A new parameter for classification and evaluation of railway ballast fouling

322 NOTE / NOTE A new parameter for classification and evaluation of railway ballast fouling Introduction Buddhima Indraratna, Li-jun Su, and Cholachat Rujikiatkamjorn Abstract: The physical condition of railway ballast should be regularly inspected and accordingly, ballast cleaning should be carried out to maintain the safe operation of a track. This paper reviews current methods commonly used for evaluating the degree of ballast fouling, and due to their limitations, a new parameter relativallast fouling ratio is proposed. Categories of fouling based on the proposed method are derived from the particle gradation curves taken from past literature. Comparisons between these methods demonstrate that the newly proposed relativallast fouling ratio would best represent the influence of the type and gradation of fouling material. Key words: ballast, fouling index, percentage void contamination, specific gravity, relativallast fouling ratio. Résumé : La condition physique des ballasts devrait être inspectée régulièrement, et lorsque requis, le nettoyage des ballasts devrait être effectué afin de maintenir une opération sécuritaire du chemin de fer. Cet article est une revue des méthodes couramment appliquées pour l évaluation du degré de colmatage du ballast, et en raison de leurs limitations, un nouveau paramètre nommé «ratio de colmatage du ballast relatif» est proposé. Les catégories de colmatagasées sur la méthode proposée proviennent des courbes de gradation des particules extraites de la littérature antérieure. Des comparaisons entre ces méthodes démontrent que le nouveau ratio de colmatage du ballast relatif représenterait le mieux l influence du type et de la gradation du matériau colmatant. Mots-clés :ballast, indice de colmatage, pourcentage de contamination des vides, densité relative, ratio de colmatage du ballast relatif. [Traduit par la Rédaction] Received 23 February 2010. Accepted 9 August 2010. Published on the NRC Research Press Web site at cgj.nrc.ca on 1 February 2011. B. Indraratna. 1 Faculty of Engineering, University of Wollongong, Wollongong, NSW 2522, Australia. L. Su. School of Civil Engineering, Xi an University of Architecture & Technology, Xi an Shaanxi 710055, PRC; School of Civil, Mining and Environmental Engineering, Faculty of Engineering, University of Wollongong, Wollongong, NSW 2522, Australia. C. Rujikiatkamjorn. School of Civil, Mining and Environmental Engineering, Faculty of Engineering, University of Wollongong, Wollongong NSW 2522, Australia. 1 Corresponding author (e-mail: indra@uow.edu.au). Railway ballast is generally composed of uniformly graded angular aggregate. As ballast ages, it can be progressively fouled by numerous fine materials, whose accumulation in the voids of ballast can result in a decrease in shear strength with reduced resiliency and drainage capability of thallast (Janardhanam and Desai 1983). An increase in moisture in fouled ballast and infiltration of water to the subgrade can promote soft soil pumping or excessive deformation of the formation. Sources of ballast fouling consist of ballast particle degradation, infiltration of fine foreign particles from the track surface, sleeper wear, as well as sub-ballast and subgrade infiltration (Selig and Waters 1994). The commonly employed parameters for evaluating ballast fouling, i.e., fouling index (FI) and percentage void contamination (PVC), are discussed and compared in this paper. A new parameter, the relativallast fouling ratio, is proposed as an alternative that best captures the nature and gradation of fouling material. Fouling index or percentage of fouling Selig and Waters (1994) proposed the FI parameter to describe ballast fouling based on gradations obtained for representative samples of ballast in North America (Fig. 1) as ½1Š FI ¼ P 4 þ P 200 where P 4 and P 200 are percentages of ballast particles passing the number 4 sieve (4.75 mm) and number 200 sieve (0.075 mm), respectively. The categories of fouling based on FI as defined by Selig and Waters (1994) are listed in Table 1. The particles passing through the 0.075 mm sieve are included twice to emphasize the adverse influence of fine particles. An index related to FI is the percentage of fouling (% fouling), which is the ratio of the dry weight of material Can. Geotech. J. 48: 322 326 (2011) doi:10.1139/t10-066

Indraratna et al. 323 Fig. 1. Gradations representing ballast conditions from clean to highly fouled (modified from Selig and Waters 1994). passing a 9.5 mm sieve to the dry weight of the total sample. The % fouling for each sample obtained from Fig. 1 is plotted against FI in Fig. 2. An earlier study by Tutumluer et al. (2008) also confirmed a linear relationship between the % fouling and FI. However, it should be noted that the above relationship is not applicable for all types of fouling due to the limited types of fouling materials used in this empirical development. Care should be taken when evaluating fouled ballast with a large percent of particles finer than 0.075 mm. Percentage void contamination Feldman and Nissen (2002) presented the PVC parameter to capture the effect of void decrease in ballast as ½2Š PVC ¼ V 2 V 1 where V 1 is the void volumetween re-compacted ballast particles and V 2 is the total volume of re-compacted fouling material (particles passing 9.5 mm sieve). The samples for PVC tests are taken from the total depth of thallast. Therefore V 1 represents the void volume of the entirallast layer. Different allowable limits of PVC haveen applied for different track standards and ballast depths. In a concrete sleeper track with a 250 mm thickness of ballast, an allowable limit of PVC at 30% is used to specify a ballast-cleaning process considering a minimum requirement for the depth of clean ballast of 100 mm. Ballast conditions based on PVC were defined by Nissen 2 and are listed in Table 2. Comparison between FI and PVC methods If the fouling material contains insufficient fine particles, thallast voids cannot be fully filled. In this case, thallast bed will sustain drainage and resilience capacities to an acceptable extent. On the other hand, if there are more fines available to fill the voids, a greater FI will result with unsatisfactory drainage and resilience capacities of thallast. However, FI cannot differentiatetween types of fouling material including the differences in specific gravity (e.g., coal compared to soil and pulverized rock). Although the PVC method is a direct measure of percentage of voids occupied by fouling particles, the measurement of volume is time consuming. Furthermore, as the total volume of fouling particles is used, the gradation of fouling particles cannot be taken into account. For example, if the contaminates are all composed of coarse particles (4.75 mm to 9.5 mm), there should still be sufficient voids between the fouling particles; hence, thallast drainage capacity would not be significantly reduced. In this regard, PVC may overestimate the extent of fouling. The authors suggest using the solid volume of fouling particles rather than the total volume in calculating the PVC. By using the solid volume, a smaller value of PVC will be obtained if there is an insufficient quantity of fine particles within the contaminates, and vice versa. Relativallast fouling ratio By comparing the % fouling and PVC methods, the authors have proposed a new parameter, i.e., the relativallast fouling ratio, (R b-f ). It is a ratio between the solid volumes of fouling particles (passing a 9.5 mm sieve) and ballast particles (particles being retained on a 9.5 mm sieve). The relativallast fouling ratio can be defined as ½3Š R b-f ¼ M f G b-f G s-f M b where M and G s are the dry mass and specific gravities, respectively, and subscripts f and b represent the fouling materials and ballast, respectively. In eq. [3], only the mass of thallast and the mass and specific gravity of the fouling material need to be measured. This will greatly speed up the measurements compared to the PVC method, and will be more attractive to practicing track engineers. R b-f is the ratio between the solid volumes of the fouling and ballast particles. Therefore, in comparison with FI, the magnitude of R b-f can better represent the degree of fouling by various materials of different specific gravities. Comparison between R b-f and PVC In the PVC method, V 2 and V 1 can be expressed as ½4Š V 2 ¼ M f G s-fr w ½5Š V 1 ¼ M b G s-br w where r w is the density of water and e is the void ratio. PVC can then be expressed as 2 D. Nissen. Network ballast PVC test results. Internal communication, 2008.

324 Can. Geotech. J. Vol. 48, 2011 Table 1. Categories of fouling based on the fouling index, percentage of fouling, and relativallast fouling ratio. ½6Š PVC ¼ G M s-b f G s-f M b Therefore, R b-f can be related to PVC by the following expression: ½7Š R b-f ¼ PVC Fouling index (Selig and Waters 1994) (%) Percentage of Category fouling (%) Clean <1 <2 <2 Moderately clean 1 to <10 2 to <9.5 2 to <10 Moderately fouled 10 to <20 9.5 to <17.5 10 to <20 Fouled 20 to <40 17.5 to <34 20 to <50 Highly fouled 40 34 50 Fig. 2. Relationship between percentage of fouling and fouling index. Table 2. Categories of fouling based on PVC. Category PVC (%) Clean 0 20 Moderately fouled 20 29 Fouled >30 Fig. 3. Relationship between fouling categories based on relative ballast fouling ratio and fouling index. Relativallast fouling ratio (%) In eq. [7], will not change significantly from sample to sample whereas e f may be quite different between samples. A smaller value of R b-f will be obtained for a sample with a larger e f, which indicates a less severe condition of fouling, and vice versa. On the contrary, a larger e f will lead to a greater PVC value, which is contradictory to thetter permeability of the material with a greater void ratio. This indicates that R b-f can reflect the influence of gradation of the fouling material on the degree of fouling compared with the PVC method. From the results of large-scale filtration tests conducted at the University of Wollongong, PVC values can be calculated based on the measured volume and the void ratio of fouling particles. For ballast fouled by sand, silt, and clay, and for the same % fouling of 30%, the values of FI were 30.7%, 60%, and 60%, respectively, whereas the calculated PVC values were 55.9%, 46.0%, and 44.5%, respectively. The assessment of fouling using PVC becomes less accurate when the particle-size distributions of the fouling materials are distinctly different. This further implies that the PVC approach cannot reflect properly the influence of fouling particle gradation. Alternatively, if the PVC is calculated using the solid volume of fouling particles, it can then be expressed using R b-f as ½8Š PVC ¼ R b-f= Comparison between R b-f and percentage of fouling According to the relationship between FI and % fouling, categories of fouling based on the % fouling and R b-f can be calculated from thosased on FI. The calculated results are listed in Table 1. Figure 3 shows the relationship between fouling categories based on R b-f and FI. For PVC tests carried out along several railway track lines in Australia, 2 the percentage passing through a series of sieves was also measured. As the fouling particles are mostly composed of coal fines along the freight lines, it is possible to calculate the R b-f based on the specific gravities of coal (1.05 to 1.4) and ballast (2.5 to 2.8), assuming a specific gravity ratio of ballast to fouling material to be 2.4. PVC values, % fouling, and the calculated corresponding R b-f values are listed in Table 3. From Table 3 it can be observed that the categories of fouling based on PVC and % fouling are quite different, while categories of fouling obtained from the PVC and R b-f are more consistent. This indicates that % fouling is not accurate as the specific gravity is not taken into account. Anbazhagan et al. (2010) measured the shear modulus of

Indraratna et al. 325 Table 3. Comparison between PVC, percentage of fouling, and relativallast fouling ratio. PVC Percentage of fouling Relativallast fouling ratio Value Category Value Category Value Category 32.78 Fouled 7.9 Moderately clean 21 Fouled 27.68 Moderately fouled 5.9 Moderately clean 15 Moderately fouled 35.69 Fouled 8.1 Moderately clean 21 Fouled 11.63 Clean 3 Moderately clean 7 Moderately clean 31 Fouled 6.2 Moderately clean 16 Moderately fouled Fig. 4. Relationships between normalized shear modulus and (a) relative fouling ratio and (b) % fouling. ballast fouled by coal and clayey sand using multi-channel analysis of surface wave (MASW). The relationships between the normalized shear modulus and relative fouling ratio and % fouling are reproduced in Figs. 4a and 4b, respectively. Before thallast particles lose their direct contacts, the fouling material could provide extra stiffness to thallast, which would result in an increase in the normalized shear modulus. However, when thallast is highly fouled, ballast particles will gradually lose their direct contact because of the reorientation of ballast particles upon the cyclic loading and due to the lubrication effects of the fouling material. In Fig. 4a, the changing condition of the two types of fouled ballast is consistent before thallast becomes highly fouled. When this happens, much of the direct contact between ballast particles decreases substantially, and the geotechnical behaviour is then governed by the fouling material. As the properties of coal dust and clayey sand (indicating specific gravity) are quite different, the curves in Fig. 4a show a divergent trend at the end. However, the two curves in Fig. 4b are totally different from the very beginning because specific gravity is not taken into account in % fouling. This demonstrates that R b-f can be used as a more informative fouling index compared to the conventional % fouling, which does not capture thasic physical properties differencetween two different fouling materials. Prediction of ballast life A rate of contamination and a ballast life can be predicted for a track section given the value of R b-f, a limit of allowable extent of fouling, a time period since undercutting, and any changes in traffic volume. An average R b-f can be calculated by performing tests every 2 km along a track section. The rate of fouling (FR) can then be calculated by dividing the average R b-f value (R Ave b-f ) by the actual ballast life (BL ACT) since the last undercutting of the track section as follows ½9Š FR ¼ R Ave b-f =BL ACT With the above calculated FR and a prescribed allowable R b-f limit (R All b-f ), the allowablallast life (BL ALL, in years) can be determined as ½10Š BL ALL ¼ R All b-f =FR The value of BL ALL can now be incorporated in track maintenance schedules as a quantitative index, in addition to standard track inspection routines and qualitative guidelines. Conclusions By comparing the fouling index, the percentage void contamination (PVC), and the newly proposed parameter, the relativallast fouling ratio (R b-f ), the following conclusions can be drawn: (1) A quick assessment of ballast conditions can be achieved using the fouling index or % fouling, neither of which can properly represent the influence of different types of fouling materials or specific gravities. (2) The PVC approach can represent the influence of specific gravity of fouling particles. However, as thulk volume of contaminates is used in the calculations, this method is too time-consuming and cannot reflect the influence of gradation of fouling particles. (3) The proposed R b-f can reflect the influence of both the specific gravity and gradation of fouling particles. In this method, only the mass of ballast and contaminates and the specific gravity of the fouling material need to be measured. Therefore, this method is both quick and realistic, making it more attractive to practicing track engineers.

326 Can. Geotech. J. Vol. 48, 2011 Acknowledgements The authors wish to thank Mr. Darryl Nissen, Mr. Andrew Wallace, and Ms. Nayoma Tennakoon for their assistance during the preparation of this paper. Financial support from the Cooperative Research Centre for Rail Innovation is gratefully acknowledged. References Anbazhagan, P., Indraratna, B., Rujikiatkamjorn, C., and Su, L. 2010. Using a seismic survey to measure the shear modulus of clean and fouled ballast. Geomechanics and Geoengineering: An International Journal, 5(2): 117 126. Feldman, F., and Nissen, D. 2002. Alternative testing method for the measurement of ballast fouling: percentage void contamination. In Proceedings of the Conference on Railway Engineering, Wollongong, Australia, 10 13 November 2002. Railway Technical Society of Australia, Canberra, Australia. pp. 101 109. Janardhanam, R., and Desai, C.S. 1983. Three-dimensional testing and modeling of ballast. Journal of Geotechnical Engineering, 109(6): 783 796. doi:10.1061/(asce)0733-9410(1983) 109:6(783). Selig, E.T., and Waters, J.M. 1994. Track geotechnology and substructure management. Thomas Telford Services Ltd., London. Tutumluer, E., Dombrow, W., and Huang, H. 2008. Laboratory characterization of coal dust fouled ballast behavior. In Proceedings of the AREMA 2008 Annual Conference and Exposition, 21 24 September 2008, Salt Lake City, Utah. American Railway Engineering and Maintenance-of-Way Association, Lanham, Md. pp. 93 101.