MSCE Thesis Final Draft (Baylon As of 02 04 2016) PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 124

Reliability Analysis of a bridge pier using

Interval Uncertainty Analysis

A Thesis presented to
the Faculty of Civil Engineering
College of Engineering
De La Salle University Manila

________________________________________
In Partial Fulfillment
of the Requirements for the Degree of
Master of Science in Civil Engineering
Major in Structural Engineering
________________________________________

BAYLON, MICHAEL BAUTISTA

Thesis Adviser:
DR. LESSANDRO ESTELITO O. GARCIANO

FEBRUARY 2016
TABLE OF CONTENTS
LIST OF FIGURES..................................................................................................................................... 4
LIST OF TABLE.......................................................................................................................................... 6
CHAPTER 1 ................................................................................................................................................ 7
PROBLEM SETTING............................................................................................................................. 7
1.1 Background of the Study ........................................................................................................... 7
1.2 Statement of the Problem........................................................................................................ 12
1.3 Significance of the Study ......................................................................................................... 12
1.4 Objective of the Study .............................................................................................................. 13
1.5 Scope and Limitations ............................................................................................................. 13
1.6 Assumptions in the Study ........................................................................................................ 14
CHAPTER 2 .............................................................................................................................................. 15
REVIEW OF RELATED LITERATURE ............................................................................................. 15
2.1 Interval Analysis ........................................................................................................................ 15
2.2 Fragility analysis ....................................................................................................................... 21
2.3 Seismic Fragility Analysis in the Philippines......................................................................... 26
2.4 Synthesis ................................................................................................................................... 28
CHAPTER 3 .............................................................................................................................................. 30
CONCEPTUAL AND THEORETICAL FRAMEWORK ................................................................... 30
3.1 Conceptual Framework............................................................................................................ 30
3.2 Theoretical Framework ............................................................................................................ 33
3.2.1 PGA Normalization ................................................................................................................. 33
3.2.2 Pushover Analysis (Nonlinear Static Analysis) .................................................................. 33
3.2.3 Time History Analysis (Nonlinear Dynamic Analysis) ....................................................... 34
3.2.4 Ductility Factors ...................................................................................................................... 35
3.2.5 Damage Index and Damage Rank ...................................................................................... 36
3.2.6 Interval Arithmetic Operations .............................................................................................. 36
3.2.7 Interval Uncertainty Analysis (IUA) ...................................................................................... 38
3.2.8 Probability of Exceedance .................................................................................................... 40

CHAPTER 4 .............................................................................................................................................. 42
RESEARCH METHODOLOGY .......................................................................................................... 42
4.1 Input data .................................................................................................................................... 42
4.4 Interval Uncertainty Analysis method of reliability analysis ................................................. 47
4.5 A library of MatLab functions implementing interval arithmetic operations ...................... 49
4.6 Research paradigm ................................................................................................................... 51
CHAPTER 5 .............................................................................................................................................. 58
RESULTS AND DISCUSSIONS ........................................................................................................ 58
5.1 Fragility Analysis ........................................................................................................................ 58
5.2 Probability of Occurrence ......................................................................................................... 63
5.3 Fragility Curves (Conventional) ............................................................................................... 73
5.4 Fragility Curves by Interval Uncertainty Analysis ................................................................. 76
5.4.1 Interval Uncertainty Analysis (Lower Bound) ..................................................................... 76
5.4.2 Interval Uncertainty Analysis (Upper Bound) ..................................................................... 80
5.5 Comparison of Fragility Curves: Conventional vs. IUA ........................................................ 85
5.6 Interval Uncertainty Analysis after First Pass ...................................................................... 105
CHAPTER 6 ............................................................................................................................................ 115
SUMMARY, CONLUSION, & RECOMMENDATION ................................................................... 115
6.1 Summary ................................................................................................................................... 115
6.2 Conclusion ................................................................................................................................ 115
6.3 Recommendation..................................................................................................................... 116
BIBLIOGRAPHY ..................................................................................................................................... 118

LIST OF FIGURES

Figure 1. Example of seismic fragility curves. ...................................................................................... 8


Figure 2. Reliability analysis results under a 4-subinterval-BPA-structure case. .......................... 15
Figure 3. Typical deformation behavior as seen in the finite element model................................. 16
Figure 4. A graph of displacement at the plastic hinge with interval holonomic elastoplastic
alpha-umax and alpha-umin responses..................................................................................................... 17
Figure 5. Single objective optimization of example wing 2 under landing loads in progress,
deterministic, probabilistic, interval analyses. ...................................................................................... 18
Figure 6. Probability of Failure for the Performance Function Z=R-S. .......................................... 19
Figure 7. Summary of Seismic Fragility Curves of some lifelines in CAMANAVA area of
Damage Rank As ..................................................................................................................................... 27
Figure 8. Summary of Seismic Fragility Curves of some lifelines in CAMANAVA area of
Damage Rank A ....................................................................................................................................... 28
Figure 9. Conceptual Framework of the Study .................................................................................... 31
Figure 10. Interval Uncertainty Analysis after the First Pass. .......................................................... 32
Figure 11. General Elevation Plan of the Bridge Showing Pier 2 ..................................................... 43
Figure 12. Elevation Plan of the Pier.................................................................................................... 43
Figure 13. Detail of Bored Pile .............................................................................................................. 44
Figure 14. Section A of the Bored Pile................................................................................................. 44
Figure 15. Section C (Coping)............................................................................................................... 45
Figure 16. Ground motion data of Bohol October 2013 earthquake normalized in 0.2g. ............ 47
Figure 17. Normal probability function assumption of an interval value. ........................................ 49
Figure 18. Research paradigm ............................................................................................................. 51
Figure 19. Pushover curve from nonlinear static analysis of SAP2000.......................................... 58
Figure 20. Pushover curve (zoomed in) to compute the energy at yield Ee................................... 59
Figure 21. Hysteresis Model of Tohoku-Kanto Fukushima-2g (1 out of 300 hysteresis model)
using SAP2000. ........................................................................................................................................ 60
Figure 22. Hysteretic energy computation using Autodesk's AutoCAD. ........................................ 61
Figure 23. Probability of Occurrence for Conventional Fragility Curves ......................................... 65
Figure 24. Probability of Occurrence for IUA-Lower Bound Fragility Curves ................................ 65
Figure 25. Probability of Occurrence for IUA-Upper Bound Fragility Curves ................................ 66
Figure 26. Plot of Lognormal of PGA to the Damage Ratio of DR="D". ......................................... 67
Figure 27. Plot of Lognormal of PGA to the Damage Ratio of DR="C". ......................................... 68
Figure 28. Plot of Lognormal of PGA to the Damage Ratio of DR="B". ......................................... 69
Figure 29. Plot of Lognormal of PGA to the Damage Ratio of DR="A". ......................................... 70
Figure 30. Plot of Lognormal of PGA to the Damage Ratio of DR="As". ....................................... 71
Figure 31. Conventional fragility curves for the different damage ranks ........................................ 75
Figure 32. IUA (Lower Bound) fragility curves for different damage ranks .................................... 80
Figure 33. IUA (Upper Bound) fragility curves for different damage ranks. ................................... 84
Figure 34. Bounded fragility curve of DR="D" c.o.v.=5% in X-direction ......................................... 85
4

Figure 35. Bounded fragility curve of DR="C" c.o.v.=5% in X-direction ......................................... 86


Figure 36. Bounded fragility curve of DR="B" c.o.v.=5% in X-direction.......................................... 87
Figure 37. Bounded fragility curve of DR="A" c.o.v.=5% in X-direction.......................................... 88
Figure 38. Bounded fragility curve of DR="As" c.o.v.=5% in X-direction........................................ 89
Figure 39. Conventional and IUA (Mean) fragility curves of DR="D" c.o.v.=5% in X-direction ... 90
Figure 40. Conventional and IUA (Mean) fragility curves of DR="C" c.o.v.=5% in X-direction ... 91
Figure 41. Conventional and IUA (Mean) fragility curves of DR="B" c.o.v.=5% in X-direction ... 92
Figure 42. Conventional and IUA (Mean) fragility curves of DR="A" c.o.v.=5% in X-direction ... 93
Figure 43. Conventional and IUA (Mean) fragility curves of DR="As" c.o.v.=5% in X-direction . 94
Figure 44. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=5% . ................. 95
Figure 45. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=5% . ............ 96
Figure 46. Bounded fragility curves for damage rank of "Moderate Damage" of c.o.v.=5% . ..... 97
Figure 47. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=10% . .............. 98
Figure 48. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=10% ........... 99
Figure 49. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=20% . ............ 100
Figure 50. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=20%. ......... 101
Figure 51. Pushover curve with lower and upper bounds................................................................ 106
Figure 52. Pushover curve (inset) with lower and upper bounds. ................................................. 107
Figure 53. Interval Uncertainty Analysis seismic fragility curve of "No Damage" rank. ............. 110
Figure 54. Interval Uncertainty Analysis seismic fragility curve of "Slight Damage" rank. ......... 111
Figure 55. Interval Uncertainty Analysis seismic fragility curve of "Moderate Damage" rank. .. 112
Figure 56. Interval Uncertainty Analysis seismic fragility curve of "Extensive Damage" rank. . 113
Figure 57. Interval Uncertainty Analysis seismic fragility curve of "Complete Damage" rank. . 114

LIST OF TABLE
Table 1. Summary of Results of Ordinary Monte Carlo Simulation................................................. 20
Table 2. Relationship between the damage index and damage rank based from HAZUS (2013).
.................................................................................................................................................................... 36
Table 3. Pertinent ground motion (East-West) data of the Bohol, Philippines earthquake ......... 46
Table 4. Pertinent ground motion (Up and Down) data of the Bohol, Philippines earthquake.... 46
Table 5. Summary of calculations of ductility factors, damage indices, and damage ranks using
MS Excel Spreadsheet. For Tohoku-Kanto Eq. (Fukushima) .......................................................... 62
Table 6. Summary of Damage Ratio for the Conventional fragility curves .................................... 63
Table 7. Summary of Damage Ratio for the IUA fragility curves (Lower Bound) .......................... 63
Table 8. Summary of Damage Ratio for the IUA fragility curves (Upper Bound) .......................... 64
Table 9. Tabulation of ln(PGA) with number of occurrences per damage rank. ........................... 72
Table 10. Tabulation of the product of number of occurrences to ln(PGA) ................................... 72
Table 11. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value. .......................................................................................................................................................... 73
Table 12. Tabulation of the values for the (X - Mean ) / Standard Deviation................................. 74
Table 13. Tabulation of the values for the Probability of Exceedance per PGA/g values. .......... 74
Table 14. Tabulation of ln(PGA) with number of occurrences per damage rank. ......................... 76
Table 15. Tabulation of the product of number of occurrences to ln(PGA) ................................... 77
Table 16. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value. .......................................................................................................................................................... 78
Table 17. Tabulation of the values for the ( X - Mean ) / Standard Deviation .............................. 79
Table 18. Tabulation of the values for the Probability of Exceedance per PGA/g values. .......... 79
Table 19. Tabulation of ln(PGA) with number of occurrences per damage rank. ......................... 81
Table 20. Tabulation of the product of number of occurrences to ln(PGA) ................................... 81
Table 21. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value. .......................................................................................................................................................... 82
Table 22. Tabulation of the values for the ( X - Mean ) / Standard Deviation................................ 83
Table 23. Tabulation of the values for the Probability of Exceedance per PGA/g values. .......... 83
Table 24. Checklist of bounded fragility curves that subscribe to the "norm". ............................... 94
Table 25. Tabulation of bounded fragility curves for DR="D","C", "B" for c.o.v.=5%.................. 101
Table 26. Tabulation of bounded fragility curves for DR="D","C" for c.o.v.=10%. ...................... 102
Table 27. Tabulation of bounded fragility curves for DR="D","C" for c.o.v.=20%. ...................... 102
Table 28. Summary of calculated c.o.v. based from the bounded fragility curves which follow
the set norm. ........................................................................................................................................... 103
Table 29. Parameters from Nonlinear Static Analysis to be used in computing ductility factors
using Interval Analysis. .......................................................................................................................... 107
Table 30. Parameters from Nonlinear Dynamic Analysis to be used in computing ductility
factors using Interval Analysis. ............................................................................................................. 108
Table 31. Ductility factors computed from Octave script. ............................................................... 108
Table 32. Damage indices and damage rank based from HAZUS. .............................................. 109
6

CHAPTER 1

PROBLEM SETTING

1.1 Background of the Study

Nowadays, an increasing social awareness was brought by nations media


practitioners on vulnerability of structures in the occurrence of every major earthquake.
As a consequence of this increased awareness, research has been the forefront to ease
the quantification of the potential social and economic losses of communities across the
globe. Bridge fragility curves have grown from this surge in research as they are essential
component to the risk assessment methodology.

The risk assessment of lifeline presented in Figure 1 shows that one of the key
links in the assessment methodology is to estimate the damage to the lifeline
components. This is done by estimating the performance of the various highway bridges
in the network as a function of a ground motion intensity parameter.

This bridge

performance is commonly represented in either a damage probability matrix or a fragility


function (Nielson, 2005).

A fragility function is a conditional probability that gives the likelihood that a


structure will meet or exceed a specified level of damage for a given ground motion
intensity measure (Ramanathan, 2012).

Figure 1. Example of seismic fragility curves.


Source: (Nielson, 2005)

In dealing with real world problems, uncertainties are unavoidable.

As Civil

Engineers, it is important to recognize and quantify the presence of all major sources of
uncertainty in the analysis and design of structures. The sources of uncertainty may be
classified into two broad types: (1) aleatory those that are associated with natural
randomness; and (2) epistemic those that are associated with inaccuracies in our
prediction and estimation of reality (Ang & Tang, 2007). Uncertainties can arise from
(Modares, Taha, & Mohammadi, 2014):

1) modeling due to oversimplification of the system and the boundary


conditions,
2) manufacturing errors in structural member dimensions,
3) load variations compiled from historical records or through instrumentations in
the field or laboratory,
4) interaction of structure leads to complexities,
5) variations in material properties, and
6) lack of sufficient test data.

Different methods of structural reliability were already in use. From the most
sophisticated method of inclusion of general method called evidence-based-theory
reliability analysis (Jiang, Zhang, Han, & Liu, 2013), a proposed interval uncertain multiobjective optimization method for structures with uncertain-but-bounded parameters (Li,
Luo, Rong, & Zhang, 2013), interval elastoplastic analysis of structures

(Yang,

Tangaramvong, Gao, & Tin-Loi, 2015), application of interval-based optimization of


aircraft wings under landing loads (Majumder & Rao, 2008), and to the relatively simple
method using interval uncertainty analysis in the reliability assessment of structures,
specifically that of probability of failure of plane truss (Modares, Taha, & Mohammadi,
2014).

In applications, interval analysis provides rigorous enclosures of solutions to model


equations. In this way, one can at least surely know what a mathematical model tells him,
and, from that, one might determine whether it adequately represents reality. Without
rigorous bounds on computational errors, a comparison of numerical results with physical

measurements does not tell how realistic a mathematical model is

(Moore, Baker

Kearfott, & Cloud, 2009).

In his dissertation, Nielson (2005) tackled the uncertainty in the seismic demand
function which was placed on highway bridges. He added that it is not always easy to
separate out the different sources of uncertainty in a given problem.

Usually, the

randomness of seismic ground motion can be considered to be aleatoric in nature. But


when one deals with synthetic ground motions, the uncertainty innate into these ground
motion cannot be only be attributed to seismological mechanisms but also path and site
characteristics. More knowledge that pertains to attenuation relationships and effects of
soil can reduce some of this uncertainty, thereby alluding to its epistemic nature. He also
added that there is uncertainty associated with each median which was defined. This
uncertainty is given in the form of a lognormal standard deviation or dispersion. His study
claimed that when enough data is unavailable for the assessment of the dispersion for
each limit state, it is still beneficial to account for some degrees of uncertainties. The
assignment of this uncertainty can be assumed in a subjective manner, that is, by
estimating the coefficient of variation or c.o.v.

In his doctoral research study,

Ramanathan (2012) considered uncertainty by citing the works of Melchers (1999) and
Ellingwood and Wen (2005), which stated that treatment of uncertainty in seismic
reliability and performance assessment has been a subject of research for many years.
In his part, Ramanathan used normal distribution in his model of reinforced concrete
materials, with the median and coefficient of variation used statistical parameters.

10

The use of Latin Hypercube sampling (LHS) is a stratified sampling procedure that
provides an efficient way of sampling variables from their distribution while limiting the
required sample size. This procedure was considered to different bridge realizations
statistically different and yet nominally identicalfor the probabilistic treatment of fragility
curves.

LHS was implemented to reduce the computational effort.

An algorithm

suggested by Iman and Conover (1982) was implemented in the frames of the proposed
methodology for the treatment of uncertainty in capacity and demand (Stefanidou &
Kappos, 2013).

Traditional reliability analysis requires probability distributions of all the uncertain


parameters. These include the computations of reliability index using First-Order SecondMoment (FOSM), First-Order Reliability Method (FORM), Second-Order Reliability
Method (SORM), Hasofer-Lind, Rackwitz-Fiessler Procedure, Monte Carlo Simulation
(MCS). All except except MCS has closed form solution; whilst MCS, gives sub-optimal
solution, depending on the observance of uncertainty in the estimate of the probability as
it decreases as the total number of simulations, N, increases (Nowak & Collins, 2013).
However, in many practical applications, the variation bounds can be only determined for
the parameters with limited information (Han, Jiang, Liu, Liu, & Long, 2014).

From the above mentioned methods of reliability analysis, there is a research gap
in bounded fragility curves. In the light of introducing a method to develop these curves
for the unknown-but-bounded uncertainty, seismic fragility curve will now have
probability of exceedance values with an overlay of quantifying how low and how high a
11

decision-maker will adopt in his design, re-design, or retrofit of built structures. The
researcher adopts interval analysis to produce this interval. Thus, an interval uncertainty
analysis form of seismic fragility curves is being developed in this present study.

1.2 Statement of the Problem

There is a need to study the application of a technique that can respond to the
uncertainties during the evaluation of damage index formula used (Park & Ang, 1985).
Interval uncertainty analysis is chosen to embed in this reliability analysis, since the
outcome of this proposed method is an interval value of probability of exceedance for a
given intensity measurement, i.e., peak ground acceleration (PGA); thus, seismic fragility
curves by interval uncertainty analysis (IUA).

1.3 Significance of the Study

This study is significant due to the following reasons:


A. Epistemic uncertainty can be quantified through interval analysis; thus, its
embedment to the probability of exceedance.
B. Re-assessing built structures using the proposed method of reliability analysis
particularly that of seismic fragility curves.
C. An alternative reliability analysis aside from the existing ones, where computer
simulation is possible through fast computing requirements.

12

1.4 Objective of the Study

This research aims to study the reliability of piers of a reinforced concrete deck
girder bridges using interval uncertainty analysis applied to fragility curves.
Specifically, this research aims:
A. To quantify the uncertainty in the construction of seismic fragility curves by
using interval analysis.
B. To develop a methodology in assessing the performance of bridge piers using
interval uncertainty analysis under shear mode of failure
C. To compare the difference between a conventional fragility curve analysis and
an interval uncertainty analysis when applied to a reinforced concrete deck
girder

1.5 Scope and Limitations

The study is focused only in the reliability analysis using the shear mode of failure
of bridge pier.
Moreover, the study limits to the fragility analysis and the non-linear static
(Pushover Analysis) and non-linear dynamic (Time History Analysis) analyses in
constructing fragility curves by interval uncertainty analysis.

13

A number of strong significant ground motion data is limited to four (4) sets which
will be obtained from PHIVOLCS, PEER, and K-net.com, namely, Mindoro December
1999 earthquake, Bohol October 2013 earthquake, The Great Hanshin Kobe earthquake
in 1995, and the Tohoku-Kanto March 2011 earthquake. Peak ground accelerations
(PGAs) are divided into ten discrete values, from 0.2g to 2.0g.

1.6 Assumptions in the Study


This study assumes that the nonlinear static analysis (pushover analysis) and
nonlinear dynamic analysis (time history analysis) are the suitable model for the shear
mode of failure.

Coefficient of variation used are 1%, 5%, 10%, and 20% for the

computation of damage indices in the interval uncertainty analysis.

14

CHAPTER 2
REVIEW OF RELATED LITERATURE
2.1 Interval Analysis
A novel evidence-theory-based reliability analysis method for structures with
epistemic uncertainty is one those methods that deals with a specific theory based on a
more general than other uncertainty modeling techniques. This was equivalent to the
classical probability theory, possibility theory, p-box approach, fuzzy sets and convex
models. In a further perspective, these basic axioms of evidence theory allowed one to
combine alleatory and epistemic uncertainty in a straightforward manner without any prior
knowledge (Jiang, Zhang, Han, & Liu, 2013). Fig. 2 is one example of the outputs of the
reliability analysis under the case of a 4-subinterval-BPA-structure, as used in the
applications of their developed general method using evidence-theory-based.

Figure 2. Reliability analysis results under a 4-subinterval-BPA-structure case.


Source: (Jiang, Zhang, Han, & Liu, 2013)

15

A proposed systematic design optimization method for structures with uncertainbut-bounded parameters is another method. Using the interval method, Li et. al used to
describe the uncertainty and the Kriging model was applied in order to generate for the
approximation model.

The interval number programming method was utilized to

transform each uncertain optimization problem with a single objective function into a
deterministic multi-objective optimization problem. Typical numerical examples applied
to engineering demonstrated that the proposed method can effectively search the Pareto
frontier using estimated approximation models. Moreover, this method has the ability to
retain an unchanged approximation space. Refer to Fig. 3, the crashworthiness vehicle
design was used as the engineering application problem, using a closed-hat beam with
interval uncertain parameters (Li, Luo, Rong, & Zhang, 2013).

Figure 3. Typical deformation behavior as seen in the finite element model.


Source: (Li, Luo, Rong, & Zhang, 2013)

16

As of this writing, a work that considers the effects of uncertainties in a classical


elastoplastic analysis was started. The uncertainties were related to the applied forces
and the plastic material capacities, both of which were taken to lie within deterministic but
bounded intervals. This proposed novel interval approach obtained full spectra of the
extreme responses under interval force and yield capacity data. The efficiency and
robustness of this method were illustrated through a number of practically-motivated
engineering structure examples. The results have been partially validated by Monte Carlo
simulations and interval limit analyses.

With these examples (see Fig. 4), it was

highlighted by this study the importance of assessing the influence of uncertain applied
forces and yield limits for practical application, with special attention to higher load levels,
that is, a sufficient number of plastic hinges had been developed (Yang, Tangaramvong,
Gao, & Tin-Loi, 2015).

Figure 4. A graph of displacement at the plastic hinge with interval holonomic


elastoplastic alpha-umax and alpha-umin responses
Source: (Yang, Tangaramvong, Gao, & Tin-Loi, 2015)

17

Reliability of structures is not only applied to vertical and horizontal structures but
also to aircraft structural components such as wings under landing loads. In order to
optimize the aircraft wing structures subjected to landing loads, an interval-based
automated optimization method was used. The interaction between landing gear and
flexible airplane structure was considered as a coupled system. Here, uncertainties were
considered in terms of the system parameters and described as interval numbers. In the
aspects of computing, the optimization procedure was illustrated using two distinct
applications, i.e., symmetric double-wedge airfoil, and supersonic airplane wing. This
interval analysis-based multicriteria optimum design of airplane wing structures under
landing loads was demonstrated.

The comparison of results indicated that, for

comparable data and for the same value of the permissible landing stress, the minimum
value of any specific objective function obtained by interval analysis was in good
agreement with the ones obtained by deterministic and probabilistic analyses (Majumder
& Rao, 2008). Fig. 5 shows the progress of individual objective functions with the number
of iterations for all the three types of analyses.

Figure 5. Single objective optimization of example wing 2 under landing loads in


progress, deterministic, probabilistic, interval analyses.
Source: (Majumder & Rao, 2008)
18

Regarded as significant in the literature of structural reliability, the Taylor-series


inspired technique has been used frequently. One of which is the First Order Reliability
Method (FORM). But with the harmonious mix with the interval analysis method, a newlyformed Interval First Order Reliability Method (IFORM) was introduced by Modares et. al.
This method was an enhanced FORM after considering the uncertainties using interval
analysis. In this method, the intervals selected for using the method can be based on a
pre-determined multiple of standard deviation of the performance function. Again, this
developed method was compared to the Monte Carlo Simulations. The results from the
traditional first-order analysis were within the lower and upper bound results from the
interval method, as seen in Fig. 6. It was found from this method that for a small amount
of uncertainty in the resistance and applied load, the traditional first-order approximation
method results in probability values that correspond to bounds equal to once standard
deviation of performance function (Modares, Taha, & Mohammadi, 2014).

Figure 6. Probability of Failure for the Performance Function Z=R-S.


Source: (Modares, Taha, & Mohammadi, 2014)

19

The author attempted to come up with an assessment of a transportation lifeline


pier using ordinary Monte Carlo Simulation under a large magnitude earthquake.
According to their paper, the Philippines capital has its mass transit, the Light Rail Transit
System (LRT), constructed in the 1980s as part of the governments modernization efforts
in the field of transportation. Over the past thirty years, the LRT has withstood a number
of natural hazards including a strong earthquake in July of 1990. Due to this event, the
Philippine government initiated the earthquake reconstruction project and made
recommendations to retrofit important bridges. The paper investigated the reliability index
of the columns of the LRT under a Level 1 (El Centro) earthquake and Level 2 (TohokuKanto) earthquake using ordinary Monte Carlo Simulation. Based from the maiden
structural plans of LRT, the slenderness ratio of columns based from the ACI 318 was
observed and checked for buckling failure. Referring to Table 1, a significant +12%
difference between the reliability indices of unconfined (3.47) and confined (3.89)
reinforced concrete column was computed in the said simulation (Baylon, Garciano, &
Koike, 2012).
Table 1. Summary of Results of Ordinary Monte Carlo Simulation
Source: (Baylon, Garciano, & Koike, 2012)

Pf

Pf

Based from NSCP 2001


Tohoku-Kanto Earthquake ( M9.0)
Unconfined
Confined
% diff
0.00146
0.00012
92%
2.976037
3.672701
23%
Based from NSCP 2010

El Centro Earthquake of 1940


Unconfined Confined
% diff
0.00029
0.00013
55%
3.440799 3.652203
6%

Tohoku-Kanto Earthquake ( M9.0)


Unconfined
Confined
% diff
0.00109
0.00039
64%
3.064547
3.359796
10%

El Centro Earthquake of 1940


Unconfined Confined
% diff
0.00023
0.00002
91%
3.503029
4.10748
17%

% diff Pf
% diff

28%
13%

20

36%
11%

2.2 Fragility analysis

Civil engineering facilities such as bridges, buildings, power plants, dams


and offshore platforms are all intended to contribute to the benefit and quality of life. It is
important that the benefit of the facility can be identified considering all phases of the life
of the facility. These facilities are established, operated, maintained and decommissioned
in such a way that was optimized or enhance the possible benefits to society and
individuals of the society (Sorensen, 2004). It is advisable to have clear expectations
about those portions of the structure that are expected to undergo inelastic deformations
and to use the analyses to (1) confirm the locations of inelastic deformations and (2)
characterize the deformation demands of yielding elements and force demands in nonyielding elements.

In this regard, capacity designs concepts are encouraged to help

ensure reliable performance (Deierlein, Reinhorn, & Willford, 2010).

Bridges are potentially one of the most seismically vulnerable structures in


the highway system. While performing a risk analysis of a highway system, it is imperative
to identify seismic vulnerability of bridges associated various states of damage
(Shinozuka, Feng, Kim, Uzawa, & Ueda, 2003).

In the probabilistic seismic risk

assessment of highway transportation networks, fragility curves are used to represent the
vulnerability of a bridge.

Because these networks have hundreds or thousands of

bridges, it is not possible to study each bridge individually. Instead, bridges with similar
properties are grouped together and represented by the same fragility curve (Gomez,
Torbol, & Feng, 2013).
21

Seismic fragility is the probability that a geotechnical, structural, and/or nonstructural system violates at least a limit state when subjected to a seismic event
of specified intensity. Current methods for fragility analysis use peak ground acceleration
(PGA), pseudo spectral acceleration (PSa), velocity (PSv), or spectral displacement (Sd)
to characterize seismic intensity (Kafali & Grigoriu, 2004). These fragility curves indicate
the evolving potential for component and system damage under seismic loading
considering time-dependent corrosion-induced deterioration. The results indicate that
while corrosion may actually decrease the seismic vulnerability of some components,
most critical components suffer an increase in vulnerability (Ghosh & Padgett, 2010).

A method was used to track changes of the structural parameters of a bridge


throughout its service life. Based on vibration data the fragility curves are updated
reflecting a change in structural parameters. Fragility curves based on vibration data,
whenever these are available, represent the vulnerability of a bridge with greater accuracy
than fragility curves based only on the geometry and material properties (Gomez, Torbol,
& Feng, 2013). Most of the studies states that fragility curves can be empirically and
analytically generated.

The empirical fragility curves are usually developed based on the damage reports
from past earthquakes (Jernigan & Hwang, 2002). While the analytical fragility curves
are developed through seismic response data from the analysis of bridges.

22

The fragility

analysis generally includes three major parts: (a) the simulation of ground motions,
(b) the simulation of bridges to account for uncertainty in bridge properties, and (c) the
generation of fragility curves from the seismic response data of the bridges.
Nonlinear analysis requires thinking about inelastic behavior and limit states that depend
on deformations as well as forces. They also require definition of component models that
capture the force-deformation response of components and systems based on expected
strength and stiffness properties and large deformations (Deierlein, Reinhorn, & Willford,
2010). The seismic response data can be obtained from nonlinear time history
analysis, elastic spectral analysis, or nonlinear static analysis (Choi, DesRoches, &
Nielson, 2004).

The nonlinear static analysis is normally used for determining the capacities
beyond the elastic limit. One type of nonlinear static analysis is the pushover analysis
wherein which it is mainly used for estimating the strength and drift capacity of a structure
when subjected to selected earthquake. It incorporates the nonlinear-deformation
characteristics of individual components and subjects the elements of a structure
to monotonically increasing lateral load within a height-wise distribution until a
predetermined displacement is attained (Requiso, Balili, & Garciano, 2013). The static
pushover analysis has no rigorous theoretical foundation. It is based on the assumption
that the response of the structure can be related to the response of an equivalent single
degree-of-freedom (SDOF) system. This implies that the response is controlled by a
single mode, and that the shape of this mode remains constants throughout the time
history response. Clearly, both assumptions are incorrect, but pilot studies carried out by
23

several investigators have indicated that these assumptions lead to rather good
predictions of the maximum seismic response of multi degree-of-freedom (MDOF)
structures, provided their response is dominated by a single mode (Krawinkler &
Seneviratna, 1998). In the pushover analysis, it is assumed that the target displacement
for the MDOF structure can be estimated as the displacement demand for the
corresponding equivalent SDOF system (Krawinkler & Seneviratna, 1998).

The pushover is expected to provide information on many response characteristics


that cannot be obtained from an elastic static and dynamic analysis. The following are the
examples of such response characteristics:
a. The realistic force demands on potentially brittle elements (axial, shear,
moment).
b. Estimates of the deformation demands for elements that have to deform
inelastically in order to dissipate the energy imparted to the structure by ground
motions.
c. Consequences of the strength deterioration of individual elements on the
behavior of the structural system.
d. Identification of the critical regions in which the deformation demands are
expected to be high and that to become the focus of thorough detailing,
e. Identification of the strength discontinuities in plan or elevation that will lead to
changes in the dynamic characteristics in the elastic range.
f. Verification of the completeness and adequacy of load path, considering all the
elements of the structural system, all the connections, the stiff nonstructural elements of significant strength, and the foundation system.

24

In contrast to the nonlinear static procedure, the nonlinear dynamic procedure,


when properly implemented, provides a more accurate calculation of the structural
response to strong ground shaking. Since the nonlinear dynamic analysis model
incorporates inelastic member behavior under cyclic earthquake ground motions, the
nonlinear dynamic procedure explicitly simulates hysteretic energy dissipation in the
nonlinear range (Nielson, 2005). In nonlinear dynamic analysis, the detailed structural
model subjected to a ground-motion record produces estimates of component
deformations for each degree of freedom in the model and the modal responses
are combined using schemes such as the square-root-sum-of-squares (Requiso D.
A., 2013).

Time-history analysis is a step-by-step integration analysis performed in the time


domain. It is the most rational method for earthquake response analysis because it can
account for all sources of nonlinearity. Although this method is computationally intensive,
it is the preferred method for investigating the response of bridges subjected to seismic
excitation. Similarly to the pushover analysis, SAP2000 can also directly perform the time
history analysis to obtain the nonlinear behavior, which includes generation of the
hysteresis models (Requiso D. A., 2013).

In establishing the seismic fragility curves, there is no universally applicable best


method for calculating fragility curves. Different methods may be preferred depending on
the circumstances (Requiso, Balili, & Garciano, 2013). The information that would be
derived from the fragility curve can be used by design engineers, researchers, reliability
25

experts, insurance experts and administrators of critical systems to analyze evaluate


and improve the seismic performance of both structural and non-structural systems
(Requiso D. A., 2013). In principle, the development of bridge fragility curves will
require synergistic use of the following methods: (1) professional judgment, (2) quasistatic and design code consistent analysis, (3) utilization of damage data associated
with past earthquakes, and (4) numerical simulation of bridge seismic response
based on structural dynamics (Shinozuka, Feng, Kim, Uzawa, & Ueda, 2003).

2.3 Seismic Fragility Analysis in the Philippines

In the local setting, seismic assessment of bridge piers and a fish port was recently
implemented by students in one of the universities in the countrys capital. Their study
was based from works of Karim-Yamazaki, Shinozuka et. al, and Ang-Park type of fragility
curves with emphasis on the nonlinear static and nonlinear dynamic analyses. Figures 7
and 8 summarize this seismic assessment depending in the damage rank classification
based from HAZUS (Baylon M. B., 2015).

26

FRAGILITY CURVES
DAMAGE RANK ="As"

1.2

PROBABILITY OF EXCEEDANCE (SHEAR)

Navotas Fishport

0.8

Lambingan
Tullahan(Mal-Val)

Bangkulasi

0.6

Tullahan (Ugong)
LRT1 South

0.4

0.2

0
0

0.5

1.5

2.5

PEAK GROUND ACCELERATION (in g)

Figure 7. Summary of Seismic Fragility Curves of some lifelines in CAMANAVA area of


Damage Rank As
Source: (Baylon M. B., 2015)

27

FRAGILITY CURVES
DAMAGE RANK = "A"

1.2

PROBABILITY OF EXCEEDANCE (SHEAR)

Navotas Fishport

0.8

Lambingan
Tullahan(Mal-Val)
Bangkulasi

0.6

Tullahan (Ugong)
LRT1 South

0.4

0.2

0
0

0.5

1.5

2.5

PEAK GROUND ACCELERATION (in g)

Figure 8. Summary of Seismic Fragility Curves of some lifelines in CAMANAVA area of


Damage Rank A
Source: (Baylon M. B., 2015)
2.4 Synthesis

Based from the previous studies reviewed, theres no definite approach of interval
uncertainty analysis to the construction of seismic fragility curves. Albeit, the interval
analysis has been in the research arena for quite a long time, reliability analysis of
28

structures using this method was already applied to sophisticated analysis, such as finite
element approach, but not yet with the fragility analysis. Some of the reliability analyses
presented in this chapter has been reviewed and motivated the present study to pursue
with one thing in mind: assessing the performance of bridge pier using fragility analysis.
The present study attempts to apply the interval analysis with the incorporation of
uncertainty in the results of nonlinear structural analyses, namely, the pushover curve
and hysteresis area; thus, an interval uncertainty analysis fragility curve or IUA-FC.

29

CHAPTER 3
CONCEPTUAL AND THEORETICAL FRAMEWORK

3.1 Conceptual Framework

Figure 9 illustrates the conceptual framework of the present study. Using the
structural model and normalized ground motion data, nonlinear static and dynamic
analyses were used in the preliminary steps for the conventional seismic fragility
development. The current study used shear as the mode of failure. The output of these
two nonlinear analyses were the parameters as inputs to the damage index formula.
Using interval uncertainty analysis, these parameters were calculated to obtain the lower
bound, upper bound, and mean value of the damage indices for every ground motion
datas peak ground accelerations (PGA) from 0.2g to 2.0g. These damage ranks were
counted as frequencies to compute the probabilities of occurrence for various peak
ground acceleration (PGA) values. For every damage level, probabilities of occurrence
were used to compute the mean and standard deviations to be used in the lognormal
equation for the fragility analysis. Plotting the cumulative lognormal probability versus the
peak ground acceleration for every damage levels creates the seismic fragility curves.
Comparing these fragility curves to the conventional seismic fragility curves creates
another plot of the difference of IUAs mean probabilities to that of conventional
probabilities versus PGA.

30

Structural Model
Ground motion data

Nonlinear Static
Analysis
(Pushover
Analysis)

Mode of Failure
SHEAR

Nonlinear
Dynamic Analysis
(Time History
Analysis)

Interval
Uncertainty
Analysis (IUA)

Parameters for
Damage Index

DAMAGE
INDICES

IUA Seismic
Fragility Curves

Checking for Optimum Coefficient of Variation to be used for


the 2nd Pass of IUA

Figure 9. Conceptual Framework of the Study


31

Structural Model
Ground motion data

Mode of Failure
SHEAR

Nonlinear Static
Analysis
(Pushover
Analysis)

Nonlinear
Dynamic
Analysis (Time
History Analysis)
Interval Uncertainty
Analysis (IUA)

Parameters for
Damage Index

DAMAGE
INDICES

IUA Seismic
Fragility Curves
Figure 10. Interval Uncertainty Analysis after the First Pass.

32

3.2 Theoretical Framework

3.2.1 PGA Normalization


Using structural model of the building and normalized ground motion data as an
input subjected to two nonlinear methods namely nonlinear static analysis and nonlinear
dynamic analysis to develop seismic fragility curves. Peak Ground Acceleration (PGA)
normalization is done by generating the original data to create another data for different
PGAs. This is basically the same graph but different extent depending on the PGA. In the
present study, the PGA normalization ranges from 0.2 g to 2.0 g.

3.2.2 Pushover Analysis (Nonlinear Static Analysis)

Nonlinear static analysis, also called as pushover analysis, is used to investigate


the force-deformation behavior of a structure for a specified distribution of forces, typically
lateral forces (Chopra, 2012). In this study, the pushover analysis is applied to produce
pushover curve showing the relationship between the force and the displacement that
would be used for further analysis.

33

3.2.3 Time History Analysis (Nonlinear Dynamic Analysis)

In the present study, the researcher is adopting the concept of time history analysis
considering the Bangkulasi Bridge as a single-degree-of-freedom (SDOF) system which
is subjected to normalized ground motion having different excitations.
The formula adopted from Karim and Yamazaki (2001) shown in equation 1 where
the ground motion data is multiplied by the ratio of the normalized and original peak
ground acceleration defines the relationship of various earthquakes while maintaining its
time history pattern.
uNEW A0uSOURCE

(1)

Where:
uNEW A0uSOURCE = the normalized ground motion data.
uSOURCE = the source ground motion data.

A0 =a coefficient factor to normalize the source of ground motion =

PGAnormalized
PGAsource

Using software, the acceleration time histories obtained was used as an input
producing another relationship between the force and displacement called the hysteresis
model (bilinear model).

34

3.2.4 Ductility Factors

From the output of the nonlinear static and nonlinear dynamic analyses which are
the push-over curve and hysteresis model respectively, the ductility factors were obtained
using the following equations, also adopted from Karim and Yamazaki (2001).

max dynamic
d
y

(2)

max static
y

(3)

Eh
Ee

(4)

Where:
= displacement ductility
= ultimate ductility
= hysteretic energy ductility
()= displacement at maximum reaction at the push over curve (static)
()= maximum displacement at the hysteresis model (dynamic)
= yield displacement from the push-over curve (static)
= hysteretic energy, i.e., area under the hysteresis model

35

= yield energy, i.e., area under the push-over curve (static) but until yield point
only
3.2.5 Damage Index and Damage Rank
Once ductility factors are obtained, damage indices for the conventional seismic
fragility curves can be determined using equation 5, taking which is the cyclic loading
factor as 0.15 according to Jiang, et. al (2012), for bridges.

ID

d h
u

(5)

After computing the damage indices, damage rank for each damage index
was determined using Table 2.
Table 2. Relationship between the damage index and damage rank based from HAZUS (2013).

Damage Index ( )

Damage Rank (DR)

Definition

0.00 < DI 0.14

No damage

0.14< DI 0.40

Slight damage

0.40 < DI 0.60

Moderate damage

0.60 < DI 1.00

Extensive damage

1.00 DI

As

Complete damage

3.2.6 Interval Arithmetic Operations

Using Interval Arithmetic Operation, defined by Moore et. al. (2009) as computing
with intervals is computing with sets, the basic arithmetic operations presented are given

36

~
by upper and lower bound of an interval in both axis. Given two interval numbers X and

~
Y whose lower and upper bound has this symbol:

~
X X

~
X ;Y Y Y

(6)

With this definition, the following interval arithmetic operations equations adopted
from Moore et al. (2009) are computed with sets;
Addition
+ = [ +

+ ]

(7)

= [

(8)

= [min

max ]

(9)

(10)

Subtraction

Multiplication

Where:
= {

Division
It can be accomplished via multiplication by the reciprocal of the second operand.
= (1)

37

(11)

Where:
1 = [1

1]

(12)

Assume that 0

3.2.7 Interval Uncertainty Analysis (IUA)

Applying the interval uncertainty analysis to the preceding topics discussed


generates an interval of damage indices. According to Modares et al. (2014) interval
analysis is widely applied in engineering especially for civil engineering. In designing,
some values are assumed knowing that it is unknown but bounded. The concept of the
interval uncertainty analysis is to obtain a range of values incorporating the uncertainties.
Using interval analysis, the ductility factors and the damage indices equations are
now bounded as shown in the following equations:

~
max dynamic
~
d
~

(13)

~
max static
~
u ~

(14)

~
E
~h ~h
Ee

38

(15)

Where:
= interval of displacement ductility
= interval of ultimate ductility
= interval of hysteretic energy ductility
()= interval of displacement at maximum reaction at the push over curve
(static)
()= interval of maximum displacement at the hysteresis model
(dynamic)
= interval of yield displacement from the push-over curve (static)
=interval of hysteretic energy, i.e., area under the hysteresis model
= interval of yield energy, i.e., area under the push-over curve (static) but until
yield point only
After the interval of ductility factors are obtained, damage indices for the IUA
seismic fragility curves can be determined using equation 5 and taking as 0.10 for
vertical structures and 0.15 for bridges.
+

Where:
;

39

(16)

The output of these equations can now be used to determine the range of damage
indices for every ground motion data from 0.2g to 2.0g peak ground acceleration (PGA).
Having the damage indices interval obtained, damage rank per PGA for upper and
lower bound can now be obtained using the Table 2.

3.2.8 Probability of Exceedance

Once the parameters have been obtained, the cumulative probability of occurrence
( ) of the damage, equal or higher than the damage rank, is computed using equation
15.

ln X
Pr

Where:
Pr =Cumulative Probability of Exceedance

= Cumulative Normal Distribution Function


X = Peak Ground Acceleration
= Mean
Standard Deviation

40

(17)

Plotting the values of cumulative lognormal probability against the peak ground
acceleration creates seismic fragility curve. After the conventional and IUA seismic
fragility curves are developed, evaluation of probability of exceedance difference would
take place.
The statistical formulas used in deriving the mean and standard deviation were
based from an ungrouped data premise.
N

f
i 1

lnxi

f
i 1

(18)

lnx

i 1

N 1

Where:

f = frequency of damage rank per PGA.


x =the PGA in cm/s2.

=the mean of the natural logarithm of PGA, in cm/s2.


=the standard deviation of the PGA, in cm/s2.

41

(19)

CHAPTER 4
RESEARCH METHODOLOGY

4.1 Input data


Structural Plans
The structural plans of Bangkulasi Bridge were obtained from Bureau of Designs
in Department of Public Works and Highways-National Capital Region.

Pier 2 of the Bangkulasi Bridge as it can be seen in its general elevation plan of
Figure 11 was assumed as the more probable to fail when subjected to earthquake
load. The elevation plan of the pier is shown in Figure 12. Referring to Figures 13 to 15,
Pier 2 is composed of 4-1200mm piles as seen in its detailed bored pile elevation, the
section of a typical pier, and that of the section of coping.

Based from these sectional properties of the structural plans, dimensions were
incorporated in both the resistance and load effect models.

42

Figure 11. General Elevation Plan of the Bridge Showing Pier 2

Figure 12. Elevation Plan of the Pier

43

Figure 13. Detail of Bored Pile

Figure 14. Section A of the Bored Pile

44

Figure 15. Section C (Coping)


Ground Motion Data
The secured ground motion data of significant earthquakes from Philippine
Institute of Volcanology and Seismology (PHIVOLCS), Pacific Earthquake Engineering
Research (PEE), and Kik-Net were used as loads for the structure. Tables 2 and 3 are
pertinent data of the ground motion used for the simulation. These ground motion data
were normalized in such a way that a total of ten (10) were created from the base data.
This is based from the previous study used in developing fragility curves by Requiso et.
al.
The ground motion data that was used in this study are summarized as follows:
1. Tohoku-Kanto-FKS March 11, 2011 Magnitude 9.0
2. Tohoku-Kanto-AIC March 11, 2011 Magnitude 9.0
3. Tohoku-Kanto-HYG March 11, 2011 Magnitude 9.0
4. Tohoku-SIT March 11, 2011 Magnitude 9.0
5. Bohol October 15, 2013 Magnitude 7.2
6. Mindoro (Cainta, Rizal station) November 15, 1994 Magnitude7.1
45

7. Mindoro (Station Quezon City) November 15, 1994 Magnitude 7.1


8. Mindoro (Station Marikina City) November 15, 1994 Magnitude 7.1
9. Kobe Shin-Osaka January 16, 1995 Magnitude 6.9
10. Kobe Takarazuka January 16, 1995 Magnitude 6.9
11. Kobe Takatori January 16, 1995 Magnitude 6.9
12. Kobe Nishi-Akashi January 16, 1995 Magnitude 6.9
13. Kobe Kakogawa January 16, 1995 Magnitude 6.9
14. Kobe KJM January 16, 1995 Magnitude 6.9
15. Kobe HIK January 16, 1995 Magnitude 6.9

Table 3. Pertinent ground motion (East-West) data of the Bohol, Philippines earthquake
FILE:
201310150012.TBPS.HNE.cor
DATE:
10/15/2013 0:12
STATION:
TBPS
Station LATITUDE:
9.691
Station LONGITUDE:
123.862
SR:
100 Hz
Earthquake Latitude:
9.8
Earthquake Longitude:
123.8
Earthquake Depth:
10.0 km
Earthquake Magnitude:
7.2
Earthquake MECH:
1 [0 = null,1 = reverse,2 = strike-slip, 3 = normal ]
RHYP:
17.1 km
PGA:
2.137010 m/s/s (0.217907 g)
PGV:
0.668706 m/s
Instrument corrected time histories filtered using a 4th order
Butterworth bandpass between 0.4-50.0 Hz
Record Processed: 2014-04-03 13:50

Table 4. Pertinent ground motion (Up and Down) data of the Bohol, Philippines
earthquake
FILE:
DATE:
STATION:
Station LATITUDE:
Station LONGITUDE:
SR:
Earthquake Latitude:
Earthquake Longitude:
Earthquake Depth:
Earthquake Magnitude:
Earthquake MECH:
RHYP:
PGA:
PGV:

201310150012.TBPS.HNZ.cor
10/15/2013 0:12
TBPS
9.691
123.862
100 Hz
9.8
123.8
10.0 km
7.2
1 [0 = null,1 = reverse,2 = strike-slip, 3 = normal ]
17.1 km
1.364399 m/s/s (0.139125 g)
0.370004 m/s

46

Instrument corrected time histories filtered using a 4th order


Butterworth bandpass between 0.4-50.0 Hz
Record Processed: 2014-04-03 13:50

This ground motion data can be plotted in MS Excel as shown in Figure 16.

Figure 16. Ground motion data of Bohol October 2013 earthquake


normalized in 0.2g.
4.4 Interval Uncertainty Analysis method of reliability analysis

The MatLab script named IUA_DI.m (Algorithm 1 in the Appendices) was then
developed for computing the damage indices using Interval Uncertainty Analysis (IUA).
Equations to calculate in the section of chapter 3 were used and implemented.
To calculate an interval, i.e., lower bound and upper values, one has to consider
the function:

47

~x [ x x] [ (1 cov) (1 cov)]
x
x

(20)

where:

~
x = a 2x1 vector, known as an interval value.

x, x = the lower bound value and the upper bound value, respectively.

= the mean value

cov

= coefficient of variation value

x x =the formula for the lower bound

x x = the formula for the upper bound

It is assumed that the probability distribution for each interval to be normal


distribution. According to Modares, Taha, & Mohammadi (2014), a unity multiplier for the
standard deviation is recommended for both the lower and upper bound of the interval.
Figure 17 shows the illustration of this assumption.
Where X is the interval, is the lower bound, and is the upper bound
To get the lower and upper bound, a coefficient of variation (c.o.v.) is assumed to
1%, 5%, 10%, and 20%, to set the value of standard deviation, , given the mean value,
.

. . . =

(21)

= (. . . )

(22)

48

= [(1 . . . )

(1 + . . . )]

(23)

Input intervals in Equations 13 to 15 were used to implement Interval Uncertainty


Analysis (IUA). To come up with the values of the damage index intervals, the parameters
which were obtained from the nonlinear static and dynamic analyses maximum and
yield displacements, hysteretic energy, and elastic energy were used to apply
applicable interval arithmetic operations.

Figure 17. Normal probability function assumption of an interval value.

4.5 A library of MatLab functions implementing interval arithmetic operations

In the implementation of interval uncertainty analysis, it was convenient to develop


separate customized functions that implements interval arithmetic operations. Algorithm
49

2 is a collection of MatLab functions created for this purpose. These functions are based
from the equations in section 3.2.2, that is, basic discussion of interval arithmetic
operations.
The following steps were implemented to Algorithm 2.
Step 1: Impose a value for the coefficient of variation (COV). In this study a value of 5%
was used.
Step 2: The COV value will be used in computing the standard deviation by multiplying
the mean value by the COV.
Step 3: The mean value and standard deviation value will be used in calculating the lower
and upper bounds of any interval value.
Step 4: Using the different fundamental arithmetic operations of intervals, one can
compute for the sum, difference, product, and quotient of an interval arithmetic operation.

50

4.6 Research paradigm

The research paradigm is shown in Figure 18.

Figure 18. Research paradigm

Once all necessary section properties have been defined the actual simulation can
now be performed. To do so, the lateral ground motion of Types 1 & 2 earthquake would

51

be considered and a nonlinear static procedure would be used to account for the shear
failure namely the pushover analysis. Using the software (SAP 2000), the implementation
of the pushover procedures as prescribed in ATC-40 and FEMA-273 and the time history
analysis would be faster, reliable and easier since it was already integrated into the
software. All the results then are obtained and be ready for the seismic fragility curve
development. Using MatLab and MS Excel Spreadsheet, damage indices were
determined using interval arithmetic operations. These damage indices in interval form
were separated into lower bound, upper bound, and mean for the processing of the
corresponding fragility curve and combination thereof. From these processed values,
probability of occurrence values were calculated thru spreadsheet solution. The statistical
parameters to obtain the probability of exceedance values, i.e., mean and standard
deviation, were computed using conventional statistical formulas used for ungrouped
data.

For given PGA, probability of exceedance values were calculated using the

lognormal transformation as proposed by Shinozuka (2000).

4.6.1 SAP2000
The following procedures, a step-by-step in conducting the push-over analysis and
time history analysis in SAP 2000. (Requiso, 2013)
1.

Create a model without the pushover analysis

2.

Define the properties and acceptance criteria for the pushover hinges.

3.

Establish the pushover hinges on the model by selecting one or more frame
member/s and assign its hinge properties and location.

52

4.

Define the load cases of the pushover analysis. These load cases are
combination of dead load and the pushover load case itself.

5.

Run the basic static analysis, if desired, dynamic analysis. Then run the static
nonlinear pushover analysis

6.

SAP2000 can demonstrate the pushover curve by simply clicking the display
menu and selecting the show pushover curve function at the topmost toolbar of
the program. This would show you a table which gives the coordinates of each
step of the curve. It also allows the user to print the pushover curve or convert it
to an excel file to analyze the results

Once done with the steps pushover analysis will now be obtained, the results
would be used in the next procedure which is the nonlinear dynamic analysis (Time history
analysis). The following step-by-step procedure of Karim and Yamazaki (2001) in
performing the nonlinear dynamic analysis (Time History Analysis) is used.

1.

Select the ground motion data records.

2.

Normalize the PGA of the strong motion data records to be used. For this step,
you can normalize the peak ground acceleration by depending on the highest
value among the records. After picking the highest value in the records, you can
now use equation 1 which is mentioned in chapter 3 of the study. The result
would now be considered as normalized record to be used.

53

3.

Create a computer model in SAP 2000, you can use the exact computer model
from the pushover model but make sure to remove all the pushover data and
load cases.

4.

Obtain the structures stiffness by nonlinear static analysis.

5.

Plot the stiffness of the structure in order to obtain its yield point and maximum
displacement.

6.

Perform the nonlinear dynamic response analysis (time-history analysis), using


the selecting strong motion records. In order to get the hysteresis area, you need
to plot the coordinates of the hysteresis area to AutoCAD.

7.

Compute the ductility factors of the structure by performing the ductility equations
2, 3 and 4.

8.

Obtain the damage indices of the structure in each excitation level using eq.5.
Use table 3.1 to calibrate the index of the damage done.

9.

Obtain the total occurrence for each damage rank and get the damage ratio.

10.

Construct the fragility curve by using the computed damage ratio and the ground
motion indices for each rank.

The damage ratio is defined as the number of occurrence of each damage rank
(no, slight, moderate, extensive, and complete) divided by the total number of records.
Once obtained, the damage ratio is plotted with the in (PGA) on a lognormal probability
paper to obtain the necessary parameters (mean and standard deviation).

54

Once the mean and standard deviation have been obtained, the cumulative
probability (PR) of occurrence of the damage equal or higher than the damage rank can
be computed using equation 6. Then by simply plotting acquired cumulative probability
with the peak ground acceleration (PGA normalized to different excitation), the fragility
curve can now be obtained.

4.6.2 Conventional Seismic Fragility Curves


Conventional seismic fragility curves are created by plotting the cumulative
lognormal probability against peak ground acceleration. The steps to develop
conventional seismic fragility curves are as follows:
1. Obtain ductility factors. After obtaining the results from the two nonlinear analyses,
the ductility factors namely displacement ductility, ultimate ductility and hysteric
energy ductility factors are computed using the equations 2,3 and 4, respectively
which was adopted from Karim and Yamazaki (2001)
2. Obtain damage indices. Using equation 5 taking as 0.15, damage indices for
every PGA are calculated considering the values obtained from the previous step.
3. Determine the damage rank. After damage indices are obtained, damage rank can
be determined referring to table 2.
4. Obtain the probability of occurrence. Determine the probability of occurrence by
dividing the number of occurrence to the total number of occurrence.
5. Obtain the cumulative lognormal probability. Using equation 16, the probability of
exceedance is calculated for every PGA.

55

6. Plot the conventional seismic fragility curves. From the statement above, the
fragility curves is developed by plotting the lognormal probability versus the peak
ground acceleration.

4.6.3 Interval Uncertainty Analysis


The process for the interval uncertainty analysis is quite the same as the
conventional procedure but the parameters obtained are bounded. The procedure for
generating IUA seismic fragility curves are the following:
1. From the results of the two nonlinear analyses; maximum displacement for static
max (static), maximum displacement for dynamic max (dynamic),

yield

displacement for static (y ), hysteretic energy (Eh ), and yield energy (Ee ), the
interval showing the lower bound and upper bound of the parameters are
computed using the following equations taking COV or coefficient of variation as
5%.

max (static) = [max (1 COV)


max (dynamic) = [max (1 COV)

max (1 + COV)]
max (1 + COV)]

(24)
(25)

y = [y (1 COV)

y (1 + COV)]

(26)

h = [Eh (1 COV)
E

Eh (1 + COV)]

(27)

e = [Ee (1 COV)
E

Ee (1 + COV)]

(28)

56

2. Obtain ductility factors. For the IUA process, the ductility factors are also obtained
with boundaries using equations 13, 14 and 15. The ductility factors for the
boundaries (lower and upper) are computed separately.
3. Obtain damage indices. Using equation 16 taking as 0.15, damage indices for
every PGA are calculated considering the values obtained from the previous step.
Damage indices are computed differently for the lower bound and upper bound.
4. Determine the damage rank. After damage indices are obtained, damage rank for
each boundary can be determined referring to Table 2.
5. Obtain the probability of occurrence. Determine the probability of occurrence by
dividing the number of occurrence to the total number of occurrence.
6. Obtain the cumulative lognormal probability. Using equation 17, the probability of
exceedance is calculated for every PGA and for each boundary. Using equations
18 and 19, the statistical parameters of the lognormal probability is computed.
7. Plot the IUA seismic fragility curves. Plot different fragility curve for the lower bound
and another for the upper bound.
8. Compare conventional fragility curves to IUA fragility curves. Determine the mean
of the lower and upper bound fragility curve to come up with a new IUA seismic
fragility curve which will be compared to the conventional seismic fragility curve.

57

CHAPTER 5
RESULTS AND DISCUSSIONS

5.1 Fragility Analysis

The results of the Nonlinear Static Analysis or Pushover analysis is a pushover


curve as seen in Figure 19.

It can be seen from this pushover curve the yield

displacement of 0.016479 m and a maximum displacement of 0.138127 m.

Pushover Curve
14000
0.138127, 11717.678

12000

Force (kN)

10000

0.016479, 10652.435

8000

6000

4000

2000

0
0

0.2

0.4

0.6

0.8

displacement (m)

Figure 19. Pushover curve from nonlinear static analysis of SAP2000.

58

1.2

The pushover curve in Figure 19 is zoomed-in, thus, Figure 20 can be seen


showing the area bounded by the origin to the yield point. This area is called the energy
yield or Ee. The computed value of the energy at yield point is 87.77073818 kJ.

Pushover Curve (zoomed in)


14000

0.016479, 10652.435
12000

Force (kN)

10000

8000

6000

4000

2000

0.02

0.019

0.018

0.017

0.016

0.015

0.014

0.013

0.012

0.011

0.01

0.009

0.008

0.007

0.006

0.005

0.004

0.003

0.002

0.001

displacement (m)

Figure 20. Pushover curve (zoomed in) to compute the energy at yield Ee.

After the researcher ran the model in the pushover analysis, he used the timehistory analysis with the result of hysteresis model as shown in Figure 21. This hysteresis
figure was then plot to Autodesks AutoCAD to compute for the area of the hysteresis.
The coordinates of the hysteresis were derived from the SAP2000 results.

Thus,

computing the area known in this literature as Hysteretic energy, Eh. This can be seen
in Figure 22. The maximum displacement in dynamic analysis for a PGA of 0.2g in the

59

Tohoku-Kanto (Fukushima station) is found to be 0.03564 m and hysteretic energy of


3.45667199 kJ.

Figure 21. Hysteresis Model of Tohoku-Kanto Fukushima-2g (1 out of 300 hysteresis


model) using SAP2000.

60

Figure 22. Hysteretic energy computation using Autodesk's AutoCAD.

Abovementioned result values are then collected as follows derived from the
nonlinear analyses of SAP2000:
static
0.138127 m; E e 87.77073818 kJ
y 0.016479 m; max

max dynamic 0.03564 m; E h 3.45667199 kJ


From these values, the following ductility factors can be solved.

max dynamic
0.03564

2.162753
y
0.016479

max static 0.138127


u

8.382001
y
0.016479

Eh
3.45667199

0.039383
Ee 87.77073818

61

The damage index can now be calculated with the cyclic loading be 0.15 for
bridges.

ID

d h 2.162753 0.15 0.039383

0.258728
u
8.382001

Based from the damage rank of Table 2, the corresponding damage rank for a
damage index of 0.258728 falls in the closed interval [0.14, 0.40], that is, D which is
equivalent to Slight Damage definition. Using MS Excel spreadsheet, these sample
computations were done as shown in Table 5.
Table 5. Summary of calculations of ductility factors, damage indices, and damage
ranks using MS Excel Spreadsheet. For Tohoku-Kanto Eq. (Fukushima)
NLSA
PGA/g

NLDA

DUCTILITY FACTORS

DAMAGE
INDEX

DAMAGE
RANK

max

Ee

max

Eh

(DI)

(DR)

0.2

0.138

0.016

87.771

0.036

0.016

3.457

2.163

8.382

0.039

0.259

0.4

0.138

0.016

87.771

0.071

0.016

13.827

4.326

8.382

0.158

0.519

0.6

0.138

0.016

87.771

0.107

0.016

31.110

6.488

8.382

0.354

0.780

0.8

0.138

0.016

87.771

0.143

0.016

55.307

8.651

8.382

0.630

1.043

As

1.0

0.138

0.016

87.771

0.178

0.016

86.417

10.814

8.382

0.985

1.308

As

1.2

0.138

0.016

87.771

0.214

0.016

124.440

12.977

8.382

1.418

1.574

As

1.4

0.138

0.016

87.771

0.249

0.016

169.377

15.139

8.382

1.930

1.841

As

1.6

0.138

0.016

87.771

0.285

0.016

221.227

17.302

8.382

2.521

2.109

As

1.8

0.138

0.016

87.771

0.321

0.016

279.990

19.465

8.382

3.190

2.379

As

2.0

0.138

0.016

87.771

0.356

0.016

345.667

21.628

8.382

3.938

2.651

As

62

5.2 Probability of Occurrence

After calculating the conventional and IUA damage indices using the Algorithm 1
and MS Excel spreadsheets, Tables 5, 6, and 7 summarize the respective damage ratios.
For this tabulation, the initial value of coefficient of variation (c.o.v.) of 5% was used.

Table 6. Summary of Damage Ratio for the Conventional fragility curves

DAMAGE RANK
PGA
0.2 g
0.4 g
0.6 g
0.8 g
1.0 g
1.2 g
1.4 g
1.6 g
1.8 g
2.0 g

As

0.4
0.2333333
0.2333333
0.1333333
0
0
0
0
0
0

0.046875
0.125
0.109375
0.0625
0.109375
0.109375
0.109375
0.109375
0.109375
0.109375

0
0
0.0909091
0.5454545
0.3636364
0
0
0
0
0

0
0
0
0.047619
0.1904762
0.3333333
0.2857143
0.0952381
0.047619
0

0
0
0
0
0
0.041667
0.083333
0.25
0.291667
0.333333

Table 7. Summary of Damage Ratio for the IUA fragility curves (Lower Bound)

DAMAGE RATIO
PGA
0.2 g
0.4 g
0.6 g
0.8 g

D
0.4285714
0.25
0.25
0.0714286

0.0461538
0
0
0.1230769
0
0
0.0923077 0.1818182
0
0.0923077 0.5454545 0.05
63

As
0
0
0
0

0
0
0
0
0
0

1.0 g
1.2 g
1.4 g
1.6 g
1.8 g
2.0 g

0.1076923 0.2727273 0.25


0
0.1076923
0
0.35 0.038462
0.1076923
0
0.3 0.076923
0.1076923
0
0.05 0.269231
0.1076923
0
0
0.307692
0.1076923
0
0
0.307692

Table 8. Summary of Damage Ratio for the IUA fragility curves (Upper Bound)

DAMAGE RATIO
PGA
0.2 g
0.4 g
0.6 g
0.8 g
1.0 g
1.2 g
1.4 g
1.6 g
1.8 g
2.0 g

As

0.5
0.5
0
0
0
0
0
0
0
0

0.1481481
0.1296296
0.1296296
0.1296296
0.1296296
0.1296296
0.1296296
0.0740741
0
0

0
0.0416667
0.25
0
0
0
0
0.125
0.2916667
0.2916667

0
0
0.1538462
0.5384615
0.3076923
0
0
0
0
0

0
0
0
0.022222
0.088889
0.177778
0.177778
0.177778
0.177778
0.177778

64

Probability of Occurrence
16

14

FREQUENCY

12
10

8
6
4
2
0
0.2 g

0.4 g

0.6 g

0.8 g
D

1.0 g
C

1.2 g
A

1.4 g

1.6 g

1.8 g

2.0 g

As

Figure 23. Probability of Occurrence for Conventional Fragility Curves

Probability of Occurrence
16

14

FREQUENCY

12
10
8
6
4
2

0
0.2 g

0.4 g

0.6 g

0.8 g
D

1.0 g
C

1.2 g
A

1.4 g

1.6 g

1.8 g

2.0 g

As

Figure 24. Probability of Occurrence for IUA-Lower Bound Fragility Curves

65

Probability of Occurrence
16

14

FREQUENCY

12
10
8
6
4
2
0
0.2 g

0.4 g

0.6 g

0.8 g
D

1.0 g
C

1.2 g
A

1.4 g

1.6 g

1.8 g

2.0 g

As

Figure 25. Probability of Occurrence for IUA-Upper Bound Fragility Curves

The lognormal plot of damage ratio can be shown in Figure 26 to Figure 30. These
lognormal plot of PGA versus the damage ratio is useful to find the statistical parameters
needed in the cumulative normal probability function of equation 17. The mean and
standardard deviation formulas, Equation 18 and Equation 19, can now be applied using
these lognormal plot of damage ratios.

66

LOGNORMAL PLOT OF DAMAGE RATIO FOR DR="D"


120%

y = -0.4338x + 3.2988
R = 0.9565

DAMAGE RATIO (in %)

100%

80%

60%

40%

20%

0%
5

5.5

6.5

7.5

LN(PGA)

Figure 26. Plot of Lognormal of PGA to the Damage Ratio of DR="D".

67

LOGNORMAL PLOT OF DAMAGE RATIO FOR DR="C"


70%

60%

y = 0.2416x - 1.194
R = 0.7379

50%

40%

30%

20%

10%

0%
5

5.5

6.5

7.5

Figure 27. Plot of Lognormal of PGA to the Damage Ratio of DR="C".

68

LOGNORMAL PLOT OF DAMAGE RATIO FOR DR="B"


25%

20%
y = 0.0867x - 0.4954
R = 0.6598
15%

10%

5%

0%
5

5.5

6.5

7.5

-5%

Figure 28. Plot of Lognormal of PGA to the Damage Ratio of DR="B".

69

LOGNORMAL PLOT OF DAMAGE RATIO FOR DR="A"


25%

20%
y = 0.0677x - 0.3997
R = 0.4577

15%

10%

5%

0%
5

5.5

6.5

7.5

-5%

-10%

Figure 29. Plot of Lognormal of PGA to the Damage Ratio of DR="A".

70

LOGNORMAL PLOT OF DAMAGE RATIO FOR DR="As"


9%
8%
7%
6%
y = 0.0378x - 0.2098
R = 0.739

5%
4%
3%
2%
1%
0%
5

5.5

6.5

7.5

-1%
-2%

Figure 30. Plot of Lognormal of PGA to the Damage Ratio of DR="As".

Table 8 shows the number of occurrence of damage rank for every PGA value.
This number of occurrence is designated here as f, which means frequency. At the
bottom of the table, the total for each damage rank was calculated. It should be noted
that total number of number of occurrence is equal to the number of ground motion data
multiplied by the number of PGA values, that is, 15 ground motion data x 10 PGAs per
ground motion data = 150.

71

Table 9. Tabulation of ln(PGA) with number of occurrences per damage rank.

PGA
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

PGA
196.2
392.4
588.6
784.8
981
1177.2
1373.4
1569.6
1765.8
1962

X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

Number of Occurrence = f
C
B
A
3
0
0
8
0
0
7
1
0
4
6
1
7
4
4
7
0
7
7
0
6
7
0
2
7
0
1
7
0
0

D
12
7
7
4
0
0
0
0
0
0

SUM

30

64

11

As
0
0
0
0
0
1
2
6
7
8

21

24

Applying Equation 18 and Equation 19 for the statistical parameters, Table 9


summarizes the result when the natural logarithm values of the PGA (which was
designated as X) is multiplied to the number of occurrence per damage rank.
Table 10. Tabulation of the product of number of occurrences to ln(PGA)
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL
MEAN
STDEV

X*f
B

63.34961
41.80597
44.64423
26.66172
0
0
0
0
0
0
176.4615
5.882051
1.143325

15.8374
47.77825
44.64423
26.66172
48.22001
49.49626
50.57531
51.51003
52.33451
53.07204
440.1298
6.877028
0.700885

0
0
6.377747
39.99257
27.55429
0
0
0
0
0
73.92461
6.720419
0.698838

72

As

0
0
0
6.665429
27.55429
49.49626
43.35027
14.71715
7.476359
0
149.2598
7.107607
0.76468

0
0
0
0
0
7.070894
14.45009
44.15146
52.33451
60.65376
178.6607
7.444196
0.95505

Table 10 shows the calculations of the square of the difference of the natural
logarithm of PGA values, X, to the mean value, . These values were needed for the
calculation of the standard deviation for each damage rank. Below this table is a row
designated to total each damage rank (x-)2 values.

Table 11. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value.
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL

(X-)2
B

0.363508
0.008142
0.245714
0.613681
1.013085
1.413348
1.803632
2.180126
2.541818
2.888873
13.07193

2.553262
0.818565
0.249281
0.044774
0.000133
0.037584
0.121116
0.231889
0.359198
0.496591
4.912394

2.077301
0.55971
0.117424
0.003024
0.028276
0.122833
0.254647
0.407244
0.571445
0.741839
4.883742

As

3.343313
1.288964
0.532697
0.195522
0.047976
0.001348
0.013792
0.062985
0.135978
0.224782
5.847357

4.687492
2.166532
1.137314
0.606479
0.308718
0.139355
0.048027
0.007331
0.001034
0.018913
9.121196

5.3 Fragility Curves (Conventional)

Table 12 and Table 13 are the needed tabulations to calculate the probability of
exceedance for every damage rank versus the PGA values. Using MS Excel spreadsheet
solution, the conventional fragility curves can now be plotted which is shown in Figure 31.

73

Table 12. Tabulation of the values for the (X - Mean ) / Standard Deviation
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

(X - )/
B

-0.52734
0.07892
0.433556
0.685175
0.880346
1.039812
1.174638
1.29143
1.394448
1.486601

-2.27982
-1.29086
-0.71236
-0.3019
0.016472
0.276603
0.49654
0.687058
0.855107
1.005432

As

-2.0624 -2.39116 -2.26696


-1.07055 -1.48471 -1.54119
-0.49035 -0.95447 -1.11664
-0.07869 -0.57825 -0.81542
0.240619 -0.28644 -0.58177
0.501511 -0.04801 -0.39087
0.722093 0.153577 -0.22947
0.913169 0.328201 -0.08965
1.08171
0.48223 0.033677
1.232476 0.620014 0.143996

Table 13. Tabulation of the values for the Probability of Exceedance per PGA/g values.
X
PGA/g
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

(X - )/ = Probability of Exceedance
C
B
A
As

0.29898
0.531452
0.667695
0.753383
0.810664
0.850786
0.87993
0.901723
0.918409
0.93144

0.011309
0.098376
0.238122
0.381363
0.506571
0.608957
0.690243
0.753977
0.803754
0.842656

74

0.019585
0.142187
0.311945
0.46864
0.595075
0.691994
0.764881
0.819423
0.860309
0.891114

0.008398
0.068811
0.169924
0.281547
0.387271
0.480854
0.561028
0.62862
0.685179
0.732376

0.011696
0.061635
0.132073
0.207416
0.280359
0.347946
0.409253
0.464283
0.513432
0.557248

Fragility Curves (Conventional)


100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

PGA (in g)

D(c)

C(c)

B(c)

A(c)

As(c)

Figure 31. Conventional fragility curves for the different damage ranks

75

2.5

5.4 Fragility Curves by Interval Uncertainty Analysis

5.4.1 Interval Uncertainty Analysis (Lower Bound)

Similar to the process of using MS Excel spreadsheet solution of section 5.3, i.e.,
the fragility curves by conventional method, Tables 14 to 18 show how Figure 31 for the
lower bound of the IUA Fragility Curve was plotted.
Table 14. Tabulation of ln(PGA) with number of occurrences per damage rank.

PGA
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

PGA
196.2
392.4
588.6
784.8
981
1177.2
1373.4
1569.6
1765.8
1962

X
Number of Occurrence = f
LN(PGA) D
C
B
A
As
3
0
0
0
5.279135 12
7
8
0
0
0
5.972282
7
6
2
0
0
6.377747
2
6
6
1
0
6.665429
0
7
3
5
0
6.888572
0
7
0
7
1
7.070894
0
7
0
6
2
7.225045
0
7
0
1
7
7.358576
0
7
0
0
8
7.476359
0
7
0
0
8
7.58172
28 65 11 20 26
SUM

76

Table 15 summarizes the result when the natural logarithm values of the PGA
(which was designated as X) is multiplied to the number of occurrence per damage rank.

Table 15. Tabulation of the product of number of occurrences to ln(PGA)


X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL
MEAN
STDEV

D
63.34961457
41.80597209
44.64422785
13.33085782
0
0
0
0
0
0
163.1306723
5.82609544
1.188228093

C
15.83740364
47.77825382
38.26648101
39.99257345
48.22000722
49.49625811
50.57531287
51.51003262
52.33451387
53.07203748
447.0828741
6.878198063
0.701031735

X*f
B
0
0
12.75549367
39.99257345
20.66571738
0
0
0
0
0
73.4137845
6.673980409
0.704949562

77

A
0
0
0
6.665428908
34.4428623
49.49625811
43.35026818
7.358576089
0
0
141.3133936
7.065669679
0.74821081

As
0
0
0
0
0
7.070894016
14.45008939
51.51003262
59.810873
60.65375712
193.4956461
7.442140236
0.953641404

Table 16 shows the calculations of the square of the difference of the natural
logarithm of PGA values, X, to the mean value, . These values were needed for the
calculation of the standard deviation for each damage rank. Below this table is a row
designated to total each damage rank (x-)2 values.

Table 16. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value.
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL

D
0.29917
0.02137
0.30432
0.70448
1.12886
1.54952
1.95706
2.3485
2.72337
3.08222
14.1189

C
2.557
0.82068
0.25045
0.04527
0.00011
0.03713
0.1203
0.23076
0.3578
0.49494
4.91445

(X-)2
B
1.94559
0.49238
0.08775
7.3E-05
0.04605
0.15754
0.30367
0.46867
0.64381
0.82399
4.96954

A
3.19171
1.1955
0.47324
0.16019
0.03136
2.7E-05
0.0254
0.08579
0.16867
0.26631
5.59819

As
4.67859
2.16048
1.13293
0.60328
0.30644
0.13782
0.04713
0.00698
0.00117
0.01948
9.09432

Table 17 and Table 18 are the needed tabulations to calculate the probability of
exceedance for every damage rank versus the PGA values. Using MS Excel spreadsheet
solution, the IUA fragility curves for lower bound can now be plotted which is shown in
Figure 32.

78

Table 17. Tabulation of the values for the ( X - Mean ) / Standard Deviation
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

D
-0.4603
0.12303
0.46426
0.70637
0.89417
1.04761
1.17734
1.28972
1.38884
1.47751

C
-2.281
-1.2923
-0.7139
-0.3035
0.0148
0.27487
0.49477
0.68524
0.85326
1.00355

( X - )/
B
-1.9786
-0.9954
-0.4202
-0.0121
0.30441
0.56304
0.78171
0.97113
1.13821
1.28767

A
As
-2.3877 -2.2682
-1.4613 -1.5413
-0.9194 -1.1161
-0.5349 -0.8145
-0.2367 -0.5805
0.00698 -0.3893
0.21301 -0.2276
0.39148 -0.0876
0.5489 0.03588
0.68971 0.14636

Table 18. Tabulation of the values for the Probability of Exceedance per PGA/g values.
X
PGA/g
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

(X - )/ = Probability of Exceedance
D
C
B
A
As
0.32264 0.01127 0.02393 0.00848 0.01166
0.54896 0.09813 0.15977 0.07196 0.06162
0.67877 0.23765 0.33716 0.17894 0.13218
0.76002 0.38075 0.49516 0.29635 0.20769
0.81438
0.5059 0.61959 0.40645
0.2808
0.85259 0.60829
0.7133 0.50279 0.34853
0.88047 0.68962 0.78281 0.58434 0.40996
0.90143 0.75341 0.83426 0.65228 0.46509
0.91756 0.80324 0.87248 0.70846 0.51431
0.93023
0.8422 0.90107 0.75481 0.55818

79

IUA Fragility Curves (Lower Bound)


100%
90%

Probability of Exceedance

80%
70%
60%
50%

40%
30%
20%
10%
0%
0.0

0.5

1.0

1.5

2.0

2.5

PGA in g
D

As

Figure 32. IUA (Lower Bound) fragility curves for different damage ranks

5.4.2 Interval Uncertainty Analysis (Upper Bound)

Similar to the process of using MS Excel spreadsheet solution of section 5.3, i.e.,
the fragility curves by conventional method, Tables 19 to 23 show how Figure 23 for the
upper bound of the IUA Fragility Curve was plotted.

80

Table 19. Tabulation of ln(PGA) with number of occurrences per damage rank.

PGA
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

PGA
196.2
392.4
588.6
784.8
981
1177.2
1373.4
1569.6
1765.8
1962

X
Number of Occurrence = f
LN(PGA) D
C
B
A
As
7
8
0
0
0
5.279135
7
7
1
0
0
5.972282
0
7
6
2
0
6.377747
0
7
0
7
1
6.665429
0
7
0
4
4
6.888572
0
7
0
0
8
7.070894
0
7
0
0
8
7.225045
0
4
3
0
8
7.358576
0
0
7
0
8
7.476359
0
0
7
0
8
7.58172
14 54 24 13 45
SUM

Table 19 summarizes the result when the natural logarithm values of the PGA
(which was designated as X) is multiplied to the number of occurrence per damage rank.

Table 20. Tabulation of the product of number of occurrences to ln(PGA)


X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL
MEAN
STDEV

D
36.9539
41.806
0
0
0
0
0
0
0
0
78.7599
5.62571
1.35579

C
42.2331
41.806
44.6442
46.658
48.22
49.4963
50.5753
29.4343
0
0
353.067
6.53828
0.73942

81

X*f
B

As

0
5.97228
38.2665
0
0
0
0
22.0757
52.3345
53.072
171.721
7.15504
0.78559

0
0
12.7555
46.658
27.5543
0
0
0
0
0
86.9678
6.68983
0.70252

0
0
0
6.66543
27.5543
56.5672
57.8004
58.8686
59.8109
60.6538
327.92
7.28712
0.85507

Table 20 shows the calculations of the square of the difference of the natural
logarithm of PGA values, X, to the mean value, . These values were needed for the
calculation of the standard deviation for each damage rank. Below this table is a row
designated to total each damage rank (x-)2 values.

Table 21. Tabulation of the square of the difference of the mean value of ln(PGA) to a PGA
value.
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172
TOTAL

D
0.12011
0.12011
0.56556
1.08102
1.59483
2.08856
2.55788
3.00283
3.42491
3.82598
18.3818

C
1.58545
0.32035
0.02577
0.01617
0.1227
0.28368
0.47164
0.67288
0.87999
1.08876
5.46741

(X-)2
B
3.51903
1.39893
0.60419
0.23972
0.07101
0.00708
0.0049
0.04143
0.10324
0.18205
6.17158

A
1.99006
0.51488
0.0974
0.0006
0.0395
0.14521
0.28646
0.44722
0.61863
0.79547
4.93541

As
4.03201
1.7288
0.82696
0.3865
0.15884
0.04675
0.00385
0.00511
0.03581
0.08679
7.31143

Table 21 and Table 22 are the needed tabulations to calculate the probability of
exceedance for every damage rank versus the PGA values. Using MS Excel spreadsheet
solution, the IUA fragility curves for upper bound can now be plotted which is shown in
Figure 33.

82

Table 22. Tabulation of the values for the ( X - Mean ) / Standard Deviation
X
LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

D
-0.2556
0.25562
0.55468
0.76687
0.93146
1.06593
1.17963
1.27812
1.36499
1.4427

C
-1.7029
-0.7655
-0.2171
0.17196
0.47374
0.72031
0.92879
1.10938
1.26867
1.41116

(X - )/
B
-2.3879
-1.5056
-0.9894
-0.6232
-0.3392
-0.1071
0.08911
0.25908
0.40901
0.54313

A
As
-2.008 -2.3483
-1.0214 -1.5377
-0.4442 -1.0635
-0.0347 -0.7271
0.2829 -0.4661
0.54242 -0.2529
0.76185 -0.0726
0.95192 0.08357
1.11958 0.22131
1.26955 0.34453

Table 23. Tabulation of the values for the Probability of Exceedance per PGA/g values.
X
PGA/g
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

LN(PGA)
5.279135
5.972282
6.377747
6.665429
6.888572
7.070894
7.225045
7.358576
7.476359
7.58172

(X - )/ = Probability of Exceedance
D
C
B
A
As
0.39912 0.04429 0.00847 0.02232 0.00943
0.60088
0.222 0.06609 0.15354 0.06206
0.71044 0.41406 0.16122 0.32844 0.14378
0.77842 0.56826 0.26656 0.48615 0.23359
0.82419 0.68216 0.36723 0.61137 0.32057
0.85677 0.76433 0.45735 0.70624 0.40018
0.88093
0.8235
0.5355 0.77692 0.47106
0.8994 0.86637 0.60221 0.82943
0.5333
0.91387 0.89772 0.65873 0.86855 0.58758
0.92545
0.9209 0.70648 0.89788 0.63478

83

IUA (Upper Bound) Fragility Curves


100%
90%

Probability of Exceedance

80%
70%
60%
50%

40%
30%
20%
10%
0%
0.0

0.5

1.0

1.5

2.0

PGA in g
D

As

Figure 33. IUA (Upper Bound) fragility curves for different damage ranks.

84

2.5

5.5 Comparison of Fragility Curves: Conventional vs. IUA

The fragility curves based from the results of the conventional and interval
uncertainty analysis are compared in the following graphs:
Type A: Conventional to IUA (Lower and Upper Bound) for every damage rank
Type B: Conventional to IUA (Mean) for every damage rank
The graphs of Figures 34 to 38 show that for a given c.o.v.=5% of Type A

Bounded Fragility Curves, DR="D"


1

Probability of Exceedance (in %)

0.9
0.8
0.7
0.6
0.5

0.4
0.3
0.2
0.1
0
0

0.5

1.5

PGA (in g)
D(c)

D(L)

D(U)

Figure 34. Bounded fragility curve of DR="D" c.o.v.=5% in X-direction

85

2.5

Bounded Fragility Curves, DR="C"


0.9

Probability of Exceedance (in %)

0.8
0.7

0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.5

1.5

PGA (in g)
C(c)

C(L)

C(U)

Figure 35. Bounded fragility curve of DR="C" c.o.v.=5% in X-direction

86

2.5

Bounded Fragility Curves, DR="B"


0.7

Probability of Exceedance (in %)

0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.5

1.5

PGA (in g)
B(c)

B(L)

B(U)

Figure 36. Bounded fragility curve of DR="B" c.o.v.=5% in X-direction

87

2.5

Bounded Fragility Curves, DR="A"


0.8

Probability of Exceedance (in %)

0.7

0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.5

1.5

PGA (in g)
A(c)

A(L)

A(U)

Figure 37. Bounded fragility curve of DR="A" c.o.v.=5% in X-direction

88

2.5

Bounded Fragility Curves, DR="As"


0.8

Probability of Exceedance (in %)

0.7

0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.5

1.5

2.5

PGA (in g)
As(c)

As(L)

As(U)

Figure 38. Bounded fragility curve of DR="As" c.o.v.=5% in X-direction

It can be observed that the bounded fragility curves differ in the norm of a
bounded conventional fragility curve, that is, the lower bound ordinate must be lower
than that of conventional, and upper bound ordinate must be higher than that of
conventional. Figures 38 to 42 are the charts of the superimposed conventional and
IUA (Mean). The IUA (Mean) values were computed as the average of the lower and
upper bound values of probability of exceedance for every given PGA. In agreement to
the norm as mentioned above, the difference of the conventional to the IUA (Mean)
must be negligible, if not equal to zero. With these conditions, only the fragility curves
of Types A and B whose damage ranks are D, C, and B passed. Thus, there was a
need to run the IUA_DI.m program to the following c.o.v.s: 1%, 10%, and 20%.
89

Fragility Curves (Conventional vs IUA-Mean) DR="D"


100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%
40%
30%

20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
D(c)

D(IUA-M)

Figure 39. Conventional and IUA (Mean) fragility curves of DR="D" c.o.v.=5% in X-direction

90

Fragility Curves (Conventional vs IUA-Mean) DR="C"


90%

Probability of Exceedance (in %)

80%
70%

60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
C(c)

C(IUA-M)

Figure 40. Conventional and IUA (Mean) fragility curves of DR="C" c.o.v.=5% in X-direction

91

Fragility Curves (Conventional vs IUA-Mean) DR = "B"


100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%

40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
B(c)

B(IUA-M)

Figure 41. Conventional and IUA (Mean) fragility curves of DR="B" c.o.v.=5% in X-direction

92

FRAGILITY CURVES (Conventional vs IUA - mean)


DR = "A"
100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
A(c)

A(IUA-M)

Figure 42. Conventional and IUA (Mean) fragility curves of DR="A" c.o.v.=5% in X-direction

93

FRAGILITY CURVES (Conventional vs IUA - mean)


DR = "As"
100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
As(c)

As(IUA-M)

Figure 43. Conventional and IUA (Mean) fragility curves of DR="As" c.o.v.=5% in X-direction

After applying different values of c.o.v. (1%, 5%, 10%, and 20%), observed
bounded fragility curves which are consistent to the set norm are summarized in Table
23.

Table 24. Checklist of bounded fragility curves that subscribe to the "norm".
DAMAGE
RANK
D
C
B
A
As

Damage
COV=1% COV=5%
Description
No Damage
X

Slight Damage
X

Moderate Damage
X

Extensive Damage
X
X
Complete Damage
X
X

94

COV=10%

COV=20%

X
X
X

X
X
X

Based from Table 24, coefficient of variation of 5% and 20% have the majority of
the bounded fragility curves as per damage rank following the set norm of the interval
uncertainty analysis.

The c.o.v. = 1% results as the least value resulted to non-

compliance of the set norm. Figures 44 to 50 are the plots of the bounded fragility
curves that passed the set norm. It can be observed from these damage ranks that only
the No Damage, Slight Damage, and Moderate Rank can be bounded by the IUA.
Table 24 to 26 summarizes these bounded fragility curves.

bounded fragility curves DR="D", c.o.v.=5%


100%

Probability of Exceedance (in %)

90%
80%

70%
60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

PGA (in g)
D(c)

D(L)

D(U)

Figure 44. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=5% .

95

2.5

bounded fragility curves DR="C", c.o.v.=5%


90%

Probabilit of Exceedance (in %)

80%
70%

60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

PGA (in g)
C(c)

C(L)

C(U)

Figure 45. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=5% .

96

2.5

bounded fragility curves DR="B", c.o.v.=5%


70%
60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
B(c)

B(L)

B(U)

Figure 46. Bounded fragility curves for damage rank of "Moderate Damage" of c.o.v.=5% .

97

bounded fragility curves DR="D", c.o.v.=10%


1

Probability of Exceedance (in %)

0.9
0.8
0.7
0.6
0.5

0.4
0.3
0.2
0.1
0
0

0.5

1.5

PGA (in g)
D(c)

D(L)

D(U)

Figure 47. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=10% .

98

2.5

bounded fragility curves DR="C", c.o.v.=10%


90%

Probability of Exceedance (in %)

80%
70%

60%
50%
40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
C(c)

C(L)

C(U)

Figure 48. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=10% .

99

bounded fragility curves DR="D", c.o.v.=20%


100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%

40%
30%
20%
10%
0%
0

0.5

1.5

PGA (in %)
D(c)

D(L)

D(U)

Figure 49. Bounded fragility curves for damage rank of "No Damage" of c.o.v.=20% .

100

2.5

bounded fragility curves DR="C", c.o.v.=20%


100%

Probability of Exceedance (in %)

90%
80%
70%
60%
50%

40%
30%
20%
10%
0%
0

0.5

1.5

2.5

PGA (in g)
C(c)

C(L)

C(U)

Figure 50. Bounded fragility curves for damage rank of "Slight Damage" of c.o.v.=20%.

Table 25. Tabulation of bounded fragility curves for DR="D","C", "B" for c.o.v.=5%.
c.o.v.=5%
PGA/
g

conventional
D(c)

C(c)

IUA (Lower Bound)


B(c)

D(L)

C(L)

IUA (Upper Bound)

B(L)

D(U)

C(U)

B(U)

0.0

0.2

0.158055

0.008539

0.008965

0.110597

0.008449

0.010008

0.208519

0.008725

0.009017

0.4

0.403439

0.073234

0.063029

0.343848

0.066497

0.061546

0.455343

0.076094

0.062878

0.6

0.578854

0.182363

0.149147

0.531282

0.162603

0.13935

0.616748

0.189722

0.14841

0.8

0.696224

0.30181

0.244273

0.662536

0.26899

0.224284

0.721474

0.313267

0.242841

1.0

0.775655

0.413376

0.336111

0.752955

0.370522

0.306648

0.791714

0.427669

0.334053

1.2

0.830714

0.510579

0.419613

0.815905

0.46125

0.382423

0.84046

0.526438

0.417063

1.4

0.869826

0.592505

0.493341

0.860493

0.539778

0.450364

0.875317

0.608931

0.490439

101

1.6

0.898242

0.660459

0.557483

0.892655

0.606673

0.510497

0.900887

0.67675

0.554355

1.8

0.919304

0.716421

0.612888

0.91626

0.663233

0.563385

0.920059

0.732124

0.609634

2.0

0.935193

0.762401

0.660605

0.933863

0.710914

0.60978

0.934705

0.777247

0.657304

Table 26. Tabulation of bounded fragility curves for DR="D","C" for c.o.v.=10%.
c.o.v.=10%
conventional

IUA (Lower Bound)

IUA (Upper Bound)

PGA/g

D(c)

C(c)

D(L)

C(L)

D(U)

C(U)

0.2

0.158055

0.008539

0.074022

0.008674

0.282313

0.011218

0.4

0.403439

0.073234

0.285408

0.064266

0.518617

0.097748

0.6

0.578854

0.182363

0.479144

0.154523

0.659433

0.236901

0.8

0.696224

0.30181

0.622775

0.254447

0.748363

0.379773

0.775655

0.413376

0.724397

0.350522

0.807795

0.504835

1.2

0.830714

0.510579

0.795973

0.437278

0.849334

0.607231

1.4

0.869826

0.592505

0.846834

0.513258

0.879413

0.688614

1.6

0.898242

0.660459

0.883448

0.578786

0.901823

0.752487

1.8

0.919304

0.716421

0.910181

0.634883

0.918916

0.802419

0.935193

0.762401

0.929972

0.682763

0.932211

0.841474

Table 27. Tabulation of bounded fragility curves for DR="D","C" for c.o.v.=20%.
c.o.v.=20%
conventional

IUA (Lower Bound)

IUA (Upper Bound)

PGA/g

D(c)

C(c)

D(L)

C(L)

D(U)

C(U)

0.2

0.158055

0.008539

0.033283

0.009733

0.368329

0.026734

0.4

0.403439

0.073234

0.191445

0.061728

0.580726

0.170019

0.6

0.578854

0.182363

0.378339

0.141241

0.698347

0.351021

0.8

0.696224

0.30181

0.535627

0.228329

0.771497

0.509151

0.775655

0.413376

0.655097

0.312744

0.820596

0.63212

1.2

0.830714

0.510579

0.742851

0.39024

0.855356

0.723899

1.4

0.869826

0.592505

0.806783

0.459515

0.880957

0.791532

1.6

0.898242

0.660459

0.853469

0.520617

0.900398

0.841337

1.8

0.919304

0.716421

0.887797

0.574159

0.915525

0.878191

0.935193

0.762401

0.913263

0.62095

0.927535

0.905661

102

Based from Equation 20, a new coefficient of variation (c.o.v.) can be calculated
for every PGA based from the lower and upper bound. These values are tabulated in
Table 28.

c.o.v.

xx
xx

(29)

Where:

x, x = the lower and upper bounds of the interval, respectively.


Table 28. Summary of calculated c.o.v. based from the bounded fragility curves which follow
the set norm.

PGA/g
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
MAXIMUM
MINIMUM
AVERAGE per DR
MEAN per set c.o.v.
ABSOLUTE
DIFFERENCE TO OLD
C.O.V.

D
0
516.1%
89.5%
29.6%
12.6%
6.0%
3.1%
1.6%
0.8%
0.3%
0.1%
516.1%
0.1%
66.0%

c.o.v.=5%
C
0
0.1%
1.2%
2.4%
3.2%
3.7%
3.9%
4.0%
3.9%
3.8%
3.5%
4.0%
0.1%
3.0%

c.o.v.=10%
D
C
0
0
58.5%
12.8%
29.0%
20.7%
15.8%
21.0%
9.2%
19.8%
5.4%
18.0%
3.2%
16.3%
1.9%
14.6%
1.0%
13.0%
0.5%
11.7%
0.1%
10.4%
58.5%
21.0%
0.1%
10.4%
12.5%
15.8%

c.o.v.=20%
D
C
0
0
83.4%
46.6%
50.4%
46.7%
29.7%
42.6%
18.0%
38.1%
11.2%
33.8%
7.0%
29.9%
4.4%
26.5%
2.7%
23.5%
1.5%
20.9%
0.8%
18.6%
83.4%
46.7%
0.8%
18.6%
20.9%
32.7%

23.7%

14.1%

26.8%

18.7%

4.1%

6.8%

B
0
-5.8%
0.9%
2.6%
3.2%
3.4%
3.5%
3.5%
3.4%
3.3%
3.2%
3.5%
-5.8%
2.1%

103

From Table 28, the nearest absolute difference of the mean c.o.v. per damage
rank to the set c.o.v. is that of c.o.v.=10%. But fairly enough, c.o.v.=20% is also near,
whereas that of c.o.v.=5%, the highest deviation to the set c.o.v. is observed, that is, an
approximate 500% of c.o.v. was calculated at a PGA of 0.2g. The derived optimum
coefficient of variation of 10% is in agreement with that of Nowak and Collins (2013) range
of values of 8% to 10% for bridges, with bias factor of 1.03 to 1.05.
It must be noted that equation 29 was derived in the following manner:
~
1) Given that an interval value of X [ x x] and the lower and upper boundary

values are x x x and x x x .


2) The mean of the lower and upper boundary values is calculated as:

x mean

xx
2

(30)

and based from Figure 16, the distance from the lower bound to the upper
bound or x x , is equal to 2 , that is:

x x 2

(31)

3) Let be the x mean , then according to Equation 21, coefficient of variation (c.o.v.)
can be calculated as:

c.o.v.

xmean

4) Using equations 30 and 31, equation 32 can be transformed to:

104

(32)

xx
c.o.v.

2
xx
2

(33)

Thus, equation 29 is derived for the coefficient of variation (c.o.v.).


5.6 Interval Uncertainty Analysis after First Pass
After determining the optimum coefficient of variation (c.o.v.), based from Figure
10, the second pass of interval uncertainty analysis was applied to the pushover curve
and the hysteresis area, which are results from SAP2000 program.
For the pushover curve in Figure 51 and inset in Figure 52, it shows the lower and
upper bounds where the maximum static displacement, max

(static),

and yield

displacement, y, was derived. In turn, these values form as the interval of the two of
three ductility factorsthe other one is the elastic energy, Eewhich was operated to
result an interval value of damage index, ID. In Table 29 and Table 30, an example of the
input to an Octave script based from a ground motion, that is, Bohol earthquake of 2013
is shown.

105

Pushover Curve (Base Shear vs. Displacement)


1.2

Displacement (meter)

1
0.8
0.6
0.4
0.2
0
0
-0.2

2000

4000

6000

8000

10000

12000

Base Shear (kN)


Displacement, m

Lower

Upper

Figure 51. Pushover curve with lower and upper bounds.

106

14000

Pushover Curve (Base Shear vs. Displacement)


0.2

Displacement (meter)

0.15

0.1

0.05

0
0

2000

4000

6000

-0.05

8000

10000

12000

14000

Base Shear (kN)


Displacement, m

Lower

Upper

Figure 52. Pushover curve (inset) with lower and upper bounds.

Table 29. Parameters from Nonlinear Static Analysis to be used in computing ductility factors
using Interval Analysis.

PGA
0.2g
0.4g
0.6g
0.8g
1.0g
1.2g
1.4g
0.6g
0.8g
2.0g

max
mean
0.116
0.116
0.116
0.116
0.116
0.116
0.116
0.116
0.116
0.116

max
lower
0.105
0.105
0.105
0.105
0.105
0.105
0.105
0.105
0.105
0.105

max
upper
0.128
0.128
0.128
0.128
0.128
0.128
0.128
0.128
0.128
0.128

STATIC NON-LINEAR ANALYSIS


y
y
y
Ee
mean
lower
upper
mean
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771
0.016
0.015
0.018
87.771

107

Ee
lower
78.994
78.994
78.994
78.994
78.994
78.994
78.994
78.994
78.994
78.994

Ee
upper
96.548
96.548
96.548
96.548
96.548
96.548
96.548
96.548
96.548
96.548

Table 30. Parameters from Nonlinear Dynamic Analysis to be used in computing ductility
factors using Interval Analysis.

PGA
0.2g
0.4g
0.6g
0.8g
1.0g
1.2g
1.4g
0.6g
0.8g
2.0g

max
mean
0.011
0.021
0.032
0.043
0.053
0.064
0.075
0.085
0.096
0.107

max
lower
0.010
0.019
0.029
0.038
0.048
0.058
0.067
0.077
0.087
0.096

max
upper
0.012
0.023
0.035
0.047
0.059
0.070
0.082
0.094
0.106
0.117

DYNAMIC NON-LINEAR ANALYSIS


y
y
y
Eh
mean lower
upper
mean
0.016
0.015
0.018
4.292
0.016
0.015
0.018
17.166
0.016
0.015
0.018
38.624
0.016
0.015
0.018
68.664
0.016
0.015
0.018
107.288
0.016
0.015
0.018
154.494
0.016
0.015
0.018
210.284
0.016
0.015
0.018
274.656
0.016
0.015
0.018
347.612
0.016
0.015
0.018
429.150

Eh
lower
3.862
15.449
34.761
61.798
96.559
139.045
189.255
247.191
312.851
386.235

Eh
upper
4.721
18.883
42.486
75.530
118.016
169.944
231.312
302.122
382.373
472.065

Using the data of Tables 29 and 30, the input were used to the Octave script to
compute the ductility factors and eventually, the damage indices, by interval arithmetic
operations. This can be shown in Table 31 and Table 32.
Table 31. Ductility factors computed from Octave script.

PGA
0.2g
0.4g
0.6g
0.8g
1.0g
1.2g
1.4g
0.6g
0.8g
2.0g

d
mean
0.648
1.296
1.944
2.592
3.240
3.889
4.537
5.185
5.833
6.481

d
lower
0.530
1.061
1.591
2.121
2.651
3.182
3.712
4.242
4.772
5.303

d
upper
0.792119,
1.584239,
2.376358,
3.168477,
3.960596,
4.752716,
5.544835,
6.336954,
7.129073,
7.921193,

DUCTILITY FACTORS
u
u
u
mean lower
upper
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,
7.068 5.783
8.639069,

108

h
mean
0.049
0.196
0.440
0.782
1.222
1.760
2.396
3.129
3.960
4.889

h
lower
0.040
0.160
0.360
0.640
1.000
1.440
1.960
2.560
3.240
4.000

h
upper
0.059760,
0.239040,
0.537839,
0.956158,
1.493998,
2.151356,
2.928235,
3.824634,
4.840552,
5.975990,

Table 32. Damage indices and damage rank based from HAZUS.

PGA
0.2g
0.4g
0.6g
0.8g
1.0g
1.2g
1.4g
0.6g
0.8g
2.0g

DAMAGE DAMAGE DAMAGE DAMAGE DAMAGE DAMAGE


INDEX
INDEX
INDEX
RANK
RANK
RANK
(DI)
(DI)
(DI)
(DR)
(DR)
(DR)
mean
lower
upper
mean
lower
upper
0.093
0.062
0.139
D
D
D
0.188
0.126
0.280
C
D
C
0.284
0.190
0.425
C
C
C
0.383
0.257
0.573
C
C
B
0.484
0.324
0.724
B
C
A
0.587
0.393
0.878
B
C
A
0.693
0.464
1.035
A
B
As
0.800
0.535
1.195
A
B
As
0.909
0.609
1.358
A
B
As
1.021
0.683
1.525
As
A
As

The First Pass of the interval uncertainty analysis was implemented again for this
new set of data. This time the IUA-FC was now established in the Second Pass. This
can be shown from Figures 53 to 57.
It can be observed that the IUA-FC are now following the norm which was set
previously. This shows that the norm was satisfied.
From the damage rank of C to A, the interval is significant, with the highest
lower-to-mean and upper-to-mean difference is attributed to damage rank C. But in the
point of view of costing, it will be practical if one can retrofit damaged structures in a
moderate to extensive damage rank.
The IUA-FC of damage rank As or the complete damage is obviously has the
smallest interval since the mean almost touches both the lower and upper bounds.

109

IUA FC for No Damage Bridge Pier


1

0.9

0.8

Probability of Exceedance (in percent)

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Peak Ground Acceleration in terms of "g"


D(m)

D(l)

D(u)

Figure 53. Interval Uncertainty Analysis seismic fragility curve of "No Damage" rank.

110

2.2

IUA FC for Slight Damage Bridge Pier


1

0.9

0.8

Probability of Exceedance (in percent)

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

2.2

Peak Ground Acceleration in terms of "g"


C(m)

C(l)

C(u)

Figure 54. Interval Uncertainty Analysis seismic fragility curve of "Slight Damage" rank.

111

IUA FC for Moderate Damage Bridge Pier


0.8

0.7

Probability of Exceedance (in percent)

0.6

0.5

0.4

0.3

0.2

0.1

0
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

2.2

Peak Ground Acceleration in terms of "g"


B(m)

B(l)

B(u)

Figure 55. Interval Uncertainty Analysis seismic fragility curve of "Moderate Damage" rank.

112

IUA FC for Extensive Damage Bridge Pier


0.8

0.7

Probability of Exceedance (in percent)

0.6

0.5

0.4

0.3

0.2

0.1

0
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

2.2

Peak Ground Acceleration in terms of "g"


A(m)

A(l)

A(u)

Figure 56. Interval Uncertainty Analysis seismic fragility curve of "Extensive Damage" rank.

113

IUA FC for Complete Damage Bridge Pier


0.7

0.6

Probability of Exceedance (in percent)

0.5

0.4

0.3

0.2

0.1

0
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

2.2

Peak Ground Acceleration in terms of "g"


As (m)

As (l)

As (u)

Figure 57. Interval Uncertainty Analysis seismic fragility curve of "Complete Damage" rank.

114

CHAPTER 6

SUMMARY, CONLUSION, & RECOMMENDATION

6.1 Summary
Seismic fragility curves were used in the reliability analysis of a bridge pier by
developing a methodology known as the interval uncertainty analysis.

Epistemic

uncertainty was quantified by this method by varying values of coefficient of variation, i.e.,
1%, 5%, 10%, and 20% that were used in the standard deviation computation, given the
mean value. With the conventional method of developing seismic fragility curves, lower
and upper bounds of the probability of exceedance. Both the nonlinear static analysis
and nonlinear dynamic analysis were used in determining these ductility factors that were
ranked using HAZUS

6.2 Conclusion
An alternative way of capturing the unknown but bounded epistemic uncertainty of
developing conventional seismic fragility curves based from nonlinear static analysis
(pushover analysis) and nonlinear dynamic analysis (time history analysis) was
developed using a methodology called interval uncertainty analysis (IUA). The behavior
of the asset (Bangkulasi Bridge, a.k.a. R-10 Bridge in Malabon City) was assessed after
subjecting one of its piers to a hazard (different ground motions, e.g., Bohol 2013

115

earthquake, etc.) and found out that at a PGA of 0.4g (design PGA) the asset is recorded
a 6.15% to 6.29% probability of exceedance of DR=B or Slight Damage, based from a
set c.o.v. of 5%, but a calculated c.o.v.=1%. Using interval uncertainty analysis, only
damage rank of No Damage, Slight Damage, and Moderate Damage based from
HAZUS have passed the set norm of applying this method, after running it through set
values of c.o.v., i.e., 1%, 5%, 10%, 20%. The optimum c.o.v. of 10% was used for the
second pass of interval uncertainty analysis. Thus, an interval uncertainty analysis
fragility curve (IUA-FC) was constructed for each damage rank, with the damage rank C
or Slight Damage was found to have the highest interval. But this is not recommended
IUA-FC damage rank, B and A can be utilized instead, since the structure can still
withstand such seismic hazard and from economical point of view, the structure can be
salvaged.

6.3 Recommendation

This study aimed to develop a new methodology for the reliability analysis of bridge
pier using interval uncertainty analysis. The following are some of the topics that were
observed whose scope is beyond the present study and it is worth mentioning for further
study:

The use of more coefficient of variation (c.o.v.) values to find the optimum c.o.v.
by considering the objective of minimizing the difference between the
conventional and IUA probability of exceedance, as constrained by the

116

condition that the conventional value shall be lower that the upper bound and
higher than the lower bound; thus, an optimization problem.

The use of other modes of failure other than shear failure of the bridge pier.

The application of interval uncertainty analysis to the development of fragility


curves of different structures, such as buildings, light railway transit, dams, etc.

The system reliability analysis of the whole bridge using interval analysis.

The finite element method application of interval analysis and apply to the
development of seismic fragility curves.

117

BIBLIOGRAPHY

Alcaraz, R. P., Cuadra, C. J., & Damian, R. S. (2015). Seismic assessment of Navotas fish port
complex. Caloocan: Undergraduate Thesis; University of the East - Caloocan.
Algura, D. O., Decal, A., Quilang, J. R., & Romero, E. J. (2015). Seismic Assessment of
Tullahan Bridge (Malabon-Valenzuela). Caloocan: Undergraduate Thesis; University of
the East - Caloocan.
Ang, A. H., & Tang, W. H. (2007). Probability Concepts in Engineering: Emphasis on
Applications to Civil and Environmental Engineering Volume 1 (2nd ed.). New Jersey:
John Wiley & Sons, Inc.
Baylon, M. B. (2015). Seismic assessment of transportation lifeline in Metro Manila. 2nd
CAMANAVA Studies Conference (pp. 1-7). Caloocan: University of the East - Caloocan.
Baylon, M. B., & Garciano, L. E. (2012). Harnessing ordinary monte carlo simulation in
computing the performance of a transportation lifeline in the Philippines . 1st
International Conference in Technology (InCiTe2012) (pp. 1-14). Manila: Far Eastern
University East Asia College.
Baylon, M., Garciano, L. E., & Koike, T. (2012). Assessing the performance of a transportation
lifeline in the philippines, the light rail transit (LRT) system, under a large magnitude
earthquake. IABSE Congress Report 18th Congress of IABSE (pp. 1710-1717). Seoul:
International Association for Bridge and Structural Engineering.
Canlas, L., Mallanao, R. N., San Diego, A., & Santiago, M. A. (2015). Seismic assessment of
Bangkulasi bridge piers. Caloocan: Undergraduate Thesis; University of the East Caloocan.
Choi, E., DesRoches, R., & Nielson, B. (2004). Seismic fragility of typical bridges in moderate
seismic zones. Engineering Structures, 187-199.
Chopra, A. K. (2012). Dynamic of Structures (Theory and Applicationsto Earthquake
Engineering). United States of America: Pearson Education, Inc.
Cruz, F. G., Gueco, F. E., Matammu, D. L., & Maglanoc, B. S. (2015). Seismic assessment of
Tullahan-Ugong bridge piers due to shear failure using fragility curves (CaloocanValenzuela) . Caloocan: Undergraduate Thesis; University of the East - Caloocan.
Deierlein, G. G., Reinhorn, A. M., & Willford, M. R. (2010). Nonlinear Structural Analysis: A
Guide for Practicing Engineers. In NEHRP, NEHRP Seismic Design Technical Brief No.
4 (pp. 2-3). USA: NEHRP.
118

Del Carmen, M. O., Kakilala, M., Santos, K., & Vicedo, N. (2015). Seismic assessment of Light
Rail Transit Line 1 South Extension. Caloocan: Undergraduate Thesis; University of the
East - Caloocan.
Deodatis, G., & Tantala, M. W. (2002). Development of seismic fragility curves for tall buildings.
15th ASCE Engineering Mechanics Conferene (pp. 1-8). New York: Columbia University.
Gao, W., Wu, D., Song, C., Tin-Loi, F., & Li, X. (2011). Hybrid probabilistic interval analysis of
bar structures with uncertainty using a mixed perturbation Monte-Carlo method. Finite
Elements in Analysis and Design, 643-652.
Ghosh, J., & Padgett, J. (2010). Aging considerations in the development of time-dependent
seismic fragility curves. Journal of Structural Engineering.
Gomez, H., Torbol, M., & Feng, M. (2013). Fragility analysis of highway bridges based on longterm monitoring data. Computer-Aided Civil and Infrastructure Engineering.
HAZUS-MH. (2013, July 26). Retrieved September 04, 2015, from A Federal Emergency
Management Agency Website: http://www.fema.gov/media-library-data/20130726-171625045-6422/hazus_mr4_earthquake_tech_manual.pdf
Jernigan, J., & Hwang, H. (2002). Development of bridge fagility curves. 7th US National
Conference on Earthquake Engineering. Boston, Massachusetts: EERI.
Jiang, C., Zhang, Z., Han, X., & Liu, J. (2013). A novel evidence-theory-based reliability analysis
method for structures with epistemic uncertainty. Computers and Structures, 1-12.
Jiang, H., Fu, B., Lu, X., & Chen, L. (2012). Constant - damage Yield Strength Spectra. 15th
World Conference on Earthquake Engineering. Lisbon.
Kafali, C., & Grigoriu, M. (2004). Seismic fragility analysis. 9th ASCE Specialty Conference on
Probabilistic Mechanics and Structural Reliability. USA: ASCE.
Karim, K. R., & Yamazaki, F. (2001). Effect of Earthquake Ground Motions on Frgility Curves of
Highway Bridge Piers Based on Numerical Simulation. Earthquake Engineering and
Structural Dynamics.
Krawinkler, H., & Seneviratna. (1998). Pros and cons of a pushover analysis of sesmic
performance evaluation. Engineering Structures Vol. 20 Nos. 4-6, 452-464.
Li, F., Luo, Z., Rong, J., & Zhang, N. (2013). Interval multi-objective optimisation of strucures
using adaptive Kriging approximations. Computers and Structures, 68-84.
Majumder, L., & Rao, S. (2008). Interval-based optimization of aircraft wings under landing
loads. Computers and Structures, 225-234.
Modares, M., Taha, R., & Mohammadi, J. (2014). Reliability assessment of structures using
interval uncertainty analysis. ASCE: Vulnerability, Uncertainty, and Risk, 204-214.
119

Moore, R. E., Baker Kearfott, R., & Cloud, M. J. (2009). Introduction to Interval Analysis.
Philadelphia: Society of Industrial and Applied Mathematics.
Moore, R.E., Kearfott, R.B., Cloud, M. J. (2009). Introduction to Interval Analysis. United States
of America: Society for Industrial and Applied Mathematics.
Nielson, B. G. (2005). Analytical fragility curves for highway bridgesi in moderate seismic zones.
Atlanta: Doctor of Philiosophy Dissertation; Georgia Institute of Technology.
Nowak, A. S., & Collins, K. R. (2013). Reliability of structures. Boca Raton, Florida: CRC Press.
Park, Y.J., & Ang, A. H.S. (1985). Mechanistic Seismic Damage Model for Reinforced
Concrete. ASCE Journal of Structural Engineering 111(4), 722739.
Ramanathan, K. N. (2012, August). Next generation seismic fragility curves for california bridges
incorporating the evolution in seismic design philosophy. A Doctor of Philosophy
Dissertation. Georgia, USA: Georgia Institute of Technology.
Requiso, D. A. (2013). The generation of fragility curves of a pier under high magnitude
earthquakes (a case study of the metro rail transit-3 pier). Manila: Undergraduate
Thesis; De La Salle University.
Requiso, D. A., Balili, A., & Garciano, L. (n.d.). Development of Seismic Fragility Curves of a
Transport Lifeline Pier in the Philippines for Flexure.
Requiso, D. T., Balili, A., & Garciano, L. E. (2013). Development of seismic fragility curves of a
transportation lifeline pier in the Philippines. 16th ASEP International Convention.
Makati: Association of Structural Engineers of the Philippines, Inc.
Schneider, J. (2006). Introduction to Safety and Reliability of Structures. Zurich Switzerland:
International Association for Bridge and Structural Engineering.
Seismostruct user manual. (2002-2014). Italy: Seismosoft Ltd. All rights reserved.
Shinozuka, M., Feng, M. Q., Kim, H., Uzawa, T., & Ueda, T. (2003). Statistical Analysis of
Fragility Curves. MCEER.
Sorensen, J. (2004). Notes in Structural Reliability Theory and Risk Analysis.
Stefanidou, S. P., & Kappos, A. J. (2013). Optimum selection of retofit easures for R/C bridges
using fragility curves. 4th ECCOMAS Thematic Conference on Computational Methods
in Structural Dynamics and Earthquake Engineering. Athens Greece: COMPDYN 2013 .
Willink, R. (2006). On using the Monte Carlo method to calculate uncertainty intervals.
Metrologia, L39-L42.
Xu Han, Chao Jiang , LiXin Liu, Jie Liu, XiangYun Long. (2014). Response-surface-based
structural reliability analysis with random and interval mixed uncertainties. Science China
Technological Sciences, 57(7), 1322-1334. doi:10.1007/s11431-014-5581-6
120

Yang, C., Tangaramvong, S., Gao, W., & Tin-Loi, F. (2015). Interval elastoplastic analysis of
structures. Computers and Structures, 1-10.

121

APPENDICES

Algorithm 1. Calculation of damage index in interval form.


clc
clear all
PGA=[0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0];
dmaxstatic=[0.138127 0.138127 0.138127 0.138127 0.138127 0.138127 0.138127
0.138127 0.138127 0.138127];
dmaxdynamic=[0.01068 0.02136 0.03204 0.04272 0.0534 0.06408 0.07476 0.08544
0.09612 0.1068];
d_y=[0.016479 0.016479 0.016479 0.016479
0.016479 0.016479 0.016479
0.016479 0.016479 0.016479];
Energy_e=[87.77073818
87.77073818 87.77073818 87.77073818 87.77073818
87.77073818 87.77073818 87.77073818 87.77073818 87.77073818];
Energy_h=[4.291503297
17.16601319 38.62352967 68.66405275 107.2875824
154.4941187 210.2836616 274.656211 347.6117671 429.1503297];
beta_coef=0.15;
cov=0.05;
for i=1:10
dms=[dmaxstatic(i)*(1-cov) dmaxstatic(i)*(1+cov)];
dmd=[dmaxdynamic(i)*(1-cov) dmaxdynamic(i)*(1+cov)];
dy=[d_y(i)*(1-cov) d_y(i)*(1+cov)];
Ee=[Energy_e(i)*(1-cov) Energy_e(i)*(1+cov)];
Eh=[Energy_h(i)*(1-cov) Energy_h(i)*(1+cov)];
mu_d=intDivi(dmd,dy); mu_u=intDivi(dms,dy); mu_h=intDivi(Eh,Ee);
DI=intDivi(intSum(mu_d,beta_coef*mu_h),mu_u);
fprintf('%f %f \n',DI(1),DI(2));
end

122

Algorithm 2. Collection of customized MatLab functions for the implementation of


Interval Arithmetic operations.
function R = intSum(X,Y)
R=[X(1)+Y(1) X(2)+Y(2)];
function R = intDiff(X,Y)
R=[X(1)-Y(2) -X(2)+Y(1)];
function R = intMult(x,y)
T1=x(1)*y(1); T2=x(1)*y(2); T3=x(2)*y(1); T4=x(2)*y(2);
R=[min([T1,T2,T3,T4]) max([T1, T2, T3, T4])];
function R = intDivi(x,y)
T1=x(1)/y(1); T2=x(1)/y(2); T3=x(2)/y(1); T4=x(2)/y(2);
R=[min([T1,T2,T3,T4]) max([T1, T2, T3, T4])];
function r=intswap(X)
r=[X(2) X(1)];
function R = intReci(X)
R=[1/X(2) 1/X(1)];
function R = intSqrt(x)
R=[x(1)^.5 x(2)^.5];
function R = intSquare(x)
R=[x(1)*x(1) x(2)*x(2)];
function R = intDI(X,Y,Z,a)
R=intDivi(intSum(X,a*Z),Y);
function R = intmud(X,Y)
R=intDivi(X,Y);
function R = intmuh(X,Y)
R=intDivi(X,Y);
function R = intmuu(X,Y)
R=intDivi(X,Y);

123

Algorithm 3. Octave program to compute for the damage indices using interval
arithmetic operations
clc
clear all
beta=0.15;
Xdd=[0.001521 0.003042 0.004563 0.006084 0.007605 0.009126 0.010647 0.012168
0.013689 0.01521];
Ydd=[0.001859 0.003718 0.005577 0.007436 0.009295 0.011154 0.013013 0.014872
0.016731 0.01859];
Xeh=[0.451207453 1.804829813 4.060867078 7.21931925 11.28018633 16.24346831
22.1091652 28.877277 36.5478037 45.12074531];
Yeh=[0.551475776 2.205903104 4.963281985 8.823612417 13.7868944 19.85312794
27.02231303 35.29444967 44.66953786 55.14757761];
Xds=[0.1048311 0.1048311 0.1048311 0.1048311 0.1048311 0.1048311 0.1048311
0.1048311 0.1048311 0.1048311];
Yds=[0.1281269 0.1281269 0.1281269 0.1281269 0.1281269 0.1281269 0.1281269
0.1281269 0.1281269 0.1281269];
Xdy=[0.0148311 0.0148311 0.0148311 0.0148311 0.0148311 0.0148311 0.0148311
0.0148311 0.0148311 0.0148311];
Ydy=[0.0181269 0.0181269 0.0181269 0.0181269 0.0181269 0.0181269 0.0181269
0.0181269 0.0181269 0.0181269];
Xes=[78.99366436 78.99366436 78.99366436 78.99366436 78.99366436 78.99366436
78.99366436 78.99366436 78.99366436 78.99366436];
Yes=[96.547812 96.547812 96.547812 96.547812 96.547812 96.547812 96.547812
96.547812 96.547812 96.547812];
for i=1:10
ddmax=[Xdd(i)
Ydd(i)];
dsmax=[Xds(i)
Yds(i)];
dy=[Xdy(i) Ydy(i)];
Ee=[Xes(i)
Yes(i)];
Eh=[Xeh(i)
Yeh(i)];
md=intmud(ddmax,dy);
mu=intmuu(dsmax,dy);
mh=intmuh(Eh,Ee);
DI=intDI(md,mu,mh,beta);
fprintf('%f %f %f %f %f %f %f %f\n',md,mu,mh,DI);
end

Output
0.083908
0.167817
0.251725
0.335634
0.419542
0.503451
0.587359
0.671268
0.755176
0.839084

0.125345
0.250689
0.376034
0.501379
0.626724
0.752068
0.877413
1.002758
1.128102
1.253447

5.783179
5.783179
5.783179
5.783179
5.783179
5.783179
5.783179
5.783179
5.783179
5.783179

8.639069
8.639069
8.639069
8.639069
8.639069
8.639069
8.639069
8.639069
8.639069
8.639069

0.004673
0.018694
0.042061
0.074775
0.116835
0.168243
0.228997
0.299098
0.378546
0.467341

124

0.006981
0.027925
0.062831
0.111700
0.174532
0.251326
0.342082
0.446801
0.565483
0.698127

0.009794
0.019750
0.029868
0.040149
0.050592
0.061197
0.071965
0.082895
0.093987
0.105241

0.021855
0.044072
0.066652
0.089593
0.112897
0.136563
0.160591
0.184981
0.209733
0.234848

You might also like