Ballast Quality and Breakdown during S. Caleb Douglas, P.E., Ph.D. Manager Special Projects Civil Geotechnical Union Pacific Railroad 1400 Douglas Street, Stop 0910 Omaha, NE 68179 Office: 402-544-3581 Email: scdougla@up.com ABSTRACT Union Pacific Railroad has prioritized optimizing the life span of ballast material utilized across the system. The quantification of ballast degradation due to tamping is an unknown variable in the life cycle of ballast. A laboratory test procedure was developed to quantify the degradation of ballast due to tamping and to evaluate the future performance of ballast materials. All thirteen sources of ballast presently used on the UP System were tested. The results of the tamping test results found no correlation to the Los Angeles Abrasion test or particle shape. This study emphasizes the difficulty of predicting the field performance of ballast in the laboratory and confirms the importance of utilizing hard, dense, and durable ballast which was properly produced, transported, placed, and maintained. INTRODUCTION The American Railway Engineering and Maintenance-of-Way Association (AREMA) describes ballast as a selected crushed and graded aggregate material that is placed on the roadbed for the purpose of providing drainage, stability, flexibility, uniform support for the track structure, distribution of track loadings to the subgrade, and facilitating maintenance (1). Ballast materials must be very durable. The ballast used on the Union Pacific Railroad (UP) is generally composed of granite, traprock, quartzite, rhyolite, and basalt. Standard laboratory testing is typically conducted to provide an initial indication of material characteristics and durability. AREMA and UP presently have recommended limiting values for the following laboratory tests: gradation, bulk specific gravity, absorption, clay lumps/friable particles, Los Angeles (LA) Abrasion, sodium sulfate soundness, and flat/elongated particles. AREMA (1) commentary states the LA Abrasion test is a factor in determining the wear characteristics of the ballast material. The LA Abrasion test has long been the primary abrasion, or degradation, laboratory test in North America (2). Selig and Waters (2) indicate that some railway engineers believe the LA Abrasion test is not sufficient, or even not useful. AREMA (1) commentary additionally states the LA Abrasion test can produce laboratory results which are not representative of the field performance of ballast materials. UP uses the LA Abrasion test plus five times the Mill Abrasion test to evaluate breakdown potential. The laboratory test procedure described in this paper provides a measure of ballast breakdown from tamping which is an activity that every tie on every railroad system will experience. A laboratory test procedure was developed to quantify the degradation of ballast due to tamping and to evaluate the future performance of ballast materials. The test procedure was developed as part of a broader ballast research program by Union Pacific Railroad to optimize 940
the life span of ballast material. The optimization requires obtaining the most durable ballast for the application coupled with characterizing ballast degradation, or breakdown, cycle. Other recent Union Pacific ballast studies were described by Gehringer et al. (3) and Wnek et al. (4). Ballast degrades primarily from traffic loading and tamping. Studies by Selig and Waters indicate that 76% of the fouling material on North American railroads is due to ballast degradation compared with 41% of the fouling material on the British Railway system being from ballast degradation (2). The portion of ballast degradation due to tamping has been studied previously with wide ranging results. British Railways found a 15 to 45% reduction in the 1.5 to 2 inch (38 to 51 particle sizes and an increase from one to five percent in the material passing the 0.5 inch (13 size after 20 consecutive tamping head insertions (2,5). Another study by British Railways found that 4.4 to 8.8 lbs (2 to 4 kg) of material less than 0.5 inch (14 was generated per each tamp, which consisted of a single insertion (2,6). A study from Association of American Railroad field tests found that 20 tamping squeezes resulted in the generation of six and ten percent fine material in a granite and limestone ballast, respectively (2,7). A 2007 study by Aursudkij found two granites to produce 0.1 and 0.6 lbs (0.05 and 0.25 kg) of minus 0.5-inch (14- material per tamp in a test which consisted of ten tamps (8). TAMPING TEST PROCEDURE A full size, Jackson tamping head was mounted inside a loading frame in a custom machine constructed by Ballast Tools. A photograph of the machine is shown in Figure 1. Not shown in the photograph are the computer which controls the test and the external hydraulic pump/reservoir system. The ballast is contained in a box which is handled by a forklift to ease loading and unloading ballast into the test machine. The tie is placed in a bracket on the ballast box frame so that the tie cannot be uplifted during repeated tamping. The initial testing consisted of a 25-cycle test on each of the 13 UP ballast sources. Additional testing included 3, 12, 50, 62, and 100 cycles on selected ballast sources. All the tests were performed in the dry, except for two 50-cycle tests where the ballast was soaked for 24 hours prior to testing. An overview of the testing procedure follows. Step 1. Obtain a 350 lb (159 kg) sample of Class 1, mainline ballast. Step 2. Complete an in-the-dry gradation analysis of the ballast sample with the following sieve sizes: 2.5 in (64, 2 in (51, 1.5 in (38, 1 in (25, 0.75 in (19, and 0.5 in (13. A hand broom was utilized to obtain the fines from the sample container, which was a 55-gallon (208-liter) metal drum. A Gilson vibratory shaker was used for this task. Multiple iterations were required to pass the 350 lb (159 kg) sample through the process. Step 3. Place the minus 0.5-in (13- material in a separate container. This material was not placed into the tamping testing box for testing. 941
FIGURE 1. testing machine. Step 4. Place all the material larger 0.5 in (13 into the 55-gallon (208-liter) metal drum. Step 5. After all the material is placed in the drum, the sample was mixed by rolling the steel drum with a fork lift for a distance of 150 ft (46 m) across the laboratory parking lot. Step 6. Place the ballast sample into the ballast box for testing as shown in Figure 2. The wood tie was clamped in place to the ballast box and the ballast was placed by hand under the wood tie during filling. Step 7. Place the ballast box in the testing machine and secure the ballast box. Step 8. Complete the required number of tamping cycles. A computer program controlled the test, the tamping head penetration, and the squeezing pressure. Some of the key considerations in the tamping test were: Each tamping cycle was a double tamp. The double tamp consisted of: o a complete insertion of the tamping tines (Position A to B as shown in Figure 3), o squeezing and releasing of the tines, o a partial retraction of the tines (Position B to C), o repenetration to depth (Position C to B), o squeezing and releasing of the tines, o and a complete retraction of the tines (Position B to A). head downward pressure setpoint: 1,800 psi 942
x head squeeze pressure setpoint: 1,800 psi Step 9. Remove the ballast box from the tamping test machine and collect the ballast sample. A hand broom was utilized to obtain the fines from the ballast box. Step 10. Complete an in-the-dry gradation analysis of the ballast sample with the following sieve sizes: 2.5 in (64, 2 in (51, 1.5 in (38, 1 in (25, 0.75 in (19, and 0.5 in (13. FIGURE 2. Placing ballast into the ballast box. Position A Position B Position C FIGURE 3. tine position cycle. 943
TEST RESULTS The initial gradations of the mainline ballast for the thirteen ballast sources are shown in Table 1, as well as the UP specification. The samples were obtained at the pits and shipped in 55-gallon, metal drums to the UP laboratory. The results of the tamping tests are shown in Table 2. Historical reference testing is also included for the LA Abrasion and Flatness/Elongation based on a 3:1 standard. The change in percent passing presented in Table 2 is the increase in finer material from the original gradation for each sieve size as a result of tamping. Figure 4 illustrates the typical ballast breakdown observed during removing the ballast from the box after tamping. TABLE 1. Initial gradation and UP specification shown in percent passing Sieve Size Gradation Description 2.5 in (64 2 in (51 UP Specification 100 90-100 1.5 in (38 1in (25 0.75 in (19 0.5 in (13 Mineralogy 50-80 10-35 0-10 0-5 Not Applicable Ballast 1 100.0% 97.9% 62.3% 19.5% 7.4% 2.0% Granite Porphyry Ballast 2 100.0% 94.5% 55.5% 10.2% 1.7% 0.4% Trachyte and Syenite Porphyry Ballast 3 100.0% 86.7% 53.1% 14.3% 3.4% 0.9% Rhyolite Porphyry Ballast 4 100.0% 89.6% 42.4% 7.4% 2.3% 0.9% Diabase and Rhyolite Ballast 5 100.0% 94.9% 73.3% 21.8% 5.0% 2.2% Syenite Ballast 6 100.0% 88.8% 43.9% 12.1% 4.3% 1.1% Quartzite Ballast 7 100.0% 99.9% 84.8% 36.0% 14.7% 1.8% Rhyolite Ballast 8 100.0% 82.4% 31.3% 6.0% 1.4% 0.8% Rhyolite Porphyry Ballast 9 100.0% 84.6% 57.3% 31.2% 18.2% 5.5% Gabbro Ballast 10 100.0% 89.7% 50.7% 10.2% 2.6% 0.8% Basalt (Traprock) Ballast 11 100.0% 97.7% 69.9% 33.3% 12.7% 0.7% Basalt Ballast 12 100.0% 92.7% 69.6% 32.6% 13.1% 2.5% Basalt/Diabase (Traprock) Ballast 13 100.0% 93.9% 62.5% 25.5% 11.3% 2.4% Granite 944
TABLE 2. Percent passing change in gradation from initial for various tamping cycles. Ballast Source Number of Cycles Sieve Size 2 in 1.5 in (51 (38 1in (25 0.75 in (19 0.5 in (13 Reference Testing Flat/ LA Elong Abrasion (%) Ballast 1 25 0.7% 5.7% 9.9% 8.4% 6.0% 9.0 3.6 Ballast 2 25 1.2% 6.0% 7.8% 6.0% 3.9% 13.6 0.9 100 1.3% 11.1% 17.9% 15.2% 10.7% 25 3.1% 7.1% 10.1% 8.5% 5.8% Ballast 3 100 2.9% 13.7% 22.4% 21.5% 16.2% 16.4 1.8 3 1.0% 2.0% 1.5% 0.9% 0.6% 12 1.5% 5.2% 4.8% 3.6% 2.6% Ballast 4 25 3.4% 8.2% 10.0% 7.8% 6.0% 9.8 3.1 62 (Dry) 4.2% 16.0% 15.9% 12.8% 9.9% 50 (Wet) 4.9% 16.7% 16.5% 14.2% 11.0% Ballast 5 25 1.1% 4.9% 10.3% 9.0% 6.0% 18.5 0.0 Ballast 6 25 0.9% 7.8% 9.2% 7.7% 5.9% 15.2 0.3 3 0.0% 0.0% 1.2% 2.0% 1.7% 12 0.0% 1.2% 4.8% 5.9% 4.9% Ballast 7 25 0.0% 3.1% 9.2% 10.1% 8.8% 12.7 5.8 62 (Dry) 0.0% 3.2% 14.3% 19.3% 16.3% 50 (Wet) 7.5% 3.1% 13.6% 15.5% 13.5% 3 1.8% 3.1% 2.2% 1.6% 1.2% 12 4.3% 8.8% 5.7% 4.0% 2.6% Ballast 8 25 5.7% 12.6% 10.0% 8.1% 5.8% 15.5 3.5 62 (Dry) 13.7% 27.6% 18.9% 14.8% 11.4% 50 (Wet) 9.8% 20.0% 13.8% 11.2% 8.4% Ballast 9 25 3.7% 6.6% 6.8% 7.3% 6.5% 11.0 0.1 Ballast 10 25 4.2% 9.9% 8.9% 6.9% 4.7% 11.9 1.8 25 0.9% 3.7% 5.4% 5.1% 3.1% Ballast 11 62 0.1% 3.6% 6.2% 7.2% 5.1% 8.8 0.1 100 1.0% 5.1% 10.3% 11.0% 7.5% Ballast 12 Ballast 13 25 1.7% 5.0% 10.6% 11.7% 9.2% 100 3.8% 11.3% 18.9% 21.6% 17.8% 10.4 0.3 25 1.4% 4.7% 7.8% 8.1% 6.5% 50 1.8% 11.1% 12.7% 11.5% 8.9% 19.1 0.7 100 1.5% 12.4% 17.7% 18.1% 15.8% 945
FIGURE 4. Typical ballast degradation observed after 50 tamping cycles. With regard to settings during the test procedure, the tamping head downward pressure setpoint was 1,800 psi (12,400 kpa) and generally required 1,300 to 1,400 psi (9,000 to 9,600 kpa) to penetrate the ballast to depth. The tamping head squeeze pressure setpoint was 1,800 psi (12,400 kpa) and was approximately 2,000 psi (13,800 kpa) during squeezing while testing. ANALYSIS AND DISCUSSION The results are presented as the change in percent passing as this allowed normalization of the data. Attempts to correlate the data based on the amount of material generated at each sieve size were skewed due to the variations in the initial gradation. The initial gradations tend to be representative of the gradations produced at the quarries and the focus of the study was on the performance of each material based on actual gradations. A laboratory prepared sample to an exact gradation from each ballast source was not the intent of the study, but would likely provide a more precise comparison between the ballast materials. Table 3 provides the average change in percent passing for 1 in (25, 0.75 in (19, and 0.5 in (13, as well as the rate of change per double tamp. Similar gradation changes and trends were observed for the three particle sizes shown in Table 3. The rate of degradation was shown to decrease as the number of tamping cycles increases. Considering the percent passing the 0.5-in (13- size at 25 cycles had an average rate of degradation of 0.24 percent and the sample weighed 350 lbs (159 kg), 0.8 to 0.9 lbs (0.36 to 0.41 kg) of minus 0.5-in (13 material is generated per tamp. This result is comparable to the results from the 1975 British Railways study (2,5), the 1989 AAR study (2,7), and the 2007 Aursudkij study (8). 946
TABLE 3. Average percent passing change in gradation and rate of change for tamping cycles. Number of Cycles Sieve Size 1 in (25 0.75 in (19 0.5 in (13 Rate (% Rate (% % % % per tamp) per tamp) Rate (% per tamp) 12 5% 0.42% 4% 0.37% 3% 0.28% 25 9% 0.36% 8% 0.32% 6% 0.24% 62 14% 0.22% 14% 0.22% 11% 0.17% 100 17% 0.17% 17% 0.17% 14% 0.14% Many researchers have attempted to correlate ballast durability to the LA Abrasion Number as discussed previously. Figures 5 and 6 illustrate relationship between the tamping testing completed in this study with the LA Abrasion Number, and flatness and elongation, respectively. No statistically valid relationship between these parameters could be established based on the tamping test data. Ballast Numbers 2 and 11 produced the least amount of minus 0.5-inch (13- material for the 25-cycle tests and the 100-cycle tests. Minus 0.5-inch (13- material for Ballasts 2 and 11 from the 100 cycle tests were passed through a No. 80 sieve and observed under a microscope. The fine sand size particles shown in Figure 7 remain sub-angular to sub-round with regard to roundness for these best performing materials. 14.0% in Percent Passing 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 LA Abrasion Number 1-inch particle size 1/2-inch particle size FIGURE 5. in percent passing in relation to LA Abrasion Number based on 25 cycle testing. 947
in Percent Passing 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 1-inch particle size 1/2-inch particle size 0.0% 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Flatness and Elongation (%) FIGURE 6. in percent passing in relation to flatness and elongation based on 25 cycle testing. Ballast 2 0.025 in (0.64 948
Ballast 11 0.025 in (0.64 FIGURE 7. Particle images for Ballast 2 and 11 at 700x magnification after 100 tamping cycles. The three tests where the ballast was soaked for 24 hours did not significantly change the test results from the dry test. This would support the hypothesis that the influence of water on degradation in the field is likely linked to the properties of the abrasive slurry that forms from the combination of the water and the fines coupled with the pumping action resulting from dynamic loading. UP presently utilizes very hard ballast from thirteen sources with the highest LA Abrasion number being 19. This study and many of the previous studies have focused on laboratory prediction of ballast performance based on hardness and durability. Obtaining harder ballast for use on the UP system does not appear feasible. Increasing ballast performance in the future will require optimization of gradation, particle shape, texture, and mineralogy. The tamping test can be extended to evaluate the influence of these factors on performance. With regard to suggested practice, the 25-cycle test results shown in Figures 5 and 6 tend to support establishing a preliminary limit of 6 percent for the 0.5-inch (13 particle size for the change in percent passing during the tamping test with regard to testing for future ballast suppliers. CONCLUSIONS A laboratory test for predicting the performance of ballast materials in track was developed at the UP laboratory. The testing resulted in a rate of change for the percent passing ranging from 0.24 to 0.36% for 1 in (25, 0.75 in (19, and 0.5 in (13 particles. A decrease in the rate of degradation was observed for increasing tamping events. The tamping test data could not be correlated to the LA Abrasion number or particle flatness/elongation. This study emphasizes the difficulty of predicting the durability of ballast in the laboratory and confirms the importance of utilizing hard, dense, and durable ballast which was properly produced, transported, placed, and maintained. 949
REFERENCES (1) AREMA (2013). Ballast, Manual for Railway Engineering, Chapter 1, Part 2, Lanham, Maryland. (2) Selig, E.T., and Waters, J.H. (1994). Track Geotechnology and Substructure Management, Thomas Telford, London. (3) Gehringer, E., Read, D., and Tutumluer, E. (2012). Characterization of Ballast Performance in Heavy Axle Loading (HAL), Proceedings of AREMA 2012 Conference, Chicago, IL. (4) Wnek, M.A., Tutumluer, E., and Moaveni, M. (2013). Investigation of Aggregate Properties Influencing Railroad Ballast Performance, 92nd Annual Meeting of the Transportation Research Board, Washington, DC, January 13-17. (5) Burks, M.E., Robson, J.D., and Shenton, M.J. (1975). Comparison of Robel Supermat and Plasser 07-16 Track Maintenance Machines, Tech. Note TN SM 139, British Railways Board R&D Division, December. (6) Wright, S.E. (1983). Damage Caused to Ballast by Mechanical Maintenance Techniques, British Rail Research Technical Memorandum TM TD 15, May. (7) Chrismer, S.M. (1989). Track Surfacing with Conventional and Stone Injection, Association of American Railroads Research Report No. R-719, Chicago, IL, March. (8) Aursudkij, B. (2007). A Laboratory Study of Railway Ballast Behavior Under Traffic Loading and Maintenance, Doctor of Philosophy Thesis, The University of Nottingham, September. BIOGRAPHICAL SKETCH Caleb Douglas is a Manager Special Projects Civil Geotechnical for the Union Pacific Railway. He has system wide responsibilities for geotechnical aspects, primarily supporting Maintenance of Way projects. Caleb joined Union Pacific after practicing as a geotechnical consultant in the southeast United States for 10 years. Dr. Douglas holds degrees from Mississippi State University and Iowa State University. Caleb is a registered professional engineer and is a member of AREMA Committee 1. 950
Ballast Quality and Breakdown during Caleb Douglas, PE, PhD Manager Special Projects Civil Geotechnical Union Pacific Railroad 951
Union Pacific System Seattle Portland Oakland LA Calexico Nogales Eastport SLC El Paso Denver Eagle Pass Laredo Twin Cities Omaha Dallas Duluth KC Chicago St. Louis Memphis Houston Brownsville New Orleans Fast Facts Commodity Revenue Route Miles $15.5 B 32,200 in 23 States Employees 49,000 Customers 25,000 Locomotives 8,700 Introduction Ballast breakdown Selig and Waters found that 76% of the fouling material on North American railroads was due to ballast degradation Traffic Loading Problem Statement Ballast Life Cycle Quantification of ballast breakdown due to tamping Ballast fouling = traffic damage + tamping damage Development of laboratory test to predict ballast performance in the field Tests, such as LA Abrasion and Mill Abrasion, have not been good predictors of performance Degradation Accumulation Degradation Trends Increased Traffic Degradation Decreased Traffic Time 952
Test Equipment Jackson tamping head Test Procedure Obtain 350 lb (159 kg) mainline ballast. Complete an in-the-dry gradation analysis of the ballast sample with the following sieve sizes: 2.5 in (64, 2 in (51, 1.5 in (38, 1 in (25, 0.75 in (19, and 0.5 in (13. Test Procedure (Cont.) Place the minus 0.5-in (13- material in a separate container. Mix by rolling in 55-gallon drum Place into the ballast box Test Procedure (Cont.) Complete the required number of tamping cycles. A computer program controlled the test, the tamping head penetration, and the squeezing pressure. head downward pressure: 1,800 psi head squeeze pressure: 1,800 psi Each tamping cycle was a double tamp Remove ballast and complete in-the-dry gradation Double tamp cycle Test in Progress 1 cycle is 1 double tamp A B C B A sequence Position A Position B Position C 953
Ballast after testing Initial Gradations Gradation Description 2.5 in (64 2 in (51 1.5 in (38 Sieve Size 1 in (25 0.75 in (19 0.5 in (13 Mineralogy UP Specification 100 90-100 50-80 10-35 0-10 0-5 - - - Ballast 1 100.0% 97.9% 62.3% 19.5% 7.4% 2.0% Granite Porphyry Ballast 2 100.0% 94.5% 55.5% 10.2% 1.7% 0.4% Trachyte and Syenite Porphyry Ballast 3 100.0% 86.7% 53.1% 14.3% 3.4% 0.9% Rhyolite Porphyry Ballast 4 100.0% 89.6% 42.4% 7.4% 2.3% 0.9% Diabase and Rhyolite Ballast 5 100.0% 94.9% 73.3% 21.8% 5.0% 2.2% Syenite Ballast 6 100.0% 88.8% 43.9% 12.1% 4.3% 1.1% Quartzite Ballast 7 100.0% 99.9% 84.8% 36.0% 14.7% 1.8% Rhyolite Ballast 8 100.0% 82.4% 31.3% 6.0% 1.4% 0.8% Rhyolite Porphyry Ballast 9 100.0% 84.6% 57.3% 31.2% 18.2% 5.5% Gabbro Ballast 10 100.0% 89.7% 50.7% 10.2% 2.6% 0.8% Basalt (Traprock) Ballast 11 100.0% 97.7% 69.9% 33.3% 12.7% 0.7% Basalt Ballast 12 100.0% 92.7% 69.6% 32.6% 13.1% 2.5% Basalt/Diabase (Traprock) Ballast 13 100.0% 93.9% 62.5% 25.5% 11.3% 2.4% Granite Findings Average increase in particle sizes Number of Cycles Sieve Size 1 in (25 0.75 in (19 0.5 in (13 % Rate (% per tamp) % Rate (% per tamp) 25 cycle testing indicate 0.8 to 0.9 lbs (0.36 to 0.41 kg) of minus 0.5 in (13 is generated per tamp. % Rate (% per tamp) 12 5% 0.42% 4% 0.37% 3% 0.28% 25 9% 0.36% 8% 0.32% 6% 0.24% 62 14% 0.22% 14% 0.22% 11% 0.17% 100 17% 0.17% 17% 0.17% 14% 0.14% Poor Correlation with LA Abrasion in Percent Passing 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 LA Abrasion Number 1-inch particle size 1/2-inch particle size Poor Correlation with Flatness and Elongation 14.0% Particle Shapes after testing Ballast 2 after 100 cycles 12.0% in Percent Passing 10.0% 8.0% 6.0% 4.0% 2.0% 1-inch particle size 1/2-inch particle size 0.0% 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Flatness and Elongation (%) 954
Particle Shapes after testing Ballast 11 after 100 cycles Influence of water Three tests were conducted after soaking the ballast for 24 hours. Soaking did not significantly change the test results from the dry test. This would support the hypothesis that: the influence of water on degradation in the field is likely linked to the properties of the abrasive slurry coupled with the pumping action resulting from dynamic loading. Conclusions Rate of change for the percent passing ranging from 0.24 to 0.36% for 1 in (25, 0.75 in (19, and 0.5 in (13 particles. 0.8 to 0.9 lbs (0.36 to 0.41 kg) of minus 0.5-in (13 material generated per tamp Decrease in the rate of degradation was observed for increasing tamping events. The tamping test data could not be correlated to the LA Abrasion number or particle shape. Further studies are required to determine test relationship with performance. Questions 955