Recycled crushed glass in road work applications

Waste Management 31 (2011) 2341–2351

Contents lists available at ScienceDirect

Waste Management journal homepage: www.elsevier.com/locate/wasman

Recycled crushed glass in road work applications M.M. Disfani a, A. Arulrajah a,⇑, M.W. Bo b, R. Hankour c a

Faculty of Engineering and Industrial Sciences (H38), Swinburne University of Technology, P.O. Box 218, Hawthorn, Melbourne, Vic. 3122, Australia DST Consulting Engineers Inc., 605 Hewitson Street, Thunder Bay, Ont., Canada P7B 5V5 c Geocomp Corporation, 125 Nagog Park, Acton, MA 01720, USA b

a r t i c l e

i n f o

Article history: Received 1 December 2010 Accepted 4 July 2011 Available online 30 July 2011 Keywords: Recycled glass Waste management Geotechnical experimentation

a b s t r a c t A comprehensive suite of geotechnical laboratory tests was undertaken on samples of recycled crushed glass produced in Victoria, Australia. Three types of recycled glass sources were tested being coarse, medium and fine sized glass. Laboratory testing results indicated that medium and fine sized recycled glass sources exhibit geotechnical behavior similar to natural aggregates. Coarse recycled glass was however found to be unsuitable for geotechnical engineering applications. Shear strength tests indicate that the fine and medium glass encompass shear strength parameters similar to that of natural sand and gravel mixtures comprising of angular particles. Environmental assessment tests indicated that the material meets the requirements of environmental protection authorities for fill material. The results were used to discuss potential usages of recycled glass as a construction material in geotechnical engineering applications particularly road works. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Waste material has been defined as any type of material byproduct of human and industrial activity that has no lasting value (Tam and Tam, 2006). The growing quantities and types of waste materials, shortage of landfill spaces, and lack of natural earth materials highlight the urgency of finding innovative ways of recycling and reusing waste material (Arulrajah et al., 2011). Additionally, recycling and subsequent reuse of waste materials can reduce the demand for natural resources, which can ultimately lead to a more sustainable environment. Recycled glass is a mixture of different colored glass particles and is often comprised of a wide range of debris (mainly paper, plastic, soil, metals, and food waste). The presence of different colored glass particles and diverse types of debris are the primary obstacles in reusing recycled glass in bottle production industries. Recycled glass particles are generally angular shaped and contain some flat and elongated particles. It is believed that the waste stream from which the glass particles have been produced controls the quality of the material, especially the amount of debris in the mixture (Landris, 2007). Furthermore the production process and the crushing procedure play the most important roles on maximum particle size, debris level and flakiness index of recycled glass which consequently influence other geotechnical characteristics

⇑ Corresponding author. Tel.: +43 613 92145741; fax: +43 613 92148264. E-mail addresses: [email protected] (M.M. Disfani), [email protected] (A. Arulrajah), [email protected] (M.W. Bo), [email protected] (R. Hankour). 0956-053X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.wasman.2011.07.003

(Landris, 2007). This causes variation in the geotechnical characteristics of recycled glass from one supplier to another. Geotechnical engineering applications of recycled glass include using it as backfill material in embankments, drainage blanket, filter media, and road pavement material (Wartman et al., 2004). Depending on the nature of the application of recycled glass, specific geotechnical parameters are of paramount importance. At the same time, certain factors affect the geotechnical characteristics of recycled glass. Insufficient knowledge on geotechnical characteristics of recycled glass is still the most important obstacle in its sustainable application in geotechnical engineering projects. Three different sample types of recycled glass obtained from recycling industries in Victoria were studied in this research work. The recycled glass types were named Fine Recycled Glass (FRG), Medium Recycled Glass (MRG) and Coarse Recycled Glass (CRG) based on their maximum particle size which is 4.75, 9.5 and 19 mm respectively. The main difference between these three samples is their gradation curve which influences other geotechnical properties. A comprehensive suite of laboratory tests were conducted on recycled glass samples to fill in the knowledge gap on the geotechnical characteristics of recycled glass in general and particularly on the one produced in Australia. The chemical and environmental tests were also executed to determine pH value, debris level, organic content and also to assess the Total Contamination (TC) level and the leachate concentration of recycled glass samples. 2. Sampling and general observations Approximately 850,000 tons of glass is used in Australia each year, with 350,000 tons recovered for recycling (Austroads, 2009).

2342

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

Nomenclature c Cc Cu wopt

epeak

cohesion (kPa) coefficient of curvature coefficient of uniformity optimum water content (%) axial strain at maximum deviator stress in CD triaxial test (%)

Recycled glass samples for this research work were collected from two recycling sites in the vicinity of Melbourne, Australia. Recycled glass in storage may become segregated by size. Additionally, segregation of material from contaminants (such as caps, labels, or miscellaneous debris) may occur (CWC, 1996). During sampling, ASTM practice for sampling aggregates was followed and all necessary precautions were taken to capture a sample containing representative particle sizes and representative amounts of all contaminants (ASTM, 2009). The CRG type sample was significantly coarser than the other two recycled glass samples and the initial visual inspection showed that it is mainly comprised of flat elongated gravel size particles and also presence of debris was noticeable. Wood, plastic caps, metal pieces, ceramic and gravel particles were some of the debris found in the CRG type sample. Initial visual inspection of the MRG sample showed that it mainly consists of gravel and sand size glass particles mixed with low percentage of fine material. It was clear that MRG sample has lower amount of gravel size particles and also an inferior percentage of flat and elongated particles compared to CRG sample. FRG sample, at first glance was also found to mainly comprise of sand size particles, and a lower percentage of coarse particles compared to MRG. Initial visual inspection of MRG and FRG samples advocate that they look similar to natural gravel and sand aggregates. 3. Soil classification and geotechnical properties Fine fraction of all recycled glass samples were identified as non plastic material and as such Atterberg limits results could not be obtained. Gradation curves of recycled glass samples are shown in Fig. 1. Based on the gradation curves and using the Australian standard for geotechnical site investigation the physical properties of recycled glass samples are presented in Table 1. Table 1 also shows the soil classification according to Australian Soil Classification System (ASCS), Unified Soil Classifications System (USCS) and AASHTO systems. The main difference between USCS and ASCS is the boundary between sand and gravel. While particles larger than 4.75 mm are considered gravel in USCS, in ASCS they are just required to be larger than 2.36 mm to be recognized as gravel (ASTM, 2010; Standards Australia, 1993). Table 2 presents the results of compaction tests along with other test results on as-received FRG, MRG and CRG samples. Table 2 indicates that all recycled glass samples possess specific gravity value of 2.5 except for FRG with specific gravity of 2.48. This is believed to be due to higher paper content of FRG. The values obtained for specific gravity of recycled glass in this research are comparable to the values reported in previous research works, including values of 2.48–2.49 reported by Wartman et al. (2004), 2.49–2.52 by FHWA (1998), 2.49 reported by CWC (1998), 2.45– 2.50 reported by Cosentino et al. (1995) and 2.51–2.52 reported by Ooi et al. (2008). Fig. 2 clearly shows that the smaller the size of crushed glass particles the more they are similar to the shape of natural aggregate. This means that larger glass particles are more flaky shaped and consequently more similar to their original shape before

cd,max rn /d /cd

r0c

maximum dry unit weight (kN/m3) normal stress applied in direct shear test (kPa) drained internal friction angle (°) internal friction angle obtained in CD triaxial test (°) effective confining pressure in CD triaxial test (kPa)

crushing. The lower flakiness index value of MRG reported in Table 2 compared to CRG implies that the quantity of flat and elongated particles (and also particles degree of angularity) depend on the degree of processing (i.e., crushing). Consequently, smaller particles, resulting from extra crushing, will exhibit somewhat less angularity and reduced quantities of flat and elongated particles (FHWA, 1998). This process seems similar to natural erosion process of granular aggregates. Debris levels were determined using American Geological Institute Data sheet 23.1 and 23.2 (CWC, 1998). Table 2 suggests that FRG and CRG samples have the lowest and highest debris level (by weight) respectively. This trend was noted to be opposite for the debris level obtained by visual method. Table 2 also indicates that the debris level determined by the weight method is less than one fifth, less than half and relatively equal to the value obtained by the visual method for FRG, MRG and CRG samples respectively. The primary reason for this is that a high percentage of debris in the FRG sample is comprised of very low density material, predominantly paper. For the MRG sample, the debris mainly consists of low density material such as wood, plastic and a lower amount of paper, while for CRG sample the debris consists more of metal, plastic, ceramic and gravel particles which have a higher density. The higher organic content value of the FRG sample compared to MRG and CRG samples and the fact that it is close to FRG debris level obtained by weight method is another evidence that the majority of FRG debris is paper which adds up to organic content value, while for CRG as the main debris are metal, ceramic, gravel and plastic they do not add up to its organic content. The pH values of the recycled glass samples were determined using the electrometric method. All recycled glass samples show a modest alkaline nature with the pH values ranging from 9.6 to 10.1. Fig. 3 presents compaction curves for FRG and MRG samples which are found to possess characteristic convex shaped curves similar to natural aggregates (Wartman et al., 2004). The curves were also found to be similar to the compaction curves of poorly graded sand. The increase in water content results in a decrease in the dry unit weight and a subsequent increase up to the optimum water content. Capillary tension in the pore water is the main reason for the decrease of dry unit weight at lower water contents (Das, 2008). The main reason that both FRG and MRG samples showed the behavior of poorly graded sand in the compaction tests despite the fact that they have been classified as well graded mixtures, would be the poor ability of the glass particles in holding and absorbing water. The low sensitivity of FRG to water content changes in comparison to natural aggregate is evident from the flatter compaction curves of this sample which enables it to have stable compaction behavior and good workability over a wide range of water contents in geotechnical engineering applications (Wartman et al., 2004). Fig. 3 also indicates that no specific curve similar to those of natural aggregate could be developed for CRG sample. This is due to the high percentage of gravel size glass particles (96.4%), negligible amounts of fine particles (0.9%) and also the nature of glass particles which results in poor capability of absorbing and holding water. During compaction test on CRG sample it was noticed that the water was not mixing with glass

2343

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

100 FRG

90

FRG After Standard Compaction

80 70

MRG MRG After Standard Compaction

60

MRG After Modified Compaction

50

CRG

40

CRG After Standard Compaction

30

CRG After Modified Compaction

Passing Percentage

FRG After Modified Compaction

20 10 0.001

0.01

0.1

1

10

0 100

Particle Size (mm) Silt

Clay

Fine

Medium

Fine Particles

Coarse

Gravel

Sand

Cobbles

According to AS 1726-1993

Fig. 1. Gradation curves of as-received and post compaction recycled glass samples.

Table 1 Classification characteristics of as-received recycled glass samples. Sample

FRG MRG CRG a b c

Classification ASCS

USCS

AASHTO

SW-SM GW-GM GP

SW-SM SW-SM GP

A-1-b A-1-a A-1-a

Cu

Cc

Fine contenta

Gravel contentb

Sand contentc

7.6 16.3 2.6

1.3 2.2 1.2

5.4 5.2 0.9

9.2 53 96.4

85.4 41.8 2.7

<0.075 mm (%). 2.36 mm <(%). 0.075–2.36 mm (%).

Table 2 Geotechnical properties of recycled glass samples.

a

Test

FRG

MRG

CRG

Specific gravity Flakiness index Debris level (visual method) (%) Debris level (weight method) (%) Organic content (%) pH value Standard proctor cd,max (kN/m3) wopt (%) Modified proctor cd,max (kN/m3) wopt (%) LA abrasion value (%)

2.48 N.Aa 7 1.23 1.3 9.9

2.5 85.4 5 2.01 0.5 10.1

2.5 94.7 3 2.98 0.23 9.6

16.7 12.5

18 9

N.A N.A

17.5 10 24.8

19.5 8.8 25.4

N.A N.A 27.7

Not Applicable.

particles and also was draining out from the bottom of the compaction mold. Table 2 shows that the values of maximum dry densities obtained for FRG and MRG samples are 10–15% lower than the values generally found for the natural aggregate within the same soil classification (Craig, 1992) which is probably the result of lower specific gravity of recycled glass compared to natural aggregates. The results of compaction tests on FRG and MRG samples are close to the values reported for recycled glass samples in previous research works, including maximum dry densities of 16.6–16.8 kN/ m3 and 17.5–18.3 kN/m3 reported by Wartman et al. (2004) for standard and modified compaction energy respectively. Pennsylvania Department of Transportation (2001) also reported maximum

dry densities of 16.9–17.6 kN/m3 and 17.6–18.4 kN/m3 for recycled glass sample classified as SW. Particles crushing and degradation can be a significant issue in certain geotechnical applications; as a result any attempt to utilize recycled material for geotechnical applications should examine this issue carefully (Sivakumar et al., 2004). To assess the durability and the abrasion resistance of recycled glass; two different methods were used; Los Angeles abrasion test and post-compaction sieve analysis. The results of LA abrasion tests presented in Table 2 indicate that FRG and MRG samples have similar LA abrasion values to that of crushed rock (24%) and lower than that of recycled crushed concrete (28%) reported by Arulrajah et al. (2009). The higher LA abrasion of CRG sample as compared to FRG and MRG samples is believed to be the result of higher debris level of CRG. The LA abrasion values reported in Table 2 are similar to the values found in previous research works including 24–25% obtained by Wartman et al. (2004) and 27–33% obtained by Ooi et al. (2008). FHWA (1998) determined higher LA values of 30–42% for the recycled glass samples they studied and also CWC (1998) obtained LA abrasion values of 29.9–41.7%. This difference is believed to be the result of discrepancy in debris level of different samples studied. The second approach to assess the durability of recycled glass samples was the post-compaction sieve analysis. Fig. 1 shows gradation curves of all three recycled glass samples in as-received and post compaction (for both standard and modified compaction tests) situations. Fig. 1 indicates that there is a noticeable change in particle size distribution of CRG sample after compaction tests. The gradation curve of CRG sample has moved toward the finer (left) end of the graph after compaction tests. This effect is lower

2344

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

Fig. 2. Change of glass particles shapes by size for CRG sample.

Fig. 3. Compaction test results for recycled glass samples.

for standard compaction tests due to the reduced number of blows and reduced imposed energy in each blow. This clear change in the gradation curve of CRG would affect all other geotechnical characteristics including shear strength. Analysis of gradation curves showed that for CRG sample the sand content of the as-received sample increased from 2.7% to 9.1% (while gravel content decreased from 96.4% to 88.8%) after standard compaction and to 24.8% (with the gravel content percentage decreased down to 71.9%) after modified compaction. This change is explained by the gravel size glass particles crushing and decreasing in size to sand size particles under compaction. On the other hand, for the

MRG sample, with 53% gravel content, the effect of compaction on changing the gradation curve is much lower. For the FRG sample, with 9.2% gravel content, no crushing can be identified after the standard compaction test. Segregation is defined as particle size separation process that results when a nominally homogeneous mixture of soil particles is spread using mechanical action (Sutherland and Grabinsky, 2003). Segregation can cause a homogeneous mixture to be divided into two different parts with completely different geotechnical characteristics from the primary homogeneous mixture. CRG sample was found to be vulnerable to segregation as a result of its poor

2345

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

particle size distribution, low friction resistance due to smooth surface of gravel size glass particles, and also poor ability in absorbing and holding moisture. 4. Hydraulic conductivity of FRG and MRG In road work applications, the hydraulic conductivity of a fill material often plays an important role in material selection. Table 3 shows results of hydraulic conductivity tests performed on FRG and MRG samples. Using Kenny et al. (1984) graphs for calculating hydraulic conductivity of granular materials based on their gradation curves, the hydraulic conductivity values of FRG and MRG samples were calculated and are shown in Table 3 which exhibit good agreement with the values obtained from permeability tests. This proves that the FRG and MRG samples have the hydraulic conductivity values similar to the values of natural aggregates (sand and gravel mixtures). The hydraulic conductivities of both FRG and MRG samples are classified as medium according to permeability classifications (Terzaghi et al., 1996). The results of this research are in agreement with previous findings in that recycled glass is more free-draining than most of other natural aggregates (CWC, 1998; Wartman et al., 1994). 5. Shear strength of FRG and MRG Shear strength parameters provide a good basis to predict the behavior of aggregate materials under the effect of imposed static or dynamic loads where the aggregates act as a load supporting medium (CWC, 1998). The most common tests to determine the shear strength parameters of the soils and other aggregate materials are the California Bearing Ratio (CBR), direct shear and triaxial shear tests. Table 4 presents the results of shear strength parameters obtained for FRG and MRG samples. 5.1. California bearing ratio Two sets of specimens were prepared by applying both standard and modified compaction efforts on FRG and MRG material mixed with optimum water content obtained in compaction tests. A surcharge mass of 4.5 kg was placed on the surface of the compacted specimens and then the samples were soaked in water for a period of four days. This is to simulate the confining effect of overlying pavement layers and also the likely worst case in-service scenario for a pavement (VicRoads, 1998). As shown in Table 4, CBR values of the FRG sample are found to be considerably lower than those of the MRG sample. This trend seems to be related to higher values of maximum dry unit weight obtained for MRG sample in compaction tests. The higher maximum dry unit weight for MRG (considering that its specific gravity is approximately equal to FRG) is an indication of better compaction which results in better particles contact and eventually better shear performance of the MRG sample. Wartman et al. (2004) and Ooi et al. (2008) reported CBR values of 47–48% and 75–80% for the recycled glass samples they studied which is comparable to values

Table 4 Shear strength parameters of FRG and MRG. Test CBR (%) Using standard compaction effort Using modified compaction effort Direct shear test /d (°) rn (30–120 kPa) rn (60–240 kPa) rn (120–480 kPa) Triaxial shear test (CD) /cd (°) r0c (30–120 kPa) r0c (60–240 kPa) r0c (120–480 kPa)

FRG

MRG

18–21 42–46

31–32 73–76

45–47° 42–43° 40–41°

52–53° 50–51° –

40° 38° 35°

42° 41° 41°

obtained for FRG samples compacted with modified compaction energy in this research. 5.2. Direct shear test The specimens for direct shear tests were compacted in 3 layers inside the shear box using a rubber-tipped tamping rod following wet compaction method (partially saturated). A 10 cm square shear box with the depth of 5 cm was used to test FRG samples and a 30 cm large square shear box with the depth of 20 cm was used for MRG samples. Generally the degree of compaction to be applied during preparation of a soil specimen for direct shear test depends on the relevant field compaction control requirement (BSI, 1990). All test specimens reached 98% maximum dry unit weight obtained in standard compaction tests. It has been recommended that the normal stress applied to the specimens in a set of direct shear tests should generally bracket the maximum stress likely to occur in the ground (Head, 1994). Three successive normal stress ranges of 30–120 kPa, 60–240 kPa and 120–480 kPa which correspond to shallow to moderate overburden pressure (Wartman et al., 2004), were chosen. Two sets of sample were prepared for each stress level which brings the total number of tests to 10 and 8 for FRG and MRG sample types respectively. Samples were allowed to consolidate under the applied normal stress for few hours under the saturated condition. Afterward a displacement controlled shear phase at a rate of 0.5–1 mm/min was applied. It took 3–6 min for FRG samples and 8–17 min for MRG samples to reach failure/peak shear stress. Shearing was continued for sufficient distance after peak shear stress to get the residual shear stress values as well. Fig. 4a shows shear stress versus horizontal deformation and suggests that the same pattern for all normal stress levels exist in which shear stress increases to a peak value and then levels off. With the increase of normal stress; the increase of the shear stress to the peak value gets sharper and the difference between the residual shear stress and peak shear stress becomes larger. Fig. 4b shows vertical deformation versus horizontal deformation of FRG samples which is similar to typical behavior of dense sand. The contraction effect at the beginning of the test becomes more

Table 3 Hydraulic conductivity of FRG and MRG. Sample

MRG FRG a b

Head (1994). Terzaghi et al. (1996).

Hydraulic conductivity (m/s) Test results

Kenny et al. (1984)

2.85  105 1.7  105

2.5  105 2  105

Drainage characteristicsa

Permeability classificationb

Good Good

Medium Medium

2346

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

(a)

Vertical deformation (mm)

Shear stress (kPa)

400

σn = 480 kPa

300 σn = 240 kPa

200

σn = 120 kPa

100

σn = 60 kPa σn = 30 kPa

0 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -1.6

Contraction Dilation σn = 480 kPa σn = 240 kPa σn = 120 kPa

0

σn = 60 kPa

σn = 30 kPa

(b) 2

4

6

8

10

12

14

Horizontal deformation (mm) Fig. 4. Direct shear test results for FRG sample.

noticeable with the increase in the normal stress level. It is also noticeable that with the increase in normal stress level, the dilation behavior gets less significant. Sharper rise to peak shear stress, larger difference between peak and ultimate shear stress, bigger contraction effect in the beginning of the shearing and lower dilation effect for the samples with higher normal stress level, indicate that as the normal stress increases, the behavior of samples turn out to be more similar to that of dense sands. Considering that all test specimens reached the same compaction degree before the test, this is the result of the increase in particle to particle contacts during the consolidation time. Linear Mohr–Coulomb envelopes drawn for FRG and MRG samples showed a small apparent cohesion. Considering the sandy nature of recycled glass particles and the fact that both FRG and MRG samples were in moist condition, this small cohesion is believed to mainly be the result of surface tension of the water between the grains (Head, 1994). Based on this interpretation, the Mohr–Coulomb envelopes were redrawn with zero cohesion intercept. While most of the available research studies on recycled glass samples adopted a linear Mohr–Coulomb failure envelope, Wartman et al. (2004), fitted a nonlinear failure envelope to the direct shear test results fairly well. To illustrate the nonlinearities in the results obtained in this research, the failure envelopes were represented using a best-fit second order polynomial in lieu of the typical linear Mohr–Coulomb envelope. Fig. 5 shows the peak shear and normal stress values for FRG and MRG samples and the Mohr–Coulomb envelopes. The coefficients of variation (R2) are extremely close to 1 which indicates that the regressions fit the data well even for assumed cohesion intercepts of zero (Wartman et al., 2004). Values of internal friction angle reported in Table 4 suggest that for both FRG and MRG samples, with increasing the normal stress level from 30 to 480 kPa the internal friction angle decreases. Table 4 also indicates that internal friction angle of MRG sample is 10–15% higher than those of FRG sample. This is in agreement with higher CBR values of MRG sample and believed to be the result of better particles contact in compacted MRG sample. Comparing internal friction angles of FRG and MRG samples to those of natural aggregates, it is clear that the recycled glass samples show internal friction angles similar to those of well graded sand, well graded

gravel or sand and gravel mixtures in dense situation with angular shape particles (Lambe and Whitman, 1969). The internal friction angles reported in Table 4 are similar to the values found in previous research works including 47–63° obtained by Wartman et al. (2004), 51–53° obtained by FHWA (1998) and 49–53° reported by CWC (1998). 5.3. Triaxial shear test A set of CD triaxial compression shear tests was performed on the compacted FRG and MRG samples. The good drainage characteristic and medium permeability classification obtained for FRG and MRG suggest that CD method is more appropriate for these samples. The tests were performed at five consecutive effective confining pressure levels of 30, 60, 120, 240 and 480 kPa. The dimensions of the samples prepared were 50  100 mm (diameter  height) for FRG sample and 70  150 mm for MRG sample. All test specimens were compacted inside a split mold in five layers using a rubber-tipped tamping rod to at least 90% of maximum dry unit weight and within ±5% difference with the optimum water content obtained in standard compaction tests. There is a high possibility of puncturing the membrane by sharp angular glass particles during placement and compacting. To avoid this, the desired unit weight was kept around 90% and also the outside of the first membrane was covered with a thin layer of silicone grease and a second membrane was placed on top of the first one. The stiffening effect of the two membranes was counted by applying a new correction factor to the results. All test specimens achieved B (Skempton’s pore pressure coefficient) value of at least 0.95 before starting a 24 h volume controlled consolidation phase followed by a straincontrolled shear phase. The strain rate has to be slow enough so no excess pore water pressure develops during the drained shear phase and were selected based on consolidation results. The loading rates of 0.02 mm/min for MRG and 0.01 mm/min for FRG were applied during shear phase. Fig. 6a shows the deviator stress versus axial strain for MRG samples and suggests that for all confining stress levels, the deviator stress gradually reaches a peak value and then levels off to a slightly lower residual value. This shear behavior is typical of dense

2347

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

Fig. 5. Mohr–Coulomb envelopes for FRG and MRG samples.

Deviator stress (kPa)

2000 (a)

1600

σ′c = 480 kPa

1200

σ′c = 240 kPa

800 σ′c = 120 kPa

400

σ′c = 60 kPa

σ′c = 30 kPa

0 6

σ′c = 480 kPa

Volumetric strain (%)

(b)

4 σ′c = 240 kPa

2 0 σ′c = 60 kPa

σ′c = 120 kPa

-2 -4

Contraction Dilation

σ′c = 30 kPa

-6 0

5

10

15

20

25

30

35

Axial strain (%) Fig. 6. CD triaxial shear test results for MRG sample.

sands, which lose strength when strained beyond peak strength (Cosentino et al., 1995) but the graduate increase of deviator stress to reach peak stress and then negligible slow fall to residual stress is more similar to the behavior of loose sand. The volumetric strain-axial strain curves of MRG samples shown in Fig. 6b suggest that for the samples tested with r0c > 120 kPa, contraction behavior happens which is consistent with the behavior of loose sands. Conversely for the samples tested with r0c 6 120 kPa, dilation behavior starts after an initial contraction behavior implying that initially loose sand (recycled glass in this case) behaves like dense sand (Holtz and Kovacs, 1981). The same trend of change for volumetric strain against axial strain was observed for FRG. Fig. 7 presents all five Mohr circles obtained for MRG sample and the Mohr–Coulomb envelope drawn for the three largest circles. Shear strength parameters of both FRG and MRG samples were obtained by drawing linear Mohr–Coulomb envelopes tangent to three consecutive Mohr circles and are presented in Table 4.

The drained internal friction angles obtained for FRG and MRG via triaxial shear tests are 10–15% and 15–20% respectively lower than the internal friction angles obtained via direct shear test. This seems to be the result of the differences in boundary conditions between direct shear and triaxial shear tests (Wartman et al., 2004) and also lower dry unit weight values reached for triaxial specimens. Considering that the triaxial test conditions best replicate most field situations, it is better to adopt these values for design or any practical applications of recycled glass material (Pennsylvania Department of Transportation, 2001). The cohesion intercepts shown in Table 5 for FRG and MRG samples are the combined result of a linear representation of a nonlinear shear strength failure envelope, and apparent cohesion resulting from adhesive label glues, food remaining, and other substances that exist in recycled glass samples (Wartman et al., 2004). Table 5 also presents the strain levels attributed to the peak deviator stresses (failure stress) which suggest that the increase in the

2348

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

Fig. 7. Mohr circles and Mohr–Coulomb failure envelop for MRG sample.

5.5. Resilient modulus and permanent strain

Table 5 Parameters obtained via triaxial shear tests on FRG and MRG. Parameter c (kPa) r0c (30–120 kPa) r0c (60–240 kPa) r0c (120–480 kPa) epeak (%) r0c : 30 kPa r0c : 60 kPa r0c : 120 kPa r0c : 240 kPa r0c : 480 kPa Modulus of elasticity (kPa) r0c : 30 kPa r0c : 60 kPa r0c : 120 kPa r0c : 240 kPa r0c : 480 kPa

FRG

MRG

0 1 24

2.2 7.1 4.7

6.65 6.75 11.7 12.3 19.4

7.95 10.1 12.9 15.5 18.5

3300 6000 8000 14,000 17,000

4500 7000 10,000 15,000 21,000

confining effective stress results in a constant increase in the strain percentage attributed to the failure stress for both FRG and MRG. CWC (1998) reported values of 42–46° for drained internal friction angle of recycled glass samples which are close to the values obtained in this research, while the 47–48° values found by Wartman et al. (2004) and 44–47° obtained by FHWA (1998) are slightly higher than those obtained in this research.

5.4. Modulus of elasticity To be able to use the concepts and formulas from the theory of elasticity to obtain the modulus of elasticity, the stress–strain curves of recycled glass samples must be linearized. Therefore, E is not constant, but rather is a quantity which approximately describes the behavior of a soil for a particular range of stresses (Lambe and Whitman, 1969). The elastic modulus derived from triaxial testing is often called Young’s modulus and the value usually quoted for soils is the secant modulus from zero deviator stress to a deviator stress equal to about 0.33–0.5 (0.33 was used in this research) of the peak deviator stress (Lambe and Whitman, 1969). Values reported in Table 5 suggest that the stiffness of both FRG and MRG samples decrease with increasing confining effective stress.

While static load testing of granular materials does not simulate the repetitive vehicular loading that occurs in pavements, the resilient modulus is the engineering property that models the behavior of material under such repetitive loading (Senadheera et al., 2005). The resilient modulus of an aggregate is determined through Repeated Load Triaxial (RLT) test (CWC, 1998) and is highly dependent on parameters such as aggregate mineralogy, particle characteristics, density, moisture content and particle size distribution (Senadheera et al., 2005). Due to the lack of cohesion between glass particles it was not possible to carry out RLT test on pure FRG and MRG sample types. Test specimens failed either prior to starting the test or after just few load cycles were applied. Resilient modulus of FRG in blends with crushed rock (FRG ratio was limited to 30% in blends) has been studied by Ali et al. (2011). The RLT results suggested that the performance of the blends in terms of permanent deformation and the resilient modulus is comparable to those of natural granular subbase aggregate (Ali et al., 2011). 6. Environmental assessment In consideration of using recycled glass in road work applications the environmental effects and any hazard that may be caused by using recycled glass needs to be investigated. A soil/recycled material blend will be recognized as fill material if the site assessment demonstrates that it is not contaminated or contamination in terms of TC is not higher than the values reported in Table 6 for fill material (EPA Victoria, 2007a). If the amount of TC in any contaminated soil or waste/recycled material is higher than those required for fill material; but both TC and leachable concentrations were lower than the values specified for category C in Table 6, then the soil/recycled material would be categorized as waste material/contaminates soil of category C (EPA Victoria, 2007a). Category C refers to prescribed industrial wastes which pose a low hazard or only exhibit any offensive aesthetic properties and which requires control and/or ongoing management to protect human health and the environment (EPA Victoria, 2005). In determining the leachate concentrations for recycled glass samples, the Australian Standard Leaching Protocol (ASLP) was followed and the results from two buffer solutions (one slightly acidic and the other slightly alkaline) were compared against the ASLP criteria (EPA Victoria, 2007b). In this research slightly acidic (pH

2349

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351 Table 6 TC and ASLP allowed in soil/waste for fill material and category C of waste material/contaminated soil (EPA Victoria, 2007a). Contaminant

Fill material

Category C

TC (mg/kg) dry weight

TC (mg/kg) dry weight

ASLP (mg/L)

Arsenic Cadmium Chromium (VI) Copper Lead Mercury Molybdenum Nickel Tin Selenium Silver Zinc Cyanide Fluoride Phenols (halogenated) Phenols (non-halogenated) Monocyclic aromatic Hydrocarbons Benzene Polycyclic aromatic hydrocarbons Benzo(a) pyrene Total petroleum hydrocarbons C6 to C9 Total petroleum hydrocarbons C9 to C36 Polychlorinated biphenyls Chlorinated hydrocarbons Hexachlorobutadiene Vinyl chloride Other chlorinated hydrocarbons Organochlorine pesticides Aldrin + Dieldrin DDT + DDD + DDE Chlordane Heptachlor Other organochlorine pesticides

20 3 1 100 300 1 40 60 50 10 10 200 50 450 1 60

500 100 500 5000 1500 75 1000 3000 500 50 180 35,000 2500 10,000 10 560

0.7 0.2 5 200 1 0.1 5 2 – 1 10 300 8 150 2 14

7 1 20 1 100 1000 2 1

70 4 100 5 650 10,000 –

– 0.1 – 0.001 – – –

2.8 1.2 10

0.07 0.03 –

1.2 50 4 1.2 10

0.03 2.0 0.1 0.03 –

1

5; which resembles rain water) and alkaline (borate) leaching fluids (pH 9.2) were prepared according to AS 4439.3-1997 (Standards Australia, 1997). FRG and MRG samples were examined for TC and ASLP (with both leaching fluids) and the results are presented in Table 7. Comparison of the TC values of FRG and MRG samples with maximum TC values allowed for fill material presented in Table 6, implies that except for chromium all other contaminant levels are far below the allowable limits. The chromium metal is found in a few oxidation states such as hexavalent chromium (chromium VI) and trivalent chromium (chromium III).

The values reported for FRG and MRG are the total chromium (chromium III + chromium VI) while the EPA Victoria requirement is on hexavalent chromium (chromium VI), so FRG and MRG will go beyond the chromium boundary only and only if all the chromium found in the test is of type chromium VI which does not seem to be the case for FRG and MRG. ASLP values of FRG and MRG samples presented in Table 7 were compared with ASLP requirement of category C shown in Table 6 and suggest that the ASLP values of FRG and MRG samples are far below the ASLP thresholds designated for category C of waste/contaminated soil.

Table 7 TC and ASLP values obtained for FRG and MRG. Contaminant

Arsenic Cadmium Chromium Copper Lead Mercury Nickel Selenium Silver Zinc Cyanide Monocyclic aromatic hydrocarbons Benzene Polycyclic aromatic hydrocarbons Benzo(a) pyrene a b

mg/kg of dry weight. mg/L.

FRG

MRG

TCa

ASLPb (Acet)

ASLPb (Borate)

TCa

ASLPb (Acet)

ASLPb (Borate)

<5 0.5 <5 6 12 <0.05 <5 <5 <5 34 <5 <0.1 <0.1 <0.1 <0.1

<0.01 0.004 <0.01 0.12 0.19 <0.001 <0.01 <0.01 <0.01 0.79 <0.05 <0.001 <0.001 <0.001 <0.001

<0.1 <0.02 <0.1 <0.1 <0.1 <0.01 <0.1 <0.1 <0.1 <0.1 <0.05 <0.001 <0.001 <0.001 <0.001

<5 <0.2 11 6 72 <0.05 <5 <5 7 70 <0.5 <0.1 <0.1 <0.1 <0.1

<0.01 0.004 0.01 0.06 0.4 <0.001 0.01 <0.01 <0.01 1.6 <0.05 <0.001 <0.001 <0.001 <0.001

<0.1 <0.02 <0.1 <0.1 <0.1 <0.01 <0.1 <0.1 <0.1 <0.1 <0.05 <0.001 <0.001 <0.001 <0.001

2350

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

Results of Toxicity Characteristic Leaching Procedure (TCLP) and Synthetic Precipitation Leaching Procedure (SPLP) tests conducted by Wartman et al. (2004) on samples of recycled glass also showed that the heavy metals concentration in recycled glass are far below that of the hazardous waste material (Wartman et al., 2004).

7. Geotechnical and road work applications The results of the laboratory experimentation conducted on samples of recycled crushed glass produced in Victoria provide good evidence that recycled glass (in the appropriate grading) has suitable geotechnical engineering characteristics to replace natural aggregates in a wide range of applications. For each application specific characteristics are of paramount importance and should be studied accurately. Recycled glass characteristics indicate that it is suitable to be used as a fill (backfill) material in structural and non-structural applications. Fills which support heavy static loads such as pier footings, fills placed under pedestrian sidewalks and also fills under bicycle trails are examples of such applications (MPSC, 2000). Recycled glass by itself or in a blend with natural or recycled aggregate has the potential to be used as retaining wall backfill materials. The lower density of recycled glass compared to natural aggregate decreases the pressure exerted to the back of retaining walls which can lead to a more economical design for the retaining walls. CWC (1998) suggests that up to 30% recycled glass can be used for stationary load backfills, while this percentage is limited to 15% for fluctuating loads. Up to 100% recycled glass can be used for non-structural backfills according to CWC (1998). Superior permeability characteristics of recycled glass advocate its usage as a drainage media in applications such as drainage media behind the retaining walls, drainage blanket, French drains and footing drains (Wartman et al., 2004; CWC, 1998). Using recycled glass for drainage purpose behind the retaining walls will reduce the risk of clogging of the drainage media and the higher permeability of recycled glass fastens the drainage time of water accumulated at the back of the wall (Wartman et al., 2004). In the event of using recycled glass for drainage purposes, the hydraulic conductivity tests and specifically the environmental assessment of the leachate are particularly important. CWC (1998) recommends use of up to 100% recycled glass for drainage purposes. Recycled glass shows the potential to be used as a filling media around the utilities or as a trench bedding media (CWC, 1998). A trial example of the use of recycled glass as a pipe embankment material was conducted in NSW, Australia (Austroads, 2009). CWC (1998) suggests using up to 100% recycled glass as a fill material around utilities or as a trench bedding although the allowable percentage is lower when the recycled glass is subjected to surcharge loads such as loads applied by a roadway or slab. Recycled glass on its own or in a mixture with natural or recycled aggregates (such as crushed rock and crushed concrete) can be used in a range of road work applications including subbase, embankments material and drainage media in roads. As of the year 2000, no data was available for use of glass in Australian roads (Austroads, 2000). The use of recycled glass in Australian roads is still limited and rare as typically only 3–5% reclaimed glass in the form of cullet is permitted in granular products (Austroads, 2009). This is believed to be the result of lack of knowledge on geotechnical engineering characteristics of recycled glass (especially its response under repeated loads) and also concerns on environmental suitability of using recycled glass in road works. Compared to Australia, several states departments of transportations in USA have published regulations and specification which allow and facilitate using recycled crushed glass in different road work applications as an alternative to natural aggregates (CWC, 1998). To

improve the shear strength performance of recycled glass in case of its application in subbase layers of roads, measures such as mixing it with natural aggregate and also stabilizing the material with additives such as lime, cement or flyash can be investigated. A footpath trial (asphalt shared path) was recently constructed in the state of Victoria. The asphalt shared path comprised a crushed rock base of nominal 100 mm thickness built in three sections with varying FRG percentage of 15% and 30% (Vuong and Arulrajah, 2010). The preliminary field results indicated that adding FRG can improve workability of the crushed rock base in terms of achieving a higher density ratio, but may reduce the material strength. Adding FRG (in ratios limited to 15%) to the crushed rock would produce a blend which is easy in workability and still produce high base strength as compared to the parent crushed rock base (Vuong and Arulrajah, 2010). In spite of all these potential applications, using recycled glass in geotechnical engineering applications has been overlooked in many countries including Australia. Easy access to virgin aggregate, vast areas available to dump the waste material, negative public perception on environmental effects of using recycled material and lack of knowledge on engineering characteristics of recycled glass are the major reasons responsible for this stance.

8. Conclusion Three different samples of recycled glass with different gradation curves produced from residential and industrial waste glass streams in Victoria were studied in this research to investigate their usage as a construction material in geotechnical applications. The Fine Recycled Glass (FRG) and Medium recycled Glass (MRG) were classified SW-SM while Coarse Recycled Glass (CRG) was classified GP according to USCS. Recycled glass specific gravity values were found to be approximately 10% lower than the values attributed to natural aggregate reported by Das (2007). Results of standard and modified proctor compaction tests showed a higher maximum dry unit weight for MRG sample compared to FRG sample while the optimum water content of MRG was found to be lower than FRG sample. LA abrasion tests proved that the abrasion resistance of FRG and MRG samples is close to those of construction and demolition material, whereas CRG showed a higher LA abrasion value. Post compaction gradation curve analysis of FRG and MRG samples proves their stability during engineering operations including handling, spreading and especially compacting. The CRG source consisted of a sizeable amount of elongated and flat shaped particles and high debris content. It was also found that the CRG source possesses little ability to absorb and hold moisture which impacts on its compaction behavior. These characteristics along with perceptible change in gradation curves of the CRG samples after compaction, and its high segregation potential led the authors to conclude that CRG source is not an ideal material for geotechnical engineering applications. On the other hand FRG and MRG samples proved appropriate characteristics close to those of natural aggregate within the same soil classification. CBR tests indicate the superior shear resistance of MRG as compared to FRG. Direct shear test results indicate that the internal friction angle of MRG is slightly higher than those of FRG. The internal friction angle values obtained for FRG and MRG samples through direct shear test are comparable to those of natural sand and gravel mixtures with angular particles. CD triaxial shear test results confirmed the findings of direct shear tests. By and large FRG and MRG samples showed the geotechnical engineering behavior of natural well graded sand and gravel mixtures. Hydraulic conductivity tests showed that FRG and MRG samples have medium permeability with good drainage characteristics. TC and ASLP assessments proved that FRG and MRG samples comply

M.M. Disfani et al. / Waste Management 31 (2011) 2341–2351

with requirements issued by EPA Victoria for using aggregates as fill material. Results of this study advocate using recycled glass in a range of geotechnical engineering applications. Further research is under progress to study the behavior of blends of FRG used in footpath and road subbase layers. Acknowledgements The authors would like to thank Alec Papanicolaou (the Geomechanics laboratory technician in Swinburne University) for his technical support during the experimental works. References Ali, M.M.Y., Arulrajah, A., Disfani, M.M., Piratheepan, J., 2011. Suitability of using recycled glass – crushed rock blends for pavement subbase applications. Proceedings of Geo-Frontiers 2011: Advances in Geotechnical Engineering, 13– 16 March, Dallas, 1325–1334. Arulrajah, A., Piratheepan, J., Aatheesan, T., Bo, M.W., 2011. Geotechnical properties of recycled crushed brick in pavement applications. J. Mater. Civil Eng. doi:10.1061/(ASCE)MT.1943-5533.0000319. Arulrajah, A., Vuong, B., Wilson, J., 2009. Laboratory testing of reclaimed demolition materials for footpaths and shared paths. Swinburne University of Technology Report for Municipal Association of Victoria, Victoria. ASTM, 2009. Standard practice for sampling aggregate. American Society for Testing and Materials, ASTM D75/D75M-09, Pennsylvania. ASTM, 2010. Standard practice for classification of soils for engineering purposes (Unified Soil Classification System). American Society for Testing and Materials, ASTM D2487-10, Pennsylvania. Austroads, 2000. Use of recycled materials and the management of roadside vegetation on low trafficked roads. Publication No. AP-R154/00, New South Wales. Austroads, 2009. Guide to pavement technology. Part 4E: recycled materials. Publication No. AGPT04E/09, New South Wales. BSI, 1990. Methods of test for soils for civil engineering purposes. Part 7: Shear strength tests (total stress). British Standards Institution, BS 1377–7:1990, London. Cosentino, P.J., Kalajian, E.H., Shieh, C.S., Heck, H.H., 1995. Developing specifications for waste glass and waste-to-energy bottom ash as highway fill materials volume 2 of 2 (waste glass). Florida Institute of Technology Report, Florida Department of Transportation; Report No. FL/DOT/RMC/06650-7754, Florida. Craig, R.F., 1992. Soil Mechanics, sixth ed. Taylor and Francis, London. CWC, 1996. Methods for sampling and testing recycled glass. Clean Washington Center, Washington. CWC, 1998. A tool kit for the use of post-consumer glass as a construction aggregate. Clean Washington Center, Report No. GL-97-5, Washington. Das, B.M., 2007. Principles of Foundation Engineering, sixth ed. PWS; Pacific Grove, California.

2351

Das, B.M., 2008. Advanced Soil Mechanics, third ed. Taylor and Francis, New York. EPA Victoria, 2005. Guidelines for hazard classification of solid prescribed industrial wastes. Environmental Protection Agency of Victoria, Publication No. 996, Victoria. EPA Victoria, 2007a. Classification of wastes. Environmental Protection Agency of Victoria, Publication No. 448, Victoria. EPA Victoria, 2007b. Information update for EPA 996. Environmental Protection Agency of Victoria, Supporting technical guidance No. 2, Victoria. FHWA, 1998. User guidelines for waste and by-product materials in pavement construction. Federal Highway Administration of U.S. Department of Transportation, Publication No. FHWA-RD-97-148, Washington. Head, K.H., 1994. Manual of Soil Laboratory Testing (volume 2): Permeability, Shear Strength and Compressibility Tests, second ed. Pentech Press, London. Holtz, R.D., Kovacs, W.D., 1981. An Introduction to Geotechnical Engineering. Prentice-Hall Inc., London. Kenny, T.C., Lau, D., Ofoegbu, G.I., 1984. Permeability of compacted granular materials. Can. Geotech. J. 21 (4), 726–729. Lambe, T.W., Whitman, R.V., 1969. Soil Mechanics. Wiley, New York. Landris, T.L., 2007. Recycled glass and dredged materials. US Army Corps of Engineers, Engineer Research and Development Center, Report No. ERDC TNDOER-T8, Mississippi. MPSC, 2000. Reusing and recycling glass in Manitoba, Challenges and Opportunities. Manitoba Product Stewardship Corporation, Winnipeg. Ooi, P.S.K., Li, M.M.W., Sagario, M.L.Q., Song, Y., 2008. Shear strength characteristics of recycled glass. J. Transplant. Res. Board. 2059, 52–62. Pennsylvania Department of Transportation, 2001. Laboratory evaluation of select engineering-related properties of glass cullet. Available from: (accessed 10.11.2010). Senadheera, S., Nash, P., Rana, A., 2005. Characterization of the behavior of granular road material containing glass cullet. Proceedings of the 7th International Conference on the Bearing Capacity of Roads, Railways and Airfields, 27–29 June, Trondheim, Norway. Sivakumar, V., McKinley, J.D., Ferguson, D., 2004. Reuse of construction waste: performance under repeated loading. ICE J. Geotech. Eng. 157 (2), 91–96. Standards Australia, 1993. Geotechnical site investigation. AS 1726–1993, New South Wales. Standards Australia, 1997. Wastes, sediments and contaminated soils. Part 3: preparation of leachates-bottle leaching procedure. AS 4439.3-1997, New South Wales. Sutherland, K., Grabinsky, M., 2003. Evaluating segregation in granular filters for earth dams. Proceedings of the 56th Canadian Geotechnical and 4th Joint IAHCNC/CGS Conference, September 28–October 1, Winnipeg, 58–63. Tam, V.W.Y., Tam, C.M., 2006. A review on the viable technology for construction waste recycling. J. Res. Conserv. Recycl. 47 (3), 209–221. Terzaghi, K., Ralph, B.P., Mesri, G., 1996. Soil Mechanics in Engineering Practice. Wiley, New York. VicRoads, 1998. Guide to General Requirements for Unbound Pavement Materials. Victoria. Vuong, B., Arulrajah, A., 2010. Field Trials of Recycled Crushed Glass in Footpaths and Shared Paths. Swinburne University of Technology Report for Municipal Association of Victoria, Victoria (unpublished). Wartman, J., Grubb, D.G., Nasim, A.S.M., 2004. Select engineering characteristics of crushed glass. J. Mater. Civil Eng. 16 (6), 526–539.