Professional Documents
Culture Documents
Abstract: Precast/prestressed concrete sandwich panels consist of two concrete wythes separated by a rigid insulation foam layer and are
generally used as walls or slabs in thermal insulation applications. Commonly used connectors between the two wythes, such as steel
trusses or concrete stems, penetrate the insulation layer causing a thermal bridge effect, which reduces thermal efficiency. Glass fiber-
reinforced polymer GFRP composite shell connectors between the two concrete wythes are used in this research as horizontal shear
Downloaded from ascelibrary.org by SERC on 03/26/14. Copyright ASCE. For personal use only; all rights reserved.
transfer reinforcement. The design criterion is to establish composite action, in which both wythes resist flexural loads as one unit, while
maintaining insulation across the two concrete wythes of the panel. The experiments carried out in this research show that hybrid
GFRP/steel reinforced sandwich panels can withstand out-of-plane loads while providing resistance to horizontal shear between the two
concrete wythes. An analytical method is developed for modeling the horizontal shear transfer enhancement using a shear flow approach.
In addition, a truss model is built, which predicts the panel deflections observed in the experiments with reasonable accuracy.
DOI: 10.1061/ASCE1090-0268200812:5570
CE Database subject headings: Composite structures; Concrete structures; Sandwich panels; Slabs; Walls; Fiber reinforced
polymers; Hybrid methods.
Fig. 2a, and others used a glass FRP bent bar as shown in Fig.
2b.
Conventional Materials
Bank et al. 1997 developed three-dimensional FRP grating
cages for reinforcing concrete elements; they recommended that Mild steel bars 13 mm diam were used as flexural reinforcement
the design of the FRP cage should be studied further and alterna- with a measured yield strength f y = 455 MPa. Normal weight con-
tive arrangements of the transverse vertical and cross reinforc- crete was used with a compressive strength f c = 28 MPa at the
ing members should be considered. Constructability and cost time of testing. The insulation layer was made of 76 mm thick
analysis including life-cycle costs of the three-dimensional FRP expanded polystyrene foam sheets with a compressive strength of
cages were studied by Shapira and Bank 1997; the results of the 230 kPa and a flexural strength of 520 kPa. A polyurethane dis-
cost analysis showed that FRP cages are a competitive alternative persion compound, which is a highly elastic water-reducible
to conventional steel reinforcement. primer and coating commonly applied to plastics was used as a
Whitehead and Ibell 2005 developed continuous FRP rectan- binder. The compound was used to immerse the GFRP composite
gular and circular helixes made of aramid FRP composites as shells to reduce, to the extent possible, the alkali reaction between
transverse reinforcement for shear resistance. The continuous he- concrete and GFRP composite. A nonabrasive, high-modulus,
lixes were used in ordinary FRP-reinforced concrete beams and high-strength, moisture-tolerant, epoxy paste adhesive was used
FRP-prestressed concrete beams. to attach the steel reinforcement to the GFRP composite shells for
Keller et al. 2006 developed a hybrid pultruded GFRP/steel construction efficiency. The adhesive had a tensile strength of
joint for transferring compression and shear forces in thermal 37 MPa, a modulus of elasticity of 2,830 MPa, and a 1.3% elon-
insulation sections of concrete slab structures. They observed that gation at break.
the hybrid joint could lead to competitive solutions despite the
relatively high material cost.
The present paper is focused on the structural performance of Assembly and Construction
hybrid GFRP/steel concrete sandwich panels; the thermal insula-
tion efficiency of the sandwich panels was not examined in this
GFRP Composite Shells
research. However, a study by Salmon et al. 1997 has shown
that sandwich panels with FRP bent bar connectors had a thermal GFRP shells were constructed from epoxy-cured or urethane-
efficiency 1.9 times that of panels with concrete stems and 1.2 cured glass fiber fabrics. The epoxy-cured fabric was cut in
times that of panels with steel truss connectors. 152 mm 1.22 m strips that were saturated with epoxy resin in a
saturator. The wet fabric strips were wrapped around 152 mm
square steel tubes covered with plastic film, with the unidirec-
Glass Fiber-Reinforced Polymer Composite tional fibers applied in the hoop direction as shown in Fig. 4a;
Systems this formed the fibers into 152 mm 152 mm 152 mm shells,
two layers thick. For the urethane-cured composite system, the
Two GFRP composite systems were used in this study: a unidi-
glass fiber fabric strip came preimpregnated with urethane resin in
rectional E-glass fibers with an epoxy matrix, and b unidirec-
a sealed container and the precut strips were used to create the
GFRP shells in a similar manner; water was sprayed during and
after wrapping to cure the urethane resin. The urethane-cured
GFRP shells were further wrapped with shrink wrap for consoli-
dation, as shown in Fig. 4a. The GFRP shells were cured in
place for three days, then separated from the steel tubes and cured
for at least seven additional days. Two 38 mm diameter holes
were drilled on the top and bottom faces of the GFRP shells, as
shown in Fig. 4b. The GFRP shells were placed in the sandwich
panels so that the holes were located in the horizontal plane of the
Fig. 2. Configuration of existing structural connectors: a steel two concrete layers. The holes facilitated the flow of concrete
connector; b FRP bent bar connector adapted from Salmon et al. during construction and created a concrete stub, which improved
1997 the bond of the concrete outside the GFRP shells to the concrete
inside. Observations after testing verified that no spalling of the 7b that of a double-cage sandwich panel. The four panels were
concrete outside the GFRP shells had occurred, which would oth- cast in a single casting operation to achieve uniform concrete
erwise be highly probable. properties. Concrete was cast into the formwork containing the
hybrid cages and vibrated until the bottom layer was 64 mm
Downloaded from ascelibrary.org by SERC on 03/26/14. Copyright ASCE. For personal use only; all rights reserved.
thick, after which the insulation was placed on the concrete sur-
Hybrid GFRP/Steel Cages
face; the top layer of concrete was then cast and vibrated and the
Mild steel bars were used as flexural reinforcement and the GFRP panel was cured for a minimum of 28 days.
shells as horizontal shear transfer connectors. A steel bar was
attached to each of the four corners of the GFRP shell using
structural adhesive to form a hybrid cage in the manner shown in Composite versus Noncomposite Section Capacity
Fig. 5a. A hybrid GFRP/steel cage comprised of four 13 mm
steel bars, 4.57 m long, attached to 15 GFRP composite shells at The criterion used for design of the GFRP shells was to establish
the four corners. The bond between steel bars and GFRP shells is composite action, in which both reinforced concrete wythes resist
advantageous for construction purposes and structural perfor- flexural loads as one unit, while providing thermal insulation
mance. Fig. 5b shows the six hybrid cages constructed after across the two wythes of the sandwich panel. A strain compatibil-
rigid foam insulation was inserted inside the GFRP shells. Addi- ity analysis of the single-cage sandwich panel of Fig. 7a shows
tional foam insulation was placed outside the GFRP shells during that when composite action is achieved Fig. 1b, the theoretical
casting to form the insulation layer between the two concrete bending moment capacity for out-of-plane loads is 24 kN/ m. If
wythes. A small amount of concrete seepage was observed be- insufficient shear transfer is achieved, then noncomposite action
tween the vertical walls of the GFRP shells and the rigid foam would occur, with both wythes resisting flexural stresses indepen-
insulation. A foam sealant and use of pultruded GFRP sections, dently Fig. 1a; the theoretical bending moment capacity in this
which have flatter surfaces, could minimize concrete seepage and case is 7 kN/ m. Similarly, a strain compatibility analysis of the
improve thermal and construction efficiency. double-cage sandwich panel of Fig. 7b shows that when com-
posite action is achieved, the theoretical bending moment capac-
Sandwich Panels ity is 43 kN/ m, and if noncomposite action were to occur, the
theoretical moment capacity is 11 kN/ m. In these calculations,
Four 4.57 m long by 0.61 m wide sandwich panels were built. the ultimate concrete strain was assumed equal to 0.003 mm/ mm
Fig. 6 shows a longitudinal section of the panel and the GFRP and a mild steel reinforcement ultimate stress of 1.25f y was as-
shell spacing. Two sandwich panels were reinforced with one sumed to account for strain hardening.
hybrid cage single-cage panels, and two sandwich panels were
reinforced with two cages double-cage panels. The cross section
of the sandwich panels consisted of two 64 mm concrete layers Design of GFRP Shell Connectors
and a middle 76 mm insulating layer made of expanded polysty- A preliminary estimate of the capacity of the shell connectors for
rene, for a total sandwich panel thickness of 204 mm. Fig. 7a resisting horizontal shear was based on two additional experi-
shows the cross section of a single-cage sandwich panel, and Fig. ments carried out on the hybrid GFRP/steel cages themselves
without any concrete wythes; the number of steel bars and GFRP
shells was identical to a single cage used in the four tests with
concrete wythes. The same four-point load setup shown in Figs. 6
and 8 was used. The capacity of the GFRP shells observed in the
tests of the cages without the concrete wythes was considered to
be a lower bound for the sandwich panels with the two concrete
Downloaded from ascelibrary.org by SERC on 03/26/14. Copyright ASCE. For personal use only; all rights reserved.
Fig. 6. Longitudinal view of insulated hybrid GFRP/steel sandwich Fig. 8. Epoxy-cured GFRP double-cage sandwich panel at the maxi-
panel showing spacing of GFRP shells mum midspan deflection of 260 mm
els, even though the bottom concrete wythe cracked near mid- was constructed using the configuration of the internal hybrid
span, as shown in Figs. 10 and 11. At the supports, even at the GFRP/steel reinforcement shown in Fig. 6, for both single-cage
maximum midspan deflections shown in Table 3, there was no and double-cage sandwich panels with the cross-sectional details
evidence of cracks in the concrete or GFRP composite laminate, given in Fig. 7. The truss shown in Fig. 12 was used for this
or buckling of the GFRP shells. This was determined by removing analysis. The weight of the panels w was included in the analysis
the concrete around the cages after testing. Thus, the GFRP shells by distributing it to the nodes of the bottom chord. In Fig. 12,
were successful as horizontal shear transfer connectors between Ts = tension tie with an area equal to the steel area of two 13 mm
the two concrete wythes. steel bars for the single-cage panel, or four 13 mm steel bars for
the double-cage panel; V f = vertical tension tie with an area equal
to four GFRP layers two layers for each of the two vertical sides
Analytical Results for the single-cage panel, or eight GFRP layers double-cage
panel, times the height of the shell 152 mm.
Horizontal Shear Transfer Capacity The upper compressive strut denoted as C, has a width of
610 mm and a depth equal to that of the uncracked compression
The ultimate moment achieved in the single-cage panel tests was zone assuming a fully cracked section. D denotes the diagonal
99% of the theoretical moment capacity assuming composite ac- compression strut, which has a thickness equal to the panel width
tion; for the double-cage panel tests, the ultimate moment of 610 mm, and a width of the diagonals equal to 157 cos mm,
achieved was 97% of the theoretical moment capacity assuming which varies according to angle 64, 34, and 27, with a cor-
composite action; thus, the GFRP shells were effective in trans- responding width of 68, 130, and 140 mm.
ferring the horizontal shear between the two concrete wythes. Eq. A bilinear approximation is adopted to model the load-
3 could be used in design to determine the number of GFRP deflection response. The midspan deflection at yield my was
shells required to achieve composite action. This is conservative evaluated using the analytical deflection expression
because of the reduced value of the allowable stress used in the
design of the GFRP shells, as obtained from the V-notch rail shear 0.5Pya
test, and the fact that concrete bond to the rigid foam insulation my = 3L2 4a2 4
24EcIcr
was ignored in the design. A multidimensional laminate layout of
the GFRP shells with fibers in the 0, +45, 45, and 90 directions where Icr = moment of inertia of the fully cracked section, Ec
from the horizontal could lead to improved performance, but this = modulus of elasticity of the concrete, 0.5Py = half of the total
was not pursued in the present study. applied load at yield, L = clear span length 4.42 m, and a
Typical demands for the same panel span height used in the = shear span 1.83 m, as defined in Fig. 6. For the postyielding
tests are calculated assuming uniformly distributed unfactored portion of the lateral load-deflection behavior, the secant modulus
loads. The uniformly distributed live load for roofs including spe- of the steel bars was used in the truss model; this was evaluated
cial purpose roofs is between 1.0 and 4.8 kPa; for office building using the measured strains of the steel bars, which reached maxi-
lobbies, corridors, and offices, the live load is between 2.4 and mum tensile strains up to 1.6%.
4.8 kPa ASCE 2005. The simplified design wind pressure for Predicted and experimental results are shown in Fig. 13 for
walls and roofs of buildings with a height less than 18.3 m, for a both single-cage and double-cage panels. For the single-cage
basic wind speed of 193 km/ h is between 1.1 and 1.5 kPa, de- panel, the deflection at yield from Eq. 4 is 92% of that shown in
pending on the roof slope ASCE 2005. Based on the experimen- Figs. 13a and b from the tests, whereas for the double-cage
tal results at the chosen deflection limit of L / 180 for roofs or panel, it is 95% of the deflection in Figs. 13c and d from the
Lc / 150 for slender walls, the capacity of the panels tested is be- tests. Thus, the midspan deflections obtained from the truss analy-
tween 8.0 and 8.6 kPa for the single-cage panels, and 15.4 and sis are found to be in good agreement with the measured deflec-
15.8 kPa for the double-cage panels. Thus, the capacity of the tions both at yield and at ultimate conditions.
panels is greater than the demand and is satisfactory for design.
Conclusions
Truss Models for Deflection
Truss analogy was used to predict the midspan deflection of the The GFRP shells developed in this research were successful in
sandwich panels up to failure. The truss analogy method has been providing greater than needed horizontal shear transfer between
Acknowledgments