PROPERTIES OF CELLULOSE FIBRES AND WASTE PLASTIC MODIFIED POROUS FRICTION COURSE MIXES

Th is paper summarises the laboratory investigation on porous friction course mixes that were modifi ed with cellulose fi bres and waste plastics. Porous friction course mixes of three diff erent aggregate gradations were tested for predetermined binder content. Th e infl uence of each modifi er on the volumetric properties, permeability, aged abrasion loss, and moisture susceptibility of porous friction course mixes were evaluated. In order to determine the signifi cance level of eff ect of modifi ers on the above properties, the tests for analysis of variance (ANOVA) and Tukey’s multiple mean comparisons were performed. Results of statistical analyses indicate that the gradations are major source of variations in all response properties. However, modifi ers too appreciably contributed in reducing the moisture-induced damages. Th e fi ndings suggest that shredded waste plastics are potentially useful as modifi ers to porous friction course mixes.


Introduction
According to some projections (Mutha et al. 2006) in India the consumption of plastics will increase about six-fold between the year 2000 and 2030. While, in the year 2030, plastic wastes for disposal (excluding recycled plastics) will increase 10 times compared to the situation in the year 2000-2001. A huge percentage of post-consumer plastic wastes are sent to landfi ll, while remaining is subjected to the process of incineration and recycling. Th e recovery of plastic bags from older landfi lls may increase the cost of resource recovery operations and landfi ll reuse options. But, potential use of waste plastics (WP), specifi cally in porous friction courses (PFCs), and generally in road construction industry can address these problems.
PFCs are typical open-graded mixes characterized by a high percentage of interconnected air voids that can ease the drainage of surface water. PFCs are also called by diff erent names by various agencies around the world, like porous asphalt (PA), open-graded friction course (OGFC), open-graded asphalt (OGA) etc. (Suresha et al. 2007). PFCs are found to off er multiple benefi ts like better skid-resistance, reduced splash and spray, and improved night-visibility during wet weather conditions, in addition to mitigation of hydroplaning. Moreover, the negativetexture of PFC surfaces enables considerable reduction in traffi c tyre-noise. Hence, these are generally recommended for surfacing high-speed road-corridors, streets with wide carriageways and runway pavements. PA layers can also be used as drainage layer sandwiched between waterproofi ng layer and wearing course, over bridge decks (Kim et al. 2009).
Th e lower surface area due to the use of uniformly graded aggregates, and the low quantity of fi ller materials used result in the draining of bitumen-mastic (draindown) from PFC mixes during mixing, storage, transport, and laying operations. To mitigate the problem of draindown, the use of fi bres as modifi ers (stabilizers) to mixes is widely recommended. Th e use of fi bres consequently requires an increase in the binder content which further improves the durability of the mix. Further, it increases the stiff ness of bitumen-mastic minimizing the amount of draindown.

Review on utilization of fi bres
Many researchers have reported the use of various types of fi bres in diff erent types of bituminous mixes. Some of the fi ndings related to use of fi bres in PFC mixes are summarised below. Decoene (1990) discussed the possibility of using cellulose fi bres (CF) as anti-draining agent in PFC mixes and his fi ndings suggested that the quantity of CF of more than 0.3% by weight of total mix is not benefi cial. Cooley et al. (2000) investigated the moisture sensitivity behaviour of PFC mixes modifi ed with cellulose and mineral fi bres. Based on laboratory and fi eld data the authors concluded that the performance of both the stabilizers were comparable. Similar fi ndings were reported by Mallick et al. (2000). Th eir fi ndings indicated that PFC mixes with slagwool failed to satisfy the min tensile strength ratio (TSR) of 80%. Hassan et al. (2005) observed that the draindown performance of PFC mixes with fi bres was better compared to that of mixes with modifi ed bitumen. Th e experimental results of Wu et al. (2006) showed that the mixes modifi ed with CF performed better than that of mixes modifi ed with polyester fi bres, in all respect. Tayfur et al. (2007) reported that the bituminous mixes modifi ed with cellulosed fi bres coated with bitumen exhibited better rutting resistance than the mixes modifi ed with uncoated cellulosed fi bres. Wu et al. (2008) used polyester fi bres to modify the bitumen, and their fi ndings indicate that the viscosity of bitumen was signifi cantly infl uenced at lower temperatures (60-135 °C).

Review on utilization of WP
Th e disposal of WP is one of the major problems causing environmental degradation and is of worldwide concern. Environmental hazards due to WP can be addressed to a large extent by using these eff ectively in road construction. A number of studies were performed on exploring the possibility of using such as, high-density polyethylene (HDPE), low-density polyethylene (LDPE), polyethylene terephthalate (PET) etc., in cement/bituminous concrete mixes (Siddique et al. 2008). Th e following section summarizes the fi ndings on utilization of WP in the form of replacements to a part of the aggregates in the bituminous mixes, or as bitumen modifi ers.
Research on the use of WP as replacements to aggregate indicated that there is a reduction in the bulk density of compacted mix, accompanied by an increase in the stability, strength, and improved deformation capacities, in comparison to mixes of similar nature where WP were not used. Based on the fi ndings of various researchers (Hinislioglu, Agar 2004;Panda, Mazumdar 2002;Punith, Veeraragavan 2007), the major advantages of using polyethylene-modifi ed bitumen in bituminous mixes when compared to that of plain bitumen mixes included an increase in the Marshall-quotient, indicating increased stiff ness and greater ability to spread the load; an increase in the resilient modulus and fatigue life; an improved resistance to moisture susceptibility; a decrease in the plastic deformation; and an increased shear resistance. Ho et al. (2006) were of the opinion that LDPE with lower molecular weights and wider molecular weight distributions were more suitable as bitumen modifi ers when compared to that of high molecular weight with narrow molecular weight distributions. Fuentes-Audén et al. (2008) suggested that the concentration of recycled polyethylene in bitumen should not exceed 5% for paving applications, while, Al-Hadidy and Yi-qiu (2009) reported that the fl exible pavement with high performance, durability, and more economic can be obtained with 6% pyrolysis LDPE.

Objectives and scope of the study
Th e literature review indicates that investigations need to be performed on the use of WP as modifi ers to PFC mixes. Th e present study was carried out with an objective of investigating the potential use of WP as modifi ers in PFC mixes, and on performing a comparative study on the behaviour of PFC mixes modifi ed with WP and CF. Laboratory studies were performed on WP and CF modifi ed PFC mixes and the results were compared to that of mixes without modifi ers. Th e mixes were evaluated for their volumetric properties, permeability measured using the falling-head method, aged abrasion loss determined using the Cantabro abrasion test method, and the test for moisture susceptibility based on the tensile strength ratio method. To study the signifi cance of WPs as modifi ers, the test results corresponding to the properties of PFC mixes were statistically analysed.

Materials
Th e PFC mixes corresponding to three diff erent aggregate gradations (G1, G2, and G3) as shown in Fig. 1 were investigated.

Fig. 1. Gradations investigated for PFC mixes
Coarse and fi ne aggregates obtained from local stone crushing plants were used in this study. Ordinary Portland cement (OPC) blended with stone dust, was used as the mineral fi ller. Th e quantity of OPC was limited to 2% by mass of the total aggregates. Straight-run paving grade bitumen used in the present investigation, was supplied by the Mangalore Refi nery and Petrochemicals Limited (MRPL), Mangalore. Th e physical properties of coarse aggregates, and paving-grade bitumen were determined in accordance with Indian Standard test methods. Th e test results are presented in Table 1.
Commercially available CF in loose (CFL) and their pellet (CFP) forms were used as modifi ers. Th e CFL are similar to cotton, in feel. Th e CFP used had been coated with a bitumen content of 36% and a CF content of 64%, as claimed by the suppliers. Th e WPs used in this study was made available in the form of shredded plastic fi bres of size smaller than 9×3 mm, obtained from shredded plastic bags reclaimed from domestic and commercial wastes. Fig. 2 shows the samples of CFL, CFP and WP, respectively.

Experimental design
PFC mixes corresponding to diff erent gradations with or without modifi ers were prepared for pre-determined binder content (BC) of 5% by weight of the total mix. Th e selection of gradations and binder content were based on the fi ndings of the previous study (Suresha et al. 2009(Suresha et al. , 2010. Table 2 shows the details of mix combinations and coding. Th e dosage of CFL and CFP were fi xed as 0.3% and 0.45% by weight of total mix respectively, based on the earlier research fi ndings Mallick et al. 2000). Th e dosage of WP was fi xed as 0.4% by weight of the total mix. Th is is in agreement with the dosages recommended for modifi ers like fi bres and polymers, to be used in hot asphalt mixes (Al-Hadidy, Yi-qiu 2009;Fuentes-Audén et al. 2008;Ho et al. 2006;Punith, Veeraragavan 2007).
Th e procedure adopted for the preparation of cylindrical PFC specimens was the same as that followed for dense graded asphalt, as suggested in the Asphalt Institute Manual Series-2. To prepare a cylindrical specimen of Ø101.4 mm, loose hot PFC mix was compacted by applying 50 blows on each end of the specimen, using a standard Marshall hammer. Each specimen thus prepared constituted 1000 g of the aggregate in addition to pre-defi ned quantity of BC and one of the modifi ers. It may be noted that modifi er was fi rst blended with pre-heated aggregates and then hot bitumen was introduced to produce PFC mix. Totally, 144 cylindrical PFC specimens that constituted 12 replicate specimens for the 12 experimental mixes (Table 2) were prepared to evaluate the volumetric properties, coeffi cient of permeability (K), aged abrasion loss (AAL) and moisture susceptibility. Th e detailed test plan is given in Table 3.

Volumetric properties
Th e volumetric properties of compacted specimens tested included the bulk specifi c gravity (G mb ), percent air voids (V a ), and the voids in coarse aggregate (VCA m ). Th e G mb was determined using the geometric measurements of the diameter and mean length, and the mass of the specimen in air. Th e theoretical max density (G mm ) of the uncompacted mix was determined in accordance with ASTM D 2041. Th e V a was then determined using the corresponding values of G mb and G mm using Eq (1). Th e presence of stone-on-stone contact condition in the compacted PFC mix was evaluated based on the VCA m and the percentage of voids in coarse aggregate of the coarse aggregate alone (VCA d ) determined using the dry-rodded test pro-  (2) and (3), respectively. Th e stone-on-stone contact condition was confi rmed when the ratio of VCA m to VCA d was found to be lesser than unity. , , , where G CA -bulk specifi c gravity of the coarse aggregate; γ w -density of water, kg/m 3 ; γ s -bulk density of the coarse aggregate fraction in the dry-rodded condition, kg/m 3 ; P CApercentage of coarse aggregate in the total mixture, %. For tests conducted on 12 mix combinations, with three replicate tested for each mix, the results corresponding to the G mb , V a , and the stone-on-stone-contact condi-tion ( ) are presented by an individual plots, as shown in Fig. 3. Th e bulk specifi c gravity of compacted mixes (G mb ) ranged between 2.038 and 2.155. Th e presence of modifi ers resulted in variation of ±4% when compared to the mixes without modifi ers. Th e use of modifi ers in the PFC mixes resulted in higher densities, especially for mixes with the CFL and the CFP. In general, the mixes with gradation-G3 exhibited low G mb , relative to that of mixes with other gradations (G1 and G2). Th is is mainly due to higher quantity of coarse aggregates. Consequently, these mixes exhibited higher V a . Th e individual values of V a for entire experimental mixes varied from 13.0% to 17.6%. While, the mean values of V a for each experimental mix varied from 13.4% to 16.9%. Th e ratios of VCA m to VCA d presented as stone-on-stone-contact, shown in Fig. 3, confi rm the presence of stone-on-stone contact condition in the coarse aggregate skeleton in all the experimented mix combinations tested. Th us, all the experimented mixes are expected to exhibit adequate stability to resist the plastic deformation. However, fi eld performance of these mixes can be assessed by frequent inspection using any kind non-destructive testing methods; like non-nuclear density gauges (Praticò et al. 2009).

Permeability
Th e hydraulic conductivity of compacted specimens tested is expressed in terms of the coeffi cient of permeability (K) determined using the falling-head method. Th e test setup used was simple and economical. Compacted PFC specimens prepared in the standard Marshall mould were subjected to this before extruded. To prevent water leakage through the joints, the circumferential contact area between the specimen and the mould was covered using paraffi n wax on either side. Care was taken to avoid clogging of voids due to paraffi n wax in the specimen. Th e collar placed on the mould-specimen assembly, acted as a water reservoir. Water was then allowed to fl ow through the specimen, and the average time (t m , s) taken for a drop in water level from 70 mm to 30 mm was recorded. Th e typical setup for permeability test using falling-head method is shown in Fig. 4. Th e coeffi cient of permeability (K, m/day) of the cylindrical specimen of Ø101.4 mm diameter (D) and of mean length (L, mm) was calculated, by applying the temperature correction factor (T C ) for the viscosity of water, using the expression as in Eq (4). .

(4)
Th e individual permeability (K) values for the 12 mix combinations tested varied in the range of 30-133 m/day. Th us, all the mix combinations satisfi ed the permeability criteria that K should be more than 8.7 m/day (0.01 cm/s) for good drainage condition. Fig. 5 shows the individual plot K values of all the 12 mixes tested. It is generally accepted that the permeability is directly proportional to the porosity (percent air voids, V a ). Here too, the variations in the per- Fig. 3. Individual plot for volumetric properties of all 12 mixes meability seem to be similar to that of trends of air voids. Th e variations in K values of CFL, CFP and WP modifi ed mixes were respectively in the ranges of 0.38-1.50, 0.52-1.68, and 0.81-2.62 times that of the respective unmodifi ed mixes. It may be observed that mixes modifi ed with WP exhibited higher K values compared to that of other mixes, similarly, the mixes of gradation-G3 exhibited higher K values compared to the mixes with other gradations.

Aged abrasion loss
Aging of PFC specimens were simulated in the laboratory. Th e Cantabro abrasion tests were then conducted on the aged specimens to evaluate the aged abrasion loss (AAL). Compacted PFC specimens of a particular mix, in triplicate, were stored in a forced draft oven at a temperature of 60 °C for a period of 168 h. Th e specimens were then taken out from the oven, allowed to cool to the ambient temperature, and stored for a period of 4 h at a temperature (25±5 °C) corresponding to the Cantabro abrasion test. Th e aged specimen was then placed in a Los Angeles abrasion drum without any abrasive charge, and the machine was operated at a speed of 30-33 revolutions per minute for 300 revolutions. Th e loss in the specimen was expressed as a percentage of the ratio of weight of disintegrated particles to the initial weight of the specimen. Fig. 6 shows pictures of the aging process of cylindrical specimens and the specimen to be subjected into Los Angeles drum.
For tests performed on 12 mix combinations, with three replicates tested for each mix, the AAL values ranged between 5.6% and 32.7%. Th e mean and individual AAL values were found not exceeded the 30% and 50% respectively, satisfying the ASTM D 7064 requirements. Th is is evident from the Fig. 7, shows the individual plot of aged abrasion losses. Th e results of AAL tests show that the use of modifi ers resulted in an improvement in the resistance to AAL. Th e mixes corresponding to gradation-G1 exhibited high resistance to AAL followed by mixes with gradations G2 and G3.

Moisture susceptibility
Th e moisture susceptibility of PFC mixes was evaluated using the retained tensile strength or tensile strength ratio (TSR) method. Totally six replicate specimens were prepared for each mix as per the experimental design. Th ree of the six replicates, were subjected to indirect tensile strength tests in drycondition (ITS d ). Th e remaining, three specimens of each mix was then subjected to wet-conditioning, and the indirect tensile strengths (ITS w ) were evaluated. Th e wet-conditioning of the compacted PFC specimens was performed as per AASHTO T 283 with minor modifi cations. Th e specimens were fi rst saturated by submerging in water, and kept at water freezing temperature for about 15 h. Th e frozen specimens were immediately transferred into the hot water bath for thawing to a temperature of 60 °C for 24 h. Aft er two such cycles of moisture-conditioning, the specimens were kept in a cold water bath to bring down the temperature to 25 °C before testing. Th e mean ITS values of each mix for the dry-and wet-conditioned specimens were used to compute the TSR. Th e following relations were used to compute ITS and TSR. , , where P u -ultimate load required to fail specimen in the indirect tension test, N; TSR -tensile strength ratio; ITS wmean indirect tensile strengths of wet-conditioned specimens, kPa; ITS d -mean indirect tensile strengths of dryconditioned specimens, kPa. Th e mean indirect tensile strengths (ITS) of dry-and wet-conditioned PFC specimens and the TSR of the respective mixes are provided in Table 4. Th e individual values of ITS d for tests on the 12 mix combinations were found to be in the range of 314-561 kPa. When, replicate specimens were subjected to wet conditioning, the individual ITS w value was found varied between 202 kPa and 498 kPa. It may be observed in these tests, variations in ITS d of modifi ed mixes were in the range of 0.96-1.16 times that of mixes without modifi ers. Similarly, variations in ITS w of modifi ed mixes were in the range of 0.97-1.82. According to ASTM D 7064, the TSR of PFC mixes should be at least 80%, so as to ensure resistance to moisture susceptibility. Th e modifi ed and unmodifi ed mixes corresponding to gradation-G1 satisfi ed this requirement, while, mixes of gradations-G2 and G3 failed to satisfy the requirement, except in the case of M8, M10 and M12.

Statistical analysis
In order to determine the signifi cance of main eff ects of modifi ers (MF) used and gradations (G), and interaction of modifi ers and gradations (MF×G), the test for analysis of variance (ANOVA) was performed using MINITAB® (Release 15, trial version). Th e details of the F-static corresponding to each source of variation on each property obtained using the ANOVA test, while F 0 values correspondent to the respective degree of freedom (DF) and 95% confi dence interval (α = 0.05) are presented in Table 5. If F > F 0 , then it indicates that each source of variation will have a signifi cance eff ect on the mean value. In order to identify the specifi c diff erences among the mean values, the Tukey's tests were conducted for 95% simultaneous confi dence intervals.
Based on the results of the ANOVA and the Tukey's tests, the following inference can be made: the eff ect of modifi ers (MF) on AAL and  ITS d were not signifi cant; the interaction of modifi ers and gradations  (MF×G) on the G mb , V a , AAL, ITS d , and the ITS w were not signifi cant; the mean  G mb and the mean V a values of mixes corresponding to CF were found to be diff erent from that of mixes without modifi ers and also, of mixes with WP; the mean  K values of mixes with CF were found to be signifi cantly lower when compared to that of other mixes; the mean  ITS w values of modifi ed mixes were signifi cantly higher when compared to that of mixes without modifi ers; except the mean values of  ITS d , the mean values of G mb , V a, K, and ITS w of the mixes corresponding to the gradation G1 and G2 remained the same.

Conclusions
The laboratory studies were conducted on PFC mixes with and without modifiers with view to reuse the WP in road constructions. The experiment results corresponding to different properties of PFC mixes were statistically analysed. In addition to the inferences of ANOVA and Tukey's tests, the following conclusions are made.
It was found that the use of CFL and CFP contributed in improving moisture susceptibility but resulted in reduced air voids content and lower permeability.
While, the mixes modifi ed with WP exhibited improved moisture susceptibility with no signifi cant reduction in the air voids content and the permeability.
Thus, WP can be effectively reused as a modifier to PFC mixes. This will enable to consume nearly about 20-35 kg of WP/100 m 2 area of PFC surfacing for a thickness varying between 25-40 mm.