Effect of Recycled Aggregate on the Compressive Behaviour of Short Concrete Columns

In this study, 12 specimens with three different lateral reinforcement ratios (ρ=0.022, 0.012 and 0.007) and four RG replacement ratios (0%, 30%, 50% and 100%) were prepared and tested under pure axial load in order to investigate the influence of recycled coarse aggregate (RG) and lateral tie reinforcement ratio (ρ) on the compressive behavior of a confined recycled aggregate concrete (RAC) column. The experimental data obtained from the tests are reported and compared with the predicted values obtained from the widely accepted Mander’s confined concrete model developed for conventional confined concrete to assess applicability of the model to recycled aggregate concrete. The results showed that the effect of RG on the compressive behavior of the RAC was negligible at low stress level, and it began to appear as the load increased at load levels above about 35% of Peak, regardless of ρ. The slope of the normalized compressive strength (f’cc/f’co) and normalized corresponding strain (εcc/εco) against ρ showed that the lateral reinforcement exerted the greatest effect on the stress and strain enhancement in the RG-30 series, and the smallest effect in the RG-50 series. The ratios of experimental values to those predicted from Mander’s model, f’cc_exp/f’cc_model and εcc_exp/εcc_model, showed that the model tended to give unconservative strength (f’cc) and strain (εcc) values for the RAC columns. *Corresponding author: Hasan Hastemoglu, Suleyman Demirel University, Faculty of Architecture 32260, Isparta, Turkey, Tel: +90 246 2118259; E-mail: hasanhastemoglu@sdu.edu.tr Received October 12, 2015; Accepted October 23, 2015; Published November 05, 2015 Citation: Hastemoglu H (2015) Effect of Recycled Aggregate on the Compressive Behaviour of Short Concrete Columns. J Civil Environ Eng 5: 194. doi:10.4172/2165784X.1000194 Copyright: © 2015 Hastemoglu H. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.


Introduction
The need to recycle old construction materials has been well established over the last few decades by government agencies and the construction industry due to the increasing cost of waste storage and the depletion of natural resources [1]. Among the numerous construction waste materials, old concrete, in particular, is a promising source of recycled aggregate with sufficient quality for civil engineering applications with moderate performance requirements [2,3]. While most of the studies on the subject thus far have focused on the processing of old concrete and the properties of recycled aggregate concrete (RAC), only a limited number of studies have reported the performance of structural members made of RAC: beam [4][5][6], column [7], joint [8], and frame [9]. Although a direct comparison of the results from different studies is difficult due to a lack of coherent concrete constituents, the results generally indicate that the strength decreases with slight increase in the corresponding strain as recycled aggregate content in the concrete increases. This strength decrement was attributed to a weak interface between the new mortar and the recycled aggregate and between the adhered mortar and the recycled aggregate [10,11]. The reduction in strength due to RG can be offset by two approaches: adjusting proportions of concrete constituents or providing proper reinforcement. The choice between the two approaches should be made based on the required structural performance and cost. Low tensile-high compressive strength cementitious materials typically exhibit brittle behavior, but their ductility and compressive strength improves significantly when subjected to favorable biaxial pressure [12]. For the past few decades, the compressive behavior of concrete confined by properly arranged lateral reinforcements has been extensively studied by many researchers. The effects of various parameters, including concrete strength, amount of lateral reinforcement, lateral reinforcement configuration, dimensions, and cross section geometry, on confined concrete behavior are now sufficiently understood. As a result, the use of confined concrete columns has become a common practice for seismic resistant design [13]. Nevertheless, as aforementioned, most of the confined concrete studies focused on concrete containing virgin aggregate. Recently, the performance of reinforced RAC frame, which satisfies the requirement set by the ACI earthquake resistant design code and the Chinese standard, has been reported by Xiao et al. [9]. Although Xiao's test demonstrated the potential use of reinforced recycled aggregate in the seismic resistance frame, few studies have investigated the effect of recycled aggregate on confined concrete members, of which behavior is critical for modern, high performance design. Hence, in an attempt to broaden the use of RAC, the influences of the recycled coarse aggregate (RG) content and the volumetric ratio of lateral reinforcement on the compressive behavior of confined columns subjected to concentric load are experimentally investigated here. In addition, the applicability of the existing design model [14] is assessed against the experimental results.

Materials
River sand and crushed gravel were used as natural fine (NS) and natural coarse aggregate (NG), respectively. The recycled coarse aggregate (RG) was obtained from 5 concrete (at least 20 years old, with a typical design compressive strength of 18 ~ 20 MPa) collected from an apartment redevelopment site in Korea. The RG was separated from the cement matrix through multiple crushing stages while the loose fine particles were washed off the surface of the RG to minimize the effect of the fines generated during the crushing [15]. The resulting RG only contained small amounts of adhered old mortar and had similar physical properties to those of the NA, as shown in Table 1 Concrete mixes were prepared by replacing NG with RG at four different replacement ratios of 0%, 30%, 50% and 100% as shown in Table 2. The amounts of high performance water reducing admixture (polycarboxylic acid type) and air entraining agents (neutralized vinsol resin type) were adjusted to achieve the target slump and air content of 200 mm and 6% respectively. The average compressive strength of the control concrete the tests was 30 ~ 34 MPa (Table 2).
For each steel reinforcement bar size, three steel samples were tested under tension. The average yield strength, yield strain, Young's modulus values from the test, and specified ultimate strength of the steel rebar are given in Table 3.

Column specimens
A total of 12 specimens of concrete columns were prepared with three different lateral reinforcement ratios (ρ=0.022, 0.012, and 0.007) and the four different RG replacement ratio. The tie hoop spacing was set at 30 mm (ρ=0.022), 65 mm (ρ=0.012) and 100 mm (ρ=0.007) to satisfy the minimum tie hoop spacing specified by the two standards: D/2 (100 mm), based on the earthquake-resistant design of the Architectural Institute of Korea; and D/4 (50 mm), based on the earthquake-resistant design of the ACI Committee 318-05. The details of the test specimens are listed in Table 4, and the detailed dimensions of the column specimens are illustrated in Figure 1. All the specimens shared the same dimensions of 200 mm × 200 mm cross section and 600 mm height. The upper and lower 100 mm of the specimens along the length were reinforced internally with tie hoops spaced at 20 mm and fitted externally with a carbon fiber reinforced plastic wrap to prevent the bearing failure of the specimen ends prior to failure of the test area of the specimens (Table 4 and Figure 2). The tie hoops were made of D6 (6 mm nominal diameter) rebar and one D13 (13 mm nominal diameter) longitudinal rebar was placed at each corner of the column. The hooks of the tie hoops were anchored with 135° bending, with an extended length of 6db, which is longer than the minimum length required by the ACI Code. The cover depth of concrete from the center of the tie hoops was 17 mm, and the cross section of the core surrounded by the center of a tie hoop was 166 mm × 166 mm. The ratio of the core area to the entire cross section was fixed at 69% for all specimens.

Test setup and measurement
A square plate with a semicircular seat was placed on the top of the specimen to prevent any moment load on the specimen by eccentric load prior to application of the axial load. The loading was     applied under load control such that the initial rate of displacement was about 0.30 mm/min, equivalent to 0.00075 strain/min. As load increased the loading rate was carefully adjusted in order to trace the unloading part of the curves. The loading was stopped when the postpeak descending curve shape became apparent or when stable loading could no longer be attained. The deformation measurements were recorded at constant time intervals using a data logger (TML TDS-303). The axial deformation of the specimen was recorded from the four linear variable differential transducers (LVDT) installed at each corner of the specimen. The average deformation value from the four LVDTs was calculated and used as the axial deformation value. We note here that the fixtures with vertical rigid rod with horizontal end plate placed above the LVDTs were attached to the top of the specimen in order to measure compressive deformation over the entire length of the specimen.

Failure modes
The maximum (Peak) and critical (Pcr) loads at which significant cover spalling occurred are listed in Table 5. The first visible crack appeared in a vertical direction near a corner edge at 60 ~ 80% of Peak, as the cover concrete deterioration worsened, and Ppeak was reached at a strain much greater than those of the unconfined concretes. Subsequently, with further increase in the applied load, the cover concrete of the specimens began to spall. For a given tie hoop spacing, the rate of cover concrete spalling was proportional to the RG replacement ratio (Figure 3).
After localized crack damage was observed on a side of the column, notable spalling of the cover concrete occurred with a gradual decrease in the applied load. Spalling of the cover concrete occurred at a lower applied load in RAC than in natural aggregate concrete (NAC). The lower Pcr of RAC can be attributed to the relatively higher water absorptiveness of RG, which should have created a steeper moisture content gradient from the cover layer to the interior layer of the specimen. In turn, the moisture content gradient due to RG can cause higher incompatible drying shrinkage between the interior and exterior RAC, compared to that of NAC [16], thereby facilitating the cover was spalling. Furthermore, the lower bond strength between the rebar and RAC than NAC [17] also contribute to the lower cover spalling load and the faster rate of cover degradation in RAC compared to those of the NAC columns.

Compressive stress-strain curves
Stress-strain curves of the plain concrete cylinders with different RG contents up to the maximum stress are shown in Figure 3. The curves show a greater loss of stiffness at a given load level with increasing RG replacement ratio. The effect of RG on the compressive behavior of concrete became apparent at about 10 MPa, which is 31% of the average strength of the four specimens, similar to the result reported by Topcu IB, et al. [18]. The axial compression tests of the laterally reinforced RAC columns showed that their compressive behavior in almost entire strain range, both the ascending and descending branches of the stress-strain curve, is affected by the RG, as shown in Figure 4. In all the specimens, the ascending branch of the stress-strain curves of the column containing RG showed a smaller increase in stress per given strain increase than that of the concrete column containing NA only, and the Ppeak value of the RAC column was lower than that of the NAC column. The lower stiffness of RAC compared to that of NAC can be attributed to presence of micro-    cracks in the RG and adhered mortar on the RG surface [19,20]. At low load below 600kN (15MPa), the specimens having the same tie hoop spacing showed almost identical compressive behavior, regardless of the RG replacement ratio. However, the compressive behavior of the specimens with different RG replacement ratio began to diverge with increasing load. In the descending branch, the shape of the curves indicated that the effect of the RG content on the post-peak behavior of the column was diminished by the lateral reinforcement as the ρ ratio increased: steady curves were recorded from the RG-X-30 series which showed a gentler slope with less RG content, while curves with abrupt change in slopes were recorded from the RG-X-65 and the RG-X-100 series. In case of RG-X-30 series, the specimens shown similar ductility effects until strain 0.03 regardless of RG replacement ratio, but in case of RG-X-65 series, the ductility effects of RAC declined than that of the RG-0-65(0.016) as the minimum 0.0098(RG-100-65) and the maximum 0.0132(RG-30-65) with increasing of RG replacement ratio. In the RG-X-100 series, the strain value of RAC declined than that of RG-0-100(0.095) as the minimum 0.0062(RG-30-100) and the maximum 0.0069(RG-50-100), and it was failure more quickly than RG-0-100 specimen. The steepness of the descending curves of the test specimens should correspond to the rate of concrete deterioration mentioned earlier (Figure 4).

Effect of lateral tie reinforcement ratio (ρ) on the confined RG concrete strength and the corresponding strain
The effect of ρ on the confined RAC behavior was examined by comparing the slope of the normalized compressive strength (f'cc/f'co) and of the normalized corresponding strain (εcc/εco) with increasing ρ. Figure 6 shows that, for all four RG contents, the strength and strain enhancements of concrete were proportional to ρ. The slopes of the fitted curve in Figure 5a indicated that the effect of lateral reinforcement on the strength enhancement was greatest in the RG-30 series, followed in order by RG-100, RG-0, and RG-50. Similarly, the slopes of the fitted curve in Figure 5b indicated that ρ exerted the greatest effect on the strain enhancement in the RG-30 series, followed in order by RG-50, RG-0, and RG-100. Although the fitted curves of the RAC showed variations in strength and strain enhancement by lateral reinforcement from those of NAC, the confinement seemed to be effective regardless of the RG replacement ratio over all. For a fixed ρ, the strength enhancement due to the tie hoops seemed to increase with increasing RG content (at ρ=0.004; f'cc/f'co=1.35 (RG-30), f'cc/f'co=1.38 (RG-50), f'cc/f'co=1.42 (RG-100)). Similarly, the strain enhancement tended to increase with increasing aggregate content (at ρ=0.006; εcc/εco=1.48 (RG-30), εcc /εco=1.53 (RG-50), εcc/εco=1.59 (RG-100)) ( Figure 6).

Comparison between the test results and values predicted by the existing confined NAC model
To assess the applicability of the existing compressive confined concrete model to the confined RAC, the experimental values of the confined RAC concrete's compressive strength (f'cc_exp) and corresponding compressive strain (εcc_exp) at f'cc_exp were compared with the values predicted by the existing model. Mander's model, equations (1) through (3), was selected as it is widely used in practice and in research.
The confinement effectiveness coefficient (ke) value in Equation (3) was calculated according to the method suggested for rectangular concrete sections confined by rectangular hoops with or without cross tie by Mander et al. [14].
Where, f'co=compressive strength of unconfined concrete; εco=compressive strain of unconfined concrete at f′co; f 'cc=compressive strength of confined concrete; εcc=compressive strain of confined concrete at f'cc; fyh=yield strength of lateral reinforcement; ke=confinement effectiveness coefficient; ρ=volumetric ratio of transverse steel reinforcement. The ratios of the experimental values to those predicted from the existing model, f'cc_exp/f'cc_model and εcc_ exp/εcc_model, are shown in Table 6. The average f'cc_exp/f'cc_model and εcc_exp/εcc_model were 0.98 and 0.79, with standard deviation of 0.069 and 0.065, respectively. These results indicated that the existing Mander's model gave the unconservative strength (f'cc) and strain (εcc) values for the RAC columns. This may be explained the fact that as Mander's model was originally developed to predict the confined strength and corresponding strain of normal concrete made of natural aggregates, it therefore does not account for different RG replacement ratio in conjunction with different ρ. The discrepancy observed between the predicted and experimental values is in agreement with Van Mier [21] and Shah and Ahmad [22], who stated that it is not sufficient to determine the mechanical behavior of confined concrete by parameters such as concrete strength, yield strength and volumetric ratio of the confining reinforcement. The importance of aggregate type was also reported by El-Dash and Ramadan [23], who found that the aggregate type influences the behavior of confined concrete, especially the roundness of the aggregate which assists to distribute the lateral confinement pressure and helps the concrete element to resist higher stress than the angular aggregate. Therefore, as well as the parameters noted above, the type of aggregate and its content shall also be considered in the model ( Table 6). The existing model reasonably estimates the compressive strength of the RAC columns with the lateral reinforcement is shown in Figure 6. The average ratio of f'cc_exp/f'cc_ model was 0.92, 1.01, and 1.04 for hoop spacing of 30 mm, 65 mm, and 100 mm, respectively. However, when the strain at f'cc is considered, Figure 6 clearly shows that the existing model overestimates the εcc value by an average of 25%, 20% and 20% with 30 mm, 65 mm and 100 mm respectively. In general, the existing equations (1) -(3) tend to overestimate the confined effect of RAC with increasing tie hoop spacing, which renders the existing Mander's model unconservative for use in the design of RAC, especially with high lateral reinforcement. Therefore, the model needs to be modified. Also the f'cc_exp/f'cc_ model and εcc_exp/εcc_model ratios are proportional to the RG content at a tie hoop spacing of 30 mm. However, this proportionality was no longer evident as the tie hoop spacing was increased to 65 mm and 100 mm, indicating that the existing model can only be successfully modified by considering effect of ρ that is dependent on the RG replacement ratio.

Conclusions
Based on the results of the RAC compressive test and comparison between the test results and the values predicted by the existing model, the following conclusions were drawn: (1) The effect of RG on the compressive confined RAC column was negligible in the ascending branch of the load-strain curve up to 600 kN, which was 35% of the average confined strength of all specimens tested.
(2) Examination of the stress-strain curves indicated that the influence of the RG content increased in both the ascending and descending branches of the stress-strain curves with decreasing ρ.
(3) The ratios of the experimental values to those predicted from the existing model (f'cc_exp/f'cc_model) ranged from 1.00 to 1.12, and the εcc_exp/εcc_model ratio from 1.00 to 1.16. The existing Mander's model tended to give unconservative strength (f'cc) and strain (εcc) values for the RAC columns.