DigitalCommons@University of Nebraska  Lincoln
Civil Engineering Theses, Dissertations, and
Student Research
Civil Engineering
1212009
Live Load Models for Long Span Bridges
Marta Lutomirska
mwolinska2@unl.edu
This Article is brought to you for free and open access by the Civil Engineering at DigitalCommons@University of Nebraska  Lincoln. It has been
accepted for inclusion in Civil Engineering Theses, Dissertations, and Student Research by an authorized administrator of
DigitalCommons@University of Nebraska  Lincoln. For more information, please contact proyster@unl.edu.
Lutomirska, Marta, "Live Load Models for Long Span Bridges" (2009). Civil Engineering Theses, Dissertations, and Student Research.
Paper 1.
http://digitalcommons.unl.edu/civilengdiss/1
LIVE LOAD MODELS
FOR LONG SPAN BRIDGES
by
Marta Lutomirska
A DISSERTATION
Presented to the Faculty of
The Graduate College at the University of Nebraska
In Partial Fulfillment of the Requirements
For the Degree of Doctor of Philosophy
Major: Engineering
Under the Supervision of Professor Andrzej S. Nowak
Lincoln, Nebraska
December, 2009
LIVE LOAD MODELS
FOR LONG SPAN BRIDGES
Marta Lutomirska, PhD
University of Nebraska, 2009
Advisor: Andrzej S. Nowak
In the doctoral dissertation a live load model for long span structures was derived.
The live load model is valid for spans between 600 ft and 5000 ft and it is intended to
reflect current traffic patterns, quantities of trucks and their weights. The live load models
available were developed for short and medium span bridges. Those models were not
appropriate for long span bridges due to different types of structure and critical traffic
patterns. Live load on long spans depends on traffic mix. One heavily overloaded truck
does not have significant influence. Moreover, the continuous increase in the number of
the trucks, their weights, and high percentage of overweight trucks led to a search for the
newest traffic data. The database includes variety of sites within many different states. A
numerical procedure was developed to process the database and simulate traffic jam
situations. From the simulation the values of uniformly distributed load were derived.
Trucks were kept in actual order, as recorded in the WIM surveys. Results of the
simulations were plotted as a cumulative distribution function of uniformly distributed
load for considered span lengths. For longer spans, uniformly distributed load decreases
and is closer to the mean value. The bias factors were calculated for the heaviest 75year
combination of vehicles. The 75year uniformly distributed loads were derived from
extrapolated distributions. It was stated that for most of the bridges current live load HL
93 is appropriate. It was also noticed that some bridges, characterized by high ADTT and
increased percentage of overloaded loaded vehicles, require special attention and
application of increased design live load. The developed live load model is recommended
to be taken into consideration in the bridge design code.
iii
© Marta Lutomirska
All rights reserved
2009
iv
DEDICATION
To My Family
v
ACKNOWLEDGMENTS
I wish to express my deepest gratitude to Professor Andrzej S. Nowak, my
academic advisor and chairperson of my dissertation committee, for his support and
guidance throughout my graduate study at the University of Nebraska. My appreciations
also go to a member of my doctoral committee, Dr. George Morcous, who advised and
encouraged me in my study. Special thanks go to Dr. Maria M. Szerszen and Dr.
Elizabeth G. Jones, members of my reading committee for their effort and time spent on
reviewing my dissertation and offering helpful suggestions.
The Department of Civil Engineering at the University of NebraskaLincoln is
acknowledged for the study and research opportunities they provided.
I am deeply grateful to my husband, Tomasz, my parents, and whole family for
their great support and encouragement.
vi
TABLE OF CONTENTS
ABSTRACT .................................................................. Error! Bookmark not defined.
DEDICATION .............................................................................................................. iv
ACKNOWLEDGMENTS ...............................................................................................v
LIST OF FIGURES ..................................................................................................... viii
LIST OF TABLES ....................................................................................................... xii
CHAPTER 1 INTRODUCTION......................................................................................1
1.1. PROBLEM STATEMENT ......................................................................................1
1.2. OBJECTIVE AND BENEFITS OF THE STUDY ..........................................................2
1.3. ORGANIZATION OF THE DISSERTATION ..............................................................4
1.4. PRIOR INVESTIGATIONS .....................................................................................5
CHAPTER 2 LIVE LOAD IN CURRENT DESIGN CODES ........................................ 11
2.1. INTRODUCTION ............................................................................................... 11
2.2. INTERNATIONAL PROVISIONS FOR LIVE LOADING ............................................ 12
2.3. PROVISIONS FOR DYNAMIC LOAD FACTOR ....................................................... 20
2.4. PROVISIONS FOR MULTILANE REDUCTION FACTORS ........................................ 21
2.5. COMPARISON OF EQUIVALENT UNIFORMLY DISTRIBUTED LOADS..................... 23
CHAPTER 3 STRUCTURAL RELIABILITY PROCEDURES .................................... 29
3.1. INTRODUCTION ............................................................................................... 29
3.2. STANDARD VARIABLES AND PROBABILITY DISTRIBUTIONS ............................... 30
3.3. LIMIT STATE FUNCTION ................................................................................... 33
3.4. RELIABILITY INDEX......................................................................................... 35
3.5. MONTE CARLO METHOD SIMULATION TECHNIQUE .......................................... 37
3.6. NORMAL PROBABILITY PAPER ......................................................................... 39
CHAPTER 4 TRAFFIC DATA ..................................................................................... 42
4.1. REGULATIONS OF TRUCK TYPES, TRUCK SIZES AND WEIGHT LIMITS ................ 42
4.2. DATA COLLECTION METHODOLOGY (WIM) .................................................... 49
4.3. WEIGH IN MOTION DATABASE ........................................................................ 51
CHAPTER 5 DEVELOPMENT OF LIVE LOAD MODEL .......................................... 61
5.1. INTRODUCTION ............................................................................................... 61
vii
5.2. MODEL BASED ON AVERAGE 5AXLE TRUCK .................................................. 61
5.3. MODEL BASED ON LEGAL LOAD TRUCKS ........................................................ 65
5.4. MODEL BASED ON TRAFFIC JAM SIMULATION USING WIM DATA .................... 66
CHAPTER 6 MULTIPLE PRESENCE ......................................................................... 77
6.1. INTRODUCTION ............................................................................................... 77
6.2. STUDIES ON PRESENCE OF MULTIPLE TRUCKS ................................................. 78
6.3. MULTIPLE PRESENCE OF TRUCKS BASED ON THE VIDEO FILES OF TRAFFIC ....... 81
6.4. APPROACHES TO MULTILANE REDUCTION FACTORS ......................................... 86
6.5. CONCLUSIONS ................................................................................................. 88
CHAPTER 7 DYNAMIC FACTOR .............................................................................. 89
7.1. INTRODUCTION ............................................................................................... 89
7.2. STUDIES ON PARAMETERS AFFECTING DYNAMIC BRIDGE RESPONSE ................ 90
7.1. BRIDGEVEHICLE INTERACTION MODEL AND DERIVATION OF DYNAMIC FACTOR
....................................................................................................................... 98
7.2. CONCLUSIONS ............................................................................................... 105
CHAPTER 8 RELIABILTY ANALYSIS OF SUSPENSION BRIDGE....................... 106
8.1. RELIABILITY ANALYSIS PROCEDURE ............................................................. 106
8.2. SELECTION OF REPRESENTATIVE STRUCTURE, ELEMENT AND LIMIT STATE
FUNCTION ................................................................................................................ 107
8.3. NOMINAL RESISTANCE .................................................................................. 108
8.4. LOAD MODEL ............................................................................................... 113
8.5. RELIABILITY RESISTANCE MODELS ............................................................... 120
8.6. LOAD MODEL ............................................................................................... 126
8.7. RELIABILITY ANALYSIS ................................................................................. 126
CHAPTER 9 SUMMARY & CONCLUSIONS ........................................................... 129
CHAPTER 10 RECCOMENDATIONS ...................................................................... 132
REFERENCES ............................................................................................................ 134
APPENDIX A CDF OF UDL FOR ALL TRUCK COMBINATIONS ........................ 141
APPENDIX B CDF OF MAXIMUM DAILY UDL ................................................... 153
APPENDIX C CDF OF maximum weekly UDL ......................................................... 161
viii
LIST OF FIGURES
Figure 2.1. HL93 Live Loading in AASHTO LRFD Code [2007]. Truck and Lane Load.
...................................................................................................................................... 12
Figure 2.2. HL93 Live Loading in AASHTO LRFD Code [2007]. Tandem and Lane
Load. ............................................................................................................................. 12
Figure 2.3. HL93 Live Loading in AASHTO LRFD Code [2007]. Alternative Load for
Negative Moment between points of contraflexure and reaction at interior piers. ........... 13
Figure 2.4. OHBD Live Loading [1991]. OHBD Truck. ................................................ 13
Figure 2.5. OHBD Live Loading [1991]. OHBD Truck and Lane Load. ........................ 14
Figure 2.6. CAN/CSAS600 Live Loading [2000]. CLW Truck .................................. 15
Figure 2.7. CAN/CSAS600 Live Loading [2000]. CLW Lane Load ........................... 15
Figure 2.8. BS 5400 Live Loading curve HA UDL [2006] ............................................. 16
Figure 2.9. Dimensions of HB vehicle. .......................................................................... 17
Figure 2.10. Eurocode 1 [2002]. Load Model 1 .............................................................. 18
Figure 2.11. ASCE Loading on Log Scale ..................................................................... 19
Figure 2.12. Equivalent Unfactored Loads, w/o IM, w/o multilane factors. .................... 24
Figure 2.13. Equivalent Factored Loads, w/o IM, w/o multilane factors. ........................ 25
Figure 2.14. Equivalent Unfactored Loads, with IM, w/o multilane factors. ................... 25
Figure 2.15. Equivalent Factored Loads, with IM, w/o multilane factors. ....................... 26
Figure 2.16. Equivalent Factored Loads, with IM, with multilane factors for 4 traffic
lanes. ............................................................................................................................. 26
Figure 3.1 PDF and CDF of a normal random variable .................................................. 32
Figure 3.2 PDF’s of load, resistance, and safety margin ................................................. 34
Figure 3.3 Reliability index defined as the shortest distance in the space of reduced
variables ........................................................................................................................ 36
Figure 3.4 Normal Distribution Function on the Normal Probability Paper. ................... 41
Figure 4.1. FHWA 13category scheme ......................................................................... 43
Figure 4.2. Longer Combination Vehicles (LCV’s) ........................................................ 45
Figure 4.3. States allowing various Longer Combination Vehicles ................................. 45
Figure 4.4. New Bridge Formula  regulation of vehicles' length and weight .................. 46
ix
Figure 4.5. Number of trucks by weight (in thousands of trucks). Transportation statistics
annual report, December 2006. ...................................................................................... 47
Figure 4.6. Freight Tonnage Moved by Truck (FHWA) ................................................. 48
Figure 4.7. Time variation of total truck weight statistic. (Gindy, M., Nassif, H.H., 2006)
...................................................................................................................................... 49
Figure 4.8. WIM data collection .................................................................................... 51
Figure 4.9. CDF's of GVW by axles Oregon I5 Woodburn ........................................... 58
Figure 4.10. CDF's of GVW by axles Oregon I84 Emigrant Hill................................... 59
Figure 4.11. CDF's of GVW by axles NY I495 EB ....................................................... 60
Figure 5.1. Critical loading. Traffic jam scenario. .......................................................... 62
Figure 5.2. CDF’s of GVW for 5axle trucks. New York WIM Data. ............................. 62
Figure 5.3. CDF of GVW for 5axle and 11axle trucks (Nowak, A.S., Laman, J. and
Nassif, H., 1994)............................................................................................................ 63
Figure 5.4. Percentage of vehicles by number of axles. FHWA WIM Data. ................... 63
Figure 5.5. Percentage of vehicles by number of axles. (Kim, SJ., Sokolik, A.F., and
Nowak, A.S., 1997) ....................................................................................................... 64
Figure 5.6. CDF’s of GVW for all types of vehicles in Oregon. ..................................... 64
Figure 5.7. AASHTO LRFD legal load trucks, Type 33 Units. ..................................... 65
Figure 5.8. Simulation of trucks moving throughout span length .................................... 66
Figure 5.9. Clearance  Gap and Spacing  Headway Concepts ...................................... 67
Figure 5.10. Interstate semitrailer WB20 (AASHTO Geometric Design of Highways and
Streets) .......................................................................................................................... 67
Figure 5.11. HL93 proposed for long span bridges ....................................................... 69
Figure 5.12. Mean value of uniformly distributed load ................................................... 72
Figure 5.13. Daily maximum mean value of uniformly distributed load ......................... 72
Figure 5.14. Weekly maximum mean value of uniformly distributed load ...................... 73
Figure 5.15. Bias (mean max 75 year to nominal value of UDL) .................................... 73
Figure 5.16. Bias (mean max 75 year to nominal value of UDL) .................................... 74
Figure 5.17. Bias (mean max 75 year to nominal value of UDL) assumed designed UDL
of 0.85 k/ft ..................................................................................................................... 74
Figure 5.18. Bias for heavily loaded localizations, assumed designed UDL of 1.25 k/ft . 75
x
Figure 5.19. Coefficient of variation of daily maximum uniformly distributed load ....... 75
Figure 5.20. Coefficient of variation of weekly maximum uniformly distributed load .... 76
Figure 5.21. Proposed coefficient of variation of uniformly distributed load .................. 76
Figure 6.1. Traffic loading pattern used for multiple truck presence statistics. ................ 79
Figure 6.2. Variation of multiple truck presence statistics with respect to truck volume.
Gindy and Nassif (2006). ............................................................................................... 80
Figure 6.3. Variation of multiple truck presence statistics with respect to bridge span
length. Gindy and Nassif (2006). ................................................................................... 81
Figure 6.4. Video 10, time: 00:00:58.............................................................................. 83
Figure 6.5. Video 1, time: 00:05:28 ............................................................................... 84
Figure 6.6. Video 1, time: 00:18:36 ............................................................................... 84
Figure 6.7. Video 2, time: 00:00:15 ............................................................................... 85
Figure 6.8. Video 8, time: 00:00:16 ............................................................................... 86
Figure 6.9. Multilane load in design codes AASHTO LRFD Code (2007), OHBDC
(1991), CAN/CSAS600 [2000], and ASCE (1981). ..................................................... 87
Figure 6.10. Multilane load in Eurocode 1. .................................................................... 87
Figure 6.11. Multilane load in actual observation. .......................................................... 87
Figure 7.1. Distribution of fundamental bridge frequencies (Cantieni 1984)....................... 90
Figure 7.2. Fundamental frequency versus span length (Cantieni 1984) ............................. 91
Figure 7.3. Fundamental frequency versus span length (Paultre 1992) ............................... 91
Figure 7.4. Impact factor versus vehicle speed and road surface condition (Wang, Shahawy,
and Huang 1993) ............................................................................................................ 93
Figure 7.5. Effects of vehicle suspension on the measured bridge response (Biggs and Suer
1955) ............................................................................................................................. 95
Figure 7.6. Impact versus span length (Fleming and Romualdi 1961) ................................ 96
Figure 7.7. Impact versus span length (Cantieni 1984) ...................................................... 96
Figure 7.8. Meshed model of the bridge........................................................................... 99
Figure 7.9. Truck model in FEM ..................................................................................... 99
Figure 7.10. Vehiclebridge interacting force. .............................................................. 101
Figure 7.11. Force due to moving truck versus time. Plot from ABAQUS. ...................... 101
Figure 7.12. Bending modes ......................................................................................... 102
xi
Figure 7.13. Torsion modes. ......................................................................................... 102
Figure 7.14. Maximal deflection due to moving truck .................................................... 103
Figure 7.15. Deflection due to a truck moving 40mil/hr versus time. ............................... 104
Figure 7.16. Deflection due to a truck moving at crawling speed versus time. .................. 104
Figure 8.1 Distribution of strains for the pure axial loading .......................................... 109
Figure 8.2 Distribution of strains distribution for the balance failure point B, the end of
the compression control zone ....................................................................................... 109
Figure 8.3 End of compression block of concrete in the bottom flange ......................... 110
Figure 8.4 End of compression block of concrete in the web. ....................................... 111
Figure 8.5 End of compression block of concrete in the top flange. .............................. 111
Figure 8.6 Stress  Strain Relationship for Reinforcing Steel ........................................ 112
Figure 8.7. Geometry of Cooper River Bridge ............................................................. 114
Figure 8.8. Load combinations ...................................................................................... 115
Figure 8.9. Envelope of bending moments for bridge tower for w=0.64k/ft .................. 116
Figure 8.10. Envelope of bending moments for bridge tower for w=0.80 k/ft ............... 117
Figure 8.11. Envelope of bending moments for bridge tower for w=1.00 k/ft ............... 118
Figure 8.12. Envelope of bending moments for bridge tower for w=1.20 k/ft ............... 119
Figure 8.13 Bias factor for compressive strength of concrete ....................................... 121
Figure 8.14 Coefficient of variation for compressive strength of concrete .................... 122
Figure 8.15 CDF’s of yield strength for Reinforcing Steel Bars, Grade 60 ksi .............. 123
Figure 8.16 Recommended material parameters for reinforcing steel bars, Grade 60 ksi
.................................................................................................................................... 124
Figure 8.17 Statistical Parameters of Resistance .......................................................... 125
Figure 8.18 Force and Moment results on the bridge tower for different live loads....... 127
Figure 8.19 Reliability indexes due to different live load ............................................. 128
xii
LIST OF TABLES
Table 2.1. Number of design lanes vs. road width in OHBDC [1991] ............................ 14
Table 2.2. Number of design lanes vs. road width in CAN/CSAS600 .......................... 15
Table 2.3. Characteristic values of load for successive road lanes .................................. 18
Table 2.4. Dynamic allowance in AASHTO LRFD [2007] ............................................ 20
Table 2.5. Dynamic allowance in CAN/CSAS600 ....................................................... 21
Table 2.6. Comparison of Multilane Reduction Factors.................................................. 22
Table 2.7. Multilane Reduction Factors for BS 5400 ...................................................... 22
Table 2.8. Values of Equivalent Unfactored Loads, w/o IM, w/o multilane factors......... 27
Table 2.9. Values of Equivalent Factored Loads, w/o IM, w/o multilane factors. ........... 27
Table 2.10. Values of Equivalent Unfactored Loads, with IM, w/o multilane factors...... 27
Table 2.11. Values of Equivalent Factored Loads, with IM, w/o multilane factors. ........ 28
Table 2.12. Values of Equivalent Factored Loads, with multilane factors for 4 traffic
lanes. ............................................................................................................................. 28
Table 3.1. Relationship between vertical scale on Normal Probability Paper and
Probability ..................................................................................................................... 40
Table 4.1. Conversion chart for vehicles’ class and number of axles .............................. 44
Table 4.2. Summary of WIM Data ................................................................................. 53
Table 4.3. Vehicles by axle in Oregon ........................................................................... 54
Table 4.4. Vehicles by traffic lane in Oregon ................................................................. 54
Table 4.5. Vehicles by axle in Florida ............................................................................ 55
Table 4.6. Vehicles by traffic lane in Florida ................................................................. 55
Table 4.7. Vehicles by axle in Indiana ........................................................................... 56
Table 4.8. Vehicles by traffic lane in Indiana ................................................................. 56
Table 4.9. Vehicles by axle in New York ....................................................................... 57
Table 4.10. Vehicles by traffic lane in New York........................................................... 57
Table 5.1. Statistical parameter for proposed uniformly distributes live load .................. 69
Table 5.2. Summary of simulated data ........................................................................... 70
Table 5.3. Heaviest truck combinations for 600 ft on I495 WB ..................................... 71
Table 6.1. Presence of multiple trucks and their location on the road lanes .................... 78
xiii
Table 8.1 Recommended Statistical Parameters for Compressive Strength, '
c
f
(Nowak
A.S. et al., 2008) .......................................................................................................... 122
Table 8.2 Statistical Parameters of Fabrication Factor. ................................................. 124
1
CHAPTER 1
INTRODUCTION
1.1. PROBLEM STATEMENT
The live load models available were developed for short and medium span
bridges. This doctoral dissertation deals with the development of a live load model for
long span structures. The developed live load model is valid for spans between 600 ft and
5000 ft. In contrast to short and medium spans, a long span live load must include the
possibility of multiple trucks being present.
The continuous increase in the number of the trucks and their weights led to a
review of traffic data for live load. Observing traffic statistics helps to realize the rate of
those changes, their importance, and to draw some conclusions regarding design. In the
last 30 years, the number of the vehicle miles logged annually on American highways has
increased 225%, with heavy truck traffic increasing 550%. Some percentage of trucks
runs overweight, particularly if it is to their economic advantage. Therefore, a new live
load model for long span bridges had to be developed and it had to be based on the
newest traffic data obtained from highway and bridge administrators.
During the AASHTO LRD calibration, the live load model for short and medium
span bridges was developed based on a set of truck weight and load effect statistics that
were presumed to be valid for any typical bridge site in the U.S. The live load model may
not represent the actual loading conditions at a particular bridge site or bridges in a state.
Nowadays, several states are using WeighInMotion (WIM) systems to collect vast
2
amounts of truck weight and traffic data that can be used to obtain sitespecific and state
specific live load models for bridge design and load capacity evaluation. This could allow
individual states to adjust the AASHTO live load factors to take into consideration the
particular truck traffic conditions throughout a state, a region, or for a particular route.
Sitespecific or statespecific live load models may be developed based on actual truck
weight and traffic data collected at the site or within the state. Traffic varies for different
sites within each state. As a result, site specific models depending on average daily truck
traffic and participation of heavily loaded vehicles seem to be more practical.
Since early publications by the American Association of State Highway Officials
(AASHO), live load was modeled as an HS20 truck. As an addition to truck load, the
uniformly distributed load of 0.64 kip/ft was introduced in 1944. Since then the original
definition of HS20 has been changed. The concentrated load was substituted with three
axial forces representing a truck. In contrast, the uniformly distributed load has never
been updated. It is still used in the current AASHTO LRFD Code as it was in 1944. The
derivation of uniformly distributed load is not clear. To amend this, a new approach to
model uniformly distributed load had to be developed and new value of uniformly
distributed load had to be proposed. Current multilane reduction factors and dynamic
allowance also may not be appropriate for long span bridges. Review of those topics was
necessary.
1.2. OBJECTIVE AND BENEFITS OF THE STUDY
The objective in this study was to develop a live load model for long span bridges.
The model is valid for spans between 600 ft and 5000 ft. It is intended to reflect current
traffic patterns, quantities of trucks and their weights. The newest available traffic
database from a variety of sites within many different states is used. Based on the
analysis of traffic records (weighinmotion and videos) the design live load is developed
and recommended to be taken into consideration in the bridge design code. Reliability
analysis is used to verify the developed live load model.
In accordance with the stated objective, the first stage was to study previous
research and current international codes’ provisions on the topic. The second stage of the
3
research was the collection of state of the art traffic data from highway and bridge
administrators. The data obtained had to be analyzed and filtered out from erroneous
readings of measurement instruments. Then a new uniformly distributed load was
derived. The value of the new live load is based on three models: an average 5axle truck,
legal load trucks and simulation of a traffic jam using WIM data. Such an extensive
actual weigh in motion database has never been used in the derivation of live load for
long span bridges. Most of the previous studies were based on measurements from
limited numbers of sites within one state. The magnitude of the database obtained for the
scope of this research has to be underlined. A derivation of uniformly distributed load
from WIM data required developing a numerical procedure of calculation to process the
extensive database. Cumulative distribution functions were plotted for all data, as well as
for maximum daily and maximum weekly uniformly distributed load. New uniformly
distributed load was proposed. Statistical parameters for live load (bias and coefficient of
variation) are derived. Relationship between site characteristics (ADTT, percentage of
overloaded loaded vehicles) and calculated values of uniformly distributed loads were
studied. The problems of multilane reduction factors and the dynamic factor were also
discussed.
The final step of this dissertation was reliability analysis. Reliability analysis was
performed in order to assess how the increase in live load influences reliability indexes.
An exemplary suspension bridge, the bridge component and a limit state function that are
the most influenced by live load were selected. The calculations were performed for the
current AASHTO LRFD design live load and increased load values obtained from real
traffic data. For the scope of this study, new statistical parameters for uniformly
distributed load were used and statistical parameters of resistance were derived based on
the newest material, fabrication and professional factors.
The outcome of this research is the recommendation of a live load model for long
span bridges. There is a recommended value of uniformly distributed load for bridges
carrying low and average ADTT. In addition, there is a recommendation for an increase
of the live load model for the long span bridges in heavily loaded urban and industrial
areas.
4
1.3. ORGANIZATION OF THE DISSERTATION
Chapter 1 of this dissertation is an introduction to the research conducted. It
presents a problem statement, objectives and the scope of the study, as well as the
benefits and limitations of this research. The chapter also includes a review of prior
research on the topic.
Chapter 2 reviews current international bridge design codes regarding live load,
dynamic load and multiple presence factors provisions. Uniformly distributed load for
wide range of spans is calculated and compared.
Chapter 3 presents the principals of the reliability theory that were applied in this
study. Definitions of standard variables, probability distributions, limit state functions
and reliability index are introduced. Methods of use of the normal probability paper and
simulation techniques are described.
Chapter 4 describes the study of traffic data, regulations of truck types, truck sizes
and weight limits, as well as the weightinmotion technology and database to be used in
this dissertation.
Chapter 5 describes development of the live load model. A model based on an
average 5axle truck, a model based on legal load trucks, and a model based on traffic
jam simulation using WIM data are presented. New values of uniformly distributed load
are derived and proposed to be applied in the code. New statistical parameters for live
load are calculated.
In the Chapter 6, the problem of the presence of multiple trucks on a bridge is
discussed. A short review of current studies, analysis of video recordings of traffic jam
situations, and discussion on different approaches to the problem are presented.
Chapter 7 presents the problem of dynamic factor. Parameters affecting bridge
dynamic response are discussed. Exemplary bridgevehicle interaction is modeled and the
dynamic factor is derived.
In the Chapter 8, the reliability procedures for a long span bridge are developed.
The analysis is performed on a bridge tower of the selected suspension bridge. Reliability
indexes for bridge loaded increasing values of live load are calculated.
5
Chapter 9 presents the summary and conclusions of research performed for the
scope of this dissertation. As well, recommendations are specified.
1.4. PRIOR INVESTIGATIONS
1.4.1. Prior Investigations on Live Loading on Short and Medium Span Bridges
Live load models for short and medium span bridges were of interest to many
researchers. Most of the studies performed on live load models were based on truck data
obtained within programs carried out by the Ontario Ministry of Transportation since the
early 1970s. This was the vastest database available until now.
Nowak and Hong (1991) formulated a procedure to calculate maximum moments
and shears for various time periods. The maximum load effects for various time periods
from one day to 75 years were derived from extrapolated distributions. Single and two
lane bridges are considered. For one lane traffic a single truck governs for shorter spans,
and two following trucks govern for longer spans. For two lanes of traffic, the maximum
effect is obtained for two trucks with fully correlated weights, travelling sidebyside. It
has also been concluded that the bias factor (ratio of the mean to nominal value) is larger
for smaller spans.
Kim, Sokolik, and Nowak (1997) studied actual truck loads on selected bridges in
the Detroit area. The measurements were taken by using a weight in motion system. It
was observed that truck loads are strongly site specific. The observed truck weights were
often heavier than legal limits. The maximum observed truck weights were up to 250
kips, causing maximum moments two times larger than AASHTO load and resistance
factor design values. Gindy and Nassif (2006) formulated a similar conclusion based on
data from New Jersey. It was found out that maximum gross vehicle weight reaches a
value of 225 kips and it shows a steady increase at an annual growth rate is 1.2%.
Nowak, Laman, and Nassif (1994) published a research report on the effect of
truck loading on bridges. The WIM measurements were taken on seven bridges in
Michigan. The researchers developed procedures for evaluation of live load spectra on
steel girder bridges with regard to fatigue. The deteriorating capacity of bridge was
6
evaluated as a function of the rate of corrosion. It has been proved that WIM
measurements show the unbiased truck weights, which are 3050 percent larger than
extreme values obtained at weight stations. The WIM data is unbiased because the drivers
are not aware of the measurements and they do not make an effort to avoid the scales.
The WIM measurements from Michigan have also been used to study dynamic load,
Nassif and Nowak (1995). It was found out that the dynamic load factor decreases with
increased static loads, and that larger values of DLF are observed in exterior girders due
to relatively smaller static load effect. Derivation of the dynamic load model is described
by Hwang and Nowak (1991).
1.4.2. Prior Investigations on Live Loading on Long Span Bridges
The most widely known researcher in the field of live loading on long span brides
is Peter G. Buckland (1978, 1980, and 1991). He concluded that traffic loading on long
span bridges can be accurately represented in the traditional manner, by one set of
uniform and concentrated loads. One of his findings was that uniform load per foot
reduces as the load length is increased. However, unlike many other studies he found out
that concentrated load increases as the loaded length increases. Four uniform loading
curves were developed for different loading cases. The load cases were distinguished
depending on the percentage of “heavy vehicles”: 2.4, 7.4, 30.0, or 100 percent, where
“heavy vehicles” are defined as trucks and buses over 12000 lb. These loading curves
were recommended by the ASCE Committee as vehicle loading of longspan bridges, in
1981. They are known unofficially as the ASCE Loading. However, they have never been
applied into the design codes.
Peter G. Buckland had also made a valuable observation regarding several loaded
lanes. He stated that if a single lane has a certain load on it, than the additional lanes
would increase the load in the lane closer to the curb, as trucks gravitate towards it.
However, load in the additional lanes can be reduced.
In the paper by Buckland (1991) the comparison of North American and British
live loads on longspan bridges is presented. The loads are compared as equivalent
uniformly distributed loads, calculated as an equivalent shear and bending moment for
7
simply supported beams. This approach can be successfully used for short and medium
span bridges. However, its application to long spans can be questioned, since long span
bridges cannot be constructed as simply supported beams. This method of deriving the
equivalent load can be used exclusively for comparison of codes.
1.4.3. Prior Investigations on Structural Reliability
The theory of structural reliability have been investigated and described by many
researchers. Several books and publications provide available knowledge regarding
reliability theory, for instance, ThorfChristtensen and Baker (1982), Ang and Tang
(1984), Madsen, Krenk and Lind (1986); ThorfChristtensen and Murotsu (1986), Ayyub
and McCuen (1997), Murzewski (1989), Nowak and Collins (2000) and Wolinski and
Wrobel (2001). The application of reliability theory has resulted in the improvement of
structural design in terms of safety, serviceability and durability. However, it was not
until the late 70’s when safety factors based on load and resistance uncertainties were
proposed for introduction into the codes. In the United States, it was the building design
code (Galambos and Ravindra 1978, and Ellingwood 1980, 1982), and in Canada, it was
the Ontario highway bridge design code (Nowak and Lind 1979). Since then, reliability
techniques have been increasingly used in modern design codes. Nowadays many
researchers keep working on further development of new methods of structural reliability
analysis, among them A.H.S. Ang, O. Ditlevesen, R.E. Melchers, H. Nielsen Faber,
A.S. Nowak, R. Rackwitz, and P. ThoftChristiansen.
For many years the random nature of various parameters influencing structural
safety has been of interest to engineers. Until they gathered more knowledge about the
laws of nature, they used to assure structural safety through ‘trial and error’ and intuition.
Mathematical theories available nowadays describe material and structural behavior
sufficiently enough to give a rational basis for structural safety evaluations (Nowak and
Collins, 2000). Early publications that quantified and presented a mathematical
formulation of structural safety problems were published by Mayer (1926) and
Wierzbicki (1936). They recognized that load and resistance parameters have random
characteristics, and that each structure has a finite and limited probability of failure. Their
8
concept of a structural reliability problem has been subsequently adopted in the
precursory publication for that field by Freudenthal (1956). In the 1960s, a new trend in
using probabilistic concepts in the analysis of limit capacity and structural resistance was
developed. It was first presented by researchers Augusti and Baratta (1973, 1973).
Subsequently, important work was done by Corotis and Nafday (1989) analyzing the
limit capacity of beamcolumn frame structures using the principles of conditional
probability.
The extensive development of practical tools and efficient methods for evaluating
the probability of structural failure has been made in the last 30 decades. Initially, the
probability of failure was defined by multidimensional integral functions of distributions
and it was cumbersome to evaluate. Pioneering studies on the first practical application of
reliability analysis were performed by Cornell and Lind in the late 1960s and early 1970s.
Their approach estimated the limit state function at mean values of random parameters
and used a linearized limit state function. A milestone was the estimation of the
probability of failure proposed by Hasofer and Lind (1974). The simplified procedure
involved a nonlinear mathematical programming problem with boundary conditions (an
estimated limit state for all variables and a defined probability in the so called “design
point”). The extension of the Hasofer and Lind approach and the transformation of
uncorrelated random variables of various distributions into standardized normal
distributions were proposed by Rackwitz and Fiessler (1978). The developed numerical
procedure of the design point estimation used to be called the RackwitzFiessler
procedure. Hohenbichler and Rackwitz (1988) used the Rosenblatt (1952) transformation
procedure for the transformation of dependent (correlated) random variables from and
into standardized form, which is currently one of the major tools used in modern
reliability analysis. Another commonly used transformation is Nataf’s transformation,
which was presented in work by Kiureghian and Liu (1986). Commonly used methods of
reliability analysis are based on the approximation of the limit state function at the design
point using first or second order functions (FORM and SORM). Advanced SORM have
been elaborated by researchers such as Fiessler, Neumann and Rackwitz (1979), Breitung
(1984), Nowak and Collins (2000). Adhikari (2004) systemized and published all of the
earlier proposed SORM procedures. Simulation techniques are another approach to
9
estimating probability of failure. The most popular is the Monte Carlo Method simulation
technique (ThoftChristensen and Baker, 1982; Hart, 1982). Determination of the
probability of structural failure with the use of simulation techniques has limited
accuracy, and a huge number of numerical simulations are required to achieve a high
accuracy of results. This method becomes very useful and an especially practical tool in
cases where physical testing is expensive.
The structural system reliability is a field of interest for many researchers. Bridges
usually consist of a combination of series and parallel systems. Identification of collapse
mode and degree of correlation between members is very difficult or often even
impossible to evaluate. Moses (1982) proposed incremental load approach and suggested
a procedure for identifying collapse mode for both ductile and brittle components. The
identification of collapse mode was also discussed by Rashedi and Moses (1988).
Reliability models applied to bridge evaluation were addressed by Nowak and
Tharmabala (1988). Moses and Verma (1987) used a load and resistance approach to
evaluate the strength of bridges with reliability principles. Tantawi (1986) developed a
grid nonlinear analysis program to calculate the momentcarrying capacity of a bridge.
Than Zhou (1987) developed an integration sampling technique to calculate the system
reliability of a bridge. Both Tantawi and Zhou, found that bridge system reliability is
higher than girder reliability.
Practical procedures for system reliability analysis were suggested by Nowak and
Zhou (1990), Zhou and Nowak (1990), and Tabsh and Nowak (1991). The procedures
assumed that the ultimate load carrying capacity is equal to the weight of a truck which
causes a collapse. A collapse was defined as an excessive, nonacceptable deflection.
Estes and Frangopol (1999) assumed that failure occurs when failure occurs in three of
five adjacent girders. Several major contributions were also made by researchers such as
Ditlevsen (1982, 1996), Grigoriu (1982, 1983), and Rackwitz (1985). Ditlevsen used
conditional probability to calculate bounds of the probability of failure, Grigoriu
discussed a parallel system with brittle elements, and Rackwitz recognized the effect of
correlation on system performance.
The present requirements for civil engineering structures primarily focus on
structural safety. New structural design codes are calibrated using the limit state analysis
10
approach to assure safety standards and to provide the required reliability of new design
structures. However, there is a major gap between the development of reliability
techniques and their application to structural engineering design and evaluation. The
system reliability analysis requires an efficient structural analysis procedure, as the
calculations have to be repeated many times. Therefore, a more comprehensive method to
assess the system reliability of a bridge needs to be developed.
11
CHAPTER 2
LIVE LOAD IN CURRENT DESIGN CODES
2.1. INTRODUCTION
While approaching the problem of live load for bridges it is necessary to review
the current codes and perform a comparative analysis. The codes were selected with the
objective to present various approaches to design live load, use of multilane factors, and
dynamic impact allowance.
The live load for bridges can be represented in many ways, including a uniform
load and a combination of truck(s), as for example in AASHTO LRFD Bridge Design
Specifications (2007), CAN/CSAS600 Canadian Highway Bridge Design Code (2000),
and Eurocode 1 (2002). Nonuniform loading curves are used in the British Standard
5400 (2006) and ASCE Recommended Design Loads for Bridges (1981).
The design live loads specified in AASHTO LRFD Code (2007), OHBDC (1991),
CAN/CSAS600 [2000], Eurocode [2002], and ASCE (1981) were briefly summarized
in the following paragraphs. The comparison of equivalent uniformly distributed loads
for a variety of spans is presented in Paragraph 2.5.
12
2.2. INTERNATIONAL PROVISIONS FOR LIVE LOADING
2.2.1. AASHTO LRFD Code [2007]
The American Association of State Highway and Transportation Officials
Standard Specifications for Highway Bridges define HL93 live load as the extreme force
effect taken as the larger of following:
1) The AASHTO LRFD 3axle design truck, and uniformly distributed design lane
load of 0.64 klf. Figure 2.1.
2) The AASHTO LRFD Design Tandem, and uniformly distributed design lane load
of 0.64 klf. Figure 2.2.
3) For negative moment between points of contraflexure and reaction at interior
piers, combination of two design trucks spaced at minimum of 50 ft, and
uniformly distributed design lane load of 0.64 klf, should be considered. All of the
forces reduced to 90 %. Figure 2.3.
Figure 2.1. HL93 Live Loading in AASHTO LRFD Code [2007].
Truck and Lane Load.
Figure 2.2. HL93 Live Loading in AASHTO LRFD Code [2007].
Tandem and Lane Load.
8.0 kip
32.0 kip 32.0 kip
0.640 klf
14’ 14’  30’
8.0 kip
32.0 kip 32.0 kip
0.640 klf
14’ 14’  30’
8.0 kip
32.0 kip 32.0 kip
0.640 klf
14’ 14’  30’
25.0 kip 25.0 kip
0.640 klf
4’
25.0 kip 25.0 kip
0.640 klf
4’
25.0 kip 25.0 kip
0.640 klf
4’
13
Figure 2.3. HL93 Live Loading in AASHTO LRFD Code [2007].
Alternative Load for Negative Moment between points of contraflexure
and reaction at interior piers.
The loads shall occupy 10 ft transversally within a design lane. Truck wheels are
assumed to be spaced 6.0 ft transversally.
2.2.2. OHBDC [1991]
The Ontario Highway Bridge Design Code (Third edition) determines live load as
a truck or a combination of truck and lane load, whichever produces the maximum load
effect:
1) The OHBD Truck, which is a 5axle truck. Figure 2.4.
2) The OHBD Lane Load consists of an OHBD Truck with each axle reduced to
70%, and superimposed centrally within the width of a 3.0 m (10 ft) wide
uniformly distributed load of 10.0 kN/m (0.685 kip/ft). Figure 2.5.
Figure 2.4. OHBD Live Loading [1991]. OHBD Truck.
14’
7.2 kip
28.8 kip
0.576 klf
14’ 14’  30’
7.2 kip
14’  30’ ≥ 50’
28.8 kip 28.8 kip 28.8 kip
14’
7.2 kip
28.8 kip
0.576 klf
14’ 14’  30’
7.2 kip
14’  30’ ≥ 50’
28.8 kip 28.8 kip 28.8 kip
13.5 kip
36.0 kip
45.0 kip
12’ 20’ 24’ 4’
36.0 kip 36.0 kip
13.5 kip
36.0 kip
45.0 kip
12’ 20’ 24’ 4’
36.0 kip 36.0 kip
14
Figure 2.5. OHBD Live Loading [1991]. OHBD Truck and Lane Load.
Table 2.1. Number of design lanes vs. road width in OHBDC [1991]
Width Number of lanes
Wc ≤ 6.0 m 1
6.0 m < Wc ≤ 10.0 m 2
10.0 m < Wc ≤ 13.5 m 3
13.5 m < Wc ≤ 17.0 m 4
17.0 m < Wc ≤ 20.5 m 5
20.5 m < Wc ≤ 24.0 m 6
24.0 m < Wc ≤ 27.5 m 7
27.5 m < Wc 8
2.2.3. CAN/CSAS600 [2000]
The Canadian Highway Bridge Design Code applies CLW loading, which
consists of the truck or the lane load:
1) The CLW Truck is 5axle truck. The number "W" indicates the gross load of the
truck in kN. For the design of a national highway network, loading not less than
CL625 shall be used. Figure 2.6.
2) The CLW Lane Load consists of CLW Truck with each axle reduced to 80%,
and a superimposed uniformly distributed lane load of 9.0 kN/m (0.617 kip/ft),
that is 3.0 m (10 ft) wide. Figure 2.7.
9.4 kip
25.2 kip
31.5 kip
12’ 20’ 24’ 4’
25.2 kip
0.685 kip/ft
25.2 kip
9.4 kip
25.2 kip
31.5 kip
12’ 20’ 24’ 4’
25.2 kip
0.685 kip/ft
25.2 kip
15
Figure 2.6. CAN/CSAS600 Live Loading [2000]. CLW Truck
Figure 2.7. CAN/CSAS600 Live Loading [2000]. CLW Lane Load
Table 2.2. Number of design lanes vs. road width in CAN/CSAS600
Width Number of lanes
Wc ≤ 6.0 m 1
6.0 m < Wc ≤ 10.0 m 2
10.0 m < Wc ≤ 13.5 m 2 or 3 (check both)
13.5 m < Wc ≤ 17.0 m 4
17.0 m < Wc ≤ 20.5 m 5
20.5 m < Wc ≤ 24.0 m 6
24.0 m < Wc ≤ 27.5 m 7
27.5 m < Wc 8
2.2.4. BS 5400 [2006]
According to British Standard the structures and its elements shall be designed to
resist the more severe effects of either design HA loading or design HA loading
11.2 kip
(0.08 W)
28.1 kip
(0.2 W)
39.3 kip
(0.28 W)
12’ 22’ 22’ 4’
33.7 kip
(0.24 W)
28.1 kip
(0.2 W)
12’ 4’
0.617 kip/ft
9.0 kip
(0.064 W)
22.5 kip
(0.16 W)
31.5 kip
(0.224 W)
22’ 22’
27.0 kip
(0.192 W)
22.5 kip
(0.16 W)
16
combined with design HB loading. HA loading represents normal traffic in Great Britain.
HB loading is an abnormal vehicle unit loading. Both loadings include impact.
HA loading consists of uniformly distributed load (UDL), and a knife edge load
(KEL), or a singlewheel load. Live Loading curve HA UDL is shown in the Figure 2.8.
Value of nominal uniformly distributed load (UDL) equal to:
• for loaded lengths up to and including 50 m:
W = 336 (1/L) 0.67 [kN]
• for loaded lengths in excess of 50 m but less than 160 0m:
W = 36 (1/L) 0.1 [kN]
• for loaded lengths above 1600 m, the UDL shall be in agree with the relevant
authority.
The nominal knife edge load (KEL) per lane shall be taken as 120 kN (27 kip).
The single nominal wheel load alternative to UDL and KEL is one 100 kN (22.5 kip)
wheel placed on a carriageway and uniform distribution over a circular contact area
assuming an effective pressure of 1.1 N/mm
2
.
Figure 2.8. BS 5400 Live Loading curve HA UDL [2006]
17
HB loading defines the minimum number of HB loading units that shall be
considered, which is 30. This number may be increased up to 45 if so directed by the
relevant authority. Figure 2.9 below shows the plan and axle arrangement for one unit of
HB loading. One axle is 10 kN (2.25 kip), i.e. 2.5 kN (0.56 kip) per wheel. The overall
length of the HB vehicle shall be taken as 10, 15, 20, 25, or 30 m for inner axle spacing
of 6, 11, 16, 21, or 26 m respectively. The effects of the most severe case shall be
adopted.
Figure 2.9. Dimensions of HB vehicle.
2.2.5. Eurocode 1 [2002]
The Eurocode 1 Part 2 is applicable to bridges with spans from 5 to 200 m (17 to
667 ft), and carriageway width up to 42 m (140 ft). It presents four models for
determining the main vertical loads from traffic:
1) Load Model 1 (LM1) consists of concentrated and uniformly distributed loads
(Figure 2.10) which cover most of the effects of the traffic of trucks and cars. The
code specifies live load to be used on each traffic lane, therefore there is no need
to introduce multilane factors. It is used for general and local verifications.
18
Table 2.3. Characteristic values of load for successive road lanes
Location Axle load Q
ik
UDL q
ik
Lane number 1 300 kN (67.5 kip) 9 kN/m
2
(0.188 kip/ft
2
)
Lane number 2 200 kN (45.0 kip) 2.5 kN/m
2
(0.052 kip/ft
2
)
Lane number 3 100 kN (22.5 kip) 2.5 kN/m
2
(0.052 kip/ft
2
)
Other lanes 0 2.5 kN/m
2
(0.052 kip/ft
2
)
Remaining area 0 2.5 kN/m
2
(0.052 kip/ft
2
)
Figure 2.10. Eurocode 1 [2002]. Load Model 1
2) Load Model 2 (LM2) consists of a single axle load of 400 kN, which covers the
dynamic effects of the normal traffic on short structural members. The distance
between wheels is 2 m. The contact surface of each wheel should be taken as a
rectangle of sides 0.35 m and 0.60 m. When relevant, only one wheel of 200 kN
may be taken into account.
3) Load Model 3 (LM3) consists of sets of axle loads representing special (carrying
heavy loads) vehicles, which can travel on routes permitted for abnormal loads. It
is intended for general and local verifications.
19
4) Load Model 4 (LM4) represents crowd loading of 5.0 kN/m
2
. It is intended only
for general verifications and it is particularly relevant for bridges in or near town
areas.
2.2.6. ASCE Loading [1981]
The live load known unofficially as the ASCE Loading is a result of the studies
performed by Peter G. Buckland, which was recommended for long span bridge by the
American Society of Civil Engineers Committee on Loads and Forces on Bridges. ASCE
(1981) specifies three levels of live load for highway bridges depending on the average
percentage of heavy vehicles in traffic flow: 7.5%, 30%, and 100% heavy vehicles of the
total vehicle population. "Heavy vehicles" were defined as buses and trucks over 12 000
lbs. It has been proved that the loading can be represented by a uniform load and a
concentrated load (Figure 2.11) to give moments and shears with a sufficient degree of
accuracy.
Figure 2.11. ASCE Loading on Log Scale
Loaded Length (ft)
50 100 200 400 800 1600 3200 6400
1
0
2
2.74
U
(
k
/
f
t
)
P
U(100% HV)
U(30% HV)
U(7.5% HV)
0
56.2
102.4
158.6
P
(
k
i
p
s
)
P U
Loaded Length (ft)
50 100 200 400 800 1600 3200 6400
1
0
2
2.74
U
(
k
/
f
t
)
P
U(100% HV)
U(30% HV)
U(7.5% HV)
0
56.2
102.4
158.6
P
(
k
i
p
s
)
P U
20
2.3. PROVISIONS FOR DYNAMIC LOAD FACTOR
There is considerable variation in the treatment of dynamic load effects by bridge
design codes in different countries. The most common approach is to apply dynamic
response as a fraction or multiple of the response that would be obtained if the same
forces or loads were applied statically. The objective of this simple approach is to not
increase complexity to the designer.
The American Association of State Highway and Transportation Officials
Standard Specifications for Highway Bridges AASHTO LRFD [2007] define that
dynamic load allowance shall be applied to static load effects of the truck or tandem, as a
percentage specified in the table below. It shall not be applied to pedestrian loads or to
the design lane load.
Table 2.4. Dynamic allowance in AASHTO LRFD [2007]
Component Dynamic allowance, IM
Deck joints,
all limit states
75%
All other components,
fatigue and fracture limit states
15%
All other components,
other limit states
33%
Until recently AASHTO Standard Specifications for Highway Bridges used to
specify dynamic load effects in terms of an impact factor that is a function of bridge span.
However relation between dynamic load factor and bridge span is a controversial issue
between researchers (see Paragraph 7.2.).
The Canadian Highway Bridge Design Code CAN/CSAS600 [2000] and
OHBDC [1991] apply dynamic allowance as a percentage to static effects of the CLW
Truck for the number of axles considered in the design lane, as shown in the table below.
21
Table 2.5. Dynamic allowance in CAN/CSAS600
for deck joints 0.50
1 axle of CLW Truck 0.40
2 axles of CLW Truck 0.30
3 or more axle of CLW Truck 0.25
According to British Standard BS 5400 [2006] and the Eurocode 1 [2002] the
effects of vibration due to live load are not required to be considered. Their effect has
already been taken into consideration in definition of design loading.
The ASCE model (1981) does not have any allowance for dynamic load on the
ground that the worst loading occurs with stationary bumpertobumper traffic.
2.4. PROVISIONS FOR MULTILANE REDUCTION FACTORS
For multilane bridges, the multiple lane reduction factors are specified in most of
the codes. The approaches to multilane factors vary significantly. They are shown in
Table 2.6 and Table 2.7. The British Standard BS 5400 developed the most compound
procedure of selection of multilane factor, which depend not only on the number of lanes,
but also on loaded length, number and width of notional lanes (Table 2.7). Eurocode does
not define multilane reduction factor, but it gives the load values to be applied on
successive road lanes directly (Table 2.3). The multilane reduction factors are further
discussed in CHAPTER 6 of this dissertation.
22
Table 2.6. Comparison of Multilane Reduction Factors
Code
Number of Lanes
1 2 3 4 5 6 or more
AASHTO LRFD (2007) 1.20 1.00 0.85 0.65 0.65 0.65
OHBDC (1983, 1991) 1.00 0.90 0.80 0.70 0.60 0.55
CAN/CSAS600 (2000) 1.00 0.90 0.80 0.70 0.60 0.55
ASCE (1981) 1.00 0.70 0.40 0.40 0.40 0.40
Table 2.7. Multilane Reduction Factors for BS 5400
Number of Lanes
loaded length [m] 1 2 3 4 or more
0<L≤20 α
1
α
1
0.6 0.6 α
1
20<L≤40 α
2
α
2
0.6 0.6 α
2
40<L≤50 1.0 1.0 0.6 0.6
50<L≤112, N<6 1.0 7.1/√L 0.6 0.6
50<L≤112, N≥6 1.0 1.0 0.6 0.6
L>112, N<6 1.0 0.67 0.6 0.6
L>112, N≥6 1.0 1.0 0.6 0.6
α
1
= 0.274 b
L
α
2
= 0.0137{b
L
(40L)+3.65(L20)}
where b
L
is the notional lane width
N is total number of notional lanes on the bridge. For a bridge carrying oneway traffic only,
the value N shall be multiplied by 2.
23
2.5. COMPARISON OF EQUIVALENT UNIFORMLY DISTRIBUTED LOADS
In this paragraph, the resulting equivalent uniformly distributed loads for a variety
of design codes and span lengths are plotted and compared. Figure 2.12 and Figure 2.14
compare the equivalent unfactored uniform loads. More valid comparison is to compare
factored loads; these are shown in Figure 2.13 and Figure 2.15. For AASHTO LRFD
(2007) live load factor is 1.75, for OHBDC (1991) it is 1.40, for CAN/CSAS600 it is
1.70, for BS 5400 it is 1.50, and for Eurocode it is 1.35. ASCE studies made no reference
to load factors to be used with its recommended loading, but since a factor of 1.80 has
been used by (Buckland 1991), the same value has been adopted for this comparison.
Equivalent loads including dynamic loads are shown in Figure 2.14, Figure 2.15,
and Figure 2.16. The design loads are increased by the dynamic load factor, DLF, which
has a value as described for each code in the paragraphs above. Since live loadings in
British Standards and Eurocode include dynamic load, they have been used only for
comparison of loads including dynamic load.
Results of the comparison show that variation between the unfactored values of
live load in different codes is significant. European values double those of North
America. The application of load factors slightly reduces the differences. The comparison
of four loaded lanes shows that the differences are reduced even more, Figure 2.16. The
importance of load factors and multilane factors cannot be underestimated.
To obtain plots of UDL, the maximum bending moment (M
max
) was calculated for
simple spans from 400 through 5000 ft. Then, the equivalent uniformly distributed load
UDL was determined from the following formula:
2
max
/ ) ( 8 L M UDL ⋅ =
(2.1)
where: L is span length.
It should be noted that most of the codes are limited to certain lengths of span,
and they are extrapolated beyond these points partially for interest and partially because
these codes are also occasionally used for longer spans due to a lack of adequate guides.
24
Figure 2.12. Equivalent Unfactored Loads, w/o IM, w/o multilane factors.
0.00
0.50
1.00
1.50
2.00
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93
OHBDC 1991
CAN/CSAS600
ASCE 7.5%HV
ASCE 30% HV
ASCE 100% HV
0.00
0.50
1.00
1.50
2.00
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93
OHBDC 1991
CAN/CSAS600
ASCE 7.5%HV
ASCE 30% HV
ASCE 100% HV
25
Figure 2.13. Equivalent Factored Loads, w/o IM, w/o multilane factors.
Figure 2.14. Equivalent Unfactored Loads, with IM, w/o multilane factors.
0.00
0.50
1.00
1.50
2.00
2.50
3.00
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93 x 1.75
OHBDC 1991 x 1.40
CAN/CSAS600 x 1.70
ASCE 7.5%HV x 1.80
ASCE 30% HV x 1.80
ASCE 100% HV x 1.80
0.00
0.50
1.00
1.50
2.00
2.50
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93
OHBDC 1991
CAN/CSAS600
ASCE 7.5%HV
ASCE 30% HV
ASCE 100% HV
BS 5400 HA loading
Eurocode LM1
26
Figure 2.15. Equivalent Factored Loads, with IM, w/o multilane factors.
Figure 2.16. Equivalent Factored Loads, with IM, with multilane factors
for 4 traffic lanes.
0.00
0.50
1.00
1.50
2.00
2.50
3.00
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93 x 1.75
OHBDC 1991 x 1.40
CAN/CSAS600 x 1.70
ASCE 7.5%HV x 1.80
ASCE 30% HV x 1.80
ASCE 100% HV x 1.80
BS 5400 HA loading x 1.50
Eurocode LM1 x 1.35
0.00
1.00
2.00
3.00
4.00
5.00
6.00
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93 x 1.75
OHBDC 1991 x 1.40
CAN/CSAS600 x 1.70
BS 5400 HA loading x 1.50
Eurocode LM1 x 1.35
27
Table 2.8. Values of Equivalent Unfactored Loads, w/o IM, w/o multilane factors.
Table 2.9. Values of Equivalent Factored Loads, w/o IM, w/o multilane factors.
Table 2.10. Values of Equivalent Unfactored Loads, with IM, w/o multilane factors.
length [ft] OHBDC 1991 CAN/CSAS600 HL93
500 1.151 1.067 0.928
1000 0.918 0.842 0.784
1500 0.841 0.767 0.736
2000 0.802 0.729 0.712
2500 0.779 0.707 0.698
3000 0.763 0.692 0.688
3500 0.752 0.681 0.681
4000 0.744 0.673 0.676
4500 0.737 0.667 0.672
5000 0.732 0.662 0.669
length [ft] OHBDC 1991 CAN/CSAS600 HL93
500 1.612 1.813 1.624
1000 1.286 1.431 1.372
1500 1.177 1.303 1.288
2000 1.123 1.240 1.246
2500 1.090 1.202 1.221
3000 1.068 1.176 1.204
3500 1.053 1.158 1.192
4000 1.041 1.144 1.183
4500 1.032 1.134 1.176
5000 1.025 1.125 1.170
length [ft] OHBDC 1991 CAN/CSAS600 HL93 BS 5400
500 1.336 1.179 1.023 1.603 2.120
1000 1.045 0.898 0.832 1.449 1.985
1500 0.948 0.804 0.768 1.375 1.941
2000 0.900 0.757 0.736 1.328 1.918
2500 0.870 0.729 0.717 1.294 1.905
3000 0.851 0.711 0.704 1.268 1.896
3500 0.837 0.697 0.695 1.246 1.889
4000 0.827 0.687 0.688 1.228 1.884
4500 0.819 0.679 0.683 1.212 1.881
5000 0.812 0.673 0.678 1.198 1.878
Eurocode
28
Table 2.11. Values of Equivalent Factored Loads, with IM, w/o multilane factors.
Table 2.12. Values of Equivalent Factored Loads, with multilane factors for 4 traffic
lanes.
OHBDC 1991 CAN/CSAS600 HL93 BS 5400 Eurocode
500 1.871 2.004 1.790 2.404 2.862
1000 1.463 1.526 1.455 2.173 2.680
1500 1.327 1.367 1.343 2.063 2.620
2000 1.259 1.288 1.288 1.993 2.589
2500 1.219 1.240 1.254 1.942 2.571
3000 1.191 1.208 1.232 1.902 2.559
3500 1.172 1.185 1.216 1.869 2.550
4000 1.157 1.168 1.204 1.842 2.544
4500 1.146 1.155 1.194 1.818 2.539
5000 1.137 1.144 1.187 1.797 2.535
length [ft] OHBDC 1991 CAN/CSAS600 HL93 BS 5400 Eurocode
500 5.238 5.612 4.655 5.771 5.309
1000 4.097 4.274 3.783 5.216 4.945
1500 3.716 3.828 3.493 4.952 4.823
2000 3.526 3.605 3.348 4.782 4.762
2500 3.412 3.471 3.261 4.660 4.726
3000 3.336 3.382 3.202 4.564 4.702
3500 3.282 3.318 3.161 4.486 4.684
4000 3.241 3.271 3.130 4.420 4.671
4500 3.209 3.234 3.106 4.364 4.661
5000 3.184 3.204 3.086 4.314 4.653
29
CHAPTER 3
STRUCTURAL RELIABILITY PROCEDURES
3.1. INTRODUCTION
The structures and their components should be designed to have a desirable level
of reliability, which would assure their good performance to account for actions applied
during construction and service. For this purpose civil engineering uses a probabilistic
evaluation of reliability. The design of new structures as well as the evaluation of existing
structures requires verification of limit states, which when exceeded lead to structural
failure (ultimate limit states) or make use of the structure impossible (serviceability limit
states).
The actions (loads, Q) and structural resistance (capacity, R ) are the variables that
decisively influence the state of a structure. They include uncertainties coming from
mechanical material properties, geometry of a structure, loads, etc. Those uncertainties
can be measured only with the use of probability. Therefore, the design of structures is a
process in which decisions are made under uncertainty and limits. Their rational
treatment, and agreement between realinput data and a mathematical model of
phenomenon, is a concern of structural reliability.
Unreliability of a structure is a state in which a structure does not fulfill design
requirements related to its function and desirable performance. It could be a collapse of a
structure, failure or other deficiency in a structural resistance, unfulfilled service demands
30
of a structure, i.e. excessive deformations, excessive vibrations, etc. Structures usually
have a number of possible failure scenarios. For most of the structures it is impossible to
examine all their failure modes. Therefore, representative failure scenarios have to be
chosen. The analysis usually includes an estimation of structural reliability with respect
to specified failure modes. All modes must be treated separately. Thus, reliability of a
structure is the probability that the system will not reach a specified failure mode related
to a specified limit state during a specified period of time.
3.2. STANDARD VARIABLES AND PROBABILITY DISTRIBUTIONS
The key probabilistic characteristics of a random variable are described in terms
of mean, variance and standard deviation. A distribution function would complete the
description of the probabilistic characteristics of random variables, but sometimes it
remains unknown. There are two types of random variables: discrete and continuous. A
discrete random variable may take on only discrete values. Its probability is given by the
probability mass function, ) (
i X
x P . A continuous random variable can take on a
continuous range of values, and its probability is defined by the probability density
function (PDF), ) (x f
X
.
Mean
The mean (expected value) is an average of all observations on a random variable. It is
also defined as the first moment about the origin. For the continuous random variables,
the mean ( µ ) can be computed as:
dx x f x
X
) ( ⋅ =
∫
+∞
∞ −
µ
(3.1)
For the discrete random variables, the mean is given by:
) (
1
i X
n
i
i
x P x ⋅ =
∑
=
µ
(3.2)
If all n observations are given equal weights ( n x P
i X
/ 1 ) ( = ), then the mean for a discrete
random variable is given by:
31
∑
=
⋅ =
n
i
i
x
n
X
1
1
(3.3)
Variance
The variance (
2
σ ) is the second moment about the mean, and it is computed as follows:
dx x f x
X
) ( ) (
2 2
⋅ − =
∫
+∞
∞ −
µ σ
(3.4)
For the discrete variable, the variance is computed as:
) ( ) (
1
2 2
i X
n
i
i
x P x ⋅ − =
∑
=
µ σ
(3.5)
If all n observations are given equal weights ( n x P
i X
/ 1 ) ( = ), the variance is as follows:
∑
=
− ⋅
−
=
n
i
i
X x
n
1
2 2
) (
1
1
σ
(3.6)
Standard Deviation
The standard deviation (
2
σ ) of a probability distribution is defined as the square root of
the variance.
Coefficient of Variation
The coefficient of variation (V ) is a dimensionless quantity defined as:
µ
σ
= V
(3.7)
Probability distributions
There are many types of discrete and continuous distributions. The most commonly used
are the continuous distributions: uniform, normal, lognormal, exponential, and gamma. In
this section the normal distribution is presented, because this is the only distribution used
in this dissertation. Further details about distributions can be found, for instance, in
Nowak and Collins (2000).
Normal distribution (Gaussian distribution) is the most widely used probability
distribution. It has a probability density function given by:
32
] ) (
2
1
exp[
2
1
) (
2
σ
µ
π σ
−
⋅ − ⋅
⋅ ⋅
=
x
x f
X
(3.8)
This function is graphically represented as shown in Figure 3.1.
Figure 3.1 PDF and CDF of a normal random variable
The standard normal distribution is a special case of Gaussian distribution, with
parameters 0 =
X
µ and 0 . 1 =
X
σ . Its PDF is denoted as ) (z φ , its CDF is denoted as
) (z Φ , and they are defined as follows:
) ( )
2
1
exp(
2
1
) (
2
z f z z
Z
= ⋅ − ⋅
⋅
=
π
φ
(3.9)
∫
∞ −
⋅ − ⋅
⋅
= Φ
z
dz z z )
2
1
exp(
2
1
) (
2
π
(3.10)
Central limit theorem states that the sum of a large number of independent observations,
without a dominating distribution type, approaches an approximate normal distribution.
The higher the number of observations the better is the approximation. This theorem is
one of the most important in probability theory. The sum of variables is often used to
model total load acting on a structure, which can be approximated as a normal variable.
random variable X
F
X
(x)
33
Mathematically it can be expressed that the sum of n random variables,
n
X X X ,..., ,
2 1
,
is equal to function Y having normal distribution:
n
X X X Y + + + = ...
2 1
(3.11)
n
X X X Y
µ µ µ µ + + + = ...
2 1
(3.12)
2 2 2 2
...
2 1 n
X X X Y
σ σ σ σ + + + =
(3.13)
3.3. LIMIT STATE FUNCTION
In most design codes, the structural design is based on the concept of limit states.
The philosophy of limit state design assumes equilibrium between applied loads and
structural response of the structure (capacity, resistance). Therefore, a specified set of
load and resistance factors is required for each limit state formulated for different
possible scenarios of structural behavior during construction as well as service life.
Three types of limit states are typically used with reference to structural reliability
analysis:
1. Ultimate limit states (ULSs), which represents the loss of structural capacity.
2. Serviceability limit states (SLSs), which represents failure due to deterioration
of functionality.
3. Fatigue limit states (FLSs), which represents the loss of strength for a structural
component under the action of repeated loading.
The limit state defines the boundary between the desired and undesired
performance of a structure, between situations when the structure is safe (a safety margin
exists) and the structure is not safe (failure occurs). The probability of the desired
performance of a structure is equal to the safety margin (
S
P ). The probability of an
undesired performance of a structure is equal to the probability of failure (
f
P ). Failure
and nonfailure states fulfill the entire probabilistic sample space ( Ω). Therefore, the
probability of occurrence is 1 ) ( = Ω P , so:
34
1 ) ( = + = Ω
s f
P P P ⇒
f s
P P − = 1
(3.14)
Reliability analysis usually begins with the formulation of a limit state function
(performance function). All loads are being incorporated into one variable ( Q) and the
resistance of the structure is being incorporated into one variable ( R ). In the general case,
the performance function of a system can be related to any possible failure scenario or
any limiting state and defined as a function of capacity and demand:
( ) Q R Q R g − = ,
(3.15)
where R is capacity representing resistance of a structural system or a structural element,
and Q is demand representing load effect in a structure or a structural component.
Both R and Q are continuous random variables having a probability density
function (PDF). The quantity Q R − is also a random variable with its own PDF as shown
in the Figure 3.2.
Figure 3.2 PDF’s of load, resistance, and safety margin
The performance function usually is a function of capacity and demand variables (
n
X X X ,..., ,
2 1
), and it adopts a format:
( ) ( )
n
X X X g X g ,..., ,
2 1
=
(3.16)
where
i
X is the collection of input parameters.
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
R  Q
Safety margin
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Q
Load
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure 58 PDFs ofload,resistance, andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
PDF
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Probability
of Failure
random value
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure58 PDFsofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R,resistance
0
Figure 58 PDFs ofload,resistance,andsafety margin.
PDF
Probability
ofFailure
RQ, safety
margin
Q,load effect R,resistance
0 R
Resistance
Figure 58 PDFs ofload,resistance,andsafetymargin.
PDF
Probability
ofFailure
RQ,safety
margin
Q,loadeffect R, resistance
0
0
35
From the definition of the performance function, it can be derived that when
0 ) ( < X g , it indicates failure and, when 0 ) ( ≥ X g , it indicates acceptable performance.
The performance defined as 0 ) ( = X g is called the failure surface. The corresponding
probability of failure can be defined as the integral of the joint density function of the
variables over the negative domain of ) ( X g (ThoftChristensen and Baker 1982):
∫ ∫
<
⋅ ⋅ =
0 ) (
1 2 1 ,..., ,
... ) ,..., , ( ....
2 1
X g
n n X X X f
dx dx x x x f P
n
(3.17)
where
X
f is the joint probability density function of
n
X X X ,..., ,
2 1
There is almost never sufficient data to define the joint probability density
function for all basic variables, thus the equation (3.17) is very difficult to evaluate. Even
knowing the joint density function, the necessary multidimensional integration may be
extremely timeconsuming. In practice, a direct calculation of the probability of failure
becomes inefficient. Therefore, indirect procedures such as the reliability index are used.
3.4. RELIABILITY INDEX
The evaluation reliability index, also called safety index, is an effective measure
of the probability of failure. There are several methods to calculate reliability of structural
components: the firstorder reliability methods (FORM), the advanced firstorder second
moment methods (FOSM), simulation techniques, etc. In the late 1960s, the firstorder
secondmoment formulation was developed and advanced by Cornell (1967) and Ang
and Cornell (1974). Further advances in these methods were made by Hasofer and Lind
(1974) and Rackwitz and Fiessler (1978). The FOSM methods can be used to solve many
practical problems. The concept of secondmoment is often used in practical
quantification of safety and reliability. It has been extensively used in calibrations of
structural design codes. The FOSM approach can be put into several categories with
regard to accuracy of results, required input data, computing cost, or simplicity of
formulation.
36
In 1974, Hasofer and Lind introduced the definition of the reliability index as the
shortest distance from the origin to the limit state function in a system of reduced
variables coordinates (Figure 3.3). Using geometry the reliability index can be calculated
as:
2 2
Q R
Q R
σ σ
µ µ
β
+
−
=
(3.18)
Figure 3.3 Reliability index defined as the shortest distance in the space of reduced
variables
All the variables should be expressed in nondimensional forms. Thus, the
reduced variables (
Q R
Z Z , ) have to be introduced:
Q
Q
Q
R
R
R
R
Z
R
Z
σ
µ
σ
µ
−
=
−
=
(3.19)
The resistance ( R ) and the load ( Q) can also be expressed in the form of reduced
variables:
Z
R
Z
Q
FAILURE
SAFE
g(Z
R
,Z
Q
) = 0
limit state function
µ
R
µ
Q
σ
Q
µ
R
µ
Q
σ
Q
β
37
Q Q Q
R R R
Z Q
Z R
σ µ
σ µ
⋅ + =
⋅ + =
(3.20)
Therefore the limit state function Q R Q R g − = ) , ( in terms of reduced variables can be
rewritten as:
Q Q Q R R R Q R
Z Z Z Z g σ µ σ µ ⋅ + − ⋅ + = ) , (
(3.21)
The reliability index recognizes the importance of uncertainty in load effects and
strength. It incorporates the four key parameters, resistance and load with their mean
values and standard deviations,
Q R
µ µ , and
Q R
σ σ , , respectively.
The limit state function used in this dissertation is linear. In case it was nonlinear,
iteration would be required to find the design point in reduced variable space such that β
corresponds to the shortest distance. Moreover, the HasoferLind approach evaluates the
reliability index for uncorrelated random variables. Thus, if the initial variables are
correlated they must be transformed into uncorrelated random variables.
The probability of failure can be calculated using the formula (3.22). The
calculation can give exact results, if the random variables are normally distributed and
uncorrelated. Otherwise it provides only an approximation.
) ( β − Φ =
f
P
(3.22)
where Φ is the standard normal distribution function.
3.5. MONTE CARLO METHOD SIMULATION TECHNIQUE
Simulation is the process of replacing reality with theoretical and experimental
models. Theoretical simulation is also called numerical or computer experimentation. It is
a practical tool that allows obtaining data, either instead of or in addition to realworld
data. Simulating a phenomenon numerically assumes events occur a finite number of
times. The frequency of occurrence of an event in the entire set of simulations
approximates its probability of occurrence. This relatively straightforward concept often
requires complex procedures. The most commonly used simulation technique is the
Monte Carlo Method (ThoftChristensen and Baker, 1982; Hart, 1982). Many other
38
simulation techniques work similarly to the Monte Carlo Method. However instead of
generating a large number of random values for variables, the values are arbitrary and
selected according to specified rules.
The Monte Carlo Method is commonly used to predict the behavior of structural
elements and systems from the probability point of view, and without physical testing. It
is used to evaluate the probability of structural failure and, indirectly, the reliability
index. A large amount of random numbers corresponding to random variables is
numerically generated. This large number of repetitions is particularly valuable in solving
problems involving rare events, where physical testing could be very expensive. This
method may be used not only to study performance of a structural system for a prescribed
set of design variables, but also to measure sensitivity in system performance due to
variations of some parameters. Therefore, for engineering purpose, it can be used to
determine optimal design.
The Monte Carlo simulation method consists of the following steps. In the first
step a simulation of the uniformly distributed random numbers
n
u u u ,..., ,
2 1
between 0
and 1 is performed. They can be generated by computer programs using a builtin option.
Then, the standard normal random values can be calculated using generated numbers and
information about the types of distributions and statistical parameters of distributions
(mean value and standard deviation) of each design variable. The standard normal
random number
i
z is calculated using the equation:
) (
1
i i
u z
−
Φ =
(3.23)
where:
1 −
Φ
is the inverse of the standard normal cumulative distribution function
Using standard random values (mean value
X
µ and standard deviation
X
σ ), the
values x
i
of sample random numbers can be generated for the random normal variable X
, as:
X i X i
z x σ µ ⋅ + =
(3.24)
39
It is important to simulate an efficient number of sets, such that the variation of
the design parameters in a single simulation will not influence the solution of the entire
process of simulations.
Performed Monte Carlo simulations allow for estimation of the probability of
failure. The probability of failure is defined as the ratio between the numbers of times the
criterion for the failure is achieved ( n ), to the total number of simulations ( N ). Each
simulated value has the same weight.
) X g(
) X g(
N
n
P f
of s simulation of number total
0 when s simulation of number total <
= =
(3.25)
where:
) ( X g defines the performance function with the limit state 0 ) ( = X g
0 ) ( < X g defines the state of failure
3.6. NORMAL PROBABILITY PAPER
Normal probability paper is used to present cumulative distribution functions
(CDF) in a convenient way. Cumulative distribution functions for the normal distribution
are “Sshape” function. Normal probability paper redefines the vertical scale so that the
normal CDF can be plotted as a straight line and allows for an easy evaluation of the
most important statistical parameters as well as type of distribution function. More
detailed information about normal probability paper can be found in textbooks (Nowak
and Collins 2000, Benjamin and Cornell 1970). The basic variable is presented on the
horizontal. The vertical axis, being the standard normal variable, represents the distance
from the mean value in terms of standard deviations. It can also be implied as the
corresponding probability of being exceeded. The relationship between the standard
normal variable and probability is given in Table 3.1.
40
Table 3.1. Relationship between vertical scale on Normal Probability Paper
and Probability
Distance from the mean value in terms of
standard deviations
Corresponding probability
4 0.9999683
3 0.99865
2 0.9772
1 0.841
0 0.5
1 0.159
2 0.0228
3 0.00135
4 0.0000317
The shape of the resulting curve representing CDF allows for analysis of the test
data plotted on the normal probability paper.
The basic properties of the normal probability paper:
A straight line represents a normal distribution function.
The mean value and standard deviation read directly from the graph.
The mean value is at the intersection of the normal CDF and horizontal axis.
The standard deviation can also be read as shown in Figure 3.4.
41
Figure 3.4 Normal Distribution Function on the Normal Probability Paper.
0
1
1
2
3
2
standard deviations
mean
Standard Normal
Variable
CDF
F(x)
42
CHAPTER 4
TRAFFIC DATA
4.1. REGULATIONS OF TRUCK TYPES, TRUCK SIZES AND WEIGHT LIMITS
Federal and state regulations limit the weight and dimensions of vehicles on U.S.
highways. These restrictions have important impacts on highway construction costs,
maintenance costs, and highway safety issues. Current Federal law includes the following
limits:
 20 000 pounds  maximum gross weight upon any one axle
 34 000 pounds  maximum gross weight on tandem axles
 80 000 pounds  maximum gross vehicle weight
 102 inches  maximum vehicle width
 48feet  minimum vehicle length for a semitrailer in a trucktractor/semi
trailer combination
 28 feet  minimum vehicle length for a semitrailer or trailer operating in a
trucktractor/semitrailer/trailer combination.
The types of the vehicles in use on American roads are classified by FHWA into 13
categories, as show in Figure 4.1. Classes 13 are passenger vehicles, classes 47 are
single unit trucks and buses, classes 810 are combination trucks, classes 1113 are multi
trailer trucks.
43
Figure 4.1. FHWA 13category scheme
44
Table 4.1. Conversion chart for vehicles’ class and number of axles
Vehicle Class
Average Number of Axles
per Vehicle
1 2
2 2
3 2
4 2.2
5 2
6 3
7 4
8 4
9 5
10 6
11 4
12 6
13 7
Several states issue overweight permits and allow higher truck loads. For example
the state of Michigan, from where some publications on field test results are used in this
study, allows trucks up to 164,000 pounds. The states which allow various longer
combination vehicles are presented in Figure 4.3. Types of longer combination vehicles
(Figure 4.2) are:
 Rocky Mountain Double (common maximum weight – 105,500  137,800 lbs)
 Turnpike Double (common maximum weight – 105,500  129,000 lbs)
 Btrain Double Trailer Combination (common maximum weight – 105,500 
147,000 lbs)
 Triple Trailer Combination (common maximum weight – 105,500  131,000 lbs)
45
Figure 4.2. Longer Combination Vehicles (LCV’s)
Figure 4.3. States allowing various Longer Combination Vehicles
Rocky Mountain Double
8Axle BTrain Double Trailer Combination
Turnpike Double
Triple Trailer Combination
States Allowing LCVs’
States Allowing Rocky Mountain Doubles States Allowing Turnpike Doubles
States Allowing Triples
46
Federal size and weight studies were established in 1982, and since then no
significant changes have been made. However, several proposals to make changes in
these regulations were presented. The most recent studies are the TRB Special Report
267 "Regulation of Weights, Lengths, and Widths of Commercial Motor Vehicles" and
the U.S. Department of Transportation "Comprehensive Truck Size and Weight Study:
Volume I Summary Report". Both documents discuss existing regulations and give
recommendations on their improvement. Elimination of the federal 80 000 pounds weight
limit on Interstate highways is recommended. It is proposed that the gross weight should
be governed by appropriate axle weight limits and the bridge formula (Figure 4.4). The
maximum weight (in pounds) carried on a group of two or more consecutive axles would
not exceed that given by the following formulas:
 W = 1000*(2L+26) for L≤24 ft
 W = 1000*(L/2+62) for L>24 ft
Figure 4.4. New Bridge Formula  regulation of vehicles' length and weight
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0 20 40 60 80 100 120
v
e
h
i
c
l
e
w
e
i
g
h
t
[
k
i
p
/
f
t
]
vehicle length [ft]
for L≤24 ft
for L>24 ft
47
In the Figure 4.5, the data from the Transportation Statistics Annual Report
December 2006 of U.S. Department of Transportation is shown. As can be observed, the
number of trucks has a trend of rapid growth. The number of heavy trucks is growing
faster than number of light trucks. The number of light trucks (under 10 000 pounds)
increased 73 percent between 1992 and 2005, and the number of heavy trucks (greater
than 10 000 pounds) increased 112 percent. In 2005 heavy trucks constituted 8% of the
volume of trucks, while in the 1990’s they were only 4% of the volume. In 2005, 95.3
million light trucks traveled 1.060 trillion vehiclemiles, and 8.5 million heavy trucks
traveled 222.29 billion vehiclemiles.
Figure 4.5. Number of trucks by weight (in thousands of trucks).
Transportation statistics annual report, December 2006.
According to Texas Transportation Institute “Over the next 20 years, truck tonnage is
expected to increase at a rate more than five times that of population growth.”
0
20000
40000
60000
80000
100000
120000
1992 1997 2002 2005
55193
68100
79760
95300
4007 4700 5400
8500
light trucks
heavy trucks
48
Figure 4.6. Freight Tonnage Moved by Truck (FHWA)
Figure 4.7 presents time variation of total truck weight statistic between the years
1993 and 2003. The data is expressed as: mean value (µ), 95th percentile (W95, 95
percent of the trucks weigh less), and maximum observed total truck weight (Max). The
study was made for the state of New Jersey, which has lower limits than the state of
Michigan. However, the observed maximum gross vehicle weight is high, and it reaches a
value of 225 kips (1000 kN). The maximum truck weight shows steady increase at an
annual growth rate of 1.2%.
0
5000
10000
15000
20000
25000
1995 2000 2005 2010 2015 2020 2025 2030 2035
T
o
n
s
(
m
i
l
l
i
o
n
s
)
Year
49
Figure 4.7. Time variation of total truck weight statistic.
(Gindy, M., Nassif, H.H., 2006)
4.2. DATA COLLECTION METHODOLOGY (WIM)
Weigh in Motion (WIM) Technology had its beginnings in the early 1950s when
the U.S. Bureau of Public Roads, the Virginia State Department of Highways, and the
Williams Construction Company installed a load cell WIM system on the Henry G.
Shirley Memorial Highway. From these early beginnings, WIM technology and
application continued to advance and spread across the nation. In 1990, the American
Society for Testing and Materials (ASTM) published the first Standard Specification for
Highway WeighinMotion (WIM) Systems with User Requirements and Test Methods
(Designation: E 131890). This document was revised in 1994, and again in 2002 to the
version (Designation: E 131802) that is used today.
ASTM Designation: E 131802 defines WIM as “the process of measuring the
dynamic tire forces of a moving vehicle and estimating the corresponding tire loads of the
50
static vehicle”. In addition to the collection of dynamic tire forces, a variety of ancillary
traffic data can also be obtained through the use of WIM systems: traffic volume, speed,
directional distribution, lane distribution, date and time of passage, axle spacing, axle
weight, and vehicle classification. Of all data collection methodologies, WIM data
collection requires the most sophisticated technology for data collection sensors, the most
controlled operating environment (smooth, level pavement), as well as the highest
equipment setup and calibration costs. The primary reason for sophistication in
technology and its high costs comes from a need to determine static weight from a
dynamic measurement. In standard weigh scale application, vehicles are stopped on a
static scale and are measured without any interaction between the vehicle and the
roadway. In WIM applications, a variety of forces are acting on the vehicle, including the
force of gravity as well as dynamic effects of influences such as: roadway roughness,
vehicle speed, vehicle acceleration and deceleration, out of balance tires and wheels, tire
inflation pressure, suspension, aerodynamics and wind; and other dynamic factors.
Several different technologies are available for WIM data collection systems. The
most commonly used are: piezoelectric cables, bending plate, load cell, quartz cables, and
Bridge WIM systems. ASTM Standard Designation: E 131802 distinguishes four types
of WIM systems (Type I, II, III, and IV) based on application and performance
requirements for data collection. Each type of WIM system has been specified to perform
its indicated functions within specific tolerances. The piezoelectric sensors, which are the
most common, offer acceptable accuracy ±15%. Strain based and load cell WIM systems
are much more expensive, but they provide more accuracy. Recently, piezoquartz
sensors were introduced in the United States. They are less sensitive to temperature
changes and generally more accurate. The majority of WIM data collection is done with
permanently installed weight sensors, although the data is not always collected
continuously.
In order to assure unbiased data, WIM sites should be localized away from weight
stations and be unknown to truck drivers’. WIM equipment should be subject to a regular
maintenance and calibration. To limit erroneous data, it is recommended to avoid sites
with numerous traffic stoppages (speed >10 mph), close to exits, and with rough surfaces.
51
Highway agencies have recognized the advantages of having automated data
collection systems that can provide information on truck weights and truck traffic
patterns for economic analysis, traffic management and various other purposes.
Therefore, the quality and quantity of WIM data has greatly improved in recent years,
and new WIM technologies continue to be developed. Due to the weighinmotion
technologies, unbiased truckloads are being collected at normal highway speeds, in large
quantity, and without truck driver’s knowledge (Figure 4.8).
Figure 4.8. WIM data collection
4.3. WEIGH IN MOTION DATABASE
For the scope of this research, the weigh in motion database was obtained from
the project NCHRP 1276 and measurements on the Throggs Neck Bridge in New York.
The database includes newest (20012006) WIM database for a variety of sites:
California (6), Florida (5), Indiana (6), Mississippi (5), New York (7+2), Oklahoma (16),
52
and Oregon (4). The variety of sites is important, because the truck traffic changes
depending on the site location: interstate or noninterstate highways, rural or urban areas,
and state. Traffic data varies also depending on time of day, day of the week, season of
the year, and direction. Therefore, it is important that the WIM database is collected
continuously (mostly oneyear data), on many traffic lanes and in both directions
(usually). The database contains date and time, lane, number of axles, spacing between
axles, axle weight, speed, and vehicle category. A summary of WIM data, including site
localizations, number of lanes, and types of sensors used is presented in Table 4.2.
Distribution of vehicles by axles and traffic lanes are presented in Table 4.3  Table 4.10.
The statistics is presented for four states selected for simulations: Oregon, Indiana,
Florida, and New York. Cumulative distribution functions of gross vehicle weights by
axles (GVW) were plotted in Figure 4.9Figure 4.11. It can be noticed, that for New York
I495, the heaviest vehicles are 6axles. Those are construction debris, gravel and garbage
haulers. They often drive overloaded above 150 kips and occasionally above 200 kips,
while NYSDOT routine permit trucks that are legal up to 120 kips.
The WIM technology is known to have certain traffic data quality problems. The
errors are due to physical and softwarerelated failures of equipment and transmission,
the difference between the dynamic weight measured and the actual static scale weight,
as well as the effect of tire pressure, size and configuration of the WIM results.
Therefore, the data quality checks have to be implemented to detect and fix/eliminate
erroneous data before processing. A standardized procedure to filter out errors is applied
to WIM data from various sites. According to Traffic Monitoring Guide (2001),
reasonableness checks were performed on the axle weights and spacing. The limits were
200 to 20,000 kilograms (0.44 to 4.41 kips) for axle weights and 0.5 to 15 meters (1.6 to
49.2 feet) for axle spacing. Moreover, all obvious errors such as zero readings for number
of axles or speed were eliminated. The percentage of filtered out data varies for different
sites, which depends on condition of equipment, its regular maintenance and
recalibration.
53
Table 4.2. Summary of WIM Data
State Site ID Route
# of Traffic
Lanes
# of WIM
Lanes
Both Dir WIM Type
CA 0001 Lodi
CA 0003 Antelope
CA 0004 Antelope
CA 0059 LA710
CA 0060 LA710
CA 0072 Bowman
FL 9916 US29 4 4 Y P
FL 9919 I95 4 4 Y P
FL 9926 I75 6 4 Y BP
FL 9927 SR546 4 4 Y BP
FL 9936 I10 4 4 Y P
IN 9511 I65 4 4 Y P
IN 9512 I74 4 4 Y SLC
IN 9532 US31 4 4 Y P
IN 9534 I65 6 6 Y P
IN 9544 I80/I94 6 6 Y P
IN 9552 US50 2 2 Y P
MS 2606 I55 4 4 Y P
MS 3015 I10 4 4 Y P
MS 4506 I55 4 4 Y P
MS 6104 US49 2 2 Y P
MS 7900 US61 4 4 Y P
NY 8280 I84 4 4 Y P
NY 8382 I84 4 4 Y P
OR Woodburn I5 3 2 N SLC
OR Emigrant Hill I84 2 1 N SLC
OR Lowell OR 58 2 2 N SLC
OR Bend US 97 2 1 N SLC
NY 9121 I81 2 2 Y P
NY 2680 8 4 4 Y P
NY I495 Y
P – Piezo, BP – Bending Plate, SLC – Single Load Cell
54
Table 4.3. Vehicles by axle in Oregon
Table 4.4. Vehicles by traffic lane in Oregon
axles
2 44507 4.6% 36959 6.04% 3333 1.56% 2273 2.48% 1942 3.28%
3 71365 7.3% 42009 6.87% 9242 4.34% 9807 10.70% 10307 17.40%
4 62025 6.4% 31066 5.08% 14728 6.91% 8032 8.76% 8199 13.84%
5 575846 59.0% 350107 57.22% 140520 65.97% 57123 62.30% 28096 47.44%
6 70639 7.2% 46792 7.65% 14441 6.78% 3789 4.13% 5617 9.48%
7 85658 8.8% 58407 9.55% 19003 8.92% 5725 6.24% 2523 4.26%
8 60907 6.2% 43947 7.18% 10041 4.71% 4764 5.20% 2155 3.64%
9 3500 0.4% 2253 0.37% 894 0.42% 131 0.14% 222 0.37%
10 697 0.1% 210 0.03% 404 0.19% 26 0.03% 57 0.10%
11 345 0.0% 46 0.01% 239 0.11% 20 0.02% 40 0.07%
12 277 0.0% 34 0.01% 172 0.08% 6 0.01% 65 0.11%
sum 975766 611830 213017 91696 59223
total I5 Woodburn (NB) US 97 Bend (NB) OR 58 Lowell (WB) I84 Emigrant Hill (WB)
lane
1 552388 90.28% 213017 100.00% 51404 56.06% 59223 100.00%
2 59442 9.72% 40292 43.94%
3
4
sum 611830 213017 91696 59223
I5 Woodburn (NB) I84 Emigrant Hill (WB) OR 58 Lowell (WB) US 97 Bend (NB)
55
Table 4.5. Vehicles by axle in Florida
Table 4.6. Vehicles by traffic lane in Florida
axles
2 2400362 25.9% 482051 65.98% 229680 10.11% 1408095 36.45% 127986 19.70% 152550 8.37%
3 779456 8.4% 67058 9.18% 124347 5.47% 402640 10.42% 115466 17.77% 69945 4.12%
4 747991 8.1% 24388 3.34% 156406 6.88% 430324 11.14% 74984 11.54% 61889 3.50%
5 5156752 55.6% 148424 20.31% 1720367 75.69% 1554039 40.23% 297593 45.81% 1436329 82.08%
6 143470 1.5% 7011 0.96% 37909 1.67% 56038 1.45% 11271 1.74% 31241 1.76%
7 18988 0.2% 888 0.12% 2847 0.13% 6990 0.18% 6296 0.97% 1967 0.11%
8 8495 0.1% 410 0.06% 867 0.04% 2475 0.06% 4051 0.62% 692 0.03%
9 15755 0.2% 422 0.06% 405 0.02% 2533 0.07% 11977 1.84% 418 0.02%
10
11
12
sum 9271269 730652 2272828 3863134 649624 1755031
total Florida 9916 Florida 9927 Florida 9919 Florida 9926 Florida 9936
lane
1 249213 34.11% 915451 40.28% 897806 23.24% 229181 35.28% 818460 46.64%
2 74085 10.14% 237260 10.44% 1119347 28.98% 88830 13.67% 114341 6.52%
3 407354 55.75% 181535 7.99% 0 0.00% 62778 9.66% 112767 6.43%
4 938583 41.30% 0 0.00% 268830 41.38% 709463 40.42%
5 1044850 27.05%
6 801070 20.74%
sum 730652 2272829 3863073 649619 1755031
Florida 9919 Florida 9926 Florida 9916 Florida 9927 Florida 9936
56
Table 4.7. Vehicles by axle in Indiana
Table 4.8. Vehicles by traffic lane in Indiana
axles
2 2527382 27.0% 44867 10.11% 16938 8.38% 738274 57.08% 1509944 24.18% 82330 9.91% 135029 37.48%
3 513522 5.5% 15135 3.41% 7400 3.66% 91398 7.07% 353721 5.67% 26523 3.19% 19345 5.37%
4 571231 6.1% 12517 2.82% 3170 1.57% 120486 9.31% 385840 6.18% 26400 3.18% 22818 6.33%
5 5654115 60.3% 364519 82.13% 171368 84.79% 334457 25.86% 3928062 62.91% 675794 81.38% 179915 49.93%
6 95770 1.0% 6534 1.47% 2993 1.48% 7359 0.57% 60292 0.97% 15688 1.89% 2904 0.81%
7 7547 0.1% 193 0.04% 159 0.08% 836 0.06% 3740 0.06% 2381 0.29% 238 0.07%
8 2967 0.0% 59 0.01% 54 0.03% 486 0.04% 1501 0.02% 829 0.10% 38 0.01%
9 945 0.0% 10 0.00% 14 0.01% 100 0.01% 396 0.01% 414 0.05% 11 0.00%
10 355 0.0% 8 0.00% 18 0.01% 46 0.00% 225 0.00% 52 0.01% 6 0.00%
11 131 0.0% 4 0.00% 6 0.00% 20 0.00% 89 0.00% 12 0.00% 0 0.00%
12 88 0.0% 0 0.00% 0 0.00% 21 0.00% 67 0.00% 0.00% 0 0.00%
sum 9374053 443846 202120 1293483 6243877 830423 360304
total Indiana 9534 Indiana 9552 Indiana 9544 Indiana 9511 Indiana 9532 Indiana 9512
lane
1 375357 84.57% 185645 91.85% 823175 63.64% 3232127 51.76% 412695 49.70% 360304 100.00%
2 68489 15.43% 16475 8.15% 470308 36.36% 2657717 42.57% 393046 47.33% 0 0.00%
3 0 0.00% 0 0.00% 354033 5.67% 24682 2.97% 0
4 0 0.00% 0 0.00% 0 0.00% 0
sum 443846 202120 1293483 6243877 830423 360304
Indiana 9511 Indiana 9512 Indiana 9532 Indiana 9534 Indiana 9552 Indiana 9544
57
Table 4.9. Vehicles by axle in New York
Table 4.10. Vehicles by traffic lane in New York
axles
2 114115 41.6% 56459 38.58% 57656 45.02% 108563 7.88% 34752 25.34%
3 40359 14.7% 21102 14.42% 19257 15.04% 135919 9.87% 23644 17.24%
4 19297 7.0% 10605 7.25% 8692 6.79% 74822 5.43% 14154 10.32%
5 82959 30.2% 47294 32.32% 35665 27.85% 1010780 73.39% 52845 38.54%
6 17426 6.4% 10716 7.32% 6710 5.24% 44357 3.22% 9720 7.09%
7 217 0.1% 131 0.09% 86 0.07% 1758 0.13% 1113 0.81%
8 27 0.0% 20 0.01% 7 0.01% 542 0.04% 560 0.41%
9 20 0.0% 19 0.01% 1 0.00% 335 0.02% 212 0.15%
10 0 0.0% 0 0.00% 0.00% 182 0.01% 98 0.07%
11 0 0.0% 0.00% 0.00% 18 0.00% 25 0.02%
12 0 0.0% 0.00% 0.00% 8 0.00% 4 0.00%
13 0.0% 0.00% 0.00% 2 0.00% 1 0.00%
sum 274420 146346 128074 1377284 137127
NY 9121 NY 2680 total I495 EB I495 WB
lane
1 52703 36.01% 43278 33.79% 618289 44.89% 46008 33.55%
2 90625 61.93% 78891 61.60% 88037 6.39% 3148 2.30%
3 3018 2.06% 5905 4.61% 74831 5.43% 8298 6.05%
4 0 0.00% 0.00% 596129 43.28% 79677 58.10%
sum 146346 128074 1377286 137131
NY 9121 NY 2680 I495 EB I495 WB
58
Figure 4.9. CDF's of GVW by axles Oregon I5 Woodburn
3.00
2.00
1.00
0.00
1.00
2.00
3.00
0 20 40 60 80 100 120 140 160 180 200 220 240
I
n
v
e
r
s
e
o
f
S
t
a
n
d
a
r
d
N
o
r
m
a
l
D
i
s
t
r
i
b
u
t
i
o
n
GVW [kips]
I5 Woodburn (NB)
2axle
3axle
4axle
5axle
6axle
7axle
8axle
9axle
10axle
11axle
12axle
59
Figure 4.10. CDF's of GVW by axles Oregon I84 Emigrant Hill
3.00
2.00
1.00
0.00
1.00
2.00
3.00
0 20 40 60 80 100 120 140 160 180 200 220 240
I
n
v
e
r
s
e
o
f
S
t
a
n
d
a
r
d
N
o
r
m
a
l
D
i
s
t
r
i
b
u
t
i
o
n
GVW [kip]
I84 Emigrant Hill (WB)
2axle
3axle
4axle
5axle
6axle
7axle
8axle
9axle
10axle
11axle
12axle
60
Figure 4.11. CDF's of GVW by axles NY I495 EB
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0 20 40 60 80 100 120 140 160 180 200 220 240
CDF's of GVW by axles
NY I495 EB
2axle
3axle
4axle
5axle
6axle
7axle
8axle
9axle
61
CHAPTER 5
DEVELOPMENT OF LIVE LOAD MODEL
5.1. INTRODUCTION
For long span bridges, the extreme live load is governed by the traffic jam
scenario. The live load is modeled as the uniformly distributed lane load and additional
axle load or single truck for deck components. Development of live load model is based
on three approaches. Two of them can be classified as initial studies. The first of them is
based on a 5axle average truck and the second one is based on AASHTO LRFD legal
load trucks. The third approach is detailed study based on truck WIM Data.
5.2. MODEL BASED ON AVERAGE 5AXLE TRUCK
For computation of the live load on the most loaded lane, the following traffic model
has been assumed:
 Traffic jam situation, Figure 5.1.
 Left lane loaded only with average trucks.
 Average trucks are 5axle trucks, which are the most popular among truck
types, Figure 5.4 and Figure 5.5. In the FHWA WIM Data, vehicle categories 1
3 FHWA are omitted, therefore the percentage of 2axle vehicles is relatively
low.
62
 An average 5axle truck:
 is 45 ft long
 weights 55 kips, Figure 5.2 and Figure 5.3.
 Clearance distance is 10 to 15 ft, therefore spacing between the last axle of one
truck and first axle of the following truck is 2025 ft.
Live load due to such a combination of vehicles is equal to:
55 kip / 70 ft = 0.79 kip/ft for clearance distance of 15 ft
55 kip / 65 ft = 0.85 kip/ft for clearance distance of 10 ft
Figure 5.1. Critical loading. Traffic jam scenario.
Figure 5.2. CDF’s of GVW for 5axle trucks. New York WIM Data.
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0 20 40 60 80 100 120 140 160 180 200
i
n
v
e
r
s
e
o
f
s
t
a
n
d
a
r
d
n
o
r
m
a
l
d
i
s
t
r
i
b
u
t
i
o
n
GVW [kips]
63
Figure 5.3. CDF of GVW for 5axle and 11axle trucks
(Nowak, A.S., Laman, J. and Nassif, H., 1994)
Figure 5.4. Percentage of vehicles by number of axles. FHWA WIM Data.
0%
10%
20%
30%
40%
50%
60%
70%
2 3 4 5 6 7 8 9 10 11 12
number of axles
Oregon
Florida
Indiana
New York
64
Figure 5.5. Percentage of vehicles by number of axles.
(Kim, SJ., Sokolik, A.F., and Nowak, A.S., 1997)
If we would like to consider all types of trucks, not only 5axles, the result would
be similar. The mean value of GVW is above 50 kips, Figure 5.6.
Figure 5.6. CDF’s of GVW for all types of vehicles in Oregon.
0%
10%
20%
30%
40%
50%
2 3 4 5 6 7 8 9 10 11 12
43.2%
10.9%
6.5%
22.4%
5.1%
2.6%
1.5%
0.8% 0.5%
6.3%
number of axles
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0 20 40 60 80 100 120 140 160 180 200 220
I
n
v
e
r
s
e
o
f
S
t
a
n
d
a
r
d
N
o
r
m
a
l
D
i
s
t
r
i
b
u
t
i
o
n
GVW [kip]
US 97 Bend
I84 Emigrant Hill
OR 58 Lowell
I5 Woodburn
65
5.3. MODEL BASED ON LEGAL LOAD TRUCKS
The second approach to model live load is based on vehicles called "legal load
types", which are developed in AASHTO. Some of the states use them for rating instead
of the traditional HS20 load. These vehicles were selected to match the federal bridge
formula (known as Formula B) for vehicles up to 80 000 lb. While HS20 was intended
to be appropriate for all span ranges, three legal AASHTO vehicles (Type 3 Unit, Type
3S2 Unit, Type 33 Unit) are supposed to adequately model short vehicles and a
combination of vehicles for short, medium and long spans respectively.
To model traffic jam situations on long span bridge the Type 33 Units have been
placed in a lane with the clearance distance of 10 to 15 ft. Therefore spacing between the
last axle of one truck and first axle of the following truck is 2025 ft. Figure 5.7. Gross
Vehicle Weight of a Type 33 Unit is 80 kips and total length of a Type 33 Unit is 54 ft,
therefore:
80 kips / (54+25) ft = 1.01 kip/ft for clearance distance of 15 ft
80 kips / (54+20) ft = 1.08 kip/ft for clearance distance of 10 ft
Since the value obtained in this way is based on heavy trucks and it is very
conservative, its value can be multiplied by factor 0.75. This approach derives from basic
philosophy used to develop lane load of 0.64 kip/ft.
0.75 x 1.01 kip/ft = 0.76 kip/ft
0.75 x 1.08 kip/ft = 0.81 kip/ft
Figure 5.7. AASHTO LRFD legal load trucks, Type 33 Units.
15.0’ 4.0’
54.0’ 20.0’  25.0’
Type 33 Unit
15.0’ 4.0’ 16.0’ 15.0’ 4.0’
54.0’
Type 33 Unit
15.0’ 4.0’ 16.0’
12 kip 12 kip 12 kip 16 kip 14 kip 14 kip 12 kip 12 kip 12 kip 16 kip 14 kip 14 kip
66
5.4. MODEL BASED ON TRAFFIC JAM SIMULATION USING WIM DATA
The considered WIM data was obtained from NCHRP 1276, described in
CHAPTER 4. The available data served as a basis for simulation of a traffic jam
situation. Starting with the first truck, all consecutive trucks were added with a fixed
clearance distance between them until the total length reached the span length (Figure
5.8). Then, the total load of all trucks was calculated and divided by the span length to
obtain the first value of the average uniformly distributed load. Next, the first truck was
deleted, and one or more trucks were added so that the total length of trucks covers the
full span length and the new value of the average uniformly distributed load was
calculated. The calculations were performed for span lengths 600, 1000, 2000, 3000,
4000, and 5000 ft. Trucks were kept in actual order, as recorded in the WIM surveys.
Clearance distance is assumed to be about 15 ft, while spacing between the last axle of
one truck and first axle of the following truck is 25 ft. Clearance concept is as defined as
in Figure 5.9, and according to literature it varies between 6 and 21 ft. Spacing between
the last axle of one truck and first axle of the following truck is clearance plus distance
from first and last axles to corresponding bumpers, based on the most common 5axle
truck WB20 defined in “AASHTO Geometric Design of Highways and Streets. Only the
most loaded lane was considered. It was assumed that in a traffic jam situation, light
vehicles are using faster lanes, therefore, vehicles of the 13 FHWA category were
omitted.
Figure 5.8. Simulation of trucks moving throughout span length
67
Figure 5.9. Clearance  Gap and Spacing  Headway Concepts
Figure 5.10. Interstate semitrailer WB20
(AASHTO Geometric Design of Highways and Streets)
To perform an extensive number of simulations on the huge databases for each
site and different span lengths, a numerical procedure based on Visual Basic coding were
developed. WIM data, such as weight and length of the trucks in actual order, had to be
inputted to the program. The desired clearance distance was added automatically before
computation. Running the computation was an extremely time consuming process.
Results of the simulations were plotted as a cumulative distribution function
(CDF) of uniformly distributed load for considered span lengths. The CDF’s for all
simulated data are presented in Appendix A, CDF’s of daily maximum combinations in
Appendix B, and CDF’s of weekly maximum combination in Appendix C.
Calculations were performed for one month, three, four, six, nine months or one
year data depending on localization. Even though it is a long period of time, it is small
compared to the actual life time of the structure. The uniformly distributed loads
68
corresponding to longer period of time, 75 years, were calculated by extrapolation of
simulated results. The extrapolated distributions are shown for maximum daily and
weekly combinations (Appendix B and Appendix C). Let N be the number of truck
combinations in time period T and assume that the traffic will remain the same. For T =
75 years, N will be larger 900 times for one month data, 300 times for 3 month, 75 times
for one year data etc. For example, for a site with 400,000 truck combinations monthly,
this will result in N = 360 million truck combinations. The probability corresponding to
N is 1/N. For 360 million, it is 1/360,000,000=2.8x10
9
, which is 5.83 on the vertical
scale of CDF plot. Probability corresponding to extrapolated maximum daily truck is
3.65x10
5
, and to maximum weekly truck is 2.56x10
4
. The number of truck combinations
N, probabilities 1/N, and inverse normal distribution corresponding to 75 years periods
are shown in Table 5.2.
From the results of simulations, the statistical parameters of live load were
obtained. It was noticed that mean value of uniformly distributed load oscillates between
value 0.50 and 0.75 k/ft (Figure 5.12). The value 0.75 k/ft is close to those obtained in
two previous models. The mean daily and weekly maximum can be found in Figure 5.12
and Figure 5.13. For longer spans uniformly distributed load decreases and is closer to
mean value. This observation confirms that for a long loaded span, one heavily
overloaded truck does not have significant influence. This is because the load depends on
a mix of traffic. The bias factors (ratio of mean to nominal) were calculated for the
heaviest 75year combination of vehicles. The 75year uniformly distributed loads were
derived from extrapolated distributions. In Figure 5.16, it can be noticed that the bias
factor values for some sites do not exceed 1.25, which is similar as for short and medium
spans, as shown in the NCHRP Report 368 (1999). It is recommended to use HL93 also
for those long spans (Figure 5.11). To keep bias value below 1.0, it would be necessary to
increase design value of uniformly distributed load to 0.85 k/ft (Figure 5.17). Brides in
localizations with high ADDT and high percentage of overloaded trucks, such as those in
New York, will require development of site specific models. For some sites with very
heavy traffic, the bias factor reaches value 2.0 (Figure 5.16). Therefore, for those bridges
the uniformly distributed load should be higher. It was found that to not exceed bias 1.25,
the uniformly distributed load should be 1.2 kip/ft (Figure 5.18). The heaviest truck
69
combinations were observed on I475, Throggs Neck Bridge in New York. They have
been presented in Table 5.3.
The coefficient of variation is calculated from the slope of transformed CDF.
Figure 5.19 and Figure 5.20 present coefficient of variation of daily and weekly
maximum uniformly distributed load. Daily maximum uniformly distributed load has
more variation due to weekends. Lighter traffic during weekends can be observed in
lower tail of CDF’s. Estimated coefficients of variation were derived from weekly
maximum values, excluding sites from Yew York. Calculated statistical parameters for
uniformly distributed load are summarized in Table 5.1.
Figure 5.11. HL93 proposed for long span bridges
Table 5.1. Statistical parameter for proposed uniformly distributes live load
span length Bias CoV
600 – 1000 ft 1.25 0.10
> 1000 ft 1.20 0.08
0.00
0.50
1.00
1.50
0 1000 2000 3000 4000 5000
e
q
u
i
v
a
l
e
n
t
U
D
L
[
k
i
p
/
f
t
]
loaded length [ft]
AASHTO HL93
0.68 kip/ft
70
Table 5.2. Summary of simulated data
State Site ID Route Number of truck
combinations
Time
period
75years
probability
FL 9916 US29 247,449 1 year 5.39x10
8
FL 9919 I95 222,368 3 months 1.50x10
8
FL 9927 SR546 225,868 1 year 5.90x10
8
FL 9936 I10 188,990 1 year 7.05x10
8
IN 9512 I74 167,630 1 month 6.63x10
9
IN 9534 I65 266,333 1 month 4.17x10
9
IN 9544 I80/I94 406,418 1 month 2.73x10
9
OR Woodburn I5 552,390 4 months 8.04x10
9
OR Emigrant Hill I84 213,019 4 months 2.09x10
8
OR Lowell OR 58 51,406 4 months 8.65x10
8
OR Bend US 97 59,225 3 months 7.50x10
8
NY 9121 I81 300,500 6 months 2.22x10
8
NY 2680 8 45,030 9 months 2.23x10
7
NY I495WB 43,200 1 month 2.57x10
8
NY I495EB 52,618 1 month 2.11x10
8
sum 3,042,444
71
Table 5.3. Heaviest truck combinations for 600 ft on I495 WB
number
of axles
W L W L W L W L W L W
Total
length
GVW
[kip] [ft] [kip] [ft] [kip] [ft] [kip] [ft] [kip] [ft] [kip] [ft] [kip]
5 20.04 17.39 41.41 4.27 35.68 22.31 38.77 4.27 37.44 0.00 0.00 73.3 173.3
2 27.97 19.69 32.82 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 44.7 60.8
5 33.26 17.72 36.12 4.27 33.04 23.62 25.11 3.94 28.41 0.00 0.00 74.6 155.9
3 30.40 16.08 26.87 4.27 27.75 0.00 0.00 0.00 0.00 0.00 0.00 45.4 85.0
2 28.19 22.97 42.73 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 48.0 70.9
5 35.68 15.42 37.44 4.27 38.55 34.45 33.04 3.94 32.38 0.00 0.00 83.2 177.1
5 33.48 12.14 43.83 3.94 38.99 28.87 43.61 3.94 39.87 0.00 0.00 74.0 199.8
5 38.11 13.12 25.33 4.27 20.48 31.50 19.60 3.94 23.57 0.00 0.00 77.9 127.1
2 33.26 15.75 39.87 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 40.8 73.1
561.9 1123.1 2.00 k/ft
2 11.23 18.70 21.15 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 43.7 32.4
5 20.04 17.39 41.41 4.27 35.68 22.31 38.77 4.27 37.44 0.00 0.00 73.3 173.3
2 27.97 19.69 32.82 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 44.7 60.8
5 33.26 17.72 36.12 4.27 33.04 23.62 25.11 3.94 28.41 0.00 0.00 74.6 155.9
3 30.40 16.08 26.87 4.27 27.75 0.00 0.00 0.00 0.00 0.00 0.00 45.4 85.0
2 28.19 22.97 42.73 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 48.0 70.9
5 35.68 15.42 37.44 4.27 38.55 34.45 33.04 3.94 32.38 0.00 0.00 83.2 177.1
5 33.48 12.14 43.83 3.94 38.99 28.87 43.61 3.94 39.87 0.00 0.00 74.0 199.8
5 38.11 13.12 25.33 4.27 20.48 31.50 19.60 3.94 23.57 0.00 0.00 77.9 127.1
564.9 1082.4 1.92 k/ft
5 38.99 14.11 35.68 4.27 31.94 17.39 28.85 3.94 31.28 0.00 0.00 64.8 166.7
5 37.00 12.14 40.09 4.27 33.04 32.48 30.84 3.94 37.22 0.00 0.00 77.9 178.2
5 37.67 10.50 22.47 4.27 24.01 26.57 15.42 10.17 18.28 0.00 0.00 76.6 117.8
2 29.07 16.08 38.11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 41.1 67.2
3 43.61 18.70 21.37 4.59 27.53 0.00 0.00 0.00 0.00 0.00 0.00 48.3 92.5
2 34.14 21.33 26.21 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 46.4 60.4
5 40.75 13.45 30.62 4.27 25.33 30.51 22.03 3.94 22.47 0.00 0.00 77.3 141.2
5 35.46 10.50 22.03 4.27 20.70 22.64 20.04 4.27 20.48 0.00 0.00 66.7 118.7
3 36.56 19.69 37.00 4.27 31.06 0.00 0.00 0.00 0.00 0.00 0.00 49.0 104.6
548.1 1047.4 1.91 k/ft
72
Figure 5.12. Mean value of uniformly distributed load
Figure 5.13. Daily maximum mean value of uniformly distributed load
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
I84 Emigrant Hill
US 97 Bend
I5 Woodburn
Florida 9936
Florida 9916
Florida 9927
Florida 9919
Indiana 9511
Indiana 9512
Indiana 9534
Idiana 9544
NY I495 EB
NY I495 WB
NY 2680
NY 9121
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
I84 Emigrant Hill
US 97 Bend
I5 Woodburn
Florida 9936
Florida 9916
Florida 9927
Florida 9919
Indiana 9512
Indiana 9534
Idiana 9544
NY I495 EB
NY I495 WB
NY 2680
NY 9121
73
Figure 5.14. Weekly maximum mean value of uniformly distributed load
Figure 5.15. Bias (mean max 75 year to nominal value of UDL)
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
I84 Emigrant Hill
US 97 Bend
I5 Woodburn
Florida 9916
Florida 9936
Florida 9927
Florida 9919
NY 2680
NY 9121
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
0 1000 2000 3000 4000 5000
B
i
a
s
span length
OR US97
OR I5
OR 58
OR I84
FL 9919
FL 9927
FL 9936
IN 9534
IN 9544
IN 9512
74
Figure 5.16. Bias (mean max 75 year to nominal value of UDL)
Figure 5.17. Bias (mean max 75 year to nominal value of UDL)
assumed designed UDL of 0.85 k/ft
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
0 1000 2000 3000 4000 5000
B
i
a
s
span length
FL 9916
NY 9121
NY 2680
NY I495 EB
NY I495 WB
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
0 1000 2000 3000 4000 5000
B
i
a
s
span length
OR US97
OR I5
OR 58
OR I84
FL 9919
FL 9927
FL 9936
IN 9534
IN 9544
IN 9512
75
Figure 5.18. Bias for heavily loaded localizations, assumed designed UDL of 1.25 k/ft
Figure 5.19. Coefficient of variation of daily maximum uniformly distributed load
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
0 1000 2000 3000 4000 5000
B
i
a
s
span length
FL 9916
NY 9121
NY 2680
NY I495 EB
NY I495 WB
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
OR I84 Emigrant Hill
OR US 97 Bend
OR I5 Woodburn
Florida 9916
Florida 9936
Florida 9927
Florida 9919
Indiana 9512
Indiana 9534
Indiana 9544
NY I495 EB
NY I495 WB
NY 2680
NY 9121
76
Figure 5.20. Coefficient of variation of weekly maximum uniformly distributed load
Figure 5.21. Proposed coefficient of variation of uniformly distributed load
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
OR I84 Emigrant Hill
OR US 97 Bend
OR I5 Woodburn
Florida 9916
Florida 9936
Florida 9927
Florida 9919
NY 2680
NY 9121
0.00
0.05
0.10
0.15
0.20
0.25
0 1000 2000 3000 4000 5000 6000
U
D
L
[
k
/
f
t
]
span length [ft]
OR 58 Lowell
OR US 97 Bend
OR I5 Woodburn
Florida 9916
Florida 9936
Florida 9927
Florida 9919
77
CHAPTER 6
MULTIPLE PRESENCE
6.1. INTRODUCTION
Determination of live loading for multiple traffic lanes is very important for
appropriate bridge design. Each traffic lane can be loaded with different live load. The
more traffic lanes the more difference in distribution of loading. Trucks tend to use right
lanes loading them heavily, while passenger cars use faster left lanes. Structural
components are strongly influenced by the location of trucks on the bridge. Those
carrying the right lane of traffic are usually subjected to more load cycles and fatigue
than the components closer the left lane. However, it has to be remembered that traffic
can be deviated and truck can be directed to left lanes, for example during maintenance
works on bridge. Presence of truck on multiple traffic lanes at the same time is critical
from the bridge design point of view.
In this chapter, a short review of current studies on presence of multiple trucks,
study of video recording of traffic jam situations, and discussion on different approaches
to multilane reduction factors are presented. Multiple presence factors applied in
international codes are discussed in CHAPTER 2.
78
6.2. STUDIES ON PRESENCE OF MULTIPLE TRUCKS
In 1993 Andrzej S. Nowak, Hani Nassif, and Leo DeFrain performed a study on
the occupation of road lanes and presence of multiple truck. For the study, the database of
over 600 000 trucks on two lanes in each direction was collected using weighinmotion
equipment. It was found that 7090% of trucks use the right lane (Table 6.1), and 6570
percent of trucks are 5axle trucks. The researchers had also found out that less than 2%
of trucks appear simultaneously with another truck in the lane, 48% side by side in
tandem or behind, with distance between front axles less than 50 ft (Table 6.1).
Table 6.1. Presence of multiple trucks and their location on the road lanes
tandem behind tandem behind
I94 1582 151 14 28 48 2073 349 14 86 93
91,3% 8,7% 0,8% 1,6% 2,8% 85,6% 14,4% 0,6% 3,6% 3,8%
U.S.23 1685 371 24 40 76 1247 601 2 38 34
82,0% 18,0% 1,2% 1,9% 3,7% 67,5% 32,5% 0,1% 2,1% 1,8%
side by side
right left
Interstate
highway
Eastbound Westbound
right left in lane
side by side
in lane
The results were later confirmed by the investigation by Gindy and Nassif (2006).
Data used in this study was collected over an 11year period between 1993 and 2003 by
the New Jersey Department of Transportation. The study gives detailed analysis of the
relation between multiple presences of trucks and four parameters: truck volume, area
type, road type, and bridge span length. An increase in truck volume results in an increase
of all multiple presence cases and a decrease in the frequency of single loading events
(Figure 6.2). The area (urban or rural) and vicinity of industry affect the frequency of
multiple truck presence. Heavier volume sites tend to be located in urban areas, and as a
consequence more cases of multiple trucks can be observed. Increasing bridge spans also
gives more opportunities for trucks to occur simultaneously (Figure 6.3). The frequency
of staggered events increases faster for shorter spans and at a steadier pace for longer
spans. Span length has almost no influence on the frequency of sidebyside trucks.
79
Assumptions of traffic patterns made for multiple presence analysis following
(see Figure 6.1):
 single – only one truck is present on the bridge
 following – two trucks on the same lane with a varying clearance distance
 sidebyside – two trucks in adjacent lanes with an overlap at least onehalf the body
length of the leading truck
 staggered – two trucks in adjacent lanes with an overlap at less than onehalf the body
length of the leading truck
Figure 6.1. Traffic loading pattern used for multiple truck presence statistics.
Comparing results of the study made by Andrzej S. Nowak, Hani Nassif, and Leo
DeFrain in 1993 and the study made by Mayrai Gindy and Hani H. Nassif in 2006, some
divergences can be observed. Some of them can be caused by the differences in the
definition of “sidebyside” and “staggered” cases. Trucks which are considered
“staggered” in one case can be considered as “side by side” in the other. However,
joining those two cases and making a sum of those two values, we obtain relatively close
results, 7.3 % for the study from 1993 and 6% for study from 2006. Regarding the
occurrence of following trucks, values obtained in 1993 vary between 0.1 and 1.2%, and
values obtained in 2006 vary between 1 and 8%. Those values cannot be compared
80
because of differences in the assumptions and definition of “following” trucks.
Furthermore, the bridges taken into consideration in the study from 2006 were longer,
which increases the probability of this occurring. The probability that the trucks will
occur as following trucks increases with the span length.
In the report NCHRP 1276 (2008), it is stated that multipresence probabilities
for permit trucks are different from those for normal traffic. The likelihood of permit
trucks exceeding the authorized weight as well as the likelihood of the presence of
multiple permit trucks is reduced.
Figure 6.2. Variation of multiple truck presence statistics with respect to truck volume.
Gindy and Nassif (2006).
81
Figure 6.3. Variation of multiple truck presence statistics with respect to bridge span
length. Gindy and Nassif (2006).
6.3. MULTIPLE PRESENCE OF TRUCKS BASED ON THE VIDEO FILES OF
TRAFFIC
FHWA has provided a DVD including video files monitoring seven traffic
situations in different sites, at different times and days of the week. The localization of
the sites was not specified. Total time of all sixteen video recordings is 2 hours 6 minutes
and 30 seconds. The list of video recordings is following:
 (1) 05.30.2008, Friday, 12.17 pm (33min44sec)
 (2) 05.29.2008, Thursday, 6.08pm (4min43sec)
82
 (3) 05.22.2008, Thursday, 11.22am (0min22sec)
 (4) 05.22.2008, Thursday, 11.20am (0min22sec)
 (5) 05.14.2008, Wednesday, 12.52pm (4min51sec)
 (6) 05.14.2008, Wednesday, 12.50pm (2min26sec)
 (7) 05.14.2008, Wednesday, 12.48pm (1min47sec)
 (8) 05.14.2008, Wednesday, 12.44pm (3min59sec)
 (9) 05.06.2008, Tuesday, 7.59am (3min49sec)
 (10) 05.01.2008, Thursday, 12.20pm (1min04sec)
 (11) 04.04.2008, Friday, 4.42pm (9min28sec)
 (12) 04.04.2008, Friday, 4.43pm (8min54sec)
 (13) 04.04.2008, Friday, 4.43pm (0min36sec)
 (14) 04.04.2008, Friday, 4.43pm (19min26sec)
 (15) 04.04.2008, Friday, 4.43pm (22min53sec)
 (16) 04.04.2008, Friday, 4.43pm (8min06sec)
The recordings show dense traffic jam situations, some of them being the result of
traffic accidents. They allow for making some observations and conclusions regarding
traffic patterns and the presence of multiple trucks moving at a crawling speed. This is
the critical case from the point of view of live loading on bridges. Despite the fact the
recordings are taken on highways, the recorded situations can be related to bridges as
well. One of the most important observations is that even in very dense traffic it is very
common to observe cars or pickup among heavy vehicles (Figure 6.4).
83
Figure 6.4. Video 10, time: 00:00:58
The video file number 1 contains the longest and the most interesting material. It
has a registered traffic accident on a highway having four lanes in one direction. The
accident takes place on the second lane (counting from the external side). The second and
the third lanes remain completely stopped by crushed cars and emergency vehicles for
approximately half an hour. Passing cars are using the first and the fourth lanes. For a
short period of time three lanes are blocked, and the passing cars can use only the fourth
lane, that intensifies jampacked traffic. For the majority of time we can observe that the
moving lanes contain a mixture of trucks and cars. However, we can also observe some
situations with multiplepresence of trucks occupying three or four lanes at the same time
(Figure 6.5 and Figure 6.6). We can also observe a situation when one lane is almost
exclusive occupied by trucks (Figure 6.6). Those cases should also be taken into
consideration during the evaluation of design live load. There is no information about
trucks’ weight. However, it can be assumed that only a limited number of the trucks are
correlated, and while some of them are fully loaded some percentage of them can be
empty.
84
Figure 6.5. Video 1, time: 00:05:28
Figure 6.6. Video 1, time: 00:18:36
85
Some of the video files show traffic jam situations at roads junction that cannot be
representative. The number of cars among trucks is increased because some of them are
entering or exiting the highway (Figure 6.7 and Figure 6.8). This observation calls our
attention to the importance of the appropriate selection of sites.
Figure 6.7. Video 2, time: 00:00:15
86
Figure 6.8. Video 8, time: 00:00:16
Video recordings from 11 to 16 show traffic situation caused by the same
accident. Three lanes are blocked and the moving vehicles are using only one lane.
However, warning signs posted adequately ahead result in the vehicles forming one lane
and do not cause multilane traffic jam situation in the vicinity of the place of the accident.
An ordered one lane of traffic is a mixture of trucks and cars. This situation does not
allow for the observation of multiple truck presence situations.
6.4. APPROACHES TO MULTILANE REDUCTION FACTORS
There are many approaches to multilane reduction factors. International codes
vary significantly in this matter (see CHAPTER 2), and all of them are much more
simplified than actual situation.
Most of the design codes decrease the value of uniformly distributed load as the
number of traffic lanes increases. The load value is the same on all traffic lanes, Figure
6.9.
87
Figure 6.9. Multilane load in design codes AASHTO LRFD Code (2007),
OHBDC (1991), CAN/CSAS600 [2000], and ASCE (1981).
According to the Eurocode one of the lanes is loaded more than the others, Figure
6.10.
Figure 6.10. Multilane load in Eurocode 1.
The observations indicate that the actual traffic is distributed differently for each
lane of traffic, Figure 6.11.
Figure 6.11. Multilane load in actual observation.
88
To reflect actual traffic situation would be very difficult and time consuming for
designers. Therefore, the approach applied in the Eurocode seems the most practical one.
Each of the lanes should be considered as the most loaded one, while the other lanes carry
equal loading.
6.5. CONCLUSIONS
Presence of multiple trucks depend on factors such as: truck volume (light,
average, heavy) area type (urban/rural), road type (major/minor), bridge span length, law
enforcement, and traffic flow control, which can cause heavy truck queues. Therefore,
multilane reduction factors could be very site specific.
Based on video recordings of traffic it was confirmed that for the majority of time
we can observe that the moving lanes contain a mixture of trucks and cars. However,
situations when one lane is almost exclusively occupied by trucks or trucks occupy three
or four lanes at the same time are also possible. Multiple reduction factors for design live
load should account for those the most critical loading cases.
In the available WIM database, the vehicles of 13 categories have not been
registered. Therefore, it does not allow for simulations and derivation of multilane factors
for all traffic lanes. Simulation of the traffic on the most loaded lane was possible with
assumption, that it is occupied exclusively by trucks. In the traffic jam situations, the
passenger vehicles are assumed to move to faster lanes.
It was concluded that the multilane reduction factors have to be an object of
additional extensive studies. They have to account for intensive traffic jam situations, as
those registered in the video recordings. As well, different distribution of loading on
multiple traffic lanes has to be considered.
Since no new multilane reduction factors were proposed, those from the current
AASHTO Code are used in this dissertation.
89
CHAPTER 7
DYNAMIC FACTOR
7.1. INTRODUCTION
The scope of this chapter is studying the origin and adequacy of the application of the
dynamic load factor in bridge design. There is considerable variation in the treatment of
dynamic load effects by bridge design codes in different countries (see CHAPTER 2). The
most common approach is to apply dynamic response as a fraction or multiple of the
response that would be obtained if the same forces or loads were applied statically. The
objective of this simple approach is to not increase complexity for the designer. This is the
approach specified in the current AASHTO Code. The live load model itself does not
account for dynamics, but the dynamic amplification is added additionally as a percentage to
static effects.
In this chapter there is a short review of the research studies. As well, a developed
exemplary vehiclebridge interaction model and the derivation of dynamic factor is
presented. In the modeling a finite element software ABAQUS was used. Threeaxle
AASHTO truck HS20 travelling over a 120 ft steel girder bridge is modeled. The truck is
assumed to travel with the velocity of 40 miles per hour and with the crawling speed. The
final result is comparison of the static the dynamic deflections, and derivation of dynamic
factor for this specific case.
90
7.2. STUDIES ON PARAMETERS AFFECTING DYNAMIC BRIDGE RESPONSE
The dynamic response of a bridge to a crossing vehicle is a complex problem affected
by the dynamic characteristics of the bridge, the vehicle, and by the bridge surface
conditions. Many of the parameters interact with one another, further complicating the issue.
Consequently, many research studies have reported seemingly conflicting conclusions. Based
on the review of past research, the effects of various parameters on the dynamic response of
bridges to vehicular loading are discussed in this paragraph.
Bridge Fundamental Frequency
The fundamental frequency of vibration for a bridge due to vertical loading has a
significant effect on the dynamic response. If the frequencies of the bridge and vehicle
converge, the dynamic response induced may be large. A majority of the fundamental
frequencies for typical bridges are in the range of 2 to 5 Hz (Figure 7.1), which corresponds
to the body bounce response frequency range of a truck.
Figure 7.1. Distribution of fundamental bridge frequencies
(Cantieni 1984)
91
Field measurements and values obtained from analytical modeling show relation
between frequencies and bridge span (Figure 7.2 and Figure 7.3).
Figure 7.2. Fundamental frequency versus span length
(Cantieni 1984)
Figure 7.3. Fundamental frequency versus span length
(Paultre 1992)
92
Bridge Damping
From basic dynamic principles, higher levels of damping reduce the dynamic
response in bridges and low levels of damping in a bridge are expected to result in high
dynamic amplification. However, damping affects impact differently at different locations
within the bridge as a result of varying modal contributions (Huang, Wang, and Shahawy,
1992). Damping values for bridges obtained from field testing can also vary considerably
based on the method of testing, level of loading, and different methods used for evaluating
damping. Reported values of damping for different types of bridges are as follows:
 concrete bridges  2 to 10 % (Tilly, 1978)
 steel bridges  2 to 6 %( Tilly, 1978), 0.4 to 1.3 % (Billing, 1984)
 composite steelconcrete bridges  5 to 10 % (Tilly, 1978)
 prestressed concrete bridges  1 to 2.2 % (Billing, 1984)
 timber bridges  3 to 4 % (Ritter, 1995)
Roadway Roughness and Approach Condition
The bridge approaches and roughness of the roadway surface have a significant
influence on the magnitude of the dynamic response. Not only do the impact forces increase
for increased roughness, but also vehicle speed affects the influence of roughness. The faster
vehicle speed has greater impact on rougher surfaces than on better maintained ones. The
results of the study by Wang, Shahawy, and Huang (1993) can be seen in Figure 7.4.
93
Figure 7.4. Impact factor versus vehicle speed and road surface condition
(Wang, Shahawy, and Huang 1993)
Experimental tests have also shown that the most severe wheel impact forces are
likely to occur adjacent to the bridge approaches, i.e., shortly after a vehicle enters the bridge
(Tilly 1978). In many experimental investigations, wooden planks were placed in the path of
the test vehicle. It was supposed to represent surface irregularities such as dropped objects or
packed snow on the roadway. Dynamic response was higher with the planks and the planks
were exciting wheel hop in the test vehicles, although excitation of the higher vibration mode
associated with wheel hop is also speed dependent.
Vehicle Speed
For most heavy trucks, natural frequencies of the vehicle typically occur in two
frequency ranges: between approximately 2 and 5 Hz for the "body bounce" response and
between approximately 10 and 15 Hz for the "wheel hop" response. However, depending on
vehicle speed, roadway surface irregularities may be effective in exciting both modes of
response (Cantieni 1983).
94
Weight of Vehicles
Many studies have shown that as the weight of the crossing vehicle increases, the
magnitude of the dynamic response expressed as a percentage of the static load decreases.
The explanation of this fact is that when the dynamic forces increase with increasing
vehicle weight, the static load increases more rapidly with increasing weight. Thus, the
impact ratio of dynamic force to live load decreases with increasing vehicle weight and
impact factors obtained from the measured dynamic response of lightly loaded vehicles
will be relatively large.
Number of Vehicles
The dynamic load factors associated with multiple vehicles are lower than those for
single vehicles. This is most likely because the total static load is larger (similar to having a
heavier vehicle) compared to the associated dynamic load, and the dynamic responses from
the two individual vehicles are likely to be at least somewhat out of phase with each other.
Vehicle suspension
The vehicle frequency ranges are a function of the suspension systems. The body
bounce frequencies in vehicles with air suspensions are lower than those for steel leafspring
suspensions, with measured frequencies in the 1.5 to 2 Hz range. Worn dampers in the
suspension systems also dramatically increased the dynamic wheel forces.
In Figure 7.5 the influence of suspension is presented. On one deflection trace the
vehicle had its normal suspension characteristics, in the other the springs were blocked so
that the truck rode directly on the axles. The increase in response is evident for the unsprung
condition.
95
Figure 7.5. Effects of vehicle suspension on the measured bridge response
(Biggs and Suer 1955)
Bridge Span
There are conflicting opinions regarding dynamic response and bridge span. While
some investigations have shown a general trend of decreasing impact in conjunction with
increasing span (Figure 7.6, Fleming and Romualdi 1961), other investigations have
concluded that considerable scatter exists in the results and there is poor correlation of impact
and span (Figure 7.7, Cantieni 1983). Some researchers have concluded that impact is not a
function of bridge span (Coussy et al. 1989). However, it should be concluded that as span
length influences bridge fundamental frequency, it also indirectly influences bridge
dynamics.
96
Figure 7.6. Impact versus span length (Fleming and Romualdi 1961)
Figure 7.7. Impact versus span length (Cantieni 1984)
97
Bridge Type and Geometry
Most of the studies on dynamic response were performed on simplespan multigirder
bridges. The investigations concluded that the total number of girders has little influence on
the maximum impact factors for each girder and that the impact at the interior supports was
larger than at external locations. The dynamic response of simplespan bridges is higher than
that for similar continuous bridges.
Cablestayed and suspension bridges are more complicated to assess than for
beam/girder bridges, mostly due to the influence of cables’ dynamics. Dynamic response
quantities are sensitive to damping, which is difficult to determinate in these types of bridges,
and may be different for different vibration modes. In analytical investigations it was found
that, with a good road surface, impact factors were generally less than 0.20. However, for
rough surfaces, impact forces increased dramatically. (Khalifa 1992)
Dynamic response of continuous and cantilever thinwalled box girder bridges under
multivehicle loading was analytically investigated by Wang, Huang, and Shahawy (1996)
and Huang, Wang, and Shahawy (1995a). It was found that the vibration characteristics of
the continuous and cantilever box girder bridges are quite different. For cantilever bridges,
the most important factor affecting impact is the vehicle speed, and they are much more
susceptible to vibration than continuous bridges. This is due to the abrupt change in loading
due to span discontinuities, when no support exists between cantilevers. For continuous
bridges, both vehicle speed and surface roughness are significant. End diaphragms were
found to provide lateral support and significantly reduce the response of the box girder
bridges. The beneficial effect of a midspan diaphragm is relatively small.
Dynamic behavior depends on curvature of the bridge. It was found that the dynamic
response in horizontally curved bridges is influenced by centrifugal accelerations, thus,
vehicle speed is particularly important. Impact forces are higher in the outer elements of the
curved bridges. Impact forces are insensitive to curvature for radii greater than 4 000 ft (1219
98
m) and markedly influenced by curvature for radii less than 800 ft (244 m). Research done by
Galdos et al. (1993) and Schilling et al. (1992) and Huang, Wang, and Shahawy (1995b).
7.1. BRIDGEVEHICLE INTERACTION MODEL AND DERIVATION OF
DYNAMIC FACTOR
7.1.1. Bridge and Vehicle Model
To study vehiclebridge interaction, with the use of ABAQUS 6.6.1 software a finite
element model of bridge and vehicle has been developed.
The bridge chosen for the analysis is a 120 ft steel girder bridge. It is modeled with 3D shell
elements:
Steel girders:
 five steel W40x264 (profile properties: area A = 77.60 in2, depth of the section
d=40 in, web thickness: tw = 0.96 in, flange width bf = 11.93 in, flange thickness
tf =: 1.73 in, moment of inertia Ixx=19400 in4)
 spaced 64 in
 fy = 60 ksi
 steel diaphragms every 30 ft
 2360 elements, 2715 nodes
Concrete slab:
 thickness: 7.5 in
 width:312 in
 f
c
’=:8 ksi
 4464 elements, 4640 nodes
Supports:
 left support  pinned
 right support  roller
Connection between girders and slab: tied (compatibility of all degrees of freedom)
Bridge model meshed with S4R elements is shown in Figure 7.8.
99
Figure 7.8. Meshed model of the bridge
A vehicle model was developed based on threeaxle AASHTO HS20 truck, which is
the design vehicle in the AASHTO Specifications. Two cases are considered, the truck HS
20 is assumed to travel with the velocity of 40 miles per hour and with the crawling speed.
Figure 7.9, shows FEM model of moving masses. To simplify the analysis each wheel is
modeled with one DOF in the vertical direction. Detailed model would include a seven
degree of freedom system. Tractor and semitrailer would have individually assigned two
DOFs corresponding to: vertical displacement (y
i
) and rotation about the transverse axis (θ
i
).
Moreover each wheelaxle set would be provided with one DOF in the vertical direction (y
i
).
Five sprung masses would be: the tractor, semitrailer, steer wheelaxle set, tractor wheelaxle
set, and trailer wheel axle set. A more detailed model could include rotations about
longitudinal axis (Φ
i
).In this case the total independent DOFs would be eleven.
Figure 7.9. Truck model in FEM
100
7.1.1. BridgeVehicle Interaction Model
There are two main sources of vibrations induced by the vehicle into the bridge. One
of them is the settling of the approach slab of the bridge. The bump at the bridge entrance
causes so called “wheel hop” that could be approximated by impulse loading. A vehicle
vibrates with a frequency between 10 and 15 Hz. Transmitted force is significant, however it
fades relatively fast converging into “body bounce” that occurs at lower frequencies. The
other source of vibrations is an undulating roadway surface. The condition of driving a
vehicle over an undulating roadway surface can be approximately idealized as an SDOF
system under harmonic loading provided the roadway varies as a sine wave. These kinds of
vibrations are called “body bounce” and usually occur at frequencies between 2 and 5 Hz.
The loading is characterized by the roughness amplitude, roughness wavelength, and vehicle
speed. If the shockabsorbing elements of the vehicle suspension system are worn, then the
vehicle's damping is fairly low and the response is quite large. Under such conditions, nearly
resonant response can develop.
Since the truck model is simplified and the scope of the project requires comparison
of the deflections in the middle of the span, the interaction model includes only “body
bounce”, which depends on the relation between suspension system and the truck. Tire
stiffness and “wheel hop” response are neglected.
The vertical interaction force acting on a bridge consists of the static interaction force
F
w
and the variation of the interaction force ∆F
w
.
F
b
= F
w
+ ∆F
w
.
F
w
= M g
)) ( ) ( ( )
) (
(
1
1
t y t y k
dt
dy
dt
t dy
c F
i i
− ⋅ + − ⋅ = ∆
The interaction force is approximated by a sinusoidal shape, as shown in the
Figure 7.10. The force F
w
transmitted by the rear wheels is 16 kips and its variation ∆F
w
is
equal to ±10%. The front wheels carrying 4 kips each are also assumed to produce force
variation ±10%. The effects of damping are small, and they will be neglected in the analysis.
101
Figure 7.10. Vehiclebridge interacting force.
Figure 7.11 is a plot from ABAQUS that shows forces being applied to the bridge versus
time. It can be noticed when the following truck axes are entering and leaving the bridge.
Their amplitudes are interfering and adding to each other. Those forces could also be
canceling each other. Such a situation of adding amplitudes was simulated on purpose, to
obtain the maximum dynamic deflections in the middle span.
Figure 7.11. Force due to moving truck versus time. Plot from ABAQUS.
0 1 2
0
5
10
15
20
t
102
7.1.1. Results of analysis
Mode shapes obtained in the analysis are shown in Figure 7.12 (bending modes) and
Figure 7.13 (torsion modes). The natural frequencies are following:
 first bending mode 2.12 Hz
 second bending mode: 8.19 Hz
 third bending mode: 17.10 Hz
 first torsion mode: 3.17 Hz
 second torsion modes: 9.09 Hz
Figure 7.12. Bending modes
Figure 7.13. Torsion modes.
103
The deflection due to dead load, weight of the concrete slab and the steel profiles, is 2.74 in.
Maximal deflection due to moving truck is 0.69 in (Figure 7.14). It corresponds to the truck
being located almost, but not in the span center.
Figure 7.14. Maximal deflection due to moving truck
Figure 7.15 and Figure 7.16 show deflections in the middle of the span due to
moving and stationary trucks. Maximum deflection due to moving truck (0.69 in) versus
maximum deflection due to stationary truck (0.65 in) shows 6% difference.
104
Figure 7.15. Deflection due to a truck moving 40mil/hr versus time.
Figure 7.16. Deflection due to a truck moving at crawling speed versus time.
105
7.2. CONCLUSIONS
The study of the topic and the developed FEM model allowed for estimating the
magnitude of the dynamic load factor and to draw some conclusions. It was concluded that
the current dynamic load factor of 0.33 is too high for bridges with longer spans. It may be
applicable in short bridges, when vibration due to “wheel hop” on the approach slab is
significant. However, for longer bridges where the influence of the approach slab decreases
and vibrations of many vehicles interfere with each other, the dynamic load factor could be
smaller. Even for the exemplary case of medium span bridge presented in this thesis the
dynamic load factor is only 6%. The FEM model was built using ABAQUS 6.6.1 software.
Finite element problem modeling, including moving load and defined interaction between
surfaces, are non linear and very time demanding. To be adequate and draw more
conclusions further studies should be performed, including a wide range of bridge types,
spans and roadway conditions. Also, a more elaborate truck model should be performed,
accounting for both “body bounce” as well as “wheel hop”. Moreover, analytical studies
should be confirmed with field tests on the representative structures.
In this dissertation, a traffic jam situation was assumed to develop the live load
model. Therefore there is no allowance for dynamic. However, to not introduce confusion
among designers it is recommended do keep dynamic factor as it is for short and medium
span bridges, which results in very small value for long span bridges.
106
CHAPTER 8
RELIABILTY ANALYSIS OF SUSPENSION BRIDGE
8.1. RELIABILITY ANALYSIS PROCEDURE
Reliability analysis was performed to verify the live load model for long span
bridges. The reliability procedure includes the following steps:
1. Selection of a representative bridge and its component
A representative suspension bridge and the structural element that is the most
influenced by live load were selected.
2. Limit state function
The limit state function was defined as the exceeding of the ultimate bending
moment capacity by the crosssection and Strength I combination.
3. Nominal resistance model
In order to find nominal resistance, software for the tower cross section was
developed.
4. Reliability resistance models
The material, fabrication and professional factors were established. Evaluation
of material, fabrication and professional factors was based on the database
available in the literature of and described by Nowak et al. (2008) and
Ellingwood et al. (1980). Based on the developed software for nominal
107
resistance and the Monte Carlo simulation method, the statistical parameters
for resistance were obtained.
5. Load model
The threedimensional FEM model of a bridge was created using Robot
Millennium software. The model was based on the actual Cooper Bridge
design made by PB World. Cross sectional axial force and bending moment
along the tower height due to dead load and live load were derived. The
statistical models for load components are defined.
6. Reliability Indices
Reliability indices were calculated in order to assess how they are influenced
by the increase in the values of live load. Reliability analysis was performed
for the several forces possible and moment in the bridge tower due to live load
cases.
8.2. SELECTION OF REPRESENTATIVE STRUCTURE, ELEMENT AND
LIMIT STATE FUNCTION
The Cooper River Bridge in South Carolina was chosen to be a representative
long span bridge for this analysis. It is a suspension bridge designed by PB Word in 2001.
For long span bridges dead load is the main loading. Dead load is also critical for
most of the structural components, such as the deck and cables. For the scope of this
dissertation, the bridge component that is the most influenced by live loading had to be
selected. The bridge tower was chosen to be such an element. Live load from all spans is
transferred through cables to the bridge tower. When not all spans are loaded evenly, the
bending moment due to live load is much higher than the bending moment due to dead
load. The bending moment due to dead load of opposite spans almost negate each other.
The limit state function was defined as the exceeding of the ultimate bending
moment capacity by the crosssection. The tower crosssection just above the deck level
was selected.
108
8.3. NOMINAL RESISTANCE
The nominal resistance can be represented by interaction diagram of force and
moment for the eccentric loaded bridge tower. To plot interaction diagram, representing
all possible cases of force and bending moment combinations, a MathCad software was
used. The mathematical description of the behavior for the exemplary bridge tower was
developed based on procedure for columns described by Lutomirski (2009). The
assumptions are based on linear strains in a distribution over the cross section and the
mechanical behavior of reinforcement. The procedure takes into account the geometry of
the tower, all layers of the reinforcement in the crosssection and the characteristics of the
materials (steel and concrete).
The first step in the procedure is the calculation of the initial compression block.
The initial compression block represents the compression over the full crosssection and
the lowest layer of the reinforcement yields due to compression. It means that strains in
concrete at the top of cross section reach value of 003 . 0 =
m
ε , while the strains in the
lowest reinforcement bar represent yielding strains due to compression in steel
00207 . 0 = =
y s
ε ε (Figure 8.2). The position of the neutral axis can be calculated based
on the linear strain distribution and strain compatibility assumptions. As a result of the
first step the size of the compression block is much bigger than the size of the cross
section. The size of the initial compression block can be calculated as follows:
y m
I m
I
d
a
ε ε
ε
β
−
⋅
=
1
(8.1)
where:
d
I
is the distance of the lowest layer of the reinforcement to the top of the crosssection.
The position of the neutral axis in the initial step can be calculated from:
1
β
I
I
a
c =
(8.2)
109
Figure 8.1 Distribution of strains for the pure axial loading
The characteristic point of end of compression control zone and beginning of
tension control zone call balance failure point can be derived. It happens when the strains
in the bottom layer of reinforcement reaches the yielding strains for the steel ε
s
=ε
y
(Figure 8.2). The size of the compression block in balance failure is represented:
y m
I m
B
d
a
ε ε
ε
β
+
⋅
=
1
(8.3)
Figure 8.2 Distribution of strains distribution for the balance failure point B, the end of
the compression control zone
β
ε
ε
110
The entire force and moment interaction diagram is calculated using the
decreasing size of the compression block of concrete, from the initial
I
a up to the point
when the compression block does not exist ( 0 = a ). For each reinforcement layer and for
each size of the compression block the strains in reinforcement are computed from the
equation:


¹

\

− ⋅ =
) (
1 ) , (
a c
d
a i
i
m s
ε ε
(8.4)
where:
i number of i
th
reinforcement bar in crosssection
a size of compression block of concrete
m
ε extreme compressive strain in concrete equal to 0.003
i
d distance of i
th
reinforcement layer for the top of crosssection
) (a c position of neutral axis due to changing size of compression block
In the procedure, four characteristic cases there can be distinguished. The first
case is when the position of neutral axis is outside of the cross section (Figure 8.1). The
second case is when the end of compression block of concrete is in the bottom flange of
the cross section (Figure 8.3). Two next cases correspond to the end of compression
block of concrete localized in the webs and in the top flange (Figure 8.4 and Figure 8.5).
Figure 8.3 End of compression block of concrete in the bottom flange
β
ε
111
Figure 8.4 End of compression block of concrete in the web.
Figure 8.5 End of compression block of concrete in the top flange.
Figure 8.6 shows material behavior of reinforcing steel, and it is described by the
following equation:
m s
m s y
y s y
y s
y
s s
y
s
for
for
for
for
f
E
f
f
ε ε
ε ε ε
ε ε ε
ε ε
ε
>
≤ <
≤ ≤ −
− <
¦
¦
¹
¦
¦
´
¦
⋅
−
=
0
(8.5)
where:
y
y
s
f
E
ε
=
(8.6)
β
ε
β
ε
112
Figure 8.6 Stress  Strain Relationship for Reinforcing Steel
Having calculated strains in every reinforcement bar, it is possible to evaluate
forces for each reinforcement layer.
( ) ( )
( )
( )
( )
¦
¹
¦
´
¦
− <
≤ ≤ −
≥
⋅ −
⋅ ⋅
⋅
=
y s
y s y
y s
y s
s s s
y s
a i
a i
a i
for
for
for
f A
a i E A
f A
a i P
ε ε
ε ε ε
ε ε
ε
,
,
,
, , (8.7)
where:
i number of i
th
reinforcement bar in crosssection
a size of compression block of concrete
The resultant reinforcement force is calculated using:
( ) ( )
∑
=
i
steel
a i P a P ,
(8.8)
The force in concrete is based on the size of compression block of concrete.
( ) ' ) ( 85 . 0
c C
f a A a P ⋅ ⋅ =
(8.9)
For each size of the compression block, the resistance force of the cross section is
expressed by the sum of all forces acting in the crosssection:
) ( ) , ( ) ( a P a i P a P
c
i
Total
+ =
∑
(8.10)
113
For each size of the compression block, the bending moment resistance is equal to
sum of all the forces in the section multiplied by the corresponding force arm to the
centroid of the crosssection:
) ( ' ) ( 85 . 0
2
) , ( ) ( a Y f a A d
h
a i P a M
C c
i
i Total
⋅ ⋅ ⋅ +
(
¸
(
¸

¹

\

− ⋅ =
∑
(8.11)
8.4. LOAD MODEL
For the scope of this dissertation Cooper River Bridge was modeled using Robot
Millennium software. The three dimensional model was based on the actual design made
by PB World. Shell elements were used to model concrete slab in the bridge. 840 3D
beam elements were used  with 12 degrees of freedom u1
x
, u1
y
, u1
z
, φ1
x
, φ1
y
, φ1
z
, u2
x
, u2
y
, u2
z
, φ2
x
, φ2
y
, φ2
z
 to represent all others members: towers leg, tower, girders and
the diaphragms. The cables elements were used to model suspension cables.
Total length of the structure is 3296 ft. Main span is 1546 ft long, two spans are
650 ft and two spans are 225 ft long. Geometry of the bridge is shown in Figure 8.7.
Bridge towers are 568 ft high, both of them have 368 ft above deck level.
114
Figure 8.7. Geometry of Cooper River Bridge
In the analysis all loading cases have been considered. For five spans, there are 31
loading combinations. They are shown in Figure 8.8.
All of the live load combinations were used to calculate load effect on the bridge
tower. The 31 combinations were used four times; for the different value of loading: 0.64
k/ft, 0.80 k/ft, 1.00 k/ft and 1.20 k/ft as a value of lane live load loading.
The resulting envelopes of bending moments due to various combinations of live
load for bridge tower are shown in Figure 8.9  Figure 8.12.
115
Figure 8.8. Load combinations
116
Figure 8.9. Envelope of bending moments for bridge tower for w=0.64k/ft
200
100
0
100
200
300
400
500
400000 300000 200000 100000 0 100000 200000 300000
P
o
s
s
i
t
i
o
n
o
n
P
y
l
o
n
(
f
t
)
Mz (kipsft)
LL1
LL2
LL3
LL4
LL5
LL6
LL7
LL8
LL9
LL10
LL11
LL12
LL13
LL14
LL15
LL16
LL17
LL18
LL19
LL20
LL21
LL22
LL23
LL24
LL25
LL26
LL27
LL28
LL29
LL30
LL31
117
Figure 8.10. Envelope of bending moments for bridge tower for w=0.80 k/ft
200
100
0
100
200
300
400
500
500000 400000 300000 200000 100000 0 100000 200000 300000 400000
P
o
s
s
i
t
i
o
n
o
n
P
y
l
o
n
(
f
t
)
Mz (kipsft)
LL1
LL2
LL3
LL4
LL5
LL6
LL7
LL8
LL9
LL10
LL11
LL12
LL13
LL14
LL15
LL16
LL17
LL18
LL19
LL20
LL21
LL22
LL23
LL24
LL25
LL26
LL27
LL28
LL29
LL30
LL31
118
Figure 8.11. Envelope of bending moments for bridge tower for w=1.00 k/ft
200
100
0
100
200
300
400
500
600000 500000 400000 300000 200000 100000 0 100000 200000 300000 400000 500000
P
o
s
s
i
t
i
o
n
o
n
P
y
l
o
n
(
f
t
)
Mz (kipsft)
LL1
LL2
LL3
LL4
LL5
LL6
LL7
LL8
LL9
LL10
LL11
LL12
LL13
LL14
LL15
LL16
LL17
LL18
LL19
LL20
LL21
LL22
LL23
LL24
LL25
LL26
LL27
LL28
LL29
LL30
LL31
119
Figure 8.12. Envelope of bending moments for bridge tower for w=1.20 k/ft
200
100
0
100
200
300
400
500
800000 600000 400000 200000 0 200000 400000 600000
P
o
s
s
i
t
i
o
n
o
n
P
y
l
o
n
(
f
t
)
Mz (kipsft)
LL1
LL2
LL3
LL4
LL5
LL6
LL7
LL8
LL9
LL10
LL11
LL12
LL13
LL14
LL15
LL16
LL17
LL18
LL19
LL20
LL21
LL22
LL23
LL24
LL25
LL26
LL27
LL28
LL29
LL30
LL31
120
8.5. RELIABILITY RESISTANCE MODELS
Due to various categories of uncertainties, the resistance of a structural
component
R
,
can be considered as a random variable being a product of nominal
resistance
n
R and three factors: the materials factor, fabrication factor and professional
factor:
P F M R R
n
= (8.12)
The materials factor represents material properties, in particular strength and
modulus of elasticity. The fabrication factor represents the dimensions and geometry of
the component, including crosssectional area, moment of inertia, and section modulus.
The professional factor represents the approximations involved in the structural analysis
and idealized stress/strain distribution models. The professional factor is defined as the
ratio of the test capacity to analytically predicted capacity (the actual insitu performance
to the model used in calculations).
The statistical parameters for material factors used in this dissertation were based
on the project "ReliabilityBased Calibration for Structural Concrete" (Nowak A.S. et al.,
2008). Because the quality of materials such as reinforcing steel and concrete has
improved over the years, the materials factors have been updated based on a new test
database. There is no new information regarding two other factors, F and P . Therefore,
in most cases, statistical parameters for F and P are taken from the previous study
(Ellingwood et al. 1980).
8.5.1. Material Factor
The material factors for concrete were based on the study within the project
"ReliabilityBased Calibration for Structural Concrete" (Nowak A.S. et al., 2008). Figure
8.13 and Figure 8.14 show the bias factor and coefficients of variation for all types of
concrete and all nominal compressive strengths of concrete. In both figures there is a
trend line of changing parameter with respect to concrete compressive strength '
c
f .
121
Recommended values are summarized in Table 8.1. In this dissertation, the concrete
compressive strength of bridge tower is 7000 psi. Statistical parameters assumed are: bias
factor λ = 1.13 and coefficient of variation V = 0.12.
Figure 8.13 Bias factor for compressive strength of concrete
(Nowak A.S. et al., 2008)
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
2
0
0
0
3
0
0
0
4
0
0
0
5
0
0
0
6
0
0
0
7
0
0
0
8
0
0
0
9
0
0
0
1
0
0
0
0
1
1
0
0
0
1
2
0
0
0
1
3
0
0
0
fc'
λ λλ λ
light weight ordinary, ready mix ordinary, plant cast
high strength approximation
122
Figure 8.14 Coefficient of variation for compressive strength of concrete
(Nowak A.S. et al., 2008)
Table 8.1 Recommended Statistical Parameters for Compressive Strength, '
c
f
(Nowak A.S. et al., 2008)
Concrete Grade
f
c
' (psi)
'
c
f
λ V
4000 1.24 0.15
6000 1.15 0.125
8000 1.11 0.11
12,000 1.08 0.11
The material factors for reinforcement steel were based on the study within the
project "ReliabilityBased Calibration for Structural Concrete" (Nowak A.S., Szerszen
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12
0.13
0.14
0.15
0.16
0.17
0.18
2
0
0
0
3
0
0
0
4
0
0
0
5
0
0
0
6
0
0
0
7
0
0
0
8
0
0
0
9
0
0
0
1
0
0
0
0
1
1
0
0
0
1
2
0
0
0
1
3
0
0
0
fc'
V
light weight ordinary, ready mix ordinary, plant cast
high strength approximation
123
M.M., et al., 2008). Data included the yield strength for the reinforcing steel bars with the
nominal yield strength of 60 ksi, and different bar sizes from No.3 to No.14. The
recommended values of statistical parameters are: bias factor λ = 1.13 and coefficient of
variation V = 0.03. Those recommended values have been used in this dissertation for the
bar sizes No.9 and No.11, which were used in the calculations. Plots of the cumulative
distribution functions (CDF) of yield strength of every reinforcement size and
recommended parameters are shown in Figure 8.15 and Figure 8.16.
Figure 8.15 CDF’s of yield strength for Reinforcing Steel Bars, Grade 60 ksi
(Nowak A.S. et al., 2008)
4
3
2
1
0
1
2
3
4
50 60 70 80 90 100 110
Yield strength of rebars [ksi]
N
o
r
m
a
l
I
n
v
e
r
s
e
P
r
o
b
a
b
i
l
i
t
y
#3 #4 #5 #6 #7 #8
#9 #10 #11 #14 All diameters Approximation
Steel rebar size:
λ λλ λ = 1.13
V = 1.03
124
Figure 8.16 Recommended material parameters for reinforcing steel bars, Grade 60 ksi
(Nowak A.S. et al., 2008)
8.5.1. Fabrication Factor
The statistical parameters of fabrication factor are based on previous studies by
Ellingwood et al., 1980. Due to lack of data for bridge towers, the parameters for
columns were assumed. They are summarized in Table 8.2.
Table 8.2 Statistical Parameters of Fabrication Factor.
Material Item Bias factor V
Concrete
Radius of column
1.005 0.04
Reinforcement cover
Steel
Reinforcement area
1.00 0.015
Reinforcement diameter
4
3
2
1
0
1
2
3
4
50 60 70 80 90 100 110
Yield strength of rebars [ksi]
N
o
r
m
a
l
I
n
v
e
r
s
e
P
r
o
b
a
b
i
l
i
t
y
All diameters Normal Approximation Lognormal Approximation
λ λλ λ = 1.13
V = 1.03
125
8.5.1. Professional Factor
The statistical parameters of professional factor are based on the previous study
performed by Ellingwood et al. (1980). The bridge tower behaves as an eccentrically
loaded column. Therefore, the statistical parameters of professional factors for columns
were used in this dissertation. The professional factors were chosen for tied columns: bias
factor is λ ൌ 1.00 and coefficient of variation is V=0.08.
8.5.1. Statistical Parameters of Resistance
Statistical parameters of resistance were obtained by 10000 Monte Carlo
Simulation: coefficient of variation V=0.16 and bias factor λ=1.17.
Figure 8.17 Statistical Parameters of Resistance
0
50000
100000
150000
200000
250000
0 200000 400000 600000 800000 1000000
F
o
r
c
e
[
k
i
p
s
]
Bending Moment [kft]
mean+std
nominal
mean
meanstd
Balance
Failure
126
8.6. LOAD MODEL
The case considered in this study is combination Strength I, which is combination
of dead load and live load. This combination has the highest load factor for live load.
For time varying loads two random variables for arbitrarypointintime and
maximum 50year load components can be considered. In this study live load is the only
varying load and it is assumed to reach the maximum 50year load value. Therefore there
is no need to use Turkstra’s rule.
Variation in the dead load, which is caused by variation of the weight of materials
(concrete and steel), variation of dimensions, and idealization of analytical models affects
statistical parameters of resistance. The assumed statistical parameters for dead load are
based on the data available in literature (Ellingwood et al. 1980; Nowak 1999). They
include for castinplace concrete elements a bias factor of 1.05 and coefficient of
variation of 0.10.
Variation of live load is derived in Paragraph 5.4 of this dissertation.
8.7. RELIABILITY ANALYSIS
Reliability analysis was performed for the considered bridge. The element
selected for the analysis was the bridge tower subjected to the bending moment and the
axial force. The bridge was design by the PB World for the design live load specified in
AASHTO LRFD CODE. The limit state function was selected as the Strength I load
combination according to AASHTO LRFD Code. There are only two major load
components: dead load and live load. However, on the real structure in a specific
localization other loads, such as wind load, influence the bridge behavior, the selected
limits state was chosen to demonstrate the sensitivity of reliability index on long span
bridges due to change of live load.
Reliability indexes were calculated for bridge loaded with AASHTO design live
load 0.64 k/f t and three other possible load cases of 0.80, 1.00 and 1.20 k/ft. For every
value of lane load the 31 load cases were analyzed according to paragraph 8.4. Each load
case, Figure 8.8, generates separate loading case to the bridge tower with different
127
eccentricity condition. The results vary depending on eccentricity. None of them exceeds
the balance failure zone, Figure 8.18 . It means that all the cases are in the compression
control and the strength reduction factor is φ=0.75, specified in AASHTO LRFD Section
5.
Figure 8.18 Force and Moment results on the bridge tower for different live loads
Figure 8.19 shows the results of reliability indexes due to different live loads. It
can be noticed that values of β are in high range for the cases with the eccentricity of
loads very small for all possible loading conditions. For the cases with the eccentricities
approaching the balance failure the reliability indexes are decreasing with the increase of
the actual loading.
0
20000
40000
60000
80000
100000
120000
140000
160000
180000
750000 500000 250000 0 250000 500000 750000
Force [kips]
Moment [kft]
Nominal Resistance
Balance Failure
Nominal Loads w=0.64 k/ft
Nominal Loads w=0.80 k/ft
Nominal Loads w=1.0 k/ft
Nominal Loads w=1.2 k/ft
128
Figure 8.19 Reliability indexes due to different live load
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
β ββ β
Uniformly Distributed Load [k/ft]
0.64
0.80
1.00
1.20
129
CHAPTER 9
SUMMARY & CONCLUSIONS
In the study, a live load model for long span structures was derived. The live load
model is valid for spans between 600 ft and 5000 ft and it is intended to reflect current
traffic patterns, quantities of trucks and their weights. The developed live load model is
recommended to be taken into consideration in the bridge design code.
Preliminary study was performed by reviewing previous research and current
provisions of international codes on the topic. Equivalent uniformly distributed load is
calculated and compared. The new live load was developed based on three models: an
average 5axle truck, legal load trucks and simulation of a traffic jam using WIM data.
The newest available traffic database from a variety of sites within many different states
was obtained. The magnitude of the database has to be underlined, because such an
extensive actual weigh in motion database has never been used in the derivation of live
load for any kind bridges. A numerical procedure was developed to filtered out WIM data
from erroneous readings and to simulate traffic jam situations. From the simulation the
values of uniformly distributed load were derived for a variety of span lengths and site
localizations. In the developed procedure, starting with the first truck, all consecutive
trucks were added with a fixed clearance distance between them until the total length
reached the span length. Then, the total load of all trucks was calculated and divided by
the span length to obtain the first value of the average uniformly distributed load. Next,
the first truck was deleted, and one or more trucks were added so that the total length of
trucks covers the full span length and the new value of the average uniformly distributed
130
load was calculated. Trucks were kept in actual order, as recorded in the WIM surveys.
Results of the simulations were plotted as a cumulative distribution function of uniformly
distributed load for considered span lengths. The obtained mean value oscillates between
values of 0.50 and 0.75 k/ft. Cumulative distribution functions were also plotted for
maximum daily and maximum weekly combinations of trucks. For longer spans,
uniformly distributed load decreases and is closer to the mean value. This observation
confirms that for a long loaded span, one heavily overloaded truck does not have
significant influence. This is because the load depends on a mix of traffic. The bias
factors (ratio of mean to nominal) were calculated for the heaviest 75year combination
of vehicles. The 75year uniformly distributed loads were derived from extrapolated
distributions. It was noticed that the bias factor values for most of the sites do not exceed
1.25, which is similar as in short and medium spans, as shown in the NCHRP Report 368
(1999). It is recommended to use HL93 also for those long spans. Two other models, an
average 5axle truck and a legal load trucks model led to similar conclusion. It was
noticed that for some sites, with very heavy traffic, the bias factor reaches a value of 2.0.
Those sites are characterized by high ADTT (usually over 3000) or increased percentage
of overloaded loaded vehicles (over about 10%). Bridges located in such sites require
special attention and application of increased design live load. The value of the design
load should be agreed with the owner of the structure. For some bridges considered, for
example in the area of New York, it was found that the uniformly distributed load should
be 1.25 kip/ft in order to obtain bias lower than 1.25. Statistical parameters for live load
are: for spans 6001000 ft bias 1.25 and coefficient of variation 0.10, for spans longer
than 1000 ft bias 1.20 and coefficient of variation 0.08.
In this dissertation, the problem of multilane reduction factors was discussed.
Multilane factors were found to be very site specific, as with the live load. Video
recordings of traffic confirmed that for the majority of the time we can observe that the
moving lanes contain a mixture of trucks and cars. However, situations when one lane is
almost exclusively occupied by trucks or trucks occupy three or four lanes at the same
time are also possible. Multiple reduction factors for design live load should account for
those most critical loading cases. In the WIM database available, the vehicles of 13
categories 13 have not been registered. Therefore, it did not allow for simulations and
131
derivation of multilane factors for all traffic lanes. Simulation of the traffic on the most
loaded lane was possible with an assumption, that in traffic jam situations passenger
vehicles merge to the left and the right lane remains occupied exclusively by trucks. It
was stated that equal reduction of load on all traffic lane does not reflect the actual
situation. At least one of the lanes should be loaded more than the others. It was
concluded that the multilane reduction factors have to be an objective of additional
extensive studies.
The study of dynamic factor was performed for the research. It was concluded that
the current dynamic load factor of 0.33 is too high for bridges with longer spans. It may be
applicable in short bridges, when vibration due to “wheel hop” on the approach slab is
significant. For longer bridges, where the influence of the approach slab decreases and
vibrations of many vehicles interfere with each other, the dynamic load factor could be
smaller. The assumption of a traffic jam situation to develop live load model induces no
dynamic allowance. However, to not introduce confusion among designers it is
recommended do keep dynamic factor as it is for short and medium span bridges, which
results in very small value for long span bridges.
132
CHAPTER 10
RECCOMENDATIONS
1. The developed live load model is valid for long spans between 600 ft and 5000 ft.
2. For long spans it is recommended to use HL93 load, specified in AASHTO
LRFD Code (2007), uniformly distributed load of 0.64 k/ft plus design truck or
tandem. The bias factor calculated for the heaviest 75year combination of
vehicles did not exceed 1.25, which is similar as in short and medium spans, as
shown in the NCHRP Report 368 (1999).
3. For some sites characterized by high ADTT or increased percentage of
overloaded vehicles (over 10%) bias factor reaches a value of 2.0. Those bridges
require application of increased site specific design live load, which should be
agreed with the owner of the structure.
4. It was proposed to use dynamic factor as specified in AASHTO LRFD Code
(2007). Developed live load model assumes traffic jam situation and does not
allow for dynamic. However, to not introduce confusion among designers it is
recommended do keep dynamic factor as it is for short and medium span bridges,
which results in very small value for long span bridges.
5. It was proposed to use multilane reduction factors as specified in AASHTO
LRFD Code (2007). It is recommended to perform further studies in this field.
133
6. Statistical parameters for live load are:
 for spans 6001000 ft: bias 1.25 and coefficient of variation 0.10,
 for spans longer than 1000 ft: bias 1.20 and coefficient of variation 0.08.
The developed live load model is recommended to be taken into consideration in the
AASHTO LRFD Bridge Design Specifications.
134
REFERENCES
[1]. AASHTO LRFD Bridge Design Specifications. (2007). American Association of
State Highway and Transportation Officials, Washington, D.C.
[2]. AASHTO Geometric Design of Highways and Streets (2001). American
Association of State Highway and Transportation Officials, Washington, D.C.
[3]. CAN/CSAS600 Canadian Highway Bridge Design Code, CSA International,
Toronto, Canada.
[4]. Manual for Condition Evaluation and load and Resistance Factor Rating (LRFD) of
Highway Bridges. 2003. American Association of State Highway and
Transportation Officials, Washington, D.C.
[5]. OHBDC (1991), Ontario Highway Bridge Design Code, Ontario Ministry of
Transportation, Downsview, Ontario.
[6]. ASCE Committee on Loads and Forces on Bridges (1981). “Recommended Design
Loads for Bridges”, ASCE Journal of Structural Engineering, Vol. 107, No. 7,
December 1981, pp. 11611213.
[7]. Ang, A. HS., and Tang, W.H., “Probability Concepts in Engineering Planning and
Design,” Volume I: Basic Principles, John Wiley & Sons, new York, 1975
[8]. Ang, A. HS., and Tang, W.H., “Probability Concepts in Engineering Planning and
Design,” Volume II: Decision, Risk, and Reliability, John Wiley & Sons, New
York, 1984
[9]. Augusti G., Barrata B. “Limit and Shakedown Analysis of Structures with
Stochastic Strengths,” Proceedings of the Second SMiRT Conference, Berlin, 1973
[10]. Augusti G., Barrata B. “Plastic Shakedown of Structures with Stochastic Local
Strengths,” Proceedings of the Symposium on Resistance Ultimate Deformability of
Structures, IABSE, Lisbon, 1973
[11]. Ayyub B.M., Halder A. “Practical Structural Reliability Techniques,” Journal of
Structural Engineering, ASCE, Vol. 110, No. 8, 1984, pp. 17071724
135
[12]. Ayyub B.M., McCuen R.H. “Probability, Statistics, & Reliability for Engineers,”
CRC Press Boca Raton, New York, 1997
[13]. Benjamin J.R., Cornell C.A. “Probability, Statistics and Decision for Civil
Engineers,” McGrawHill, New York, 1970
[14]. Biggs, J.M., Suer, H.S. 1955. “Vibration Measurements on SimpleSpan Bridges”,
Highway Research Board, Bulletin 124.
[15]. Breitung K. “Asymptotic Approximations for Multinormal Integrals,” Journal of
Engineering Mechanics, ASCE, Vol. 110, 1984, pp. 357366
[16]. Buckland, P.G., Navin, F.P.D, Zidek, J.V., and McBryde, J.P, 1978, “Traffic
loading of longspan bridges”, Transportation Research Record, Vol. 665, pp. 146
154.
[17]. Buckland, P.G., Navin, F.P.D, Zidek, J.V., and McBryde, J.P., 1980, “Proposed
Vehicle Loading of LongSpan Bridges”, Journal of Structural Division, ASCE,
Vol. 106, pp. 915932.
[18]. Buckland, P.G., 1991, “North American and British LongSpan Bridge Loads”,
Journal of Structural Division, ASCE, Vol. 117, pp. 29722987.
[19]. Cantieni, R. 1984. “Dynamic Load Testing of Highway Bridges”, Transportation
Research Record No 950, pp. 141148.
[20]. Cebon, D. 1999. “Handbook of vehicleroad interaction”, Department of
Engineering, University of Cambridge. London.
[21]. Cebon, D. 1993. “Interaction between heavy vehicles and roads”, Department of
Engineering, University of Cambridge. London.
[22]. Cornell, C. A., “Bounds on the Reliability of Structural Systems,” Journal of the
Structural Division, ASCE, 93 (ST1), 171200, 1967
[23]. Corotis R.B., Nafday A.M. “Structural Systems Reliability using Linear
Programming and Simulations,” Journal of Structural Engineering, ASCE, Vol. 115,
No. 10, 1989, pp. 24352447
136
[24]. Ellingwood, B.; Galambos, T.V.; MacGregor, J.G., and Cornell, C.A.,
“Development of a Probability Based Load Criterion for American National
Standard A58,” NBS Special Publication 577, Washington, DC, National Bureau of
Standards, 1980
[25]. Ellingwood, B., MacGregor, J.G., Galambos, T.V., and Cornell, C.A., “Probability
Based Load Criteria: Load factors and Load Combinations,” Journal of Structural
Engineering, ASCE, Vol. 108, No. ST5, 1982
[26]. Fiessler B., Naumann HJ., Rackwitz R. “Quadratic Limit States in Structural
Reliability,” Journal of Engineering Mechanics, ASCe, Vol. 105, 1979, pp. 661676
[27]. Fleming, F.J., Romualdi, J.P. 1961. “Dynamic Response of Highway Bridges”,
Journal of the Structural Division, Vol. 87.
[28]. Fraudenthal A.M. “Safety and Probability of Structural Failure,” Transactions,
ASCE, Vol. 121, 1956, pp. 13371375
[29]. Galambos, T.V., and Ravindra, M.K., “Load and Resistance Factor Design,”
Journal of Structural Devision, ASCE, ST9, Proc. Paper 14008, Sep. 1978
[30]. Galdos, N.H., Schelling, D.R., Sahin, M.A. 1993. “Methodology for Impact Factor
of Horizontally Curved Box Bridges”, Journal of Structural Engineering, Vol. 119,
No. 6, pp. 19171934.
[31]. Gindy, M., Nassif, H.H., 2006, “Multiple Presence Statistics for Bridge Live Load
Based on WeighinMotion Data”, TRB annual meeting
[32]. Hasofer A.M., Lind N.C. “An Exact and Invariant First Order Reliability Format,”
Journal of Structural Mechanics Division, ASCE, Vol. 100, No. 1, 1974
137
[33]. Hohenbichler M., Rackwitz R. “Improvement of SecondOrder Reliability Estimates
by Importance Sampling,” Journal of the Engineering Mechanics, ASCE, Vol. 114,
1988, pp. 21952199
[34]. Hwang, ES. and Nowak, A.S., “Simulation of Dynamic Load for Bridges”, ASCE
Journal of Structural Engineering, Vol. 117, No. 5, May 1991, pp. 14131434.
[35]. Khalifa, M.A. 1992. “Parametric Study of CableStayed Bridge Response Due to
Trafficinduced Vibration”, Computers & Science, Vol. 47, No. 2, pp. 321339.
[36]. Khisty, C.J. Prentice Hall, Englewood Cliffs, New Jersey, “Transportation
Engineering”, 1990.
[37]. Kim, SJ., Sokolik, A.F., and Nowak, A.S., 1997, “Measurement of Truck Load on
Bridges in the Detroit Area”, Transportation Research Record, No.1541, pp. 5863.
[38]. Kiureghian A., Liu PL. “Structural Reliability under Incomplete Probability
Information,” Journal of Engineering Mechanics, ASCE, Vol. 112, 1986, pp. 85104
[39]. Kulicki J.M., Prucz Z., Clancy C.M., Mertz D.R., Nowak A.S., “Updating the
calibration report for AASHO LRFD”, NCHRP 207/186, January 2007
[40]. Lutomirski, T.A., “Reliability models for circular concrete columns”, Doctoral
Dissertation, 2009.
[41]. MacAdam, C.C., Fancher, P.S., Hu, G.T., Gillespie, T.D. 1980. “A computerized
model for simulating the breaking and steering dynamics of trucks, tractor
semitrailers, doubles, and triples combinations”, MVMA Project 1197.
[42]. McLean, D.I., Marsh, M.L. 1998. “Dynamic Impact Factors for bridges”, NCHRP
Synthesis 266.
[43]. Nassif, H.H., and Nowak, A.S., 1995, “Dynamic Load Spectra for Girder Bridges”,
Transportation Research Record, No. 1476, pp. 6983.
[44]. Navin, FPD, 1982, “Road vehicles on bridges”, Canadian Journal of Civil
Engineering, Vol. 9, pp. 468476.
138
[45]. Nowak, A.S., “Calibration of LRFD Bridge Design Code,” NCHRP Report 368,
Transportation Research Board, Washington, D.C., USA, 1999
[46]. Nowak, A.S., and Lind , N.D., “Practical Bridge Code Calibration,” Journal of
Structural Division, ASCE, pp. 24972510, December 1979
[47]. Nowak, A.S., 1993, “Live Load Model for Highway Bridges”, Journal of Structural
Safety, Vol. 13, Nos. 1+2, December, pp. 5366.
[48]. Nowak, A.S., “Calibration of LRFD Bridge Design Code”, Final Report, NCHRP
Project 1233, December 1993.
[49]. Nowak, A.S., 1994, “Load Model for Bridge Design Code”, Canadian Journal of
Civil Engineering, Vol. 21, pp. 3649.
[50]. Nowak, A.S., 1995, “Calibration of LRFD Bridge Code”, ASCE Journal of
Structural Engineering, Vol. 121, No. 8, pp. 12451251.
[51]. Nowak, A.S. and Hong, YK., “Bridge Live Load Models”, ASCE Journal of
Structural Engineering, Vol. 117, No. 9, Sept. 1991, pp. 27572767.
[52]. Nowak, A.S., Hong, YK and Hwang, ES., “Modeling Live Load and Dynamic
Load for Bridges”, Transportation Research Record, No. 1289, 1991, pp. 110118.
[53]. Nowak, A.S., Kim, SJ., Laman, J., Saraf, V. and Sokolik, A.F., “Truck Loads on
Selected Bridges in the Detroit Area”, Final Report submitted to MDOT, November
1994.
[54]. Nowak, A.S., Laman, J. and Nassif, H., “Effect of Truck Loading on Bridges”, Final
Report submitted to MDOT, December 1994.
[55]. Nowak, A.S. and Collins, K.R., “Reliability of Structures,” McGrawHill New
York, New York, USA, 2000
139
[56]. MacGregor J.G., Mirza S.A., Ellingwood B., “Statistical Analysis of Resistance of
Reinforced and Prestressed Concrete Members”, ACI Structural Journal, Title
no.8016, MayJune, 1983, pages 167 to 176.
[57]. Madsen H.O., Krenk S., Lind N.C. “Methods of Structural Safety,” PrenticeHall,
1986
[58]. Murzewski J. “Reliability of Engineering Structures” (in Polish), Arkady, Warsaw,
Poland, 1989
[59]. Paultre, P., Chaallal, O., Proulx, J. 1992. “Bridge Dynamics and Dynamic
Amplification Factors: A Review of Analytical and Experimental Findings”,
Canadian Journal of Civil Engineering, Vol. 19, No. 2, pp. 260278.
[60]. Rackwitz, R., Fiessler, B. “Structural Reliability under Combined Random Load
Sentences,” Computers & Structures, Vol. 9, 1978, pp. 489494
[61]. Rosenblatt, M. “Remarks on Multivariate Transformations,” The Annuals of
Mathematical Statistics, Vol. 23, 1952, pp. 470472
[62]. Schelling, D.R., Galdos, N.H., Sahin, M.A. 1992. “Evaluation of Impact Factors for
Horizontally Curved Steel Box Bridges”, Journal of Structural Engineering, Vol.
118, No. 11, pp. 32033221.
[63]. Sivakumar, B., Ghosn, M., Moses, F., “Protocols for Collecting and Using Traffic
Data in Bridge Design,” Final Report, NCHRP Project 1276, July 2008.
[64]. ThoftChristiansen, P., Baker, M.J. “Structural Reliability Theory and Its
Applications,” SpringerVerlag, Berlin, Heidelberg, New York, 1982
[65]. ThoftChristiansen, P., Murotsu, Y. “Application of Structural Systems Reliability
Theory,” SpringerVerlag, Berlin, Heidelberg, New York, Tokio, 1986
140
[66]. Wang, T.L., Shahawy, M., Huang, D.Z. 1993. Dynamic Response of Highway
Truck Due to Road Surface Roughness. Computers and Structures, Vol. 49, No. 6,
pp. 10551067.
[67]. Wang, T.L, Hunag, D.Z. 1992. “Cablestayed Bridge Vibration Due to Road Surface
Roughness”, Journal of Structural Engineering, Vol. 188, No. 5, pp. 13541374.
[68]. Wang, T.L., Huang, D.Z., Shahawy, M. 1996. “Dynamic Behavior of Continuous
and Cantilever ThinWalled Box Girder Bridges”, Journal of Bridge Engineering,
Vol. 1, No. 2, pp. 6775.
[69]. Winkler, C.B., Fancher, P.S., MacAdam, C.B. 1983. “Parametric Analysis of heavy
duty truck dynamic stability”, UMTRI8313.
[70]. Wolinski Sz., Wrobel K. “Reliability of Engineering Structures” (in Polish),
Publishers of Rzeszow Technical University (Oficyna Wydawnicza Politechniki
Rzeszowskiej), 2001
[71]. Zhu, X.Q., Law, S.S. 2002. “Dynamic Load on Continuous Multilane Bridge Deck
from moving vehicles”, Journal of Sound and Vibration, Vol. 251, No. 4, pp. 697
716.
[72]. Transportation Statistics Annual Report, December 2006, U.S. Department of
Transportation.
[73]. “Regulation of Weights, Lengths, and Widths of Commercial Motor Vehicles”.
Special Report 267. TRB, National Research Council, Washington, DC, 2002.
[74]. “Comprehensive Truck Size and Weight Study. Volume I: Summary Report”.
Publication FHWAPL00029. FHWA, U.S. Department of Transportation,
Washington, DC, 2000.
[75]. Truck Driver's Guidebook, Michigan Center for Tuck Safety, 2008.
[76]. “Model for Highway Bridge Impact”, Journal of Structural Engineering, Vol. 116,
No. 7, pp. 17721793.
[77]. Traffic Monitoring Guide, U.S. Department of Transportation, Federal Highway
Administration, 2001.
141
APPENDIX A
CDF OF UDL FOR ALL TRUCK COMBINATIONS
Figure A.1. CDF of UDL for Oregon I5 Woodburn, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
142
Figure A.2. CDF of UDL for Oregon I84 Emigrant Hill, lane 1
Figure A.3. CDF of UDL for Oregon OR 58 Lowell, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
143
Figure A.4. CDF of UDL for Oregon OR 58 Lowell, lane 2
Figure A.5. CDF of UDL for Oregon US 97 Bend, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
144
Figure A.6. CDF of UDL for Florida 9916, lane 1
Figure A.7. CDF of UDL for Florida 9919, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
1st lane
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
145
Figure A.8. CDF of UDL for Florida 9927, lane 1
Figure A.9. CDF of UDL for Florida 9936, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
146
Figure A.10. CDF of UDL for Indiana 9534, lane 1
Figure A.11. CDF of UDL for Indiana 9534, lane 2
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
147
Figure A.12. CDF of UDL for Indiana 9534, lane 3
Figure A.13. CDF of UDL for Indiana 9544, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
148
Figure A.14. CDF of UDL for Indiana 9512, lane 1
Figure A.15. CDF of UDL for New York 9121, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
149
Figure A.16. CDF of UDL for New York 9121, lane 4
Figure A.17. CDF of UDL for New York 2680, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
150
Figure A.18. CDF of UDL for New York 2680, lane 4
Figure A.19. CDF of UDL for New York I495 EB, lane 1
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
151
Figure A.20. CDF of UDL for New York I495 EB, lane 2
Figure A.21. CDF of UDL for New York I495 EB, lane 3
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
152
Figure A.22. CDF of UDL for New York I495 WB, lane 1
Figure A.23. CDF of UDL for New York I495 WB, lane 2
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
5.00
4.00
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
5.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
153
APPENDIX B
CDF OF MAXIMUM DAILY UDL
Figure B.1. CDF of maximum daily UDL for Oregon I5 Woodburn, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrpolated
154
Figure B.2. CDF of maximum daily UDL for Oregon I84 Emigrant Hill, lane 1
Figure B.3. CDF of maximum daily UDL for Oregon OR 58 Lowell, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
155
Figure B.4. CDF of maximum daily UDL for Oregon US 97 Bend, lane 1
Figure B.5. CDF of maximum daily UDL for Florida 9916, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.50 1.00 1.50 2.00
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
156
Figure B.6. CDF of maximum daily UDL for Florida 9919, lane 1
Figure B.7. CDF of maximum daily UDL for Florida 9927, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
157
Figure B.8. CDF of maximum daily UDL for Florida 9936, lane 1
Figure B.9. CDF of maximum daily UDL for Indiana 9534, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
158
Figure B.10. CDF of maximum daily UDL for Indiana 9544, lane 1
Figure B.11. CDF of maximum daily UDL for Indiana 9512, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolation
159
Figure B.12. CDF of maximum daily UDL for New York 9121, lane 1
Figure B.13. CDF of maximum daily UDL for New York 2680, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
160
Figure B.14. CDF of maximum daily UDL for New York I495 EB, lane 1
Figure B.15. CDF of maximum daily UDL for New York I495 WB, lane 1
3.00
2.00
1.00
0.00
1.00
2.00
3.00
4.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
161
APPENDIX C
CDF OF MAXIMUM WEEKLY UDL
Figure C.1. CDF of maximum weekly UDL for Oregon I5 Woodburn, lane 1
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
162
Figure C.2. CDF of maximum weekly UDL for Oregon I84 Emigrant Hill, lane 1
Figure C.3. CDF of maximum weekly UDL for Oregon OR 58 Lowell, lane 1
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
163
Figure C.4. CDF of maximum weekly UDL for Oregon US 97 Bend, lane 1
Figure C.5. CDF of maximum weekly UDL for Florida 9916, lane 1
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
164
Figure C.6. CDF of maximum weekly UDL for Florida 9919, lane 1
Figure C.7. CDF of maximum weekly UDL for Florida 9927, lane 1
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
165
Figure C.8. CDF of maximum weekly UDL for Florida 9936, lane 1
Figure C.9.CDF of maximum weekly UDL for New York 9121, lane 1
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolatred
166
Figure C.10. CDF of maximum weekly UDL for New York 2680, lane 1
3.0
2.0
1.0
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
600 ft
1000 ft
2000 ft
3000 ft
4000 ft
5000 ft
extrapolated
LIVE LOAD MODELS FOR LONG SPAN BRIDGES
by Marta Lutomirska
A DISSERTATION
Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of the Requirements For the Degree of Doctor of Philosophy
Major: Engineering
Under the Supervision of Professor Andrzej S. Nowak
Lincoln, Nebraska December, 2009
LIVE LOAD MODELS FOR LONG SPAN BRIDGES Marta Lutomirska, PhD University of Nebraska, 2009
Advisor: Andrzej S. Nowak
In the doctoral dissertation a live load model for long span structures was derived. The live load model is valid for spans between 600 ft and 5000 ft and it is intended to reflect current traffic patterns, quantities of trucks and their weights. The live load models available were developed for short and medium span bridges. Those models were not appropriate for long span bridges due to different types of structure and critical traffic patterns. Live load on long spans depends on traffic mix. One heavily overloaded truck does not have significant influence. Moreover, the continuous increase in the number of the trucks, their weights, and high percentage of overweight trucks led to a search for the newest traffic data. The database includes variety of sites within many different states. A numerical procedure was developed to process the database and simulate traffic jam situations. From the simulation the values of uniformly distributed load were derived. Trucks were kept in actual order, as recorded in the WIM surveys. Results of the simulations were plotted as a cumulative distribution function of uniformly distributed load for considered span lengths. For longer spans, uniformly distributed load decreases and is closer to the mean value. The bias factors were calculated for the heaviest 75year combination of vehicles. The 75year uniformly distributed loads were derived from
It was stated that for most of the bridges current live load HL93 is appropriate.extrapolated distributions. The developed live load model is recommended to be taken into consideration in the bridge design code. It was also noticed that some bridges. . require special attention and application of increased design live load. characterized by high ADTT and increased percentage of overloaded loaded vehicles.
© Marta Lutomirska All rights reserved 2009
iii
DEDICATION
To My Family
iv
ACKNOWLEDGMENTS
I wish to express my deepest gratitude to Professor Andrzej S. Nowak, my academic advisor and chairperson of my dissertation committee, for his support and guidance throughout my graduate study at the University of Nebraska. My appreciations also go to a member of my doctoral committee, Dr. George Morcous, who advised and encouraged me in my study. Special thanks go to Dr. Maria M. Szerszen and Dr. Elizabeth G. Jones, members of my reading committee for their effort and time spent on reviewing my dissertation and offering helpful suggestions.
The Department of Civil Engineering at the University of NebraskaLincoln is acknowledged for the study and research opportunities they provided.
I am deeply grateful to my husband, Tomasz, my parents, and whole family for their great support and encouragement.
v
...1........................................................... xii CHAPTER 1 INTRODUCTION..... 20 PROVISIONS FOR MULTILANE REDUCTION FACTORS ............... 35 MONTE CARLO METHOD SIMULATION TECHNIQUE ....................1......................... TRUCK SIZES AND WEIGHT LIMITS ....................... 30 LIMIT STATE FUNCTION ...3.............................................. Error! Bookmark not defined...........................................3............................................................................. 29 STANDARD VARIABLES AND PROBABILITY DISTRIBUTIONS ......................... 11 INTERNATIONAL PROVISIONS FOR LIVE LOADING ...1.......................1 1. REGULATIONS OF TRUCK TYPES............................ 23 CHAPTER 3 STRUCTURAL RELIABILITY PROCEDURES .................................................................1.................... 2...............................2......................................................... 37 NORMAL PROBABILITY PAPER ........... 51 CHAPTER 5 DEVELOPMENT OF LIVE LOAD MODEL ............1......................5......... DEDICATION ... 3................................................................................................................ 1....................................................... 3....................................................3.................... 4.................................... 29 3.................................. 42 DATA COLLECTION METHODOLOGY (WIM) .......2........... 4..4..... 3...2...............2 ORGANIZATION OF THE DISSERTATION .................v LIST OF FIGURES ............. 21 COMPARISON OF EQUIVALENT UNIFORMLY DISTRIBUTED LOADS........................ 33 RELIABILITY INDEX.... INTRODUCTION .............................2............. 61 vi ..........TABLE OF CONTENTS ABSTRACT ............................... 39 CHAPTER 4 TRAFFIC DATA .......................................................................................6.................................................................. 11 2........................... 1...................................................1 OBJECTIVE AND BENEFITS OF THE STUDY ................4....... INTRODUCTION ........... 61 5..............4 PRIOR INVESTIGATIONS ....................................... iv ACKNOWLEDGMENTS ................. 1................................................ 49 WEIGH IN MOTION DATABASE ...................................................................5 CHAPTER 2 LIVE LOAD IN CURRENT DESIGN CODES...................................................................3..... 2............................................................................... viii LIST OF TABLES .................... 3.............................................................. 3........... 42 4.......................5........ 2........... 12 PROVISIONS FOR DYNAMIC LOAD FACTOR................................................ PROBLEM STATEMENT ..................... 2........................................................... INTRODUCTION ...........4..
.............. 88 CHAPTER 7 DYNAMIC FACTOR .. 107 8......................................................2.... 8..........6................... 90 BRIDGEVEHICLE INTERACTION MODEL AND DERIVATION OF DYNAMIC FACTOR .................7................. 108 LOAD MODEL ........................... 5............ 106 8.......................................................................................................................................78 MULTIPLE PRESENCE OF TRUCKS BASED ON THE VIDEO FILES OF TRAFFIC ..2............... 86 CONCLUSIONS............................. 7.......................... 89 STUDIES ON PARAMETERS AFFECTING DYNAMIC BRIDGE RESPONSE .............. 126 CHAPTER 9 SUMMARY & CONCLUSIONS .... 105 CHAPTER 8 RELIABILTY ANALYSIS OF SUSPENSION BRIDGE.............. 66 CHAPTER 6 MULTIPLE PRESENCE ... 141 APPENDIX B CDF OF MAXIMUM DAILY UDL .............. 8...........1.... RELIABILITY ANALYSIS PROCEDURE ....................................4.............................................. INTRODUCTION ......... INTRODUCTION ............................................................3........................ 77 6......................................................................................................... 113 RELIABILITY RESISTANCE MODELS ........................................................................................................................2.................. 98 7....5............ 8.......2.......................61 MODEL BASED ON LEGAL LOAD TRUCKS ........... 161 vii .......................................1...........................................3.......................................................................................................................... 8. 6.. CONCLUSIONS ............................................................. 134 APPENDIX A CDF OF UDL FOR ALL TRUCK COMBINATIONS .......................... 5...................... 126 RELIABILITY ANALYSIS ....................................89 7........ 81 APPROACHES TO MULTILANE REDUCTION FACTORS ..........................2...............1............................. 77 STUDIES ON PRESENCE OF MULTIPLE TRUCKS ............................................. 8............................................ 153 APPENDIX C CDF OF maximum weekly UDL. 65 MODEL BASED ON TRAFFIC JAM SIMULATION USING WIM DATA ........ 132 REFERENCES ... 6................. 120 LOAD MODEL .................4.............. 6..........................................................................5..... MODEL BASED ON AVERAGE 5AXLE TRUCK .............................................................. 106 SELECTION OF REPRESENTATIVE STRUCTURE.................................................... ELEMENT AND LIMIT STATE FUNCTION ................................... NOMINAL RESISTANCE ............................................... 129 CHAPTER 10 RECCOMENDATIONS ..............................1.............. 7.............3.......4......5................................. 6.................................
......... ............................................. resistance.................. 25 Figure 2. Equivalent Factored Loads.. with multilane factors for 4 traffic lanes.................. Truck and Lane Load.......................... HL93 Live Loading in AASHTO LRFD Code [2007].... 45 Figure 4............................ w/o multilane factors.... w/o IM.................................... HL93 Live Loading in AASHTO LRFD Code [2007].. Equivalent Factored Loads.......... CAN/CSAS600 Live Loading [2000]............................... Tandem and Lane Load...................................................... 12 Figure 2....9................................13...11....................1...............10.....4.............. Equivalent Unfactored Loads........ ASCE Loading on Log Scale ............7................. 15 Figure 2......................................... ....1 PDF and CDF of a normal random variable ....... 26 Figure 3... 34 Figure 3........ with IM....................................... 18 Figure 2..2..4...... ..........8..16............... Equivalent Unfactored Loads... .................................... 13 Figure 2. 14 Figure 2............................ and safety margin .........3. ........................... Eurocode 1 [2002]................regulation of vehicles' length and weight ...... OHBD Truck.................... 45 Figure 4......... Equivalent Factored Loads....................... ....... . 19 Figure 2... 25 Figure 2...... 36 Figure 3..........5.................... 43 Figure 4...... 32 Figure 3.............. OHBD Live Loading [1991]...1.............................. New Bridge Formula . Longer Combination Vehicles (LCV’s) ...... HL93 Live Loading in AASHTO LRFD Code [2007].... 24 Figure 2............... Alternative Load for Negative Moment between points of contraflexure and reaction at interior piers................... Dimensions of HB vehicle... w/o multilane factors............. 26 Figure 2......LIST OF FIGURES Figure 2........ States allowing various Longer Combination Vehicles .............. CLW Lane Load ... Load Model 1 ............. ............4 Normal Distribution Function on the Normal Probability Paper.......... CLW Truck . OHBD Truck and Lane Load............................3 Reliability index defined as the shortest distance in the space of reduced variables ............................ BS 5400 Live Loading curve HA UDL [2006] .............. 12 Figure 2.......................... ....... 46 viii ........................15............. ........................3.. 13 Figure 2.......... with IM........ FHWA 13category scheme ...........................................................2 PDF’s of load...... 41 Figure 4...6...2....... CAN/CSAS600 Live Loading [2000].......... 17 Figure 2........... w/o multilane factors...............14................ OHBD Live Loading [1991].................... w/o IM..................... w/o multilane factors... with IM...12............... .... 15 Figure 2.. 16 Figure 2.... ..........
... Interstate semitrailer WB20 (AASHTO Geometric Design of Highways and Streets) ...18.... Laman... 67 Figure 5..... 62 Figure 5...... assumed designed UDL of 1..........F. Clearance ...... 69 Figure 5........... 72 Figure 5...... M.......... SJ...................... Freight Tonnage Moved by Truck (FHWA) .................................. 63 Figure 5....... Critical loading.........................................................25 k/ft ................14..............H...............................6................................ CDF's of GVW by axles Oregon I84 Emigrant Hill............................... Weekly maximum mean value of uniformly distributed load .......16......5............ Percentage of vehicles by number of axles...... A................................. Bias (mean max 75 year to nominal value of UDL) ...............85 k/ft .................... 1997) ....... FHWA WIM Data..15.... Traffic jam scenario....... (Kim. 66 Figure 5................. CDF’s of GVW for 5axle trucks........................................ J....11...........4...... CDF's of GVW by axles Oregon I5 Woodburn ...... H................... . Bias (mean max 75 year to nominal value of UDL) ............ 51 Figure 4.S...................................... H................................6.. CDF’s of GVW for all types of vehicles in Oregon.... and Nowak...... 67 Figure 5...... 73 Figure 5. 2006) .......2........................8.7........... 59 Figure 4.... HL93 proposed for long span bridges ...... and Nassif......7.... 74 Figure 5.... 49 Figure 4................Gap and Spacing ............................................... 74 Figure 5..................................... CDF of GVW for 5axle and 11axle trucks (Nowak......................................................... 64 Figure 5.......... 65 Figure 5...........10... 62 Figure 5.......10............S... 63 Figure 5...... Transportation statistics annual report...........12... Bias (mean max 75 year to nominal value of UDL) assumed designed UDL of 0......... WIM data collection .9...........8.................................. 73 Figure 5.............3...... 58 Figure 4........Headway Concepts ... ................................................. Time variation of total truck weight statistic.................. A....... 75 ix ... 60 Figure 5.................. 72 Figure 5... New York WIM Data............. Daily maximum mean value of uniformly distributed load ...................11........................ CDF's of GVW by axles NY I495 EB ......... December 2006....... ...... ........ Simulation of trucks moving throughout span length.. Percentage of vehicles by number of axles............... AASHTO LRFD legal load trucks.. Number of trucks by weight (in thousands of trucks).............. 48 Figure 4........5................................. 1994)................................. Mean value of uniformly distributed load ...... (Gindy....... Bias for heavily loaded localizations.......... 47 Figure 4........ A...... 64 Figure 5............................17.. .........13........ Sokolik........................................ . Nassif...............9............. Type 33 Units...........................Figure 4.......................1......
......... and ASCE (1981)....... Gindy and Nassif (2006)................ 85 Figure 6....................... Multilane load in design codes AASHTO LRFD Code (2007)................. OHBDC (1991)......... Fundamental frequency versus span length (Cantieni 1984) ....................................................................5........................ 80 Figure 6.......... 99 Figure 7. Proposed coefficient of variation of uniformly distributed load .......................87 Figure 6... 79 Figure 6.................................... time: 00:00:15 . Effects of vehicle suspension on the measured bridge response (Biggs and Suer 1955) ................. Multilane load in Eurocode 1. Gindy and Nassif (2006).... Fundamental frequency versus span length (Paultre 1992) .................9........................................................ and Huang 1993) .......... 101 Figure 7.... Coefficient of variation of weekly maximum uniformly distributed load ............................................. Plot from ABAQUS.....20............................... Bending modes .. 102 x .... time: 00:18:36 ...... Meshed model of the bridge......... Traffic loading pattern used for multiple truck presence statistics.................................. time: 00:05:28 ..........21. Vehiclebridge interacting force.7....................................................... ... Variation of multiple truck presence statistics with respect to truck volume............................. Variation of multiple truck presence statistics with respect to bridge span length.......................... Impact versus span length (Fleming and Romualdi 1961) .....10...........................................................6.....3............................................................................................ Video 1....... 87 Figure 6............. Multilane load in actual observation.... Video 10. CAN/CSAS600 [2000]......... .. 101 Figure 7.........5. Video 1. 87 Figure 7.......4. 81 Figure 6............ time: 00:00:16 ......1.Figure 5.................... 86 Figure 6.. .............................7. 76 Figure 6...................................... 99 Figure 7... .6............... 90 Figure 7..............................2...... Impact versus span length (Cantieni 1984)...................... 76 Figure 5.......... ....... Shahawy..............................3... Video 2........................................................ 91 Figure 7......................... .. Video 8..............2.................... ........... 84 Figure 6.....83 Figure 6..................................................11.............................................................1...... 93 Figure 7............... 91 Figure 7.....10.................. Impact factor versus vehicle speed and road surface condition (Wang........................... time: 00:00:58...........9............... Distribution of fundamental bridge frequencies (Cantieni 1984)................................. 95 Figure 7........ Truck model in FEM ................ Coefficient of variation of daily maximum uniformly distributed load ....19..........8..........................................11.4.......... 96 Figure 7......................8.. Force due to moving truck versus time.. .12.. 84 Figure 6........ 75 Figure 5...................... 96 Figure 7............
........................................... 121 Figure 8.4 End of compression block of concrete in the web...................................... the end of the compression control zone ... 118 Figure 8........................... 127 Figure 8................................. 123 Figure 8.............18 Force and Moment results on the bridge tower for different live loads....... ...... Geometry of Cooper River Bridge ............15....16 Recommended material parameters for reinforcing steel bars...................................................9........5 End of compression block of concrete in the top flange........ 109 Figure 8.. ....... 115 Figure 8..........1 Distribution of strains for the pure axial loading .... .... 104 Figure 7...........................8.................................... 124 Figure 8................ 111 Figure 8.....................7..... 110 Figure 8.. Torsion modes.. Maximal deflection due to moving truck ........................................................... 112 Figure 8........Figure 7.. 103 Figure 7...14......... 111 Figure 8....... 102 Figure 7.......... 117 Figure 8.......80 k/ft . 114 Figure 8.............13 Bias factor for compressive strength of concrete ...15 CDF’s of yield strength for Reinforcing Steel Bars.........16......... Grade 60 ksi ....6 Stress .. 122 Figure 8.....................3 End of compression block of concrete in the bottom flange ........Strain Relationship for Reinforcing Steel .........10.....12................. Load combinations ........................ 125 Figure 8.............64k/ft .....11....... Grade 60 ksi ........................2 Distribution of strains distribution for the balance failure point B...19 Reliability indexes due to different live load ........ Envelope of bending moments for bridge tower for w=0................... 109 Figure 8.... Envelope of bending moments for bridge tower for w=1........14 Coefficient of variation for compressive strength of concrete ..... Deflection due to a truck moving at crawling speed versus time......20 k/ft ........................ Envelope of bending moments for bridge tower for w=0......................................................... 128 xi ................................. 104 Figure 8....................................................13........ 119 Figure 8..... 116 Figure 8................................17 Statistical Parameters of Resistance .... Deflection due to a truck moving 40mil/hr versus time... ......00 k/ft ........... ... Envelope of bending moments for bridge tower for w=1.................................................................
.........3..........................6................................ Values of Equivalent Factored Loads...... 71 Table 6............ 57 Table 5....................... Values of Equivalent Unfactored Loads..........................5..... 27 Table 2.....4.......................... 22 Table 2................. 54 Table 4............ Multilane Reduction Factors for BS 5400 .............1. ......... w/o multilane factors...... Summary of simulated data ...................... 22 Table 2...... with IM.......................................... road width in OHBDC [1991] ..... w/o multilane factors............ Vehicles by axle in Oregon ...... Values of Equivalent Factored Loads.................... w/o multilane factors................................................ 20 Table 2........... Values of Equivalent Unfactored Loads.... w/o IM............................... w/o multilane factors..... 56 Table 4...12.......... Vehicles by traffic lane in Oregon .......................... Characteristic values of load for successive road lanes ......................... 40 Table 4...........................2........... with IM........ 53 Table 4.. 14 Table 2.........................................2........................... Vehicles by traffic lane in Indiana ..... Vehicles by axle in New York ............7............... w/o IM...8.......... Vehicles by traffic lane in New York....... Relationship between vertical scale on Normal Probability Paper and Probability ....... Vehicles by traffic lane in Florida .......... 15 Table 2............ with multilane factors for 4 traffic lanes................. 69 Table 5...... ........ 54 Table 4...........................2. 78 xii ...............................................9.... Summary of WIM Data ... 28 Table 3...1.................................. Number of design lanes vs......1.....................4......... 55 Table 4.......3..... Dynamic allowance in CAN/CSAS600 .. 56 Table 4.................. 28 Table 2...9....... Presence of multiple trucks and their location on the road lanes .......10.......LIST OF TABLES Table 2....................... 21 Table 2............ 44 Table 4................... 57 Table 4.................. Number of design lanes vs.....................11.......................... .............3.. 18 Table 2.................................. 27 Table 2......................... Comparison of Multilane Reduction Factors......... 27 Table 2............... road width in CAN/CSAS600 ...................................7...................................................5...... Heaviest truck combinations for 600 ft on I495 WB .1.......... Vehicles by axle in Florida ............. Vehicles by axle in Indiana ........ Values of Equivalent Factored Loads...... 55 Table 4................. Statistical parameter for proposed uniformly distributes live load ................10...................................... Dynamic allowance in AASHTO LRFD [2007] ...8........................................1...... 70 Table 5................ Conversion chart for vehicles’ class and number of axles ................6......
.............................................. 2008) ..2 Statistical Parameters of Fabrication Factor.. et al..........................1 Recommended Statistical Parameters for Compressive Strength.....Table 8... 122 Table 8. ....................S.......................... f c ' (Nowak A................ 124 xiii .................
a long span live load must include the possibility of multiple trucks being present.1. PROBLEM STATEMENT The live load models available were developed for short and medium span bridges. Therefore. Nowadays. the live load model for short and medium span bridges was developed based on a set of truck weight and load effect statistics that were presumed to be valid for any typical bridge site in the U. Observing traffic statistics helps to realize the rate of those changes. with heavy truck traffic increasing 550%. their importance. The continuous increase in the number of the trucks and their weights led to a review of traffic data for live load. In the last 30 years. During the AASHTO LRD calibration. The developed live load model is valid for spans between 600 ft and 5000 ft.S. and to draw some conclusions regarding design. particularly if it is to their economic advantage. the number of the vehicle miles logged annually on American highways has increased 225%. Some percentage of trucks runs overweight. The live load model may not represent the actual loading conditions at a particular bridge site or bridges in a state. In contrast to short and medium spans.CHAPTER 1 INTRODUCTION 1. several states are using WeighInMotion (WIM) systems to collect vast 1 . This doctoral dissertation deals with the development of a live load model for long span structures. a new live load model for long span bridges had to be developed and it had to be based on the newest traffic data obtained from highway and bridge administrators.
This could allow individual states to adjust the AASHTO live load factors to take into consideration the particular truck traffic conditions throughout a state. the uniformly distributed load of 0. The newest available traffic database from a variety of sites within many different states is used. OBJECTIVE AND BENEFITS OF THE STUDY The objective in this study was to develop a live load model for long span bridges. site specific models depending on average daily truck traffic and participation of heavily loaded vehicles seem to be more practical. a new approach to model uniformly distributed load had to be developed and new value of uniformly distributed load had to be proposed. In accordance with the stated objective. the first stage was to study previous research and current international codes’ provisions on the topic. As an addition to truck load. The model is valid for spans between 600 ft and 5000 ft. Reliability analysis is used to verify the developed live load model. The second stage of the 2 . Review of those topics was necessary. live load was modeled as an HS20 truck. Current multilane reduction factors and dynamic allowance also may not be appropriate for long span bridges. It is still used in the current AASHTO LRFD Code as it was in 1944. 1. a region. The derivation of uniformly distributed load is not clear. the uniformly distributed load has never been updated. The concentrated load was substituted with three axial forces representing a truck. In contrast.amounts of truck weight and traffic data that can be used to obtain sitespecific and statespecific live load models for bridge design and load capacity evaluation. To amend this. As a result. Traffic varies for different sites within each state. Based on the analysis of traffic records (weighinmotion and videos) the design live load is developed and recommended to be taken into consideration in the bridge design code. Since early publications by the American Association of State Highway Officials (AASHO). Sitespecific or statespecific live load models may be developed based on actual truck weight and traffic data collected at the site or within the state. or for a particular route. quantities of trucks and their weights. Since then the original definition of HS20 has been changed.64 kip/ft was introduced in 1944.2. It is intended to reflect current traffic patterns.
The data obtained had to be analyzed and filtered out from erroneous readings of measurement instruments. the bridge component and a limit state function that are the most influenced by live load were selected. Reliability analysis was performed in order to assess how the increase in live load influences reliability indexes. Statistical parameters for live load (bias and coefficient of variation) are derived. 3 . Relationship between site characteristics (ADTT. fabrication and professional factors. New uniformly distributed load was proposed. A derivation of uniformly distributed load from WIM data required developing a numerical procedure of calculation to process the extensive database. new statistical parameters for uniformly distributed load were used and statistical parameters of resistance were derived based on the newest material. For the scope of this study. The problems of multilane reduction factors and the dynamic factor were also discussed. The value of the new live load is based on three models: an average 5axle truck. legal load trucks and simulation of a traffic jam using WIM data. The outcome of this research is the recommendation of a live load model for long span bridges. there is a recommendation for an increase of the live load model for the long span bridges in heavily loaded urban and industrial areas. as well as for maximum daily and maximum weekly uniformly distributed load. The final step of this dissertation was reliability analysis. The calculations were performed for the current AASHTO LRFD design live load and increased load values obtained from real traffic data. The magnitude of the database obtained for the scope of this research has to be underlined. Most of the previous studies were based on measurements from limited numbers of sites within one state. Such an extensive actual weigh in motion database has never been used in the derivation of live load for long span bridges. percentage of overloaded loaded vehicles) and calculated values of uniformly distributed loads were studied.research was the collection of state of the art traffic data from highway and bridge administrators. In addition. An exemplary suspension bridge. Then a new uniformly distributed load was derived. Cumulative distribution functions were plotted for all data. There is a recommended value of uniformly distributed load for bridges carrying low and average ADTT.
In the Chapter 6. objectives and the scope of the study. Definitions of standard variables. It presents a problem statement. regulations of truck types. Uniformly distributed load for wide range of spans is calculated and compared. as well as the benefits and limitations of this research. New values of uniformly distributed load are derived and proposed to be applied in the code. truck sizes and weight limits. a model based on legal load trucks. as well as the weightinmotion technology and database to be used in this dissertation. the reliability procedures for a long span bridge are developed. Exemplary bridgevehicle interaction is modeled and the dynamic factor is derived. ORGANIZATION OF THE DISSERTATION Chapter 1 of this dissertation is an introduction to the research conducted. Reliability indexes for bridge loaded increasing values of live load are calculated. the problem of the presence of multiple trucks on a bridge is discussed. A short review of current studies.3. probability distributions. The chapter also includes a review of prior research on the topic. A model based on an average 5axle truck. The analysis is performed on a bridge tower of the selected suspension bridge. Chapter 5 describes development of the live load model. Methods of use of the normal probability paper and simulation techniques are described. New statistical parameters for live load are calculated. In the Chapter 8. Parameters affecting bridge dynamic response are discussed. Chapter 7 presents the problem of dynamic factor.1. and discussion on different approaches to the problem are presented. Chapter 2 reviews current international bridge design codes regarding live load. 4 . analysis of video recordings of traffic jam situations. Chapter 3 presents the principals of the reliability theory that were applied in this study. limit state functions and reliability index are introduced. dynamic load and multiple presence factors provisions. Chapter 4 describes the study of traffic data. and a model based on traffic jam simulation using WIM data are presented.
The deteriorating capacity of bridge was 5 . The WIM measurements were taken on seven bridges in Michigan. It has also been concluded that the bias factor (ratio of the mean to nominal value) is larger for smaller spans. and Nowak (1997) studied actual truck loads on selected bridges in the Detroit area. The maximum load effects for various time periods from one day to 75 years were derived from extrapolated distributions. recommendations are specified.2%. and Nassif (1994) published a research report on the effect of truck loading on bridges. The measurements were taken by using a weight in motion system. travelling sidebyside.1. 1. This was the vastest database available until now. Nowak and Hong (1991) formulated a procedure to calculate maximum moments and shears for various time periods. Kim. The researchers developed procedures for evaluation of live load spectra on steel girder bridges with regard to fatigue.4. and two following trucks govern for longer spans.4. It was observed that truck loads are strongly site specific. It was found out that maximum gross vehicle weight reaches a value of 225 kips and it shows a steady increase at an annual growth rate is 1. Gindy and Nassif (2006) formulated a similar conclusion based on data from New Jersey. Single and two lane bridges are considered. For one lane traffic a single truck governs for shorter spans. For two lanes of traffic. Laman. The maximum observed truck weights were up to 250 kips. the maximum effect is obtained for two trucks with fully correlated weights. Nowak. causing maximum moments two times larger than AASHTO load and resistance factor design values. Sokolik. As well. The observed truck weights were often heavier than legal limits. PRIOR INVESTIGATIONS 1. Most of the studies performed on live load models were based on truck data obtained within programs carried out by the Ontario Ministry of Transportation since the early 1970s.Chapter 9 presents the summary and conclusions of research performed for the scope of this dissertation. Prior Investigations on Live Loading on Short and Medium Span Bridges Live load models for short and medium span bridges were of interest to many researchers.
The loads are compared as equivalent uniformly distributed loads. Nassif and Nowak (1995). they have never been applied into the design codes. and that larger values of DLF are observed in exterior girders due to relatively smaller static load effect. which are 3050 percent larger than extreme values obtained at weight stations. They are known unofficially as the ASCE Loading. and 1991). Peter G. Prior Investigations on Live Loading on Long Span Bridges The most widely known researcher in the field of live loading on long span brides is Peter G. Buckland had also made a valuable observation regarding several loaded lanes. The load cases were distinguished depending on the percentage of “heavy vehicles”: 2. However. However. 7. 30. 1. where “heavy vehicles” are defined as trucks and buses over 12000 lb. in 1981. One of his findings was that uniform load per foot reduces as the load length is increased. 1980.4. by one set of uniform and concentrated loads. unlike many other studies he found out that concentrated load increases as the loaded length increases. It has been proved that WIM measurements show the unbiased truck weights. Buckland (1978. He concluded that traffic loading on long span bridges can be accurately represented in the traditional manner. Derivation of the dynamic load model is described by Hwang and Nowak (1991). Four uniform loading curves were developed for different loading cases. The WIM data is unbiased because the drivers are not aware of the measurements and they do not make an effort to avoid the scales. as trucks gravitate towards it.4. In the paper by Buckland (1991) the comparison of North American and British live loads on longspan bridges is presented. The WIM measurements from Michigan have also been used to study dynamic load. It was found out that the dynamic load factor decreases with increased static loads. He stated that if a single lane has a certain load on it. load in the additional lanes can be reduced.0. However.4. calculated as an equivalent shear and bending moment for 6 .evaluated as a function of the rate of corrosion.2. than the additional lanes would increase the load in the lane closer to the curb. or 100 percent. These loading curves were recommended by the ASCE Committee as vehicle loading of longspan bridges.
H. 1. Ang and Tang (1984). serviceability and durability. However. Rackwitz. reliability techniques have been increasingly used in modern design codes. O. R.simply supported beams.E. For many years the random nature of various parameters influencing structural safety has been of interest to engineers. This approach can be successfully used for short and medium span bridges. Nowak and Collins (2000) and Wolinski and Wrobel (2001).Faber. Prior Investigations on Structural Reliability The theory of structural reliability have been investigated and described by many researchers. 2000). Early publications that quantified and presented a mathematical formulation of structural safety problems were published by Mayer (1926) and Wierzbicki (1936). ThorfChristtensen and Baker (1982). it was not until the late 70’s when safety factors based on load and resistance uncertainties were proposed for introduction into the codes. Murzewski (1989). They recognized that load and resistance parameters have random characteristics.3. Nowak. Ang. Krenk and Lind (1986). and that each structure has a finite and limited probability of failure. H. However. since long span bridges cannot be constructed as simply supported beams. and P. ThorfChristtensen and Murotsu (1986). Madsen. it was the Ontario highway bridge design code (Nowak and Lind 1979). for instance. Melchers. and Ellingwood 1980. The application of reliability theory has resulted in the improvement of structural design in terms of safety. it was the building design code (Galambos and Ravindra 1978.S. its application to long spans can be questioned. Until they gathered more knowledge about the laws of nature. ThoftChristiansen. Ditlevesen.S. Their 7 . among them A. Several books and publications provide available knowledge regarding reliability theory. R. Since then. Ayyub and McCuen (1997). Nowadays many researchers keep working on further development of new methods of structural reliability analysis. In the United States. 1982). Mathematical theories available nowadays describe material and structural behavior sufficiently enough to give a rational basis for structural safety evaluations (Nowak and Collins. A. they used to assure structural safety through ‘trial and error’ and intuition. Nielsen.4. and in Canada. This method of deriving the equivalent load can be used exclusively for comparison of codes.
Neumann and Rackwitz (1979). The extensive development of practical tools and efficient methods for evaluating the probability of structural failure has been made in the last 30 decades. Nowak and Collins (2000). Advanced SORM have been elaborated by researchers such as Fiessler. It was first presented by researchers Augusti and Baratta (1973. Subsequently. Breitung (1984). Their approach estimated the limit state function at mean values of random parameters and used a linearized limit state function. Adhikari (2004) systemized and published all of the earlier proposed SORM procedures. the probability of failure was defined by multidimensional integral functions of distributions and it was cumbersome to evaluate. The extension of the Hasofer and Lind approach and the transformation of uncorrelated random variables of various distributions into standardized normal distributions were proposed by Rackwitz and Fiessler (1978).concept of a structural reliability problem has been subsequently adopted in the precursory publication for that field by Freudenthal (1956). 1973). which was presented in work by Kiureghian and Liu (1986). A milestone was the estimation of the probability of failure proposed by Hasofer and Lind (1974). Simulation techniques are another approach to 8 . Another commonly used transformation is Nataf’s transformation. The developed numerical procedure of the design point estimation used to be called the RackwitzFiessler procedure. The simplified procedure involved a nonlinear mathematical programming problem with boundary conditions (an estimated limit state for all variables and a defined probability in the so called “design point”). Pioneering studies on the first practical application of reliability analysis were performed by Cornell and Lind in the late 1960s and early 1970s. Hohenbichler and Rackwitz (1988) used the Rosenblatt (1952) transformation procedure for the transformation of dependent (correlated) random variables from and into standardized form. a new trend in using probabilistic concepts in the analysis of limit capacity and structural resistance was developed. Commonly used methods of reliability analysis are based on the approximation of the limit state function at the design point using first or second order functions (FORM and SORM). important work was done by Corotis and Nafday (1989) analyzing the limit capacity of beamcolumn frame structures using the principles of conditional probability. Initially. In the 1960s. which is currently one of the major tools used in modern reliability analysis.
Than Zhou (1987) developed an integration sampling technique to calculate the system reliability of a bridge. Estes and Frangopol (1999) assumed that failure occurs when failure occurs in three of five adjacent girders. Zhou and Nowak (1990). and Tabsh and Nowak (1991). Both Tantawi and Zhou. and a huge number of numerical simulations are required to achieve a high accuracy of results. Several major contributions were also made by researchers such as Ditlevsen (1982. nonacceptable deflection. 1982). Moses (1982) proposed incremental load approach and suggested a procedure for identifying collapse mode for both ductile and brittle components. Practical procedures for system reliability analysis were suggested by Nowak and Zhou (1990). Grigoriu discussed a parallel system with brittle elements. Bridges usually consist of a combination of series and parallel systems. 1982. Determination of the probability of structural failure with the use of simulation techniques has limited accuracy. 1983). and Rackwitz recognized the effect of correlation on system performance. Grigoriu (1982. The present requirements for civil engineering structures primarily focus on structural safety. Identification of collapse mode and degree of correlation between members is very difficult or often even impossible to evaluate. A collapse was defined as an excessive. 1996). The structural system reliability is a field of interest for many researchers. New structural design codes are calibrated using the limit state analysis 9 . Reliability models applied to bridge evaluation were addressed by Nowak and Tharmabala (1988). This method becomes very useful and an especially practical tool in cases where physical testing is expensive. Hart. found that bridge system reliability is higher than girder reliability. The most popular is the Monte Carlo Method simulation technique (ThoftChristensen and Baker. Moses and Verma (1987) used a load and resistance approach to evaluate the strength of bridges with reliability principles. The identification of collapse mode was also discussed by Rashedi and Moses (1988). Tantawi (1986) developed a grid nonlinear analysis program to calculate the momentcarrying capacity of a bridge. Ditlevsen used conditional probability to calculate bounds of the probability of failure. and Rackwitz (1985).estimating probability of failure. The procedures assumed that the ultimate load carrying capacity is equal to the weight of a truck which causes a collapse.
approach to assure safety standards and to provide the required reliability of new design structures. 10 . However. there is a major gap between the development of reliability techniques and their application to structural engineering design and evaluation. as the calculations have to be repeated many times. a more comprehensive method to assess the system reliability of a bridge needs to be developed. Therefore. The system reliability analysis requires an efficient structural analysis procedure.
Nonuniform loading curves are used in the British Standard 5400 (2006) and ASCE Recommended Design Loads for Bridges (1981). CAN/CSAS600 [2000]. INTRODUCTION While approaching the problem of live load for bridges it is necessary to review the current codes and perform a comparative analysis. The codes were selected with the objective to present various approaches to design live load.1. OHBDC (1991). The comparison of equivalent uniformly distributed loads for a variety of spans is presented in Paragraph 2. and ASCE (1981) were briefly summarized in the following paragraphs. The design live loads specified in AASHTO LRFD Code (2007). and Eurocode 1 (2002).CHAPTER 2 LIVE LOAD IN CURRENT DESIGN CODES 2. The live load for bridges can be represented in many ways. use of multilane factors. Eurocode [2002]. including a uniform load and a combination of truck(s).5. 11 . as for example in AASHTO LRFD Bridge Design Specifications (2007). CAN/CSAS600 Canadian Highway Bridge Design Code (2000). and dynamic impact allowance.
2.2. Figure 2. and uniformly distributed design lane load of 0. Tandem and Lane Load.64 klf.3. INTERNATIONAL PROVISIONS FOR LIVE LOADING 2. and uniformly distributed design lane load of 0.640 klf 4’ Figure 2. 32. AASHTO LRFD Code [2007] The American Association of State Highway and Transportation Officials Standard Specifications for Highway Bridges define HL93 live load as the extreme force effect taken as the larger of following: 1) The AASHTO LRFD 3axle design truck. Figure 2. 25. 2) The AASHTO LRFD Design Tandem.2. should be considered.64 klf.0 kip 0. combination of two design trucks spaced at minimum of 50 ft. All of the forces reduced to 90 %.640 klf 14’ 14’ .2. 3) For negative moment between points of contraflexure and reaction at interior piers. and uniformly distributed design lane load of 0.0 kip 32.1.1.64 klf.2. Truck and Lane Load.0 kip 8. 12 . Figure 2.0 kip 25.1.30’ Figure 2. HL93 Live Loading in AASHTO LRFD Code [2007].0 kip 0. HL93 Live Loading in AASHTO LRFD Code [2007].
whichever produces the maximum load effect: 1) The OHBD Truck. Figure 2.0 kip 36. 36.30’ Figure 2. 2) The OHBD Lane Load consists of an OHBD Truck with each axle reduced to 70%.5.8 kip 7.0 kN/m (0. 2.0 m (10 ft) wide uniformly distributed load of 10. OHBD Truck. OHBDC [1991] The Ontario Highway Bridge Design Code (Third edition) determines live load as a truck or a combination of truck and lane load. 13 . HL93 Live Loading in AASHTO LRFD Code [2007].0 kip 13. which is a 5axle truck.576 klf 14’ 14’ .4.2.4.685 kip/ft). OHBD Live Loading [1991]. The loads shall occupy 10 ft transversally within a design lane.8 kip 7. and superimposed centrally within the width of a 3.8 kip 0.3.8 kip 28. Figure 2. Alternative Load for Negative Moment between points of contraflexure and reaction at interior piers.0 kip 36.0 kip 12’ 4’ 20’ 24’ Figure 2.28.0 ft transversally.2 kip 28.5 kip 45.2 kip 28.30’ ≥ 50’ 14’ 14’ .2. Truck wheels are assumed to be spaced 6.
and a superimposed uniformly distributed lane load of 9.617 kip/ft).5 m < Wc ≤ 24.5 m 20. Figure 2. OHBD Live Loading [1991]. The number "W" indicates the gross load of the truck in kN.5 m < Wc ≤ 17. For the design of a national highway network.0 m (10 ft) wide.0 m 10. 14 . that is 3.0 m < Wc ≤ 10.4 kip 31.6. 2) The CLW Lane Load consists of CLW Truck with each axle reduced to 80%.2 kip 9. road width in OHBDC [1991] Width Wc ≤ 6.2 kip 25. Table 2.2.25.2 kip 0.0 m < Wc ≤ 13. CAN/CSAS600 [2000] The Canadian Highway Bridge Design Code applies CLW loading.685 kip/ft 12’ 4’ 20’ 24’ Figure 2.3.5 m 13.5.5 kip 25. loading not less than CL625 shall be used. Figure 2.0 m < Wc ≤ 20. OHBD Truck and Lane Load.0 m 17.0 kN/m (0.0 m < Wc ≤ 27.1.7.5 m < Wc Number of lanes 1 2 3 4 5 6 7 8 2.0 m 24. which consists of the truck or the lane load: 1) The CLW Truck is 5axle truck.0 m 6. Number of design lanes vs.5 m 27.
617 kip/ft 12’ 4’ 22’ 22’ Figure 2.2. Number of design lanes vs.1 kip 28.16 W) 9.2.0 m 17.224 W) 27.0 m < Wc ≤ 20.4.7.064 W) 31.5 m 13.5 m 20.5 m < Wc ≤ 17.28 W) 33.5 m 27.2 kip (0. CLW Lane Load Table 2.08 W) 28.0 m 6.5 kip (0.2 W) 39.0 kip (0.5 m < Wc ≤ 24. BS 5400 [2006] According to British Standard the structures and its elements shall be designed to resist the more severe effects of either design HA loading or design HA loading 15 .0 m < Wc ≤ 13.24 W) 12’ 4’ 22’ 22’ Figure 2.11.16 W) (0.6. road width in CAN/CSAS600 Width Wc ≤ 6. CAN/CSAS600 Live Loading [2000].2 W) (0.0 m 24.5 kip 22.192 W) 0.0 m < Wc ≤ 27.5 kip (0. CAN/CSAS600 Live Loading [2000].1 kip (0.0 m < Wc ≤ 10.0 m 10.5 m < Wc Number of lanes 1 2 2 or 3 (check both) 4 5 6 7 8 2. CLW Truck 22.3 kip (0.7 kip (0.0 kip (0.
the UDL shall be in agree with the relevant authority. Both loadings include impact.1 [kN] • for loaded lengths above 1600 m.combined with design HB loading. HA loading consists of uniformly distributed load (UDL).1 N/mm2. or a singlewheel load. HB loading is an abnormal vehicle unit loading.67 [kN] • for loaded lengths in excess of 50 m but less than 160 0m: W = 36 (1/L) 0.5 kip) wheel placed on a carriageway and uniform distribution over a circular contact area assuming an effective pressure of 1. Figure 2. Value of nominal uniformly distributed load (UDL) equal to: • for loaded lengths up to and including 50 m: W = 336 (1/L) 0. The nominal knife edge load (KEL) per lane shall be taken as 120 kN (27 kip).8.8. BS 5400 Live Loading curve HA UDL [2006] 16 . The single nominal wheel load alternative to UDL and KEL is one 100 kN (22. Live Loading curve HA UDL is shown in the Figure 2. and a knife edge load (KEL). HA loading represents normal traffic in Great Britain.
One axle is 10 kN (2.HB loading defines the minimum number of HB loading units that shall be considered. This number may be increased up to 45 if so directed by the relevant authority. Dimensions of HB vehicle. The overall length of the HB vehicle shall be taken as 10. It presents four models for determining the main vertical loads from traffic: 1) Load Model 1 (LM1) consists of concentrated and uniformly distributed loads (Figure 2.e. or 30 m for inner axle spacing of 6.9 below shows the plan and axle arrangement for one unit of HB loading.5. The code specifies live load to be used on each traffic lane. 2. 11. 2.25 kip). therefore there is no need to introduce multilane factors. The effects of the most severe case shall be adopted. or 26 m respectively. 20.2. It is used for general and local verifications. Eurocode 1 [2002] The Eurocode 1 Part 2 is applicable to bridges with spans from 5 to 200 m (17 to 667 ft). 25. and carriageway width up to 42 m (140 ft). Figure 2.56 kip) per wheel.9. Figure 2.10) which cover most of the effects of the traffic of trucks and cars. 17 . i. 15. 21. which is 30. 16.5 kN (0.
052 kip/ft2) 2.052 kip/ft2) Figure 2.5 kip) 200 kN (45. The distance between wheels is 2 m.188 kip/ft2) 2. 3) Load Model 3 (LM3) consists of sets of axle loads representing special (carrying heavy loads) vehicles.052 kip/ft2) 2.5 kip) 0 0 UDL qik 9 kN/m2 (0. which covers the dynamic effects of the normal traffic on short structural members.35 m and 0. When relevant.5 kN/m2 (0. The contact surface of each wheel should be taken as a rectangle of sides 0.60 m.052 kip/ft2) 2. which can travel on routes permitted for abnormal loads. only one wheel of 200 kN may be taken into account.10. Load Model 1 2) Load Model 2 (LM2) consists of a single axle load of 400 kN.5 kN/m2 (0. It is intended for general and local verifications.Table 2. 18 .5 kN/m2 (0. Eurocode 1 [2002]. Characteristic values of load for successive road lanes Location Lane number 1 Lane number 2 Lane number 3 Other lanes Remaining area Axle load Qik 300 kN (67.3.0 kip) 100 kN (22.5 kN/m2 (0.
5%.2 U (7.74 P U 2 U(100% HV ) 158. ASCE Loading on Log Scale 19 . 2.11) to give moments and shears with a sufficient degree of accuracy.11.0 kN/m2. It is intended only for general verifications and it is particularly relevant for bridges in or near town areas.5% HV ) U (30% HV ) 0 0 50 100 200 400 800 1600 3200 6400 Loaded Length (ft) Figure 2. It has been proved that the loading can be represented by a uniform load and a concentrated load (Figure 2. ASCE Loading [1981] The live load known unofficially as the ASCE Loading is a result of the studies performed by Peter G.4) Load Model 4 (LM4) represents crowd loading of 5. and 100% heavy vehicles of the total vehicle population. Buckland. 2.2. "Heavy vehicles" were defined as buses and trucks over 12 000 lbs. 30%.6. ASCE (1981) specifies three levels of live load for highway bridges depending on the average percentage of heavy vehicles in traffic flow: 7. which was recommended for long span bridge by the American Society of Civil Engineers Committee on Loads and Forces on Bridges.4 P (kips) 56.6 U (k/ft) P 1 102.
other limit states Dynamic allowance.2.3. The most common approach is to apply dynamic response as a fraction or multiple of the response that would be obtained if the same forces or loads were applied statically. 20 . all limit states All other components. Dynamic allowance in AASHTO LRFD [2007] Component Deck joints. PROVISIONS FOR DYNAMIC LOAD FACTOR There is considerable variation in the treatment of dynamic load effects by bridge design codes in different countries. The Canadian Highway Bridge Design Code CAN/CSAS600 [2000] and OHBDC [1991] apply dynamic allowance as a percentage to static effects of the CLW Truck for the number of axles considered in the design lane. The American Association of State Highway and Transportation Officials Standard Specifications for Highway Bridges AASHTO LRFD [2007] define that dynamic load allowance shall be applied to static load effects of the truck or tandem. The objective of this simple approach is to not increase complexity to the designer.2. as a percentage specified in the table below.). Table 2. fatigue and fracture limit states All other components. as shown in the table below.4. It shall not be applied to pedestrian loads or to the design lane load. IM 75% 15% 33% Until recently AASHTO Standard Specifications for Highway Bridges used to specify dynamic load effects in terms of an impact factor that is a function of bridge span. However relation between dynamic load factor and bridge span is a controversial issue between researchers (see Paragraph 7.
the multiple lane reduction factors are specified in most of the codes. Dynamic allowance in CAN/CSAS600 for deck joints 1 axle of CLW Truck 2 axles of CLW Truck 3 or more axle of CLW Truck 0.40 0. PROVISIONS FOR MULTILANE REDUCTION FACTORS For multilane bridges. which depend not only on the number of lanes. number and width of notional lanes (Table 2. Their effect has already been taken into consideration in definition of design loading.30 0.6 and Table 2.5. The multilane reduction factors are further discussed in CHAPTER 6 of this dissertation. The approaches to multilane factors vary significantly. Eurocode does not define multilane reduction factor.50 0.Table 2. The British Standard BS 5400 developed the most compound procedure of selection of multilane factor. but it gives the load values to be applied on successive road lanes directly (Table 2. 21 . 2.4. but also on loaded length. They are shown in Table 2.3). The ASCE model (1981) does not have any allowance for dynamic load on the ground that the worst loading occurs with stationary bumpertobumper traffic.7.7).25 According to British Standard BS 5400 [2006] and the Eurocode 1 [2002] the effects of vibration due to live load are not required to be considered.
1 α1 α2 1. N≥6 L>112.Table 2.55 0.65 0.70 0.40 6 or more 0.274 bL α2 = 0.0 0. For a bridge carrying oneway traffic only.70 0.70 3 0.6 0.40 Table 2.40 4 0.0 2 α1 α2 1.0 1. N<6 50<L≤112.7.67 1.0 1.65(L20)} where bL is the notional lane width N is total number of notional lanes on the bridge.80 0.0137{bL(40L)+3.00 0.0 1.40 5 0.0 1.6 0. 1991) CAN/CSAS600 (2000) ASCE (1981) 1.0 7.55 0.6 α2 0.6 0.6 0.6.6 α1 0.00 1.85 0.65 0.00 1.90 0.00 2 1.65 0.20 1.90 0.80 0. Multilane Reduction Factors for BS 5400 Number of Lanes loaded length [m] 0<L≤20 20<L≤40 40<L≤50 50<L≤112.6 0.6 0.6 0.6 0. Comparison of Multilane Reduction Factors Number of Lanes Code 1 AASHTO LRFD (2007) OHBDC (1983.0 3 0.6 22 .1/√L 1. N<6 L>112.6 0.60 0.60 0.6 4 or more 0. the value N shall be multiplied by 2.6 0. N≥6 α1 = 0.
Figure 2. Equivalent loads including dynamic loads are shown in Figure 2. for BS 5400 it is 1.15.80 has been used by (Buckland 1991). they have been used only for comparison of loads including dynamic load.16. and for Eurocode it is 1.14 compare the equivalent unfactored uniform loads. DLF.70. The comparison of four loaded lanes shows that the differences are reduced even more. but since a factor of 1. Then.5. the same value has been adopted for this comparison. ASCE studies made no reference to load factors to be used with its recommended loading. It should be noted that most of the codes are limited to certain lengths of span. and they are extrapolated beyond these points partially for interest and partially because these codes are also occasionally used for longer spans due to a lack of adequate guides.13 and Figure 2. The design loads are increased by the dynamic load factor. European values double those of North America. 23 .12 and Figure 2.40. the maximum bending moment (Mmax) was calculated for simple spans from 400 through 5000 ft.1) where: L is span length.35. More valid comparison is to compare factored loads.75. for CAN/CSAS600 it is 1. COMPARISON OF EQUIVALENT UNIFORMLY DISTRIBUTED LOADS In this paragraph. Figure 2. To obtain plots of UDL. and Figure 2. The application of load factors slightly reduces the differences. Results of the comparison show that variation between the unfactored values of live load in different codes is significant. the resulting equivalent uniformly distributed loads for a variety of design codes and span lengths are plotted and compared.2. Figure 2. the equivalent uniformly distributed load UDL was determined from the following formula: UDL = 8 ⋅ ( M max ) / L2 (2. For AASHTO LRFD (2007) live load factor is 1.16.50. The importance of load factors and multilane factors cannot be underestimated.15. these are shown in Figure 2. which has a value as described for each code in the paragraphs above.14. Since live loadings in British Standards and Eurocode include dynamic load. for OHBDC (1991) it is 1.
5%HV ASCE 30% HV ASCE 100% HV equivalent UDL [kip/ft] 1. Equivalent Unfactored Loads.00 0 1000 2000 3000 4000 5000 loaded length [ft] 2.00 AASHTO HL93 OHBDC 1991 CAN/CSAS600 1. 24 .2. w/o multilane factors.00 AASHTO HL93 OHBDC 1991 CAN/CSAS600 1.50 0.00 equivalent UDL [kip/ft] 0.00 0.12. w/o IM.50 0.50 ASCE 7.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 2.50 ASCE 7.5%HV ASCE 30% HV ASCE 100% HV 1.
00 ASCE 30% HV ASCE 100% HV BS 5400 HA loading 1.5%HV 2.80 1. 25 .75 OHBDC 1991 x 1.50 0. w/o multilane factors.00 ASCE 100% HV x 1.00 0.80 equivalent UDL [kip/ft] 2.50 Eurocode LM1 equivalent UDL [kip/ft] 1. Equivalent Factored Loads.00 0. w/o IM.40 CAN/CSAS600 x 1.13. w/o multilane factors.50 ASCE 7.50 0.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 2.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 2.3. Equivalent Unfactored Loads. with IM. 2.5%HV x 1.00 AASHTO HL93 x 1.14.80 ASCE 30% HV x 1.70 2.50 AASHTO HL93 OHBDC 1991 CAN/CSAS600 ASCE 7.50 1.
with IM.00 5.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 2. Equivalent Factored Loads.00 AASHTO HL93 x 1.75 OHBDC 1991 x 1.50 CAN/CSAS600 x 1.16. Equivalent Factored Loads.00 2.00 0.35 1.40 CAN/CSAS600 x 1.00 equivalent UDL [kip/ft] 4. 26 .00 ASCE 100% HV x 1.70 ASCE 7.50 1. 6.00 0.75 OHBDC 1991 x 1.70 BS 5400 HA loading x 1.3.5%HV x 1.00 3.15.40 2. with IM. with multilane factors for 4 traffic lanes.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 2.50 Eurocode LM1 x 1.00 AASHTO HL93 x 1.50 0.35 1.80 ASCE 30% HV x 1.80 equivalent UDL [kip/ft] 2. w/o multilane factors.50 Eurocode LM1 x 1.80 BS 5400 HA loading x 1.
802 0.896 1.158 1.673 HL93 1.303 1.898 0.900 0.767 0.10.246 1. w/o multilane factors.763 0.023 0.717 0.905 1.612 1.688 0.832 0.692 0.819 0.123 1.375 1.204 1.679 0.985 1.827 0.683 0.268 1.841 0.240 1.662 HL93 0. w/o IM.736 0.041 1.729 0.125 HL93 1.8.134 1. w/o IM.698 0.151 0. w/o multilane factors.768 0.179 0.228 1.372 1. with IM.941 1. length [ft] 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 OHBDC 1991 1.681 0.881 1.603 1. Values of Equivalent Factored Loads.032 1.202 1.948 0.053 1.784 0.842 0.813 1.732 1.804 0.757 0.695 0.177 1.170 Table 2.Table 2.120 1.676 0.736 0.176 1.729 0.681 0.870 0.624 1. Values of Equivalent Unfactored Loads.294 1.744 0.737 0.045 0. length [ft] 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 OHBDC 1991 CAN/CSAS600 1.288 1.672 0.918 1.025 CAN/CSAS600 1.192 1.688 0.687 0. w/o multilane factors.144 1.336 1.221 1. Values of Equivalent Unfactored Loads.328 1.779 0.851 0.697 0.176 1.878 27 .712 0.889 1.884 1.449 1.673 0.812 CAN/CSAS600 1.707 0.704 0.669 Table 2.431 1.246 1.090 1.928 0.837 0.198 Eurocode 2.752 0.667 0.068 1.918 0.286 1.183 1.9.711 0.067 0.212 1. length [ft] 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 OHBDC 1991 1.678 BS 5400 1.
828 3.526 3.241 3.660 4.208 1.168 1.288 1. with multilane factors for 4 traffic lanes.716 3.404 2.382 3.097 3.063 1.364 4. with IM.605 3.862 2.818 1.185 1.187 BS 5400 2.455 1.Table 2.842 1.12.309 4.564 4.771 5.240 1.902 1.993 1.343 1.184 CAN/CSAS600 5.173 2.259 1.161 3.137 CAN/CSAS600 2.559 2.612 4.172 1.274 3. Values of Equivalent Factored Loads.653 28 .571 2.945 4. length [ft] 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 OHBDC 1991 5.783 3.216 1.219 1.493 3.144 HL93 1.348 3.232 1.550 2.702 4.336 3.412 3.762 4.782 4.204 HL93 4.486 4.871 1.869 1.471 3.106 3.11.254 1.544 2.539 2. OHBDC 1991 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1. w/o multilane factors.726 4.463 1.004 1.318 3.086 BS 5400 5.204 1.823 4.680 2.202 3.288 1.216 4.589 2.661 4.191 1.671 4.684 4.157 1.655 3.194 1.620 2.420 4.238 4.234 3.261 3.952 4.314 Eurocode 5.146 1.367 1.942 1.327 1.155 1.790 1.282 3.526 1.271 3.535 Table 2.130 3.209 3.797 Eurocode 2. Values of Equivalent Factored Loads.
They include uncertainties coming from mechanical material properties. which would assure their good performance to account for actions applied during construction and service. and agreement between realinput data and a mathematical model of phenomenon. The actions (loads. Therefore.CHAPTER 3 STRUCTURAL RELIABILITY PROCEDURES 3. R ) are the variables that decisively influence the state of a structure. failure or other deficiency in a structural resistance. which when exceeded lead to structural failure (ultimate limit states) or make use of the structure impossible (serviceability limit states). loads. geometry of a structure. Unreliability of a structure is a state in which a structure does not fulfill design requirements related to its function and desirable performance. For this purpose civil engineering uses a probabilistic evaluation of reliability. the design of structures is a process in which decisions are made under uncertainty and limits. INTRODUCTION The structures and their components should be designed to have a desirable level of reliability.1. It could be a collapse of a structure. is a concern of structural reliability. Q ) and structural resistance (capacity. Those uncertainties can be measured only with the use of probability. etc. The design of new structures as well as the evaluation of existing structures requires verification of limit states. Their rational treatment. unfulfilled service demands 29 .
Mean The mean (expected value) is an average of all observations on a random variable. For the continuous random variables. All modes must be treated separately.2) If all n observations are given equal weights ( PX ( xi ) = 1 / n ). It is also defined as the first moment about the origin. Therefore. the mean ( µ ) can be computed as: +∞ µ = ∫x⋅ n f X ( x ) dx (3. etc. then the mean for a discrete random variable is given by: 30 . the mean is given by: µ = ∑ xi ⋅ PX ( xi ) i =1 (3. There are two types of random variables: discrete and continuous. f X (x) . The analysis usually includes an estimation of structural reliability with respect to specified failure modes. For most of the structures it is impossible to examine all their failure modes. PX ( xi ) . reliability of a structure is the probability that the system will not reach a specified failure mode related to a specified limit state during a specified period of time. and its probability is defined by the probability density function (PDF). Thus. A continuous random variable can take on a continuous range of values. A distribution function would complete the description of the probabilistic characteristics of random variables. Structures usually have a number of possible failure scenarios. A discrete random variable may take on only discrete values.of a structure. variance and standard deviation. Its probability is given by the probability mass function.2. STANDARD VARIABLES AND PROBABILITY DISTRIBUTIONS The key probabilistic characteristics of a random variable are described in terms of mean. representative failure scenarios have to be chosen. excessive vibrations.1) −∞ For the discrete random variables. 3. excessive deformations.e. i. but sometimes it remains unknown.
X= Variance 1 n ⋅ ∑ xi n i =1 (3.6) Standard Deviation The standard deviation ( σ 2 ) of a probability distribution is defined as the square root of the variance. exponential. because this is the only distribution used in this dissertation.7) Probability distributions There are many types of discrete and continuous distributions.3) The variance ( σ 2 ) is the second moment about the mean. Coefficient of Variation The coefficient of variation ( V ) is a dimensionless quantity defined as: V = σ µ (3. Normal distribution (Gaussian distribution) is the most widely used probability distribution. the variance is as follows: σ2 = n 1 ⋅ ∑ ( xi − X ) 2 n − 1 i =1 (3. the variance is computed as: σ 2 = ∑ ( xi − µ ) 2 ⋅ PX ( xi ) i =1 n (3. In this section the normal distribution is presented.5) If all n observations are given equal weights ( PX ( xi ) = 1 / n ). and it is computed as follows: +∞ σ 2 = ∫ ( x − µ ) 2 ⋅ f X ( x)dx −∞ (3. and gamma. Further details about distributions can be found. for instance.4) For the discrete variable. It has a probability density function given by: 31 . The most commonly used are the continuous distributions: uniform. lognormal. in Nowak and Collins (2000). normal.
and they are defined as follows: φ ( z) = Φ( z ) = 1 ⋅ exp(− ⋅ z 2 ) = f Z ( z ) 2 2 ⋅π 1 z (3.1. without a dominating distribution type. Its PDF is denoted as φ (z) .f X ( x) = 1 x−µ 2 ⋅ exp[ − ⋅ ( ) ] 2 σ σ ⋅ 2⋅π 1 (3. approaches an approximate normal distribution. its CDF is denoted as Φ(z) .1 PDF and CDF of a normal random variable The standard normal distribution is a special case of Gaussian distribution. FX(x) random variable X Figure 3. which can be approximated as a normal variable.10) Central limit theorem states that the sum of a large number of independent observations. 32 .9) ∫ −∞ 1 ⋅ exp(− ⋅ z 2 )dz 2 2⋅π 1 (3. The higher the number of observations the better is the approximation.0 .8) This function is graphically represented as shown in Figure 3. This theorem is one of the most important in probability theory. The sum of variables is often used to model total load acting on a structure. with parameters µ X = 0 and σ X = 1.
. Therefore. The probability of an undesired performance of a structure is equal to the probability of failure ( P f ). Three types of limit states are typically used with reference to structural reliability analysis: 1. 2.11) n µY = µ X + µ X + . Ultimate limit states (ULSs). resistance). + µ X 1 2 (3. + σ X n (3. between situations when the structure is safe (a safety margin exists) and the structure is not safe (failure occurs).. + X n (3...Mathematically it can be expressed that the sum of n random variables. LIMIT STATE FUNCTION In most design codes. which represents failure due to deterioration of functionality. which represents the loss of strength for a structural component under the action of repeated loading. The philosophy of limit state design assumes equilibrium between applied loads and structural response of the structure (capacity. which represents the loss of structural capacity.. a specified set of load and resistance factors is required for each limit state formulated for different possible scenarios of structural behavior during construction as well as service life.. The probability of the desired performance of a structure is equal to the safety margin ( PS ). the structural design is based on the concept of limit states.3.. Fatigue limit states (FLSs).12) 2 σY 2 =σ X 2 +σ X 1 2 2 + . X 1 . X 2 ..13) 3. is equal to function Y having normal distribution: Y = X 1 + X 2 + . Serviceability limit states (SLSs). the probability of occurrence is P ( Ω ) = 1 . The limit state defines the boundary between the desired and undesired performance of a structure. X n . so: 33 .. 3. Therefore. Failure and nonfailure states fulfill the entire probabilistic sample space ( Ω )..
safety margin Q.P ( Ω ) = P f + Ps = 1 ⇒ Ps = 1 − P f (3. safety Q. load effect RQ. and safety margin. Figure 58 PDFs of load. Q) = R − Q PDF R. resistance. load effect R. resistance. resistance. In the general case. resistance. resistance. resistance Resistance 58 PDFs of load. resistance margin PDF margin g (R. All loads are being incorporated into one variable ( Q ) and the resistance of the structure is being incorporated into one variable ( R ). X 2 .Q Q 0 0 0 R Figure 58 PDFsofload. and safety margin. X 2 . and safety margin. X n ). resistance PDF the performance function of a system can be related to any possible failure scenario or (3. 0 Fig random value 00 Figure 58 PDFs of load. The quantity R − Q is also Probability a random variable with of Failure its own PDF as shown Probability Probability of Failure 0 of Failure ofFailure of Failure Probability in the Figure 3. resistance R. safety PDF Q. PDF PDF PDF PDF RQ.. margin Probability of Failure Both R and Q are continuous random variables having a probability density Probability of Failure Probability function (PDF). safety Q. safety Q. safety margin Q. load effect RQ. and safety margin. and safety margin. safety Q. 34 . and safety ma Figure 58 PD 0 Probability Probability Probability Failure ofofFailure Probability ofFailure of Failure Figure 58 PDFs of load. load effect R.. load effect R. safety RQ.16) where X i is the collection of input parameters. load effect R PDF RQ. resistance. safety or R is capacity representing resistance of a structural system margin R. Figure 3. Figure 58 PDFs of load. and safety margin.. load margin RQ.2.. and safety margin Figure 58 PDFs of load.2 PDF’s of load. resistance PDF PDF Probability of Failure any limiting state and defined as a function of capacity and demand: where and Q RQ. safety Q. margin RQ. PDF Q. safety Safety margin Q. X n ) (3.. resistanceR. andsafetymargin. and safety margin. resistance. loadeffect effect R. resistance element. resistance. resistance. resistance 0 Figure 58 PDFs of load. Figure 58 PDFs of Figure 58 PDFs of load.load. load effect margin margin RQ. resistance. resistance.. and it adopts a format: g ( X ) = g ( X 1 . resistance of Failure margin is demand representing load effect in a structure or a structural component.. 0 0 0 R. Load Figure R.14) Reliability analysis usually begins with the formulation of a limit state function (performance function)..15) RQ. load effect a structural Probability RQ. and safety margin. The performance function usually is a function of capacity and demand variables ( X 1 .
X n (x1 ... is an effective measure of the probability of failure.4.17) where f X is the joint probability density function of X 1 . also called safety index... thus the equation (3. It has been extensively used in calibrations of structural design codes. 35 .. the firstorder secondmoment formulation was developed and advanced by Cornell (1967) and Ang and Cornell (1974). X 2 . it indicates failure and. the necessary multidimensional integration may be extremely timeconsuming.. X n There is almost never sufficient data to define the joint probability density function for all basic variables. Even knowing the joint density function.. The corresponding probability of failure can be defined as the integral of the joint density function of the variables over the negative domain of g ( X ) (ThoftChristensen and Baker 1982): Pf = ∫ . xn )dx1 ⋅ . Further advances in these methods were made by Hasofer and Lind (1974) and Rackwitz and Fiessler (1978). simulation techniques. required input data. In the late 1960s. ⋅ dxn (3.17) is very difficult to evaluate. X 2 . it indicates acceptable performance. etc... The performance defined as g ( X ) = 0 is called the failure surface. or simplicity of formulation. In practice. 3. indirect procedures such as the reliability index are used.. it can be derived that when g ( X ) < 0 . The concept of secondmoment is often used in practical quantification of safety and reliability.∫ f g ( X ) <0 X 1 . a direct calculation of the probability of failure becomes inefficient. There are several methods to calculate reliability of structural components: the firstorder reliability methods (FORM).From the definition of the performance function. x 2 .. Therefore... The FOSM methods can be used to solve many practical problems. when g ( X ) ≥ 0 . RELIABILITY INDEX The evaluation reliability index.. computing cost. the advanced firstorder secondmoment methods (FOSM).. The FOSM approach can be put into several categories with regard to accuracy of results...
In 1974. Hasofer and Lind introduced the definition of the reliability index as the shortest distance from the origin to the limit state function in a system of reduced variables coordinates (Figure 3. the reduced variables ( Z R .19) σQ The resistance ( R ) and the load ( Q ) can also be expressed in the form of reduced variables: 36 . Thus. Z Q ) have to be introduced: ZR = ZQ = R − µR σR R − µQ (3.18) ZR g(ZR.3 Reliability index defined as the shortest distance in the space of reduced variables All the variables should be expressed in nondimensional forms. Using geometry the reliability index can be calculated as: β= µ R − µQ 2 2 σR +σQ (3.ZQ) = 0 limit state function SAFE µ Rµ Q σQ β µ Rµ Q σQ FAILURE ZQ Figure 3.3).
22) where Φ is the standard normal distribution function. Simulating a phenomenon numerically assumes events occur a finite number of times. the HasoferLind approach evaluates the reliability index for uncorrelated random variables. either instead of or in addition to realworld data. The limit state function used in this dissertation is linear. resistance and load with their mean values and standard deviations. This relatively straightforward concept often requires complex procedures. σ Q . Thus. 1982. 3.21) The reliability index recognizes the importance of uncertainty in load effects and strength. if the random variables are normally distributed and uncorrelated. Pf = Φ ( − β ) (3. if the initial variables are correlated they must be transformed into uncorrelated random variables. Hart. In case it was nonlinear.22).R = µR + ZR ⋅σ R Q = µQ + Z Q ⋅ σ Q (3. It is a practical tool that allows obtaining data. Many other 37 . Z Q ) = µ R + Z R ⋅ σ R − µ Q + Z Q ⋅σ Q (3. µ Q and σ R . Theoretical simulation is also called numerical or computer experimentation. The calculation can give exact results. 1982). The frequency of occurrence of an event in the entire set of simulations approximates its probability of occurrence. It incorporates the four key parameters. Otherwise it provides only an approximation. The most commonly used simulation technique is the Monte Carlo Method (ThoftChristensen and Baker. µ R . Moreover. Q) = R − Q in terms of reduced variables can be rewritten as: g (Z R . MONTE CARLO METHOD SIMULATION TECHNIQUE Simulation is the process of replacing reality with theoretical and experimental models.20) Therefore the limit state function g ( R. respectively. The probability of failure can be calculated using the formula (3.5. iteration would be required to find the design point in reduced variable space such that β corresponds to the shortest distance.
but also to measure sensitivity in system performance due to variations of some parameters. This method may be used not only to study performance of a structural system for a prescribed set of design variables. the reliability index. The standard normal random number zi is calculated using the equation: z i = Φ − 1 (u i ) (3. u n between 0 and 1 is performed.simulation techniques work similarly to the Monte Carlo Method.. indirectly. for engineering purpose. u 2 .. the values are arbitrary and selected according to specified rules. the values xi of sample random numbers can be generated for the random normal variable X . It is used to evaluate the probability of structural failure and. it can be used to determine optimal design. The Monte Carlo Method is commonly used to predict the behavior of structural elements and systems from the probability point of view. In the first step a simulation of the uniformly distributed random numbers u1 . and without physical testing. where physical testing could be very expensive.24) 38 . as: xi = µ X + z i ⋅ σ X (3. The Monte Carlo simulation method consists of the following steps.. A large amount of random numbers corresponding to random variables is numerically generated. However instead of generating a large number of random values for variables. Then. Therefore.. the standard normal random values can be calculated using generated numbers and information about the types of distributions and statistical parameters of distributions (mean value and standard deviation) of each design variable. This large number of repetitions is particularly valuable in solving problems involving rare events. They can be generated by computer programs using a builtin option.23) where: Φ −1 is the inverse of the standard normal cumulative distribution function Using standard random values (mean value µ X and standard deviation σ X ).
represents the distance from the mean value in terms of standard deviations. The basic variable is presented on the horizontal. 39 .6. It can also be implied as the corresponding probability of being exceeded.25) g ( X ) defines the performance function with the limit state g ( X ) = 0 g ( X ) < 0 defines the state of failure 3. Pf = where: n total number of simulations when g( X ) < 0 = N total number of simulations of g( X ) (3. More detailed information about normal probability paper can be found in textbooks (Nowak and Collins 2000. Performed Monte Carlo simulations allow for estimation of the probability of failure.It is important to simulate an efficient number of sets. such that the variation of the design parameters in a single simulation will not influence the solution of the entire process of simulations. The vertical axis. Benjamin and Cornell 1970). Cumulative distribution functions for the normal distribution are “Sshape” function. The probability of failure is defined as the ratio between the numbers of times the criterion for the failure is achieved ( n ). The relationship between the standard normal variable and probability is given in Table 3. being the standard normal variable. Normal probability paper redefines the vertical scale so that the normal CDF can be plotted as a straight line and allows for an easy evaluation of the most important statistical parameters as well as type of distribution function. Each simulated value has the same weight. to the total number of simulations ( N ). NORMAL PROBABILITY PAPER Normal probability paper is used to present cumulative distribution functions (CDF) in a convenient way.1.
40 . Relationship between vertical scale on Normal Probability Paper and Probability Distance from the mean value in terms of standard deviations 4 3 2 1 0 1 2 3 4 Corresponding probability 0. The mean value is at the intersection of the normal CDF and horizontal axis.Table 3.1.0228 0. The mean value and standard deviation read directly from the graph.00135 0. The basic properties of the normal probability paper: A straight line represents a normal distribution function.159 0.841 0.4.9999683 0.9772 0.5 0.99865 0. The standard deviation can also be read as shown in Figure 3.0000317 The shape of the resulting curve representing CDF allows for analysis of the test data plotted on the normal probability paper.
Standard Normal Variable CDF F(x) 3 2 1 mean 0 1 standard deviations 2 Figure 3.4 Normal Distribution Function on the Normal Probability Paper. 41 .
and highway safety issues.48feet . 42 . maintenance costs. TRUCK SIZES AND WEIGHT LIMITS Federal and state regulations limit the weight and dimensions of vehicles on U.102 inches .1.1.maximum gross weight on tandem axles .34 000 pounds .20 000 pounds .maximum gross vehicle weight . Current Federal law includes the following limits: .80 000 pounds . Classes 13 are passenger vehicles.28 feet . These restrictions have important impacts on highway construction costs.maximum vehicle width .maximum gross weight upon any one axle . classes 810 are combination trucks. classes 1113 are multitrailer trucks. REGULATIONS OF TRUCK TYPES. as show in Figure 4.CHAPTER 4 TRAFFIC DATA 4. The types of the vehicles in use on American roads are classified by FHWA into 13categories.minimum vehicle length for a semitrailer in a trucktractor/semitrailer combination . highways. classes 47 are single unit trucks and buses.minimum vehicle length for a semitrailer or trailer operating in a trucktractor/semitrailer/trailer combination.S.
1.Figure 4. FHWA 13category scheme 43 .
500 . from where some publications on field test results are used in this study.3. allows trucks up to 164.131.137.500 147.500 . For example the state of Michigan.Table 4. The states which allow various longer combination vehicles are presented in Figure 4. Conversion chart for vehicles’ class and number of axles Vehicle Class 1 2 3 4 5 6 7 8 9 10 11 12 13 Average Number of Axles per Vehicle 2 2 2 2.000 lbs) Triple Trailer Combination (common maximum weight – 105.500 .800 lbs) Turnpike Double (common maximum weight – 105.1.129.000 lbs) 44 .2 2 3 4 4 5 6 4 6 7 Several states issue overweight permits and allow higher truck loads. Types of longer combination vehicles (Figure 4.2) are: Rocky Mountain Double (common maximum weight – 105.000 pounds.000 lbs) Btrain Double Trailer Combination (common maximum weight – 105.
3. States allowing various Longer Combination Vehicles 45 .2. Longer Combination Vehicles (LCV’s) States Allowing LCVs’ States Allowing Triples States Allowing Turnpike Doubles States Allowing Rocky Mountain Doubles Figure 4.Rocky Mountain Double Turnpike Double 8Axle BTrain Double Trailer Combination Triple Trailer Combination Figure 4.
Federal size and weight studies were established in 1982, and since then no significant changes have been made. However, several proposals to make changes in these regulations were presented. The most recent studies are the TRB Special Report 267 "Regulation of Weights, Lengths, and Widths of Commercial Motor Vehicles" and the U.S. Department of Transportation "Comprehensive Truck Size and Weight Study: Volume I Summary Report". Both documents discuss existing regulations and give recommendations on their improvement. Elimination of the federal 80 000 pounds weight limit on Interstate highways is recommended. It is proposed that the gross weight should be governed by appropriate axle weight limits and the bridge formula (Figure 4.4). The maximum weight (in pounds) carried on a group of two or more consecutive axles would not exceed that given by the following formulas:  W = 1000*(2L+26) for L≤24 ft  W = 1000*(L/2+62) for L>24 ft
6.0
5.0
for L≤24 ft for L>24 ft
vehicle weight [kip/ft]
4.0
3.0
2.0
1.0
0.0 0 20 40 60 80 100 120
vehicle length [ft]
Figure 4.4. New Bridge Formula  regulation of vehicles' length and weight
46
In the Figure 4.5, the data from the Transportation Statistics Annual Report December 2006 of U.S. Department of Transportation is shown. As can be observed, the number of trucks has a trend of rapid growth. The number of heavy trucks is growing faster than number of light trucks. The number of light trucks (under 10 000 pounds) increased 73 percent between 1992 and 2005, and the number of heavy trucks (greater than 10 000 pounds) increased 112 percent. In 2005 heavy trucks constituted 8% of the volume of trucks, while in the 1990’s they were only 4% of the volume. In 2005, 95.3 million light trucks traveled 1.060 trillion vehiclemiles, and 8.5 million heavy trucks traveled 222.29 billion vehiclemiles.
120000 95300 100000 80000 55193 60000 heavy trucks 40000 20000 0 1992 1997 2002 2005 4007 4700 5400 8500 68100 79760 light trucks
Figure 4.5. Number of trucks by weight (in thousands of trucks). Transportation statistics annual report, December 2006. According to Texas Transportation Institute “Over the next 20 years, truck tonnage is expected to increase at a rate more than five times that of population growth.”
47
25000 20000
Tons (millions)
15000 10000 5000 0 1995 2000 2005 2010 2015 2020 2025 2030 2035
Year
Figure 4.6. Freight Tonnage Moved by Truck (FHWA)
Figure 4.7 presents time variation of total truck weight statistic between the years 1993 and 2003. The data is expressed as: mean value (µ), 95th percentile (W95, 95 percent of the trucks weigh less), and maximum observed total truck weight (Max). The study was made for the state of New Jersey, which has lower limits than the state of Michigan. However, the observed maximum gross vehicle weight is high, and it reaches a value of 225 kips (1000 kN). The maximum truck weight shows steady increase at an annual growth rate of 1.2%.
48
.Figure 4. (Gindy. the American Society for Testing and Materials (ASTM) published the first Standard Specification for Highway WeighinMotion (WIM) Systems with User Requirements and Test Methods (Designation: E 131890).. M. From these early beginnings. Bureau of Public Roads. WIM technology and application continued to advance and spread across the nation. In 1990. Shirley Memorial Highway.2.H.7. 2006) 4. ASTM Designation: E 131802 defines WIM as “the process of measuring the dynamic tire forces of a moving vehicle and estimating the corresponding tire loads of the 49 . H. and again in 2002 to the version (Designation: E 131802) that is used today. This document was revised in 1994. DATA COLLECTION METHODOLOGY (WIM) Weigh in Motion (WIM) Technology had its beginnings in the early 1950s when the U. the Virginia State Department of Highways. and the Williams Construction Company installed a load cell WIM system on the Henry G. Time variation of total truck weight statistic.S. Nassif.
Recently. and with rough surfaces. WIM data collection requires the most sophisticated technology for data collection sensors. II. aerodynamics and wind. axle weight. and Bridge WIM systems. the most controlled operating environment (smooth. lane distribution. out of balance tires and wheels. In WIM applications. piezoquartz sensors were introduced in the United States. bending plate. which are the most common. as well as the highest equipment setup and calibration costs. They are less sensitive to temperature changes and generally more accurate. date and time of passage. ASTM Standard Designation: E 131802 distinguishes four types of WIM systems (Type I. vehicle speed. and other dynamic factors.static vehicle”. a variety of ancillary traffic data can also be obtained through the use of WIM systems: traffic volume. directional distribution. The most commonly used are: piezoelectric cables. Several different technologies are available for WIM data collection systems. WIM equipment should be subject to a regular maintenance and calibration. Strain based and load cell WIM systems are much more expensive. suspension. although the data is not always collected continuously. speed. load cell. 50 . vehicles are stopped on a static scale and are measured without any interaction between the vehicle and the roadway. level pavement). III. and IV) based on application and performance requirements for data collection. The piezoelectric sensors. In order to assure unbiased data. a variety of forces are acting on the vehicle. tire inflation pressure. To limit erroneous data. The majority of WIM data collection is done with permanently installed weight sensors. vehicle acceleration and deceleration. The primary reason for sophistication in technology and its high costs comes from a need to determine static weight from a dynamic measurement. WIM sites should be localized away from weight stations and be unknown to truck drivers’. In addition to the collection of dynamic tire forces. Of all data collection methodologies. but they provide more accuracy. it is recommended to avoid sites with numerous traffic stoppages (speed >10 mph). including the force of gravity as well as dynamic effects of influences such as: roadway roughness. and vehicle classification. quartz cables. In standard weigh scale application. offer acceptable accuracy ±15%. Each type of WIM system has been specified to perform its indicated functions within specific tolerances. axle spacing. close to exits.
the quality and quantity of WIM data has greatly improved in recent years. 51 . Due to the weighinmotion technologies. traffic management and various other purposes. Oklahoma (16). the weigh in motion database was obtained from the project NCHRP 1276 and measurements on the Throggs Neck Bridge in New York. The database includes newest (20012006) WIM database for a variety of sites: California (6). in large quantity.Highway agencies have recognized the advantages of having automated data collection systems that can provide information on truck weights and truck traffic patterns for economic analysis. Figure 4. Therefore. Indiana (6). Mississippi (5). Florida (5). WIM data collection 4. and new WIM technologies continue to be developed.8.8). New York (7+2). and without truck driver’s knowledge (Figure 4.3. unbiased truckloads are being collected at normal highway speeds. WEIGH IN MOTION DATABASE For the scope of this research.
and state. It can be noticed.2 feet) for axle spacing.5 to 15 meters (1. the data quality checks have to be implemented to detect and fix/eliminate erroneous data before processing. The limits were 200 to 20. 52 .2. number of axles. its regular maintenance and recalibration. The variety of sites is important. and types of sensors used is presented in Table 4. Traffic data varies also depending on time of day. the heaviest vehicles are 6axles. spacing between axles. Distribution of vehicles by axles and traffic lanes are presented in Table 4. size and configuration of the WIM results. because the truck traffic changes depending on the site location: interstate or noninterstate highways. gravel and garbage haulers. day of the week. that for New York I495.44 to 4. reasonableness checks were performed on the axle weights and spacing. A standardized procedure to filter out errors is applied to WIM data from various sites.11. A summary of WIM data. and vehicle category. Florida. The statistics is presented for four states selected for simulations: Oregon. all obvious errors such as zero readings for number of axles or speed were eliminated. which depends on condition of equipment.6 to 49.3 . Those are construction debris. The errors are due to physical and softwarerelated failures of equipment and transmission. The WIM technology is known to have certain traffic data quality problems. it is important that the WIM database is collected continuously (mostly oneyear data). rural or urban areas.Table 4.000 kilograms (0. including site localizations.10. Therefore. and New York. According to Traffic Monitoring Guide (2001). The percentage of filtered out data varies for different sites. the difference between the dynamic weight measured and the actual static scale weight. Therefore. number of lanes. as well as the effect of tire pressure.41 kips) for axle weights and 0. Indiana. Cumulative distribution functions of gross vehicle weights by axles (GVW) were plotted in Figure 4. They often drive overloaded above 150 kips and occasionally above 200 kips. lane.9Figure 4. on many traffic lanes and in both directions (usually). speed. and direction. while NYSDOT routine permit trucks that are legal up to 120 kips. axle weight.and Oregon (4). The database contains date and time. season of the year. Moreover.
Table 4. Summary of WIM Data State CA CA CA CA CA CA FL FL FL FL FL IN IN IN IN IN IN MS MS MS MS MS NY NY OR OR OR OR NY NY NY Site ID 0001 0003 0004 0059 0060 0072 9916 9919 9926 9927 9936 9511 9512 9532 9534 9544 9552 2606 3015 4506 6104 7900 8280 8382 Woodburn Emigrant Hill Lowell Bend 9121 2680 Route Lodi Antelope Antelope LA710 LA710 Bowman US29 I95 I75 SR546 I10 I65 I74 US31 I65 I80/I94 US50 I55 I10 I55 US49 US61 I84 I84 I5 I84 OR 58 US 97 I81 8 I495 4 4 6 4 4 4 4 4 6 6 2 4 4 4 2 4 4 4 3 2 2 2 2 4 4 4 4 4 4 4 4 4 6 6 2 4 4 4 2 4 4 4 2 1 2 1 2 4 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y N N N N Y Y Y P P BP BP P P SLC P P P P P P P P P P P SLC SLC SLC SLC P P # of Traffic Lanes # of WIM Lanes Both Dir WIM Type P – Piezo.2. SLC – Single Load Cell 53 . BP – Bending Plate.
28% 213017 100.37% 894 0.4% 59.70% 31066 5.0% I5 Woodburn (NB) I84 Emigrant Hill (WB) OR 58 Lowell (WB) 36959 6.65% 14441 6.76% 350107 57.18% 10041 4.28% 10307 17.92% 5725 6.4.48% 2523 4.71% 4764 5.06% 59442 9.2% 0.1% 0. Vehicles by axle in Oregon axles 2 3 4 5 6 7 8 9 10 11 12 sum total 44507 71365 62025 575846 70639 85658 60907 3500 697 345 277 975766 4.03% 46 0.03% 404 0.08% 6 0.48% 42009 6.78% 3789 4.4% 0.01% 239 0.44% 5617 9.0% 0.37% 57 0.11% 20 0.34% 9807 10.02% 34 0.01% 611830 213017 91696 US 97 Bend (NB) 1942 3.07% 65 0.22% 140520 65.84% 28096 47.87% 9242 4.94% US 97 Bend (NB) 59223 100.13% 58407 9.04% 3333 1.40% 8199 13.64% 222 0.55% 19003 8.56% 2273 2.01% 172 0.8% 6.72% 40292 43.11% 59223 Table 4. Vehicles by traffic lane in Oregon lane 1 2 3 4 sum I5 Woodburn (NB) I84 Emigrant Hill (WB) OR 58 Lowell (WB) 552388 90.14% 210 0.26% 2155 3.10% 40 0.2% 8.97% 57123 62.91% 8032 8.0% 7.6% 7.19% 26 0.24% 43947 7.30% 46792 7.3% 6.Table 4.00% 611830 213017 91696 59223 54 .00% 51404 56.08% 14728 6.42% 131 0.20% 2253 0.3.
31% 7011 0.07% Florida 9927 127986 19.44% 181535 7.81% 11271 1.28% 88830 13.1% 5156752 55.11% 692 0.23% 56038 1.66% 268830 41.2% 8495 0.18% 2475 0.50% 1436329 82.88% 1720367 75.43% 709463 40.74% 3863073 Florida 9927 229181 35.Table 4.38% Florida 9936 818460 46.42% 430324 11.00% 0 0.08% 31241 1.02% Florida 9926 1408095 36.24% 1119347 28.42% 730652 2272829 649619 1755031 55 .12% 410 0.06% Florida 9919 229680 10.1% 15755 0.13% 867 0.75% Florida 9919 915451 40.74% 6296 0.06% 2533 0.45% 6990 0.14% 1554039 40.76% 1967 0.67% 62778 9.14% 407354 55.18% 24388 3.6. Vehicles by axle in Florida axles 2 3 4 5 6 7 8 9 10 11 12 sum total 2400362 25.98% 67058 9.34% 148424 20.04% 405 0.84% Florida 9936 152550 8.05% 801070 20.6% 143470 1.77% 74984 11.37% 69945 4.30% Florida 9926 897806 23.03% 418 0.96% 888 0.2% Florida 9916 482051 65.28% 237260 10.70% 115466 17.62% 11977 1.52% 112767 6.5% 18988 0.45% 402640 10.47% 156406 6.9% 779456 8.99% 938583 41.54% 297593 45.11% 124347 5.11% 74085 10.06% 422 0.64% 114341 6.5.98% 0 0.67% 2847 0. Vehicles by traffic lane in Florida lane 1 2 3 4 5 6 sum Florida 9916 249213 34.12% 61889 3.97% 4051 0.4% 747991 8.00% 1044850 27.02% 9271269 730652 2272828 3863134 649624 1755031 Table 4.69% 37909 1.
57% 171368 84.00% 21 0.00% 6 0.07% 120486 9.01% 46 0.66% 3170 1.03% 14 0.1% 2967 0.00% 0.57% 68489 15.67% 6243877 Indiana 9544 412695 49.00% 202120 Indiana 9532 738274 57.00% 202120 Indiana 9532 823175 63.67% 385840 6.00% 1293483 Indiana 9534 1509944 24.82% 364519 82.00% 20 0.47% 193 0.0% 513522 5.00% 0 0.5% 571231 6.0% 7547 0.0% 945 0.76% 2657717 42. Vehicles by axle in Indiana axles 2 3 4 5 6 7 8 9 10 11 12 sum total 2527382 27.85% 16475 8.37% 22818 6.81% 238 0.00% 4 0.48% 159 0.38% 7400 3.00% 67 0.15% 0 0.02% 396 0.Table 4.00% 830423 Indiana 9552 135029 37.0% 131 0.91% 60292 0.00% 0 0.7. Vehicles by traffic lane in Indiana lane 1 2 3 4 sum Indiana 9511 375357 84.57% 354033 5.19% 26400 3.01% 10 0.00% 443846 Indiana 9512 185645 91.0% 355 0.11% 15135 3.07% 38 0.00% 360304 Table 4.93% 2904 0.97% 0 0.86% 7359 0.13% 6534 1.33% 179915 49.79% 2993 1.36% Indiana 9534 3232127 51.3% 95770 1.08% 91398 7.41% 12517 2.05% 52 0.00% 6243877 Indiana 9544 82330 9.91% 26523 3.01% 18 0.04% 59 0.57% 836 0.10% 414 0.00% 0 0.00% 89 0.48% 19345 5.8.00% 443846 Indiana 9512 16938 8.04% 100 0.00% 830423 Indiana 9552 360304 100.43% 0 0.00% 8 0.06% 486 0.08% 54 0.18% 675794 81.0% 9374053 Indiana 9511 44867 10.1% 5654115 60.33% 24682 2.18% 3928062 62.00% 0 0 360304 1293483 56 .00% 0 0.89% 2381 0.18% 353721 5.64% 470308 36.01% 11 0.01% 12 0.38% 15688 1.31% 334457 25.01% 225 0.70% 393046 47.97% 3740 0.00% 0 0.00% 0 0.29% 829 0.0% 88 0.00% 0 0.01% 6 0.06% 1501 0.
06% 0 0.4% 217 0.28% 1377286 NY 2680 46008 33.6% 40359 14.24% 86 0.41% 212 0.0% 82959 30.43% 1010780 73.00% 0.32% 10716 7.01% 1 0.60% 5905 4.05% 79677 58. Vehicles by traffic lane in New York lane 1 2 3 4 sum I495 EB 52703 36.01% 18 0.2% 17426 6.00% 0.0% 0 0.32% 131 0.09% 20 0.81% 560 0.87% 74822 5.04% 8692 6.0% 0 0.54% 9720 7.02% 182 0.0% 0 0.39% 44357 3.00% 0.00% 128074 NY 9121 108563 7. Vehicles by axle in New York axles 2 3 4 5 6 7 8 9 10 11 12 13 sum total 114115 41.88% 135919 9.24% 14154 10.58% 21102 14.01% 19 0.13% 542 0.7% 19297 7.42% 10605 7.07% 7 0.00% 146346 I495 WB 57656 45.00% 128074 NY 9121 618289 44.55% 3148 2.39% 74831 5.34% 23644 17.00% 0.00% 1 0.10% 137131 57 .89% 88037 6.0% 0.02% 19257 15.0% 20 0.02% 4 0.61% 0.15% 98 0.25% 47294 32.1% 27 0.01% 90625 61.79% 35665 27.00% 137127 Table 4.0% 274420 I495 EB 56459 38.01% 0 0.00% 8 0.04% 335 0.10.09% 1113 0.32% 52845 38.00% 0.79% 78891 61.30% 8298 6.85% 6710 5.9.00% 146346 I495 WB 43278 33.00% 1377284 NY 2680 34752 25.22% 1758 0.43% 596129 43.07% 25 0.00% 2 0.Table 4.00% 0.93% 3018 2.00% 0.
00 2.I5 Woodburn (NB) 3. CDF's of GVW by axles Oregon I5 Woodburn 58 .00 Inverse of Standard Normal Distribution 1.00 GVW [kips] Figure 4.00 3.00 0 20 40 60 80 100 120 140 160 180 200 220 240 7axle 8axle 9axle 10axle 1.00 11axle 12axle 2.9.00 2axle 3axle 4axle 5axle 6axle 0.
00 0 20 40 60 80 100 120 140 160 180 200 220 240 7axle 8axle 9axle 10axle 1.00 3axle 4axle 5axle 6axle 0. CDF's of GVW by axles Oregon I84 Emigrant Hill 59 .00 11axle 12axle 2.10.00 GVW [kip] Figure 4.00 3.I84 Emigrant Hill (WB) 3.00 2.00 Inverse of Standard Normal Distribution 2axle 1.
00 3.00 3axle 4axle 0.00 20 40 60 80 100 120 140 160 180 200 220 240 5axle 6axle 7axle 8axle 9axle 2.00 3.00 4.00 Figure 4.00 2.11.00 0 1.00 4. CDF's of GVW by axles NY I495 EB 60 .00 5.CDF's of GVW by axles NY I495 EB 5.00 2axle 1.
5. The live load is modeled as the uniformly distributed lane load and additional axle load or single truck for deck components.Left lane loaded only with average trucks.Traffic jam situation. Figure 5. the extreme live load is governed by the traffic jam scenario. In the FHWA WIM Data. the following traffic model has been assumed: . therefore the percentage of 2axle vehicles is relatively low.1. INTRODUCTION For long span bridges. vehicle categories 13 FHWA are omitted. which are the most popular among truck types. . Development of live load model is based on three approaches. Figure 5. . 5. The first of them is based on a 5axle average truck and the second one is based on AASHTO LRFD legal load trucks.4 and Figure 5. MODEL BASED ON AVERAGE 5AXLE TRUCK For computation of the live load on the most loaded lane. Two of them can be classified as initial studies.Average trucks are 5axle trucks. 61 .CHAPTER 5 DEVELOPMENT OF LIVE LOAD MODEL 5.2. The third approach is detailed study based on truck WIM Data.1.
00 5. Critical loading. .00 4.00 0 1.00 3. therefore spacing between the last axle of one truck and first axle of the following truck is 2025 ft. New York WIM Data.00 4.00 1. 62 . Traffic jam scenario.3.79 kip/ft for clearance distance of 15 ft 55 kip / 65 ft = 0.85 kip/ft for clearance distance of 10 ft Figure 5.00 inverse of standard normal distribution 3.2 and Figure 5. 5. Live load due to such a combination of vehicles is equal to: 55 kip / 70 ft = 0.00 2.2.00 2. CDF’s of GVW for 5axle trucks..Clearance distance is 10 to 15 ft.00 20 40 60 80 100 120 140 160 180 200 GVW [kips] Figure 5.00 0.1.An average 5axle truck: is 45 ft long weights 55 kips. Figure 5.
Percentage of vehicles by number of axles.S. FHWA WIM Data.3..Figure 5. J.4. H. CDF of GVW for 5axle and 11axle trucks (Nowak. A. 63 . and Nassif.. 1994) 70% 60% 50% 40% 30% 20% 10% 0% 2 3 4 5 6 7 8 9 10 11 12 number of axles Oregon Florida Indiana New York Figure 5. Laman.
9% 10% 0% 2 3 6.5% 0.S.5% 4 5 6 7 8 9 10 11 12 number of axles Figure 5.6% 1.50% 40% 30% 43.00 4. the result would be similar.00 0 1. not only 5axles.00 20 40 60 80 100 120 140 160 180 200 220 2. 64 . Sokolik.00 GVW [kip] Figure 5.00 2.3% 2.00 5.8% 0. 5. and Nowak. SJ..5. The mean value of GVW is above 50 kips.00 Inverse of Standard Normal Distribution 3.6.00 4..F. A. A. (Kim.1% 6. Percentage of vehicles by number of axles.00 3.00 0.4% 20% 10.5% 5. 1997) If we would like to consider all types of trucks. CDF’s of GVW for all types of vehicles in Oregon.6. Figure 5.00 US 97 Bend I84 Emigrant Hill OR 58 Lowell I5 Woodburn 1.2% 22..
64 kip/ft.0’ 15. Type 33 Unit) are supposed to adequately model short vehicles and a combination of vehicles for short.0’ 4. To model traffic jam situations on long span bridge the Type 33 Units have been placed in a lane with the clearance distance of 10 to 15 ft. 65 . Gross Vehicle Weight of a Type 33 Unit is 80 kips and total length of a Type 33 Unit is 54 ft.0’ Type 33 Unit 16. MODEL BASED ON LEGAL LOAD TRUCKS The second approach to model live load is based on vehicles called "legal load types". three legal AASHTO vehicles (Type 3 Unit. Figure 5.0’ 15.75 x 1.0’ 4.01 kip/ft = 0.0’ 20.81 kip/ft 12 kip 12 kip 12 kip 16 kip 14 kip 14 kip 12 kip 12 kip 12 kip 16 kip 14 kip 14 kip 15. Type 3S2 Unit. therefore: 80 kips / (54+25) ft = 1.08 kip/ft = 0. Type 33 Units. Some of the states use them for rating instead of the traditional HS20 load.7. AASHTO LRFD legal load trucks. Therefore spacing between the last axle of one truck and first axle of the following truck is 2025 ft.5.0’ 4.0’ 4.0’ . This approach derives from basic philosophy used to develop lane load of 0.08 kip/ft for clearance distance of 10 ft Since the value obtained in this way is based on heavy trucks and it is very conservative.3.75 x 1.25.0’ Figure 5. its value can be multiplied by factor 0. These vehicles were selected to match the federal bridge formula (known as Formula B) for vehicles up to 80 000 lb. 0.76 kip/ft 0.01 kip/ft for clearance distance of 15 ft 80 kips / (54+20) ft = 1.75.7.0’ Type 33 Unit 16.0’ 54. medium and long spans respectively.0’ 15.0’ 54. which are developed in AASHTO. While HS20 was intended to be appropriate for all span ranges.
9. as recorded in the WIM surveys. It was assumed that in a traffic jam situation. 1000. Figure 5. the first truck was deleted. Simulation of trucks moving throughout span length 66 .8). 4000. 2000. Then.8. therefore. the total load of all trucks was calculated and divided by the span length to obtain the first value of the average uniformly distributed load. Starting with the first truck. and one or more trucks were added so that the total length of trucks covers the full span length and the new value of the average uniformly distributed load was calculated. light vehicles are using faster lanes. described in CHAPTER 4. and 5000 ft. Only the most loaded lane was considered.5. 3000. and according to literature it varies between 6 and 21 ft. Clearance concept is as defined as in Figure 5. based on the most common 5axle truck WB20 defined in “AASHTO Geometric Design of Highways and Streets. vehicles of the 13 FHWA category were omitted. Trucks were kept in actual order. Clearance distance is assumed to be about 15 ft. The calculations were performed for span lengths 600. MODEL BASED ON TRAFFIC JAM SIMULATION USING WIM DATA The considered WIM data was obtained from NCHRP 1276. while spacing between the last axle of one truck and first axle of the following truck is 25 ft. The available data served as a basis for simulation of a traffic jam situation. all consecutive trucks were added with a fixed clearance distance between them until the total length reached the span length (Figure 5.4. Spacing between the last axle of one truck and first axle of the following truck is clearance plus distance from first and last axles to corresponding bumpers. Next.
Results of the simulations were plotted as a cumulative distribution function (CDF) of uniformly distributed load for considered span lengths. it is small compared to the actual life time of the structure. six. had to be inputted to the program. Calculations were performed for one month. Interstate semitrailer WB20 (AASHTO Geometric Design of Highways and Streets) To perform an extensive number of simulations on the huge databases for each site and different span lengths. a numerical procedure based on Visual Basic coding were developed.Gap and Spacing . CDF’s of daily maximum combinations in Appendix B. The CDF’s for all simulated data are presented in Appendix A. WIM data. and CDF’s of weekly maximum combination in Appendix C. Clearance . Running the computation was an extremely time consuming process. three. The uniformly distributed loads 67 . The desired clearance distance was added automatically before computation. four.10. such as weight and length of the trucks in actual order.9. Even though it is a long period of time.Headway Concepts Figure 5. nine months or one year data depending on localization.Figure 5.
The number of truck combinations N.corresponding to longer period of time.56x104.83 on the vertical scale of CDF plot. will require development of site specific models. The mean daily and weekly maximum can be found in Figure 5. It was noticed that mean value of uniformly distributed load oscillates between value 0. and inverse normal distribution corresponding to 75 years periods are shown in Table 5. Probability corresponding to extrapolated maximum daily truck is 3. It is recommended to use HL93 also for those long spans (Figure 5.0 (Figure 5.13. This observation confirms that for a long loaded span. The value 0.17). for those bridges the uniformly distributed load should be higher.75 k/ft (Figure 5. as shown in the NCHRP Report 368 (1999). This is because the load depends on a mix of traffic. For 360 million. N will be larger 900 times for one month data. the uniformly distributed load should be 1. For T = 75 years. To keep bias value below 1. 300 times for 3 month. The probability corresponding to N is 1/N. For longer spans uniformly distributed load decreases and is closer to mean value. The bias factors (ratio of mean to nominal) were calculated for the heaviest 75year combination of vehicles. such as those in New York. In Figure 5.16. It was found that to not exceed bias 1. Therefore.000. The 75year uniformly distributed loads were derived from extrapolated distributions. this will result in N = 360 million truck combinations. Brides in localizations with high ADDT and high percentage of overloaded trucks.25.8x109. For some sites with very heavy traffic. 75 times for one year data etc. it is 1/360. the statistical parameters of live load were obtained. and to maximum weekly truck is 2.12 and Figure 5.75 k/ft is close to those obtained in two previous models.25. it can be noticed that the bias factor values for some sites do not exceed 1. The heaviest truck 68 .000 truck combinations monthly. were calculated by extrapolation of simulated results.0. probabilities 1/N.11). 75 years. For example.65x105. Let N be the number of truck combinations in time period T and assume that the traffic will remain the same. the bias factor reaches value 2. it would be necessary to increase design value of uniformly distributed load to 0.16). one heavily overloaded truck does not have significant influence.50 and 0.2 kip/ft (Figure 5.12). From the results of simulations. for a site with 400. The extrapolated distributions are shown for maximum daily and weekly combinations (Appendix B and Appendix C).18).000=2. which is 5.85 k/ft (Figure 5. which is similar as for short and medium spans.2.
combinations were observed on I475.50 AASHTO HL93 equivalent UDL [kip/ft] 1. Throggs Neck Bridge in New York. They have been presented in Table 5. Daily maximum uniformly distributed load has more variation due to weekends.50 0.68 kip/ft 0.19 and Figure 5.1. excluding sites from Yew York.25 1. Calculated statistical parameters for uniformly distributed load are summarized in Table 5.10 0.3. Statistical parameter for proposed uniformly distributes live load span length 600 – 1000 ft > 1000 ft Bias 1.00 0.1. Lighter traffic during weekends can be observed in lower tail of CDF’s.20 CoV 0.08 69 . The coefficient of variation is calculated from the slope of transformed CDF.20 present coefficient of variation of daily and weekly maximum uniformly distributed load. Estimated coefficients of variation were derived from weekly maximum values.00 0 1000 2000 3000 4000 5000 loaded length [ft] Figure 5. Figure 5.11. HL93 proposed for long span bridges Table 5. 1.
23x107 2.225 300.418 552.57x108 2.11x108 247.2.019 51.50x108 5.73x109 8.63x109 4.05x108 6.030 43.406 59.500 45.444 70 .200 52.Table 5.22x108 2.390 213.449 222.990 167.618 3.09x108 8. Summary of simulated data State Site ID Route Number of truck combinations FL FL FL FL IN IN IN OR OR OR OR NY NY NY NY 9916 9919 9927 9936 9512 9534 9544 Woodburn Emigrant Hill Lowell Bend 9121 2680 US29 I95 SR546 I10 I74 I65 I80/I94 I5 I84 OR 58 US 97 I81 8 I495WB I495EB sum Time period 1 year 3 months 1 year 1 year 1 month 1 month 1 month 4 months 4 months 4 months 3 months 6 months 9 months 1 month 1 month 75years probability 5.630 266.333 406.17x109 2.90x108 7.868 188.39x108 1.042.65x108 7.50x108 2.368 225.04x109 2.
14 10.45 10.1 564.00 4.48 38.68 33.00 0.68 0.8 73.00 0.00 28.00 0.99 20.00 23.9 178.27 31.00 0.04 4.00 30.00 0.00 k/ft 43.00 4.00 0.60 0.77 0.03 3.00 0.84 3.27 4.68 33.9 83.00 0.73 37.00 0.27 3.8 166.00 0.87 L [ft] 4.00 0.00 2 5 2 5 3 2 5 5 5 11.00 L [ft] 4.0 70.00 0.8 74.00 32.83 25.4 85.37 26.9 83.00 38.26 30.00 0.87 23.83 25.9 127.00 0.87 42.97 33.41 0.7 60.00 38.27 4.00 28.0 199.4 85.62 0.00 0.6 155.40 28.00 3.7 60.00 4.4 77.00 0.94 3.27 0.1 1047.3 44.14 13.62 0.41 32.56 14.00 4.00 0.8 77.00 0.57 0.27 0.41 0.00 0.3.62 22.87 31.33 39.17 18.00 25.00 0.44 0.61 19.82 36.7 118.91 k/ft [ft] [kip] 0.00 0.19 35.00 0.00 0.00 0.68 0.00 0.27 0.41 32.94 0.94 31.55 38.00 3.8 74.04 43.00 0.27 4.00 4.92 k/ft 64.3 44.72 16.94 0.97 15.00 37.61 19.00 4.11 12.7 49.4 1.45 28.04 27.7 32.3 92.68 40.48 38.33 13.51 22.00 0.67 29.50 0.9 1123.7 77.3 173.9 1082.87 31.00 0.00 34.3 141.06 17.11 0.00 35.08 22.50 19.27 4.39 32.48 0.94 3.38 39.00 0.00 33.75 0.00 0.00 4.00 0.94 3.42 12.87 23.94 0.04 27.12 26.94 0.00 0.27 4.70 31.27 0.00 0.00 0.82 36.00 34.00 33.27 20.00 33.07 43.00 W [kip] 37.59 0.00 0.00 0.00 0.00 0.94 4.00 0.00 25.00 71 .15 41.85 3.12 26.00 0.27 4.12 21.27 0.69 17.00 L W Total GVW length [ft] [kip] 73.22 15.00 0.04 27.27 0.00 0.19 35.50 0.26 L [ft] 17.0 48.0 48.00 22.Table 5.1 74.00 0.04 24.94 22.69 35.00 4.6 155.00 0.48 26.00 0.1 40.00 0.14 40.11 18.00 3.1 67.9 45.00 0.44 43.50 16.00 0.00 0.08 22.87 42.00 0.00 0.94 37.00 0.9 45.47 20.75 35.97 15.75 W [kip] 41.2 48.00 25.00 28.27 0.2 76.00 33.26 30.00 4.21 30.00 0.9 127.04 27.00 37.00 W [kip] 38.00 W [kip] 35.14 13.70 17.44 43.60 0.39 19.42 10.04 43.5 46.00 0.00 38.99 20.99 37.39 19.00 0.00 27.97 33.00 0.2 66.73 37.72 16.77 0.47 38.64 0.11 33.27 3.2 177.33 20.70 21.69 17.28 0.38 39.00 0.55 38.27 0.01 0.00 0.00 0.00 0.48 0.0 199.00 32.27 0.48 0.03 37.1 74.46 36.00 0.00 0.61 34.09 22.75 0.00 3.44 0.4 60.57 0.00 0.31 0.4 73.53 0.1 561.57 0.33 0.6 117.00 0.31 0.12 15.8 77.00 0.28 30.3 173.00 23.00 0.2 177.4 1.45 28.42 12.1 2.23 20.00 L [ft] 22.00 5 5 5 2 3 2 5 5 3 38. Heaviest truck combinations for 600 ft on I495 WB number of axles 5 2 5 3 2 5 5 5 2 W [kip] 20.11 21.0 104.8 41.40 28.00 0.00 0.08 18.00 0.94 33.6 548.11 0.00 0.0 70.00 0.00 0.94 4.00 22.94 3.
20 NY I495 WB 0.0.60 0.80 0.80 Florida 9919 Indiana 9512 Indiana 9534 0. Mean value of uniformly distributed load 1.70 0.40 US 97 Bend I5 Woodburn Florida 9936 1.10 0.50 0.40 0.30 0.60 OR 58 Lowell I84 Emigrant Hill US 97 Bend I5 Woodburn Florida 9936 Florida 9916 Florida 9927 Florida 9919 Indiana 9511 Indiana 9512 Indiana 9534 Idiana 9544 NY I495 EB NY I495 WB NY 2680 0 1000 2000 3000 4000 5000 6000 NY 9121 UDL [k/ft] 0.13. Daily maximum mean value of uniformly distributed load 72 .40 Idiana 9544 NY I495 EB 0.00 0 1000 2000 3000 4000 5000 6000 NY 2680 NY 9121 span length [ft] Figure 5.12.00 span length [ft] Figure 5.00 Florida 9916 Florida 9927 0.20 0.20 UDL [k/ft] 1.60 OR 58 Lowell I84 Emigrant Hill 1.
40 2.15.20 2. Weekly maximum mean value of uniformly distributed load 2.00 0.80 0.60 2.80 1.00 1.20 0.40 US 97 Bend 1.60 OR 58 Lowell I84 Emigrant Hill 1.40 0.20 1.60 NY 2680 NY 9121 0.1.20 0.60 span length Figure 5.20 I5 Woodburn Florida 9916 UDL [k/ft] 1.00 Florida 9936 Florida 9927 Florida 9919 0.14.80 0.60 1.00 0 1000 2000 3000 4000 5000 IN 9512 Bias 1.40 0. Bias (mean max 75 year to nominal value of UDL) 73 .40 OR US97 OR I5 OR 58 OR I84 FL 9919 FL 9927 FL 9936 IN 9534 IN 9544 0.00 0 1000 2000 3000 4000 5000 6000 span length [ft] Figure 5.
Bias (mean max 75 year to nominal value of UDL) assumed designed UDL of 0.2.20 2.40 2.60 1.80 0.40 2.17.60 0. Bias (mean max 75 year to nominal value of UDL) 2.20 1.80 0.00 1.40 0.20 2.16.80 1.20 1.85 k/ft 74 .20 0.00 0.60 span length Figure 5.00 0 1000 2000 3000 4000 5000 IN 9512 Bias 1.20 0.40 FL 9916 NY 9121 NY 2680 NY I495 EB NY I495 WB Bias 1.80 1.40 0.00 0 1000 2000 3000 4000 5000 span length Figure 5.60 2.60 1.60 2.40 OR US97 OR I5 OR 58 OR I84 FL 9919 FL 9927 FL 9936 IN 9534 IN 9544 0.00 0.00 1.
20 0.80 0.15 0.18.00 1.80 1.00 0 1000 2000 3000 4000 5000 6000 OR 58 Lowell OR I84 Emigrant Hill OR US 97 Bend OR I5 Woodburn Florida 9916 Florida 9936 Florida 9927 Florida 9919 Indiana 9512 Indiana 9534 Indiana 9544 NY I495 EB NY I495 WB NY 2680 NY 9121 UDL [k/ft] span length [ft] Figure 5. Coefficient of variation of daily maximum uniformly distributed load 75 .25 k/ft 0. Bias for heavily loaded localizations.40 FL 9916 NY 9121 NY 2680 NY I495 EB NY I495 WB Bias 1.40 0.20 1.25 0.40 2.60 0.30 0.05 0. assumed designed UDL of 1.00 0 1000 2000 3000 4000 5000 span length Figure 5.2.60 2.19.60 1.00 0.20 0.20 2.10 0.
05 0.25 0.15 0.10 0. Proposed coefficient of variation of uniformly distributed load 76 .20 OR 58 Lowell OR I84 Emigrant Hill OR US 97 Bend OR I5 Woodburn Florida 9916 Florida 9936 UDL [k/ft] 0.10 Florida 9927 Florida 9919 UDL [k/ft] 0.0.00 0 1000 2000 3000 4000 5000 6000 span length [ft] Figure 5.20.15 0.05 0.30 0. Coefficient of variation of weekly maximum uniformly distributed load 0.25 OR 58 Lowell OR US 97 Bend 0.20 OR I5 Woodburn Florida 9916 Florida 9936 0.00 0 1000 2000 3000 4000 5000 6000 Florida 9927 Florida 9919 NY 2680 NY 9121 span length [ft] Figure 5.21.
Multiple presence factors applied in international codes are discussed in CHAPTER 2. Those carrying the right lane of traffic are usually subjected to more load cycles and fatigue than the components closer the left lane. Structural components are strongly influenced by the location of trucks on the bridge. study of video recording of traffic jam situations. INTRODUCTION Determination of live loading for multiple traffic lanes is very important for appropriate bridge design. a short review of current studies on presence of multiple trucks. and discussion on different approaches to multilane reduction factors are presented. 77 . for example during maintenance works on bridge. The more traffic lanes the more difference in distribution of loading. Each traffic lane can be loaded with different live load. However. it has to be remembered that traffic can be deviated and truck can be directed to left lanes.CHAPTER 6 MULTIPLE PRESENCE 6. Trucks tend to use right lanes loading them heavily. while passenger cars use faster left lanes. Presence of truck on multiple traffic lanes at the same time is critical from the bridge design point of view. In this chapter.1.
8% right 1582 91. and as a consequence more cases of multiple trucks can be observed.2% right 2073 85.6% 2 0.6% 38 2. 78 . Presence of multiple trucks and their location on the road lanes Eastbound Interstate highway side by side tandem 28 1. The study gives detailed analysis of the relation between multiple presences of trucks and four parameters: truck volume.8% 76 3. Table 6. An increase in truck volume results in an increase of all multiple presence cases and a decrease in the frequency of single loading events (Figure 6. The area (urban or rural) and vicinity of industry affect the frequency of multiple truck presence. road type. and 6570 percent of trucks are 5axle trucks. For the study. and bridge span length.7% 371 18.5% in lane 14 0.1.8% 24 1. Data used in this study was collected over an 11year period between 1993 and 2003 by the New Jersey Department of Transportation.6.23 The results were later confirmed by the investigation by Gindy and Nassif (2006). Heavier volume sites tend to be located in urban areas.6% 1247 67. area type.3).7% Westbound side by side tandem 86 3. Hani Nassif.S. Span length has almost no influence on the frequency of sidebyside trucks.8% 34 1. 48% side by side in tandem or behind.1).1% behind 93 3.1% I94 U.5% left 349 14. and Leo DeFrain performed a study on the occupation of road lanes and presence of multiple truck. the database of over 600 000 trucks on two lanes in each direction was collected using weighinmotion equipment.3% 1685 82.0% in lane 14 0. Increasing bridge spans also gives more opportunities for trucks to occur simultaneously (Figure 6. The researchers had also found out that less than 2% of trucks appear simultaneously with another truck in the lane.2.0% left 151 8.6% 40 1. The frequency of staggered events increases faster for shorter spans and at a steadier pace for longer spans. STUDIES ON PRESENCE OF MULTIPLE TRUCKS In 1993 Andrzej S. Nowak.1).2). It was found that 7090% of trucks use the right lane (Table 6.9% behind 48 2.4% 601 32. with distance between front axles less than 50 ft (Table 6.
and Leo DeFrain in 1993 and the study made by Mayrai Gindy and Hani H.1 and 1. some divergences can be observed. Nowak. Hani Nassif. Comparing results of the study made by Andrzej S.2%. Nassif in 2006.1): single – only one truck is present on the bridge following – two trucks on the same lane with a varying clearance distance sidebyside – two trucks in adjacent lanes with an overlap at least onehalf the body length of the leading truck staggered – two trucks in adjacent lanes with an overlap at less than onehalf the body length of the leading truck Figure 6. 7. Trucks which are considered “staggered” in one case can be considered as “side by side” in the other.3 % for the study from 1993 and 6% for study from 2006. Traffic loading pattern used for multiple truck presence statistics.1. values obtained in 1993 vary between 0. we obtain relatively close results. Those values cannot be compared 79 . Some of them can be caused by the differences in the definition of “sidebyside” and “staggered” cases. However. and values obtained in 2006 vary between 1 and 8%.Assumptions of traffic patterns made for multiple presence analysis following (see Figure 6. Regarding the occurrence of following trucks. joining those two cases and making a sum of those two values.
Figure 6. The likelihood of permit trucks exceeding the authorized weight as well as the likelihood of the presence of multiple permit trucks is reduced. it is stated that multipresence probabilities for permit trucks are different from those for normal traffic. the bridges taken into consideration in the study from 2006 were longer. The probability that the trucks will occur as following trucks increases with the span length.because of differences in the assumptions and definition of “following” trucks.2. which increases the probability of this occurring. Gindy and Nassif (2006). Furthermore. 80 . In the report NCHRP 1276 (2008). Variation of multiple truck presence statistics with respect to truck volume.
6. Thursday.29. Variation of multiple truck presence statistics with respect to bridge span length. Total time of all sixteen video recordings is 2 hours 6 minutes and 30 seconds. Gindy and Nassif (2006).(1) 05. MULTIPLE PRESENCE OF TRUCKS BASED ON THE VIDEO FILES OF TRAFFIC FHWA has provided a DVD including video files monitoring seven traffic situations in different sites.3.2008. Friday. 12.Figure 6. The localization of the sites was not specified.30. 6. The list of video recordings is following: .08pm (4min43sec) 81 .(2) 05.3.17 pm (33min44sec) .2008. at different times and days of the week.
(10) 05.22. some of them being the result of traffic accidents. 11. Wednesday.(4) 05. the recorded situations can be related to bridges as well.43pm (0min36sec) .20pm (1min04sec) .52pm (4min51sec) .04.43pm (19min26sec) .43pm (8min54sec) . Friday.06. Friday.(5) 05.22am (0min22sec) .(7) 05.2008. Friday. Thursday. 12..2008.(14) 04. 11.(6) 05.43pm (8min06sec) The recordings show dense traffic jam situations.(8) 05.2008. Thursday.01.22. 12.2008. Friday.2008.(9) 05.14.2008. Wednesday.(3) 05. Despite the fact the recordings are taken on highways.48pm (1min47sec) . 7. Friday. 4.(13) 04. Wednesday. 82 . Wednesday. 4. One of the most important observations is that even in very dense traffic it is very common to observe cars or pickup among heavy vehicles (Figure 6.42pm (9min28sec) .43pm (22min53sec) .2008.04. Tuesday.20am (0min22sec) .14. 12. 4. 12.2008.2008.04.(11) 04.(16) 04.04.2008.2008.4).14. They allow for making some observations and conclusions regarding traffic patterns and the presence of multiple trucks moving at a crawling speed.2008.2008. 4.50pm (2min26sec) . 4.59am (3min49sec) .04. Thursday. Friday.44pm (3min59sec) .04.(12) 04. This is the critical case from the point of view of live loading on bridges.2008.(15) 04.14. 12. 4.
Figure 6. However. Video 10. The second and the third lanes remain completely stopped by crushed cars and emergency vehicles for approximately half an hour.5 and Figure 6. The accident takes place on the second lane (counting from the external side).4. Those cases should also be taken into consideration during the evaluation of design live load. and the passing cars can use only the fourth lane.6). There is no information about trucks’ weight. We can also observe a situation when one lane is almost exclusive occupied by trucks (Figure 6. Passing cars are using the first and the fourth lanes.6). that intensifies jampacked traffic. it can be assumed that only a limited number of the trucks are correlated. 83 . and while some of them are fully loaded some percentage of them can be empty. For the majority of time we can observe that the moving lanes contain a mixture of trucks and cars. time: 00:00:58 The video file number 1 contains the longest and the most interesting material. However. we can also observe some situations with multiplepresence of trucks occupying three or four lanes at the same time (Figure 6. For a short period of time three lanes are blocked. It has a registered traffic accident on a highway having four lanes in one direction.
time: 00:18:36 84 .6.Figure 6.5. Video 1. Video 1. time: 00:05:28 Figure 6.
Video 2.7 and Figure 6. The number of cars among trucks is increased because some of them are entering or exiting the highway (Figure 6. This observation calls our attention to the importance of the appropriate selection of sites. Figure 6.8).Some of the video files show traffic jam situations at roads junction that cannot be representative. time: 00:00:15 85 .7.
Figure 6. time: 00:00:16 Video recordings from 11 to 16 show traffic situation caused by the same accident. An ordered one lane of traffic is a mixture of trucks and cars. 6. and all of them are much more simplified than actual situation. warning signs posted adequately ahead result in the vehicles forming one lane and do not cause multilane traffic jam situation in the vicinity of the place of the accident.8. The load value is the same on all traffic lanes. 86 . However. International codes vary significantly in this matter (see CHAPTER 2).9. Three lanes are blocked and the moving vehicles are using only one lane. APPROACHES TO MULTILANE REDUCTION FACTORS There are many approaches to multilane reduction factors. This situation does not allow for the observation of multiple truck presence situations.4. Figure 6. Most of the design codes decrease the value of uniformly distributed load as the number of traffic lanes increases. Video 8.
The observations indicate that the actual traffic is distributed differently for each lane of traffic. Multilane load in design codes AASHTO LRFD Code (2007).10. Figure 6. OHBDC (1991). CAN/CSAS600 [2000].11. According to the Eurocode one of the lanes is loaded more than the others. Figure 6. Figure 6.11. Figure 6.Figure 6.9. Multilane load in actual observation. 87 .10. and ASCE (1981). Multilane load in Eurocode 1.
multilane reduction factors could be very site specific. However. Therefore. it does not allow for simulations and derivation of multilane factors for all traffic lanes. Since no new multilane reduction factors were proposed. road type (major/minor). average. As well. 6. heavy) area type (urban/rural). Multiple reduction factors for design live load should account for those the most critical loading cases. the vehicles of 13 categories have not been registered. They have to account for intensive traffic jam situations. CONCLUSIONS Presence of multiple trucks depend on factors such as: truck volume (light. situations when one lane is almost exclusively occupied by trucks or trucks occupy three or four lanes at the same time are also possible. In the available WIM database. bridge span length. Therefore. the passenger vehicles are assumed to move to faster lanes. the approach applied in the Eurocode seems the most practical one. Each of the lanes should be considered as the most loaded one. different distribution of loading on multiple traffic lanes has to be considered. those from the current AASHTO Code are used in this dissertation.To reflect actual traffic situation would be very difficult and time consuming for designers. It was concluded that the multilane reduction factors have to be an object of additional extensive studies. which can cause heavy truck queues. that it is occupied exclusively by trucks. Simulation of the traffic on the most loaded lane was possible with assumption. while the other lanes carry equal loading. In the traffic jam situations. law enforcement. 88 . as those registered in the video recordings. Therefore. and traffic flow control. Based on video recordings of traffic it was confirmed that for the majority of time we can observe that the moving lanes contain a mixture of trucks and cars.5.
89 . but the dynamic amplification is added additionally as a percentage to static effects. As well. The objective of this simple approach is to not increase complexity for the designer. There is considerable variation in the treatment of dynamic load effects by bridge design codes in different countries (see CHAPTER 2). INTRODUCTION The scope of this chapter is studying the origin and adequacy of the application of the dynamic load factor in bridge design.CHAPTER 7 DYNAMIC FACTOR 7. and derivation of dynamic factor for this specific case. This is the approach specified in the current AASHTO Code. a developed exemplary vehiclebridge interaction model and the derivation of dynamic factor is presented. Threeaxle AASHTO truck HS20 travelling over a 120 ft steel girder bridge is modeled. The live load model itself does not account for dynamics. In this chapter there is a short review of the research studies.1. In the modeling a finite element software ABAQUS was used. The most common approach is to apply dynamic response as a fraction or multiple of the response that would be obtained if the same forces or loads were applied statically. The truck is assumed to travel with the velocity of 40 miles per hour and with the crawling speed. The final result is comparison of the static the dynamic deflections.
which corresponds to the body bounce response frequency range of a truck. STUDIES ON PARAMETERS AFFECTING DYNAMIC BRIDGE RESPONSE The dynamic response of a bridge to a crossing vehicle is a complex problem affected by the dynamic characteristics of the bridge. many research studies have reported seemingly conflicting conclusions. Based on the review of past research.2.1. Bridge Fundamental Frequency The fundamental frequency of vibration for a bridge due to vertical loading has a significant effect on the dynamic response.1). and by the bridge surface conditions. the vehicle. A majority of the fundamental frequencies for typical bridges are in the range of 2 to 5 Hz (Figure 7. the effects of various parameters on the dynamic response of bridges to vehicular loading are discussed in this paragraph. further complicating the issue. If the frequencies of the bridge and vehicle converge. Many of the parameters interact with one another.7. Consequently. the dynamic response induced may be large. Figure 7. Distribution of fundamental bridge frequencies (Cantieni 1984) 90 .
Field measurements and values obtained from analytical modeling show relation between frequencies and bridge span (Figure 7.2 and Figure 7.3).
Figure 7.2. Fundamental frequency versus span length (Cantieni 1984)
Figure 7.3. Fundamental frequency versus span length (Paultre 1992) 91
Bridge Damping
From basic dynamic principles, higher levels of damping reduce the dynamic response in bridges and low levels of damping in a bridge are expected to result in high dynamic amplification. However, damping affects impact differently at different locations within the bridge as a result of varying modal contributions (Huang, Wang, and Shahawy, 1992). Damping values for bridges obtained from field testing can also vary considerably based on the method of testing, level of loading, and different methods used for evaluating damping. Reported values of damping for different types of bridges are as follows:  concrete bridges  2 to 10 % (Tilly, 1978)  steel bridges  2 to 6 %( Tilly, 1978), 0.4 to 1.3 % (Billing, 1984)  composite steelconcrete bridges  5 to 10 % (Tilly, 1978)  prestressed concrete bridges  1 to 2.2 % (Billing, 1984)  timber bridges  3 to 4 % (Ritter, 1995)
Roadway Roughness and Approach Condition
The bridge approaches and roughness of the roadway surface have a significant influence on the magnitude of the dynamic response. Not only do the impact forces increase for increased roughness, but also vehicle speed affects the influence of roughness. The faster vehicle speed has greater impact on rougher surfaces than on better maintained ones. The results of the study by Wang, Shahawy, and Huang (1993) can be seen in Figure 7.4.
92
Figure 7.4. Impact factor versus vehicle speed and road surface condition (Wang, Shahawy, and Huang 1993)
Experimental tests have also shown that the most severe wheel impact forces are likely to occur adjacent to the bridge approaches, i.e., shortly after a vehicle enters the bridge (Tilly 1978). In many experimental investigations, wooden planks were placed in the path of the test vehicle. It was supposed to represent surface irregularities such as dropped objects or packed snow on the roadway. Dynamic response was higher with the planks and the planks were exciting wheel hop in the test vehicles, although excitation of the higher vibration mode associated with wheel hop is also speed dependent.
Vehicle Speed
For most heavy trucks, natural frequencies of the vehicle typically occur in two frequency ranges: between approximately 2 and 5 Hz for the "body bounce" response and between approximately 10 and 15 Hz for the "wheel hop" response. However, depending on vehicle speed, roadway surface irregularities may be effective in exciting both modes of response (Cantieni 1983).
93
Weight of Vehicles
Many studies have shown that as the weight of the crossing vehicle increases, the magnitude of the dynamic response expressed as a percentage of the static load decreases. The explanation of this fact is that when the dynamic forces increase with increasing vehicle weight, the static load increases more rapidly with increasing weight. Thus, the impact ratio of dynamic force to live load decreases with increasing vehicle weight and impact factors obtained from the measured dynamic response of lightly loaded vehicles will be relatively large.
Number of Vehicles
The dynamic load factors associated with multiple vehicles are lower than those for single vehicles. This is most likely because the total static load is larger (similar to having a heavier vehicle) compared to the associated dynamic load, and the dynamic responses from the two individual vehicles are likely to be at least somewhat out of phase with each other.
Vehicle suspension
The vehicle frequency ranges are a function of the suspension systems. The body bounce frequencies in vehicles with air suspensions are lower than those for steel leafspring suspensions, with measured frequencies in the 1.5 to 2 Hz range. Worn dampers in the suspension systems also dramatically increased the dynamic wheel forces.
In Figure 7.5 the influence of suspension is presented. On one deflection trace the vehicle had its normal suspension characteristics, in the other the springs were blocked so that the truck rode directly on the axles. The increase in response is evident for the unsprung condition.
94
5. 1989). it should be concluded that as span length influences bridge fundamental frequency. Effects of vehicle suspension on the measured bridge response (Biggs and Suer 1955) Bridge Span There are conflicting opinions regarding dynamic response and bridge span.7.6. However. other investigations have concluded that considerable scatter exists in the results and there is poor correlation of impact and span (Figure 7. Cantieni 1983).Figure 7. it also indirectly influences bridge dynamics. 95 . Some researchers have concluded that impact is not a function of bridge span (Coussy et al. While some investigations have shown a general trend of decreasing impact in conjunction with increasing span (Figure 7. Fleming and Romualdi 1961).
Impact versus span length (Cantieni 1984) 96 . Impact versus span length (Fleming and Romualdi 1961) Figure 7.6.7.Figure 7.
Dynamic response quantities are sensitive to damping. the most important factor affecting impact is the vehicle speed. and Shahawy (1995a). Dynamic behavior depends on curvature of the bridge. End diaphragms were found to provide lateral support and significantly reduce the response of the box girder bridges. Impact forces are insensitive to curvature for radii greater than 4 000 ft (1219 97 . (Khalifa 1992) Dynamic response of continuous and cantilever thinwalled box girder bridges under multivehicle loading was analytically investigated by Wang. when no support exists between cantilevers. It was found that the dynamic response in horizontally curved bridges is influenced by centrifugal accelerations. Wang. vehicle speed is particularly important. and they are much more susceptible to vibration than continuous bridges. For continuous bridges. The beneficial effect of a midspan diaphragm is relatively small. Cablestayed and suspension bridges are more complicated to assess than for beam/girder bridges. and may be different for different vibration modes.Bridge Type and Geometry Most of the studies on dynamic response were performed on simplespan multigirder bridges. However.20. The dynamic response of simplespan bridges is higher than that for similar continuous bridges. which is difficult to determinate in these types of bridges. It was found that the vibration characteristics of the continuous and cantilever box girder bridges are quite different. mostly due to the influence of cables’ dynamics. impact factors were generally less than 0. Huang. and Shahawy (1996) and Huang. In analytical investigations it was found that. This is due to the abrupt change in loading due to span discontinuities. thus. Impact forces are higher in the outer elements of the curved bridges. for rough surfaces. The investigations concluded that the total number of girders has little influence on the maximum impact factors for each girder and that the impact at the interior supports was larger than at external locations. with a good road surface. both vehicle speed and surface roughness are significant. impact forces increased dramatically. For cantilever bridges.
Wang. Bridge and Vehicle Model To study vehiclebridge interaction. Research done by Galdos et al. 7.m) and markedly influenced by curvature for radii less than 800 ft (244 m). and Shahawy (1995b). with the use of ABAQUS 6.73 in.1 software a finite element model of bridge and vehicle has been developed. 4640 nodes Connection between girders and slab: tied (compatibility of all degrees of freedom) Bridge model meshed with S4R elements is shown in Figure 7.8.93 in. BRIDGEVEHICLE INTERACTION MODEL AND DERIVATION OF DYNAMIC FACTOR 7. The bridge chosen for the analysis is a 120 ft steel girder bridge. flange thickness tf =: 1. (1993) and Schilling et al.1.96 in.60 in2.1. moment of inertia Ixx=19400 in4) spaced 64 in fy = 60 ksi steel diaphragms every 30 ft 2360 elements.pinned right support . 98 . depth of the section d=40 in. It is modeled with 3D shell elements: Steel girders: five steel W40x264 (profile properties: area A = 77. (1992) and Huang.5 in width:312 in fc’=:8 ksi 4464 elements. 2715 nodes Concrete slab: Supports: left support .6. flange width bf = 11. web thickness: tw = 0.roller thickness: 7.1.
tractor wheelaxle set. semitrailer. Figure 7. Moreover each wheelaxle set would be provided with one DOF in the vertical direction (yi). Meshed model of the bridge A vehicle model was developed based on threeaxle AASHTO HS20 truck. A more detailed model could include rotations about longitudinal axis (Φi). Figure 7. shows FEM model of moving masses.8. Two cases are considered. Detailed model would include a seven degree of freedom system. Tractor and semitrailer would have individually assigned two DOFs corresponding to: vertical displacement (yi) and rotation about the transverse axis (θi).9.In this case the total independent DOFs would be eleven. steer wheelaxle set. and trailer wheel axle set. Five sprung masses would be: the tractor.Figure 7. Truck model in FEM 99 . which is the design vehicle in the AASHTO Specifications. the truck HS20 is assumed to travel with the velocity of 40 miles per hour and with the crawling speed.9. To simplify the analysis each wheel is modeled with one DOF in the vertical direction.
as shown in the Figure 7. The effects of damping are small. Tire stiffness and “wheel hop” response are neglected. nearly resonant response can develop. and they will be neglected in the analysis. Under such conditions.7. BridgeVehicle Interaction Model There are two main sources of vibrations induced by the vehicle into the bridge.1. roughness wavelength. 100 . The other source of vibrations is an undulating roadway surface. the interaction model includes only “body bounce”. If the shockabsorbing elements of the vehicle suspension system are worn. then the vehicle's damping is fairly low and the response is quite large.1. A vehicle vibrates with a frequency between 10 and 15 Hz. One of them is the settling of the approach slab of the bridge. The bump at the bridge entrance causes so called “wheel hop” that could be approximated by impulse loading. The front wheels carrying 4 kips each are also assumed to produce force variation ±10%. and vehicle speed. The force Fw transmitted by the rear wheels is 16 kips and its variation ∆Fw is equal to ±10%. Since the truck model is simplified and the scope of the project requires comparison of the deflections in the middle of the span. These kinds of vibrations are called “body bounce” and usually occur at frequencies between 2 and 5 Hz. The condition of driving a vehicle over an undulating roadway surface can be approximately idealized as an SDOF system under harmonic loading provided the roadway varies as a sine wave. Fb = Fw + ∆Fw. Transmitted force is significant. however it fades relatively fast converging into “body bounce” that occurs at lower frequencies.10. The loading is characterized by the roughness amplitude. The vertical interaction force acting on a bridge consists of the static interaction force Fw and the variation of the interaction force ∆Fw. Fw = M g ∆F = ci ⋅ ( dy(t ) dy1 − ) + k i ⋅ ( y (t ) − y1 (t )) dt dt The interaction force is approximated by a sinusoidal shape. which depends on the relation between suspension system and the truck.
It can be noticed when the following truck axes are entering and leaving the bridge.11 is a plot from ABAQUS that shows forces being applied to the bridge versus time.20 15 10 5 0 0 1 t 2 Figure 7. Their amplitudes are interfering and adding to each other. 101 .11. Such a situation of adding amplitudes was simulated on purpose. to obtain the maximum dynamic deflections in the middle span. Figure 7. Figure 7. Plot from ABAQUS.10. Force due to moving truck versus time. Those forces could also be canceling each other. Vehiclebridge interacting force.
09 Hz Figure 7.13 (torsion modes). Torsion modes.13.10 Hz first torsion mode: 3. The natural frequencies are following: first bending mode 2.1.17 Hz second torsion modes: 9.12. 102 .19 Hz third bending mode: 17.1. Results of analysis Mode shapes obtained in the analysis are shown in Figure 7.12 (bending modes) and Figure 7. Bending modes Figure 7.12 Hz second bending mode: 8.7.
weight of the concrete slab and the steel profiles. Maximum deflection due to moving truck (0.65 in) shows 6% difference. but not in the span center.74 in. Maximal deflection due to moving truck Figure 7.16 show deflections in the middle of the span due to moving and stationary trucks.14).The deflection due to dead load.15 and Figure 7. 103 . Figure 7.69 in (Figure 7.69 in) versus maximum deflection due to stationary truck (0. It corresponds to the truck being located almost. Maximal deflection due to moving truck is 0.14. is 2.
Figure 7. Deflection due to a truck moving 40mil/hr versus time. 104 .16. Deflection due to a truck moving at crawling speed versus time. Figure 7.15.
105 . However. for longer bridges where the influence of the approach slab decreases and vibrations of many vehicles interfere with each other. Even for the exemplary case of medium span bridge presented in this thesis the dynamic load factor is only 6%. To be adequate and draw more conclusions further studies should be performed.7. a more elaborate truck model should be performed. CONCLUSIONS The study of the topic and the developed FEM model allowed for estimating the magnitude of the dynamic load factor and to draw some conclusions. Therefore there is no allowance for dynamic. It may be applicable in short bridges. Also. accounting for both “body bounce” as well as “wheel hop”. a traffic jam situation was assumed to develop the live load model. including moving load and defined interaction between surfaces. when vibration due to “wheel hop” on the approach slab is significant. It was concluded that the current dynamic load factor of 0.1 software.33 is too high for bridges with longer spans. In this dissertation. Moreover.6. are non linear and very time demanding. including a wide range of bridge types. spans and roadway conditions. However. to not introduce confusion among designers it is recommended do keep dynamic factor as it is for short and medium span bridges. analytical studies should be confirmed with field tests on the representative structures. which results in very small value for long span bridges. The FEM model was built using ABAQUS 6. Finite element problem modeling. the dynamic load factor could be smaller.2.
RELIABILITY ANALYSIS PROCEDURE Reliability analysis was performed to verify the live load model for long span bridges. (1980). (2008) and Ellingwood et al. software for the tower cross section was developed. 2. fabrication and professional factors was based on the database available in the literature of and described by Nowak et al. Nominal resistance model In order to find nominal resistance. 3.CHAPTER 8 RELIABILTY ANALYSIS OF SUSPENSION BRIDGE 8. The reliability procedure includes the following steps: 1. Evaluation of material. Reliability resistance models The material. Based on the developed software for nominal 106 . 4. Selection of a representative bridge and its component A representative suspension bridge and the structural element that is the most influenced by live load were selected. fabrication and professional factors were established.1. Limit state function The limit state function was defined as the exceeding of the ultimate bending moment capacity by the crosssection and Strength I combination.
Load model The threedimensional FEM model of a bridge was created using Robot Millennium software. 8. SELECTION OF REPRESENTATIVE STRUCTURE. the bridge component that is the most influenced by live loading had to be selected. Cross sectional axial force and bending moment along the tower height due to dead load and live load were derived. ELEMENT AND LIMIT STATE FUNCTION The Cooper River Bridge in South Carolina was chosen to be a representative long span bridge for this analysis. such as the deck and cables. 107 . For the scope of this dissertation. Reliability analysis was performed for the several forces possible and moment in the bridge tower due to live load cases. The bending moment due to dead load of opposite spans almost negate each other. 6. It is a suspension bridge designed by PB Word in 2001. For long span bridges dead load is the main loading.resistance and the Monte Carlo simulation method. the bending moment due to live load is much higher than the bending moment due to dead load. The bridge tower was chosen to be such an element. The tower crosssection just above the deck level was selected. Live load from all spans is transferred through cables to the bridge tower. Reliability Indices Reliability indices were calculated in order to assess how they are influenced by the increase in the values of live load.2. 5. The limit state function was defined as the exceeding of the ultimate bending moment capacity by the crosssection. When not all spans are loaded evenly. The statistical models for load components are defined. the statistical parameters for resistance were obtained. The model was based on the actual Cooper Bridge design made by PB World. Dead load is also critical for most of the structural components.
The mathematical description of the behavior for the exemplary bridge tower was developed based on procedure for columns described by Lutomirski (2009).3. The first step in the procedure is the calculation of the initial compression block. representing all possible cases of force and bending moment combinations. It means that strains in concrete at the top of cross section reach value of ε m = 0.2) 108 .8.00207 (Figure 8. NOMINAL RESISTANCE The nominal resistance can be represented by interaction diagram of force and moment for the eccentric loaded bridge tower. The assumptions are based on linear strains in a distribution over the cross section and the mechanical behavior of reinforcement. The position of the neutral axis in the initial step can be calculated from: cI = aI β1 (8. a MathCad software was used. The initial compression block represents the compression over the full crosssection and the lowest layer of the reinforcement yields due to compression. To plot interaction diagram.2). The size of the initial compression block can be calculated as follows: a I = β1 εm ⋅ dI εm −ε y (8. The procedure takes into account the geometry of the tower. all layers of the reinforcement in the crosssection and the characteristics of the materials (steel and concrete).1) where: dI is the distance of the lowest layer of the reinforcement to the top of the crosssection. while the strains in the lowest reinforcement bar represent yielding strains due to compression in steel ε s = ε y = 0. The position of the neutral axis can be calculated based on the linear strain distribution and strain compatibility assumptions.003 . As a result of the first step the size of the compression block is much bigger than the size of the crosssection.
1 Distribution of strains for the pure axial loading The characteristic point of end of compression control zone and beginning of tension control zone call balance failure point can be derived. the end of the compression control zone 109 . The size of the compression block in balance failure is represented: a B = β1 εm ⋅dI εm + εy β (8.Figure 8.2).2 Distribution of strains distribution for the balance failure point B.3) ε ε Figure 8. It happens when the strains in the bottom layer of reinforcement reaches the yielding strains for the steel εs=εy (Figure 8.
4 and Figure 8.3). For each reinforcement layer and for each size of the compression block the strains in reinforcement are computed from the equation: ε s (i. a) = ε m ⋅ 1 − where: i di c( a ) (8.5). The first case is when the position of neutral axis is outside of the cross section (Figure 8. four characteristic cases there can be distinguished.003 distance of ith reinforcement layer for the top of crosssection position of neutral axis due to changing size of compression block a εm di c (a ) In the procedure. The second case is when the end of compression block of concrete is in the bottom flange of the cross section (Figure 8. from the initial aI up to the point when the compression block does not exist ( a = 0 ).The entire force and moment interaction diagram is calculated using the decreasing size of the compression block of concrete.4) number of ith reinforcement bar in crosssection size of compression block of concrete extreme compressive strain in concrete equal to 0. Two next cases correspond to the end of compression block of concrete localized in the webs and in the top flange (Figure 8.1).3 End of compression block of concrete in the bottom flange 110 β . ε Figure 8.
ε β Figure 8.5) where: Es = fy εy (8. ε β Figure 8. and it is described by the following equation: − fy E ⋅ ε fs = s s fy 0 ε s < −ε y for for − ε y ≤ ε s ≤ ε y for ε y < ε s ≤ ε m εs > εm for (8.6 shows material behavior of reinforcing steel.5 End of compression block of concrete in the top flange. Figure 8.6) 111 .4 End of compression block of concrete in the web.
a ) ≤ ε y −A ⋅ f for ε s (i.85 ⋅ A( a ) ⋅ f c ' (8. a ) for − ε y ≤ ε s (i. a ) ≥ ε y P(i.10) 112 . a ) = As ⋅ E s ⋅ ε s (i.8) The force in concrete is based on the size of compression block of concrete. a ) + Pc ( a ) i (8. a ) i (8. the resistance force of the cross section is expressed by the sum of all forces acting in the crosssection: PTotal ( a ) = ∑ P (i .Figure 8.7) where: i a number of ith reinforcement bar in crosssection size of compression block of concrete The resultant reinforcement force is calculated using: Psteel (a ) = ∑ P (i.Strain Relationship for Reinforcing Steel Having calculated strains in every reinforcement bar.9) For each size of the compression block.6 Stress . it is possible to evaluate forces for each reinforcement layer. a ) < −ε y s y (8. As ⋅ f y for ε s (i. PC (a ) = 0.
Main span is 1546 ft long. Geometry of the bridge is shown in Figure 8.For each size of the compression block. φ2x . φ1z . The cables elements were used to model suspension cables. Shell elements were used to model concrete slab in the bridge. Total length of the structure is 3296 ft.to represent all others members: towers leg. φ1x . two spans are 650 ft and two spans are 225 ft long. tower. φ2y . 113 .85 ⋅ A(a) ⋅ f ' c ⋅YC (a) 2 i (8. The three dimensional model was based on the actual design made by PB World. the bending moment resistance is equal to sum of all the forces in the section multiplied by the corresponding force arm to the centroid of the crosssection: h M Total (a) = ∑ P(i. u2y . u2x .with 12 degrees of freedom u1x .7. LOAD MODEL For the scope of this dissertation Cooper River Bridge was modeled using Robot Millennium software.11) 8. a) ⋅ − d i + 0. u1z . both of them have 368 ft above deck level. u2z . φ2z . 840 3D beam elements were used . φ1y . Bridge towers are 568 ft high. u1y .4. girders and the diaphragms.
The 31 combinations were used four times.7.00 k/ft and 1. Geometry of Cooper River Bridge In the analysis all loading cases have been considered. The resulting envelopes of bending moments due to various combinations of live load for bridge tower are shown in Figure 8. They are shown in Figure 8. for the different value of loading: 0.80 k/ft. 0. All of the live load combinations were used to calculate load effect on the bridge tower.9 .Figure 8.12.8. there are 31 loading combinations. For five spans.20 k/ft as a value of lane live load loading.Figure 8. 1.64 k/ft. 114 .
Figure 8. Load combinations 115 .8.
500 Possition on Pylon (ft) 400 LL1 LL2 LL3 LL4 LL5 LL6 300 LL7 LL8 LL9 LL10 LL11 LL12 200 LL13 LL14 LL15 LL16 LL17 100 LL18 LL19 LL20 LL21 LL22 LL23 0 400000 300000 200000 100000 0 100000 200000 Mz (kipsft) 300000 LL24 LL25 LL26 LL27 LL28 LL29 100 LL30 LL31 200 Figure 8.9. Envelope of bending moments for bridge tower for w=0.64k/ft 116 .
500 Possition on Pylon (ft) 400 LL1 LL2 LL3 LL4 LL5 LL6 300 LL7 LL8 LL9 LL10 LL11 LL12 200 LL13 LL14 LL15 LL16 LL17 100 LL18 LL19 LL20 LL21 LL22 LL23 0 500000 400000 300000 200000 100000 0 100000 200000 300000 Mz (kipsft) 400000 LL24 LL25 LL26 LL27 LL28 LL29 100 LL30 LL31 200 Figure 8. Envelope of bending moments for bridge tower for w=0.10.80 k/ft 117 .
500 Possition on Pylon (ft) 400 LL1 LL2 LL3 LL4 LL5 LL6 300 LL7 LL8 LL9 LL10 LL11 LL12 200 LL13 LL14 LL15 LL16 LL17 100 LL18 LL19 LL20 LL21 LL22 LL23 0 600000 500000 400000 300000 200000 100000 0 100000 200000 300000 400000 500000 Mz (kipsft) LL24 LL25 LL26 LL27 LL28 LL29 100 LL30 LL31 200 Figure 8.11.00 k/ft 118 . Envelope of bending moments for bridge tower for w=1.
20 k/ft 119 .12.500 Possition on Pylon (ft) 400 LL1 LL2 LL3 LL4 LL5 LL6 300 LL7 LL8 LL9 LL10 LL11 LL12 200 LL13 LL14 LL15 LL16 LL17 100 LL18 LL19 LL20 LL21 LL22 LL23 0 800000 600000 400000 200000 0 200000 400000 Mz (kipsft) 600000 LL24 LL25 LL26 LL27 LL28 LL29 100 LL30 LL31 200 Figure 8. Envelope of bending moments for bridge tower for w=1.
14 show the bias factor and coefficients of variation for all types of concrete and all nominal compressive strengths of concrete.S. in most cases. 1980).13 and Figure 8. Therefore. F and P . the resistance of a structural component R . The statistical parameters for material factors used in this dissertation were based on the project "ReliabilityBased Calibration for Structural Concrete" (Nowak A. 120 . 2008).12) The materials factor represents material properties. et al. can be considered as a random variable being a product of nominal resistance Rn and three factors: the materials factor.5.1. There is no new information regarding two other factors. Figure 8. 2008). The professional factor represents the approximations involved in the structural analysis and idealized stress/strain distribution models. Material Factor The material factors for concrete were based on the study within the project "ReliabilityBased Calibration for Structural Concrete" (Nowak A. and section modulus.8. In both figures there is a trend line of changing parameter with respect to concrete compressive strength f c ' . statistical parameters for F and P are taken from the previous study (Ellingwood et al. in particular strength and modulus of elasticity. fabrication factor and professional factor: R = Rn M F P (8.S. moment of inertia. 8.. RELIABILITY RESISTANCE MODELS Due to various categories of uncertainties.5. including crosssectional area.. The fabrication factor represents the dimensions and geometry of the component. et al. The professional factor is defined as the ratio of the test capacity to analytically predicted capacity (the actual insitu performance to the model used in calculations). the materials factors have been updated based on a new test database. Because the quality of materials such as reinforcing steel and concrete has improved over the years.
Statistical parameters assumed are: bias factor λ = 1.7 0. the concrete compressive strength of bridge tower is 7000 psi.12. In this dissertation.2 λ 1.4 1.8 high strength 0. 1.6 1. 2008) 121 .9 light weight 0.5 1.13 Bias factor for compressive strength of concrete (Nowak A.3 1.S.13 and coefficient of variation V = 0..0 0.Recommended values are summarized in Table 8. plant cast fc' Figure 8.1. ready mix ordinary. et al.1 1.6 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 approximation ordinary.
15 0. Szerszen 122 13000 .14 Coefficient of variation for compressive strength of concrete (Nowak A.08 0.S.S.11 0. plant cast 10000 11000 12000 fc' Figure 8.08 The material factors for reinforcement steel were based on the study within the project "ReliabilityBased Calibration for Structural Concrete" (Nowak A.14 0.17 0.10 V 0.06 0. 2008) Table 8.05 0.18 0. 2008) Concrete Grade fc' (psi) 4000 6000 8000 12.13 0.11 1.07 0.02 0.15 0. et al.S.125 0.04 0..12 0.16 0..24 1. f c ' (Nowak A.01 0.00 2000 3000 light weight high strength 4000 5000 6000 ordinary.. ready mix approximation 7000 8000 9000 ordinary. et al.11 0.15 1.03 0.0.1 Recommended Statistical Parameters for Compressive Strength.000 λ fc ' V 0.09 0.11 1.
03 2 Normal Inverse Probability 1 0 1 2 3 4 50 60 70 80 90 100 110 Steel rebar size: #3 #9 #4 #10 #5 #11 #6 #14 Yield strength of rebars [ksi] #7 All diameters #8 Approximation Figure 8.15 CDF’s of yield strength for Reinforcing Steel Bars.9 and No.13 V = 1.. et al. Those recommended values have been used in this dissertation for the bar sizes No. which were used in the calculations..03.S. 4 3 λ = 1.3 to No. et al. Grade 60 ksi (Nowak A. and different bar sizes from No.13 and coefficient of variation V = 0. Data included the yield strength for the reinforcing steel bars with the nominal yield strength of 60 ksi.M. 2008). The recommended values of statistical parameters are: bias factor λ = 1. Plots of the cumulative distribution functions (CDF) of yield strength of every reinforcement size and recommended parameters are shown in Figure 8. 2008) 123 .M..11.16.14.15 and Figure 8.
13 V = 1.1.2.2 Statistical Parameters of Fabrication Factor. Material Concrete Reinforcement cover Reinforcement area Steel Reinforcement diameter 124 1.00 0.. the parameters for columns were assumed. 2008) 8.S.16 Recommended material parameters for reinforcing steel bars. Table 8. et al.04 Bias factor V .03 Normal Inverse Probability 2 1 0 1 2 3 4 50 60 70 80 90 100 110 Yield strength of rebars [ksi] All diameters Normal Approximation Lognormal Approximation Figure 8.. Due to lack of data for bridge towers. Grade 60 ksi (Nowak A.005 0. They are summarized in Table 8.4 3 λ = 1. Fabrication Factor The statistical parameters of fabrication factor are based on previous studies by Ellingwood et al.5. 1980.015 Item Radius of column 1.
(1980). 8. the statistical parameters of professional factors for columns were used in this dissertation. 250000 mean+std 200000 nominal mean meanstd Force [kips] 150000 Balance Failure 100000 50000 0 0 200000 400000 600000 800000 1000000 Bending Moment [kft] Figure 8. The bridge tower behaves as an eccentrically loaded column.1.16 and bias factor λ=1. Therefore. Statistical Parameters of Resistance Statistical parameters of resistance were obtained by 10000 Monte Carlo Simulation: coefficient of variation V=0.5.8.5.17.17 Statistical Parameters of Resistance 125 . Professional Factor The statistical parameters of professional factor are based on the previous study performed by Ellingwood et al.08. The professional factors were chosen for tied columns: bias factor is λ 1.00 and coefficient of variation is V=0.1.
10. Reliability indexes were calculated for bridge loaded with AASHTO design live load 0. Nowak 1999). 8.80. and idealization of analytical models affects statistical parameters of resistance. influence the bridge behavior. LOAD MODEL The case considered in this study is combination Strength I. However.05 and coefficient of variation of 0. The assumed statistical parameters for dead load are based on the data available in literature (Ellingwood et al. For every value of lane load the 31 load cases were analyzed according to paragraph 8.6.7. 1.4.00 and 1. the selected limits state was chosen to demonstrate the sensitivity of reliability index on long span bridges due to change of live load.8.4 of this dissertation. RELIABILITY ANALYSIS Reliability analysis was performed for the considered bridge. The limit state function was selected as the Strength I load combination according to AASHTO LRFD Code. The bridge was design by the PB World for the design live load specified in AASHTO LRFD CODE. which is caused by variation of the weight of materials (concrete and steel). They include for castinplace concrete elements a bias factor of 1. This combination has the highest load factor for live load. 1980. Therefore there is no need to use Turkstra’s rule. variation of dimensions. which is combination of dead load and live load. Variation in the dead load. There are only two major load components: dead load and live load.64 k/f t and three other possible load cases of 0. generates separate loading case to the bridge tower with different 126 . Figure 8.20 k/ft. Each load case. such as wind load.8. In this study live load is the only varying load and it is assumed to reach the maximum 50year load value. on the real structure in a specific localization other loads. The element selected for the analysis was the bridge tower subjected to the bending moment and the axial force. For time varying loads two random variables for arbitrarypointintime and maximum 50year load components can be considered. Variation of live load is derived in Paragraph 5.
Nominal Resistance 180000 Force [kips] 160000 140000 120000 100000 80000 60000 40000 20000 0 750000 500000 250000 0 Moment [kft] 250000 500000 750000 Balance Failure Nominal Loads w=0.0 k/ft Nominal Loads w=1. specified in AASHTO LRFD Section 5.80 k/ft Nominal Loads w=1. It means that all the cases are in the compression control and the strength reduction factor is φ=0. The results vary depending on eccentricity.64 k/ft Nominal Loads w=0. 127 . For the cases with the eccentricities approaching the balance failure the reliability indexes are decreasing with the increase of the actual loading. None of them exceeds the balance failure zone.75.eccentricity condition. It can be noticed that values of β are in high range for the cases with the eccentricity of loads very small for all possible loading conditions.18 .19 shows the results of reliability indexes due to different live loads. Figure 8.18 Force and Moment results on the bridge tower for different live loads Figure 8.2 k/ft Figure 8.
00 4.64 0.20 2.60 0.00 1.00 1.00 β 3.80 1.00 6.19 Reliability indexes due to different live load 128 .40 Uniformly Distributed Load [k/ft] Figure 8.80 1.7.20 1.00 0.00 5.00 0.00 0.00 0.00 1.40 0.20 0.
The newest available traffic database from a variety of sites within many different states was obtained. because such an extensive actual weigh in motion database has never been used in the derivation of live load for any kind bridges. the first truck was deleted. Then. In the developed procedure. and one or more trucks were added so that the total length of trucks covers the full span length and the new value of the average uniformly distributed 129 . The developed live load model is recommended to be taken into consideration in the bridge design code. quantities of trucks and their weights. legal load trucks and simulation of a traffic jam using WIM data. The live load model is valid for spans between 600 ft and 5000 ft and it is intended to reflect current traffic patterns. Equivalent uniformly distributed load is calculated and compared. From the simulation the values of uniformly distributed load were derived for a variety of span lengths and site localizations. starting with the first truck. the total load of all trucks was calculated and divided by the span length to obtain the first value of the average uniformly distributed load. all consecutive trucks were added with a fixed clearance distance between them until the total length reached the span length. Preliminary study was performed by reviewing previous research and current provisions of international codes on the topic.CHAPTER 9 SUMMARY & CONCLUSIONS In the study. Next. The magnitude of the database has to be underlined. a live load model for long span structures was derived. A numerical procedure was developed to filtered out WIM data from erroneous readings and to simulate traffic jam situations. The new live load was developed based on three models: an average 5axle truck.
Those sites are characterized by high ADTT (usually over 3000) or increased percentage of overloaded loaded vehicles (over about 10%). It was noticed that the bias factor values for most of the sites do not exceed 1. Multilane factors were found to be very site specific. the bias factor reaches a value of 2. an average 5axle truck and a legal load trucks model led to similar conclusion. for spans longer than 1000 ft bias 1.25.20 and coefficient of variation 0. Cumulative distribution functions were also plotted for maximum daily and maximum weekly combinations of trucks. which is similar as in short and medium spans. Two other models. as with the live load. Results of the simulations were plotted as a cumulative distribution function of uniformly distributed load for considered span lengths. Bridges located in such sites require special attention and application of increased design live load. However. the problem of multilane reduction factors was discussed. For some bridges considered.25 and coefficient of variation 0. In this dissertation.load was calculated. In the WIM database available. This observation confirms that for a long loaded span. as recorded in the WIM surveys.75 k/ft.0. uniformly distributed load decreases and is closer to the mean value.10. The 75year uniformly distributed loads were derived from extrapolated distributions. the vehicles of 13 categories 13 have not been registered. The value of the design load should be agreed with the owner of the structure. Multiple reduction factors for design live load should account for those most critical loading cases. The bias factors (ratio of mean to nominal) were calculated for the heaviest 75year combination of vehicles. situations when one lane is almost exclusively occupied by trucks or trucks occupy three or four lanes at the same time are also possible. for example in the area of New York.25 kip/ft in order to obtain bias lower than 1. For longer spans. It is recommended to use HL93 also for those long spans. Statistical parameters for live load are: for spans 6001000 ft bias 1. it was found that the uniformly distributed load should be 1. Video recordings of traffic confirmed that for the majority of the time we can observe that the moving lanes contain a mixture of trucks and cars.50 and 0. The obtained mean value oscillates between values of 0. as shown in the NCHRP Report 368 (1999). it did not allow for simulations and 130 . Trucks were kept in actual order. one heavily overloaded truck does not have significant influence.08. Therefore. It was noticed that for some sites. This is because the load depends on a mix of traffic. with very heavy traffic.25.
which results in very small value for long span bridges. the dynamic load factor could be smaller. Simulation of the traffic on the most loaded lane was possible with an assumption. It was concluded that the current dynamic load factor of 0.derivation of multilane factors for all traffic lanes. to not introduce confusion among designers it is recommended do keep dynamic factor as it is for short and medium span bridges. when vibration due to “wheel hop” on the approach slab is significant.33 is too high for bridges with longer spans. 131 . that in traffic jam situations passenger vehicles merge to the left and the right lane remains occupied exclusively by trucks. The study of dynamic factor was performed for the research. For longer bridges. It may be applicable in short bridges. At least one of the lanes should be loaded more than the others. However. where the influence of the approach slab decreases and vibrations of many vehicles interfere with each other. It was stated that equal reduction of load on all traffic lane does not reflect the actual situation. The assumption of a traffic jam situation to develop live load model induces no dynamic allowance. It was concluded that the multilane reduction factors have to be an objective of additional extensive studies.
The bias factor calculated for the heaviest 75year combination of vehicles did not exceed 1. uniformly distributed load of 0. specified in AASHTO LRFD Code (2007). which should be agreed with the owner of the structure. 3. to not introduce confusion among designers it is recommended do keep dynamic factor as it is for short and medium span bridges.CHAPTER 10 RECCOMENDATIONS 1. The developed live load model is valid for long spans between 600 ft and 5000 ft.25. It was proposed to use dynamic factor as specified in AASHTO LRFD Code (2007). which results in very small value for long span bridges.64 k/ft plus design truck or tandem. It was proposed to use multilane reduction factors as specified in AASHTO LRFD Code (2007). 4. However. as shown in the NCHRP Report 368 (1999). which is similar as in short and medium spans. 132 . Developed live load model assumes traffic jam situation and does not allow for dynamic. 2. 5. Those bridges require application of increased site specific design live load.0. For long spans it is recommended to use HL93 load. For some sites characterized by high ADTT or increased percentage of overloaded vehicles (over 10%) bias factor reaches a value of 2. It is recommended to perform further studies in this field.
6. The developed live load model is recommended to be taken into consideration in the AASHTO LRFD Bridge Design Specifications.08. for spans longer than 1000 ft: bias 1.10.25 and coefficient of variation 0.20 and coefficient of variation 0. Statistical parameters for live load are: for spans 6001000 ft: bias 1. 133 .
Barrata B. W. IABSE. Berlin. Lisbon.. ASCE Committee on Loads and Forces on Bridges (1981). [4].” Proceedings of the Symposium on Resistance Ultimate Deformability of Structures. Augusti G.C.” Proceedings of the Second SMiRT Conference. New York.M. “Recommended Design Loads for Bridges”. Ontario Highway Bridge Design Code. 7. 17071724 134 . Washington. 1984 [9].. CAN/CSAS600 Canadian Highway Bridge Design Code. Canada. “Probability Concepts in Engineering Planning and Design. A. new York. Toronto.. pp.. W. Risk. ASCE Journal of Structural Engineering. [2]. [7]. “Probability Concepts in Engineering Planning and Design. pp. No. and Tang. Ayyub B. Ontario Ministry of Transportation. 110. 11611213. 2003. 1975 [8].C. Vol. OHBDC (1991).. and Tang.H.” Volume I: Basic Principles.C.H. John Wiley & Sons. HS.” Volume II: Decision. D. “Limit and Shakedown Analysis of Structures with Stochastic Strengths. American Association of State Highway and Transportation Officials.. No. 107. CSA International. 1973 [10]. AASHTO Geometric Design of Highways and Streets (2001). “Plastic Shakedown of Structures with Stochastic Local Strengths. Washington.REFERENCES [1]. D.. Ang. [6]. (2007). Ontario. Washington. Augusti G. [3]. A. Manual for Condition Evaluation and load and Resistance Factor Rating (LRFD) of Highway Bridges. American Association of State Highway and Transportation Officials. 1973 [11]. December 1981. ASCE. Halder A. HS. “Practical Structural Reliability Techniques. John Wiley & Sons.” Journal of Structural Engineering. Barrata B. Downsview. American Association of State Highway and Transportation Officials. D. AASHTO LRFD Bridge Design Specifications. 1984. Vol. Ang. 8. and Reliability. [5].
“Handbook of vehicleroad interaction”. Vol. 1970 [14]. “North American and British LongSpan Bridge Loads”. University of Cambridge.. 29722987. [15]. 146154. Cantieni. ASCE.V. Cebon. pp. 141148.P. Benjamin J. Cornell. Vol.. Buckland. 915932.P.R.A. “Traffic loading of longspan bridges”. “Probability. 1967 [23]..V. ASCE.” Journal of the Structural Division. 171200. 1984. Highway Research Board. “Vibration Measurements on SimpleSpan Bridges”.” Journal of Structural Engineering. P.. No. 1978. [22]. pp. and McBryde. Biggs. Buckland. [21]. McCuen R. 115. 1999. London. 24352447 135 . [20].” CRC Press Boca Raton. University of Cambridge. “Dynamic Load Testing of Highway Bridges”. Corotis R... Vol. London. A.. 1993. 117. 1997 [13]. New York. 1980. pp. Cebon. Suer. pp. 665. ASCE. Breitung K. 106.G.. P. C. “Probability.. pp. Statistics and Decision for Civil Engineers. F. Ayyub B.. Transportation Research Record. Statistics.” Journal of Engineering Mechanics.M. Bulletin 124. F. J. Zidek. H. Vol. 1955. D. “Interaction between heavy vehicles and roads”.M. Journal of Structural Division. Navin. J. Zidek.. ASCE. D. 357366 [16].G. “Bounds on the Reliability of Structural Systems. P. J. “Proposed Vehicle Loading of LongSpan Bridges”. Navin. and McBryde.G.” McGrawHill.M. 10. “Asymptotic Approximations for Multinormal Integrals. Department of Engineering. Journal of Structural Division. [18]. R. 1991. 1989. & Reliability for Engineers.D. 1984. 110.P.D. Vol. ASCE. 93 (ST1).P.B. J. Cornell C. J. [17]. Nafday A. Department of Engineering. New York.H. “Structural Systems Reliability using Linear Programming and Simulations.[12].S. Transportation Research Record No 950. pp. [19]. Buckland.
ASCE. C.. T. 13371375 [29]. Naumann HJ.” NBS Special Publication 577. “An Exact and Invariant First Order Reliability Format.. TRB annual meeting [32].. and Cornell. [28]. Vol.” Journal of Structural Mechanics Division. 19171934.. DC. ST5.. Vol. J. Washington. 1. No.A. Proc. 100. National Bureau of Standards. Sahin.” Journal of Structural Engineering. 1982 [26]. 1980 [25]. Fraudenthal A. ASCE. Galambos. MacGregor. 108. ST9. N.P.M. “Development of a Probability Based Load Criterion for American National Standard A58.M. Vol. Vol... and Cornell. H. No. Vol. C. Hasofer A.V. Journal of Structural Engineering. pp. MacGregor. M. Galambos.H. “Methodology for Impact Factor of Horizontally Curved Box Bridges”. F.V. Fleming. and Ravindra.. Nassif...A. “Quadratic Limit States in Structural Reliability. T... 1961. “Load and Resistance Factor Design. M. pp. 1956. Ellingwood. Lind N.C.K. 6. B. 119. 1978 [30]. pp. Rackwitz R. Sep. ASCe.. 2006..A. D. J..” Transactions. “Probability Based Load Criteria: Load factors and Load Combinations. 1993. Ellingwood. Fiessler B. “Dynamic Response of Highway Bridges”. Romualdi.[24].V.. ASCE. Vol. T. 121. “Multiple Presence Statistics for Bridge Live Load Based on WeighinMotion Data”.” Journal of Engineering Mechanics. 87.H.G. Galambos. Gindy.J.G. 105.” Journal of Structural Devision.R. Galdos. ASCE. Schelling. 661676 [27]. Paper 14008. 1974 136 . “Safety and Probability of Structural Failure. B. J. No. Journal of the Structural Division. [31]. M... 1979.
[33].H. Vol. ASCE. [37]. 1990.S. pp. ASCE. pp. NCHRP 207/186. “Parametric Study of CableStayed Bridge Response Due to Trafficinduced Vibration”. “A computerized model for simulating the breaking and steering dynamics of trucks. 21952199 [34]. doubles. [42]. Hwang. tractorsemitrailers. Gillespie.S. New Jersey. A.. D. “Improvement of SecondOrder Reliability Estimates by Importance Sampling. M. [44]. “Transportation Engineering”. 6983.. 1476. Khisty. H. Marsh. pp. and Nowak. “Dynamic Impact Factors for bridges”. P. “Reliability models for circular concrete columns”. 114.” Journal of Engineering Mechanics. C. C. 85104 [39].T. Sokolik.M. 2009.C. Mertz D. Navin. Kim. No. 5. 137 . 2.1541.. January 2007 [40]. and triples combinations”.. Khalifa. pp. Rackwitz R.. 1992. 321339. [36]. 1995. [41].L. A. [43]. Transportation Research Record.D.. “Dynamic Load Spectra for Girder Bridges”. MacAdam. pp. Nassif. Kiureghian A. A. No. Kulicki J.F. Prucz Z.. Vol. Transportation Research Record. “Structural Reliability under Incomplete Probability Information.. Computers & Science. pp. 9. “Updating the calibration report for AASHO LRFD”. Clancy C. “Measurement of Truck Load on Bridges in the Detroit Area”. 117. Canadian Journal of Civil Engineering. T. 1980.J. No.. [38]. ES.S. No. 468476. 1998.A. 5863. [35]. 47. Hu. Doctoral Dissertation.M.R. and Nowak.S. Englewood Cliffs. NCHRP Synthesis 266.. “Simulation of Dynamic Load for Bridges”. “Road vehicles on bridges”.. 1986. Vol. 1997. May 1991. FPD. Liu PL. Hohenbichler M. Nowak A.. 1988. pp.I. Vol. 1982. and Nowak. M.. Fancher. McLean.S. ASCE Journal of Structural Engineering.. Prentice Hall. MVMA Project 1197.. A.. Vol.A. SJ. Lutomirski. T.” Journal of the Engineering Mechanics... G. 112. 14131434.
. and Nassif. Final Report submitted to MDOT. A. New York. A. [52]. and Lind . Transportation Research Record. 12451251. “Practical Bridge Code Calibration. Nowak.” McGrawHill New York. Nowak. YK and Hwang. ASCE Journal of Structural Engineering. SJ.S. Vol. D. 1289. [55]. 9. Final Report. Laman.. pp.S. December. “Bridge Live Load Models”. “Live Load Model for Highway Bridges”. Nowak. ASCE Journal of Structural Engineering. pp. N. Kim..S. H.. 5366. 3649. Final Report submitted to MDOT.. [54]. Laman. pp. Journal of Structural Safety. pp.S. November 1994. No. Nowak. “Reliability of Structures.[45].R.. and Collins.S. A. 1999 [46]. No.S. Transportation Research Board. 1994. Nowak. J. “Load Model for Bridge Design Code”. ASCE. A. ES. Sept.. “Calibration of LRFD Bridge Code”. V.S. 1991. “Modeling Live Load and Dynamic Load for Bridges”. Nowak. Hong.S. Vol. 27572767. USA. December 1993..” NCHRP Report 368. 121. “Calibration of LRFD Bridge Design Code”.D. 110118. USA. No. A.. Washington. [50]. “Truck Loads on Selected Bridges in the Detroit Area”.. 1993. Nowak. A..S. Vol. 13.. 8.C. Nowak. pp. 2000 138 . Nowak. Nos. “Calibration of LRFD Bridge Design Code. A. [53]. December 1979 [47].. [51]. J. “Effect of Truck Loading on Bridges”.F. 117. 21. YK.” Journal of Structural Division.S. December 1994. [48]. K.. Vol.. A. Nowak. Saraf. pp. A.S. A. and Sokolik. Nowak. 1991. Canadian Journal of Civil Engineering. A. and Hong. 1995.. A. NCHRP Project 1233. 24972510. 1+2... [49].
ThoftChristiansen. Poland. Title no. “Evaluation of Impact Factors for Horizontally Curved Steel Box Bridges”. 2. Vol. pages 167 to 176. M. No. NCHRP Project 1276. New York.” Computers & Structures. Berlin. B.H. Lind N. F.A. [64]. 1982 [65]. Canadian Journal of Civil Engineering. Y. P. Paultre. Heidelberg.. ACI Structural Journal.C.” SpringerVerlag. Mirza S. 1978. Fiessler. 489494 [61]. [57]. M. 1986 [58]... Madsen H. P. Rackwitz.” Final Report.. Chaallal. B.. “Application of Structural Systems Reliability Theory. Proulx. “Reliability of Engineering Structures” (in Polish). Ghosn. No. Sahin. Berlin.. “Bridge Dynamics and Dynamic Amplification Factors: A Review of Analytical and Experimental Findings”. J. 19. “Protocols for Collecting and Using Traffic Data in Bridge Design... pp. 11.[56]. Krenk S. ThoftChristiansen. MacGregor J. “Remarks on Multivariate Transformations.O. “Structural Reliability Theory and Its Applications. Schelling. Warsaw. Vol. [63]. 1952. “Statistical Analysis of Resistance of Reinforced and Prestressed Concrete Members”.8016. pp.R. New York. P.... Tokio. 1992. M. pp. “Structural Reliability under Combined Random Load Sentences.A. Galdos. Sivakumar. “Methods of Structural Safety. R. N. Arkady. Ellingwood B. Baker. 23. O. 1989 [59].. MayJune.J. Vol. Moses. 118. 260278. [60]. Rosenblatt. D. Heidelberg.. 1986 139 .” PrenticeHall. Murotsu. 1983. Murzewski J. 9. pp.” The Annuals of Mathematical Statistics. 470472 [62]. Vol. July 2008. Journal of Structural Engineering.” SpringerVerlag.G.. 1992. 32033221.. M.
S.S. 2001 [71]. 2.Q. pp. Washington. Winkler. [77]. Shahawy. Journal of Bridge Engineering. pp. 5. Journal of Structural Engineering. 116..B. T. Volume I: Summary Report”. Lengths.. C. Huang. Wolinski Sz. 4. Journal of Sound and Vibration.B. 140 . Vol.. D. [68].Z.S. Shahawy. M. 2002. 1983. “Cablestayed Bridge Vibration Due to Road Surface Roughness”. [69]. and Widths of Commercial Motor Vehicles”. Vol. 1993. DC. Wang. [75]. “Comprehensive Truck Size and Weight Study. Computers and Structures. pp. T. 17721793. Department of Transportation. D. 7. [72]. No. M. Fancher. Publication FHWAPL00029.L. Vol. Law. Special Report 267. Traffic Monitoring Guide. TRB. Michigan Center for Tuck Safety. 1. P. [76]. S. No. Wrobel K. No. 13541374.. “Model for Highway Bridge Impact”. Department of Transportation. Transportation Statistics Annual Report. Washington. [67]. Journal of Structural Engineering.Z. pp. Hunag. [73]. 697716. 6775. No. 10551067. Wang. C. Wang.S. 2000.. [74]. 6. “Reliability of Engineering Structures” (in Polish).S.[66]. “Dynamic Load on Continuous Multilane Bridge Deck from moving vehicles”. 1996. Department of Transportation. UMTRI8313. “Dynamic Behavior of Continuous and Cantilever ThinWalled Box Girder Bridges”. “Parametric Analysis of heavy duty truck dynamic stability”. National Research Council. Dynamic Response of Highway Truck Due to Road Surface Roughness. 2002. 1992. Federal Highway Administration.L. Zhu. Vol. No. December 2006. 2001. Vol. Publishers of Rzeszow Technical University (Oficyna Wydawnicza Politechniki Rzeszowskiej). 2008.. Truck Driver's Guidebook. FHWA. MacAdam. X.. “Regulation of Weights. U. 49.L. DC. 188. T. D. U. U.Z.. pp. Huang. [70]. 251.
00 0.40 0.00 5.00 4.APPENDIX A CDF OF UDL FOR ALL TRUCK COMBINATIONS 5.00 1.00 2. lane 1 141 .00 1.00 3000 ft 4000 ft 5000 ft 2.60 1.00 3.00 0.00 Figure A.20 0.00 4.40 1. CDF of UDL for Oregon I5 Woodburn.80 2.60 0.80 1.00 600 ft 1.1.00 1000 ft 2000 ft 0.00 3.20 1.
20 1.00 1.00 4.00 5.20 1.60 0.40 0. lane 1 5.00 2.80 2.00 1.5.00 0.00 Figure A.20 0.80 1.40 1.00 4.00 3. CDF of UDL for Oregon OR 58 Lowell.00 3. CDF of UDL for Oregon I84 Emigrant Hill.40 0.60 1.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 0.00 600 ft 1.00 3.00 2.20 0.00 4.00 3.40 1.00 0.00 1.2.00 2.00 600 ft 1000 ft 0.3. lane 1 142 .00 5.00 Figure A.00 1.00 0.00 3000 ft 4000 ft 5000 ft 2.60 0.80 1.00 1000 ft 2000 ft 0.00 4.60 1.80 2.
80 2.00 0.00 1.60 1.00 3.00 1000 ft 2000 ft 0.40 1.40 0.00 3.80 2.80 1.00 0.20 1.60 0. lane 1 143 .80 1.00 5.20 0.60 0.00 0.00 3.00 600 ft 1.00 0.00 Figure A. CDF of UDL for Oregon US 97 Bend.20 0.40 1.00 Figure A.5.00 600 ft 1000 ft 0.00 3000 ft 4000 ft 5000 ft 2.00 1. CDF of UDL for Oregon OR 58 Lowell.00 2.4.00 5.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 1.00 4.00 1.20 1.00 4.60 1.00 2.00 2.00 4.5.40 0.00 3.00 4. lane 2 5.
00 0. lane 1 5.80 2.00 1000 ft 2000 ft 0.00 600 ft 1. CDF of UDL for Florida 9916.7.00 4.00 600 ft 1.00 1. lane 1 144 .00 5.00 0.40 1.00 1.00 2.00 5.80 2.6.20 1.00 4.80 1.00 1.20 0.00 3.00 3. CDF of UDL for Florida 9919.80 1.00 4.40 1.00 0.00 Figure A.00 Figure A.00 1.40 0.00 0.00 3000 ft 4000 ft 5000 ft 2.60 0.00 0.00 2.00 2.00 3.00 3000 ft 4000 ft 5000 ft 1000 ft 2000 ft 3.40 0.60 0.1st lane 5.60 1.60 1.00 4.20 0.20 1.
00 1.00 3.00 4.00 1000 ft 2000 ft 0.60 1.00 3.00 3000 ft 4000 ft 5000 ft 2.40 1.00 Figure A.00 5.00 1000 ft 2000 ft 0.20 0.00 0.20 1. lane 1 145 . CDF of UDL for Florida 9936. lane 1 5.00 3000 ft 4000 ft 5000 ft 2.20 0.80 2.00 4.00 0. CDF of UDL for Florida 9927.40 0.5.80 2.60 0.00 1.00 5.00 2.80 1.00 4.00 600 ft 1.00 2.80 1.00 4.00 0.00 3.00 0.60 1.40 1.9.60 0.00 Figure A.20 1.00 1.00 1.00 3.00 600 ft 1.40 0.8.
80 2.40 0.00 3.00 5.00 3000 ft 4000 ft 5000 ft 2.00 1.00 4.60 1. lane 1 5. lane 2 146 .00 600 ft 1.00 3000 ft 4000 ft 5000 ft 2.11.00 600 ft 1.00 1000 ft 2000 ft 0.5.00 0.00 3.20 0.40 1.20 1.80 1.00 0. CDF of UDL for Indiana 9534.40 1.80 1.00 Figure A.60 1.80 2.20 1.60 0.40 0.00 0.00 2.00 0.00 2.00 Figure A.00 3. CDF of UDL for Indiana 9534.00 1.00 4.60 0.00 4.00 1.00 5.20 0.10.00 1.00 3.00 1000 ft 2000 ft 0.00 4.
00 4.00 600 ft 1000 ft 2000 ft 0.40 1.00 0.00 2.40 0.80 1.00 0.00 2.5.00 1.12.80 2.00 5.80 2.60 1.20 0. lane 1 147 .00 3000 ft 4000 ft 5000 ft 1.00 3.20 0.60 1.13.40 1. CDF of UDL for Indiana 9544. CDF of UDL for Indiana 9534.00 4. lane 3 5.00 1000 ft 2000 ft 3000 ft 4000 ft 5000 ft 2.00 1.00 3.00 600 ft 1.00 2.00 3.00 0.00 Figure A.00 0.00 1.60 0.80 1.00 Figure A.00 5.00 3.20 1.60 0.00 4.40 0.00 0.20 1.00 4.00 1.
40 0.60 1.00 600 ft 1000 ft 0.20 0.60 0.00 1.5.40 1. lane 1 5.40 0.00 1.00 5.00 1.20 1.15.00 3000 ft 4000 ft 5000 ft 1.00 2. CDF of UDL for Indiana 9512.20 1.00 5.00 4.40 1.00 Figure A.00 0. CDF of UDL for New York 9121.80 2.00 600 ft 1000 ft 2000 ft 0.00 1.00 4.00 3.00 2.00 0. lane 1 148 .00 3.00 3.60 0.00 0.00 2.80 2.00 2.80 1.60 1.00 4.20 0.00 0.00 4.14.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 3.00 Figure A.80 1.
60 1.80 2.00 2.00 4.00 3.40 1.80 2.5.00 2.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 3. CDF of UDL for New York 9121.20 1.00 2.00 0.20 1.00 3.00 1.00 4.40 0.40 0.16.00 600 ft 1000 ft 0.00 1.00 600 ft 1000 ft 0.60 1.00 3.00 0. lane 4 5.00 4.60 0.17.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 Figure A.20 0.20 0. lane 1 149 .60 0.00 Figure A.00 5.40 1. CDF of UDL for New York 2680.00 1.00 4.00 5.00 2.00 1.80 1.00 0.80 1.00 0.
40 0.00 3.00 1.00 2000 ft 3000 ft 4000 ft 5000 ft 1.00 4.00 600 ft 1000 ft 0.40 1.20 1.00 5.18.60 1.00 0.00 4.80 2.80 1.00 1.00 1.00 3000 ft 4000 ft 5000 ft 1.5.00 2.00 0.00 Figure A.00 2.00 3.40 0.20 1. lane 4 5.80 2.00 3.60 0. CDF of UDL for New York 2680.40 1.60 0.00 4.20 0.00 4.00 0.20 0.80 1.00 0.60 1.19. CDF of UDL for New York I495 EB. lane 1 150 .00 600 ft 1000 ft 2000 ft 0.00 5.00 Figure A.00 3.00 2.00 2.00 1.
00 4.80 2. lane 2 5.00 3.80 2.00 600 ft 1000 ft 0. CDF of UDL for New York I495 EB.20 1.00 2000 ft 3000 ft 4000 ft 5000 ft 1.40 0.00 Figure A.20. CDF of UDL for New York I495 EB.00 2.80 1.21.00 5.00 0.00 3.40 1.00 5.00 0.60 1. lane 3 151 .00 2.00 4.40 0.00 3.00 0.00 1.00 0.00 2.60 0.20 1.00 600 ft 1000 ft 0.5.00 4.00 1.00 Figure A.00 3.20 0.00 1.00 2.40 1.00 2000 ft 3000 ft 4000 ft 5000 ft 1.60 1.80 1.00 4.60 0.00 1.20 0.
23.00 0.00 1.40 0.00 0.00 5.60 1.00 600 ft 1000 ft 2000 ft 0.00 4.00 3000 ft 4000 ft 5000 ft 1.00 600 ft 1.00 0.80 1.00 2.00 4.5.60 0.00 1000 ft 2000 ft 0.00 2.00 3. lane 2 152 .20 1.20 0.00 1.00 3.20 1.00 5.00 3. lane 1 5. CDF of UDL for New York I495 WB.20 0.00 4.00 Figure A.00 0.00 4.22.00 1.80 1.80 2.00 1.60 1. CDF of UDL for New York I495 WB.40 1.00 3.00 2.80 2.00 3000 ft 4000 ft 5000 ft 2.60 0.00 Figure A.40 1.40 0.
APPENDIX B CDF OF MAXIMUM DAILY UDL 4.60 1.0 2000 ft 3000 ft 1.80 1.40 1.00 1.80 2.0 Figure B. CDF of maximum daily UDL for Oregon I5 Woodburn.0 4000 ft 5000 ft extrpolated 0.0 0.20 1.1.40 0.20 0.60 0.0 3.0 0.00 1.0 3.0 600 ft 1000 ft 2. lane 1 153 .
0 4000 ft 5000 ft extrapolated 0.0 Figure B.0 2000 ft 3000 ft 4000 ft 1.00 1.0 600 ft 1000 ft 2.60 1.0 0. CDF of maximum daily UDL for Oregon OR 58 Lowell.00 1.0 3.00 1.0 2000 ft 3000 ft 1.60 0.80 2. lane 1 4.0 600 ft 1000 ft 2.40 0.0 5000 ft extrapolated 0.4.20 0.0 Figure B.40 1.20 0. CDF of maximum daily UDL for Oregon I84 Emigrant Hill.60 0.0 3.00 1.40 1.0 0.2.80 2.20 1.80 1.80 1.40 0.0 3.3. lane 1 154 .0 3.60 1.0 0.0 0.20 1.
00 1.0 3.0 1000 ft 2000 ft 3000 ft 1.50 2.0 0.0 3.80 1.0 0. lane 1 155 .80 2.0 Figure B.00 1.0 600 ft 2.0 2.60 1.60 0.0 2000 ft 3000 ft 0.4.0 3.0 3.40 0. lane 1 4.20 1.0 Figure B.40 1.5.50 1.00 1.0 4000 ft 5000 ft extrapolated 0.0 0.4. CDF of maximum daily UDL for Florida 9916.20 0. CDF of maximum daily UDL for Oregon US 97 Bend.00 1.0 600 ft 1000 ft 1.0 0.00 4000 ft 5000 ft extrapolated 2.
0 0.0 4000 ft 5000 ft extrapolated 0.0 2000 ft 3000 ft 0.00 1.20 0.40 0.80 4000 ft 5000 ft extrapolated 2.0 0.0 600 ft 1000 ft 2.7.40 1. CDF of maximum daily UDL for Florida 9919.0 3.0 3.0 Figure B.0 Figure B.60 0.20 0.00 1.0 2000 ft 3000 ft 1. CDF of maximum daily UDL for Florida 9927.4. lane 1 156 .80 1.80 2.00 1.0 600 ft 1000 ft 1.60 1. lane 1 4.40 1.60 0.0 2.60 1.6.20 1.0 0.80 1.20 1.40 0.00 1.0 3.0 3.0 0.
60 0.80 1.0 3.4.0 3.60 1.0 600 ft 1000 ft 1.0 4000 ft 5000 ft extrapolated 0.40 1.80 1.60 1.0 3. lane 1 4.20 1.0 0.60 0. lane 1 157 .00 1.0 0.0 0.0 2.20 0. CDF of maximum daily UDL for Florida 9936.9.8.00 1.40 1.20 0.20 1.00 1.80 4000 ft 5000 ft extrapolated 2.0 3.0 Figure B.00 1. CDF of maximum daily UDL for Indiana 9534.40 0.40 0.0 1000 ft 2000 ft 3000 ft 1.80 2.0 0.0 600 ft 2.0 2000 ft 3000 ft 0.0 Figure B.
0 2000 ft 3000 ft 0.0 0.40 1.20 1.0 3.0 2.40 1. CDF of maximum daily UDL for Indiana 9512.80 1.0 3.60 0.0 600 ft 1000 ft 1.80 4000 ft 5000 ft extrapolated 2.40 0.0 Figure B.0 2000 ft 3000 ft 0.60 1.0 2.0 3.0 Figure B.10.0 0.60 1.20 1. lane 1 4.11.40 0.80 4000 ft 5000 ft extrapolation 2. CDF of maximum daily UDL for Indiana 9544.0 3.80 1.0 0.00 1.4.60 0.20 0.00 1.00 1.0 0.20 0. lane 1 158 .00 1.0 600 ft 1000 ft 1.
0 4000 ft 5000 ft extrapolated 0.0 1000 ft 2000 ft 3000 ft 1.0 600 ft 1000 ft 2.0 600 ft 2.20 0.00 0. lane 1 4.40 1.20 2. lane 1 159 .60 1.80 1. CDF of maximum daily UDL for New York 9121.60 0.0 5000 ft extrapolated 0.13.40 1.0 2.20 1.0 3.80 2.00 1. CDF of maximum daily UDL for New York 2680.0 2.0 0.00 2.20 1.40 1.0 0.40 0.40 1.80 2.60 0.00 2.00 1.0 3.0 3.00 0.0 Figure B.0 2000 ft 3000 ft 4000 ft 1.20 2.60 1.0 Figure B.12.80 1.4.40 0.0 3.20 0.
0 4000 ft 5000 ft extrapolated 0.00 4000 ft 5000 ft extrapolated 0.00 600 ft 1000 ft 2.00 2.80 2.40 0.00 0.40 0.00 0.14.20 0. lane 1 4.00 Figure B.0 3.0 2000 ft 3000 ft 1.60 1.20 1.00 2.40 1. lane 1 160 .00 0.80 2.0 600 ft 1000 ft 2.40 1.60 0.0 0.60 0.0 3.00 1. CDF of maximum daily UDL for New York I495 WB.00 2000 ft 3000 ft 1.4.80 1.00 2.20 0.15. CDF of maximum daily UDL for New York I495 EB.0 2.20 1.20 2.80 1.00 1.60 1.0 Figure B.00 3.40 1.00 3.20 1.
lane 1 161 .60 1.00 0.APPENDIX C CDF OF MAXIMUM WEEKLY UDL 4.0 2.20 0.0 Figure C.0 0.0 3000 ft 4000 ft 5000 ft 0.00 1.0 600 ft 1000 ft 2000 ft 1.40 1.80 1.1.20 1.80 extrapolated 1. CDF of maximum weekly UDL for Oregon I5 Woodburn.40 0.0 3.0 2.60 0.
40 0.80 1. CDF of maximum weekly UDL for Oregon OR 58 Lowell.0 0.00 1.0 3.0 2.80 1.60 0. lane 1 4.00 0.60 0.0 0.0 2.40 1.60 1.0 600 ft 2.20 0.60 1.80 1.20 0.20 1.4.0 5000 ft extrapolated 0.2.0 3.20 1.00 0.0 2000 ft 3000 ft 4000 ft 1. lane 1 162 .0 Figure C.40 1.0 1000 ft 2000 ft 3000 ft 4000 ft 1.3.40 0.0 5000 ft extrapolated 0.80 1.0 600 ft 1000 ft 2. CDF of maximum weekly UDL for Oregon I84 Emigrant Hill.00 1.0 Figure C.
4.4.0 2000 ft 3000 ft 0.40 1.0 3.0 0.00 1.0 0.0 3.60 1.0 2.0 extrapolated 0.80 1.5.60 1.20 0. lane 1 4.0 2000 ft 3000 ft 4000 ft 5000 ft 1.20 1.0 600 ft 1000 ft 1.40 0.40 1.0 Figure C.20 0.60 0.0 Figure C.0 2. CDF of maximum weekly UDL for Florida 9916. lane 1 163 .80 1.0 3.00 1.80 4000 ft 5000 ft extrapolated 2. CDF of maximum weekly UDL for Oregon US 97 Bend.60 0.20 1.0 0.0 600 ft 1000 ft 2.40 0.00 0.80 1.00 1.
0 2. lane 1 4.0 3.0 600 ft 1000 ft 2000 ft 1. CDF of maximum weekly UDL for Florida 9919.00 1. CDF of maximum weekly UDL for Florida 9927.0 3.60 0.00 0.80 1.20 1.20 0.0 3000 ft 4000 ft 5000 ft 0.60 1.60 0.0 0.0 3.40 0.20 1.0 Figure C.7.0 0.0 1000 ft 2000 ft 3000 ft 1.40 1.4.80 2.0 600 ft 2.60 1.80 extrapolated 1. lane 1 164 .00 1.0 0.00 1.40 1.0 Figure C.40 0.20 0.0 4000 ft 5000 ft extrapolated 0.6.0 2.80 1.
4.00 1.20 1.0 2000 ft 3000 ft 0.80 1.CDF of maximum weekly UDL for New York 9121.0 600 ft 1000 ft 1. CDF of maximum weekly UDL for Florida 9936.20 0.40 1.00 0.80 1.8.60 1.40 1.80 extrapolated 1.0 4000 ft 5000 ft extrapolatred 2.00 0.0 0.40 0.0 3. lane 1 165 .20 2.0 2.0 2.0 Figure C.0 0.0 2.20 0.00 1.20 1.0 3.0 3.40 0.00 2.40 1.60 0.60 1.80 2.0 Figure C.0 3000 ft 4000 ft 5000 ft 0.9.60 0.0 600 ft 1000 ft 2000 ft 1. lane 1 4.
60 0.80 1.10.0 4000 ft 5000 ft extrapolated 0.40 1.0 0. CDF of maximum weekly UDL for New York 2680.20 1.0 Figure C.0 1000 ft 2000 ft 3000 ft 1.80 2.0 3. lane 1 166 .4.0 3.00 0.0 2.20 0.40 1.20 2.40 0.60 1.00 2.00 1.0 600 ft 2.