548

SHEAR STRENGTH OF MASONRY WALLS

J.R. RIDDINGTON, B.Sc . , Ph . D. , C.Eng., MICE University of Sussex, Brighton BN1 9QT, England

M.Z . GHAZALI, M.Sc, Ph.D . , P.Eng, MIEM Universiti Teknologi Malaysia,

54100 Kuala Lumpur

ABSTRACT

This pape r presents some experimental resul ts from a research programme which is attempting to develop a fundamental theory for shear failure in masonry . The results from a series of brick triplet tests with precompression stress leveIs up to 2 N/mm' and from a series with precompression stress leveIs up to 7 N/ mm' were reported earlier. This test work resulted in a general hypothesis for shear failure being proposed in which it was suggested that the basic mo de of failure changes from joint slip to mortar tensile failure as precompression increases. As the triplet has only two mortar layers and is always loaded symmetrically, the failure pattern can always be predicted. Consequently, there was considered to be a need to investigate the behaviour of larger masonry structures. In this paper experimental results from small wall samples are reported, and these are shown to support the previously presented hypothesis. The average shear stresses in the walls at failure were found to be Iower than the stresses in the corresponding triplet samples, but this was to be expected since the walls were subjected to bending as well as shear Ioading.

INTRODUCTION .

In order to understand better the mechanics of shear wall behaviour, while

at the same time avoiding expensive full scale tests, many investigations

[1 - 7] have been carried out into the strength of masonry bedjoints. The

forms of test sample used were the triplet prism formed from three bricks,

the couplet prism formed from two bricks, the small square shear panel ar

small discs cut out from a larger brickwork prism . Results from triplet

and couplet tests have always been presented as average stresses acting on

the joints. Experimental results from triplet samples subjected to

549

precompression, shear and bending have led to a hypothesis being proposed

[8,9] that the failure of masonry in shear is caused by either shear slip

at the brick-mortar interface ar by diagonal tensile failure in the mortar

layer, and that such fa1lure 1s initiated when local failure cri teria are

violated. As the triplet sample has only two mortar layers and is always

loaded symmetrically, the failure pattern can always be predicted . This is

not the case, however, wi th larger masonry structures, which have many

layers of mortar, both in the vertical and the horizontal directions .

There are more possibilities, and local failure can be propagated in any

direction along the brick-mortar interface.

Consequently, there was considered to be a need to carry out tests on

larger masonry structures. Three types of wall sample were tested, 5-brick

high walls using mortars whose water-cement ratio were 0.9 (WH series) and

1.1 (WL series) and 9-brick high walls using mortar whose water-cement

ratio was 0.9 (TWH series). These are shown in Figure 1. AlI the walls

were formed from solid-solid bricks with mortar of designation ii (BS

5628) . For the WH and TWH walls, the consistency of the mortar was

nominally 10 mm as determined by the dropping ball test whilst for the WL

walls, it was 18 mm.

~-L __ r-~ __ .-~L--. __ ~ __ ~ _SHEAR LOAD

UNIFORM PRECOMPRESSION

la) S - BRICK HIGH I WH 11 WL)

REACTION R -I-~--l---~--+--,---+--,.----!

UNIFORM PRECOMPRESSION

I b) 8 -BRICK HI8H I TWH)

Figure 1. Wall samples.

550

As experimental testing of a wall as a cantilever is difficult . due to

the fact that the base has to be prevented from sliding and overturning .

the walls were tested as deep beams using three point loading. together

wi th uniform normal precompression . The most important aspect of the

loading arrangement was to ensure that both shearing and bending were

produced in the wall samples. with the shearing being dominant .

WALL TEST PROCEDURE

The wall samples were tested in the apparatus shown in Figure 2. This

consisted basically of two universal beams. 457 mm x 152 mm. of 900 mm

length. a hydraulic jack with a 500 KN load cell . a 1000 KN load cell

attached to the vertical hydraulic ram of a Mitchell testing machine and

two 3.5 metre lengths of threaded rod o The two universal beams acted as

bearing plates. exerting approximately unlform normal compressive force on

the wall . whilst the 1000 KN load cell and the vertical hydraul1c ram

provided the shear force . The hor izontal force necessary to produce the

normal compression was obtained by the self-reacting mechanism of the

threaded steel rods and the hydraulic ramo Shear force and normal

compressive fo r ce at failure were measured by the two load cells and were

recorded using a Solatron Orion data logger .

LOAD FROM MITCHELL r---'

6mm PLYWOOD ::s::z:

04 x I05x 30 PLATE

I- f- I-

12mm WEB f- f- f- r- 500kN LOAD CELL

STIFFENER "- I- f- I- \ .J . " .. ~ " ;,,, , , .. '" , " I I , •• I I I • , , •• " • 1 1 ."" I

f- I- !I I I LlI UB 457 x 152x52 I- l- r- f.! HYDRAULlC RAM

o g 32

L f- IsR lc.t .~ LJ-40mm 0 high ylelc threaded steel rod

I- f- I--

ERS --º- U n p O O 12 mm END PLA"

STEEL SUPPORTS

Figure 2 . Apparatus for testing wall samples

551

The walls were tested with the bed joints vertical. They were

supported on two steel supports , such that the loading was symmetrical . In

principIe the tests were similar to the triplet tests. Precompression was

applied to the walls by the hand operated hydraulic ram until the required

stress leveI was obtained. Subsequent1y, shear load was applied to the

walls through the vertical hydraulic ram which was operated by the Mitchell

testing machine. The failure pattern of the wall was noted.

RESULTS ANO DISCUSSION

The results from the shear tests on the three types of wall WH, WL and TWH

are shown in Figures 3, 4 and 5 . From Figure 3, it can be seen that the

relationship between the average failure shear stress TU and normal

precompression stress 0c can be expressed by two linear relationships.

(a) TU = 0.25 + 0 . 70oc N/mm 2

(for 0c < 1.5 N/mm 2 approximately)

where the coefficient of correlation r

(b) TU = 0.96 + 0.250c N/mm 2

(for 0c > 1.5 N/mm 2)

where r 1.0.

0.95, and

The same resul ts could, however, also be expressed by just one linear

relationship, as:

TU = 0.40 + 0.52oc N/mm 2

where r = 0.96 .

The apparent reduction in slope at high precompression supports the

hypothesis that diagonal tensile cracking in the mortar plays some role in

the ultimate shear failure of masonry structures. At much higher

precompression stress leveIs, it would be expected that the mo de of failure

would be completely due to tensile mortar cracking.

A similar trend although not conclusive can also be detected from the

results from the WL walls, as shown in Figure 4. The relationship between

the shear strength and the average normal precompression can be expressed

as:

TU 0 . 58 + 0 . 50oc N/ mm 2

(r 0.93)

N S S ..... z

552

Wall WH Serl .. exper ImentaI

~ 2

:z: .... I!)

z UI a: .... (1)

a:

"' UI :z: fi)

ar/ d

/ /

/1:.

.V • V·

<;f '

I:. /~ /

/ ,AI

v · Via I I WH 1:.-- Wall WH , , , ,

'2

AVERAGE PRECOMPRESSION IN N/MM2

Figure 3 . Results of wall tests (WH Series)

N :i :i ..... z

3TI'-rT-r~-r~-rTõrr'-õT'-~-.,,-r,,-rTõ~ Wall WL Serl .. experImentaI

~ 2 :z: .... I!) z UI a: ti a:

"' UI :z: (1)

.............. 6 \0-- ...........

"' ......

..... "'6

.......... I:. ............... 6

..........

o ~ t. - - Wall WI. Õ tI I ' , I '~ , , , I J

AVERA6E PRECOMPRESSION IN N/ MM2

Figure 4. Results of wall tests (WL Series)

553

The results of tests on TWH walls are shown in figure 5. The first

crack load, which was observed from the wall tested with a precompression

leveI of 0.83 N/mm' (nominal) is also shown in the figure:

slightly lower than the ultimate failure load.

3

N 2 2

...... z

~ 2 :J: ~ C!)

z UI a: ~ fi)

a:

"' UI :J: fi)

n

Wall TWH Serias experimentai

V ·A

I

V A

Ultlmate Fallur. 1 .t. era ek

2

AVERAGE PRECOMPRESSION IN N IMM2

Figure 5. Results of wall tests (TWH Series)

this is

The results of the wall tests are shown together in Figure 6. From the

figure, no significant difference seems to be noticeable between the

results of WH walls and WL walls. This suggests that mortar consistency

(and hence water-cement ratio) is not a significant factor in determining

the shear strength at lower precompression leveIs of walls formed wi th

these solid bricks. One reason for this is that the solid bricks used have

low water absorption. Hence, there was no rapid absorption of water from

the mortar to the brick, which for the drier mortar mix might have made the

mortar locally stiff and the bond weaker. If the· test had gone up to

higher prestress leveIs where tensi le cracking would be more important,

slightly lower strengths from the WL walls would be expected. The results

of TWH walls are slightly lower than those of WH walls, especially for the

resul t when 0c = 0.83 N/mm'. Because of their shape the TWH walls were

554

subjeeted to relatively greater bending than the WH and WL walls and would

therefore be expeeted to fail at lower shear loads.

3rl~rr.-rr"rT'-"'-"-'''-'~'''-ro.-,,~

t\I ::E ::E ..... z 2 :J: ~ (!)

z

Experimental R •• u Ita

Wall WL, WH a TWH

A ... W IY. ~

... A+ ... (f)

a: ct W :J: (f)

'{ o_

I

t:. ... ...

t:. ...

+

"

+ TWH ... WH A WL

2

AVERAGE PRECOMPRESSION IN N/ MM2

Figure 6. Results of wall tests

The results of the WH wall tests were eompared with the results from

the triplet tests reported earlier [8-9]. For the WL walls a set of

triplets were prepared using the same mortar, and triplet tests were

earried out. The eomparisons are shown in Figure 7 for the WH walls and

Figure 8 for the WL walls.

The results for WL triplets shown in Figure 8 may be expressed by 2

linear relationships:

(a) TU 0.48 + 0.82oe N/mm 2

(for 0e < 2.0 N/mm 2)

where r 0.93

and

(b) TU = 1.7 + 0.320e N/mm 2

(for 0e > 2.0 N/mm 2)

where r 0.90 . ..

N 2 2 -.... z 2 :z: ... CP Z UI Q: ... (/)

Q: cf UI :z: (/)

A

O O

555

EKperlmental R .. ulfs Wall WH

A

A A

A

A A

A WH results

2

AVERAGE PRECOMPRESSION IN N/ MM2

Figure 7. Comparison of wall WH and triplet test results.

N 2 2

-.... z

~

:z: ... CD Z UI Q: ... (/)

Q: cf UI :z: (/)

V

Wall WL Serl .. 8lperlmental

+ __ Trlplet hlgh SX v ... Trlple' low SX &--WaIlWL

2 3

AVERAGE PRECOMPRESSION IN N/ MM2

Figure 8. Comparison of wall WL and triplet test results.

4

556

From the results shown in Figures 7 and 8, it can be seen clearly that

the wall resul ts are lower than the triplet results . This reflects the

limi tation of the average stress method of analysis. The shear stress

dlstribution within larger masonry structures is more complicated than

within a triplet sample with the structures normally being subjected to a

significant degree of bending in addition to shear loading. Furthermore,

the triplet has only one symmetrical possibly line of failure which is

along the brick-mortar interface of the mortar joint. Larger masonry

structures, like walls, have a number of vertical and horizontal mortar

joints along which failure may take place . Failure will develop along the

mortar joints, where shear stress conditions (or diagonal tensile stress)

are worst. The typical failure pattern for the WH (similar to WL) wall is

shown in Plate 1 and for the TWH wall in Plate 2. The failure line follows

basically the imaginary line joining the applied shear load to the

supports; but affectlng only the brick-mortar interfaces.

Plate 1 Failure pattern for WH wall ..

557

Plate 2 Failure pattern for TWH wall

CONCLUSIONS

The relationship between average shear strength and average normal

precompression for walls may be expressed by a Coulomb type of expression.

At higher precompression leveIs, the slope of the linear expressions seems

to be reducing, indicating the possibility of a change in the failure mode.

The shear strength of the walls tested were lower than the strength of

the corresponding triplet samples . Therefore the avp.rage shear strength of

walls cannot normally be accurately predicted by the triplet equation. The

triplet equation may however be used as a guide to the upperbound value of

wall strength.

558

The failure patterns for the walls tested suggest that the worst shear

conditions exist along the brick-mortar interfaces nearest to the imaginary

line connecting the supports to the concentrated shear load.

1. Johnson, F . B. Procedures to Assemblages , Products , ed.

REFERENCES

and Thompson, J .N. , Development of Diametral Testing Provi de a Measure of Strength Characteristics of Masonry Designing, Engineering and Constructing with Masonry F.B. Johnson, Houston , Texas , Gulf Publishing, 1969.

2 . Stafford Smith , B. and Carter, C., Hypothesis of Shear Failure of Brickwork , Jnl . of the Structural Div . , ASCE, Vol. 97, No . ST4, Proc. paper 8029, April 1971 .

3 . Hamid, A. A. and Drysdale, R.G., The Shear Behaviour of Brickwork Bed Joints , Proc . of the British Ceramic Soc., no . 30 , Sept . 1982 .

4 . Hamid, A.A . and Drysdale , R. G. , Behaviour of Brick Masonry under Combined Shear and Compression Loading, Second Canadian Masonry Symposium , Ottawa , Canada, June, 1980 .

5. Jain, A.R. Tests on Brick Couplets, Proc. Instn. Civ. Engrs, Part 2, 1978, 65 , Dec., pp.909-915 .

6. Hofmann , P . and Stockl, S . , Tests on the Shear-Bond in the Bed Joints of Masonry , Proc . Sixth International Brick Masonry Conference , Rome, 1982.

7. Polyakov, S.V., 'Masonry in Framed Buildings', Moscow, 1956. Translated from Russian by G.L. Cairns (1963). Publ i shed by National Lending Library, Boston Spa, U.R.

8 . Ghazali, M. Z. and Riddington, J.R . Shear Strength of Brickwork, Proc . First East Asian Conference on Structural Eng., Asian Inst . Tech . , Bangkok, Vol.1, 1986 .

9. Riddington , J .R. and Ghazali , M. Z. , Shear Strength of Masonry Joints at High Normal Stress LeveIs , 2nd Int . Seminar on Structural Masonry for Developing Countries, Ruala Lumpur, Malaysia, March 1987 .