You are on page 1of 13

Engineering Structures 36 (2012) 4960

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Seismic retrotting methods for the jack arch masonry slabs


Mahmoud R. Maheri , S. Pourfallah, R. Azarm
Department of Civil Engineering, Shiraz University, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Jack arch masonry slab, developed in the post-industrial revolution Britain has been used extensively to
Received 6 April 2010 oor and roof industrial and residential buildings in many parts of the world. It is still in use in parts of
Revised 3 November 2011 Europe, the Middle East and Indian subcontinent. Collapse of a large number of these composite slabs
Accepted 9 November 2011
during past earthquakes, including those belonging to some buildings of historic importance, has
Available online 27 December 2011
pointed out the weakness of this type of ooring to seismic loading. It has also highlighted the need
for developing appropriate retrotting schemes. Two different retrotting methods are currently in
Keywords:
use. One method uses a thin reinforced concrete layer over the slab to provide integrity and the
Jack arch slab
Retrotting
necessary strength. Another method utilizes transverse beams to form a steel grid with the main beams
Concrete layer to achieve the same objectives. In this paper, out-of-plane pushover tests are rst conducted on a num-
Masonry ber of full-scale ordinary and retrotted jack arch slabs. Results of the tests are then used to compare
Pushover test the strength capacity and other seismic performance parameters of the slabs including; ductility and
Seismic performance the behaviour factor. Numerical analyses are also carried out in support of the experimental investiga-
tion. It is concluded that the steel grid method of retrotting, in addition to being easier and faster to
perform and much less costly, addresses all the strength and performance requirements of the slab,
whereas, the concrete layer method, though effective in increasing the strength, considerably increases
the weight of the slab, which, in turn, may increase strength demand on other structural elements. The
benets of using the concrete layer in improving the seismic performance are also less than those of
the steel grid method.
2011 Elsevier Ltd. All rights reserved.

1. Introduction Jack arch masonry ooring system consists of a number of clo-


sely-spaced, parallel steel I-beams and a number of shallow brick
The masonry jack arch ooring was developed by early arches lling the spaces between the steel beams (Fig. 1). The
Victorian architects in Britain and used extensively to cover large spaces over the arches are then lled to create a at oor. The roof
oor areas at factories and other industrial buildings [1]. Fine and oor slabs constructed using the jack arch system are stable un-
examples of this type of ooring can still be seen in the Albert Dock der normal static conditions, as the brick arches transfer the gravity
warehouses in Liverpool, designed by the early Victorian architect loads, mainly in compression, along the arch to the supporting
Jesse Hartley and the late Victorian kiln buildings and warehouses beams. The load is then transferred along the parallel steel beams
around London [2,3]. The ooring technique spread westwards to to the supporting walls or girders. The geometric form of the jack
North Americas and eastwards to mainland Europe and, by the arch and the load path through the steel beams, make the slab act
middle of the twentieth century, it became a popular ooring sys- as a one-way system. The Victorian jack arch oors had larger brick
tem in parts of East Europe, the Middle East and the Indian subcon- arches (around 1.5 m wide with a maximum rise of 20 cm) [2]. The
tinent. Presently, there are a large number of post-industrial later, 20th century oors, however, were constructed using smaller
revolution buildings of historical importance which are oored brick arches, 80 cm1.0 m wide with an average rise of 5 cm so that
by jack-arch system. Due to its technical simplicity, speed in con- a at surface could also be created for the ceiling by direct plaster-
struction, low cost and the ability to alter or provide openings in ing. The Victorian builders also employed transverse tie-rods to join
the slab after construction, jack arch slabs are still very popular the steel beams and provide lateral support for the beams; a prac-
in the Middle East, where, not only industrial buildings and ordin- tice ignored by the later builders.
ary dwellings but also many multi-storey steel framed buildings Taking into consideration the widespread use of the jack arch
are oored by this method. ooring and its advantages compared to the more modern, con-
crete-based slabs, it is rather surprising that there is no mention
Corresponding author. Tel.: +98 917 7167274; fax: +98 711 6473161. of the system in codes of practice. Indeed, save for the works pub-
E-mail addresses: maheri@shirazu.ac.ir, mmaheri@hotmail.com (M.R. Maheri). lished by the rst author and his colleagues [1,46], and another

0141-0296/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2011.11.018
50 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

Fig. 1. The ordinary, jack arch roong system.

recent study carried out by others [7] a search of the related liter- become part of an inter-connected steel grid, allowing the vertical
ature reveals almost no other references to any particular scientic loads to be transferred in two directions, also enabling the transfer
research directed at studying this roong system or any attempts of the in-plane forces (Fig. 2). In fact, by using a steel grid, the grid
to provide an engineering basis for its design and construction. will act as the main load-carrying element in the slab while the
For gravity design, the engineers consider the brick arches as brick arches act mainly as in-ll panels. The proposed two-way,
merely dead loads, carried by the steel beams, and sufce with steel grid, engineered jack arch system, therefore, addresses all
designing the steel beams. This assumption ignores the role of the weaknesses of the ordinary one-way slab as discussed in refer-
brick arches in transferring slab loads and the resulting large stres- ence [1]. The effectiveness of the method in removing the seismic
ses developed in them. There are also no provisions for their seis- weaknesses of the jack arch slab was investigated both experimen-
mic design. On the other hand, the performance of the jack arch tally and numerically [5] and procedures for engineered design and
oors in a number of recent earthquakes in East Europe and the construction of the slab were also introduced [1].
Middle East, particularly in Iran, has generally been poor. Collapse
of a large number of jack arch slabs and damage to many more was 2. Retrotting of jack arch oors
reported from the Qir, Iran earthquake of 1972 [8], the Romanian
earthquake of 1990 [9], the Manjil, Iran earthquake of 1990 [10], Following the Romanian earthquake of 1990 and the Iranian
the Golbaf, Iran earthquake of 1998 [11] and the Bam, Iran earth- earthquake of 1990, the need to retrot the earthquake-damaged
quake of 2003 [6,12]. The three latter earthquakes are of particular and the existing jack arch slabs became apparent. In Romania, after
interest in this regard as they provided real testing grounds for the earthquake, a method was developed and used to retrot some
many jack arch oors and roofs. old jack arch oor slabs. In this method, a reinforced concrete layer
Typical weaknesses and modes of failure of the traditional one- is placed over the jack arch slab. The retrotting procedure
way slab include [1]; (i) movement of simply supported steel includes; removing the slab ooring nish, then placing over the
beams from their position under earthquake shaking, causing the slab a mesh of reinforcement consisting of minimum 8 mm bars
collapse of brick arches, (ii) inability of the brick arches to transfer with maximum bar spacing of 50 cm, connected to the top anges
in-plane loads in the direction perpendicular to the steel beams, of the slab beams by either direct welding or through shear keys
(iii) concentration of stresses in the stiff brick arches due to out- (Fig. 3). Finally, the slab is covered with a layer of concrete having
of-plane vibration of the slab, (iv) weakness of the slab system in a minimum thickness of 5 cm. This method, although effective, is
transferring in-plane shear, (v) dynamic interaction between the costly and time consuming. The addition of the concrete layer also
stiff brick arches and the more exible steel beams under vertical adds to the weight of the slab, resulting in increased gravity and
vibration and (vi) inability of the slab to act as a diaphragm as is seismic loading.
required for good seismic performance. Restraining the ends of In Iran, after the Manjil earthquake of 1990, the use of diagonal
the I-beams with transverse steel beams, or xing the ends of tie bracing of the jack arch slabs by steel bars, as a minimum
the beams to the concrete ring beam and using diagonal steel tie requirement, became mandatory. However, the subsequent exper-
bars over the span are two earlier recommendations, presented imental and numerical investigations carried out by Maheri and
by Moinfar [13] and incorporated in the Iranian seismic code colleagues [1,5] showed the inadequacy of the code-recommended
[14] for improving the seismic performance of the ordinary one- technique in improving the seismic response of the slab. The per-
way system. This form of anchored jack-arch slab has a better seis- formance of the jack arch slabs during Bam earthquake of 2003
mic response because the relative movements of the oor beams [6], on one hand, reinforced the above notion and on the other
are somewhat prevented. It should be noted that the contemporary hand, showed the high potentials of the jack arch slab as an earth-
jack-arch slab construction is still considered a non-engineered quake resistant ooring system if certain simple criteria are met.
slab in the Iranian seismic code, and there are no particular proce- Fig. 4 shows an example of the resilience of the anchored slab
dures for their engineered design. under earthquake loading. The slab shown in this gure had sur-
To eliminate the seismic weaknesses of the one-way jack arch vived the dynamic loading in the Bam earthquake 0f 2003 and re-
oors and roofs, the rst author has proposed to use a number of tained its integrity after collapse.
transverse steel beams spanned between the main I-beams to form Following the Bam, Iran earthquake of 2003 a vast program of
a steel grid [1,5]. In this way the unconnected parallel steel beams retrotting the old government buildings including schools has
M.R. Maheri et al. / Engineering Structures 36 (2012) 4960 51

Fig. 2. The two-way, steel grid, engineered jack arch ooring.

Fig. 3. Retrotting of jack arch slab by laying a concrete layer on top.

been undertaken. Many of these buildings are roofed and oored


by ordinary jack arch system. Both the Romanian method of adding
a concrete layer (here abbreviated as CL method) and the two-way
steel grid (SG) method proposed by Maheri [1] are being exten-
sively applied to retrot the oors of these buildings.
One of the main shortcomings of the ordinary jack arch slab is
that the non-homogeneous, composite system cannot act as a dia-
phragm as it is expected from a seismic resisting slab. Both the
concrete layer and the steel grid retrotting methods can provide
enough in-plane stiffness and strength to make the slab act as a
diaphragm. In the case of the CL retrotting method, the RC layer,
when fully xed to the steel beams of the jack arch oor, can pro-
vide the necessary in-plane stiffness and in the SG retrotting tech-
nique, the steel grid should be able to act as a moment resisting
frame to sustain the in-plane loads and provide in-plane stiffness.
For this, the connections between the main and the transverse
beams should be appropriately designed and constructed. This
point is discussed later. As the requirement for the in-plane dia-
phragm action is met by both retrotting methods, it is the out-
of-plane seismic performance and the cost-effectiveness of the
Fig. 4. Resilience of the anchored ordinary jack arch slab under earthquake loading methods which need to be investigated.
(Bam, Iran earthquake of 2003). In this paper, the gravity and out-of-plane seismic strength and
performance of jack arch masonry slabs retrotted by the above
52 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

two methods are explored and compared. The strength and seismic Rl V e =V y ; Rs V y =V s 2
performance parameters are evaluated both experimentally and
where, Ve, Vy and Vs denote the elastic response strength of the sys-
numerically using forcedisplacement performance curves
tem, the idealized yield strength and the rst signicant yield
obtained from force-controlled nonlinear static pushover analyses.
strength, respectively.
The seismic performance parameters of the slabs retrotted by the
Ductility reduction factor, Rl , is a function of the characteristics
two methods are also compared with those of the existing code-
of the structure, including ductility, damping and fundamental
recommended diagonal tie-braced slab and the ordinary slab.
period of vibration, T, as well as, the characteristics of earthquake
ground motion [17,18]. Therefore, it cannot directly be evaluated
3. Seismic performance parameters from the relation given in Eq. (2), as the elastic response strength,
Ve, itself is a dependent variable. A representing relation for Rl is
The nonlinear static pushover analysis is known to represent well proposed by Nassar and Krawinkler [19] in the following form:
the seismic performance of structures and structural elements. The
forcedisplacement performance curves obtained from the push- Rl cl  1 11=k 3
over analyses are the basis for vulnerability studies and perfor-
where,
mance-based seismic design of structures and structural elements.
In the nonlinear static pushover procedure, monotonically increas- Ta b
cT; a 4
ing loads or displacements are applied to a real structural system 1 Ta T
or a nonlinear mathematical model of that system until the displace-
and l is the structure ductility ratio, dened in terms of the ratio of
ment of a control point (usually the point of maximum displace-
the maximum structural displacement at failure, Dmax, to the dis-
ment) reaches a specied target displacement. The target
placement corresponding to the idealized yield strength, Dy. In Eq.
displacement can be either code-specied or the displacement at
(4), a, is the post-yield stiffness given as a percentage of the initial
the maximum sustainable load in the force-controlled analysis;
stiffness of the system and a and b are parameters given as func-
known as the maximum displacement (Dmax). The loaddisplace-
tions of a [19]. Although the relations given by Nassar and Krawin-
ment curve thus obtained can be used to determine the performance
kler are developed for a structure, they are used here to determine
level (point) of the system to a specied earthquake design spectrum
the behaviour factor for jack arch slabs. In fact, in line with the
or it can be used to evaluate seismic performance parameters of the
design procedure for the two-way jack arch slab, outlined in refer-
system including; strength, ductility factor, toughness and the seis-
ence [1], in the present paper the jack arch slab is considered as a
mic behaviour factor.
separate entity to its supporting structure (i.e. simply supported
In forced-based seismic design procedures, seismic behaviour
on rigid supports). This is a viable assumption as only the out-of-
factor, R, is a force reduction factor used to reduce the linear elastic
plane response of the slab to vertical ground motion is considered.
response spectra to the inelastic response spectra. In other words,
With reference to Fig. 5, the position of the horizontal line of the
seismic behaviour factor is the ratio of the strength required to
idealized bilinear forcedisplacement curve is obtained by approx-
maintain the structure or the structural element elastic to the
imately balancing the areas above and below the curve. The equiv-
inelastic strength of the structure or the structural element. The
alent yield displacement, Dy and yield strength, Vy are obtained
seismic behaviour factor, R, therefore, accounts for the inherent
using this idealized bilinear curve. After evaluating the ductility ra-
ductility and overstrength of a structural system. Taking into
tio, l, and the idealised yield strength, Vy, the ductility reduction
account these two components, it is generally expressed in the fol-
factor, Rl can be obtained from Eqs. (3) and (4) and the over-
lowing form,
strength factor, Rs is determined from Eq. (2).
R Rl  Rs 1
4. Experimental program
where, Rl is the ductility-dependent component, also known as the
ductility reduction factor and Rs is the overstrength factor.
The main experimental program consists of testing four, full-
With reference to Fig. 5, in which the actual forcedisplacement
scale jack arch slabs. All four slabs were initially made as ordinary
response curve is idealized by a bilinear response curve [15,16], the
jack arch slabs with the same size and of the same materials. Three
seismic behaviour factor parameters may be dened as:
frames were subsequently retrotted by, respectively, tie-braces,
steel grid and concrete layer. A supporting frame for the slabs
was rst manufactured. The frame was designed to rigidly support
the slabs in the vertical direction. The test slabs were then, in turn,
constructed over the supporting frame.

4.1. Test specimens and setup

The jack arch slabs were 3.2 m  4 m in dimensions, each hav-


ing 5 main steel beams, spanning the longer direction and spaced
80 cm apart. The slabs were considered to be the most vulnerable
forms used in practice with IPE-12 steel I-beams as the main
beams and brick arches made of traditionally manufactured, clay
bricks. The rise of all brick arches was kept constant as 5 cm.
The mortar used in constructing brick arches should be a fast
setting mortar so that individual bricks could be placed in position
without the need for a supporting formwork. Traditionally, equal
amounts by weight of a high clay content soil and gypsum are used
for this purpose. The claygypsum mortar sets almost as it is ap-
Fig. 5. Determination of seismic response parameters using idealized bilinear plied and can support the weight of the bricks before the arch is
response curve. completed (Fig. 6). Prior to testing the slabs, materials used in con-
M.R. Maheri et al. / Engineering Structures 36 (2012) 4960 53

The SG specimen was retrotted according to the guidelines gi-


ven in Ref. [1]. IPE-12 steel I-sections were used for transverse
beams. The brick arches were rst removed to a length of 30 cm
at two equidistant locations along the span to allow for placing
the transverse beams. The transverse beam sections were then
cut, in appropriate lengths, placed in position and connected to
the main beams, using right-angle sections to join the webs, and
short strips to join the anges. The remaining small empty gaps
in the brick arches at the sides of the transverse beams were nally
relled with bricks and mortar.
In retrotting jack arch slabs with reinforced concrete layer, if
the main steel beams are deep enough (IPE-16 and larger), the
reinforcing mesh can be directly placed on and welded to the top
ange of the beams. For shallower main beams, such as those con-
sidered in this study, the apex of the brick arches stands higher
than the level of top ange of the beams, therefore, reinforcing
Fig. 6. Construction of brick arches using fast-setting claygypsum mortar. mesh cannot be directly welded to the top anges. In this case,
similar to the technique used in composite RC-steel slabs, the rein-
forcing mesh should be xed to the steel beams through shear
structing and retrotting the slabs were tested. The tests conducted keys. To retrot the CL slab, 5 cm long, 5  5  0.5 cm right-angle
include; compressive strength of brick units, compressive strength sections were rst welded to the top anges of the beams at
of claygypsum mortar, tensile strength of claygypsum mortar, 30 cm spacing to act as shear keys. The reinforcing mesh, consist-
tensile (exural) strength of brickwork, compressive strength of ing of 8 mm bars at 30 cm spacing and corresponding to a mini-
brickwork, Youngs modulus of brickwork, yield strength of steel mum steel ratio of qmin = 0.002, was then welded to the shear
beams, yield strength of steel reinforcement and compressive keys (Fig. 7(a)). The concrete was then laid over the slab to a depth
strength of concrete. The results of material tests are listed in Table of 5 cm, measured from the apex of the brick arches (Fig. 7(b)).
1. The specimen preparation, test procedure and the number of The loading frame used to apply the incremental out-of-plane
specimens used in testing brickwork, steel and concrete conform load to the slabs is shown in Fig. 8. The out-of-plane seismic load
to the relevant ASTM standards listed in Table 1. However, no spe- exerted on the slab during an earthquake is a varying distributed
cic standard exists for testing claygypsum mortar. For this mor- inertia load, relative intensity of which follows the bowl-shaped
tar, the compressive strength was determined using 12, 50 mm mode of vibration. Such varying distributed load cannot practically
cube specimens, tested after 7 days of dry curing. The results of be applied to the slab in an incremental pushover test. The out of
compressive strength test averaged at 6.0 MPa with a standard plane load was therefore applied on a line in the middle of the span
deviation of 0.66. For determination of the direct tensile strength to distribute the load in one direction. This was done by using two
of claygypsum mortar, 12 standard briquette specimens were pre- hydraulic jacks transferring the load through a deep steel beam
pared and tested after 7 days of dry curing. The direct tensile and a 5 cm thick concrete layer to the slab (Fig. 8). The correlation
strength of the mortar averaged at 0.8 MPa with a standard devia- between the results obtained using this method of line loading and
tion of 0.93. the actual surface-distributed loading is discussed in Section 5.
Of the four slabs; one was kept unchanged as an ordinary jack The masonry jack arch ooring is inherently a non-symmetric
arch slab (specimen OD); another was tie-braced as recommended system having different out-of-plane behaviour in downward and
by the Iranian Seismic Code (specimen TB); the third slab, was sub- upward directions. This is essentially due to the geometry and
sequently retrotted using the steel grid method (specimen SG) behaviour of the brick arches; being stronger and stiffer in the
and the fourth one was retrotted using the concrete layer method downward direction and weaker in the upward direction. During
(specimen CL). the tests, pushover loading of the slabs was carried out in down-
To prepare the specimen TB, after constructing the brick arches, ward direction. It might be thought prudent to have carried out
U14 reinforcing bars were used to tie-brace this slab as it is recom- the one-directional pushover test on the slabs in the more critical,
mended by the Iranian Seismic code [14]. The cross diagonal bars weaker, direction. However, as the permanent gravity loads act
were welded to the top anges of the main beams. In a similar downwards and it is unlikely that the upward vertical seismic
manner to the ordinary jack arch slab (OD); this slab was then n- loading on the slab exceeds the gravity loads, the most vulnerable
ished by covering top of the brick arches by claygypsum plaster. direction would indeed be downwards.
The incremental pushover test of each slab was carried out at
load steps of 5.7 kN. In each load step, the out of plane displace-
ments of the slabs were recorded at six different location using
Table 1 mechanical gauges (Fig. 9).
Material test results.

Property The standard Test value 4.2. Test observations and results
used (MPa)
Compressive strength of brick units ASTM C67 13.8 4.2.1. Ordinary jack arch slab (OD)
Compressive strength of claygypsum 6.0 The force-controlled pushover test of this slab was carried out
mortar as described above. The forcedisplacement curves obtained for
Tensile strength of claygypsum mortar 0.8
Tensile (exural) strength of brickwork ASTM E518 0.25
the specied six locations of the slab are plotted in Fig. 10a. During
Compressive strength of brickwork ASTM C1314 8.4 the test, the rst cracks in the brick arches appeared in the form of
Youngs modulus of brickwork ASTM C1314 2500 diagonal cracks at corners of the slab at the load of 39 kN (marked
Yield strength of steel beams ASTM E8 250 as crack type (1) in Fig. 10b). A look at the forcedisplacement
Yield strength of steel reinforcement ASTM E8 400
curves (Fig. 10a) however, shows that these cracks had actually
Compressive strength of concrete ASTM C39 21
occurred at 35 kN but did not become visible until the load had
54 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

resulted in increased deections. At the ultimate load, the steel


beams had yielded. Yielding and permanent deformation of the
steel beams was visually veried after unloading the slab. An inter-
esting point to note was that the slab retained its integrity
throughout the test and no brittle failure or spalling of brick arches
occurred during the test, even at deections twice the deection
recorded at the maximum load. This indicates the ductile behav-
iour of the jack arch slab, as also was observed in similarly
anchored jack arch slabs during Bam earthquake of 2003 [6].

4.2.2. Tie-braced slab (TB)


The response of this slab to the pushover loading was some-
what similar to the ordinary slab. The rst crack occurred at
34 kN at one corner of the slab. Similar diagonal cracks developed
in the outer brick arches and at higher loads parallel cracks also
developed at the interface between the brick arches and the main
beams (Fig. 11a). The test continued until the maximum load of
74.4 kN, corresponding to a maximum deection of 25.7 mm at
the centre of the slab. Similar to the ordinary slab, no brittle failure
was noted in the brick arches. Also, the steel beams had yielded
during the loading and developed permanent deformation. The
loaddeection curves recorded at the specied locations on the
slab are presented in Fig. 11b.

4.2.3. Slab retrotted by the steel grid method (SG)


In this retrotted slab, the rst diagonal crack developed at one
of the outer brick arches at 51.5 kN. After this initial crack, further
parallel diagonal cracks occurred in the brick arches at the corners
of the slab at 85.8 kN (Fig. 12a). Prior to the ultimate load some
cracks also developed in the middle of outer brick arches, parallel
to the main beams. The loading was continued until the ultimate
load of 103 kN at which the maximum deection was measured
as 30 mm. Similar ductile response was noted for this slab; with
yielding of the main steel beams and no spalling or brittle failure
of brick arches. The loaddeection curves for this retrotted slab
are presented in Fig. 12b.

4.2.4. Slab retrotted by concrete layer (CL)


The rst cracks occurred in brick arches at the corners of the slab
Fig. 7. Retrotting the jack arch slab using the concrete layer method (a)
connecting the reinforcement mesh to the slab beams and (b) concreting. in a diagonal pattern similar to the other slabs. The load correspond-
ing to these cracks was measured as 143 kN. However, unlike the
other slabs, the cracks could only be seen in the brick arches under
the slab and they did not appear on top of the slab (i.e. in the concrete
layer). Therefore, these cracks were only associated with the brick
arch failure and not the concrete layer. A small number of further
diagonal cracks developed in the brick arches at higher loads
(Fig. 13b). The loading was continued up to 194.5 kN at which load

Fig. 8. Test set-up for out-of-plane pushover loading of the slabs.

reached 39 kN. These failures appear to be due to torsion at the


corners of the slab. At higher loads similar parallel diagonal cracks
appeared in the outer brick arches (crack type 2). The maximum
load which could be applied to the slab was 63 kN resulting in a
maximum displacement at the middle of the slab (gauge 2) of
23 mm. At this load cracks also appeared in the central brick arches
parallel to the main beams. Attempts at increasing the load only Fig. 9. Location of gauges recording the deections of the slabs.
M.R. Maheri et al. / Engineering Structures 36 (2012) 4960 55

4.3.1. Ultimate strength


The strength capacities of the four slabs were compared in Table
2. The ultimate strengths of the slabs are also compared graphi-
cally in Fig. 15(a). In this gure, the strengths of the retrotted
slabs are normalized to the strength of the ordinary slab. Strength
increases of 18%, 63% and over 200% are obtained for, respectively,
the tie-braced, steel grid and the concrete layer slabs. In order that
a better judgment of these strength increases could be made, it
would be appropriate to rst discuss the capacity demand for such
slabs. The slab beams are normally designed for gravity loads. The
increased capacity demand due to vertical seismic loading is a
function of the vertical design base acceleration and the slabs re-
sponse amplication factor. Based on the Iranian seismic code for
a high intensity seismic zone and the response amplication factor
for the slab evaluated as outlined in reference [1], the spectral
acceleration for the out-of-plane vibration of the slab is calculated
as around 0.5 g. As the slab is already designed for 1.0 g gravity
load, therefore the increased capacity demand will be 50%; a de-
mand that cannot be met by the tie-braced slab and is far less than
the capacity that can be provided by the concrete layer slab. Only
the steel grid slab can cost-effectively meet this capacity demand.

4.3.2. Ductility ratio (l) and ductility reduction factor (Rl)


The ductility ratio for the ordinary jack arch slab is evaluated as
l = 1.7. This is a relatively high value for a masonry structural
member. In Fig. 15(b), ductility ratios of the four slabs are com-

Fig. 10. Test results for the ordinary slab; (a) forcedeection curves and (b) failure
pattern of brick arches.

the slab could sustain no further loads. The maximum deection at


this load was measured as 45 mm. At the maximum load, some local
brittle failure and spalling occurred in the brick arches. Yielding of
the steel beams is assumed to have caused the inability of the retro-
tted slab to sustain further loads; however, this could not be visu-
ally veried on unloading of the slab. The loaddisplacement curves
for this slab are plotted in Fig. 13a. The response of this slab was evi-
dently different to the other three slabs. It behaved in a manner sim-
ilar to a concrete-steel composite slab rather than a masonry jack
arch slab. A comparison of the results for the four slabs follows.

4.3. Comparison of test results

In this section, the strength and seismic performance parame-


ters of the four tested slabs are extracted from their respective
pushover forcedeection curves and compared. For this purpose,
the forcedeection curves for the reference point, close to the
middle of the slabs (gauge 2), are used. The seismic performance
parameters are evaluated as outlined in Section 3. The parameters
evaluated and compared include; ultimate strength, ductility ratio,
ductility reduction factor, overstrength factor and behaviour factor.
The evaluated parameters for the four slabs are listed in Table 2.
The forcedeection curves for the reference point of the four slabs
are also compared in Fig. 14. In the following, these parameters are
further discussed. Other important factors; including the added
weight due to retrotting and the cost of retrotting are also eval- Fig. 11. Test results for the tie-braced slab; (a) forcedeection curves and (b)
uated and compared. failure pattern of brick arches.
56 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

Fig. 12. Test results for the steel grid slab; (a) forcedeection curves and (b) failure Fig. 13. Test results for the concrete layer slab; (a) forcedeection curves and (b)
pattern of brick arches. failure pattern of brick arches.

pared, relative to the ductility of the ordinary slab. It is evident that the tested slabs, except for the tie-braced case which shows a high-
diagonal tie-bracing of the slab does not have much effect on this er value of overstrength. It seems that in the tie-braced slab, failure
parameter, whereas, 30% and 67% increases in ductility could be has been associated with yielding of the tie-braces as well as the
seen for the retrotted steel grid and the concrete layer slabs, steel beams, whereas, in the other slabs, failure has been associ-
respectively. Similarly, no change in the ductility reduction factor ated with the yielding of steel beams only. It was expected that
Rl is noted for the tie-braced slab (Fig. 15(c)). However, it is inter- the concrete layer slab exhibit a higher overstrength factor. How-
esting to note that although the increased ductility provided by the ever, as the concrete layer did not yield under loading, the failure
concrete layer retrotting method is much higher than that pro- mechanism of this slab appears to be somewhat similar to that
vided by the steel grid method, the ductility reduction factor Rl of the ordinary slab with yielding of the main steel beams domi-
for both systems are the same. The reason for this being the depen- nating the response.
dence of Rl, not only on the ductility ratio, but also on the funda-
mental period of vibration (T) of the system as discussed in Section 4.3.4. Behaviour factor (R)
3. The stiffer concrete layer slab is of lower period of vibration Seismic behaviour (or force reduction) factor, as the product of
compared to the steel grid slab. the ductility reduction factor and the overstrength factor, is per-
haps the best indicator of the seismic performance of a structural
4.3.3. Overstrength (Rs) system. This parameter is presented comparatively for the four
With reference to Table 2 and Fig. 15(d) in which the over- slabs in Fig. 15(e). As expected, little difference is seen in the R-fac-
strength factor of the four slabs are compared relative to that of tors for the tie-braced slab and the ordinary slab. However, the two
the ordinary slab, little difference is noted in this parameter for retrotted slabs show higher values compared to the ordinary slab,
M.R. Maheri et al. / Engineering Structures 36 (2012) 4960 57

Table 2
Strength and seismic performance parameters of the tested slabs.

Slab Yield strength Yield displ. Dy (mm) Ultimate strength Maximum displ. Ductility ratio (l) Ductility reduction Overstrength Behaviour factor
Fy (kN) Fu (kN) Dmax (mm) factor (Rl) factor (Rs) (R)
Ordinary (OD) 53 13.4 63 22.8 1.7 1.5 1.55 2.33
Tie-braced (TB) 62 17.6 74 25.7 1.46 1.4 1.8 2.52
Steel grid (SG) 84 13.5 103 29.6 2.2 1.96 1.63 3.20
Concrete layer (CL) 166 15.9 194 45.2 2.8 1.96 1.53 3.00

5. Numerical investigations

Numerical analyses were carried out with the aim of investigat-


ing some of the variable parameters affecting the performance of
the retrotted slabs and the test results, which could not be stud-
ied through testing. These include; (i) the effects of using an area-
distributed incremental loading compared to the strip-distributed
loading used in the tests, (ii) the effects on the response of the
tie-braced slab of xing diagonal tie braces to the bottom anges
of the main beams, compared to the top anges as was performed
in the tests and (iii) the inuence on the response of the concrete
layer retrotted slab of the thickness of the concrete layer. For
these investigations, ANSYS software program [20] was utilized
to create numerical models of the slabs and conduct numerical
analyses. Prior to investigating the above three parameters, accu-
racy of the numerical models was veried by comparing the results
of numerical analyses with those obtained from the tests.
Fig. 14. Comparison of the forcedeection curves at the middle of the four tested
slabs. The 3D, 8-noded, SOLID65 element in ANSYS was used to model
the brick arches. The most suitable material model in this software
to model the behaviour of the jack arch brickwork is the Concrete
indicating their better performance. Comparing the two retrot- material. This nonlinear material uses a smeared crack model to
ting methods, the steel grid system exhibits a higher R-factor, take into consideration material cracking as well as the orthotropic
due mainly to its higher overstrength. nature of brickwork. It is also capable of modeling material crush-
ing. Hognestad model is also used to represent the material stress
strain relationship in compression and tension [20]. To model the
4.3.5. Added weight
steel beams, the 3D, SOLID45 element was utilized. Also, in the
Negligible changes in the weight of the slab occur when diago-
tie-braced slab, Link8 elements were used to model the tie-bars
nal tie-braces or transverse beams are added to the slab. However,
and to model the concrete layer, the same SOLID65 element and
adding a concrete layer increased the weight of the slab by 25%.
Concrete material were used. The material properties used in the
This is a considerable amount of weight increase for the slab. A
numerical investigations are the same as those obtained experi-
large increase in the weight of the slab due to retrotting is not
mentally (Table 1). To best model the actual loading exerted on
suitable, because the added weight increases the gravity and seis-
the slabs during the tests, in the pushover analysis the incremental
mic loads on the slab, as well as, on the other supporting structural
load was assumed to apply uniformly on a 30 cm wide central
members such as walls, beams, columns and foundation. This, in
strip, perpendicular to the main beams. The numerical model for
turn, increases the strength demand on these elements and may
the ordinary slab is shown in Fig. 16.
necessitate retrotting the supporting elements.

4.3.6. Retrotting costs 5.1. Verication of the numerical models


The cost is a key factor when choosing a retrotting scheme. A
cost analysis was carried out on the two retrotted slabs based on Nonlinear pushover analysis was carried out on the numerical
the tariffs of the Iranian Management and Programming Organiza- model of each of the four slabs and results, in the form of force
tion. The cost of retrotting an existing masonry jack arch slab deection curves, were compared with those recorded in the tests.
using the steel grid method is 54% of constructing a new slab The comparisons are shown graphically in Fig. 17(ad) for the OD,
(Fig. 15(f)). In using the concrete layer method, the retrotting TB, SG and CL slabs, respectively. The numerical pushover curve for
costs should be considered in two parts; rst, the cost of retrot- the ordinary slab matches relatively well with the experimental
ting the slab, which is calculated as 163% of the cost of constructing curve; the numerical model showing a somewhat stiffer post-yield
a new slab. The second part is the cost of retrotting other ele- behavior (Fig. 17(a)). The numerical and experimental results also
ments, due to weight increase. This part is a function of the type compare well in the tie-braced slab, with the numerical model
of building and the available strength of its structural members. exhibiting a relatively stiffer response throughout (Fig. 17(b)). A
It seems that the concrete layer method, though effective in con- fair match between the numerical and experimental pushover
siderably increasing the strength of the slab (far in excess of the curves can also be noted in Fig. 17(c) for the steel grid retrotted
strength demand), it suffers from being more time consuming, slab. However, for the concrete layer retrotted slab (Fig. 17(d)),
technically more difcult to perform and particularly far more although the numerical analysis has predicted the post-yield state
costly. The benets of using the concrete layer method in improv- of the slab well, it shows a markedly stiffer initial response. Com-
ing the seismic performance of the slab are also less than those of parisons made in Fig. 17 for the four slabs show the reliability of
the steel grid method. the numerical models for further investigations.
58 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

Fig. 15. Comparison of (a) strength, (b) ductility ratio, (c) Rl, (d) Rs, (e) R and (f) cost parameters of the four slabs, normalized to those of the ordinary slab.

5.2. Area-loading versus strip-loading 5.3. Lower ange, versus upper ange tie-bracing

As it was mentioned earlier, the out-of-plane seismic load The Iranian Seismic Code [14] is not specic about the location
exerted on the slab during an earthquake is a varying distributed to x the diagonal bracing. For the new jack arch constructions, it is
inertia load, relative intensity of which follows the bowl-shaped common practice to place the bracing on top of the steel beams. To
mode of vibration. Such varying distributed load could not be upgrade the existing ordinary jack arch slabs to code recommenda-
applied to the slab in an incremental pushover test. The out of tions, bracing is, however, xed to the lower anges of the beams
plane load was therefore applied on a narrow strip in the middle as ooring does not allow for the former. The effects of tie-bracing
of the span to distribute the load in one direction. A pushover the lower anges were investigated by changing the numerical
numerical analysis on the ordinary slab was carried out using a lin- model of TB slab accordingly and repeating the pushover analysis
ear varying load in both directions with the load at the apex of the on the slab. The pushover forcedeection curve obtained from
pyramid type loading used as the incremental loading intensity. this analysis is compared to that of the TB slab braced at top
The forcedeection curve obtained using this area-distributed anges in Fig. 18. It can be seen that bracing the lower anges mar-
load cannot directly be compared to that from the strip-loading. ginally increases the strength of the jack arch slab compared to
However, the effects of using the two types of loading on the seis- bracing the upper anges. A higher increase of 15% was, however,
mic performance parameters such as ductility ratio and behavior calculated for ductility of the slab due to bracing the lower anges.
factor can be compared as carried out in Table 3. This table shows
that the differences in the seismic performance parameters evalu- 5.4. The Effects of concrete layer thickness
ated using strip loading and area loading are small, being within
5%. Therefore, using strip loading in the nonlinear pushover tests For retrotting the slab by a concrete layer, there are no scien-
and analyses is justied. tic studies available on the optimum thickness of the concrete
M.R. Maheri et al. / Engineering Structures 36 (2012) 4960 59

Fig. 16. Numerical model of the ordinary jack arch slab.

Fig. 17. Comparison of the experimental and numerical response curves of the four slabs.

layer or the required steel reinforcement. A minimum thickness of


5 cm is mandatory to allow for sufcient cover for the reinforce- Table 3
ment mesh. This is the thickness used for the retrotted slab in this Seismic performance parameters evaluated using strip loading and area
loading.
study. To investigate the level of possible gains by increasing the
thickness of concrete, two further thicknesses of 7 cm and 9 cm Type of Maximum Ductility Behaviour
were also studied. The numerical models used for the new analyses loading displacement Dmax ratio (l) factor (R)
(mm)
were exactly the same as that used for the 5 cm thick concrete
layer model except for the different thicknesses of the concrete. Strip 21.1 1.57 2.4
Area 22.3 1.65 2.5
Around 15% and 20% increases in strength are calculated for 7 cm
60 M.R. Maheri et al. / Engineering Structures 36 (2012) 4960

more time consuming, technically more difcult to perform


and particularly far more costly. It also changes the form and
considerably increases the weight of the slab, increasing
strength demand on other structural elements. The benets
of using the concrete layer method in improving the seismic
performance of the slab are also less than those of the steel
grid method.

Acknowledgement

The authors wish to acknowledge the nancial support


provided by, the Iranian Building and Housing Research Centre, un-
der the Grant No. BHRC-1-8010.

References
Fig. 18. Response curve of the TB slab, with tie-braces attached to the top anges of
the steel beams compared to bottom anges. [1] Maheri MR, Rahmani H. Static and seismic design of one way and two way jack
arch masonry slabs. Eng Struct 2003;25(13):163954.
[2] Parkinson G, Curtin WG. Albert Dock, LiverpoolStructural survey, appraisal
thick and 9 cm thick concrete layer models, respectively. It should and rehabilitation. Struct Eng 1986;64A(10):28391.
be noted that these gains in strength, are at the expense of, respec- [3] Coatsworth AM. Letter to the Editor on Static and seismic design of one way
tively, 40% and 80% increases in weight of the layer. Increasing the and two way jack arch masonry slabs, [Eng Struct, 2003; 25(13); 16391654].
Eng Struct 2004;26(8):1163.
thickness of the layer for improved strength in a retrotting meth- [4] Maheri MR, Bahar O. Analytical studies of the seismic behavior of the I-beam
od which is already well overdesigned is evidently not justied. jack arch system, Proc. 2nd Int. Conf. on Seism. and Earthq Eng (SEE2), Tehran,
Iran 1995; 1; 819828.
[5] Maheri MR, Imanipour A, Seismic evaluation of proposed two way jack arch
6. Conclusions slab, Proc. 3rd Int. Conf. on Seism. and Earthq Eng (SEE3), Tehran, Iran 1999; 2;
605612.
Based on the results of the experimental and numerical investi- [6] Maheri MR. Performance of building roofs in the 2003, Bam, Iran earthquake.
Earthquake Spectra 2005;21(S1):S41124.
gations presented in this paper, the following conclusions can be [7] Mirjalili A, Shakib H, Mazrooee A, Maheri MR. Experimental evaluation of
drawn regarding the performance of ordinary jack arch slabs and different methods of retrotting jack arch roofs. Shiraz, Iran: 8th Int. Congress
the effectiveness of different retrotting method; on Civil Engineering; 2009.
[8] Razani R, Lee KL. The engineering aspect of the Qir earthquake of April 10.1972
in Southern Iran. Washington, DC: National Academy of Engineering; 2009.
(1) The engineered masonry jack arch slab is robust under [9] Pomonis A, Coburn W, Ledbetter S. The Vrancea, Romania earthquake of 3031
earthquake loading with a fair amount of ductility and an May 1990, a eld report by Earthquake Engineering Field Investigation Team
(EEFIT). London, UK: Institute of Civil Engineers; 1990.
ability to sustain high out-of-plane deformations. It is easy [10] Maheri MR. The engineering aspects of Manjil, Iran earthquake of June 1990,
and fast to construct and less costly than the concrete-based Earthquake Engineering Field Investigation Team (EEFIT) report. London,
slabs. UK: Institute of Civil Engineers; 1990. Oct.
[11] Maheri MR. Lessons from Golbaf, Kerman earthquake of 14 March 1998, Proc.
(2) Tie-bracing of the ordinary jack arch slab, as recommended
1st IranJapan workshop on recent earthquakes in Iran and Japan, Tehran:
by the Iranian Seismic Code, does not provide the required 1998 p. 319330.
integrity and the necessary strength for the slab to sustain [12] Ahmadizadeh M, Shakib H. On the December 26, 2003, southeastern Iran
earthquake in Bam region. Eng Struct 2004;26(8):105570.
earthquake loading. The improvements in seismic perfor-
[13] Moinfar AA, Earthquake engineering trend in Iran, Proc. 3rd World Conf. on
mance parameters gained by tie-bracing the slab beams earthquake Engineering. New Zealand: vol. 3, 1965.
are minimal. [14] Building and Housing Research Center, Iranian code of practice for seismic
(3) As discussed in previous publications, the engineered jack resistant design of buildings; Standard No 2800, Tehran: 2005.
[15] Maheri MR, Akbari R. Seismic behavior factor, R, for steel X-braced and knee-
arch slab, in which transverse beams are used to form a steel braced RC buildings. Eng Struct 2003;25(12):150513.
grid, overcomes all the seismic weaknesses of the ordinary [16] Niroomandi A, Maheri A, Maheri MR, Mahini SS. Seismic performance of
jack arch slab. It is also shown here that the transverse beams ordinary RC frames retrotted at joints by FRP sheets. Eng Struct
2010;32(8):232636.
can be used as an effective retrotting measure, enhancing [17] Miranda E, Bertero VV. Evaluation of strength reduction factors for
the out-of-plane strength capacity of the slab to the required earthquake-resistant design. Earthquake Spectra 1994;10(2):35779.
level and highly improving its seismic performance. It is the [18] Levy R, Rutenberg A. Qadi Kh. Equivalent linearization applied to earthquake
excitation and the RlTo relationship. Eng Struct 2006;28(2):21628.
most cost-effective method of retrotting jack arch slabs. It is [19] Nassar AA, Krawinkler H. Seismic demands for SDOF and MDOF systems.
easy and fast to perform with little change to the original Report No. 95, The John A. Blume earthquake engineering center. Stanford,
form of the slab and almost no extra weight. California: Stanford University; 1991.
[20] ANSYS Manual. Canonsburg, PA 15317, USA: ANSYS INC.; 2005.
(4) The concrete layer method of retrotting considerably
increases the out-of-plane strength of the slab; far in excess
of the strength demand on the slab. It suffers from being
ID Title Pages

267845 Seismic retrofitting methods for the jack arch masonry slabs 12

http://fulltext.study/journal/192

http://FullText.Study

You might also like