The objective of the work presented in this paper is to investigate the direct shear strength of a combined sand and rubber tyre mixture for possible use as a lightweight fill material. The paper concentrates on the shear strength of the mixture under relatively high normal stresses in an attempt to assess its suitability as a fill material for use in relatively large geotechnical engineering projects. The paper presents the results of direct- shear tests on both small particles of 'granular' rubber and small particles of 'shredded' rubber waste, representing the sizes and shapes currently available in the UK. An example slope stability calculation is performed to demonstrate changes in material strength using the widely available computer program SLOPE in the OASYS computer package.
In industrialised countries, rubber tyres alone account for 60% of the total rubber consumption (Ahmed & Van de Klundert1). Typically, rubber tyres have a unit weight of approximately 5kN/m3 and disposal of the whole tyre in landfills is being increasingly banned due to their 'bulky' shape and tendency to migrate towards the surface with time. In 1994 it was estimated that 2 billion scrap tyres were stockpiled across the United States and that the stockpiles would continue to grow at the rate of 200- 250 million tyres per year (Edil & Bosscher2). In Germany and the UK the rate of growth is around 0.6 million tyres per year and 0.74 million tyres per year respectively, constituting around 1% to 2% of the total municipal solid waste1. The remaining 40% of rubber waste represent discards from the shoe making industry, rubber belts, tubes and so on.
Currently in the UK waste rubber is available in particle sizes no greater than 5mm and is produced in either 'chipped' or 'shredded' forms. The object of this study was to investigate the performance of UK waste rubber when combined with fine sand to form a composite in an attempt to assess its possible use as a lightweight construction material. The results presented in the paper should not be used as a definitive document on UK rubber mixture design, but as an initial guide to its possible behaviour and use. More research is required before any formal design guidelines can be established and it is hoped that this paper will lead to more general discussions and research in this area.
Direct-shear tests were carried out on a series of rubber-sand composites to determine how the percentage rubber content in the composite influenced the shear strength. Tests were performed at (relatively) high normal stress to assess the possible performance of the material in relatively large scale construction projects as this will lead to conservative results. The parameters examined include initial density, increasing rubber content in the composite, and a comparison of the influence of two different shapes of rubber particles on the overall performance. In an attempt to compare results and determine an 'optimum' mixture, all results are presented later in the form of 'figures of merit'.
The paper concludes by presenting an example slope stability problem, using a standard computer package, to illustrate its possible use as lightweight fill as a method of improving the slope factor of safety.
Research into new methods for the recycling of scrap tyres has been increasing at a modest rate. Probably the two most popular new methods of disposal are through the burning of ground-up tyres in power plants, and the mixing of rubber and asphalt in pavement construction (Ahmed & Lovell3 and Ahmed4). Small proportions of rubber are also used as an energy absorbing material in children's play areas to prevent injury.
Edlin & Senouci5 performed tests using rubber as a concrete aggregate. Three sets of tests were performed, the first set on chipped tyres of sizes between 19-38mm in size and the second and third sets using smaller diameter chips of 6mm and 2mm respectively. They found that when mixed with cement the rubber aggregate tends to act as a large pore and did not have a significant role in the resistance to applied external loading. The compression and tensile strengths of the concrete were strongly dependent on the volume of rubber aggregate. Reductions in strength of up to 85% of the compressive strength and 65% of the tensile strength were observed. The rubber-concrete did not experience brittle failure and was able to absorb a significant amount of plastic energy. They proposed that the rubber-concrete had several possible applications including false facades, low-strength lightweight driveways and crash barriers.
The majority of the previous research carried out on soil-rubber composites examined the reinforcing effect of 'long' tyre shreds between 50-450mm in length. Foose et al6 performed direct shear tests on dry sand and tyre composites consisting of shreds between 50-150mm in length and found that the addition of tyre shreds tended to increase the shear strength of the sand. Edil and Bosscher2 performed direct-shear tests on soil-shredded rubber composites focusing on tyre shred sizes ranging from 25mm wide by 50mm long to 100mm wide by 450mm long (the most common chip sizes being 50mm wide by 75mm long). They found that the composite exhibited a significant plastic compression under load (up to 40% of initial placement thickness), but in certain cases some improvement in frictional strength was observed. It has also been suggested that shredded tyres do not appear to show any adverse effects on groundwater quality (Bosscher et al7). However, more research is required to verify this as ongoing work by the authors has found that the rubber appears to produce some discoloration in kaolin clay composites.
Direct shear machine and sample preparation
Direct shear machine
All the shear strength tests were performed on a 300mm square direct- shear machine ('shear-box'). The machine represents a typical commercial machine, originally designed for testing free-draining materials containing particles up to 37.5mm in diameter. All the shear tests and sample preparation were performed in accordance with British Standard BS1377:1990. The normal stress is applied through a hydraulic loading system and is designed to provide normal stresses from 111-1100kPa. The horizontal shear stress is applied through a motorised hydraulic loading system providing a constant rate of displacement of the box (between 0.001mm/min to 5mm/min). The horizontal force is measured by a calibrated load ring and vertical displacement is measured by a calibrated dial gauge.
Sample preparation and test procedure
The required amount of sand or pre-mixed sand-rubber composite was poured steadily into the shear box using a circular motion. The material was then spread as evenly as possible and the required density achieved through vibration as discussed by Poulin8.
The normal force was applied to the specimen to give the desired vertical (normal) stress n measured in kPa. Preliminary tests on the material indicated that approximately 15 minutes was required to ensure that full settlement of the sand-rubber composite had occurred. The settlement of the material was recorded and the vertical strain calculated. A constant rate of displacement was then applied to the material and the test stopped when the horizontal displacement reached 45mm (close to the limit of the machine) corresponding to a horizontal strain of 15.0%. Failure was assumed to occur at the peak of the stress-strain curve, however; for some of the samples a peak was not obtained at 45mm. For these samples, it was noted that the increase in shear stress was very small at this displacement and this point was taken as the shear strength of the material.
The sand used in this study was 'Levenseat' sand which is a naturally occurring fine grained granular material and was chosen due to its availability and relatively low cost. Furthermore, its unit weight was found to be relatively easy to control to provide different densities for comparison. Figure 1 shows the particle size distribution curve for the material and Table 1 shows the results of standard classification tests. The sand was classified as uniformly graded.
Two different types of discarded tyres were used; 'granulated' rubber (often called 'chipped') and 'shredded' rubber, representing the sizes available for recycle purposes in the UK (Photo 1). The granulated rubber consisted of small angular particles that passed the 5mm sieve in any orientation, whereas the shredded rubber had a length of approximately 30mm and passed the 3.35mm sieve in its longitudinal direction only. The two rubbers had generally clean cuts and only a small percentage of steel wires were exposed. This made triaxial compression tests difficult due to membrane penetration and hence a magnet was used to remove the steel component. 'Free' steel was not present in the rubber. Figure 2 shows the grading curves for the rubbers and Table 2 shows the results of standard classification tests.
Tables 1 and 2 show that the rubber has a very low dry density compared to the sand. At the minimum dry densities the granulated rubber is 3.22 times lighter and the shredded rubber is 5.70 times lighter. It was observed that in the shredded rubber there was a large degree of 'interlocking' between the shredded particles that tended to create large voids, giving rise to large deformation during normal load application.
The two types of rubber were mixed with Levenseat sand to form a composite material. The different percentages of rubber tested (in terms of their mass in the composite) were 0%, 10%, 20% and 40%.
In an attempt to create a 'uniform' mixture the sand and rubber were mixed in a standard soil mixer for approximately 5 minutes. It was observed that as the amount of rubber in the composite increased the amount of segregation of the particles also increased this was particularly noticeable in the shredded composite. Photos 2 & 3 show the two composites at 10%, 20% and 40% and Table 3 shows the results of classification tests. As expected, the maximum density achieved during compaction decreased with increasing rubber content. Figure 3 shows curves of maximum and minimum dry densities for the two composites and illustrates a marked change from the 0% to 20% rubber content samples to the 20% to 40% rubber content samples (the 100% point is for reference only in Figures 3 & 5). The first region represents the rubber contents at which the density was influenced more strongly by the sand. In the second region the influence of the rubber increases at a greater rate and segregation of the mix can occur (particularly when a dense state is attempted). The graph also shows that the shape of the rubber particle influenced the dry density, lower densities being achieved using the shredded rubber.
The normal stress range considered was set between 333kPa and 777 kPa, roughly representing lower fill surcharges in an embankment over 20m in height (for an example unit weight of 15 kN/m3). The large shear box is designed to test at these large normal stresses and the results of the testing programme should yield conservative results for any lower stresses. Tests on pure Levenseat sand samples were performed on a 100mm by 100mm direct-shear machine at a range of normal stresses and the friction angles attained were found to be in good agreement with the large shear box at higher normal loads. Initially three different densities were chosen, namely loose, medium-dense and dense. However, the direct-shear test results presented in this paper only concentrate on the loose and medium-dense states as high initial densities for the composite became difficult to obtain due to significant segregation of the particles during compaction, leading to unreliable results.
Direct shear strength behaviour
Previous work carried out on rubber shreds greater than 5mm in length (Foose et al6) demonstrated that the angle of friction envelopes for dense sand-rubber composites exhibit a bilinear relationship and those for loose and medium-dense sands are approximately linear. This paper confirms that the angle of friction for the smaller rubber chips and shreds gives an approximately linear relationship in the loose and medium-dense states. Work on longer shreds (Foose et al6 & Humphrey et al9) reported an apparent 'cohesion' (ie a shear strength at zero mean stress) when rubber samples alone were tested at low normal stresses. However, no evidence of this in either the rubber chip or shred samples has yet to be established in this work using higher normal stresses. It would appear that the small particle sizes follow purely granular behaviour; cohesion seems to develop as the rubber particle size increases.
Figure 4 shows a typical coulomb envelope for the rubber chip-sand composite at the medium-dense state and suggests a linear relation as commented above. The various envelopes are summarised in Figure 5 as angles of friction with percentage of rubber chip or shred content in the composite for both loose and medium-dense samples, and Table 4 gives the variation of the friction angle with initial density and unit weight for each composite tested. The angle of friction of the Levenseat sand was found to be 22degrees in its loosest state and 31degrees in the medium-dense state. It was found that values of for the pure rubber samples appeared not to vary appreciably with change in density of the sample. The rubber chip samples had an angle of friction of 19degrees in their loose state and 18degrees in the medium-dense state. Pure rubber shreds gave slightly higher angles of 22degrees and 22degrees in the loose and medium dense states respectively. These values of compare favourably with those obtained by Humphrey et al9 who found for shredded tyres to lie between 19degrees and 25degrees.
Angles of repose for rubber chips and shreds were also obtained for comparison and were found to be 37degrees and 45degrees respectively; significantly higher than the friction angles obtained from the direct shear testing. The angle of repose should not therefore be used as a lower estimate of for engineering purposes.
In the loose state, both the chips and shreds were found to improve the angle of friction of the sand at rubber contents of 15% or less. Beyond this level the angle of friction begins to reduce as the influence of the rubber begins to dominate the composite behaviour. In the medium dense state improvement in the friction angle is limited to 10% rubber content or less. The maximum improvement obtained () was approximately 5degrees using 10% rubber shreds in the loose state and approximately 1.5degrees using 10% rubber chip content in the medium-dense state.
Samples in the loose state did not demonstrate a definite peak stress, instead exhibiting a gradual build up in shear stress to the 15% strain limit (45mm horizontal displacement). In the medium-dense state however a peak shear stress was observed. At both densities and using either of the rubber materials, the initial vertical strain experienced by the composite on loading was high. The magnitude of this initial strain increased with increasing rubber content and normal load. By comparison, during direct- shear testing relatively low vertical strains of the composite were observed.
Summarising, it would appear that between 10% and 20% rubber content the material is behaving as a 'true' composite in which the engineering properties of both materials are enhanced. In general, these results reflect those obtained by Black and Shakoor10 who found that the shredded tyre content required to optimise the composite's engineering properties was also between 10% and 20%.
Comparison of composites
In an attempt to compare the results obtained from the direct shear strength testing into a form in which properties can be compared a set of normalised dimensionless figures of merit are proposed. The figures of merit (Table 5) are defined in such a way that any value greater than unity represents a better material than the sand alone (in relation to either the loose or medium-dense state).
Shear strength figure of merit
Let be the figure of merit for angles of friction, such that tansrtans where, s is the friction angle of the sand and sr is the friction angle of the composite.
Unit weight figure of merit Let be the figure of merit for unit weights, such that = s sr where, s is the unit weight of the sand and sr is the unit weight of the composite
General shear strength and unit weight figure of merit Let be the figure of merit for shear strength and unit weight combined (Table 5)
The figures of merit are not intended as a definitive solution to composite mix design; they are merely included as a means of comparison of the results obtained to date. The figures of merit offer a useful comparison to the feasibility of the various mixes and can offer information on their behaviour.
Figures of merit for the composites
It appears that the inclusion of small amounts of rubber particles do have a benefit on the shear strength of the Levenseat sand (given by the increase in the parameter ). Benefits resulting from the reduction in unit weight of the composite () become more significant by the inclusion of 40% rubber chips or 20% rubber shreds.
The combined shear strength and unit weight figure of merit () for loose state samples suggests that the rubber can improve the performance of the Levenseat sand. Apparent improvement increases with rubber content due to the large effect of the parameter.
In the medium-dense state, samples show much less improvement over the performance of the sand only. Values of , in general, show an overall reduction in angle of friction for rubber contents 10%. Unit weights also show a less significant reduction using 40% rubber chips or 20% to 40% shreds than the loose state. The resulting values of show an improvement of less than 10% with the one exception being a 41% improvement for a 40% rubber shred content.
As an example, consider the 30m high embankment shown in Figure 6 constructed over a 20m deposit of firm clay of undrained shear strength of Cu=65 kPa and a saturated unit weight of =18 kN/m3 at a slope of gradient 1:2.5. The slope stability program SLOPE11 (part of the OASYS computer package) was used with BISHOP's Simplified Method with Horizontal Interslice Forces12 to find the minimum factor of safety of the slope. Using medium-dense Levenseat sand as the slope fill material (=30.9degrees & =15.4 kN/m3) F=0.95 (the slip mechanism is shown in Figure 6). If the Levenseat sand was replaced using a composite of sand plus 20% shredded rubber at the medium-dense state (=27degrees & =12.4 kN/m3) the factor of safety increases to F=1.109, representing a 16% increase in stability due to the lightweight nature of the material. Greater increases in stability would be possible if longer rubber shreds were used due to the expected increase in the composite friction angle.
Previous research carried out in the United States on 'long' rubber shreds has shown that, in general, they are able to improve the shear strength of sands. Considerably smaller waste rubber chips, of the type currently produced in the UK, have been tested in combination with dry Levenseat sand in direct shear. The conclusions from this preliminary investigation into the factors affecting the engineering properties of such composites can be summarised as follows:
In general, large rubber contents in the composites lead to a reduction in shear strength and an increase in initial compressibility. However, significant reductions in unit weight can be achieved.
The 'optimum' rubber content, in terms of a lower unit weight and improvement to the shear strength of the sand, was found to be in the 10% to 20% range.
This paper represents the results of a feasibility study into a relatively lightweight composite material comprising fine sand and rubber. Further work is required to ascertain the full extent of the material behaviour, including its compressibility and dynamic behaviour, before general design guidelines can be determined. In particular an investigation into the performance of the composite in the triaxial cell would prove invaluable in completing the overall picture. Liaison with the UK rubber recycling industry will be essential to determine the economic feasibility of producing 'longer' rubber shreds which have been shown by other researchers to produce a greater increase in shear strength, including the appearance of an apparent 'cohesion'.
The second author gratefully acknowledges the support of the Mott MacDonald Charitable Trust and the UK Engineering & Physical Sciences Research Council. The authors would also like to acknowledge Poulin8 for his work in connection with this paper.
1.AHMED, R and Van de KLUNDERT, A, Rubber recycling. 20th WEDC conference, Colombo, Sri Lanka, (1994).
2.EDIL, TB and BOSSCHER, PJ, Engineering properties of tire chips and soil mixtures. Geotechnical Testing Journal, Vol 17, No 4, 453-464, (1994).
3.AHMED, I and LOVELL, CW, Use of waste materials in highway construction: State of the practice and evaluation of shredded waste products. Transportation Research Board, Washington DC, 1-9, (1992).
4.AHMED, I, Laboratory study on properties of rubber-soils. PhD thesis, School of Civil Engineering, Perdue University, W Lafayette, (1993).
5.ELDIN, NN and SENOUCI, AB, Rubber-tire particles as concrete aggregate. Journal of Materials in Civil Engineering, Vol. 5, No. 4, 478-496, (1993).
6.FOOSE, GJ, BENSON, CH and BOSSCHER, PJ, Sand reinforced with shredded waste tires. Journal of Geotechnical Engineering, Vol. 122, No. 9, 760- 767, (1996).
7.BOSSCHER, PJ, EDIL, TB and ELDIN, NN, Construction and performance of a shredded waste tire test embankment. Transportation Research Board, Washington DC, Transportation Research Board No 1345, 44-52, (1992).
8.POULIN, F, Large shear box testing on rubber/sand composites, MSc dissertation, Heriot-Watt University, (1996).
9.HUMPHREY, DN and MANION, WP, Properties of tire chips for lightweight fill. Grouting, soil improvement and geosynthetics, American Society of Civil Engineers, New York, Geotechnical special publication, No 30, Vol. 2, 1344-1355, (1992).
10.BLACK, BA and SHAKOOR, A, A Geotechnical investigation of soil-tyre mixtures for engineering applications. First international congress on environmental geotechnics, (1994).
11. SLOPE user manual version 3.5. Copyright Oasys, London, UK, (1991).
12. BISHOP, AW The use of the slip circle in the stability analysis of slopes. Geotechnique Vol. 5, No 1, 7-17, (1955).