Vous êtes sur la page 1sur 8

Journal of Building Engineering 20 (2018) 663–670

Contents lists available at ScienceDirect

Journal of Building Engineering


journal homepage: www.elsevier.com/locate/jobe

Investigating a reinforced lightweight foamed concrete walling system for T


low-rise residential buildings in moderate seismic regions

Trevor P.A. Dunn, Gideon P.A.G. van Zijl , Algurnon S. van Rooyen
Department of Civil Engineering, Stellenbosch University, Corner of Joubert and Banhoek Streets, Stellenbosch 7600, South Africa

A B S T R A C T

Historically, lightweight foam concrete (LWFC) has been used in non-structural applications. However, low self-weight, insulating properties and fire resistance make
LWFC an ideal construction material for residential infrastructure, if suitable mechanical properties of strength, stiffness and ductility, as well as durability can be
assured. The development of LWFC for structural use has created a need for a structural system that harnesses these beneficial material properties. In this paper a
LWFC walling system is proposed, tested and analysed with finite elements. Results of quasi-static monotonic pull-over tests as well as cyclic tests indicate that the
proposed precast walling and connection system is likely to both provide predictable structural behaviour as well as adequate resistance for low-rise residential
buildings in regions of low to moderate seismicity.

1. Introduction to a 150% increase in fracture energy, Gf – identified by De Villiers et al.


[3] as the main cause of poor bond in reinforced LWFC. This material
Lightweight foamed concrete is a low-density concrete in develop- development of LWFC now requires the development of a system for its
ment for structural use. The goal for the structural use of LWFC is to structural use
harness the insulation and fire risk reduction properties of the material Current low-rise residential infrastructure in the Western Cape of
for residential construction. The current uses of LWFC are geotechnical South Africa comprise mainly of URM buildings. These three-four
applications and non-structural insulation purposes, but an overview by storey buildings are unlikely to conform to the SANS 10160-4:2017
Ramamurthy et al. [16] reflects a trend in mechanical and durability seismic design criteria and are susceptible to the low-moderate peak
properties that hold potential for structural application. LWFC is man- ground acceleration of ag = 0.15∙g predicted for the area. A steel re-
ufactured by adding a pre-formed foam of entrapped air and protein inforced LWFC (R/LWFC) walling system in place of URM construction
based agent, or synthetic agent, (Panesar [15]) Falliano et al. [7] to a aims to take advantage of the thermal, acoustic and fire-resistant
cement, fly-ash and water paste. Both the low-density foam and zero- properties of LWFC. This structural system aims to avert the govern-
aggregate concrete contribute to the lightweight nature of the material. ment housing backlog that currently exists through high quality, high
Recent developments by Kearsley and Mostert [10], and the in- output factory production. In addition, the proposed system aims to
vestigation into the structural use of LWFC by Jones and McCarthy [9] ensure adequate seismic resistance of the residential structures.
have required further investigation into specific structural engineering This study proceeds to identify a prototype LWFC building and
application of LWFC. propose a structural system of load-bearing precast LWFC walls. The
The material properties of LWFC are considerably different to proposed walling system will be explained alongside the rationale for
normal weight concrete (NWC) and differ with regards to tensile the connection design. The target building and scope of the study is
strength, compressive strength and fracture energy. Direct substitution focussed on the Western Cape in South Africa and aims to provide a
of LWFC in place of NWC is not advised. The work of De Villiers et al. structural system for the use of LWFC that is seismically sufficient.
[3] confirmed that although the compressive strength of LWFC is within This paper reports on a study of three precast LWFC wall tests to
a tolerable range to that of NWC, the low fracture energy and tensile better understand both the mechanical behaviour of reinforced LWFC
strength will lead to brittle failure and poor reinforcement bond. De and the proposed structural system. The walls were tested under pull-
Villiers concluded by suggesting that Polypropylene fibre addition may over monotonic and quasi-static loads, according to the precast con-
improve brittle bond behaviour by bridging micro cracks at the re- crete connection testing regime developed in the Negro and Toniolo
inforcement-LWFC interface. This hypothesis was confirmed by Grafe [13] . The physical test results were later interrogated via numerical
[8] who found that adding 0.35% by volume Polypropylene fibres lead analysis and normalisation comparison. The paper is concluded with a


Corresponding author.
E-mail address: gvanzijl@sun.ac.za (G.P.A.G. van Zijl).

https://doi.org/10.1016/j.jobe.2018.09.011

Available online 15 September 2018


2352-7102/ © 2018 Elsevier Ltd. All rights reserved.
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

proposed analytical design method for reinforced LWFC wall panels. mass representation. In this contribution, the intention is to demon-
strate the activation of appropriate, energy-dissipating failure me-
2. Precast LWFC walling system chanisms in the walling system through quasi-static laboratory tests, as
has become usual e.g., [13,14]. Also, attention is given here to the role
The proposed LWFC walling system consists of connected precast of inter-element connections in mobilising favourable failure mechan-
reinforced LWFC panels. The factory cast panels are manufactured to isms. In continuation of the results reported here, a computational
fine tolerances while all the electrical conduits and required plumbing model is calibrated and used to extrapolate the results to full-scale
are cast into the panels. The lightweight panels are then transported to three-dimensional buildings, including various arrangements of open-
site and speedily erected on top of a cast NWC slab and protruding ings for doors and windows, fracture-mechanics based consideration for
reinforcement. Bolted mechanical connection boxes (Fig. 2b,e) connect size effect [1], strain rate and appropriate mass distribution. Here, it
the vertical joints between the walls. These mechanical connections has merely been ensured that seismic resistance is adhered to according
utilise embedded anchors and reinforcement within the wall to tie two to standardised simplified equivalent push-over loading [17], and si-
walls together. Reinforcement dowels, grouted across the horizontal milar for both the scale model and the actual building. Successful
connection (Fig. 2a,c,d), bind the wall panels to both the foundation mobilisation of energy-dissipating failure mechanisms leading to duc-
below and floor panels above. These horizontal connections follow the tile shear behaviour of the laboratory walls tested here, demonstrate the
precast connection guidelines outlined in the New Zealand building potential appropriate response of the walling system on full scale. It is
codes [14,18]. This guideline suggests that reinforcement across a likely, but remains to be proven to what extent such mechanisms will be
connection be less that the reinforcement within the wall, to ensure activated on full scale.
failure at the connections where ductility can be confirmed. The test setup schematic is shown in Fig. 3. Two normal weight
It is proposed that the first-floor slab of the LWFC building be concrete (NWC) panels were placed on top of the connected LWFC wall
constructed of hollow-core slabs, grout connected to the top of the panels to replicate the ground slab below and hollow-core floor slab
LWFC panels. The first-floor walls would be placed atop the ground- above the walls. The green arrows represent the self-weight load of the
floor walls and first-floor hollow-core floor slab now in place. This structure above, applied via springs to the top of the wall and dis-
process would be repeated for the second-floor, above which a standard tributed through a 254x 254x 89 (width (mm) x height (mm) x mass (kg/
roof can be built. m length)) structural steel H-section [19]. To simulate the effect of
upper storey weight in the low-rise building, the walls were initially
loaded to an average 0.1 MPa compressive vertical stress, by pre-ten-
3. Physical tests
sioning each pretensioning rod passing through each of the six (three
pairs) of springs at the positions indicated in Fig. 3. During the pull-over
3.1. Wall 1 and 2 pull-over tests
tests, the effective average compressive stress increased to 0.2 MPa
(Wall 2) and 0.4 MPa (Wall 1), which is calculated from the load-cell
Two monotonic pull-over tests and a single quasi-static, cyclic test
recorded force in each of the six rods. The setup simulates an upper
were performed on a pair of connected LWFC wall panels. The selected
restraint of finite stiffness, which is argued to represent flexural stiffness
wall was that of the lowest storey in the prototype LWFC building,
of upper storeys. However, more accurate simulation of the upper re-
outlined in red in Fig. 1. It should be noted that openings are not
straint would be presented by full-scale analysis of the building, in
considered in this experimental programme, but will be included in
follow-up research endeavours. The walls were displaced through a
future research. Due to the nature of the testing and laboratory space
500kN Instron Materials Testing Machine (MTM) and transferred across
available, the overall wall size was scaled down 1:3. Each scaled LWFC
the structural steel section by embedded shear bolts into the top NWC
panel was of the dimensions 1380x 920x150 (mm). The wall width of
panel. The vertical dashed lines indicate the locations of grouted dowels
150 mm was not scaled due to the top-heavy nature of the test setup and
along the horizontal connections, each consisting of a 12 mm diameter,
emphasis on connection testing. It is acknowledged that careful scaling
520 MPa yield stress steel bar, encased in a 40 mm diameter PVC pipe
is required for the experimental results to be representative of actual
and grouted with SikaGrout®212. Wall 1 had three dowels per panel per
behaviour e.g., [12]. The purpose of applying scaling laws is to cor-
interface. The four horizontal connections (boxed in red) are the con-
rectly replicate stresses and failure mechanisms. For seismic action, this
nections that were removed as a variable for the second pull-over test,
requires not only strength and stiffness scaling, but also appropriate
in wall 2. The vertical connection boxes between the two wall panels
are also indicated in Fig. 3. The concrete mix design for both the LWFC
and NWC panels are given in Table 1. Each wall panel was reinforced
with two layers of a local South African (Ref 0.193) steel mesh, com-
prising of 5.6 mm diameter, 520 MPa yield strength steel bar at a grid
spacing of 200 mm x 200 mm, placed with 40 mm cover from each wall
face.
Displacement of the wall was measured at various points via
LVDT’s. These measurements were taken relative to the wall, so as to
limit the effect of unexpected slip within the laboratory setup. The
notation in the legend of Fig. 4, where wall load-deformation responses
are show, is defined as follows: Global displacement is considered as the
overall displacement of the wall including the top and bottom NWC
panels, measured by the difference in LVDT readings A and D of Fig. 3.
In contrast to this, local deformation is the deformation of the LWFC
panels themselves, measured by the difference in LVDT readings B and
C. Finally, top and bottom slip represent the interfacial slip between the
upper NWC panel (LVDT A in Fig. 3a) and the wall upper corner (LVDT
B), and between the wall lower corner (LVDT C) and the bottom NWC
panel (LVDT D) respectively. Positive displacement was decided to be
in the pull-over direction for movement towards and with the Instron
Fig. 1. Proposed LWFC building. MTM.

664
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Fig. 2. Proposed LWFC precast connec-


tions. a) Elevation and sectional view of
grouted horizontal connection. b) Steel
connection box and attached embedment
anchors. c) Reinforcement and mould for
top and bottom NWC panels. NOTE: Extra
shear reinforcement at grout duct and shear
transfer bolts. d) Grouting of a single con-
nection in the top NWC panel, (shear
transfer bolts for applied load are shown
protruding from the member, the top of a
grouted reinforcement bar is visible). e)
LWFC wall panel section showing eventual
placement of connection box within cast
voids for the vertical connection, embedded
connection anchors are shown at the centre
of the wall.

Fig. 3. (a) Physical test setup schematic, showing top and


bottom NWC panels and connection boxes. Green arrows
indicate points of applied vertical, upper self-weight load.
Black arrows indicate direction and placement of lateral
displacement transducers. Dashed lines indicate points of
grouted- dowel connections, the red rectangles indicate
which connections are removed for wall 2. (b)
Reinforcement of each connector along the central vertical
joint, with 12 mm diameter steel bar of 520 MPa yield
strength. (c) Illustration of full setup, including spring-
supported vertical pretensioning.

665
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Table 1
Mix volumes utilised for this study detailing both LWFC and NWC for ground
and floor slab panels per cubic metre.
Concrete: LWFC NWC

Target casting density [kg/m3] 1600 2400


Base mix density [kg/m3] 1705 N/A
CEM II−52.5 [kg] 371.8 378.0
Fly-ash, class S [kg] 743.7 122.5
Coarse aggregate 6–9 mm [kg] 0 769.5
Polypropylene fibre [kg] 4.1 0
Malmesbury sand [kg] 0 972.0
Water [Litre] 479.7 180.0
Foam [kg] 4.8 N/A
Super plasticiser [kg] 0 7.6

The pull-over load-displacement results of the two walls are shown


in Fig. 4. Four distinct regions of nonlinear displacement were observed
in the displacement graphs for the walls with three dowels per panel,
per interface. These regions are: 1. cracking at the bottom horizontal
connection, 2. initial rotation and lifting of both wall panels 3. in-
Fig. 5. (a) Separate wall panel rotation with 3 dowel connections, versus (b)
dividual panel rotation and 4. slip along the top and bottom horizontal
wall panels rotating together as a single unit in two dowel connections per
connections, the failure mechanisms within the walling system. The panel per interface.
separate rotation of the two wall panels (stage 3) is shown in Fig. 5a. In
the second wall (Fig. 4b) with only two dowels per panel per interface,
stage 3 of individual panel rotation, requiring slip along the vertical The seismic resistance of the prototype low-rise LWFC building was
joint, was significantly less, and the wall panels rotated as a unit, as analysed according to the static lateral force procedure in [17]. The
shown in Fig. 5b. reader is directed to Dunn et al. [6] for detailed calculations. For the
It is hypothesised that the individual panel rotation of wall 1 is a scaled building in the Cape Town region of the Western Cape, with
positive design feature, allowing extra local displacement and energy moderate earthquake acceleration, local soil type, reinforced concrete
dissipation without global structural failure. The final nonlinear dis- wall behaviour and low estimated first vibration period, or high fre-
placement region can be attributed to dowel action and crushing of the quency, the base shear is calculated to be 19.4 kN , compared with the
neighbouring grout and LWFC at the grouted horizontal connections. ultimate resistance of the weakest wall configuration of 132kN .
This was later confirmed via numerical analysis.
In the wall with two dowels per panel per interface (Fig. 4b), an
3.2. Wall 3 cyclic test
overall reduction in peak lateral force resistance from 240kN for to
132kN is seen, and ascribed to reduced dowel resistance. The total
The second physical test, wall 3, was conducted to evaluate the
shearing displacement before horizontal connection slip failure, i.e.
energy dissipation capabilities of the connected walls. This quasi-static
stage 4 on Fig. 4a and b was reduced from 10mm (Figure 4a) to 7mm for
test was conducted according to the same Joint Research Centre
the less ductile wall (Fig. 4b). Apart from the difference in wall panel
European Commission (JRC) scientific and policy report [13]. Three
rotation and slip along the vertical joint, the nonlinear displacement
push-pull cycle groups were applied to the wall via the same test setup
mechanisms compare to those of the three-dowel wall, with partici-
in Fig. 3. Each group of three cycles increased in amplitude to gauge the
pating slip along bottom and top interfaces, as well as shear deforma-
force-displacement response where the initial amplitude and increase
tion of the LWFC walls. The wall tests therefore indicate that the lo-
were based on the monotonic response of the first physical test, wall 1.
cation and number of connections has a significant influence on the
The wall total displacement (LVDT A, Fig. 3) and wall shear (B-C)
resistance, and displacement behaviour, and likely energy dissipation
deformation are compared to the Instron MTM crosshead displacement
capacity of the precast wall system.
in Fig. 6. The latter was used to control the test, as it was not possible to

Fig. 4. Wall responses with (a) three dowel connections per panel, and (b) two dowel connections per panel.

666
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Fig. 6. Instron vs overall wall displacement applied through cyclic loading according to JRC Scientific and Policy Report [13] for 18 half cycles.

Results indicate that global specific energy dissipation is relatively


constant, while the local (wall shear) energy dissipation portion of
elastic-plastic energy increases steadily as the wall is further displaced.
Note that the global (total) wall energy dissipation includes the wall
shear as well as the slip in the upper and lower interfaces between the
wall and the beams/panels representing floors/foundation. In contrast
the local (wall shear deformation) wall energy excludes these me-
chanisms, and includes only the cracking within the wall panel itself
that enables its distortion or shearing deformation. Had the cycles been
increased further, i.e., beyond the overall displacement where slip in
the upper and lower interfaces become dominant as observed in Walls 1
and 2 (stage 4 in Fig. 4), the portion of specific energy dissipated by the
Fig. 7. Force-displacement hysteresis loops for cycle groups 1, 2 and 3. local wall might reduce, emphasising the various contributions to in-
elasticity in the wall system.
control with the LVDTs mounted on the wall. Clearly a significant dif-
ference was recorded between the Instron and the actual wall dis- 4. Numerical analysis
placement, due to flexibility and gaps in the total loading chain.
Nevertheless, reasonable wall total displacement and shearing de- The scaled wall was modelled in Diana FEA © via a total-strain
formation cycles were achieved. rotating crack nonlinear material model, under a displacement con-
Fig. 7 shows the wall load versus total displacement cycles. Rea- trolled analysis. The purpose of the analyses was to study sensitivities to
sonable energy dissipation is indicated by the open load-displacement LWFC material strength, fracture energy, friction along the LWFC wall
cycles at load reversal. The amount of energy dissipated was calculated and foundation/floor interface, as well as the number of dowels and
as the area of each force-displacement hysteresis loop (Fig. 7). It was dowel diameter, without the need to perform a large number of time
observed that due to slip in the test setup, the push half-cycle was of a consuming, expensive physical experiments. Obtaining reasonable
lower amplitude than the pull half-cycle. In Fig. 8 each of these half- agreement between the two physical wall tests and computed results
cycles are converted into a normalised, dimensionless energy dissipa- was considered sufficient, whereupon sensitivity studies were per-
tion bar graph. For this purpose, the integrated area enclosed within formed to ascertain main mechanisms for the derivation of design
each force-displacement half cycle is divided by that cycle’s elasto- guidelines.
plastic (rectangular) half-cycle area, defined by the cycle’s peak force A plane stress analysis was carried out under phased analysis. In the
times the corresponding ultimate displacement, following Negro and first phase the vertical applied and self-weight load of the structure was
Toniolo [13]. The graph expresses the portion of fully plastic load-de- applied while the lateral displacement applied step-wise in the second
formation dissipated in each half cycle. phase [4].
The LWFC material model was characterised to material test results
of de Villiers et al. [3] and Grafe [8], summarised by Van Zijl et al. [20]
and detailed in Dunn [5]. Compressive strength, tensile strength and
fracture energy values were determined via laboratory tests according
to SANS specifications. The final density and compressive strength of
the wall was determined through core drilling of the walls post testing.
A summary of these mechanical properties as well as the relevant finite
element model (FEM) input parameters are listed in Table 2.
The connection interfaces were assumed to be high friction, zero-
tension interfaces to allow for tensile gapping, this unitary friction
coefficient was not tested. Other simplifications in the numerical model
were the connections and angle of embedded reinforcement at the
vertical connection boxes. The overall scale of the wall and 25mm
Fig. 8. Specific energy histogram providing a comparison between local (wall quadrilateral finite element mesh made it difficult to model the con-
shear) and global (wall total displacement) energy dissipation. Energy values nection box within the wall. As a simplification, straight embedded
are normalised by their respective elastic-plastic cycles. reinforcement bars were prescribed and evaluated across the

667
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Table 2 connection in a similar way to that of the horizontal grouted connec-


LWFC Material properties and Finite Element model input values [4]. tion. The vertical load was applied as a distributed load across the
LWFC Panel: Total strain rotating crack model structural steel member and virtual springs were placed in accordance
with the physical tests. The wall FE schematisation is shown in Fig. 9.
Young’s Modulus,E 9.59 Fracture Energy, Gf 69.32 The crack band theory was used [2] for objective energy dissipation,
[GPa] [N/m]
i.e., independent of the element size. The selected finite element size of
Poisson’s Ratio, ν 0 Density, ρ [kg/m ]3
1579
Tension: Hordijk softening curve Compression: Maekawa compressive stress-
25 mm complies with the maximum equivalent length required to
(Diana, 2012) strain curve (Diana, 2012) prevent snap-back within an element, governed by
Tensile Strength ftux 1.41 Compressive Strength 22.02
= ftuy [MPa] fcux = fcuy [MPa] Gf E
Horizontal and vertical connections: Von Mises plasticity model h≤
ft2 (1)
Young’s Modulus, E 210 Yield stress, σy [MPa] 520
[GPa]
Steel Density, ρ [kg/ 7850 Hardening function None with Gf the fracture energy, E is the Young’s modulus and ft is the
m3] tensile strength. Element size dependence analyses were performed in
Horizontal and vertical connections: bond-slip interface
the form of the in-plane shearing response of a wall, converging to
Area, A [mm2] 113.1 Y12 Rebar, Diameter 12
[mm]
insignificant change in the solution for element sizes below 30 mm.
Normal Stiffness 1.3392e10 Shear Stiffness 1.3392e9 The LWFC wall mesh reinforcement and NWC reinforcement bar
Modulus [N/m3] Modulus [N/m3] reinforcement were modelled as embedded, fully bonded reinforcement
Bond-slip interface failure model [4], and assigned Von Mises plasticity material behaviour with para-
Parameter C, [N/m2] 3.5e9 Shear slip at start [m] 0.0028
meters as given in Table 2. The dowels for connection along the hor-
Grid reinforcement
Bar Diameter x [m] 0.0112 Bar Diameter y [m] 0.0112 izontal interfaces were also assigned Von Mises plasticity behaviour,
Bar spacing x [m] 0.2 Bar spacing y [m] 0.2 but a bond-slip formulation was used with the parameters given in
Horizontal and vertical connections: frictional interface material properties Table 2. This is considered essential to capture the wall uplift and slip
Interface Type 2D Line Opening Model Gapping-brittle, along the dowels during the observed wall rocking. The bond-slip
zero tension
Normal Stiffness 5.44e10 Shear Stiffness 2.72e10
model was calibrated against beam-end bond tests on similar LWFC
Modulus [N/m3] Modulus [N/m3] performed by de Villiers et al. [3].
Coulomb Friction model at connections Fig. 10 summarises wall 1 computed response sensitivities to shear
Cohesion [N/m2] 0 Friction Angle [deg] 45 pull-over, firstly to the compressive strength of the LWFC (Fig. 10a),
Dilatancy Angle [deg] 0
and to friction along the interfaces. The initial peak resistance, as well
Concrete foundation: linear elastic isotropic
E0 [GPa] 36 V 0.2 as the residual shearing resistance are shown to be reduced with lower
ρ [kg/m³] 2400 compressive strength, ascribed to crushing at the dowel connections –
see Fig. 11. In contrast, lower friction does not reduce the initial peak
resistance, but leads to significantly lower residual resistance.
Sensitivity to the number of dowels was shown by FE analyses to be
significant, as was clear from the physical experiments, comparing wall
1 to wall 2 behaviour. For more detail, the reader is referred to Dunn
[5]. Sensitivity to LWFC tensile strength and fracture energy, dowel
diameter was also analysed, but shown to be less significant than
compressive strength, friction and number of dowels.
The FE analyses simulated the observed mechanisms of inelasticity,
i.e., wall shear cracking, slip along the top and bottom interface. The
difference in individual panel rotation of wall 1 versus the rotation of
the two panels together in wall 2 due to reduced number of dowels, was
also captured [6]. Figs. 12 and 13 give a visual description of the rea-
Fig. 9. Un-meshed FE model of the laterally displaced wall. listic wall response of the numerical models. The first and second peak
crack output is shown alongside the final cracked state of the respective
physically tested wall.

Fig. 10. Computed wall 1 sensitivities to (a) LWFC compressive strength for a fixed friction angle of 45°, and (b) friction along the interface for a fixed compressive
strength of 25 MPa.

668
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Utilising the analytical model input values in Table 3, the analytical


model results were 20% lower than the measured peak force value of
wall 1 and 6% higher than that of wall 2. The two analytical models had
degree of reinforcement values of ρwall1 = 0.16% and ρwall2 = 0.11%
which provided the substantial difference in resistance across the in-
terface. Table 4 compares the peak lateral resistance for the physical
tests and analytical models.

6. Conclusions

Fig. 11. Compressive principal stress contours, showing crushing at the dowel The response of steel reinforced lightweight foam concrete walls to
connection in the horizontal interfaces. in-plane quasi-static pull-over has been investigated and reported, in
order to demonstrate that it holds potential as construction and mate-
rial. Interconnection of foundations and floors to wall parts, and wall
5. Design model
parts to each other has been a particular focus, in order to demonstrate
its role in mobilising multiple failure mechanisms of rocking, diagonal
Following the results of the sensitivity study, it was hypothesised
cracking, crushing and shear/frictional slipping, whereby significant
that the shear slip failure of the horizontal connections dominate the
energy is dissipated. Experimental testing has been supplemented with
lateral force resistance of the connected precast LWFC walls. Based on
computational modelling. However, calibration to actual full-scale
these results and the overwhelming influence of shear friction and
building response, with realistic window and door opening patterns,
dowel-action due to changes in compressive strength, an analytical
interconnecting orthogonal walls and appropriate mass distribution is
design model has been proposed to predict the lateral force resistance of
currently undertaken in an ongoing research program. From the test
connected LWFC walls.
results reported here on three wall systems, each comprising two in-
The fib Model Code [11] Section 6.3.4 details an expression for the
terconnected wall parts, a dowel-connected foundation and floor with
shear stress design limit ΤRdi across a shear interface intersected by
simulated upper load and constraint, the following conclusions are
dowels. The three-part equation incorporates chemical adhesive
drawn:
bonding and mechanical interlock between the two surfaces , shear
friction across the two surfaces and dowel-action due to the tensile
force and bending moment in the reinforcement across the connection. • The walls exhibited failure mechanisms of rocking, diagonal
cracking, crushing and shear-sliding. More cracking, and thereby
Due to the nature of this structural system it was assumed that the
more energy dissipation is mobilised by increasing the number of
effect of initial adhesive bonding and mechanical interlock across the
dowel connections from two to three per wall panel. Shear-sliding
NWC-LWFC horizontal connection would be overcome before shear
along the wall vertical interconnection plane is also more pro-
friction and dowel-action would dominate the shear stress and re-
nounced with three dowel connections, allowing separate rocking
sistance across the interface. Thus, the original equation was refined to
and an increased number of cracks per wall part.
that in Eq. (2) where the ductile combination of shear friction and
dowel-action are used to determine the lateral force resistance of a • Significantly higher (75%) pull-over resistance and more than
double the lateral sway at ultimate resistance are observed for the
single horizontal connection, giving the lateral force capacity of a
wall with three dowel connections compared with the wall con-
connected wall.
nected with only two dowel connections to the floor and founda-
tions beams per panel.

ΤRdi = μ⋅σn + κ1⋅ρ⋅f yd ⋅(μ⋅sinα + cosα ) + κ2⋅ρ⋅ (fyd⋅fcd ) ≤ βc⋅ν⋅fcd (2) No brittle, localised failure occurred, demonstrating sound beha-
viour of the lightly reinforced lightweight foam concrete and con-
Where: nection layouts. The weakest wall shear resistance (132 kN) is a
κ1 is the interaction ("effectiveness") factor factor of more than six times that which is required from standar-
κ2 is the interaction coefficient for flexural resistance ≤ 1.6 dised lateral push-over earthquake design calculations for the re-
σn is the lowest compressive stress, resulting from the normal force gion.
on the interface.
ρ is the degree of reinforcement across the interface ρ = As / Ac
• Quasi-static cyclic testing highlighted the effect on energy dissipa-
tion of dowel connection number and location. It is recommended
f y is the reinforcement steel yield strength that both local and global displacement (rocking) be ensured by
fck characteristic value of compressive strength of concrete balanced connections.
f yd design yield strength of reinforcing steel in tension

and
fcd design value of the cylinder compressive strength of concrete,
• The calibrated computational models capture the failure mechan-
isms of rocking, cracking and shear-sliding, as well as the larger
α is the inclination of the reinforcement crossing the interface energy dissipation by increased cracking and separate rocking al-
βc is the coefficient for the strength of the compression strut lowed by shear-sliding of the separate wall panels along joint in-
1
κ1 =
σs
fy
≤1.0 and ν = 0.55 ( )
30 3
fck
≤ 0.55 terfaces. The models confirm that a balanced connection

Fig. 12. Peak 1, peak 2 and final cracked state for numerical model of wall 1.

669
T.P.A. Dunn et al. Journal of Building Engineering 20 (2018) 663–670

Fig. 13. Peak 1, peak 2 and final cracked state for numerical model of wall 2.

Table 3 1828–1844.
Material factors for interface behaviour. [2] Z.P. Bazant, B.H. Oh, Crack band theory for fracture of concrete, Mater. Struct. 93
(16) (1993) 155–177.
ν 0.55 fck 20 MPa σn 0.065 MPa [3] J.P. De Villiers, G.P.A.G. Van Zijl, A.S. Van Rooyen, Bond of steel reinforcement in
µ 0.6 f yd 520 MPa ρwall1 0.16 % lightweight foamed concrete (LWFC) (Available online, 31 October 2016), Struct.
Concr. Int. J. fib (2017), https://doi.org/10.1002/suco.201600019.
κ1 0.5 fcd 22 MPa ρwall2 0.11 %
[4] 10 1 Diana, Diana (Displacement Analyser) user’s manual, in: J. Manie (Ed.),
κ2 1.1 α 90 deg βc 0.4 MPa Element Library. TNO-DIANA, Netherlands, 2016.
[5] T.P.A. Dunn, Precast lightweight foamed concrete walling, a structural system for
low-rise residential buildings Master’s thesis, Stellenbosch University, 〈http://hdl.
handle.net/10019.1/102870〉.
Table 4
[6] T.P.A. Dunn, A.S. Van Rooyen, G.P.A.G. Van Zijl, Reinforced lightweight foamed
Summary of peak lateral force resistances. concrete for seismically resistant low-rise residential buildings, 71st RILEM Annual
Week & ICACMS 2017, Chennai, India, 3rd – 8th September 2017, Vol 3, pp.
Wall 1 Wall 2
269–277.
[7] D. Falliano, D. De Domenico, G. Ricciardi, E. Gugliandolo, Experimental in-
Physical peak [kN] 240 132
vestigation on the compressive strength of foamed concrete: effect of curing con-
Analytical peak [kN] 202 140 ditions, cement type, foaming agent and dry density, Constr. Build. Mater. 165
(2018) 735–749.
[8] J. Grafe, Report issue s0117: optimization of fibre content in foam concrete for
arrangement of vertical and horizontal joints exist and should be the improved fracture behaviour, Tech Report, Institute of Structural Engineering,
Stellenbosch University, South Africa, 2017.
design objective for optimal ductility of in-plane response of the [9] M. Jones, A. McCarthy, Preliminary views on the potential of foamed concrete as a
wall system. structural material, Mag. Concr. Res. 57 (1) (2005) 21–32.
• Sensitivity studies reveal compressive strength to control the initial [10] E. Kearsley, H. Mostert, Designing mix composition of foamed concrete with high
fly ash contents. in: Use of Foamed Concrete in Construction: Proceedings of the
peak pull-over resistance, while the friction coefficient controls the International Conference held at the University of Dundee, Thomas Telford
residual resistance. Publishing, Scotland, UK on 5 July 2005, pp. 29–36.
• The computational results informed an analytical expression, for [11] Model Code, Special activity group 5, fib Bulletin, no. 55, 2010.
[12] P.D. Moncarz, H. Krawinkler, Theory and application of experimental model analyis
lateral resistance, which captures the actual resistance of shear walls
in earthquake engineering, Stanford, California: Report number 50, June 1981, The
manufactured from LWFC as construction material with reasonable John A. Blume Earthquake Engineering Center, 1981.
accuracy. [13] P. Negro, G. Toniolo, JRC scientific and policy report: design guidelines for con-
nections of precast structures under seismic actions, Joint Research Centre
European Commission, Luxembourg, Publications Office of the European Union,
Testing and characterisation of durability properties of R/LWFC are 2012.
essential for appropriate durability design with this new construction [14] NZCS, Guidelines for the use of structural precast concrete in buildings, New
material, which is the focus of current research by the authors. Zealand Concrete Society, Christchurch, University of Canterbury, 1999.
[15] D.K. Panesar, Cellular concrete properties and the effect of synthetic and protein
foaming agents, Constr. Build. Mater. 44 (2013) 575–584.
Acknowledgements [16] K. Ramamurthy, E.K. Nambiar, G.I.S. Ranjani, A classification of studies on prop-
erties of foam concrete, Cem. Concr. Compos. 31 (6) (2009) 388–396.
[17] SANS 10160-4, Basis of Structural Design and Actions for Buildings and Industrial
The research is funded by The Concrete Institute (TCI) and the Structures -Part 4: seismic Actions and General Requirements for Buildings, South
Department of Trade and Industry of South Africa under THRIP African Bureau of Standards, Pretoria, 2011.
Research Grant TP14062772324. The original prototype LWFC re- [18] P. Seifi, R. Henry, J. Ingham, Panel connection details in existing New Zealand
precast concrete buildings, Bull. N.Z. Soc. Earthq. Eng. (2016).
sidential building design is property of Mr Thomas Swana and [19] SAISC, South African Steel Construction Handbook, 8th edition, Institution of Steel
Ukuzwana Project Management Solutions (Pty) Ltd. Polypropylene fi- Construction (SAISC), South African, 2016.
bres were sponsored by Sapy (Pty) Ltd. The grout used at the dowel [20] G. Van Zijl, A. Van Rooyen, M. Mubatapasango, T. Dunn, J. Grafe, Durability and
bond of reinforced lightweight foamed concrete, in: High Tech Concrete: Where
connections was sponsored by Sika South Africa (Pty) Ltd.
Technology and Engineering Meet, Springer, 2017, pp. 2185–2193.

References

[1] Z.P. Bazant, Scaling laws in mechanics of failure, ASCE J. Eng. Mech. 119 (9) (1993)

670

Vous aimerez peut-être aussi