Sie sind auf Seite 1von 9

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S44

Nonlinear Analysis of Membrane Elements


by Fixed-Angle Softened-Truss Model

by Thomas T. C. Hsu and Li-Xin “Bob” Zhang

A fixed-angle softened-truss model, which assumes the cracks to be analyzing reinforced concrete was initiated in the late 1960s
oriented in the principal compression direction of the externally when Nielson (1967) and Lampert and Thurlimann (1968)
applied stresses, has been proposed for nonlinear analyses of rein- derived the three fundamental equilibrium equations for
forced concrete membrane elements (Pang and Hsu 1996). This shear based on the theory of plasticity. The interaction rela-
new smeared-crack model takes into account the “concrete contri-
bution” which is produced by the shear resistance of concrete
tionship of bending, shear, and torsion was further developed
along the initial crack direction. This “concrete contribution” by Elfgren (1972). These theories were known collectively
cannot exist in the various rotating-angle models because they as the equilibrium (plasticity) truss model.
assume the cracks to be oriented in the principal compression The compatibility condition of reinforced concrete element
direction of the post-cracking concrete. subjected to shear was first introduced by Collins (1973).
The fixed-angle softened-truss model is capable of predicting the Because the average strain condition must satisfy Mohr’s
entire load-deformation history because it takes into consideration circle, this type of theory is called the Mohr compatibility truss
all 12 equations governing the equilibrium condition, the compati-
bility condition, and the constitutive laws of materials. This paper
model. Three compatibility equations can be established from
proposes an efficient algorithm for solving these 12 governing the geometric relationships in a Mohr's circle.
equations. This algorithm is used to analyze the behavior of A fundamental breakthrough in the understanding of shear
reinforced concrete panels tested at the University of Houston. The was the discovery of the softening of the concrete struts by
analysis shows that the new fixed-angle model is much more Robinson and Demorieux (1968) and the first quantification
powerful than the prevailing rotating-angle models. of this phenomenon by Vecchio and Collins (1981, 1982).
Prior to 1968, the stress-strain curve of the concrete struts
Keywords: constitutive laws; compatibility; equilibrium; fixed-angle;
was assumed to be the same as that obtained from the
membrane elements; membrane stresses; nonlinear analysis; post-cracking
behavior; reinforced concrete; rotating-angle; shear stress; shear strain;
uniaxial compression tests of standard concrete cylinders.
softened truss model; stress; strain; stress-strain relationship. This assumption led to a severe overestimation of the shear
strength. Robinson and Demorieux observed that the prin-
INTRODUCTION cipal compressive stress in a reinforced concrete panel was
Wall-type and shell-type reinforced concrete structures, softened by principal tension. This softening phenomenon
such as shear walls, box bridges, I-girders, shell roofs, was quantified by Vecchio and Collins, who proposed a
offshore platforms, and nuclear containment vessels, can be stress-strain curve incorporating a softening coefficient.
visualized as assemblies of membrane elements subjected to By combining the equilibrium, compatibility, and the soft-
in-plane stresses. The key to a rational analysis of these ened stress-strain relationship of concrete, Vecchio and
structures is to first thoroughly understand the behavior of Collins (1981, 1982) developed a compression field theory
reinforced concrete membrane elements (or panels) isolated to predict the post-cracking relationship between shear stress
from such structures. and shear strain. This rational approach was generalized by
The modern truss model treats a cracked reinforced Hsu and his colleagues to establish a softened-truss model
concrete element as a continuous material. The stresses and (Hsu 1988) that is applicable to both shear and torsion. This
strains of concrete and steel in an element are evaluated by unified theory could predict with good accuracy the behavior
the average or smeared values crossing several cracks. Using of various types of structures subjected to shear or torsion
these average stresses and strains (or smeared stresses and (Hsu and Mo, 1985a,b,c; Mau and Hsu, 1986, l987a,b; Hsu,
strains) in cracked reinforced concrete allows us to apply the Mau and Chen 1987). A unified nomenclature for all rational
principles of mechanics of materials. This modern way of theories in reinforced concrete is proposed by Hsu (1996).

ACI Structural Journal, V. 94, No. 5, September-October 1997.


This paper is the eighth in a series of papers dealing with the devel- Received October 2, 1995, and reviewed under Institute publication policies. Copy-
opment of a unified theory for reinforced concrete sponsored by the right © 1997, American Concrete Institute. All rights reserved, including the making
National Science Foundation and conducted at the University of of copies unless permission is obtained from the copyright proprietors. Pertinent
Houston. discussion will be published in the July-August 1998 ACI Structural Journal if
received by March 1, 1998.

ACI Structural Journal/September-October 1997 483


The three rotating-angle models have a common weak-
Thomas T. C. Hsu, FACI, is a professor in the Department of Civil and Environmental
Engineering at the University of Houston, Texas. The author of two books and many ness: they are incapable of predicting the so-called “contri-
journal publications, Dr. Hsu was the recipient of ACI’s Anderson Award for Research bution of concrete.” Tests have shown that the shear strength
in 1991 and the Wason Medal for Materials Research in 1965. He is a member of ACI
Committee 215, Fatigue; of joint ACI-ASCE Committee 343, Concrete Bridges; and of
of a membrane element is made up of two terms: the “major
445, Shear and Torsion. term” attributed to the steel, and the “minor term” attributed
to the concrete. The existence of the minor term is apparently
Li-Xin “Bob” Zhang is a senior structural engineer at McDermott Engineering
Houston, LLC. He received his BS and MS degrees from Tianjin University, Tianjin, caused by the shear stresses that develop along the cracks
China, and his PhD from the University of Houston in 1995. His current interest is in oriented in the 2-direction (Dei Poli et al. 1987, 1990; Kupfer
the design and analysis of deep water structures and the nonlinear behavior of rein- and Bulicek 1992). To take into account this “minor term” or
forced concrete structures.
“contribution of concrete,” a softened-truss model was
developed which assumes that the cracks follow the 2-direc-
FIXED-ANGLE SOFTENED-TRUSS MODEL tion, Fig. 1(f). This new model, which is based on the
A reinforced concrete membrane element subjected to in- smeared-crack concept, is called the fixed-angle softened-
plane stresses σl, σt, and Tlt is shown in Fig. 1(a). This truss model (Pang and Hsu 1996).
element is composed of concrete, Fig. 1(b), and orthogonal The fixed-angle softened-truss model becomes identical to
steel reinforcement, Fig. 1(c), which defines the orientation the rotating-angle softened-truss model when the d-r coordi-
of l-t coordinate system. The applied principal stresses for
nate coincides with the 2-1 coordinate. This coincidence will
the element are designated as σ2 and σl, which define the 2-
occur in the special case of panels that are reinforced with
1 coordinate system as shown in Fig. 1(d). The angle
equal smeared steel stresses in the l- and t-directions (i.e.,
between the l-axis and the 2-axis is the fixed-angle α2. If the
ρ l f l = ρt ft in Fig. 1(c)) and subjected to pure shear (i.e. σl =
“smeared steel stresses” ρl f l and ρt ft in Fig. 1(c) are
σt = 0 in Fig. 1(a)). Therefore, this type of test panel is
different, then the concrete after cracking in Fig. 1(b) will
have principal stresses that deviate from the 2-1 coordinate. useless in determining the differences between the fixed-
The new direction of post-cracking principal stresses in angle softened-truss model and the rotating-angle softened-
concrete, σd and σr, are defined by the d-r coordinate system truss model (e.g. Series A in Pang and Hsu, 1995). Useful
shown in Fig. 1(e). panels for determining the differences between the two
The three truss models (equilibrium truss model, Mohr models are those with unequal smeared steel stresses in the
compatibility truss model, and the softened truss model) all l- and t-directions (e.g. Series B in Pang and Hsu [1995]).
assumed that the direction of the cracks follows the d-axis, Extensive experiments have been conducted to generate the
Fig. 1(g). Since the angle α between the d-axis and the l-axis constitutive law of concrete in shear that is required in the
is called the rotating angle, these three truss models are fixed-angle softened-truss model (Pang and Hsu, 1996; Zhang
collectively referred to as the rotating-angle theory. A soft- 1992). Incorporation of this new shear stress-strain relation-
ened truss model based on a rotating angle is called the ship of concrete into the model results in four constitutive
rotating-angle softened-truss model. The rotating-angle soft- laws, in addition to the three equilibrium conditions and three
ened-truss model is valid only within a range of 33 deg <α < compatibility conditions. To satisfy all these conditions
57 deg if full ductility is desired (Pang and Hsu 1995). requires the simultaneous solution of 12 equations. The

Fig. 1—Reinforced concrete membrane elements subjected to in-plane stresses

484 ACI Structural Journal/September-October 1997


primary purpose of this paper is to present an efficient algo- The fixed-angle α2 used in Eq. (1) to (3) is also used here
rithm for solving these 12 equations. in Eq. (4) to (6). This implies the assumption of coaxiality
between the principal stresses and the principal strains.
RESEARCH SIGNIFICANCE
A fixed-angle softened-truss model, which is capable of Constitutive laws
producing the “concrete contribution” without external inter- There are two kinds of finite element modeling for cracked
ference, is proposed for the nonlinear analysis of reinforced reinforced concrete: the microscopic models and the macro-
scopic models (ASCE/ACI, 1993). Microscopic models are
concrete structures composed of membrane elements. In this
based on the stress-strain relations of plain concrete and
paper, an efficient algorithm is proposed for the solution of
steel, focusing on their interactions through bond slipping
this new model. This algorithm is used to analyze the behavior along the reinforcing bars and through shear-sliding along
of reinforced membrane elements with concrete strength of 42 cracked surfaces. This microscopic modeling was found to
MPa (6000 psi) and 65 MPa (9500 psi). From this analysis, the be unsuitable for large structures, because the strong and
range of applicability of the fixed-angle softened-truss model complicated interaction between concrete and steel can be
is determined. It was found that this new model is much more reproduced neither by the simple superposition of the mate-
powerful than the existing rotating-angle models. rial laws of plain concrete and steel, nor by the introduction
of bond and shear interface elements. To overcome this diffi-
GOVERNING EQUATIONS culty, the macroscopic models were developed. This kind of
The fixed-angle softened-truss model (Pang and Hsu, modeling (also called the smeared crack concept) is based on
1996) requires three equilibrium equations, three compati- the average stress-strain relationships of concrete and steel
bility equations, and four constitutive laws for concrete and where bond slipping and shear sliding are implicitly
steel as follows: included. The constitutive laws of concrete and steel were
then obtained directly from tests of reinforced concrete
panels subject to biaxial loading.
Equilibrium equations
The constitutive laws given below were established by the
The three equilibrium equations, which relate the applied macroscopic modeling. All the stresses in Eqs. (1) to (3)
stresses (σl, σt and τlt) to the internal stresses of reinforce- must be the average (or smeared) stresses, and all the strains
ment (fl and ft) and concrete (σc2, σc1 and τc21) in a in Eqs. (4) to (6) must be the average (or smeared) strains.
membrane element, are written with the help of the coordi- An average stress means an average value of the stresses
nate transformation of average concrete stresses (Hsu1993) from a crack to midway between two cracks. An average
strain is measured from the displacement over a length that
c 2 c 2 c traverses several cracks, thus including the gaps that consti-
σ l = σ 2 cos α 2 + σ 1 sin α 2 + τ 21 2 sin α 2 cos α 2 + ρ l f l (1)
tute the crack widths.
(1) Average compressive stress-strain relationship of
c 2 c 2 c concrete in 2-direction:
σ t = σ 2 sin α 2 + σ 1 cos α 2 + τ 21 2 sin α 2 cos α 2 + ρ t f t (2)
This relationship of σc2 vs. ε2 is assumed to be parabolic
as follows:
c c c 2 2
τ lt = ( – σ 2 + σ 1 ) sin α 2 cos α 2 + τ 21 ( cos α 2 – sin α 2 ) (3) ε2 ⎞ ⎛ ε2 ⎞ 2 ε2
σ 2 = ζ f c′ 2 ⎛ -------
c
- – -------- -≤1
------- (7a)
⎝ ζε ⎠ ⎝ ζε ⎠ ζε o
o o
The fixed-angle α2 is considered a known value, deter-
mined from the applied stresses, σl, σt , and τlt. ε 2 ⁄ ζε o – 1 ⎞ 2 ε2
σ 2 = ζ f c′ 1 – ⎛ -------------------------
c
- ->1
------- (7b)
⎝ 2⁄ζ–1 ⎠ ζε o
Compatibility equations
The compatibility equations deal with the 2-D relation- where εo is the strain at the peak stress of concrete
ships of average strains in a membrane element. In the fixed- cylinder. The softened coefficient ζ was found to be a
angle softened-truss model, the compatibility equations function of the tensile strain ε1, the parameter η, and the
represent Mohr’s relationship between the strains (εl, εt and concrete strength fc′ (Zhang 1995):
γlt) in the l-t coordinate of the reinforcement and the strains
(ε2, ε1 and γ21) in the 2-1 coordinate of the applied stresses. 5.8 1
These three compatibility equations are expressed by the ζ = -------------------------- ---------------------------------- (8)
f c′ ( MPa ) ⎛ 400ε
coordinate transformation of average strains (Hsu 1993): 1 + -------------1- ⎞
⎝ η ⎠

2 2 γ 21
- 2 sin α 2 cos α 2
ε l = ε 2 cos α 2 + ε 1 sin α 2 + ------ (4) where
2
ρ t f ty – σ t
η = ---------------------
- (8a)
2 2 γ 21 ρ l f ly – σ l
- 2 sin α 2 cos α 2
ε t = ε 2 sin α 2 + ε 1 cos α 2 + ------ (5)
2
In Eq. (8), 5.8/ f c′ = 0.9 for a normal strength concrete of
42 MPa (6000 psi). Assuming η to be approximately unity,
γ lt γ 21 2 2 Eq. (8) is degenerated into the softened coefficient given by
----- = ( – ε 2 + ε 1 ) sin α 2 cos α 2 + ------
- ( cos α 2 – sin α 2 ) (6)
2 2 Pang and Hsu (1995) for rotating-angle softened-truss model.

ACI Structural Journal/September-October 1997 485


(2) Average tensile stress-strain relationship of concrete The shear stress of concrete τc21 can be derived from the
in 1-direction: two equilibrium equations, (1) and (2):
This relationship between σc1 and ε1 is the same as the one
used in the rotating-angle softened-truss model: c 1 c c
τ 21 = -------------------- [ ( σ l – σ t ) – ( ρ l f l – ρ t f t ) ] – ( σ 2 – σ 1 ) cot 2 α 2 (17)
2 sin 2α 2
c
σ1 = Ec ε1 ε 1 ≤ 0.00008 (9a) For the case of pure shear (α2 = 45 deg and σl = 0) the
shear stress is
0.00008 0.4
σ 1 = f cr ⎛ ------------------- ⎞
c
ε 1 > 0.00008 (9b)
⎝ ε ⎠ c 1
1 τ 21 = --- ( ρ t f t – ρ l f l) (17a)
2
The elastic modulus of concrete Ec in Eq. (9a) is taken as
3875 f c′ (fc′ and f c′ are in MPa) This Ec should also serve In Eq. (17) the maximum shear stress of concrete τc21m is
as a maximum limit for the secant modulus in compression the shear stress τc21 that corresponds to the maximum
when using Eq. (7a). The concrete cracking stress fcr in Eq. (9b) applied shear stress τltm. The stresses fl and ft would then
is taken as 0.31lfc′ ( fc′ and lfc′ are in MPa). The tensile represent the average steel stresses at maximum load and are
stress-strain curve of concrete (σc1 -ε1 curve) in the 1-direc- calculated from Eq. (10) and (11) at the maximum load.
tion is somewhat different from the tensile stress-strain curve The test value of shear strain γ21 can be calculated from the
of concrete (σr-εr curve) in the r-direction. Since the tensile four measured strains, εl , εt , ε2, and ε1 by the following
stresses of concrete are very small compared to the compres- compatibility equation:
sive stresses of concrete, this small difference between σc1
and σr is ignored. γ 21 = – ( ε t – ε l csc 2α 2 + ( ε 1 – ε 2 ) cot 2α 2 (18)
(3) Average tensile stress-strain relationship of steel
embedded in concrete: and for the case of pure shear, α2 = 45 deg, the strain is
This relationship of fs vs. εs is expressed by a bilinear model:
γ 21 = – ( ε t – ε l ) (18a)
fs = Esεs εs ≤ εn (10a) or (11a)
At the maximum shear stress, the corresponding shear
strain γ21o is mainly affected by the severity of cracking and
ε 2 – α 2 /45 deg ⎞
f s = f y ⎛ 0.91 – 2B ) + ⎛ 0.02 + 0.25B ) ----s ⎛ 1 – ------------------------------ the difference of smeared steel stresses in the l- and t-direc-
⎝ ⎝ εy ⎝ 1000ρ ⎠
tions. The severity of cracking can be measured by the corre-
εs > εn (10b) or (11b) sponding tensile strain ε1o at maximum shear stress, while
the difference of smeared steel stresses in the l- and t-direc-
where tions can be evaluated by the parameter η defined in Eq. (8a).
Based on the test results of panels with a strain-control
2 – α 2 /45 deg ⎞ procedure (Zhang 1995), the empirical relationship for pure
ε n = ε y ( 0.93 – 2B ) ⎛ 1 – --------------------------------
- (12)
⎝ 1000ρ ⎠ shear was found to be

and γ 21o = – 0.85ε 1o ( 1 – η ) (19)

1 f cr ⎞ 1.5 SOLUTION METHOD


B = --- ⎛ ----- (13)
ρ ⎝ fy ⎠ The fixed-angle softened-truss model is considerably
more complicated than the rotating-angle softened-truss
Eqs. (10) and (11) are two identical equations, one for the model because the fixed-angle softened-truss model must
longitudinal steel and one for the transverse steel. The factor incorporate the constitutive law that relates the average
[(0.91 – 2B) + (0.02 + 0.25B)(εs /εy)] in Eq. (10b) or (11b) concrete shear stress τc21 to the average shear strain γ21 in
takes care of the averaging of steel stresses in the post-yield the 2-1 coordinate. The 12 governing equations, (1) to (11)
branch. The factor [1 – (2 – α2/45 deg)/1000ρ] takes into and (16), involve 15 unknown variables. These 15 unknown
account the “kinking” of reinforcing bars at the cracks. When variables include eight stresses (σl, σt, τlt, σc2, σc1, τc21, fl
α2 = 45 deg, this “kinking” factor is equal to (1 – /1000ρ). and ft), and six strains (εl, εt , γlt , ε2, ε1 and γ21), as well as
When α2 = 90 deg, it becomes unity. the material coefficient ζ. If three of these 15 variables are
given, the rest of the 12 variables can be determined mathe-
(4) Average shear stress-strain relationship of concrete in
matically by using the 12 governing equations.
2-1 coordinate:
For the case of pure shear studied here, the two applied
This relationship between τc21 and γ21 has been deter- normal stresses, σl and σt, are given as zero. The compres-
mined experimentally by Pang and Hsu (1996): sive strain ε2 is usually selected as the third given variable
because it varies monotonically from zero to maximum. For
γ 21 ⎞ 6
τ 21 = τ 21m 1 – ⎛ 1 – -------- each given value of ε2, the remaining 12 unknown variables
c c

- (16)
γ 21o ⎠ can be found by solving the 12 equations. The series of solu-
tions for various ε2 values will allow us to trace the loading
where τc21m and γ21o are the maximum shear stress of history by plotting the τlt vs. γlt curve.
cracked concrete and the corresponding shear strain, Notice that the τc21-γ21 relationship in Eq. (16) can be used
respectively. The power of 6 was obtained from tests only if τc21m and γ21o are known at the maximum shear stress
(Pang and Hsu 1996). τltm. Therefore, a method must be devised to find τc21m and

486 ACI Structural Journal/September-October 1997


Table 1—Mechanical properties and principal variables of test panels
Concrete Steel
fc′ εo l-direction t-direction
Panel MPa mm/mm Rebars ρl fly, MPa Rebars ρt fty, MPa η
B1 45.3 0.00215 15M@188mm 0.0120 463 10M@188mm 0.0060 445 0.480
B2 44.1 0.00235 20M@188mm 0.0179 447 15M@188mm 0.0120 463 0.690
B3 44.9 0.00215 20M@188mm 0.0179 447 10M@188mm 0.0239 445 0.332
B4 44.8 0.00205 25M@188mm 0.0298 470 10M@188mm 0.0359 445 0.189
B5 42.9 0.00220 25M@188mm 0.0298 470 15M@188mm 0.0120 463 0.394
B6 43.0 0.00220 25M@188mm 0.0298 470 20M@188mm 0.0179 447 0.570
HB1 66.5 0.00230 15M@188mm 0.0120 409 10M@188mm 0.0060 445 0.543
HB3 66.8 0.00240 20M@188mm 0.0179 447 10M@188mm 0.0060 445 0.332
HB4 62.9 0.00235 25M@188mm 0.0298 470 10M@188mm 0.0060 445 0.189
Note: All panels have a size of 1397 x 1397 x 178 mm (55 x 55 x 7 in.); ρl = Al/bsl, ρt = At/bst; η = ρt fty/flyρl

γ21o without using Eq. (16). This process to find τc21m and involve 15 unknown variables (σl, σt, τlt, σc2, σc1, τc21, fl, ft,
γ21o will be called the first phase of the solution procedure. εl, εt, γlt, ε2, ε1, γ21, and ζ). When three variables (σl and σt,
Once τc21m and γ21o are found, the process of solving the 12 and ε2) are given, the remaining 12 unknown variables are
governing equations, (1) to (11) and (16) will be called the solved by the 12 equations using an iteration procedure as
second phase of the solution procedure. shown by the flow chart in Fig. 2. Selecting a series of ε2
In order to formulate these two phases of solution proce- values and solving τlt for each ε2 value will give a τlt vs. ε2
dures, the following two useful equations are first derived curve, then search for a maximum shear stress τltm in the τlt vs.
based on the equilibrium equations. Adding Eqs. (1) and (2) ε2 relationship to obtain the corresponding τc21m and γ21o The
gives the summation of the smeared steel stresses in the l- step-by-step solution procedure of the first phase is explained
and t-directions as in detail in the Appendix.*

c c Second phase of solution procedure


ρl fl + ρt ft = ( σl + σt ) – ( σ2 + σ1 ) (20)
Once the maximum shear stress of concrete τc21m and the
shear strain at maximum load γ21o are determined, we can
Subtracting Eq. (2) from Eq. (1) gives the remainder of now solve the 12 unknown variables (τlt, σc2, σc1, τc21, fl, ft,
the smeared steel stresses in the l- and t-directions as εl, εt, γlt, ε1, γ21, and ζ) by the 12 governing equations, (3) to
(11), (16), (20) and (21), using an iteration procedure as
c c c
ρ l f l – ρ t f t = ( σ l – σ t ) – 2 ( σ 2 – σ 1 ) cos 2α 2 – 2τ 21 sin 2α 2 (21) shown by the flow chart in Fig. 3. Selecting a series of ε2
values will allow us to plot the σlt vs. γlt curve. The step-by-
Equations (20) and (21) will be used in the second phase of solu- step solution procedure of the second phase is also explained
tion procedure in lieu of Eqs. (1) and (2). Equations (20) and in the Appendix.
(17) will be used in the first phase of solution procedure.
In the first phase of the solution procedure, we need to EXAMPLE CALCULATIONS
device an approximate method for calculating γ21. This The above solution procedure will be used to analyze nine
method is obtained by assuming that Eq. (19) is applicable to panels, six in the B-series with normal strength concrete of 42
the entire loading history, rather than just the level of the MPa (6000 psi) (Pang and Hsu 1995) and three in the HB-
maximum load. Replacing the shear strain γ21o and the tensile series with medium-high strength concrete of 65 MPa (9500
strain ε1o at the maximum applied load by γ21 and ε1 gives: psi) (Zhang 1992). The test program for these nine panels are
given in Table 1. To illustrate the algorithm of solution
γ 21 = – 0.85ε 1 ( 1 – η ) (22) procedures, panel B1 will be analyzed in detail in this section.
Particular interest is placed in the stress and strain conditions
Eq. (22) will be used to calculate an approximate shear at the first yield of steel and at the ultimate load stage.
strain γ21 at every stage of loading in the first phase of the
solution procedure. Find τc21m and γ21o (first phase)
A simple and useful compatibility equation can also be The maximum shear stress of concrete in the 2-1 coordi-
obtained from the invariance of the trace of strain tensor (or nate τc21m and the corresponding shear strain γ21o for panel
by summing Eq. (4) and (5)): B1 can be found from the first phase of the solution proce-
dure. Selecting a series of compressive strains ε2 allows us
ε1 = εl + εt – ε2 (23) to plot the τlt-ε2 curve as shown in Fig. 4. The step-by-step
calculations are summarized in Table 2 for four typical
Eq. (23) will be used to check the assumed ε1 in the first strains of ε2 = –0.0002, –0.0003, –0.00034, and –0.00035.
phase of the solution procedure. Figure 4 and Table 2 show that the maximum shear stress τltm
occurs at a stress of 3.88 MPa (563 psi) with a corresponding
First phase of solution procedure strain ε2 = –0.00034. The concrete shear stress τc21m and
In the first phase of the solution procedure the maximum
concrete shear stress τltm is determined from the 12 governing *The Appendix is available in xerographic or similar form from ACI headquarters,
where it will be kept permanently on file, at a charge equal to the cost of reproduction
equations, (3) to (11), (17), (20) and (22). These 12 equations plus handling at time of request.

ACI Structural Journal/September-October 1997 487


Fig. 3—Flow chart of solution procedure for second phase.

Table 2—The first phase of solution procedure for


Fig. 2—Flow chart of solution procedure for first phase. panel B1
Variables Eqs. Calculated values
shear strain γ21o corresponding to τltm are then determined to –3
ε2 (10 ) selected — –0.20 –0.30 –0.34 –0.35
be –1.19 MPa (173 psi) and –0.0134, respectively (shown by
–3
bold types in Table 2). ε1 (10 )last assumed — 11.4 28.9 30.2 30.4
–3
γ21 (10 ) Eq. (22) –5.04 –12.77 –13.41 –13.47
Plot τlt vs. γlt curve (second phase) ζ Eq. (8) 0.266 0.172 0.168 0.168
Now that the maximum concrete shear stress τc21m and σc2 (MPa) Eq. (7) –6.95 –7.51 –7.57 –7.56
the corresponding shear strain γ21o are known, we can solve
the 12 unknown variables (τlt, σc2, σc1, τc21, fl, ft, εl, εt, γlt, σc1 (MPa) Eq. (9) 0.29 0.20 0.19 0.19
ε1, γ21, and ζ) by the 12 governing equations, (1) to (11) εl (10 ) –3 Eq. (4) 3.08 7.91 8.30 8.39
and (16) using the second phase of the solution procedure. εt (10 ) –3 Eq. (5) 8.12 20.1 21.7 21.8
The results of the calculations shown for ε2 values of –0.00010,
fl (MPa) Eq. (10) 379 401 403 403
–0.00012, -0.00019, and –0.00036 are summarized in
Table 3. A comparison of these four cases clearly illustrates ft (MPa) Eq. (11) 338 402 407 408
the trends of all the variables. ε2 = –0.00010 represents a (ρl fl + ρt ft )1 — 6.53 7.18 7.24 7.24
point before the yielding of both transverse steel and longi- (ρlfl + ρt ft )2 check Eq. (20) 6.66 7.31 7.37 7.37
tudinal steel. ε2 = –0.00012 is the point of first yield of the –3
ε1 (10 ) check Eq. (23) 11.4 28.3 30.3 30.5
transverse steel (εt = 0.00171), but before the yielding of
longitudinal steel. ε2 = –0.00019 is the second yield point τc21 (MPa) Eq. (17) –1.24 –1.19 –1.19 –1.19
of the longitudinal steel (εl = 0.00194). Finally, ε2 = –0.00036 γ21 (10–3) check Eq. (18) –5.04 –12.19 –13.40 –13.41
gives a point at which the maximum shear stress is obtained. τl t (MPa) Eq. (3) 3.62 3.83 3.88 3.87
The relationship between the shear stress τlt and shear –3
γl t (10 ) Eq. (6) –11.6 –29.2 –30.5 –30.8
strain γlt for panel B1 is shown in Fig. 5. The shear strains,

488 ACI Structural Journal/September-October 1997


Table 3—The second phase of solution procedure
for panel B1
Variables Eqs. Calculated values
–3
ε2 (10 ) selected — –0.10 –0.12 –0.19 –0.36
–3
ε1 (10 )last assumed — 2.31 2.90 9.70 30.7
–3
γ21 (10 ) last assumed — –0.49 –0.63 –4.77 –9.56
ζ Eq. (8) 0.503 0.466 0.286 0.167
σc2 (MPa) Eq. (7) –4.02 –4.86 –6.82 –7.56
σc1 (MPa) Eq. (9) 0.543 0.496 0.306 0.193
c
τ 21 (MPa) Eq. (16) –0.254 –0.318 –1.216 –1.263
–3
εl (10 ) Eq. (4) 0.86 1.07 1.94 10.4
Fig. 4—The τlt-ε2 curve for Panel B1 in the first phase of εt (10 ) –3 Eq. (5) 1.35 1.71 7.57 19.7
solution procedure (1 MPa = 145 psi). fl (MPa) Eq. (10) 167 206 373 413
ft (MPa) Eq. (11) 245 305 335 398
γlt, at first yield, at second yield and at ultimate are 0.00302, (ρlfl + ρtft )1 — 3.52 4.36 6.48 7.30
0.00989, and 0.0311, respectively. (ρlfl - ρtft)1 — 0.52 0.71 2.46 2.55
(ρl fl +ρt ft)2 check Eq. (20) 3.55 4.41 6.54 7.37
ANALYSIS OF TEST RESULTS (ρl fl – ρt ft)2 check Eq. (21) 0.52 0.71 2.44 2.53
Figure 5 shows the applied shear stress vs. shear strain
relationships (τlt-γlt curves) predicted by the fixed-angle τl t (MPa) Eq. (3) 2.31 2.66 3.56 3.88
softened-truss model for the six panels in B-series (Pang and γl t (10 ) –3 Eq. (6) 2.47 3.02 9.89 31.1
Hsu 1995, 1996) and the three panels in HB-series (Zhang
1992). A comparison of these nine theoretical predictions
(solid lines) with the corresponding experimental results Table 4—Comparison of theoretical and
(solid dots) shows excellent agreement. The predicted values experimental maximum applied shear stresses τltm
of the maximum applied shear stress τltm and the applied and applied shear stresses at first yield τlty
shear stress at first yield τlty are given in Table 4, and are
compared with their corresponding experimental values. It τlty (MPa) τltm (MPa)
can be seen that the agreements are consistently very good Test Test
Panel η Test Theory Theory Test Theory Theory
for all nine panels.
The agreement between theory and tests for the six panels B1 0.480 2.798 2.660 1.05 3.972 3.877 1.01
in B-series is expected, because data from these panels were B2 0.690 5.330 5.425 0.98 6.131 6.481 0.95
part of the information used in establishing the constitutive B3 0.332 3.455 3.344 1.03 4.367 4.711 0.93
laws of mild steel bars and normal strength concrete (42 MPa B4 0.189 4.202 4.091 1.03 5.076 4.808 1.06
or 6000 psi). However, the three panels with concrete strength B5 0.394 5.841 5.894 0.99 7.166 7.057 1.02
of 65 MPa (9500 psi) in the HB-series (Zhang 1992) were B6 0.570 7.200 7.217 1.00 9.152 9.067 1.01
tested after the development of constitutive laws for normal HB1 0.543 2.874 2.736 1.05 4.322 4.274 1.01
strength concrete. The excellent agreement between the theo- HB3 0.332 3.338 3.208 1.04 4.889 4.859 1.01
retical and experimental curves in Fig. 5 for these three panels HB4 0.189 4.191 4.230 0.99 5.334 5.113 1.04
could not be foreseen and are particularly meaningful. Recent
tests of 14 panels with concrete strength of 100 MPa (14,500
psi) (Zhang 1995) also confirm the adaptability of the fixed- for the rotating-angle softened-truss model, the fixed-angle
angle softened-truss model to very high strength concrete. softened-truss model is a much more powerful analytical
Pang and Hsu (1995) showed that the rotating-angle soft- tool for the nonlinear analysis of reinforced concrete
ened-truss model cannot predict the complete τlt-γlt curves if membrane elements. A comparison of these two limiting
the steel yield forces in the l- and t-directions differ greatly. ranges of applicability is shown by a failure modes diagram
For panel B3 with η (= ρt fty /ρl fly) value of 0.332, the in Fig. 6 for A- and B-series (Pang and Hsu 1995) with
predicted τlt-γlt curve terminated prematurely before the concrete strength of 42 MPa (6000 psi).
maximum applied shear stress was reached. For panel B4 It is interesting to examine this new limiting range, 0.2
with η value of 0.189, the predicted τlt-γlt curve terminated < η < 5, in terms of the balanced steel ratio ρb and the
even before the first yield of steel. Since the η parameters for minimum steel ratio ρmin for elements subjected to pure
these two panels are less than 0.4, a limiting range of 0.4 < η shear. The balanced steel ratio ρb is determined when the
< 2.5 was proposed for the rotating-angle softened-truss concrete compression strut crushes simultaneously with
model if full ductility of an element is desirable. In contrast, the yielding of steel; and the minimum steel ratio ρmin is
the analysis of these two panels, Fig. 5, using the fixed-angle obtained when the first tensile cracking of concrete occurs
softened-truss model does provide the entire τlt-γlt curves simultaneously with the yielding of steel. These two
with almost full ductility. limiting steel ratios have been derived by Zhang (1995) to
In view of the success of the fixed-angle softened-truss be ρb = 1.89 f c′ /fy and ρmin = 0.35 f c′ /fy. Since the ratio
model, we propose that the range of applicability be widened ρmin/ρb = 0.35/1.89 = 0.18 and the ratio ρb /ρmin = 1.89/
to 0.2 < η < 5. Compared to the 0.4 < η < 2.5 limiting range 0.35 = 5.4, the practical range of applicability is about

ACI Structural Journal/September-October 1997 489


Fig. 5—Shear stress vs. shear strain curves of panels in B-series and HB-series (1 MPa = 145 psi).

0.18 < η < 5.4. From this point of view the 0.2 < η < 5
range of applicability for fixed-angle softened-truss
model covers essentially the whole practical range.

CONCLUSIONS
1. A softened-truss model based on a fixed-angle is success-
fully developed to analyze the nonlinear behavior of reinforced
concrete membrane elements. This new smeared-crack model
has the capability to predict the “concrete contribution.”
2. The fixed-angle softened-truss model requires 12
governing equations derived from the two-dimensional equilib-
rium condition, the Mohr’s circular compatibility condition, and
the biaxial constitutive laws of concrete and steel bars. An
efficient algorithm is developed to solve these equations.
3. The range of applicability to ensure full ductility is
proposed to be 0.2 < η <5 for the fixed-angle softened-truss
model. This range is much greater than the 0.4 < η < 2.5
range previously determined for the rotating-angle softened-
truss model.

ACKNOWLEDGMENTS
This research was supported by National Science Foundation Grants No.
Fig. 6—Failure modes diagram for elements subjected to BCS-9213707 and MSS-9114543, as well as State of Texas Grant
proportional shear stresses (panels in Series A and B). 003652020-ATP. These generous supports are gratefully acknowledged.

490 ACI Structural Journal/September-October 1997


NOTATION Dei Poli, S.; Gambarova, P. G.; and Karakoc, C., 1987, “Aggregate Inter-
lock Role in R. C. Thin Webbed Beams in Shear,” Journal of Structural
1 ⎛ f cr ⎞
1.5 Engineering, ASCE, V. 113, No. 1, Jan., pp. 1-19.
B = a parameter defined as --- ⎜ -----
-⎟
ρ ⎝ fy ⎠ Dei Poli, S.; Di Prisco, M.; and Gambarova, P. G., 1990, “Stress Field in
Web of R. C. Thin-Webbed Beams Failing in Shear,” Journal of Structural
Ec = modulus of elasticity of concrete, taken as 3875 f c′ (fc′ and
Engineering, ASCE, V. 116, No. 9, Sept., pp. 2496-2515.
f c′ are in MPa), or 47,000 f c′ (fc′ and f c′ are in psi)
Elfgren, L., 1972, “Reinforced Concrete Beams Loaded in Combined
Es = modulus of elasticity of steel bars
Torsion, Bending and Shear,” Publication 71:3. Division of Concrete Struc-
fcr = cracking stress of concrete, taken as 0.31lf'c ( fc′ and lfc′ are in
MPa), or 3.75 f c′ ( fc′ and f c′ are in psi) tures, Chalmers University of Technology, Goteborg, Sweden, 249 pp.
Hsu, T. T. C., and Mo, Y. L., 1985a, “Softening of Concrete in Torsional
f 'c = maximum compressive strength of standard 6 in. by 12 in. (15.2
cm by 30.5 cm) concrete cylinder Members-Theory and Tests,” ACI JOURNAL, Proceedings, V. 82, No. 3,
fl, ft = average steel stresses in the l- and t-directions, respectively May-June, pp. 290-303.
fn = average stress of mild steel bars embedded in concrete at the Hsu, T. T. C., and Mo, Y. L., 1985b, “Softening of Concrete in Torsional
beginning of yielding, defined at the intersection of the two Members-Prestressed Concrete,” ACI JOURNAL, Proceedings, V. 82, No. 5,
straight lines in the bilinear model Sept.-Oct., pp. 603-615.
fs = average stress in mild steel bars. fs becomes fl or ft when applied Hsu, T. T. C., and Mo, Y. L., 1985c, “Softening of Concrete in Low-Rise
to longitudinal steel or transverse steel, respectively Shear Walls,” ACI JOURNAL, Proceedings V. 82, No. 6, Nov.-Dec. pp. 883-889.
fy = yield stress of bare mild steel bars, representing the local yield Hsu, T. T. C.; Mau, S. T.; and Chen, B., 1987, “A Theory on Shear
stress of 90 deg steel bars at cracks. fy becomes fly or fty when Transfer Strength of Reinforced Concrete,” ACI Structural Journal, V. 84,
applied to longitudinal steel or transverse steel, respectively No. 2, Mar.-Apr., pp. 149-160.
fy′ = average yield stress of mild steel bars embedded in concrete. It Hsu, T. T. C., 1988, “Softening Truss Model Theory for Shear and
is not a constant, but is a linear function of the average steel Torsion,” ACI Structural Journal, V. 85, No. 6, Nov.-Dec., pp. 624-635.
strain εs. fy′ becomes fly′ or fty′ when applied to longitudinal Hsu, T. T. C., 1991, “Nonlinear Analysis of Concrete Membrane
steel or transverse steel, respectively Elements,” ACI Structural Journal, V. 88, No. 5, Sept.-Oct., pp. 552-561.
α = rotating-angle between the principal compressive stress of post- Hsu, T. T. C., 1993, Unified Theory of Reinforced Concrete, CRC Press,
cracking concrete (d-direction) and the longitudinal steel bars Inc., Boca Raton, 336 pp.
(l-direction) Hsu, T. T. C., 1996, “Toward a Unified Nomenclature for Reinforced
α2 = fixed-angle between the applied principal compression stress Concrete Theory,” Journal of Structural Engineering, ASCE, V. 122, No. 3,
(2-direction) and the longitudinal steel bars (l-direction) Mar., pp. 275-283.
εo = concrete strain at maximum compressive strength of standard Kupfer, H., and Bulicek, H., 1992, “A Consistent Model for the Design
concrete cylinder of Shear Reinforcement in Slender Beams with I- or Box-Shaped Cross
ε2, ε1 = average normal strains in the 2- and 1-directions, respectively Section,” Concrete Shear in Earthquake, Hsu and Mau, eds., Elsevier
εl, εt = average normal strains in the l- and t-directions, respectively Science Publishers, Inc., London-New York, Jan., pp. 256-265.
(positive for tension) Lampert, P., and Thurlimann, B., 1968, “Torsion Tests of Reinforced
εn = average yield strain of mild steel bars embedded in concrete at Concrete Beams (Torsionsversuche an Stahlbetonbalken),” Bericht No. 6506-
the beginning of yielding; εn = εy(0.93 – 2B) 2, Institute fur Baustatik, ETH, Zurich, 101 pp.
εs = average strain in the mild steel bars embedded in concrete. εs Mau, S. T., and Hsu, T. T. C., 1986, “Shear Design and Analysis of Low-
becomes εl or εt, when applied to the longitudinal and trans- Rise Structural Walls,” ACI JOURNAL, Proceedings V. 83, No. 2, Mar.-Apr.,
verse steel, respectively
pp. 306-315.
εy = yield strain in bare mild steel bars; εy = fy /Es
Mau, S. T., and Hsu, T. T. C., 1987a, “Shear Behavior of Reinforced
γlt = average shear strains in the l-t coordinate (positive as shown in
Concrete Framed Wall Panels with Vertical Loads,” ACI Structural
Fig. 1 for τlt)
Journal, V. 84, No. 3, May-June, pp. 228-234.
γ21 = average shear strain in the 2-1 coordinate
Mau, S. T., and Hsu, T. T. C., 1987b, “Shear Strength Prediction for
γ21o = average shear strain in the 2-1 coordinate at maximum shear
Deep Beams with Web Reinforcement,” ACI Structural Journal, V. 84,
stress τc21m
No. 6, Nov.-Dec., pp. 513-523.
ρ = reinforcement steel ratio; ρ becomes ρl or ρt when applied to
Nielsen, M. P., 1967, “Om Forskydningsarmering i Jernbetonbjaelker,”
the longitudinal and transverse steel, respectively
(On Shear Reinforcement in Reinforced Concrete Beams), Bygn-
ρl, ρt = mild steel ratios in the l- and t-directions, respectively
ingsstatiske Meddelelser, V. 38, No. 2, pp. 33-58.
σl, σt = applied normal stresses in the l- and t-directions, respectively
(positive for tension) Pang, X. B., and Hsu, T. T. C., 1995, “Behavior of Reinforced Concrete
Membrane Elements in Shear,” ACI Structural Journal, V. 92, No. 6, Nov.-
σc2 , σc1 = average normal stresses of concrete in the 2- and 1-directions,
respectively Dec., pp. 665-679.
τlt = applied shear stress in the l-t coordinate (positive as shown in Pang, X. B., and Hsu, T. T. C., 1996, “Fixed-Angle Softened-Truss
Fig. 1(a)) Model for Reinforced Concrete,” ACI Structural Journal, V. 93, No. 2,
τltm = maximum applied shear stress in the l-t coordinate Mar.-Apr., pp. 197-207.
τc21 = average shear stress of concrete in the 2-1 coordinate Robinson, J. R., and Demorieux, J. M., 1968, “Essais de Traction-
τc21m = average maximum shear stress of concrete in the 2-1 coordinate Compression sur Modeles d'ame de Poutre en Beton Arme,” IRABA
ζ = softening coefficient of concrete in compression Report, Institut de Recherches Appliquees du Beton de L'ame, Part 1, June
1968, 44 pp., and Part 2, “Resistance Ultimate du Beton de L'ame de
Poutres en Double Te en Beton Arme,” 53 pp.
REFERENCES Vecchio, F., and Collins, M. P., 1981, “Stress-Strain Characteristics of
ASCE/ACI Committee 447, 1991, “Finite Element Analysis of Reinforced Reinforced Concrete in Pure Shear,” Final Report, IABSE Colloquium on
Concrete Structures II,” Proceedings of the International Workshop, Advanced Mechanics of Reinforced Concrete, Delft, 1981, pp. 211-225.
Columbia University, NY, June 2-5, J. Isenberg, ed., ASCE, New York, 1993. Vecchio, F., and Collins, M. P., 1982, “The Response of Reinforced
Belarbi, A., and Hsu, T. T. C., 1994, “Constitutive Laws of Concrete in Concrete to In- Plane Shear and Normal Stresses,” Publication No. 82-
Tension and Reinforcing Bars Stiffened by Concrete,” ACI Structural Journal, 03, Department of Civil Engineering, University of Toronto, Toronto,
V. 91, No. 4, July-Aug., pp. 465-474. 1982, 332 pp.
Belarbi, A., and Hsu, T. T. C., 1995, “Constitutive Laws of Softened Zhang, L. X., “Constitutive Laws of Reinforced Elements with Medium-
Concrete in Biaxial Tension-Compression,” ACI Structural Journal, V. 92, High Strength Concrete,” M. S. Thesis, Department of Civil and Environ-
No. 5, Sept.-Oct., pp. 562-573. mental Engineering, University of Houston, Houston, 1992, 214 pp.
Collins, M. P., 1973, “Torque-Twist Characteristics of Reinforced Zhang, L. X., “Constitutive Laws of Reinforced Elements with High
Concrete Beams,” Inelasticity and Non-Linearity in Structural Concrete, Strength Concrete,” Ph. D. Dissertation. Department of Civil and Environ-
Study No. 8, University of Waterloo Press, pp. 211-231. mental Engineering, University of Houston, Houston, 1995, 303 pp.

ACI Structural Journal/September-October 1997 491

Das könnte Ihnen auch gefallen