1. Introduction
As a main part of building structures, the floor system withstands live loads and its self-weight and transfers them to vertical load-bearing components [
1]. According to the structural forms, they could be categorized into two main types: beam-slab floor and beamless floor [
2].
Theoretically, the slab and the beam are calculated separately by simplifying them into beam and one-way or two-way plate elements, respectively [
3]. Thus, with the increase of span, the heights or depths of the beam and slab must be enlarged to bear the increasing self-weight of the floor system [
4]. This would result in two economic concerns in real estate, which are material consumption and labor costs [
5].
To balance the economic costs and mechanical needs, four main methods have been proposed in recent decades. Hollow slab was the first method conceived by engineers and researchers [
6,
7,
8]. It was recognized that the parts with small stress of the slab should be eliminated to reduce its self-weight and construction costs without impairing its mechanical behavior. Furthermore, the cavity could be infilled with some insulation materials to improve its thermal and sound insulation properties. Experimental and theoretical research of this type, such as plastic bubble ball, foamed concrete infills, have been carried out and satisfactory results have been reached [
9,
10,
11,
12].
Two-way multi-ribbed floor systems turned out to be another effective solution [
13]. By substituting beams with ribbed beams of lower height and smaller spacing, the story height could be reduced and the average costs of the building could be lowered.
To further reduce the total costs and protect the environment at the same time, assembled integral and assembled floor systems were proposed. With all or main components manufactured in factory, the construction period could be prominently shortened and the demands of casting template and labor could be greatly reduced. Ma proposed a Vierendeel-sandwich-plate floor system [
14], which is composed of two layers of ribbed reinforced concrete slab connected by reinforced concrete shear keys at the intersection. Through the design and construction of a real-life project, the feasibility of this new floor system was verified. Other kinds of assembled or assembled integral in floor systems, such as PK prestressed composite slab floor system [
15], the assembled monolithic hollow-ribbed floor [
16], prefabricated PC floor system [
17], and other new systems [
18,
19,
20,
21,
22] were all investigated through experimental and theoretical methods. Though the assembled techniques could greatly shorten the construction time and enhance the construction efficiency, there is still an urging worry about its stiffness and anti-crack properties under designed load.
In this paper, a new type of assembled integral two-way multi-ribbed composite floor system is proposed inheriting the advantages of both two-way multi-ribbed floor and composite floor systems. The innovative floor system is composed of four main parts, which are precast ribbed bottom slab (PRBS), lightweight infills, cast-in-situ upper slab (CUS) and joints. The precast ribbed bottom slab is illustrated in
Figure 1, where the plane size is 2050 mm × 2050 mm and the thickness is 50 mm. In the middle of the bottom slab there is a 200 mm × 150 mm convex ribbed beam designed in both directions. The ribbed beam is reinforced by two longitudinal rebars and stirrups to undertake the shear force and bending moment, of which the areas of the longitudinal rebars are calculated by analogue beam or analogue slab methods (see JGJ/T 268-2012 [
23] “Technical specification for cast-in-situ concrete hollow floor structure”). The bottom slab is reinforced by 8 mm rebar mesh with the spacing of 100 mm in both directions. The extension lengths of the rebar mesh are all 175 mm for binding with adjacent panels. To prevent cracking in demolding, the corners are strengthened with diagonal rebars of 8 mm.
The main function of the lightweight infills is to ensure the hollow cavity and reduce its self-weight.
Figure 2 illustrates the detailed sizes of the lightweight infill, which are identical to those of the cavity to ensure there is no concrete penetration in the casting.
With the PRBSs and the lightweight infills arranged, they could serve as lateral and bottom templates for the CUS and joints (see
Figure 3). In this way, the hollow ratio and assemble ratio could reach 43% and 87%, respectively, which could bring promising economic and social benefits.
The three different parts above and adjacent bottom slab are integrated into a reliable system through three methods. (1) The connection between the PRBS and the CUS is realized by the shear strength of the rough interface, cast-in-situ joints and stirrups. (2) The lightweight infills are fixed on the precast bottom slab through the positioning rebars and recessed cavities. (3) The rebars of adjacent PRBSs are connected by couplers for squeezing and splicing of rebars and cast-in-situ joints where there are shear keys equally arranged alongside the outsides of the PRBS.
In order to study the mechanical performance of the assembled integral two-way multi-ribbed composite floor system, a static load test of a 9.2 m × 9.2 m full-scale multi-ribbed composite floor was carried out. The mechanical performance of this floor system in normal service state and maximum design load according to Chinese design code GB50010-2010 were studied. Mechanical performance (i.e., crack distribution, deformation and stress redistribution) were analyzed and compared with finite element method.
3. Experimental Results
3.1. Experimental Observation
The deflection and crack development were monitored and logged during the static loads, which were concluded as follows:
In the initial loading stage, the deflection in the middle of the slab increased with the increase of the load linearly, and no cracks were found at other parts of the slab bottom except for some initial fractures. When the load reached 9.83 kN/m2 (the 6th load, accumulated load including its self-weight), a small number of tiny cracks parallel to the direction of the joint appeared on the interface of the new and old concrete. The cracks mainly concentrated in the middle position of the cast-in-situ joints. It is an interesting phenomenon, which will be discussed in the following.
When the floor was exerted with 3.4 kN/m
2 (the 4th load), the accumulated area load reached 7.89 kN/m
2, which is beyond the service load of GB50010-2010 [
24]. There were no cracks and the deflection was only about 2 mm. According to the load protocol, when the additional load reached its self-weight, the experiment was maintained for 24 h. The maximum deflection of the floor increased about 0.5 mm during the maintaining time.
With the load increasing from 9.83 kN/m2 to 12.38 kN/m2, the cracks further developed and the number of cracks increased. It was noticed that the cracks did not extend to the joint or the shear keys. Besides that, the distribution of cracks from the 6th load to the 8th load was not symmetrical and not in the sections with maximum movement. Thus, it was predicted that the cracks of the 6th–8th load were mainly caused by initial defects.
When the floor was exerted with 12.38 kN/m2 (the 9th load), the accumulated area load exceeded the maximum load of GB50010-2010. The maximum strain of steel rebars was only 450 με and the maximum deflection was 5.32 mm, indicating that the floor did not reach its load-carrying capacity. Thus, additional area load should be exerted to explore its mechanical performance.
In the 10th load, the accumulated area load reached 13.23 kN/m2 and diagonal cracks appeared in the middle section of the floor. It is worth noting that the diagonal cracks only developed within one single PRBS. With the area load increased, the length and number of diagonal cracks increased rapidly, which was a distinct difference from the cracks that appeared in 6th to 8th load. Some cracks along the shear key of the PRBS appeared during the 11th load, which was a characteristic and typical phenomenon of the innovative floor system. When the 12th load was exerted, a diagonal crack appeared at the intersection of the floor and column, extending from the column to the mid span direction. As the load increased, the width of cracks along the shear key of the PRBS increased.
For the 14th load, the cracks concentrated on the ribbed beam of PRBS and joints and the overall shape of cracks turned out to be a “X” shape (see
Figure 12 and
Figure 13), where the numbers in the figures represented the load grade numbers. And the accumulated area load reached 16.63, which was about 1.5 times the maximum load of GB50010-2010 [
24].
3.2. Load–Deflection Relationship
The load–deflection curve of the innovative floor is presented in
Figure 14, where the maximum deflection is obtained from D5. According to GB50010-2010, the maximum deflection of normal service load (7.89 kN/m
2) should be calculated with the long-term effects, and its deflection limit is L
0/300 (L
0 is the calculated span of the floor). The maximum deflection was calculated through the displacement sensors D1~D9, as presented in Equation (1).
According to the test results, the floor system is still in an elastic state until the 5th level load. And the deflection is measured from 4.49 kN/m
2, which is its self-weight. Based on this, it can be concluded that the true deflection under normal service load is 2.40 × (7.99/4.49) = 4.27 mm. According to GB50010-2010 [
24], the long-term load correction coefficient is 1.856. The calculated long-term load deflection under normal use load is 7.92 mm, which is less than L
0/300 (30.67 mm).
3.3. Stress Distribution
Due to the large number of strain gauges installed in the integral multi-ribbed composite floor, only the representative strain gauges were selected for analysis in this section, and the selected strain gauge locations were strain of the bottom reinforcements according to
Figure 11.
It was found that the strain of rebars along one cross section shared identical load–strain configurations. And the maximum value of the strain gauge was still less than 400 με (see
Figure 15), indicating the rebars were still in an elastic working state. Even when the test load increased to 1.5 times the designed load of GB50010-2010, the maximum strain was only 800 με.
4. Finite Element Simulation
Even when the total load of the experiment reached 110 tons, beyond the limits of the laboratory, the floor did not reach the load-carrying capacity. To further study its mechanical behaviors beyond 17 kN/m2, the finite element method turns out to be a supplementary method.
The application of finite element method (FEM) could be explained as follows. First, the calculated results, including load–deflection relationship, stress distribution and failure mode, were compared with the experiment to verify the effectiveness of parameter settings. Then, the finite element model was exerted with a larger area load to explore its nonlinear mechanical behaviors.
The finite element simulation contains six main steps, which are FEM model establishment and meshing, constitutive relationship settings of materials, element selection, contact of different components, load and boundary conditions, analysis steps and calculation.
4.1. Establishment and Meshing of Finite Element Model
A commercial software, ABAQUS 2017 (Dassault Systemes, Paris, France), was utilized to obtain numerical simulation of the assembled integral two-way multi-ribbed composite floor under static load. Due to the sophisticated design of the floor, the FE model was identical to
Figure 7 to avoid the influence of oversimplification. The FE model contained four main parts, i.e., the 16 PRBSs, the CUS, cast-in-situ beams and columns, steel rebars, as presented in
Figure 16.
For the extreme geometrical irregularity of precast ribbed bottom slab (PRBS), it was necessary to divide it into two parts when meshing. The green parts indicate it was geometrically regular and could be meshed into a hexahedron, while the pink parts suggest the geometrical irregularity and the need to be meshed into a tetrahedron. From the detailed geometry of
Figure 17, it can be easily seen that the pink parts were mainly shear keys around the PRBS and the ends of ribbed beams.
The same meshing methods were applied to CUS and joint. The meshing of the FE model is presented in
Figure 18.
4.2. Constitutive Relationship Settings and Element Selection
Four different materials were used in the experiment, which were concrete, steel rebar, lightweight infills and coupler for squeezing and splicing of rebars. As previously mentioned, the lightweight infill was designed to ensure the cavity and improve the sound or heat-insulation capabilities of the floor. Thus, its constitutive relationship was not considered when calculating the numerical results. While the behavior of concrete, rebars and coupler might work in the range of elastic-plastic, so the whole life stage, i.e., elastic stage, softening stage and strengthening stage, should be included in the strain–stress relationship.
Figure 19 and
Figure 20 illustrate the stress and strain relations of concrete and steel bars, respectively. The nominal strain and stress were calculated by Equations (2) and (3), where
and
were the real strain and stress of rebars,
and
were the nominal strain and stress of rebars obtained from the experiment in
Table 1 and
Table 3.
The element type of C3D8R (three-dimensional eight-node linear brick elements with reduced integration) in ABAQUS was chosen to simulate the geometrically regular part. The geometrically regular part was simulated by C3D10 (three-dimensional ten-node quadratic tetrahedron) element. The rebars were all simulated by T3D2 (three-dimensional two-node truss) element.
4.3. Contact and Boundary Conditions
The contact surface mainly included two different categories: (1) the direct contact between the upper surface of the PRBS and the CUS and (2) the contact of the shear key round the PRBS and cast-in-situ joint, which was connected through rebars and squeezing. Considering the connection types, the former contact relationship was set as a tie. For the latter one, the contact relationship of the normal direction was set as “hard contact”, and the tangential direction was set as “penalty friction”. Boundary conditions were identical with the experiment.
4.4. Load and Analysis Steps
According to the load protocol, the specimen was divided into 49 sections, where mass block was evenly stacked to simulate the area load. The same division method and area load were applied on the upper surface of the FE model.
The calculation contains two load procedures, which were self-weight and additional area loads. The maximum area load was set as 60 kN/m2 to fully investigate its nonlinear behaviors. The solver of “static general” in ABAQUS was selected to calculate its numerical results.
4.5. Simulation Results Analysis and Load-Carrying Capacity Calculation
4.5.1. Simulation Results
Figure 21 and
Figure 22 show the displacement and stress distribution of the assembled integral two-way multi-ribbed composite floor system under an area load of 16 kN/m
2 (including the self-weight of the floor). The mid span deflection of the floor was 15.91 mm, and the absolute error of the experimental and finite element results was 7.86%.
To further compare the crack distribution, a parameter “DAMAGET” was selected to represent the degree of tensile damage of the assembled integral two-way multi-ribbed composite floor system in ABAQUS. The calculation method of DAMAGET at a specific strain ε
1 is explained as Equation (4) [
20], where
E0 represents the original undamaged Young’s modulus of concrete,
f(ε) is the stress function of a specific node and
stands for the strain energy at the given stress. The range of DAMAGET is within [0,1], indicating the extent of damage of materials. When the concrete works in elastic stage, the strain energy equals
and DAMAGET could be calculated as zero. For elements undergoing nonlinear behaviors, the stiffness degrades and the accumulated strain energy is less
, resulting a larger DAMAGET.
It could be found from the maximum damage factor of the bottom surface mainly focused on the central four PRBSs, which was similar to results of
Figure 12. Comparison of DAMAGET at the top and bottom surface of the assembled integral two-way multi-ribbed composite floor system and the experimental results are summarized in
Figure 23. The distribution of concrete damage was in an “X” shape, which was basically consistent with the crack distribution in the experiment. The simulation results showed that the damage in the middle of the floor slab is the largest, which is also consistent with the phenomenon observed in the experiment.
Figure 24 shows the comparison of the load–deflection curves between finite element analysis and testing. The trends of the two curves are basically consistent, and both include two working stages: elastic working stage, where the entire cross-section worked together without cracks. When cracks appear, the stiffness decreased, and as the cracks gradually increased, the stiffness continued to decrease. The maximum mid span deflection error of the two curves was less than 10%, within the acceptable range.
4.5.2. Load-Carrying Capacity Calculation
From the comparison above, it could be found that the finite element simulation could efficiently simulate the mechanical performance of the assembled integral two-way multi-ribbed composite floor system.
To furtherly study its load-carrying capacity, the area load when the rebars yield (i.e., reach 400 MPa) was defined as its load-carrying capacity. Based on the same finite element model above, a larger area load was applied and the ultimate load obtained was 50.98 kN/m
2. The calculation results of this model are shown in
Figure 25 and
Figure 26.
4.5.3. Error Discussion
As illustrated in
Figure 26, it is evident that the finite element simulation is consistent with the experiment. In the overall configuration, the trend of the finite element simulation and experimental results in the initial stage were basically consistent. With the area load increases, its deflection exhibited more obvious nonlinearity.
But there were still some errors in both load–deflection relationship and failure mode. They mainly came from three aspects. (1) The finite element method ignored the squeezing between the elements and resulted in deviation from the actual deformation. (2) The reinforcement was simulated by truss elements embedded in the concrete, where the bond-slip relationship between the two materials was not considered in the simulation. (3) There were construction defects that could not be precisely simulated. The absolute error was within the acceptable range.
In summary, the finite element modeling method based on this section could serve as an effective tool for studying the ultimate bearing capacity of a new type of assembled integrated bidirectional multi-ribbed composite floor, further compensating for the limitations of experimental conditions.