*Article* **Flexural Strength of Internally Stiffened Tubular Steel Beam Filled with Recycled Concrete Materials**

**Ahmed W. Al Zand 1,\* , Mustafa M. Ali 1,2, Riyadh Al-Ameri <sup>3</sup> , Wan Hamidon Wan Badaruzzaman <sup>1</sup> , Wadhah M. Tawfeeq <sup>4</sup> , Emad Hosseinpour <sup>1</sup> and Zaher Mundher Yaseen <sup>5</sup>**


**Abstract:** The flexural strength of Slender steel tube sections is known to achieve significant improvements upon being filled with concrete material; however, this section is more likely to fail due to buckling under compression stresses. This study investigates the flexural behavior of a Slender steel tube beam that was produced by connecting two pieces of C-sections and was filled with recycled-aggregate concrete materials (CFST beam). The C-section's lips behaved as internal stiffeners for the CFST beam's cross-section. A static flexural test was conducted on five large scale specimens, including one specimen that was tested without concrete material (hollow specimen). The ABAQUS software was also employed for the simulation and non-linear analysis of an additional 20 CFST models in order to further investigate the effects of varied parameters that were not tested experimentally. The numerical model was able to adequately verify the flexural behavior and failure mode of the corresponding tested specimen, with an overestimation of the flexural strength capacity of about 3.1%. Generally, the study confirmed the validity of using the tubular C-sections in the CFST beam concept, and their lips (internal stiffeners) led to significant improvements in the flexural strength, stiffness, and energy absorption index. Moreover, a new analytical method was developed to specifically predict the bending (flexural) strength capacity of the internally stiffened CFST beams with steel stiffeners, which was well-aligned with the results derived from the current investigation and with those obtained by others.

**Keywords:** stiffened CFST beam; recycled concrete; finite element; flexural strength; cold-formed tube

#### **1. Introduction**

The usefulness of various types of concrete-filled steel tube (CFST) has been proven in modern composite structural projects since they have achieved better performances (in terms of strength, ductility and stiffness) than the corresponding conventional concrete/steel members under axial and flexural loading [1–4]. In general, the structural performance of concrete-filled steel composite members, under different loading scenarios, has been experimentally and numerically examined in several studies; for example, those presented in [5–9].

Particularly, the adequacy of using the Slender steel tube section (Class 4; as per Eurocode classification) in the concept of CFST beams has been previously investigated [10–14]. This section (Slender) has achieved more strength improvement percentages than those of the Noncompact and Compact sections, after being filled with concrete materials, as

**Citation:** Al Zand, A.W.; Ali, M.M.; Al-Ameri, R.; Badaruzzaman, W.H.W.; Tawfeeq, W.M.; Hosseinpour, E.; Yaseen, Z.M. Flexural Strength of Internally Stiffened Tubular Steel Beam Filled with Recycled Concrete Materials. *Materials* **2021**, *14*, 6334. https://doi.org/10.3390/ma14216334

Academic Editor: Stelios K. Georgantzinos

Received: 26 September 2021 Accepted: 18 October 2021 Published: 23 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

compared to their corresponding hollow tube sections [10,15]. In addition, the steel tubes of CFST members with Slender cross-sections usually have lighter self-weight than the Noncompact and Compact sections; thus, they are more favourable in terms of reducing the impact of cost in construction projects. However, the Slender tube section is more likely to buckle outward at the compression zone stress because of Slender structural sections [16,17]. Therefore, internal steel stiffeners were provided to delay and restrict the outward tube's buckling of the Slender CFST beams' cross-section [18–21]. For example, Al Zand et al. [19] experimentally and numerically examined the effect of using varied shapes of internal steel stiffeners that were welded along all sides of high-slenderness coldformed square CFST beams. Their investigation showed that these CFST beams' flexural stiffnesses and strength capacities were improved by about 22–29% when a single stiffener was welded along each tube's internal sides, and these values were increased by a further 48–59% when double steel stiffeners were provided. Furthermore, in some cases, the CFST beam's cross-sections were externally stiffened, either by using additional steel plates that were fixed mechanically along the beam's flanges [5], or by preparing a cold-formed tube's cross-section with V-grooves that were provided along the sides of the CFST beam [22]. However, the welding of additional steel stiffeners for the stiffening of the CFST beam's cross-section could require an additional fabrication process and extra labour costs. Thus, using pre-fabricated cold-formed tubular steel sections (C-sections/C-Purlins) without extra welding and/or mechanical fixation could be considered as a new concept for use in the stiffened Slender CFST beam system, particularly for lightweight flooring structures. For example, in Malaysia, the cost of 1 ton of fabricated steel sections is about 1000 USD, and the equivalent cost of hot-rolled sections is about 1200 USD, while the cost of 1 ton of concrete is about 40–50 USD. This material's cost comparison shows that it is very useful if the engineers can utilise CFST beams that have been prepared from cold-formed sections filled with concrete in the lightweight composite structures, even though the self-weight of these composite beams can increase by 5–8 times as compared to the equivalent steel beams due to the effect of concrete infill material.

The research topic of CFST has received considerable attention over the past few years from a wide range of structural engineering scholars. A dependent stress–strain model was developed for the CFST column [23]. The flexural behaviour of steel-fibre-reinforced self-stressing recycled-aggregate CFST was studied by [24]. The behaviour of heated CFST stub columns, containing steel fibre and tire rubber, was tested by [25]. Reinforced CFST with steel bars was studied experimentally in terms of its seismic behaviour [26]. Several other related research studies on the advanced application of CFST can be reviewed through [27,28]. Nevertheless, this review of the literature evidenced the importance of the enhancement of concrete properties for the achievement of better reliability and robustness in construction projects [29–31].

In recent years, engineers have been increasingly utilizing the recycled aggregate generated from the demolition of existing materials and waste materials in the field of construction [24,32–34]. There have been several studies on the behaviour of Recycled-Aggregate Concrete (RAC) material, which is prepared by crushing the old structural elements of concrete. For example, variations in the RAC replacement percentages (0%, 25% and 50% of the raw aggregate) were adopted in the concrete infill material for CFST members under flexural and compression loads [11,24,35–40]. The results achieved were quite similar to those obtained for CFST members filled with normal concrete, albeit with slightly lower strength capacities [38]. Recently, Liu et al. [24] used RAC and sulphoaluminate cement in varied concrete mixtures, which were reinforced with steel fibre and used as infilling material for 54 CFST beams. Their experimental study reveals that the reinforcement of the recycled-aggregate concrete mixture with 1.2% steel fibre can achieve similar flexural behaviour to that of the CFST beams that are filled with normal concrete. Furthermore, using the lightweight concrete infill material can significantly reduce the overall self-weight of the CFST members [17,41–43], which is significant given that the self-weight represents one of the most important challenges in terms of adopting composite

members of this type in modern structures. Generally, using expanded polystyrene (EPS) beads as a replacement for the raw aggregate is the most effective method that is usually used to achieve lightweight concrete mixtures [19,44–46]. For instance, EPS can also be sourced from waste polystyrene materials (recycled materials). However, the performance of cold-formed steel tubes (the Slender tube section) filled with lightweight and recycled concrete material under pure flexural loading has not yet been extensively investigated. In such cases, as explained earlier, the overall self-weight and the cost impact constitute important factors in the Engineer's decision to adopt this system in modern composite structures. Therefore, at present, finding an alternative CFST composite structural design concept is a matter of urgency.

The main objectives of this research can be highlighted as follows: The first aim was to examine the suitability of using the new concept of cold-formed tubular cross-sections fabricated from two pieces of galvanized C-sections (face-to-face connection) in the Slender CFST beam's system, in which the lips of these C-sections are expected to behave as internal steel stiffeners along the top and bottom flanges. Second, this research project aimed to explore the behaviour of the proposed Slender CFST beams, specifically when recycled-aggregate concrete mixture is used instead of the normal concrete mixture, in order to reduce the self-weight and cost of the beam's section. Lastly, to date, all of the theoretical methods for predicting the flexural strength of CFST beams were developed according to different standards, and they were mostly intended to be applied for beams with conventional tube sections (unstiffened sections). As such, this study aimed to develop a new analytical method that can theoretically predict the flexural strength capacity (ultimate bending moment; *Mu*) of the internally stiffened CFST beam's cross-section with varied numbers/sizes of steel stiffeners; this constitutes one of the main novelties of the current research work. Therefore, five CFST specimens were prepared for this purpose and were experimentally tested under pure flexural loading, including one specimen without concrete filling material (hollow tube section). The finite element (FE) ABAQUS software was used to develop and analyse an additional 20 models that were designed to examine the effect of further parameters that were not explored experimentally.

#### **2. Experimental Approach**

### *2.1. Preparation of Samples*

In order to suggest a new concept for the design of Slender steel tube cross-sections that are internally stiffened, in this research, five specimens were produced from two steel C-sections connected by means of tack welding to obtain the suggested steel tubular section, as shown in Figure 1. These prefabricated tubular sections had 20 mm lips that functioned as internal stiffeners along the top and bottom flanges of the tubular beams. A single specimen was tested as a hollow tube beam (HB), while the other four tubes' specimens were filled with different concrete mixtures (filled beam; FB). The raw coarse aggregate of the concrete mixtures was replaced with different recycled-aggregate materials (EPS and RAC), since one of the main objectives of this research was to reduce the self-weight and cost of the mixture. The replacement percentages of raw coarse aggregate (by volume) were equal to 0% (filled beam designation; FB-RC0), 30% (FB-RC30), 50% (FB-RC50), and 70% (FB-RC70), as presented in Table 1. All steel tubes specimens were placed vertically, after which the concrete material was poured from the top ends in multiple stages, while the bottom ends were temporary sealed to prevent water leakage.

#### *2.2. Material Properties*

Steel tube: Three coupons were cut from the cold-formed C-sections and prepared in accordance with the ASTM-E8/E8M-2009 standard. The average results of the yield tensile strength (*fy*), maximum elongation (%), ultimate tensile strength (*fu*), and elastic modulus (*Es*) were 489 N/mm<sup>2</sup> , 27.4%, 558 N/mm<sup>2</sup> , and 201 <sup>×</sup> <sup>10</sup><sup>3</sup> N/mm<sup>2</sup> , respectively, and these results were obtained from the direct tensile of coupons test.

Concrete mixtures: the recycled aggregates used in the suggested concrete mixtures were EPS beads (4–6.3 mm) with a density of 9.5 (kg/m<sup>3</sup> ), and RAC (4.75–16 mm) with a density of 1278 kg/m<sup>3</sup> ; these densities were much lower than that of the raw coarse aggregate (1490 kg/m<sup>3</sup> ). As explained earlier, the first concrete mixture was prepared without the use of any recycled aggregate (normal concrete; 0% replacement aggregate: RC0). The second concrete mixture (RC30) was prepared by replacing 30% (by volume) of the raw coarse aggregate by EPS beads only. Meanwhile, the concrete mixtures RC50 and RC70 were also made by replacing 30% of the raw aggregate with EPS beads and adding another 20 and 40% (in terms of total volume) of RAC, respectively, bringing the total replacement to 50 and 70%, respectively. In addition, silica fume (SF) material was used to enhance the bonding performance between the cement and EPS beads, as previously advised in [45]. Lastly, for all concrete mixtures, a water/cement (w/c) ratio of 0.46 was used together with the superplasticizer liquid (Real Flow 611). The proportions of the concrete mixtures are presented in Table 2. For each concrete mixture, three samples of concrete cubes (150 mm) were prepared, cured for up to 28 days, and tested in accordance with BS 1881: 1983. results were obtained from the direct tensile of coupons test. Concrete mixtures: the recycled aggregates used in the suggested concrete mixtures were EPS beads (4–6.3 mm) with a density of 9.5 (kg/m3), and RAC (4.75–16 mm) with a density of 1278 kg/m3; these densities were much lower than that of the raw coarse aggregate (1490 kg/m3). As explained earlier, the first concrete mixture was prepared without the use of any recycled aggregate (normal concrete; 0% replacement aggregate: RC0). The second concrete mixture (RC30) was prepared by replacing 30% (by volume) of the raw coarse aggregate by EPS beads only. Meanwhile, the concrete mixtures RC50 and RC70 were also made by replacing 30% of the raw aggregate with EPS beads and adding another 20 and 40% (in terms of total volume) of RAC, respectively, bringing the total replacement to 50 and 70%, respectively. In addition, silica fume (SF) material was used to enhance the bonding performance between the cement and EPS beads, as previously advised in [45]. Lastly, for all concrete mixtures, a water/cement (w/c) ratio of 0.46 was used together with the superplasticizer liquid (Real Flow 611). The proportions of the concrete mixtures are presented in Table 2. For each concrete mixture, three samples of concrete cubes (150 mm) were prepared, cured for up to 28 days, and tested in accordance with BS 1881: 1983.

Steel tube: Three coupons were cut from the cold-formed C-sections and prepared in accordance with the ASTM-E8/E8M-2009 standard. The average results of the yield tensile strength (*fy*), maximum elongation (%), ultimate tensile strength (*fu*), and elastic modulus (*Es*) were 489 N/mm2, 27.4%, 558 N/mm2, and 201 × 103 N/mm2, respectively, and these

*Materials* **2021**, *14*, x FOR PEER REVIEW 4 of 26

*2.2. Material Properties* 

**Figure 1.** Cold-formed steel C-sections (all dimensions in mm). **Figure 1.** Cold-formed steel C-sections (all dimensions in mm).


**Table 2.** Proportions of the concrete mixtures (kg/m<sup>3</sup> ).


**Mixture** 

**Designations Cement Fine** 

#### *2.3. Test Setup 2.3. Test Setup*

**Agg.** 

**Coarse Agg.** 

**Silica Fume** 

All specimens were tested under four-point static loads, as shown in the schematic of the test setup in Figure 2. A hydraulic jack with a maximum capacity of 500 kN was used to apply the static load on the suggested CFST specimens. Utilizing linear variable differential transducers (LVDTs) type KYOWA, Osaka, Japan, the vertical deflections of the specimens were recorded at different locations. The strain gauges SG1 to SG5 were fixed vertically at the mid-span of each CFST specimen. A data logger was used for collection of the data from the load cell, strain gauges, and LVDTs during the tests, and these data were recorded in a computerized system. All specimens were tested under four-point static loads, as shown in the schematic of the test setup in Figure 2. A hydraulic jack with a maximum capacity of 500 kN was used to apply the static load on the suggested CFST specimens. Utilizing linear variable differential transducers (LVDTs) type KYOWA, Osaka, Japan, the vertical deflections of the specimens were recorded at different locations. The strain gauges SG1 to SG5 were fixed vertically at the mid-span of each CFST specimen. A data logger was used for collection of the data from the load cell, strain gauges, and LVDTs during the tests, and these data were recorded in a computerized system.

*Materials* **2021**, *14*, x FOR PEER REVIEW 5 of 26

**Table 2.** Proportions of the concrete mixtures (kg/m3).

RC0 390 700 1115 - - - - - 180 128 2295 RC30 350 700 781 40 30 2.1 - - 180 147 1881 RC50 350 700 558 40 30 2.1 20 190 180 151 1813 RC70 350 700 335 40 30 2.1 40 380 180 154 1772

**(%) EPS RAC** 

**(%) RAC Water Slump** 

**(mm) Density** 

**EPS** 

**Figure 2.** Schematic of specimens' test setup (all dimensions in mm). **Figure 2.** Schematic of specimens' test setup (all dimensions in mm).

#### **3. Discussion of Experimental Results 3. Discussion of Experimental Results**

#### *3.1. Failure Modes 3.1. Failure Modes*

All specimens were tested beyond their strength capacities in order to study their extreme failure performance. The hollow steel tube specimen (HB) showed the inward tube's buckling failure at the support points only, wherein the failure of tube's webs was increased gradually by increasing the applied load, as shown in Figure 3a. Meanwhile, all filled specimens (FB-RC0, FB-RC30, FB-RC50, and FB-RC70) showed almost typical failure modes regardless of the types of their concrete mixtures. It was found that, for all filled specimens, outward buckling failure started occurring at the top tube's flange, between the two-point loads of these filed specimens, particularly when the applied load values reached about 70–80% of their ultimate loading capacity, as shown in Figure 3b. This performance was due to the effects of the concrete infill materials, which mainly All specimens were tested beyond their strength capacities in order to study their extreme failure performance. The hollow steel tube specimen (HB) showed the inward tube's buckling failure at the support points only, wherein the failure of tube's webs was increased gradually by increasing the applied load, as shown in Figure 3a. Meanwhile, all filled specimens (FB-RC0, FB-RC30, FB-RC50, and FB-RC70) showed almost typical failure modes regardless of the types of their concrete mixtures. It was found that, for all filled specimens, outward buckling failure started occurring at the top tube's flange, between the two-point loads of these filed specimens, particularly when the applied load values reached about 70–80% of their ultimate loading capacity, as shown in Figure 3b. This performance was due to the effects of the concrete infill materials, which mainly prevented the occurrence of inward buckling failure of the steel tube. Figure 3c presents all filled specimens after testing at the extreme failure limits.

prevented the occurrence of inward buckling failure of the steel tube. Figure 3c presents

Furthermore, the 20 mm lips located at the top and bottom edges of the prefabricated cold-formed C-sections were well bonded with the concrete core, thus implying that these lips functioned sufficiently as internal stiffeners for the top and bottom flanges of the steel tubes of these filled specimens. Therefore, the outward buckling failure occurred only at the half-width of their top flanges (see Figure 3b). These lips achieved similar contributions to those of the additionally welded steel stiffeners that were included in previous research to stiffen the Slender CFST members [19,47]. Moreover, there was no slippage failure between the steel tube and the concrete core observed at both opened ends of the tested filled specimens. Unlike the unfilled specimen (HB), the filled specimens (FB-RC0, FB-RC30, FB-RC50 and FB-RC70) displayed a smooth deflection behaviour during the loading stages, which was quite similar to the half-sine curve, as illustrated in Figure 4. This performance allowed the CFST beams to distribute the point loads with almost uniform stress along the beam's span, which was very similar to the beams that were loaded in a uniform loading scenario during the practical application of the members. Based on this finding, the above hypothesis was confirmed insofar as the currently suggested prefabricated cold-formed tubular steel beam filled with recycled concrete materials performed very similarly to the conversional CFST beams, which were

all filled specimens after testing at the extreme failure limits.

tested earlier, for example, in [11,17,19,48].

**Figure 3.** *Cont*.

testing.

(**c**) **Figure 3.** Typical failure modes: (**a**) unfilled specimen (HB); (**b**) filled specimen (FB-RC30); (**c**) all filled specimens after testing.

**Figure 3.** Typical failure modes: (**a**) unfilled specimen (HB); (**b**) filled specimen (FB-RC30); (**c**) all filled specimens after **Figure 3.** Typical failure modes: (**a**) unfilled specimen (HB); (**b**) filled specimen (FB-RC30); (**c**) all filled specimens after testing.

Furthermore, the 20 mm lips located at the top and bottom edges of the prefabricated cold-formed C-sections were well bonded with the concrete core, thus implying that these lips functioned sufficiently as internal stiffeners for the top and bottom flanges of the steel tubes of these filled specimens. Therefore, the outward buckling failure occurred only at the half-width of their top flanges (see Figure 3b). These lips achieved similar contributions to those of the additionally welded steel stiffeners that were included in previous research to stiffen the Slender CFST members [19,47]. Moreover, there was no slippage failure between the steel tube and the concrete core observed at both opened ends of the tested filled specimens. Unlike the unfilled specimen (HB), the filled specimens (FB-RC0, FB-RC30, FB-RC50 and FB-RC70) displayed a smooth deflection behaviour during the loading stages, which was quite similar to the half-sine curve, as illustrated in Figure 4. This performance allowed the CFST beams to distribute the point loads with almost uniform stress along the beam's span, which was very similar to the beams that were loaded in a uniform loading scenario during the practical application of the members. Based on this finding, the above hypothesis was confirmed insofar as the currently suggested prefabricated cold-formed tubular steel beam filled with recycled concrete materials performed very similarly to the conversional CFST beams, which were tested earlier, for example, in [11,17,19,48].

#### *3.2. Flexural Behaviour and Strength Capacity*

This section discusses the flexural behaviour, ultimate moment capacity (*Mu*), and the moment vs. strain relationships of the tested specimens. The relationships between the bending moment and the deflection at mid-span for the tested specimens are compared in Figure 5, including the hollow specimen (HB). In general, for the concrete filled specimens, the moment–deflection curves continued to show linear behaviour until an archive of about 50–60% of their *M<sup>u</sup>* values, after which these curves behaved as elasto-plastics up to the loading limits of about 70–80% of their *M<sup>u</sup>* values (at this point, buckling failure of the top flange started to occur). Thus, due to further outward buckling failure, the same moment– deflection curves showed fully plastic behaviour with continual slow decreases. The *M<sup>u</sup>* value of the filled specimens was recorded at the deflection limit of *Le*/50 − *Le*/60 [49], which was about 46 mm to 56 mm. However, the moment–deflection curve of the unfilled specimen (HB) showed an almost linear behaviour until the peak loading point (*Mu*) was achieved, at which the bottom flanges started to buckle inwardly at the supporting points; furthermore, as a result of extreme tube buckling failure, the loading curve began to descend.

**Figure 4.** Deflection shapes of the filled specimens: (**a**) FB-RC0; (**b**) FB-RC30; (**c**) FB-RC50; (**d**) FB-RC70. **Figure 4.** Deflection shapes of the filled specimens: (**a**) FB-RC0; (**b**) FB-RC30; (**c**) FB-RC50; (**d**) FB-RC70.

**Figure 5.** Moment vs. mid-span deflection relationships. **Figure 5.** Moment vs. mid-span deflection relationships.

The *Mu* values of the tested specimens are presented in Table 1. The HB specimen achieved the ultimate flexural capacity of 14.1 kN·m; this capacity increased to 57.7 kN·m when the same fabricated steel tubular specimen was filled with normal concrete (FB-RC0), achieving an improvement of 409%. The tubular specimen filled with varied The *M<sup>u</sup>* values of the tested specimens are presented in Table 1. The HB specimen achieved the ultimate flexural capacity of 14.1 kN·m; this capacity increased to 57.7 kN·m when the same fabricated steel tubular specimen was filled with normal concrete (FB-RC0), achieving an improvement of 409%. The tubular specimen filled with varied recycled

recycled concrete achieved *Mu* values that were slightly lower than those of the FB-RC0 specimen (57.7 kN·m); these values were 53.7 kN·m, 52.6 kN·m and 51.5 kN·m, for

regardless of the aggregate replacement percentages, the concrete filled tubular specimens achieved much higher flexural strength capacity values as compared to the capacity of the hollow specimen (HB), with a strength improvement of about 363 to 381%. It is worth mentioning that the self-weight of the tube beam was substantially increased due to the addition of the filled concrete; however, the flexural strength was remarkably enhanced, as stated above (363–409%). From the structural engineering point of view, this represents a very important finding regardless of the increment of the weight, and, in this case, the role of the structural engineer, who can prospectively determine scenarios in which this would be reliable for the targeted purpose of construction projects. On the other hand, the cost was also found to be a vital element in the construction design when we compared the concrete cost with the steel cost; this was the case because, in global terms, the cost of

Furthermore, the moment vs. longitudinal tensile strain relationships obtained at the mid-span of the steel tube of the filled specimens are presented in Figure 6. In this figure, it can be seen that the moment–strain relationships showed a fairly typical behaviour for all of the filled specimens regardless of the percentages of replacement aggregate in the concrete material [11,24]. The strain gauges SG1 and SG2 (see Figure 2) showed gradually increasing negative values (compression stress) with the increasing of the bending loads, while the strain gauges located at the bottom-half of beam's cross-section (SG4 and SG5) showed gradual increases in their tension-related stress values. In contrast, the strain gauge SG3 (located in the middle of the specimen's mid-span) showed a slight increase in tension-related stress, confirming that a positive correlation was discovered with the upward movement of the neutral axis in the tested filled specimens as the bending loads increased. Figure 7, as an example, shows the maximum strains distributed at the tube's mid-span depth along its cross-section for specimens FB-RC0 and FB-RC30 during a variety of loading stages. In the current study, the aforementioned moment–strain relationships of the fabricated filled-steel tubular beams behaved similarly to those

1 ton of steel is about 15–20 times the cost of 1 ton of concrete material.

obtained for the conventional cold-formed CFST beams [11,48].

concrete achieved *M<sup>u</sup>* values that were slightly lower than those of the FB-RC0 specimen (57.7 kN·m); these values were 53.7 kN·m, 52.6 kN·m and 51.5 kN·m, for specimens FB-RC30, FB-RC50 and FB-RC70, respectively. In addition, it was noticed that, regardless of the aggregate replacement percentages, the concrete filled tubular specimens achieved much higher flexural strength capacity values as compared to the capacity of the hollow specimen (HB), with a strength improvement of about 363 to 381%. It is worth mentioning that the self-weight of the tube beam was substantially increased due to the addition of the filled concrete; however, the flexural strength was remarkably enhanced, as stated above (363–409%). From the structural engineering point of view, this represents a very important finding regardless of the increment of the weight, and, in this case, the role of the structural engineer, who can prospectively determine scenarios in which this would be reliable for the targeted purpose of construction projects. On the other hand, the cost was also found to be a vital element in the construction design when we compared the concrete cost with the steel cost; this was the case because, in global terms, the cost of 1 ton of steel is about 15–20 times the cost of 1 ton of concrete material.

Furthermore, the moment vs. longitudinal tensile strain relationships obtained at the mid-span of the steel tube of the filled specimens are presented in Figure 6. In this figure, it can be seen that the moment–strain relationships showed a fairly typical behaviour for all of the filled specimens regardless of the percentages of replacement aggregate in the concrete material [11,24]. The strain gauges SG1 and SG2 (see Figure 2) showed gradually increasing negative values (compression stress) with the increasing of the bending loads, while the strain gauges located at the bottom-half of beam's cross-section (SG4 and SG5) showed gradual increases in their tension-related stress values. In contrast, the strain gauge SG3 (located in the middle of the specimen's mid-span) showed a slight increase in tension-related stress, confirming that a positive correlation was discovered with the upward movement of the neutral axis in the tested filled specimens as the bending loads increased. Figure 7, as an example, shows the maximum strains distributed at the tube's mid-span depth along its cross-section for specimens FB-RC0 and FB-RC30 during a variety of loading stages. In the current study, the aforementioned moment–strain relationships of the fabricated filled-steel tubular beams behaved similarly to those obtained for the conventional cold-formed CFST beams [11,48].

#### *3.3. Flexural Stiffness*

The initial stiffness (*K<sup>i</sup>* ) and serviceability stiffness (*Ks*) levels of tested specimens are usually measured from the moment vs. mid-span curvature relationships [50–53], which are based on moment values of 0.2*M<sup>u</sup>* and 0.6*Mu*, respectively. The *K<sup>i</sup>* and *K<sup>s</sup>* values of the tested specimens are presented in Table 1. The HB specimen achieved the lowest flexural stiffness values as compared to those of the filled specimens (FB), which were 1207 kN·m<sup>2</sup> and 1121 kN·m<sup>2</sup> for *K<sup>i</sup>* and *K<sup>s</sup>* , respectively. These values increased to 2216 kN·m<sup>2</sup> and 1924 kN·m<sup>2</sup> for the FB-RC0 specimen due to the influence of the normal concrete infill material. In general, similar flexural stiffness improvement behaviours were recorded for all of the filled recycled concrete materials (FB-RC30, FB-RC50 and FB-RC7), but with slightly lower values than that of FB-RC0, since they had lower concrete modulus (*Ec*) values. For example, compared to the FB-RC0 specimen (0% aggregate replacement), the FB-RC70 specimen, which was filled with recycled concrete (30% EPS plus 40% RAC) achieved lower *K<sup>i</sup>* and *K<sup>s</sup>* values (−6.7% and −12.1%, respectively). From the above discussion, it can be concluded that even when using the recycled aggregate in the concrete infill materials of the cold-formed CFST beams, the flexural stiffness can nevertheless be significantly improved, to a far greater degree than that of the corresponding hollow tube beam.

**Figure 6.** Moment vs. longitudinal strain relationships of the filled specimens: (**a**) FB-RC0; (**b**) FB-RC30; (**c**) FB-RC50; (**d**) FB-RC70. **Figure 6.** Moment vs. longitudinal strain relationships of the filled specimens: (**a**) FB-RC0; (**b**) FB-RC30; (**c**) FB-RC50; (**d**) FB-RC70. **Figure 6.** Moment vs. longitudinal strain relationships of the filled specimens: (**a**) FB-RC0; (**b**) FB-RC30; (**c**) FB-RC50; (**d**) FB-RC70.

**Figure 7.** Depth of steel tube cross-section vs. longitudinal strain relationships: (**a**) FB-RC0; (**b**) FB-RC50. **Figure 7.** Depth of steel tube cross-section vs. longitudinal strain relationships: (**a**) FB-RC0; (**b**) FB-RC50. **Figure 7.** Depth of steel tube cross-section vs. longitudinal strain relationships: (**a**) FB-RC0; (**b**) FB-RC50.

#### *3.4. Energy Absorption Index*

The ability of CFST beams to dissipate energy compared to that of the hollow steel tube beams was confirmed in several previous research studies [19,54]. Most commonly, the energy absorption index (EAI) of CFST beams was estimated from the cumulative area under the load vs. the deflection curves, up to the point at which the maximum beam strength limit had been reached [22,54,55]. The EAI values of the tested specimen in the current research are presented in Table 1; these values clearly confirm that the energy dissipation ability of the newly fabricated hollow specimen was extremely enhanced when filled with normal/recycled concrete materials. For example, the EAI value of the HB specimen increased from 201.8 kN·mm to 5996 kN·mm (29.7 times) when it was filled with normal concrete (FB-RC0). Moreover, compared to the tubular specimens that were filled with normal concrete (0% replacement aggregate), the specimens that were filled with recycled aggregate (EPS and RAC) had a slightly lower ability to dissipate energy, which can be considered a logical outcome since they achieved slightly lower loading capacities (see Figure 6). It is interesting to note that the EAI values of specimens FB-RC30, FB-RC50, and FB-RC70 were approximately 5539 kN·mm (−7.6%), 5493 kN·mm (−8.3%), and 439 kN·mm (−9.2%), respectively, as compared to the value of the FB-RC0 specimen (5996 kN·mm). These values represent significant achievements when compared to the value obtained for the corresponding unfilled HB specimen (201.8 kN·mm), being approximately 26.9 to 27.5 times higher

#### **4. Numerical Approach**

#### *4.1. Development and Verification of the Numerical Model*

The suggested CFST beam was further investigated in terms of its flexural behaviour by using the ABAQUS 6.14 software to conduct non-linear finite element (FE) analysis. In order to reduce the time conception of the analyses of the models, a typical 3D quarter model was built to simulate the actual tested CFST specimen, as shown in Figure 8, which had the advantages of the symmetric cross-section and the beam's loading scenario already having been prepared [15,19,56–58]. The actual test loading scenario was implemented in the FE modelling scenario by allowing the nodes that were positioned at the upper tube's flange (loading point) to gradually move downwards during the FE analysis, using the incremental downward displacement option (displacement control approach), which is available in the software. The nodes placed at the support's location were restricted from horizontal and vertical movement, but they were allowed to freely rotate around the X-axis to simulate the roller support. Then, the loading value of the FE model was obtained from the reaction forces that were observed at the support's nodes [19,56].

The main component materials of the currently developed FE CFST models are the concrete infill and the steel tube (double C-sections). For the concrete component, the C3D8R element type was used, which has eight nodes integrated with six degrees of freedom, while, for the steel tube components, the type S4R shell element was used. The penalty friction coefficient of 0.75 was adopted in the current FE analysis to realise the mechanical interaction between the surfaces of the steel and concrete parts. In general, several parameters that mainly affected the proper friction coefficient values, such as the loading type, the size/shape of the beam's cross-section, and the properties of the materials were selected [10,19,56–58]. Thus, in the current study, a convergence study was adopted in order to establish a suitable friction coefficient value (0.75), where several preliminary FE CFST models, which had varied friction coefficient values ranging from 0.4 to 0.9, were utilised. A finer mesh size was used for the distance between the two applied loads to sufficiently represent the failure modes of the analysed CFST models.

Consequently, the material properties of the steel tube and concrete components of the FE model were the same as those of the corresponding tested specimen. Generally, the concrete material is considered to be a brittle material since it cracks under tension-related stress and is crushed under compression stress [15]. Thus, the "Concrete Damage Plasticity" approach was adopted for the current FE models in order to ensure the compressive and

tensile performance of the concrete infill component [15,57,59,60]. The elastic modulus and Poisson's ratio of the steel component were identified in the elastic-isotropic section, while the plastic-isotropic option was used to identify the steel-yielding strength and the relevant strain values. For both the steel and concrete materials of the developed FE CFST models, the constitutive stress vs. strain relationships were estimated by adopting the same expressions that were used earlier in [15,56]. *Materials* **2021**, *14*, x FOR PEER REVIEW 12 of 26 model was obtained from the reaction forces that were observed at the support's nodes [19,56].

**Figure 8.** Typical finite element CFST model. **Figure 8.** Typical finite element CFST model.

The main component materials of the currently developed FE CFST models are the concrete infill and the steel tube (double C-sections). For the concrete component, the C3D8R element type was used, which has eight nodes integrated with six degrees of freedom, while, for the steel tube components, the type S4R shell element was used. The penalty friction coefficient of 0.75 was adopted in the current FE analysis to realise the mechanical interaction between the surfaces of the steel and concrete parts. In general, several parameters that mainly affected the proper friction coefficient values, such as the loading type, the size/shape of the beam's cross-section, and the properties of the materials were selected [10,19,56–58]. Thus, in the current study, a convergence study was adopted in order to establish a suitable friction coefficient value (0.75), where several preliminary FE CFST models, which had varied friction coefficient values ranging from 0.4 to 0.9, were utilised. A finer mesh size was used for the distance between the two applied loads to sufficiently represent the failure modes of the analysed CFST models. The validity of the current FE models was confirmed via the corresponding experimental results; almost all of the filled specimens with varied concrete mixtures achieved similar behaviours and close flexural strength capacities. Thus, the FB-RC30 specimen (a steel tubular beam filled with a recycled concrete mixture of 30% EPS beads) was selected to verify the relevant FE model. The numerical analysis showed that the moment vs. mid-span deflection relationship of the FB-RC30 (FE) model sufficiently agreed with that of the corresponding tested specimen (FB-RC30), as shown in Figure 9a. The flexural strength capacity value of the numerical model of FB-RC30 was overestimated by about 3.1% (55.4 kN·m) as compared to that obtained from the corresponding tested specimen, which was an acceptable degree of deviation. In addition, the outward buckling failure occurring at the top-half cross-section of the tested FB-RC30 specimen, for the distance between the two-point loads, was numerically simulated by the relevant FE model, as shown in Figure 9b.

#### Consequently, the material properties of the steel tube and concrete components of the FE model were the same as those of the corresponding tested specimen. Generally, the *4.2. Parametric Studies*

concrete material is considered to be a brittle material since it cracks under tension-related stress and is crushed under compression stress [15]. Thus, the "Concrete Damage Plasticity" approach was adopted for the current FE models in order to ensure the compressive and tensile performance of the concrete infill component [15,57,59,60]. The elastic modulus and Poisson's ratio of the steel component were identified in the elasticisotropic section, while the plastic-isotropic option was used to identify the steel-yielding strength and the relevant strain values. For both the steel and concrete materials of the After confirming the validity of the developed CFST model, additional models were built and analysed to investigate the influence of further parameters. In particular, the effects of varied tube thickness (1.0 mm to 3.0 mm), concrete infill strength (14.6 MPa to 55.0 MPa), steel yield strength (275 MPa to 550 MPa), and steel tube depth (150 mm to 250 mm) were studied, and these were categorized into groups A, B, C and D, respectively. Table 3 presents the model's designation along with the physical properties adopted for the 20 CFST models, including the *Mu*, *K<sup>i</sup>* and *K<sup>s</sup>* values derived from their FE analyses.

developed FE CFST models, the constitutive stress vs. strain relationships were estimated

achieved similar behaviours and close flexural strength capacities. Thus, the FB-RC30

by adopting the same expressions that were used earlier in [15,56].

relevant FE model, as shown in Figure 9b.

specimen (a steel tubular beam filled with a recycled concrete mixture of 30% EPS beads) was selected to verify the relevant FE model. The numerical analysis showed that the moment vs. mid-span deflection relationship of the FB-RC30 (FE) model sufficiently agreed with that of the corresponding tested specimen (FB-RC30), as shown in Figure 9a. The flexural strength capacity value of the numerical model of FB-RC30 was overestimated by about 3.1% (55.4 kN·m) as compared to that obtained from the corresponding tested specimen, which was an acceptable degree of deviation. In addition, the outward buckling failure occurring at the top-half cross-section of the tested FB-RC30 specimen, for the distance between the two-point loads, was numerically simulated by the

**Figure 9.** Verification of the FB-RC30 FE model with the corresponding tested specimen: (**a**) flexural behavior; (**b**) numerical failure mode. **Figure 9.** Verification of the FB-RC30 FE model with the corresponding tested specimen: (**a**) flexural behavior; (**b**) numerical failure mode.


**Table 3.** FE model designations and analysis results.
