Sei sulla pagina 1di 12

Engineering Structures 37 (2012) 193–204

Contents lists available at SciVerse ScienceDirect

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

Experimental tests and numerical modelling of wall sandwich panels


Fabrizio Gara ⇑, Laura Ragni, Davide Roia, Luigino Dezi
Dept. of Civil and Construction Engineering and Architecture, Università Politecnica delle Marche, via Brecce Bianche, 60131 Ancona, Italy

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

Article history: This paper presents the first part of an experimental investigation carried out on a construction system
Received 23 July 2011 based on completed in situ sandwich panels with non-shear connectors, concerning the study of vertical
Revised 15 November 2011 panels used as structural walls. Compression tests with axial and eccentric loads were carried out on sev-
Accepted 2 December 2011
eral full scale panel specimens with different slenderness ratios in order to study the behaviour of panels
Available online 4 February 2012
under vertical in-plane forces. Additionally, diagonal compression tests were performed on square spec-
imens in different configurations in order to study the behaviour of panels under horizontal in-plane
Keywords:
forces. The most significant load–displacement diagrams for increasing load are illustrated and the failure
Sandwich panels
Bearing wall panels
modalities are discussed. The semi-composite behaviour of the panels, guaranteed by the internal layer of
Full scale compression test polystyrene and the reinforced concrete beams at the panel ends, is highlighted. Finally, some numerical
Diagonal compression test simulations are performed with non-linear finite element models and some useful design indications are
Finite element analysis given.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction on construction details, such as reinforced concrete regions at


the panel ends.
Construction systems based on sandwich panels are commonly The structural behaviour of such a sandwich panel can be theo-
used worldwide for intensive building production. Sandwich pan- retically studied by means of analytical models which are usually
els are typically constituted by two concrete layers which are sep- referred to in the literature as models for multilayered beams
arated by an internal insulation layer of various materials (i.e. [1,2] or two-layers composite beams [3,4] with deformable inter-
expanded and extruded polystyrene, rigid polyurethane foam) layer connection. As this paper deals with experimental investiga-
and are usually joined with ‘‘shear connectors’’ (i.e. truss connec- tion, a literature review of theoretical models is beyond the scope
tors) able to transfer the longitudinal interface shear between the of this paper and only some of the most recent works are cited as
layers so as to ensure a fully-composite or a semi-composite example from which a comprehensive list of references describing
behaviour of the sandwich panel. these models may be founded. However the behaviour of single
This paper deals with a construction system that utilises sandwich panels is obviously much easier to predict than the
sandwich panels, both for structural walls and floors, which are behaviour of panels constituting real building walls. To evaluate
obtained by self-supporting reinforced insulation layers completed the structural performances of buildings constructed with this kind
in situ with spritz-beton. The prefabricated modular elements are of construction system, in addition to specific modelling taking
made of an undulated (corrugated) layer of expanded polystyrene, into account the semi-composite behaviour of the sandwich pan-
with suitable density, reinforced by two metallic meshes con- els, other aspects need to be considered in the numerical evalua-
nected by means of orthogonal steel wires welded to the meshes tions, like (i) the restrain degree of wall-floor node depending on
(steel connectors). Thanks to the easy and fast mounting proce- the connection details and (ii) the bi-dimensional behaviour of
dures, this construction system presents some technical advanta- the panels which are each-other connected in real buildings. For
ges that make it often competitive in comparison with traditional this purpose, experimental results of full scale tests, are an essen-
methods or precast systems. From a structural point of view these tial instrument to calibrate both theoretical methods and numeri-
panels are characterised by orthogonal connectors (‘‘no-shear con- cal models.
nectors’’) so that their semi-composite behaviour depends on the In the technical literature several experimental campaigns on
shear stiffness of the expanded polystyrene layer and, above all, precast sandwich panels with shear connectors can be found
[5–13]. On the contrary, very few experimental tests have been
performed on sandwich panels with in situ sprayed concrete and
⇑ Corresponding author. Tel.: +39 071 2204550; fax: +39 071 2204576. no-shear connectors [14,15] and on 3D full scale mock-up [16] in
E-mail addresses: f.gara@univpm.it (F. Gara), laura.ragni@univpm.it (L. Ragni), order to study the behaviour of the panels in real structures. Con-
d.roia@univpm.it (D. Roia), dezi@univpm.it (L. Dezi). sequently, general conclusions on the structural behaviour of this

0141-0296/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2011.12.027
194 F. Gara et al. / Engineering Structures 37 (2012) 193–204

construction system cannot be drawn and further experimental Table 1


investigations are needed. For this reason, an extensive experimen- Mean values of material properties.

tal campaign, including a large number of tests on floor and wall Prismatic specimens Cored specimens Metallic meshes
panels, cyclic tests on wall-floor connections and a load test on (MPa) (MPa) (MPa)
the floor of a full scale 3D mock-up, has been carried out. fcu = 21.95 fcu = 25.10 fm = 769.00
This first paper refers only to tests performed on wall panels. In fcfm = 5.52 fct = 2.40 Agt = 7.62
particular, the results of compression tests with axial and eccentric
loads carried out on panels with an internal layer of different thick-
ness, are presented. After that, the experimental results obtained on
panels with non-undulated polystyrene sheet and on panels with The concrete was characterised by means of tests on
undulated polystyrene sheet and half the number of connectors 40  40  160 mm specimens sampled during the cast, and tests
are discussed. Additionally, the results of diagonal compression on cored specimens with a diameter of 94 mm and length of
tests carried out on square specimens of different configurations 250 mm sampled from the reinforced concrete beams at the ends
in order to study the behaviour of panels under horizontal in-plane of the panel after the concrete curing. In accordance with EN ISO
forces are illustrated: wall standard panels, as well as wall panels 12504-1 [17], bending tests were first carried out to evaluate the
flexural strength of the rectangular specimens, then the two resul-
externally prestressed to simulate the effects of vertical loads and
panels stiffened with four orthogonal walls to simulate the behav- tant parts of the specimens were used for compression tests. A to-
tal number of eight specimens were prepared so that eight flexural
iour of the wall in a real building are considered. For each test,
the load–displacement diagram and the failure modalities are tensile tests and sixteen compression tests were performed. Table 1
reports the mean values of the compression (fcu) and flexural ten-
examined. Finally some numerical simulations, performed with
sile concrete strength (fcfm). The cored specimens were divided into
non-linear finite element models, are also reported and some useful
two sets, one of which was subjected to the compression test and
design indications are given.
the other to the indirect tensile splitting test, in accordance with
EN 12390-6 [18]. Four cored specimens were sampled and, conse-
2. Experimental campaign quently, four compression tests and four tensile splitting tests
were performed. The average values of compression strength (fcu)
The sandwich panels considered in this study are made of a and tensile concrete strength (fct) are reported in Table 1. From
sheet of polystyrene reinforced by two 80 mm  75 mm metallic these specimens, a mean value of 10500 MPa for the concrete
meshes assembled by means of steel connectors. The sheet of poly- elastic modulus (Ec) was also estimated.
styrene has an undulated profile and density of about 15–25 kg/ Tensile tests and weld shear strength tests were carried out on
m3. The galvanised welded wire meshes and the connectors six samples of metallic meshes following EN ISO 15630-2 [19].
welded orthogonally to the meshes, are made with U3 wires of Four samples reached the yield stress showing a very low ductility
high yield steel. Wall panels (WP) were completed simply by (less than 2). The mean values of tensile strength (fm) and percent-
spraying concrete onto the external surfaces of the sheet, first up age elongation at failure (Agt) are reported in Table 1.
to the metallic mesh and then up to the final thickness of the con- The other two samples showed a brittle fracture and a strength
crete layer, using manual tools or pumps (Fig. 1). A ready-mixed value which was about 20% lower than the yielding strength. It is
concrete, with sand no greater than 3 mm and specific additives worth noticing that the failure of all the mesh samples occurred
to improve adhesion and workability, was used. at a welded joint as a consequence of the welding disturbs. Finally,
the weld shear strength tests gave an average shear force of
2.1. Mechanical properties of materials 2.64 kN, which is 1.25 times greater than the wire yielding force,
as prescribed by the code. The mechanical properties of the inter-
In order to evaluate the mechanical properties of the used nal layer were obtained by means of shear tests according to the
materials, several tests were carried out on concrete, metallic Standard Test Method for Shear Properties of Sandwich Core Mate-
meshes and internal layer consisting of a polystyrene sheet and rials [20]. Samples with dimensions of 400  445 mm made of
metallic connectors. three concrete layers and two internal layers were tested. In

wire meshes concrete layers

steel connectors
polystyrene

(a) (b) (c)


Fig. 1. ‘‘Concrewall’’ wall sandwich panels: (a) schematic sketch of the components; (b) concrete spraying onto the external surfaces up to the metallic mesh and (c) up to the
final thickness.
F. Gara et al. / Engineering Structures 37 (2012) 193–204 195

(a)

30 30
1S (80 mm) 3S (160 mm) 1S (80 mm) 5S (160 mm)
Load [kN]

Load [kN]

0 0
Displacement [mm] 5 Displacement [mm] 5

(b)
Fig. 2. Shear tests: (a) test configuration and (b) load–displacement cycles.

particular, eight samples were considered with internal layers of Table 2


different thickness and with a different number of connectors. Shear tests: samples and results.
Subsequently, the polystyrene sheets of four samples were dis- Sample hi (mm) Polystyrene Connectors Ki (kN/mm) Gi (N/mm2)
solved in order to obtain samples with metallic connectors only.
1S 80 Yes Double 15.70 3.45
Load and unload cycles were performed up to failure (Fig. 2a). Slip- 2S 120 Yes Double 9.20 3.03
ping between the concrete layers was measured by means of two 3S 160 Yes Double 7.30 3.20
LVDTs and the applied load was determined by means a pressure 4S 80 Yes Single 12.80 2.81
transducer. In Fig. 2b load–displacement graphs are illustrated 5S 80 No Double 0.47 0.10
6S 120 No Double 0.15 0.04
with reference to samples 1S and 3S, with polystyrene sheet thick-
7S 160 No Double 0.07 0.03
ness of 80 mm and 160 mm respectively, and sample 5S which is 8S 80 No Single 0.20 0.04
similar to sample 1S but without the polystyrene sheets. By com-
paring the results of the tests on 1S and 3S samples it is evident
that the stiffness decreases considerably when the thickness of
the internal layer increases, whereas by comparing the results of 2.2. Panel geometry
1S and 5S samples it can be observed that the contribution of con-
nectors is negligible with respect to the contribution of the A total of twenty two panels were built: sixteen for compres-
polystyrene. sion tests with axial and eccentric load and six for diagonal com-
For each test the initial shear modulus was calculated by means pression tests.
of The panels for compression tests had a total height of 2940 mm,
a width of 1120 mm and concrete layer thickness of 35 mm. To
Ki
Gi ¼ hi ð1Þ avoid stress concentrations and to facilitate the handling
2Ai
operations two reinforced concrete beams were built at the ends
where Ki is the initial stiffness of the sample, and hi and Ai the thick- of the panels by dissolving a portion of the polystyrene sheet
ness and the area of each internal layer. Table 2 shows all the in order to obtain a proper anchorage of the meshes (Fig. 3). For
obtained results. the compression tests, wall panels (WP) with three different
196 F. Gara et al. / Engineering Structures 37 (2012) 193–204

Table 3

120
Geometric characteristics of panels for compression tests.

Specimen Panel Compression loading c (mm) h (mm)


2a.1 WP08 Axial 80 150
4φ8 2a.2 WP08 Axial 80 150
A A φ6/200mm 3a.1 WP12 Axial 120 190
3a.2 WP12 Axial 120 190
4a.1 WP16 Axial 160 230
4a.2 WP16 Axial 160 230
X.2 WPN08 Axial 80 150
Y.2 WPH8 Axial 80 150
2940 mm

2700 mm
2b.1 WP08 Eccentric 80 150
2b.2 WP08 Eccentric 80 150
3b.1 WP12 Eccentric 120 190
3b.2 WP12 Eccentric 120 190
4b.1 WP16 Eccentric 160 230
4b.2 WP16 Eccentric 160 230
X.1 WPN08 Eccentric 80 150
Y.1 WPH08 Eccentric 80 150

and one with eccentric load were performed for each kind of panel.
The list of all the panels with the overall thickness (h), the internal
120

layer thickness (c), and the kind of test performed are reported in
Table 3.
35 35 mm
For the diagonal compression tests, only 1120 mm  1120 mm
1120 mm WP08 panels were considered since the thickness of the internal
c
h layer does not influence the panel behaviour under in-plane forces.
In addition, prestressed and transversally stiffened panels were
Fig. 3. Panel for compression tests. considered in order to simulate the effects of vertical load and
the stiffening contribution of walls and floors orthogonal to the
panels, respectively. To better distribute the compression load,
thicknesses of the internal layer, 80 mm (WP08), 120 mm (WP12) two triangular reinforced concrete regions at two opposite corners
and 160 mm (WP16), were built. For each thickness, two panels of the standard wall panels (Fig. 5a) and two reinforced concrete
were tested under axial load and two under eccentric load. In addi- L-shaped beams in the stiffened panels (Fig. 5b) were built.
tion, two different kinds of wall panel were prepared for the com- In the prestressed panels two steel threaded bars are applied
pression tests: the WPN panel with a non-undulated polystyrene with prestressing loads of 30 kN and 90 kN. The list of all the
layer and the WPH panel with half the number of connectors specimens with overall thickness (h), internal layer thickness (c),
(Fig. 4). In these cases only one compression test with axial load prestressing forces and loading type are reported in Table 4.

Section B-B
B WP08 WPH08

A A
φ3
20
20

150
150
75

40 40 80 B
1120 80 15
1245

concrete polystyrene wire meshes φ3 Section A-A


150
80

WP08
150

WPN08
80

Fig. 4. Details of WP, WPN and WPH panels.


F. Gara et al. / Engineering Structures 37 (2012) 193–204 197

4φ8 Section A-A


A φ6/200mm
A Section A-A

1420 mm
1120 mm
1120 mm

A
A
1120 mm 150 1120 mm
(a) (b)
Fig. 5. Panels for diagonal compression tests: (a) wall panel and (b) transversally stiffened panel.

Table 4
Geometric characteristics of panels for diagonal compression tests.

Specimen Panel Compression loading Prestressing load (kN) c (mm) h (mm)


5.1 WP08 Diagonal – 80 150
5.2 WP08 Diagonal – 80 150
5.3 WP08 Diagonal 30 80 150
5.4 WP08 Diagonal 90 80 150
C.1 WP08a Diagonal – 80 150
C.2 WP08a Diagonal – 80 150
a
With traversal stiffening walls.

2.3. Test configuration and instrumentation displacement. This static scheme simulates the restraint condition
of panels in real multi-storey buildings when connections between
For the compression tests the configuration of Fig. 6 was floor and wall panels produce negligible bending moments.
adopted: panels were placed vertically with the bottom end pinned In the axial compression tests the load is applied at the panel
(cylindrical pin) and the top end restrained so as to prevent lateral axis, while in the eccentric compression test the load is applied

reaction frame

S4

hydraulic
S5 jacks

S1

Svf Svb
Sf
S2

S3
cylindrical pin

axial load eccentric load

Fig. 6. Compression tests with axial and eccentric load: test configuration and instrumentation.
198 F. Gara et al. / Engineering Structures 37 (2012) 193–204

Svf Svf

Shf Shf

Fig. 7. Compression tests: overview and details of the top and bottom restraints.

at the axis of an external concrete layer. In both cases the load is


applied by means of four hydraulic jacks of 500 kN fixed to a reac-
tion frame (Figs. 6 and 7). The hydraulic jacks are managed by
Fig. 9. Diagonal compression test without and with prestressing load: test
means of a hydraulic control unit equipped with a pressure trans- configuration and instrumentation.
ducer to measure the applied load. A steel plate is placed between
the panel and the actuators in order to distribute the load uni-
formly. Furthermore, metallic profiles are used to confine the rein- load is measured using a pressure transducer. Each panel is instru-
forced concrete beams at the panel ends. mented with four LVDTs, working in a range of ±50 mm, placed
Each panel is instrumented with two LVDTs (Svf and Svb), work- vertically (Svf and Svb) and horizontally (Shf and Shb) on the front
ing in a range of ±50 mm, applied in an extensometric configura- and back panel surfaces, in an extensometric configuration to mea-
tion to measure strain over a base length of 0.9 m. Three other sure strain over a base length of 500 mm.
transducers (S1, S2, S3) are placed horizontally at 1/4, 1/2 and 3/4
of the panel height to measure horizontal displacement. Finally, 2.4. Test results
two transducers (S4, S5) are placed within the panel thickness, at
an angle of 45° with respect to the vertical direction, at 3/4 of In this section the main results of the tests carried out on the
the panel height, in order to measure the relative displacement be- wall panels are illustrated: first the results of the axial and eccen-
tween the concrete layers (Fig. 6). tric compression tests and then the results of the diagonal
Diagonal compression tests are carried out by means of a slide compression tests are reported and discussed. As regards the com-
pushed by six hydraulic jacks. The panels, rotated 45°, are placed pression tests, Fig. 10 shows the lateral deflections recorded by the
between the slide and the reaction frame (Figs. 8 and 9). To avoid LVDT at mid-height of the panels (S2) on the sixteen panel speci-
any stress concentration, metallic L-shaped profiles are used to dis- mens under axial (continuous lines) and eccentric (dashed lines)
tribute the load at the panel corners. Also in this case the applied increasing load. The firsts three graphs report the results of four

Fig. 8. Diagonal compression tests of panels without and with prestressing load.
F. Gara et al. / Engineering Structures 37 (2012) 193–204 199

1000 1000
WP08 WP12
Load [kN]

Load [kN]
500 500

2a.1 2b.1 3a.1 3b.1


2a.2 2b.2 3a.2 3b.2
0 0
1000 1000
WP16 WPN08
WPH08
Load [kN]

Load [kN]
500 500

4a.1 4b.1 Y.2 Y.1


4a.2 4b.2 X.2 X.1
0 0
5 10 15 20 25 5 10 15 20 25
Lateral deflection [mm] Lateral deflection [mm]

Fig. 10. Axial and eccentric compression tests: load-lateral deflection diagrams at mid-height of the panel.

compression tests, two with axial and two with eccentric load, 1000
carried out on specimens of the same typology (standard panels)
axial load

Ultimate Load [kN]


eccentric load
WP08, WP12, and WP16, respectively. The fourth graph shows
the results of tests on panels with non-undulated polystyrene
sheet (WPN08) and half a number of connectors (WPH08), one 650
under axial load and the other under eccentric load.
The maximum load (Ultimate Load) achieved in each test is
reported in Table 5, together with the mean value of the ultimate
300
load (Mean U.L.) reached by two specimens of the same typology, 12 14 16 18 20
and the mean ultimate uniformly distributed load (Mean U.U.D.L.), Slenderness ratio L/h
i.e. the ultimate load divided by the panel width. In both the cases,
with axial and eccentric load, the ultimate loads decrease by Fig. 11. Influence of panel slenderness ratio on axial and eccentric ultimate loads.
increasing the panel slenderness ratios, defined as L/h, where L is
the total height and h the overall thickness of the panel, as shown
in Fig. 11. It is worth noticing that in the case of the axial compres- due to compression, followed by the crushing of the concrete layer
sion test the ultimate load of the panels is strongly influenced by in compression and the rupture of the metallic mesh inside the
any small undesired eccentricity due to imperfections in the spec- other concrete layer, subjected to tension (Fig. 13a). As regards
imen and test set-up (load, restraints, etc.). However, due to the the eccentrically loaded specimens, Fig. 10 shows that the load
large difference between the values of undesired and imposed vs lateral deflection plots are nearly linear at the earlier stages of
eccentricity, the behaviour of the axially and eccentrically loaded loading; later, after the first crack has appeared, the panels exhibit
panels are significantly dissimilar. a non-linear behaviour. The failure of the panel occurs, in this case,
In particular, as regards axially loaded specimens, the lateral because of the rupture of the metallic mesh in the concrete layer in
deflection of the panel remains generally small under increasing tension. However, a not very ductile behaviour was observed, since
loading (Figs. 10 and 12a) up to load values close to the ultimate mesh failure occurs at the joints where the effective cross-section
load. Failure occurs as a result of overall buckling of the specimen, of the metallic wires may be reduced and the steel strength and
ductility are lower due to welding (Fig. 13b and c).
The different behaviour and different failure modes between
Table 5 wall panels under axial and eccentric loading are also shown in
Compression tests with axial and eccentric loading: ultimate loads. Fig. 14a which reports the displacement measured by the trans-
Specimen Panel Loading Ultimate Mean U.L. Mean U.U.D.L. ducers Svf and Svb placed in a vertical position on the front and back
type Load (kN) (kN) (kN/m) faces of specimens 2a.1 and 2b.2. Continuous lines refer to tests
2a.1 WP08 Axial 701 742 662.5 with axial load (2a.1), dashed lines to tests with eccentric load
2a.2 WP08 Axial 783 (2b.2). In the case of axially loaded specimens, the two concrete
3a.1 WP12 Axial 806 825 736.6 layers initially behave in the same way, both characterised by
3a.2 WP12 Axial 844
shortening deformations; only during a second phase does the
4a.1 WP16 Axial 855 881 786.6
4a.2 WP16 Axial 907 behaviour of the two concrete layers become different, with one
X.2 WPN08 Axial 736 657.1 concrete layer characterised by shortening deformation and the
Y.2 WPH08 Axial 765 683.0 other by elongation. In the case of eccentrically loaded specimens
2b.1 WP08 Eccentric 375 388 346.4
a different behaviour is observed from the beginning of the test.
2b.2 WP08 Eccentric 401
3b.1 WP12 Eccentric 460 503 448.7
In fact, with a low load level, the two concrete layers behave
3b.2 WP12 Eccentric 545 differently, one with shortening deformation and the other with
4b.1 WP16 Eccentric 524 577 515.2 elongation, and this denotes a predominant flexural behaviour of
4b.2 WP16 Eccentric 630 the panel.
X.1 WPN08 Eccentric 461 411.6
In Fig. 14b the longitudinal (slip) and transversal (separation)
Y.1 WPH08 Eccentric 591 527.7
components of the relative displacement between the two
200 F. Gara et al. / Engineering Structures 37 (2012) 193–204

3.0 3.0
150 kN 4a.2 150kN 4b.2
300 kN 300kN
450 kN 450kN
600 kN 600kN

Wall height [mm]


Wall height [mm]

0 0
20 0 20 20 0 20
Lateral deflection [mm] Lateral deflection [mm]
(a) (b)
Fig. 12. Lateral deflection at different load stages: (a) axial load and (b) eccentric load.

Fig. 13. Specimens after failure (a) axially and (b) eccentrically loaded; (c) mesh failure.

800 800
Svb Svf
Load [kN]

Load [kN]

400 Svf 400


Svb
2a.1 2a.1
2b.2 0 2b.2
0
-3 0 3 -0.5 0 2
Displacement [mm] Displacement [mm]

(a) (b)
Fig. 14. (a) Vertical deformation of the two concrete layers and (b) slip and separation between the concrete layers in axially and eccentrically loaded specimens.

concrete layers of the above mentioned specimens 2a.1 and 2b.2 longitudinal slip exhibits an initial nearly linear behaviour fol-
are plotted, for increasing axial (continuous line) and eccentric lowed by a non-linear behaviour until ultimate values of about
(dashed line) loadings. 1.5 mm. Compared to the slip, the separation is characterised by
These components are calculated from the vectorial decomposi- much lower values, which are practically negligible, meaning that
tion of the displacement recorded by transducers S4 and S5. The the two concrete layers deflect together.
F. Gara et al. / Engineering Structures 37 (2012) 193–204 201

Table 6 reaching a diagonal tensile failure. In fact, for the other specimens
Diagonal compression tests: cracking load, ultimate load, and failure modes. concrete crushing occurred at the load application point. Specimen
Specimen First cracking load (kN) Failure load (kN) Failure modes C1, which is similar to specimen 5.2 but with transversal stiffening
5.1 144 302 a* walls, nearly simulates a pure shear test thanks to the diffusion of
5.2 129 342 a the vertical load along the panel perimeter guarantee by the trans-
5.3 118 332 a versal walls. The highest stress values are reached in the central
5.4 168 306 a part of panel C1, where tensile and compression stress values are
C1 103 341 b
C2 137 225 b*
similar. Diagonal tensile failure occurred with a load of 341 kN.
In Fig. 15a the largest crack is marked with a thick dashed line.
a – localised concrete crushing; b – failure due to diagonal tension. On the contrary, for specimen 5.2, a concentration of the compres-
*
Failure of one of the two concrete layers.
sion stresses occurred around the load application point, causing
the crushing of the concrete at a load value of 342 kN, as shown
The high values of the ultimate loads obtained, the influence of in Fig. 15b.
the slenderness ratio on the ultimate loads as well as the low val- The influence of transversal walls can also clearly be seen in
ues of relative displacement between the concrete layers confirm Fig. 16 where both the vertical shortening deformation (Svf and
that these wall panels behave as semi-composite elements. How- Svb) and the horizontal elongation (Sof and Sob) measured on the
ever, some aspects deserve to be discussed. First of all, it is worth two sides of specimen C1 (with transversal walls) (Fig. 16a) and
noticing that the slip between the two concrete layers is restricted specimen 5.2 (without transversal walls) (Fig. 16b) are reported.
not only by the shear deformable internal layer but also by the In the panel with transversal walls, at the first stage of loading,
solid reinforced concrete beams at the bottom and top ends of the average horizontal elongation is nearly equal to the vertical
the panels. Consequently, the results of the tests presented in this shortening deformation; later, after the cracking of the concrete
paper may be considered as representative only for real buildings layer, it becomes larger. On the contrary, in panels without trans-
in which the connections between floor and wall panels are built versal walls, at the first stage of loading, the average horizontal
with solid reinforced concrete regions. Furthermore, it is important elongation is lower than the vertical shortening and becomes sim-
to underline that the reinforced concrete beams at the panel ends ilar after concrete cracking.
also cause a higher degree of flexural restraint between the floor The effects of prestressing can be observed in the results of tests
and wall panels, which may lead to high values of vertical load on specimens 5.3 and 5.4, prestressed with forces of 30 kN and
eccentricity and, thus, to ultimate loads significantly lower than 90 kN respectively. The first concrete cracking appeared at a
the values obtained in the tests. slightly higher load for specimen 5.4 than for specimen 5.3. How-
With regard to the tests on WPN08 and WPH08 wall panels, ever, a lower failure load was achieved by specimen 5.4 than by
under both axial and eccentric load, the values of ultimate loads specimen 5.3, since the prestressing force incremented the
are similar to or higher than those obtained by standard panels compression stresses around the load application point where
WP08. These panels can therefore be considered as valid alterna- the failure occurred.
tives for standard panels even if a greater number of specimens Finally, it may be noticed that in all the tests performed, failure
should be tested to arrive at some general conclusions. did not occur suddenly but was always preceded by extensive
As regards the diagonal compression tests, Table 6 reports the diffuse concrete cracking. All the specimens were in fact already
first cracking load, the failure load and the failure modes. It is micro-cracked before the tests, due to concrete shrinkage and
worth noticing that specimens 5.1 and C2 were characterised by specimen handling. The micro-cracks constitute weak zones in
premature failure of one of the two concrete layers of the panels, the concrete where cracks may preferentially occur. Nevertheless,
due to a small undesired eccentricity of the axial load. For this rea- the specimens revealed a high capacity for stress redistribution
son, these results are not taken into consideration in the following thanks to the metallic mesh inside the concrete layers. However,
comments. High first cracking loads were observed for all the it should be underlined that the results considered here are
specimens, while concerning the ultimate load the only significant relevant to tests on symmetrically loaded panels, with the two
result is that provided by specimen C1, which is the only one concrete layers equally loaded; in reality this is an ideal condition

Fig. 15. Diagonal compression test: crack pattern at failure for panels (a) with and (b) without transversal stiffening walls.
202 F. Gara et al. / Engineering Structures 37 (2012) 193–204

400 400
Sof Svf
Load [kN] Sof Svf

Load [kN]
200 200 Sob Svb
Sob Svb

C1 5.2
0 0
-4 0 2 -4 0 2
Displacement [mm] Displacement [mm]

(a) (b)
Fig. 16. Diagonal compression test: vertical shortening and transversal elongation of panels with (C1) and without (5.2) transversal stiffening walls.

that can only occur when concrete layers are connected by rein- A non-linear constitutive law [22] was considered for the con-
forced concrete beams. crete and a symmetric elasto-perfectly-plastic bilinear model for
the steel. The values of the mechanical parameters were deter-
mined from the results of the tests performed on the materials
3. Numerical simulation (paragraph 2.1). In particular the shear stiffness of the links joining
the two concrete layers was calculated by the equation
Compression tests were numerically simulated with a displace-  
GAl hl
ment based non-linear static analysis taking into account both Ki ¼ ð2Þ
geometrical and material non-linearities, performed with the c c
structural analysis programme Seismostruct [21]. Specimens were where G = 3.2 N/mm2 is the mean value among those obtained with
modelled with non-linear finite element models. In particular 20 the shear tests for material characterisation, Al is the influence area
beam elements were used for each concrete layer. The nodes at of the links, c is the internal layer thickness and the factor hl/c takes
each end of the two concrete layers were joined with two rigid into account the difference between the length (hl) of the links and c
elements to simulate the reinforced concrete beams while the (Table 7).
internal nodes were joined with shear elastic links. The links are In Fig. 18 the load vs lateral deflection graphs obtained from
axially rigid and shear deformable with the shear stiffness pro- compression tests are compared with the results obtained from
vided by the internal layer (Fig. 17). the numerical analysis. The behaviour of eccentrically loaded pan-
The specimen is restrained with a cylindrical pin at the base and els is well-approximated by the numerical model while for axially
a horizontal support at the top. In simulating axial compression loaded panels a lower agreement between experimental and
tests a small eccentricity of the vertical was considered in order numerical results is achieved. In fact, the behaviour and the ulti-
to simulate geometrical imperfections of panels and uncertainties mate loads of real panels are largely influenced by geometrical
of the load position. To simulate eccentric compression tests the imperfections (not perfectly flat concrete layers, variability of
load was applied at the axis of a concrete layer (Fig. 17). In both thicknesses, etc.) that are difficult to evaluate and take into consid-
cases the vertical load was applied incrementally until failure of eration in a numerical model. However the numerical simulations
the sample was reached. may be considered satisfactory. Furthermore, in order to evaluate
the critical load Pb1 of panels, a buckling analysis was also carried
out using the same numerical model but considering a linear elas-
axial eccentric tic behaviour of the materials. Values of the buckling loads ob-
load load tained with these analyses (Pb1) are reported in Table 8 and also
in Fig. 18.
It can be observed that the Pb1 values seem to be approached by
the curves obtained with the non-linear models, considering
approximately axial loads. In order to highlight the semi-compos-
beam ite behaviour of the panels, the values of the Euler buckling load
element (Pb2), calculated in the hypothesis of zero shear stiffness of the
links are reported in Table 8 (values of Pb2 are twice the Euler buck-
ling load for a single concrete layer). Furthermore, the buckling
300 cm

load Pb3 are reported in the same table, where Pb3 was calculated
15 15

by considering the panel to be entirely made of concrete. The coef-


shear ficient a = Pb1/Pb3 was introduced to easily estimate the buckling
deformable load of the panels. As expected, the values of this coefficient are
link ( Kl) less than 1 and decrease as the thickness of the internal layer
increases.
However, only in the case of undesired eccentricity, buckling
loads are close to the ultimate loads. In fact, due to the pronounced
rigid non-linear behaviour of the materials, the ultimate load of the
link eccentrically loaded panels is significantly lower than the buckling
load and can be estimated only with a non-linear analysis which
hl hl considers both geometrical and material non-linearities. Obviously
the reduction in the ultimate load will be more significant for
Fig. 17. Finite element model for compression tests. greater values of load eccentricity. As already mentioned, the value
F. Gara et al. / Engineering Structures 37 (2012) 193–204 203

Table 7 400
Characteristics of the numerical model.
experimental
WP08 WP12 WP16

Load [kN]
f.e.m.
hl (mm) 115 155 195
200 (Ec)
Kl (N/mm) 9660 5786 4095

f.e.m.
(0.4Ec) C1
0
1300 2
Displacement [mm]
Pb1 = αPb3
Load [kN]

f.e.m. 2a.1 Fig. 19. Diagonal compression tests: comparison between experimental and
numerical results.
650
2a.1
2b.2
f.e.m. 2b.2 well-simulated by the model with the original concrete elastic
modulus Ec.
0
1300 At this load a first crack appeared which was associated with a
distinct horizontal segment in the load and displacement plot.
Pb1 = αPb3 After the first crack, the experimental curve is non-linear due to
Load [kN]

3a.2 f.e.m. 3a.2 the progressive cracking of the concrete. The numerical model with
650 3b.2 a reduced elastic modulus (0.4Ec) simulates quite well the global
behaviour of the cracked panel.
f.e.m. 3b.2 The ultimate load may be estimated by simplified formulas
found in technical scientific literature. In particular for the diagonal
0 compression test the following formula may be used:
1300
Pb1 = αPb3
f.e.m. 4a.1 2  1120  70  2:3
Pu ¼ 2Bstot fct ¼ ¼ 360 kN ð3Þ
1000
Load [kN]

4a.1
4b.2
650 where B is the width of the panel and stot is the overall thickness of
f.e.m. 4b.2 the concrete layers. The result is close to the ultimate load experi-
mentally evaluated for specimen C1. However, in real buildings,
the strength of panels under vertical and horizontal forces involves
0 not only the shear resistance but also the bending resistance,
5 10 15 20 25
Displacement [mm] depending on the overall dimensions of the wall. Moreover, the
influence of openings for doors and windows must be considered.
Fig. 18. Axial and eccentric load tests: comparison between experimental and
numerical results.
4. Conclusions

The results of an experimental campaign on completed in situ


Table 8
sandwich panels with no-shear connectors, used as wall panels,
Critical loads and reduction coefficient.
have been presented. In particular, compression tests with axial
Specimen Mean U.L. (kN) Pb1 (kN) Pb2 (kN) Pb3 (kN) a and eccentric load and diagonal compression tests were per-
WP08 742 931 92.8 3653 0.25 formed. Some numerical simulations with linear and non-linear fi-
WP12 825 1082 92.8 7424 0.15 nite element models were also carried out.
WP16 881 1221 92.8 13,169 0.09
As regards compression tests, wall panels with different inter-
nal layer thickness (WP08, WP12, WP16) and with two different
configurations (WPN and WPH) were tested. High ultimate loads,
of the load eccentricity depends on the rotational restraint be- decreasing for increasing values of the slenderness ratios, were ob-
tween the wall and floor panels. tained. The numerical simulations indicated that the ultimate
With regard to the diagonal compression tests, the behaviour of loads of axially loaded panels are close to the buckling loads which
the specimen with the transversal concrete wall (C1) was numeri- can be determined by performing a linear buckling analysis or by
cally simulated with an elastic finite element model using the using the coefficient a. Differently, the ultimate loads of eccentri-
structural analysis programme SAP2000 [23]. The panel was mod- cally loaded panels, which are significantly lower than the buckling
elled with shell elements with a thickness equal to the overall loads, can be simulated only by performing a non-linear analysis.
thickness of the two concrete layers and with an elastic modulus Additional research is needed to develop simple, effective and ra-
equal to that of the concrete used. The transversal concrete walls tional methods for predicting the ultimate load of wall panels for
were modelled with beam elements. Vertical static loads were ap- different values of load eccentricity. The results of the experimen-
plied on several nodes for a total force equal to the ultimate load tal tests and numerical simulations indicated that a partial degree
experimentally obtained (341 kN). The value of the horizontal ten- of composite behaviour was attained by the tested panels even if
sion in the central node of the model is 2.3 N/mm2 which is nearly non-shear connectors are used in the interior layer. However, this
equal to the ultimate tensile strength of the concrete. In Fig. 19 the semi-composite behaviour is due not only to the internal layer, but
load-shortening plot experimentally obtained is compared with also to the reinforced concrete beams at the ends of the panels.
the results obtained with the numerical model, considering an Additional investigations are needed to develop simple, effective
elastic modulus Ec and a reduced modulus 0.4Ec to take into ac- and rational methods for predicting the ultimate load of wall pan-
count the cracking of the concrete. Up to a load of about 100 kN els for different values of load eccentricity and to study the behav-
the behaviour of the specimen is typically linear elastic and is iour of panels without reinforced concrete beams.
204 F. Gara et al. / Engineering Structures 37 (2012) 193–204

As regards diagonal compression tests, simple wall panels, pre- [5] PCI Committee on Precast Concrete Sandwich Wall Panels. State of the art of
precast/prestressed sandwich wall panels. PCI J 1997;42(2):92–133.
stressed wall panels and panels with transversal stiffening walls
[6] Benayoune A, Aziz A, Samad A, Trikha DN, Abdullah Abang Ali A, Ashrabov AA.
were tested. In all these cases high cracking loads were observed. Structural behaviour of eccentrically loaded precast sandwich panels. J Constr
The panels also showed a high capacity for stress redistribution Build Mater 2006;20:713–24.
thanks to the metallic mesh inside the concrete layers. However, [7] Benayoune A, Samad AA, Abang Ali AA, Trikha DN. Response of pre-cast
reinforced composite sandwich panels to axial loading. J Constr Build Mater
only one specimen with transversal stiffening walls showed a ten- 2007;21:677–85.
sile diagonal rupture while the other specimens showed a com- [8] Benayoune A, Abdul Samad AA, Trikha DN, Abang Ali AA, Ellinna SHM. Flexural
pression failure at the load application region. The numerical behaviour of pre-cast concrete sandwich composite panel – experimental and
theoretical investigations. J Constr Build Mater 2008;22:580–92.
simulation of the test reaching the tensile diagonal rupture showed [9] Salmon DC, Einea MK, Tadros A, Culp TD. Full scale testing of precast concrete
that an effective concrete modulus of elasticity may be considered sandwich panels. ACI J 1997;94:354–62.
to simulate the global behaviour of the cracked panel and that the [10] Einea A, Salmon DC, Tadros MK, Culp TD. A new structurally and thermally
efficient precast sandwich panel system. PCI J 1994;39(4):90–101.
ultimate load may be estimated on the basis of the tensile strength [11] Bush TD, Stine GL. Flexural behaviour of composite precast sandwich panels
of the concrete. However, since in real buildings the behaviour of with continuous truss connectors. PCI J 1994;39(2):112–21.
the panels under vertical and horizontal in-plane forces is strongly [12] Kabir MZ. Mechanical properties of 3D wall panels under shear and flexural
loading. In: 4th Structural specialty conference of the Canadian society for civil
influenced by the overall dimensions of the wall and by openings engineering, Montreal; 2002.
for doors and windows, further investigations on panels with dif- [13] Giacchetti R, Menditto G. Indagini sperimentali su pannelli sandwiches
ferent configurations are recommended. realizzati con la tecnica dello spritzbeton. Atti V Convegno CTE, Firenze;
1984. p. b.13–b.24 [in Italian].
[14] Ceccoli C, Mazzotti C, Savoia M, Dallavalle G, Perazzini G, Tomassoni C.
Acknowledgments Indagini sperimentali su una tipologia di pannelli in c.a. alleggeriti gettati in
opera. Atti XIV Convegno CTE, Mantova; 2002. p. 557–67 [in Italian].
The financial support provided by ‘‘Schnell House’’ S.p.A., based [15] Bassotti O, Ricci M. Caratteristiche di pannelli sandwich in cls. alleggerito e
loro applicazioni costruttive. Atti XIV Convegno CTE, Mantova; 2002. p. 579–
in San Marino, is gratefully acknowledged. The technical support of 87 [in Italian].
the laboratory staff at the Dept. of Architecture, Construction and [16] Refaifar O, Kabir MZ, Taribakhsh M, Tehranian A. Dynamic behaviour of 3D-
Structures, Università Politecnica delle Marche, is greatly appreci- panel single-storey system using shaking table testing. Eng Struct
2008;30:318–37.
ated. The opinions, findings and conclusions contained in this pa- [17] EN ISO 12504-1. Testing concrete in structures. Cored specimens. Taking,
per are those of the authors, and do not necessarily reflect the examining and testing in compression. CEN European Committee for
views of the sponsors. Standardization; 2000.
[18] EN 12390-6. Testing hardened concrete. Tensile splitting strength of test
specimens. CEN European Committee for Standardization; 2000.
References [19] EN ISO 15630-2. Steel for the reinforcement and prestressing of concrete – test
methods – part 2: welded fabric. CEN European Committee for
[1] Ranzi G. Locking problems in the partial interaction analysis of multi-layered Standardization; 2002.
composite beams. Eng Struct 2008;30:2009–911. [20] ASTM C273/C273M-07. A standard test method for shear properties of
[2] Sousa Jr JBM, da Silva AR. Analytical and numerical analysis of multilayered sandwich core materials. Committee D30.09 on Sandwich Construction; 2007.
beams with interlayer slip. Eng Struct 2010;32:1671–80. [21] Seimosoft Seimostruct. A computer program for static and dynamic nonlinear
[3] Gara F, Ranzi G, Leoni G. Displacement-based formulations for composite analysis of framed structures; 2004. <http://www.seismosoft.com/>.
beams with longitudinal slip and vertical uplift. Int J Numer Methods Eng [22] Mander JB, Priestley MJN, Park R. Theoretical stress–strain model for confined
2006;65:1197–220. concrete. J Struct Eng 1988;114(8):1804–26.
[4] Schnabl S, Saje M, Turk G, Planinc I. Analytical solution of two-layer beam [23] SAP2000. Computer and Structures, Inc. CSI analysis reference manual. SAP
taking into account interlayer slip and shear deformation. J Struct Eng, ASCE 2000, Berkeley, California; 2004.
2007;133(6):886–95.

Potrebbero piacerti anche