Experimental Tests on Bending Behavior of Profiled Steel Sheeting Dry Board Composite Floor with Geopolymer Concrete Infill

Profiled Steel Sheet Dry Board (PSSDB) system is a lightweight composite structure comprises Profiled Steel Sheeting and Dry Board connected by self-drilling and self-tapping screws. This study introduced geopolymer concrete, an eco-friendly material without cement content as an infill material in the PSSDB floor system to highlight its effect onto the PSSDB (with full and halfsize dry boards) floor system’s stiffness and strength. Experimental tests on various full scale PSSDB floor specimens were conducted under uniformly distributed transverse loads. Results illustrate that the rigidity of the panel with geopolymer concrete infill with half-size dry board (HBGPC) increases by 43% relative to that of the panel with normal concrete infill with full-size dry board (FBNC). The developed finite-element modeling (FEM) successfully predicts the behavior of FBGPC model with 94.8% accuracy. Geopolymer concrete infill and dry board size influence the strength panel, infill contact stiffness, and mid-span deflection of the profiled steel sheeting/dry board (PSSDB) flooring system.


INTRODUCTION
This paper deals with a lightweight, slim, easily constructed composite load-bearing structural system known as the Profiled Steel Sheet Dry Board (PSSDB) system.The system consists of profiled steel sheet (PSS) connected to dry board (DB) by means of self-drilling and selftapping screws, see Figure 1.The PSSDB in its original form was envisaged by Wright et al. (1989) as an alternative to the traditional timber plank and joist floor system in the United Kingdom.Research works have since been conducted on the PSSDB system in its applications as flooring, roofing, and walling units (Ahmed et al. (2000 and2002), Wan Badaruzzaman et al. (2003), Ahmed and Wan Badaruzzaman (2003, 2005and 2013), Akhand et al. (2004), Gandomkar et al. (2011Gandomkar et al. ( , 2012Gandomkar et al. ( , 2013aGandomkar et al. ( and 2013b)), Seraji et al. (2012)).
This paper reports on the research work conducted on the PSSDB floor system introducing geopolymer concrete (Abdullah et. al. 2011) infill in the trough of the PSS, as an alternative to the previously adopted normal concrete.The geopolymer concrete, in addition to being eco-friendly is high strength and expected to improve the structural performance of the PSSDB floor system in terms of stiffness and strength.Previous research involving normal concrete infill in PSSDB floors have been reported in Shodiq (2004), Wan Badaruzzaman et al. (2003), Seraji et al. (2013), and Gandomkar et al. (2011 and2012).Shodiq (2004) has demonstrated that the stiffness and strength of the PSSDB floor system with normal concrete infill is 20.2% and 61.0% higher than the one without infill.This was due to the increased stiffness contributed by the concrete infill and its effect in delaying the onset of local buckling of the upper flange under compression of the PSS.The PSSDB system contains PSS that has high strength and low stiffness.This characteristic makes the PSSDB structure more vulnerable to fail as a result of deformation and local buckling of the steel rather than reaching the ultimate strength (Seraji et al. 2013).Structures 13 (2016) 272-295 The present study aims to understand the behavior in terms of stiffness and strength of various configurations of the PSSDB floor system (with full and half-size dry boards) under the effect of normal concrete and geopolymer concrete.A PSSDB panel with normal concrete infill will be used for comparison.Full scale experimental tests on simply supported PSSDB floor specimens were conducted under uniformly distributed transverse loads to achieve this aim.Results from one of the experimental specimen were used to validate the finite element model (FEM) which will be utilized for more parametric studies in future.

GEOPOLYMER CONCRETE
Ordinary Portland cement (OPC) is a concrete material that is extensively used in the construction industry worldwide.However, OPC has a negative effect on the environment.The global production of OPC contributes approximately 1.35 billion tons annually, or approximately 7% of the total greenhouse gas emissions, to the atmosphere (Malhotra, 2002).To mitigate the total amount of carbon dioxide released by the cement industry, pulverized fuel ash can be used as an alternative cement material to produce geopolymer concrete.
A geopolymer binder can be used as a substitute to cement binder.The use of fly ash minimizes the dependency on cement materials to produce concrete.Geopolymer has emerged as a new eco-friendly engineering material and serves an important function in the construction industry and in the manufacture of environmentally sustainable construction products (Davidovits 2002).
A geopolymer is a binder that consists of a non-organic material formed from solid matter synthesized by the reaction of "aluminosilicate" minerals such as pulverized fuel ash; thus, the type of activator used is significant in the geopolymer production process (Palomo et al. 1999, Fernández-Jiménez dan Palomo 2005, Ruiz-Santaquiteria et al. 2012).To activate the PFA and POFA, an available commercial of sodium silicate (Na2SiO3) and sodium hydroxide (NaOH) solution was chosen.
A high-rate reaction occurs with the use of dissolved silicate as opposed to the use of alkaline hydroxide.Xu (2000) pointed out that the reaction between materials, as well as amount of solution, both increased when sodium silicate was added to sodium hydroxide (NaOH).Sodium metasilicate (Na2SiO3), which is a mixture of NaOH and sodium silicate, is a widely used alkaline activator (Pacheco-Torgal et. al. 2008;Rangan 2008).NaOH solution is considerably better than potassium hydroxide (KOH) solution because of the solvency degrees of these substances.In addition, the purity of the solid form of NaOH is between 97% and 99% (Criado et. al. 2010, Rangan 2008, Zheng et. al. 2010).Palomo et al. (1999) found that the molarity of NaOH is optimal at 8, 10, 12, 14, and 16 M.Raijiwala and Patil (2011) suggested that the compressive strength of geopolymer concrete increases with increasing NaOH molarity.The results of a compressive strength test on samples with 8 M to 16 M NaOH show that the sample with 12 M NaOH has a compressive strength (33.16 MPa) close to that of normal concrete grade 30.However, 8 M to 16 M NaOH can also be used to produce geopolymer depending on the type of the primary material used (Hardjito & Rangan 2005).
Latin American Journal of Solids and Structures 13 (2016) 272-295 Hardjito (2005) proposed that the suitable ratio of silicon dioxide (SiO3) to sodium oxide (Na2O) is 2. The sodium silicate/NaOH solution was prepared 24 hours earlier to achieve satisfactory geopolymer concrete characteristics.Hardjito and Rangan (2005) stated that the ratio of the silicate/hydroxide solution determines the strength of geopolymer compression.The molar ratio of 2.5 reportedly increases the strength of a geopolymer (Hardjito & Wallah 2004).

Sample Preparation
To investigate the behavior of various configurations of simply supported (pin-roller supports) PSSDB floor system of 2500 mm span under the influence of geopolymer concrete infill, five (5) different full scale floor specimens were tested in the laboratory, see Table 1.Flexural bending tests under uniformly distributed transverse loads were performed to obtain the load deflection graphs to facilitate the experimental stiffness values and ultimate load (strength) of these panels.
All samples consist of 1 mm thick Peva 50 profiled steel sheet (PSS) with dimensions 2600 mm x 1000 mm.The dry board (DB) used was 12 mm thick PRIMAflex; the size of the fullboard was 2600 mm × 1000 mm, and that of the half-board 1300 mm × 1000 mm.The DB was attached to the top flanges of the PSS at a distance of 200 mm center-to-center through self-drilling and self-tapping screws (type DS-FH 432).Geopolymer concrete (12M) and normal grade 30 concretes were used as the infill materials.The cross-sectional shape and dimensions of the Peva 50 PSS is shown in Figure 2.
Table 2 gives the specification and material properties of every component of the PSSDB

Testing and Observations
The samples were tested under uniformly distributed load via a whiffle tree set up utilizing rectangular hollow 100 mm x 100 mm steel sections as shown in Figures 6 and 7   The load was generated using a manually operated hydraulic jack until the ultimate failure of the test panels.The deflection transducers and load cell readings were recorded directly by a data logger.Local buckling was observed initially developed at the under the compressed PSS top flanges, propagating to parts of the webs under compression of all samples (see Figure 9) Latin American Journal of Solids and Structures 13 (2016) 272-295 within the vicinity of the mid-span (where compressive stress is the highest) as the samples were loaded until failure.It was noticed that the concrete infill had moved out horizontally, slipping off the PSS at both ends of the samples (see Figure 10).Longitudinal crack-lines on concrete top surface for the half-board samples along the span direction were also observed (see Figure 14).

Symmetrically Deflection
As mentioned before, the use of transducers T3 and T4 is to record the symmetrical deflection at quarter span locations.The load-deflection curves for the five cases are shown in Figures 15 till 19.From the graphs, the deflections are seen almost identical for symmetrically positioned T3 and T4 for all samples.This situation shows the deflections recorded by the transducers at quarter span locations are considered symmetrical, hence verifying the reliability of the experimental data.

Load-Mid-Span Deflection Curves
The results of the experimental tests were obtained as shown in the plotted graphs below.Based on BS 8110 Part 1 (1997), the floor limiting deflection (for non-brittle partitions) is span (L)/250.Therefore, the serviceability deflection limit in this case is 10 mm for the span of 2500 mm.All samples underwent linear elastic, followed by non-linear ductile load-deflection characteristics before finally reaching ultimate failure.Upon failure, there was a big increase in deflection accompanied by reduction in load.Figure 20 shows the load-deflection plots for panels with full-board (FB).It can be seen that the rigidity and ultimate load of the FBGPC panel increased by 20% and 3.4% respectively relative to that of the FBNC panel, and by 83% and 35% respectively relative to that of the CS panel.This full-board geopolymer concrete, FBGPC panel can sustain up to 3.6 kN/m 2 (which is much higher than the normal 1.5 kN/m 2 loading for domestic housing) before reaching the limiting deflection.
Figure 21 shows the load-deflection plots for the panels with half-board (HB).The rigidity and ultimate load of HBGPC panel exhibited a remarkable increase of 117% and 46% respectively higher than the CS panel and 30% and 7.33% respectively higher than the HBNC panel.This half-board geopolymer concrete, HBGPC panel can sustain up to 4.0 kN/m 2 (higher that the full-board panel) loading before reaching the limiting deflection.Figure 22 shows the comparison of the PSSDB panel with full and half-boards infilled with normal concrete.It can be seen that the rigidity and ultimate loads of the HBNC panel increased by 10% and 4% relative to that of the FBNC panel, and by 67% and 36% relative to that of the CS panel.
Figure 23 shows the comparison of the PSSDB panel with full and half-boards infilled with geopolymer concrete.It can be seen that the rigidity and ultimate loads of the HBGPC panel increased by 19% and 8% relative to that of the FBGPC panel, and by 117% and 46% relative to that of the CS panel.The better performing samples are the ones with geoploymer concrete as expected, and this is due to two main reasons, firstly, the E value of the geoploymer concrete and the natural adhesion/bonding in between the geopolymer concrete and the DB and PSS surfaces are better than that of the normal concrete.These factors helped increased the composite EI value of the PSSDB floor system.The better performing half-board compared to full-board is mainly due to the higher rigidity of the PSSDB floor that has more concrete since concrete has higher E value than the DB.

Latin
In terms of the floor design, as has been mentioned earlier, the main criteria governing the design of the PSSDB floor would be serviceability limit state.From the above graphs, if the deflection of the floors is limited to L/250, the ratios of the load at that level compared to the ultimate load are given in Table 3, which indicates that the PSSDB floor systems are very safe in terms of structural safety.

Local Buckling Characteristic
The non-linear characteristics observed in all the load-deflection plots are due to the load buckling of the PSS under compression as described in section 3.2.It can be seen that the control sample developed local buckling only at the mid span position.On the other hand the infill model (full-board and half-board) demonstrated quarter span position load buckling as well as that took place sometime after the mid span local buckling.
The local buckling occurred various critical load depending on the composition of the specimen.It can be seen from Table 3 that the control sample (CS) buckled at the lower load of 7.15 kN/m 2 kN/m 2 compared to the other infill samples where samples FBNC, HBNC, FBGPC and HBGPC buckled at , 8.00 kN/m 2 , 8.48 kN/m 2 , 8.59 kN/m 2 and 9.33 kN/m 2 respectively.
The precise of concrete infill further delayed the onset of local buckling compared to the presence of the dry board (PRIMAflex) alone.Wright et. al. (1989) attributed the delay the onset of local buckling due to the presence of the dry board when comparing bears profile steel sheeting to profiled steel sheeting dry board system.
The samples with infill of all types were observed to reduced local buckling almost at the same critical load.It is important to note that the local buckling in the case of these samples also took place at the quarter span positions the ultimate loads in these cases as much higher that the control sample.

Load-Concrete Infill End Horizontal Movement Curves
The measurement of the load-concrete infill end horizontal movements are only recorded for full-board involving normal and geopolymer concrete samples.These horizontal movements in a way reflect the strength of the natural bond in between the concrete infill, PSS and the DB, where the case of higher horizontal deflection value of the concrete infill at the end support indicate lesser bonding strength in between the concrete infill and the other components.
Figure 24 shows the plots of load vs concrete infill end horizontal movement for the FBNC and FBGPC panels.The curves show an increasing outward movement from the early stage of loading until the load was reduced with a drastic increase in the horizontal deflections, implying that the panel had failed.The FBGPC panel recorded an ultimate concrete infill end horizontal movement of 0.68 mm, whereas the FBNC panel recorded a value of 3.82 mm.These results clearly show that the natural bonding in between the geopolymer concrete infill and the PSS and DB is stronger than that of the normal concrete.This justified the increase in rigidity of the PSSDB system with geopolymer concrete infill as compared to the normal concrete infilled PSSDB system.

FEM Development for FBGPC Model
The experimental test involving PSSDB floor system constructed from Peva 50 PSS, Primaflex DB, and DS-FH432 self-drilling and self-tapping screws at 200 mm c/c with a span of 2.5 m, infilled with 12M geopolymer concrete (FBGPC) was chosen for the verification of the nonlinear FEM.This FBGPC experimental model had been presented in Section 3. The ABAQUS Latin American Journal of Solids and Structures 13 (2016) 272-295 (2008) software was used for the FEM.The verified FEM would be utilized for further parametric studies in future.

Materials Modeling and Elements Selection
This section presents the selections of appropriate elements for the finite element modeling of the PSSDB flooring.Thin-shell S4R-type element in ABAQUS software was applied to both the PSS and DB (Seraji et. al., 2013).The S4R is a three-dimensional four-node (with six degrees of freedom per node), doubly curved, general-purpose shell, reduced integration and finite membrane strain elements (Figure 25) with bilinear interpolation element.The infill materials applied to the PSSDB floor in this study was normal grade 30 and geopolymer concretes.The uniaxial tension of the concrete's stress-strain relationship is also assumed to be linear up to the failure point, which is followed by the beginning of the macrocracking in the concrete (Crisfield 1982).Based on Hardjito & Rangan (2005).Concrete is a material with high compressive strength but low tensile strength.Despite its low tensile strength as an infill material in the PSSDB floor system, concrete serves a function in supporting the applied load (Akhand 2004).In the ABAQUS model, the concretes were assumed to have two failure mechanisms: compressive crushing and tensile cracking.The constitutive curve of the concrete for uniaxial compression was assumed to be linear up to the point of the initial yield stress.The plastic region begins at the point characterized by strain hardening and ends at the ultimate strength and falling branch of the curve known as concrete softening (ABAQUS 2008).In the ABAQUS model, a C3D8R element (3D) and a continuum element with an eight-node linear FEM were selected for the infill concrete (Figure 27).

Modeling the Connections
There are two types of connections considered in this modeling; firstly the DB to PSS connection via the self-drilling and self-tapping screws, and secondly, the connection provided by the natural adhesion in between the infill concrete and the PSS and DB.Both these types of con-Latin American Journal of Solids and Structures 13 (2016) 272-295 nections contributed to the composite action of the PSSDB floor system through the transfer of shear between the connected layers of components.The stiffer the connection would result in higher interaction in between the components, thus resulting in more enhanced composite actions.The system may reach a full interaction or partial behavior depending on the connectors' stiffness, strength and numbers.The stiffer the connection is, the less slip and deflection the floor will experience (Ahmed et al. 2000).
The CARTESIAN connection type in ABAQUS showed in Figure 29 provides independent behavior between two nodes in three local CARTESIAN directions had been adopted by Seraji et al. (2012) to model the connectors.It has twelve nodal degrees of freedom to facilitate three displacements and three rotations in element local direction of both end of the element.The isotropic connector element represents the biaxial shear deformability of the connectors.The required input data for the connector were obtained from past studies (Shodiq 2004 andAkhand 2001).The values of spring stiffness respectively the screws are 710 N/mm in X and Z direction, whilst 2.9 10 6 N/mm is adopted in Y direction Akhand 2001 .The stiffness connection provided by the natural adhesion in between the infill concrete and the PSS and DB are 3.82 x 10 8 N/mm in X and Z direction, whilst 1.5 10 7 N/mm in Y direction Shodiq 2004 .

Boundary Conditions
The experimental models were all supported on pin-roller supports.These boundary conditions were modelled in the FEM.This setup indicated that the slab with simply supported conditions exhibited outward or inward movements at the support (Vecchio and Tang, 1990).

Loading
The four equal line loads of the loading beams adopted in the experimental test were utilized to simulate a uniformly distributed loading.The loading beams were arranged such that the point load could be applied on the system.The loading beams then transferred the four equal line loads to the model.This strategy can be similarly adopted in the nonlinear FEM; thus, the displacement control method can be easily applied by providing displacement when the point load is applied through a hydraulic jack.

Convergence Study
Convergence study is a process of finite element analysis.Accurate results must be produced without wasting computer memory and analysis time.Increasing the number of elements causes the obtained results to approach the analytical results in mesh convergence.No significant change occurs in the resulting deflection and stresses at a certain refinement level.Thus, they are considered a converged solution to a particular loading, geometry, and constraint.Mesh convergence has been applied in both linear and non-linear finite element analyses.The accuracy of its results is affected by the aspect ratio of each element.In general, the higher the division of the continuum, the smaller the size of each element; thus, accurate results are indirectly produced (Akhand et. al. 2004).Various mesh sizes were applied to obtain the best mesh with already converged result.The sizes of elements employed ranges from 15 mm to 35 mm. Figure 30 shows the converged result for deflection was reached when the mesh size adopted was 30 mm for all elements in the FEM.When the aspect ratio of each element increases, the accuracy of the results decreases.Therefore, the selection of the aspect ratio of the elements is important.A trial-and-error method was used in this study to determine the number of meshes, until the results for the mid-span deflection indicated convergence.

FEM Results Validation and Discussion
Figure 31 shows the results of the laboratory experiments for the FBGPC model and the results of the FEM under bending loads.Both the experimental and FEM results showed the linear turning into non-linear load-deflection relationship.The FEM modeling was verified by the experimental test results, as can be seen in Figure 17 with an accuracy within 94.8% for EI and 97.46% for the ultimate load.Based on the ASCE Standard (1984) and American Iron and Steel Institute (AISI) (1996), all the discrepancies were still within the limits of permissible variation, which was less than 15%.The results were verified through the FEM, and the model was found to be valid and applicable for parametric studies in future.

CONCLUSION
This study has highlighted a new finding that the use of geopolymer concrete and half size DB are able to enhance the performance of the PSSDB structural behavior compared to the previously adopted normal concrete and full-board PSSDB floor system.This finding has now introduced a modified PSSDB floor system utilizing geopolymer concrete infill and half size DB (instead of the normally adopted full size) that has been proven possessing enhanced structural performance that would lead to a more cost effective system.Longer span PSSDB can now be introduced in the construction industry.
The HBGPC panel exhibits a higher by 43% in rigidity and 12% in ultimate load compared to the normal FBNC.This increase in performance is due to the better natural adhesion or bonding in between the geopolymer concrete with the DB and PSS compared to the normal concrete.This was what the authors anticipated prior to embarking on this research.The halfboard concept would be still provide the advantages provided by the PSSDB system compared to traditional flooring systems without losing its structural performance; in fact increasing it.The PSSDB floor when compared to conventional composite slab system offers advantages such as lighter, slimmer, easier to construct (no temporary propping required) floor system, thus leading to more cost effective floor system.
Finally, the FEM model developed in this paper has been verified by the experimental results and thus could be confidently used to conduct more parametric studies on the PSSDB floor system.The results of the FEM were in very good agreement with the experimental results with an accuracy level of 94.8% for rigidity and 97.46% for ultimate load.

Figure 6 :
Figure 6: Schematic view of experimental test setup.

Figure 7 .
Figure 7. Photo of a test panel under testing.

Figure 8 Figure 8 :
Figure 8 show position of displacement transducers T1, T3 and T4 were used to record the vertical deflection along the span of the sample.T2 was placed to record the horizontal displacement of the concrete infill.

Figure 11 :
Figure 11: PSS local buckling positions for Control Panel.

Figure 13 :
Figure 13: PSS local buckling positions for half-board with infill panel.
The cracks were probably due to the side-arching Latin American Journal of Solids and Structures 13 (2016) 272-295 effect of the PSSDB floor under test that has created tensile stress on the top surface of the concrete in the width.

Figure 14 :
Figure 14: Top view of the half-board sample with crack lines on the concrete top surface at failure.

Figure 15 :
Figure 15: Symmetrical deflections at quarters span positions for CS sample.

Figure 18 :
Figure 18: Symmetrical deflection at quarter span positions for FBGPC sample.

Figure 19 :
Figure 19: Symmetrical deflection at quarter span positions for HBGPC sample.

Four
different types of load vs mid-span deflection (T1) graphs are drawn from the experiments, as illustrated in Figures 20 till 23 to highlight the effect of the various parameters on the behavior of the PSSDB samples.

Figure 20 :
Figure 20: Graph of the load vs deflection for the CS and various full-board panels.

Figure 21 :
Figure 21: Graph of the load vs deflection for the CS and various half-board panels.

Figure 22 :
Figure 22: Graph of the load deflection for the CS and various normal concrete panels.

Figure 23 :
Figure 23: Graph of the load deflection for the CS and various geopolymer concrete panels.

Figure 31 :
Figure 31: Load-mid panel deflection curve of the FBGPC model, experiment and FEM.

Figure 32 :
Figure 32: Local buckling of the PSS.

FiguresFigure 33 :
Figures 33 (a) and (b) show the effect of the slip of the infill at one end of the sample for both the experiment and FEM.

Mohd Isa Jaffar a* Wan Hamidon Wan Badaruzzaman b Shahrizan Baharom c
a Ph.D candidate b Professor c Senior Lecturer

Table 3 :
Table 3 gives the experimental rigidity and ultimate load values and load at local buckling load for all the tested samples.Experimental rigidity and ultimate load values.

Table 4 :
Table 4 presents the element selections for the FEM of the FBGPC model.Element selection for FEM.