Sunteți pe pagina 1din 8

Statistical Distribution of Bridge Resistance Using

Updated Material Parameters


Sarah L. Orton, Ph.D., M.ASCE1; Oh-Sung Kwon, Ph.D., M.ASCE2; and Timothy Hazlett3

Abstract: Resistance (load-carrying capacity) of a bridge girder is a random variable and can be determined by considering the uncertainty in
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

material, fabrication, and professional/analysis properties. Previous calibrations of load and resistance factor design (LRFD) determined the
distribution of bridge resistance on the basis of data from more than 30 years ago. This study uses the latest Material properties available in,
the literature to update the resistance distribution. The statistical distribution of the resistance was determined through Monte Carlo
simulation. The results of the analysis show an increase in bias and a decrease in the coefficient of variation (COV) for all types of bridges
in comparison with those used in previous calibration studies. The changes in bias and COV are the result of higher bias and lower
COV in material properties owing to better quality control in concrete and steel manufacturing. Steel and concrete bridges saw the greatest
change in resistance distribution. Prestressed bridges saw little change because the material properties of prestressing steel, which is the most
sensitive parameter in the prestressed bridges, did not change significantly since the previous calibration study. With these resistance dis-
tributions, it is expected that the calibration of the load factor in the AASHTO specification will lead to a lower live load factor, thereby
possibly reducing the material cost of the bridge. In addition, the ratio of actual to required (design) resistances of representative bridges in
Missouri was determined. The analysis showed that almost all representative bridges had a capacity-to-demand ratio greater than 1 according
to current AASHTO standards. DOI: 10.1061/(ASCE)BE.1943-5592.0000278. © 2012 American Society of Civil Engineers.
CE Database subject headings: Bridges; Prestressed concrete; Reinforced concrete; Steel; Reliability; Load and resistance factor
design; Material properties; Statistics.
Author keywords: Bridges; Prestressed concrete; Reinforced concrete; Steel; Reliability; LRFD; Material properties.

Introduction Previous projects have determined the distribution of the loads


and resistance and calibrated factors for a uniform probability of
Load and resistance factors in load and resistance factor design failure. The National Cooperative Highway Research Progam
(LRFD) are determined to produce a certain reliability, which (NCHRP) Project 12–33 produced NCHRP Rep. 368 (Nowak
can be related to the probability of failure pf . Particularly, LRFD 1999), which is the basis of the current LRFD specification. Before
evaluates the difference between required capacity from the loads NCHRP Rep. 368, several studies were undertaken to determine the
Q, and the available capacity from the resistance R. Both loads and distribution of the resistance of bridge elements (Tabsh and Nowak
resistance are random variables that can be described with proba- 1991; Nowak et al. 1994; Tabsh 1996) using the material, fabrica-
bility density functions, as shown in Fig. 1. The probability content tion, and professional factors reported in Ellingwood et al. (1980).
of the region where the load is greater than the resistance gives the The factors reported in Ellingwood et al. came from earlier studies
on the variations in material and fabrication parameters (Mirza et al.
probability of failure. To calibrate load and resistance factors, it is
1979; Mirza and MacGregor 1979; Galambos and Ravindra 1978).
necessary to have an accurate prediction of the load-carrying capac-
Therefore, the resistance uncertainty data used in the LRFD cali-
ity (resistance) of structural elements. The resistance is a random
bration is more than 30 years old. The Final Rep. Project No.
variable owing to uncertainties in material properties, dimensions, NCHRP 20-7/186 (Kulicki et al. 2007) revised the load factors
and methods of analysis. Accurate determination of the uncertain- on the basis of new truck weight information, but the same data
ties in each of these properties is essential for accurate determina- for the resistance was used. The Michigan Department of Transpo-
tion of the distribution of resistance and, hence, the probability of tation Rep. RC-1466 (Van de Lidt et al. 2005) also calculated new
failure. load factors on the basis of heavier truck weights in Michigan, but
used the same source for the resistance uncertainty. Although these
1
Assistant Professor, Univ. of Missouri Columbia, E2503 Lafferre projects were thorough at the time, more recent data are now avail-
Hall, Columbia, MO 65211 (corresponding author). E-mail: ortons@ able to update uncertainties in material properties and produce a
missouri.edu more accurate determination of the distribution of the resistance.
2
Assistant Professor, Univ. of Toronto, 242A 35 St. George St, Toronto, This research is part of an overall project sponsored by the Mis-
ON M5S 1A4. E-mail: kwon@utoronto.ca souri Department of Transportation (MoDOT) to calibrate the live
3
Graduate Research Assistant, Univ. of Missouri Columbia, E2509 load factor for the Strength I limit state in the AASHTO LRFD bridge
Lafferre Hall, Columbia, MO 65211. E-mail: tahtgb@mizzou.edu
design specifications (AASHTO 2008) considering the typical
Note. This manuscript was submitted on August 25, 2010; approved on
June 3, 2011; published online on June 6, 2011. Discussion period open
bridge configurations in Missouri and state-specific traffic environ-
until October 1, 2012; separate discussions must be submitted for indivi- ment. The determination of the loads and calibration of the live load
dual papers. This paper is part of the Journal of Bridge Engineering, factor is presented in Kwon et al. (2010, 2011). The objective of this
Vol. 17, No. 3, May 1, 2012. ©ASCE, ISSN 1084-0702/2012/3-462– research is to find new resistance distributions using the latest
469/$25.00. information available for the material and fabrication uncertainties.

462 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012

J. Bridge Eng., 2012, 17(3): 462-469


ratio is greater than 1, as expected; however, for some girders
the value is slightly less than 1, with the worst case being 0.85.
This is likely because the 2008 version of the AASHTO code
(and associated loads) was used rather than the actual version
the bridge was designed under or because of improper design of
that bridge. On average, the reinforced concrete bridge girders have
an actual/required resistance value of 1.18, indicating that their
strength is relatively close to the required 2008 factored load par-
ticularly if the resistance factors are taken into account. Fig. 3
shows the actual/required moment resistance values for prestressed
bridge girders. The negative moment capacity was calculated over
Fig. 1. Probability density functions for load and resistance the support. The positive moment capacity was calculated at the
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

center of each span. The average actual/required resistance for neg-


ative moment is 1.42 and for positive moment is 1.33. All of the
In addition, the research determines the ratio of required/actual re- ratios are greater than (or close to) 1 and generally less than 2. This
sistance for representative bridge girders to give an indication of an indicates a fairly consistent prestressed girder design and a level of
additional factor of safety possessed by most bridges. over strength. For composite steel bridge girders (Fig. 4), the aver-
age ratios are 1.56 and 2.54 for negative and positive moments,
respectively. These ratios are greater than those for the reinforced
Comparison of Required and Actual Strengths
concrete or prestressed bridges, likely owing to other factors such
An analysis was conducted to gauge the over strength (strength as fatigue or deflection controlling the design of the bridge. Fig. 5
greater than that required for design) of bridge girders according shows the relationship between span length and the actual/required
to the Strength I limit state. For this analysis, all bridges were moment capacity. As shown in Fig. 5, the bridges with the greatest
viewed through the 2008 version of the AASHTO code (AASHTO ratios have the shortest spans. This is because those spans are often
2008). This assumes that the 2008 version of the AASHTO code is the short spans in a multiple-span bridge. For example, one three-
the most accurate way to predict both the actual loads and resistan- span bridge with lengths of 18.1, 36.2, and 16.9 m had ratios of
ces of the bridge. Although this over strength cannot be used in the 5.85, 1.67, and 6.67, respectively. Therefore, it is likely that the
calibration of load and resistance factors, it can give an accurate
indication of the required/actual resistance under the Strength I
limit state. It is expected that almost all bridges will be over
designed (owing due to constraints of a different code, other limit
states, or design considerations controlling the design).
To conduct the research, representative bridges were selected on
the basis of statistical analyses of the National Bridge Inventory
(NBI) database. A total of 100 representative bridges in Missouri
were selected considering randomness in number of spans, span
lengths, and number of lanes. The actual strength of each sample
bridge was determined from provided design drawings and nominal
material properties. The required strength of the bridge girder was
also determined using the using Strength 1 limit state loads. For the
analysis, resistance factors were not used in calculating the actual
strength of the bridge girder. Load factors were used in calculating
the required resistance.
Fig. 3. Comparison of actual/required strength on the basis of bridge
Fig. 2 shows the ratios of actual-to-required moment resistance
build year for prestressed bridge girders
for the reinforced concrete bridges. In most cases, the resistance

Fig. 2. Comparison of actual/required strength on the basis of bridge Fig. 4. Comparison of required/actual strength on the basis of bridge
build year for reinforced concrete slab bridges build year for steel bridge girders

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012 / 463

J. Bridge Eng., 2012, 17(3): 462-469


subsequent calibrations are from data from the 1970s and
1980s. These parameters come from Ellingwood et al. (1980),
Mirza et al. (1979), and Mirza and MacGregor (1979) and are
shown in Table 1 in the previous column. For the concrete strength,
0:85f ’c was used as the maximum compressive strength allowed in
the American Concrete Institute (ACI) code (Ellingwood et al.
1980). All distributions were treated as normal variables, with
the exception of the yield strength of the steel, which was treated
as lognormal.
For this research, the material and geometric parameters were
updated using the latest published research. Nowak and Szerszen
Fig. 5. Comparison of required/actual strength on the basis of span
(2003) found material parameters for reinforced and prestressed
length for steel bridge girders
concrete buildings. Although the report was meant for buildings,
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

the material parameters for the concrete and steel should be


applicable to bridges. For concrete compressive strength, Nowak
design of the shorter spans was influenced by the longer span (web and Szerszen (2003) give the bias on the basis of concrete compres-
depth generally remains constant throughout the bridge). sive cylinder strength as

λ ¼ 0:0081f 03 02 0
c þ 0:1509f c  0:9338f c þ 3:0649 ð3Þ
Material and Geometric Parameters

The statistical distribution of resistance considers the uncertainties where f 0c is in ksi. This relationship indicates a decreasing bias with
in the materials (e.g., strength modulus), fabrication (e.g., geometry increasing concrete strength. Bias values for certain strengths are
dimensions), and professional/analysis (accuracy of analysis equa- shown in Table 1. These bias values do not include the 0.85 factor
tions) parameters. Therefore, the resistance R is considered the on f ’c because the 0.85 factor is accounted for in the calculations.
product of the nominal resistance Rn and three factors: materials To relate the compressive cylinder strength to in-place strengths,
M, fabrication F, and professional/analysis P (Nowak et al. 1994) Bartlett and MacGregor (1996) give the ratio of concrete compres-
sive strength measured in-place to that measured from cylinder test
R ¼ Rn MFP ð1Þ as 0.95 for elements less than 431 mm in depth (generally slabs)
and 1.03 for elements greater than 431 mm in depth (generally
For small coefficients of variation (COVs), the COV can also be beams). Bartlett and MacGregor further say that the concrete com-
found as the square root sum of the squares pressive strength increased by approximately 25% from 28-days to
1 year. However, for this study, only the 28-day compressive
COVR ¼ ðCOV2M þ COV2F þ COV2p Þ2
1
ð2Þ strength is used. The updated material parameters used for rein-
forced and prestressed bridges are shown in Table 1.
The statistical distribution of each parameter can be described in Compared with the previous parameters, the most significant
terms of a bias factor λ and the COV. The bias factor (referred to as changes for reinforced and prestressed bridges were that the bias
“ bias” later in this paper) is the ratio of the mean to the nominal for concrete and steel increased and the COV decreased. This is
design value. For example, the design value of a concrete mix may as expected because improvements have been made in the batching
be specified to be 27 MPa; however, the concrete that is actually of concrete and manufacture of steel. Furthermore, the bias and
delivered is a bit stronger, for example 31 MPa. Therefore, the con- COV for prestressing tendons have remained relatively unchanged.
crete strength would have a bias of 1.125. The COV is the ratio of A study done on steel by Schmidt and Bartlett (2002) found
the standard deviation to the mean. It gives an indication of the updated material parameters for steel girders. In general, the size
uncertainty of the parameter. of the plate (thickness and height) is close to deterministic, with a
To determine the distribution of resistance, the uncertainties (or bias close to 1 and a low COV. The yield strength of the steel de-
bias and COV values) for the material, fabrication, and professional pends on the thickness of the plate. Although it would be assumed
parameters must be defined. Previous material and geometric that thinner plates should have a more bias, that is not the case.
parameters used for the NCHRP Rep. 368 (Nowak 1999) and Schmidt and Bartlett found that each plate thickness range uses

Table 1. Material Parameters for Reinforced Concrete and Prestressed Bridges


Previous Updated
Variable Nominal Bias Coefficient of variation Bias Coefficient of variation
Concrete strength (f 0 c) 20.7 MPa 0.92 0.18 1.4029 0.1
27.6 MPa — — 1.2257 0.1
34.5 MPa 0.805 0.15 1.1559 0.1
41.4 MPa — — 1.1449 0.1
Ratio of in-place strength to cylinder strength < 431 mm — — 0.95 —
> 431 mm — — 1.03 —
Yield stress of steel (f y ) 275 MPa 1.125 0.12 1.145 0.05
414 MPa 1.12 0.1 1.145 0.05
Strength of prestressing steel (f pu ) 1,861 MPa 1.04 0.025 1.045 0.025
Modulus of elasticity 200 GPa 1 0.06 1 0.06

464 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012

J. Bridge Eng., 2012, 17(3): 462-469


a slightly different composition of steel; the differences in the com- Table 3. Fabrication Parameters for Reinforced and Prestressed Bridges
position lead to greater bias values in the thicker plates. The pre- Variable Bias Coefficient of variation
vious parameters used in the calibration by Kulicki et al. (2007) for
the yield strength of the steel were a bias of 1.05 and a COVof 0.10. Area of prestressing steel 1 0.0125
As shown in Table 2, the new bias can be as great as 1.2. Therefore, Area of reinforcing bars 1 0.015
the yield strength of the steel has increased in bias by as much as Effective depth 1 0:7∕day
14% and lowered in COV. Section height 1 0:4∕h
For the fabrication (geometry) parameters for reinforced and Slab thickness 1 0:4∕t
prestressed concrete and for the professional factors, no updated Note: h = section height; t = slab depth.
information was found. Thus, in this study, the same parameters
used in NCHRP Rep. 368 (Nowak 1999) are used, as shown in
Tables 3 and 4.
Table 4. Professional Parameters
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

Item Bias Coefficient of varition


Statistical Distribution of Resistance
Reinforced concrete moment 1.02 0.06
To determine the statistical distribution of the moment Prestressed moment 1.01 0.06
resistance, plans from 100 representative bridges (14 reinforced Steel moment 1.05 0.06
concrete slab, 58 prestressed girder, 28 steel girder) in Missouri
were analyzed, and the strength of the bridge girders was deter-
mined using the 2008 version of the AASHTO code (AASHTO
Table 5. Comparison of Test Case Using Previous Material Parameters
2008). The updated variations in the material and geometric param-
eters are then included by a Monte Carlo analysis to determine the Nowak et al. 1994 Crystal Ball
overall statistical distribution of resistance for each bridge. The ef- Coefficient of Coefficient
fect of professional/analysis uncertainty is included by using Beam Section Bias variation Bias of varition
Eqs. (1) and (2) after the Monte Carlo analysis.
Reinforced A 1.115 0.120 1.100 0.098
For this task, the program Crystal Ball (Oracle Crystal Ball
Fusion) was used in conjunction with Excel to perform a Monte concrete B 1.117 0.119 1.099 0.096
Carlo analysis. Crystal Ball is a spreadsheet-based application suite C 1.118 0.119 1.101 0.096
for predictive modeling, forecasting, simulation, and optimization. Prestressed II 1.036 0.039 1.026 0.046
Crystal Ball performs a Monte Carlo analysis by randomly III 1.037 0.035 1.022 0.047
assigning numerical values to selected cells following a defined IV 1.037 0.035 1.018 0.047
statistical distribution. Using Crystal Ball, the material and fabri-
cation properties in the Excel spreadsheets used to calculate the
bridges. The increase in COV is owing to the reduction in depth of
actual strengths of the bridge girders were given statistical distri-
the bridge slab and greater relative influence of the uncertainty of
butions, as shown in Tables 1–3. The program then performed a
Monte Carlo analysis while keeping track of the forecast value, reinforcing bar placement. The average bias and COV for positive
the moment resistance of the girder. The resistances were then plot- and negative moment in a reinforced concrete bridge are 1.15 and
ted and fitted to a log-normal distribution to obtain the bias and 0.067. The materials and fabrication factors can then be combined
COV for the resistance distribution. A log-normal curve best fit with the professional factors (Table 4) to give the distribution of
the data on the basis of a goodness of fit test and was also used resistance for the bridge. The combined resistance (Table 6) bias
in the original NCHRP Rep. 368 (Nowak 1999) calibration. To and COV for reinforced concrete bridges are 1.17 and 0.09, which
check the accuracy of the procedure, the results of the analysis us- are a slightly greater bias and a lower COV than the previous
ing the previous material parameters are compared with those pre- NCHRP report (Nowak 1999). The change is because the material
sented in Nowak et al. (1994) for a series of different reinforced parameters, particularly the steel strength, have improved (reduced
concrete (A, B, C) and prestressed (II, III, IV) girder sizes and show variability). A graphical representation of the change in the bias and
a fairly close agreement on the bias and COV, as shown in Table 5. COV factors in Fig. 7 shows that the distribution resistance has
The results for bias and COV accounting for the materials and narrowed and moved to the right.
fabrication uncertainty in reinforced concrete slab bridges are For prestressed bridge girders (Fig. 8, Table 7), the average bias
shown in Fig. 6. The bias falls in a fairly tight range between and COV for positive moment are 1.044 and 0.033. For negative
1.16 and 1.14 and does not seem to depend on the span length moment, the average bias and COV are 1.151 and 0.057, which are
of the bridge. For the COV, slight increase occurs for shorter span close to those of reinforced concrete bridges because most

Table 2. Updated Material and Fabrication Parameters for Plate Steel


Previous Updated
Variables Bias COV Variables Bias COV
Plate thickness 1 0.05 Plate thickness 1.04 0.025
Plate height 1 0.05 Plate height 0.999 0.002
Yield stress—for flange 1.05 0.1 Yield stress—0 to 30 mm 1.11 0.053
Yield stress—for web 1.1 0.11 Yield stress—30 to 40 mm 1.16 0.063
Yield stress— larger than 40 mm 1.2 0.055
Young’s modulus 1 0.06 Young’s modulus 1 0.06

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012 / 465

J. Bridge Eng., 2012, 17(3): 462-469


For composite steel bridge girders (Fig. 9, Table 8), the average
bias and COV for positive moment are 1.17 and 0.057. For negative
moment, the average bias and COV are 1.18 and 0.053. Including
the professional parameter, the combined resistance bias and COV
for steel bridges are 1.23 and 0.081; the bias is much greater than
the previous NCHRP report (Nowak 1999). The difference between
the two distributions is shown in Fig. 7. The substantial difference
is because the material parameter for the yield strength of the
steel changed from 1.05 or 1.1 to 1.11, 1.16, or 1.2 (depending
on plate thickness). Because the overall resistance parameter is
very sensitive to the material property of yield strength, the overall
resistance parameter also increased significantly.
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

Sensitivity Study

In an effort to determine the effect of changes in material param-


eters and which parameters have the greatest influence on the bias
and COV factors for the bridges, a sensitivity study was conducted.
The study changed each bias shown in Tables 2–4 by 50% and
then ran the Monte Carlo simulation to determine the resulting
change in the material/fabrication bias of the bridge. For the study,
only one bridge girder with the average span length for that
Fig. 6. Bias and coefficient of variation for moment in reinforced bridge type was used.
concrete slab bridges The results for the reinforced concrete slab bridges are shown in
Fig. 10. The most sensitive parameters are the slab thickness (depth
to reinforcement + cover), steel area, and yield strength. Because
these parameters changed only slightly (yield strength of steel bias
prestressed bridges in Missouri are made continuous by a rein- increased from 1.12 to 1.145), the statistical parameters of the re-
sistance did not change greatly when the updated material param-
forced concrete section at the supports. The resistance bias and
eters were used. Furthermore, the material parameter with the
COV for prestressed concrete bridge girders, including the profes-
greatest change f ’c does not greatly influence the material/fabrica-
sional parameter in positive moment, are 1.055 and 0.068, which tion bias of the bridge. The results of the prestressed bridge girders
are nearly the same as the previous NCHRP report (Nowak 1999). (Fig. 11), show a similar trend. The material/fabrication bias is not
The similarity between the previous and the updated distributions is sensitive to the strength of the concrete or rebar in the slab. How-
also shown in Fig. 7. The lack of significant change is because the ever, it is very sensitive to the area and strength of the prestressing
material parameters, particularly the prestressing steel strength, did tendons and effective depth. Because these updated material prop-
not change significantly, and therefore, the bias and COV remained erties for prestressing steel had little change, little change occurred
similar. in the statistical parameters for resistance of prestressed bridges.

Table 6. Statistical Parameters for Moment Resistance of Reinforced Concrete Bridges


Material/fabrication Professional Resistance
Bias Coefficient of variation Bias Coefficient of variation Bias Coefficient of variation
Previous 1.12 0.12 1.02 0.06 1.14 0.13
Updated 1.15 0.074 1.02 0.06 1.17 0.09

Table 7. Statistical Parameters for Moment Resistance of Prestressed Girders


Material/fabrication Professional Resistance
Bias Coefficient of variation Bias Coefficient of variation Bias Coefficient of variation
Previous 1.04 0.045 1.01 0.06 1.05 0.075
Updated 1.045 0.033 1.01 0.06 1.055 0.069

Table 8. Statistical Parameters for Moment Resistance of Steel Girders


Material/fabrication Professional Resistance
Bias Coefficient of variation Bias Coefficient of variation Bias Coefficient of variation
Previous 1.07 0.08 1.05 0.06 1.12 0.10
Current 1.18 0.055 1.05 0.06 1.23 0.081

466 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012

J. Bridge Eng., 2012, 17(3): 462-469


Fig. 7. Previous and updated distribution of moment resistance for reinforced concrete slab bridges
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Bias and coefficient of variation for moment in steel bridge


Fig. 8. Bias and coefficient of variation for moment in prestressed girders
bridge girders

For steel bridge girders (Fig. 12), the plate dimensions and yield
strength play a significant role in determining the material/fabrica-
tion bias. Because the plate yield strength changed significantly
using the updated parameters (from 1.05 to 1.16 in some cases),
the change in the material/fabrication bias was also significant.
Properties of the composite concrete deck do not have a significant
effect on the material/fabrication bias.
The sensitivity study also highlights which improvements to
standards specifying accepting material tolerances would have
the greatest effect. For example, specifications reducing the toler-
ance on concrete strength would not have a significant effect in the
overall strength of the bridge; however, specifications that reduce
the tolerance on reinforcing bar placement would have a greater Fig. 10. Sensitivity parameters for reinforced concrete bridges
effect.

were calculated. All parameters other than the resistance factors


Effect of Updated Resistance Parameters on Bridge were identical in both cases. Details on bridge selection and reli-
Reliability ability analysis procedures can be found in Kwon et al. (2010,
2011). The analysis showed that for steel bridges and reinforced
To understand how the updated resistance factors affect reliability concrete bridges, the significant changes in the newly calculated
of bridges, reliability indices of representative bridges in Missouri bias and COVs for resistance increased the reliability index by

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012 / 467

J. Bridge Eng., 2012, 17(3): 462-469


ratios indicated that according to the Strength I limit state, steel
bridges are the most over designed. This is likely because the de-
sign of steel bridges may not be controlled by the Strength I
limit state.
The distribution of resistance was determined on the basis of
material parameters for resistance updated using the latest research
available. The resistance distribution calculation was achieved
through Monte Carlo simulation. The results show an increase
in bias and a decrease in COV for all types of bridges evaluated,
with the most drastic changes being for the concrete and steel
bridges. The bias of the reinforced concrete slab bridges changed
from 1.14 to 1.17, and the COV changed from 0.13 to 0.09. The
Fig. 11. Sensitivity parameters for prestressed concrete bridges improved bias and COV were the result of improved concrete com-
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

pressive strengths and steel yield strengths. The bias of the pre-
stressed bridge girders changed only from 1.05 to 1.055, and
the COV changed from 0.075 to 0.069. The prestressed bridges
had little change because their most sensitive parameter, strength
of prestressing steel, did not change with the updated material
parameters. For the steel bridge girders, the bias changed from
1.12 to 1.23, and the COV changed from 0.10 to 0.081. The drastic
improvement in the bias for the steel bridges was owing to an in-
crease in the bias for the yield strength of steel.
With these new biases and COVs, the reliability increased as
much as 24% for the steel bridges. The improved reliability can
lead to a new lower live load factor for bridges, possibly reducing
the material cost of a bridge.

Fig. 12. Sensitivity parameters for steel bridges


Acknowledgments
approximately 24 and 15%, respectively. For prestressed concrete This research was supported through MoDOT and the National
bridges, the newly calculated bias and COVs for resistance in- University Transportation Center (NUTC). Undergraduate students
creased the reliability index only approximately 3%. The small (Issac Hop, Ashley Schmidt, Greg Neihus, and Joe Kirby), who
change in the reliability index in prestressed concrete bridges is helped to process the data on the many bridges were greatly
the result of the small change in the bias and COV factors for appreciated.
the material parameters. Because the updated resistance factors
on the basis of the latest material parameters increase the reliability
of bridges with varying degree, it may be appropriate to recalibrate References
factors in bridge design specifications on the basis of the updated
resistance factors. For steel and concrete bridges, the significant AASHTO. (2008). LRFD bridge design specifications, 2008 interim revi-
increase in reliability owing the material parameters may allow sions, 4th Ed., Washington, DC.
for reduction of load factors to reduce the required design strength Bartlett, F., and MacGregor, J. (1996). “Statistical analysis of the compres-
sive strength of concrete in structures.” ACI Mater. J., 93(2), 158–164.
and cost of the bridge.
Ellingwood, B., Galambos, T., MacGregor, J., and Cornell, A. (1980).
“Development of a probability based load criterion for American Na-
tional Standard A58.” National Bureau of Standards, Washington, DC.
Conclusions
Galambos, T. V., and Ravindra, M. K. (1978). “Properties of steel for use in
LRFD.” J. Struct. Div., 104(ST9), 1459–1468.
Previous resistance distributions for bridges were calculated using
Kulicki, J. M., Prucz, A., Clancy, C. M., Mertz, D. R., and Nowak, A. S.
material data from the 1980s and earlier. This research was conducted (2007). “Updating the calibration report for AASHTO LRFD code.”
to update the distribution of resistance to coincide with the most cur- Final Rep. Project No. NCHRP 0-7/186, Transportation Research
rent material properties available. This research was done as part of an Board of the National Academies, Washington, DC.
overall project sponsored by MoDOT to calibrate the live load factor, Kwon, O., Kim, E., and Orton, S. (2011). “Calibration of the live load fac-
considering the typical bridge configurations in Missouri and state- tor in LRFD bridge design specifications based on state-specific traffic
specific traffic environment, for Strength I limit state in AASHTO environments.” J. Bridge Eng., 16(6), 812–819.
LRFD bridge design specifications (AASHTO 2008). Kwon, O., Orton, S., Kim, E., Hazlett, T., and Salim, H. (2010). “Calibration
Concrete, prestressed, and steel bridges representative of of live load factor in LRFD design guidelines.” Final Rep. TRyy0913,
Missouri were selected, and their actual and required resistances 〈http://library.modot.mo.gov/RDT/reports/TRyy0913/or11003revised
were determined. The analysis showed that almost all .pdf〉 (Mar. 19, 2012).
Mirza, S. A., Hatziuikolas, M., and MacGregor, J. G. (1979). “Statistical
representative bridge girders were generally over designed accord-
descriptions of the strength of concrete.” J. Struct. Div., 105(ST6),
ing to current AASHTO (2008) standards, indicating that the true 1021–1037.
probability of failure will be less than the two in 10,000 that the Mirza, S. A., and MacGregor, J. G. (1979). “Variations in dimensions of
calibration procedure uses. On average, concrete bridges had an reinforced concrete members.” J. Struct. Div., 105(ST4), 751–766.
actual/required strength ratio of 1.18, prestressed bridges had ratios Nowak, A. (1999). Calibration of LRFD bridge design code NCHRP Rep.
of 1.33 for positive moments, and composite steel bridges had 368, Transportation Research Board, National Research Council,
ratios of 1.56 for negative and 2.54 for positive moments. These Washington, DC.

468 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012

J. Bridge Eng., 2012, 17(3): 462-469


Nowak, A., and Szerszen, M. (2003). “Calibration of design code for build- Data collection.” Can. J. Civ. Eng., 29(1), 109–118.
ings (ACI 318). Part 1: Statistical models for resistance.” ACI Struct. J., Tabsh, S. (1996). “Reliability of composite steel bridge beams designed
100(3), 377–382. following AASHTO’s LFD and LRFD specifications.” Struct. Saf.,
Nowak, A., Yamani, A., and Tabsh, S. (1994). “Probabilistic models 17(4), 225–237.
for resistance of concrete bridge girders.” ACI Struct. J., 91(3), Tabsh, S., and Nowak, A. (1991). “Reliability of highway girder bridges.”
269–276. J. Struct. Eng., 117(8), 2372–2388.
Oracle. (2008). Oracle Crystal Ball Fusion Edition Release 11.1.1.1.00 Van de Lindt, J. W., Fu, G., Pablo, R., and Pei, S. (2005). “LRFD Load
[computer software]. Redwood City, CA. Calibration for State of Michigan truckline bridges.” Rep. RC-1466,
Schmidt, B., and Bartlett, F. (2002). “Review of resistance factor for steel: Michigan Dept. of Transportation, Lansing, MI.
Downloaded from ascelibrary.org by University of Central Florida on 12/07/20. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2012 / 469

J. Bridge Eng., 2012, 17(3): 462-469

S-ar putea să vă placă și