You are on page 1of 138

I.U.S.S.

Istituto Universitario Università degli


di Studi Superiori Studi di Pavia

EUROPEAN SCHOOL OF ADVANCED STUDIES IN


REDUCTION OF SEISMIC RISK

ROSE SCHOOL

NON-LINEAR SEISMIC RESPONSE OF


CIRCULAR PRESTRESSED RESERVOIRS

A Dissertation Submitted in
Partial Fulfillment of the Requirements for the
Master Degree in Earthquake Engineering

By

Christos Baltas

Advisors:
Prof. M.J. N. PRIESTLEY
Dr. R. PINHO

December, 2004
The dissertation entitled “Non-Linear Seismic Response of Circular Prestressed Reservoirs”, by
Christos Baltas, has been approved in partial fulfillment of the requirements for the Master Degree in
Earthquake Engineering.

M. J. N. Priestley________________________

R. Pinho________________________
Abstract

ABSTRACT

The scope of this project is to investigate the non-linear seismic response of circular prestressed
concrete reservoirs tanks of different configurations designed under current design codes [1,2,8,9]
using a beam on elastic foundation model [10,14,15] and determine effective force reduction factors
for seismic design cases and safe PGA for the non seismic design case.

Modeling assumptions are presented for the analysis of cylindrical concrete storage tanks as those are
covered by standard texts on the design of circular reservoirs. A comprehensive review and application
of the Force-Based Design approach of circular prestressed reservoirs structures when subjected to
seismic action in the transverse direction is presented. The design procedure to use with this type of
structure is presented first, followed by a series of applications to different possible tank
configurations.

Assessment of the tanks designed is carried out using pushover and time-history analyses. Models
including inelasticity for pushover and time-history analyses are presented. The pushover analysis
indicates the multiple of code seismic load that each tank can sustain revealing the effectiveness of
code reduction factors used for their design. Incremental dynamic analysis is carried out by using
multiple of time-histories that codes impose indicating the conservatism of the seismic design of the
circular prestressed concrete reservoir.

Finally some design and analysis aspects of sliding base tanks are presented.

-i-
Acknowledgements

ACKNOWLEDGEMENTS

I would particularly like to thank Professor Nigel Priestley - being really an inspiration to me – for
sharing his vast knowledge with me and for his wise advice during the development of this work. I
would like to express my gratitude to Dr. Rui Pinho - the busiest person in Pavia - for his support and
for being always willing to listen and help me during my staying in Pavia. I would like also to thank
Professor Calvi, director of the ROSE School, who gave me the opportunity of studying and being a
member of this promising school and all the Professors that came during my staying to Pavia and
helped me to understand the concepts of Earthquake Engineering.

Thanks to all my friends at ROSE School, (the Greeks - Aggelos, Argyroula, Lena, Sofia and Panos,
the Japanese - Masahiro, the Italians - Carlo and Sandra, the Kiwis - Damian, Didier and Tim, the
Latins - Agustin, Ana, Camilo, Luis, Catalina and Bea) for all the great times we shared. I would
particularly like to thank Katrin Beyer - my German neighbour - who apart from being a very good
friend was always willing and capable to give me answers to my questions.

I would like also to thank my dear friend and colleague from the National Technical University of
Athens - Mario Panagiotou - who encouraged me to follow graduate studies on the subject of
Earthquake Engineering.

This work and this MSc Degree are dedicated to my family - Panagiotis, Vasiliki and Stavros - for
their love, patience and unconditional support during all the last year being outside my country,
Greece.

- ii -
Table of Contents

TABLE OF CONTENTS

ABSTRACT ............................................................................................................................................. i

ACKNOWLEDGEMENTS .................................................................................................................... ii

TABLE OF CONTENTS ....................................................................................................................... iii

LIST OF TABLES ................................................................................................................................. vi

LIST OF FIGURES.............................................................................................................................. viii

1. INTRODUCTION........................................................................................................................1

2. MODELING OF CIRCULAR PRESTRESSED RESERVOIRS ...............................................3

2.1. INTRODUCTION .................................................................................................................3

2.1.1. HOOP TENSION ...........................................................................................................4

2.1.2. VERTICAL BENDING .................................................................................................5

2.1.3. SURFACE CIRCUMFERENTIAL STRESS ................................................................5

2.2. FRAME ANALOGY .............................................................................................................7

2.2.1. CYLINDRICAL TANKS WITH AXISYMMETRIC LOADING ................................7

2.2.1.1 PRESTRESS SIMULATION................................................................................8

2.2.2. CYLINDRICAL TANKS WITH ROTATIONALLY UNSYMMETRICAL


LOADING .................................................................................................................................10

3. DESIGN OF CIRCULAR PRESTRESSED RESERVOIRS.....................................................15

- iii -
Table of Contents

3.1. WALL-FOUNDATION CONNECTION TYPES ..............................................................15

3.2. DESIGN PROCEDURE ......................................................................................................15

3.2.1. NON-SEISMIC DESIGN CASE .................................................................................16

3.2.2. SEISMIC DESIGN CASE ...........................................................................................18

3.2.2.1 IMPULSIVE, CONVECTIVE AND VERTICAL MODE PERIODS................18

3.2.2.2 BEHAVIOR FACTORS......................................................................................22

3.2.2.3 SEISMIC ACTIONS ...........................................................................................23

3.2.2.4 SLOSHING HEIGHT..........................................................................................31

3.2.3. DESIGN FOR VERTICAL AND HOOP DIRECTION..............................................31

3.3. APPLICATION TO DIFFERENT TANK CONFIGURATIONS.......................................33

3.3.1. MEMBER INFORMATION........................................................................................33

3.3.2. MATERIAL PROPERTIES.........................................................................................34

3.3.3. DESIGN CASES ..........................................................................................................35

4. ASSESSMENT OF TANK RESPONSE ...................................................................................39

4.1. PUSHOVER ANALYSIS....................................................................................................39

4.1.1. VERTICAL BEAM MEMBERS .................................................................................39

4.1.2. RADIAL SPRINGS .....................................................................................................43

4.1.3. PUSHOVER ANALYSIS DETAILS...........................................................................51

4.1.4. PUSHOVER ANALYSIS RESULTS ..........................................................................51

4.2. TIME-HISTORY ANALYSIS ............................................................................................55

4.2.1. VERTICAL BEAM MEMBERS .................................................................................55

4.2.2. RADIAL SPRINGS .....................................................................................................56

4.2.3. TIME-HISTORY ANALYSIS DETAILS ...................................................................57

4.2.4. TIME-HISTORY ANALYSIS RESULTS...................................................................59

5. DESIGN AND ANALYSIS OF SLIDING BASE TANKS ......................................................61

5.1. SLIDING BASE TANKS ....................................................................................................61

6. CONCLUSIONS........................................................................................................................68

BIBLIOGRAPHY ..................................................................................................................................70

- iv -
Table of Contents

APPENDIX A ........................................................................................................................................72

APPENDIX B.........................................................................................................................................77

APPENDIX C.........................................................................................................................................86

APPENDIX D ......................................................................................................................................115

-v-
List of Tables

LIST OF TABLES

Table 3.1 : Flexible Base Period to Rigid Base Period Ratios for different H/R, t/R and E/Gs ratios due
to Soil-Structure Interaction...........................................................................................................20
Table 3.2 : Damping % for different H/R, t/R and E/Gs ratios due to Soil-Structure Interaction..........20
Table 3.3 : Design Behavior Factors for Impulsive, Convective and Vertical Mode.............................22
Table 3.4 : Values of the Parameters Describing the Recommended Type 1 Elastic Response
Spectrum. .......................................................................................................................................26
Table 3.5 : Values of the Parameters Describing the Recommended Type 2 Elastic Response
Spectrum. .......................................................................................................................................26
Table 3.6 : Values of the Parameters Describing the Vertical Elastic Response Spectrum. ..................27
Table 3.7 : Member Dimensions. ...........................................................................................................33
Table 3.8 : Material Properties. ..............................................................................................................34
Table 3.9 : Increase of Initial Prestress for Non Seismic Design Case. .................................................36
Table 3.10 : Non-Seismic Design Values for Vertical Bending Moment in kNm. ................................36
Table 3.11 : Impulsive Period in sec. .....................................................................................................37
Table 3.12 : Convective Period in sec. ...................................................................................................37
Table 3.13 : Vertical Period in sec. ........................................................................................................37
Table 3.14 : Difference of Total Force Using the SRSS Combination and the Impulsive Mode Force.38
Table 3.15 : Increase of Prestress for Seismic Design Relative to Non-Seismic Design Case. .............38
Table 3.16 : Seismic Design Values for Vertical Bending Moment in kNm. ........................................38
Table 4.1 : Multiple of Seismic Code Intensity that Cracking occurs for the Non-Seismic Design Case.
.......................................................................................................................................................53
Table 4.2 : Multiple of Seismic Code Intensity that Cracking occurs for the Seismic Design Case......53

- vi -
List of Tables

Table 4.3 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the
Non-Seismic Design Case. ............................................................................................................53
Table 4.4 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the
Seismic Design Case......................................................................................................................54
Table 4.5 : Multiple of Seismic Code Intensity that Failure occurs for the Non-Seismic Design Case.54
Table 4.6 : Multiple of Seismic Code Intensity that Failure occurs for the Seismic Design Case. ........54
Table 4.7 : Multiple of Seismic Code Intensity that Cracking occurs for the Non-Seismic Design Case.
.......................................................................................................................................................59
Table 4.8 : Multiple of Seismic Code Intensity that Cracking occurs for the Seismic Design Case......60
Table 4.9 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the
Non-Seismic Design Case. ............................................................................................................60
Table 4.10 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the
Seismic Design Case......................................................................................................................60

- vii -
List of Figures

LIST OF FIGURES

Figure 2.1 : Pinned-Base Tank Under Fluid Pressure. .............................................................................6


Figure 2.2 : Beam on Elastic Foundation and Frame Analogy Simulations of Cylindrical Tank Wall. ..6
Figure 2.3 : Forces Applied to Tank Wall by Prestressing Tendons. .......................................................9
Figure 2.4 : Mechanical Analogue of Response of Fluid Contents of Tank. .........................................10
Figure 2.5 : Curves for Obtaining Factors WI/WT and WC/WT for the Ratio R/Hw................................11
Figure 2.6 : Curves for Obtaining Factors hI/Hw and hC/Hw for the Ratio R/Hw. ...................................12
Figure 2.7 : Seismic Pressure Distribution in Cylindrical Tanks. ..........................................................14
Figure 2.8 : Force on the Tank Wall.......................................................................................................14
Figure 3.1 : Wall-Foundation Connections. ...........................................................................................16
Figure 3.2 : Loading Patterns for the Non-Seismic Case. ......................................................................17
Figure 3.3 : Loading Patterns for the Seismic Case................................................................................18
Figure 3.4 : Period Coefficient, Cw.........................................................................................................20
Figure 3.5 : Angle of Cable Inclination and Spacing between the Cable Sets. ......................................21
Figure 3.6 : Horizontal Elastic Spectrum for Types 1&2.......................................................................27
Figure 3.7 : Vertical Elastic Spectrum for Types 1&2. ..........................................................................28
Figure 3.8 : Seismic Coefficients. ..........................................................................................................30
Figure 3.9 : Seismic Coefficients for the Convective Mode. .................................................................30
Figure 3.10 : Models Used for the Analysis...........................................................................................35
Figure 3.11 : Water and Initial Prestress Pressures. ...............................................................................36
Figure 4.1 : Moment-Curvature Curve of Vertical Beam Members for Pushover Analysis. .................42
Figure 4.2 : Stress-Strain Curve for Low Relaxation Prestressed Steel .................................................44
Figure 4.3 : Radial Spring Force-Displacement Curves of Radial Springs for Pushover Analysis........50
Figure 4.4 : Moment-Curvature Curve of Vertical Beam Members for the Time-History Analysis. ....55

- viii -
List of Figures

Figure 4.5 : Radial Spring Force-Displacement Curve of Radial Springs for the Time-History Analysis.
.......................................................................................................................................................56
Figure 4.6 : Time-Histories Response Spectra Matching EC8 Type 1 Design Spectrum. .....................57
Figure 4.7 : Time-History of EC805GA (ATH1) Artificial Accelerogram............................................58
Figure 5.1 : Modeling of Bearing Pad (Friction-Pendulum Isolator Link Element of SAP)..................63
Figure 5.2 : Modeling of Vertical Resistance of Seismic Cables (Hook Spring Link Element of SAP).
.......................................................................................................................................................63
Figure 5.3 : Modeling of Horizontal Resistance of Seismic Cables (Wen Plasticity Link Element of
SAP)...............................................................................................................................................64
Figure 5.4 : Pushover Analysis of Sliding Base Tank Using SAP Program. .........................................66

- ix -
Chapter 1 – Introduction

1. INTRODUCTION

The seismic performance of circular reservoirs is a matter of special importance. Without an assured
water supply, uncontrolled fires occurring as a result of a major earthquake may cause more damage
than the earthquake itself. If the outbreaks of disease that frequently follow destructive earthquakes are
to be avoided, it is essential that safe supplies of drinking water are available. Spillage or toxic
chemicals and highly inflammable petroleum products from damaged tanks can cause damage many
times the value of the affected tank and contents.

There are many recorded failures of cylindrical tanks under seismic attack. Most failures have resulted
from buckling or uplift of walls of steel storage tanks and there are very few reports of damage to
concrete structures.

The scope of this project is to investigate the non-linear seismic response of circular prestressed
concrete reservoirs of different configurations designed under current design codes [1,2,8,9] using a
beam on elastic foundation model [10, 14, 15] and determine effective force reduction factors for
seismic design cases and safe PGA for the non seismic design case.

In Chapter 2 modeling assumptions are presented for the analysis of cylindrical concrete storage tanks
as those are covered by standard texts on the design of circular reservoirs.

In Chapter 3 a comprehensive review and application of the Force-Based Design approach of circular
prestressed reservoirs structures when subjected to seismic action in the transverse direction is
presented. The design procedure to use with this type of structure is presented first, followed by a
series of applications to different possible tank configurations.

-1-
Chapter 1 – Introduction

In Chapter 4 the assessment of the tanks designed on Chapter 3 is carried out using pushover and time-
history analyses. Models including inelasticity for pushover and time-history analyses are presented.
The pushover analysis indicates the multiple of code seismic load that each tank can sustain revealing
the effectiveness of code reduction factors used for their design. Incremental dynamic analysis is
carried out by using multiple of time-histories that codes impose indicating the conservatism of the
seismic design of the circular prestressed concrete reservoir tanks.

In Chapter 5 some design and analysis aspects of sliding base tanks are presented.

Some conclusions that were drawn from the studies are presented in Chapter 6.

-2-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

2. MODELING OF CIRCULAR PRESTRESSED


RESERVOIRS

2.1. INTRODUCTION

Analysis of cylindrical tanks under axisymmetric loads, such as fluid or gas pressure, is covered by
standard texts on the design of circular reservoirs, and is only briefly summarized here [10, 14, 15].
Loads are carried by two contributing mechanisms: hoop stress and bending of the wall along the
surface generators. Figure 2.1 illustrates the radial deflection of a pinned-base ground supported
cylindrical tank under internal fluid pressure.

Consideration of compatibility requirements results in the characteristic equation for radial


deformation:

d4y φ ( h)
4
+ a4 y = (2.1)
dh K

where φ ( h ) is the loading function [ φ ( h ) = ρ gh for fluid loading, where ρ is the mass density and g

is the acceleration due to gravity,

Ec I
K= (2.2)
1 −ν 2

-3-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

t3
I= (2.3)
12

and

1 −ν 2
a= 4 (2.4)
R 2t 2

is the characteristic of the equation, where R is the tank radius, t is the wall thickness, and ν is
Poisson’s ratio for the concrete.

Equation (2.1) has the typical solution:

 ah  ah ah 
− ah
2  ah ah  
y = PI +  e 2  A sin + B cos  + e  C sin + D cos  (2.5)
  2 2  2 2  

where PI is the load-dependent particular integral, and A,B,C,D are constants of integration dependent
on boundary conditions at the wall base and top. Once Equation (2.5) is solved for y, the complete
stress distribution in the tank may be calculated as follows.

2.1.1. HOOP TENSION

Since circumferential length and radius R are related by the constant 2π, a radial deformation y causes
a circumferential length increase of 2πy. The circumferential strain will thus be:

ε h = 2π y 2π R = y R (2.6)

and hence hoop tension stress is given by:

f h = Ec y R (2.7)

where Ec is the concrete modulus of elasticity.

-4-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

2.1.2. VERTICAL BENDING

Vertical bending moments are as:

Ec I d 2 y
Mv = (2.8)
1 − ν 2 dh2

If moments have been calculated for unit wall width, then surface bending stresses in the vertical
direction can be found from:

6
fv = ± M v (2.9)
t2

2.1.3. SURFACE CIRCUMFERENTIAL STRESS

Since the loading and structure are both axisymmetric, radial lines connecting points on the inside and
outside surfaces of the tank wall must remain radial after application of loading. This effectively
means that transverse Poisson’s ratio strains from vertical bending moments cannot develop, and in
consequence, the wall is in a state of plane strain for vertical bending. Consequently, circumferential
Poisson stresses, νfv, are developed in the walls.

These Poisson stresses are commonly ignored in design, but can quite significant, particularly when
high bending moments develop from base fixity. Final circumferential stresses at the wall surface are
thus:

Ec y 6M
fc = fh +ν fv = ±ν 2 v (2.10)
R t

-5-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

Figure 2.1 : Pinned-Base Tank Under Fluid Pressure.

Figure 2.2 : Beam on Elastic Foundation and Frame Analogy Simulations of Cylindrical Tank Wall.

-6-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

2.2. FRAME ANALOGY

2.2.1. CYLINDRICAL TANKS WITH AXISYMMETRIC LOADING

In the introduction it was established that the response of cylindrical tank walls under rotationally
symmetric loads involved load sharing between two mechanisms: hoop tension and vertical bending.
Equation (2.1) is developed from the compatibility requirement that radial deformations of the two
mechanisms must be identical at all points. Behavior can be described as basic vertical beam action,
where radial deformations are further constrained by the stiffness of the hoop action.

The analogy to Beams on Elastic Foundation (BEF) analysis [10] is obvious and has been used in the
past to generate design tables for tanks. Vertical bending of the tank wall is represented by beam
action, and the elastic radial stiffness of hoop action is represented by the spring stiffness of the
Winkler foundation.

Consider a unit height of tank wall subjected to pressure p, and freed from cantilever action. From
considerations of simple statics, the radial expansions will be:

pR 2
y= (2.11)
tEc

Figure 2.2α shows a unit circumferential length of tank wall supported by a horizontal Winkler
foundation. Deflection of this foundation under pressure p will be:

p
y= (2.12)
k

where k is the subgrade modulus in kN/m3. Thus Figure 2.2α can exactly model the wall behavior,
providing the subgrade modulus is given by:

tEc
k= (2.13)
R2

A more convenient method of simulation, capable of modeling thickness variation of the tank walls
and complex loading, is to replace the continuous foundations of Figure 2.2α by the discrete system of
Figure 2.2b, where the tank wall is divided into a number of vertical beam elements whose connecting
nodes are supported by lateral pin-ended struts from a rigid foundation [14].

-7-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

This system can be solved by simple frame analysis. Beam members are given the local vertical
bending stiffness properties of the section of wall represented, while strut properties model the radial
stiffness of that portion of wall extending midway to adjacent nodes above and below, shown shaded
in Figure 2.2b, and in detail in Figure 2.2c for a constant thickness wall.

Thus if unit circumferential wall width is again considered, simulation of hoop stiffness requires:

pR 2 p∆hli
y= = (2.14)
tEc Ai Ei

Hence, if for convenience the strut modulus of elasticity Ei is set equal to Ec, then:

Ai t ∆h
= 2 (2.15)
li R

Either the cross-sectional area Ai, or the length li of the strut (or both), may be varied to obtain the
required similitude. Loads applied to the vertical beam must be commensurate with the width of wall
adopted for analysis (e.g. unit width in the above development). Vertical bending moments in the tank
wall are directly modeled by the moments developed in the beam members, and circumferential
stresses are found from the node radial displacements, using Equation (2.10).

Loading applied to the frame members corresponds to the load applied to the width of wall modeled,
and will generally be applied as resultant joint loads without significant loss of accuracy. Greater
accuracy will result by application of the actual distributed loads to the beam members as member
loads.

2.2.1.1 PRESTRESS SIMULATION

The frame analogy can also be used to calculate the stresses in the tank walls by prestressing. Consider
the case of a tank constructed of precast wall panels, and stressed by tendons in internal ducts as
shown in Figure 2.3. If F is the tendon force, then the radially inward line load applied to the tank by a
single tendon, as shown in Figure 2.3α, is:

F
P= (2.16)
R

-8-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

The stresses induced by prestress can be found by analyzing the tank wall under the loads P at the
appropriate heights, as shown in Figure 2.3b. Alternatively, if the spacing s is a small fraction of the
wall height, the prestress may be simulated as an equivalent radially inwards pressure of:

P F
p= = (2.17)
s Rs

In the case of a tank stressed by wire wrapping, this would be the appropriate approach. However, for
a tank prestressed by individual tendons of high prestress force, the local stresses induced by the
discrete nature of the prestress line loading may be significant, or the stresses induced during stressing
operations by stressing a single tendon may be of interest. In such cases the prestress should be
modeled by discrete radial load to the frame analogy.

It is clear that simulation of prestress as a series of radial loads will provide a more realistic simulation
than that provided by an equivalent pressure, using tables or design charts. This method of analysis
also enables the prestress design to be optimized since variation in individual tendon positions can be
easily modeled.

Figure 2.3 : Forces Applied to Tank Wall by Prestressing Tendons.

-9-
Chapter 2 – Modeling of Circular Prestressed Reservoirs

Figure 2.4 : Mechanical Analogue of Response of Fluid Contents of Tank.

2.2.2. CYLINDRICAL TANKS WITH ROTATIONALLY UNSYMMETRICAL LOADING

It is not commonly realized that loads that are not rotationally symmetric can often be analyzed with
adequate accuracy by use of a rotationally symmetric analysis. Of particular interest is the case of
seismic loading.

The dynamic behavior of the fluid contents of tanks can be approximated by a mechanical analogue of
springs and masses, as shown in Figure 2.4. mI represents that portion of the total fluid mass mf which
acts as though rigidly attached to the tank walls, and is thus subject to the same accelerations as the
tank walls. This is generally referred to as the impulsive mass of the fluid. The sloshing or convective
response is represented by a number of masses and springs simulating the different antisymmetrical
slosh-modes of the fluid. In practice it is used to regard one slosh-mode as sufficient to represent the
convective forces on the walls.

Expressions for impulsive mass mI and first mode convective mass mC1 and the heights at which these
act (hI and hC1 respectively) have been developed by Jacobsen [12] and Housner [11]. The weight of
the impulsive and convective masses and the heights at which they act can be obtained from the
expressions:

- 10 -
Chapter 2 – Modeling of Circular Prestressed Reservoirs

(
WI tanh 3R / H w
=
) (2.18)
WT 3R / H w

Wc
=
(
3.375 R tanh 3.375 H w / R ) (2.19)
WT 4H w

hI 0.50 − 0.1875 R / H w , for R / H w ≤ 0.6667


= (2.20)
H w 0.375 , for R / H w > 0.6667

hc

= 1− 
cosh ( )
3.375 H w / R − 1 
 , for all R / H w (2.21)
Hw 
 ( )
3.375 H w / R sinh ( 3.375 H w / R 
)

where WT is the total of the fluid and is:

WT = γπ R 2 H w (2.22)

as these are presented in [2] and are shown graphically in Figure 2.5 and Figure 2.6.

Figure 2.5 : Curves for Obtaining Factors WI/WT and WC/WT for the Ratio R/Hw.

- 11 -
Chapter 2 – Modeling of Circular Prestressed Reservoirs

Figure 2.6 : Curves for Obtaining Factors hI/Hw and hC/Hw for the Ratio R/Hw.

The circumferential distribution of pressure follows a cosine distribution with maximum pressure
increase and decrease occurring at opposite ends of the diameter parallel to the earthquake excitation
as:

pθ = p0 cosθ (2.23)

The vertical distribution of pressure on the wall at any location consists of the components from
impulsive response and convective response (sloshing) of the tank and its contents.

For analytical purposes it is common to approximate the curved pressure distributions of Figure 2.7b
and Figure 2.7c by linear approximations, as shown, with the same total force FIθ or FCθ over the wall
height for unit circumferential distance, and with the same height of centre of force, hI or hC.

From Figure 2.7:

( pt + pb )θ
Fθ = H (2.24)
2

- 12 -
Chapter 2 – Modeling of Circular Prestressed Reservoirs

and

( 2 pt + pb )θ
h= H (2.25)
3 ( pt + pb )θ

Consequently if Fθ and h are known, the pressures at top and bottom of the wall, pt and pb, can be
calculated. The maximum value of Fθ occurs at θ=0 and θ=180. From Equation (2.23):

Fθ = F0 cos θ (2.26)

From Figure 2.8, the force on the tank wall for an incremental angle dθ will be:

dVθ = F0 cosθ Rdθ (2.27)

Resolving parallel to the direction of seismic attack:

dV0 = dVθ cos θ = F0 cos 2 θ Rdθ (2.28)

Integrating round the circumference:

2π π /2
V= ∫ dV0 = 4 F0 R ∫ cos θ dθ = π RF0
2
(2.29)
0 0

Thus the total force up the wall per unit circumference at θ=0 is:

V
F0 = (2.30)
πR

where V is:

V = mSag (2.31)

for either impulsive or convective mode of response and Sa is the seismic coefficient obtained from a
design response spectrum for each case respectively.

- 13 -
Chapter 2 – Modeling of Circular Prestressed Reservoirs

It would be initially appear that a full shell analysis would be necessary to model the effect of the
circumferential variation of pressure. In fact, a solution of sufficient accuracy for maximum stresses
may be obtained by assuming the maximum pressure distribution (po in Figure 2.7α) to be distributed
with rotational symmetry around the tank.

This is because at the location of maximum pressure, the rate of variation of pressure in the
circumferential direction is small, and stresses induced by shell distortion resulting from pressure
differences along adjacent generators are negligible. This being the case, the frame analogy approach
or standard design tables or charts may be used to predict maximum stresses induced by earthquake
loading.

Figure 2.7 : Seismic Pressure Distribution in Cylindrical Tanks.

Figure 2.8 : Force on the Tank Wall.

- 14 -
Chapter 3 – Design of Circular Prestressed Reservoirs

3. DESIGN OF CIRCULAR PRESTRESSED RESERVOIRS

This chapter presents a comprehensive review and application of the Force-Based Design approach of
circular prestressed reservoirs when subjected to seismic action in the transverse direction. The design
procedure to use with this type of structure is presented first, followed by a series of applications to
different possible tank configurations.

3.1. WALL-FOUNDATION CONNECTION TYPES

The project is investigating the non-linear seismic response of circular prestressed reservoirs with three
different types of wall-foundation connections (see Figure 3.1) [1]:

A. Fixed Base
B. Hinged Base
C. Anchored Flexible Base

3.2. DESIGN PROCEDURE

The design ignoring creep or shrinkage assumes water and EQ loading. Loading from prestressed
tendons - which are modeled as shown in Chapter 2 - must be provided to ensure that stress limits for
the non-seismic and seismic case are not exceeded.

- 15 -
Chapter 3 – Design of Circular Prestressed Reservoirs

Figure 3.1 : Wall-Foundation Connections.

3.2.1. NON-SEISMIC DESIGN CASE

The stress criterion for the design case ignoring seismic effects or creep and shrinkage for the
circumferential direction is 1.4MPa residual compression [1].

The water pressure loading has a triangular pattern (see Figure 3.2). The initial prestress force
provided to balance the water loading assumes the hoop water tension plus 1.4MPa . So, the initial
prestress force is:

F = pR + 1.4t (3.1)

- 16 -
Chapter 3 – Design of Circular Prestressed Reservoirs

and the equivalent radially inwards pressure pf is:

pf = F = p + 1.4t ( MN / m height ), p = water pressure in MPa (3.2)


R R

Assuming water pressure from top of the tank the water pressure is:

p = wH , w water unit weight , H wall height (3.3)

50% of the above force is applied with the tank free to slide and the other 50% of circumferential
prestress is applied with the base fixed or pinned for the case A and B respectively. The prestress force
may be increased proportionally to provide the 1.4MPa residual compression stress criterion.

1.4t/R

H Hw

wHw p=wH p+1.4t/R


Actual Water Water Pressure Prestressed Tendons
Pressure Used Pressure

Figure 3.2 : Loading Patterns for the Non-Seismic Case.

- 17 -
Chapter 3 – Design of Circular Prestressed Reservoirs

>1.4t/R
pt

H Hw

p=wH p’>p+1.4t/R pb
Water Pressure Prestressed Tendons EQ Pressure
Used Pressure Used

Figure 3.3 : Loading Patterns for the Seismic Case.

3.2.2. SEISMIC DESIGN CASE

The stress criterion for the design case taking into account seismic effects for the circumferential
direction is zero hoop tension [1].

The water pressure loading is as that is defined for the non-seismic case (see Figure 3.3). The prestress
force provided to balance both the water and EQ loading has the same pattern shape as that is defined
for the non-seismic case but increased proportionally to ensure zero hoop tension in the
circumferential direction. Again, 50% of the above force is applied with the tank free to slide and the
other 50% of circumferential prestress is applied with the base fixed or pinned for the case A and B
respectively.

3.2.2.1 IMPULSIVE, CONVECTIVE AND VERTICAL MODE PERIODS

Because of soil-structure interaction effects, a stiff tank resting on a soft soil may have an earthquake
response significantly different from the case of a tank on a rigid foundation. Soil-structure interaction
may lengthen the period of vibration of the impulsive modes and also increase the damping. Generally
an increase in the impulsive mode periods will result in an increase in response but the additional
damping will tend to counteract this effect. Strictly speaking, soil-structure interaction will change

- 18 -
Chapter 3 – Design of Circular Prestressed Reservoirs

both the impulsive and convective periods but the effects on the convective modes are small and can
be neglected [13].

The impulsive period for concrete tanks is related to soil structure interaction and to a lesser extent to
shell flexibility. An impulsive period of 0.1sec is appropriate for tanks on rock and stiff soils and 0.3
sec for moderately soft soils. Most code spectra are based on 5% damping. However soil damping
effects can result in much higher values for tanks with large diameters and large D/H ratios. Damping
values of 15% or more are appropriate for most concrete tanks with diameters greater than 30m. Soil
damping effects for concrete storage tanks on soft soils is an area that needs further review [17].

For this work since rock soils are assumed for which a shear velocity of Vs=800m/s is realistic [8], the
shear modulus given from:

Gs = ρ sVs2 (3.4)

is around Gs = 2 /1000 ⋅ 8002 = 1280 MPa .(a density of 2t/m3 is assumed).

For tanks with elastic modulus for concrete around Ec=37000MPa and hence a Ec/Gs=29.2, a thickness
of the wall t=0.18m, a height of the wall equal to 5m and radiuses equal to 5, 10 and 15m the flexible
base period to rigid base period ratios and the damping due to soil-structure interaction [18, see also
APPENDIX A] are given in Table 3.1 and Table 3.2. The period increase is quite small and since the
periods of the tanks are close to zero can be neglected. The damping values obtained for these
properties are not far more than 5% and hence, conservatively, a value of 5% damping is assumed for
the impulsive mode and a value of 0.5% damping for the convective mode for the purposes of this
project.

The impulsive mode period of vibration of a concrete tank on a rigid base (Type A & B) can be
calculated using approximate equations given in Veletsos [16]. The impulsive mode, rigid base,
approximate period of vibration Tr is given by:

2π H ρw
Tr = (3.5)
Cw Ew

- 19 -
Chapter 3 – Design of Circular Prestressed Reservoirs

where
H height of liquid
Cw the concrete tank period coefficient which is a function of the H/R and t/R ratios as shown in
Figure 3.4.
ρw the wall density, and
Ew the elastic modulus of the wall material

Table 3.1 : Flexible Base Period to Rigid Base Period Ratios for different H/R, t/R and E/Gs ratios due to
Soil-Structure Interaction

R H/R t/R Period Ratio


5 1 0.36 ~1.30
10 0.50 0.18 ~1.10
15 0.33 0.12 ~1.05

Table 3.2 : Damping % for different H/R, t/R and E/Gs ratios due to Soil-Structure Interaction

R H/R t/R Damping


5 1 0.36 ~10%
10 0.50 0.18 ~7%
15 0.33 0.12 ~5%

Rigid Base Period Coefficient For Concrete Tanks


0.25

t/Rw = 0.020
0.20
0.015
Period Coefficient, Cw

0.15 0.010

0.10
0.005

0.05

0.00
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Liquid Height to Wall Radius Ratio, H/Rw
Figure 1

Figure 3.4 : Period Coefficient, Cw.

- 20 -
Chapter 3 – Design of Circular Prestressed Reservoirs

The impulsive mode period for Type C case is [1]:

Ws + Wr + WI
TI = 2π ≤ 1sec (3.6)
gπ Rk

For tanks with strand cables embedded in the wall and in footing and with an elastomeric bearing pad
below the tank wall, the spring stiffness k* is determined as:

 Es As cos 2 α 2GwLp 
k = +  (3.7)
 Lc Sc t p S p 

where
Es the elastic modulus of the cable
As the area of the tendon
α the angle of tendon inclination, usually taken as 30o (see Figure 3.5)
Lc length of cable assumed to stretch, taken as sleeve length plus 35 times strand diameter
Sc spacing between the cable sets (see Figure 3.5)
G shear modulus of elastomeric pad
w width of elastomeric bearing pad in radial direction
Lp length of bearing pad along circumference
tp thickness of elastomeric pad after compression due to dead load
Sp spacing of elastomeric pads

Sc

Figure 3.5 : Angle of Cable Inclination and Spacing between the Cable Sets.

*For typical design a value k=33992kN/m2 can be obtained using properties and design values from
Table 3.7, Table 3.8 and Chapter 5.

- 21 -
Chapter 3 – Design of Circular Prestressed Reservoirs

The convective mode period is [2]:

R
Tc =
1.5 tanh ( 3.375 H / R ) (3.8)

The vertical mode period is [1]:

γ DH w2
TV = 2π (3.9)
2 gtEc

3.2.2.2 BEHAVIOR FACTORS

Currently seismic design of circular prestressed concrete storage tanks is based on allowable stress
limits in the tank wall, implying elastic response. However, many design standards, particularly those
in North America [1, 2], allow calculated elastic response force levels to be reduced by a force-
reduction factor applied to impulsive, convective and vertical components. This implies a significant
reserve capacity above the design force level.

The design behavior factors used for all modes considered and are given in Table 3.3 [1].

Table 3.3 : Design Behavior Factors for Impulsive, Convective and Vertical Mode.

Behavior Factor
Type
Impulsive Convective Vertical
A 3 3 3
B 3 3 3
C 2.5 2.5 2.5

Generally reserve capacity above elastic response levels is attributed to the ability of a structural
system to absorb energy by ductility. It is known, however, that prestressed concrete systems have
very low energy absorption capacity, and as a consequence it has been argued that reduction-factors
should not be applied to compute elastic response force levels. It is clear, however, that reserve
capacity will still exist above elastic stress limits, regardless of energy absorption capacity, as a
consequence of a number of factors, briefly outlined below [15]:

- 22 -
Chapter 3 – Design of Circular Prestressed Reservoirs

• Stress limits are based on a “no-cracking” performance level as these were presented. Although it
may seem extreme not to permit any tension strength capacity in the hoop direction, joints between
precast panels, or at vertical construction joints for in-situ construction will often have little, and
undependable tension strength. However, significant increases in hoop tension capacity are
possible without recourse to true inelastic response. Due to friction and strain losses, average
tendon hoop stress at cracking will be about 50-55% of UTS, implying quite high potential
increases in hoop tension capacity above the “no-tension” criterion, before inelastic tension strains
develop, typically at 0.80-0.85 fpu.
• Vertical bending stresses in the wall are rarely critical under seismic loading. Hence a reserve of
vertical moment capacity may exist above the moments developed at code seismic force levels.
• Internal pressures - both seismic and hydrostatic - are resisted by a combination of hoop tension
and vertical bending actions. In the initial elastic phase of response the relative magnitudes of the
two actions is exactly fixed. However, as cracking develops, the ratio between hoop action and
vertical bending can be expected to change, as may the distribution of hoop force and bending
moment up the wall. It is conceivable that a considerable reserve of strength (an “overstrength”
factor) may exist as a consequence of this force redistribution.
• Although elastic response in bonded prestressed structural systems should not be relied upon for
energy dissipation, it is well established that non-linear elastic response can be relied on if the
tendons do not exceed stress levels corresponding to their limit of proportionality. Non-linearity
modifies the response period, and can reduce resonant build-up.

3.2.2.3 SEISMIC ACTIONS

As it was mentioned before the vertical distribution of pressure on the wall at any location due to
ground excitation consists of the components from impulsive response and convective response
(sloshing) of the tank and its contents. Apart form these modes the vertical mode must be taken into
account. From Equation (2.30), it is clear that the design force for each mode (FI or FC) is computed
from the base shear of each mode. The base shear of each mode can be computed from any response
spectra. The base shear as it is given in [1] is:

ZICI
VI = (Ws + Wr + WI ) , for the impulsive mod e (3.10)
Rw

ZICC
VC = WC , for the convective mod e (3.11)
Rw

- 23 -
Chapter 3 – Design of Circular Prestressed Reservoirs

where
Z the peak ground acceleration (in g) of the design spectrum
I importance factor
CI, CC the horizontal acceleration seismic coefficients which corresponds to the impulsive
and convective mode period TI and TC respectively
Ws, Wr the weights of the tank walls and roof respectively
Rw the design behavior factor

The design force FV of the vertical mode is computed directly from:

ZICV β
FV = WV (3.12)
Rw

where
CV the vertical acceleration seismic coefficient which corresponds to the vertical mode
period TV
β The value of β (that portion of the horizontal acceleration to be used as vertical
acceleration) shall be 0.25 for seismic zones1, 2A and 2B; 0.50 for zone 3; and 0.60
for zone 4
WV water load, equal to:

1
WV = γΗ 2w (3.13)
2

The seismic coefficients of each mode are:

1.25S
CI = ≤ 2.75, for the impulsive mod e (3.14)
TI2 / 3

S
CC = , for the convective mod e ( AWWA D115 − 95)
TC
(3.15)
4S
CC = , for the convective mod e ( AWWA D110 − 95)
TC2

1.25S
CV = ≤ 2.75, for the vertical mod e (3.16)
TV2 / 3

- 24 -
Chapter 3 – Design of Circular Prestressed Reservoirs

where the coefficient S accounts for the soil class of the site of interest.

The design forces for the impulsive and convective mode are given from the following expressions:

VI
FI = (3.17)
πR

VC
FC = (3.18)
πR
To compute the total force applied to the tank FT, the SRSS combination rule can be used as follows:

FT = FI2 + FC2 + FV2 (3.19)

The above seismic coefficients are compatible with US based spectra and probably accounts for the
different damping ratios of each mode. The values of the impulsive and vertical seismic coefficients
for short periods are constant and equal to 2.75, a value which corresponds to the constant acceleration
period range. The seismic coefficient of the impulsive mode used for the design is the maximum
obtained in the period range 0 to 0.3 sec, unless a more rational computation of the impulsive mode
period is made, which also results in constant seismic coefficient of the impulsive mode period for
short period tanks.

In order to obtain seismic coefficients compatible with European ground excitations and obtain more
realistic seismic coefficients for short periods, one could use the EC8 design spectrum.

For the horizontal components of the seismic action the horizontal elastic response spectrum, Se(T), is
defined by the following expressions [8]:

  T 
α g S 1 + ( n 2.5 − 1)  , for 0 ≤ T ≤ TB
  TB 
α Sn 2.5 , for TB ≤ T ≤ TC
 g
S d (T ) =   TC  (3.20)
α g Sn 2.5   , for TC ≤ T ≤ TD
 T 
 T T 
α g Sn 2.5  C 2D  , for TD ≤ T
  T 

- 25 -
Chapter 3 – Design of Circular Prestressed Reservoirs

where
T vibration period of a linear single-degree-of-freedom system
αg design ground acceleration on Type A ground
TB, TC limits of the constant spectral acceleration branch
TD value defining the beginning of the constant displacement response range of the spectrum
S soil factor
η damping correction factor with reference value η=1 for 5% viscous damping given as:

10
n= ≥ 0.55 (3.21)
5 +ζ

The values of the periods TB, TC and TD and of the soil factor S describing the shape of the elastic
response spectrum depend on ground type and are shown in Table 3.4 and Table 3.5.

Table 3.4 : Values of the Parameters Describing the Recommended Type 1 Elastic Response Spectrum.

Ground Type S TB (sec) TC (sec) TD (sec)


A 1.0 0.15 0.40 2.0
B 1.2 0.15 0.50 2.0
C 1.15 0.20 0.60 2.0
D 1.35 0.20 0.80 2.0
E 1.4 0.15 0.50 2.0

Table 3.5 : Values of the Parameters Describing the Recommended Type 2 Elastic Response Spectrum.

Ground Type S TB (sec) TC (sec) TD (sec)


A 1.0 0.05 0.25 1.2
B 1.35 0.05 0.25 1.2
C 1.50 0.10 0.25 1.2
D 1.80 0.10 0.30 1.2
E 1.60 0.05 0.25 1.2
*Note: The values to be ascribed to TB, TC, TD and S for each ground type and type (shape) of spectrum to be used in a
Country may be found in its National Annex. The recommended choice is the use of two types of spectra: Type 1 and Type 2.
If the earthquakes that contribute most to the seismic hazard defined for the site for the purpose of probabilistic hazard
assessment has a surface-wave magnitude, Ms, not greater than 5.5, it is recommended that the Type 2 spectrum is adopted.

- 26 -
Chapter 3 – Design of Circular Prestressed Reservoirs

Table 3.6 : Values of the Parameters Describing the Vertical Elastic Response Spectrum.

Ground Type αvg/αg TB (sec) TC (sec) TD (sec)


Type 1 0.90 0.05 0.15 1.0
Type 2 0.45 0.05 0.15 1.0

For the vertical component of the seismic action the elastic spectrum used for design is given by the
Equation (3.20), with the design ground acceleration in the vertical direction αvg replacing αg, S taken
equal to 1.0 and TB, TC and TD as in Table 3.6.

The horizontal and vertical elastic spectra normalized to peak ground acceleration are shown in Figure
3.6 and Figure 3.7 respectively.

Figure 3.6 : Horizontal Elastic Spectrum for Types 1&2.

- 27 -
Chapter 3 – Design of Circular Prestressed Reservoirs

Figure 3.7 : Vertical Elastic Spectrum for Types 1&2.

The reduction factor q’ is given as:

 T 
1 +   ( q − 1) , for 0 ≤ T ≤ TB
q ' (T ) =   TB  (3.22)
q , for TB ≤ T

where
q behavior factor

The design seismic coefficients from Equation (3.20) using the reduction factor q’ are modified to:

S d (T )
S 'd (T ) = (3.23)
q ' (T )

The seismic coefficients for the impulsive mode obtained from AWWA are constant for a period range
between 0.0 and 0.3 sec. Since the period of concrete tanks is very small the seismic coefficients are
obtained from this plateau which has an amplification value equal to 2.75. Using a reduction factor of
3.00 the seismic coefficient for the design of the tanks is reduced to 2.75/3.00=0.92 and hence the

- 28 -
Chapter 3 – Design of Circular Prestressed Reservoirs

horizontal acceleration for which the tanks are designed is in the order of 0.92 PGA for the tanks
which lie in the period range mentioned.

EC8 [9] makes use of a behavior factor equal to 1.00 and up to 1.50 (accounting for overstrength)
unless a dissipative mechanism is used whose action shall be properly substantiated. The design uses a
horizontal acceleration equal to the peak ground acceleration (PGA) for the case of rigid tanks
(assuming Tr =0) a value very close to the previously mentioned 0.92 PGA which AWWA uses.

Figure 3.8 shows also that if a behavior factor equal to 3 is used for the design of the tanks then the
design horizontal acceleration will have a value between 2.50/3.00=0.833 and 1.00 times the peak
ground acceleration depending on the period of the tank.

If instead of using a behavior factor equal to 3, a reduction factor equal to 3 is used independent of the
period of the tank, the design horizontal acceleration will be far smaller for the case of short period
tanks and hence is not giving a realistic estimate of the seismic loading of the tank (see Figure 3.8).

It is known that at short period range the reduction factor used from current codes is less than the
reduction factor used for the design of structures with higher period because the short period range
structures have very high ductility demand (especially the elasto-plastic systems, see [6]) if a behavior
factor is used. On the other hand the reserve capacity of the tanks always exists from the factors
mentioned (limit stress, distribution of resistance between springs and vertical beams etc) and hence a
design value for the horizontal acceleration less than the peak ground acceleration may be reasonable.

For the purposes of this work the seismic coefficients obtained from AWWA are used which are very
close to the values that EC8 suggest (using a behavior factor of 1.00 and adopting a zero period for the
case of rigid tanks, the response of which this work is investigating).

The seismic coefficients used for the convective mode are obtained from AWWA D110-95 and not
from D115-95 since they seem to be more realistic for the period range above 5, common period
values for the convective mode (see Figure 3.9). The behavior factors implied from AWWA are used
also for the convective mode for consistency (for completeness it is mentioned that EC8 [9] does not
support any reduction of the seismic coefficients for the convective mode).

- 29 -
Chapter 3 – Design of Circular Prestressed Reservoirs

Figure 3.8 : Seismic Coefficients.

Figure 3.9 : Seismic Coefficients for the Convective Mode.

- 30 -
Chapter 3 – Design of Circular Prestressed Reservoirs

3.2.2.4 SLOSHING HEIGHT

During a seismic event, the maximum water surface displacement (sloshing height) may impinge on
the underside of the roof slab. The sloshing height taking into account only the first convective mode
is [13]:

d max = 0.841RSe (TC1 ) (3.24)

If a freeboard height is not provided to prevent uplift forces due to sloshing, the tank roof and its
connections shall be designed for the uplift.

3.2.3. DESIGN FOR VERTICAL AND HOOP DIRECTION

The model used for the analysis under the described loadings is shown in Figure 2.2b. Assuming unit
length for the struts the cross-sectional area of each strut is given as:

t ∆h
Ai = (3.25)
R2

The forces of each loading pattern occur as:

Force = pressure x ∆h (3.26)

The results of the analysis can be taken from a static analysis program. The analysis is stopped when
prestress loading ensures the stress limit for the circumferential direction by computing the maximum
hoop stresses using the Equations (2.7) and (2.10).

The maximum vertical bending moment M max occurred is used to design the vertical beam members.

The vertical bending stress is:

6M max
f max = (3.27)
t2

- 31 -
Chapter 3 – Design of Circular Prestressed Reservoirs

Assuming a percentage α* of losses of the pretension force of the wire strand Pi the strand spacing is:

s=
(1 − α ) Pi
(3.28)
f max t

The applied circumferential prestress force varies linearly from top to bottom and can be found from
the radial displacements ∆r occurred under the initial 50% prestress force as:

∆ r Ec t
F= ( x 2 for 100% prestress ) (3.29)
R

*a value of 20% is used for α in this work

- 32 -
Chapter 3 – Design of Circular Prestressed Reservoirs

3.3. APPLICATION TO DIFFERENT TANK CONFIGURATIONS

The dimensions of the tanks adopted for this study are based on common tank dimensions [5]. Three
different ratio of radius to height of water in the tank are adopted, R/H=1, 2, 3 with water height equal
to 5m, and hence with tank radius of 5m, 10m and 15m respectively. Because of the displacement of
the water due to the sloshing mode of the tank the actual tank height adopted for all the cases is 5.5m.

3.3.1. MEMBER INFORMATION

The tanks assumed for the design consist of precast panels with prestressed tendons in vertical and
circumferential direction. The wall, bearing pad and seismic cable dimensions used are presented in
Table 3.7

Table 3.7 : Member Dimensions.

Tank Walls

Wall Thickness, t 180mm

Roof Thickness, tr 140mm

Panel Length, Lpanel 2m

Seismic Cables

Length of Cable, Lc 600mm

Elastomeric Bearing Pads

Width, w 120mm

Bearing Pad Thickness, tb 30mm

Spacing of Bearing Pads, Sp 1m (for continuous pad)

Length of Bearing Pads, Lp 1m (for continuous pad)

- 33 -
Chapter 3 – Design of Circular Prestressed Reservoirs

3.3.2. MATERIAL PROPERTIES

Concrete, prestress steel, elastomeric bearing pad and seismic cable properties used for design
purposes are presented in Table 3.8.

Table 3.8 : Material Properties.

Prestress Tendons

Ultimate Strength, fpu 1860 MPa

Modulus of Elasticity, Es 200000 MPa

Circumferential Tendons 12 x 0.276” (601kN initial)


½” 7 wire strand (115.2kN Pretension)
Vertical Tendons Cross-Sectional : Area98.7mm2

Losses
Circumferential 15% strain losses
20% curvature losses

Vertical 20% total

Concrete
Concrete Design Strength, f c 40MPa

Probable Strength at EQ , f c' 1.4 f c = 56 MPa

Modulus of Elasticity, Ec 5000 f c' = 37417 MPa

Unit Weight, kN/m3 23.5

Seismic Cables

Ultimate Strength, fpu 1860 MPa

Modulus of Elasticity, Es 200000 MPa

Cable Cross-Sectional Area 98.7mm2

Elastomeric Bearing Pads

Modulus of Elasticity, Ep 6895kPa (=1000psi x 6.895)

Shear Modulus, G 1500kPa

- 34 -
Chapter 3 – Design of Circular Prestressed Reservoirs

3.3.3. DESIGN CASES

Non-seismic design and seismic design for PGA=0.2, 0.3, 0.4 0.5g are carried out for all three wall-
foundation-connection types and the three different ratios assumed. The importance factor I is set to 1.

Figure 3.10 : Models Used for the Analysis

The hydrostatic and the initial prestress pressure are shown in Figure 3.11. The models which are used
for the analysis are shown in Figure 3.10 where the struts with unit length have a spacing of 0.5m. The
program software RUAUMOKO [4] is used for the static analysis under the forces obtained from the
pressure loading using Equation (3.26). The increase of the initial prestress pressure for each tank type
and ratio considered to provide the 1.4MPa residual compression stress limit in hoop direction is
shown in Table 3.9.

The maximum vertical bending moment occurs for the case of 100% prestress before losses. The
maximum vertical bending moment used for the design of vertical direction for each design case is
shown in Table 3.10.

(All results for radial displacements, vertical bending moment and hoop stresses are shown in
APPENDIX B)

- 35 -
Chapter 3 – Design of Circular Prestressed Reservoirs

1.4t/R=252/R kN/m

H Hw

p=53.96kN/m p+1.4t/R=53.96+252/R kN/m


Water Pressure Used Initial Prestressed Tendons
Pressure Used

Figure 3.11 : Water and Initial Prestress Pressures.

Table 3.9 : Increase of Initial Prestress for Non Seismic Design Case.

R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

0.5% 0.5% 1.5% 2.0% 2.6% 3.0% 5.7%

Table 3.10 : Non-Seismic Design Values for Vertical Bending Moment in kNm.

R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

13.64 5.12* 20.46 7.69 34.05** 10.76


*t=0.13 minimum from AWWA and increased moment to have smaller spacing of the strands=0.39m
from 3.25kNm
**t=0.22 due to high compression strain

- 36 -
Chapter 3 – Design of Circular Prestressed Reservoirs

For each design PGA the design forces of each mode are computed using Equations (3.17), (3.18),
(3.12) for the impulsive, convective and vertical mode respectively. The impulsive mode periods for
the Type A, B, C tanks are shown in Table 3.11. The impulsive period for Type A and B are assumed
the same using Equation (3.5). The period is not affecting the design since being very small (less than
0.30 sec) the same seismic coefficient would be obtained if the impulsive period for the Type B tanks
was assumed greater. The convective and vertical mode periods for each R/H ratios are shown in Table
3.12 and Table 3.13 respectively.

The SRSS combination rule is used to obtain the total force applied to the tank wall. The total force is
slightly larger than the one obtained from the impulsive mode (see Table 3.14). To simplify the
analysis the force used for the design assumes that seismic pressure is coming only from the impulsive
mode but replacing the impulsive mode force with the total force obtained from the SRSS combination
rule. To compute the seismic pressure, the impulsive height is used from Equation (2.20).

The prestress pressure used for the non-seismic design case is increased to ensure the zero hoop
tension for each of the seismic design case (see Table 3.15).

Table 3.11 : Impulsive Period in sec.

R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

0.036 0.036 - 0.045 0.045 - 0.066 0.066 0.168*


*See Chapter 5 for number of cables used to find this period.

Table 3.12 : Convective Period in sec.

R/Hw Ratio

1 2 3

3.39 5.49 7.75

Table 3.13 : Vertical Period in sec.

R/Hw Ratio

1 2 3

0.032 0.038 0.047

- 37 -
Chapter 3 – Design of Circular Prestressed Reservoirs

The design of the tanks show that for almost all the configurations of Type A and B tank, the static
case governs the design. Seismic design for 0.40g and 0.50g is needed only for the case of a ratio
R/H=3 of Type B tank (see Table 3.15 and Table 3.16). For Type C tanks only the seismic design case
for 0.50g with a ratio R/H=3 is carried out (see Chapter 5 for comments on Type C tanks).

Table 3.14 : Difference of Total Force Using the SRSS Combination and the Impulsive Mode Force.

R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

7% 8% 5.6% 5.6% 4.5% 4.8% 4.8%

Table 3.15 : Increase of Prestress for Seismic Design Relative to Non-Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

0.3 NO* NO - NO NO - NO NO -

0.4 NO NO - NO NO - NO 5% -

0.5 NO NO - NO NO - NO 15% 25%


*NO means that static design governs the design and hence no seismic design is needed according to
the code used.

Table 3.16 : Seismic Design Values for Vertical Bending Moment in kNm.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B C A B C A B C

0.3 NO NO - NO NO - NO NO -

0.4 NO NO - NO NO - NO 11.30 -

0.5 NO NO - NO NO - NO 12.38 -

- 38 -
Chapter 4 – Assessment of Tank Behavior

4. ASSESSMENT OF TANK RESPONSE

In this chapter the inelastic seismic response of the tanks designed under current design codes (see
Chapter 3) is assessed using inelastic analyses.

4.1. PUSHOVER ANALYSIS

In order to carry out the pushover analysis using the program RUAUMOKO the same model is used
but using members which can go inelastic deformations.

4.1.1. VERTICAL BEAM MEMBERS

To model the vertical beam members the moment-curvature curve of the section used is required. The
vertical beam members exhibit elastic linear behavior till cracking. After cracking the stiffness of these
members is reduced to cracked stiffness till the limit of proportionality of the prestress tendons. After
this point the stiffness is reduced even more and the only factor that attributes to the stiffness of the
members is the stiffness of the prestress tendons. If before reaching the limit of proportionality - but
cracking has occurred - the member is unloaded follows the same path back to cracking point and then
to the initial unstressed point due to the non linear elastic behavior of the member. To account for
these characteristics the hysteretic model MUTO from the library of RUAUMOKO is used for the
vertical beam members of the model. Each beam member is assumed to have a plastic hinge length of
half the member length to both ends. The moment-curvature curve of these members is obtained as
follows.

- 39 -
Chapter 4 – Assessment of Tank Behavior

The vertical members of the model were designed for the maximum vertical bending obtained from the
design which assumes zero tension. The cracking moment is obtained as:

I
M cr = ( f t + f max ) (4.1)
t/2

where ft is the concrete tension strength equal to

f t = 0.6 f c' (4.2)

The moment at tendon limit of proportionality can be obtained as follows:

As (= 98.7 mm 2 )
Tendon Area = (4.3)
spacing

The stress at limit of proportionality is:

f LP = 0.85 f pu (4.4)

The force of the tendon at fLP is:

T = Tendon Area ⋅ f LP (4.5)

The compression zone length is:

T
α= (4.6)
0.85 f c' ⋅1.0

The neutral axis depth is:

α
c= (4.7)
0.65

- 40 -
Chapter 4 – Assessment of Tank Behavior

Finally the moment capacity at the limit of proportionality is:

M LP = T ( t / 2 − α / 2 ) (4.8)

To find the moment-curvature curve of the section (see Figure 4.1), the cracking and at limit of
proportionality curvature are computed using the following procedure. The incremental steel stress is:

∆f s = f pu − f si (4.9)

where fsi is the initial prestress after losses 20% computed as:

0.80 Pini
f si = (4.10)
As

where Pini is the initial pretension.


The incremental steel strain is:

∆f s
∆ε s = (4.11)
Es

The concrete compression strain must be always checked not to exceed the value of 0.003 as:

c
concretecompressionstrain = ∆ε s < 0.003 (4.12)
(t / 2 − c )

The “plastic” curvature is:

∆ε s
φp = (4.13)
(t / 2 − c )

The cracking curvature is:

εc
φcr = (4.14)
t/2

- 41 -
Chapter 4 – Assessment of Tank Behavior

where εc is the concrete strain at cracking equal to:

( f t + f max )
εc = (4.15)
Ec

The uncracked stiffness of the curve is obtained as:

M cr
kucr = (4.16)
φcr

The cracked stiffness of the curve is obtained as:

M LP − M cr
kcr = rkucr = (4.17)
φp

and the ratio of cracked to uncracked stiffness is:

kcr
r= (4.18)
kucr

The stiffness after the limit of proportionality is exceeded is assumed to be 0.05 times the uncracked
stiffness.

Moment

MLP

Mcr 0.05kucr
kcr

kucr

φcr φLP
Curvature

Figure 4.1 : Moment-Curvature Curve of Vertical Beam Members for Pushover Analysis.

- 42 -
Chapter 4 – Assessment of Tank Behavior

4.1.2. RADIAL SPRINGS

To model the radial springs the radial force-displacement curve of each spring is required. The radial
spring exhibit the same behavior described for the vertical members. Hence the MUTO hysteretic
model could be used again replacing the moment with the radial force and curvature with radial
displacement. Even if RUAUMOKO supports the MUTO hysteretic model for springs, it was found
that using this hysteretic model the stiffness of the springs is overestimated. To model each radial
spring, two springs in parallel are used with the BI-LINEAR hysteretic model with their characteristics
shown in Figure 4.3. The sum of the bi-linear curves gives a tri-linear curve. Because the pushover
analysis uses monotonic loading, modeling of the unloading path after cracking and before the limit of
proportionality is not required.

To find the radial springs’ characteristics first the total tri-linear radial force-displacement curve is
required using the following procedure. The assumptions made are that, conservatively, cracking
occurs at precast panel joints at zero tensions stress and cracking is taken place only at panel joints and
not inside panels. It is assumed a length of 30db for bond slip either side of crack (average stress). The
prestress after losses are:

f si = ( 0.80 ⋅ 0.85 ) 0.80 f pu = 0.544 f pu




(4.19)
losses initial

To find the characteristics of the total tri-linear radial force-displacement curve till and after the limit
of proportionality till failure the following assumptions are made. It is assumed that, conservatively,
failure occurs at strain 0.03 of the prestressed tendons [7]. The stress strain curve for low relaxation
prestress steel given from Equation (4.20) is simplified with a BI-LINEAR one (see Figure 4.2).The
post-yield stiffness of the bi-linear curve is assumed to be equal to the exact “post-yield” stiffness
(~5000MPa) and the prestress “yields” at stress equal to the limit of proportionality which is obtained
from the intersection of the two curves consisting the bilinear one and has a value of around
f LP = 1695MPa=0.91f pu and not equal to f LP = 0.85f pu . This stress value is obtained for a strain
very close to the exact strain of the limit of the proportionality and the small loss of prestress which
this implies is reasonable for seismic design. This assumption is not made for the beam members since
bending moments are not critical and the point of the limit of proportionality is not reached.

 
 0.975 
f p = 200000ε p 0.025 + 0.10  ≤ 1860 MPa (4.20)
 1 + (118ε )10  
  p
 

- 43 -
Chapter 4 – Assessment of Tank Behavior

Figure 4.2 : Stress-Strain Curve for Low Relaxation Prestressed Steel

Hence, assuming f LP = 0.91 f pu , the incremental stress is:

∆f s = f LP − 0.544 f pu = 0.367 f pu (4.21)

The incremental steel strain is:

∆f s
∆ε s = (4.22)
Es

The extension (crack width) is:

∆l = 2 ⋅ 30db ∆ε s ( d b ≈ 25mm ) (4.23)

The effective circumferential strain is:

ε s ,eff = ∆l / Lpanel (4.24)

- 44 -
Chapter 4 – Assessment of Tank Behavior

The equivalent radial deformation is:

∆R = ε s ,eff R (4.25)

The incremental tendon force is:

∆P = Α sh ∆f s (4.26)

where Ash is the tendon area per m height.


But the incremental tendon force can be expressed also as:

∆f s 0.367 f pu
∆P = P =P = 0.675 P (4.27)
f si 0.544 f pu

The uncracked circumferential stiffness is:

Ec Ai E t ⋅1.0
kucr = ∆h = c ∆h = Ec t ∆h (4.28)
li 1.0

The cracked circumferential stiffness is:

∆P 0.675 f h ∆ht
kcr = = (4.29)
ε s ,eff ε s ,eff

The post-cracking stiffness ratio is:

kcr 0.675 f h
a= = (4.30)
kucr ε s ,eff Ec

where fh is the residual hoop stress.

The cracking hoop force is:

Pcr = f h ∆ht (4.31)

- 45 -
Chapter 4 – Assessment of Tank Behavior

The radial cracking force is:

Pcr
Pcr ,r = (4.32)
R

The radial force at the limit of proportionality is:

0.675 f h ∆ht
PLP ,r = Pcr ,r + (4.33)
R

The yield strain is of the bi-linear curve is:

f LP
εy =
Es (4.34)

The plastic strain to failure is:

ε pl = ε u − ε y (4.35)

The failure stress is equal to:

f u = f LP + 5000ε pl (4.36)

and the incremental stress is:

∆f s , pl = f u − f LP (4.37)

The extension (crack width) - with the same assumptions for bond slip as before - is:

∆l pl = 2 ⋅ 30dbε pl ( d b ≈ 25mm ) (4.38)

- 46 -
Chapter 4 – Assessment of Tank Behavior

The effective circumferential strain is:

ε s , pl ,eff = ∆l pl / L (4.39)

The equivalent radial deformation is:

∆R pl = ε s , pl ,eff R (4.40)

The incremental tendon force is:

∆Ppl = Α sh ∆f s , pl (4.41)

But the incremental tendon force can be expressed also as:

∆f s , pl f u − f LP f − 0.91 f pu
∆Ppl = P =P =P u (4.42)
f si 0.544 f pu 0.544 f pu

The post limit of proportionality circumferential stiffness is:

∆Ppl f u − 0.91 f pu f h ∆ht


ku = = (4.43)
ε s , pl ,eff 0.544 f pu ε s , pl ,eff

The post limit of proportionality circumferential stiffness ratio is:

ku
r= (4.44)
kucr

The radial force at failure is:

f u − 0.91 f pu f h ∆ht
Pu ,r = PLP , r + (4.45)
0.544 f pu R

- 47 -
Chapter 4 – Assessment of Tank Behavior

To find the radial springs’ characteristics the Equations (4.46), (4.47), (4.48) must be used.

kucr = k1 + k2 (4.46)

where k1, k2 are the initial Stiffnesses of the bi-linear springs 1,2 respectively.

kcr = k1 + r2 k2 = akucr (4.47)

where r2 is the post-yield stiffness ratio for spring 2.

ku = r2 k2 = rkucr (4.48)

since the post-yield ratio stiffness for spring 1 r1 is zero.

k1, k2 and r2 are obtained using the Equations (4.49), (4.50), (4.51) after solving the system of
Equations (4.46), (4.47), (4.48).

k1 = kucr − k2 (4.49)

k2 = (1 − a + r ) kucr (4.50)

ku
r2 = (4.51)
k2

The yield forces Fy1, Fy2 for the bi-linear springs are found as follows:

Pcr ,r = ( k1 + k2 ) δ cr (4.52)

PLP ,r = k1δ LP + k2δ cr + r2 k2 (δ LP − δ cr ) (4.53)

- 48 -
Chapter 4 – Assessment of Tank Behavior

where the cracking displacement is:

Pcr ,r
δ cr = (4.54)
kucr

and the displacement at the limit of proportionality is:

PLP ,r − Pcr ,r + kcrδ cr


δ LP = (4.55)
kcr

The yield force F1 is:

F1 = k1δ LP (4.56)

The yield force F2 is:

k2
F2 = Pcr ,r − k1δ cr = Pcr ,r (4.57)
ku

- 49 -
Chapter 4 – Assessment of Tank Behavior

k1

k2

Radial Force

Fy2

+ r2k2
Fy1
k2
k1

δcr δLP δu
Radial Displacement

||
Radial Force
Pu,r
PLP,r

Pcr,r ku
kcr

kucr

δcr δLP δu
Radial Displacement

Figure 4.3 : Radial Spring Force-Displacement Curves of Radial Springs for Pushover Analysis.

- 50 -
Chapter 4 – Assessment of Tank Behavior

4.1.3. PUSHOVER ANALYSIS DETAILS

The pushover analysis is carried out using the load pattern coming from seismic code loading. The
pushover analysis using the program RUAMOKO is carried out as a dynamic analysis with a slow
linear ramp time function. This pattern multiplied by 4 is applied using a linear ramp with duration of
40 sec. Hence every 10 sec the design load pattern is applied to the model.

Due to the instability observed during preliminary analysis without using masses - probably because
the pushover analysis is carried in a dynamic loading manner - masses are assigned to each node of the
model (see Equation (4.59)) and hence a - very small - time step of 0.00005 is used due to the very
small higher mode period of the model.

To account for the static loading (prestress and water) the initial vertical bending moment are
subtracted from both positive and negative moment capacities of the beam elements and these
moments are added at the results obtained form the pushover analysis. The spring characteristics are
not changed but the static displacements are added to the seismic displacements to obtain the total
displacement profile of the model.

For completeness it is worth mentioning that the stiffness of radial springs after the point of limit of
proportionality is exceeded sometimes may be very small ( < 0.001kucr ) and RUAUMOKO cannot

model a so small post-“yield” stiffness. In this work in case of exceedance of the point of limit of
proportionality the total stiffness of some springs is zero and hence the results obtained from the
analysis are assumed conservative. From preliminary analyses, it was found that if a big number of
springs exceed the point of limit of proportionality instability occurred during the analysis. Failure of
the model is assumed when the first radial spring exceeds the failure point and at this time a small
number of springs has exceeded the point of limit of proportionality and hence instability does not
occur till failure of the tank.

4.1.4. PUSHOVER ANALYSIS RESULTS

The multiple of the seismic code Intensity that cracking, exceedance of the limit of proportionality and
failure occurs for the non-seismic and seismic design cases are shown in Table 4.1 - Table 4.6 (see
also APPENDIX C). Results and comments for Type C tanks are given in Chapter 5.

For the non-seismic design cases the multiple of seismic code Intensity obtained are proportional to a
seismic code loading for 0.50g. Hence for the case of the ratio R/H=3 Type B tank a safe PGA is

- 51 -
Chapter 4 – Assessment of Tank Behavior

0.75*0.5g=0.375g. For all the other cases, since static design governs the design, the results show that
tanks can sustain a higher value of PGA in terms of forces till cracking.

The multiple of seismic code Intensity that tanks can sustain till the point of the limit of proportionality
is for the tanks with ratios R/H=2 and 3 around 3 and with R/H=1 higher than 4. Hence the current
design is realistic for ratios R/H=2 and 3 and conservative for a ratio smaller or equal to R/H=1.

The multiple of seismic code Intensity that tanks can sustain till the failure point as that is defined
previously in this chapter is always higher than 3 and hence failure of a tank seems not be obtained
even for the cases where static design governs.

- 52 -
Chapter 4 – Assessment of Tank Behavior

Table 4.1 : Multiple of Seismic Code Intensity that Cracking occurs for the Non-Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.5 2.36 1.08 1.29 1.07 1.23 0.75

Table 4.2 : Multiple of Seismic Code Intensity that Cracking occurs for the Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.4 NO NO NO NO NO 1.07

0.5 NO NO NO NO NO 1.07

Table 4.3 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the Non-
Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.5 >4 >4 3.59 3.23 2.65 2.53

Time-history analyses of the tanks designed will be carried out to obtain a better understanding of the
seismic response of circular prestressed reservoirs.

- 53 -
Chapter 4 – Assessment of Tank Behavior

Table 4.4 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the Seismic
Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.4 NO NO NO NO NO 3.39

0.5 NO NO NO NO NO 3.12

Table 4.5 : Multiple of Seismic Code Intensity that Failure occurs for the Non-Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.5 >4 >4 >4 3.95 3.46 2.85

Table 4.6 : Multiple of Seismic Code Intensity that Failure occurs for the Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.4 NO NO NO NO NO 3.79

0.5 NO NO NO NO NO 3.44

- 54 -
Chapter 4 – Assessment of Tank Behavior

4.2. TIME-HISTORY ANALYSIS

To obtain the response of the tanks under dynamic loading time history analyses are carried out using
the program RUAUMOKO. The time-history analyses are investigating the multiple of seismic code
intensity that the tank can sustain without exceeding the limit of proportionality.

The model for the time-history analysis is the same with the one used for the pushover analysis, but the
hysteretic models used for both vertical beam members and radial springs are different.

4.2.1. VERTICAL BEAM MEMBERS

To model the vertical beam members for the time-history analysis instead of using the MUTO
hysteretic model and since the response till the point of the limit of proportionality is achieved is being
investigated, the NON-LINEAR ELASTIC model is used. This model loads elastically till a “yielding”
point with a stiffness equal to the uncracked stiffness of the section. After “yielding” the model
remains elastic but the loading curve has a reduced stiffness which is equal to the cracked stiffness of
the section (see Figure 4.4).

Moment

MLP

Mcr
kcr

kucr
-φLP -φcr

φcr φLP
Curvature

kcr
-Mcr

-MLP

Figure 4.4 : Moment-Curvature Curve of Vertical Beam Members for the Time-History Analysis.

- 55 -
Chapter 4 – Assessment of Tank Behavior

The “yielding” moment of this model is equal to the cracking moment given from Equation (4.1).
Since the model is not modeling the stiffness reduction after the point of the limit of proportionality is
exceeded, the moment given from Equation (4.8) is checked not to be exceeded after each analysis.
For both positive and negative moments the vertical beam members exhibit the same behavior.

4.2.2. RADIAL SPRINGS

To model the radial springs for the time-history analysis each radial spring is not modeled with two
springs but with one spring with the NON’LINEAR ELASTIC hysteretic model. The radial spring
exhibit the same behavior described for the vertical members (see Figure 4.5). The “yielding” radial
force is obtained from Equation (4.32). Since, again, the model is not modeling the stiffness reduction
after the point of the limit of proportionality is exceeded, the radial force given from Equation (4.33) is
checked not to be exceeded after each analysis.

Radial Force

PLP,r

Pcr,r
kcr

kucr

δcr δLP
Radial Displacement

Figure 4.5 : Radial Spring Force-Displacement Curve of Radial Springs for the Time-History Analysis.

- 56 -
Chapter 4 – Assessment of Tank Behavior

The difference for the radial springs compared to the vertical beam members, is that they are assumed
not to yield in the negative direction due to high compression resistance in the circumferential
direction. The limit for the compression stress in the circumferential direction is checked against [1]:

f compr ,lim it = 0.45 f 'cd = 0.45 ⋅ 56 = 25.2 MPa (4.58)

4.2.3. TIME-HISTORY ANALYSIS DETAILS

The time-history analysis is carried out using five (5) artificial accelerograms obtained using the
program SIMQKE [5]. The artificial accelerograms are generated to match the EC8 design spectrum
for a value of peak ground acceleration of 0.5g (see Figure 4.6) assuming duration of 45 sec, using
trapezoidal envelope parameters with acceleration rise time 10 sec and acceleration level time of
20sec. The artificial records (see Figure 4.7) are rich in high frequency pulses (also the number is quite
high) and hence the response obtained from the time-history analysis can be assumed conservative.

The time-history analysis for tanks designed for smaller peak ground acceleration is carried out with
the same accelerograms but reduced proportionally to match the peak ground acceleration needed for
each case [3]. For the case of the ratio R/H=3 Type B tank the time-history assumes a peak ground
acceleration of 0.30g for which is the value that static design governs against seismic design.

Figure 4.6 : Time-Histories Response Spectra Matching EC8 Type 1 Design Spectrum.

- 57 -
Chapter 4 – Assessment of Tank Behavior

Figure 4.7 : Time-History of EC805GA (ATH1) Artificial Accelerogram.

To account for the static loading (prestress and water) the initial vertical bending moment are
subtracted from both positive and negative moment capacities of the beam elements and these
moments are added at the results obtained form the pushover analysis. The spring characteristics are
not changed but the static displacements are added to the seismic displacements to obtain the total
displacement profile of the model.

The mass assigned to each node is proportional to the seismic load pattern. The total mass to be
distributed is obtained as:

Ws + Wr + WI 1
mtot = (4.59)
g πR

The damping considered in the analysis is a Rayleigh damping of 2% assuming tangent stiffness. This
value is appropriate for prestressed concrete structures and the results obtained are conservative
assuming that higher damping levels may exist.

- 58 -
Chapter 4 – Assessment of Tank Behavior

4.2.4. TIME-HISTORY ANALYSIS RESULTS

The multiple of the seismic code intensity that cracking and exceedance of the limit of proportionality
occurs for the non-seismic and seismic design cases are shown in Table 4.7 - Table 4.10 (see also
APPENDIX D). The final results are based on the mean value of maximum values obtained from the
analysis for each of the five artificial accelerograms used. Time-history analyses for Type C tanks are
not carried out (see Chapter 5).

The results obtained from the analysis show that cracking occurs at 0.5-1.0 times the intensity the
design assumes for each case. The ratio R/H=3 tanks seem to crack at 0.5 times the intensity the design
assumes. The design of tanks with this ratio should use greater values of seismic loading if cracking
needs to be avoided. This is may be due to the flexibility of these tanks which imposes higher
horizontal acceleration than the peak ground acceleration. A design horizontal acceleration equal to
1.00 times the peak ground acceleration might prevent cracking at Intensity level less than 1.0.

The limit of proportionality for each case is exceeded for Intensity level at least 2.5 times greater than
the design intensity. This Intensity level is increasing with decreasing radius to height ratio. Especially
for the ratio R/H=1 and the Type A tank this intensity reaches a value of 6.5.

Table 4.7 : Multiple of Seismic Code Intensity that Cracking occurs for the Non-Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.5 >1.0 ~1.0 1.0 0.5-1.0 0.5-1.0 ~0.5

- 59 -
Chapter 4 – Assessment of Tank Behavior

Table 4.8 : Multiple of Seismic Code Intensity that Cracking occurs for the Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.4 NO NO NO NO NO ~0.5

0.5 NO NO NO NO NO ~0.5

Table 4.9 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the Non-
Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.5 ~6.5 ~4.0 ~3.5 ~3.0 ~2.5 ~3.0*


*Intensity o f the analysis is referred to PGA=0.3g

Table 4.10 : Multiple of Seismic Code Intensity that the Limit of Proportionality is exceeded for the
Seismic Design Case.

PGA(g) R/Hw Ratio

1 2 3

Type Type Type

A B A B A B

0.4 NO NO NO NO NO ~2.5

0.5 NO NO NO NO NO ~2.5

- 60 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

5. DESIGN AND ANALYSIS OF SLIDING BASE TANKS

This chapter describes the design and analysis of sliding base tanks (Type C Tanks).

5.1. SLIDING BASE TANKS

This type of tank cannot be analyzed with the model used for the analysis of Type A and B tanks. The
connection of the tank-wall foundation consists of a bearing pad (continuous or not) along the wall
base of the tank and seismic cables (strands) which offer only tangential (and not any radial) resistance
in the horizontal direction and hook action in the vertical direction, again, along the wall base of the
tank since the cables are embedded diagonally in the tank walls.

The wall is free to slide on the bearing pad. Hence the shear force from the wall to the foundation is
transferred through friction till the tanks slides. The axial load imposed on the bearing pad due to the
overturning moment is different along the wall base and hence the friction force is different in each
point of the wall base and the tank is not sliding simultaneously to all the points of the base.

The seismic cables are not offering any resistance for static loading, since static loading is symmetrical
around the wall tanks and imposes only radial deformation on the tank walls. During seismic
excitation, loading is not any more symmetrical and hence the seismic cables are activated but their
contribution is different along the wall base and depends on the seismic excitation angle. If seismic
loading is significant the seismic cables may yield.

- 61 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

All these reasons make a 3D modeling of the tank compulsory. For analysis purposes the Non-Linear
SAP program is used. The model used is described as follows.

The tank walls consist of shell elements with thickness for both membrane and bending behavior equal
to the thickness of the tank wall. The element used is an elastic shell element assuming thick plate.

For simplicity the wall-foundation connection consist of 10 nodes for every quarter of the tank
perimeter. Hence one base node exists every 9 degrees. The wall-foundation connection on each base
node consists of a friction-pendulum isolator link element, hook spring link element and a Wen
Plasticity Spring link element.

The friction-pendulum isolator link element is a biaxial friction-pendulum isolator that has coupled
friction properties for the two shear deformations, post-slip stiffness in the shear directions due to the
pendulum radii of the slipping surfaces, gap behavior in the axial direction, and linear effective-
stiffness properties for the three moment deformations. This element is used for the bearing pads’
modeling. In vertical direction the vertical stiffness of the bearing pad is assigned. In the two shear
directions the horizontal stiffness of the bearing pad is assigned (the same in both DOFs). The friction
coefficient is set to 0.5 (maximum possible [1]) and the radii of the sliding surface are set to zero since
bearing pad has a flat surface and does not have restoring capacity (see Figure 5.1). The local axes of
this element matches the global axes of the model so that the x-direction of the element matches the
direction to which seismic loading is applied. The horizontal stiffness of the bearing pad per m is
obtained from Equation (5.1) and the vertical stiffness of the bearing pad is obtained from Equation
(5.2).

Gw
kh , pad = (5.1)
tp

Epw
kv , pad = (5.2)
tp

The hook spring link element is used for the vertical stiffness of seismic cables’ modeling (see Figure
5.2) to capture the hook action of seismic cables in case of overturning.

The Wen Plasticity Spring link element is used for the transverse stiffness of seismic cables’ modeling
which acts tangentially to each base node (see Figure 5.3). The properties of the model are obtained
using the bilinear curve of Figure 4.2.

- 62 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

The horizontal stiffness of seismic cables is obtained assuming one of the diagonal cables of each pair.
The vertical stiffness of seismic cables is obtained assuming both the diagonal cables of each pair (see
Figure 3.5).

P
No pendulum
Action

Friction & Hook Horizontal


Action Stiffness of
Bearing Pad

P
1
3
2 P
P
Modeling of each horizontal direction

Figure 5.1 : Modeling of Bearing Pad (Friction-Pendulum Isolator Link Element of SAP).

Opening = 0 (assumed)

Vertical Stiffness
of Seismic Cable
P

Figure 5.2 : Modeling of Vertical Resistance of Seismic Cables (Hook Spring Link Element of SAP).

- 63 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

Figure 5.3 : Modeling of Horizontal Resistance of Seismic Cables (Wen Plasticity Link Element of SAP).

The horizontal stiffness of each cable is obtained from Equation (5.3) and the vertical stiffness of each
cable is obtained from Equation (5.4). The yield load of each cable is obtained from Equation (5.5).

Es As cos 2 α
kh ,cable = (5.3)
Lc

Es As sin 2 α
kv ,cable = (5.4)
Lc

Fy ,cable = f LP As cos α , f LP = 1695MPa ( assumed ) (5.5)

The design procedure for the sliding base tanks is the same with the design procedure followed for the
fixed and hinged tank cases. Hence prestress must be applied so as to provide 1.4 MPa residual
compression for the case of static design and zero hoop tension for the case of seismic design. For the
case of seismic design seismic cables are designed not to exceed a limit stress of 70%fpu.

From preliminary analysis it was found that the tank is not sliding for the code seismic loading. Hence
the cables can be designed assuming the horizontal stiffness of the bearing pad and the sum of
horizontal stiffness from the contribution of each seismic cable. Equation (3.7) implies that the
contribution of seismic cables to the total horizontal stiffness is equal to half the number of cables
times the horizontal stiffness of each cable:

- 64 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

kh ,total ,cable = 0.5nkh ,cable (5.6)

The total horizontal stiffness of the system is:

kh = kh , pad + 0.5nkh ,cable (5.7)

Since the bearing pad and seismic cables act in parallel, the shear which seismic cables sustain can be
found. The number of the cables can be found after two-three iterations using the following equations:

0.5nkh ,cable
V = Vtot ,cable (5.8)
kh

Vtot ,cable
< 70% f pu (5.9)
As 0.5n cos α

After the design of the tank the model can be analyzed by applying each load (water prestress and
seismic) as surface pressure to the shell elements. Gravity loads coming from wall and roof weights
must be assigned to base nodes.

The number of cables used for the design of a tank with R/H=3 and properties given at Table 3.7 and
Table 3.8 is 84*2=168.

The scope of the analysis of the tank is to find the multiple of seismic code loading that the tank can
sustain. As it is mentioned before, the tank is not sliding for 1.00*code seismic loading. This is due to
the flexibility of the bearing pad of the model. The bearing pad deforms and also due to the axial load
from gravity loads the friction force is quite high and hence, sliding is not occurring. This results in
increasing stresses of the walls due to the constraint action of the bearing pad (till sliding). After
applying a multiple of code seismic loading using a pushover analysis, it is found that sliding occurs at
around 1.40*code seismic loading. Sliding occurs only to half of the wall base since the axial load is
increasing to the other half and sliding is avoided. At around 1.675*code seismic loading the analysis
is stopped since seismic cable have failed. SAP program does not support nonlinear shell elements
with prestress. Using the elastic shell element it is found that the walls have high tensile stresses and
hence cracking of the walls may be occurring before the seismic cables fail (see Figure 5.4).

- 65 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

Tank under Prestress and


Water Loading
Radial Deformation Only

Tank at 0.00*Code
Zero Hoop Tension Limit
at Tank Walls
No sliding has occurred

Tank at 1.00*Code

Tension Stresses at Tank Walls


Sliding is occurring only at the left
part of the wall base

Tank at 1.40*Code
The analysis is stopped
at this point since
Seismic Cables Failed
All the left part of the wall base
is sliding

Tank at 1.675*Code

Figure 5.4 : Pushover Analysis of Sliding Base Tank Using SAP Program.

- 66 -
Chapter 5 – Design and Analysis of Sliding Base Tanks

To capture the inelastic seismic response of a sliding base tank a more accurate model is needed
involving the prestress action and cracking of the concrete for the reasons mentioned at this chapter.

- 67 -
Chapter 6 – Conclusions

6. CONCLUSIONS

This project investigates the inelastic seismic response of circular prestressed concrete reservoirs tanks
of different configurations designed under current design codes [1,2,8,9] using a beam on elastic
foundation model [10, 14, 15].

Modeling assumptions are presented for the analysis of cylindrical concrete storage tanks as those are
covered by standard texts on the design of circular reservoirs. A comprehensive review and application
of the Force-Based Design approach of circular prestressed reservoirs structures when subjected to
seismic action in the transverse direction using simplified load shapes of seismic loading is presented.
The design procedure to use with this type of structure is presented first, followed by a series of
applications to different possible tank configurations considering stiff soils to neglect soil-structure
interaction effects. Design of cylindrical prestressed reservoirs including soil-structure effects with the
model adopted in this project needs further investigation since this model does not account for soil
flexibility.

The design of cylindrical prestressed reservoirs including seismic loading reveals that seismic design is
not critical, especially for low radius to height ratios (R/H=1 to 2), even for high values of seismic
intensity the design uses (PGA=0.50g).

Assessment of the tanks designed using pushover and time-history analyses is carried out. Models
including inelasticity for pushover and time-history analyses are presented. Members which contribute
to flexural and hoop resistance after cracking can sustain large deformations till the point of limit of
proportionality and a failure limit defined. Having non-linear elastic characteristics makes cracks to
close if the point of limit of proportionality is not exceeded.

- 68 -
Chapter 6 – Conclusions

Pushover analysis indicates the multiple of code seismic load that each tank can sustain revealing the
effectiveness of code reduction factors used for their design.

Incremental dynamic analysis is carried out by using multiple of time-histories that codes impose
indicating the conservatism of current seismic design procedure of the circular prestressed concrete
reservoir tanks especially for a safety limit state due to the non-linear characteristics and the reserve
capacity of tanks of both flexural and hoop resistance of tank model members.

Tanks with higher radius to height ratios (R/H>3) are more vulnerable to seismic excitation. Due to
their greater flexibility compared to tanks with lower radius to height ratios smaller reduction factors
should be used for a serviceability limit state which assumes no cracking at intensity the design
assumes.

Some design and analysis aspects of sliding base tanks are presented which due to their geometrical
and material properties a sophisticated analysis using more complex models is needed. Further
investigation is required on the inelastic seismic response of this type of tanks.

- 69 -
Bibliography

BIBLIOGRAPHY

1. American Water Works Association [1996]. AWWA Standard for Circular Prestressed Concrete
Water Tanks with Circumferential Tendons, ANSI/AWWA D115-95, First Edition.
2. American Water Works Association [1996]. AWWA Standard for Wire and Strand-Wound
Circular, Prestressed Concrete Water Tanks, ANSI/AWWA D110-95, Revision of
ANSI/AWWA D110-86.
3. Bommer, J. J., Acevedo. A. B. [2004]. “The Use of Real Earthquake Accelerograms as Input to
Dynamic Analysis”. Journal of Earthquake Engineering, 8 (Special Issue 1).
4. Carr, A. [2002]. RUAUMOKO Users Manual. University of Canterbury, Chirstchurch, New
Zealand, May 2002.
5. Carr, A., [2002]. SIMQKE Users Manual. University of Canterbury, Chirstchurch, New
Zealand, May 2002.
6. Chopra, A. K. [2000]. Dynamics of Structures. Prentice-Hall international Series in Civil
Engineering and Engineering Mechanics, Second Edition, Prentice-Hall, Upper Saddle River,
NJ 07458.
7. Collins, Michael P and Mitchell, Denis [1997]. Prestressed Concrete Structures, Response
Publications, Ontario, Canada. 766 pag.
8. Eurocode 8: Design of Structures for Earthquake Resistance [2002]. Part 1: General Rules,
seismic actions and rules for buildings. Draft No5, Revised Final Project Team Draft (pre Stage
49), May 2002.
9. Eurocode 8: Design of Structures for Earthquake Resistance [2003]. Part 4: Silos, tanks and
pipelines. Draft No2, September 2003.
10. Hetenyi, M. [1946]. Beam on Elastic Foundations, University of Michigan Press, Ann Arbor,
Michigan 1946.

- 70 -
Bibliography

11. Housner, G.W. [1957]. “Dynamic Pressures on Accelerated Fluid Containers”, Bulletin,
Seismological Society of America, V.47, 1957, pp 15-35.
12. Jacobsen, L.J. [1949]. “Impulsive Hydrodynamics of Fluid Inside a Cylindrical Tank, and of
Fluid Surrounding a Cylindrical Pier”, Bulletin, Seismological Society of America, V.39, 1949,
pp 189-204.
13. New Zealand National Society for Earthquake Engineering, [1986]. Seismic Design of Storage
Tanks. Recommendations of a Study Group of the New Zealand National Society for
Earthquake Engineering, December 1986.
14. Priestley, M. J. N. [1985]. “Analysis and Design of Circular Prestressed Concrete Storage
Tanks”. PCI Journal, July-August 1985.
15. Priestley, M. J. N. [2002]. Improvements to Seismic Design of Circular Prestressed Concrete
Storage Tanks. Draft1, March 4, 2002.
16. Veletsos, AS [1984]. Seismic response and design of liquid storage tanks. Guidelines for the
seismic design of oil and gas pipeline systems. Technical Council on Lifeline Earthquake
Engineering ASCE NY 1984:255 –370, 443 –461.
17. Wood, J. H. [1999]. “Improvements to Seismic Design of Circular Prestressed Concrete Storage
Tanks”. Progress Report, 11 October, 1999.
18. Wood, J. H. [2000]. “Improvements to Seismic Design of Circular Prestressed Concrete Storage
Tanks”. Progress Report, 23 March, 2000.

- 71 -
Appendix A

APPENDIX A

- 72 -
Appendix A

Flexible Base Period to Rigid Base Period Ratios for different H/R, t/R and E/Gs ratios due to
Soil-Structure Interaction

Concrete Tanks:
Period Ratios for t/Rw = 0.02
2.2
Flexible Base Period / Rigid Base Period, T f/Tr

2.0 Ew/Gs = 200

1.8

1.6 100

1.4
50

1.2
25
1.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Liquid Height / Foundation Radius Ratio, H/R

Figure 2

Concrete Tanks:
Period Ratios for t/Rw = 0.01
1.7
Flexible Base Period / Rigid Base Period, T f/Tr

1.6

1.5 Ew/Gs = 200

1.4

1.3 100

1.2
50

1.1
25
1.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Liquid Height / Foundation Radius Ratio, H/R

Figure 3

- 73 -
Appendix A

Concrete Tanks:
Period Ratios for t/Rw = 0.005
1.40
Flexible Base Period / Rigid Base Period, T f/Tr

1.35

1.30 Ew/Gs = 200


1.25

1.20
100
1.15

1.10
50
1.05

1.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Liquid Height / Foundation Radius Ratio, H/R

Figure 4

- 74 -
Appendix A

Damping % for different H/R, t/R and E/Gs ratios due to Soil-Structure Interaction

Concrete Tanks:
Damping Values for t/Rw = 0.02

40%

35%
Ew/Gs = 200
30%

25%
Damping, %

100
20%

15% 50

10%

5% 25

0%
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Figure 5 Liquid Height / Foundation Radius Ratio, H/R

Concrete Tanks:
Damping Values for t/Rw = 0.01

30%

Ew/Gs = 200
25%

20%
Damping, %

100
15%

10% 50

5%
25

0%
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Figure 6 Liquid Height / Foundation Radius Ratio, H/R

- 75 -
Appendix A

Concrete Tanks:
Damping Values for t/Rw = 0.005

20%

Ew/Gs = 200
15%
Damping, %

10%
100

50
5%

0%
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Figure 7 Liquid Height / Foundation Radius Ratio, H/R

- 76 -
Appendix B

APPENDIX B

- 77 -
Appendix B

DESIGN RESULTS
HINGED BASE R/H=3 NON-SEISMIC DESIGN

- 78 -
Appendix B

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.50G

- 79 -
Appendix B

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.40G

- 80 -
Appendix B

HINGED BASE R/H=2 NON-SEISMIC DESIGN

- 81 -
Appendix B

HINGED BASE R/H=1 NON-SEISMIC DESIGN

- 82 -
Appendix B

FIXED BASE R/H=3 NON-SEISMIC DESIGN

- 83 -
Appendix B

FIXED BASE R/H=2 NON-SEISMIC DESIGN

- 84 -
Appendix B

FIXED BASE R/H=1 NON-SEISMIC DESIGN

- 85 -
Appendix C

APPENDIX C

- 86 -
Appendix C

PUSHOVER ANALYSIS RESULTS


HINGED BASE R/H=3 NON-SEISMIC DESIGN

- 87 -
Appendix C

- 88 -
Appendix C

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.50G

- 89 -
Appendix C

- 90 -
Appendix C

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.40G

- 91 -
Appendix C

- 92 -
Appendix C

HINGED BASE R/H=2 NON-SEISMIC DESIGN

- 93 -
Appendix C

- 94 -
Appendix C

HINGED BASE R/H=1 NON-SEISMIC DESIGN

- 95 -
Appendix C

- 96 -
Appendix C

FIXED BASE R/H=3 NON-SEISMIC DESIGN

- 97 -
Appendix C

- 98 -
Appendix C

FIXED BASE R/H=2 NON-SEISMIC DESIGN

- 99 -
Appendix C

- 100 -
Appendix C

FIXED BASE R/H=1 NON-SEISMIC DESIGN

- 101 -
Appendix C

- 102 -
Appendix C

HINGED BASE R/H=3 NON-SEISMIC DESIGN

- 103 -
Appendix C

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.50G

- 104 -
Appendix C

- 105 -
Appendix C

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.40G

- 106 -
Appendix C

HINGED BASE R/H=2 NON-SEISMIC DESIGN

- 107 -
Appendix C

- 108 -
Appendix C

HINGED BASE R/H=1 NON-SEISMIC DESIGN

- 109 -
Appendix C

FIXED BASE R/H=3 NON-SEISMIC DESIGN

- 110 -
Appendix C

- 111 -
Appendix C

FIXED BASE R/H=2 NON-SEISMIC DESIGN

- 112 -
Appendix C

FIXED BASE R/H=1 NON-SEISMIC DESIGN

- 113 -
Appendix C

- 114 -
Appendix D

APPENDIX D

- 115 -
Appendix D

TIME-HISTORY RESULTS
HINGED BASE R/H=3 NON-SEISMIC DESIGN

- 116 -
Appendix D

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.50G

- 117 -
Appendix D

- 118 -
Appendix D

HINGED BASE R/H=3 SEISMIC DESIGN PGA=0.40G

- 119 -
Appendix D

HINGED BASE R/H=2 NON-SEISMIC DESIGN

- 120 -
Appendix D

- 121 -
Appendix D

HINGED BASE R/H=1 NON-SEISMIC DESIGN

- 122 -
Appendix D

FIXED BASE R/H=3 NON-SEISMIC DESIGN

- 123 -
Appendix D

- 124 -
Appendix D

FIXED BASE R/H=2 NON-SEISMIC DESIGN

- 125 -
Appendix D

FIXED BASE R/H=1 NON-SEISMIC DESIGN

- 126 -
Appendix D

- 127 -

You might also like