Вы находитесь на странице: 1из 11

Out-of-plane resistance of concrete masonry infilled panels

J . L. DAWEAND C. K. SEAH
Department of Civil Engineering, University of New Brunswick, Fredericton, N.B., Canada E3B 5A3

Received September 2, 1988 Revised manuscript accepted May 11, 1989 Nine large-scale concrete masonry infilled panels (3.6 X 2.8 m) were tested to destruction under uniformly distributed lateral pressure applied in small increments. Load-deformation curves of the infills and the enclosing steel frame showed linear elastic behaviour prior to first cracking. Nonlinear behaviour due to cracking and arching action of infill was prominent in the postcracking range. Parameters investigated experimentally included the effects of boundary supports, joint reinforcement, panel thicknesses, panel opening, and characteristics of construction. In parallel with the testing program, computer-aided analytical techniques were developed to predict the first crack and ultimate loads. First crack prediction was based on a finite element analysis for bending of thick plates, while ultimate load prediction was based on a yield-line technique modified to account for the arching action of infill confined within a flexible frame. Having been verified by comparison with test results, the postcracking analysis program was used to conduct a parametric study. It was found that infill compressive strength, panel dimensions, and frame rigidity have significant effect on ultimate loads. While central openings do not affect the ultimate strength, they do, however, reduce postcracking ductility. Key words: masonry, infill panel, steel frame, experimental, out-of-plane,behaviour, strength, arching, yield-line technique, cracking. Neuf panneaux de grandes dimensions avec remplissage de maqonnerie (3,6 X 2,8 m) ont CC soumis, dans le cadre d'essais de t destruction, A une pression lattrale rCpartie uniformCment.Les courbes charge-deformation des matkriaux de remplissage et du cadre en acier rkvklent un comportement Clastique lineaire avant la premikre fissuration. Le comportement non linCaire dii A la fissuration et A l'effet de voiite du remplissage Ctait evident durant la postfissuration. Les paramktres vCrifiCs expCrimentalement incluaient les effets des appuis de pourtour, de l'armature des joints, de 1'Cpaisseurdes panneaux, de l'ouverture des panneaux et des caractCristiquesde construction. Des techniques d'analyse informatisCes ont C ClaborCes en mime temps que le programme t C d'essai afin de prCdire la premikre fissuration et les charges ultimes. La prediction de la premiere fissuration Ctait basCe sur l'analyse des ClCments finis pour la flexion des plaques Cpaisses, alors que la prkdiction des charges ultimes Ctait basCe sur la technique de la ligne d'ClasticitC modifiCe pour tenir compte de l'effet de voiite du matCriau de remplissage entour6 d'un cadre flexible. Aprks comparaison avec les rksultats d'essai, le programme d'analyse de la postfissuration a et6 utilisC dans le cadre d'une Ctude paramCtrique. on a constate que la rksistance A la compression du matkriau de remplissage, les dimensions du panneau et la rigidit6 du cadre avaient un effet important sur les charges ultimes. Bien que les ouvertures centrales n'affectent pas la resistance ultime, elles rkduisent cependant la ductilitk de la postfissuration. Mots cle's : maqomerie, panneau de remplissage, cadre d'acier, expkrimental, hors-plan, comportement, resistance, effet de voiite, technique de la ligne d'elasticiti, fissuration. [Traduit par la revue]
Can. I. Civ. Eng. 16, 854-864 (1989)

Introduction
While masonry infilled structural steel or reinforced concrete frames are a common form of construction, there remains a general lack of conclusive research and design information on the lateral strength and behaviour of the panels themselves. For example, the Canadian code for masonry design for buildings (Canadian Standards Association 1984) contains no specific guidelines for the design of infilled panels. A critical review of current design methods conducted by Essawy and Drysdale (1987) reveals a great international diversity in design philosophies, ranging from working stress design based on elastic analysis to ultimate limit states design based on yield-line analysis. The majority of experiments conducted (Haseltine and Hodgkinson 1973; West and H o d m s o n 1976; West et al. 1979; Anderson 1976, 1985; Drysdale and Essawy 1988) to validate possible design methods also vary greatly. In most cases, test specimens were fabricated with well-defined, but somewhat unrealistic, boundary conditions. In the experimental works presented herein, all test specimens were fabricated by experienced masons using commonly used techniques. It is believed that these specimens can realistically reflect the behaviour of practical infilled panels. NOTE:Written discussion of this paper is welcomed and will be received by the Editor until April 30, 1990 (address inside front cover).

Tests of clay brick infilled panels conducted by Thomas (1953) showed considerable additional lateral capacity beyond first cracking due to arching of an infill confined within a frame. Hendry (1973) also recognized the strength enhancement due to arching action and West et al. (1973) have demonstrated the importance of frame-to-panel interface conditions on the lateral strength of brickwork panels. Theoretical treatment of the arching sh-ength of masonry panels was scanty. McDowell et al. (1956) proposed an analytical method to predict the lateral strength of one-way spanning brickwork beams with rigid supports due to arching. Anderson (1984) proposed a theory for predicting the behaviour of one-way spanning unreinforced masonry walls subjected to out-of-plane loading. Anderson's theory may also include the effects of shrinkage, initial boundary gaps, and abutment stiffness. Prior to the work presented herein, there has been no analytical method to determine cracking and ultimate capacity of a two-way spanning masonry infilled panel confined within a flexible steel frame. In the investigation presented herein, nine specimens, each consisting of a hollow concrete masonry block infill panel in a steel frame, were subjected to gradually increasing uniform normal pressure applied to the wall surface. A finite element elastic analysis was used to predict the response up to first cracking, while a yield-line technique, modified to account for the interaction of infill with a flexible frame, was used to predict

DAWE AND SEAH . I

855

W200x46

Panel infill

sections welded continuously along their flange tips and fastened to the laboratory strongfloor by anchor bolts a.t 1220 mrn intervals. The webs of the floor beam and roof beam were parallel to the plane of the panel and those of the columns were perpendicular to it. Figure 1b shows an end profile of the test setup. The masonry infill was loaded by inflating air bags enclosed by a reaction panel. Supports for the reaction panel were provided by a series of longitudinal angles and short adjusting angles bolted to the reaction frame. Air bag pressure was monitored with a water manometer and infill and frame deformations were monitored with dial gauges, with precisions of '0.1 kPa and '0.01 mm, respectively. Description of test specimens A typical test .panel consisted of ungrouted, unreinforced hollow concrete blocks laid up in a steel frame in running bond pattern with Type S mortar. ace-shell mortar bedding was used throughout. Table 1 summarizes the properties of the concrete block units used in this investigation. Type S mortar used was proportioned by volume using 1 part masonry cement, 1/ 2 part portland cement, and 44 sand. Water was added to achieve the required workability as determined by masons. All specimens were continuously kept moist by spraying the specimens with water on both faces for approximately 2 min each day for a period of 7 days. They were then left to cure in laboratory condition for an additional week before testing. Table 2 summarizes the important characteristics for each of the nine specimens. Specimens WE1 and WE2 were designed to evaluate the effect of bed joint reinforcement for infilled panels snugly fitted to the surrounding steel frame with mortar. Specimen WE3 was a dry stacked panel designed to assess the contribution of tensile bond strength of mortar joints in the infill. Effect of panel thicknesses was determined by specimens WE4 and WE5, while specimens WE6, WE7, and WE8 were used to
u

///////////////////////////////////

i t

4 T ~ r d a n sat 1200 m m

FIG. la. Test setup: front view.

the postcracking behaviour. Verified by comparisons with test results, the proposed analytical technique for ultimate load prediction was used to conduct a study of several design parameters. Based on this parametric study, expressions suitable for design purposes are proposed.

Experimental program
Test setup A typical test specimen consisted of a 3.6 m x 2.8 m masonry panel laid up in a steel frame. The steel frame members shown in Fig. 1a consisted of W250 x 58 columns base-supported on a rigid floor beam and connected to the top to a W200 X 46 roof beam. The rigid floor beam was fabricated from two W3 10 x 52

Test

Specimen Reaction Panel

1;

<( airbags enclosed

W250 x 50 Column

L 150x 150 x 10 Angles

Adjusting Angles

Column Brace

Out-of-plane Support

React ion Frame

-4

FIG. 1b. Test setup: end profile.

856

CAN. J. CIV. ENG . VOL. 16, 1989

TABLE Properties of concrete masonry units* 1. Nominal wall thickness (mm) Parameter

TABLE Auxiliary test results: material properties 3. Orientation Property Normal to Parallel to bed joint bed joint

Actual dimensions: Length (mm) Width (mm) Height (mm) Face shell (mm) Minimum web thickness (mm) % solid Mass density (kg/m3) Unit compressive strength (MPa) *Based on 10 test samples for each block type. TABLE Characteristics of test specimens 2.
-

Compressive strength Mean (MPa) No. of test specimens Coefficient of variation (%) Elastic modulus Mean (MPa) No. of test specimens Coefficient of variation (%) Modulus of rupture - Plain panel Mean (MPa) No. of test specimens Coefficient of variation (%) Modulus of rupture-Bed joint reinforcement Mean (MPa) No. of test specimens Coefficient of variation (%) Poisson's ratio Mean No. of test specimens Coefficient of variation (%) *No experimental data available. 0.21 45 22.3

Test specimen

Test specimen characteristics Thickness = 190 mm (190 X 190 X 390 mmunits). Trust type joint reinforcement at alternate courses. All sides mortared to frame members. Thickness = 190 mm (190 X 190 X 390 mm units). Plain masonry panel. All sides mortared to frame members. Thickness = 190 mm (190 X 190 X 390 mm units). Dry-stack panel. All four sides of panel mortared to frame members. Thickness = 140 mm (140 X 190 x 390 mm units). Plain masonry panel. All sides of panel mortared to frame members. Thickness = 90 mm (90 X 190 X 390 mm units). Plain masonry panel. Vertical edges restrained from slipping. Thickness = 190 mm (190 X 190 x 390 mm units). Plain masonry panel. Vertical edges restrained from slipping. 20 mm gap at roof beam to panel interface. Same as specimen WE6 except for provision of truss type joint reinforcement at alternate courses. Same as specimen WE4 except for provision of restraints against slipping on four sides of panel. Thickness = 190 mm (190 X 190 X 390 mm units). Plain masonry panel. Central window opening (1.6 m X 1.2 m). All sides restrained from slipping.

*
-

Additionally, crack patterns and cracking loads corresponding to the appearance of each new crack were recorded.
Auxiliay tests Quality control tests on mortar (Canadian Standards Association 1976) and concrete blocks (Canadian Standards Association 1985) were performed to ensure standards were met. Compressive strength and elastic modulus in directions parallel and normal to bed joints were determined by compression tests on two-unit high concrete block prisms. Flexural tensile strength normal to bed joints was determined using a bond wrench (Brown and Palm 1982). Test specimens consisted of two-unit high prisms with face-shell mortar bedding. A truss-type joint reinforement was embedded into the bed joint to evaluate the effect of such inclusion. Flexural tensile strength (modulus of rupture) parallel to a bed joint was evaluated using two-unit-long and two-unit-high test specimens. The test specimens were tested as beams with third-point loading and simple end supports. Effect of joint reinforcement was evaluated by specimens with truss-type joint reinforcement embedded in the bed joint. A summary of all auxiliary test results is given in Table 3.

determine the effect of boundary conditions. The effect of a panel with a central window opening was evaluated by specimen WE9.
Testing procedures Uniform pressure normal to the panel surface was applied gradually in increments of 0.2 kPa prior to first cracking and 0.4 kPa thereafter. At the end of each load increment, the pressure was maintained at a constant level to allow the system to stabilize before deflection readings were taken. Recorded readings consisted of the deflections of the infill at various locations and deformation of the columns at mid-height.

Experimental results A typical curve of uniform pressure versus deflection of the infill central point, CP, is shown in Fig. 2. CP does not necessarily correspond to the point of maximum deflection. The infill load-deflection behaviour can be decomposed into four stages. Stage I is characterized by linear elastic behaviour prior to initial cracking, while in stage 11, propagation of initial cracks and development of a yield-line failure mechanism occurs. In stage 111, arching of infill confined in the flexible steel frame causes the load to increase to a maximum above that predicted by standard yield-line analysis. In stage IV, the load drops off

DAWE AND SEAH : I

0 0 20 40 60
( mm

80

100

10 2

Deflection

FIG. 2. Typical load-deflection of infill.

Air -bag pressure

4 'P

!,

8 - Rotation

( a ) Leeward face
N = North
Column S = South Column

crushing of infill

-8 -C

Deflection

C ( mm

8 1

-20 -10
Rotation
(a

0 1 0 20 0.0001 rad. I

(b) Windward

face

FIG. 3. Column deformations.

FIG. 4. Typical crack patterns.

due to gradual crushing of masonry at the yield lines and boundaries and, finally, ultimate collapse occurs. Load-deformation behaviour of the columns illustrated in Fig. 3 indicates relatively small deformation prior to first cracking. The abrupt increase in the rates of lateral deflection and rotation occurs at the onset of first cracking. It is believed that the infill resists applied lateral loads by flexural action prior to first cracking, while arching action is the main load-resisting

mechanism in the postcracking range. The column load-deformation behaviour gives a good indication of the level of lateral in-plane thrust developed in the infill due to arching. A typical crack pattern as shown in Fig. 4 exhibits the characteristics of yield-line mechanisms for reinforced concrete slabs. The major crack pattern was fully formed at the end of stage 11, while subsequent increase in loading increased the crack width on the leeward face and the depth of crushing on

858

CAN. J. CIV. ENG. VOL. 16, 1989

the windward face. Occasionally, some secondary cracking (remote from the major yield lines) also occurred, but no permanent load drop was associated with these. Ultimate load was reached when a critical limit of crushing of concrete blocks at the crack lines and at the interface with the confining frame was reached. Specimen WE3 consisted of a dry-stacked infilled panel designed to qualitatively evaluate the effect of arching. The entire load-deflection curve exhibits stage I11 type behaviour (Seah 1988). Its ultimate load, although lower than the other panels, was nevertheless significant.
u

Ril -

1
I

First iteration

.-

i \.J
Ri2

'

12:

2 0
M 12:

1 \. -----+----.

j !
:
I

:.
I

iteration

First crack load prediction Elastic behaviour prior to thejrst cracking A finite element analysis using a thick plate element incorporating flexural and transverse shear deformations (Hinton and Owen 1984; Cope and Clark 1984) was adapted herein. The quadrilateral element has x- and y-rotational and transverse translational degrees of freedom at each of the four comer nodes, four mid-side nodes, and one-mid-element node. The masonry panel was treated as a voided plate, (i.e., a void sandwiched between two face shells). Plate rigidity factors proposed by Basu and Dawson (1978) were modified to account for web misalignment typical of face-shell bedded hollow concrete blocks in running bond.
Failure criteria The four failure criteria relating to masonry tensile cracking as developed by Essawy and Drysdale (1986) were used in the analysis. These criteria correspond to (1) debonding along bed joints, (2) simultaneous bond failure through head joints and units, (3) a stepped failure through head and bed joints, and (4) splitting directly through the units. These proposed criteria were rearranged as functional relations of the form [l] Fi(u,r)=O where u represents a set of applied stresses, r represents strength parameters, and i = 1, 2, 3, 4 for the four different failure modes. Each relation is satisfied identically in the critical situation, while in a noncritical condition, a residual, Ri, corresponding to an amount by which the critical failed to be met, can be determined. Further details of the implementation of these failure criteria are available elsewhere ( ~ e a h 1988). Analytical procedure In the analytical procedure, an arbitrary assumed applied pressure and the weight of the panel are considered in calculating shear stresses and extreme fibre stresses normal and parallel to bed joints. A mesh size that guarantees satisfactory convergence is used. Once the four failure criteria are checked and the residuals calculated, the initial applied pressure is incremented by a small amount and the stress analysis repeated to obtain a new set of residuals for each failure mode. When two sets of residuals are obtained, the true critical load is approached iteratively using a secant method as illustrated in Fig. 5. In this figure, residuals Ri for each failure mode i are plotted against applied pressure. A solution is found when Riis less than or equal to a prescribed minimum representing the precision of the technique. This is repeated for all four failure modes, and the failure mode giving the lowest critical load is taken as the governing mode of failure. First crack load, mode of tensile cracking, and the location of first cracking can be predicted using this technique.

\. ,.

P2

P3

4 '

'5

'cr

Applied Pressure FIG. 5. Secant method for determining critical load. TABLE4. First crack loads First crack load ( H a ) Test specimen Experiment Theory R *

Failure mode

WE1 WE2 WE3 WE4 WE5 WE6 WE7

2.6 3.8
-

2.9 3.4

1.12 Debonding of bed joint 0.89 Debonding of bed joint


-

2.4 2.2 5.2 6.8 4.8 3.2

2.2 1.8 4.1 7.2 3.7 3.4

WE8
WE^

0.92 Debonding of bed joint 0.82 Debonding of bed joint 0.79 Splitting through head joints and blocks 1.06 Splitting through head joints and blocks 0.77 Splitting through headjoints and blocks 1.06 Stepped failure through head and bed joints

* R = theoretical value divided by experimental value.

Analysis of test results The predicted first crack loads and failure modes are compared with those obtained experimentally in Table 4. The ratios of predicted to experimental values vary from 0.77 to 1.12 with a mean of 0.93 and a coefficient of variation of 15%. First cracking appeared as bond failure between mortar and units along bed joints when vertical edges of the panel along the column webs were not restrained against slippage. Splitting through head joints and blocks was the predominant cracking mode when such slippage was prevented. Since tensile bond strength of masonry construction is highly variable and cannot be determined precisely, the accuracy of first crack load predictions similarly suffers random variability.

Postcracking analysis Arching action of injlled panels Many researchers (Park 1964; Desayi and Kulkarni 1977; Breastrup 1980) have employed a yield-line technique to analyze

DAWE AND SEAH : I

859

I1

Type 1 S r p ti

( a ) Horizontal strips

4 Lfracture
FIG. 7. Type I strip in a deformed configuration.

-------:
depth of contact strain ( a ) Ultimate strain not exceeded

k l fk k
stress

cult =0.0015 ( from test resutts )

( b ) Vertical strips
FIG. 6. Subdivision into horizontal and vertical strips.

-----,? p

e p t h of crushing

reinforced concrete slabs with rigid edge restraints. In the analytical procedure presented herein, the conventional yieldline method is modified for the analysis of voided concrete masonry panels including arching action within flexible steel frame boundaries. To facilitate computations, the panel is subdivided into a series of horizontal and vertical strips as shown in Fig. 6 (Park 1964). The error involved in this approximation can be minimized by increasing the number of strips. A typical Type I strip is shown in plan view in its deformed configuration in Fig. 7. In this figure, E is the compressive strain in the strip, q is the applied uniformly distributed pressure, tl and t2 are outward displacements of the columns at the strip elevation, c is the depth of contact at the fracture location and at strip ends in contact with column webs, and p, a fraction between 0 and 1, defines the location of the fracture line in a strip. Over regions of contact defined by c in Fig. 7, a rectangular plastic stress block as used in reinforced concrete analysis is adopted (Park 1964). The strain and stress distributions on the face shell for various strain ranges are shown in Fig. 8. The masonrv material is conse~ativelvassumed ineffective when loaded beyond the ultimate strain, ,I,. An average ultimate strain of 0.0015, as determined experimentally from prism

--- 4

depth of contact

st rain

stress

(b) Ultimate strain exceeded

FIG. 8. Stress block at region of contact for various strain ranges. (tf = face-shell thickness.)

compressive tests, was used in the analysis of all large-scale specimens. Based on the assumed stress distribution as well as geometric and equilibrium considerations, the following relationship can be established for a single horizontal strip (Seah 1988): [21 {Wan $1 + tan $2) + (klk2f6LIAEm) x [p cos + (1 - p) cos $21)~= d(tan $1 + tan $2) - L{1 - p(c0s $1 - cos $2) - COS $2) - (tl + t2 + G1 + G2)

where, as shown in Figs. 7 and 8, L is the length, d is the thickness of a strip, kl and k2 are stress block factors each equal

CAN. J. CIV. ENG. VOL. 16. 1989

; ; :r ', \ I I
1 1
I

11 1 /I I 11 I I 11

F~ axial deformation of beam

N;

N;

bending of column twisting of column

Strip i
N; +

! I
1 1
1 : ;

N;
Strip j

1 1 1

I 1

;I

'

t- F~~
Nj = 1 unit

FIG. 9. Evaluation of flexibility coefficients.

to 0.85 (Park 1964), A is the cross-sectional area of a strip, f; is the masonry compressive strength, Em is the modulus of elasticity of the infill, and +2 are rotations of strip segments = (Fig. 7), z is the strip deflection at the fracture line, tan z/pL, and tan +2 = z/[(l - p)L]. Also included in the above expression are G I and G2,the initial boundary gaps between the panel and the frame due to shrinkage or construction errors. For a given horizontal strip, tl and t2 are dependent on the flexibilities of the frame members and in-plane thrusts developed in horizontal strips. Typical strips, i and j, with in-plane thrusts, Ni and Nj respectively, are shown in Fig. 9. The total flexibility at an end of strip i can be expressed as [3] F.. = F a . + F b + F ! . 1 J rl rl I J where superscripts a, b, and t indicate flexibility contribution due to axial elongation of the roof beam, flexural deformation of the column, and torsional deflection of the column, respectively. The flexibility coefficient, Fij, is the total frame displacement at an end of strip i due to a unit thrust in strip j. These quantities have been evaluated elsewhere (Seah 1988) for both horizontal and vertical strips. Using these flexibility coefficients, t , and t2 for use in [2] for a horizontal strip i are computed as follows:

( a ) actual system

( b ) equivalent system

( c ) equivalent system with bed joint reinforcement

FIG. 10. In-plane thrust developed in a strip due to arching and its
equivalent actions.

where n is the number of strips and F ; and Ff are the flexibility coefficients for the left and the right side of strip i respectively. A similar procedure is followed for vertical strips. Solution procedure A yield-line pattern consistent with boundary conditions is first established. By assuming rigid plate rotation within yield-line boundaries, the entire deflected configuration is defined by a given lateral deflection, z, at a convenient location on the selected yield-line pattern. An iterative technique, similar to a successive displacement technique used for solving large systems of equations (Al-Khafaji and Tooley 1986), is used to find the applied load for a given deflected configuration. The solution procedure is as follows: 1. Axial thrusts for all strips and frame displacements at all strip locations are set to zero and a value of z is assumed. 2 . For strip i , the depth of contact and axial thrust are calculated based on the assumptions in step 1.

3. Boundary displacements in all previous strip locations due to the force computed in step 2 are calculated from [4] and [5]. 4. Boundary displacements computed in step 3 change axial thrusts for all strips, and therefore the axial thrusts of all previous strips are updated based on the most recent total boundary displacement computed in step 3. 5. Steps 3 and 4 are repeated until no significant change in axial thrust of strips occurs between two successive iterations. The axial thrust in strip i + 1 is then computed and the procedure repeated until all strips have been considered. 6. By equating the external virtual work of the applied pressure to the total internal virtual work at the yield lines for all strips, the applied load is computed for the value of a assumed in step 1. Internal virtual work of a typical strip can be computed by considering the statically equivalent system of forces as shown in Fig. 10. This results in

DAWE AND SEAH : I

TABLE Ultimate loads 5. Ultimate load (kPa) Test specimen Experiment Theory R* Remarks Slippage at roof beam to panel interface; two-way arching at ultimate. Slippage at roof beam to panel interface; two-way arching. Dry-stacked panel; two-way arching.

WE1 WE2 WE3 WE4

22.3 19.2 7.8 11.2

21.8 (21.8) 19.9 (19.3)


-

0.98 (0.98) 1.04 (1.01)


-

11.7 (10.4) 8.9 (8.6) 10.7 (9.9) 13.8 (13.O) 13.5 (13.5) 17.4 (17.4)

1.04 (0.93) 1.14 (1.10) 1.O1 (0.93) 0.94 (0.88) 1.O1 (1.01) 1.OO (1.00)

140 mm panel; slippage at column webs at low load; slippage at room beam to panel interface at ultimate; two-way arching. 90 mm panel; no slippage at frame to infill interface; two-way arching. 20 mm top gap; no interface slippage; one-way arching. Joint reinforcement; 20 mm top gap; no interface slippage; one-way arching. 140 mm panel; no interface slippage; slippage of upper bed joints at ultimate; two-way arching. 1.6 m x 1.2 m central window opening; no interface slippage; slippage of upper bed joints at ultimate; two-way arching.

WE5 WE6 WE7 WE8 WE9

7.8 10.6 14.7 13.4 17.4

NOTE: Values enclosed within parentheses are theoretical minimum collapse loads based on failure patterns obtained by trial and error. *R = theoretical value divided by experimental value.

.
where A is the virtual displacement at a fracture line, M and M' are resisting moments at a boundary and a yield line, respectively, N is the axial thrust, p is a fraction between 0 and 1, and L is the total length of a strip. For a strip with joint reinforcement, the internal work is approximated as
Crack pattern
i

'\ .

(Spec.WE7)
1

0 j 0

20

where M, = A,Fyd, (moment resistance of a cross section due to reinforcement), A, is the sectional steel area of joint reinforcement, Fy is the yield strength of steel, and d, is the moment a m of the steel wire. r The external work for a strip of width w, is given by [81 Wext = qLwsAI2 where q is the external applied pressure and L is the length of a strip. The load resisted by a fractured panel at an assumed deflection, z , can be obtained by equating
[91

80 L O 60 Deflection ( m m I

100

120

FIG. 11. Comparison of a typical theoretical and experimental load-deflection response of infill.

dures. This process is repeated until the entire load-deflection response, including an unloading portion, is obtained.
Comparison with test results Predicted ultimate loads are compared with the corresponding experimental values in Table 5. Ratios of predicted to experimental values vary from 0.94 to 1.14 with a mean of 1.02 and a coefficient of variation of 6%. A typical comparison of theoretical and experimental load-deflection curves is shown in Fig. 1 1 , where applied pressure is plotted against mid-panel deflection. Actual observed failure patterns were used in the analysis and, in most cases, excellent correlation between experimental and predicted behaviour was obtained. In cases where failure patterns are unknown, patterns that yield minimum collapse loads may be found by trial-and-error procedures. Results of such analyses are shown in parentheses in Table 5.

Wint

Wext

where the summation is performed for all horizontal and vertical strips. Shear forces have no net work component and therefore are neglected in [6]-[8]. The yield-line deflection, z , is next incremented and the corresponding applied pressure computed by the above proce-

862

CAN. 1. CIV. ENG. VOL. 16, 1989

Parametric studies and experimental observations Effects of bed joint reinforcement Specimens WE1 and WE7 incorporated 190 mm thick panels with bed joint reinforcement placed at alternate courses. Specimens WE2 and WE6 were identical to specimens WE1 and WE7 respectively, except for the omission of bed joint reinforcement. Both specimens without joint reinforcement failed in a sudden, explosive manner, while specimens with joint reinforcement showed considerable ductility for the entire loading history. Specimens WE6 and WE7 were constructed with a 20 mm gap between roof beam and top edge on panel. Since the panels were restrained along the other three edges, the horizontal span was mobilized in resisting applied pressure. Comparison of specimens WE6 and WE7 shows that inclusion of joint reinforcement results in a higher first crack load. This can be attributed to the higher modulus of rupture parallel to bed joints for panels with joint reinforcement (Table 3). The two vertical edges of specimens WE1 and WE2 were not prevented from slipping at the column wed-to-infill interface, and hence bending in the vertical span caused debonding failure of bed joints at relatively low loads. Inclusion of bed joint reinforcement had little effect on first crack loads of these two specimens.
Effects of boundary supports Slippage at panel-to-column interfaces effectively causes the vertical span to resist the applied loading. This results in higher stresses normal to the bed joint and causes debonding failure at substantially lower first crack loads. Slippage may be prevented by using panel-to-frame ties or by filling the gaps between panel and column flanges with mortar or wooden strips. Higher first crack loads, effected by the latter type of construction, were obtained for specimens WE6, WE7, and WE8. Effects of lateral edge restraint on ultimate load are indicated by the load-deflection curves shown in Fig. 12 for specimen WE4. This specimen was constructed with 140 rnm thick hollow concrete blocks, and a 100 mrn gap between the panel and column flanges was left unfilled. Failure mode "a" indicated in Fig. 12 was eminent in the initial stage of loading. When the leeward face of the panel was restrained from further slippage by the column flanges, failure mode "b" ensued and substantial increase in load capacity was obtained. Figure 13 shows theoretical load-deflection curves of four panels with different boundary conditions. Case 1 is represented by panels supported in a nonslip mode at the top and bottom. Case 2 consists of panels supported along three sides and free at the top. Cases 1 and 2 allow only one-way arching to develop and the ultimate loads are therefore lower. Case 3 is represented by panels supported on three sides, but the absence of a gap at the top boundary allows arching action to develop in the vertical span. Slippage at the top interface, however, is not restricted. This boundary resulted in higher arching strength as compared with the two previous cases. A panel with all four sides supported, as shown by Case 4 of Fig. 13, results in a substantially higher ultimate load. Effects of panel thickness and panel aspect ratio Thick panels are more effective in developing arching action and, as a result, the ultimate loads are higher. However, this effect is diminished very rapidly with increasing panel length. Results of a parametric study of these effects are shown in Fig. 14. Panels of three different types of block enclosed in a steel

0 0
40

80
Deflection
(

I20

160

mm

FIG. 12. Load-deflection curves for specimen WE4: effect of boundary supports.

-. , c-. --. I ,

_-

i
/
t

I '

.. .-4--.. -conditions frame interface gap

Boundary

/I

/
;

I '

2
interface slippage

,
I

I I I I

4
/.A

crack pattern

/
I I

/.-.-.-.-.-.-

. '

--. -.
50 100
(

\. I 3 . . .
1 -

0 0 150

Deflection

mm

FIG. 13. Effect of boundary conditions on ultimate loads.

frame, with a stiffness comparable to that of the test frames used herein, were used in this study. Theoretical ultimate loads are plotted on the vertical axis and the corresponding aspect ratios are plotted on the horizontal axis. The ultimate strength due to arching diminishes rapidly with decreasing panel thicknesses, and for a thin panel with a large aspect ratio, arching strength is insignificant.

Effects of frame rigidity Figure 15 shows the results of a parametric study, conducted for a 190 mm thick panel restrained in a steel frame along the floor beam and both columns. Ultimate loads increase parabolically with in-plane flexural stiffness of the columns. However, the rate of increase in ultimate load decreases rapidly with increasing frame flexural stiffness. An increase in torsional stiffness of the columns was also found to increase ultimate load, but only for frames with relatively high flexural stiffness. Effects of panel opening A 1.6 m X 1.2 m opening was placed at the center of specimen WE9. The opening was covered with a stiffened

DAWE AND SEAH : I

863

(a) For a panel supported on three sides and free at the top, [lo] where
quit =

800(fk)0.75 t 2 a / ~ 2 . 5

(b) For a panel supported on four sides,

+ [ l 11 quit = 800(f &)0.75t2{a/~2.5P/H'.~} where

0
1.O

1.2

1.4 16 . Aspect Ratio ( L / H 1

1.8

2.0

FIG. 14. Effect of panel aspect ratio and panel thicknesses on ultimate loads. ( t = panel thickness.)

J = L o ~ =m m L ! ~

E.200 GPa G= 77 GPa

Column Moment of Inertia ( 8 1 )

FIG. 15. Effect of frame stiffness on ultimate loads.

plywood sheathing during the test. Load acting on the closed window was therefore transferred to the four sides of the opening. Test results indicated no significant decrease in ultimate load as a result of the opening, but postultimate ductility was reduced somewhat. The postultimate ductility of masonary panels infilled in steel frames may be attributed to the ability of central strips to resist part of the applied loading when crushing of perimeter strips occurs. A central opening in a panel effectively destroys the ability of central strips to resist load and, as a result, postultimate ductility is reduced. Panel deflections at ultimate load were significantly lower than for specimens without openings.

In [lo] and [ l l ] , quit is the ultimate load (kPa), t is the panel thickness (mm), L is the panel length (mm), H is the panel height (mm), E and G are Young's modulus and shear modulus of the frame members respectively, I, and Ib are moments of inertia of columns and beam respectively, and Jc and Jb are torsional constants of columns and beams respectively. The above equations are applicable to hollow concrete block panel infills within a pinned frame. Application of these equations to infills in moment-resisting frames may result in conservative estimates of ultimate loads. Similarly, for infills sandwiched between two adjacent panels as frequently occurs in a multi-bay infilled frame system, the upper limits of a and P of [lo] and [ l l ] may be used for a conservative estimate of collapse loads. The upper limits of a and P were established based on a results of a parametric study conducted for infilled panels enclosed within infinitely stiff frames. It should be noted that quit is the resistance due to arching action of the panel as it deflects. At low loads, lateral restraints preventing the panel from moving out of the frame are required to enable arching action to begin. Frame members are assumed to exhibit adequate ductility and strength so that failure of frame members may not occur before failure of the infill. For infill strengths and thicknesses comparable to those used in the test program presented herein, frame failure would not be a major concern. The above relations are applicable for ultimate load prediction due to arching. Elastic analysis and failure criteria proposed by Essaway and Drysdale (1986) and as implemented herein can be used to predict first crack loads.

Design recommendations An extensive parametric study was conducted to evaluate the effect on ultimate load of several parameters. The more important parameters identified were panel thickness, infill strength, boundary conditions, frame rigidity, and panel dimensions. Based on the results of this study, the following empirical relations suitable for design are proposed:

Conclusions As a result of this investigation, the following conclusions have been reached: 1. Prior to first major cracking, the main lateral load resisting mechanism for infilled panels is by flexural action, and in the postcracking range it is by arching action. 2. Elastic analysis combined with the failure theories of Essawy and Drysdale as implemented herein may be used to predict first crack loads. 3. Inclusion of arching action in yield-line analysis improves predicted capacities over that obtained by the conventional yield-line technique. 4. Deformations of a flexible frame should be included in the analysis to reflect the true behaviour of a system. 5. Ultimate loads increase parabolically with increasing panel thickness, but decrease with increasing panel length and height.

864

CAN. J. CIV. ENG. VOL. 16, 1989

6. The effectiveness of central strips is not as significant as that of perimeter strips in developing arching action. Thus, a relatively small central opening does not reduce the arching strength significantly.
Acknowledgements
The authors would like to thank L. E. Shaw Ltd. for providing financial support for this research and to Mr. George Forsyth of the Atlantic Masonry Association for his support and encouragement.

AL-KHAFAJI, W., and TOOLEY, R. 1986. Numerical methods in A. J. engineering practice. Holt, Rinehart and Winston, Inc., New York, NY. pp. 118-120. ANDERSON, 1976. Lateral loading tests on concrete block walls. C. Structural Engineer, Part 2,54(7): 239-246. 1984. Arching action in transverse laterally loaded masonry wall panels. Structural Engineer, 62B(1): 12-23. 1985. Test on walls subjected to uniform lateral loading and edge loading. Proceedings of the Seventh International Brick Masonry Conference, Melbourne, Australia, pp. 889-899. BASU, K., ~ ~ ~ D A W J. M.N , A. S O 1978. Orthotopic sandwichedplates - Part 1: dynamic relaxation treatment, Part 2: analysis and application to multicell and voided bridge decks. Proceedings, the Institution of Civil Engineers, Supplementary, pp. 87-1 15. BREASTRUP, W. 1980. Dome effect in RC slabs: rigid plastic M. analysis. ASCE Journal of the Structural Division, 106(ST6): 1237- 1253. BROWN, H., and PALM,B. D. 1982. Flexural strength of brick R. masonry using the bond wrench. Proceedings, Second North American Masonry Conference, University of Maryland, College Park, MD, pp. 1.1-1.15. CANADIAN STANDARDS ASSOCIATION. Mortar and grout for unit 1976. masonry. National Standard of Canada CAN3-A179 M76, Rexdale, Ont. 1984. Masonry design for buildings. National Standard of Canada CAN3-S304 M84, Rexdale, Ont. 1985. Standard for concrete masonry units. National Standard of Canada CAN3-A 165 M85, Rexdale, Ont . COPE,R. J., and CLARK, A. 1984. Concrete slabs: Analysis and L. design. Elsevier Applied Science Publishers, New York, NY. DESAYI, and KULKARNI, B. 1977. Load-deflection behaviour of P., A. restrained R/C slabs. ASCE Journal of the Structural Division, 103(ST2): 405-419. R. A. DRYSDALE, G., and ESSAWY, S. 1988.Out-of-plane bending of concrete block walls. ASCE Journal of the Structural Division, 114(ST1): 121-133. ESSAW,A., and DRYSDALE, G. 1986. Microscopic failure criterion R. for masonry assemblages. Proceedings, Fourth Canadian Masonry Symposium, Fredericton, N.B., pp. 263-277. 1987. Evaluation of available design methods for masonry walls subject to out-of-plane loading. Proceedings, Fourth North American Masonry Conference, University of California, Los Angeles, CA, pp 32.1-32.18. HASELTINE, A., and HODGKINSON, R. 1973. Wind effects upon B. H. brick panels walls - design information. Proceedings, Third International Brick Masonry Conference, Essen, Germany, pp. 399-406. HASELTINE, A., WEST,H. W. H., and TUTT,J. N. 1977. Design of B. walls to resist lateral loads. Structural Engineer, Part 2, 55(10): 422-430.

HENDRY, W. 1973. The lateral strength of umeinforced brickwork. A. Structural Engineer, 51(2): 43-50. HINTON, and OWEN,D. R. J. 1984. Finite element software for E., plates and shells. Pineridge Press Limited, Swansea, United Kingdom. MCD&ELL, E. L., MCKEE,K. E., and SEVEN, 1956. Arching E. action theory of masonry walls. ASCE Journal of the Structural Division, 82(ST2): 915.1-915.18. PARK, 1964. Ultimate strength of rectangular concrete slabs under R. short-term uniform loading with edges restrained against lateral movements. Proceedings, the Institution of Civil Engineers, 28: 125-150. SEAH,C . K. 1988. Out-of-plane behaviour of concrete masonry infilled panels. M.Sc.E. thesis, Department of Civil Engineering, University of New Brunswick, Fredericton, N.B. THOMAS, G. 1953. The strength of brickwork. Structural Engineer, F. Part 2,31: 35-41. WEST, H. W. H., and HODGKINSON, R. 1976. The resistance of H. brickwork to lateral forces - a review of work at BCRA. First CanadianMasonry Symposium, The University of Calgary, Calgary, Alta., pp. 275-291. WEST, H. W. H., HODGKINSON, R., and WEBB,W. F. 1973. H. Lateral loading test on walls with different boundary conditions. Proceedings, Third International Brick Masonry Conference, Essen, Germany, pp. 180-186. . WEST,H. W. H., HODGKINSON, R., and HASELTINE, A. 1977. H. B. The resistance of brickwork to lateral loading. Structural Engineer, Part2,55(10): 411-421.

List of symbols
cross-sectional area of a strip depth of contact at fracture line strip thickness elastic modulus of frame members elastic modulus of masonry infill failure criteria function flexibility coefficient masonry compressive strength shear modulus of frame members initial boundary gaps panel height moment of inertia of frame members torsional constants of frame members stress block factors panel length resisting moments in-plane thrusts ultimate lateral pressure residual of failure criteria function masonry panel thickness lateral boundary displacements external work internal work width of a strip deflection of fracture line virtual displacement compressive strain ultimate strain a fraction between 0 and 1 strip rotations

Вам также может понравиться