For Non-Statistical Loads
VI.3 Comparison Of Failure Envelopes For Tubular Beam-Column With Fixed
Ends
vI.4 Post-Failure Modes Of Members
VI.5 Safety Index For Composite Structures, Conditional Upon Known Loads
VI.6 -
Results Of Analyses Platform Subjected To Known Loads
VI.7 Axial Foundation Overturning Resistance
VI.8 Results Of Analysis - Foundation Subjected To Known Loads
VI.9 Combined Structure And Foundation Reliability For Model Subjected To
Known Loads
VI.10 Spread Of Applied Lateral Force (i) For Two Values Of Hindcast Lateral
Force (SH)
VI.11 Reliability Of Hindcast Lateral Force
VI.12 Calculated Reliability Of Model Structure As A Function Of Hindcast Laterai
Force
VI.13 Family Of Possible True Resistance Curves
VI. 14 Calibrated Platform Reliability
VI.15 Increasing The Design Load Decreases The Average Annual Risk R a t e

vi i
SECTION MI
VIL1 GOA Study Area, Sites, And Seismic Model
VIL2 Comparisons of Peak Ground Accelerations And Velocities Predicted By McGuire
Relationships With Recorded Data From Three Earthquakes
VII.3 Expected Firm Ground Motion Parameters Based On Historical Data
VII.4 Expected Firm Ground Motion Parameters A t Study Sites Due To Random
Earthquakes
VII.5 Expected Firm Ground Motion Parameters A t Study Sites Due To Non-Random
Earthquakes
VII.6 Taft (1952) Earthquake Ground Motion Time Histories
VTI.7 Borrego Mountain (19681 Earthquake Ground Motion Time Histories
VII.8 Expected Spectral Response Velocities For Pamplona Ridge and Kayak Island
Sites For Random Events
VIL9 Response Spectra For Random Events
VII. 10 Response Spectra For Nonrandom Events
VIL1 1 Correlation Of Maximum Ground Velocity With Platform Response

SECTION Mn
VIII.l Schematic Load And Deformation Behavior Of Platform Under Wave And
Earthquake Loading Conditions
VIL2 Truncation Of Earthquake Loading
VIII.3 Truncated Expected Maximum Resultant Base Shear Force Characterization
For Study Platform
VIII.4 Base Shear And Overturning Moment Time Histories
VIII.5 Influence Of Pile Lateral Stiffness On Resultant Base Shear
VIII.6 Influence Of Pile Axial Stiffness On Resultant Base Shear
VIII.7 Influence Of Damping Assumed For The Study Platform System
VIII.8 DCHARM Response Spectra For Soil Profile B
VIII.9 Comparison Of Maximum Velocity Input (-100 Ft) And Output (-45 Ft) For
Soil Profile B
VIII.10 Study Platform Maximum Resultant Base Shear Versus Peak Ground Velocities
For Modified Input Motions
V I L I 1 Study Platform Maximum Resultant Overturning Moment Versus Peak Ground
Velocity For Modified Input Motions

viii
S T D - A P I / P E T R O 1 1 7 - E N G L 1978 9 0 7 3 2 2 7 0 Ob02207 413 D

VIII.12 Correlations Of Base Shear And Overturning Moment With Input Peak Ground
Velocity
ViII. i 3 Static-Inelastic-To-Collapse Analysis Of Study Platform

-
VIII. 14 Fundamental Elements For Evaluation Of Primary Factors Contributing
To Uncertainties In Predicted Elastic Response

SECTION I%
IX.l Probabilistic Distribution Of Actual Vs. Predicted Maximum Base Shear
For Storm And Quake Loadings
IX.2 Resistance Characterization For Study Platform - Storm & Earthquake Loading
IX.3 Characterization Of Annual Probability Of Failure In Relation To Elastic
Design Load For Offshore Platforms
IX.4 Characterization Of Lifetime Probability Of Failure Of A Gulf Of Alaska
Platform Due To Random And Non-Random Events

SECTION X
X.l Tangible Costs Evaluation For Study Platform
X.2 Sensitivity Of Expected Cost Of Failure
X.3 Cost Ratio Versus Optimum Probability Of Failure
X.4 Evaluation Of Life Loss Rates

SECTION X I
XI. 1 Elastic Non-Ductile, And Inelastic Ductile Load-Deform tion System I

X1.2 Seismic Risk Map Of United States Coastal Waters


X1.3 Response Spectra, Spectra Normalized To 1.0 Gravity
XI.4 Normalized Mean + 1 Standard Deviation Acceleration Response Spectra
For 5 Percent Of Critical Damping

- ..
GROUND MOTION INTENSITY FACTORS AND ACCUERATION RESPONSE
-
SPECTRA Am RP 2A EARTHQUAKE DESIGN PROVISIONS
EASTERN GULF OF ALASKA REGION
R. G. Bea, M. R. Akky, A. R. Dover, and J. A. Egan

I. INTRODUCTION

I. 1 PERSPECTIVE
This report documents a study of earthquake design criteria for the eastern Gulf of
Alaska. This study backgrounded a portion of t h e development of t h e Ninth Edition of
API RP 2A's Section 2.10 earthquake design guideline effective ground acceleration
and response spectra.

This study was performed during 1976 and 1977, with additional details developed
during early 1978.

This study is based on a statistical and probabilistic platform system performance


analysis (risk or reliability analysis). In this analysis, an assessment was made of
environmental exposures (oceanographic and seismic), t h e loadings imposed or induced
in a given type of platform system, and t h e ability of t h e platform system (designed
for given amounts of loading and according t o API element sizing criteria) t o resist
such loadings. Environmental criteria parameters were determined from an
assessment of reliabilities (for different design strategies) and cost impacts.

A conventional drilling and production, steel, tubular membered, pile supported, 12-
leg, template-type platform for 300 feet of water, sited on s o f t and stiff soil profiles
was used to characterize t h e platform system. Environmental conditions for the
eastern Gulf of Alaska were studied. Elements of t h e platform (legs, braces, joints,
piles) were sized according t o the element sizing criteria contained in the Ninth
Edition of R P 2A.

The results of this study are contained in two primary elements of the Ninth Edition
R P 2A earthquake guidelines:

a Seismic toning of the eastern Gulf of Alaska offshore area (Zone 51, and
t h e associated effective ground acceleration (0.4 g), and
=
~ ~ ~

STD.API/PETRO 117-ENGL 1978 0732290 Ob02209 2 7 b

o Elastic response spectra appropriate for different soil conditions, and


for calculation of elastic design forces (to be used in sizing platform
elements together with API elenient sizing criteria).

I. 2 API RP 2A

As stated in the forward t o the API RP 2A:

This Recommended Practice . . . contains engineering design principles


and good practices t h a t have evolved during t h e development of offshore
oil resources Good practice is based on good engineering; therefore, this
recommended practice consists essentially of good engineering
recommendations. in no case is any specific recommendation included
which could not be accomplished by presently available techniques and
equipment. Consideration is given in all cases to t h e safety of personnel,
compliance with existing regulations, and antipollution of water bodies.

Offshore technology is growing rapidly. In those areas where t h e


committee felt that adequate d a t a were available, specific and detailed
recommendations are given. In other areas general statements are used t o
indicate t h a t consideration should be given t o those particular points.
Designers are encouraged t o utilize all research advances available t o
them. As offshore knowledge continues to grow, this recommended
practice will be revised.

Until t h e 1976 edition, API R P 2A earthquake provisions were limited to the static
design guidelines provided in the Uniform Building Code. However, t h e last three
editions contain earthquake design guidelines which include:

o Definition OF five seismic zones and associated effective ground


accelerations (see Fig. I. i), and normalized acceleration response
spectra for three local site conditions (see Fig. 1.2).

o Analytical methods and considerations for calculating forces induced in


t h e platform system by t h e prescribed ground motions.

1-2
~
~

. .
STD.API/PETRO 1 1 7 - E N G L 1778 D 0 7 3 2 2 7 0 Ob022LO T O B D

o Considerations and analytical procedures t o guide t h e designer in


providing sufficient energy absorption capacity and ductility to prevent
platform collapse during rare, intense earthquakes.

o Member and joint sizing guidelines appropriate for t h e forces and


displacements obtained using t h e prescribed analytical methods and
consider ations.
These four groups of provisions a r e intended to assist t h e designer in determining a
platform system t h a t will provide sufficient strength and ductility t o resist t h e forces
and displacements induced by rare, intense earthquake ground motions (a single
"level" criteria). Providing acceptable safety and preventing undesirable performance
during extreme events is t h e fundamental objective.

Ground motion characteristics of r a r e earthquake events a r e not explicitly addressed


in t h e API RP 2A provisions. Instead, ground motion descriptions and factors of
safety a r e coupled t o elastic strength level provisions, inelastic ductility level
provisions, and member-joint sizing guidelines t o develop a tractable design process
and a platform system which will provide acceptable reliability against significant
damage or collapse during t h e life of t h e platform.

The API R P 2A (1977) earthquake provisions consist of t h r e e basic components of


design criteria:

i) Characterization of earthquake ground motions for use in the design of


fixed platforms (Environmental Criteria);
2) Description of analytical methods t o determine forces induced in t h e
platform by ground motions; and
3) Definition of guidelines to size elements for t h e forces determined by
components 1 and 2 (Member Sizing Criteria).

In t h e absence of a detailed site-specific seismic exposure study, t h e design


earthquake ground motions are characterized by strength and ductility level design
earthquake provisions. The strength level earthquake is described through a
normalized major horizontal component response sp.ectra (soil dependent), an
effective ground acceleration design coefficient (G-scaling factor) as a function of
zone relative seismicity, and provisions for t h e orthogonal weaker horizontal and
vertical components as a fraction of t h e specified major component.

1-3
L

- ..
~
~ ~~ ~

STD.API/PETRO 1 1 7 - E N G L 1978 H 0 7 3 2 2 9 0 O b O 2 2 L 1 4 4 4

The ductility level ground morions are specified as being double those of t h e strength
level provisions. In lieu of this requirement, some static stability guidelines (double
def Iection) a r e recommefided.

1.3 DEVELOPMENT

Fundamentally, these guidelines have been developed through the integration of three
philosophicai approaches: an analogy approach where onshore experience is
extrapolated to offshore applications; a judgement approach where offshore experi-
ence, intuition and analytical results a r e combined; and a platform system
performance (risk, reliability) analysis encompassing both previous approaches.

The analogy approach involves extrapolating onshore experience to offshore applica-


tions. Analogies a r e drawn from experience with onshore structural performance
under earthquake loading and onshore building codes such as SEAOC, ATC-3, and
regulatory guides from t h e nuclear industry. Inherent t o this approach is t h e
I t must be noted,
assumption t h a t onshore and offshore conditions a r e similar.
however, that onshore building codes are written for ordinarily stiff and highly
redundant buildings generally founded on good soils. The structurai characteristics,
operational needs, and failure consequences of offshore platforms are vastly
different. Furthermore, seismic transmission characteristics in offshore geological
environments may be different from those in onshore environments.

The second philosophicai approach, t h e judgement approach, requires the use of


experience, intuition, and technology. I t encompasses technological background, past
experience, and personal bias. Although the bulk of technological background and
past experience in earthquake engineering has been with onshore structures, there has
..
been considerable experience with t h e performance of platforms subjected t o wave
loading. Because of the similar cyclic nature of t h e wave and earthquake loading,
this experience is helpful in understanding t h e potential behavior of a platform
subjected t o earthquake ground motions.

The third approach on which the API R P 2A is based is the analytical one. I t involves
both of t h e other approaches as well as data and numerical results. A logical
framework is developed, from which design decisions a r e made.

1-4
=
~

= 0732230
~

STD.API/PETRO I L S - E N G L I978 Ob02232 880

A platform system performance analytical framework was used in developing portions


of t h e earthquake provisions of tb,e API RP2A. Bea (1975, 1976) has reported t h e use
of risk and reliabiiity analyses in t h e development of design criteria. The value
analysis designates t h e critical components t h a t comprise risk, quantifies t h e tangible
and intangible consequences of risk components, and weighs t h e consequences to
arrive at a level of tolerable risk or acceptable reliability. The reliability analysis
etimates t h e environmental loading, structural strength, and t h e attendant uncertain-
ties and variabilities in these quantities.

1.4 OBJECTIVES
A primary purpose of API's response spectra and associated effective ground
accelerations is t o produce a given level of forces 'for a variety of platform systems.
These forces, when combined with t h e element (joints, braces, legs, piles, etc.) design
algorithms and factors of safety a r e intended t o develop t h e strength and ductility
necessary t o prevent collapse of t h e platform system during r a r e intense earthquakes.
Thus, t h e elastic design response spectra a r e fundamentally derivatives of inelastic
performance requirements, sufficient yield strength and adequate ductility for t h e
entire platform system (including soils, foundation, and superstructure).

API response spectra a r e inexorably linked t o t h e platform modeling and analytical


methods which are involved in determining forces induced (not imposed) by
earthquake ground motions. The method used t o combine modai responses (currently
root-sum-square for ail platform elements) influences t h e design spectra ordinates.
In addition, t h e guidelines which specify t h e other two components of ground motions
influence t h e spectral ordinates (two-thirds for other horizontal and one-half for
vertical, applied simultaneously).

Also, t h e methods and parameters used to model foundation effects (principally


stiffness and damping) influence t h e response spectra. For this reason, for pile-
supported platforms which receive major horizontal motions effects below t h e
mudline, t h e motions are representative of those close to (but not at) t h e soil surface.
For determination of vertical motion effects, t h e motions a r e intended to be
representative of those at t h e greater depths where t h e piles will receive major
vertical motion effects.

1-5
- ..
....
STD.API/PETRO 117-ENGL 1978 0 7 3 2 2 9 0 CJbR22L3 7 1 7

Most importantly, t h e API response spectra are tied to a n implicit level of safety
against collapse or unacceptably large amounts of damage. The design response
spectra a r e but one link in a very long chain of parameters and procedures which
develop this desirable level of safety. Given t h a t one would want to keep all of t h e
other links unchanged, t h e response spectra ordinates could be increased or decreôsed
to fit t h e specific situation a t hand (e+, larger for high exposure operations and
smaller for low exposure operations). As stated in t h e API provisions (Section 1.5):

"Selection of t h e design environmental conditions to which platforms


are designed shall be t h e prerogative of t h e owner. Factors to b e
considered include t h e following: .. .
Personnel ... Prevention of
pollution ...Intended use .. .
Planned life ..
. Cost."
In general, a perceptive understanding of t h e API earthquake provisions can be gained
only when one considers t h e entire system of environmental loading and force effects
and platform performance guidelines. Again, quoting API:

"The selected design environmental conditions of the platform should be


based on a risk (reliability) .analysis which considers all of t h e above
factors (previously quoted). The analysis should include: t h e estimated
cost of t h e platform designed to environmental conditions for several
average expected recurrence intervals; t h e probability of platfcrm
damage or loss when subjected to environmental conditions with various
recurrence intervals; t h e financial loss due t o platform damage or loss
including lost production, cleanup, replacing platform and redrilling of
wells, etc. As a guide, analyses have indicated that one optimum average
recurrence interval is several times t h e planned life of t h e plztform."

1.5 SEISMIC ZONING

The API RP 2A utilizes the seismic zoning concept to describe t h e relative intensity
of ground motions in a given geographical zone. Seismic zoning maps of t h e United
States coastal a r e a s developed by t h e U. S. Geological Survey (Algermissen and
Perkins, 1976) and the Applied Technology Council (1977) were considered by APL

1-6
For offshore Alaska (Figure 1.11, t h e coastal waters a r e divided into eight areas and
t h e relative seismicity in each a r e a is Characterized using a factor of O through 5, t h e
relative seismicity being higher for higher factor numbers.

1.6 EFFECTIVE GROUND ACCELERATIONS


The seismicity in each of t h e zones indicated above is characterized using a design
coefficient (effective ground acceleration). The values of these coefficients are:

z=o 1 2 3 4 5
G = O 0.05 o. 10 0.20 0.25 0.40

where: Z = Zone or relative seismicity factor from seismic risk map (Fig. 1.1).
G = Design coefficients (termed effective ground accelerations to scale
associated response spectra or time histories) expressed as a ratio of
gravitational acceleration.

I. 7 NORMALIZED ELASTIC RESPONSE SPECTRA

Along with the seismic risk map and t h e associated design coefficients, t h e API
provides response spectra t o characterize t h e distribution of energy with frequency.
The response spectra shown in Figure 1.2 are referenced to three types of local soil-
geological conditions defined as follows:

(A) Rock - crystalline, conglomerate, or shale-like materiais generally


having shear wave velocities in excess of 3000 ft/s (914 m/s).

(B) Shallow Strong Alluvium - Competent sands, silts, and stiff


clays having
shear strengths in excess of about 1500 psf (71.8 kPa), limited to depths
of less than about 200 f t (61 m), and overlying rock-like materials.

(C) Deep Strong Alluvium - Competent sands, silts, and stiff clays with
thicknesses in excess of about 200 f t (61 m) and overlying rock-like
materials.

I- 7
STD.API/PETRO 117-ENGL 1978 0 7 3 2 2 9 C Ob02215 59T

These spectra consist of three straight-line segments on a log-log scale as follows:

(1) In t h e period range 0.07 t o 0.125 seconds, t h e recommended spectral


values are represented by a line whose equation is SA/C = 20 T.

(2) In t h e period range of 0.125 t o 0.3 seconds (or up to 0.7 seconds,


depending on t h e soil type), t h e spectral acceleration is kept at a
constant value of SA/G = 2.5.

(3) For periods beyond those listed in (2) and up t o 5 seconds, t h e spectral
values reflect a constant spectral velocity. The recommended spectral
values are a set of straight lines having a slope of one on a log-log scale
and are represented by t h e following equations:

-'A- - 0.8 Soil Type A


G - T

-'A- - 1.2 Soil Type B


G -T

'A --1.8
-
G 'T Soil Type C

1.8 STRENGTH (ELASTIC) LEVEL RESPONSE SPECTRA


The API provides a description of three components of ground motion t o be used in
designing a platform for strength requirements. The major horizonta! component
(applied along a major axis of t h e platform) consists of t h e response spectrum
(appropriate t o t h e specific soil condition) shown in Figure 1.2 and scaled by a design
coefficient obtained form the seismic risk map shown in Figure 1.1. The minor
horizontal component (acting in an orthogonal direction t o t h e major axis of t h e
structure) is taken as two-thirds of that assigned to the major horizontal Component.
A response spectrum of one-half t h e major horizontal component is assigned t o t h e
vertical component.

The API R P 2A guidelines provide for the simultaneous application of these three
components. In t h e current 1978 edition, no guidelines regarding development of
design ground motion time histories are provided.

1-8
=
~

. . .
..... . STD.API/PETRO 1 1 9 - E N G L 1978 0 7 3 2 2 9 0 Ob022Lb 42b

These response spectra are intended to represent ground motions at elevations where
foundation elements receive t h e major effects of ground motion. In general, these
elevations are considered to be t h e center of soil lateral resistance f o r the horizontal
spectra and the firm ground f o r the vertical spectrum,

I, 9 STRENGTH (ELASTIC) LEVEL ANALYTICAL CONSIDERATIONS

To determine design forces meeting t h e strength requirements, t h e system is treated


as a series of single-degree-of-freedom systems having t h e same natural frequencies
as t h e platform system. The computed response is taken as the weighted linear
superposition of t h e responses of these multiple single-degree-of-freedom systems.
The weighting factors are composed of t h e ground motion participation factors and
t h e natural mode shapes. The maximum total response of t h e platform system is
calculated as t h e square root of t h e sum of t h e squares of t h e individual modal
responses. The provisions recognize that accuracy of t h e method is dependent on t h e
proper superposition of t h e modal individual responses due to different modal
maximum response time and direction.

Since t h e response spectrum method is an energy-based concept, t h e provisions


provide for t h e use of sufficient modes to represent at least 90 percent of t h e energy
of all t h e modes (exact representation of the system). Generally, this is attained by
considering a minimum of six modes with the highest energy content.

In modeling t h e platform system to compute the natural periods, modes, and modal
response, t h e provisions suggest using a thresdimensional model or special
considerations given to t h e torsional effect if a two-dimensional analytical model is
used.

The provisions recommend t h e consideration of t h e effects of soil-foundation


interaciton on t h e natural periods and response of t h e structure. However, no
guidelines or specific recommendations are provided for such considerations.

I. 1O DUCTILITY (INELASTIC) LEVEL PROVISIONS


To satisfy t h e ductility requirements, t h e API R P 2A recommends designing t h e
platform system so t h a t it is stable under static deflections twice those calculated
using t h e response spectrum for t h e strength levei. The stability under these

1-9
-
- ..
-

S T D S A P I I P E T R O 1 1 9 - E N G L 1778 W 0 7 3 2 2 9 0 0 3 0 2 2 1 7 3b2 S

deflection requirements could be examined by checking t h e static stability of t h e


platform structure.

For purposes of these requirements, the deflecticns at t h e centroid of the deck mass,
including equipment, storage, supplies and structure, may be used. An acceptable
method is to apply a system of lateral and vertical loads, derived from t h e strength
requirement seismic analysis and proportionately scaled, to induce limited yielding or
buckling of t h e structure and deformation of the supporting soils until t h e required
deflections a r e reached with the structure remaining stable. Account should be taken
of t h e limited strength of framing t h a t either buckled or yielded. The effects of the
vertical loads acting through inelastic deflections (the P-A effects) should also be
considered.

In lieu of an inelastic analysis, structures may be investigated f o r ground motions


twice those specified for t h e strength requirement. In such cases, t h e stress and
stability limitations specified for the strength requirements should be met.

The provisions also propose general guidelines t o provide t h e energy absorption


capacity desired by t h e ductility requirements. These are:

Designing t h e connections of primary members t o develop a strength in


excess of t h a t of t h e member.

Providing redundancy in t h e structure and foundation so t h a t alternate


load redistribution paths may be developed in t h e event of primary
member failures.

Avoiding designs and details which incorporate abrupt changes in


stiffness or strength and which utilize materials or combinations of
materials t h a t may behave in a brittle manner during intense earthquake
shakin gs .
Considering loading reversal imposed by earthquake motions.

1-1 o
STD-API/PETRO 11S-ENGL 1978 = 0732290 Ob02218 2T9

1.1 1 REPORT OUTLINE AND CONTENTS


The remainder of this report will be divided into eleven primary sections. To give t h e
reader an overail perspective, a summary of t h e contents of t h e eleven sections
follows.

Section II - Reliability Assessments

This section outlines t h e primary parts of a reliability assessment of a platform


system. Reliability is defined as t h e probability of not experiencing a '<failure1'of t h e
platform system during its lifetime. Failure refers to those states of damage to t h e
platform system which would constitute a significant loss of serviceability.
Characterization of environmental exposures (probabilistic description of
oceanographic and earthquake parameters), their translation t o loadings and forces
induced in platform systems, and t h e characterization of t h e platform system
resistance (probabilistic description of damage states, for a given design strategy) are
I

discussed.

Background on reliability computations and t h e basic concepts of probabilistic and


statistical methods a r e given. Perspective on t h e interpretation and application of
reliability assessments is developed.

Section III - Value Assessments


A value assessment is a process in which alternatives (design strategies) are weighed
or considered t o develop an acceptable or desirable reliability of a platform system.
The process considers tangible (easily valued in monetary terms) and intangible costs,
and takes into account the utility preferences of the decision maker or makers.

This section develops these concepts in a generalized format. In addition, t h e


assessment of cost ratios (ratio of initial costs to failure costs) to develop optimum
(lowest total cost) design strategies and reiiabilities, and t h e use of "historical"
reliabilities (computed for operations in other offshore areas) t o develop t a r g e t
reliabilities are discussed.

Section IV - Study Platform


This section defines t h e components of t h e platform system which was studied. The
system is defined in terms of its superstructure elements (legs, braces, joints, deck
sections), and substructure elements (soils, piles, conductors) and t h e static and
dynamic modeling characteristics.

- ...
.+
1-1 1
~ ~~ ~~

STD.API/PETRO 117-ENGL 1778 0 7 3 2 2 9 0 ü b 0 2 2 1 7 135

Section V - Oceanographic Loadings


Characterization of the oceanographic (wind, wave, current) exposure was based on a
hindcast study by Marine Advisers and i t s calibration with measured wave data. For
t h e range of extreme oceanographic conditions which would provide a significant test
of platform performance, resultant loadings were develcped utilizing API hydrody-
namic loading criteria (Morison Force Equaiton, Stokes' Fifth Order Waves, CD, = 0.6,
C M = 1.5, etc.) and a full three-dimensional analytical model of t h e platform system.
The platform served as t h e mechanism t o combine t h e multiple effects of wind, wave,
and current into meaningful platform response parameters (e+, maximum resultant
base shears and overturning moments).

Section VI - Platform Resistance t o Oceanographic Loadings

This section focuses on a characterization, in probabilistic terms, of the ability of t h e


platform to resist collapse level oceanographic loadings. The process focuses on t h e
synthesis of the tubular brace, leg, and joint elements t o characterize the strength of
t h e superstructure; and on t h e synthesis of the soil, pile, and conductor elements t o
characterize t h e strength of the substructure. These characterizations are based on
available laboratory and field test data of t h e elements and subsystems concerned.
The ultimate strength is referenced t o t h e design strength as determined by API
element sizing guidelines or criteria. The computed oceanographic strength
characterization was updated using t h e survival experience of platforms subjected to
intense hurricane wind, wave, and current loadings.

Section VI1 - Earthquake Conditions

Based on a study of t h e geology, seismology, and tectonics of t h e eastern Gulf of


Alaska, an analytical model was developed t o characterize t h e seismic exposure of
this area. Historical data, supplemented with judgement and guided by results from
sensitivity studies, were used as input to this model. The analytical model results
were used to describe the occurrence and characteristics of earthquake caused strong
ground motions: peak ground accelerations, velocities, displacements, and eiastic
response spectra ordinates in probabilistic terms.

Section VI11 - Platform Response t o Earthquakes

A three-dimensional elastic, time domain analytical model was built t o describe t h e


dynamic response of t h e study platform system when i t was subjected t o a variety of
earthquake ground motion characterizations. Potential effects of s o f t soils in

1-12
- ..
~ ~ ~ ~~ ~~ ~~

S T D - A P I / P E T R O 119-ENGL 1978 m O732290 Ob02220 7 5 7 m

modifying local ground motions were determined using an inelastic, time domain, one-
dimensional wave propagation analytical model. In addition, an a t t e m p t was made to
recognize the projected effects of surface waves on ground motions at the platform
site. The overall effects of strong ground motions on platform response were
expressed in’terms of maximum resultant base shears and overturning moments.

Inelastic response characteristics of t h e platform systems were determined from t h e


results of static-inelastic analyses and from the results of inelastic analyses of single-
degree-of-freedom systems having fundamental periods, inelastic behavior and
damping similar to those projected f o r the real plat form systems.

Section IX - Reliability Assessments


In this study, reliability is defined as the probability of not experiencing collapse of
t h e platform system during its lifetime. The focus was on t h e most serious and costly
damage state - collapse. Oceanographic loadings and t h e forces induced by
earthquake ground motions are convoluted with t h e various possible collapse level
strengths to determine the relative frequency with which platform failures might be
expected. The collapse level strengths are determined by various possible elastic
design loadings, the ductility requirements (a ductility of 2.0, on a force, or
alternatively, on a displacement basis), and t h e element sizing design criteria.
Recognition is given to the inability of the substructure t o transmit forces to t h e
superstructure which are greater than t h e ultimate strength of t h e substructure (a
limiting or truncated base shear and overturning moment process).

Section X - Value Assessments


in this section, an attempt is made t o develop a design strategy (total elastic level
lateral force coupled with API member sizing design criteria) which arrives at an
equitable balance between potential costs (tangible and intangible) and platform
system reliability. For tangible costs, a minimum cost strategy was used. Human life
loss was taken as the highest impact cost, and an equivalent voluntary risk strategy
was used. Acceptable or historical reliability (computed) levels determined from
previous studies of offshore operations (Gulf of Mexico) were used as reference
values.

1-13
~~ ~
~
~

STD.API/PETRO LLS-ENGL 1778 W 0 7 3 2 2 7 0 O b 0 2 2 2 1 ô 9 3

Section XI - Design Strategies

Given a level of lateral force, which when coupled with API elastic member sizing
design criteria, results in a desirable or acceptable level of reliability of t h e platform
system, a n appropriate normalized elastic response spectrum can be coupled t o zn
effective ground acceleration to result in t h a t level of iatcral force. This section
describes this process. The impor?ant interplays of oceanographic criteria,
operational criteria, and element sizing criteria a r e highlighted. The contrasts of the
analysis process (intended to describe complex reality) with t h a t of t h e design process
(intended to result in adequate reliability and strength a t equitable cost utilizing
highly simplified processes) a r e developed.

Section XII - Closure


This concluding section focuses on the general perspective and interpretation of
environmental criteria, reliability, and value assessments in t h e context of this study.
Primary limitations of t h e work documented in this report are listed and discussed.

1-14
. ..
Relative Seismicity Effective Ground
Acceleration - g
~

Factor, Z
0.05
0.70
0.20
0.25
0.40

FIG, 1, 1 SEISMIC RISK MAP OF ALASKAN COASTAL WATERS

I' WOODWARD-CLYDE CONSULTANTS


-.
c c

* .
~~ ~ ~ -~
l

i
l
I
+ I
2
W
i
I
I 1
W
>. u i
I- U
- 1 l
o a Ii
z
-
O
O -
m
n
I!

W
I-
a > -J
a J U
W a a
-I (5: I-
W I- u
u o w
u W a
II
a m
0 v,
II
\ -.I
U u1
U II U

cnU
o a
m U 4
I- cn a
o
W
a
v) .-IS II

11

v,
a v,
>

__ ._ WOODWARD-CLYDE CONSULTANTS
- *
~~
~ - ~

STD.API/PETRO 117-ENGL 1778 0732290 0b0222q 5 T Z m

II, RELfABILITY ASSESSMENT

II. 1 BASIC CONCEPT5


A Reliability Assessment is a logic framework in which one examines in a systematic
and detailed way t h e uncertain and variable aspects of platform system loads,
capacity, and consequences associated with alternative design strategies. Alternative
terms are Risk Analysis or Platform System Performance Analysis.

A fundamental application of such an assessment is development of balanced design


formulations for offshore facilities. The intent is to define a system of consistent
loading and capacity-determining factors and algorithms. When properiy applied, this
system should develop an equitable balance between costs and reliability of t h e
facilities.

Thus, economics, safety, and design formulation consistency are the major motivating
factors behind a reliability assessment.

Figure 11.1 outlines the major elements of a reliability assessment. The assessment is
initiated with two elements. The first is t h e characterization of t h e loading
environment. This is a description, in statistical terms, of t h e extreme environmental
conditions (e+, storm wind, waves, and currents or earthquake ground motions) which
may confront a structure placed at a given location or in a given locale during its
lifetime. I t is at this point that we define the appropriate expected maximum wave
height or peak ground velocity. W e must also define t h e environmental parameters
associated with the full range of possibilities of envircnmentai severities (e.g., 2, 50,
500, 1000 year events).

The second initial element is t h e specification of a preliminary design strategy and


facility system. This means t h a t a facility configuration is developed and sized for a
given set of environmental and operational conditions, and design load and capacity
algorithms. Thus, t h e process becomes focused immediately on a given class of
structures, proportioned f o r an initial set of criteria.

11-1
~

S T D - A P I I P E T R O 119-ENGL 1 9 7 8 m 0 7 3 2 2 9 0 Ob02225 4 3 9 m

criteria must b e revised, t h e structure resized, and t h e process repeated to see i f a


desirable levei of reliability is developed by t h e revised design strategy.

The statistical probability theory upon which t h e reliability assessment methodology


is based is presented in t h e following section.

11.2 APPLICABLE PROBABILITY THEORY

11.2.1 Probability Definitions

Probability deals with projections of future outcomes. It is an approximation of t h e


relative frequency with which outcomes occur. Probabilities can only be estimated.
Only an infinite amount of d a t a will determine probabilities exactly.

If a f u t u r e outcome is known, then the underlying process is said t o be deterministic.


If t h e future outcome is unknown, then the process is said to be probabilistic or
stochastic.

The ciassicai definition of probability is: If an event can occur in N equally likely and
different ways, and if n of these ways have an attribute A, then t h e probability of t h e
occurrence of A, denoted P(A) is defined as n/N (Hahn and Shapiro, 1967). This is an
objective definition, based on relative frequency.

Probabilities may also be defined on a subjective basis: The probability P(A) is a


measure of t h e degree of belief one holds in a specified proposition A. Because of
limited, and frequently misleading d a t a for most offshore situations, this second
concept becomes an important part of statistical characterizations for offshore
platform systems.

Statistics relate to d a t a from past events, as contrasted with probabilistics t h a t


relate to future events. Statistics become an important part of probabilistic
analyses.

The calculus of probability and t h e mathematics of statistics have been well outlined
in several texts; for example, Hahn and Shapiro (19671,Haugen (19681,Benjamin and
Cornell (19701,a n d King (1971).

A few of t h e essential and key concepts of statistics and probabilistics as applied to


fundamental reliability assessments will be outlined in the following paragraphs.

-.
L _ .I
=
~~ ~~~~
~

S T D - A P I I P E T R O 117-ENGL 1778 07322’70 Ob0222b 375

11.2.2 Key Concepts

Probabilities are positive and lie between z e r o and one. Probability equals one when
an event is sure to happen and is equal t o zero when it is impossible.

- is t h e probability of the event


If P(A) is the probability of a n event occurring and P(A)
not occurring, then:

- -
P(A) = 1 P(A)

For example, t h e probability of success (or reliability) is equal t o the complement of


the probability of failure (or risk).

If events A and B a r e independent (occurrence of one event does not affect t h e


chance of occurrence of the other); then t h e probability of occurrence of both A and
B is:

Two events A and B are mutually exclusive events if t h e occurrence of one negates
t h e possibility of occurrence of t h e other. The probability t h a t either A or B (denoted
by P(A+B) will occur is equal t o t h e sum of their respective probabilities of
occurrence:

If two events A and B are independent and not mutually exdusive, then t h e
probability of occurrence of A or B or A and B is:

For three independent, but not mutually exclusive events, this concept is extended as
f oll0w s:

P(A and/or B and/or C = P(A)+ P(B)+P(C)- P(AB)- P(BCb P(AC )+P(ABC)

If A and B are not independent from one another, then:

...
II-3
~ ~

S T D - A P I / P E T R O 117-ENGL 1978 0 7 3 2 2 9 0 Ob02227 201 W

For two non-independent events A and B, t h e probability t h a t event B occurs knowing


that t h e event A has already occurred i s called a conditional probability and is
designated as P(B A). This is called a conditional probability because the probability
so determined is conditioned on t h e fact that a known event has already occurred and
t h e probability of occurrence of this event is known. Thus,

In summary, given t h a t two events A and B a r e independent [P(AB) = PA)P(B)] (the


probability of occurrence of event B is not influenced by the occurrence or non-
- denote the non-occurrence of event A. Thus:
occurrence of event A), let A

And:

-
The probability of occurrence of both A and B is given by:

P(AB) = P(A) P(B)

The probability t h a t just event A occurs (Le., A but not B),

-
- = P(A) P(B) = P(A) [ l
P(AB) - P(B)J
The probability t h a t neither A nor B occurs is (i.e., not A and not 8)

- P(B)
- = P(A)
P(A$ -
- = I: 1 P(A)I [l-P(B) I

The probability t h a t Itat least" one of the two events A and B occur (Le.,
either A or B or both) (not mutually exclusive),

11-4
o The probability t h a t "at most' one of t h e two events A or B occurs is 1
minus t h e probability of the occurrence of both events together

- + -BI = I
P(A - P(A) P(B) = I- P(AB)

o The probability of occurrence of either A or B, but not both A and By

-
P(A or B) = 1 P(A) - P(Ej) + P(A) P(B)
-
= P(A) + P(B) 2P(A) P(B).

An important class of offshore probability problems may be cast in t h e following


framework. Several different and mutually exclusive causes are involved in an event,
A (e+, t h e failure of a platform). These causes will be noted as C 1,C2, Ci, Cn. ..., ...
The probability of cause Ci occurring knowing t h a t event A has happened is:

where P(Ci A) is the probability t h a t Ci and A occur in conjunction with one another.
Furthermore, P(Ci A) is given by:

P(Ci A) = P(A ]Ci) P(Ci).

...,
For event A t o occur, one of t h e mutually exclusive events Ci, C2, C n must have
occurred and this is the Oniy way in which event A can occur. Thus, for A to occur at
all, one of t h e mutually exclusive events C1 A, C 2 A, ...,
C n A must occur. Using t h e
addition rule for the probability of mutually exclusive events,

P(A) = P(A C i ) + P(A C2) + ...+ P(A Cn)


= P(C1) P(A IC I) + P(C2) P(A I C2) + ...+ P(Cn) P(A I Cn)
n

Substituting gives:

II-5
= 0732270
~ ~~ ~ ~~

S T D - A P I I P E T R O Z Z 7 - E N G L 1778 Ob02227 O B ' i

The facility system resistance i s also characterized in statistical terms to reflect the
uncertainties associated with t h e projected performance and associated states of
damage. Such a characterization is shown in Figure 11.2(b). In terms of resistance
probabilities, P(R). The elastic design resistance, R,,, is shown f o r reference.

The next step consists of characterizing t h e probabilities of experiencing t h e various


possible states of damage, given t h e environmental intensity and facility resistance
statitics.

In the simplest terms of loads (SI and resistance or capacity (RI, reliability is t h e
probability of not exceeding t h e system capacity during a given period of exposure.
.
Its complement is risk. Such a concept is expressed in Figure II.2(c) where both the
loading and resistance distributions a r e shown.

The reliability, or probability of success (PSI, is roughly inversely proportional t o t h e


area of overlap of the two distributions (high loads and low resistances).

In t h e case of earthquakes, t h e loading distribution is controlled by design strategy -


the yield level resistance, R, and t h e ductility of the system. The resistance
distribution is also directly controlled by t h e design engineer. The elastic design
loads, methods, materials, configurations, and factors-of-safety determine this
distribution. Thus, t h e resistance distribution may be moved t o t h e right (higher
resistances) by increasing or changing a variety of factors involved in t h e design
process. For example, increasing t h e elastic level design load would move t h e
resistance distribution t o t h e right, decrease t h e a r e a of overlap, and thereby
increase reliability.

The next step (returning t o Figure 11.1) is a decision point. I t is at this point t h a t we
compare t h e reliability for a given candidate facility and design strategy with t h e
desirable target level of reliability. This is fundamentally an optimization process in
which one is searching for a n equitable balance between costs (impacts) and t h e
reliability which is achieved.

If the candidate facility and design strategy results in a desirable level of reliability,
then t h e criteria initially used t o size and proportion the facility may be accepted as
finai design criteria. If t h e reliability is less than t h a t desired, then t h e design

11-6
= 0732270 =
~~
~~ ~

S T D - A P I / P E T R O L I S - E N G L L77ô Ob02230 8Tb

The process then proceeds to t h e characterization of t h e facility components -a


detailed description of t h e elastic and inelastic strength and deformation properties
of the soils, foundation elements (piles, mats), and structural elements (joints, braces,
legs, shear walls, etc.).

The next step consists of developing force-deformation transfer functions for t h e


various categories of environmental and operational loadings. For example, t h e
imposed forces developed by various possible com binations of winds, waves and
currents must be determined, and t h e deformation of elements characterized. The
forces induced in t h e facility system by earthquake ground motions are similarly
de term ined.

The following step combines t h e information on t h e characterization of the


environment with t h e results of t h e force-deformation transfer functions to develop a
statistical characterization of the imposed (e.g., due t o waves) and induced (e.g., due
t o earthquake) forces in t h e facility system developed by the various possible
environmental and operational condition combinations. I t is important to no te t h a t
for extreme conditions, ail or many parts of t h e system (soils, foundation,
superstructure) may be developing significant inelastic responses. These responses
must b e recognized in development of the force-deformation transfer functions, as
well as in the following step of t h e process.

The steps to this point have served primarily t o define statistical descriptions of the
loadings and forces t o which t h e system may be subjected. Figure II.2(a) illustrates
such a description in terms of a frequency or probability function, P(S), for various
possible magnitudes of loodings or force (S).

The next step focuses on determination of system resistances, (RI. This is essentially
a projection of the damage or loss of utility incurred by t h e facility system when it is
subjected to t h e various possible intensities of environmental conditions.

In t h e case of earthquake ground motions, the resistance is stated in terms of t h e


yield strength of t h e system, for a given level of design ductility in t h e system and
for a given design loading.

II- 7
S T D - A P I I P E T R O 119-ENGL 1 9 7 8 0 7 3 2 2 9 0 O b 0 2 2 3 1 732

This formula has some very important engineering implications in updating prior
estimates of causes. Bayes' Theorem is based on this formula and may be interpreted
as providing a mechanism for combining t h e initial or prior probability concerning t h e
occurrence of some event, P(Ci), with related subsequent data or experience to obtain
I
a revised or posterior probability, P(Ci A). The right-hand side of the formula is seen
t o consist of two terms:

-
P(Ci) t h e prior probability? and

I
P(A Ci) - a weighting factor by which t h e prior probability
n
Z HCi) P(A I c i ) is revised on t h e basis of t h e experience, A.
i

This formulation should be used with great c a r e since any condusions may be very
sensitive to t h e assumed prior distribution, particularly for small sets of updating
experiments or experiences. However, t h e formulation does provide an important
ingredient in decision problems when d a t a is scarce.

11.2.3 Random Variables

Generally, t o characterize facility loading and resistance, t h e engineer is concerned


with numbers o r quantities t h a t represent the result of events or observations. When
some quantity is measured and varies from observation t o observation? t h e quantity is
cailed a random variable.

The probability t h a t a random variable X is less than o r equal t o some specified


number x is cailed t h e distribution function for X and is expressed by FX(x). It is
written as:

-.
.L.. c
11-8
= 0732270 üb02232 b79
~~ ~~ ~~

STD.API/PETRO 119-ENGL 1978

The density function, fX(x), is the probability t h a t X falls in the interval x to x + dx,
and therefore is the derivative of the distribution function:

The curves shown in Figures II.2a and II.% a r e such density functions - expressing t h e
probabilities (P) t h a t a given value of load ( S ) or resistance (R)is a given magnitude.

A distribution (or density) function can be characterized with t h e moments of t h e


random variable. If we visualize the density function distributed about t h e x-axis as a
weight distribution, then these moments correspond to the area, moment of inertia,
etc., of t h e distribution.

The f i r s t moment is called the mean, while t h e second moment about t h e mean is
called t h e variance. The square root of t h e variance is called the standard deviation.
These definitions are expressed in symbols as follows:

-
Mean = X = E[X] =

where: E [ ] refers t o expected value.

/(. - El2
4-03

Variance = cr2X = var I: 1 = f(x) dx

Standard Deviation = K= üx

The coefficient of variation is a non-dimensional characterization of the scatter in X:

Also used to describe t h e centrai tendencies of t h e distribution (in addition t o t h e


mean) are:

Mode - value of t h e variable which occurs most frequently; in Figure


II.2a, it is t h e load, S, which corresponds to t h e peak in t h e distribution
function, X'.

11-9
-.
e- . c

- < .

* .
S T D * A P I / P E T R O 1 1 7 - E N G L 1778 = 0 7 3 2 2 7 0 Ob02233 5 0 5 w

a Median - value of the variable which divides the frequency into two
parts; t h e 50th percentile value, X" = X50.

A coefficient which characterizes the shape of t h e density function is the coefficient


of skewness:

Y is z e r o for symmetric distributions (e.g., normal distribution), positive if t h e density


function has a dominant tail to t h e right, and negative if t h e dominant tail is t o t h e
left.

For purposes of fundamental reliability assessments the mean (XI - and coefficient of
variation (V ) are two of t h e primary quantities used t o characterize a distribution
X
function and are equivalent t o a measure of centrai tendency and scatter,
respectively.

Typical ranges of Vx for strength of materiais are:

o 5 to 12% for steel


o 10 t o 20% for concrete
o 20 t o 40% for soils

There are a variety of distribution functions which are intended t o model t h e


characteristics of random variables. Two important models a r e t h e normal and
lognormal distributions.

A distribution of a -
sum of random variables f o r general conditions approaches a
normal distribution. The mean, variance, and distribution function a r e defined as
follows:
f

Mean = -
X = MODE = MEDIAN.(symmetrical distribution)

a Variance =cr X2

11-1 o
- ..*
c-
~

.:.. . .,.
STD.API/PETRO 117-ENGL 1978 0 7 3 2 2 9 0 üb02234 4q3
.. ...'.I.....
. .)<

For a normal distribution, t h e chance t h a t a particular value of a random variable, x,


lies between:

a k O is 68.26%
e * 2 0 is 95.45%
a k 3 0 is 99.74%

The chance that:

a x <-
X + O.. is 84.13%
o x<-X - O is 15.87%

A distribution of a product of random variables for general conditions approaches a


lognormal distribution. This is a very important model for many natural processes.

The lognormal distribution (or density) function is given by:

2
f,(x) =
1
exp - (in x - a)
xB JE 2ß2
L J

where a and ß are distribution shape parameters:

ß -
0.4 (In Xgo ln Xio)

where: Xgo = 90th percentile vaiue of variable (90-percent of values of x a r e


equal t o or less than this value)

10th percentile value of variable (10-percent of values of x are


Xlo =
equal t o or less than this value)

In terms of t h e two shape parameters the mean, variance, median and mode are:

a x = e x p a + % ß2
-

-.
c .% 11-1 1
S T D * A P I / P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 7 0 Ob02235 388

O X" = X5* = exp (c)

11.2.4 Probability Papers

Application of t h e foregoing information on modeling of random variables is greatly


facilitated through t h e use of probability papers (King, 1971). Through distortion of
t h e scales, i t is possible to make cumulative frequency data appear as a straight line.
Scaled graph papers for such use in distribution analyses are called probability papers.

For every different type of probability distribution, t h e r e is a different probability


scale. The intent is t o be able t o t a k e a set of data, rank it, calculate t h e plotting
position f o r each element of data, and plot i t on t h e type of probability paper which
will likely best model t h e data. How well t h e data set is matched by a straight line
on a given type of probability paper is a measure of how well t h a t particular
distribution models t h e data.

For example, assume t h a t a series of soil borings through a given layer of soil a r e
taken. The samples are returned t o t h e laboratory and a large number of strength
tests on this layer a r e conducted with the following results:

Shear Num ber of Cumulative


Strength Tests With This (less than)
(psf 1 Strength Frequency (%)
370 380- 8 0.8
380 - 390 17 2.5
390 - 400 50 7.5
400 - 410 113 18.8
410 - 420 269 45.7
420 - 430 232 68.9
430 - 440 138 82.7
. 440 - 450 110 93.7
450 - 460 50 98.7
460 - 470 13 100.0

11-1 2
.. I

. .

From this data set t h e mean, standard deviation, and coefficient of variation would
be calculated as:

a Mean = 423 psf


e Standard Deviation = 16.9 psf
a Coefficient of Variation = 4% (unusually low)

W e would now take t h e cumulative frequencies and shear strengths and plot them on
normal probability paper and see if t h e d a t a is well-fitted by a straight line. Such a
plot and f i t a r e shown in Figure 11.3. A normal distribution is a satisfactory f i t for
this example set of data.

11.3 RELIABILITY COMPUTATIONS

The previous section has introduced t h e terms and concepts involved in reliability
assessments. To perform a reliability assessment it is necessary t o determine t h e
probability of failure, pf, of t h e platform system. The basis for this evaluation exists
within the discipline of statistical probability theory przviously discussed. For t h e
sake of continuity, a review of t h e previous discussion will be induded in t h e
determination of probability of failure.

In a strictly statistical sense, t h e r e exists variation between measured quantities of


t h e same property. This variation is due t o different things such as: variability in t h e
physical measurement of t h e property, uncertainty in equipment used, limitations of
any theory used, and uncertainty in human perception of any measured quantity.
Thus, t h e perceived value of something may vary considerably, and therefore any
table of d a t a will show scatter. This scatter is clearly seen if a graph of frequency of
occurrence versus any property of concern, PC, is constructed (Figure 11.4). The
graph is termed a frequency distribution graph, and can be defined in terms of a mean
value -x and a coefficient of variation (cov). The cov is defined as t h e ratio of the
standard deviation to t h e mean value. A smaller value of cov is indicative of more
uniformity in PC and is shown by a more narrow frequency distribution.

Citing Benjamin and Cornell (1970):

II-13
"Since occurrences in different intervals are mutually exclusive events, i t
follows t h a t t h e probability that a random variable takes on a value in an
interval of finite length is t h e cisumlTof probabilities or t h e integral of
fx(x) dx over t h e interval. Thus the area under t h e PDF (probability
density function) in an interval represents the probability t h a t the random
variable will take on a value in that interval
"
A
2
p[xl < X < x2] = [fx(x) dx
%
where: p = "probability of1'

"The probability t h a t a continuous random variable X takes on a specific


value x is zero, since t h e length of the interval has vanished. The value of
-
fX(x) is not itself a probability; i t is only t h e measure of the density or
intensity of probability at the point. it follows t h a t fX(x) need not be
restricted to values less than 1, but two conditions must hold:

"The cumulative distribution function, or CDF, is an alternate form by


which t o describe t h e probability distribution of a random variable. Its
definition is, for a continuous random variable

T h e right-hand side of this equation may be written p < X < x 1and


[-a

thus, for continuous random variables,


A
X

"where u has been used as t h e dummy variable of integration to avoid


confusion with t h e limit of integration x (the argument of the function
FX(x) 1-

11-14
- .
.L_..i

- .
~
S T D - A P I I P E T R O 1 1 9 - E N G L 1978 0732290 Ob02238 O97 -

Yn addition, t h e PDF can be determined if t h e CDF is known, since fX(x)


is simply t h e slope or derivative of FX(xh

L J
Further details of t h e elements of probability theory can be found in Benjamin and
Cornell (1970), and Haugen (1968).

These probability concepts can now be utilized t o explain t h e reliability assessment


for an offshore facility. A probability density function can be established for both
t h e environmental loading conditions ( S ) and soil resistance or strength (R)as shown
in Figure 11.2, and mean values for each (-S , R
)- determined and displayed. I t is here
t h a t t h e probabilistic approach differs from t h e deterministic and semi-probabilistic
approaches.

e Deterministic: This approach is t h e classical 'lfactor of safety" method


whereby a load, S, and resistance, R, are compared in t h e ratio FS =
R
The values usually used a r e t h e mean values of t h e associated PDF, Le.,

in reality, FS =
R -
-(see Figure Iï.5).
-5
e Semi-Probabilistic: If t h e resistance is a t'designf1value (RD)represent-
ing t h e mean value of resistance, and t h e loading distribution, s, is used,
t h e resulting analysis can be termed semi-probabilistic. The resistance
distribution can be visualized as being on rollers, and being pushed to
t h e l e f t or right depending upon t h e design value chosen.

o Probabilistic: A truly probabilistic approach incorporates t h e use of


density function for both loading and resistance, and a range of values
of both is therefore considered. Reliability is t h e probability t h a t t h e
resistance R is g r e a t e r than t h e load, S, and this can be expressed as:

reiiability = p (R > S)
where: p = ltprobability that."

II-15
~ ~ ~
~ ~ ~ ~~

S T D . A P I / P E T R O 117-ENGL 1778 0732290 0 6 0 2 2 3 9 T 2 3

The a r e a of overlap between S and R is proportional t o t h e probability


of failure, pf. Near the overlap area, t h e load S is relatively large while
resistance, R, is relatively small. The problem then is t o determine t h e
desired amount of overlap between t h e two distributions. The
reliability is determined from t h e basic concept t h a t a no-failure
probability exists when resistance, R, is not exceeded by load, S.
Examining Figure 11.6 (from Haugen, 19681, t h e probability of a load
value, Slyis equal t o t h e a r e a of element dS or to Al:

prob. [(SI -dS


2 4 S 4 S1 +-)2
dS 1 = f(S1) dS = A 1

with: resistance = f(R) = f(X1, X2, ....st


‘n)
load = f(S) = g(Y I, Y2, .....,Y,)
Since these a r e density functions, t h e probability t h a t R > SI equals to
t h e shaded area, A2 under resistance curve:

3
1
The reliability, RB, is t h e product of these two probabilities:

And thus,

+a,

dRB = f (S dS l’,f (RI dR

Since t h e reliability is t h e probability of resistance being greater than


t h e possible values (over the range) of load, t h e basic equation is:

R B =/.,RB =l? (?It) dR] dS

+a,

(Note also that: l m f ( R ) dR = 1 and

- .
Exact analytical solutions t o t h e above equation are known if t h e
distribution is normal.

Next, t h e distributions of loading, S, or resistance, R, can be expressed


as cumulative distribution functions, as explained previously, and this is
schematically shown in Figure 11.7.

Finally, if t h e environmental loadings and facility resistances are each


idealized as lognormally distributed variables, t h e probability of failure,
pf, can be computed as follows (Ang and Corneil, 1973):

where: [X] -- tabulated normal cumulative probability


of t h e standard variate, X, in t h e interval
-0)t o x.
mean value of resistance
mean value of loading
coefficient of variation of t h e resistance
coefficient of variation of t h e loading

and,

= 50th percentile load


'50
a = 1st lognormal parameter = In S50
8 = 2nd lognormal parameter = 0.4 (in Sgo - ln
sie) g2
-S = exp (a+ T )

2 2
d2
S
= variance = Cexp (2a + B )I [(exp (B - I3
d = standard deviation =
S
CT
vS = coefficient of variation- S
-S
II-1 7
- ..
- .
Figure 11.8 shows the variation of t h e probability of failure for different values of t h e
- - for different values of t h e coefficient of
ratio of mean resistance t o mean load (R/S),
variation of t h e resistance
R = 30%), and for lognormally-distributed load and
(CO~

resistance functions. A similar illustration is given in Figure 11.9 for a coefficient of


variation of t h e resistance equal t o 50 percent.

The reliability of t h e system is then defined as t h e probability of survival or:

-
reliability = 1 probability of failure

If t h e coefficients of variation f o r loading and resistance are known, or assumed, and


if t h e ratio of R / 3 is known, t h e probability of failure is easily Calculated.
Conversely, if t h e probability of failure is specified and t h e coefficients of variation
are known, then t h e ratio, R / 3, c a n be determined, and t h e required resistance
established. Further, if we define a design load as %, then -RD =R
where: Ü = mean value of ultimate resistance
K = factor of safety f o r an elastic non-ductile system
K = ductility factor f o r an inelastic ductile system

I t is then possible, for various values of %, to construct a relationship between -%


and Pf as shown in Figure 11.10. Thus, management can specify a desirable level of
reliability, and the designer can then directly choose t h e design resistance by which
t h e facility is designed. I t is emphasized t h a t management's choice of a value of t h e
probability of failure reflects a g r e a t amount of judgement and experience, rather
than a belief in the absolute accuracy of t h e value.

11.4 PERSPECTIVE

A t this point, it is important to emphasize t h a t at t h e present state of development,


and f o r t h e general problem of offshore f a u l i t y systems, a reliability assessment
likely will not result in determining t h e actual reliability of a facility system. The
calculated reliabilities a r e likely only reflectors or gauges of t h e true reliabilities.
This is so because of imperfections in knowledge and data, and t h e need for
judgement t o fill in t h e gaps. The reliability gauges or reflections a r e significant

11-18
-

=
~

S T D - A P I / P E T R O 1 1 9 - E N G L 1778 O732290 0b022'42 518

when used f o r comparisons - f o r example, what is t h e difference in reliability of a


platform designed with strategy A compared with one designed with strategy B and
what are the comparable projected consequences. Meaningful decisions, however, can
be based on such comparisons. The absolute value of t h e reliabilities should be used
only with great caution, if at d.

In s p i t e of its weaknesses, if properly implemented, interpreted, and applied, a


reliability assessment offers t h e possibility of asking t h e right questions and
developing good answers and decisions on design formulations and strategies for
offshore facility systems.

11-19
S T D - A P I / P E T R O 1 1 9 - E N G L 1978 0 7 3 2 2 9 0 Ob02243 '454

/
I Charoc ter i z o 1ion
i O f The Environment
I 1
Waves, Earthquakes,Gravity

-
I i
I
Characteriro t ion Of
Prei iminaiy Plot f o m Components
1i Design
Soiis,Foundotion Elements
i' Strategy
S t r uc t u ra 1 E I e men t s

I
Characterizat ion Of
Forces Tronsfer Function
Waves, Earthquakes
Operational Load
'I

I
>
O
Q)
I Chora c t e riza tio n of
Forces in Plotform System
Waves, Earthquakes
L Gravity
L
L

v)

al
Y)
I

>
i
al
LT
Charocteriza tion Of
i Platform System
I Resistance
'I I
Q
Ch ara c te riz a t ion O f
Tot o i Re1 ia b iI i t y

Srna Ile r
Final Design

- ..
STD.API/PETRO 117-ENGL 1778 9 0732290 üb02244 390 9

:.:

(a)

FOR C E

œ
a J

U
R E S I S T A NC E

FIG, II,2 STATISTICAL CHARACTERIZATION OF FACILITY


SYSTEM FORCES AND RESISTANCES
c-.%
WOODWARD-CLYDE CONSULTANTS ~ ~ ~ ~ ~
~~~

- *
- ._
. .. S T D . A P I / P E T R O 1 1 7 - E N G L 1778 O73227[1 Ob022Li5 2 2 7 M

01

OZ

05 -
I - 99

2 - 98

- 95

- 90
2
w
- BO æ0w
P
- m k
- 60
P

- 50
c
a
a
- 40
w
2
-30w
1
- c520
a
f
- IO 3
o

:I
-s

-
99 -
O. I

998
I I 4 1 1 I l I
4 0.2

99.9 L 45.5
10.1
465
573 3a5 395 405 415 425 435 443

SHEAR STRENGTH ípif)

FIG, II,3 DATA FIT ON NORMAL PROBABILITY PLOT

WOODWARD-CLYDE CONSU LTANTS


- ..
PROPERTY OF CONCERN - X

..

FIG. I I ,4 GENERAL FREQUENCY DISTRIBUTIONOR


PROBABILITY DENSITY FUNCTION
OVERLAP
9 Oc
AREA
J"

m
O
a t
n
I

1OAD ( S ) OR ULTIMATE STRENGTH (R)

FIG, I I .5 R E L I A B I L I T YANALYSIS

I
I
---.- WOODWARD-CLYDE CONSULTANTS
STD-API/PETRO 117-ENGL 1 7 7 8 D 0 7 3 2 2 7 0 Ob02248 T 3 b D

i'

dS, 4F d S
I

'f(R.1 S,R -

FIG, IL6 OVERLAP OF s AND R DISTRIBUTIONS

A- k WOODWARD-CLY DE CONSULTANTS ~~ ~ ~

- 1

I
~~ ~ ~~
~

S T D - A P I I P E T R O 1 1 7 - E N G L 1778 0732270 Oh02249 772 W

VI 99

log S,R -
I

S,R -
FIG, IL7 CUMULATIVE DISTRIBUTION FUNCTIONS FOR S OR R
WOODWARD-CLYDE CONSULTANTS i!
-.
c- *
- .
. . -.
. . S T D - A P I / P E T R O 117-ENGL I778 m 0732270 Ob02250 b74 m

aox

IO%

1.5 20 2.5 u) 3.5 4.0 4.5

. -R

FIG, IL8 RELATIONSHIP BETWEEN PFJ v S J AND B//


FOR VR = 30%
STD.API/PETRO LIS-ENGL 1778 m 0732270 Ob0225L 5 2 0 R

SO% I 1 I I I I I I I I l l 1

IO%

&-

1%

0.iY.

FIGf IIfg RELATIONSHIP BETWEEN PFJ v s J AND R/S


FOR VR = 5%

WOODWARD-CLYDE CONSULTANTS
S T D - A P I / P E T R O 1 1 7 - E N G L 1778 m 0732290 Ob02252 Lib7 m

-
PROBABILITY OF F A I L U R E Pi, percent p e r 'year
(tog s t o l e )

FIG, 11, 10 TYPICAL RESULTS OF A RELIABILITY


LITY ASSESSMENT
ASSESSMENT -
PI= VS DESIGN LOAD

WOOD WARD-CLY DE CONSULTANTS


-.c

~
~
S T D - A P I / P E T R O 119-ENGL 1978 0 7 3 2 2 9 0 Ob02253 3T3

IlI. VALUE ASSESSMENT

A value assessment is a decision process t h a t is intended to indicate a tolerable levei


of reliability which achieves a n equitable balance in benefits and costs. Two types of
costs can b e associated with a value assessment: tangible and intangible.

III. 1 TANGIBLE COSTS

Tangible costs a r e those which c a n b e adequately evaluated in "monetary" or


"equivalent" terms. T h e monetary "equivalent" can be conceived as t h e "value" of
costs to t h e decision-maker or group of concern. An expected value tangible cost
assessment is a process in which initial costs and potential f u t u r e failure costs are
weighted and combined. T h e objective is to find t h e reliability which produces the
minimum total of initial cost and potential f u t u r e failure costs (i.e., maximizes
benefits).

Figure 111.1 shows typical results of a tangible cost assessment. T h e abscissa is t h e


probability of failure (which is expressed as percent per year of life) and t h e ordinate
is t h e total expected cost. It can be reasonably assumed :hat t h e first or initial cost
of a n offshore facility, E(I), is proportional to its size or calculated capacity (curve
I)), while t h e risk cost, E(F), decreases with increasing capacity, or initial facility
size (curve (2)). Likewise, t h e various costs c a n b e linked with t h e probability of
failure of t h e facility as shown in Figure 111.2. Costs are represented in t e r m s of cost
units to show t h e general applicability of t h e value assessment.

T h e family of curves represents t h e sum of initial cost and t h e present-valued, risk-


weighted failure costs. This total cost has been computed as follows:

where: E(C) = expected total costs


E(I) = expected initial cost. The variation of cost with probability of
failure (linked to the design load or capacity) is shown as a dashed
line in Figure 111.2.
E(F) = expected present value of potentia! failure costs.

The risk cost or expected present value of potential failure costs c a n b e computed as
follows:

111-1
~ ~~ ~ ~~

S T D - A P I / P E T R O 117-ENGL 1978 0 7 3 2 2 9 0 übD225Y 2 3 T

where: E(PL)= expected cost of potential losses in event of a failure. In this


example a range of 10 t o 30 cost units has been used.
PVF = a design life, interest and inflation r s t e dependent function
intended to present-value f u t u r e costs.
t h e probability OÍ failure during t h e life of t h e facility.
Pf =

The expected cost of potential losses in t h e event of a facility failure is computed as


follows:

n
E(PL) =Z PL^ x pli
i= 1

where: PLi = t h e potential loss due to cause i, given a facility failure.


Pli = T h e probability of experiencing t h e potential loss due to cause i,
given a failure.

The present value function can b e computed as follows:

-(r-i)L
PVF = 1 - e
'T
where: r = t h e interest r a t e s e d for discounting profit and loss from f u t u r e
t o present value
i = t h e inflation r a t e on ;he cost of failure consequences
L = t h e design life of t h e platform.

Each of t h e solid curves in Figure 111.2 represents t h e results for a different expected
cost of potential loss (10, 20, and 30 cost units). The average potential loss costs
usually include t h e tangible costs of facility salvage, environmental damage, pollution
control, and clean-up, etc. Several important points should be noted in examining this
figure. The first is t h a t basically this analysis can b e developed independent of
specific details on the facility or its loadings. At this point, t h e probability of failure
is only a numerical quantity, arbitrarily chosen a t various levels to compute t h e
resulting t o t a l expected costs.

111-2
The sensitivity of t h e minimum cost probability of failure (called herein acceptable
reliability) to t h e expected cost of failure is clearly evident in Figure 111.2. For a n
expec ted cost of failure of 10 cost units, t h e tolerable risk is about 1.4 percent per
year, while for failure costs of 20 and 30 cost units, it is about 0.75 and 0.45 percent
per year, respectively.

Similarly, Figure 111.3 illustrates t h e sensitivity of t h e results to changes in the


expected initial costs (indicated by dashed lines in both Figures 111.2 and III. 3). In this
illustration t h e initial cost increment has been increased, in t h e range of interest,
from about 3.5 cost units per percent change in probability of failure t o 5.5 cost units
per percent. The tolerable risks a r e indicated to be 2.0, 1.1, and 0.85 percent for t h e
failure costs of 10, 20, and 30 cost units, respectively. Thus, increasing t h e initial
cost increment associated with increasing t h e reliability, results in increasing t h e
tolerable risk.

111.2 RISK REDUCTION

If the consequences of failure mentioned previously a r e too great, or the level of risk
too high, t h e concept of risk reduction may be employed. The risk associated with
failure of a n offshore facility is assumed by a combination of people depending upon
t h e type of contract under which t h e structure is built and operated, and must be at
an acceptable level for.all concerned. The owners, design engineers, contractors, or
regulatory bodies may ail carry a part of t h e risk during t h e design, construction, or
service phases of t h e structure and foundation. Failure of any type can result in
complex legal problems involving assumed risk, negligence, and insurance claims. The
effort to minimize these problems as well as prevent direct financial loss from failure
has resulted in t h e theoretical analysis of uncertainty or reliability analysis previously
discussed.

Risk reduction involves reducing, controlling, or avoiding various risks through siting
studies, design procedures, construction procedures, soil parameter investigations,
and/or general surveillance and quality control techniques. Thus, with proper thought
and action, t h e hazard potential of t h e situation can be reduced. A step-by-step
procedure for accomplishing this goal can be established as follows (Vanmarcke,
1977):

1. Consider all events of concern and consequences of events (i.e., t h e


hazard potential, Cl.

III-3
-.
A- .c
2. For each major alternative, t h e risk or probability of occurrence, p,
must be assessed.

3. Conceive measures or strategies f o r added protection. These would


include site location, exploration and testing, alternative structural
systems, foundation types and sizes, construction operations, and type
and frequency of inspection a n d repair.

4. Evaluate t h e effectiveness, r, and added cost, h c of each measure or


strategy.

5. Choose and implement t h e most acceptable strategy.

If hazard potential, C, is defined in terms of costs; probability of occurrence, p, in


terms of "per year"; and effectiveness, r, defined as a range from O t o 1; a n
economic/ value analysis of risk reduction can be performed. A reduced risk, p1, c a n
b e conceived as:

where: r=O represents existing or status quo condition


r=l represents a situaiton of complete effectiveness and hence, zero
risk

The expected benefit of risk reduciton can be expressed by t h e following expression:

CP - Cp' = C p - C p (i-r) = Cpr

and, Cpr = (hazard potential) x (original risk) x (effectiveness of protection


measur es)

Figure 111.4 shows this concept schematically. For the case of r = O, t h e r e is no added
protection (status quo) and p' = p. Thus, there is no change in risk a t no added cost.
For t h e case of r 1, t h e r e exists 100% effective protection (p' = O), but at a very
e=

g r e a t added cost. However, if t h e case of r = 0.9 is examined, t h e risk, p, is


decreased tenfold (p' = 0 . 1 ~ )at a modest increase in initial cost Ac. At this point, t h e
total expected loss is near a minimum value.
111.3 INTANGIBLE COSTS

To this point only tangible costs have been considered. The other category of costs,
intangible, becomes even more difficult with which t o deal. There a r e two primary
schools of thought concerning intangible costs: (1) evaluate t h e intangible costs in
monetary terms, no matter how difficult or objectionable, and include these costs in
t h e tangible costs value analysis; and (2) use history and experience as a basis for
judging what is tolerable and acceptable.

It is our opinion that the second School of thought is the most reasonable at t h e
present time. Intangible costs a r e characteristically difficult t o define or quantify.
Quantifications become embroiled in emotional issues. The second approach has t h e
advantage that history or society defines what is tolerable and thus avoids t h e
situation of a small group telling society what it should accept.

Discussion of a specific example will help clarify some of t h e points. For offshore
platforms in severe environments, t h e potential loss of human lives as a result of a
structure failure is an important type of intangible cost.

In t h e Gulf of Mexico, hurricanes pose t h e primary environmental loading threat.


These storms a r e closely monitored from birth t o death. Personnel a r e evacuated
from t h e platforms prior to their arrivai. However, in an area where severe storms
develop more frequently and quickly and facilities a r e not available t o monitor their
progress as closely, t h e evacuation of personnel is not practical. They must ride it
out with t h e platform. Thus, t h e potential loss of their lives must be integrated into
t h e decisions on platform design criteria.

As noted earlier, there a r e two attacks to evaluate the impact of such potential
losses. In this case, t h e first attack would consist of equating t h e loss of a worker's
life to perhaps someinsurance valuation, court case settlement amount, or to t h e loss
in gross national product of the country. These costs in i974 were estimated t o be in
t h e range of $100 t o $500 thousand per life (Rosenblueth, 1974; Wiggins, 1974).

III-5
-.
.i-.I
In a given platform failure, the number of lives lost might range between none (the
platform didn't collapse to t h e point where men could not find refuge from the storm,
life-saving equipment was used, or the platform was unmanned at the time) to 20 to
30. It is important t o note that we want t o focus on an average life exposure during
t h e entire platform history, spanning from t h e drilling phase when the largest crew is
onboard t o t h e production phase when there may be few or no (automated operations)
personnel onboard.

Using 20 lives lost as an example and $250 thousand per life, t h e total loss would be
valued at $5 million. Adding such a cost t o t h e previously discussed tangible costs of
failure would result in a very small change in the acceptable risk level.

This analysis doesn't properly recognize t h e importance of potential life loss. For
very large tangible costs, life loss value has almost no influence on t h e decision. In
addition? no one wants t o be put in the position of placing a value on human lives.

The second attack is based on the premise that society has historically determined
what is an acceptable or tolerable level of life loss. This level varies with t h e
percentage of t h e population exposed, t h e value of the activity? and most important,
whether or not i t is a voluntary activity.

Such considerations of the impact of technology on life are not new. They have been
considered in developing criteria for design of ships (Abrahamsen, 1963; Carsten,
19731, air traffic control centers (Raisbeck, 19711, public buildings Wiggins, 1972;
. . Wilson, 19731, and nuclear power plants (Starr, 1969; Sagan, 1972).

The background and life loss statistics have been extensively sttidied by Starr (1971).
This work indicates that life loss risks divide themselves into t w o categories:
voluntary and involuntary. Voluntary risks a r e those taken by miners, divers, airplane
pilots, etc. Involuntary risks are those due to natural disasters that affect cities, t h e
modern aspects of life (medicine, canned foods), and disease.

The best present basis on which t o compile and compare risk rates in these two
categories is to normalize them by the expected fatalities per year of exposure to t h e
particular activity being discussed. Thus, miners who a r e exposed to their work risk
about 8 hours each day (1760 hours per year) and the general public is exposed t o t h e
risks associated with power generation 24 hours each day (8760 hours per year).

111-6
- .
i-.c
~~ ~ ~ ~~
~

S T D * A P I / P E T R O 1 1 7 - E N G L 1778 0732270 0b02257 811

Starr's studies indicate that there is a general boundary between voluntary and
involuntary life loss rates. It is about equal t o t h e risk of death due to disease, or
about one death per 100 person-years of exposure. Voluntary risks fail in t h e range of
10 t o 1000 times this figure.

In t h e application of t h e foregoing data and rationale to the offshore worker and


facility, it will be assumed for purposes of illustration t h a t personnel will have no
evacuation potential and that they will face the risk of platform failure in storms
each day throughout the year. Further, it will be assumed that every platform failure
will be catastrophic in nature with a total loss of 20 t o 30 crew members.

The results of these assumptions and t h e application of the life loss data a r e shown in
Figure 111.5. ï h e variation of expected annual casualties with the platform risk level
is shown relative to t h e upper level of involuntary risk and to a tolerable voluntary
range 10 to 100 times this amount. Within t h e mid-range of t h e tolerable voluntary
range, t h e acceptable facility risk level is about one percent per year.

There is a very important contrast between this historical life loss analysis and t h e
life loss valuation analysis previously discussed. In t h e historical analysis, the impact
of life loss has been examined as a separate parameter. It has not been mixed with
t h e costs of steel and equipment involved in t h e tangible cost analysis. The difficult
decision of placing a value on human life has been avoided. I t is only a fortunate
coincidence for t h e offshore facility problem t h a t t h e results of t h e tangible and life
loss analysis indicate approximately t h e same acceptable risk level (0.5 to 1.0 percent
per year).

111.4 UTILITY EVALUATIONS


It was shown previously that, in a value analysis, cost units rather than money can be
used. These cost units are based on t h e concept of utility evaluations, in which
-
is an expression of the value of costs to a particular decision maker or group.
Both tangible costs and intangible costs can be assigned values of utility. For
example, to some decision makers, future costs, even if present-valued, have less
"utility" or value than present costs because of their future and problematical nature.
In effect, t h e bias that exists regards future failures as less important than present

111-7
S T D * A P I / P E T R O 127-ENGL 2 7 7 8 D 0 7 3 2 2 9 0 Ob022b0 533 H

initial costs. These types of deusion-makers are therefore termed "risk takers". The
opposite evaluation is t r u e for t h e risk averter. Likewise, in t e r m s of intangible
costs, a bias toward the present effects of intangible costs indicates a lower value of
utility t o t h e decision-maker. Through experience or interrogation of these decision-
makers, t h e variation of utility units (arbitrary in magnitude) for t h e vairous types of
costs and with t h e various levels of monetary costs, c a n be developed. Then, rather
than working in monetary terms, the calculations or weighing process is carried out in
utility units.

Recently t h e concept of "utility theory" has been introduced as a n evaluation process


of costs (Keeney and Raiffa, 1976; Keeney and Nair, 1975). Utility theory is used t o
represent t h e preferences for possible consequences, such as costs, loss of life, and
environmental impact. The theory is based upon t h e use of multi-attribute utility
functions.

In t h e value assessment previously described, i t was assumed t h a t t h e deusion-maker


is neither taking or averting risk; this resulted in an expected value process in which
utility units and monetary units are equivalent measures.

111.5 COST RATIO CONCEPT


Application of t h e value assessment does not eliminate. t h e risks of failure, but
instead, a t t e m p t s t o optimize t h e reliability and t h e system of initiai costs and
failure costs. Some balance between higher reliability with associated higher cost
and lower reliability with associated high future failure costs must b e established.
The previous section has explained t h e initial and failure costs, and t h e sensitivity of
a tangible cost analysis. The next s t e p is a method by which t h e optimization of risk
and cost can be identified.

As was previously explained, initiai c o s t is related to t h e design criteria and


ultimately t h e reliability of t h e facility. I t has been indicated t h a t a linear variation
of initial cost with t h e logarithm of probability of failure has been suggested as the
general relationship (Moses, 1975). Thus, equal cost increments a r e associated with
e a c h order of magnitude change in probability of failure, as shown in Figure 111.6.

111-8
S T D - A P I / P E T R O 117-ENGL 1978 = 0732270 ObO22b1 ‘47T =
Failure probability is usually expressed as the r a t e of failures per year, but lifetime
effects must be included in t h e cost. Assuming failures a r e independent from year t o
year, then t h e failure cost can be time valued to give its equivalent present worth
cost, E(F) as was shown previously:

E(F) = E(PL) x PVF x Pf


L

where: E(PL)= expected cost of potential losses in event of a failure


PVF = an interest-rate-and-time function used to present-value future
cost (summed over the life of t h e facility)
= probability of failure over the life of t h e facility.

If, instead of evaluating t h e structure on a lifetime basis, t h e probability of failure on


an annual basis (Pf ) is used as a reference, t h e present value function (PVF) can be
described as t h e present value of an annuity for each year over t h e life of t h e
structure, and can be expressed as:

PVF = i - I: i + (r-i)] -L
(r-i)

where: r = interest r a t e used for discounting profit and loss from the future
to t h e present
i = inflation r a t e
L = design life of structure,

The assumption t h a t a facility failure r a t e is independent from year to year is


generally conservative and is fundamentally derived from t h e influence of indepen-
dent random occurrences of severe storms or earthquakes from year to year.

A schematic example of total cost vs. risk rate, similar to Figure 111.2 is shown in
Figure 111.6. It indicates a relatively flat minimum with small changes in cost over a
considerable range of values of probability of failure, Pf. To find t h e optimum
probability of failure, and hence optimize the cost and reliability, t h e expected total
cost equation can be differentiated with respect to Pf, as follows:

III-9
- ..
..-
~~

S T D - A P I I P E T R O 1 1 7 - E N G L L978 II 0 7 3 2 2 9 0 Ob022b2 30b D

Since t h e damage cost, E(PL), is independent of Pf,t h e derivative of E(PL) is zero,


and

--E(') - -E(PL)
dpf
If t h e initial cost varies with t h e log of Pf (previously assumed), t h e initial cost can
be expressed as:

E(1) = ACi In Pf + Co

where: AC.1 = t h e slope of t h e initial cost curve


CO = t h e intercept of t h e curve

Taking t h e derivative of EO) with respect to Pf gives:

ACi
-d -E(1)
- -
pf pf

Substituting into t h e equation above gives:

ACi
-- - -E(PL)
pf

ACi ACi
-
Or:
pf optimum -m)=cf
Note t h a t t h e coefficient ACi refers to t h e slope of t h e c o s t curve when plotted
against natural log of Pf. If t h e log t o the base 10 is used, then ACi should be divided
by lne10 (=2.3). In addition, if Pf is the annual failure rate, rather than lifetime
failure rate, then t h e cost of failure must be modified by t h e present worth factor of
a uniform series of payments.

The resulting equation is:

ACi - 1
ao = EWL) x PVF x 2.3 - 2.3 x PVF x Cost Ratio
'f.

111-1 o
S T D * A P I / P E T R O 117-ENGL 1778 0 7 3 2 2 9 0 Ob022b3 2 4 2 D

where: f‘ - t h e ‘annual probability of failure determined on t h e basis of


ao -
minimizing t h e tangible costs associated with placement of t h e
facility and those potential costs associated with a loss of
serviceability (defined as failure here).
Cost Ratio = t h e ratio of E(PL) t o ACi, or Cf to.ACi
ACi = t h e slope of t h e initial cost curve when plotted against t h e
common log of the annual probability of failure
E(PL) = Cf = t h e expected cost of facility loss of serviceability (cost of failure)
PVF = t h e present value function (annual basis) intended to bring t o
present cost terms t h e costs of potential future losses.

If i t is further assumed that’the occurrence of the loss of serviceability events can be


adequately modeled as a Poisson process; then the optimum lifetime reliability (P, )
Lo
becomes:

P = e4Pf x L)
‘Lo ao

The optimum annual probability of failure vs. the Cost Ratio for conventional
template-type offshore facilities is shown in Figure 111.7.

Experience with offshore template platforms located in the North Sea and the Gulf of
Mexico indicates a reasonable range in t h e Cost Ratio of between 2 and 10 (Figure
111.7).
.
In basic terms, utilizing a higher cost ratio indicates an acceptance of the rationale
of providing greater reliability through increased initial costs, while a lower cost
ratio indicates willingness to accept higher future failure cost at a reduction in initial
costs and a higher annual probability of failure. The cost ratio concept can be used
within a reliability analysis to help establish a set of design criteria.

III. 6 PERSPECTIVE

In t h e context of analysis and design of offshore structures, error manifests itself in


the form of uncertainties in environmental loading conditions and platform system
resistance, and, also in the mathematical methods t h a t a r e used to describe and/or
relate loading and resistance.

c .z
111-1 1
- *
Generally, exposures (which cause risk) associated with offshore facilities can be
thought of as either technical or nontechnical (Vanmarcke, 1977).

Technical exposures a r e listed as follows:

Natural Loads: Wind, Hurricanes, Tornados, Waves, Currents, Tides,


Earthquakes, Tsunamis, Ice, Submarine Slides, Scours,
Liquefaction

Accidental Loads: Impact from ships, planes, etc.; Explosions, Fire

Material Problem: Corrosion, Fatigue, Soil Strength Degradation

Nontechnical exposures are:

Time and cost overruns, labor problems, material supply, construction


equipment problems, contract and legal problems, gross errors and/or
omission.

For t h e purposes of offshore facility reliability and value assessments, technical


exposures involving environmental loading and platform system resistance a r e of
primary concern. Other nontechnical aspects enter the problem in the analysis of
acceptable risk levels and economic considerations.

Errors which give rise to risk can be classified as follows (Fjeld, 1977):

gross errors: blunders such as result from miscalculation or


omissions by engineers, designers, etc.

systematic errors: related to ignorance of real physical phenomena.


For example, neglecting t h e effects of cyclic loading
of piles because of lack of acceptable theory is a
systematic error.

111-12
- .
~~~~ ~ ~~

S T D - A P I I P E T R O 1 1 9 - E N G L 1978 0 7 3 2 2 9 0 Ob022b5 015

random errors: these a r e associated with statistical variations. For


example, soil shear strength analysis of a given clay
under identical conditions may result in significantly
different strength values.

Reliability analysis and value assessment a r e primarily concerned with systematic and
random errors. There are consequences associated with t h e various exposures which
c a n be described as:

loss of human life or injury,


physical property/structure damage or loss,
environmental damage.

Normally, these consequences are represented in economic terms, and some


acceptable risk level is defined. A cost or value analysis is made to determine t h e
optimum design strategy.

In t h e current State-of-the-Practice, acceptable risk is inherent with t h e inclusion of


t h e factor of safety. This is a n obviously subjective or intuitive procedure. Likewise,
a management decision which assigns a n acceptable risk t o a project on t h e basis of
past data can be subjective if no cost/value analysis is included in t h e decision-
making process. Through t h e use of t h e reliability assessment in conjunction with a
value assessment, t h e uncertainty representing risk can be rationally included in
design criteria development.

III-13
T
r
-o
I-

a
a
a
u

I
18

17

16

15

14

13

I2

IO

.9

c.
I I 1 i I I I
O. I 0.2 0.4 0.6 1.0 2 3 5

PROBABILITY OF FAILURE-PERCENT PER YEAR.

FIG. 111.2 TYPICAL TANGIBLE COST ANALYSIS

WOOD WARD-CLY OE CONSU LTANTS -

-.
~
~~ ~~ ~~

c- .I

- ..
-
~~ ~~~ ~

STD.API/PETRO 117-ENGL 1778 0 7 3 2 2 9 0 Ob022bû 8 2 4 H

16

15

14

13

12
-z
V)
I-
II
3
I- 10
V)
O
o
9
I

6
O
u
o
Ew
o.
X
W
J
s
O
I-

o. I O3 0 5 0.7 1.0 3 5
PROBABILITY OF FAILURE-PERCENT PER YEAR

FIG, II1 3 I EFFECTS OF C H A N G I N G P A R A M E T E R S IN


T A N G I B L E COST A N A L Y S I S

-__ . WOODWARD-CLYDE CONSULTANTS


CP
p 1
II

Expected
Benefit of
pe x p e c t e d loss
I I

Risk expected
Reduction loss ($1
a d d e d initio1
i- cost Ac
Co'

O 0.9 ' I f-

FIG, IIL4 ECONOMICS OF R I S K REDUCTION

WOODWA R D-CLY0E CONSU LTANTS


-.
A- .t

- ..
~
~ ~

S T D - A P I I P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 9 0 Ob02270

1 I EXPECTED LIFE LOSS GIVENA FAILURE y 1 I


I
a
U
W
>
a
w
a
v)
!!i
3
.a 0.1
3
v)
a
O
n
W
b
o
. W
n
X
W
0.0 I

0.0 o I
.o1 . o. I I IO

PROBABILITY OF FAILURE-PERCENT PER YEAR

FIG, 111,s INTANGIBLE COST (HUMAN LIFE) ANALYSIS


(AFTER STARR, 1971)

WOODWARD-CLYDE CONSU LTA NTS


- .
~- ~-

STD.API/PETRO 117-ENGL 1978 0732290 Ob02271 317 D

I I I
*
0.0I o. I 1.0 10.0
--
LOG Pf - P E R C E N T P E R Y E A R

FIG, 111, 6 GENERAL VARA


I TO
IN OF COST WITH PROBABILIN OF FAILURE

L
-. WOODWARD-CLYDE CONSULTANTS ---=-
I
3
>
ki-
-z 2
d W
m u
Q a
m W
O 0
a
i O 1

-I:
a 1
3;
’ 0.8
2:
z -2 0.6

: Da
-
=ELL
+lL

$ 0
O.4

O.3
2 4 6 0 IU
I COST R A T I O : -
CF
ac1

FIG, III 7 I CûST RATIO VEFISUS ANPiUAL OPTIMUM PROBABILIN


l OF FAILURE FOR CONVENTIGNAL TLVUTE-TYPE -
OFFSHORE STRXTIJRES
l

I WOODWARD-CLYDE CONSULTANTS
- ..-
~

L_

- .
~
IV, STUDY PLATFORM

IV.1 COMPONENTS OF THE PLATFORM SYSTEM

The study platform is a 12-leg, 30-pile, 36-well steel t e m p l a t e platform in 300 f t of


water, and consists basically of a superstructure and substructure capable of
supporting 35,000 kips of drilling and production equipment. The superstructure is
comprised of decks, which support drilling and production equipment, and a template
(or tower) which connects t h e decks with t h e substructure. The superstructure also
provides súpport and protection for t h e large number of conductors, through which
wells are drilled and produced. In many ways, these groups of conductors are
analogous to elevator-facilities cores of conventional buildings.

The substructure is comprised of the soils and embedded foundation elements (piles,
mats, well conductors) which support t h e superstructure, deck loadings, and
environmental forces. During earthquakes, i t is this subsystem which acts t o transmit
ground motions t o t h e superstructure. In turn, i t must support t h e induced forces and
assist in dissipation of superstructure motions.

The platform was designed to 1977 API Guidelines for a design wave height of 100 f t
and a current of 4 feet per second (fps), decreasing linearly to zero at t h e mudline.
Such conditions produce a maximum resultant base shear on this structure of 25,000
kips.

IV.2 SUPERSTRUCWRE

Broadside and end-on elevations of t h e platform superstructure t o be discussed are


shown in Figure IV.1. The 12-leg, 3û-pile, 36-wei! conductor platform is composed of
small diameter, steel, truss-framed members. Leg diameters a r e 78 in. and brace
diameters range from 24 in. t o 48 in.

The superstructure is also:

e -
Highly redundant many alternate load redistribution paths exist.
e Truss framed - relying primarily on axial capacity of horizontal and
diagonal braces in t h e vertical and horizontal trusses f o r lateral
resistance.

IV-1
-.
L- ..-
- ..
o Multi-batter legged - the 12 legs which act to transmit verticai loads,
contain some of the piles, and due t o their batter, develop significant
iaterai resistance through axial loadings of the legs and piles.

Total weight of t h e jacket and deck elements is approximately 15,000 tons (not
including piles and conductors). Platform mass distribution (including entrained and
parasite water masses) a r e given in Table IV.l by components (Columns 1 and 2) and
by elevation (Columns 3 and 4).

Operational conditions (combined drilling and production) indicate an imposed vertical


loading of 35,000 kips (156,000 kN) to be appropriate for analysis of this structure.

IV.3 SUBSTRUCTURE
Two generalized soil profiles were developed t o characterize two potential ranges of
soils expected at offshore sites. The variation of soil shear strength with depth below
t h e mudline is shown in Figure IV.2.

The pile systems and conductors designed for these two soil conditions a r e shown in
Figures IV.2 and IV.3.

Foundation piles a r e 72 in. in diameter and have 1.5 in. wall thicknesses in t h e
vicinity of t h e mudiine, and extend 200-300 f t into t h e soil. The well conductors are
24 in. in diameter and have an equivalent thickness (recognizing effects of internai
well tubulars) of 2 in.

Soil-pile system "A" is taken as representative of locations where very firm alluvium
sediments a r e exposed at t h e mudline. Pile driving likely wouid not be feasible in
these soils, and drilling-grouting methods would have t o be employed.

Soil-pile system "Bfl is taken as representative of locations where there is a 1OO-ft


deep mantle of soft underconsolidated clays overlying t h e strong alluvium.

IV -2
- ..
=
~~ ~ ~~~

STD.API/PETRO 117-ENGL I778 0732290 Ob02275 Tb4

IV.4 DESIGN PROCEDURE FOR SIZING PLATFORM ELEMENTS


IV.4.1 Lateral Pile Behavior

The design lateral load-moment-deformation characteristics of t h e two 72-in.


diameter pile systems (pile section is elastic) are shown in Figure IV.4. The behavior
is based on the Matlock (1970)criteria for soil resistance, assuming fully cycled P-Y
(soil load-deformation) characteristics.

The imposed moments and axial loads (design value of 8,000 kips) require pile wall
thicknesses of 1.5 in. and 2.0 in. for piles "A" and 'lBfi, respectively (50ksi yield steel).

Utilizing t h e inelastic beam-column computer code, BMCOL, developed by Matlock,


et al (1967), and static P-Y characteristics, t h e pile response, elastic-plastic load
deformation behavior shown in Figure IV.5 was developed for t h e two pile-soil
systems. First and complete yield load-deflection points for the piles a r e noted. The
piles were designed to behave as compact sections so that large lateral deformations
could be developed without significant loss in pile capacity, i.e., full plastic hinges
could occur over a significant number of cycles of loading.

It will become apparent in the subsequent sections of this report, that these elastic
stiffness and plastic load-deformation characteristics will be very important in
determining t h e response of the platform system to earthquakes, much more so than
for platform response t o wave loadings.

IV.4.2 Axial Pile Behavior


. Design pile penetration curves for soil profile ftA1fare shown in Figure IV.6. Capacity
level loadings composed of t h e static operation loads and those developed by dynamic
wave and earthquake forces are referenced in these results.

The axial properties were developed using t h e findings of k a and Doyle (1975) and
t h e analytical model developed by Coyle and Reese (1966). The piles are treated as
being essentially skin-friction piles with only minor amounts of capacity derived from
- ,
point bearing or suction.

IV-3
-.
L .c

- ..
- .
STD.API/PETRO LLS-ENGL 1978 m 0 7 3 2 2 9 0 Ob0227b 9 T O D

Capacity curves for grouted piles, which would be applicable to t h e "A" soils, a r e
referenced to those applicable for driven piles. The upper and lower bounds are based
on different assumptions concerning end bearing (pile tip) resistance. The upper
bound is based on t h e assumption of end bearing in a dense sand (limiting bearing
pressure of 150 tons per square foot). The lower bound is based on the assumption of
end bearing in t h e "At1soils.

For design level maximum axial pile load of 8,000 kips and a capacity level factor-of-
safety of 1.5 (ultimate capacity of 12,000 kips), pile penetrations of 150 t o 220 feet
are required. For t h e llB1l
soil profile, penetrations of 250 t o 320 f t would be required.

Utilizing t h e foregoing lengths of piles, t h e shaft friction-deformation (T-2)


recommendations of Bea and Doyle (1975), and t h e pile tip load-deformation
recommendations of Coyle and Reese (19661,pile head load-deformation curves have
been developed. The beam-column computer code (BMCOL) developed by Bogard and
Matlock (1977) was used t o perform t h e computations, and t h e results are shown in
Figure IV.7.

IV.4.3 Structurai Eiements and Member Joints

All superstructure legs, braces, and joints were designed according t o t h e guidelines
provided by API R P 2A (1977) Sections 2.18-2.23. Use of t h e guidelines in member
design results in essentially elastic behavior under design loadings, i.e., t h e platform
was designed f o r elastic-strength level loadings.

IV.5 DYNAMIC MODELING AND CHARACTERISTICS

The primary physical elements of t h e study platform superstructure, foundation, and


soils have been outlined.

The dynamic response characteristics of t h e study platform system were determined


using a three-dimensional, elastic, time-domain computer program. The computer
model lumps self-weight, imposed, and dynamic masses at t h e nodes, determines
stiffness characteristics from t h e elastic properties of t h e interconnected members.
The model utilizes uniform modal damping t o recognize energy dissipation developed
within t h e structure and foundation, and by interaction with t h e water. In this model,
support t o the structure from t h e piles and conductors was developed by a series of
linear coupled elastic springs.

IV-4
- ..
~ ~ ~~ ~~

... STD.API/PETRO 11'7-ENGL 1978 m 0 7 3 2 2 9 0 Ob02277 ô37 m

Earthquake induced motions can be input to t h e superstructure model (supported on


elastic foundation simulation springs) base in t h e form of a three-dimensionai
acceleration-time history. At each time step in t h e analysis, t h e equations of
motions can be solved for t h e six components of mode displacements. Member forces
and stresses can then be determined from t h e displacements and elastic properties of
t h e members.

Mass distribution for t h e study platform system is again referred to in Table IV.1.
The total weight of t h e platform as modeled was 105,100 kips. 39,500 kips of this
weight was due to contained fluid (within t h e legs and braces) and virtuai water mass.
An "effective" mass of foundation elements and soils below t h e base of t h e
superstructure model was not included.

Fundamental bending and extensional mode periods of t h e platform system models


(for soil-pile systems "A" and ttBt')are given in Table IV.2.

An important consideration in t h e modeling is t h e characterization of energy


dissipation which comes from deformations of t h e structural members and joints, t h e
foundation-soil interaction, and tne superstructure-water interaction. Energy
dissipation is represented in t h e form of uniform, velocity dependent damping. The
damping coefficient (Cl is expressed as a damping ratio (8) which is t h e percent of
t h e coefficient of critical damping (Cc):

ß =- C
cc
For a simple single-degree-of-freedom system having a stiffness (k) and mass (m), t h e
coefficient of critical damping may be expressed as:

Thus, t h e damping coefficient is:

Consequently, when t h e damping coefficient is determined from a given value of t h e


damping ratio, t h e coefficient is dependent on t h e mass of t h e system.

IV-5
S T D - A P I I P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 9 0 Ob02278 7 7 3

Rainer (i975) and Bielak (1976)have developed energy dissipation related methods for
combining t h e various potential sources of damping in a complex structure-foundation
system t o result in a correct (proper amount of energy dissipation) formulation of
uniform, velocity dependent, modal damping coefficients.

Accounting for potential ranges in strain-dependent hysteretic damping and radiation


damping sources and effective masses lead t o the adoption of damping ratios of 4
percent for low-level excitations and 10 percent for high-level excitations. High
level excitations can be defined as those in which t h e induced loads equalled or
exceeded the elastic level design loads.

IV-6
- ..
TABLE IV. 1: CHARACTERISTICS OF STUDY PLATFORM

WEIGHT ELEVATION WEIGHT


COMPONENT -
(KIPS) (+ABOVE M W L ) (KIPS)
Jacket 20, 500
+ 75 40,600
Deck- Structural 4,000
+ 30 7,000
Deck-Operational 35,000
- 70 13,400
Conductors 2,100
-140 13,600
Contained Water 11,500
-220 20,600
Parasite Water 28,000 -300 9,900
Appendages 4,000 Total 105,100
Total 105,100

TABLE IV.2: FUNDAMENTAL BENDING & EXTENSIONAL


MODE PERIODS OF STUDY PLATFORM

PILE SYSTEM A PILE SYSTEM B


END-ON BDSIDE VERTICAL END-ON BDSIDE VERTICAL
-
MODE SEC SEC SEC SEC SEC SEC

1 1.77 2.08 0.36 2.43 2.45 0.36


2 0.69 0.70 0.24 0.95 0.95 0.23
3 0.36 0.34 O. 16 0.48 0.38 O. 16
~ ~

STD.API/PETRO 1 1 7 - E N G L 1778 0 7 3 2 2 7 0 Ob02280 3 2 1 =

TABLE IV.3: DAMPING RATIOS

MODERATE TO
LOW LEVEL HIGH LEVEL
EXCITATION EXCITATION
SOURCE ß - % OF CRITICAL -
B % OF CRITICAL
Steel Truss Frames 0.5 - 1.0 1-2+
(Welded Tubulars and
Joints )

Hydrodynamic 0.5 - 1.0 5+


(Small Diameter Tubulars)
Foundation Piles 1.0 - 2.0 5++
(Geometric h Hysteretic,
Lateral & Vertical

Overall Effect 2.0 - 4.0 1I++


Used in this Study 4.0 10

- ..
~ ~~ ~~ ~~

STD-APIIPETRO 119-ENGL $978 0 7 3 2 2 9 0 Ob02281 2bô m

LL
I-
à
0,

>-
ci
3
I-
V)

u,
O
cn
æ
O
œ
I-
a
3
A
W
æ
O
1
t
a
æ
W

la
z
a

--- .-. WOODWARD-CLYDE CONSULTANTS =


~-
~
~~

- - I

-
- ..
SOIL SHEAR STRENGTH- K S F
O 2 4 6
C l i

I-
W
IOC
W Q:
LL
I
W

-J 1%
a
-e
J

o
3 w
z -
3 20( e
J
W
m
I
I- 2x
a
W
0

30C

S O I L SHEAR STRENGTH P R O F I L E
= 0732290
~~

S T D - A P I / P E T R O 1 1 7 - E N G L 1978 Ob02283 O30

1Z M i m Piii II* %II? Ru


1.I*D.

FIG, IV, 3 PILE C O N F I G U R A T I O N S

'NOODWARD-CLYDE CONSULTANTS
-. -: ~~

~~

- ..
STD.API/PETRO 117-ENGL 1978 m 0 7 3 2 2 9 0 Ob02284 T 7 7

0
a
W
I
W
=1
a

v)
-x
a

I
O
z
rn
-
O

I
I-
Z
W
5
O
P
o
Q
W
I
W
-
J
a
O 2 4 6 0 10 12
PILE HEAD DEFLECTION - INCHES

F IG I IV 4
I P I LE LOAD-MOMENT-DEFORMAT I ON CHARACTER I S T I C S

WOODWARD-CLYDE CONSULTANTS
S T D - A P I I P E T R O 1 1 9 - E N G L 1978 D 0732290 Ob02285 903 D

20001 I I I I

v,
-
a l 500 -
Y
I
0
a
O
J I 000-
-I
a
a
w
I-
a
A
w 500-/ FOR DESIGN LOAD=25,000 KIPS
-
J
a + F I R S r YIELD Of PILE .
0 FULLY YIELDED SECTION

I I A
O I
10 20 30 40
LATERAL DEFLECTION AT M U O L I N E - IN.

FIG, I V , 5 PILE-SOIL SYSTEMS A AND B LOAD-DEFLECTION


CHARACTERISTICS AT PILE HEAD

- .
FIG, IV, 6 A X I A L P I L E CAPACITIES FOR D R I V E N A N D
DRILLED AND GROUTED PILES

---.- WOODWARD-CLYDE CONSULTANTS


15

IO

cl I 2 3 c
AXIAL DEFLECTION A T MUDLINE -* IN.

FIG, I V , 7 AXIAL PILE LOAD-DEFORMATION CURVES

-. WOODWARD-CLYDE CONSULTANTS
~~ ~Ö

-
~

- ..
- ~

S T D - A P I I P E T R O L19-ENGL 1978 D 0 7 3 2 2 9 0 O b 0 2 2 8 8 6 1 2 W

v. OCEANOGRAPHIC LOADING

Oceanographic loading conditions have been developed f o r t h e area of t h e study


platform, as well as other areas of t h e United States coastal waters. In order t h a t
subsequent reasonable comparison between oceanographic loadings and seismic
induced loadings be made, t h e Guif of Alaska region was selected as t h e focus of t h e
study.

The loads and forces produced on and in offshore platforms by storms and earthquakes
represent the effects of very complex processes, many of which a r e still imperfectly
understood.

The intent of this section is t o develop those indices of t h e environment which most
nearly capture t h e essence of important load-force effects on t h e class of structure
t o be studied. The effects of interest a r e those which threaten the performance of
t h e platform system (extreme conditions producing overload effects).

In t h e case of storms, t h e maximum wave height will be used as the primary index of
imposed load effects of this environmental threat. The influence of other storm
parameters (e+, wave period, directionality, currents) will be introduced as second
order effects.

v. 1 GULF OF ALASKA WAVES, WINDS, AND CURRENTS


The Gulf of Alaska is recognized as one of t h e worid?s roughest stretches of water,
rivaling t h e North Sea and t h e Southern Ocean. Large, slow moving low pressure
systems frequently sweep in from t h e North and West, pile-up against t h e Alaskan
Mountain Range bordering t h e study area. Thus, persistent, high velocity winds act
on very large stretches of water. In addition, t h e mountain passes act to concentrate
parts of t h e general wind field, causing high local winds, which in turn cause y e t
higher localized storm waves (superimposed on swell and storm waves generated by
t h e general wind system).

Unfortunately, t h e r e are insufficient measured d a t a on which to base reliable


statistics for extreme wave conditions. And, as in the Gulf of Mexico and North Sea,
one is forced into either trying an admittedly shaky extrapolation of t h e scant
measured d a t a o r in trying t o calibrate hindcast storm wave-wind models, and then
use these models t o generate t h e necessary d a t a (from historical wind field
information).
v- I
The extensive historical hindcast study by Marine Advisers (1970) formed a major
base of information. Unfortunately, this study preceded wave height measurements
in this area, and wind-wave field calibrations had to- be performed with North
Atlantic storm wave data. At this level of calibration, comparisons of t h e predicted
with measured maximum wave heights (during a given storm) indicated a n
unconservative bias of 10 to 15 percent.

Table V.l summarizes t h e Marine Advisers (1970) study results for sites in t h e study
a r e a (reference is made t o Station No. 16 hindcast results). For t h e 23 year period,
1947 through 1969, hindcast maximum wave heights for the most severe 23 storms a r e
listed (Column 3). The storm current component in direction of wave travel (at free
surface and at t h e time of occurrence of t h e maximum wave height) is also listed
(Column 4). The hindcast average (one minute, at 30-ft above water level) wind speed
(at t i m e of occurrence of maximum wave height) is listed in Column 5.

The worst storm during t h e 23-year period generated a maximum wave height of 92 f t
(period of about 15 seconds), a surface current of 4.6 fps, and a wind speed of !O5
mph (time, direction, etc. as previously defined).

A plot of t h e wave height d a t a from Table 1 vs. t h e cumulative percentage of values


t h a t are equal to or less than a given wave height is shown in Figure V.1. The d a t a
has been plotted on Log Normal Probability paper (if d a t a plots on a straight line,
process is well modeled by a lognormal distribution).

The lognormal fitting of t h e data indicates a 100-year (average return period) annual
expected maximum wave height of 1OU f t , and a n annual value of about 40 f t .
Recalling t h e 10 t o 15 percent unconservative bias in hindcast heights, these results
would be approximately 110-115 f t and 44-46 f t , respectively. These results a r e
summarized in Figure V.2.

Analysis of measured wave period d a t a gathered in t h e a r e a indicates t h a t


steepnesses (ratio of wave height t o length, H/L) in t h e range of 1/14 t o 1/16 would
be appropriate for t h e maximum storm waves. For a deepwater wave height of 100 f t
(L = 5.12 T21, wave periods (TIof the 16 t o 18 seconds would be indicated.

L .c
v-2
..... . ,.
..-. ..
............ ...
. . . ,. ..

Figure V.3 shows a comparison of results from other studies of extreme condition
wave heights for this area. The lowermost shaded band is the range of results
developed from t h e Marine Advisers study (1970) (LOCKWAVE is the acronym given
t o the hindcast computer code utilized in this study).

.._
The uppermost shaded band (Thom) was developed by Thom (1973) from measured
wind speed data in this area. In this work, a direct statistical correlation was
developed between wind speed and wave height (based on observed wave heights and
measured winds in t h e north Atlantic). Thus, t h e development of complete wind
fields for individuai storms and hindcast of wave fields from this information was
avoided.

The solid line (Quayle and Fulbright, 1975) was developed by NOAA investigators
utilizing more recent and complete wind speed statistics for this area, and t h e Thom
wind-wave correlation.

The dashed line is that from Figure V.2. These results fall about midway between t h e
other study results.

The resultant probabilistic distribution of expected maximum wave heights for the
Gulf of Alaska Study Area are given in Figure V.4.

This characterization of the probabilities of experiencing different levels of expected


maximum wave heights essentially represents t h e results of a data calibrated
(Cardone, et al, 1975) hindcast Statistical study (Marine Advisers, 1970) of
oceanographic conditions in this portion of the Gulf of Alaska.

The expected maximum %year (8ûth percentile) wave height at a given site is about
80 ft. The expected maximum 10û-year (99th percentile) wave height at a given site
is about 110 ft. Comparable values reported by Quayle and Fulbright (1975) were 69
and I i 6 f t , respectively.

The storm associated currents hindcast during t h e Marine Advisers (1970)study were
tabulated in Table V.1. These currents were those in t h e direction of wave travel
(those which potentially add significantly t o the platform forces) experienced at the
time of arrivai of t h e maximum wave heights. The currents tabulated were those at
the free surface. The study indicated that a linear decrease of these currents t o zero
at the mudline would be an appropriate approximation.
v-3
-.
L .<
In t h e wave-current force analysis which follows, the expected annual maximum wave
height characterization in Figure V.4 will be used. In addition, a storm surface
current of 4 fps in t h e direction of wave travel, decreasing linearly to zero at t h e
mudline, will be used. Platform loadings assuming a range of wave periods (15 t o 20
seconds) and currents (zero and 4 fps) will be developed for wave heights in the range
of 60 to 120 f t .

V.2 PREDICTED IMPOSED LOADING ON STUDY PLATFORM

In this study t h e three-dimensional, t i m e domain SESAS wave loading computer code


was utilized to compute hydrodynamic loadings. The 12-leg study platform was
modeled in complete %dimensional detail, including appurtenances and conductors.

Stokes' Fifth Order Wave Theory was used t o compute wave kinematics for the range
of heights and periods studied. A drag coefficient (C,) of 0.6 and inertia coefficient
(C,) of 1.5 were used in the Morison Force Equation. Current velocities were added
t o those of t h e wave field before forces were computed.

A summary of t h e gross loadings on the platform are given in Figure V.5. Figure V.5
shows t h e variation of maximum resultant t o t a l lateral force (wave and current) (BS
max) with wave height. The influence of wave period (T) has been shown as height (H)
2
t o length (L)ratios of 1/12 and 1/15 (L = 5.12 T ). The steeper waves have been
associated with t h e heights above 90 f t . The lower steepness waves and wind-surge
currents (4 fps at f r e e surface) have been combined t o describe the forces associated
with t h e lower wave heights (60 f t and less) which might propagate into deep water
essentially parallel t o t h e coast, and thus, align with t h e current components cited.

Wind loadings were determined using t h e results of previous studies of platform wind
loadings during Gulf of Mexico hurricanes. This work indicated for t h e deck height,
response period (2.5 t o 3 seconds), and projected deck area of t h e 12-leg platform a
wind loading of 1,000 kips (150 mph wind speed).

The next s t e p i s t o take t h e wave loading relationships given in Figure V.5 and with
t h e statistical description of wave heights given in Figure V.4, develop a
characterization of realizing the various levels of total wind, wave, and current
loadings associated with t h e projected storm environment (for t h e purposes of t h e
exercise wind loads were taken as five percent of t h e wave-current loadings).

v-4
- .
The result is shown in Figure V.6. The line labeled "Fc'~represents t h e computed
expected maximum resultant loading on t h e study platform.

V.3 UNCERTAINTIES IN PREDICTED LOADING

The major cause of uncertainties in t h e predicted forces shown in Figure V.6 is t h e


variability from year t o year in t h e wave heights associated with storms. However,
t h e r e are several additional causes of uncertainties which need to be accounted for.
These include:

1) Uncertainties in predicted significant wave heights. This u n c e r t a i n v


may be characterized based on d a t a developed by comparing hindcast
wave heights associated with extreme storms with those actually
measured or experienced. Line 1 of Figure V.7 presents such a
characterization.

2) Uncertainties in predicted maximum wave heights. Line 2 of Figure V.7


presents a characterization of such uncertainties where t h e ordinate of
this line is t h e ratio of the actual maximum wave height (experienced or
measured) to the maximum height predicted f o r a given location.

3) Uncertainties in predicted maximum wave heights at a given site. This


uncertainty is a location variability caused by t h e variation in t h e paths
of future storms. These uncertainties a r e characterized by Line 3 of
Figure V.7 as t h e ratio of t h e actual maximum wave height experienced
at a given site to t h a t predicted based on t h e results hindcast wave
heights.

4) Total uncertainties in predicted maximum wave height and in t h e


storm-circulation currents associated with t h e maximum wave height.
The uncertainties are characterized in Line 4 of Figure V.7. Line 4 was
developed through a multiplication of t h e three lognormal distributions
shown in Lines 1, 2 and 3.

v-5
-.
k .c

.-
~

S T D - A P I I P E T R O 117-ENGL 1 9 7 8 0 7 3 2 2 7 0 Ob02273 T B T 111

Uncertainties in predicted resultant (wave and current) water particle


velocities. These uncertainties a r e characterized by Line 5 of Figure
V.7. The resultant variance has been computed as the summation of the
variances associated with the waves and current associated velocities.

Uncertainty in square of predicted resultant water particle peak


velocities. This uncertainty directly affects t h e drag-related forces
which represent the major portion of the extreme condition peak wave
forces. The characterization of this uncertainty is shown by Line 6 of
Figure V.7. This line is based on the variance of Line 5.

Uncertainty in force equation and related drag coefficient. This


uncertainty, characterized by Line 7 of Figure V.7, is associated with
t h e application of the Morison force equation and t h e use of a constant
drag coefficient of 0.6. The characterization was developed from
measurements of water particle peak velocities (Sarpkaya, 1976).

Resultant uncertainty in predicted peak resultant lateral force. This


uncertainty is characterized by Line 8 of Figure V.7. The uncertainty is
characterized in the form of t h e ratio of actual t o predicted peak
resultant lateral force on the study platform, was derived from t h e
product of Lines 6 and 7.

These factors have been studied by Marshall (1969) and Marshall and Bea (1974) and
the resultant expression of uncertainty is given in Figure '1.6 as t h e "correction ratio
distribution (E )I?.

The characterization indicates that at the 50th-percentile t h e ratio of actual t o


predicted lateral force is 1.0; at the 1-percentile the ratio is 2.5 (or for 1 case out of
100 will t h e actual lateral force be 2.5 times that predicted).

ushg t h e relationships given in Eea (1974, 1975; Marshall and k a , 19761, the two Log
Normal variates in Figure V.6 a r e multiplied t o derive the characterization of actual
lateral forces (Fa) thereby including t h e natural (basically irreducible) and
computational uncertainty variabilities.

V-6
~~~

.. . , . ._,
..
.. . . -.,. S T D - A P I / P E T R O 1 1 7 - E N G L A978 m 0732270 Ob0227'i 7Lb m

As an example, computation of wave loadings from what are essentially two-


dimensional models of t h e sea with theories we know fail to properly describe t h e
confused, turbulent processes of a n intense storm sea. Similarly, computation of
earthquake ground motions with models which fail to realistically describe t h e
fundamental processes of energy release and transmission at both epi- and
hypocentral distances.

It is important t o recognize t h a t it is at this point in t h e analysis t h a t it becomes


essential t o e n t e r t h e best available judgement, experience, and even intuition, into
t h e analysis. The validity of t h e results may hinge on this step.

V.4 EVALUATION OF ACTUAL LOADINGS


As shown in t h e previous paragraph, t h e result is given in Figure V.6 as Fa. While t h e
50th-percentile expected annual maximum lateral force remains unchanged (Figure
V.6 indicates an unbiased estimator), t h e I-percentile level force (100-year storm has
been increased from 34,000 kips to 60,000 kips.

The uncertainties contributed by modeling variabilities obviously have an important


influence on the loading. characterizations, as they should.

v-7
~

S T D - A P I / P E T R O 117-ENGL 1978 m 0 7 3 2 2 9 0 Ob02275 8 5 2 m

TABLE V.l
HINDCAST HISTORICAL EXTREME STORM
OCEANOGRAPHIC CONDITIONS 1947-1969

Hmax vsc vw
-
RANK DATE -.
ft fps meii
1 1/16/59 92 4.6 105
2 12113/67 89 4.7 120
3 4/8/59 76 3.7 105
4 9/23/50 73 3.9 120
5 i/ 19/61 61 4.6 105
6 12/22/59 59 4.7 120
7 12/26/50 56 4.2 95
8 11/17/60 56 4.6 105
9 10128162 54 4.6 105
10 1/22/51 54 4.6 105
11 11/16/58 54 4.6 105
12 9/26/67 54 3.5 105
13 11/1/51 52 4.6 105
14 12/23/60 52 4.2 95
15 1/3/61 52 4.2 95
16 4/24/52 52 3.4 95
17 10/25/60 49 4.7 115
18 11/16/61 50 4.0 85
19 4/1/51 49 3.2 85
20 11/3/60 49 3.5 65
21 11/4/58 49 3.2 75
22 11/24/68 48 3.5 65
271 2/6/47 48 3.5 65
.. . ..
.. _ .. .
. <
- ..
_.:_.:i

9 9.5 200
99 io0
90 50
25
95

90 IO
VI
v)
w
80 5
5 70
P
IA 60
O
w 50
W
s
æ
40
w 30
o
œ
W
a 20
W I
>
i= IO
43
z 5
3
o
2
I
30 40 60 80 100 200
ANNUAL EXPECTED MAXIMUM
WAVE HEIGHT -FEET

FIG, v, 1 STATISTICAL CHARACTERIZATION OF


WAVE HEIGHT DATA

WOODWARD-CLYDE CONSI
-.
STD.API/PETRO 1 1 7 - E N G L 1778 0732290 ü b 0 2 2 9 7 b25

AVERAGE RETURN INTERVAL - YRS.

F
L
50
1

20

X
a
I

IO
9 9.9 99 95 90 70 50 20

CUMULATIVE PERCENTAGE OF
VALUES =ã

FIG, v, 2 ANNUAL EXPECTED MAXIMUM WAVE H E I G H T FOR STUDY AREA


BASED ON EXTRAPOLATION OF MEASURED DATA

WOODWARD-CLYDE CONSULTANTS
Q
W
I-
u
W
o.
x
w

CUMULATIVE PERCENTAGE
OF VALUES 4

FIG, v, 3 ANNUAL EXPECTED MAXIMUM WAVE HEIGHTS FOR STUDY


AREA BASED ON MARINE ADVISORS HINDCAST STUDY

--- ._. WOODWARD-CLYDE CONSULTANTS


\
-
~~

-
~ ~ . . -
~ ~ ~ ~ ~

S T D - A P I I P E T R O 1 1 7 - E N G L 1778 0 7 3 2 2 7 0 Ob02277 4T8

I,

I/

AVERAGE RETURN PERIOO- Y E A R S


l o , I I 1
loo0 iooi5QB K) 5 2
200 , I 1 I l I [ u l -
+
LL
I
I

201 I I I I I I l I I I
99 9 99 9s 90 70 50 20
CUMULATIVE PERCENT 5

FIG, v, 4 EXPECTED M4XIMUM WAVE HEIGHTS AT STUDY SITES

-. WOODWARD-CLYDE CONSULTANTS
.. .
. ...
._
.._.
-..
.. ..
. - ... ..
S T D = A P I / P E T R O 1 1 7 - E N G L 1978 0732290 Ob02300 T 4 T W
..... ~

1%

i2U

il0

c
WIOO
h.
I
c
I:
-W s o
=.
w
>

70

TOTAL WAVE FORCE - Fc - I O O d t KIPS

FIG, v, 5 WAVE AND CURRENT FORCES ON STUDY PLATFORM


S T D - A P I I P E T R O 1 1 9 - E N G L 1978 9 0 7 3 2 2 9 0 Ob02301 9db

RATIO: ACTUALICOMPUTED = e

-
-

æ
o 10-
5-

2-

4 6 IO 20 Bo
ANNUIIL EXPECTED MAXIMUM TOTAL WIND,
WAVE, 8 CURRENT U)AD-F,-KXX)'s KIPS

FIG, V, 6 PROBABILIN OF PLATFORMSTORMLOADINGS

L- :- WOODW ARD-C LY DE CON SU LTANTS


- *
S T D - A P I I P E T R O 117-ENGL 1978 0732290 Ob02302 A L 2 =

O .5
0.01 o. 1 1 2 5 IO 20 40 60 80
CUMULATIVE PERCENTAGE OF VALUES 2

FIG, v, 7 CHARACTERIZATION OF ADDITIONAL UNCERTAINTIES IN


PREDICTED LATERAL WAVE FORCES

WOODWARD-CLYDE CONSULTANTS
-.
L- -c
~ ~~~

- -
~~
S T D - A P I I P E T R O 117-ENGL 1778 O732270 Ob02303 7 5 9 D

YI. PLATFORM RESISTANCE TO OCEANOGRAPHIC LOADINGS

VI.1 GENERAL

The basic structural and foundation systems of the study platform are illustrated in
Figure VI.l. Each of these systems is a composite of many individual members and
joints, having a variety of failure modes. The expected performance of each of the
members or their systems c a n be characterized in terms of i t s probability of failure
versus total lateral force (failure is considered t o be collapse or t h e inability of the
structure to' support vertical loadings).

Basic d a t a on t h e performance of individual elements have been presented by


Marshall (1969) and are summarized in Figure VI.2. These d a t a are presented in terms
of the safety index, ß . The Safety index is t h e ratio of t h e central safety margin (E-
3, mean resistance - mean loading) t o t h e resultant uncertainty in loading and
resistance (i=}, U
: is resistance variance and 02 is loading variance), or
S

The safety index may be thought of as a "sanitized" measure of failure probability, Pf


(probability of actual load exceeding collapse load), with t h e following rough
correspondence:

Pf (approx.

50%
15%
2%
1 in 1000
i in lo5

The bias and s c a t t e r in these basic strength elements result from omissions and
simplifications in analysis, variation in material qualities, and workmanship errors.
The distributions given result from comparisons between assumed (or calculated)
strength and measured strength, in t h e various test series used t o develop and/or

VI-1
S T D . A P I / P E T R O 1 1 7 - E N G L 1778 D 0 7 3 2 2 7 0 O b 0 2 3 0 4 b y 5

justify design criteria for tubular members (including welded tubular connections at
each end). They represent only those sources of bias and uncertainty which are known
and can be quantified.

VI2 SYSTEM RELIABILITY


The information on t h e performance of individual elements (Figure VI.2) must b e
translated to a description of overail system reliability. Failure of t h e individual
elements rarely will b e catastrophic. Redundancy (alternate load redistribution
paths) allows t h e system to continue to perform satisfactorily.

In t h e subsequent discussion of platform system reliability, several new t e r m s are


introduced:
O Series System - a weak link structure or system in which t h e overall
s t r u c t u r e fails if any element fails.
O Parallel System - a fail-safe; system in which more than one member
must fail before t h e structure fails.
O Ductile Element - an element which will continue t o carry a major
portion of its imposed loading when i t reaches its elastic limit.
O Brittle Element - such a n element will cease t o carry load when it
reaches its elastic limit.
O Correlated Elements - t h e r e is a positive relationship between t h e
strengths of t h e elements, if one has a high strength, then t h e others
a r e likely to have high strengths, and vice versa.
O Independent Elements - t h e r e is no correlation of t h e strengths of t h e
elements.

The complete platform is an extremely complex assemblage of elements with many


failure modes. Two simplified models were adopted from earlier studies Marshall and
Bea (1976): (1) Brittle Redundant, and (2)Ductile Redundant.

In t h e Brittle Redundant analysis, t h e members a r e assumed brittle so t h a t when a


member fails, i t drops all of its load. The structure is then reanalyzed with t h e first
member removed. Progressive failure occurs as t h e s t r u c t u r e is unzipped. In this
analysis t h e overall failure probability was taken as:

VI-2
-.....
.%

- .
~~~ ~

S T D - A P I / P E T R O L17-ENGL 1978 m 0732270 Ob02305 521 m

Pf = Pfl x Pf2 x . .. .'fn


where Pfl , - probability of any member failing in t h e original structure

pf2
- probability of any member failing in t h e second stage, with t h e
first member removed

'fn
- failure probability of t h e last stage leaving a n unstable structure
with no load carrying capacity.

In the ductile redundant analysis, it was assumed t h a t t h e members would continue to


carry their load after t h e member reached its elastic limit. Further it was assumed
t h a t there was full load sharing between parallel members or units of t h e platform,
t h a t members at a given level were perfectly correlated, and t h a t failure of various
levels or elements in series were independent. The safety factors for t h e platform
elements were based on a space f r a m e analysis and a beamcolumn ultimate load
analysis of each member. The overall safety index was found by assuming t h a t
elements in parallel had their safety factors averaged (correlation) and t h a t elements
in series have their failure probabilities added (independent).

After the resistance distribution of each primary type of failure mode was found in
t h e platform (superstructure, laterally loaded piles, axially loaded piles), t h r e e were
combined into an overall strength distribution as follows:

FR(') = P Strength - r

where FR(') - platform system strength distribution

FSS(r) - strength distribution of superstructure (jacket and deck)

FA# - strength distribution of axially loaded foundation

FLF(r) - strength distribution of laterally loaded foundation

VI.3 SUPERSTRUCTURE
The indicated performance of tubular structures is very sensitive t o how t h e safety
factor is computed. Although A I X plus 1/3 corresponds to a nominal safety factor of

.. .
VI-3
-
~ ~ ~ ~

S T D - A P I / P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 9 0 Ob023üb 4b8

1.25 against first yield, analytical studies (Fowler, et al, 1973) of ultimate strength of
compact tubular beam columns (Sherman and Glass, .1974) indicate considerable
plastic reserve, as shown in Figure VI.3.

Beyond t h e influence of failure strength of individual members, behavior of t h e


composite structural frame also depends on t h e failure modes of t h e members.
Several types of post-failure behavior are compared in Figure VI.4. A t one extreme is
t h e brittle mode in which t h e member is rendered ineffective once failure occurs.
Examples include failure by brittle f r a c t u r e and local buckling of large diameter
unstiffened tubes. At t h e other extreme is t h e behavior of ideally ductile tension
members. The failure behavior tubular beam-columns (Bouwkamp, 19741, with some
yielding in t h e joints, falls somewhere in between.

To examine t h e collapse probability of t h e structure using t h e brittle mode (Fig.


VI.41, f i r s t consider t h e original intact structure (Stage 1). Individual brace failure
probabilities are obtained from Figure VI.2, using calculated safety factors
determined for a given loading system. The probability of a n initial failure
somewhere in t h e structure is obtained by combining individual brace failure
probabilities:

where: P[a] P[b] = P[a] + P[b] - P l a l P[b] =1 - ( - P r a l ) (l-P[b]) .


Assume brace "attsuffers t h e initial failure. By removing brace "a"and re-analyzing
t h e structure, new safety factors and failure probabilities can be obtained which are
conditional upon t h e failure of "a". Thus, Stage 2 represents t h e conditional
probability of additional failures.

The process is repeated for progressive failure of braces "b", llcll,and "dV1,until a
mechanism for collapse develops. Collapse requires t h e combined occurrence of

VI-4
.L .c
- *
- .
Stages 1, 2, 3, and 4, so that the overall failure probability is the product of the stage
probabilities. Thus, for a redundant, fail-safe structure, t h e probability of collapse -
with the attendant large dollar losses and pollution potential - is lower than t h e
probability of initial, localized failure.

For a series structure in which any failure leads to collapse, increasing t h e number of
members increases the risk of failure, corresponding t o a reduced value of B for the
overall structure. On the other hand, where many members in parallel share the load,
if one fails, the remaining members may still carry the load; for such fail-safe
structures, the reduced risk of complete collapse is reflected in increasing values of
B. For the offshore structures, where initial failures introduce eccentricity of
loading and load sharing is less than ideal, intermediate results a r e obtained (load
variability still being excluded).

These results, using basic AISC allowables, for simple 4, 8, 12, and 16-pile offshore
structures are illustrated in Figure V1.5, along with those for idealized series and
parallel systems. The results for AISC plus 1/3 (on allowable stress) a r e also given.
These indicate that the overall safety index B tends to level off, meaning that the
expected performance of more complex structures (i.e., 12 or 16 piles) is comparable
t o the simpler 6 or %pile structures.

For the ductile mode, safety factors for each member can be computed using the
results of a structural analysis and a family of ultimate strength failure envelopes
similar to Figure VI.3. These safety factors were converted t o failure probability or
safety index using the "single tubular member" curve of Figure VI.2. individual
members are combined into successively larger and larger assemblages as follows:

a for load sharing braces, the capacities add, and the combined safety
factor is the weighted average of individual safety factors.
e for assemblages in series, t h e failure Probabilities are @ added.
a similarly for K-braces in series.
o assemblages in parallel combine to make a substructure of a given level
or truss bay.
o substructures are stacked up vertically in series to make up t h e overall
structure.

VI-5
- -
- Y
Results for t h e ultimate strength ductile model a r e considerably more optimistic
(higher failure load at a given level of risk) than t h e f i r s t yield, brittle design results.
This can be seen in Figure VI.6 where t h e results of several analyses of overall failure
probability versus total lateral load a r e illustrated. For t h e given design load (3200
kips), t h e first yield brittle design has t h e highest failure probability. The ultimate
strength ductile design model gives a failure probability several orders of magnitude
lower than t h e first yield case,with t h e ultimate strength/six worst members case as
a n intermediate situation. A similar result can b e obtained for a given failure
probability (say Pf = 0.1) - the ultimate strength ductile design has approximately
twice t h e resistance as t h e first yield brittle design, with t h e ultimate strength/six
worst members case in between. The ultimate strength model can be used as a
realistic bound on structural reliability.

V1.4 SUBSTRUCTURE

The foundation may be visualized in terms of laterally and axially loaded piles. The
laterally loaded piles transmit t h e structure's base shear to t h e seafloor soils. They
a r e usually analyzed with a finite increment computer program which considers t h e
flexural stiffness of t h e pile and t h e nonlinear loaddeformation characteristics of
lateral soil support. Although intimate soil-pile interaction is involved, serviceability
is usually defined in terms stresses in t h e steel pile. Axially loaded piles resist
vertical and overturning loads in t h e structure. Usually, their safety factors a r e
defined relative t o bearing (including friction) or pull-out failure in t h e soil.

Failure envelopes for overturning and vertical loads - resisted by axially loaded piles -
a r e shown in Figure v1.7. Normally, piles are designed to have a safety factor of 1.5
against t h e most heavily loaded corner pile reaching its ultimate capacity (maximum
load carrying capability) under t h e influence of diagonal wave loads, using elastic
analysis. However, considering t h e generally ductile load-deformation characteristics
of axially loaded piles limit analyses of t h e example plot form 30 -pile group,
indicate t h a t t h e ultimate failure envelope shows considerable reserve strength.

The lateral resistance of t h e pile foundation is also analyzed as a unit for purposes of
risk evaluation. In conventional elastic design we calculate a safety factor against
localized yielding in t h e most heavily loaded pile. However, limit state or ultimate
analysis again shows a large plastic reserve for this platfo-rm configuration.
Considerable additional capacity can be obtained from t h e most heavily loaded pile by
considering fully plastic interaction between axial load and moment, as well as
moment redistribution as hinges form in t h e pile. Still further increases come from
t h e remaining piles; since they carry less axial load, they can take on more lateral
load as collapse mechanism for t h e entire lateral foundation develops.

Both axial and lateral foundations have significant reserve strengths, not considered
by t h e usual design procedures. Foundation ultimate behavior and failure risk were
analyzed at several levels of applied lateral force on t h e model structure, with the
results summarized in Figure VI.8. Median failure loads on t h e order of twice the
design level a r e indicated.

VI.5 PLATFORM SYSTEM RESISTANCE EVALUATIONS

in this evaluation, storms will be characterized in terms of hindcast total lateral


force on t h e 12-leg structure. This places t h e structural and oceanographic sides of
t h e problem on a common basis, and is much more manageable than t h e joint
probability distribution for wave height, period, and direction; current velocity and
direction; wind speed and direction.

Given a hindcast storm intensity, SHY t h e probability of structural failure (collapse) is


given by:

o o r
load level (1)
load pattern f(l)dl
rest of system
uncertainty
in force
given considered levei (i)
storm given storm

The summation term inside the brackets indicates that we first combine structural
and foundation failure modes, recalling that these are not quite independent, but
rather conditional on a common load level, load pattern, and the rest of the system
functioning. The result for combining failure modes for structural frame @ axial piles

VI-7
@ lateral piles are shown in l i g u r e VI.9. The range of results corresponds t o ductile
vs. brittle of the structural frame.

After t h e structure is fully synthesized, t h e uncertainty in applied loads is then


introduced. Figure VI.10 illustrates t h e spread of applied lateral force for t w o values
of hindcast lateral force. The probability density function f(l) is derived from t h e
cumulative distribution function of Figure VI. 11.

The quantification of t h e resultant uncertainty in predicted maximum wave forces is


given in Fig. VI.ll. The result has been formulated from t h e uncertainty in t h e
predicted wave force, given a wave height (indicated ak t h e uncertainty in t h e drag
coefficient, C 1, and f r o m t h e uncertainty in t h e predicted wave height (indicated as
D
t h e uncertainty in t h e square of t h e wave height, H
2>.The force reliability profile
indicates t h a t given a sea state prediction or hindcast, t h a t t h e maximum resultant
lateral forces due to waves will be greater than hindcast (on t h e unsafe side) for only
26 percent of t h e cases. The actual maximum force shows a median bias on t h e s a f e
side, equal t o 0.8 times t h e hindcast force. The resultant coefficient of variation is
about 40 percent.

The result of the integration of structural and oceanographic uncertainty for various
levels of hindcast lateral force is given in Figure VI.12. This indicates an even chance
of t h e structure surviving 200% of t h e design lateral force; t h e principal e f f e c t of
including oceanographic uncertainty is t h a t of now employing a wider range on t h e
f o r c e scale (compare Figures VI.9 and VI.12). This is a priori evaluation of t h e
reliability of t h e structure and design technology. It is as y e t uncalibrated, and still
excludes random storm occurrence.

VI.6 CALIBRATION

-
With these results, t h e next s t e p is calibration using platform survival experience in
severe hurricanes to update t h e platform reliability results developed thus far.

The method used t o update t h e assessment of platform and design technology


reliability is t h a t suggested by Cornell (1972) as well as Tang and Ang (19731, as
outlined here. The t r u e resistance is taken t o be t h e product of a correction factor
and the calculated resistance, i.e.

VI-8
~ ~~~ ~~

STD*API/PETRO LL7-ENGL 1778 0732270 Ob02311 8 2 5

TRUE
[RESISTANCE] = [ CORRECTION
FACTOR
CALCULATED
RESISTANCE I
The calculated resistance Rc will be taken on t h e conservative side of t h e Figure
V1.12. This includes sources of uncertainty and bias which is known to be present.
The correction factor X represents professional uncertainty, representing effects of
uncertainty not considered (e.g., variations in load pattern), bias in data interpreta-
tion, and errors in extrapolation from t h e empirical d a t a base to the design situation.
The correction factor X is represented by a random variable, giving a whole family of
possible true resistance curves, as shown in Figure VI.13. A I9prior" distribution for
t h e correction factor, S, is developed and then updated using platform proof test
experience and Bayes Rule. Note that it is the correction factor distribution which is
being updated; that is, Bayes Rule is applied only t o the uncertainty in X which can be
reduced or eliminated by experience. After it is updated, it will then be multiplied
times the original reliability characterization.

As discussed by many investigators (MacFarland, 19721, t h e success of the Bayesian


updating process hinges on the development of a reasonable Prior. Results reported
in the literature on ultimate loading of braced steel frames have been used to guide
t h e description of t h e correction factor Prior. The Prior X shown in Figure VI.13, was
discretized to facilitate numerical application of Bayes Rule, as given by:

This is applied successively for each of the platform survivals using t h e hindcast force
ratio SH/RD together with Figure VI.13 t o obtain t h e indicated probabilities. The
Posterior X distribution based on hindcast results is shown in Figure VI.13 also. The
hindcast Posterior indicates a median X correction factor of 1.6.

VI-9
- .-
I
S T D . A P I / P E T R O 1 1 7 - E N G L 1978 m 0 7 3 2 2 9 0 Ob023L2 7 b L

The same Bayesian updating can be applied using "actual" load data based on observed
wave heights, filtering o u t t h e oceanographic uncertainty of wave hindcasting
relationships, t o obtain a calibrated platform resistance. The experience Posterior
indicates a median X correction factor of 1.3. Although we started out with t h e
brittle model, this brings us very close to the ductile model.

The calibrated reliability assessment, obtained by multiplying t h e correction factor


distribution times t h e resistance distribution is shown in Figure VT.14. At t h e 50th
percentile level, the Posterior collapse load is indicated to be about 3.5 times t h e
design load if t h e hindcast loadings a r e used; it is about 2.5 times t h e design load if
t h e experienced loadings are used. For t h e hindcast loading "systemt1, t h e platform is
about 99 percent reliable for a loading equal to its design loadings. Based on t h e best
d a t a available, the actual resistance results in less than a 2 percent probability of
failure for loads equal to t h e design loading. The resulting coefficient of variation,
which includes the loading uncertainties for a given sea state, is 55 percent.

The computed annual probability of failure for various levels of design load using t h e
previously described loading and resistance distributions is shown in Figure VI. 15. The
average annual risk rate R is computed by summing over all t h e ranked storms
occurring at a typical site in t h e Guif of Mexico,

1.
max
Where Tmax is a sufficiently long t i m e interval (in years) for which ranked storm d a t a
is avaiiabie (or has been generated by simulation), and Pf values a r e taken from t h e
calibrated (based on hindcasts) reliability curve of Figure VI.14. The results for t h e
Prior R refer t o t h e original approximation. The Posterior R results refer t o t h e
resistance distribution updated by hindcast loadings. This range should bound t h e
reasonable resu 1ts.

VI-1 o
= 0732270
~~

S T D - A P I I P E T R O 117-ENGL 1778 Ob02313 b T B

I STRUCTURAL
I FRAME

FOUNDATION

FOUNDATION AND STRUCTURAL SYSTEMS


FOR STUDY PLATFORM
~
~

- _.
:*...
)..,
.
..,:
.__. S T D - A P I I P E T R O 117-ENGL 1778 0732270 Ob0231'4 53'4
_. ' -

I
i -
1

4
-0.5 1 S

CALCULATED SAFETY FACTOR

.. .

FIG, V I , '2 R I S K FACTOR FOR NON-STATISTICAL LOADS

WOODWARD-CLYDE CONSULTANTS
-..
c -%

- .
I
E
a
W
a

I-
œ
O
IL

v)
W
a
O
W
>
æ

æ!-
0-
cris
U

OVO7 W i X V 3 A I l V l 3 ü

--- .
.c WOODW ARD-C LY DE CONS ULTANTS
STD.API/PETRO 119-ENGL 1778 0 7 3 2 2 7 0 O b 0 2 3 L b 307

MAX.
CAPY.

IDEALLY DUCTILE
TENSION MEMBER

n
U
MEMBER

sz
.25 f-- CATASTROPHIC

VI BRITTLE FAILURE

O .O01 .O02 ,003 .O04


AXIAL STRAIN

FIG, VI, 4 POST-FAILURE MODES OF MEMBERS


BRITTLE
PARALLEL

/ REDUNDANT
" EXACT" O ES I GN

. "EXACT" DESIGN
1/3 INCREASE
I

1 1 1 1 1 1 1 il il^
1 2 3 4 5 6 7 8 9 101112
NUMBER OF ELEMENTS

FIG, VI, 5 SAFETY I N D E X FOR COMPOSITE STRUCTURES,


C O N D I T I O N A L UPON KNOWN LOADS

WOODWARD-CLYDE CONSULTANTS
-.
& .c
i
~~ ~~ ~
~~

STD.API/PETRO 119-ENGL 1778 0 7 3 2 2 7 0 Ob02318 L B T H

.5

.1

Pf
.o 1

,001

.o001

O O

c
o
?
c

APPLIED LATERAL FORCE

FIG, V I , 6 RESULTS OF ANALYSES PLATFORM-


SUBJECTED TO KNOWN LOADS

& .c
WOODWARD-CLYDE CONSULTANTS
.
.... ... ......:.
.. .-...~....<_.
.....

o
O 1 2 3 4 X lo6 K-FT,

FIG, V I , 7 AXIAL FOUNDATION OVERTURNING RESISTANCE


S T D - A P I I P E T R O 117-ENGL 1778 W 0 7 3 2 2 9 0 Ob02320 838 W

O O O
O
O
c
0
c
-

APPLIED LATERAL FORCE

FIG, VI, 8 RESULTS OF A N A L Y S I S - FOUNDATION


SUBJECTED TO KNOWN LOADS

WOODWARD-CLYDE CONSULTANTS
- .
L -c
S T D - A P I / P E T R O 1 1 7 - E N G L I978 D 0 7 3 2 2 9 0 Ob02321 774

.5

COMBINED
STRUCTURAI
MODES
.1

Pf
.o

.o01

.o001
O O
h O O
O O O O
O
w a s O
0-
c

APPLIED LATERAL FORCE


(COLLAPSE LOAD)

COMBINED STRUCTURE AND FOUNDATION .RELIABILITY


FOR MODEL SUBJECTED TO KNOWN LOADS

-. WOODWA RD-CLYDE CO NSULTANTS


-
,
.
~ ~

S T D . A P I / P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 7 0 0 b 0 2 3 2 2 bOO

f (I) FOR f ( I ) FOR


H INDCAST MAXIMUM
LATERAL FORCE CREDIBLE
S, =3,300k STORM
SH=15,OOO k

\
-i \
\
m
a \
\
8œ \
e

O 0 O O
O
O
c
- O
O
N
O
O
O
O

APPLIED LATERAL FORCE (I)

FIG, V I 10 SPREAD OF A P P L I E D LATERAL FORCE ( ) FOR TWO


I

VALUES OF H I N D C A S T LATERAL FORCE t SH)

WOODWARD-CLYDE CONSULTANTS
......_
......
i ._. -..-
......f: .
.2.
. ... :
.
.
S T D - A P I I P E T R O 117-ENGL L778 0 7 3 2 2 7 0 Ob02323 5Y7 D

...
. .

FWAVEa Co t i 2

99
90

.. 95
VI
... . . ..
_. v>
W
90
. .
3
Q 80
>
5 70

U 60
. ..
5
Lu
50

W
40
a.
30
-
UJ
>
2
d
20
3
s
æ 10
V

2
1
0:3 0.6 1.o 2.0 3.o

RATIO: F ACTUAL
.. F HINDCAST

FIG, VI, 11 R E L I A B I L I T Y OF HINDCAST LATERAL FORCE

---. .- W O0 DWAR D-CLYD E CONSULTANTS


-
~

.
-~
STD.API/PETRO LIS-ENGL 1778 0 7 3 2 2 9 0 Ob02324 483 =

( SH/RD)
HINDCAST LATERAL FORCE
AS PERCENT OF DESIGN

50% 100% 200%


O .5

1
.1

SAFETY 2
INDEX Pf
ß .o1
3 .o01

.o001
4
O
O
O
O
O
O 8
O
F
s O
v)
O
0-
F

H i NDCAST LATERAL FORCE ( SH)


KIPS

FIG, V I , 12 CALCULATED R E L I A B I L I T Y OF MODEL STRUCTURE


AS A FUNCTION OF H I N D C A S T L A T E R A L FORCE

L .%
WOODWARD-CLYDE CONSULTANTS
.
.. .... . . ....
.. .....
..
.__. ...
_..
... .
. .._ ..
.--i
. .
STD.API/PETRO 1 1 7 - E N G L 1978 H U732270 Ob02325 3 L T D .

. ... .
, .
.. ..
.......... .:

._
.. .

FIG, ‘JI, 13 F A M I L Y OF POSSIBLE T R U E RESISTANCE C U R V E S

-. WOODWARD-CLYDE CONSULTANTS =--- - -- r - v

-
- .
S T D . A P I / P E T R O 117-ENGL 1778 = 0732270 Ob0232b 2 5 b

99

98

95

90

80
70
60
50
40
30
20

10

2
1
60 80 100 200 400 600 800
HINDCAST LOAD AS o
/' OF DESIGN LOAD
(SH /R D )

FIG, V I 14.
I CALIBRATED PLATFORM R E L I A B I L I T Y

L- ..- WOODWARD-CLYDE CONSULTANTS


S T D - A P I I P E T R O LLS-ENGL 1 7 7 8 m 0 7 3 2 2 7 0 üb02327 1 7 2 =

FIG, V I , 15- INCREASING T H E DESIGN LOAD DECREASES


. .THE AVERAGE ANNUAL RISK RATE
~ ~ ~~~ ~

=
~~ ~

S T D - A P I / P E T R O 119-ENGL 1978 m 0732290 O b 0 2 3 2 8 O29

VIL EARTHQUAKE CONDITIONS

VII.1 GULF OF ALASKA SEISMIC EXPOSURE


VIL 1.1 Seismic Modeling

The paper by Page (1975) describes t h e tectonic, seismic, and geologic background f o r
t h e portion of t h e Gulf of Alaska t o be discussed here - t h e area between Kayak
Island and Pamplona Ridge shown in Figure V1I.la.

A simplified model of potential sources of earthquakes has been developed f o r this


area and is shown in Figure VII.lb. The model consists of six major parts:

I. Chugach-St. Ellias System - primary and branch faults which parallel


t h e coast.

2. Shelf Edge System -


lies at t h e base of t h e Continental Slope and
parallels t h e Chugach system in this area.

3. Kayak Island System - oriented generally perpendicular t o t h e Chugach


and Shelf Edge Systems and located at t h e western end of t h e study
area.

4. Pamplona Ridge System - located at the east end of the study area, i t is
a system of primary and branch faults which lie generally perpendicular
t o t h e Chugach and Shelf Edge Systems.

5. Shelf Fault System - a randomly oriented system of small faults


underlying t h e e n t i r e study area.

6. Malispina Block System - a large block of essentially unbroken Holocene


sediments extending from Pamplona Ridge t o Cross Sound - an area of
I about 60,000 square kilometers. The vertical perimeters of this block
1
are formed by t h e Chugach and Shelf Edge systems and t h e horizontal
perimeter at a depth of 15 to 20 kilometers by t h e underlying Queen
Charlotte Islands formation.

VII-1
~
The first five parts of this model are taken t o be the potential sources of quakes
which c a n reasonably be identified as being random - their locations, magnitudes,
numbers, effects, and times of occurrence being very uncertain. The seismic risk
model proposed by Cornell and Vanmarcke (1975) will be applied to this portion of t h e
model.

The sixth p a r t of t h e model, t h e Malispina Block System, i s taken t o be t h e source of


t h e non-random event postulated by Sykes (1971) and Kelleher, et al (19731, and
discussed by Page (1975). In addition to the tectonic and seismic history analyzed by
Sykes and Kelleher in developing t h e proposal of a "seismic gap" for this portion of
t h e Circum-Pacific Belt, t h e geology and past record of intense events left in the
Holocene sediments, particularly in the a r e a of Pamplona Ridge, lends much
credibility t o this proposal. I t will be assumed t h a t this p a r t of t h e model will be the
s i t e of a very large magnitude quake, in the Richter Magnitude range of 7.5 t o 8.5,
within t h e next 25 to 50 years.

These two major parts of t h e seismic model will be analyzed as independent loading
conditions for platforms t o be sited in the area. Their potential effects on two sites
will be studied - a site adjacent t o t h e Kayak Island System (KI) and another adjacent
t o t h e Pamplona Ridge System (PR)(see Figure ViLlb).

MI. 1.2 Seismicity


The historical d a t a on locations and quake magnitudes cited by Page (19751,
supplemented with other d a t a (U. S. Dept. of Commerce, 1966; NOAA, 19751, has
been used to develop t h e activity rates and magnitude probabilities for each of t h e
five sources which comprise the random quake model. Table VII.1 details t h e
-
seismicity parameters based on t h e historical d a t a file covering the period 1899
through 1974.

The table lists f o r e a c h of t h e five sources, t h e number of events having Richter


Magnitudes equal t o or greater than 3.5, t h e magnitude of t h e largest event recorded,
t h e average number of quakes experienced each year, and a distribution coefficient
(b) which relates t h e frequency of occurrence of quakes of different magnitudes
(Cornell and Vanmarcke, 1975). The b-coefficient given h e r e is t h e one t h a t forms a
best fit t o t h e e n t i r e range of magnitudes. In t h e detailed modeling work, (b) was var-
ied according t o t h e different magnitudes (Merz and Cornell, 1973).

--.... VI 1-2
..
. ....... ....- i
.
S T D - A P I / P E T R O 117-ENGL 2778 0 7 3 2 2 7 0 üb02330 787 D
,... .-:
.-Y-
~
.:.. . j

Commonly, t h e b-coefficient applicable t o t h e e n t i r e range of significant magnitudes


will fall in t h e range of 0.5 t o 1.0 (Evernden, 1970). Smaller values indicate a
tendency toward experiencing larger magnitude events; given t h e ' occurrence of a
quake. The d a t a certainly indicates such a tendency for this area. Similar
conclusions have been reactied f o r this general region (Evernden, 1970).

Unfortunately, t h e data summarized in Table VIL1 has many deficiencies. I t was not
until t h e 1940's t h a t instruments were installed in Alaska which were capable of
accurately determining t h e magnitude, location, and occurrence of significant events.
Thus, prior to t h e 1940's many of the lower magnitude events are missing from t h e
historical record. Further, many of t h e larger events are only very approximately
known in terms of their magnitudes and locations.

There is a further inconsistency. During 1964, t h e Prince William Sound quake (M = 8


to 8.5) virtually flooded this area with fore- and aftershocks. During the four months
after t h e main event, some 1260 aftershocks affected this region (Algermissen, et al,
1969). For some of t h e sources shown in Figure VII.lb, more than half of t h e number
of historical events were contributed by t h e Prince William Sound aftershocks.

Consequently, t h e data from t h e period 1899-1974 is a very mixed lot. The parameter
of primary concern is t h e activity rate, since it is likely t h a t a sufficient quantity of
d a t a is available from this and similar areas t o determine t h e appropriate ranges for
t h e magnitude distribution coefficients (Evernden, 1970). Shown in Table M1.2 are
t h e activity rates f o r the five random quake sources, subdivided into four stretches of
history:

1. 1899-1974 -
t h e entire historical record, including four events with
magnitudes equal t o or g r e a t e r than 8.0.

2. 1940-1974 - t h e record since installation of Alaskan seismometers,

3. 1964-1974 - t h e record since and including t h e Prince William Sound


earthquake,

4. 1965-1974 - therecent record, excluding t h e Prince William Sound


event and aftershocks.

VII-3
=
~~ ~ ~~ ~

STD.API/PETRO 117-ENGL 1 9 7 8 0 7 3 2 2 9 0 Ob02331 b 1 3

For these periods of history, t h e activity rates range nearly over an order of
magnitude. The second s t r e t c h of history, 1940-1974, will b e used as a base case, and
t h e other periods of history utilized t o illustrate other possible óutcomes during t h e
forecast period (next 25 to 50 years).

VII. 1.3 Attenuation


The previous discussion has centered on identifying t h e potential locations of quakes,
their activity rates, and probabilities of experiencing various magnitudes given t h a t
a n event occurs. The next s t e p in t h e process is t o identify the relationships which
will be used t o define t h e expected effects of a quake on a given site which is a
distance (R)from t h e focus of a n earthquake of magnitude (M).

This is a particularly difficult p a r t of t h e problem for mapy reasons. First, no single


parameter competently reflects t h e potential effects of a n earthquake at a given site
on a given structure. Acceleration, velocity, displacement, frequency content, and
duration all play potentially major roles in determining t h e effects on soils,
foundations, and superstructure which comprise a platform system. Of predominant
influence in determining t h e parameters of importance . are t h e response
characteristics of a particular platform system.

In addition, for this particular area, we are forced t o rely on ground motion
measurements made onshore, primarily in Southern California. There have been no
measurements of strong ground motions made at offshore sites. W e know t h a t t h e
unique characteristics of a n event at its source, its intensity, travel path geology,
location of t h e s i t e relative to t h e source or sources, and local geology-soil conditions
all influence t h e results. Thus, t h e r e a r e many sound reasons to treat with caution
t h e relationships which have been developed from studies of data from Southern
California as they a r e extrapolated t o t h e Gulf of Alaska.

The attenuation relationships developed by McGuire and Cornell (1974) will be


utilized to describe t h e expected effects of an earthquake at a given site. These
relationships will be taken as applicable t o firm ground sites, typical of those from
which t h e measured d a t a were obtained. The error terms associated with these
relationships will be included t o recognize t h e very large uncertainties in predicted
ground motion parameters associated with t h e use of these relationships.

VII-4
~ ~~ ~

'.
4-
. ..; ..-.i.. S T D * A P I / P E T R O LLS-ENGL 1778 W 0732270 Ob02332 5 5 T

Source-to-site transmission relations (commonly known as attenuation laws) can be


conveniently expressed in t h e form:

Y = c l e-c2M (R + 25)-c3 E

where: Y = measure of ground motions at a site caused by an event of


Magnitude, M, at a focal distance of R (kilometers); for example,
peak ground acceleration (A max), velocity (V max), displacement
(Dmax), or response spectra velocity (Sv)
E = e r r o r term which expresses t h e ratio between observed and
predicted ground motion parameters
C = parameters whose values are estimated from past measured strong
ground motions data.

McGuire (McGuire and Cornell, 1974) has studied measured strong ground motions
d a t a from Southern California sites. He has included records from some 33 large
magnitude events, including those from t h e 1974 San Fernando quake. His extensive
and consistent t r e a t m e n t of these data justified use of his attenuation law
coefficients in this study. The coefficients are tabu1a;ed in Table WI.3.

McGuire's relationships are taken as applicable to firm ground sites, typical of those
from which t h e measured d a t a were obtained. The error terms associated with these
relationships were included to recognize t h e very large uncertainties in predicted
ground motion parameters associated with t h e use of these relationships. As shown in
Table Vi1.3, t h e coefficients of variation (cov) f o r these relationships range from 55
t o 88 percent.

A comparison of McCuire's relationships for peak ground acceleration and velocity


versus measured d a t a form three well-documented and measured earthquakes is given
in Figure WI.2. D a t a from three events are shown:

a Parkfield 1970 earthquake (Magnitude 5.5)


a Lytle Creek 1970 earthquake (Magnitude 5.Q)
a Daly City 1957 earthquake (Magnitude 5.3)

vn-5
-.
L .c
~

S T D m A P I I P E T R O 1 1 7 - E N G L 1778 m 0 7 3 2 2 9 0 O b 0 2 3 3 3 47b m

McGuire's mean relationships (a and v) and plus one standard deviation relationship
(ala and via) a r e shown.

In general, this data indicates t h a t t h e McGuire relationships are tending t o over-


predict t h e peak ground motions at distances in excess of 50 to 100 kilometers. In
part, this tendency to over-predict at these g r e a t e r distances has been caused by t h e
large preponderance of records in t h e d a t a analysis from t h e San Fernando quake (dip-
slip, thrust fault source). This event released a large part of its energy dose t o t h e
earth's surface and energy decay with distance was not as rapid as t h a t associated
with t h e deeper focus, strike-slip quakes, which comprised t h e minority d a t a form
other events.

There are two major problems associated with t h e available measured data. First,
t h e r e are very few measurements of strong ground motions within t h e epicentral
region (within 10 km) of t h e epicenter of intense earthquakes. Thus, we know very
poorly t h e expected ground motions within such a region.

Secondly, t h e relationships are based primarily on recorded d a t a from intermediate


-
intensity events an average magnitude of 6.3. The application of t h e relationships
assumes t h a t t h e coefficients which determine t h e various ground motion parameters
do not vary with magnitude. Work by Trifunac and Brady (1975) indicates t h a t such
a n assumption f o r larger magnitude events may not be appropriate - "peak
accelerations of strong ground motion do not grow linearly with magnitude and this
rate of growth becomes very small for magnitudes greater than 6.5 to 7.0. For a
magnitude of 7.5 shock t h e peaks of near-field strong motion effectively reach t h e
maximum amplitudes."

Thus, f o r some portions of t h e ground motion characterizations which a r e attempted


here, t h e assumption of ground motion relationships which remain constant with
increasing magnitude may be conservative, certainly for t h e very large magnitude
events.

VII-6
-.
L .%

- .
~~~ ~

...
. ..- ......
. _... ,
:i..::..:.3
A..
:i

-. . I..
STD.API/PETRO 117-ENGL 1978 D 0732270 0 b 0 2 3 3 q 322

ViI.1.4 Random Quake Seismic Exposure

To this point, t h e building of a seismic exposure model for t h e study area has
proceeded through he following steps:

a Identification of sources - utilizing t h e geologic, tectonic, and seismic


history of the area, five major sources of future events were identified.

a Determination of t h e activity rates and frequency of occurrence of


various possible magnitude events for each of t h e sources has been
. ._. developed from historical seismicity data, geology, and analogy with
. .-
other similar areas. Ranges of these parameters were identified for
study to determine the influence of recognized deficiencies in the data.

a Selection of a source-to-site transmission model (attenuation law) which


would describe how selected strong ground motion parameters (peak
ground acceleration, velocity, displacement, and response spectra
velocities) would be changed with distance between t h e quake source
and a given site. These parameters would be taken as applicable t o the
surface of t h e strong soils (identified earlier as the Yakataga
Formation). Localized effects of overlying soft soils have not been
considered to this point; they will be in the section of Platform
Response. The very large uncertainties in the attenuation laws (50 t o
150 percent coefficients of variation) would be explicitly entered at this
point in the analysis.

These components have been used to build a seismic exposure analytical model for the
random quake effects characterization using the statistical framework developed by
Cornell (i968).

Figure VII.3 contains t h e results of t h e seismic modeling of the ground motion


parameters associated with t h e random quake (RQ) component. Figure VII.3 shows
t h e annual probabilities of experiencing t h e various magnitudes of expected maximum
ground acceleration (Ama), velocity (Vmax), and displacement (Dmax) at the
Pamplona Ridge site. I t is to be noted that these distributions include the attenuation
relationship error terms, and thus reflect both t h e natural or inherent variability and
that contributed by our uncertainty in predicting t h e ground mo tion parameters, given
this type of model.

VII-7
=
~

S T D - A P I I P E T R O 1 1 7 - E N G L 1778 0 7 3 2 2 9 0 Ob02335 2 b 7

The marked influence of the activity rates determined from the various periods of
historical d a t a is apparent. As one would suspect, including the multitude of events
from t h e Prince William Sound event (spread over the 11-year period of 1964-19741,
indicates the larger expected maximum ground motion parameters. Spreading this
and t h e earlier data over the 76-year time period of 1899-1974 produces the lower
-
expected ground motion parameters.

The truth of t h e matter for the next 25 to 50 years likely lies somewhere between
these two extremes. For t h e purposes of this report the ground motion
characterizations based on the 1940-1974 period will be used.

Figure VIL4 compares the annual probabilities of experiencing t h e various magnitudes


of expected Amax, Vmax, at the Pamplona Ridge and Kayak Island sites.
and Dmax
In all cases, t h e Kayak Island site is indicated t o have more intense ground motions at
a given probability level. This is primarily due t o its proximity t o the Prince William
Sound area, as compared t o the Pamplona Ridge site's proximity to t h e far less
active Malispina Block system.

For purposes of general reference, t h e random earthquake expected firm ground


motion parameters at t h e one percentile (equivalent to a s i t e average return period of
100 years) are summarized in Table VII.4. I t is interesting t o note that the largest
value of the expected maximum ground acceleration at t h e one percentile level (0.5
g's) is equal t o that proposed by Milne and Davenport (1969) for this same area and
probability level. It is substantially (1.5 to 2.5 times as large) greater than that found
by Donovan (19731, Kallberg and Cornell (19731, and Der Kiureghian and Ang (1975)
for comparable California sites (comparable site conditions and average return
period).

VII. 1.5 Non-Random Quake Seismic Exposure


As discussed previously, t h e seismic exposure (strong ground motions at a site)
characterizations have been divided into two major components - random and non-
random.

VII-8
~~

'.. ...
.~_,.__.:
_.. I
....
... .
. ,., .,. . * S T D - A P I I P E T R O 117-ENGL 1978 0732290 Ob0233b I T 5

In the non-random earthquake model proposed here, it is assumed that:

o The gap-filling earthquake will be in the Magnitude range of 7.5 to 8.5.

o I t is a certainty to occur during t h e first 25 years of platform life. The


risk or exposure during this time will be assumed t o be uniformly
distributed. While this assumption is not important t o this part of the
modeling, it will become important in the distribution and combination
of random and non-random exposures.

o The energy of t h e event will be uniformly distributed over t h e area of


t h e Malispina Block and concentrated at a depth of 15 kilometers below
t h e surface.

o The source-to-site transmission relationship (McGuire's) will be


adequate to describe t h e effects of distance between the sites and t h e
multiple sources in t h e Malispina Block.

The basic elements of this model are in substantial agreement with those proposed by
Kelleher, et aí (19731, Sykes (1971), and Page (1975). The physical characteristics of
area, volume of strained rock and depth of focus a r e reasonable representations of
similar past large magnitude events, such as the Chilean earthquake (M = 8.5) of May
22, 1960 (Kanamori and Cipar, 1974).

The results of t h e non-random earthquake model a r e given in Figure VII.5. This figure
presents the expected firm ground A max, V max and D max at various levels of
probability, given that the llgap-filling'l event has occurred. Results are shown for the
Pamplona Ridge (PR)and Kayak Island (KI) sites.

The probability levels are developed from the uncertainties in t h e attenuation


relationships (error terms) and t h e uncertainties associated with the possible range in
. magnitude of t h e event (range of 7.5 to 8.5).

Due to t h e proximity of t h e Pamplona Ridge site to t h e source of t h e non-random


event, t h e magnitude of ground motion parameters a r e substantially larger than those
for the Kayak Island site. The 50th percentile expected firm ground motion
parameters a r e summarized in Table VII.5.

. .-.
.
VII-9
- .
...
=
~

STD.API/PETRO 1 1 7 - E N G L 1778 9 0 7 3 2 2 7 0 Ob02337 O31

VIL2 FIRM GROUND MOTION CHARACTERISTICS

VII.2. I Time Histories

To this point, t h e seismic exposure analyses have developed information on firm


ground peak accelerations, velocities, and displacements for two types of exposure
(random and non-random) for various probability levels. The next step in t h e analysis
requires t h a t these values be translated into three-dimensional ground motion t i m e
histories.

In this study, four recorded (on firm ground) earthquake acceleration-time histories
were chosen t o form t h e basic input to t h e platform models. The four records used
are:

o 1952 Kern County Quake, Taft Lincoln School Tunnel Site (Taft 52)
e 1940 Imperial Valley Quake, El C e n t r o Site (El C e n t r o 40)
o i949 Western Washington Quake, Olympia Site (Olympia 49)
o i968 Borrego Mountain Quake, El Centro Site (Borrego Mountain 68)

The digitized accelerograms were obtained from t h e California Institute of


Technology Earthquake Engineering Research Laboratory (197 1).

These four records were chosen to determine t h e influence on platform response of a


wide range of possible ground motion frequency contents, amplitude variations, and
amplitude sequencing.

Figures VIL6 and VII.7 contain t h e three-dimensional acceleration, velocity, and


displacement t i m e histories for two of t h e recordings - Taft 52 and Borrego Mountain
68. The Taft record might be taken as representative of t h e characteristics of a
nearby event; t h e Borrego Mountain record might be taken as representative of t h e
characteristics of a distant event in which quake energy was released close t o t h e
earth's surface and in which t h e r e is much energy in the low frequency range.

The extremely complex nature of the ground motions is immediately obvious. I t is


clear why a single ground motion parameter (e+, peak ground acceleration) can not
describe t h e essential features of t h e ground motions.

VII-1 o
-.
.L .c
STD.API/PETRO 1 1 9 - E N G L 1978 0 7 3 2 2 9 0 ü b 0 2 3 3 8 T78

The intensity of the ground motions was varied by scaling t h e peak recorded
accelerations to various levels, scaling ail of t h e other acceleration amplitudes by t h e
same factor, and keeping ail other parts of t h e record (fréquency content) intact.

It will be seen subsequently, t h a t t h e peak ground acceleration is a poor measure of


t h e intensity of effects on platform system response, peak ground velocity being a f a r
superior measure.

As will be discussed in Section XI, much insight into the dynamic response of complex
systems can be developed by study of t h e response of simple single degree of freedom
(SDOF) systems. The closer t h e response of the complex system comes to being
dominated by the first mode response, then t h e more direct is t h e application of t h e
SDOF system results.

As mentioned earlier, t h e seismic exposure modeling included t h e development of


response spectra information (spectral velocities, Table VII.3). Presentation of t h e
results for the random earthquake model a r e given in Figure VII.8.

The expected annual maximum spectral response velocities for iû-percent of critical
damping, for three SDOF system periods ( i , 3 and 8 seconds), and t h e two study sites
a r e given for various probability levels.

An alternative presentation of these results is given in Figure VII.9 for the random
earthquake exposure. A t the 99th-percentile, and periods in t h e range of 1 to 6
seconds, spectral velocities of 20 to 30 inches per second a r e indicated.

Comparable information for t h e non-random earthquake exposure is given in Figure


VII. 10. Note the much narrower plateaus (maximum spectral velocities experienced
over a smaller range of periods) of these spectra. 50th-percentile spectral velocities
a r e approximately 15 inches per second in t h e period range of 1 to 4 seconds.

VIL3 SOFT GROUND MOTION CHARACTERIZATION


At this point, one might ask how likely it is for a platform located on soil B to
experience grcund motions which a r e characteristic of those recorded at firm ground
sites. A t present, there a r e several possible answers t o such a question:

VII-11
1. If the pile foundation receives a large part of the input energy from t h e
soils close t o t h e firm ground interface at 100 f t (Penzien, 19701, then
t h e firm ground records may be a correct characterization.

2. If t h e recorded ground motions a r e typical of those of one class of


earthquakes - events in which a preponderance of t h e energy is released
close t o t h e ground surface and which have propagated over significant
distances to the recording site primarily in t h e form of surface waves
(Hasegawa, 1974; Trifunac, 1971; Duke and Mal, 1975; Swanger and
b o r e , 1974), then t h e motions of the upper soft sediment may be very
similar t o those associated with t h e firm ground interface. The El
Centro (1968 Borrego Mountain event) record was chosen in an attempt
t o recognize the unique characteristics of such an event.

3. If the recorded ground motions a r e representative of those expected at


t h e firm ground interface as developed by earthquake energy
transmitted t o the site primarily in the form of vertically propagating
shear waves, then the motions of the upper sediments might best be
characterized by results form shear-beam analyses which recognize t h e
nonlinear, hysteretic, strain and cyclic sensitive response of the soft
soils (Idriss, et al, 1975; Idriss, et al, 1976; Schnabel, et ai, 1972;
Streeter, et al, 1974; Joyner and Chen, 1975).

Platform response results for the third possibility a r e shown in Table VII.6. The upper
100 f t of soft soils were modeled as a soft clay having strength and modulus
properties typical of soft clays found in the Gulf of Alaska (Idriss, et al, 1975; Seed
and Idriss, 1970). The Taft (1952) record was scaled to three different intensities
(0.5, 0.33, and 0.10 g's) and input at t h e base of the soft soil column. The lateral
motions of t h e soil at the -45 f t elevation were used t o characterize input motions to
t h e piles. This location is t h e approximate center of lateral soil bearing pressure on
the piles.

The ground motion results indicate t h a t t h e Amax's have been severely attenuated,
t h e Vmars apprbximateìy preserved, and the Dmax,s about doubled over those input
at t h e underlying firm ground level. Again, the platform response quantities
........-.. ...
-
:/. I
STD.API/PETRO 1 1 7 - E N G L 1778 0732270 Ob02340 b 2 b 9

correlate well with the value of maximum ground velocity. These points a r e plotted
as open circles on Figure MI.11. .While not detailed here, t h e input of other firm
ground time histories at t h e base of t h e soft soil column develóps similar platform
response correlations with Vmax.

The last case in Table WI.6 typifies the potential effects on platform response of an
effective combination of two elevation levels of energy input to the platform piles -
vertical excitation coming in t h e lower portions of t h e piles and lateral excitation
coming in t h e upper portions close to t h e seafloor. The vertical motions input to the
platform response model were assumed to be those of the input time history at the
base of t h e soft soil column while those of t h e lateral motions were those determined
as in Case 9. Significant increases (50 to 100 percent) in all response quantities a r e
developed. Thus, t h e realistic characterization of vertical excitation is as important
as t h e lateral excitation.

By this point, it should be apparent that t h e response of the platform is determined In


a very major way by t h e manner in which quake energy is transferred to the piles, and
t h e effective capacity, stiffness, and energy dissipation characteristics of t h e piles.
The proper modeling of this very complex soil-pile-structure interaction problem has
been only very crudely attempted here (superstructure supported by linear coupled
elastic springs). This complex analysis problem warrants much additional attention t o
develop t h e tools needed by t h e design engineer.

VII-13

- .
- .
~ ~
~
~

S T D - A P I / P E T R O 117-ENGL 1978 0732290 Ob02341 5b2

TABLE VII.1
SEISMICITY PARAMETERS 1899- 1974

NO. MAGNITUDE NO. COMBINED


QUAKES OF MAX QUAKE QUAKES b
FAULT SYSTEM M 2 3.5 M PER YEAR

Pamplona Ridge 18 6 0.24 -0.49

Kayak Island 96 5.5 1.26 -0.47


Chugach 86 8.5 1.13 -0.37

Shelf Edge 11 7.5 O. 15 -0.26

Shelf Area 18 6.5 O. 24 -0.42

TABLE VIL2
EARTHQUAKE OCCURRENCE RATES

AVERAGE NUMBER OF QUAKES PER YEAR


(M 2 3.5)
FAULT SYSTEM 1899-74 1940-74 I 1964-74 1965-74
Pamplona Ridge 0.24 0.43 1.36 1.20
Kayak Island 1.26 2.70 8.54 4.40
Chugach 1.13 2.34 7.45 3.20
Shelf Edge O. 15 0.29 0.91 0.30
Shelf Area 0.24 0.49 1.45 0.70
TABLE VII.3
TRANSMISSION LAW TERMS

PARAMETER cl c2 c3 C.O.V.

PEAK GROUND
ACCELERATION2 -4.72 0.54 1.30 0.51 0.55
( A max, cm/sec

PEAK GROUND
VELOCITY 5.64 0.92 1.20 0.63 0.70
( V rnax, cm/sec)

PEAK GROUND
DISPLACEMENT 0.39 1 .o 0.88 0.75 0.88
(D max, cm)

(DAMPING = 10%)
PERIOD = 0.5 sec 4.80 O. 364 1.236 0.24 0.60

= 1.0 sec 0.431 0.422 O. 836 0.28 0.71

= 4 . 0 sec O. 092 0.518 O . 885 0.40 1.15

= 8 . 0 sec O, 1318 O. 449 0.831 0.35 0.97

. .,'
~

S T D . A P I / P E T R O 1 1 7 - E N G L 1778 9 0 7 3 2 2 7 0 0b023Li3 335

TABLE V11.4: RANDOM EARTHQUAKE EXPECTED GROUND


MOTION PARAMETERS FOR AVERAGE SITE
RETURN PERIOD O F 100 YEARS

A MAX2 V MAX D MAX


CM/SEC CMISEC CM
or or Of
SITE (g's 1 (IN/SEC -
(IN)

Pamplona Ridge 280 27 23


(0.3) (10.6) ( 91

Kayak Island 500 45 33


(0.5) (17.7) (13)

TABLE VII.5: NON-RANDOM EARTHQUAKE EXPECTED FIRM


GROUND MOTION PARAMETERS FOR THE
50TH PERCENTILE

A MAX2 V MAX D MAX


CMISEC CM/SEC CM
or or or
SITE (g's) (IN/SEC 1 (IN)
Pamplona Ridge 250 42 25
(O. 25) (17) ( 10)

Kayak Island 72 15 12
(0.07) (6) (5)
I .

Cu
vi u
. . m. .
vi N
vi h
Co
h m
. .
O0 h
cr\
m
h
h
.
vi
Co
m
.
u
25-

h O m*
vi
O m
3
. d

N
. h
ch
3
O0
d

O
. m
h
h
*.
d
vi
\o
h \o 0:
i(
m Cu m 00 \o N h
4

d'

2
.
vi O

2
.
O0

c1
3.
rrl
N
h
.
(u
Cu
v\
d

0
O
N
(v
Q\
m
d
h
n
3
ö
m

E
o, 6/ O0 m N
. . \o
h
0 d

0
h
In
. 0
d
0
h
0
d
. O
d

2 O0
4 2 m u\
N u m - \o

\o
N
m
vi
N
v\
.
ril

\o
=l.
m
m
0
. . h
O
*
h
m
m m N a- 4

m
m
O
. m
m
O
Vl
m
O
m
m
O
O
vi
O
O
d

O
. O
d

O
h
4

O
.

..

N
ril
S T D - A P I I P E T R O 119-ENGL I978 0732290 Ob02345 L O 8

ALASKA I

. .

/SEA FLOOR

FIG, VII, 1 ' G O A STUDY AREA, SITES, AND SEISMIC MODEL


..............,*.
. -.
_.
. < . . .

.~.., STD.API/PETRO 117-ENGL 1 9 7 8 1 0 7 3 2 2 9 0 Ob0234b O 4 4 m

a
u
W
u W
O
-I J-rr
w t-
>
n
z
a
a
t-
a
Ca
ci
W
cl
o:
O
u
W
nz

I I I 1
- - .. . .
-
I
I-

--l
1 O 3
lx
(3 v)
o.
‘t ci.
a I
Lu
n

z
X

I -

-
c1
U

>
. Q -
(3
m
u,

Q
I I
-01 1 I u O
9
o

-.
L- .c WOODWARD-CLYDE CONSULTANTS
. .
Y
. :--..
.. .-.. . -.
..

--_.
.~ WOODWARD-CLYDE CONSULTANTS
-.
-Li -< WOODWARD-CLYDE CONSULTANTS
-
*
-
-
~

.-
STD.API/PETRO 1 1 7 - E N G L 1778 0 7 3 2 2 7 0 Ob02344 853

I"" I ' 1 1 1

I I I I I I I I I I i I I I !

-
I

8- 35 2 rn

Vi 3 'xvnû ' l N 3 N 3 3 V l d S I O O N i l O ü 3 WflWIXVW 0 3 1 3 3 d X l

i 3 3 S / . W 3 ' x v w V ' N O I l V ä 3 1 3 3 3 V ONnOü3 H í ì W i X V W 031336x3


WOODWARD-CLYDE CONSULTANTS -
-.
.z
... . ,. . .. ..
.. . . . . .
. ..-,x:
~

~ .I..,
,<"i,
S T D - A P I / P E T R O 117-ENGL 1978 m 0732290 ü b 0 2 3 5 0 575

SAW 52.002 TACT LINCOLN smoa NNMLL coyc s69E


. O.mU vuucs: YCLL - 1 7 3 s cY/s€vscc VLroCtrr h l 7 7 9 i / S E C DISC * - 9 ? CM
2 -2Sr
0

, ....
. .
.. .. . .
I...

20' I
-10-
2 Y
w
5 I
- .
=;d e o
P
. .
: I

20 30 40
t

50
I
U

t-
'O0 IO
.TIWE - SECONDS z
O
U
11AOO4 52.002 TAFT LINCOLN SCHOOL TUNNEL UiMP Y 2 l C
QPCIN VhLUES: ICCLL -Is.?CU6ECISEC vttoclTT-tl.7 C W S E C OISP .-CU
I.?
-2SOr

;U
* 2 I 1 1 I I I I . 1

w
Y

i-I
U

TINE - SECONDS I-
IL
a
l l A O O 4 52.002 T A F T LINCOLN SCHOOL N N N E L COMP VERT c
DCCh.* VALUES: LCCCL*OZ.9 CUILCISEC VCLOCITT"c.7 C W Y C OISC'-50 CY
-250r

-8r

-i' ...

-. iL- -< WOODWARD-CLYDE CONSULTANTS


S T D - A P I / P E T R O 1 1 7 - E N G L 1778 m 0 7 3 2 2 7 0 O b 0 2 3 5 1 401 m

I-
œ
L a
- u
zo :. CI w
as
>
Y U
O 4 n
W

2d ' I ' 0 8 . I t I I I , , , , , ,
O o 20 Y)
TIME
40
- 5a
SECONOS
60 m m 90 O
E
O
116019 ao05 EL CENTRO Sm I U P V A W 3 IRR OIST COYP VERT. c3
O-CCA. VALUCS: ACCEL -29 7 CWSECISCC VCLDCITT). 3 CrrSCC DISC-3 S CY
w
a
nr:
O
m

-4 P

4r R

A _ .t
WOOD WARD-CLY DE CONSULTANTS
LI-
Z

m a
>ar
WCC
W O
zu,
O
CLm
W W
wI-
CC-
v)
-!
an
œ z
I-a
o-!
Wv)
Q u
I 1 1 . 1 I I I I
cn

v)

--.-. WOODWARD-CLYDE CONSULTANTS


= WhE 25E2090 Ob22ELO W L b T 13N3-bTT O i I 1 3 d / I d V ' Q I S ..-
. ... .
. .
-. :'I
S T D - A P I / P E T R O 117-ENGL I978 W U 7 3 2 2 9 0 ü b 0 2 3 5 3 2 8 4 m

OS 1.0 IO
PERIOD -SECONDS

.-

FIG, VII, 9 RESPONSE SPECTRA FOR RANDOM EVENTS


WO 0 0 WAR D-CLYDE CONSU LIANTS
...

. .
~.
i .
-.. .
.... -.. ...
....
.. ..
..
..... .

0.1 1.0 IO
-
PERIOD SECONDS
(0)

, .....

.... ...-

0.1 1.0 IO
PERIOD -SCONOS
(b1

-. WOODWARD-CLYDE CONSULTANTS ~ ~~
-
. _ S T D - A P I / P E T R O 117-ENGL 1978 0 7 3 2 2 7 0 Ob02355 O57
- . ....

..

fn 80- 1 I I 1 1
0
Y
9

s
0 7 0

I L I M I T LOADS D ~ R M I N E D ie-_ .

w
u3

O 20 4 0 6 0 80 100 120
MAXIMUM GROUND VELOCITY- V m - c m / s e c

FIG, VII, CORRELATION OF W I M ü M GROUND VELOCITY


W I T H PLATFORM RESPONSE

- .
~

S T D - A P I / P E T R O 117-ENGL 1978 W 0 7 3 2 2 9 0 Ob0235b T93

VIL PLATFORM RESPONSE TO EARTHQUAKES

VIII.l INDUCED FORCES DUE TO EARTHQUAKES CONTRASED WITH


IMPOSED FORCES DUE TO OCEANOGRAPHIC CONDITIONS

Wave loadings generate forces within t h e platform which a r e relatively independent


of t h e stiffness, strength, and substructure (foundation and soils) characteristics of
t h e platform. As shown in the upper portion of Figure VIII.l, as t h e wave height (H)
increases, t h e r e is an increase in t h e forces generated within t h e platform, indicated
here by t h e maximum resultant base shear (BS max). This type of environmental
condition (WWC) develops what is essentially a n untruncated force system; as wave
height increases, t h e forces increase without limit until platform collapse is
experienced.

As shown in t h e lower portion of Figure VIII.1, t h e same f o r c e system is not developed


by earthquake conditions. As t h e intensity of t h e ground motions increases (expressed
in t h e figure as ground displacement, AG), t h e reactive forces transmitted from t h e
superstructure (BS max) increase t o a point at which the substructure (foundation
elements and soils) will no longer transmit increasing forces (RU 1.
F .
Thus, t h e reactive forces developed within t h e superstructure a r e limited by t h e
ability of t h e substructure system t o transmit forces to t h e superstructure. This type
of environmental condition (RQ or NRQ) may develop what is essentially a truncated
force system, truncated by the ultimate strength of t h e substructure system. This
truncated force limit is determined by t h e loadings used t o design t h e foundation
system (R ),
DF
For t h e study platform system, t h e truncated force limit for t h e foundation system
was alluded t o in t h e section on t h e study platform; refer t o Figure IV.5. For t h e
system designed for a t o t a l base shear of 25,000 kips, t h e ultimate strength of t h e l'A''
foundation (piles and conductors)-soil system was 70,000 kips (including b a t t e r
effects). The ultimate strength of t h e "8: foundation-soil system was 40,000 kips.
These relationships between design base shear (RD) and ultimate strength base shear
(RU) will be used in t h e subsequent sections as follows:

VIII-1
- .
~

S T D - A P I I P E T R O 119-ENGL 1978 m 0732290 üb02357 92T m

o Foundation-Soil System "A" %/R = 2.8


o Foundation-Soil System "BI' R d R D= 1.6

The importance of t h e ability of t h e substructure system t o limit t h e loads


transmitted t o t h e superstructure will be entered in this analysis by truncating
earthquake loads which exceed the ultimate strength of the foundation, as determined
by a given design load, R D' As shown in Figure vIII.2, this will have t h e e f f e c t of
dramatically reducing t h e probability of failure (roughly proportional to t h e area of
overlap of t h e loading and resist-ce distributions) for cases where t h e r e is a
significant probability of experiencing loads in t h e range of those used t o design t h e
substructure system.

A potentially important design strategy point is raised in this discussion. The


platform superstructure and substructure systems must be made sufficiently strong t o
resist t h e anticipated wind, wave, and current loadings. Given t h e platform in an
earthquake environment, excessive strengthening and stiffening of t h e substructure
system beyond t h a t required f o r the W W C environment may be detrimental.

In addition, t h e foundation elements must be designed so t h a t they will be able to


develop t h e large ductilities required during intense earthquake excitations without
significant losses in capacity.

Previous study results (Bea, 1978) indicated t h a t t h e important e f f e c t s of force


truncation were for t h e Pamplona Ridge, non-random quake reactive force condition.
Figure VI11.3 shows t h e truncated expected maximum resultant base shear force
characterization for t h e study platform at t h e Pamplona Ridge site. The truncated
limits have been shown f o r the design load (RD)of 25,000 kips. The large amount of
loading distribution which is deleted is obviously significant.

Two important points are observed here. The first is t h a t t h e truncation limits must
vary with t h e design loads t o be studied. The second is t h a t t h e a r e a under t h e
loading probability distribution must be equal t o one. This requires re-scaling t h e
probabilities in t h e truncated distribution. Both of these points were recognized in
implementing t h e probabilistic algorithms used to compute t h e platform reliabilities
discussed in t h e section on platform reliability.

VIII-2

- .
S T D - A P I / P E T R O LLS-ENGL 1778 0732290 Ob02358 Bbb

VIII.2 FIRM GROUND MOTION EARTHQUAKE FORCES

In this study, t h e maximum, total, resultant base shear (BS max) acting on the
platform has been used t o characterize t h e effects of ground motions on the platform
system as a whole. As in attempting to characterize ground motions with only one or
two parameters, t h e use of base shear fails to reflect many of the important details
of response. However, these analyses and similar past analyses associated with wave
loadings indicate base shear to be a n acceptable characterization of the gross effects
of ground motions on t h e structure and foundation system.

Plots of the variation of base shears (BSX and BSY refer t o X and Y components) and
overturning moments at the mudline (BMX and BMY) during t h e first 15 seconds of a
scaled El Centro excitation are given in Figure VIII.4a. The baseline corrected
digitized record stronger component was scaled to a peak acceleration of 0.5 g.
Other components and amplitudes were scaled proportionally.

In all cases to be discussed in this section, analyses were performed for t h e first $0
seconds of t h e input ground motion time histories. Without exception, peak response
(measured in terms of total base shears and moments) occurred during t h e first 15
seconds of ground motions.

The BSY peak occurs at about 5.5 sec. and has a magnitude of 21,100 kips. The BSX
peak occurs at about 10 sec. and has a magnitude of 17,200 kips. The peak resultant
base shear of 26,000 kips is developed at t h e same time as BSY.

A similar, but f a r less irregular variation of BMX and BMY with time is indicated.
Note t h e potentially important influence of a slight change in t h e phasing of these
responses. In t h e case of t h e overturning moments, changing t h e time of occurrence
of t h e peak X and Y responses which occur at about 6.5 sec. by 0.2 sec. would
increase t h e resultant response $y i 0 percent. A similar, although somewhat smaller,
change is seen possible in t h e resultant base shears.

Plotting or visualizing t h e time histories of the resultant responses indicates t h a t the


platform is obviously undergoing a complex rotary motion, with peak resultant
motions developing in t h e diagonal directions. This three-dimensionai e f f e c t was
found t o be of major importance for ail of t h e 12 ground motion time histories
studied. "Turning off" one or t h e other of the input- horizontal ground motions could
influence peak response by 30 to 50 percent.

VIII-3
-
.
.
~

S T D - A P I / P E T R O LLS-ENGL 1978 D 0 7 3 2 2 9 0 Ob02359 7 T 2 D

For purposes of comparison, plots of t h e variation of BS and BM during t h e first 15


seconds of a scaled Taft (Kern County 1952) three-component ground motion is given
in Figure VIII.4b. As before, t h e stronger component of- t h e record was scaled to a
peak acceleration of 0.5 g. Other components and amplitudes were scaled
proportionally.

The BS and BM peaks are experienced early in the ground motion time history. The
peak resultant BS is 20,800 kips and occurs at 5.0 seconds. The peak resultant BM is
66.9 x 106 in.-kips and occurs at 6.0 seconds.

Recalling t h a t the platform was designed t o perform "elastically" f o r a peak resultant


BS of 25,000 kips, these responses a r e fully within the regime of. elastic response.
However, inelastic response in t h e soft clays of Soil B have been recognized in
developing t h e input motion t i m e histories. These effects will be discussed further in
a later p a r t of this section.

During t h e course of t h e work on t h e study platform, some 100 cases of dynamic


platform responses were developed. The primary variables were as follows:

o Firm Ground Motions (4 primary used, 4 others studied)


Soft Ground Motions (2 primary used, 2 others studied)
o intensity (3 levels)
o Damping (4 vaiues)
o Axial Pile Stiffness (2 primary, 2 others studied)
a Lateral Pile Stiffness (2 primary, 2 others studied)
a Deck Loading (1 primary, 2 others studied)

Figure VIL5 shows t h e influence of pile lateral stiffness on maximum resultant base
shear (BS max). Ail of t h e input records have been scaled to a peak ground
acceleration of 0.33 g (g = acceleration due t o gravity).

Pile lateral stiffness may be important. For example, f o r t h e Borrego Mountain


excitation, t h e BS max is increased from 31,000 t o 38,000 kips by increasing t h e
lateral pile stiffness from 85 kips/in. (soil-pile "B") t o 315 kips/in. (soil-pile )'Att).

vm-4
~ ~

..--.
.
,
..-
_ S T D * A P I / P E T R O 1 1 7 - E N G L 1978 H 0732290 Ob023b0 4 3 4

Conversely, for other inputs, pile lateral stiffness may be unimportant; for example,
t h e response results f o r t h e El Centro and Olympia excitations.

Of primary significance is t h e difference in response due to t h e different character of


t h e input excitations. There is a factor of three difference in BS max between t h e
platform's response to t h e Olympia event and t h a t of the Borrego Mountain event.

Similar conclusions are indicated f o r t h e influence of pile axial stiffness as shown in


Figure WII.6. However, two factors may significantly amplify t h e influence of axial
stiffness - eccentricity of deck loadings or a marked change in t h e relationship
between horizontal and vertical excitations (proportionally more of t h e earthquake's
total energy being concentrated in the vertical direction).

In development of these results, t h e deck loadings have been assumed to be uniformly


distributed over t h e deck area. In t h e following section on platform response to soft
ground motions, t h e influence of changing the relationship between horizontal and
vertical excitations will be recognized.

Figure VIII.7 summarizes t h e results of t h e influence of t h e damping assumed for t h e


system. PS-A refers t o t h e pile-soil "Af1system and PS-B t o the pile-soil 'vB1l system
(axial and lateral stiffness). At damping. ratios in excess of 2 t o 3 percent, t h e r e is a
relatively minor influence on BS max of changing damping. Changing t h e damping
ratio from 4 to 10 percent, decreases the BS max by about 20 percent.

VIII.3 SOFT GROUND MOTION EARTHQUAKE FORCES

Section VII.3 has explored t h e characterization of soft ground motions as might be


expected for soil profile B. Some of t h e information presented here will be a
reiteration of t h a t previously described.

The soil profile B dynamic analysis was performed using t h e DCHARM computer code
(Idriss, et al, 1976). This code fa derivative of CHARSOIL (Streeter, et al 1974) is
able to analyze t h e influence of strain-cycling caused degradations in t h e soil's stress-
strain properties using a Ramberg-&good model for soil behavior. For weak ground
motions, such a n analyses appears well confirmed by measurements (Joiner, et al
(1976).

VIII-5
=
~~ ~ ~ ~~

S T D . A P I / P E T R O LLS-ENGL 19'78 0 7 3 2 2 9 0 Ob023bL 3 5 0

The components were scaled t o produce t h e effects of a range of excitation


intensities. Peak input accelerations of up to 0.5 g and peak velocities of up to 50
cm/sec were input. For t h e range of excitations, t h e soil maximum strains remained
below 5 percent, well below failure strains or strains large enough when combined
with t h e number of cycles to result in significant cumulative deformations.

Figure VIL8 shows t h e response spectra developed by t h e DCHARM computer code


for soil profile IiB1' at the mudline, -45 f t , and -100 f t (level of input motions). The
spectra a r e for t h e El C e n t r o 40 (North-South component) record, scaled t o a peak
acceleration of 0-5 g.

Note t h a t in t h e range of t h e fundamental lateral response periods of the soil (2 t o 3


seconds) and platform (1.8 t o 2.5 seconds) t h e r e is a significantly amplified response.
at t h e mudline, In contrast, t h e r e is a de-amplified response at the -45 elevation.
Correct characterization of t h e effective elevation of input motions is obviously very
imp0 r t an t.

Shown in Figure VIII.9 is a comparison of the input (at -100 f t ) maximum velocity and
t h a t output (at -45 f t ) for t h e six records (two components of t h r e e events) scaled to
various intensity levels. There is a reasonably good clustering of t h e d a t a points
about t h e line of equality.

The acceleration-time histories developed by t h e DCHARM analysis at t h e -45 f t


elevation (modified motions) were input to t h e base of t h e platform analytical model
as previously described,

A plot of t h e maximum resultant base shears for the modified input motions versus
t h e peak ground velocities associated with these input motions is given in Figure
The modified component (two lateral components modified, and vertical
VIII. 10.
component preserved) results are indicated as i'modulated.tl The unmodified as
recorded (but scaled) input motion results are indicated as

A striking and very important correlation between peak ground velocity (input t o t h e
base of the superstructure model) and maximum resultant base shear is developed.
Similar a t t e m p t s to develop such a correlation for peak ground accele.rations or
displacements failed.

VIII-6
. .
_ . .
....
.....
.. .... . :..'..
. ... . .1
A....
S T D - A P I / P E T R O 117-ENGL 1978 O732290 Ob023b2 297 m

A similar correlation between peak ground velocity and maximum resultant


overturning moment is shown in Figure VIII.11. The data from which these plots were
prepared are tabulated for ease of reference in Table VIII;I.

VI11.4 ELASTIC RESPONSE SUMMARY

Shown in Figures VIII.12a and b is a plot of the input peak ground velocity (input at
the base of t h e spring supported superstructure analytical model) versus t h e range of
maximum resultant base shear and overturning moment for cases evaluated.

The data includes a wide variety of assumptions concerning:

e Axial pile stiffness (4,000 to 8,000 kips per inch)


o Lateral pile stiffness (85 to 315 kips per inch)
o Points of effective input of excitations (mudline, -45 f t , vertical at
mudline and -100 f t )
e Influences of soft soils (firm ground with no modification and soft
ground with no modifications for surface wave conditions or shear wave
modifications for body wave conditions.
e Frequency content, amplitudes, and amplitude sequencing of ground
excitations (Taft, El Centro, Borrego Mountain and Olympia)
o Damping (4 to 10 percent of critical)
e Deck loads (12,000 t o 35,000 kips)

The results indicate that as long as the gross response correlation is based on a peak
ground velocity, that all of the foregoing factors have a minor influence on t h e
response.

This is not to imply t h a t t h e range in assumptions will not produce important effects
in some parts of t h e structure (e+, certain brace loadings, pile loadings, etc.); only,
that as long as t h e focus is kept on the overall response characteristics of t h e
platform (maximum resultant base shear and overturning moment), they have a minor
influence. The factor of primary importance is t h e peak ground velocity (or some
similar alternative measure, e+, RMS ground velocity).

It is important to remember that all of the discussion to this point has dealt
fundamentally with t h e elastic response of t h e platform system (only t h e soft soils
were analyzed recognizing significant inelas tic action). The following section will
discuss an extremely important influence on response developed by inelastic action in
the substructure (fwndation elements and soils).
VIII-7
VIII.5 INELASTIC RESPONSE AND DAMAGE STATES (STATIC ANALYSES)
Here, as in t h e foregoing section, we will direct- attention to development of an
understanding of gross measures of response (base shears and moments,
displacements) of offshore platforms subjected t o intense ground motions

Examination of t h e published literature indicates that the current State of Practice


of detailed analyses of earthquake response is largely limited to elastic methods
(Penzien and Kaul, 1972; Penzien, 1976; Kirkley, 1975; Hasselman, et al, 1975) While
much background has and is being developed on response of yielding (Veletsos and
Newmark, 1960; Veletsos, 1969), interacting (Veletsos and Verbic, 1973; Veletsos,
1975), and degrading (Shibata and Sozen, 1977; Gates, 19771, simple systems, i t is only
very recently t h a t the analytical methods, and equally important, data on element
(joint, brace, pile, soil) performance have begun t o become available t o implement
detailed, dynamic, inelastic analyses (Idriss, et al, 1976; Mondkar and Powell, ,1975;
Gates, et al, 1977).

Early results from these currently implemented inelastic analyses (Gates, et ai, 1977;
Marshall, et al, 19771, static-inelastic-to-collapse analyses (Bea, 1977; Gates, et al,
1977; Kallaby and Millman, 1975; ATC, 1977), and knowledge of the response of SDOF
yielding, interacting, and degrading systems provides our best glimpse of inelastic
performance during intense ground motions.

The focus in this section will be on response resulting from three-dimensional, static,
inelastic analyses of platform systems. In such analyses, t h e pattern of inertial
forces in t h e system a r e determined from results of elastic, dynamic, analyses (e+,
the pattern of inertial forces which produces peak BS or BM). This pattern of force is
progressively and proportionally scaled up. The static, inelastic response of t h e
system is based on the inelastic performance descriptions for the elements, and
recognizes t h e progressive changes in stiffness and development of load
redistributions.

Such an analysis fails t o reflect t h e important changes in magnitude and distributions


of the induced inertial forces caused by significant changes in periods, modes of
response, and energy dissipation. The indicated response and perf ormance a r e simiiar
t o that developed by a single dominant lurch of the ground (a single impulse).

VI II- 8
-
i_

...-
.. -.:
....
. ......., :-
.. ..
._...
~
..
-.. .
. ..;.-.-

Initial results from much more advanced'and realistic time-domain, two-dimensional,


inelastic analyses (Gates, et al, 1977; Marshall, et al, 1977) indicate t h a t t h e static
inelastic analysis does give us some valuable clues ' t o t h e platform's expected
inelastic behavior and performance. - -

Shown in Figure VIII.13 are results of a static-inelastic-to-collapse analysis of t h e 12-


leg platform system previously discussed. The analysis has attempted to t a k e into
account elastic-plastic response of tubular members, piles, and joints. Potential
cyclic degradation has not been taken i n t o account.

refers to t h e elastic API (strength level) resultant base shear used


In this figure, llRdl
to design t h e platform system; "Ry" refers t o the total lateral load at which first
significant yielding of t h e system is experienced; "Ru1' refers t o t h e ultimate lateral
load developed on t h e platform at t h e time a collapse mechanism (very large
permanent displacements or marked reductions in load-carrying capacity) is formed.

Ru is a f a c t o r of two times Rd; Ry is a factor of 1.6 times Rd. The ductility, defined
as t h e ratio of displacement at Ru t o t h a t at Ry, is approximately 1.7.

Shown f o r reference are t h e ATC (1977) and UBC (1973) elastic design shears for both
moment resisting frames (MF) and braced frames (BF). The fundamental period and
total dynamic weight (including entrained and parasite water weights) of t h e system
has been used in t h e computations of design shears. The ATC and UBC elastic level,
design shears are 10 t o 35 percent of those of API. Further explanation of these
dramatic differences will be attempted at the conclusion of this section.
. .

For this particular platform system, t h e separation of yield and ultimate level base
shears and displacements from those of t h e design level are due to a combination of
t h e foilowing factors:

o -
Redundancy During load increments developed during t h e transition
from R d to Ry, four braces in two external trusses are buckled. The
slight reduction in stiffness is due t o a transfer of t h e loads from these
braces and trusses t o t h e much stiffer and stronger internal trusses.
Further buckling of members (Ry to Ru) is reflected as a dramatic

VIII-9
STD.API/PETRO 117-ENGL 1978 D 0732270 O b 0 2 3 b 5 TTb W

decrease in stiffness. Sixteen members have buckled by t h e time Ru is


developed. Collapse, as previously defined, is developed by t h e
formation of a mechanism in the truss legs. -

o -
Element Performance Factors-of-safety are in tentionaily incorporated
into the algorithms used t o design t h e braces, legs, joints, and piles.
Coupled with other often unrecognized factors-of-safety (e+, nominal
versus expected t r u e yield strength, a 20% factor), t h e margin of safety
between t h a t of design and first substantial yield is large.

o Designer's Prerogative - Extra strength is often added to the platform


system by t h e designer. This can be in the form of brace configurations
and bracing schemes, and/or t h e e x t r a sizing of members and braces
when numbers close t o ailowables a r e indicated. This factor can have
very marked effects on both strength and ductility of the overall
system.

o Other Design Loadings - It


is t h e overall system of design loads t h a t
determines strength and ductility. Operational loadings (e+, due to
deck equipment, launching, fabrication) and other environmental loading
conditions (e+, those due t o storm winds, waves, and currents) cause a
distribution of strength within t h e platform. In this case, a 100 f t
design wave height, heavy deck loading (35,000 kips) and a large
package of stiff and strong well conductors influences not only t h e
strength, but as weil, the stiffness, mass, and response of t h e platform
t o earthquake ground motions. As discussed elsewhere (Bea, 1977; Bea,
19761, these other loading factors have a n .important influence on
earthquake design forces; and of course, ultimately on the overall
performance of the system during intense loadings.

The dashed line in Figure WII.13 shows t h e response of t h e platform when i t is


supported by t h e Soil-Pile B substructure system. The increased flexibility is due t o
t h e much softer lateral and axial resistance o€ t h e Soil-Pile B substructure system
(see Figures IV.5, IV.7).

VIII-1 o
Due t o inelastic response developed in t h e B soil-pile system at a BS of 30,000 kips no
significant inelastic action is developed in t h e superstructure. The soil-pile elements
of t h e system act to limit t h e amount of load which can be developed on t h e
superstructure. The piles (compact sections able t o develop large rotations) and soft
soils behave as very ductile elements.
. .
...:.:.. . ... ... ...:

Such potential response has important implications in effectively truncating t h e loads


which can be transmitted to t h e superstructure by ground motions (Bea, 1976). This,
as well suggests an interesting design strategy similar t o the much maligned 'Isoft
story1' (Fintel and Kahn, 1968) concept for conventional building structures. In t h e
case of the offshore platform system, large permanent displacements can be
tolerated in the substructure without seriously impairing utility (Garrison and Bea,
1977; Bea, 1971).

VIII.6 UNCERTAINTY IN PREDICTED ELASTIC RESPONSE

Thus far we have established t h e platform earthquake loading transfer function. This
is given in terms of the correlation between input peak velocity and output maximum
resultant base shear. The next step is to evaluate uncertainties associated with such
correlations.

Shown in Figure WII.14 a r e t h e fundamental elements for t h e evaluation of t h e


primary factors contributing to uncertainty in predicted (elastic) response. Recalling
t h a t t h e uncertainty in t h e transmission laws has already been included in t h e
characterization of expected peak ground velocities, the line labeled V A is intended
to express the additional uncertainties associated with prediction of t h e firm ground
peak velocities (e+, t h e stretch of seismic history .which will adequately
characterize the next 25 to 50 years, Figure VII.3).

The line labeled VB/VA is intended to express the uncertainties in predicting t h e peak
ground velocities in t h e soft soil, given t h e peak ground velocity in t h e underlying
firm soil. It was drawn from recent results of DCHARM measured versus predicted
response study. The evaluation indicates that given the firm ground peak velocity
t h a t in only 1 case out of 100 will t h e actual soft soil peak velocity be greater than
1.5 times the predicted velocity.

VIII-11
-.
.
i .c
~

STD.API/PETRO 119-ENGL 1 9 7 8 0 7 3 2 2 7 0 Ob023b7 8 7 9 D

The line labeled BS max/V is intended t o express t h e uncertainties in predicting t h e


maximum resultant base shear acting on the platform, given t h e effective input peak
ground velocity. This evaluation was developed from the'data shown in Figure VIII.12
and tabulated in Table VIII.1, and includes t h e uncertainties in response results
developed by potential ranges in pile-conductor stiffnesses, damping, deck loads, etc.
The evaluation indicates t h a t given t h e input peak velocity, t h a t in only 1 case out of
100 will t h e actuai BS max be greater than 2.5 times t h e predicted BS m a .

All of the foregoing evaluations indicate unbiased predictions; t h e ratio of t h e actual


to predicted value at t h e 50th-percentile is 1.0.

The lines labeled as BS max)A and BS max)N indicate t h e resultant uncertainties in


t h e predicted response contributed by the previously discussed uncertainties. The
combination of t h e lognormal variates was accomplished using t h e formulation
discussed by Bea (1974).

The resultant uncertainties indicate t h a t predicted response is slightly more uncertain


for t h e soft soils (B) than for t h e stiff soils (A). At the 1-percentile level, t h e ratio of
actual t o predicted response ranges from 3.4 t o 3.8. These uncertainties are included
in evaluations of annual expected maximum response, thus adding significantly t o t h e
natural variabilities expressed in Figures VII.4 and VII.5.

--... VIII- 12
~ ~~ ~
~

...., ..
-..--- STD.API/PETRO 117-ENGL 1778 0732270 Ob023b8 705

TABLE VIII. 1: PLATFORM RESPONSE SUMMAR-Y

PEAK
GROUND RESULTANT OVERTURNING
VELOCITY BASE SHEAR MOPENT
QUAKE (CM/SEC) ( X 10 KIPS) ( X 10 K-IN.)

Taft 49.4 16.9 42

El Centro 53.5 24.4 60


Taft 32.6 11.1 28
El Centro 35.3 16.2 40
Borrego Mountain 66.1 25.0 77
Olympia 25.2 9.6 30
Taft 9.9 3.4 9
Taft - Modified 37.4 16.7 47
Ei Centro - Modified 75.0 23.6 61
Taft - Modified 28.7 12.3 33.5
El Centro - Modified 47.6 16.0 42

Damping = 10%

Axial Pile Stiffness = 4000 K/in


Lateral Pile Stiffness = 85 K/in
S T D - A P I I P E T R O 1 1 9 - E N G L L778 0 7 3 2 2 9 0 Ob023b9 b 4 1 9

BS

m BS rnax
H

AG

FIG, vIII.1 S C H E M A T I C LOAD A N D DEFORMATION B E H A V I O R OF


PLATFORM UNDER W A V E AND EARTHQUAKE LOADING
CONDITIONS

WOODWARD-CLY DE CONSULTANTS
-..
L .c
S T D - A P I / P E T R O 1 1 7 - E N G L 1978 = O732290 Ob02370 3b3

I kRu

INDUCED LOAD (BSmax) & ULTIMATE i


RESISTANCE (RU)

>
-
c
=!
m
Q
m
O
a
Q

i
i

:
i
I!

FIG, V III Z TRUNCATION OF EARTHQUAKE LOADING

-. WOODWARD-CLYDE CONSULTANTS
~
S T D - A P I / P E T R O 1 1 7 - E N G L 1778 m 0 7 3 2 2 7 0 O b 0 2 3 7 1 2 T T m

1000
80

60

40

30

20

B S TRUNC.
I
8 2.8 ,
DR
IOC 1

6
I
4 - 4-I)

3
ES TRUNC.= 1.6 RA,

IO
0.01 O.! I .2 5 IO 20 40 60
CUMULATIVE PERCENTAGE: OF VALUES 2

FIG, V I 1I , 3 TRUNCATED EXPECTED MAXIMUM RESULTANT B A S E


SHEAR FORCE CHARACTERIZATION FOR STUDY PLATFORM

WOODWARD-CLYDE CONSULTANTS
.-
-.
L_
. . .. .. .
.....
. ... ., . . ,
- .......- -.....,.. ...,
... . '...

u)
U
C
O
u
a2
v,
n
I
.Q
Y
W
-
5
I-

+
LL
a
I-
9 0
O
8
(u
O
-

/t
Ix

O O O

. . .. .
._

** This document has been reproduced from = WOODWARD-CLYDE CONSULTANTS ~ ~~ -

the best hardcopy available ** -


~~~~~~ ~
--
STD.API/PETRO 1 1 7 - E N G L 1778 m 0 7 3 2 2 7 0 Ob02373 072 m

40
I
O p e a k = 0.33 g 1
/3= 4 v e

35

30

25

20

15
I
I -
I
IO
O 100 200 300 400
. PILE LATERAL STIFFNESS - KI- K I P S / I N
( A X I A L STIFFNESS = 4000 KlPS/IN 1

FIG, VIII, 5 INFLUENCE OF PILE LATERAL STIFFNESS


ON RESULTANT BASE SHEAR

WOODWARD-CLYDE CONSULTANTS
-. L .%
.. .. . . . -..
._....
.... .-.: ,...,::. .
<--..y.
. ... S T D * A P I / P E T R O 1 1 9 - E N G L 1978 0732290 0 b 0 2 3 7 q T O 9 D

40
I
Opeok = 0.33

v) 40’0~
n 35
z
I
K
E 30
v,
m
I
a
a
W
I 25
cn

I-20
z
U
5
2
w
a

10
2 4 6 8 10
PILE AXIAL STIFFNESS - K a -
IO3 KIPSAN
( LATERAL STIFFNESS = 85 KIPS/ IN )

FIG, VIII, 6 INFLUENCE OF P I L E A X I A L S T I F F N E S S


ON RESULTANT BASE SHEAR
S T D - A P I I P E T R O 11’7-ENGL 1978 = 0 7 3 2 2 9 0 Ob02375 945

3 10 I I I

DAMPING RATIO -p - PERCENT I

l
i
i
i
i

FIG, VIIi, 7 INFLUENCE OF DAMPING ASSUMED FOR THE


i
STUDY PLATFORM SYSTEM iI//

WOODWARD-CLYDE CONSULTANTS
-.
L .c
~ ~~

..,. .. .
.
.. .. . .
.

-
1 -

EL CENTRO 1940
S. COMPONENT
- - -- -A--

--2 --_

0.1 I .o IO
PER100 -SECONDS

FIG, VIII, 8 DCHARM RESPONSE SPECTRA FOR SOIL PROFILE B

--. ._ WOODWARD-CLYDE CONSULTANTS


-
,
-
. ...
... ,
STD.API/PETRO 117-ENGL 1778 O732270 0 b 0 2 3 7 7 718

f t
ELCENTRO 40 S
II 40 W
TAFT 69 E
70 ll 21 E-
OLYMPIA 86 E
I I
!
60

50
A

40 A
L

30

20

IO

O
O IO 20 30 40 50 60
INPUT (-100FT) MAXIMUM VELOCITY - CM/SEC

FIG, VI11 , 9 COMPARISON OF INPUT (-180 FT)


Y IMUM VELOCIPI
AND OüTPüT (-4 FT) FOR SOIL PROFILE B

-.
2 _.% WOODWARD-CLYDE CONSULTANTS
- .
S T D * A P I / P E T R O 1 1 7 - E N G L 1778 H 0732270 Ob02378 b5'i

... .
. .
.

40 I I l t

30

20

IO

O
O 20 40 60 80 100
, .. PEAK GROUND VELOCITY -V- CM/SEC

FIG, V I 1I STUDY PLATFORM MAXIMUM RESULTANT BASE SHEAR VERSUS


PE4K GROUND V E E I T I E S FOR MODIFIED INPUT MOTIONS

-. WOODWARD-CLYDE CONSULTANTS ~~

-
-~

-
I
STD-API/PETRO 117-ENGL 1978 m 0 7 3 2 2 9 0 Ob02377 5’70

b=10%
I
Ka = 4000 KIPS/IN
K i = 85 KIPS/ IN

8 RAW
0 MODULATED

-0 1
O 20 40 60 80
PEAK GROUND VELOCITY - V- CM /SEC

FIG, vIIItu STUDY PLATFORM M I M U M RESULTANT OVERTURNING MOMENT


VERSUS PEAK GROUND VELOCITY FOR MODIFIED INPüT MOTIONS

WOODWARD-CLY DE CONSULTANTS
-.
L C
S T D * A P I / P E T R O 117-ENGL 1778 m 0 7 3 2 2 7 0 Ob02380 202
siwmsNo3 3am-awMaooM

MAXIMUM RESULTANT BASE SHEAR - 103Kips


Iu u P 0
O O O O O O

TI
m
D
x

L
z
O
<
m
r
O
-
c,
+
<
I

, r
o
7 D 3
\
3 4u
-
4 0 c)
öO l
z

MAXIMUM RESULTANT OVERTURNING MOMENT - IO6 in - kips


U
m
D
x

c
z
. . O
æ <
G
I
tc:
O
E OF
3 o
-
m -I
2
-I -<
I
c,
3

** This document has been reproduced from


the best hardcopy available **
60

50

40

30

20

' IO

O
-
O 0.5 1.0 1.5 2.0 2.5
LATERAL DISPLACEMENT A T DECK - Feet

FIG, VIII I 13 STATIC-ZNELASTIC-TO-COLLAPSE


ANALYS IS OF STUDY PLATFORM

WOODWARD-CLY DE CON SU LTANTS


-.
&- *
- .
._
.,. .
._. ..-
.. ....--..
. ....-.-
.
._
,.... ..._.._ _.!

0.5
0.01 0.1 1 2 5 io 20 40 60 80
CUMULATIVE PERCENTAGE OF VALUES

FIG, VIII, 14 ~ N T A ELEMENTS


L FOR EVALUATIONOF PRIMARY FACTORS
CûNTRIBLVING TO UNCERTAINTIES IN PREDICTED ELASTIC RESPONSE

-.
i- -% WOODWARD-CLYDE CONSULTANTS
- .
~ ~ ~

I0 7 3 2 2 7 0 ü b 0 2 3 8 3 T L L
STD.API/PETRO 1 1 7 - E N G L 1978 .

Ix RELIABILITY ASSESSMENT

Given the platform loading and resistance probabilistic charactennations developed in


the previous four sections (Sections V through VIII), i t is possible t o compute t h e
reliability of the platform system designed for variable levels of elastic design loads.

IX. 1 OCEANOGRAPHIC LOADINGS AND PLATFORM RESISTANCES

The oceanographic loadings (wave and current) acting on t h e study platform have
been computed and presented in Section V of this report. Figure V.5 presented t h e
total wave forces on the platform versus wave height. Combining this wave loading
relationship with the statistical description of wave heights given in Figure V.4 has
resulted into a characterization of realizing t h e various levels of total wind, wave and
current. loadings associated with t h e projected storm environment (wind loads were
taken as five percent of the wave-current loading). This Characterization was
presented in Figure V.6. The results of t h e analysis of modeling deficiencies for
storm loading has been presented in Section V.

The analysis has attempted to recognize uncertainties in hindcast maximum wave


heights and concurrent currents, t h e flow theory drag coefficients, effective
projected area of the platform, and wave periods.

Results of t h e analysis are given in Figure IX.1. Given is a probabilistic expression of


t h e ratio of actual t o predicted BSmax for storm loadings. An unbiased force
estimator (Xso = 1.0) and a cov of 66 percent a r e indicated.

Section Vi focused on t h e resistance characteristics, in both a load and deformation


sense of t h e study platform system t o storm loadings. Derivation of this
characterization was based on work done by Marshall and Bea (1974) and Marshall
(1974). Figure IX.2 presents the statistical description of t h e resistance ratio RR
(collapse load/elastic design load) for intense storm loading. As can be seen from this
figure, t h e median RR is 3.0 and the cov is 53 percent.

IX-1
- .
IX.2 OCEANOGRAPHIC RELIABILITI’ ASSESSMENT

The reliability (probability t h a t resistance exceeds loading) of the platform system


under oceanographic loading has been quantified using the formulition documented in
Section II. I t should be noted t h a t both t h e annual maximum loading-force and
resistance characterizations have been described using lognormal distribution models.

The of fshore study platform annual reliability computations under oceanographic


results are presented in Figure IX.3 by t h e line identified with t h e word “waves.” The
abscissa is t h e total resultant lateral elastic design load which has been used t o size
-
t h e platform (R 1. The ordinate i s t h e annual probability oi failure (Pf = 1 annual
D a
r eiiabilit y).

For example, this figure shows t h a t t h e projected annual probability of faiiure under
oceanographic loading for t h e study platform located in t h e Gulf of Alaska is 0.18 for
a n elastic design lateral base shear of 35,000 kips.

The values of elastic design load (R,,) required to develop a n annual probability of
failure of 0.5 and 1.0 percent under oceanographic loading are 26,000 and 20,000 kips,
respectively.

IX.3 EARTHQUAKE INDUCED FORCES AND PLATFORM RESISTANCES

Sections VI1 and VI11 focused on t h e details of earthquake environmental conditions


and platform elastic and inelastic response t o them. The computed variation of
maximum resultant base shear with peak ground velocity is shown in Figure VIII.12a.
The result is based on some 80 three-dimensional, elastic, time-domain analyses of
t h e study platform system.

The range of t h e results (the shaded band) illustrate t h e sensitivity of maximum


resultant base shear t o a wide range of input motion t i m e histories (ten 3-D sets),
different values of damping (ranging from 5 to 10 percent of critical) and different
foundation stiffness assumptions (fixed t o very soft).

Again, as for storm loading, t h e earthquake induced force transfer function shown in
Figure ViII.12a is used t o transform projected ground motion conditions (Section VI11
t o an estimate of future expected annual maximum resultant base shear imposed on
S T D * A P I / P E T R O 119-ENGL 1978 0732290 Ob02385 89Li H

t h e study platform at the location under consideration. Analysis has been performed
to estimate the resultant uncertainties in computed induced forces due to earthquake
ground motions.

The analysis has recognized t h e uncertainties in forecasted free-f ield ground motions,
in developing ground motions appropriate for analysis of platform response (effects of
soil-foundation-superstructure interactions), in computing maximum resultant base
shear with a given ground motion time history scaled to an effective peak ground
velocity and with t h e analytical model of t h e platform used.

The result of this analysis is also given in Figure IX.l. An unbiased force estimator
(Xso = 1.0) and a cov of 82 percent a r e indicated.

Load and force uncertainties subsequently will be combined with t h e natural or


inherent variabilities. As one would expect, the result is a substantially increased
overall variability or uncertainty.

The results of t h e static, inelastic analysis of t h e study platform system a r e


summarized in Figure VIII.13. This analysis formed the basis for the resistance
estimate.

The analysis has indicated that t h e overall system ductility, referenced to the
deformation at the design load is in the range of 2 t o 3. The margin between the
elastic design load (RD)and inelastic collapse load (RU)is about 2.

Shown with a dashed line in Figure VíII.13 is t h e response anticipated for the study
platform supported by a soft soil foundation (soil B). By the time the first damage
state in t h e superstructure is reached (Ry),t h e system has displaced almost 3.5 f t at
t h e deck level. At this point, plastic hinges have begun to form in most of t h e piles.
Adáitional displacements of several feet are required to form the additional hinges
which would lead to a collapse mechanism in t h e substructure. In the process, t h e
forces in the superstructure elements are limited to a total of about 40,000 kips.
Thus, t h e %oft" substructure (soils, piles, conductors) may potentially function like a
shock isolation device, limiting t h e forces and relative displacements developed in t h e
superstructure system by t h e ground motions.

-.
c .e IX-3
STD.API/PETRO iLî-ENGL Lî78 W 0 7 3 2 2 7 0 Ob023ôb 7 2 0 m

T h e earthquake resistance .characterization for t h e study platform system has been


summarized in Figure IX.2. Ultimate or collapse resistance (R (defined on a force,
U
not deformation, basis), has been referenced to t h e elastic design load (RD)through
t h e Resistance Ratio (RR).

Uncertainties in computed resistance were developed using d a t a and methods outlined


by Marshall and Bea (19741, Marshall (19691, and Marshail (1976). Variabilities are due
primarily t o those of materials, fabrica tion, element strength, and performance of
subsystems (composed of elements and joints). The median RR is 2.0 and t h e cov is
28 percent.

IX.4 EARTHQUAKE RELIABILITY ASSESSMENT

The reliability of t h e platform system under earthquake loading has been quantified
using t h e formulation documented in Section II of this report.

The annual reliability assessment of t h e study platform f o r t h e Gulf of Alaska


earthquake environment is presented in Figure IX.3 by t h e line identified with t h e
word "quakes1'. The abscissa is t h e total resultant lateral elastic design load (RD)
used t o size t h e platform. The ordinate is t h e annual probability of failure (Pf = 1-
a
annual reliability).

I t should be mentioned t h a t this reliability assessment is based on random earthquake


occurrences. Non-random ltgap-fiïlingll events occurrence will be accounted for in a
l a t e r p a r t of this section.

The results shown in Figure IX.3 indicate t h a t t h e projected annual probability of


failure f o r t h e study platform located in t h e Gulf of Alaska at a design load (RD)of
25,000 kips is 0.18 percent.

The values of elastic design loads required to develop a n annual probability of failure
of 0.5 and 1.0 percent under random earthquake occurrence loading are 13,000 and
10,000 kips, respectively.

IX.5 COMBINED ENVIRONMENT RELIABILITY ASSESSMENT

In t h e previous subsections, t h e platform reliability for t h e random Ioading


components (earthquakes and waves) have been presented.

IX -4 - .
.. ...-..
.
--. . >:._..
. ;. ?:
.. . , S T D - A P I I P E T R O 1 1 7 - E N G L 1978 m 0732290 Ob02387 b b 7 m

It is obvious t h a t t h e loadings due o t winds, waves, and currents are the primary
environmental loading effect examined here. The average annual probability of
platform failure (Pf ) is almost an order of magnitude larger for the storm loadings
than those of t h e ranadom earthquakes.

The platform reliability which reflects t h e overall safety of t h e platform subjected to


t h e loadings expected during its lifetime is the quantity of primary interest. Its
compliment, t h e lifetime probability of failure (Pf provides the basis for combining
t h e ef f e a s of t h e random and non-random loading komponents as follows:

Pf = Pf
ar wwc + pfRQ
pf
=I - exp (-Pr XL)
Lr ar

where: Pf = Average annual probability of failure due to random loading


ar threats
= Average annual probability of failure due to,wind, wave,
f' wwc
and current loadings
= Average annual probability of failure due t o random
PfRQ earthquake loadings
= Probability of failure due to rândom loading threats during
pf
Lr the platform lifetime, L
= Probability of failure due to t h e non-random earthquake
pf
NRQ loading, assumed to occur during t h e platform lifetime, L
= The total probability of failure due .to both random and non-
Pfr random loadings during t h e platform lifetime, L.

There a r e three major assumptions which underlie t h e combination of probabilities


outlined in these equations. The first is that the dasses of loadings are statistically
independent.

IX-5
- - ~

S T D - A P I / P E T R O I L S - E N G L 1978 0 7 3 2 2 7 0 Ob02388 5 T 3

Due t o t h e frequency of moderate (BSmax 10,000 kips) storm loadings (an average
of about once every t e n years) as compared t o moderate quake ground motion
loadings (an average of about once every 20 years), this is felt to be a reasonable
representation.

The second is t h a t t h e occurrence of the random events during t h e lifetime (L)can be


described by a Poisson Model (Borgman, 1963, 1972). This is indicated t o be a n
acceptable assumption based on results of previous analyses (Bea, 1975; Rosenblueth,
1973).

The third assumption is t h a t i t is a certainty t h a t t h e non-random event, the '!gap-


filling" quake, will occur during t h e platform lifetime (LI.

This third assumption is certainly t h e most tenuous assumption of the three.


However, until more is known of t h e temporal, spacial, and magnitude characteristics
of "gap-filling" quakes, this seems t o be a good assumption.

The results of t h e foregoing discussion as zpplied to t h e component reliability


characterizaiton a r e shown in Figures IX.3 and IX.4. Figure IX.3 shows t h e combined
annual reliability (Pf of the random earthquakes' (RQ)and winds, waves and current
(WWC)environments. V i g u r e IX.4 shows t h e results of t h e analysis of t h e influence of
t h e non-random earthquake (NRQ) induced forces on projected platform reliability.

The analysis has assumed t h a t t h e "gap-fillingfl earthquake will occur during t h e 25


year economic life of the platform. Given a knowledge of t h e probability distribution
of waiting times for such an event, one could modify this assumption.

The line labeled 'INRQ' indicates t h a t t h e platform designed for an RD of 30,000 kips
would have a 10 percent lifetime probability of failure, given t h a t the non-random
event occurred. The line labeled "RQ + WWC'I indicates that, at a n RD of 30,000
kips, t h e total lifetime probability of failure of the platform due to the random
earthquake and storm (wind, wave, current) loadings is 5.3 percent.

1x4
-..
The line labeled llTotai'l indicates the summed contributions to platform failure of the
independent loading threats - random storm and earthquake loadings plus t h e single
occurrence of t h e non-random, gapfilling earthquake. The two -arrows to t h e right
designate the lifetime (25 years) probabilities of failure which are equivalent to 1.0
. ... .
and 0.5 percent annual probabilities of failure.
i'..
. ... . ......
- .. ..--.
..... ....
.. ..
. <.
Choosing elastic design loads (RD)on t h e same basis as previously described would
-
indicate values of 27,000 to 32,000 kips about a 15 percent increase in R,, as
compared t o t h e cases developed neglecting t h e effects of t h e non-random
earthquake loading threat.

Such design loadings would be developed by wave heights of 104 t o 113 feet, or
alternatively peak ground velocities of 63 to 74 cm/sec.

IX-7
~

STD.API/PETRO L L S - E N G L 1778 D 0 7 3 2 2 7 0 Ob02370 151 D

O.5
0.0I O.I 1 2 5 10 20 40 60 80

CUMULATIVE PERCENT 1

FIG, I X , 1 PROBABILISTIC DISTRIBUTION OF ACTUAL V S , PREDICTED


MAXIMUM BASE SHEAR FOR STORM AND QUAKE LOADINGS

L- t
WOODWARD-CLYDE CONSULTANTS
= 0732290 0 b 0 2 3 7 1 O78 W
~

S T D * A P I / P E T R O 117-ENGL 1778

I
i
l

99 l

l
98

95

90

VI
80
i-
- 2
w 70
o
a 60
W
e
50

IO

2
I
' I 2 3 4 5 678910
COLLAPSE LOAD
RESISTANCE RATIO =
ELASTIC DESIGN LOAD

FIG, Ix, 2 R E S I S T A N C E C H A R A C T E R I Z A T I O N FOR STUDY PLATFORM -


STORM 8 EARTHQUAKE LOADING

WOODWARD-CLYDE CONSULTANTS
STD.API/PETRO 1 1 7 - E N G L 1778 0 7 3 2 2 7 0 Ob02392 T 2 4

R D : ELASTIC DESIGN LOAD - IO3 K i p s

FIG, Ix, 3 CHARACTERIZATION OF ANNUAL PROBABILITY OF


FAILURE IN RELATION TO ELASTIC DESIGN LOAD
FOR OFFSHORE PLATFORMS

-.
i_.% WOODWARD-CLYDE CONSULTANTS
- ~~~~ ~

.i e.
. S T D - A P I / P E T R O 117-ENGL 1778 0732290 Ob02373 Yb0

I .O

IL
O

o. I

0.0I
.
. '. . :
IO 20 30 40 50
ELASTIC DESIGN LOAD - IO3Kips
CHARACTERIZATION OF LIFETIME PROBABILITY OF
FAILURE OF A GULF O F ALASKA PLATFORM DUE TO
RANDOM AND NON-RANDOM EVENTS

__ ,,.\ISULTANTS
~
~

STD.API/PETRO 1 1 7 - E N G L I778 0732270 Ob02374 B T 7

X. VALUE ASSESSMENT

Before a design load can be selected from t h e foregoing informatlon, a decision must
be made on t h e level of desirable reliability, or conversely, on t h e acceptable
probability of failure. This decision process is essentially a value analysis (Section III)
which a t t e m p t s t o develop an equitable balance between t h e benefits of safety
(failures and failure consequences not experienced) and costs of achieving safety.

x. 1 INITIAL COSTS

The initial tangible cost estimate for t h e 12-leg study platform (1974 price base) is
summarized as follows:

' o Materiais and Fabrication, $27 million


o Installation, $12 million (could increase t o $20 million)
a Transportation, $7 million
a Contingencies and Miscellaneous, $4 million
a Total, $50 million (to $58 million)

I t is important t o recall t h a t these figures exist in conjunction with platforms


designed f o r a total lateral load of 25,000 kips.

For a platform designed f o r a total lateral load of 35,000 kips, initial costs were
estimated as:

o Stiff soil conditions, $120 miiiion


a Soft soil conditions, $100 miiiion ~.

The increase in design load of 10,000 kips cost approximately $50 million.

X.2 DAMAGE STATE COSTS

The potential future tangible failure costs were developed by Bea (1976). The
following summarizes the estimates:

x-1
~~ ~ -~ ~ ~ ~~

STD.API/PETRO LLS-ENGL 1978 0 7 3 2 2 7 0 Ob02375 733 D

Salvage Costs, $30 t o $50 million


Well Control Costs (assuming 95% reliability downhole shut in
equipment) $10 to $20 million
Pollution Control and Cleanup Costs, $10 t o $20 milhion
Platform Replacement Costs (assumed platform replacement in all
cases), $100 to $120 million
Well Re-drilling Costs, $20 t o $40 million
Production Equipment Replacement, $40 t o $50 milbion
Deferred Production Costs, $0 t o $200 million
Total, $210 to $500 million

X.3 TANGIBLE COST ANALYSIS

The following equations express the evaluation system components utilized t o assess
total tangible costs (see Section III):

where: E(C) = expected total costs


E(1) = expected initial cost. The variation of platform cost with
probability of failure (linked t o t h e design load) is shown as a
dashed line in Figure X. 1.
E(F) = expected present vaiue of potential failure costs.

The expected present vaiue of potential failure costs is computed as follows:

E(F) = E(PL) x PVF x Pf

where: E(PL)= expected cost of potential losses in event of a platform


failure. In this study platform example, a range of $10 to
$30 million has been used.
PVF = a design life, interest and inflation r a t e dependent function
intended to present-value future costs.
Pf = t h e probability of platforni failure during its life.

The expected cost of potential losses in t h e event of a platform failure is computed


as follows:

x-2
. . ... .:..
...i..,
.... ... ... ...... . ..
"

n
E(PL) = 2 PL^ x pii
i=l

where: PLi = t h e potential loss due to cause i, given a platform failure.


Pli = t h e probability of experiencing t h e potential loss due to
cause i, given a platform failure.

The present value function is computed as follows:

-(4-i)L
PVF = 1 - er-i

where: r = t h e interest r a t e used for discounting profit and loss form


future to present value.
i = t h e inflation r a t e on t h e cost of failure consequences. In
this example, a n e t r a t e (r-i) of 10 percent has been used.
L = t h e design life of the platform. A 25-year life has been used
in this study.

The results of applying t h e foregoing tangible cost evaluation system t o t h e first and
failure costs estimated for t h e study platform operaitons are given in Figure X.1.
The ordinate expresses t h e expected total costs (initiai + present-valued, risk
weighted, failure costs). The abscissa expresses t h e average annual probability of
failure f o r the platform.

Results for total expected costs of failure of $200 and $500 million are shown. The
l1optirnumt1or desirable (minimum total expected costs) average annual probability of
failure (fioa) falls in t h e range of 0.6 to 1.4 percent.

Previous work by Eka (1974) has shown that for a given total expected cost of failure,
it is t h e slope of t h e initiai cost curve (change in initial cost for a given change in
design load or probability of failure) which determines t h e value of t h e optimum
probability of failure. The results of such an evaluation, in which the slope of t h e
initial cost curve is changed from $65 million (per order of magnitude change in t h e
probability of failure, e.g., from 1 to 0.1 percent) to $98 million (+JO%) t o $33 million
(-50%) are shown in Figure X.2.

- ...
x-3
L
~ ~ - - ~ ~-

- ...*-- STD.API/PETRO 117-ENGL 1778 D 0 7 3 2 2 7 0 Ob02377 Sub

The evaluation indicates t h a t at a n expected cost of failure of $200 million, a 50


percent increase or decrease in t h e initial cost curve slope changes t h e optimum
annual probability of failure from about 1.4 percent t o 2.1 p e r c e n t and 0.7 percent,
r espectivel y.

For t h e $500 million cost of failure, t h e optimum probability of failure falls in t h e


range of 0.3 t o 0.8 percent, with a value of about 0.5 percent for t h e projected slope
in t h e initial cost curve and t h e total expected costs of platform failure (or
catastrophic loss of serviceability). It has been shown t h a t the annual probability of
failure which produces the lowest combination of initial c o s t and potential future
losses due to platform failure is:
AC.
I
Pfao = Cf x pvf x 2.3

- 1
-2.3 x pvf x Cost Ratio

where: The annual probability of fa,*Jre determined on t h e basis of


minimizing t h e tangible costs associated with placement of
t h e platform and those potential costs associated with
platform loss of serviceability (defined as failure here).
ACi = The slope of t h e initial cost curve when plotted against t h e
common log of t h e annual probability of failure.
Cf = The expected cost of platform loss .of serviceability of
failure.
Cost Ratio = The ratio of Cf to ACi
PVF = A present value function intended to bring t o present cost
terms t h e costs of potential future losses. In this work, i t
was assumed t h a t PVF is t h a t associated with a uniform
series of payments over 25 years at a n effective interest
r a t e of t e n percent. Thus, PVF = 9.077.

If i t is f u r t h e r assumed t h a t t h e occurrence of t h e loss of serviceability events is


adequately modeled as a Poisson Process (Borgman, 1972; 19631, then t h e optimum
lifetime reliability (PrLo) becomes:

PrLo = exp -(Pfao x i)

-.
.L .c x -4
~~
~~ ~

..-.....
. . ....... .
S T D * A P I / P E T R O 119-ENGL L978 0732270 Ob02398 Li112
. -.-
...,.....
I

The variation of optimum annual probability of failure and optimum lifetime (25
years) probability of failure with the Cost Ratio is shown in Figure X.3.

X.k INTANGIBLE COSTS ANALYSIS


To this point, we have considered only tangible costs. The other category of costs,
intangible, becomes more difficult to deal with. There are two primary schools of
thought: (I) evaluate t h e intangible costs in monetary terms, no matter how difficult
or objectionable, and include these costs in the tangible costs vaiue analysis; and (2)
. .
use history and experience as a basis for judging what is tolerable and acceptable.

The most reasonable approach at the present time appears t o be the second school of
thought. Intangible costs a r e characteristically difficult to define or quantify.
Quentifications become embroiled in emotional issues. The second approach has t h e
advantage that history or society defines what is tolerable and thus t h e situation of
having a small group tell society what i t should accept is avoided.

In the Gulf of Mexico hurricanes pose t h e primary environmental loading threat.


These storms a r e closely monitored from birth to death. Personnel a r e evacuated
from t h e platforms prior t o their arrival.

However, in an area such as the Gulf of Alaska, where severe storms develop more
frequently and quickly and facilities a r e not available to monitor their progress as
closely, t h e evacuation of personnel is not practical. They-must ride it out with the
platform. Thus, t h e potential loss of their lives must be integrated into t h e decisions
on platform design criteria.

As noted earlier, there a r e two approaches to evaluate t h e impact of such potential


losses. In this case, t h e first approach would consist of equating t h e loss of a worker's
life to perhaps some insurance valuation, court case settlement amount, or t o the loss
in gross national product of t h e country. These costs were estimated in 1974 to be in
t h e range of $100 to $500 thousand per life (Rosenblueth, 1974; Wiggins, 1974).

In a given platform failure, t h e number of lives lost might range between none (the
platform didn't collapse to t h e point where men could not find refuge from t h e storm,
life-saving equipment was used, or the platform was unmanned at the time) t o say 20
to 30. It is important to note that we want to focus on an average life exposure
during t h e entire platform history, spanning from t h e drilling phase when t h e largest

x-5
~

STD.API/PETRO 117-ENGL 1978 W 073Z270 ü b 0 2 3 9 9 3 8 9

crew is onboard t o t h e production phase when there may be few or no (automated


ope rations) personnel onboar d.

Using 20 lives lost as an example, and $250 thousand per life, the total loss owüld be
valued a t $5 million. Adding such a cost t o t h e previously discussed tangible costs of
failure would result in a very small change in t h e acceptable risk level for the study
platform.

However, this analysis doesn't properly recognize the importance of potential life
loss. For very large tangible costs, life loss value has almost no influence on t h e
decision.

The second approach is based on the premise that society has historically determined
what is an acceptable or tolerable level of life loss. This level varies with t h e
percentage of the population exposed, t h e value of t h e activity and, most important,
whether or not i t is a voluntary activity.

Such considerations of the impact of technology on life a r e not new. They have been
considered in developing criteria for design of ships (Abrahamsen, 1963; Carsten,
19731, air traffic control centers (Raisbeck, 19711, public buildings (Wiggins, 1972;
Wilson, 1973), and nuclear power plants (Starr, 1969; Sagan, 1972).

The background and life loss statistics have been extensively studied by Starr (1971).
This work indicates that life loss risks divide themselves into two categories:
voluntary and involuntary. Voluntary risks a r e those taken by miners, divers, airplane
pilots, etc. Involuntary risks a r e those due t o natural disasters that affect cities, t h e
modern aspects of life (medicine, canned foods), and disease..

The best present basis on which t o compile and compare risk rates in these two
categories is t o normalize them on the basis of expected fatalities per year of
exposure t o t h e particular activity being discussed. Thus, miners a r e exposed t o their
work risk about 8 hours each day (1760 hours per year) and t h e general public t o the
risks associated with power generation 24 hours each day (8760 hours per year).

X -6
. . ,.,
.-::.,. ..
..-..
_... .. . .
._.
.. ... . STD.API/PETRO 1 1 7 - E N G L 1778 = 0732290 Ob02400 720

Starr's studies indicate that there is a generai boundary between voluntary and
-
involuntary life loss rates. It is about equal to the risk of death due t o disease, or
about one death per 100 person-years of exposure. Voluntary risks fall in t h e range of
10 to 1000 times this figure.

In t h e application of the foregoing data and rationale t o the offshore worker and
platform, it will be assumed for purposes of illustration, t h a t personnel will have no
evacuation potential and that they will face the risk of platform failure in storms
each day throughout t h e year. Further, it will be assumed that every platform failure
will be catastropic in nature with a total loss of 10 t o 40 crew members.

The results of these assumptions and the application of t h e life loss data a r e shown in
Figure X.4. The variation of expected annual casualties with t h e platform risk level
is shown relative to t h e upper level of involuntary risk and t o a tolerable voluntary
range 10 to 100 times this amount. Within the mid-range of the tolerable voluntary
range, t h e acceptable platform risk level is about one percent per year (range of 0.8
t o 3 percent per year).

There is a very important contrast between this historical life loss analysis and the
life loss valuation analysis previously discussed. In the historical analysis, the impact
of life loss has been examined as a separate parameter. It has not been mixed with
the costs of steel and equipment involved in the tangible cost anaiysis. The difficult
decision of placing a value on human life has been avoided. It is a fortunate situation
for the offshore platform value analysis that the results of the tangible and life loss
analysis indicate about the same acceptable risk level (1.0 percent per year).

X.5 UTILITY EVALUATIONS

The concept of utility has been previously discussed in Section 111.0. It was shown
that, in a value analysis, cost units rather than money can be used. These cost units
are based on t h e concept of utility evaluations, in which llutilityllis an expression of
-
t h e value of costs to a particular decision maker or group. Both tangible costs and
intangible costs can be assigned values of utility. For example, to some decision
makers, future costs, even if present-valued, have less "utility" or value than present
costs because of their future and problematical nature. In effect, t h e bias that exists
regards future failures as less important than present initial costs. These types of

x-7
...
...
S T D * A P I / P E T R O 119-ENGL 1778 W 0732290 Ob02401 Ab7

decision-makers a r e therefore termed "risk takers." The opposite evaluation is true


for t h e risk averter. Likewise, in terms of intangible costs, a bias toward the present
e f f e c t s of intangible costs indicates a lower value of utility t o t h e decision maker.
Through experience or interrogation of these decision makers, t h e variation of utility
units (arbitrary in magnitude) for t h e various types of costs and with t h e various
levels of monetary costs, can be developed. Then, rather than working in monetary
terms, t h e calculations or weighing process is carried out in utility units.

In t h e value analysis for t h e study platform previously described, it was assumed t h a t


t h e decision maker is neither taking or averting risk; this resulted in a n expected
value process in which utility units and monetary units were equivalent measures.

X.6 HISTORICAL TARGET RELIABILITY ANALYSIS

Results of t h e tangible value analysis f o r t h e study platform operations would


indicate a possible range of Cost Ratios from 2 to about 10, with a n expected range
of 3 ($200 million/$65 million) to 5 ($500 million/$98 million). Cost ratios for Gulf of
Mexico and North Sea platforms also fall in this same range.

The expected range of Cost Ratios would indicate t h e following optimum probabilities
of failure:

o Average Annual, O.% to 1.4 percent


e Lifetime (25 years), 18 t o 30percent.

Therefore, t h e average annual t a r g e t reliability range would be 98.6% t o 99.2%, and


t h e lifetime t a r g e t reliability range would be 70% t o 82%.

X-8 ,
- ..*
c
- .
.. ,.
. ..... .
..... ..............
..
-.. ..
. . . . . . ....,
.. .
.

i80

# 160

o
W
t;
wa 80
X
W

60
0.5 I 2 4
AVERAGE ANNUAL PROBA8lLITY
OF FAILURE - P f a d PERCENT

FIG, X, 1 TANGIBLE COSTS EVAWTION FOR STLDY PLATFORM

1
-. WOODWARD-CLYDE CONSULTANTS
S T D - A P I I P E T R O 1 1 7 - E N G L L778 0 7 3 2 2 9 0 O b 0 2 4 0 3 b3T
-.

IO 1 I I 1 I I I 1
8

4
I

O.
I O0 200 300 400 500
EXPECTED COST OF FAILURE
-CF- ILL LIONS (1975 PRICES)

FIG, x, 2 SENSITIVITY OF EXPECTED COST Of FAILURE

- . WOOD WA RD-CLYDE CONSULTANTS


. . ~. .
.-..~..... :-... .
- ..... ._
.._. .i.-
, .,.. ,
-. .,. ,

i>

II 2 4 6 0

COST RATIO: A
Cfr

FIG, Xe 3 COST RATIO VERSUS OPTIMUM PROBABILITY OF FAIWRE

WOODWAR D-CLYDE CO NSU LTANTS


-.
~ ~~~

k .c
S T D . A P I / P E T R O 11’7-ENGL 1778 = 0732270 Ob02405 402 D

IO

1.0

a
U
u
>
œ
w o. 1
e
#
W

5a
3
u)
a
o
n
W
i-
u
w 0.01
a
X
W

0.001
0.01 O.I I.o 10
ANNUAL PROBABILITY OF FAILURE
-
Pf PERCENT PER YEAR
-

FIG, 4 EVALUATION OF LIFE LOSS RATES

-. WOODWAR D-CLYD E CON SU LTANTS


XI, DESIGN STRATEGIES

XI.l ANALYSIS AND DESIGN PROCESSES


The overall objective of this study has been to document t h e basis for a portion Óf t h e
API seismic design guidelines. A key word in t h e foregoing is design. It is important
t o make a clear distinction between the processes of design and analysis:

o The analysis process is intended t o develop t h e details of response,


given a platform system and environmental system (e+, ground motions .
or wave heights). Intelligent, detailed, and perceptive analysis should
precede t h e design process.

o The design process is intended to be a simplified, y e t realistic, process


which t h e design engineer can use t o develop desirable (cost efficient)
performance (strength and deformations) and reliability of t h e platform
system

Design guidelines are intended to result in desired strength and reliable performance
of the facility. Strength or performance is taken to be comprised of two primary
elements: ultimate capacity, and ductility. Ultimate capacity reflects t h e load
carrying capabilities of t h e facility. Ductility reflects t h e deformation capabilities
of t h e facility.

Strength is determined by t h e design constraints (e.g., specified seismic forces,


allowable steel stresses), the analytical process used to determine
loads/forces/stresses, t h e construction process, and t h e verification/quality control
processes, f o r a given type of structure.

The designation of a desirable level of strength is determined as t h e result of a


complex process of decisions which a t t e m p t t o reach a balance between the costs of
achieving a given level of strength and those costs which might be incurred if t h e
strength is not enough. Costs in t h e tangible sense of dollars and in t h e intangible
sense of impact and e f f e c t s are included in such a process.

XI-1
~
Earthquake ground motions and platform system response are very complex
phenomena. There a r e significant uncertainties involved in t h e specification of t h e
ground motion characteristics and t h e potential structure system response to
earthquakes. Seismic design guidelines should explicitly deal with these complex
phenomena and t h e uncertainties associated with them.

Yet, design guidelines should provide t h e designer with a simple, readily applied
process and set of parameters which will guide him in engineering a facility system
(superstructure, foundation, soils) t o have acceptable performance characteristics.
To be able to develop design guidelines, one must understand the design process,

The specification of seismic design guidelines is only one part of a process which has
as its objective t h e attainment of a given Ievel of strength in t h e facility. To be
reasonable and rational, t h e environmental conditions specification must be coupled
t o t h e remaining parts of t h e process.

XI.2 ELEMENT SIZING PROCEDURE

The potential effects of a n earthquake on a platform are greatly dependent upon t h e


particular characteristics of t h e superstructure and substructure elements, and t h e
local soils which act to both input energy t o the platform and absorb energy from the
reacting platform.

The example platform used in this report was designed to API guidelines for a wave
height of 100 f t and a current of 4 feet-per-second, decreasing linearly t o zero at t h e
mudline. Such conditions produce a maximum resultant base shear on this structure
of 25,000 kips. The platform w a s checked for earthquake conditions using t h e API R P
2A (1976)guidelines,

X1.3 DESIGN ENVIRONMENTAL CONDITIONS REQUIRED


TO ACHIEVE TARGET RELIABILITIES

The platform cost and example value analysis have indicated a n average annual
reliability of 99.5 to 99% (average annual probability of failure of 0.5 to 1%).

XI-2
Considering a platform lifetime of 25 years, t h e average lifetime reliability is found
t o be equal t o 70 to 78% (average lifetime probability of failure of 22 to 30%).

From t h e relationships developed between the designated probability of failure and


design load, it is indicated t h a t t h e platform needs to be designed for an average base
shear of 30,000 kips t o m e e t t h e target reliabilities.

This elastic design load is associated with a lifetime (25 years) probability of failure
of approximately 30%.

Thus, from t h e correlations between maximum ground velocity or wave height with
maximum resultant base shear and assuming the platform would be designed for
either value independently (no interaction between t h e two loading systems), t h e
following desirable design levels are indicated:

Wave Height 100 f t


Peak Ground Velocity 70 cm/sec

XI.4 GROUND MOTION TIME HISTORIES FOR


ELASTIC AND INELASTIC ANALYSES

The correlations between peak ground velocity and maximum resultant base shear
were based on three-dimensional, elastic dynamic response analyses. Three
component firm-ground recorded time histories were used in these analyses.

Therefore, t h e computed design earthquake peak ground velocity should be used to


scale firm-ground recorded time histories. The major components a r e to be scaled to
t h e design peak ground velocity and t h e two minor components a r e to be scaled
proportionally.

In using t h e design peak ground velocity, it should be emphasized t h a t it is intended


for three-dimensional dynamic analyses, where t h e input motions consist of three
components t o be input at t h e base of a three-dimensional, elastic, dynamic response
model of t h e superstructure supported on springs which simulate the range of
stiffness characteristics of t h e foundation elements.

XI-3
S T D - A P I I P E T R O I L S - E N G L 1978 W 0 7 3 2 2 9 0 Ob02409 O 5 8

XI3 RESPONSE SPECTRA FOR ELASTIC AND INELASTIC ANALYSES

If t h e response analysis of t h e offshore platform is t o be achieved through a spectral


modal analysis, t h e level of design loading needs to be specified through a design
response spectra.

The design conditions may be expressed in terms of elastic response spectra for
platform systems incorporating t o ductility or overload capacity, or in terms of
inelastic response spectra derived from t h e elastic response spectra for systems with
given ductility or overload capacity.

The inelastic design spectra may be obtained from t h e elastic response spectral
ordinates and the system ductility factor as follows:

-- 1
'A.
in. in. 2Fi -1 e

sv* =- s
I
SV.in. in, Fi e'

sD
SD = -1s
in. in. Fi De

where SA. , Sv. , and Sr, are t h e inelastic response spectral acceleration,
velocity Lkd dfalacernent i"drdinates in their respective frequency ranges
(approximately 0.1-0.3, 0.3-3.0 and 3-5 Hertz). SA , Sv and S,, are t h e elastic
response acceleration, velocity and displacement or%nat& in theesarne frequency
ranges. p is t h e system ductility factor defined in l i g u r e XI.1.

The analysis carried out in this report is based on the assumption t h a t the platform
system has a ductility factor of 2. The results of such an analysis are equivalent t o
t h a t utilizing a load factor (factor-of-safety) of 2. As wiIl be shown l a t e r in this
report, results of pseudo-static inelastic analysis have indicated t h e appropriateness
of this concept.

XI.6 DESIGN RESPONSE SPECTRA AND EFFECTIVE GROUND ACCELERATIONS

There are several approaches t o t h e dynamic analysis of offshore platforms subjected


t o earthquake loading conditions. One of the most commonly used approach, one t h a t
we a r e concerned with here, is the spectral analysis method.

X 1-4
.,. ... .. -
. ... - .
::
I
S T D - A P I / P E T R O 1 1 7 - E N G L 1978 D 0732290 ObO2LiLO 87T

The method can be best defined as t h e weighted superposition of t h e response of


several single degree of freedom (SDOF) systems.

The development of t h e dynamic response using t h e spectrum method begins with


idealizing t h e structure into a lumped mass multiple degree of freedom (MDOF)
system.

The force equilibrium equations for each of the system's masses, written in a matrix
form, a r e t h e same simple basic equaitons that control the single degree of freedom
(SDOF) system. The devefopment then proceeds t o determine t h e undamped, natural
frequencies of the system. Then, t h e natural mode shapes of t h e system a r e
developed. These mode shapes are the shape the system would vibrate in if i t were
vibrated at a given natural frequency.

After the system's natural frequencies and mode shapes, the participation factors for
the system are developed. These participation factors a r e the constants which
determine the amount that a given mode shape will contribute to t h e total motion of
t h e system.

The response then may be treated as t h e weighted superposition (by some given
method) of the response of SDOF systems having frequencies equal to the natural
frequencies of the MDOF system. The weighting factors are the natural mode shape
and the participation factors.

The development and application of t h e response spectra relationships a r e presented


in Appendix A.

Generally, two methods of superposition of the individual spectral response a r e used.


The simple superposition method where all the absolute values of t h e individual
spectral responses a r e added (upper bound estimate) and the root mean square (RMS)
method where the maximum response value is given by t h e square root of t h e sums of
t h e squares of t h e individual spectral responses.

A primary requirement for a spectral response analysis is the response spectrum


itself. The response spectrum is defined as the maximum responses of many single
degree of freedom oscillators having different natural periods, but the same degree of
internai damping.

XI-5
- .
..
The original concept of response spectrum was developed by Biot (1943) and l a t e r
developed by Housner (1959).

. -
To develop a %pectrumt', a SDOF system with a given coefficient of damping and
period is subjected to a n earthquake acceleration record and t h e maximum value of
its response is determined. Next, the period is changed t o some other value and t h e
s a m e process repeated (damping and the acceleration record kept t h e same). When
we have computed enough points, these can be plotted on a graph whose horizontal
scale is t h e natural period and whose vertical scale is t h e maximum value of t h e
response (acceleration, velocity or displacement). When these points are connected,
we have constructed a response spectrum. This same process may be repeated for
other values of damping. From this discussion, we can see t h a t the earthquake
response spectrum is a plot showing t h e maximum value of response for a SDOF
system (having a given value of damping) versus t h e natural period of vibration when
its base is subjected t o a given acceleration-time history.

To simplify t h e inter-relationships of deflection, velocity, and acceleration, two


principal response characteristics were defined: t h e Spectral Velocity and t h e
Spectral Acceleration (sometimes called t h e pseudo-velocity and t h e pseudo-
acceleration). The relations between these quantities are:

where:
'A = Spectral Acceleration

Sv =
Spectral Velocity
SD = Spectral Displacement
T = Period

The spectral velocity is nearly equal t o t h e t r u e maximum relative velocity for


systems with moderate or high frequencies, but i t may differ considerably for very
low frequency systems. The spectra[ acceleration is exactly equal t o the t r u e
maximum acceleration f o r systems with no damping, and is not greatly different for
moderate amounts of damping common t o structures. These definitions have been
used primarily for computational and interpretive ease and justified on the basis t h a t
they a r e sufficiently close to the exact values (410 percent) for t h e earthquake
analysis of most structurai systems.
XI-6
-.
L- c

~ ~~
S T D * A P I / P E T R O 117-ENGL 1778 D 0 7 3 2 2 7 0 Ob02412 b y 2 D

In general, the spectral shape is a measure of the frequency content of an earthquake,


while t h e spectral amplitudes are a measure of t h e energy (or intensity) content of
such earthquakes.

The concept of response spectra as defined above becomes a convenient way to


specify seismic design conditions such as those of the Nuclear Regulatory Commission
(NRC), American Technology Council (ATC), and the American Petroleum Institute
(API RP 2A).

These criteria a r e expressed generally in terms of normalized response spectra and


associated scaling factors. As explained above, the combination of these two
elements represent the design loading level.

The rest of this section is going to be devoted to explaining the basis for the
development of the design response spectra of t h e API RP 2A (1977).

The API R P 2A utilizes the seismic zoning concept t o describe the relative intensity
of ground motions in a given geographical zone. Seismic zoning maps of t h e United
States coastal areas developed by t h e U. S. Geological Survey (Algermissen and
Perkins, 1976) and the Applied Technology Council (1977) were considered by API.

In this map (shown in Figure XI.2) t h e United States coastal waters a r e divided into
six zones, and the relative siesmicity in each zone is characterized using a factor of O
through 5, t h e relative siesmicity being higher for higher factor numbers.

The seismicity in each of t h e zones indicated above is characterized using a scaling


coefficient (effective ground acceleration). The values of these coefficients are:

.. .
..
<
. z=o 1 2 3 4 5
G = O 0.05 o. 10 0.20 0.25 0.40

where: Z = Zone or relative siesmicity factor from seismic risk map (Fig.
~.
i.
. .
vI.2)
G = Design coefficients (termed effective ground accelerations t o
scale associated response spectra or time histories) expressed as a
ratio of gravitational acceleration.

.. . .
XI-7
STD-APIIPETRO 117-ENGL 1978 0 7 3 2 2 7 0 Ob02413 589

Along with this seismic risk map and t h e associated design coefficients, t h e API
provides response spectra t o characterize t h e distribution of energy with frequency.
The response spectra shown in Figure VI.2 are referenced to three types of local soil-
geological conditions defined as follows:

Rock - crystalline, conglomerate, or shale-like materials generally


having shear wave velocities in excess of 3000 ft/sec (914 m/sec).

Shallow Strong Alluvium - Competent sands, silts, and stiff clays having
shear strengths in excess of about 1500 psf (71.8 kPa), limited t o depths
of less than about 200 f t (61 m), and overlying rock-like materials.

Deep Strong Alluvium - Competent sands, silts, and stiff clays with
thicknesses in excess of about 200 f t (61 m) and overlying rock-like
materials.

These spectra consist of t h r e e straight-line segments on a log-log scale as shown in


Figure XI.3.

The API R P 2A 1977 recommended design spectral shape (normalized spectra) were
based on results of a study reported by Seed, et al (1976 a, b).

In this study, statistical analyses were performed on strong ground motion records
obtained from t h e United States and Japan. Normalized mean, mean minus one, and
mean plus one standard deviaiton cf these records grouped into four soils subsurface
conditions were obtained. The soils subsurface condition groups were rock, shallow
strong alluvium and deep strong and deep firm ailuvium.

The API 1977 spectral shape was based on t h e mean plus one standard deviation
shallow strong alluvium normalized spectrum of this study (Figure XI.4). This
spectral shape is t h e same as t h a t by ATC (3).

The scaling horizontal ground motion acceleration coefficients were selected such
t h a t t h e design loads resulting from t h e scaled design response spectra m e e t t h e
t a r g e t reliability (99 percent annually).

XI-8
... ... .. . ... ... S T D - A P I I P E T R O 1 1 7 - E N G L 1778 H 0 7 3 2 2 9 0 Ob02414 415 H
.... ..
. ... ... .,

.:. This selection was performed using t h e results of t h e three-dimensional elastic t i m e


history analyses of t h e example platform and simple MDOF systems t o model
potential platforms similar t o t h e one analyzed.

For example, t h e results of t h e three-dimensional elastic time history analyses have


indicated t h a t for t h e Gulf of Alaska environment, a platform needs to be designed
. .I”:..
. .. ~

for an average maximum resultant shear of 30,000 kips t o m e e t t h e t a r g e t annual


reliability of 1 percent.

This resultant base shear (BSR) represents t h e combined effect of two component
base shear in t h e two principal horizontal directions (BXS and BSY) as follows:

BSR =*BSX)2 + (BSY)*

Generally, t h e two horizontal base shear components may be related using t h e rule of
two thirds:

BSY = 2y BSX

then:
BSR =d(BSX12 + ($ BSX)2

BSR = 1.202 BSX

Therefore, f o r a one-dimensional loading component, t h e design maximum shear may


be obtained as:
*.

RD = BSX = BSR/1.202 . .

RC = 30,000/1.202 = 24,958 kips

‘Assuming t h e r e is a good balance between t h e expected earthquake and wave


loadings, t h e target reliability may be m e t then by selecting design earthquake
loadings capable of inducing maximum base shear of 24,958 kips.

-.
iL- .c XI- 9
~- ~~ ~ ~

S T D - A P I I P E T R O LIS-ENGL IS78 W 0 7 3 2 2 9 0 Ob02415 351

For cases where t h e f i r s t fundamentai mode response represents t h e major portion of


t h e platform modal response, t h e base shear may be expressed in terms of t h e product
of t h e mass and spectral acceleration ordinate a t the periód of this first myode (SDOF
system):

BSX = RD = M(Sa) = M (a' x ) G


T1

where: R = design maximum base shear


D
(S 1 = spectral acceleration ordinate at first fundamental mode
a period, T
'G
=a 1
= normalized
1
spectral acceleration ordinate at first
T1 fundamental mode period, TI
G = scaling coefficient (effective ground acceleration)

For the example platform t h e total mass of t h e system was indicated t o be 105,100
kips and the first fundamental mode to be 1.95 sec. The API (1977) normalized design
spectra indicate a value of Sa of 0.59 at a period of 1.95 sec., then
-
G
24,958 = 105y100 (0.59) G
g

G = 0.402 g

-.
If we a r e t o assume t h a t t h e f i r s t mode response represented 90 percent of t h e t o t a l
response of t h e modes, then

24,958 = 105y100 (1.1) (0.59) G


g

G = 0.37 g

XI-1 o
-.
.L .c

- .
~~ ~
If we a r e to consider t h e first three modes response of t h e system (three degrees of -
freedom systems), then from t h e computed fundamental periods of t h e system and t h e
API i977 normalized spectra, t h e spectral acceleration-ät each of these periods may
be written as:

Mode Period (sec) 'a/c


1 1.95 0.59
2 0.72 I .4
3 0.52 1.85

It could be shown t h a t the participation factors of these modes are:

-
Mode Participation Factor
1 O. 622
2 O. 333
3 O . 034

and that ?he natural mode shapes of this three degrees of freedom system:

Then t h e response of this system may be obtained using root mean square (RMS)
method to combine t h e modal responses as:

2 2
RD = MGd(0.622 x 0 ~ 5 9+) (0.333
~ x 1.4)* + (0.034 x 1.85)
_ .
Solving this equation yields t h e following value of t h e G factor:

G = 0.398 g

The above computations indicate that a scaling coefficient of about 0.40 g associated
with normalized spectrum (soil type B) of t h e API 1977 would produce the level of
design loading t o meet t h e target reliability.

XI-1 1
-. iL.- .c
*
S T D * A P I / P E T R O 1 1 7 - E N G L 17-78 9 0 7 3 2 2 9 0 Ob02417 124 =

Ry= Rc

lflrnox
\

DISPLACEMENT A

Ry = R g =Rc
/-
-/- -
u I/
z
a se

DISPLACEMENT A

rIc NON-DUCTILE, A N D INELASTIC DUCTILE


li LOAD-DEFORMATION SYSTEMS
1 -
- .
.% WOODWARD-CLYDE CONSULTANTS
- .
S T D * A P I / P E T R O 119-ENGL 1978 W 0 7 3 2 2 9 0 Ob02q1A Ob0

..
I I I l i l I l I 1 I l 2
- I I
5 Percent of -
21" I-
Critical Damping
-
I!

2.0 -

1.0
--
-
-
0.5 -
- SA = Spectral Acceleration ar
Base of .Structure
-
-
sv -- T
2n SA = Spectral Velocity

0.2 T2
SD = - SA = Spectral Displacement
- 4n

I I I 1 I i i i I I I I I l
I). 05 o. 1 9.2 0.5 1.c 2.0 5.0
Period - T - Seconds

FIG. XI, 3 RESPONSE SPECTRA, SPECTRA NORMALIZED TO


GRAVITY

WOODWARD-CLY DE CONSULTANTS
-.
L_ %

- .
~
. . . .
.. ,,I.',
..... .
i
...
.
-:.
S T D - A P I / P E T R O 119-ENGL 1978 W O732290 Ob02420 719 =

v>
n
z
O
u
W
v,
I
n
-Oa
W
II

- ..
- O
. .

WOODWARD-CLYDE CONSULTANTS
-.
L -%

-
I

.
- - .
~

. .
, _
S T D . A P I / P E T R O I L S - E N G L 1778 0 7 3 2 2 9 0 Ob02Li2L b 5 5

XII CLOSURE

XII.1 SUMMARY
- .
This report h a s summarized and documented a study done t o develop earthquake
ground motion descriptions appropriate for design of template-type platforms for t h e
eastern Gulf of Alaska. The results of this study were expressed as t h e effective
ground acceleration, G, for API Zone 5 , and as t h e response s p e c t r a ordinates
appropriate for two soil conditions (shallow and deep firm alluvium) a n d for structures
having fundamental periods in t h e range of 1 to 2 seconds.

The API RP 2A earthquake provisions consist of t h r e e basic components of design


criteria; characterization of earthquake ground motions for use in t h e design of fixed
offshore platforms (environmental criteria), description of analytical procedures to
determine forces induced in t h e platform system by ground motions, and guidelines to
size elements and joints for these computed forces.

T h e study assumed t h a t t h e design ground motion characterization was t h e sole


degree of freedom of change in t h e API guidelines. The provisions dealing with t h e
basic analytical procedure (elastic response spectra, specified major horizontal
component with other horizontal being two-thirds of this value and t h e vertical, one-
half of this value, RMS method of mode superposition, etc.), member sizing and
tubular joint design guidelines, and ductility requirements (double deflection
requirements leading t o a ductility in excess of 2 for t h e structure) were assumed to
remain unchanged.

Thus, t h e intent was to consider a conventional template-type platform sized using


current API guidelines, and then for t h e oceanographic and earthquake environment
of t h e eastern Gulf of Alaska, determine t h e scaled elastic response spectra ordinates
which would lead to a given strength or force resistance of t h e s t r u c t u r e (ductility
was defined by other provisions). T h e strength or force resistance desirable for t h e
s t r u c t u r e was determined based on results of reliability and value analyses.

The reliability analyses focused on t h e abilities of t h e s t r u c t u r e studied to sustain


large overloads developed by either e x t r e m e oceanographic or intense earthquake
conditions, without collapse. Collapse was defined as t h a t levei of loading (and its

XII-1
- ...-
i-
~- ~ ~~
~

STD.API/PETRO L I S - E N G L L778 0 7 3 2 2 7 0 ü b 0 2 4 2 2 591

associated deformation) at which t h e structure could support no further increases in


loading (or deformation) and which was accompanied by extensive permanent damage
within the structure which would render it unserviceable. Such a state was generally
characterized by formation of mechanisms in t h e legs and piles, extensive buckling of
diagonal and horizontal braces, and excessive permanent displacements in t h e
structure.

The value analyses attempted t o recognize important tangible and intangible costs
which were projected to be associated with platform operations, design, and
performance. First costs which reflected the costs of achieving given levels of design
strength (resistance and ductility) were balanced against risk costs associated with
t h e same levels of design strength. The objective was t o find t h e range of
reliabilities which would result in minimizing the total first and risk costs (maximize
benefits). The reliability range associated with maximized benefits defined a range
of desired strength for t h e platform.

in addition, results of previous studies of t h e computed or analytical reliabilities


associated with present selfcontained platform drilling and production operations in
t h e Gulf of Mexico were used t o define t h e range of reliabilities which t h e general
industry had found to b e acceptable (associated with current API guidelines for t h e
Gulf of Mexico). Thus, both code-calibration based reliabilities and value cost-based
reliabilities were used to define t h e desirable range of reliability.

For t h e eastern Gulf of Alaska operations, t h e reliability of the platforms designed


for a given strength were taken to be a linear combination of t h e reliability of t h e
platform in the oceanographic environment and of t h e platform in t h e earthquake
environment. Simultaneous occurrences of significant oceanographic and earthquake
loadings were ignored. Also, significant deterioration in t h e strength of t h e platform
due t o fatigue effects were ignored (assumed t h a t API member and joint sizing
guidelines, inspection and repair practices would negate such effects). The reliability
which was achieved for a given design strength thus reflected the effects of forces
and deformations induced by both wave, wind, and current loadings, and ground
motions.

XII-2
-
The platform strength can b e associated with a given magnitude of elastic design
force. The elastic design force could thus be determined to result in a given platform
strength (force resistance and ductility), and thus be linked with the range of desired
reliabilities. This elastic design force (actually chosen from a range of acceptable
forces) could then be translated to a non-unique combination of effective ground
acceleration and major component response spectra ordinates, given t h e API response
spectra analysis formulation.

The normalized response spectra ordinates were chosen t o correspond closely with
those developed by Seed, Newmark, Mohraz, and others, and contained in current
guidelines for design of buildings (ATC-3). The objective was to choose a set of
response spectra ordinates which would reflect t h e basic characteristics of the ground
motions which were projected for the eastern Gulf of Alaska, and which would be
generally acceptable t o the design and research communities.

The effective ground acceleration, G, which is in reality a scaling factor or design


coefficient, was then chosen t o dimensionalize t h e normalized response spectra
ordinates t o values which would result in the desired elastic design force.

XII.2 PERSPECTIVE
The. description of the eastern Gulf of Alaska earthquake environment contained in
Section 2.10 of API RP 2A consists of two fundamental parts: t h e effective ground
accelerations (G-factors) and the normalized response spectra appropriate for t h e
elastic response spectrum approach formulated by API.

The results of applying these two fundamental camponents to a given platform system
is a given magnitude and distribution of inertial loadings. These loadings when
combined with the remaining provisions of Section 2.10 and the other sections of API
RP 2A are intended to produce a platform. system which has desirable performance
-
capabilities force resistance and ductility or deformation capacity.

Thus, t h e combination of effective ground accelerations and response spectra a r e


fundamentally derivatives of t h e requirements to provide acceptable: safety and
preventing undesirable performance during extreme force or deformation producing

XII-3
events such as severe storms and intense earthquakes. Per se, .the characteristics of
these extreme events a r e not explicitly addressed in t h e provisions. instead,
environmental criteria (parameters and procedures used to determine forces and
deformations) and elastic analysis based structural criteria (parameters and proce-
dures used t o configure, size, and join structural and foundation elements) arexoupled
with factors-of -safety and inelastic performance provisions (e.g. double deflection
and tubular member compactness provisions of Section 2.10) t o develop a workable
design process and platform system which will provide acceptable or desirable
reliability against significant damage or unserviceability during t h e life of t h e
platform.

Consequently, in the case of Section 2.10 provisions for t h e eastern Gulf of Alaska,
t h e effective ground acceleration and t h e response spectra ordinates a r e not only
derived from t h e considerations of the earthquake and soil environment of this area,
but as well are derivatives of inelastic performance requirements of t h e structure
(levels of force resistance and associated ductilities or deformations capacities) and
if the structure sizing criteria .and prescribed analysis proceedures.

in an important way, these ground motion provisions are also influenced by t h e


oceanographic environment of t h e eastern Gulf of Alaska. The reliability of t h e
structure is a function of the effects of both force-deformation producing
environments (oceanographic and earthquake); a portion of the unreliability or
probability of failure is produced by t h e storm loadings and a portion by t h e
earthquake induced forces. Thus, for a given level of desired reliability or t a r g e t
reliability, t h e design load used t o configure and size t h e platform t o resist
earthquakes and storm loadings (not imposed simultaneously). must be such as t o result
in t h a t reliability, recognizing both force-deformation producing environments.

For example, if there were no storms in t h e eastern Gulf of Alaska, all of t h e


platform's unreliability could be attributed to t h e earthquake induced forces, and t h e
design force could be reduced. Such a reduction could be accomplished by reducing
t h e effective ground acceleration, or alternatively, by reducing t h e normalized
response spectra ordinates. Thus, t h e ground motion provisions become linked t o t h e
oceanographic environment through t h e desired resultant t a r g e t reliability for t h e
platform operations.

XII -4
-.
A - .%
. -....
....-. . .
--;.>i

The platform cost and example value analysis have indicatëd an average annual -

reliability of 99.5 to 99% corresponding to an average annual probability of failure of


. .. . .
0.5 t o 1%. Consideration of a platform lifetime of 25 years resulted in an average
lifetime reliability of 70 to 78% corresponding to an average lifetime probability of
. ...
:
failure of 22 to 30%.

On this basis, using correlations between maximum ground velocity or wave height
with maximum resultant base shear for t h e example platform and assuming the
platform to be designed for either environmental condition independently (no
interaction between the two conditions) a desirable design wave height of 100 f t and
peak ground velocity of 70 cm/sec were obtained.

However., it should be emphasized that this design peak ground velocity is intended
for three-dimensional dynamic analyses, where t h e input motions consist of three
components to be input at the base of a three-dimensional, elastic, dynamic response
model of t h e superstructure supported on springs which simulate the range of
stiffness characteristics of the foundation-soil elements.

Therefore, the computed design earthquake peak ground velocity should be used to
scale firm-ground recorded time histories. The major components ace to be scaled to
t h e design peak ground velocity and t h e two minor components are to be scaled
proport ionally. .* .

For cases in which the response evaluation is to be performed using a spectral modal
analysis, t h e level of design loading needs t o be specified in terms of a design
response spectra. This may be expressed in terms of elastic response spectra for
platform systems incorporating no ductility or overload capacity, or in terms of
inelastic response spectra derived from t h e elastic response spectra for systems with
given ductility or overload capacity.

The results of this study have indicated that for t h e class of offshore platorms studied
subjected to t h e combined wave and earthquake environment in eastern Gulf of
Alaska region, target or acceptable reliability may be achieved by designing platform
systems for design response spectra derived from API (1977) normalized response
spectra and average effective ground acceleration of 0.4 g.

XII-5
STD.API/PETRO 117-ENGL 1778 0 7 3 2 2 7 0 O b 0 2 ~ 2 b 137 =

xII.3 LIMITATIONS

Several limitations do exist relative t o t h e validity of t h e results of t h e study


summarized in this report. These pertain to:

Platform System Characteristics: Only one t y p e of platform was studied; a pile-


supported drilling and production template-type structure and foundation system.
Other systems can be anticipated to perform differently than t h e one studied here.
How differently they respond and perform will determine whether or not these
criteria are suitable for such systems. Design criteria a r e dependent on t h e type of
platform system as well as on t h e physical environment which is involved.

Furthermore, only one specific platform was studied; a 12-leg, 3O-pile, 36-well
platform for 300 f t of water, located on stiff and soft soils, supporting 35,000 kips of
drilling and production loadings. These physical characteristics will determine t h e
static and dynamic response characteristics of t h e platform. How similar these
characteristics may be t o other systems will determine whether or not these criteria
are applicable to such systems. Design criteria a r e dependent on t h e static and
dynamic response characteristics of t h e system which is involved.

Analytical Procedures: The performance characteristics projected for this platform


are based on results of analyses of elastic 2nd inelastic behavior. The results are
dependent on t h e fidelity of the analyses. In addition, t h e results are dependent on
t h e procedures which are used to configure, size, and join t h e piatform elements. In
this study, i t is explicitly assumed t h a t current API guidelines will b e used t o define
t h e structural criteria. Should t h e s e criteria change; for example, changing t h e
ductility requirements or tubular member compactness requirements, changes in t h e
environmental criteria may be necessary.

In t h e main, results from static, inelastic analyses have been used to characterize
platform overload behavior. These are t h e best results currently available. However,
they leave much to b e desired even though they are backed-up by experiences with
overloaded platforms and by results from dynamic analyses of simple inelastic
systems. Much remains to be learned about the. overload characteristics of t h e basic
elements which comprise such systems - tubular members, joints, piles, conductors,
and soiis.

XII-6
-.
L.. z
~~

S T D * A P I / P E T R O 1 1 9 - E N G L 1978 0732290 0b02Li27 O73

Oceanographic Design Loading Conditions: Conventional static, elastic methods have


been used to determine t h e loadings imposed and forces produced by storm elements -
winds, waves, and currents. Results from historical hindcast models which a t t e m p t to
develop a sufficient amount of information for projection into t h e near future have
been used. Large uncertainties in t h e results a r e indicated for both of these
components, in addition to t h e variabilities contributed by t h e unpredictability of
future storm characteristics. However, again t h e best information available at the
t i m e of t h e study have been used to characterize t h e oceanographic loading
environment. Changes in elements of this characterization; for example, finding t h a t
t h e t r u e wave forces in extreme sea states a r e only a fraction of t h a t computed using
these analytical models, will result in changing t h e design criteria.

Earthquake Design Loading Conditions: A similar, but even more uncertain picture is
developed in t h e projection of earthquake-induced forces. Fundamentally, historical
hindcast models a r e likely not sufficient in themselves, due both t o a lack of
sufficient historical data, and to a suspicioned long-period cyclicity in earthquake
activity. Uncertainties in t h e specific characteristics of ground motions for this
unique seismic-tectonic-geologic enviornment and uncertainties associated with t h e
general energy release characteristics of intense earthquakes and t h e associated
ground motions in the epicentral region are very large.

in addition, two major assumptions were made in deriving appropriate design loading
conditions to m e e t target reliability:

I) The spectral shape recommended by t h e API (1977) is appropriate to the


eastern Gulf of Alaska seismic environment; This spectral shape is
based on statistical studies of earthquake recorded motions in t h e past
40 years. in t h e main, these recorded ground motions are
representative of ground motions recorded from small to intermediate
magnitude earthquakes (5-6 Richter magnitude) and at relatively large
(20 to 60 km) epicentral distances. Furthermore, these statistical
studies were limited to accelerograms recorded in t h e onshore western
United States.

XII-7
S T D * A P I / P E T R O 1 1 7 - E N G L 1978 W 0 7 3 2 2 9 0 Ob02Li28 T O T

This study, however, is concerned with t h e offshore eastern Gulf of


Alaska region where t h e dominant tectonic environment is one of large
magnitude earthquakes at small epicentral distances.

Recent seismic exposure studies indicate significant differences be-


tween t h e spectral shape appropriate to t h e Alaskan region and western
United States. Very little is known at present concerning t h e potential
e f f e c t of offshore environment seismic transmission characteristics
relative to t h a t of t h e onshore environment.

2) Design earthquake forces were determined from three-dimensional


elastic t i m e domain analysis results and correlation between an elastic
and simple elasto-plastic one-degree-of-freedom systems. This approxi-
mation had to b e used because of our inability t o analyze t h e inelastic
dynamic response of a three-dimensional platform system. Recent
advances in t h e field (SPSS-INTRA) might shed some light on t h e
validity of such an approximation.

It is important to recognize these general limitations for they form a backdrop for
interpretation and application of t h e results documented in this report. However, one
should not become overwhelmed by t h e uncertainties, certainly not to t h e point of
contending t h a t they must b e reduced by orders of magnitude. For if one were t o
adopt such a view, platforms could not b e built for many decades. It is t h e purpose of
t h e design criteria to accommodate these uncertainties with additional strength in
t h e facilities so t h a t desirable performance is realized. As t h e uncertainties a r e
changed, or as certainties are redefined, t h e design criteria must be redefined.

XiI.4 ACKNOWLEDGEMENT

A major part of t h e analytical work presented in this report was carried out while t h e
senior author was with Shell Oil Company during t h e period 1975-1976. Many staff
members of Shell Oil Company have contributed t o t h e analytical background, and
this contribution is gratefully acknowledged.

XII-8
- .
.-..
.. . . .
......
.. . ,
~

- Additional analytical studies and interpretation and development of design strategies


and criteria were developed in 1977 and 1978 by t h e senior author ahd members of t h e
staff of t h e Ocean Engineering Division of Woodward-Clyde Consultants, Houston,
Texas.

The timely reviews and penetration discussions by t h e API Earthquake Minicommittee


provided vital direction andinput to this effort. The members of t h e committee who
participated were W. W. Mitchell, Chairman (Standard Oil Co. of California, San
Francisco, California), P. Arnold (Shell Development Co., Houston, Texas), L. A.
Boston (J. Ray McDermott & Co., inc., New Orleans, Louisiana), J. T. kick (Exxon
Co. U.S.A., Houston, Texas), and J. Kallaby (Brown & Root, Inc., Houston, Texas).
The contributions of these men are gratef ully acknowledged.

i
:

. . ..
.I

-i
~ ~ ~ ~ ~ ~

STD.API/PETRO 117-ENGL 1978 0 7 3 2 2 9 0 Ob02430 b b ô

REFERENCES

Abrahamsen, E. (19631, "Structurai Safety of Ships and Risks to Human Life," D e t


Norske Veritas Publication No. 34.

Algermissen, S. T. and Perkins, D. M. (19761, "A Probabilistic Estimate of Maximum


Acceleration in Rock of the Contiguous United Sates, U. S. Geologic Survey, Open
File Report 76-416.

Algermissen, S. T., Rinehart, W. A., Sherburne, R. W. and Dillinger, W. H. (i969),


"Preshocks and Aftershocks of t h e Prince William Sound Earthquake of March 28,
1964," The Prince William Sound, Alaska.

American Petroleum institute (19771, "API Recommended Practice for Planning,


Designing, and Constructing Fixed Offshore Platforms," API RP 2A, Eighth Edition,
April.

American Petroleum institute (19771, "API Recommended Practice for Planning,


Designing and Constructing Fixed Offshore Platforms," Proposed Additions and
Revisions to API RP 2A, Ninth Edition, November.

Ang, A. H.-S. and Cornell, C. A. (19731, "Reliability Bases of Structural Safety and
Design," Modern Concepts of Structural Safety and Design, ASCE National Structural
Engineering Meeting, Meeting Preprint 2023.

Applied Technology Council (19771, "Final Review Draft of Recommended Compre-


hensive Seismic Design Provisions for Buildings,'< Palo Alto, California, January.

Bea, R. G. (19741, "Gulf of Mexico Hurricane Wave Heights," Journal of Petroleum


Technology, Sept.; also Preprints, Offshore Technology Conference, Paper O K 21 10,
Houston, Texas, May.

Bea, R. G. (19751, "Development of Safe Environmental Criteria for Offshore


Structures," Proceedings, 1975 Oceanology international Conference, Brighton,
England.

Bea, R. G. and Doyle, E. H. (19751, "Parameters Affecting t h e Axial Capacity of


Driven Piles in Clays," Proceedings, Offshore Technology. Conference, Vol. ïï, Paper
NO. 1899, pp. 731-742.

Bea, R. G. (1976), "Earthquake Criteria for Platforms in t h e Gulf of Alaska,I1


Proceedings,
Off shore Technology Conference, Paper OTC 2675.
Bea, R. G. (19771, "Earthquake Design Criteria for Offshore Platforms," Presented at
Session No. 50, '>Probabilistic Methods in Earthquake Engineer ing,fl ASCE Annual
Convention, San Francisco, California, October.

Bea, R. G., Audibert, 3. M. E. and Akky, M. R. (19771, "Earthquake Response of


Offshore Platforms,1t Presented at Session No. 12, "Earthquake Response of Special
-
Structures State of the Art," ASCE Annual Convention, San Francisco, California,
October.

.<
- -
L
~~

S T D - A P I I P E T R O LLS-ENGL 1778 0732290 Ob02431 5 T 4 _.

Bea, R. G. (19781, "Earthquake Criteria for Platforms in the Gulf of Alaska," Journal
of Petroleum Technology, March.
Benjamin, J. R. and Cornell, C. A. (19701, Probability, Statistics, and DeciS.ion for
Civil Engineers, McGraw-Hill, New York.

Bielak, J. (19761, llModal Analysis for Building-Soil interation," Journal of t h e


Engineering Mechanics Division, ASCE, October 1976.

Biot, M. A. (19431, "Analytical and Experimental Methods in Engineering Seismology,


Transactions, American Society of Civil Engineers, 108 p. 365.

Bogard, D. and Matlock, H. (19771, "A Computer Program for t h e Analysis of Beam-
Columns Under Static Axial and Lateral Loads," Proceedings, Offshore Technology
Conference, Vol. III, pp. 581-588.

Borgman, L. E. (19631, "Risk Criteria," Journal of Waterways and Harbors Division,


ASCE, WW3.

Borgman, L. E. (19721, "Risk Evaluation in Engineering," McGraw-Hill Yearbook of


Science and Technology, New York.
Cardone, V. J., Pierson, W. J., and Ward, E. G. (19751, "Hindcasting the Directional
Spectra of Hurricane Generated Waves," Proceedings, Offshore Technology Confer-
ence, Paper OTC 2332,. Houston, Texas.

Carsten, B. (19731, "Some Thoughts on Safety and Risk," Veritas Magazine, DnV, No.
76, Vol. 19, October.
. CIT (1971-751, California Institute of Technology, Earthquake Engineering Research
Laboratory, "Strong Motion Earthquake Accelerograms, Digitized and Plotted Data,"
Vol. II, Parts A through Y, Corrected Accelerograms and Integrated Ground Velocity
and Displacement Curves, Pasadena, California.

Cornell, C. A. (19681, "Engineering Seismic Risk Analysis," BSSA, Vol. 58, No. 5,
October.

Cornell, C. A. (19721, "Bayesian Statistical Decision Theory and Reliability Based


Design," Structural Safety and Reliability, Pergamon Press. '

Cornell, C. A. and Vanmarcke, E. H. (19751, llSeismic Risk Analysis for Offshore


Structures," Proceedings, Offshore Technology Conference, Paper OTC 2350, Hous-
ton, Texas.

Coyle, H. M. and Reese, L. C. (19661, %oad Transfer for Axially Loaded Piles in
Clay,11 Journal of t h e Soil Mechanics and Foundations Division, ASCE, Vol. 92, No.
SM2.

Der Kiureghian, A. and Ang, A. H-S. (19751, A Line Source Model for Seismic Risk
Analysis, University of Illinois, St. Res. Series No. 419, (UILO-ENG-75-2023).

Donovan, N. C. (19731, Tarthquake Hazards for Buildings," Building Practices for


Disaster Mitigation, U. S. Dept. of Commerce, Nat. Bureau of Standards, BSS 46.
=
-

S T D * A P I / P E T R O 119-ENGL 1778 0 7 3 2 2 9 0 Ob02932 430

Duke, C. M. and Mal, A. K. (1975), "A Model for Analysis of Body and Surface Waves
in Strong Ground Motion," Proceedings, U. S. National Conference on Earthquake
Engineering.

Evernden, J. F. (19701, "Study of Regional Seismicity and Associated Problems,"


Prince William Sound, Alaska, Earthquake of 1964 and Aftershocks, Vol. 70, No,. 2.
Fintel, M. and Kahn, F. R.' (19681, "Shock-Absorbing Soft Story Concept for Multistory
Earthquake Structures," Preprints, 64th Annual AC1 Convention, Los Angeles, March.

Fjeld, S. (19771, "Reliability of Offshore Structures," Proceedings, Offshore Techno-


logy Conference, Paper OTC 3027, Houston, Texas, May.

Fowler, D. W., Erzurumulu, H. and Toprac, A. A. (19731, Wltirnate Strength of


Tubular Columns Under Combined Bending and Axial Load," paper presented at ASCE
Structural Engineering Meeting, San Francisco, April.

Garrison, L. E. and Bea, R. G. (1977), "Bottom Stability as a Factor. in Platform Siting


and Design," Proceedings, Offshore Technology Conference, Paper OTC 2893,
Houston, Texas, May.

Gates, W. E., Marshall, P. W. and Mahin, S. A. (19771, "Analytical Methods for

+
Determining the Ultimate Earthquake Resistance of Fixed Offshore Structures,"
Proceedin s, Offshore Technology Conference, Paper OTC 2751, Houston, Texas,

Gates, N. C. (19771, l'The Earthquake Response of Deteriorating Systems," California


institute of Technology, Report EERL 77-03, Pasadena.

Hahn, G. J. and Shapiro, S. S. (19671, Statistical Models in Engineering, John Wiley &
Sons, inc., New York.
Hasselman, T. K., Bronowicki, A. and Chrostowski, J. '719751, "Probabilistic Response
of Offshore Platforms t o Seismic Excitations," Proceedings, Offshore Technology
Conference, Paper OTC 2353, Houston, Texas, May.

Haugen, E. G. (19681, Probabilistic Approaches t o Design, John Wiley & Sans, Inc.,
New York.

Hasegawa, H. S. (1974), 'Theoretical Synthesis and Analysis of Strong Motion Spectra


of Earthquakes, Canadian Geotechnical Journal, Vol. i 1.
Housner, G. W. (19591, "Behavior of Structures During Earthquakes," Journal of t h e
Engineering Mechanics Division, Proceedings of t h e American Society of Civil
Engineers, Vol. 85, No. EM4, October.
Idriss, I. M., Dobry, R. and Power, M. S. (19751, "Soil Response Considerations in
Seismic Design of Offshore Platforms," Proceedings, Off shore Technology Confer-
ence, Paper OTC 2355, Houston, Texas, May.
.,. . ...
. .,.
.._....
...-:..
. .II. .
1 .-,
.
I
S T D - A P I / P E T R O L L S - E N G L 1778 D 0732270 0b02Li33 377

Idriss, 1. M., Dobry, R.,. Singh, R. D. and Doyle, E. H. (1976), "Behavior of Soft Clays
Under Earthquake Loading Conditions,11 ProceedinRs, Off shore Technology Confer-
ence, Paper OTC 2671, Houston, Texas, May.

International Conference of Building Officials (19731, YJniform Building Code," 1973


Edition, Whittier, California.

Joyner, W. B. and Chen, A. T. F. (19751,"Calculation of Nonlinear Ground Response in


Earthquakes," Bul. Seis. Soc. of Amer., Vol. 65, No. 5.

Kalberg, K. T. and Cornell, C. A. (19691, 5 e i s m i c Risk in Southern California," MIT


Dept. of Civil Engineering, Research Report R69-31

Kallaby, 3. and Millman, D. N. (19751, "Inelastic Analysis of Fixed Offshore Platforms


for Earthquake Loading," Proceedings, Off shore Technology Conference, Paper OTC
2357, Houston, Texas, May.
Kanamori, H. and Cipar, J. J. (19741,llFocal Process of the Great Chilean Earthquake
May 22, 1960," Phy. of the Earth and Planetary Interiors, Vol. 9, Netherlands.

Keeney, R. L. and Nair, K. (19751,"Decision Analysis for the Siting of Nuclear Power
Plants: t h e Relevance of Multiattribute Utility Theory," Social Systems Engineering
Proceedings, Institute for Electrical and Electronic Engineers, Special Issue.

Keeney, R. L. and Raiffa, H. (19761,Decisions with Multiple Objectives, John Wiiey &
Sons, inc., New York.
Kelleher, J., Sykes, L. and Oliver, J. (19731, "Possible Criteria for Predicting
Earthquake Locations and Their Application to Major Plate Boundaries of t h e Pacific
and the Caribbean," Journal of Geological Research, Vol. 78, No. 14.
King, J. R. (19711,Probability Charts for Decision Making, Industrial Press, Inc., New
York.
Kirkley, O. W. (19751, "Earthquake Response Spectra for Offshore StructuresY1'
Proceedings, Offshore Technology Conference, Paper OTC 2356, Houston, Texas,
May.
McFarland, W. 3. (19721, "Bayes Equation, Reliability. -and Multiple Hypothesis
Testing," IEEE Transactions on Reliability, Vol. R-21, No. 3, August.

Marine Advisers, Inc. (19701, %roup Oceanographic Survey, Gulf of Alaska, Phase I
Area, Final Report, Vol. 4, Compensated Extreme Storms," A Report on Industry
Cooperative Study to Standard Oil Co. of California.

Marshail, P. W. (19691, "Risk Evaluations for Offshore Structures, Journal of t h e


Structurai Division, ASCE, Vol. 95, ST12, Proc. Paper 6981.

Marshall, P. W. and Bea, R. G. (19741, lgFailure Modes of Offshore Platforms,"


presented at the ASCE-EMD Specialty Conference on Probabilistic Methods in
Engineering, Stanford University; also BOSS '76 Conference, Vol. iI, Trondheim,
Norway, pp. 579-635.
~
~

S T D - A P I I P E T R O 1 1 7 - E N G L 2 7 7 8 W 0 7 3 2 Z 7 0 Ob0243Li 203 D

Marshall, P. W., Gates, W. E. and Anagnostopoulos, S. (19771,. "Inelastic Dynamic


Analysis of Tubular Offshore Structures," Proceedings, Offshore Technology Confer-
ence, Paper OTC 2908, Houston, Texas, May.
Marshall, P. W. (19741, "Basic Considerations for Tubular Joint Design in Offshore
Structures," Welding Research Council Bulletin 193,

1
Matlock, H. (19701, 'Correlations for Design of Laterally Loade Piles in Soft Clay,"
Preprints, Offshore Technology Conference, Paper OTC 1204, Ho ston, Texas, May.

Matlock, H., Kelly, A. E., Hudson, W. R., Dawkins, W. P. and Panak, J. J. (1967),
"Field Tests of t h e Lateral-Load Behavior of Pile Groups in Soft Clay," Report to
Shell Development Co., December.

McGuire, R. K. and Cornell, C. A. (19741, "Seismic Structural Response Risk Analysis,


Incorporating Peak Response Regressions on Earthquake Magnitude and Distance,"
MIT Dept. of Civil Engineering Research Report R74-51.
Mert, H. A. and Cornell, C. A. (19731, "Seismic Risk Analysis Based on a Quadratic
Magnitude-Frequency Law," Bul. Seis. Soc. of Amer., Vol. 63, No. 6 .

Milne, W. G. and Davenport, A. G. (19691, "Distribution of Earthquake Risk in


Canada," Bul. Seis. Soc. of Amer., Vol. 59, No. 2.
Mondkar, D. P. and Powell, G. H. (19751, "ANSR-1 General Purpose Program for
Analysis of Nonlinear Structural Response," Earthquake Eng. Res. Center Report No.
7537, University of California, Berkeley.

Mohraz, B. (1976), "A Study of Earthquake Response Spectra for Different Geological
Conditions," Bul. of Seis. Soc. of Amer., Vol. 66, No. 3, June.

Moses, F. ( 19761, "Reliability of Structural Systems," Proceedings, BOSS '76, Trond-


heim, Norway, Vol. 1, 1976.

National Oceanic and Atmospheric Administration (19751, World Earthquakes 1961 -


a Present, published bi-monthly since Jan. 1, 1961.
NRC (19771, National Research Council Draft Report from. t h e Panel on Earthquake
Problems Related to the Siting of Critical Facilities of the Committee on Seismology.

Page, R. A. (19751, "Evaluation of Seismicity and Earthquake Shaking a t Offshore


Sites," Proceedings, Offshore Technology Conference, Paper OTC 2354, Houston,
Texas, May.

Penzien, J. (1970), "Soil-Pile Foundation Interaction," Earthquake Engineering,


Prent ice-Hall.
Perizien, J. (19751, "Seismic Analysis of Platform Structure-Foundation Systems,"
Proceedings, Offshore Technology Conference, Paper 0°C 2352, Houston,. Texas,
May.

, Penzien, J. and Kaul, J. (19721, "Response of Offshore Towers t o Strong Motion


Earthquakes,'I Earthquake Engineering and Structural Dynamics, Vol. 1.
... ...
.... ....
.. ..._.. .
t,.<

.. . . . .

Penzien, J. (19761, Wructural Dynamics of Fixed Offshore Structures," Proceedings,


BOSS '76 Conference, Vol. I, Trondheim, Norway.
Quayle, R. G. and Fulbright, D. C. (19751, "Extreme Wind and Wave Return Periods
for the U. S. CoastYf1
Mariners Weather Log, NOAA,. March.

Rainer, J. H. (19751, "Damping in Dynamic Structure-Foundation Interaction,"


Canadian Geotechnical Journal, 12.

Raisbeck, G. (1971), "Problems in the Establishment of Practical Safety Goals in


Transportation," Earthquake, Risk, Conference Proceedings, Joint Committee on
Seismic Safety to the California Legislature, September.

Richart, F. E. (19751, "Some Effects of Dynamic Soil Properties on Soil-Structure


Interaction," Journal of Geotechnicai Engineering, Proc. ASCE, Vol. 101, No. GT12,
December.
Rosenblueth, E. (19741, "Evaluation of Life Loss Impact," An Informal Discussion,
Probabilistic Methods in Engineering, ASCE-EMD Specialty Conference, Stanford
University, California.

Sagan, L. A. (19721, "Human Costs of Nuclear Power," Science, Vol. 177, August.
Sarpkaya, T. (19761, 'Vortex Shedding and Resistance in Harmonic Flow About Smooth
and Rough Circular Cylinders at High Reynolds NumbersY1l National Science
Foundation, Februar y.
-
Schnabel, P. By Lysmer, J. and Seed, H. B.(19721, "SHAKE A Computer Program for
Earthquake Response Analysis of Horizontally Layered Sites,l' Report No. EERC 72-
12, University of California, Berkeley.

Seed, H. B., Murarka, R., Lysmer, J, and Idriss, I. M. (1976a1, "Relationships of


Maximum Acceleration, Maximum Velocity, Distance from Source and Local Site
Conditions for Moderately Strong Earthquakes," Bul, Seis. Soc. of Amer., Vol. 66, No.
4, August.

Seed, H., Ugas, C., and Lysmer, J. (19761, , "Site Dependent Spectra for Earthquake-
Resistant Design," Bul. Seis. Soc. of Amer., Vol. 66, No. 1, February.

Shibata, A. and Sozen, M. A. (19771, "Substitute-Structure Method t o Determine


Design Forces in Earthquake-Resistance Reinforced Concrete Frames," Proceedings,
Sixth World Conference on Earthquake Engineering, New Delhi, India.

Starr, C. (19691, ''Social Benefit Versus Technological Risk," Science, Vol. 165,
. . September.
.."
. -._... ...:.
I Starr, C. (i9711, ''Social Benefits vs. Risk," Earthquake Risk, Conference Proceedings,
Joint Committee on Seismic Safety t o t h e California Legislature, September.

Sweeter, V. L., Wylie, E. B. and Richart, F. E. (19741,50il Motion Computations by


Characteristic Method," Journal of the Geotechnical Engineering Division, ASCE, Vol.
100, No. GT3.

Swanger H. and Boore, D. (19741,T h e Surface Wave Contribution t o Design Motion,"


A Report to Shell Oil Company, November.

-.
.c-.c
= =
~

S T D - A P I I P E T R O 1 1 7 - E N G L 1978 0 7 3 2 2 9 0 0b0243b OBb

Sykes, L. R. (19711, "Aftershock Zones of Great Earthquakes, Seismicity Gaps and


Earthquake Prediction for Alaska and the Aleutians," Journal of Geological Research,
Vol. 76, No. 32.

Tang, W. H. and Ang, A. H.-S. (19731, "Modeling Analysis and Updating of


Uncertainties," ASCE National Structural Engineering Meeting, Preprint 2016, April.

Thom, H. C. S. (19731, "Extreme Wave Height Distributions Over Journal of


t h e Waterways, Harbors and Coastal Engineering Division, ASCE.

Trifunac, M. D. (19711, "Response Envelope Spectrum and Interpretation of Strong


Earthquake Ground Motion, Bul. Seis. Soc. of Amer., Vol. 61.
Trifunac, M. D. and Brady, A. G. (19751, W n the Correlation of Peak Acceleration of
Strong Motion with Earthquake Magnitude, Epicentral Distance and Site Conditions,"
Proceedings, U. S. National Conference on Earthquake Engineering, EERI.

U. S. Department of Commerce (19661, "Seismicity of Alaska," The Prince William


f l Vol. 1, 2 and 3.
U. S. Geological Survey (1973), "Draft Environmental Statement for the Proposed
1974 Outer Continental Shelf Oil and Gas General Lease Sale, Offshore Louisiana, U.
S. Geological Survey, Department of the Interior.
Vanmarcke, E. H. (19771, "Risk Analysis for Offshore Structures," Lecture Notes for
Course 1.75 S, Massachusetts Institute of Technology, Sumemr Session, June.
Veletsos, A. S. (19691, "Maximum Deformations of Certain Nonlinear Systems,"
Proceedings, Fourth World Conference on Earthquake Engineering, Santiago.
Veletsos, A. S. and Verbic, B. (19731, "Dynamics of Elastic and Yielding Structure-
Foundation Systems," Proceedings, Fifth World Conference on Earthquake Engineer-
ing? Rome.
Veletsos, A. S. (19751, "Dynamics of Structure-Foundation Systems," Proceedings,
Symposium on Structural and Geotechnical Mechanics, University of Illinois, Urbana.

Veletsos, A. S. and Newmark, N. M. (19601, "Effect of inelastic Behavior on t h e


Response of Simple Systems t o Earthquake Motions," Proceedine, Second World
Conference on Earthquake Engineering, Tokyo, Japan.

Wiggins, J. H. (19721, T h e Balanced Risk Concept: New Approach t o Earthquake


Building Codes," Civil Engineering Magazine, ASCE, August 1972.

Wiggins, J. H. (19741, "Budgeting Justification for Earthquake Engineering Research,"


Probabilistic Methods in Engineering, ASCE-EMD Specialty Conference, Stanford
University, California.

Wilson, A. G. (19731, "Design of Large Buildings for Safety and Health," Building
Research Division, National Research Council, Ottawa, Engineering Journal, April.
~~
~
~

S T D - A P I / P E T R O 1 1 7 - E N G L 1778 0 7 3 2 2 7 0 Ob02437 T12

APPENDIX A

DEVELOPMENT AND APPLICATXON OF RESPONSE SPECTRA

RELATIONSHIPS FOR SDOF AND MDOF SYSTEMS


~
~ ~~ ~
~ ~

STD.API/PETRO 117-ENGL 1778 D 0732290 Ob02438 757 D


AO DEVELOPMENT AND APPLICATION OF RESPONSE SPECTRA
RELATIONSHIPS FOR SDOF AND MDOF SYSTEMS

Y FW

Definitions:

P... .......External Force


W.. ........System's Weight
k..........System's Stiffness Coefficient

C..........Syctem's Damping Coefficient

X..........Deflection from Reference Axis (consider o n l y horiz.)

Fc.........Damping Force (Assume proportional to velocity, X)

F, .........Stiffness Force (Assume proportional to deflections, x)

F, .........Inertial Force (iiesistance to accelerations, 2)

m.....,....Mgss (Equal to W/g; g - acceleration due to gravity)

Force System:

F, : CX c dx/dt ( t - time)
F, : kx

F, :: mx : m d2x / d tZ

L- .I
- *
~ ~~~~ ~~

._ .. . . .
.:,_ -.
... . . ... .... S T D - A P I / P E T R O 119-ENGL 1978 m 0732290 U b 0 2 4 3 9 895 m

~.

.-
Force System Equilibrium:

2 F Horii = 0 : F, -t 5 + F , - P

ZFVert O = FW - W (Assume no v e r t . motion)

... . mus from Z F,, ,


mx + cx + kx = P
This is t h e governing e q u a t i o n of motion for t h e s i n g l e degree-of-
freedom ( x ) , e l a s t i c (k), damped ( c ) , system.

' Free Vibrations: (no e x t e r n a l f o r c e , P, a c t i n g on t h e system.)

mx + cx + kx = 0
For no damping:

mx f k x = O which has a s o l u t i o n

x(t) = A COS Pt + 8 sin p t or

p z J km c i r c u l a r n a t u r a l frequency

or
f = -P
27
= n a t u r a l frequency ( c y c l e s p e r sec.)

T = 2- n= -I - - naturai period-
P - f
A,8 = C o n s t a n t s t o be e v a l u a t e d from i n i t i a l conditions ( at t =O 1

t = O

>,-
at X = X ,
i n i t i a 1 d i s p 1acement
and v e l o c i t y
i = vo -0
A = K, and 8 =
P
Thus
X(t) = x, cos pt
Y4
+- sin p t
P
_..... .. .. ."

Free V i b r a t i o n s (Cont'd.):

Now cons i d e r i n g damping


mx
.. + ci + kx = O
let ß C 1 ?mP and o b t a i n s o l u t i o n f o r

(This i s t h e most usual s t r u c t u r a l c o n d i t i o n . The s o - u t ,n is:

x ( t )= [ A , cosp t
d
+ 8, sin p t ]
d
ê - Base of n a t u r a l logs.

ß- Dimensionless c o e f f i c i e n t known as p r o p o r t i o n of
c r i t i c a l damping, o r ß = c / c cr
C cr - damping c o e f . f o r @ = I . f o r t h i s case
Ccr = 2 m p

pd
- damped c i r c u l a r n a t u r a l frequency

Pd = pJI-ßZ
For most s t r u c t u r a l systems 5 0.10
thus P 2 (approx. equals) P
.d
A, , 8, = I n i t i a l c o n d i t i o n c o n s t a n t s

at t = O, x = x 0 and x = V,
-
- A +ß
A, = X
O
8 8,
'd m o X

Thus, we have an e q u a t i o n which wiL1 d e s c r i b e the t i m e h i s t o r y of motions


for o u r system (damped, e l a s t i c , SDF), s u b j e c t e d t o an I n i t i a l d i s p l a c e m e n t
and v e l o c i t y , and no e x t e r n a l f o r c e s .
Base Movements :
Now, we will look at the condition when
the base of the system is moved rather than ~ r--i
I
placing an external force on the system or I I
I ! i
subjecting the system to an intital displacement
and velocity.
I
Consequently, giving the base of the
system an acceleration

.i

x!a

The mass will experience a total acceleration


*.
'7' = **
'P + i
Only the inertial force will be changed:

F~ = m ( X g + i )

As before 2 FH = O :.
m(l, + i ) + C i + kx = O

or rnx + Ci + kx
..
:-mxp ,

Before (top p 2) we had

m i f C i i- .kX = f'when the system was subjected


to an external force P.

These two governing equations of motion will be equivalent if we let


..
- mx,
From this development we see that the ground motion, Xg, may be replaced
by a force acting on the mass equal to the ground acceleration, ki,multiplied
by the system's mass.

This means if we can find the solution for the equation of motion for
the system subjected to an external force (as we did for no force),
then the problem of ground motions can be simply handled by translating
the ground acceleration-time history into a force-time history (by
multiplying times the m a s s ) .

. -
-
I

.
System S u b i c c t e d t o A r b i t r a r y E x t e r n a l Force Time H i s t o r y :

We w i l l now c o n s i d e r our SDF system


s u b j e c t e d t o some a r b i t r a r y e x t e r n a l f o r c e ~

which may v a r y w i t h t i m e a8 shown t o t h e r i g h t . +p

P r e v i o u s l y , (bottom p. 3) w e o b t a i n e d
t h e s o l u t i o n f o r t h e SDF system s u b j e c t e d t o t
an i n i t i a l d i s p l a c e m e n t , xo, and i n i t i a l v e l o c i t y ,
Vo, and no e x t e r n a l f o r c e .
-P

[ X,COS ß -
v a+ x sin p t ]
x ( t ) = e-?Pt p t +(
d d ma d

Now, c o n s i d e r t h e s o l u t i o n f o r o n l y an
i n i t i a l v e l o c i t y , Vo, a c t i n g :
x ( t ) = ërrt -.A
V sin p t i
Pd d
+X
This f u n c t i o n would look l i k e t h e diagram on t h e
r i g h t . I t would be a decaying motion (due t o
damping), and would have a n i n i t i a l s l o p e of
Vo - the i n i t i a l velocity.
-X

We may c o n s i d e r t h a t t h i s i n i t i a l
v e l o c i t y is caused by an impulse. An impulse
is a force a p p l i e d o v e r a s h o r t p e r i o d o f
time - any may b e i n t e r p r e t e d as t h e area
under a load-time diagram.

From p h y s i c s , w e know t h a t a system


w i l l have an i n i t i a l v e l o c i t y , Vo, t h a t i s +P
e q u a l t o t h e a p p l i e d impulse, I , d i v i d e d
by t h e mass OF t h e system. Thus:
vo= I/m

Now, we w i l l t a k e t h e load-time diagram -P


shown a t t h e t o p of t h e page and d i v i d e i t
i n t o a number of s h o r t p u l s e s . Consider one
of t h è s e as shown t o t h e r i g h t . For some
I
p a r t i c u l a r time, T , w e w i l l look a t t h e motion l

caused by t h e impulse which results from t h e I


f o r c e which is a c t i n g a t t h a t time, P (T), 1
a c t i n g f o r a s h o r t p e r i o d , dT. I +x
The impulse w i l l be: I
PíT)dT I
I(T) = P ( T ) d T â Vo= m I
And t h e r e s u l t i n g motion I
. .
* i _1 - STD.API/PETRO 1 1 9 - E N G L 1978 0 7 3 2 2 9 0 Ob02LiLi3 2Lb D

ARBITRARY EXTERNU FORCE-TIME HISTORY (CONT'D. )

Note in t h e last e q u a t i o n on t h e p r e c e d i n g page, t h e t i m e , t ' ,


is measured from t h e time t h e impulse is a p p l i e d . To-refer t h a t time
back t o t h e o r i g i n a l b a s e on which t h e f o r c e - t i m e h i s t o r y is drawn, t h e n
t = t ' + T or . tl= t - T
Now., t o g e t t h e r e s p o n s e f o r t h e g i v e n load-time h i s t o r y , we can add up
t h e r e s p o n s e s due t o t h e approximating impulses ( s i n c e s u p e r p o s i t i o n holds
e l a s t i c system) as shown below.
-

-P

..
~

S T D - A P I / P E T R O 1 1 7 - E N G L 1 9 7 8 W 0 7 3 2 2 9 0 Ob02444 1 5 2 W

ARBITRARY EXTERNAL FORCE-TIME HISTORY (CONT'D. )

Thus, by odding up t h e individual r e s p o n s e s :

And

Using on integro1 t o sum w e hove (I, = P ( T I dT)


T* t

For o g r o u n d m o t i o n , we r e p l o c e P ( T ) by
mx,(T) and pd = p ( c l o s e approximotion)
Then,

This i s known as DUHAMEL'S INTEGRAL. I t p r o v i d e s t h e t e c h n i q u e whereby


we may e v a l u a t e t h e disp1.acement of t h e system a t any time x ( t ) , when i t i s
.. m,
s u b j e c t e d t o any b a s e a c c e l e r a t i o n - t i m e h i s t o r y . Xg

IT IS VERY IMPORTANT TO REMEMBER THE PHYSICAL MEANING OF THIS


EQUATION: W e are s i m p l y a d d i n g up t h e e f f e c t s of t h e i r i d i v i d u a l impulses
which approximate t h e f o r c e - t i m e h i s t o r y t h a t r e s u l t s from t h e base a c c e l e r a t i o n -

any p a r t i c u l a r time, t, Refer back t o t h e p r e v i o u s page for h e l p i n visualizing


-
t i m e h i s t o r y ( d i r e c t t r a n s f o r m a t i o n by m u l t i p l y i n g by t h e s y s t e m ' s mass) for

t h i s principle.

Also, note t h a t only two system c h a r a c t e r i s t i c s - the circular natural


f r e q u e n c y , p , anti t h e $amping r a t i o , ß , e n t e r t h i s e q u a t i o n . Thus, g i v e n t h e s e
-
two q u a n t i t i e s and a g i v e n a c c e l e r a t i o n - t i m e r e c o r d the system's displacements
are determined.
. . .. .....
.. . . .
..
. .. ...
....
.. .. ......
~

..,
........ . .
. . - . ..

AKBITURY EXTERNAL FORCE-TIME HISTORY (CONT'D. )

With Duhamel's integral, we have the ability to handle the complex


earthquake base acceleration records. We do this by-approximating the
record with a series of very closely spaced impulses determine.the motion -
which results from each of these impulses at any time, t then we choose -
some time of interest and add up all of the motion components that exist at
..
.
.
- that time from the individual impulses. Thus, Duhamel's integral could be
L

. . .. .. . . written as follows f o r numerical analysis on a computer.

For some arbitrary time - t,,

7-01
I
Xotion which results from
SDF system at application of the nfh -
( X, ( T , ) )-impulse.

impulses.

s PECTIIAL DIS PLACEMENT, 'VELOCITY, AND ACCELERATION:


Before we proceed to the discussion of the computation of the
spectral displacement, velocity, and acceleration, it will be necessary
to return briefly to the equation we obtained for undamped, free vibration
(at the bottom of the second page):

x(t) : x, cos p t + + sin p t

Which is the equation that g i v e s the motion of the mass at any time, X ( t ) ,

Xo, and velocity, Vo -


when the system ia subjected to the initial conditions of a displacement,
and has no dampfng or external forces acting.
_ .
Thir equation i s equivalent to the one which foilows (from geometry):

where
tan 8 : x,, / ( Vo /P - .A constant
or
x(t) = zo sin ( P t + 8)
Now iuccersivaly differentiating this ;quation, we find:

h)= P [ t ocos ( p t + 811


~ ~~

STD.API/PETRO 119-ENGL L97ô 0732290 Ob0244b T25 m

SPECTRAL X , i . AND i (CONT'D.)


These l a s t t h r e e e q u a t i o n s w i l l have maximum'values when t h e
sin ( ) and c o s ( ) = 1
Or

. I

*mm = -p2 t o{or neglecting the sign] = pz t o

Thus

= = p-* M A X . DISPI .
..
xmax PXmax Or MAX. VELOCITY
x,i - pg Xma5 or MAX. ACCL. = p2* M A X , o isp~
We have c o n s e q u e n t l y d i r e c t l y r e l a t e d t h e maximum v a l u e s of X , #, and
..
X.

With t h i s background, we now t o proceed t o d e f i n e t h e d e f i n i t i o n


of a spectral q u a n t i t y as b e i n g t h e maximum v a l u e of t h a t q u a n t i t y . Thus,
w e c o u l d t a k e t h e Duhamel i n t e g r a l and determine t h e maximum v a l u e of t h e
d i s p l a c e m e n t , Xmax., f o r a g i v e n a c c e l e r a t i o n r e c o r d , and f o r a g i v e n v a l u e óf
c i r c u l a r n a t u r a l freqiiency, p , and damping r a t i o , ß - It is important t o
r e c o g n i z e t h a t p and .ß are t h e o n l y two v a r i a b l e s needed t o d e s c r i b e the
d i s p l a c e m e n t s of t h e system f o r a g i v e n a c c e l e r a t i o n r e c o r d .

Consequently, i f we w o u l d ' t a k e t h e maximum v a l u e of t h e i n t e g r a l


q u a n tit y :

T: t
f

f:O

( Note that we have dmpped out -


as b e i n g e q u a l t o :
S" or ;\
Then, t h e maximum v a l u e of t h e d i s p l a c e m e n t ( o r t h e S p e c t r a l Displacement)
would be:
SD = 1
-
P
- S"

From t h e e q u a t i o n s a t t h e t o p of the page, we c a n see t h a t Sv c o u l d be


i n t e r p r e t e d as a Spectral V e l o c i t y -
and this is how i t is d e f i n e d . A l s o ,
from t h e s e same r e l a t i o n s h i p s it follows t h a t t h e S p e c t r a l A c c e l e r a t i o n may
be d e f i n e d ,
s, = P2 so
S T D - A P I i P E T R O L I S - E N G L 1978 D 0732290 IIb02447 Yb1 m

SPECTRAL.X. X. AND a' (CONT'D.)

We have thus d e f i n e d t h e s p e c t r a l displacement -


which is t h e t r u e
?
maximum displacement of t h e system r e l a t i v e t o - t h e ground -
t h e spectral v e l o c i t
r e l a t i v e t o t h e ground, and t h e s p e c t r a l a c c e l e r a t i o n , which is t h e maximum
v a l u e of t h e a b s o l u t e mass a c c e l e r a t i o n (or n e a r l y s o ) .

To be e x a c t , t h e maximum values of v e l o c i t y and a c c e l e r a t i o n are


n o t a c t u a l l y determined. T h e . q u a n t i t y used i n s t e a d of t h e t r u e r e l a t i v e
v e l o c i t y is v e r y n e a r l y e q u a l t o t h e maximum r e l a t i v e v e l o c i t y e x c e p t f o r
v e r y low frequency systems. I t i s e x a c t l y e q u a l t o t h e maximum v e l o c i t y
if t h e maximum v e l o c i t y occurs a f t e r t h e ground m o t i o n ceases.
F u r t h e r , t h e q u a n t i t y d e s i g n a t e d as t h e spectral a c c e l e r a t i o n is
a c t u a l l y t h e maximum a c c e l e r a t i o n for a system without damping and is v e r y
n e a r l y e q u a l t o t h e maximum a c c e l e r a t i o n f o r a system w i t h normal amounts
of damping.

Summarizing:

so = -
I
P
* s,
s,= p*s0
s = O PI' so = p s,
where

- s,. r] x&T) e-JP


'* tt - r) sin[ p ( t - T ) ] d T

1.0
\

We are o f t e n I n t e r e s t e d i n t h e maximum s h e a r f o r c e developed a t


t h e base of t h e structure. This can be developed as follows:

or ka-!-
P
S,. -
P
P
- Ok S,
P'

8 Q,,, = mp S,
But p * s, = sa
. . .
. ..
SQ 1
Qmax = mSa - as would be e x p e c t e d
. .. .
. . . _ ,--
. .
1
~ ~ ~

S T D - A P I / P E T R O 127-ENGL 2 9 7 8
~

- ~~

0732270 Ob02448 8 T B
~ ~

Two Degree-of -Freedom S y 8 tem:


I r-1
I 1

! k2

I; kl

LET THIS.. ........EQUAL.. ......THIS..


EQUAL
- - - a -

7
3
- -+J

FORCE EQUILIBRIUM:

Weight ++ I
On
c

FI, +
c

FC, +
-

FKI -
-
FKz
-
- Q,
L
On Weight 4k 2 9
c c c 4 -O
FI, + Fc2 + FK, = p2 c
C

.-ON
Or 8
-r

m , i , + c, X, + k, x, - F:

TP., i,-C, XI + C, X , - k, X, + k,X, =


Force E q u i l i b r i u m (Cont'd.)

This c a n be w r i t t e n as:

.. . ...
i .

- b-

since

oA bB

db]*[B]=
+

_ .
A + d 8
o r more g e n e r a l l y

h f e r r i n g t o page 2 of Appendix B - This is t h e same e q u a t i o n as


for t h e SDF system.

Free Vibrations:

No damping:

-
1 [m] [i] + [k] [x] = 0
o r expanding
m,
..
x, + ( k , + k,) x, - k, x, = O
..
m, XI - k, XI + k, x2 = O
c
These e q u a t i o n s w i l l have a s o l u t i o n o f . t h e form

..
x, ( t ) =XI sin ( p t +fi)

..x 2 ( t ) = X sin ( p t + 4

As w a s t h e case f o r t h e SDF system, when t h e s e r e l a t i o n s are s u b s t i t u t e d


i n t o t h e two governing e q u a t i o n s , we s h o u l d g e t a n answer which is e q u a l
t o zero.

.
I

- -.
c- .I
Free Vibration8 (Cont’d.)

[-ki] XI + [k2- m2 p2] x2 = O


Or
1
[-k21 !
I O
I
L
C-kzl [kt-m, P’j j
Now

# O

Thus

S t nce

We can expand this determinant and find

Which can be solved (From Quadratic Eqn.) to yield

m2 m, mz
This last equation w i l l g i v e us two values for the undamped circular
natural frequency -
t h e s e are tb two natural frequencies of t h e system.

- .
L .c
.... ..
.... .. .
....
.. ....
=
-
.
a
~.
.,
..i' .. ..

F r e e V i b r a t i o n (Cont'd.)

As a g e n e r a l r u l e , WB w i l l f i n d t h a t a system has as many n a t u r a l


f r e q u e n c i e s as i t does degreee-of-freedom. Thus, we 'have found t w o n a t u r a l
f r e q u e n c i e s f o r a two degree-of-freedom system.

It has been t h e o b j e c t of a l l of t h e work t o t h i s p o i n t t o d e t e r m i n e


t h e n a t u r a l f r e q u e n c i e s of t h e system. h i s i s always t h e f i r s t o r d e r of
b u s i n e s s when performing anv dynamic a n a l y s i s . A f t e r t h i s is done, t h e n e x t
s t e p w i l l be t o determine t h e ' s h a p e s d e s c r i b e d by t h e system as i t v i b r a t e s
with the natural frequencies ........These are known as N a t u r a l Mode Shaoes.

I n a r r i v i n g a t t h e s o l u t i o n f o r t h e n a t u r a l f r e q u e n c i e s , i t was
assumed (and c o r r e c t l y so) t h a t t h e displacement-time f u n c t i o n s f o r t h e
i n d i v i d u a l masses would be of t h e form:

X(t) = X sin ( p t 8 a )
/
I n t h i s e q u a t i o n , X , r e p r e s e n t s t h e amplitude (maximum d i s p l a c e m e n t ) o f
motion. The v a l u e s f o r t h e X's may be determined by s u b s t l t u t i n g t h e
v a l u e s of t h e n a t u r a l f r e q u e n c i e s back i n t o t h e governing e q u a t i o n s ( a t t h e
t o p of t h e p r e v i o u s page). Thus,

As b e f o r e :

- m, p2 X, + ( k, t k,) X, - k , . X ; = O
- m2 P' xt - k, xi + k, X2 = O
Now t a k i n g t h e f i r s t e q u a t i o n and s o l v i n g for

_.

We can s u b s t i t u t e the f i r s t n a t u r a l frequency, p1 and d e t e r m i n e

And s i m i l a r l y s u b s t f t u t i n g p2 Ln t h e second e q u a t i o n

X -
2
-
I

N2
XI
Y
Now, if we t a k e XL II 1.0 w e c a n d e s c r i b e t h e n a t u r a l mode s h a p e s as shown
on t h e n e x t page,

..
STD.API/PETRO 1 1 3 - E N G L L778 0732290 ü b 0 2 4 5 2 223

t i m e of i t s maximum
d isp iacement s ) when -3

i t is v i b r a t i n g a t
its. two n a t u r a l

X i is t a k e n e q u a l t o .
1.0. This is OK s i n c e
at t h i s p o i n t we are
.
L
only i n t e r e s t e d i n

For Pl For P,
( Flrst Not. Freq.) (Second Nat. Freq.)
.,
Now, w e w i l l determina t h e motions of each of t h e masses a t
any t i m e . . . . f o r t h e f r e e v i b r a t i o n c o n d i t i o n . We will assume ( l u c k i l y
t h e mathematicians have a l r e a d y ahown i t t o be a v a l i d assumption)
t h a t t h e motion can be determined for each mass by superimposing (in
some "weighted" manner) t h e simple harmonic motion of t h e two moda shapes.

It)
Notation : Let X, =' -
Firs? Mode Amplitude of MaSs#l
X:"= Second Mode Amplitude of Moss* I
Thus,
x, ( e ) = R , X',''sin (pit + a,) + R, X, sin ( p z t t a , )

R, 0 R, are the unknown "Weighting" factors

Now, Let

[$1 P~I i ,I
xy
x:"' rxl
-..._ .. ... .. ... ..
. ....
_:- ,..:. S T D * A P I / P E T R O 1 1 7 - E N G L 1778 0732270 Ob02453 Lb5
c

Free Vibcations'(Cont'd.)

Thus, the expression can be written :


c--

At this point we hove 4 unknowns: R I , 4 , a , , a ,


These will be determined (later) from the 4 initial
conditions for our system -
2 for each mass,

Now from sin ( a + b ) = sin o cos. b + cos. o . sin b ,


expond the sin functions

NOW L e t :

C, = k l cos o, (Which is constant - not a time variable )

c, = k, cos , a1

d, = R I sin a,

d, = R 7 sin a2 .-.
L

Thus,

i
Nota t ion :

= [A.] Just using

I [:,I
=
Shapes of Modes !
Free V i b r a t i o n s (Cont'd.)

We will now e v a l u a t e t h e e x p r e s s i o n for (x) -for a g i v e n s e t


of i n i t i a l c o n d i t i o n s . I n t h i s case w e w i l l assume t h a t t h e two masses
are g i v e n i n i t i a l displacements and t h a t t h e i n i t i a l v e l o c i t i e s are zero.

Initio1 ' Displacements

For massX I : ( xJ0 ,( = O


For mass'2 1 ( , ( ~ ~ 1 =, O

At t =O sin's. = O 8 cos's. = I

Thus

Or expanding

Which can be solved for di 8 d2 to yield

dl = Nz(x,), - ( X210
H- I

N2 - N,
d, = Ni ( xilo - ( XZ).,
N, - N,

These are t h e i n i t i a l displacement p a r t i c i p a t i o n f a c t o r s . Note


f i r s t t h a t they are dependent o n l y upon t h e c h a r a c t e r i s t i c s of t h e system
( t h e y determine t h e N's) and t h e i n i t i a l c o n d i t i o n s of displacements.
Secondly, n o t e t h a t when they are s u b s t i t u t e d back i n t o t h e o r i g i n a l
e q u a t i o n s (as shown below), they are t h e c o n s t a n t s t h a t determine t h e
amount of t h e mode shapes t h a t a c t in t h i s motion....hence the name of
Particioation Factors.

For i n i t i a l displacement and no initial velocities


~ ~- ~~~

-_
....... .. .-.
. .. ... .. .. STD.API/PETRO LLS-ENGL 1978 D 0 7 3 2 2 9 0 Ob02V55 T38

Free Vibrations (Cont'd.)


Lr
A similar development can be shown for the initial conditions of the
two m a s s a subjected t o initial velocities and no initial displacements. The
solution for the two participation factors are shown below.

Initial velocities
. .
For mass *1 ( x , ) ~-= O ~ (VI )o

íor mass#2. (x2l0 = O , (v,)~

Y
\ '' =
N, CV, lm + ( V z ) o
NI - N
--
F
1
(-)
p2

These are the initial velocity participation factors. And, as


before:

Now, since superposition of effects still holds, we can determine the


motion of the system resulting from conditions of initiai displacements and
initial velocities by addition of the two expressions:

[x] := [ I ] + [x] ' = Displacement of system a t cny


Vo
t i m e t ,due to initial condition
of dispiocement and velocities
of the mosses

Y
Forced V l b r a t l o n s of TDF Systems: (assuming no damping)

The s o l u t l o n t o t h e problem of t h e f o r c e d v i b r a t i o n (P D O) is
a r r i v e d a t i n t h e same manner as f o r the s i n g l e degree-of-freedom system.
T h a t Is, w e assume t h a t t h e f o r c e - t i m e h i s t o r y a c t i n g on each mas8 c a n
be r e p l a c e d by a s e r i e s of Impulses. We t h e n l e t each o f t h e impulses
g e n e r a t e a n i n i t i a l v e l o c i t y c o n d i t i o n (which we a l r e a d y have a s o l u t i o n
f o r ) , and sum t h e r e s u l t i n g r e s p o n s e s (by i n t e g r a t i o n ) .

The s o l u t i o n which r e s u l t s i s as f o l l o w s :

dT

r-O

This solutiotion assumes thot the t i m e wise variation of the forces


a c t i n g on the two masses is the same, ¡.e.,

/r,or/(c.
f (t ) - Describes the t i m e - w i s e variation

i f they ore not the same, the solution i s obt'ained by super


position of t w o seporate solutions.

Note t h a t the
iT=
T
(
=
) expression
~ for ( @ =
is the same as before
O- no damping )

Thus, p,
Ta0
T i t

i s the Duhamel i n t e g r a l for undamped vibrations


STD.API/PETRO 117-ENGL 1978 0732290 Ob02457 A00 D

Forced Vibrations of TDF Systems: (Cont'd.)


- The equation shown at the top of the previous page can be
simplified as follows:

L--

..

These are the undamped, forced vibration participation factors.

And
TZ Jdescribes time-wise variation of forces
c p, f ( T J sin [p (t-T)] dT
. .
t=O

These are interpreted as the Instantaneous Amplification Factors (IM)


for single degree-of-freedom systems having frequencies p1 and p2.

For the ground motion condi tion


.. . _ .
Pei = m, x g ( 9 )
..
Pen = m, x g ( t )

A nd

ZGMI = - NI -- N1
N,
.\
Which are the ground motion Participation Factors
J .
............
.. . . .
-

. . STD.API/PETRO 1 1 9 - E N G L 1978 W 0 7 3 2 2 7 0 O b 0 2 4 5 8 7 4 7
.
.
. .. ..,

Forced Vibrations of TDF' Svstema (Cont'd.)

The corresponding Duhamel Intemals for ground motion (undamped) are M


followr:

GMF - Ground Motion Factor


Consequently ,
Motion of TDF
System due to
Ground Motions

\
Or ex pond i ng
c

Thu8, with these last two equations we can compute the relative
displacementa of masa #I or masa #2 at any time, t, due t o a prescribed
ground motion having an acceleration*time history, Xg (t).

The GME"8 are the saaie factors found f o r the displacement history
of a SDF systam,..,,using Duhatuel's Integral.

There last aquatitma for the displacement-time history of a TDF system

,..
can be interpreted a8 a "weighted" superpodition of the responses of two
8 ingle degree-of-freedom (SPP) ayateras The "weighting" factors ara the
Ground Motion Participation Factor8, , .. , and the Natural
Mode S h a p a a . l . . . . . . , ~ . , ~ . ~ ~ ~ o . ~ o o ~ , o ~ ~ ~ ~
THIS IS A MûST IMPORTANT CONCEPT,
The m a x i m m value o f the relative displacements of tha two masse3
is of primary importance ta the engineer, The expressions above give the
displacemant at any time, t. Many succeseive tries could be made during
the response period ta determine the maximum displacement values. However,
senerally this i s not practical, Consequently, approximate techniques ara used
to determine the maximum response valuea.
.. .
. . .- . ..
...... ... .. . S T D - A P I I P E T R O 1 1 7 - E N G L 1978 m 0732290 Ob02459 b83 m
.. . . .. -. .. .

Forced Vibrations of TDF Systems: (Cont'd.)

Noting that.:
ot frequency pl

. . (- y
- (SO),
S V )-
Pl

Then an upper bound to the maximum displacements


would be

( S o )1 = Spectral Displacernent ot p, or f, = -Pl


2n

(So l2 = Spectral Displacement 'at pl or f2 = -p2


27

A good s t a t i s t i c a l approximation (Valid if maximum volues úistributed


randomly with time )

Thus, we have developed an-approximate method for finding the


maximum relative displacements of a TDF system subjected to an arbitrary
ground motion (for no damping) .....
which is a "weighted suamation of the
responses of SDF systems having natural frequencies equal to those of
the TDF system. Further, this is a good approximation for "lightly" damped
( f i - 1 0 ~systems.
~)
~~
~

S T D - A P I I P E T R O ILS-ENGL 1 9 7 8 U 0 7 3 2 2 9 0 ObU24bO 3 T 5

TDF Systems S u b j e c t e d to Ground Motions:

SUMMARY

T h u s , in summary the steps are as follows:


( For ground motions)

I. Find system's natural frequencies ( p.3)

57 4 ( f , = Q 277 f,=

2 Find systems natural mode shape ( p . 4

3 Find ground motion participation factors

' NI- I
tGM2 =
NI - N2

4 Find system's approximate maximum relative displacements

(SO), , SO)^ - Spectral displacements


for given ground motions - elevated a t frequencies f, 8 f,

Or more. generally

-.
c .c
- .
, _.
. .
. . :p.
... _>. .
~_.e
..,. .: , ... .

b F Systems Subjected to Ground Motions: (Cont'd.)


5. Similarly, the approximate maximum velocities and
accelerations

(These latter three equations are all upper bound expressions.)


Thus, we are obtaining good estimates of the maximum values of
displacements, velocities, and accelerations of each of the
system's masses .......
by the "weighted" superposition of the
responses of SDF systems having equivalent natural frequencies
the "weighting" factors are functions of the natural mode shapes
...
of the system.

Multiple Degree-of-Freedom (MDF) Systems Sublected to Ground Motions:

The reader can visualize that i t will be an extension of the


dewl.opatent for a TDF system to handle MDF systems. The mathematics
are much more involved, but the techniques are the same.

Def in¡t ions:

[m] = MOSS MO
A.

N 8 No. of masses I No. of Degrees-of-Freedom

Note: M has non-zero terms only on diagonal in this case.

IX I
1x1
= Displocement Vector =

= Velocity Vector
rxI
*2

x3
IX I = Accelerot.ion Vector
~

~~

S T D - A P I / P E T R O Z Z 7 - E N G L 1978 R 0 7 3 2 2 9 0 ObCI2Lib2 178 W


. ...

MDF Systems Subjected to Ground Motions: (Cont'd.)


Dcf inítlonr:
[k] = Stiffners matrix =

'
kn; knn

k - Motrix is symmetrical about the diagonal


k2,- Interperted os load o t point *2 caused by a
unit deflection of point $4 3 ( Ibdn = %
' = I ,,t#3 )
Free Vibrations (no damping) - object is to find the natural frequencies
of the system:

As before we take the equation of motion (in matrix form):

And aseume a form o f motion for the system which will satisfy this
equation:
AS befüre-x, = X n sin ( p t + * )
The fol lowing results

= P* = The unkown circular natural frequencies


Or

This last equation says that there are some factors, , which
when multiplied times the mass matrix -
and the result subtracted
from the stiffness matrix will yield a zero matrix. This is commonly
known as an "eigenvalue" problem. There will be as many values of A
as there are degrces-of-freedom in the system. There are many standard
computer programs available for handling problems of this kind most of -
which will handle in excess of twenty eigenvalues.

Once the natural frequencies are determined, then the natural


mode shapes, Iq
, can be determined by substituting back into the original
express Lona which contained x.
.
S T D - A P I I P E T R O I L S - E N G L 1978 0732290 Ob024b3 O04

HDF Systems Sublected t o Ground Motions: (Cont'd.)

System Subjected t o I n i t i a l Displacements and V e l o c i t i e s :

As w a s done i n t h e development f o r t h e TQF system, t h e next s t e p


i-
a f t e r developing a technique t o determine t h e n a t u r a l frequencies and
mode s h a p e s , is t o develop mode p a r t i c i p a t i o n f a c t o r s f o r t h e c o n d i t i o n
of i n i t i a l displacements -
and then f o r t h e c o n d i t i o n of i n i t i a l v e l o c i t i e s
- ( f o r no damping and f o r no e x t e r n a l f o r c e s a c t i n g ) . The r e s u l t i n g equations
are shown below.
n - N u m b e r o f degrees of f r e e d o m

At. t = o
IVOI = o

Also, O? t = O

T h e participation f a c t o r s f o r the rth m o d e ore as follows

Thus, there w i l l 'be N . p a r t i c i p a t i o n f,actors ( c & d)- one f o r


Li each mode. These f a c t o r s may be s u b s t i t u t e d back i n t o t h e equation a t
t h e t o p of t h e page to o b t a i n t h e d e s i r e d s o l u t i o n s f o r . t h i s case.

O
*"Tonspose of v e c t o r l A l . =
: is , IAl' = lobcd I
. . L
=
~

S T D - A P I / P E T R O 117-ENGL 1 9 7 8 0 7 3 2 2 9 0 Ob024b'i T Y O

MDF Systems S u b j e c t e d t o Ground Motions: (Gont'd.)


System S u b j e c t e d t o E x t e r n a l Force System:

I n e x a c t l y t h e same manner as w a s done f o r t h e SDF and TDF s y s t e m s ,


a s o l u t i o n is o b t a i n e d f o r t h e case of t h e MDF system s u b j e c t e d t o an
a r b i t r a r y e x t e r n a l force system. As was done f o r t h e TDF system, i t is
assumed t h a t t h e f o r c e s a c t i n g for a g i v e n c o n d i t i o n a l l have t h e same
t i m e - w i s e v a r i a t i o n . S o l u t i o n s f o r o t h e r time-wise v a r i a t i o n s may be
superimposed. The r e s u l t i n g e q u a t i o n s f o r t h i s case are g i v e n below:

WHERE:

P a r t i c l p o t ion Foctor
For I t h Mode

)he loading is assumed of the form :

1 P1 - Characteristic Value Of Loading,


' f (t) - Time-Wise Variation O f Looding

I djl = j
th
Natural Mode Shape
t-t
f
p i f ( T ) sin

J
t:O

This i s a g e n e r a l i z a d s o l u t i o n f o r t h e r e s p o n s e , / x ( t ) / , of a m u l t i d e g r e e -
of-freedom system (undamped) s u b j e c t e d t o a system of e x t e r n a l f o r c e s which
a l l have t h e sama time-wise v a r i a t i o n .

Nota t h a t t h e s o l u t i o n c o n s i s t s of t h r e e p a r t s : (1) t h e P a r t i c i p a t i o n
F a c t o r s which are f u n c t i o n s of t h e p h y s i c a l c h a r a c t e r i s t i c s of t h e system
( n a t u r a l f r e q u e n c i e s and mode s h a p e s ) and t h e r e l a t i v e magnitudes of t h e
a p p l i e d load system ( n/); (2) t h e n a t u r a l mode shapes ( /0/) which are
a p r o p e r t y of the g i v e n syatern ( n o t i n c l u d i n g any e x t e r n a l l o a d i n g s ) ; and
(3) t h e I n s t a n t a n e o u s A m p l i f i c a t i o n F a c t o r s which d e s c r i b e t h e response of
a SDF s stem (having the same n a t u r a l f r e q u e n c i e s as t h e MDF system) t o
t h e l o d i n g s y s t e m ' s time-wirie v a t i a t i o n s ,

- .
MDF Systems Subjected t o Ground Motiona: (Cont'd.)
System Subjected t o Only 'Ground Motions:
To handle t h e case of t h e MDF system s u b j e c t e d t o ground a c c e l e r a t i o n s ,
w e w i l l r e p l a c e t h e e x t e r n a l loading system from t h e previous page -
IP(t)l = IPl .f (t)
b y ' a loading system which w i l l produce a response e q u i v a l e n t t o t h e ground
accelerations:
IPe ( t ) l = I m l . x o ( t )
This is t h e same technique used f o r both the SDF and TDF systems. The
r e s u l t i n g equations are shown below.

Y=t
/

-
Where :
I x ( t-) IGhC= x, ( t ) = Response of system t o ground
a c c e l e r a t i o n s :displacement
5, 0) I ( r e l a t i v e t o ground) of each
i, (t) system mass w i t h t i m e .

The ground motion p a r t i c i p a t i o n f a c t o r


for the moda (a c o n s t a n t ) .

= Ground motion a m p l i f i c a t i o n f a c t o r (varies


with t h e )

l+JI = Notural Mode Shade of ja Mode =

L
-.
& -8
S T D - A P I / P E T R O 1 1 9 - E N G L 1978 m CI732290 U b 0 2 4 b b 813 D

MDF Systems S u b i e c t e d t o Ground Motions: (Cont'd.)

Where ( con't ) :
pj = Noturol frequency of j 'hmode

1m 1 = MOSS vector

E
j=i
= Summation of responses of individual modes
from jth = I t o j t h = n , where n is the number of
degree of -freedom in the system

Approximate Maximum Response Due t o Ground Motions:

The maximum response from each of t h e modes o c c u r s at a d i f f e r e n t


time. An upper-bound estimate t o t h e maximum r e s p o n s e may be o b t a i n e d
by a d d i n g all of t h e modal rnaximums....a c o n s e r v a t i v e approach, A
g e n e r a l l y reasonab-le estimate may b e made of the maximum r e s p o n s e by
t a k i n g a root-mean-square (RMS) v a l u e of t h e modal m a x i m u m s . These two
approaches are o u t l i n e d below.
_ I

1 xMAx I", 5
- 2 ( Zj lGM I $jl ( S D ) j = Upper-bound response

Where

(SD)j = j t h Mode S p e c t r a l Displacement -


which is e q u a l
t o t h e S p e c t r a l Displacement of a SDF system
having a n a t u r a l frequency e q u a l t o t h a t of t h e
j t h mode, Pj.
. . .. - .
. . .-.?.... .,
....y;..t.,.....
.:
. .... .
. . - .'
..
L

MDF Systems S u b j e c t e d t o Ground Motions (Cont'd.)


Maximum Response (Cont'd.)

Similarly

I
i

..

Where,
SV)^ = j t h Mode S p e c t r a l V e l o c i t y - which i s e q u a l
t o t h e S p e c t r a l A c c e l e r a t i o n of a SDF system
having a n a t u r a l frequency e q u a l t o t h a t o f
t h e j t h mode, p
1.
(SOIj = j t h Mode S p e c t r a l A c c e l e r a t i o n - which i s e q u a l
t o t h e S p e c t r a l A c c e l e r a t i o n of a SDF system
h a v i n g a n a t u r a l frequency e q u a l t o t h a t of t h e
j t h mode, pj.
From t h e s e e q u a t i o n s , we c a n see t h a t t h e Upper-Bound Maximum
Responses are o b t a i n e d by a "weighted" a d d i t i o n of t h e r e s p o n s e s of
8 series of SDF s y r t e m s h a v i n g f r e q u e n c i e s e q u a l t o t h o s e of t h e MDF
o r i g i n a l system. The "weighting" f a c t o r s are t h e P a r t i c i p a t i o n
F a c t o r r , Z j , and t h e Mrlde Shapes, /0j/. The primary c o m p u t a t i o n a l
o b s t a c l e s are p r e s e n t e d by t h e s o l u t i o n of t h e e q u a t i o n s which d e t e r m i n e
t h e n a t u r a l f r e q u e n c i e s and mode s h a p e s .
c
The root-mean-square approach t o e s t i m a t i n g t h e maximum r e s p o n s e
is shown below. See t h e main text f o r a d i s c u s s i o n of t h e a c c u r a c y o f .
t h i s estimate.
L

L
MDF Syatems Subjected t o Ground Motions (Cont'd.)
Damped MDF Systems:
Thus f a r , t h e response of damped MI)F systems h a s not been d i s c u r r e d .
There are t h r e e primary types which have been developed analytically.
Absolute Damping - a damping c o e f f i c i e n t , which -is taken as
c o n s t a n t throughout t h e syrtem, is m u l t i p l i e d times t h e
v e l o c i t y (with r e s p e c t t o the ground) of a given maas t o
o b t a i n t h e damping force.
R e l a t i v e Damping - a damping c o e f f i c i e n t , which is taken as
c o n s t a n t throughout t h e system, is m u l t i p l i e d time8 t h e
r e l a t i v e v e l o c i t y (with r e s p e c t t o t h e a d j a c e n t masr) of
a given mass t o o b t a i n t h e damping f o r c e .
Combined Dmnping - t h e damping c o e f f i c i e n t s are taken as
a linear combination of t h e mass and s t i f f n e s s c h a r a c t e r i s t i c s
o f t h e system - and are m u l t i p l i e d times t h e v e l o c i t y (with
r t 8 p c c t t o t h e ground) of a given mass t o o b t a i n t h e damping
force.
hi^ l a t t e r method has a major computational-ease advantage over t h e o t h e r
two when t h e mode-superposition technique i r used. This 7s because the
assumption t h a t t h e damping c o e f f i c i e n t is a l i n e a r combination of t h e maas and
s t i f f n e s s c o e f f i c i e n t , a i l o u r u s t o u s e t h e damped response spectra of SDF
sys terns d i r e c t Ly.

igid Rigid

A bsolufe
dam ping
-1 --R e l a t i y e
damping
1
r

Combined d a m p i n g 2~
~~

=
~

. . ..
S T D - A P I / P E T R O 117-ENGL 1778 0732290 Ob024b9 5 2 2

.. ..... MDF Systems Subitcted to Ground Motions (Cont'd.)


Combined Damping:
For damped SDF systems, it w a s shown in Appendfx C, that the
resulting equation of motion would be as follows.

Since, p2= -
k2
p = f i or mz
. .

. .
~ n d ,i f w e assume ß = -
2mP
C

Then, .the equation o f motion becomes

..

assuming combimtion damping will' be a s f o l l o w s :

Where T is a scalar quantity resulting from the assumption

. .
. I
Natural t
Scalar
mod* quantity
rho p e
_ .

The important point is that B for the ' S D F system i s

directly analogous to ( 7 + -
- ap ) f o r the m OF system.
.. . .
. .
a? 2
. ..
..
Further, for p u r e absolute damping

for pure r e l a t i v e damping

- .
~ ~ ~ ~~

STD.API/PETRO 1 1 7 - E N G L 1978 0 7 3 2 2 9 0 üb02470 24Li

MDF Syatems Subjected to Ground Mations (Cont'd.)

Combined Damping:
Frcnn the foregoing, t h e conclusion reached is- that i f combination
dampinR ir, a s u r æ d - then-the response of the system may be taken aa
a l i n e a r combination of the individual modal responses or:

7'
J
t"O

OR

(So 1 = The Damped S p e c t r a l Displacement For The


tk
ßi 1 Mode Where
. _ ~~~ ~ ~ ~ ~~~ ~~ ~~ ~

..... .
-:... ._
..
..
.... _. ..
,..-
..
-. __..
,~ i
.,
S T D - A P I / P E T R O 119-ENGL 1 9 7 8 0 7 3 2 2 9 0 Ob02471 180 m
L

A3

. . .
- Level No.
Given: Mass Properties - 3 -
from a given system;

s t i f f n e s s properties from a 2 - = -W2


c
9
s t a t i c analysis; assume

combination damping - -W I
Q
Find:
.
Maximum response to c r i t e r i a
earthquake

Idealized S w t e m

A. Determine the ßy8teu1'8 natural frequencies.

-k
2k
-k
O
-k
k
I
'I2
-I
1' I

= Ix,~ sin
O
m
O
- 'O
I
O ' 1/2 I:
-.
Also Remembering
I j r J = op2 Ix, I= op' 1x.l sin ( p t +o<)
~
~

STD.API/PETRO 1 1 7 - E N G L 1978 w 0 7 3 2 2 9 0 D b 0 2 4 7 2 017

A. Notural Frequencies (cont.)


Then

If an eigenvalue computer program were available, w e would formulate


this equation as:

-
A = P' m
k
and let the program produce the values for ~,,~,,~,,
and thun for P,, Palp,,
However, we will asume none ia available and we will write the equation8
at the top of t h i s page i n expanded form.

- k X, O = 01
-mp' X, + 2kX, -kX,
- kX,
Or
í ~ù -mp2) + (-k 1 +
(-k) t (2k-mp2) + =O
O '+ (-k) +

Or
STD.API/PETRO 117-ENGL 1778 0732270 üb02473 T 5 3 D

A. Natural Frequencier (Cont'd.)


Now to rolve thir ryrtcm of three equations

Let N = (2k-mp')

Exponding 'the determinant we find

N'- 3k'N = 0
Dividing b y N

. N2 - 3k2= 0 Now k # O .'. N =O

k
Thus p*: = 27n

A Is0

Or

p,:
= (2 +fiIr
k = 3 . 1 3 k2 ~
_ -
' p2
a3
= (2 0 sk ) =~0 . 2 6 k8 ~
Consequent I y ,

P,* = 0.268 k ( lowest n a t u r o l frequency)

P! = 3 , 7 3k2 ~
.
There ara the deiired noéural'frequancier.
STD.API/PETRO 1 1 9 - E N G L 1778 0732290 Ob02474 79T

B. Determine the System's Natural Mode Shapes

Returning to the equation

And with N = 2 k - m p 2 , Expanding we h a v e


- kX2 = O --- Eqn. I
N XI
-kXI-NX2 - kX, = O --- Eqn. 2
- 2 k X2 + NX1 = O---Eqn. 3

.
Now, express .the mode amplitudes &and&in
amplitude XI
tem8 of the f i r s t l e v e l

From Eqn. I -=-


x2N
XI k
From Eans. l a n d 2

For the f i r s t natural frequency -

* T h u s f o r XI = I
x2 =Js
xt =2

or
IQ,I = a
I

2
First mode shope

_.
E
~ ~~

.
.. . ..:.._..
:..._ ....
-- :.
.I

.. .., .. ..
-*:i-.-: S T D = A P I / P E T R O 1 1 7 - E N G L 1978 0732290 0b02‘475 82b
.-.

8 Mode Shapes ( C o n t . )

Or -I

[921 = [ij
Second Mode S h a p e
k
For p i = 3.732 -
m

x, N2 3k2

Or

Third Mode’ S h a p e

I
9, = O

J -I
~

STD.API/PETRO 117-ENGL 1978 9 0 7 3 2 2 9 0 Ob0247b 7 b 2 9

B. Natural Modes (Cont'd.)

These modes of natural vibration may be verified by the


requirement that all of the natural modes- must be-orthogonal. This
requirement of orthogonality is the result of the properties of the
matrices involved in the differential equation of motion for the system.
It may be. stated and illustrated as follows;

I #n I T
= Transpose of ?he nth m o d e s h o p e

I +,,I = The mth mode shape

For this
1 I I
I +I
-
- 6
2
I l= +2
-I
O 4
2

O r for n m = 2

-I
I I
O1
'12
O
-I
I

O = o which checks

-0.5

For n = 2 ond m = 3

' I I 0 - 1 1 I O I
0 1 -fi =
L
O 0 2

I
-fi O which checks
I
... _... . . _
.
- . _._'._.
~..
..
..: . S T D - A P I I P E T R O 117-ENGL 1978 0732270 ûb02Y77 bT7 M
.... - .
.-..-,-a.

C. Determine the ParticiPotion Factors

For j = I

0.5H 0 -
0.333
(i2)gm = I.5P

.-

i
(Z3)gm = 0.268 p :-
0.034
8~
From these participation factors, we~may now see that 62.22 of the first mode,
33.32 o f the second moda, and 3.5% of the third mode (note total o f 100%)
responses w f l l comprise the total response o f the system to ground motiona.
D. Determine the Spectral Response Factors

Combination Damping Factors

Assume pure absolu te dam ping

T o k ng 5 % domping for t h e f i r s t mode

6 , = 0.05
n O.0 5 O. 0 5

O. 05 -- -“ O 5 = 0.013
3. 7 2
Thus, we can use 52 damping i n the first mode, 29, i n the second,
and 1% i n the third.

W e hove previously found

( f . = - P1
2x

I t is given that,

k = 200 k i p s l i n . W = 2000 k i p s
2
2000 k kip -sec.
r n r = 5.18
386in. in.
sec?

Thus,

5, = - ‘
2 1
O. 268 = 0.512 cps

T, = ï.95 sec.
O. .Spectral Response Factors (cont.)
L

-- f, : 1.40 C P S , f2 = 0.715 Sec.


f, = 1.91 cps , T3 = 0.524 sec.
The criteria earthquake i s given as the average of the E l Centro
(1940) horizontal acceleration records.

From the response spectra shown on the next page, the spectral
displacements are:

Mode Period Domping S, SD

i Sec. O/. Ft. I n.


I l .95 5 O.54 6.5
2 0.72 2 O.25 3 .O
3 0.52 I 0.28 3.4

E. Determine the Maximum Displacements


Upper bound es timate

f (0.622) 6
I

2
(6.5 in) -
. . First
Mode
Contribution
mor
ub

..
+ (0.3331 O
I

-I
(3.0 in) - Second
Mode
Contribution

+ (0.034) ' (3.4in) - Third


Mode
Contribution
PER I OD, T( sec.) --c

f
n t
o
Y
a

w
>

P E R I O D , T(sec.1-

AVERAGE RESPONS€ SPECTRA FOR EARTHQUAKE GROUND MOTION


OF INTENSITY RECORDED AT EL CENTRO, CALIFORNIA 1940
(Maximum Ground A c c e l e r a t i o n 33"/.g)

Converted to logarithmic plot from spectra given in


TI.0. 7024, 'Nuclear Reactors and Earthquakes'
U. S.Atomic Energy Commission. August 1963
E. ,Maximum Displacements (Cont.)

XI 4.05 1 .o0 o.i2


x2
-4
702 + O + -0.20
x3 1 Mar
UB 8. IO - 1.00 O,23

[ N o t e : Use A b s o l u t e Values O f Mode C o n t r i b u t i o n s I


*, I, 5.17in. U p p e r Bound Estimote O f
rz 2 7,22 in. Maximum Displacements Of

x3 2 9.33in Each Level

Root - M e a n - S q u a r e ( R M S) Estmate

F. Determine Maximum Shears

Base : (5.17 in ) (200 ikips


n.) 6 1034 k i p s

Total system weight = 5000 kips

For, a c o d e . latem force f a c or = 0.1


Shear ( c o d e ) = cw = (0.1) (5000)= 500 k i p s
F

(x2- xi) k S 17.02 - 4.051 + (-0.20 - 0.l2U k


(12.971 + 1-0.321) k S 3.29 k S 658 kips

( x 3 - x ~ k) I
ci 8.10 - 7.021 + 1 - i . O O l + 10.23 + 0.201
1
(2.51) k 2 502 kips

.Y

S-ar putea să vă placă și