I.ABSTRACT

11th INTERNA TIONAL BRICKlBLOCK MASONRY CONFERENCE

TONGJI UNIVERSITY, SHANGHAI, CHINA, 14 - 16 OCTOBER 1997

SEISMIC BEHA VIOR OF CONFINED MASONRY BUILDINGS: AN EXPERiMENTAL STUDY

Miha Tomaievic1 and Iztok Klemenc2

Seismic behavior of a three story confined masonry building, conforming to the requirements of Eurocode 8 for simple buildings, has been studied by subjecting two models of building to simulated seismic ground motion on a shaking-table. As a resuIt of relatively high wall/floor area ratio, the observed resistance of the models was high, indicating that prototypes of the tested type and size will be able to withstand, with moderate damage to the walls, strong earthquakes with peak ground acceleration 0.8 g, and will not collapse when subjected to repeated shaking with PGA more than 1.3 g. The measured response and observed mechanism of structural behavior have been used to develop a rational method for seismic resistance verification of the tested type of masonry structures, by modeling the confmed masonry shear walls as frames . Good correlation between experimental and calculated envelopes has been obtained, indicating the validity ofthe proposed method.

2. INTRODUCTION

The basic feature of confined masonry construction systems are vertical reinforced­concrete or reinforced-masonry elements, tie-colurnns, which confme the masonry walls along their vertical edges. Tie-colurnns are usuaIly placed at aIl corners and waIl intersections, as well as along the vertical edges of door and window openings, and their reinforcement should be well connected with the reinforcement of horizontal bonding elements. Traditionally, r.c. confining elements do not represent the load-bearing part of

Keywords: Confined Masonry; Seismic Resistance; Testing; Modeling.

1 Professor, 2 Research Associate, Slovenian National Building and Civil Engineering Institute, Dirniceva 12, 1000 Ljubljana, Slovenia

542

the structure. In most cases, the amount of reinforcing steel is determined fY experience and depends on the height and size of the building. Even the recent Eurocodes 6 and 8 [1 and 2] require that no contribution of vertical confinement to vertical and lateral resistance of the structure should be taken into account in the calculation, although the experiments indicate that tie-columns improve both, the ductility and lateral resistance of structural walls and buildings [3-7].

To investigate the mechanism of action of tie-columns and seismic behavior, two models of a typical three-story conuned masonry building have been tested on the shaking-table at Slovenian National Building and Civil Engineering Institute (ZAG) in Ljubljana, Slovenia. In addition, a number of plain and confmed masonry model walls have been tested to study the behavior of a single, confmed masonry wall element.

3. EXPERIMENTAL PROGRAM

The tested models represented a typical three story confmed masonry building with wall/floor area ratio 5.0 % in longitudinal and 5.6 % in transverse direction. The structural layout of the prototype building in plan conformed to the requirements of EC 8 for "simple buildings" , in the case of which an explicit verification of the seismic resistance is not mandatory. The dimensions of the models in plan and elevation are shown in Figs. 1 and 2, respectively.

80 7 7:! 6

164 7

Fig. 1. Disposition oftie-columns and dimensions of models; plan

The structural system consists of perimetral walls, pierced with window and/or door openings on ali four sides, and two internai walls, which separate the plan of the building into four units. Whereas the building is symmetrical in the shorter direction, a slight asyrnmetry in the longer direction is a result of non-syrnmetric position of openings along the walls, as well as non-syrnmetric position of internai wall in that direction. Tie columns are provided at the corners and wall intersections, as well as along the vertical edges of ali opening. Floors are monolithic reinforced-concrete, cast in-situ slabs, supported by structural walls in both orthogonal directions .

543

Models have been built at 1:5 scale by taking into account the basic requirements for modeling the dynamic behavior and failure mechanism. Special model masonry materiais with mechanical characteristics corresponding as c1ose1y as possible to the ones required by true modeling, have been made. Mechanical properties of assumed prototype and actual model materiais are given in Table I.

3.8 :! :! 3.8:2 2 3.8 - .. " 2o~i ·" 32.6 -.~ w.:f"28.2"202-·-

. -- --------,1C764'". 7~---- 142.6

Fig. 2. Disposition of tie-colurnns and dimensions of shear-walls; elevation

Table 1. Mechanical properties of prototype and model materiais

Prototype Model Prot.lModel

Compressive strength of masonry 5.0 MPa 1.27 MPa 3.9 Tensile strength ofmasonry 0.35 MPa 0.12 MPa 2.9

Modulus of elasticity of masonry 3500 MPa 942MPa 3.7 Shear modulus of masonry 500 MPa 185 MPa 2.7

Compressive strength of concrete 25.0 MPa 10.8 MPa ... 2.3 Yield stress of reinforcing steel 240.0 MPa 266 MPa ;: 1.0

Compression strength ofblocks, mortar and concrete was 1.09 MPa, 0.45 MPa, and 10.8 MPa, respectively. Fully annealed wire has been used to reinforce the tie-colurnns. 4 bars, 3.2 mm diameter, have been used at the comers and wall intersections, and 2 bars have been used at the vertical edges of door and window openings. In order to model the missing live load, lead bricks (160 kg at each floor levei) were fixed to the floors, so that the resulting masses concentrated at fust, second, and third floor levei of the models amounted to 501 kg, 501 kg, and 377 kg, respectively.

The models have been instrumented with LVDT's and accelerometers placed at all floor leveis, as well as with strain-gauges glued on tie-colurnns' reinforcement (Fig. 3). They have been tested by subjecting them to a sequence of seismic excitation, with gradually increased intensity of motion in each successive test run, up until the fmal collapse of the models. Modeled earthquake accelerogram and absolute acceleration response spectrum are shown in Fig. 4. Whereas mode1 MI has been tested longitudinally, model M2 was subjected to excitation along its transverse axis.

544

R

Figo 30 Instrumentation ofmodels

.------------------------, Accelerations

6.0 ,...------------:D=-a-m-,pi:-ng- y-a:-Iu-e =[0;.:-, -'of=-c-=rit:-ic-'al:ll

0.25

0.0 2.0 4.0 6.0

t[s) 8.0

(a)

10.0 12.0

'õil -;; 4.5 o ." ~

" ~ 3.0

u '5 g 1.5 .D -<

0.25 0.50

Period (5) 0.75

(b)

1.00

Figo 40 Modeled earthquake accelerogram and absolute acceleration response spectrum

Within the experimental program, a series of 380/240/38 mm model walls, confmed with 20/38 mm r. Co tie-columns, reinforced with two 302 mUl diameter reinforcing bars (specimens A), or without tie-columns (specimens B) have been also tested (Figo 5) by compression (specimens A V and BV), and by a combination of constant vertical and cyclic lateralload (specinlens AH and SR)o Average compression stress 0'0 = 0028 MPa, approximately 22 % of the masonry's CQmpression strength, has been induced as a constant gravity load in the case of testing of specimens AH and BHo

IV IV -.-.!::!. +H ~ Dy ' ..-" .. ~

I I I ' DH I I

I I I I I i

<a) , I I I i I I

l AH I

.4 PDv

~ . ! I I ~. ' ~o I"I! l

I '"

t

• "' . ~ '" R .. oç . ,

~r: O ~ ; I .

}.8* 13 24 13 3.8 ~ ..

Figo 50 Dimensions and instrumentation ofwalls during lateral resistance testso AR: confmed masonry walls, BH: plain masonry walls

545

4. SEISMIC RESPONSE AND RESISTANCE

The behavior of both models was similar, although different position of models MI and M2 on the platform and resulting different resistance, stiffness, and non-symmetry, caused slightly different damage propagation in the two cases. Diagonally oriented cracks in the middle part of perimetral walls in the direction of motion have developed in the initial phases of testing. During the test runs to follow, the existing cracks propagated and new diagonal cracks formed, oriented in the other diagonal direction. Horizontal cracks, passing through mortar joints, have been observed in the parapets.

Models have been severely damaged when subjected to shaking, which exceeded the intensity, corresponding to the prototype earthquake. In ali stories and in ali walls standing in the direction of seismic excitation, a system of cracks developed, oriented in both diagonal directions. Ali parapets have been also damaged. At ultimate state, heavy structural damage occurred in the middle sections ofthe walls, located in the first fioor, where masonry units crushed, and the walls separated from the confming elements, and disintegrated. Parts of the central wall failed in shear, in some parts, however, sliding shear failure was the reason of collapse. The amplitudes of vibration immensely increased, so that the walls orthogonal to seismic excitation broke. Horizontal cracks at the connection of the walls to the foundation and first fioor slab, as well as diagonal cracks in the comer parts of these walls developed. Severe damage to vertical tie­columns has been also observed. Heavy damage occurred to r.c. tie-columns, whereas no damage to fioor slabs has been observed. At the connection of vertical tie-columns to bond-beams or slabs, crushing of concrete and buckling of reinforcing steel took place. Because of large lateral displacements, damage to transverse walls in the first story increased significant ly (Fig. 6).

Fig. 6. Mechanism of collapse of model M2

To correlate the results of experiments with calculations and verify the proposed numerical model, three limit states have been defined in the behavior f)fthe models:

546

• Elastic limit, defmed by the occurrence of the first major damage to the model's structural elements that caused noticeable decay of the first natural frequency of vibration,

• Maximum resistance, defmed by the maximum base shear resisted, and • Ultimate state, representing the characteristics of the models just before collapse (i.e.,

at the moment when the instruments have been removed from the models).

2 0 mmr---------------------~3~,d~noo~, 800 mm

OO mm fi 3rd floor

VVVV

A 2nd flocr 20mm~----~--------------~~~

2nd fl oor 800 mm

OOmm VVV V

A 1st floor

V V

1 0 mm 1-----------------------':-"-:;"noo-1, 800 mm

OO mm

,"1. I .Ii. Shakmg u ble 08g l---------------------S~h.7~-"B-"~b" 4 0 s

OOg

(a) " " 1'

00 40 60 8 0 00 , 4 0 60

I [secJ 1 [5«J

Fig. 7. Typical measured displacement response time-histories in linear (a) and (b) in non-linear range ofvibration

(b) 80

As the analysis ofrecorded dynamic response ofthe models (Fig. 7) has shown, the first vibration mode shape and shear-beam type shape of vibration prevailed during shaking. Assuming that the shape of vibration is stable during the shaking, and neglecting the damping, base shear and story shear forces at each instant of time have been determined as a sum of inertia forces acting at floor leveIs.

o 10 15 20 25 30 35 40 Relative story cIri fi [nun J

Fig. 8. Base shear - first story drift hysteresis envelopes

The relationship between the base shear and relative story displacement for both models is shown in Figure 8. As can be seen, beca use of different distribution of structural walls in longitudinal (Ml) and transverse direction (M2), different strength and rigidity have been measured. Mainly beca use of longitudinal central wall, not pierced with window and door openings, the rigidity of the model tested in longitudinal direction was greater than that of the model tested transversely. As regards the lateral resistance attained in longitudinal direction, the differences were not so significant. Obviously, the ultimate mechanism and contribution of vertical LC. confming elements to the resistance of model buildings reduced the differences observed in the rigidity during the monolithic phase ofbehavior.

547

On the basis of story shear - relative story drift hysteresis loops, the values of story stiffness defmed as the average value of effective secant stiffness between two maximurn amplitude ,peaks (peak-to-peak stiffness) have been evaluated for each testing phase. As can be seen in Fig. 9, stiffness degradation is in good correlation with the observed fust natural frequency decay (Table 2).

Table 2. First natural frequency decay in correlation with testing phases (Hz)

ModelMI ModelM2 Test run (1) (2) (3) (1) (2) (3)

Virgin model 22.0 21.9 - 21.4 21.9 -R50 21.4 20.3 15.6 15.9 15.8 15.2 R75 17.4 17.2 14.3 13.5 12.9 10.9

RI00 15.6 14.1 11.9 8.5 8.2 8.8 RISO 7.1 6.3 7.0 4.5 4.1 3.3

(1) impact hammer response - after the test (in Hz), (2) Fourier spectrurn of impact hammer response - after the test (in Hz), (3) Fourier spectrurn of seismic response- during the test (in Hz).

40 ,----------------------------,

Ml

.-"~, o ~~--~--~--~----~====~

o 25 50 75 100 150 200 Test run

Fig. 9. Equivalent stiffness degradation in dependence on testing phases

Since the values of the base shear and story shear have been determined on the basis of the measured accelerations, the calculated values should be multiplie(i by acceleration scale factor Sa = 0.58 when referred to as prototype values. The values of maximurn ground acceleration PGA, base shear coefficient BSC (base shear to weight of the building ratio) and dynamic amplification factor DAF (maximurn response to shaking­table acceIeration ratio) at elastic limit, maximurn resistance and at ultimate state, are given in Table 3.

As ean be seen, the maximurn resistance of prototype buildings is relatively high. Longitudinally and transversely, the three-story contined masonry building will be able to withstand, with limited damage to the walls, a strong earthquake with peak ground aeceleration (PGA) corresponding to 0.8 g. Energy dissipation capacity of the structure wilI prevent the building from collapse even in the case when subjected to subsequent stronger shaking with PGA exceeding 1.3 g. However, at this levei of earthquake ground motion, the damage caused to masonry wall panels wilI be beyond repair.

548

Table 3. Seismic resistance ofprototype buildings

Direction Limit state PGA BSC DAF (g)

Elastic limit (R75) 0.57 1.14 2.05 Longitudinal Maximum resistance (RI 00) 0.84 1.73 2.99

Ultimate state (R200) 1.67 0.61 0.43

Elastic limit (R50) 0.42 0.61 2.40 Transverse Maximum resistance (RI 00) 0.82 1.25 1.83

Ultimate state (R200/1) 1.38 0.64 0.66

The analysis of seismic resistance of the tested building type indicates that, even if c1assified as "simple building" according to EC 8, and without any verification of seismic resistance, the structural configuration will ensure the buildings of the tested type and size the required degree of seismic resistance. It should be borne in mind that the required wall/floor area for simple confined masonry buildings is 5 % (actual values

"' are 5.0 % in longitudinal and 5.6 % in transverse direction ofthe building).

5. CALCULATION OF SEISM1C RESISTANCE OF CONFINED MASONRY W ALLS AND BUILDINGS

The results of tests of wall elements show that, by confining the wall with r.c. tie­colurnns, lateral resistance and deformation capacity is significantly improved. In the particular case studied, the resistance Hmax has been improved by more than 1,5-times and the deformation capacity dmax by almost 9-times with respect to the unreinforced wall (Fig. lO). Tie-colurnns increased the energy dissipation capacity of the plain masonry wall by 6-7 times.

2,5

2

zl ,5 ~ :c I

0,5 \ BH-I

O O 5 lO 15

d[mm] '

Fig. 10. Comparison between experimental resistance envelopes of plain (BH) and confmed masonry walls (AH)

On the basis ofthe observed behavior, a model for the calculation of seismic resistance of a confmed masonry wall has been proposed, by idealizing the experimentally obtained resistance envelope curve as a trilinear relationship. The predominant shear behavior of the rnasonry wall has been considered. As indicated by experiments, the contribution of tie-colurnns ' reinforcement due to dowel action is added to the

549

resistance of the wall at ultimate state. Stiffness degradation is modeled by an empirical relationship between the initial, effective stiffness ofthe wall and damage indexo As can be seen in Fig. 11, good agreement between experimental and theoretical resistance envelopes has been obtained in the particular case studied.

2,5

2

~I , 5

:r: I

0,5

o o

,.' \: Calculated

/~ l7~

: / AH.3 AH· I .~ "

AH·2(·)

5 lO d[rnrn]

15

Fig. 11. Correlation between experimental and calculated resistance envelopes for the case of confined masonry walls

In the case that the resistance of the building is to be assessed, the shear walls, pierced by window and door openings, are modeled as frames . Assuming triangular distribution of displacements along the height of the building, the structure is displaced by a small value and the resisting forces in structural members are calculated. The calculation is repeated step-by-step by increasing the imposed displacements. Once the walls enter into the non-linear range, the structural system ofthe building and stiffness matrices are modified. Stiffness and resistance of individual walls in each step of calculation are deterrnined considering the calculated story displacement and idealized resistance envelopes of structural walls in each story. As a result of calculation, relationship between the resistance of the criticai story and interstory drift, i.e. the resistance envelope is obtained.

Resistance envelope, calculated by applying the above described method to model Ml, is correlated with the experimental1y obtained one in Fig. 12. As can be seen, good correlation between both curves has been obtained.

35~~--------------------------~

30

25

~20

~ 15

10

5 R25 calculated

O R5

O 10 15 20 25 30 35 40 d[mm]

Fig. 12. Comparison between experimental and calculated story resistance envelopes of the first story of model M2

550

6. CONCLUSIONS

Model test results show that prototype buildings of the tested type and size will be able to withstand, with moderate damage to the walls, strong earthquakes with peak ground acceleration of 0,8 g, and will not collapse when subjected to repeated shaking with PGA of more than 1,3 g. This indicates that the actual resistance by far exceeded the required resistance for simple confined masonry buildings in most severe seismic conditions (ag = 0,3 g in the zones ofhigh expected seismicity, according to EC 8).

On the basis of test results a rational method for seismic resistance verification of confined masonry walls and buildings has been developed. Taking into consideration the predominant fust vibration mode shape and shear-beam type shape of vibration, the story resistance envelope is calculated by modeling the shear walls as frames. By imposing triangularly distributed displacements along the height of the building, the structurt: is displaced step by step and the resistance of shear walls to imposed displacements is calculated. Good correlation between experimental and calculated envelopes has been obtained in the particular case studied, indicating the usability ofthe proposed method.

7. ACKNOWLEDGEMENTS

The research discussed in this contribution has been financed by the Ministry of Science and Technology of the Republic of Slovenia (project J2-5208-1502). As regards the experimental part of the project, the contribution of the author's colleague Mr.Ljubo Petko~ié is gratefully acknowledged.

8. REFERENCES

I. Eurocode 8. Design provisions for earthquake resislance of structures. Part 1-3: General rules - specific rules for various materiais and elements. ENV 1998-1-3: 1995,1995.

2. Eurocode 6. Design of masonry structures. Part l-I: General rules for buildings. Rules for reinforced and unreinforced masonry. ENV 1996-1-1: 1995, 1995.

3. Y. Wenzhong and J .Zhaohong, "Functions of ties concrete colwnns in brick walls". Proc. 9th world conl earthquake eng., Tokyo-Kyoto, 6, 139-144 (1988).

4. Z.Bolong., W.Mingshun and Z.Deyuan, "Shaking table study of a five-story unreinforced block masonry model building strengthened with reinforced concrete colwnns and tie bars". Proc., US-PRC joint workshop on seismic resistance of masonry struct., Harbin, IV-lI; 1-11 (1988).

5. M.O.Moroni, M.Astroza and S.Tavonatti, "Nonlinear models for shear failure in confmed masonry walls". The masonry soc.j. (Boulder) 12 (2) 72-78 (1994).

6. G.Aguilar, R.Meli, R.Diaz and R.Vasquez-del-Mercado, "Influence of horizontal reinforcement on the behavior of confined masonry walls". Proc. 11th world conl earthquake eng., Acapulco, paper no.1380 (1996).

7. M.liba, H.Mizuno, T.Goto and H.Kato, "Shaking table test on seismic performance of confined masonry wall". Proc. 11th world conl earthquake eng., Acapulco, paper no. 659 (1996).

551