Behavior and Flexural Strength of Composite Steel-Reactive Powder Concrete Decks

The need for construction materials with superior chemical, physical, and mechanical properties has led to important developments in the concrete industry. One of the most recent advancements is reactive powder concrete (RPC), which is a special high-strength cementitious material with favorable behaviour for civil engineers.

Ordinary concrete and RPC are considerably different in their composite materials and physical properties. By eliminating coarse aggregate and incorporating materials like steel fiber, silica fume, and very fine sand, mixed with a low W/C ratio, RPC can show ultra-high performance in tension and compression compared to ordinary concrete. Tension resistance is developed as a result of steel fiber, which is also responsible for preventing or reducing cracks [1-3]. Sections made of RPC are nearly impermeable due to the high density achieved by adopting fine materials. Due to its low permeability and resistance to aggressive environmental agents, the corrosion resistance of reinforcement in RPC is better than that in ordinary concrete [4]. The most important feature of the RPC is its remarkably high compressive strength, which can be greatly developed through optimized mix design and suitable curing procedure [5].

However, in some cases, using a single construction material is insufficient to meet structural requirements. Therefore, two or more materials are mixed or combined to take full advantage of their properties. The resulting structural section is known as a composite member.

The application of composite sections covers a range of areas, including buildings with long spans between columns [6], where load-bearing capacity can be increased, and cross-sectional dimensions can be reduced. Moreover, the response of bridge structures, including durability and deflection, can be enhanced by providing composite girders [7]. Also, composite members are a favorable choice in industrial, underground, and seismic-resistance structures [8].

Although composite beams or decks with normal concrete are extensively researched [9-12], the performance of composite structures employing RPC has received relatively limited attention [13, 14]. The special components and the produced microstructure densification of RPC may affect interface response, crack distribution, and shear transfer mechanisms.

One essential aspect in composite action is the behaviour of shear connectors, which are provided to prevent or reduce slip and enable one-unit behavior of the composite structure. Various types of shear connectors can be used in composite structures, such as bolts, studs, and steel sections.

Dakhil and Dawood [15] investigated the response of composite RPC beams, focusing on the effect of the number of studs. It was found that this parameter had a negligible impact on the deflection values. On the other hand, it had a great influence on the ultimate load of the composite beam.

Ali et al. [16] carried out a numerical investigation to study the influence of the number of studs and their distribution on the response of composite beams. The study showed that the asymmetry in the distribution of shear connectors involved in staggered distribution could lead to fluctuations in deflection values from one pattern to another.

The current study addresses the flexural performance of RPC composite decks. Experimental investigations were utilized for this purpose. A total of 7 tests were conducted on simply supported composite decks aimed to study the influence of several parameters, such as the thickness of the RPC deck, the number of studs used, and the type of shear connectors (studs, bolts and angle steel-sections.

3.1 Effect of the reactive powder concrete deck thickness

The thickness of RPC in composite structures may have a big impact on the overall response of the structure, including ultimate load, cracking load, maximum deflection, crack pattern, and slip [19]. In order to investigate the influence of RPC deck thickness on the behavior of the composite decks, three specimens (CRB1, CRB2, and CRB3) were considered. The specimens had deck thicknesses of 60, 80, and 100 mm, respectively, with all other parameters kept unchanged.

Figure 5 illustrates the load-deflection plots for the three specimens. Each curve showed a two-stage response starting with linear elastic behavior for both RPC and steel sections, followed by a decrease in the slope of the linear portion up to a nonlinear curve. From the figure, it can be concluded that the ultimate load increases as the deck thickness increases. The thicker specimen (CRB3) recorded an ultimate load of 288 kN, which is higher than the two other specimens by about 9% for CRB2 (264 kN) and 28% for CRB1 (224 kN). This result reflects the well-known engineering concept that a thicker deck or slab can enhance section stiffness (EI), increase the compression zone, and provide a larger moment arm.

Figure 5. Load-deflection curves for the tested specimens

The corresponding mid-span deflections of each specimen at the value of ultimate load are listed in Table 4. The table also shows the values of the overall stiffness of the deck (P_u / δ_u) up to the ultimate load. It can be seen that the stiffness values rise as the ultimate load values rise (i.e., thickness increases). This explains the reduction in ultimate deflection values as deck thickness increases, although higher loads are recorded, as a result of the fact that higher stiffness members show less deflection [20].

Table 4. Test results of specimens CRB1, CRB2 and CRB3

To describe the relationship between deck thickness and the recorded ultimate load, two models have been fitted (linear and quadratic), as shown in Figure 6. It can be seen that linear fitting gives a reasonable fit with (R² = 0.98) for the range of thicknesses adopted. However, the quadratic fitting (nonlinear) represented by the equation shown in the figure passes exactly through the points, reflecting a strong nonlinear relationship between thickness and ultimate load. The nonlinear relationship was also noticed by Ibrahim and Ahmed [21].

For a rectangular slab (not a composite slab), the cracking load is proportional to t² (P_cr ∝ t²), which is the case in the current study, as can be seen by the quadratic equation of the trend line illustrated in Figure 6.

Figure 6. Ultimate load – deck thickness relationship

However, for composite slabs or beams, the steel I-section contributes to a portion of the tensile force, leading to a reduction in the contribution of the RPC deck. This could be the possible reason for the approximated linear or near-linear response, which can be used with reasonable results.

The visual observations of the first crack loads have recorded the same trend of ultimate load, with almost a direct proportionality between deck thickness and initial cracking load P_cr (P_cr = 65, 94, and 129 kN for CRB1, CRB2, and CRB3, respectively). Figure 7 shows the trend line of the recorded initial cracking load as a function of deck thickness, which can be approximated as a linear line with R² = 0.9843. However, this trend is for the small range of thicknesses considered (t = 60 to 100 mm).

Figure 7. Cracking load – deck thickness relationship

Figure 8 shows the crack pattern of the specimens at the ultimate load. The crack was first observed at the bottom of the deck, then developed up to the upper boundary of the deck thickness. It can be seen that the cracks were concentrated in the mid-point of the slab, where the load was applied.

Crack patterns showed that only a few cracks with a limited network were developed in the RPC decks. This can be attributed to the nature of RPC materials, where its high density is achieved by fine contents and the presence of steel fibers enhances its tensile strength even at ultimate load. Furthermore, the big contribution of the steel section in carrying the tension is also an important factor in limiting the number and distribution of cracks. The limited or no crack appears in RPC structures was also noted by Zhong et al. [22] and Cao et al. [23].

Figure 8. Crack pattern of the specimens

Figure 9 shows the relative end slip at the ultimate load for the three specimens as recorded by the dial gauges attached to the deck and steel section, as an indicator of the composite action provided by the studs. Although all specimens had the same number of studs (10), they showed varying relative slip. The comparison reveals that the thicker deck experienced relatively less value of end slip. The reduction of end slip with increasing deck thickness may be a result of increasing overall stiffness of the RPC deck, as the concrete effective zone surrounding each stud increases, causing better restraint to the slip deformation.

Figure 9. End slip at ultimate load

3.2 Effect of the number of studs

Two specimens named CRB1 and CRB5 with 10 and 5 studs, respectively, were tested and compared, with all other properties kept the same. Figure 10 illustrates the comparison of the load-deflection curve up to ultimate load for both specimens. Table 5 lists a summary of the test results.

Figure 10. Load-deflection relationship of CRB1 and CRB5

Table 5. Test results of specimens CRB1 and CRB5

It is obvious from the figure. that the ultimate load of specimen CRB1 is greater than that of specimen CRB5. The ultimate load recorded for CRB1 was 224 kN, which is about 15% greater than that recorded for CRB5 (194 kN). This reflects the contribution of increasing the number of studs in enhancing the behavior of composite structures. This response can be attributed to the fact that more shear connectors lead to an increase in the composite action between concrete and steel, in which the composite structure behaves as one unit. However, both specimens experienced the same performance with an initial linear followed by a nonlinear response.

The deflection at the ultimate load recorded by the mid-span dial gauge was 10.4 mm for CRB1 and 11 mm for CRB5.

The crack pattern of specimen CRB5 (Figure 11) was slightly different from that observed in CRB1. The top surface of the deck showed longitudinal cracks along its length, i.e., along the axis along which the studs were distributed. These cracks have appeared before reaching the ultimate load (at load 130 kN). The possible reason for such a type of shear crack is the insufficient number of shear connectors used (5 studs), which may cause a reduction in bond along the axis of studs. This phenomenon was also observed by Dakhil and Dawood [15]. It is important to mention that the first crack, in CRB5, was appeared at a load of 50 kN, which is less than that of CRB1 (65 kN).

Figure 11. Crack pattern of specimen CRB5

The relative slip between the RPC deck and the steel section at the end of specimen CRB6 was 2.4 mm. This value is only a little bit greater than that observed in specimen CRB1 with 10 studs (2.1 mm). Although the longitudinal crack that appeared in CRB5 may indicate a slip problem and a reduction in composition, this slip may occur locally and does not propagate towards the ends of the deck.

As for specimen CRB1, the transverse cracks that appeared in specimen CRB5 were few in number and narrow in distribution. This indicates that the reduction in the number of studs does not negatively affect the transverse cracks. This result can be argued by (1) with all other material and section properties are the same, changing the number of studs will not chang the tensile strain caused by bending stress at the concrete mid-span, and (2) it appears that the composite action is reduced but not completely lost by using 5 studs, especially at early stage of loading when first and few other cracks were appeared. In this stage, both specimens behave the same way, as the induced shear stress can be carried effectively by 10 or 5 studs. After that, the shear stress applied to the 5 studs exceeds their shear capacity, causing local slip and longitudinal cracks, with no big effects on the transverse cracks.

3.3 The response when thread bolts are used as shear connectors

Welded bolts are not commonly used in real structures, although some researchers, e.g., researcher [11], have used this type of connection in their tests. However, this approach was adopted to ensure full connection in a way that eliminates any possible slip or looseness that may happen during loading at the bolt-steel section interface.

In order to study the influence of using bolts as shear connectors, a composite specimen (CRB4) with 10 bolts welded to the top flange of steel was tested and compared to another specimen of 10 studs (CRB2), with all other parameters kept unchanged.

Figure 12 portrays the load-deflection response for both specimens. It can be noticed that both specimens had almost the same behavior in the early stages of the linear part. However, the difference in load-deflection response gradually increased up to the value of the ultimate load. At this point, the values of ultimate load recorded were 264 kN and 249 kN for CRB2 and CRB4, respectively. Also, the final deflection for CRB2 was 9.0 mm, while that for specimen CRB4 was 9.5 mm.

Figure 12. Load-deflection relationship of CRB2 and CRB4

In spite of the similarity in load-deflection curves, specimen CRB4 showed additional longitudinal cracks that were not visible in specimen CRB2, as can be seen in Figure 13. The possible reason for this behaviour can be attributed to the shape of the bolts. Although the threaded shanks of the bolts can provide high bond and interaction with the RPC compared to the smooth stud, they may have a negative influence. In addition to the slight reduction in the bolt diameter at its roots (Figure 14), the thread crest can show high stress concentration, which may cause damage to the concrete surrounding, leading to cracks along the longitudinal axis of bolts to appear.

Figure 13. Longitudinal cracks of specimen CRB4

Figure 14. Details of the threaded bolt

The internal cracks that may develop in the RPC around the bolt thread can lead to loss of interaction and, then, to slip. This may explain why the magnitude of relative slip in CRB4 was 5.7 mm, which is higher than that noted in CRB2 (1.5 mm).

It is important to mention that the first crack appeared at the top surface of the deck and not at the tension zone, as it was expected. The local stresses generated around the bolts may be developed to longitudinal shear along the bolt’s axis, causing cracks to the upper compression surface before flexural cracks appear in the tension zone.

3.4 The response when angles are used as shear connectors

In order to investigate the efficiency of steel sections as shear connectors, a test with angle steel sections welded transversely to the top flange of the steel beam was carried out (specimen CRB6). Different shapes and orientations of steel sections can be used to transfer shear between the RPC deck and the steel I-section [24]. In the current study, 5 angle steel sections distributed equally over the entire 1000 mm of the beam length were used.

The load-deflection curve for specimen CRB6 is illustrated in Figure 15. For the purpose of comparison, the figure also portrays the response of specimen CRB5, which had 5 studs and a 60 mm deck thickness. It can be seen that there was no significant difference between the responses of the two specimens, especially at early stages (linear portion). The ultimate load of specimen CRB6 (204 kN) was slightly greater than that for the specimen with 5 studs (194 kN).

In spite of the similarity in load-deflection curve, a significant benefit of using angle shear connectors can be identified in preventing the longitudinal cracks that were propagated on the top surface of specimen CRB5, as can be seen in Figure 16. This can be attributed to the geometry of the angle shear connectors, which can distribute the shear over a greater contact area compared to the contact area provided by studs, making the shear stress less concentrated around the shear connectors. On the other hand, tension cracks were, as in specimen CRB5, few in number, narrow in distribution, and concentrated at the mid-span.

Figure 15. Load-deflection relationship of CRB5 and CRB6

Figure 16. Crack pattern of specimen CRB6

For the first crack records, the specimen CRB6 showed the first crack at 80 kN, which is greater than that observed in CRB5 (50 kN). The likely reason for this specimen to show the first crack at a higher load is the large contact area that the steel angles are acting on. This may lead to spreading the shear over a larger area, making the local tensile stresses decrease in the RPC deck.

Concerning the slip, specimen CRB6 recorded only 1.0 mm for the value of relative slip, which is the least value among all other specimens. This result suggests that the force causing slip can be resisted effectively by employing transversal channels as shear connectors.

3.5 Modes of failure

There are several scenarios in which a composite RPC deck can fail, depending on factors such as shear connection, material characteristics, and applied loads. Failure may be related to the RPC deck (concrete) when crushing or excessive flexural/longitudinal/shear cracking appears. On the other hand, studs may fail as a result of exceeding their shear capacity, pulling out the studs, or slip failure. Another way of failure is related to the steel I-section, which can be due to buckling of the steel section or yielding, which may happen in the flange or web. It is important to mention that experimental tests [25], for example, showed a combination of the above failure modes with varying sequence.

In the current study, no visible separation or slip was noticed between the RPC deck and the steel section at the ends of the specimen. This means that the studs used have provided an adequate interaction and composition between the steel and concrete, with only limited relative slip recorded. Moreover, no crushing was observed at the top compression surface of the RPC deck.

In general, all specimens tested with a central load applied on the RPC deck experienced a flexural mode of failure. This type of failure was characterized by an initial linear elastic response for both concrete and steel sections. This stage includes the appearance of the first crack at a certain point of loading. Then the composite deck behaves nonlinearly. The ultimate load was reached prior to the crushing of concrete, slipping of studs, and buckling of the steel section.

3.6 Sources of experimental limitations

Before conducting the tests, it was decided not to strengthen the steel sections for all specimens. However, while carrying out the first trial test, an unexpected scenario occurred. After completing the elastic stage, the steel section showed a gradual buckling at one end. The test has been stopped when buckling is rapidly increased (Figure 17). This case of failure was also noticed by Dakhil and Dawood [15].

Figure 17. Crack pattern of specimen CRB6

In order to prevent this defect from recurring, a decision was made to strengthen the ends of the steel sections at the supports by placing two steel angles back-to-back on both sides of the steel section, as can be seen in Figure 16.

Moreover, it is known that the mixing of RPC needs a high shear mixer to wet particles. Otherwise, the mix stays dry and clumpy with no workability. However, the rotating mixers available in the laboratory were of low speed and, thus, can not develop the fluidity required to prepare a concrete mix with workability that would allow the concrete to be poured into molds. To overcome this issue, small mixers with medium to high speeds were used as alternatives. Three mixers working at the same time were used to prepare the mix.