Characterization of concrete composites with recycled plastic aggregates from postconsumer material streams

Construction and Building Materials 182 (2018) 561–572

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Characterization of concrete composites with recycled plastic aggregates from postconsumer material streams Colin Jacob-Vaillancourt, Luca Sorelli ⇑ Department of Civil and Water Engineering, Université Laval, 1065 avenue de la Médecine, Quebec City, Canada

h i g h l i g h t s
Optical sorting allowed to reduce the impurity of sorted plastic to about 5%
A fifth of the aggregate of concrete can be replaced with plastics from postconsumer material streams.
Comparing different plastics, PVC allowed the lowest reduction of compressive strength.
PAG allowed to increase thermal insulation of 5% and reduce water adsorption.
The seasonal variations had an acceptable impact on concrete performances.
The presence of surface contamination and entrapped air are critical factors to consider.
Correlation analysis showed the importance of bulk density and PAG shape.

a r t i c l e

i n f o

Article history: Received 26 April 2017 Received in revised form 9 June 2018 Accepted 9 June 2018

Keywords: Plastic Concrete Recycling Recycled aggregates Waste stream characterization Post-cracking strength Thermal conductivity Water absorptivity

a b s t r a c t Plastic waste is today an crucial environmental issue for which innovative recycling techniques are needed. This research aims at investigating the feasibility of replacing fine aggregate in concrete by recycled plastics from real post-consumer streams. Different kind of plastics were sorted in Material Recovery Facilities (MRF) to realize different Plastic Aggregates (PAG) at different seasonal time. The concrete composites were characterized in terms of compressive strength, post-cracking compressive strength, toughness indeces, thermal conductivity, density, and water absorptivity. Correlation tables were used to understand the key material parameter of PAG governing concrete properties. The effects of the replacement percentage, the PAG kind, the level of impurity, and the time-related variations on the quality of the concrete are discussed. The presented results indicate that concrete composite with PAG from postconsumer waste is a promising research direction for developing eco-responsible construction materials with enhanced thermal insulation and water absorptivity. Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction Worldwide plastics production reached 322 million metric tons (MT) and an average yearly growth rate of nearly 4% in 2015, and Abbreviations: PAG, Plastic aggregates; PET, Polyethyene Teraphtalate; PE, Polyethylene; PVC, Polyvinyl chloride; PP, Polypropylene; PS, Polystyrene; PLA, Polylactic acid; PC, Polycarbonate; PU, Polyurethane; MRF, Materials recovery facility; MT, metric ton; E, Young’s modulus; fc, compressive strength; ft, tensile strength; k, thermal conductivity; TGA, Thermogravimetric analysis; EOL, End of linearity; We, Elastic toughness; I3, Toughness index at 3 time linear deformation; I5, Toughness index at 5 times linear deformation; PCS, Post-cracking strength. ⇑ Corresponding author at: Université Laval, Département de Génie Civil, Pavillon Adrien-Pouliot 2928-A, 1065 avenue de la Médecine Québec (Québec) G1V 0A6, Canada. E-mail address: [email protected] (L. Sorelli). https://doi.org/10.1016/j.conbuildmat.2018.06.083 0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.

the associated environmental pressure has reached a critical point [1]. Despite persistent efforts by governments, businesses and individuals to increase plastic recovery and recycling rates, it is estimated that between 5 and 13 million metric tons end up as debris in rivers and oceans [2]. Even in regions with a developed waste management infrastructure, the global recycling rate is still limited. For instance, postconsumer plastic recycling rates are below 40% in all European countries [1]. In Quebec province (Canada) household plastics recovery rate is below 32%, although the actual recycling rate is expected to be even lower as only 17% of postconsumer plastics have been reportedly sold for recycling [3,4]. There exist several challenges to recycle postconsumer plastic waste [5,6], e.g.: the wide variety of polymer resins found in

562

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

municipal recyclable streams; the general mechanical incompatibility of molten plastics, the high levels of impurities requiring extensive sorting, conditioning and upgrading processes [7]. At Material Recovery Facilities (MRF), different kinds of plastics are generally sorted manually (based on packaging visual appearance) or with optical sorting units [8–10]. The sorted plastic streams are then granulated, washed and often pelletized by conditioners to obtain an easily marketable commodity [11]. Door-to-door collection programs are generating increasing volumes and kinds of plastic [5]. There is today a critical need for recycling technologies that can process large volumes of plastic at low cost with minimal prior conditioning, as an alternative to landfilling or incineration of plastic. In this context, concrete offers a promising opportunity for recycling postconsumer plastic waste due to its extensive worldwide volume production of about 10 km3/y, which is about 5, 7.7 and 10 times more than the yearly use of fired clay, timber, and construction steel, respectively [12]. It is worth mentioning that recycled construction and demolition concrete waste are already used as aggregate for structural concrete [12,14]. Recycling PAG in concrete may help tackle three major environmental issues: (i) the ever-increasing amount of anthropogenic waste in landfills and the environment [13]; (ii) the concerns associated with extracting limited natural resources; (iii) improving the ecological footprint of concrete. A ‘‘systematic reuse of anthropogenic materials from urban areas” may result in a successful urban mining application [15]. In the last decades, numerous research on the use of PAG as sand replacement in concrete mixes have been undertaken [16–18], which demonstrated their potentially beneficial effect on concrete properties, such as: ductility, flexural toughness, density and thermal resistance [19–21]. Table 1 summarizes some results available in open literature on the effect of PAG on concrete properties. In particular, Table 2 reports typical Young’s modulus (E), tensile strength (ft) and thermal conductivity (k) of the polymers considered in this study along with typical concrete raw materials such as aggregates, sand and cement paste. The introduction of PAG in concrete usually negatively affects the Young’s modulus E and the compressive strength (fc). It has been reported

Table 2 Some properties of high volume polymers, and concrete materials. 1 column width. Material

E (GPa) [29,31]

ft (MPa) [29]

k (W/mK) [30,31]

PET PE PVC PP PS Quartzite sand Limestone gravel Cement paste (w/c = 0.5)

2.1–3.1 0.6–1.4 2.7–3.0 1.3–1.8 3.1–3.3 70 70 36–40

55–80 18–30 50–60 25–40 30–55 – – –

0.15 0.33–0.52 0.17–0.21 0.12 0.105 4.45 2.29–2.78 1

that E decreases proportionally with PAG volume [17], e.g., Correira et al. [22] found that E decreased by 13–31% for mixes where sand was substituted by polyethylene terephthalate (PET) aggregates at 7.5% volume. This is somehow expected as PET aggregates have a much lower E value than those of typical mineral aggregates (Table 2). It is also worth mentioning that the range in results was due to the variation of the water-to-cement ratio as the mix-designs were formulated to have a similar workability in terms of slump. Notably, contrarily to round sand particles or pellet-shaped PAG, more angular PAG decreased the workability of fresh concrete [22]. Moreover, PAG size can significantly impact fresh concrete workability, and thus porosity, compaction and mechanical performances in hardened concrete [17,23]. For the compressive strength fc, previous studies have found that replacing sand by PAG also leads to a significant strength loss. Hannawi et al. [24] reported a decrease in fc of respectively 30% and 28% by replacing 10% of fine aggregate volume with PET and polycarbonate (PC) aggregates. The loss of fc was well correlated with the loss of E, hinting to similar degradation mechanisms. Visual inspections of the interface zones under an electron microscope showed large gaps around hydrophobic PAG, which weakens the bond with the cement paste [24–26]. The interface zones between PAG and the surrounding cement matrix resulted to be more porous than that of a cement-mineral aggregates interface [24]. Compressive PostCracking Strength (PCS) of concrete has several beneficial effects

Table 1 Summary of the reported effects of PAG on key material parameters. Polymer used for PAG

Size1

PAG volume2

fc loss3 (%)

E loss3 (%)

PET

<8 mm <4 mm <2 mm

7.5 7.5 10 20 10 50 10 50 20 20 15 30 13.1 33.7 10 30 100

31 14 48.5 75.7 30.5 69 27.2 63.9 49 38 18.6 21.8 56 94 0 0 86

31 13

RPOMIX (compound) PET PC PET

a

<10 mm <5 mm

PVC

<11.4 mm <2.6 mm <5 mm

PU

<4 mm

PET

<2.5 mm

Hydraulic lime mortars. Lightweight concrete with expanded clay aggregates. PUR foam lightweight mortars. d Mortars. e Concrete. 1 Estimated from a 95% passing value from size analysis curves. 2 Expressed as % of fine aggregates in concrete, and as % total volume in mortars. 3 Under wet curing conditions, when specified. 4 Dried specimens. b c

k loss4 (%)

Slump loss (%)

Reference

Notes

8 0

[22]

e

[21]

a

[24]

e

[24]

e

[23]

e

[20]

b

[27]

c

[28]

d

50 68.4 61.9 48 45 13.8 18.9

100 44

55 85 10.3 20.6 73.8

-14 57

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

on concrete structures, e.g.: a better confinement to the longitudinal reinforcement, lower risk of instability of the longitudinal reinforcement, and enhanced rotational capacity of a beam section [32]. However, the effect of PAG on compressive PCS has not been investigated previously. In previous studies, flexural tests were often used to characterize the fracture energy of concrete and mortars incorporating PAG [17,24,33]. Interestingly, the addition of PAG in concrete was found to decrease strength, but to enhance fracture energy. Compared to conventional sand and cement mortar, Hannawi et al. [24] reported a slight reduction of the tensile strength ft in mortar mixtures incorporating PET or PC as fine aggregate, but an increase of up to 6 times of the fracture energy. The ductile PAG inclusions allow delaying crack propagation making more resilient structures [24]. For thermal properties, PAG could reduce the heat conduction in concrete thanks to the low conductivity k of polymers [28,34,35] (see Table 2). For instance, Marzouk et al. found that the conductivity of mortars reduced from 1.26 to 1.13 and 0.63 W/(mK) by replacing fine aggregates with 10% and 50% PET aggregate, respectively [28]. As for thermal insulation, concretes with PAG could enhance the building energy efficiency [35]. In one of the few work investigating the effect of PAG on concrete durability, Kou et al. [20] found that polyvinyl chloride (PVC) aggregates significantly reduced the drying shrinkage and chloride ion penetration of concrete, which are important properties to reduce the crack widths in concrete and enhancing structural durability. From a waste management perspective, recycled material streams are prone to variations in their physico-chemical properties due to factors such as time, industrial processes and sourcing of material [36]. Although of its practical importance, few studies have investigated the effects of the impurities of different sorted polymers and their variations in postconsumer material on concrete properties. Hannawi et al. [24] combined Polyethyene Teraphtalate (PET) and Polycarbonate (PC) for PGA obtaining significant differences in concrete properties (Table 1). Quality control specifications on sorted plastic material properties employed for PAG need to be developed to limit the effect of waste stream variations on the concrete performances. By considering plastic streams sorted in a MRF, the scope of this research is threefold: (i) to study the effect of different PAG on some basic properties of concrete, such as: compressive strength, post-cracking resistance, conductivity, dry density, and water absorptivity; (ii) to study the effect of the polymer kind and volume fraction; (iii) to consider the effect of time-related fluctuations and impurity of sorted postconsumer plastic streams on the concrete properties. 2. Materials and methods 2.1. Cement and aggregates A quarry sand with a maximum size of 4 mm and limestone aggregate with a maximum size of 2.5–10 mm were employed. Fig. 1 shows the size distributions of the sand and limestone aggregates. A Portland cement GUP type (Lafarge, St. Constant) was employed. An air-reducing agent (Eucon Air-Out, Euclid Canada) was used to control the volume of entrapped air in concrete samples. 2.2. Plastic material The samples of mixed postconsumer plastics were collected from a MRF in Quebec (Gaudreau Environment Inc., Victoriaville,

563

Fig. 1. Size grading analysis of fine aggregates used in this study.

Canada) and labelled #1 to #7 by sampling from March 2015 to February 2016. The sorting procedure was two-steps, such as: (i) mixed plastic packaging was manually picked-out from the recyclable materials stream; (ii) the selected samples were further sorted by infrared optical sorting unit (Eagle Vizion, Sherbrooke) which selects plastics based on the different polymer kinds, such as: Polypropylene (PP), Polyethylene (PE), Polystyrene (PS), Polyvinyl chloride (PVC). The sorting process is illustrated in Fig. 2. The residual material is herein called ‘‘Others” and consisted of mixed plastic items that could not be ejected by infrared optical sorting unit (e.g., plastic with irregular size or shape or other polymer types). Fig. 3 shows the variation of polymer composition of the plastic waste due to the seasoning effect. One can note that PP is the major plastic waste (between 56% and 62%), followed by PE (between 14 and 18%), PP (between 10 and 12%), and PVC (between 8% and 10%). 2.3. Plastic aggregates (PAG) As shown in Table 3, the plastic PP, PE, PVC, PS were extracted by infrared optical sorting from the stream samples #3, #5, #6, #7, respectively. From stream sample #1, the different plastic kinds were sorted and recombined into a mixed sample, which is called MIX. As PS and PVC have the highest E-moduli of the considered polymers (Table 2), the polymers PS and PVC were sorted from stream samples #3 to #7 and combined into binary samples called PS-PVC (1 to 5). Those polymers have the lowest economical value on current recycling markets. The selected polymers were ground in a rotary knife granulator with a mesh size of 5 mm to generate coarse graded PAG. Furthermore, PS-PVC fractions were sieved at 1.7 mm to generate fine graded PAG. The size distribution was measured according to standard method ASTM C136 [37]. Finally, part of the sieved MIX fraction was recombined to generate an optimal graded size distribution (called ‘‘Graded PAG”) to match a Fuller distribution with maximum packing in accordance to standard CSA A23.1.14 for fine aggregates (Appendix A) [38], as shown in Fig. 1. Un-compacted bulk density of PAG were measured by means of a calibrated stainless steel vessel of known volume. The PAG particle shape, which affects the bulk density, the fresh concrete workability and the mechanical performances [39], may strongly depends on kind of plastics due to the thicknesses of packaging items [9]. In this work, a shape index u which describes the porosity within un-compacted PAG as follows:

564

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

Fig. 2. Separation of plastic samples into polymeric fractions by optical sorting.

Fig. 3. Variations of the polymer compositions of the seven stream samples collected at the MRF (the legend colors are in the same vertical order of the histogram bar).

Table 3 Samples selected for this study with a single polymer or combination of them (n.c. = not considered). Mixed stream sample ID ? ? ? ? ? ? ?

#1 #2 #3 #4 #5 #6 #7

u¼1

ðm=qÞ V bulk

Sorted single polymer

Mixes of sorted polymers

– n.c. PP – PE PVC PS

MIX (PP, PE, PS, PVC and Others) n.c. PS-PVC1 PS-PVC2 PS-PVC3 PS-PVC4 PS-PVC5

ð1Þ

where u is the % volume of voids between the uncompacted plastic particles, V bulk is the vessel volume, m is the mass of PAG in the

vessel and q is the experimentally determined plastic apparent density. The latter was determined on 30 g samples by helium gas pycnometer (AccuPyc II 1340, Micromeritics) and standard procedure ASTM D5550 [40] with 3 repetitions. Thermogravimetric analysis (TGA) was used to estimate the remaining inorganic contamination (parameter: % polymer) after optical sorting. TGA curves were performed in an oven (Q5000 TGA, TA Instruments) on homogenized PAG samples of 20–50 mg with 3 repetitions according to ASTM E 1131 [41]. Samples were heated to 600 °C at a rate of 10 °C/min in a N2-inert atmosphere, then combusted in the presence of O2 at a rate of 10 °C/min to reach 750 °C. Fig. 4 shows a typical TGA curve (mass vs. temperature). At a given heating rate, anaerobic thermal degradation (pyrolysis) of a specific polymer occurs within known temperatures in the 200–600 °C range. The derivative TGA curves was used to identify different degradation steps, according to a method detailed by Ehrenstein et al. [29]. The largest continuous mass loss was associated with the polymer content in the

565

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

the standard deviation is indicated). Table 4 reports also PAG shape index (from Eq. (1), where a higher value means flakier or more needle-like particles. For the MIX fraction, two shape indices are reported: the first was measured on ‘‘coarse” PAG, while the second value is for ‘‘graded” PAG. As explanation, the 1.7 mm sieving might have favored the sorting of small needle-like particles. Shape index varied from 62.4 (PE) to 73.6 (PS). A visual inspection of the sorted samples showed that PP and PE streams (such as buckets, pails or food tubs and lids) were characterized with thicker injection molded items, whereas PVC and PS streams consisted mostly of thin thermoformed items, such as clamshells and food trays. The PS fraction consisted mostly of thin-walled thermoformed items such as yogurt containers or disposable cups. Thus, the shape index may strongly depend on the packaging kind. Furthermore, the fast chopping action of the powdering process could have created several PAG morphologies. Finally, Table 4 reports the bulk density parameter, which depends on both the apparent density and the shape index.

Fig. 4. Example of TGA results for PAG from PS fraction with 3 test repetitions.

2.4. Concrete mix designs sample, and used to estimate a polymer percentage (‘‘% Polymer”) parameter. The experimental temperatures associated with this mass loss were corroborated with polymeric pyrolysis temperature ranges reported in literature [42]. In the specific case of PVC, polymer pyrolysis occurs in two steps and generates carbon-based residual compounds (‘‘char”) [29]. Therefore, it was necessary to combine mass losses for 2 pyrolysis steps and the mass loss immediately following combustion at 600 °C to better estimate the polymer content. The remaining mass after combustion mainly consists of inorganic material from polymer additives, reinforcement material or surface contamination. As for the studied PAG, Table 4 reports the polymer concentration estimated through TGA in terms of mean value and standard deviation. The impurity concentration (which is defined here as the non polymeric part of sorted plastics) was below 10% with a mean of 5% of weight. The contaminants in a plastic stream may be papers (e.g., cellulose, lignin), adhesive labels, plastic additives (e.g., pigments and reinforcement materials), and other residues such as food or detergents. Table 4 reports also the PAG apparent density measurements with satisfactory reproducibility. The optically sorted polymers exhibited a density between 0.919 and 1.335 g/cm3. Plastic streams PVS-PS 1 to 5 have different apparent density due to seasonal effect. A preliminary analysis of PS and PVC fractions showed that a slight variation in the apparent density may exist between samples of the same polymer kind. Furthermore, the concentration of PS and PVC varied between mixed samples 1 to 7 : based on Fig. 3, the volume content of PS and PVC was 54%±3% and 46%±3% for PS and PVC, respectively (where

The considered concrete consisted of 400 kg/m3 cement; 1024 kg/m3 limestone aggregate; 724 kg/m3 quarry sand and 200 kg/m3 water (% air = 3.7, slump = 125 mm). A volumetric fraction of sand was replaced by PAG samples, while ensuring that final concrete volume remained the same for all mixes. The water-to-cement ratio (w/c = 0.5) was kept constant in this study without any corrections for slump. Table 5 summarizes the 14 mix designs studied in this work which were classified in 3 series: (i) Series ‘‘a” consisted of 4 mix designs with the same plastic material (MIX fraction, i.e., mixed plastic material stream prior to sorting), while varying the PAG volume and size distribution; (ii) Series ‘‘b” consisted of 5 mix design at the same replacement volume content of sand (20%) by PAG, but different plastic kinds. Additional a bMIX sample was fabricated by mixing all the sorted plastics to check the effect of separating them; (iii) Series ‘‘c” consisted of 5 mix design with 10% sand replacement with PAG fractions PS-PVC 1 to 5 to verify the effect of seasonal variations in the plastic stream. Mix-designs were cast according to CSA A23.22C and CSA A23.23C standard procedures [38]. Each mixture was cast into 6 cylinders (height of 150 mm and diameter of 75 mm) for mechanical testing. Additionally, certain mixtures were cast into 3 cylinders (height of 200 mm and diameter of 100 mm) for thermal conductivity, water absorption and density analyses. All cylinders were cured at 100% humidity for 28 days prior to mechanical testing, and 15 days for thermal testing. The curing temperature was about 24 ± 2 °C. The air content and slump of fresh concrete were also characterised according to standard procedures CSA A23.24C and CSA A23.25C, respectively [38]. All cylinders were cured at

Table 4 Characteristics of the plastic streams used in this study. 1.5 column width.

*

Fraction

% Polymer

PAG Apparent density (g/cm3)

PAG Bulk density (g/cm3)

PAG Shape index

MIX

97.76 ± 0.31

0.972

PP PE PS PVC PS-PVC1 PS-PVC2 PS-PVC3 PS-PVC4 PS-PVC5

95.85 ± 0.20 97.11 ± 0.37 91.09 ± 1.31 97.44 ± 0.13 93.79 ± 0.53 93.39 ± 0.25 94.02 ± 0.20 95.07 ± 0.11 94.34 ± 0.44

0.919 0.952 1.086 1.335 1.182 1.208 1.200 1.197 1.195

0.334 * 0.226 0.330 0.358 0.287 0.433 0.380 0.395 0.377 0.366 0.349

66.0 *77.0 64.1 62.4 73.6 67.6 67.9 67.3 68.6 69.5 70.8

Graded size distribution

566

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

Table 5 Concrete mix-designs studied in this work. Tests Series

Fraction

Substitution rate (%)

PAG Size distribution

Air reducing agent (mL/m3)

Concrete ID

Series a

MIX MIX MIX MIX PP PE PVC PS MIX PS-PVC1 PS-PVC2 PS-PVC3 PS-PVC4 PS-PVC5

5 10 10 20 20 20 20 20 20 10 10 10 10 10

Coarse Coarse Graded Coarse Coarse Coarse Coarse Coarse Coarse Fine Fine Fine Fine Fine

0 0 0 0 50 50 50 50 50 25 25 25 25 25

a5 a10 a10_graded a20 bPP bPE bPVC bPS bMIX c1 c2 c3 c4 c5

Series b

Series c

100% humidity for 28 days prior to mechanical testing, while only 15 days for density, water absorption and thermal testing. 2.5. Concrete characterization 2.5.1. Elastic modulus E and compressive strength fc A hydraulic press was employed to carry out the compressive tests. To measure the deformation during compressive testing, 3 linear variable differential transformers were mounted on the cylindrical samples by means of two aluminium rings at a distance of 100 mm. The E-modulus was determined by averaging the slopes of the last two of three cycles in the elastic regime at a controlled rate of 1.1 kN/s. The stress-strain curve of each cylinder was obtained by imposing a displacement of 0.25 mm/min. The test was stopped when the residual stress was equal to 30% of the strength (fc). Results reported in Section 3 are averages of 4–5 test repetitions. 2.5.2. Water absorption and concrete dry density Dry densities were determined by drying the cylinders in an oven at 60 °C until the recorded mass change was <0.5% in 24 h. The cylinder dimensions measured with an accuracy of 10 mm were employed to estimate their volumes. Water absorption was calculated by sinking the cylinders in boiling water for 5 h in accordance to CSA A23.2-11C [38]. 2.5.3. Thermal conductivity Thermal conductivities were obtained using oven-dried cylinders with 2 test repetitions under thermal equilibrium conditions at an average temperature of 10 °C. Each cylinder was fitted with two disc-shaped flowmeters in a fully isolated chamber, and the power flux in W/m2 was automatically acquired for a controlled temperature gradient set at 5 °C between the top and bottom disks. A full description of the apparatus can be found in [43]. Thermal conductivity of the sample was estimated by assuming a simplified Fourier law as follows:

k ¼ q h =rT ½W=m=K

ð2Þ

where h is the specimen height, q is the power flux and rTis the temperature gradient. 2.5.4. Toughness indices The compressive stress-strain curve of each mix design was measured with 4–5 test repetitions for mix-design, while 10 repetitions for the reference concrete. The deformation of the End Of Linearity (EOL) point was conventionally defined on stress vs. strain curve when the tangent modulus differs from the initial E-modulus by more than 15%. Moreover, the area underneath the

stress-strain curve up to the EOL strain was defined as the elastic energy (We). Two toughness indices, I3 and I5, were defined by integrating the stress strain-curve to a total strain of 3 and 5 times the EOL strain, respectively, while normalizing by the elastic compression toughness, as follows : Ii = Wie/We with i = 3 or 5. 3. Results and discussion 3.1. Fresh concrete properties Table 6 reports the fresh properties of mix designs in terms of air content and slump. The replacement of spherical sand grains with flakier PAG reduced slump as found in previous work [19,23]. For Series a, which did not include an air-reducing agent, the air content increased significantly with the increase of PAG volume content. The volume of entrapped air seems to be rather proportional to the PAG volume. The hydrophobic nature of polymers caused air bubbles to form on the surface of PAG. Such trapped air layer on PAG surface could explain the gap observed at the PAGcement interface by previous authors [25,26]. The stability of these trapped air bubbles on the surface of PAG has not been investigated, although the effect of entrapped air on the compressive strength is previously found [39]. As for the rheology of fresh concrete, the air content compensated the slump reduction due to PAG by a lubrication effect. For Series b, an air-reducing agent was used in mix designs to stabilize the air content between 2.5 and 3.5 %, which is comparable to the reference mix (3.7%). The slump values, which were considerably lower than the reference, varied from 40 mm to 10 mm. As for Series c, an air-reducing agent was also

Table 6 Air content and slump of concrete mixes. 1 column width. Concrete mixtures

PAG volume content (%)

Air content (%)

Slump (mm)

Ref a05 a10 a10_graded a20

0 5 10 10 20

3.7 7.2 8 7 11.2

125 110 75 35 80

bPP bPE bPVC bPS bMIX

20 20 20 20 20

2.8 3.1 3 3.5 2.5

40 40 25 10 35

c1 c2 c3 c4 c5

10 10 10 10 10

5.8 2.7 3.1 3.8 7.6

60 55 50 50 85

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

used, but the entrapped air ranged from 2.7 to 7.6%. For Series c, it is possible that surface contamination of the flakes (e.g., leftover soap or detergent) generated air bubbles, especially in the case of concrete c5. In this study, plastic samples did not go through a cleaning process as it is not common practice in material recovery facilities. 3.2. Mechanical properties 3.2.1. E-modulus and compressive strength Fig. 5 shows the effect of the kind of PAG on the elastic modulus E and compressive strength fc in terms of mean values and standard deviation of 5 test repetitions. In Series a (Fig. 5a), the increase of PAG volume lowered both fc and E. At maximum replacement of 20% sand by PAG, fc and E losses are 46.9% and 32.1%, respectively, compared to the reference mix. The lower strength of a Series mixes depended on 3 factors: (i) PAG volume content. PAG with lower compressive strength could act as stress concentration zones favoring damage propagation; (ii) PAGcement weakened interfaces; (iii) increased air content. Notably, the PAG size distribution did not seem to significantly affected fc and E (see a10 and a10_graded). For all the mixtures of Series b, the replacement of 20% sand volume by PAG resulted in a relatively lower loss of fc and E with respect to mix design a20 thanks to the use of an air-reducing agent. Series b showed that polymer kind can significantly affect the mechanical performances of concrete with a reduction of fc varying from 13 to 38 % (Fig. 5b). Interestingly, PVC and PS aggregates respectively gave the best and the worst performances in Series b. In series b, the mean E for mix designs with recycled polymers is 26.6 GPa with a coefficient of variation of 5.9%, while mean fc is 31.8 MPa with a coefficient of variation of 11.4%. It is worth noting that bMIX, simulating the use of PAG that were not polymer-sorted, resulted in the second best mechanical properties for Series b. As for Series c, the volume of PAG was relatively low (10% of total sand volume) and an air-reducing agent was used (Fig. 5c) so that the reduction of E was limited to 12–18% with respect to the reference mix. Series c showcased a good repeatability in E with PAG collected from the same material streams over a period of 11 months. Values ranged between 27.68 GPa (c5) and 30.08 GPa (c3). However, the compressive strength fc showed to be more sensitive to time-related variations by ranging from 31.81 MPa (c5) to 40.85 MPa (c3). In particular, mix design c5 had a very high air

567

content (7.6%), which may hint that this uncertainty is mainly due to variations in entrapped air. 3.2.2. Compressive stress-strain curves and toughness indices Fig. 6a–d show mean compression stress-strain curves used to evaluate PCS. The average stress-strain curves of Series a behaved similarly to that of reference mix curve, although the maximum strength is significantly reduced at 20% replacement of sand (Fig. 6a). Fig. 6b shows the 90% confidence interval bands (based on 5 test repetitions), highlighting the beneficial effect of PAG on PCS dispersion after peak load (the same phenomenon was also observed for Series b and c). In the reference mix, the confidence interval (with 10 samples) quickly expanded after peak strength because of randomly occurring brittle failure. It is possible that PAG acted as reinforcement fibers, somehow delaying damage propagation [21]. It is also noteworthy that graded PAG were slightly more efficient than coarse PAG in maintaining PCS. If residual strength and ductility in PAG mixes is attributed to plastic flakes delaying crack coalescence and propagation, smaller flakes could reduce crack propagation by generating more reinforcing particles for the same volume of plastic. One should also note that lower-strength concretes exhibit often a less brittle behavior. Series b and c, which are characterized by lower entrapped air displayed significantly higher fracture energy. In particular, PVC aggregates in Series b (and their combined use with PS in Series c) generated mixes that exhibit PCS after peak strength even greater than that of the reference mix. At 10% PAG volume, the toughness indices I3 and I5 of Series c were equal or superior to that of the reference mix, except for high entrained-air c5 mixture (Table 7). In the same table, the embedded figure compares the toughness indices. One can observe that the PAG affects more the toughness index I5 than the I3, meaning that the effect on the dissipated energy is more at higher inelastic deformation. In more details, as for series a, the PAG volume increases the toughness index I5. As for series b, the mix bPVC shows also the highest toughness I5 hinting for a better bond between PVC particles and cement paste. Finally, as for series c, the seasonal effect on the toughness indices are limited. 3.2.3. Dry density, water absorption and thermal conductivity (k) In order to study the water absorption and k parameters over a large gradient of concrete dry density, the following 6 mix designs were chosen for further analyses: (i) Ref; (ii) a10; (iii) a10_graded;

Fig. 5. Mean values of elastic moduli E and compressive strength fc for (a) series a; (b) series b and (c) series c. The standard deviation bars are shown.

568

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

Fig. 6. Mean compressive stress-strain curve for (a) Series a with confidence interval bands (CI = 90%); (c) Series b; (d) Series c.

Table 7 Toughness indices for reference and experimental concrete mixtures. Mix

We

I3

I5

Ref a10 a10_graded a20

0.64 0.44 0.39 0.19

7.22 7.11 7.26 7.06

14.39 14.34 15.37 15.63

bMIX bPE bPP bPS bPVC

0.50 0.48 0.51 0.45 0.62

6.93 7.03 6.98 6.57 7.10

14.14 14.34 13.70 13.23 14.51

c1 c2 c3 c4 c5

0.63 0.72 0.74 0.65 0.48

7.29 7.07 7.20 7.16 7.10

14.70 13.78 13.84 14.43 14.50

Fig. 7. For selected mix-designs: (a) Dry density and % absorbed water; (b) Dry density and thermal conductivity.

(iv) a20; (v) bMIX; and (vi) c2. As shown in Fig. 7, dry density values progressively decreased with an increase of air and plastic content by ranging between 2.26 g/cm3 for the reference mix (3.4% air, 0% PAG) to 2.00 g/cm3 for A20 (11.2% air, 20% PAG). Mixtures with high air content such as Series a should normally be more porous and permeable with higher water absorption [17,25]. As shown in Fig. 7a, the mix a20 with the highest air void content was characterized by the lowest water absorption. The air-reducing surfactant may have had the effect of reducing the hydrophobicity of PAG surface. Such results are in agreement with those of Hannawi et al. [24] who reported lower coefficients of water sorptivity with higher PAG contents. The authors attributed these results to the hydrophobic nature of the polymer flakes, which reduces the water

imbibition. It is worth mentioning that water immersion methods do not guarantee a complete saturation for mixes containing PAG due to the entrapped air bubbles caused by polymer flakes [44]. The thermal conductivity k is proportional to the dry density as evinced in Fig. 7b. Besides the effect of the different conductivity of polymers [21,28], the Series a mixtures with higher air contents exhibited lower conductivities. As found by Dermiboga et al. [34], k was mostly influenced by the increased air voids generated by PAG. Mix designs c2 and bMIX, have significantly lower thermal conductivity than that that of the reference, despite the fact they have similar air contents. Thus, thermal conductivity also depends on the conductivity of concrete constituents, volume of air content and PAG volume. Furthermore, Côté and Konrad suggested that

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

thermal conductivity in cement materials is also influenced by the pore shape and the ratio of water-to-entrapped air [43]. 3.3. Correlation study This section investigates the Pearson correlation between the mechanical performances and the characteristics of studied PAG. All the correlation tables are presented in section A for the 3 series and the 6 mix designs investigated in the Section 3.2.3. In particular, Figs. 8 and 9 shows the most relevant correlations. Fig. 8a shows that, for the Series a without air reducer, the compressive strength fc and the volume of entrapped air are strongly correlated to PAG volume. The hydrophobic plastic particles may explain the strong correlation between air content and PAG volume. In Series b, PAG shape index varied strongly as plastic fractions were characterized by different packaging items. Fig. 8b shows the strong dependency of the slump on the shape (angularity) of aggregates for Series b. Although the same granulation process and sieves were applied to all PAG, the difference in container wall thickness generated particles of different morphologies, which in turn affected the slump of fresh concrete. Finally, the entrapped air is also strongly correlated to slump, due to its effect on the viscosity of fresh concrete batches. Fig. 9 shows the effect of the aggregate bulk density on fc for the Series b and Series c. The former which have similar air content, shows a strong correction which likely depends on the intrinsic apparent density (q) of the different plastics. In general, spherical aggregates of graded size distribution lead to higher packing density and better mechanical resistance, while lower packing density are achieved by randomly distributed elongated (or flaky) particles [39]. Based on empirical data, aggregate shape alone may explain up to 22% of concrete fc and 31% of con-

569

crete ft [39,45]. Thus, aggregate mechanical strength and shape are key factors for concrete mechanical properties. Fig. 9b shows that entrapped air is correlated to mechanical properties in Series c (R2 = 0.7). The wide range of entrapped air in Series c does not seem to be strongly dependent on any of the measured PAG properties (shape, % polymer, apparent and bulk densities). Thus, the explanation of such varying higher air content may be inferred to surface contamination of PAG from material stream, such as residues in household detergent of soap bottles. This highlights the importance to add a washing step in the preparation of PAG in order to achieve consistent results in further research. The positive relationship between slump and % air content is also evidenced in Fig. 9b. Finally, Fig. 9c shows a good correlation (R2 = 0.96) between concrete dry density and thermal conductivity k. The changes in dry density resulted from both variation of PAG volumes and variation of percentage of entrapped air. Air has a low k (0.024 W/m/K) and polymers’ k are about one order of magnitude lower than conventional concrete materials (Table 2). 4. Concluding remarks Facing the urgent need to find eco-friendly applications for recycled polymers, this work analyses some thermo-mechanical properties of concrete incorporating PAG from real postconsumer material streams. In particular, 3 effects were considered, such as: (i) the volume and size of PAG; (ii) the kind of recycled polymer; (iii) seasonal or time effect on stream quality. Based on the present results, the following conclusions can be drawn: 1. As for the mechanical properties, the replacement of 20% sand volume by PAG (series a) reduced E-modulus and compressive strength (fc) of concrete by 32% and 47%, respectively;

Fig. 8. (a) For Series a, fc and % air content vs. PAG volume; (b) For Series b, slump vs. PAG shape index.

Fig. 9. Correlation curves: (a) fc vs. PAG bulk density for Series b and Series c; (b) Slump and fc vs. % entrapped air for Series c; (c) Concrete dry density vs. k. (Standard deviation bars is shown).

570

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

2. The addition of air-reducing agent reduces the strength loss due to the PAG addition. Considering the lack of information on measured air content in previous studies on PAG concrete [17], this result highlights the importance to consider the effect of entrapped air induced by PAG and the use of an air-reducing concrete additive; 3. Considering the kind of recycled polymer at 20% replacement of sand (series b), PVC aggregates were found to perform the best (fc reduction of 14%, E reduction of 18%); 4. Replacing 10% of sand with a combination of recycled PVC and PS (Series c) engendered a relatively smaller reduction of compressive strength (fc reduction of 11%) and stifness (E reduction of 13% on average), while maintaining similar post-cracking strength and toughness (i.e. in terms of toughness index I5); 5. As benefit, PAG reduced thermal conductivity (k) by about 4.4% in concrete composites with similar air content, which can provide an interesting improvement for building isolation; 6. PAG exhibited a remarkable reduction of the water adsorption capacity of concrete, in spite of the increased porosity, perhaps due to their hydrophobic surface and electrostatic forces. This offers new potentials for impermeable concrete applications where durability is a concern; 7. Sampling plastic material over time generated significant variations in material composition and PAG properties. In general, time-related variations had a rather minimal impact on concrete performances (differences of about 3% in terms of E and fc), with the exception of a mix design (series c5) which was characterized by an unexpected high volume of entrapped air. This suggested that surface-cleaned post-consumer plastic waste could be systematically used as concrete aggregate replacement. Also, the use of an air-reducing agent will help stabilise the effect of variable entrapped air. Eventually, uncertainties in concrete properties can be further reduced by a careful monitoring of variations in plastic packaging types in the material stream;

8. A correlation study revealed that the PAG bulk density is the key factor behind compressive strength (fc), while the shape index had a strong influence on the concrete workability (slump). This explains why the PVC aggregates, which had the highest bulk density, generated mixes with significantly higher compressive strngth fc. On the other hand, particle size of PAG were found to have an appreciable effect only on concrete PCS, but almost no influence on the studied properties E, fc, k or water absorption; Upcycling of postconsumer plastics as aggregate replacement in concrete appears to be a feasible and promising endeavour for addressing today’s environmental issues and improving concrete mechanical resilience and thermal properties. Our next works will extend this study to other mechanical properties (e.g., flexural behaviour, drying shrinkage, etc.) and durability properties (e.g., freeze–thaw resistance, salt scaling, etc.), especially considering specific concrete applications. 5. Conflict of interest None. Acknowledgements and Funding sources The authors wish to thank Gaudreau Environnement Inc. and the National Science and Engineering Research Council (NSERC) CRDPJ 477126-14 for funding this research. The first Author also extends his gratitude to the Fonds Québécois de Recherche en Nature et Technologies (FQRNT) and RECYC-QUEBEC for scholarships. Appendix See Tables A1, A2, A3 and A4.

Table A1 Correlation table for Series a (including Reference). Pearson coefficients above 0.90 are in bold.

E (GPa) fc (MPa) W0 (Nm) I3 (Nm) I5 (Nm) Entrapped air (%) Slump (mm) PAG volume (%)

E (GPa)

fc (MPa)

W0 (Nm)

I3 (Nm)

I5 (Nm)

Entrapped air (%)

Slump (mm)

PAG volume (%)

– 0.99 0.97 0.76 0.61 0.98 0.55 0.99

– – 0.98 0.74 0.57 0.98 0.57 0.98

– – – 0.72 0.52 0.95 0.63 0.98

– – – – 0.95 0.83 0.07 0.83

– – – – – 0.70 0.34 0.67

– – – – – – 0.39 0.96

– – – – – – – 0.53

– – – – – – – –

Table A2 Correlation table for Series b. Pearson coefficients above 0.90 are in bold.

E (GPa) fc (MPa) We (Nm) I3 (Nm) I5 (Nm) Entrapped air (%) Slump PAG density PAG shape index PAG bulk density PAG % polymer

E (GPa)

fc (MPa)

We (Nm)

I3 (Nm)

I5 (Nm)

Entrapped air (%)

Slump (mm)

PAG density (g/cm3)

PAG shape index

PAG bulk density (g/cm3)

PAG % polymer

– 0.92 0.66 0.96 0.86 0.78 0.78 0.05 0.82 0.74 0.98

– – 0.90 0.92 0.87 0.59 0.49 0.43 0.55 0.92 0.92

– – – 0.70 0.69 0.29 0.08 0.76 0.14 0.93 0.68

– – – – 0.89 0.59 0.74 0.14 0.79 0.82 1.00

– – – – – 0.47 0.53 0.32 0.64 0.87 0.93

– – – – – – 0.72 0.29 0.70 0.24 0.79

– – – – – – – 0.56 0.99 0.24 0.75

– – – – – – – – 0.48 0.67 0.16

– – – – – – – – – 0.33 0.82

– – – – – – – – – – 0.85

– – – – – – – – – – –

571

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572 Table A3 Correlation table for Series c. Pearson coefficients above 0.90 are in bold.

E (GPa) fc (MPa) We (Nm) I3 (Nm) I5 (Nm) Entrapped air Slump PAG density PAG shape index PAG bulk density PAG % polymer

E (GPa)

fc (MPa)

We (Nm)

I3 (Nm)

I5 (Nm)

Entrapped air (%)

Slump (mm)

PAG density (g/cm3)

PAG shape index

PAG bulk density (g/cm3)

PAG % polymer

– 0.97 0.92 0.40 0.41 0.85 0.99 0.14 0.71 0.71 0.11

– – 0.95 0.42 0.48 0.83 0.93 0.14 0.82 0.81 0.33

– – – 0.12 0.72 0.95 0.92 0.45 0.77 0.82 0.34

– – – – 0.48 0.13 0.28 0.81 0.23 0.07 0.03

– – – – – 0.78 0.45 0.87 0.43 0.58 0.44

– – – – – – 0.90 0.64 0.66 0.75 0.20

– – – – – – – 0.26 0.66 0.67 0.02

– – – – – – – – 0.13 0.31 0.15

– – – – – – – – – 0.98 0.72

– – – – – – – – – – 0.72

– – – – – – – – – – –

Table A4 Correlation table for 6 mix designs selected for dry density, water absorption and thermal conductivity characterization. Pearson coefficients above 0.90 are in bold.

Thermal conductivity Water absorption (% mass) Dry density Entrapped air (%) Slump PAG volume (%) PAG density PAG shape index PAG bulk density

Thermal conductivity [W/(mK)]

Water absorption (% mass)

Dry density (g/cm3)

Entrapped air (%)

Slump (mm)

PAG volume (%)

PAG density (g/cm3)

PAG shape index

– 0.77 0.97 0.77 0.41 0.84 0.70 0.09 0.31

– – 0.71 0.59 0.29 0.69 0.16 0.09 0.19

– – – 0.88 0.18 0.72 0.70 0.12 0.27

– – – – 0.14 0.32 0.51 0.11 0.33

– – – – – 0.59 0.00 0.48 0.42

– – – – – – 0.43 0.52 0.14

– – – – – – – 0.08 0.63

– – – – – – – – 0.82

References [1] PlasticsEurope, Plastics – the Facts 2014/2015. Accessed March 4, 2017. Available at: http://www.plasticseurope.org/documents/document/ 20150227150049final_plastics_the_facts_2014_2015_260215.pdf [2] J.R. Jambeck, R. Geyer, C. Wilcox, T.R. Siegler, M. Perryman, A. Andrady, et al., Plastic waste inputs from land into the ocean, Science 347 (6223) (2015) 768– 771. [3] RECYC-QUEBEC, Bilan 2012 de la gestion des matières résiduelles au Québec (in French), ISBN 978-2-550-71080-6 (pdf), 2014. [4] Eco-Entreprises Quebec and RECYC-QUEBEC, Caractérisation des matières résiduelles du secteur résidentiel 2012-2013 (in French), ISBN 978-2-92395506-3 (pdf), 2015. [5] J. Hopewell, R. Dvorak, E. Kosior, Plastics recycling: challenges and opportunities, Philos. Trans. R. Soc. B Biol. Sci. 364 (1526) (2009) 2115–2126. [6] S.M. Al-Salem, P. Lettieri, J. Baeyens, Recycling and recovery routes of plastic solid waste (PSW): a review, Waste Manage. 29 (10) (2009) 2625– 2643. [7] F. Vilaplana, S. Karlsson, Quality Concepts for the Improved Use of Recycled Polymeric Materials: A Review, Macromol. Mater. Eng. 293 (4) (2008) 274–297. [8] B. Luijsterburg, H. Goossens, Assessment of plastic packaging waste: Material origin, methods, properties, Resour. Conserv. Recycl. 85 (2014) 88–97. [9] B. Hu, S. Serranti, N. Fraunholcz, F. Di Maio, G. Bonifazi, Recycling-oriented characterization of polyolefin packaging waste, Waste Manage. 33 (3) (2013) 574–584. [10] S. Serranti, A. Gargiulo, G. Bonifazi, Characterization of post-consumer polyolefin wastes by hyperspectral imaging for quality control in recycling processes, Waste Manage. 31 (11) (2011) 2217–2227. [11] Kosior, E., Dvorak, R., Large Scale HDPE Recycling Trial, Waste and Resources Action Program (WRAP), ISBN 1-84405-308-3, 2007. [12] E.M. Gartner, Potential improvements in cement sustainability, in: 31st Cement and Concrete Science Conference Novel Developments and Innovation in Cementitious Materials, 12–13 September. 2011. [13] R.J. Flatt, N. Roussel, C.R. Cheeseman, Concrete: An eco-material that needs to be improved, J. Eur. Ceram. Soc. 32 (11) (2012) 2787–2798. [14] M. Batayneh, I. Marie, I. Asi, Use of selected waste materials in concrete mixes, Waste Manage. 27 (12) (2007) 1870–1876. [15] P.H. Brunner, Urban Mining: A Contribution to Reindustrializing the City, J. Ind. Ecol. 15 (3) (2011) 339–341. [16] R. Sharma, P.P. Bansal, Use of different forms of waste plastic in concrete – a review, J. Cleaner Prod., 2016;112, Part 1:473–482. [17] N. Saikia, J. de Brito, Use of plastic waste as aggregate in cement mortar and concrete preparation: a review, Constr. Build. Mater. 34 (2012) 385– 401.

[18] R. Siddique, J. Khatib, I. Kaur, Use of recycled plastic in concrete: a review, Waste Manage. 28 (10) (2008) 1835–1852. [19] L. Ferreira, J. de Brito, N. Saikia, Influence of curing conditions on the mechanical performance of concrete containing recycled plastic aggregate, Constr. Build. Mater. 36 (2012) 196–204. [20] S.C. Kou, G. Lee, C.S. Poon, W.L. Lai, Properties of lightweight aggregate concrete prepared with PVC granules derived from scraped PVC pipes, Waste Manage. 29 (2) (2009) 621–628. [21] F. Iucolano, B. Liguori, D. Caputo, F. Colangelo, R. Cioffi, Recycled plastic aggregate in mortars composition: effect on physical and mechanical properties, Mater. Des. 52 (2013) 916–922. [22] J.R. Correia, J.S. Lima, J. de Brito, Post-fire mechanical performance of concrete made with selected plastic waste aggregates, Cem. Concr. Compos. 53 (2014) 187–199. [23] C. Albano, N. Camacho, M. Hernández, A. Matheus, A. Gutiérrez, Influence of content and particle size of waste pet bottles on concrete behavior at different w/c ratios, Waste Manage. 29 (10) (2009) 2707–2716. [24] K. Hannawi, S. Kamali-Bernard, W. Prince, Physical and mechanical properties of mortars containing PET and PC waste aggregates, Waste Manage. 30 (11) (2010) 2312–2320. [25] O.Y. Marzouk, R.M. Dheilly, M. Queneudec, Valorization of post-consumer waste plastic in cementitious concrete composites, Waste Manage. 27 (2) (2007) 310–318. [26] Y.-W. Choi, D.-J. Moon, J.-S. Chung, S.-K. Cho, Effects of waste PET bottles aggregate on the properties of concrete, Cem. Concr. Res. 35 (4) (2005) 776–781. [27] P. Mounanga, W. Gbongbon, P. Poullain, P. Turcry, Proportioning and characterization of lightweight concrete mixtures made with rigid polyurethane foam wastes, Cem. Concr. Compos. 30 (9) (2008) 806–814. [28] O.Y. Marzouk, R.M. Dheilly, L. Boubker, M. Queneudec, Valorization of plastic waste: thermal conductivity of concrete formulated with PET, in: 1st International conference on engineering for waste treatment (École de Mines d’ Albi-Carmaux, France), 2005. [29] G.W. Ehrenstein, G. Riedel, P. Trawiel, Thermal Analysis of Plastics: Theory and Practice, Carl Hanser Verlag GmbH, München, 2004. [30] J.E. Mark, Physical Properties of Polymers Handbook, Springer Science & Business Media, New York, 2007. [31] K.-H. Kim, S.-E. Jeon, J.-K. Kim, S. Yang, An experimental study on thermal conductivity of concrete, Cem. Concr. Res. 33 (3) (2003) 363–371. [32] R. Park, T. Paulay, Reinforced concrete structures, John Wiley & Sons, New York, 1975. [33] Z.Z. Ismail, E.A. AL-Hashmi, Use of waste plastic in concrete mixture as aggregate replacement, Waste Manage. 28 (11) (2008) 2041–2047. [34] R. Demirboga, A. Kan, Thermal conductivity and shrinkage properties of modified waste polystyrene aggregate concretes, Construct. Build. Mater. 35 (2012) 730–734.

572

C. Jacob-Vaillancourt, L. Sorelli / Construction and Building Materials 182 (2018) 561–572

[35] M. Elzafraney, P. Soroushian, M. Deru, Development of energy-efficient concrete buildings using recycled plastic aggregates, J. Archit. Eng. 11 (4) (2005) 122–130. [36] F. Vilaplana, A. Ribes-Greus, S. Karlsson, Analytical strategies for the quality assessment of recycled high-impact polystyrene: a combination of thermal analysis, vibrational spectroscopy, and chromatography, Analytica Chimica Acta. 604 (1) (2007) 18–28. [37] ASTM C136 / C136M-14, Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates, ASTM International, West Conshohocken, PA, 2014, www. astm.org. [38] CSA Group, CSA A23.1-14/A23.2-14: Concrete materials and methods of concrete construction, 2014. [39] A.M. Neville, Properties of concrete, 5th ed., Harlow, England; New York: Pearson, 2011. [40] ASTM D5550-14, Standard Test Method for Specific Gravity of Soil Solids by Gas Pycnometer, ASTM International, West Conshohocken, PA, 2014, www. astm.org.

[41] ASTM standard E1131-08, Standard test method for compositional analysis by thermogravimetry ASTM International, West Conshohocken, PA, 2014, www. astm.org. [42] C.L. Beyler, M.M. Hirschler, Thermal Decomposition of Polymers, SFPE Handbook of Fire Protection Engineering 2, Section 1, Chapter 7, pp. 111–131, 2002. [43] J. Côté, J.-M. Konrad, Thermal conductivity of base-course materials, Can. Geotechn. J. 42 (1) (2005) 61–78. [44] R.V. Silva, J. de Brito, N. Saikia, Influence of curing conditions on the durabilityrelated performance of concrete made with selected plastic waste aggregates, Cem. Concr. Compos. 35 (1) (2013) 23–31. [45] Material Selection Charts, Accessed March 5, 2017, Available at: http://wwwmaterials.eng.cam.ac.uk/mpsite/interactive_charts/default.html