Abstract
Concrete was made with four different types of coarse aggregate, i.e., natural stone aggregate concrete (NSAC), crushed virgin clay brick aggregate concrete (VBAC), crushed recycled brick concrete aggregate concrete (RBAC), and crushed recycled stone concrete aggregate concrete (RSAC). Beam specimens prepared from these four types of concrete were subjected to pure torsional moment up to failure. From this experimental procedure, ultimate torque at failure along with twisting angle were observed. From the torque vs. twisting angle curves, torsional stiffness and torsional toughness were also evaluated for these four types of concrete. It was observed that ultimate torque of VBAC and RBAC was 95% and 90% of that of NSAC, respectively. The torsional toughness of VBAC was found to be 68%–72% of that of NSAC. In addition, experimental torques were compared with predictions of torsional strength as per five commonly used models. For VBAC, the ultimate torque prediction made by skew bending theory was found to be the closest to the experimental findings.
1. Introduction
The use of alternative aggregate as opposed to natural aggregate has been on the agenda for construction companies, regulatory bodies, researchers, and academicians for the last few decades. Several investigations have been reported in the existing literature regarding the use and applicability of such materials in the building industry [1–3]. Various waste materials like concrete from old demolished buildings and structures, industrial wastes and byproducts, etc. are now being used in the production of concrete as a replacement for aggregate and supplementary cementitious material for two primary reasons: to devise ways to dispose of waste materials to promote a circular economy and to reduce burden on existing natural resources like natural aggregate as their supply is limited [4–8]. Crushed clay bricks have been used extensively in Bangladesh and adjacent areas of India for long time as natural stone is in short supply. For constructing new structures, these old brick concrete buildings are being teared down that is generating a large amount demolished concrete made from brick aggregate. Numerous studies are available in existing literature regarding the recyclability of such old brick aggregate concrete as aggregate for fresh concrete casting [9–11]. Similar studies are also available for virgin brick aggregate concrete [12–15]. It has been observed from these studies that the physical, mechanical as well as durability properties of brick and recycled brick aggregate concrete are inferior to those of natural stone aggregate concrete (NSAC) of equivalent compressive strength.
Though many aspects of mechanical properties have been studied and reported in the available literature for these two types of concrete (brick and recycled brick concrete), extensive examination of torsional behavior has not been conducted in detail. Modern analytical and design techniques with the advent of finite element based software have led to relatively smaller members, lowering conservatism that sometimes barely satisfies torsional requirements. Hence, understanding the details of torsional behavior for any type of concrete has become an important aspect of the analysis of concrete structures and structural elements. Torsional behavior for different types of plain and reinforced normal and high-strength concrete have been examined experimentally that have been reported in the existing literature. These studies have examined effect of concrete properties as well as effect of longitudinal and transverse reinforcement on torsional behavior [16–19]. A few studies on torsional behavior of recycled stone aggregate concrete have also showed that torsional properties of such concrete are considerably inferior when 100% natural stone aggregate is replaced by the recycled aggregate [20–22]. Most of these studies employed special loading arrangement that generated pure torsional stresses in their respective beam specimens.
In line with these, a testing scheme was carried out at the Bangladesh University of Engineering and Technology to observe the comparative torsional properties of normal strength concrete made from natural stone, crushed brick, recycled brick concrete, and recycled stone as coarse aggregate. Beam specimens were prepared from these four types of coarse aggregate and were subjected to torsional load, and subsequently, their twisting pattern, failure torque, and twisting angles were observed. Information found from testing operations was analyzed and compared to get insight into the difference in torsional behavior of concrete prepared from four different types of aggregate. Further, experimental results were compared with standard torsional models widely utilized to estimate the torsional strength of concrete. This process identified the most suitable models to predict torsional strength for crushed brick and recycled brick concrete.
2. Materials and Methods
2.1. Cement and Fine Aggregate
Throughout the experimental program, ASTM Type 1 Ordinary Portland Cement that complied with the ASTM C150 [23] standard was employed as the binder. The sand used in this investigation was collected from the northeastern area of Bangladesh, which had an F.M. of 2.6, in order to make all concrete samples uniform in terms of fine aggregate. Gradation of this sand was such that it fitted within the limiting boundaries specified in ASTM C33 [24]. The fine aggregate’s water absorption and specific gravity were evaluated as per ASTM C29 [25] and ASTM C128 [26], respectively, and were found to be 1.34% and 2.62, respectively.
2.2. Coarse Aggregates
Four different types of coarse aggregate, namely, natural stone aggregate (NSA), virgin brick aggregate (VBA), recycled brick concrete aggregate (RBA), and recycled stone concrete aggregate (RSA), were utilized in this work to prepare respective concrete samples. Natural crushed limestone aggregate was procured from the local market. Bangladesh standard BDS 208:2002 [27] categorizes brick into three groups. Of these three, “S” type of bricks is particularly suitable for aggregate production. Hence “S” type of brick was procured and was subsequently crushed in crusher machine to produce VBA. For the two types of recycled aggregate, two separate construction sites were selected with old buildings that were to be demolished to make way for new buildings. One of the old buildings, 37-year-old, was constructed using brick aggregate concrete, whereas, the other 23-year-old building was made of natural aggregate concrete. Concrete chunks were properly collected from these two buildings and crushed in a crusher machine. As the mechanical properties of concrete made from recycled aggregate depend on the quality of the recycled aggregate, debris, dust, and other undesirable particles were removed through careful screening [28].
In order to ensure a size distribution such that the gradation curve depicted in Figure 1 remained within the predetermined limit as set forth in ASTM C33 [24], all four types of aggregate were sieved using a 25 mm or smaller sieve. Figure 2 shows the four different types of coarse aggregate used in this experiment. Clear distinction between VBA (Figure 2(a)) and RBA (Figure 2(b) can be made because the latter’s surface is obviously covered in cement mortar. Similar differences can be found between RSA (Figure 2(c)) and NSA (Figure 2(d)). The fundamental characteristics of the NSA, RSA, VBA, and RBA aggregates are shown in Table 1, along with the standard used to assess those characteristics. It is obvious from this table that NSA had higher density and substantially less water absorption capacity than that of the VBA, RBA, and RSA. The densities of VBA and RSA were found to be similar. However, RBA had much lower density, which might be caused by the aggregate particles being attached to old cement mortar [29]. Further, water absorption was found to be highest for RBA. On the other hand, it was observed that VBA’s water absorption was substantially closer to the first-class brick’s terminal limit as specified in BDS 202:2008 [27]. Table 1 also reports that NSA has the lowest Los Angeles (LA) abrasion value, indicating it to be the toughest of these four types of aggregates. On the other hand, RBA was found to be softer compared to the other aggregates, as its LA abrasion value was the highest. The flakiness index was also measured to give an indication of the shape of the aggregate. The flakiness index of these four different aggregates varied between a small range of 18% and 23%, demonstrating that the shapes of the aggregates were mostly spherical in nature. All types of coarse aggregates (NSA, VBA, RSA, and RBA) were soaked in water for 24 hr and were made saturated surface dry condition before being mixed into the concrete mixer.


(a)

(b)

(c)

(d)
3. Concrete Mix Design, Sample Preparation, and Experimental Program
Concrete samples were prepared using NSA, VBA, RSA, and RBA aggregates and are represented as NSAC, VBAC, RSAC, and RBAC, respectively. Four different water-to-cement (w/c) ratios of 0.596, 0.569, 0.527, and 0.504 were used in this work in order to produce concrete with compressive strength of 18.96 MPa (2,750 psi), 20.68 MPa (3,000 psi), 22.41 MPa (3,250 psi), and 24.13 MPa (3,500 psi) considering slump of 75–100 mm, respectively. Accordingly, concrete samples were designated both with type of coarse aggregate and target compressive strength, i.e., NSAC18.9, VBAC20.68, RBAC22.41, and so forth. The w/c ratios were on the higher side in this work since no admixture was used to achieve the desired workability of concrete as well as the range of relatively low-compressive strengths that were examined. Table 2 shows the concrete mix design ratios calculated as per American Concrete Institute (ACI) 211.1-91 [34]. Table 2 also reports the achieved slump measured as per ASTM C143 [35] and fresh density of concrete. The slump values were found to be within acceptable limits, but for VBAC and RSAC, slump values were higher compared to NSAC and RSAC. As expected, highest fresh density was found for NSAC, followed by RSAC, VBAC, and RBAC.
Beam specimens of 150 mm × 150 mm × 400 mm were prepared for evaluating torsional capacity, angle of twist, crack pattern, etc. for concrete with different coarse aggregates and different strengths. Size of beam were made to remain constant for all types of concrete to negate size effect related issues on torsional behavior of concrete [36–38]. Apart from this, nine-cylinder specimens (100 mm × 200 mm) were also prepared for evaluating the compressive strength, splitting tensile strengths, and density void ratios of each type of concrete. All specimens were cured by immersion in water for 28 days before they were subjected to respective testing. Table 3 shows a tally of the concrete specimens that were used in this experimental program. Altogether, a total of 160 specimens were used in this work.
3.1. Experimental Setup
Figure 3(a) shows beam specimens (150 mm × 150 mm × 400 mm) being loaded at their diagonally opposite ends with steel I joists (painted in red) on the top surface. The load was applied through a column fixed at the center of this joist. This column was attached to a universal testing machine (Brand: Shimadzu; Capacity 1,000 tons) that applied loads in displacement control mode. From the I joist, load was transferred to the concrete beam through spherical balls. Again, these spherical balls rested on stainless steel bearing plates (37.5 mm × 37.5 mm × 6 mm) that were attached to the beam surface using epoxy (Figure 3(b)). Figure 3(a) also shows a laser displacement sensor in front of the beam specimen that recorded the specimen’s displacement.

(a)

(b)
At the center zone of the beam, this opposing load application and support system produced pure torsion. In order to avoid indeterminacy and prevent the generation of axial forces, one spherical ball was allowed to roll in both directions while the other ball had the ability to roll in the transverse direction only. The experimental procedure followed here is in line with that followed by Bažant et al. [36].
3.2. Sensor, Device, and Software Setup for Torsional Test
The load was applied at the midpoint of the I joist under displacement control regime, which transmitted the load equally at the two diagonal ends of the top surface of the beam. As the support at the beam bottom surface was on two opposite diagonal ends, this special arrangement created a pure torsional moment in the beam by creating opposite movement at all four ends of the beam. Four laser displacement measurement sensors were used at four different sides of the specimen, as shown in Figure 4, to capture this displacement. Specification of the sensor was, Laser Radiation, Maximum Output (peak): 1.2 Mw, Pulse Duration: 22 ms, Wavelength: 650 mm, Class 2 Laser Product, satisfying EN 60825-1:2007. Magnetic stands were used to mount the laser sensor so that it could be positioned wherever it was most appropriate. The displacement of the beam was determined using Lab View software [42]. The data collector (NI USB-6211, 16 inputs, 16 bits, 250 KS/s, multifunction) device collected the signal from the laser displacement measurement device and sent it to the lab view software. These measured displacements were processed to determine the rotation of the beam. The laser displacement sensor, data collector, and necessary power input devices are shown in Figure 5.


(a)

(b)

(c)

(d)

(e)
4. Results and Discussion
4.1. Compressive and Splitting Tensile Strength
The 28-day compressive and splitting tensile strengths for each kind of concrete are shown in Table 4. The achieved compressive strength was only somewhat greater than the target strength for the case of NSAC. For the remaining three, i.e., VBAC, RSAC, and RBAC, the observed compressive strength at 28th was lower than the target strength. The LAA test showed that VBA, RBA, and RSA are relatively softer than NSA (Table 1); as a result, using the ACI code mix design approach had difficulties in reaching target strength for concrete prepared from these aggregates [9, 12–14]. The physical characteristics of each form of concrete, including density, porosity, and absorption, are listed in Table 5. This table shows that NSAC and RSAC have the highest apparent and bulk densities, followed by VBAC and RBAC. In contrast to NSAC and RSAC, voids and absorption were much higher in RBAC and VBAC.
4.2. Torsional Parameters
The average torsional properties obtained from the two tested beams for each case are reported in Table 6. According to the results for a given target strength, NSAC had the highest ultimate torque, followed by VBAC, RSAC, and RBAC. In the case of VBAC, ultimate torque was around 95% of that of NSAC. The ultimate torque of RSAC ranged between 92% and 93% of that of NSAC. RBAC’s ultimate torque was significantly lower, ranging from 87% to 90% of that of NSAC. Taking into account the sample size and concrete compressive strength, the torsional capacity observed for NSAC in this study was found to be equivalent to that reported in the existing literature [21, 43]. The twisting angle at ultimate torque changed with torque, i.e., the larger the twisting angle, the greater the torque. While preparing concrete with a specific target compressive strength, all parameters were kept constant except the coarse aggregate. Hence, the difference in ultimate torque for NSAC, VBAC, RSAC, and RBAC was clearly related to changes in coarse aggregate characteristics. Table 1 shows that VBA is substantially softer, has a higher LAA value, and is more porous than NSA. As a result, VBAC’s torsional properties were found to be inferior to those of NSAC. Similar reasonings applied to concrete with recycled aggregates, i.e., RSAC and RBAC.
Figure 6(a)–6(d) shows the torque vs. twisting angle diagram up to failure for all the beams tested in this experimental scheme. Two distinctive behaviors of torque versus twisting curves was observed from these figures. For NSAC and RSAC, torque versus twisting angle curves had a steeper initial slope that progressively became flatter as it approached failure. Though RSA is softer than NSA due to the preexistence of microcracks as evident from LAA value (Table 1), the torque vs. twisting graph was found to be similar for NSAC and RSAC. This was probably due to the fact that residual cementing properties and better interlocking due to preexisting microcracks in RSA enhanced the overall stiffness of RSAC. In contrast, for brick aggregate concretes, i.e., both VBAC and RBAC, the initial slope of torque vs. twisting angle was much flatter compared to stone (both NSAC and RSAC) aggregate concrete. After the initial zone, the slope of the torque vs. twisting curve for VBAC and RBAC abruptly increased and stayed quite steady up to the failure zone. At close to failure, the slope of the torque vs. twisting curve for VBAC and RBAC decreased rapidly, and the curve became much flatter. The initial slope in VBAC and RBAC was lower due to the initiation and progression of microcracks at a much lower stress level than that for NSAC. Another reason may be that the porosity, voids and absorption in VBAC and RBAC are much higher than those in NSAC (Table 4). At initial stages of loading, filling up of these voids due to applied stresses might have taken place that reduced the stiffness. After a sufficient portion of voids were rearranged and filled due to applied stresses, the stiffness started to increase again. The increase in stiffness afterward might also be due to the interlocking action between the rough internal crack surfaces. Table 5 lists the initial stiffness of all types of concrete, from which it can be seen that for VBAC and RBAC, the initial stiffness was only 47%–52% of that of NSAC. Although VBAC had a lower initial slope than RSAC, ultimate torque was always higher in VBAC.

(a)

(b)

(c)

(d)
Additionally, Table 6 provides information on the torsional toughness, i.e., the energy absorbed up to torsional failure for concrete made from different types of coarse aggregates. As can be seen, VBAC had a torsional toughness of 68%–72% of that of NSAC. Whereas, the torsional toughness of RBAC was only about 58%–61% of that of NSAC. Torsional toughness for RSAC, on the other hand, was between 84% and 88% of that of NSAC. The torque vs. twisting diagram (Figure 6(a)–6(d)) reveals that for the cases of VBAC and RBAC, the initial torsional rigidity was much lower, resulting in torque vs. twisting curves that had substantially smaller areas under them, i.e., absorbed energy was much less for the case of VBAC and RBAC. On the basis of this, it can be concluded that concrete built with VBA and RBA requires significantly less energy to fail in torsion.
4.3. Crack Pattern
Figure 7 shows a three-dimensional schematic diagram of a beam with a failure crack, along with the location and orientation of the cracks. The crack angle after failure was measured for all beams. Pure torsion creates biaxial compression–tension situation along diagonal directions that cause failure at around 45° crack angles [44]. Figure 8 shows crack orientation with distance from the corresponding end for all cracks. From this figure, it can be seen that for NSAC beams, failure angles varied between 44.3° and 46.5°. For VBAC beams, the failure angle was found to be to varied between 43° and 56.5°. Whereas, for RSAC beams, the failure angle was between 44.2° and 46.2°, and for RBAC beams, the failure angle had a large variation between 39.2° and 56.9°, respectively. As can be seen, for NSAC and RSAC, the angle of the crack is close to 45°, but for VBAC and RBAC, the range of variation of the crack angle was much higher. Greater variation in the crack angles for VBAC and RBAC compared to NSAC and RSAC possibly caused by the presence of more microdefects and increased porosity in VBA and RBA compared to NSA and RSA.


Figure 9 shows typical crack surfaces for NSAC and VBAC beams. It may be observed that the crack surface of the VBAC beam is comparatively smoother than that of the NSAC beam, which is irregular with protruding aggregates over the surface.

(a)

(b)
5. Result Comparison with Different Theories and Codes
Torsional behavior of concrete made different coarse aggregate were next compared with the available torsional strength model. Five widely used models, i.e., elastic theory [45, 46], plastic theory [46, 47], ACI code method [48], Turkish standard TS500 [49], and Skew bending theory [50] were examined for this purpose.
According to elastic theory, which is from St. Venant’s principle, torsional capacity of plain concrete member is expressed as follows:where T = failure torque, kN m;
ft = tensile strength of concrete, MPa;
x, y = length of short and long sides of the section, respectively;
α = a coefficient with values ranging from 0.208 to 0.333 and dependent on y/x ratios. Since, in the present experimental configuration, y/x = 1, the value of α was taken as 0.208 [45].
Plastic theory incorporates concrete properties in the plastic range. Likewise in the elastic theory, failure is considered when the maximum principal tensile stress becomes equivalent to tensile strength of concrete, ft. The failure torque following plastic theory is expressed as per following equation:
ACI adopted space truss analogy and according to ACI 318-19, torsional strength (cracking torque) is given by:
Acp = area bounded by external concrete cross-sectional perimeter, mm2;
Pcp = perimeter of concrete cross-section, mm. For this experimental scheme, since normal weight concrete was used, λ was taken as 1.
Cracking or failure torque T as per Turkish Standard (TS500) is calculated from the following equation:
S in Equation (4) is the shape factor, and for rectangular section, may be taken as y2x/3. x and y are long and short sides of the rectangle.
fctd = axial tensile strength of concrete and may be considered as fctd = 0.35();
f′c = compressive strength of concrete.
Hsu et al. [51] discovered that when subjected to pure torsion, plain-concrete beams failed by bending. The idea postulated that failure would occur when the tensile stresses generated by a 45° bending component of the torque reached a reduced value of modulus of rupture at wider face of the beam. According to this theory, ultimate torque of plain concrete members can be expressed as in following equation:
Here, represents modulus of rupture of concrete.
Table 7 displays the estimated torsional capacity of these five models along with the actual ultimate torque obtained from the testing scheme. The errors for each data set and the root-mean-square errors (RMSE) for each model and for each concrete type are also presented in Table 7. All models underestimated the experimental results for all types of concrete except for VBAC, which had the lowest experimental torsional capacity. As the prediction from skew bending theory was always higher than all other models, its prediction was also closest to the experimental results except for VBAC. The accuracy of predictions using the plastic theory, the elastic theory, the TS500 Turkish theory, and the ACI space truss analogy decreased in that order. The utilization of various stress levels at which cracks in concrete begin to appear under torsional stress is the main cause of the variations in predictions made by these models. Elastic and TS500 Turkish theory use tensile strength as the stress in concrete when cracks begin. Plastic theory incorporates concrete properties of the plastic range and considers as cracking stress. ACI space truss analogy conservatively considers as the maximum tensile stress for cracks to start. And for skew bending theory, this is set as , the modulus of rupture of concrete. Accordingly, prediction from skew bending theory was highest and the ACI space truss analogy was lowest.
For the case of RBAC, predictions from skew bending theory were actually higher than the experimental results. However, other models underpredicted the experimental results. For RBAC, considering RMSE, the predicted capacity by plastic theory was closest to the experimental results. For all types of concrete, the greatest discrepancy between experimental results and the predicted capacity was found for the case of ACI space truss theory. Skew bending theory may be employed to conservatively predict the torsional capacity of plain concrete made from all types of aggregate except RBA.
6. Conclusion
In this work, a comprehensive testing scheme was carried out to examine and compare the torsional behavior of concrete formed from crushed clay bricks and recycled brick concrete as coarse aggregate. For this, concrete with target strength of 18.9, 20.7, 22.4, and 24.1 MPa were prepared with four different types of coarse aggregate namely natural stone, crushed clay brick, recycled stone, and recycled brick aggregate concrete. Cylinder (Ø 100 mm× 200 mm) and beam (150 mm× 150 mm × 400 mm) specimens were made from those concretes. Cylinder samples were tested to measure compressive strength, splitting tensile strength, density, porosity, and absorption of respective concrete. Beam samples were subjected to pure torsion up to failure to observe ultimate torque and maximum twisting angle as well as torque vs. twisting angle behavior. Based on these experimental results and analysis following conclusions may be stated:(1)Crack due to torque and its angle with surface for NSAC, RSAC was found to remain close to 45° but for VBAC and RBAC, crack angle varied between wide range of 39° and 56°. Crack surface of VBAC and RBAC was relatively smooth compared to NSAC and RSAC. The later two had rough surface with protruding aggregate distributed over the surface.(2)All beams failed with a loud noise, but for VBAC and RBAC beams the noise at failure was much louder compared to the NSAC and RSAC beams.(3)For a given goal strength, NSAC had the highest ultimate torque, followed in descending order by VBAC, RSAC, and RBAC. For VBAC, the maximum torque was roughly 95% of that of NSAC. Whereas for RBAC, maximum torque was between 87% and 90% of that of NSAC. The initial slope of torque vs. angle of twist curves of VBAC and RBAC was only 47%–52% of that of NSAC. VBAC always had a lower initial slope than RSAC, but the ultimate torque for VBAC was found to be higher than that of RSAC.(4)Torsional toughness of VBAC was 68%–72% of that of NSAC. On the other hand, torsional toughness of RBAC was only about 58%–61% of that of NSAC. Concrete built with VBA and RBA requires significantly less energy to fail in torsion compared to equivalent concrete made from NSA.(5)Torsional strength prediction by five widely used models, namely, elastic theory, plastic theory, the ACI code space truss analogy method, the Turkish Standard TS500, and skew bending theory were compared to the experimental results. It was found that, except for one instance (Table 6), in general, all models underestimate the torsional strength of concrete prepared from the four types of aggregates utilized in this work. As per RMSE for VBAC beams, the ultimate torque estimation made by skew bending theory was shown to be the most accurate. For RBAC beams, prediction by the plastic method was found to be nearest to the experimental findings.
Data Availability
The experimental data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The work was funded by Bangladesh University of Engineering and Technology for graduate level research of Mr. Md. Ashraful Alam.