Sunteți pe pagina 1din 9

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 100-S42

Calibration of Design Code for Buildings (ACI 318): Part 2


Reliability Analysis and Resistance Factors
by Maria M. Szerszen and Andrzej S. Nowak

Calibration of the design code for concrete structures is presented from material tests performed in 2000 and 2001 (Nowak and
in two studies. The first one focused on the development of resistance Szerszen 2003). The quality of workmanship can have a strong
models. This paper deals with the reliability analysis and selection influence on materials, and it is assumed to be of average level.
of resistance factors. The structural types considered in this study
include beams, structural slabs, and columns. The analysis is
performed for reinforced concrete and prestressed concrete elements. CALIBRATION PROCEDURE
A wide range of materials is covered: ordinary concrete, high- The calibration procedure used for selection of resistance
strength concrete, lightweight concrete, reinforcing bars No. 3 factors is based on the structural reliability theory (Nowak and
through 11, and two grades of prestressing strands. The reliability Collins 2000). The calibration procedure includes five steps.
analysis requires the knowledge of the statistical parameters for First, the types of structural elements and materials covered
load and resistance. Resistance models are described in the first by ACI 318 Code are identified. It is important to select the
paper. In addition to the material parameters based on the new test representative dimensions and reinforcement ratios for
results, the parameters used in previous studies (1970s) were structural elements. Depending on the structural types, the
considered. Load models are based on the data available in literature. typical load component ratios D/(D + L) are determined,
Reliability indexes are calculated using various load combinations, where D = dead load effect and L = live load effect.
the basic combination of dead load and live load, and other
combinations with snow, wind, and earthquake for the two
The second step is to select the statistical models for load
resistance models considered (older database and new material components that are needed for reliability analysis. For each
database). The resulting values of the reliability index calculated time-varying load, two load components are considered: an
for the old data base and old load models, and potential consequences arbitrary-point-in-time value and the maximum lifetime value.
of failure, served as a basis for the selection of the target reli- The models include the mean value and coefficient of variation.
ability index. For each type of structural element and load combination It is assumed that the database for loads available in the
case, several possible values of the resistance factor were considered literature is sufficient for the purpose of this calibration. For
(rounded to the nearest 0.05). The recommended values provide a load combinations, Turkstras rule was applied (Turkstra 1970;
close fit to the target reliability levels. Turkstra and Madsen 1980; Nowak and Collins 2000).
The third step, selection of the resistance models, is one of
Keywords: analysis; load; resistance. the most important tasks in the calibration process. The
available database for material properties was reviewed. The
INTRODUCTION quality of materials (concrete and steel) has improved over the
This is the second of two papers presenting the calibration years and it is not reflected in the design formulas, particularly
of the Building Code Requirements for Structural Concrete in the resistance factors. Therefore, this step in the calibration
(ACI 318). The first study (Nowak and Szerszen 2003) focused process included the collection of new material test results
on the development of the statistical models for resistance and redevelopment of statistical models for resistance.
including strength of materials: concrete (ordinary concrete, The next step in the calibration process focused on the
high-strength concrete, and lightweight concrete), reinforcing reliability analysis procedures. Various available procedures are
steel, and prestressing steel. This paper deals with the reliability presented in the literature (for example, Nowak and Collins
analysis and selection of resistance (strength reduction) factors. 2000; Thoft-Christensen and Baker 1982; Madsen, Krenk,
The 2002 edition of the Building Code Requirements for and Lind 1986; and Melchers 1987). In this study, safety was
Structural Concrete (ACI 318-02) specifies loads and load measured in terms of the reliability index.
combinations consistent with ASCE 7, Standard on Minimum The developed procedure was applied to calculate the
Design Loads for Buildings and Other Structures (1998). reliability indexes for the considered structural types and
The objective of this calibration is to calculate a new set of materials, for various ratios of load components, designed
resistance factors (strength-reduction factors) based on a exactly according to ACI 318-99 (no under-design or over-
probabilistic analysis using the currently available statistical design). The analysis was performed for two sets of statistical
data, so that the reliability of designed elements is consistent models for resistance: 1) based on the material test data from
with a predetermined target level. 1970s and early 1980s; and 2) based on new material test data
The calibration procedure follows the basic steps as described provided by the industry and processed in conjunction
by Nowak and Collins (2000). Load and resistance parameters with this study.
are treated as random variables. Therefore, structural perfor-
mance is measured in terms of the reliability index (Nowak ACI Structural Journal, V. 100, No. 3, May-June 2003.
MS No. 02-177 received May 28, 2002, and reviewed under Institute publication
and Collins 2000). The statistical parameters of load are policies. Copyright 2003, American Concrete Institute. All rights reserved, including
based on the available literature, and for resistance, two the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors closure, if any, will be published in the March-
databases are considered: one from the 1970s, and the other April 2004 ACI Structural Journal if the discussion is received by November 1, 2003.

ACI Structural Journal/May-June 2003 383


In addition, wind load is multiplied by the directionality
ACI member Maria M. Szerszen is an assistant research scientist in the Department
of Civil and Environmental Engineering, University of Michigan, Ann Arbor, Mich. factor 0.85. The objective of this calibration is to determine
She is a member of ACI Committee 348, Structural Safety, and Joint ACI-ASCE the value of factors corresponding to Eq. (2). In addition,
Committee 343, Concrete Bridge Design. the values of load factors are also verified for consistency.
ACI member Andrzej S. Nowak is a professor in the Department of Civil and Envi- Some of the load factors specified in ACI 318-02 for the
ronmental Engineering, University of Michigan. He is a member of ACI Committees ultimate limit states and load combinations including live
318, Structural Concrete Building Code; 348, Structural Safety, and Joint ACI-ASCE
Committee 343, Concrete Bridge Design. load, snow, and wind are different than load factors in ASCE
7-98. In particular, the live load factor of 0.5 in ASCE 7-98
The next important step in the calibration process is the is changed to 1.0 in ACI 318-02. A live load factor of 0.5 is
selection of the target reliability index for all design cases, still permitted except for garages, areas of public assembly, and
including all of the considered structural types and materials. for a live load greater than 100 lb/ft2. The reliability analysis was
The reliability indexes corresponding to the current practice, performed for live load factor of 0.5, and an increased load factor
as calculated in the previous step, were used as important will result in higher reliability indexes. Therefore, the recom-
references in the selection process. The target reliability level mended resistance factors are acceptable for ACI 318-02.
depends on consequences of failure and cost of increasing/
decreasing the safety margin by a unit (marginal cost of safety). LOAD MODELS
The final step in this calibration is the calculation and The considered load components include dead load, live
selection of resistance factors for the considered limit states load, snow, wind, and earthquake. The statistical parameters
and the design cases corresponding to the load factors for load components are taken from the available literature
specified in ASCE 7-98 Standard. The acceptance criterion (Ellingwood et al. 1980; Ellingwood and Rosowsky 1996;
was closeness to the target reliability index. The resistance and Nowak 1999). For each load combination, the statistical
factors were rounded to the nearest 0.05. To check the parameters are determined using the so-called Turkstras
consistency of the results, reliability indexes were calculated rule (Turkstra 1970; Turkstra and Madsen 1980; and Nowak
using the proposed resistance factors and the results were and Collins 2000). Turkstra observed that the extreme value
compared with the target values. of load combinations corresponds to the occurrence of an
extreme value of only one load component, while all the other
DESIGN FORMULAS load components take the corresponding average (arbitrary-
The ultimate limit state load cases specified by ACI 318-99 point-in-time) values.
Code are: Therefore, for each load component, two sets of statistical
parameters are considered. The first corresponds to the
1.4D + 1.7L < R (1) maximum expected value of the load component during the
lifetime of the structure, and the other corresponds to the
0.75(1.4D + 1.7L +1.7W) < R average value (arbitrary-point-in-time value).
Dead load is the weight of structural and nonstructural
0.9D + 1.3W < R elements permanently connected to the structure. The bias
factor (ratio of mean-to-nominal) value of dead load is = 1.05,
0.75(1.4D + 1.7L + 1.87E) < R and coefficient of variation V = 0.10 for cast-in-place concrete,
and = 1.03, and coefficient of variation V = 0.08 for precast
where D = dead load; L = live load; W = wind, E = earthquake; concrete (Nowak and Collins 2000). Because dead load is
R = resistance (load-carrying capacity); and = resistance assumed to be time-invariant, only one set of parameters
factor (capacity-reduction factor). Snow S is not mentioned is needed.
in the Code, and it is implied that snow load is included in Live load is the weight of people, furniture, partitions, and
the live load L. other items. For the maximum 50-year live load, the bias factor
Resistance factors are specified depending on structural is = 1.0, and the coefficient of variation V varies depending
type, material, and limit state; = 0.90 for flexure, = 0.85 for on the influence area. V decreases with an increasing influence
shear, = 0.70 for tied columns, = 0.75 for spiral col- area. In this study, the calculations were performed for sev-
umns, and = 0.60 for plain concrete. eral influence areas, but the results are presented for 40 m2
The ultimate limit state load cases specified in the and V = 0.18. For average live load, = 0.24 and V = 0.65.
ASCE 7-98 Standard are Snow is specified in terms of ground snow, and the design
value is modified using several additional parameters. The
1.4D < R (2) parameters for the maximum snow are = 0.82 and V = 0.26.
For the average snow load, = 0.20 and V = 0.87.
1.2D + 1.6L < R Wind load is specified in terms of wind velocity at 10 m
above the ground level, and the design value of pressure due
1.2D + 1.6L + 0.5S < R to wind is modified using several additional parameters,
including gust factor and exposure factor. The parameters
1.2D + 0.5L + 1.6S < R for the maximum 50-year wind are = 0.78 and V = 0.37. It
was assumed that the average (arbitrary-point-in-time) wind
1.2D + 1.6W + 0.5L + 0.5S < R load is negligible, with = 0.
Earthquake load is specified in terms of ground acceleration,
1.2D + 1.0E + 0.5L + 0.2S < R and the design value is modified using several additional
parameters. The parameters for the maximum earthquake
0.9D (1.6W or 1.0E) < R are = 0.66 and V = 0.56. For the average earthquake, = 0.

384 ACI Structural Journal/May-June 2003


The basic load combination includes dead load and live structural slabs, and plain concrete. Three types of concrete
load. The reliability analysis is performed for a full range of are considered: ordinary, high-strength, and lightweight. A
load ratios that are expressed as D/(D + L), varying from 0 to wide range of reinforcing bar diameters and prestressing
1. The practical range of D/(D + L), however, is between 0.3 strands are included.
and 0.9. Because of the load duration and a negligible The statistical parameters of resistance were derived by
probability of a simultaneous occurrence, the combination Monte Carlo simulations, based on material test data and
of W and E is not considered. other parameters. The procedure is presented by Nowak and
For load combinations including S, W, and E, the analysis Szerszen (2003), and the results are summarized in Table 2.
is performed for a full range of D/(D + L) in addition to one The parameters shown in Table 2 are bias factor and coefficient
of the environmental loads. For combinations including of variation. For each design case considered, the mean value of
snow, the results are presented for S = 0.5D. For combina- resistance is calculated as a product of the nominal (design)
tions including wind, the results are presented for W = 0.5D. value and bias factor. The standard deviation is calculated as
For combinations including earthquake, the results are the product of the mean and coefficient of variation.
presented for E = 0.5D.
The load statistical parameters are summarized in Table 1. RELIABILITY ANALYSIS
The parameters shown in Table 1 are bias factor and coeffi- Load and resistance parameters are random variables;
cient of variation. For each design case considered, the mean therefore, it is convenient to measure the structural perfor-
value of load is calculated as a product of the nominal (design) mance in terms of the reliability index . Various procedures
value and bias factor. The standard deviation is calculated as for calculation of are presented by Nowak and Collins
the product of the mean and coefficient of variation. (2000). The general format of the limit state function g is

RESISTANCE MODELS g = RQ0 (3)


The reliability analysis is performed for reinforced concrete
and prestressed concrete components such as beams, columns, where g = safety margin; R = resistance; and Q = load effect.
In this study, Q is a combination of load components.
Table 1Statistical parameters for The reliability index can be considered as a function of
load combinations the probability of failure PF
Arbitrary-point-in- Maximum
time load 50-year load 1
= ( PF ) (4)
Load component Bias COV Bias COV
Dead load (cast-in-place) 1.05 0.10 1.05 0.10
Dead load (plant-cast) 1.03 0.08 1.03 0.08
where 1 = inverse standard normal distribution function.
The reliability analysis procedure used in this calibration
Live load 0.24 0.65 1.00 0.18
includes the following steps:
Snow 0.20 0.87 0.82 0.26
1. Prepare input data:
Wind 0.00 0.00 0.78 0.37 Structural type and limit state;
Earthquake 0.00 0.00 0.66 0.56 Nominal values of load components: D, L, S, W, and E; and
Note: COV = coefficient of variation. Load and resistance factors: D, L, S, W, E, and .

Table 2Statistical parameters of resistance R


Statistical parameters of resistance
Existing design (based on Proposed design (based on new material data)
material data from
Ellingwood et al. [1980]) Ordinary concrete High-strength concrete Lightweight concrete
Structural type and limit state Bias factor V Bias factor V Bias factor V Bias factor V
RC beam cast-in-place, flexure 1.114 0.119 1.190 0.089 1.160 0.090 1.180 0.090
RC beam plant-cast, flexure 1.128 0.133 1.205 0.081
RC beam cast-in-place, shear 1.159 0.120 1.230 0.109 1.190 0.110 1.230 0.110
RC beam plant-cast, shear 1.170 0.116 1.242 0.105
PS beam plant-cast, flexure 1.034 0.081 1.084 0.073
PS beam plant-cast, shear 1.130 0.105 1.194 0.103
RC slab cast-in-place 1.052 0.169 1.077 0.146 1.070 0.145 1.080 0.150
RC slab plant-cast 1.146 0.116 1.174 0.082
PS slab plant-cast 1.053 0.070 1.075 0.070
Post-tensioned slab cast-in-place 0.961 0.146 0.982 0.145 1.030 0.110
RC column cast-in-place, tied 1.107 0.136 1.260 0.107 1.200 0.120 1.260 0.130
RC column plant-cast, tied 1.102 0.134 1.252 0.103
RC column cast-in-place, spiral 1.163 0.124 1.316 0.097 1.260 0.110 1.330 0.120
RC column plant-cast, spiral 1.156 0.122 1.323 0.091
PS column plant-cast, tied 1.017 0.094 1.080 0.090
PS column plant-cast, spiral 1.068 0.076 1.133 0.071
Plain concrete, flexure, shear 1.004 0.082 1.105 0.082 1.240 0.080 1.400 0.080

ACI Structural Journal/May-June 2003 385


where mR = mean value of resistance; mQ = mean value of
the total load effect; R = standard deviation of resistance;
and Q = standard deviation of the total load effect.

RELIABILITY INDEXES FOR SELECTED


STRUCTURES AND MATERIALS
The reliability analysis is performed following the procedure
described previously for the considered structural types and
limit states. The resistance is determined using formulas in
Eq. (1) and (2) for ASCE 7-98 and ACI 318-99, respectively.
The former is denoted as proposed design, and the latter as
existing design. The statistical parameters used in the
computations are as described in this paper and in Nowak
Fig. 1Reliability indexes calculated for D + L load com- and Szerszen (2003). For the existing design, the analysis is
bination; reinforced beam made of ordinary concrete (flexure). carried out for old material data. For the proposed design,
new statistical parameters for materials are considered.
The analysis is performed for the following load com-
binations:

D (6)

D+L

D+L+S

D+L+W

D+L+E
Fig. 2Reliability indexes calculated for D + L load com-
bination; reinforced beam made of ordinary concrete (shear). D+L+S+W

D+L+S+E

Reliability indexes are calculated for each type of structural


component and material, and for the full range of the D/(D + L)
ratio, including the case of dead load without any live load
(L = 0). The reliability indexes are shown for the full range
of the D/(D + L) ratio. For existing design (ACI 318-99), the
old factors were used. For the proposed design (ASCE 7-98),
s were calculated for specified in ACI-318-99, and in
addition for ( 0.05) and ( + 0.05).
The average reliability indexes, determined for ordinary
concrete, high-strength concrete, and lightweight concrete,
are presented in Table 3. These average values are calculated
Fig. 3Reliability indexes calculated for D + L load combina- based on the range of reliability indexes selected depending
tion; reinforced slab made of ordinary concrete (flexure). on structural component and D/(D + L) ratio. The most probable
load ratios are selected as follows: for beams, D/(D + L) =
The load factors are available, but the factor is to be deter- from 0.3 to 0.7, for slabs, D/(D + L) = from 0.3 to 0.6, and
mined. There are a limited number of possible values for , for columns, D/(D + L) = from 0.4 to 0.9. Reliability indexes
however, (they are rounded to the nearest 0.05); therefore, calculated for existing design and old factors served as refer-
calculations are carried out for several possible values of . ences for the selection of the target reliability indexes. The
selection criterion for new factors was closeness to the
2. Calculate load parameters: the mean total load, the corre-
safety margin calculated for the existing design.
sponding coefficient of variation, and the standard deviation.
Examples of the reliability indexes plotted versus D/(D + L)
3. Calculate the nominal resistance for two cases: ASCE 7-98 for the basic load combination of D + L and for combinations
from Eq. (2), using new statistical data for materials (Nowak with environmental loads are shown in Fig. 1 to 20. The
and Szerszen 2002), and ACI 318-99 from Eq. (1), using old resulting s vary, depending on type of component, limit
statistical data for materials (based on Ellingwood et al. 1980). state, and load ratio.
4. Determine the statistical parameters of R (Table 2). Reliability indexes calculated for beams vary depending
5. Calculate the reliability index on the type of beam (reinforced concrete, prestressed concrete,
cast-in-place or plant-cast) and limit state (flexure or shear).
2 2 0.5 Most of the s are close to 4 or over. Values calculated
= ( m R m Q ) ( R + Q ) (5) for old materials data and existing design are slightly

386 ACI Structural Journal/May-June 2003


Table 3Average values of reliability indexes for load combination, D + L
Old material data, existing
design 1.4D + 1.7L New material data, proposed design, 1.4D or 1.2D + 1.6L
Resistance factor Ordinary concrete High-strength concrete Lightweight concrete
Structural type and limit state
0.95 3.83 3.63 3.75
RC beam cast-in-place, flexure 3.54 0.90 4.19 3.99 4.10
0.85 4.55 4.36 4.46
0.95 4.31
RC beam plant-cast, flexure 3.34 0.90 4.69
0.85 5.09
0.90 3.78 3.58 3.75
RC beam cast-in-place, shear 3.95 0.85 4.07 3.87 4.04
0.80 4.36 4.17 4.33
0.90 4.03
RC beam plant-cast, shear 4.18 0.85 4.33
0.80 4.63
0.95 3.75
PS beam plant-cast, flexure 4.34 0.90 4.20
0.85 4.65
0.90 3.88
PS beam plant-cast, shear 4.37 0.85 4.19
0.80 4.51
0.95 2.25 2.24 2.21
RC slab cast-in-place 2.45 0.90 2.48 2.47 2.44
0.85 2.72 2.71 2.67
0.95 4.13
RC slab plant-cast 3.84 0.90 4.51
0.85 4.90
0.95 3.82
PS slab plant-cast 4.90 0.90 4.27
0.85 4.73
0.95 1.85 2.54
Post-tensioned slab cast-in-place 2.41 0.90 2.09 2.85
0.85 2.35 3.17
0.75 4.68 4.03 3.94
RC column cast-in-place, tied 3.98 0.70 4.99 4.32 4.20
0.65 5.30 4.61 4.45
0.75 4.93
RC column plant-cast, tied 4.09 0.70 5.25
0.65 5.57
0.80 4.97 4.27 4.19
RC column cast-in-place, spiral 4.26 0.75 5.30 4.57 4.45
0.70 5.64 4.87 4.71
0.80 5.44
RC column plant-cast, spiral 4.38 0.75 5.78
0.70 6.13
0.75 4.67
PS column plant-cast, tied 5.21 0.70 5.09
0.65 5.51
0.80 5.45
PS column plant-cast, spiral 6.05 0.75 5.96
0.70 6.48
0.70 5.37 6.21 6.93
Plain concrete, flexure, shear 5.98 0.65 5.84 6.66 7.34
0.60 6.32 7.11 7.74

ACI Structural Journal/May-June 2003 387


Fig. 4Reliability indexes calculated for D + L load combina-
Fig. 8Reliability indexes calculated for D + L + S load
tion; reinforced column made of ordinary concrete (tied).
combination (snow is dominating load); reinforced slab made
of ordinary concrete (flexure).

Fig. 5Reliability indexes calculated for D + L load combina-


tion; reinforced column made of ordinary concrete (spiral).
Fig. 9Reliability indexes calculated for D + L + S load
combination (snow is dominating load); reinforced column
made of ordinary concrete (tied).

lower. Reliability indexes calculated for axially loaded


columns are higher then those for beams by approximately 10
to 20%.
It is observed that reliability indexes for slabs are lower than
for beams, and this applies both to existing design and proposed
design code. In slabs, there is a considerable uncertainty about
the actual effective depth, and the reliability index is very
sensitive to any departure from the specified value of the
depth. Effective depth in concrete slabs is usually very small
(compared with that of a beam), and even a small reduction can
drastically reduce the reliability index. The overall reliability
Fig. 6Reliability indexes calculated for D + L + S load of the slab, however, is considerably higher than the calculated
combination (snow is dominating load); reinforced column value due to load sharing. The reliability analysis is performed
made of ordinary concrete (flexure). for a 1 ft (or 1 m) wide segment of the slab. The slab, as a
structural system, can be considered as a parallel system of
interacting (load-sharing) segments. The reliability index of the
slab treated as a system is similar or larger than that of a beam.
Some of the reliability indexes for load combinations
including wind and earthquake are low. This is due to very
large coefficients of variation of W and E.

TARGET RELIABILITY INDEX


The optimum value of the target reliability index T can be
determined based on two parameters: consequences of failure,
and incremental cost of safety (Nowak and Collins 2000;
Madsen, Krenk, and Lind 1986; and Melchers 1987). In
general, the larger the expected cost of failure, the larger T
Fig. 7Reliability indexes calculated for D + L + S load is. However, T also depends on the financial considerations.
combination (snow is dominating load); reinforced beam made If additional reliability can be achieved at a low cost, then T
of ordinary concrete (shear). can be larger than an otherwise acceptable minimum level,

388 ACI Structural Journal/May-June 2003


Fig. 10Reliability indexes calculated for D + L + S load
combination (snow is dominating load); reinforced column Fig. 13Reliability indexes calculated for D + L + S + W
made of ordinary concrete (spiral). load combination (wind is dominating load); reinforced slab
made of ordinary concrete (flexure).

Fig. 11Reliability indexes calculated for D + L + S + W Fig. 14Reliability indexes calculated for D + L + S + W
load combination (wind is dominating load); reinforced beam load combination (wind is dominating load); reinforced column
made of ordinary concrete (flexure). made of ordinary concrete (tied).

Fig. 12Reliability indexes calculated for D + L + S + W Fig. 15Reliability indexes calculated for D + L + S + W
load combination (wind is dominating load); reinforced beam load combination (wind is dominating load); reinforced column
made of ordinary concrete (shear). made of ordinary concrete (spiral).

and if it is very costly to increase , then even T that is lower in this study, however. Because of insufficient input data in
than an otherwise required value can be acceptable. The this study, the target reliability levels were selected on the
selection of the optimum T, however, requires a considerable basis of the existing code (ACI 318-99).
database. The code provisions in ACI 318 apply to various The calculated reliability indexes, presented in Table 3 and
types of structural components, and it is assumed that T is to be Fig. 1 to 20, represent element reliability as opposed to the
selected for primary members (important components), and the system reliability. The relationship between the element
failure of a component can cause failure of other components. reliability index e and system reliability index s depends on
For secondary members, T can be reduced. There is practically type of the system (parallel, series, or mixed) and degree
no basis available to determine the consequences of failure of correlation between the elements. In general, statically
and cost of safety for the structural components considered determinate structures can be treated as series systems, and s

ACI Structural Journal/May-June 2003 389


Fig. 19Reliability indexes calculated for D + L + S load
Fig. 16Reliability indexes calculated for D + L + S load combination (live load is dominating load); reinforced column
combination (live load is dominating load); reinforced beam made of ordinary concrete (tied).
made of ordinary concrete (flexure).

Fig. 20Reliability indexes calculated for D + L + S load


Fig. 17Reliability indexes calculated for D + L + S load combination (live load is dominating load); reinforced column
combination (live load is dominating load); reinforced beam made of ordinary concrete (spiral).
made of ordinary concrete (shear).
of load ratio is an indication that the load factors are not
properly selected.
The new material data represents the material properties
determined in conjunction with this study, as also shown in
Table 3. For most cases, the new parameters are improved
compared to old data. The target reliability indexes selected
based on the old material data and ACI 318-99 Code design
formula are shown in Table 4. These values are conserva-
tively selected as upper rather than lower limits of the range
of obtained in calculations.
The target is 3.5 for most of the components except of
columns. Special consideration is required for slabs. In cast-
in-place slabs, there is a considerable degree of load sharing,
and the system reliability is much larger than for a segment of
Fig. 18Reliability indexes calculated for D + L + S load 1 ft (0.305 m). This justifies a reduced value of T = 2.5. For
combination (live load is dominating load); reinforced slab precast slab panels, the degree of load sharing can be similar
made of ordinary concrete (flexure). to beams.
Further consideration is required for columns. In this study,
< e. For parallel systems, s > e. The difference between only axially loaded columns were included. Table 4 presents the
range of reliability index based on selected D/(D + L) ratios
s and e depends on the coefficient of correlation ; and s
typical for particular types of structure, calculated for proposed
increases for increased and for series systems, and s new resistance factors. These values can be compared with
decreases for increased for parallel systems. target reliability indexes selected for considered structural
It is assumed that the reliability indexes for components types and design cases.
designed using the ACI-318-99 Code are acceptable. The
Code provisions have been used for over 30 years. Therefore, LOAD AND RESISTANCE FACTORS
the corresponding values of are calculated using the old The reliability indexes corresponding to various categories of
material data, as shown in Table 3. These reliability indexes structural types and materials were reviewed and compared to
are assumed to be a lower limit for acceptable values of T. the target values. Based on that analysis, the recommended
For each type of component, a large variation of s as a function values of the resistance factor are given in Table 5.

390 ACI Structural Journal/May-June 2003


Table 4Selected target reliability indexes Table 5Recommended resistance factors
Structural type and limit state Range of T Structural type and limit state Resistance factors
RC beam cast-in-place, flexure 3.4 to 3.6 3.5 RC beam cast-in-place, flexure 0.90
RC beam plant-cast, flexure 3.2 to 3.4 3.5 RC beam plant-cast, flexure 0.90
RC beam cast-in-place, shear 3.8 to 4.0 3.5 RC beam cast-in-place, shear 0.85
RC beam plant-cast, shear 4.1 to 4.2 3.5 RC beam plant-cast, shear 0.85
PS beam plant-cast, flexure 4.2 to 4.4 3.5 PS beam plant-cast, flexure 0.90
PC beam plant-cast, shear 4.3 to 4.4 3.5 PC beam plant-cast, shear 0.85
RC slab cast-in-place 2.3 to 2.5 2.5 RC slab cast-in-place, flexure 0.90
RC slab plant-cast 3.8 to 3.9 3.5 RC slab plant-cast, flexure 0.90
PS slab plant-cast 4.7 to 5.0 3.5 PS slab plant-cast, flexure 0.90
Post-tensioned slab cast-in-place 2.3 to 2.5 2.5 Post-tensioned slab cast-in-place, flexure 0.90
RC slab cast-in-place, tied 3.8 to 4.1 4.0 RC column cast-in-place, tied 0.75
RC column plant-cast, tied 3.9 to 4.2 4.0 RC column plant-cast, tied 0.75
RC column cast-in-place, spiral 4.0 to 4.4 4.0 RC column cast-in-place, spiral 0.80
RC column plant-cast, spiral 4.2 to 4.5 4.0 RC column plant-cast, spiral 0.80
PS column plant-cast, tied 5.0 to 5.3 4.0 PS column plant-cast, tied 0.75
PS column plant-cast, spiral 5.8 to 6.2 4.0 PS column plant-cast, spiral 0.80
Plain concrete, flexure, shear 5.7 to 6.2 4.0 Plain concrete, shear 0.65

It was observed that the reliability indexes are low for the ACKNOWLEDGMENTS
load combinations with dead load being approximately 80 to The research reported in this paper (PCA R&D Serial No.2634) was
90% of the total load, and with live load being 10 to 20%. To conducted by the authors, with the sponsorship of the Portland Cement
achieve uniform values of , it is recommended to replace Association (PCA Project Index No. 99-10) and the Precast/Prestressed
Concrete Institute. The contents of this paper reflect the views of the authors,
the first load combination in Eq. (2) (1.4D < R), with who are responsible for the facts and accuracy of the data presented. The
contents do not necessarily reflect the views of the Portland Cement Association
1.4 ( D + L ) < R (7) or the Precast/Prestressed Concrete Institute.

CONCLUSIONS AND RECOMMENDATIONS REFERENCES


The reliability-based calibration was performed for structural ACI Committee 318, 1999, Building Code Requirements for Structural
Concrete (ACI 318-99) and Commentary (318R-99), American Concrete
types and materials covered by the ACI 318-99 Code. The Institute, Farmington Hills, Mich., 391 pp.
objective was to calculate the resistance factors for the design of ASCE 7-98, 1998, Minimum Design Loads for Buildings and Other
concrete structures corresponding to load and load combination Structures, American Society of Civil Engineers, Washington, D.C., 352 pp.
factors specified by the ASCE 7-98 Standard. Ellingwood, B.; Galambos, T. V.; MacGregor, J. G.; and Cornell, C. A.,
Three categories of concrete: ordinary, high strength, and 1980, Development of a Probability Based Load Criterion for American
lightweight; and two categories of steel: reinforcing bars and National Standard A58, NBS Special Report 577, U.S. Department of
Commerce, National Bureau of Standards, 222 pp.
prestressing strands, were considered. Statistical parame-
Ellingwood, B., and Rosowsky, D., 1996, Combining Snow and Earthquake
ters were determined for the new material test data provided Loads for Limit States Design, ASCE Journal of Structural Engineering,
by the industry. By comparison with previous studies, it was V. 122, No. 11, pp. 1364-1368.
observed that the quality control of material has improved in Madsen, H. O.; Krenk, S.; and Lind, N. C., 1986, Methods of Structural
the last 20 to 30 years. In particular, this applies to concrete Safety, Prentice-Hall, N.J., pp. 403.
and reinforcing bars. Prestressing strands continue to exhibit Melchers, R. E., 1987, Structural Reliability Analysis and Prediction,
a very low degree of variation. Ellis Horwood Limited, 400 pp.
The reliability indexes were calculated for a wide variety Nowak, A. S., 1999, Calibration of LRFD Bridge Design Code,
of structural types and limit states (beams in flexure, shear, NCHRP Report 368, Transportation Research Board, Washington, D.C.
slabs, tied and spiral columns, tension members, cast-in- Nowak, A. S., and Collins, K. R., 2000, Reliability of Structures, McGraw-
Hill New York, 360 pp.
place, and plant-cast). A variation of s is observed for various
Nowak, A. S., and Szerszen, M. M., 2001, Reliability-Based Calibration
design cases. In general, s are lower for concrete slabs. for Structural Concrete, Report UMCEE 01-04, Department of Civil and
Reliability indexes are also lower for load combinations Environmental Engineering, University of Michigan, Ann Arbor, Mich., Nov.
involving wind and earthquake. Nowak, A. S., and Szerszen, M. M., 2003, Calibration of Design Code
The final selection of resistance factors for ACI 318 Code for Buildings (ACI 318): Part 1Statistical Models for Resistance,
is based on the results of the reliability analysis, comparison ACI Structural Journal, V. 100, No. 3, May-June, pp. 377-382.
with previous practice, and simplicity of the code (convenience Thoft-Christensen, P., and Baker, M. J., 1982, Structural Reliability Theory
and Its Applications, Springer-Verlag Berlin Heidelberg, New York, 267 pp.
of the designer). Therefore, there is a need to minimize the
Turkstra, C. J., 1970, Theory of Structural Design Decisions, Solid
number of different resistance factors in the code. Mechanics Study No. 2, University of Waterloo, Waterloo, Canada.
In addition to the change in resistance factors, it is recom- Turkstra, C. J., and Madsen, H. O., 1980, Load Combinations in Codified
mended to change the load combination factors for dead load Structural Design, Journal of the Structural Division, ASCE, V. 106,
and live load, as shown in Eq. (7). No. ST12, Dec., pp. 2527-2543.

ACI Structural Journal/May-June 2003 391

S-ar putea să vă placă și