Professional Documents
Culture Documents
REINFORCED-CONCRETE COLUMNS
By Sonia E. RuizI and J. Cipriano Aguilar2
INTRODUCTION
RELIABILITY ANALYSIS
Structural reliability is defined as the probability that a structure does not
fail or does not cease to adequately provide the service it was intended for.
~Prof. of Civ. Engrg., Inst. of Engrg., National Univ. of Mexico, Apdo Postal 70-
472, 04510 Mexico D.F., Mexico.
2Res. Asst., Dept. of Appl. Mech., Inst. of Engrg., UNAM.
Note. Discussion open until November 1, 1994. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on
January 14, 1993. This paper is part of the Journal of Structural Engineering, Vol.
120, No. 6, June, 1994. 9 ISSN 0733-9445/94/0006-1850/$2.00 + $.25 per
page. Paper No. 5263.
1850
The reliability index 13 is a parameter related to the probability of failure.
In the present paper, 13 equals m,/~r,, where m, is the mean value of the
ratio In(R/S), and % is its standard deviation (Rosenblueth and Esteva
1971). Here R and S stand for structural resistance and load effect on the
structure, respectively. They are random variables that are simulated by
means of the Monte Carlo technique. For the numerical analysis the fol-
lowing steps are taken:
1851
all cases, the statistical characteristics of loads and quality control are typical
of Mexican construction. In (1) and (2), the values 1.4 and 1.7 are associated
with the load factors specified in the codes under study. Load factors are
equal to 1.4 for dead and live loads in the latter code; and equal to 1.4 for
dead-load and 1.7 for live-load actions in U.S. regulations.
7. The loads S;, i = 1, 2, . . . , N are simulated.
8. The mean m,, and standard deviation ~ru of u~ are calculated from the
simulated values ul = ln(R;/Si), i = 1, 2, . . . , N. From these, 13 =
m , k r , is obtained.
RESISTANCE CONSIDERATIONS
where Eco = initial tangent modulus that is equal to 2fco/~co, and ~c rep-
resents the concrete strain.
1852
ks fc
0.85 k s fc
0.30 k s fc
t t I
~cl Ec2 ~c85 E
FIG. 1. Stress-Strain Curve for Confined Concrete
Confinement Model
The contribution of the transverse reinforcement is taken into account
by means of the overresistance factor Ks (Sheikh and Uzumeri 1982). The
stress-strain relation for confined concrete (fc - e) assumed in the present
paper is shown in Fig. 1. Its main parameters are (Sheikh and Uzumeri
1982)
f~ = KJcp (6a)
Ks = 1 + - - - - ~ o 1 5 . - ~ 2 ) ~ 1 - ~-~ ~ (6b)
a ec = kAco (6c)
nc 2
k = 1 (6d)
5.5Aco
sc1 = 0.55K~f'(1 • 10 -6) (7a)
1853
fsfsut
fy~ I
Ey Esh Esu E
FIG,2. Stress-Strain
CurveforSteelBars
[m(G- ash) + 2
f" = fY L60(a, ash) + 2
Slenderness Effect
Two main criteria are normally used for the structural analysis of slender
columns: moment magnification and second-order analysis. The former is
generally adopted by design codes, including ACI 318-89 and RCDF-87.
Design Resistance
The amplified moment, used to calculate the design resistance in the
present paper, is given by
M c = [Cm/(1 - P,/qb, Pc)]Ma >- M 2 (10)
in which M2 = largest of the factored bending moments applied at the ends
of the slender column; P, = factored axial load acting on the column; Pc
= critical buckling strength of the column; Cm = equivalent uniform-mo-
ment diagram factor; and ~bs = stability resistance factor. In the present
paper, Cm = 1, since the members studied are assumed to be in single-
curvature bending in nonsway frames with equal moments applied at both
ends; +s is specified as 1 in RCDF-87, and is equal to the cross-sectional
resistance factor in ACI 318-89 (0.75-0.90).
1854
RCDF-87
60 - - - - - ANSI-82
40
0
0
_I
20
I I I
2000 4000 6000
Influence area, ft 2
a) B e a m s
601 b RCDF-87
--.mANSI-82
"'-k
Q..
\
40 - \
"10
0
0
_.1
20
I I I
2000 4000 6000
Influence area, ft 2
b) Columns
FIG. 3. Code Loads for Offices: (a) Beams; (b) Columns
Limit States
The second-order analysis is based on the estimation of the column's
lateral deflections and second-order moments. These are added to the pri-
mary moments in order to calculate the amplified moments. A numerical
integration method to calculate the column's lateral deflections can be used;
however, it can result in a high computing cost. In the present paper, a
trial-and-error procedure (Newmark 1943) is used. For a column bending
in single curvature under equal end moments, a second-order parabola has
been suggested to represent its shape. The lateral deflection A is assumed
to occur at midheight of the column. This is given by
A,, = 12(4),, + 0.25+e)/10 (11)
where l = column height; and qb,, and +e = curvatures at midheight and
at the ends of the column. For a given eccentricity e and axial load P, the
1855
moment-curvature diagram is calculated and used to obtain the deflected
shape of the column (11) and the amplified moment M = P(e + 2~m). This
moment is compared with that obtained in the preceding iteration cycle,
until the calculated shape matches the moments and curvatures at the ends
and midheight of the column (Newmark 1943). Several iterations are re-
quired at each load level. The maximum moment in the column, the cor-
responding axial load, and the selected eccentricity define one point of the
interaction diagram of the slender column. This procedure is used in the
present paper to define the limit state for the analysis of column reliability.
LOAD CONSIDERATIONS
In the present study, the reliability associated with AC1318-89 and RCDF-
87 is examined. ACI 318-89 does not specify design loads; these are given
by the governing building code. It is assumed here that the live loads are
given by the ASCE 7-88 standard.
It is worth emphasizing the differences between ASCE 7-88 and RCDF-
87 specifications in connection with the lifetime exceedance probabilities of
the design live loads specified in those documents:
It has been shown (Chalk and Corotis 1980) that the mean of the lifetime
(50-years) maximum total live loads corresponds approximately to the ASCE
7-88 values.
On the other hand, RCDF-87 loads correspond to a fractile of about 0.9-
0.999 of the 50-year maximum load. For the special case of office loads
with influence areas of approximately 92.9-185.8 m 2 (1,000-2,000 sq. ft.),
the corresponding exceedance probability is approximately 2%.
The differences between exceedance probabilities assumed here for ASCE
7-88 and RCDF-87 are reflected in (1) and (2). The factor 1/(1 + 2C,) is
contained only in the first equation, whereas a value of unity appears in the
second. Figs. 3 ( a - b ) represent the comparison between the two code for-
mulations for office loads. Curves labeled RCDF-87 are the design values
from RCDF-1987. Since this code is based on a tributary area rather than
the influence area, two sets of curves are shown. For a column, the influence
area is four times the tributary area (McGuire and Cornell 1974). Curves
labeled ANSI-82 correspond to ASCE 7-88 design values.
ANALYSIS OF COLUMNS
Nominal cross sections of 400 mm • 400 mm (15.75 in. • 15.75 in.) and
400 mm x 800 mm (15.75 in. • 31.5 in.) are analyzed. The longitudinal
steel-reinforcement percentages pg of 1%, 3%, and 6% of the section area
are distributed along two and four bar layers. Normalized tie spacing is
taken equal to s/h* = 0.125, 0.20, and 0.25. The assumed eccentricities
divided by the nominal section depth e/h* are equal to 0, 0.001, 0.01, 0.015,
9 .0.1, 0.15, . . . , 7.5, 8.0 (see Figs. 5-8). Relative lengths of l/h* of 0,
.
10, and 15 are studied9 A total of 216 cases were analyzed. The statistical
characteristics of the columns analyzed are in Tables 1 and 2.
The geometric and material characteristics mentioned in Tables 1 and 2
represent the variabilities of Mexico City construction (Meli 1985).
The mean value and standard deviation of the concrete resistance that
appear in Table 2 correspond to concrete in the structure rather than to
cylinders.
In other papers (Mirza and MacGregor 1989; Mirza and Shrabek 1992),
the yield stress of the reinforcement bars was assumed to have a 13probability
1856
TABLE 1. Geometric Characteristics of Columns Analyzed
Specified Standard
value Mean value deviation Probability
Variable [ram (in.)] [turn (in,)] [mr" (in,)] distribution
(1) (2) (3) (4) (5)
b 400 (15.75) 395.4 (15.56) 3.82 (0.1503) Normal
h 400 (15.75) 395.4 (15.56) 3.82 (0.1503) Normal
h 800 (31.50) 802.3 (31.58) 2.08 (0.0818) Normal
r 40 (1.57) 45.0 (1.77) 10.5 (0.4133) Normal
distribution. This has shown an influence on columns with high e/h* values
(Mirza and MacGregor 1989); however it does not influence the overall
slender beam-column strength. A normal distribution is assumed in the
present paper.
The influence of the following parameters on 13 is analyzed for each code:
load ratio re, slenderness ratio l/h*, concrete confinement of the elements,
longitudinal reinforcement pg, cross-sectional width/depth ratio b/h, number
of bar layers, and tie spacing s/h*.
Balanced point
7 u.m Pure compression
~ Pure bending
9 - 9 ~ o
t 1 1 I
0.4 0.5 0.6 0.7 0.8 0,9 1.0
re
and B, are identified in these and the following figures. The former cor-
responds to elements failing in compression, the latter to those subjected
to compression loads and failing in flexural tension. The boundary between
zones A and B corresponds to the balanced eccentricity. Fig. 5 shows that
the general trends of the curves corresponding to both codes are similar.
Both regulations give place to higher values of I~ indexes as the slenderness
ratio increases. In Fig. 5 it is shown that for each code, the reliabilities for
slender and short columns tend to be equal as the ratio e/h* increases (pure
bending).
Slenderness ratios l/h* of 10 and 0 give place to similar values of 13, except
in the interval 0.06 < e/h* < 2, where columns with l/h* = 10 have 13values
greater (approximately 0.4 13) than those with l/h* = O.
I/h*
- - 0
-----10
----15
,5 . . . . . ""'---- ~ , j A ~ r
,~ _ %/~.Z./.~,.r.-.-,.~\ /',..
.~..~ .~,.
1859
7
A B
/3
6 RCDF
-. ~ o ~
~ . o ,~o
h l~./''~'"~'k,"~
*
- - Confined
Unconfined
2
-O.OOl 0,01 0.1 1 10
e/h*
FIG. 6. Effect of Confinement on 13 (RCDF-87 Designs) with Cross Sections of
400 mm x 800 mm (15.75 in, x 31.5 in.); pg = 3%; and l/h* = 10
G E N E R A L V I E W OF 13 I N D E X E S F O R O T H E R LIMIT S T A T E S
A general view of the mean values of 13 for rc = 0.7, including those
corresponding to slender columns, is presented in Table 3. The values as-
sociated with flexure, shear, and torsion limit states were calculated and
discussed previously by the first writer (Ruiz 1993). The method of analysis
and the assumptions of the previous work are the same as those used in the
present paper. Therefore the t3 values obtained in that study and in the
present paper can be compared.
From Table 3, it is concluded that 13 indexes associated with slender
columns (I/h* = 15) are the highest. For small and intermediate e/h* values,
those indexes are higher than those corresponding to short columns (l/h*
= 0). For higher normalized eccentricities e/h*, f3 indexes get closer to those
associated to failure in bending. 13 values for shear and torsion should be
greater than for bending; nonetheless, the reverse is true for ACI 318-89 13
1860
8
(a)
A B
p,
- - 3 % A S
-----6% =
P~ (b)
--3% A B
5 --'--6% = : =-
indexes. This has been already pointed out by Israel et al. (1987) and Ruiz
(1993).
Load factors intended to apply to all structural materials used in U.S.
building construction have been developed by Ellingwood et al. (1981) under
the sponsorship of the National Bureau of Standards (NBS). For example,
they recommend load factors equal to 1.2 and 1.6 for dead and live loads,
instead of 1.4 and 1.7 as used in the present study (see ACI 318-89, section
9.2.1). Based on those load factors, values of resistance factors for concrete
design have been calculated (MacGregor 1983). The results show that those
safety factors give place to more consistent reliabilities for reinforced-con-
crete elements.
1861
7 0.125 A = ~ =- B
0.20
6 0.25
3 - l/h*=O/
TABLE 3. Mean Values of 13 for rc = 0.7 [see Ruiz (1993)]. Compression Plus
Bending Limit State
RCDF-87
Confined Unconfined ACI 318-89
Column size e/h* column column tied column
(1) (2) (3) (4) (5)
Short, l/h* = 0 0 5.80 5.10 4.30
Short, l/h* = 0 0.3 5.57 5.00 4.10
Short, I/h* = 0 2.0 5.03 5.00 3.90
Short, I/h * = 0 8.0 4.50 4.50 3.20
Slender, l/h* = 15 0 6.50 5.80 5.00
Slender, l/h* = 15 0.3 6.00 5.80 4.40
Slender, l/h* = 15 2.0 5.10 5.10 4.10
Slender, l/h* = 15 8.0 4.50 4.50 3.20
Note: For RCDF-87, flexure = 4.25; shear = 4.75 and torsion = 5.21. For ACI 318-
89, flexure = 3.20; shear = 3.00; and torsion = 2.80.
CONCLUSIONS
Values of the reliability index 13have been calculated for short and slender
columns, complying with A C I 318-89 and R C D F - 8 7 codes. From this study
the following conclusions were obtained.
The results reveal that the following parameters are significant for the
reliability evaluation of reinforced-concrete slender columns: load and slen-
derness ratios, longitudinal steel, and end eccentricity ratio. This is con-
sistent with results obtained by Mirza and MacGregor (1989) and Mirza and
Skrabek (1992).
13-index values corresponding to slender columns are the highest with
1862
respect to the following limit states: flexure, torsion, and shear of beams,
as well as compression plus bending of short columns (Table 3).
As expected, in all cases, RCDF-87 13 values are higher than those as-
sociated with A C I 318-89. There are three reasons for this.
The probabilities of exceedance of the maximum loads adopted for both
codes are drastically different. It seems advisable to check if the exceedance
probability specified in RCDF-87 is adequate. On the other hand, it is
believed that in many buildings in downtown Mexico City the live loads
significantly exceed those recommended by the code. This became evident
during the 1985 earthquake, when it was discovered that a large number of
the constructions damaged by the earthquake were loaded in excess of the
loads specified in RCDF-87 code. This fact makes advisable the adoption
of a low-exceedance probability in the Mexico City Code.
To evaluate the strength of the structural elements, A C I 318-89 defines
concrete strength as f'c, while RCDF-87 uses f* = 0.8f'c in the design equa-
tions. That is, the Mexican code provides higher protection against uncer-
tainties related to concrete properties and to quality control during its pro-
duction and placement, because this control is often poorer than is desirable.
Load factors and strength-reduction factors differ for both codes. This
also contributes to the differences in 13 values.
ACKNOWLEDGMENT
The writers wish to thank Santiago Loera for his valuable suggestions.
The manuscript review by L. Esteva is acknowledged. Comments offered
by R. Meli and R. G. Tremari are appreciated. This study was partially
financed by the D e p a r t m e n t of the Federal District, Mexico City under
grant DDF-UNAM-CA-016-91.
APPENDIX I. REFERENCES
"Building code requirements for reinforced concrete." (1989). ACI318-89, American
Concrete Institute (ACI), Detroit, Mich.
Chalk, P. L., and Corotis, R. B. (1980). "Probability model for design live loads."
J. Struct. Div., ASCE, 106(10), 2017-2033.
Desayi, P., and Krishnan, S. (1964). "Equation for the stress-strain curve of con-
crete." J. of the Am. Concrete Inst., 61(3), 345-350.
Ellingwood, B. (1977). "Statistical analysis of RC beam-column interaction." J.
Struct. Div., ASCE, 103(7), 1377-1388.
Ellingwood, B., Galambos, T. V., MacGregor, J. G., and Cornell, C. A. (1981).
"A probability-based load criterion for structural design." Civ. Engrg., ASCE,
51(7), 74-76.
Floris, C., and Mazzucchelli, A. (1991). "Reliability assessment of RC column under
stochastic stress." J. of Struct. Engrg., ASCE, 117(11), 3274-3292.
Israel, M., Ellingwood, B., and Corotis, R. (1987). "Reliability-based code for-
mulations for reinforced concrete buildings." J. ofStruct. Engrg., ASCE, 113(10),
2235-2252.
MacGregor, J. G. (1983). "Load and resistance factors for concrete design." ACI
J., Title No. 80-27, 279-287.
McGuire, R. K., and Cornell, C. A. (1974). "Live loads effects in office building."
J. Struct. Div., ASCE, 100(7), 1351-1366.
Meli, R. (1985). "Estudio estadistico de propiedades de materiales y estructuras de
concreto reforzado en el Distrito Federal." Res. Rep. Inst. of Engrg., National
University of Mexico, Mexico, D. F., (in Spanish).
Mendoza, C. J. (1983). "Propiedades mec~nicas de los concretos fabricados en el
Distrito Federal." Res. Rep., Inst. of Engrg., National University of Mexico,
Mexico, D. F. (in Spanish).
1863
"Minimum design loads for buildings and other structures." (1990). ASCE 7-88.
ASCE, New York, N.Y.
Mirza, S. A., and MacGregor, J. G. (1989). "Slenderness and strength reliability of
reinforced concrete columns." A C I Struct. J., 86(4), 428-438.
Mirza, S. A., and Skrabek, B. W. (1992). "Statistical analysis of slender composite
beam-column strength."./. Struct. Engrg., ASCE, 118(5), 1312-1332.
Newmark, N. M. (1943). "Numerical procedure for computing deflections, moments
and buckling loads." Trans., ASCE, 108, 1161.
"Normas t6cnicas complementarias para disefio y construcci6n de estructuras de
concreto." (1987). Official Bull. of the Federal District Dept., No. 44, Mexico, D.
F., (in Spanish).
Park, R., and Paulay, T. (1975). Reinforced Concrete Structures, John Wiley & Sons,
New York, N.Y.
"Reglamento de construcciones para el Distrito Federal." (1987). RCDF-87, Official
Bull. of the Federal District Dept., No. 9, Mexico D. F., (in Spanish).
Rosenblueth, E., and Esteva, L. (1971). "Reliability basis for some Mexican codes."
Probabilistic design of reinforced-concrete building. SP31 Special Publication,
American Concrete Institute (ACI), Detroit, Mich.
Ruiz, S. E. (1993). "Reliability associated to safety factors of the ACI 318-89 and
the Mexico City concrete design regulations." ACI Struct. J., 90(3), 262-268.
Sheik, S. A., and Uzumeri, S. M. (1982). "Analytical model for concrete confinement
in tied columns." J. Struct. Div., ASCE, 108(12), 2703-2722.
1865