Centrifuge Tests on Comixing of Mine Tailings and Waste Rock Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Nonika Antonaki, Ph.D., A.M.ASCE 1; Tarek Abdoun, M.ASCE 2; and Inthuorn Sasanakul, P.E., A.M.ASCE 3 Abstract: A series of centrifuge tests were conducted to examine self-weight consolidation, dynamic response, liquefaction potential, and slope stability of mine tailings alone and in mixtures with waste rock. Mixture ratio of waste rock to tailings by dry mass varied around the region of a rock skeleton just-filled with liquefiable mine tailings (roughly 2.3∶1 in this case). The performance of three mixtures was compared to that of tailings in terms of pore pressure dissipation, shear wave velocity, soil acceleration, lateral displacement, pore pressure buildup, liquefied depth, settlement, and slope stability. Comixing proved to be a promising alternative to conventional separate disposal. Consolidation time and settlement were reduced significantly, and shear wave velocity increased with increasing rock content. The improvement in dynamic response was not pronounced when the tailings proportion was larger than required to fill the rock skeleton but became very apparent when that ratio was reached or surpassed. DOI: 10.1061/(ASCE)GT.1943-5606.0001783. © 2017 American Society of Civil Engineers. Introduction Vast quantities of mine waste are generated annually and disposed of by the mining industry. The produced waste can typically be divided into two categories: highly processed mine tailings and coarse waste rock. The different types of waste are normally stored separately (Blight 2009). Failures of tailings storages can pose an enormous threat to human safety as well as the environment because of the large material volumes that may be released. Earthquake-induced liquefaction is one of the major causes of failure in seismically active areas (Rico et al. 2008; Blight 2009). Tailings from hard rock mines are often very loose, saturated, nonplastic sandy silts and are thus vulnerable to liquefaction (Vick 1990). Their generally low permeability can result in slow consolidation rates, and shear strength typically remains low even after consolidation. Conversely, waste rock dumps possess high shear strength and permeability and are mostly dry. Yet water seepage through the sizeable voids can cause acid mine drainage (AMD) and contamination, depending on the chemical composition of the mine waste. Comixing [initially proposed by Brawner (1978)] is defined as the combination of tailings and waste rock prior to disposal (Bussiere 2007). Taking advantage of the benefits of both waste streams is a promising way of addressing the issues of conventional separate disposal. Keeping the waste rock saturated reduces AMD, and adding large frictional particles to the mine tailings can both 1 Geotechnical Engineer, Geotechnical and Tunneling, Technical Excellence Center, WSP; 1 Penn Plaza, New York, NY 10119 (corresponding author). ORCID: https://orcid.org/0000-0003-4226-3231. E-mail: firstname.lastname@example.org; email@example.com; firstname.lastname@example.org 2 Professor, Dept. of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, 110 8th St., JEC 4049, Troy, NY 12180. E-mail: email@example.com 3 Assistant Professor, Dept. of Civil and Environmental Engineering, Univ. of South Carolina, 300 Main St., C227, Columbia, SC 29208. E-mail: firstname.lastname@example.org Note. This manuscript was submitted on July 26, 2016; approved on May 18, 2017; published online on October 25, 2017. Discussion period open until March 25, 2018; separate discussions must be submitted for individual papers. This paper is part of the Journal of Geotechnical and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. © ASCE reinforce and aid drainage. Using the voids of the waste rock to store the tailings additionally serves to utilize otherwise unexploited space. Increased shear strength can lead to steeper and more stable beach slopes in the impoundments and further conserve space. A few key considerations are chemical compatibility of the materials, degree of mixing, method of placement, and mixture ratio of waste rock to mine tailings (Wickland et al. 2006). Ratios between 0.5∶1 and 8∶1 (by volume) were considered, and ratios higher than 4∶1 were suggested by Leduc and Smith (2003). Wickland and Wilson (2005) performed a meso-scale column study on the consolidation of mixtures at ratios of 4∶1 and higher, relating waste rock to tailings by dry mass. That would translate to somewhat lower ratios of dry rock to wet tailings by volume. They blended clay to sand-size tailings with rock particles smaller than 15 cm using a concrete transit mixer. Codisposal by combined pumping at a ratio of 5.25∶1 has been tried by the Jeebropilly colliery in southeastern Queensland, Australia (Morris and Williams 1997). The authors did not clarify whether the ratio referred to volume or mass. According to Wickland et al. (2010), codisposal was listed in the services of major Canadian geotechnical firms in 2003 and was considered as an option since 2007, even though the concept had barely been applied or investigated. Bussiere (2007) subdivided codisposal of tailings and waste rock into comixing, layering codisposal and waste rock in impoundments, depending on the degree of mixing of the two materials. Waste rock is commonly a coarse run-of-mine product created from the blasting of hard rock. Leduc and Smith (2003) talked about practical issues of mixing, such as the selected method, and said that it should be evaluated for the specific site and mine waste. They specifically mentioned three methods of blending: mixing at the face by placing both tailings and waste rock near the crest of the active dump and then pushing both over the face with a dozer; placing the tailings in a haul truck preloaded with waste rock and then dumping the un-mixed material at the dump face; and mixing the waste and tailings together on a conveyor belt. Costs associated with these new approaches can initially be higher than with traditional tailings management methods, but the difference may be worth it when considering site rehabilitation costs and long-term environmental risks (Bussiere 2007). 04017099-1 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Centrifuge Testing Program and Description A set of nine centrifuge tests was conducted to study the behavior of fine mine tailings alone and in mixtures with waste rock. Tests 1–6 were performed with tailings, and Tests 7–9 were performed using mixtures. The first two tests were static selfweight consolidation tests, and the second two examined the dynamic response of partially consolidated mine tailings (Antonaki et al. 2013, 2014). Only Tests 5–9 are presented herein for the purpose of comparing the behavior of fully consolidated mine tailings and waste rock–tailings mixtures in terms of their consolidation, as well as their response to dynamic loading. Dynamic loading was applied by the in-flight shaker installed on the centrifuge basket. A one-dimensional stacked-ring laminar container was used to allow soil profile deformation in the direction of dynamic loading as illustrated in the cross-sectional view of Fig. 1. The container included rigid end walls that were removed before the model was subjected to an earthquake. A centrifugal acceleration of 80 g was deemed appropriate to simulate a deposit of small-to-medium height, and all quantities, including the container dimensions in Fig. 1, are reported in the prototype scale. The mine tailings were shipped for study from the metallurgical pilot plant for a planned mining project (by Golder Associates Ltd.) located in the earthquake-prone Andean region of South America. The material consisted of 60% fines and was classified as CL-ML (LL of 22% and PL of 17%). A water content of 59% was selected as the pumping water content at disposal and was applied to all tests. The corresponding density was 1,650 kg=m3 . According to Taylor (1995), water is most commonly used in geotechnical centrifuge tests. The velocity of water is scaled up by scale factor N, and water is sometimes replaced with fluid at a viscosity N times higher than that of water. That, however, is generally not practically applicable to fine-grained soils, and the slight increase in permeability when using water in centrifuge tests is considered acceptable. Because of the very low shear strength of the material, special miniature plates had to be designed and glued to all sensors to make instrumentation feasible. Excessive lateral movement of the sensors during a calibration test led to the development of a supplementary means of stabilization in the form of thin vertical wires running through the sensor plates. The wires were loosely tied to the crossbeams and the bottom of the container. Waste rock particles with dimensions in the range of 20–40 cm were considered to be adequately representative of field conditions based on a literature review (Wickland and Wilson 2005; Wickland 2006; Wickland et al. 2010) and internal discussions with experienced mining engineers (Golder Associates Ltd., personal communication, 2014). Crushed stone with an average size of 0.3 cm was purchased and sieved to achieve that range when a scaling factor of approximately 80 was taken into account. The density and void ratio of the dry waste rock were measured at 1,400 kg=m3 and 0.85, respectively. Using basic material properties and soil phase relations, the mixture ratio to accomplish a just-filled rock skeleton was calculated at approximately 2.3∶1 waste rock to mine tailings by dry mass. Ratios between 2∶1 and 3∶1 were therefore used in Tests 7–9. The two materials were blended using a power mixer, as illustrated in Fig. 2(a). While the tailings’ initial water content was equal to 59% for all tests, water content of mixtures at disposal was between three and four times lower. Models were built in four layers, at approximately 6–7.5-m initial height per layer depending on the material. Layers were numbered in the order that they were placed, as shown in Fig. 1. Layer heights varied slightly between tests and testing phases. Mine tailings models were built higher because they were expected to settle more than mixtures. Brief consolidation at 20 g under the material’s own weight followed deposition of every layer, whereas drainage was allowed only at the surface. The following layer was placed after fully instrumenting the previous one. Sensors embedded in the soil model included miniature pore pressure transducers, modified pore pressure transducers that served as 52.5 m 4% 26.6 m Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Evaluation of mine waste properties is undoubtedly essential, but obtaining representative samples and performing laboratory tests under realistic conditions is a challenge, particularly because waste rock particles can reach dimensions larger than 3 m (Wickland et al. 2010). Centrifuge modeling allows scaling of the rock particles and enables the simulation of layered disposal, consolidation and earthquake loading of the materials under prototype stress levels. In this study, tailings alone and uniformly mixed deposits were constructed, instrumented, and tested in terms of consolidation, liquefaction potential, and slope stability on the centrifuge at the Center for Earthquake Engineering Simulation at Rensselaer Polytechnic Institute in Troy, New York. Mixtures were designed in terms of ratio of waste rock to mine tailings by dry mass, aiming to create a rock skeleton “just-filled” with tailings. Slow cyclic and postcyclic monotonic triaxial tests performed by Wijewickreme et al. (2010) demonstrated that the rock skeleton predominantly carried the loads in that configuration. Wickland (2006) suggested that mixture ratio and tailings water content are the most manageable mixture design variables. The latter was not varied for the purposes of this research. Layer 4 Layer 3 Layer 2 4.7 m Layer 1 Direction of Dynamic Loading Pore Pressure Transducer Accelerometer Settlement Gauge LVDT Pair of Bender Elements Fig. 1. Key prototype dimensions and model configuration in stacked-ring rectangular container mounted on shaker prior to dynamic loading © ASCE 04017099-2 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 2. (a) Comixing mine tailings and waste rock at specified ratio prior to disposal; (b) pair of bender elements attached to individual aluminum mounts and embedded in the mixture during layer disposal and consolidation settlement gauges (Antonaki et al. 2013, 2014), and accelerometers. Additional fixed-depth pore pressure transducers were attached to a vertical ruler that was then loosely taped against the middle of one of the long container walls, as depicted in Fig. 1. Bender element pairs were installed in vertical aluminum columns and glued to opposite sides of a rigid container during static Tests 1 and 2. Shear wave velocity measurements were thus obtained for the mine tailings, and bender elements were not used again until Tests 8 and 9. Because a deformable container was used in these tests, bender elements had to be installed in individual mounts that were embedded in the soil as exhibited in Fig. 2(b). As shown in Fig. 1, bender elements were not placed in the top (fourth) layer because the confinement was not sufficient to guarantee reliable data. The 2∶1 mixture was not stiff enough to hold the mounts in place, and shear wave velocity data is not available for Test 7. After construction was complete, a centrifugal acceleration of 80 g was applied to all models to simulate the desired height and corresponding stress level. Consolidation was monitored until completion, and models were subsequently spun down. Consolidation duration was in the range of 1.3–3.4 years, or a few hours on the centrifuge. Table 1 summarizes test parameters including consolidation time for each of the presented tests. As expected, mine tailings exhibited slower consolidation rates than mixtures. The necessary steps before dynamic loading included stopping the centrifuge, draining the water that collected at the surface, recording the observed settlement, removing surficial material for a 4% slope excavation, detaching the rigid end walls to allow deformation, and placing additional instrumentation. In practice, thickened tailings can be deposited to form a gentle slope (between 2 and 6%) (Sanin et al. 2012). Beach slopes typically range between 0.2 and 1.5% when tailings are not highly thickened and between 2 and 10% for high-density thickened tailings (HDTT) and paste (Engels 2002). A view of the mild slope created at the surface of a layered mixture model after consolidation is given in Fig. S1(a). Sensors added included accelerometers and LVDTs glued on opposite ends of the sliding rings that constituted the container. The final configuration is presented in Fig. 1. Additionally, four targets were placed at the slope surface and monitored via video cameras for slope displacement. A harmonic motion consisting of 50 cycles at a frequency of 0.9 Hz was selected as the base excitation. Acceleration amplitude increased during the first 10 cycles, was maintained constant for the next 30, and then decreased during the last 10 cycles. Maximum acceleration could be adjusted, and two values were targeted in this testing sequence: roughly 0.10 and 0.25 g. The motion adjusted to approximately 0.10 g will henceforth be referred to as Motion 1, and the second © ASCE Table 1. List of Presented Centrifuge Tests and Main Parameters Test number 5 6 7 8 9 Material Consolidation time (years) Applied motions amax (g) Mine tailings Mine tailings 2.0∶1 mixture 3.0∶1 mixture 2.4∶1 mixture 3.3 3.4 1.3 1.3 1.3 1 2 1 and 2 1 and 2 1 and 2 0.10 0.26 0.07 and 0.24 0.1 and 0.21 0.13 and 0.24 and stronger motion will be referred to as Motion 2. Time was allotted for pore pressure dissipation before and after the motions were applied. Because of low permeability and shear strength, only one motion was applied to each mine tailings model, whereas both motions were applied to all three mixed models, as stated in Table 1. Slight discrepancies in acceleration amplitude were attributed to small differences in model mass and stiffness as well as expected variation in shaker response. Fig. S1(b) depicts a vertically excavated mixture deposit after a test was concluded. The bottom pore pressure transducer was revealed as the excavation progressed. Results Results from the consolidation and dynamic loading phases of Tests 5–9 are presented in prototype scale and discussed in the following sections. Consolidation All deposits consolidated until excess pore pressure dissipated, although some evaporation was still taking place as the centrifuge was spinning. Pore pressure transducers fixed at the bottom of the container and on the vertically placed ruler were used to assess the end of consolidation. As stated in Table 1, consolidation of mine tailings lasted approximately 3.4 years, whereas consolidation of mixed deposits ended markedly sooner, approximately 1.3 years after construction was concluded. Waiting time on the centrifuge was kept the same for all three mixtures, even though mixtures with higher waste rock content consolidated slightly faster. The final dissipation rate was close to 0.01 kPa=day at the bottom of all deposits. Fig. 3 compares the pore pressure dissipation rate of a mine tailings deposit and three mixed deposits at two different depths. Pore pressure has been normalized with its initial value to facilitate the comparison between different materials and model heights. Fig. 3(a) shows pore pressure measured by a fixed sensor 04017099-3 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 4. Shear wave velocity profiles of mine tailings and 2∶1, 2.4∶1, and 3∶1 mixtures after 1.25 years of consolidation (prototype scale) Fig. 3. Pore pressure dissipation normalized with maximum value during consolidation of mine tailings and 2.4∶1, 3∶1 mixtures at depth of (a) Layer 3; (b) Layer 2 (prototype scale) in the proximity of the third layer surface, which corresponded to approximately 7 m from the surface of all deposits with the exception of the 2∶1 mixture. The thickness of the 2∶1 deposit was a few meters less than that of the other mixtures, and thus the sensor was located 4.1 m from the surface. All normalized curves start at 100%, but different rates and proportions of dissipation were experienced by each material. Similarly, Fig. 3(b) shows pore pressure measured at the level of the second layer, or 12 m from the surface. The depth of the same sensor was 9.3 m in the 2∶1 mixture. Some initial dissipation that occurred in the 2∶1 deposit was not captured because of a technical issue; hence, the 2∶1 mixture experienced somewhat more dissipation at the beginning than could be plotted, which would move the curve slightly lower than shown but would not markedly alter the observed trends. Pore pressure in the mine tailings dissipated 34–35% of its starting value at both presented depths [Figs. 3(a and b)]. The corresponding value was 7–8% for the 2∶1 mixture and approximately 7% for the 2.4∶1 mixture. The 3∶1 mixture data was corrected for evaporation because the top of the deposit was partially saturated by the end of the consolidation phase. Dissipation was within the range of 4–6% after the correction was applied. Similar findings were derived at other depths in terms of dissipation amounts as well as general trends. Mixtures consolidated rapidly because of lower fines and water content, which led to higher permeability and less development of excess pore pressure. There appears to be a shift in behavior when the mixture ratio rises above the just-filled value. Both the 2∶1 and 2.4∶1 mixture curves look qualitatively similar to those corresponding to tailings. They all consist of an initial steep fragment and a noticeably less steep second fragment. The 3∶1 mixture curves, however, are moderately sloped and resemble a straight line near parallel to the final part of the tailings dissipation © ASCE curves. The behavior approaches that of a coarse rock with very little pore pressure to dissipate. Consolidation that occurred while the centrifuge was accelerating was most likely a substantial portion of the total dissipation for the 3∶1 mixture. Therefore, the dissipation rate was already very low when centrifugal acceleration finally became constant. The 2∶1 and 2.4∶1 curves are enclosed by the tailings and 3∶1 curves but lie in closer proximity to the 3∶1 curves. Shear wave velocity was evaluated from bender element readings, and a comparison between tailings alone and mixtures at approximately 1.25 years after construction is presented in Fig. 4. Data is unavailable up to a 5–6-m depth below ground surface level (GSL) because of low confining stress and shear strength. Even though the tailings deposit continued to consolidate, shear wave velocity more or less remained constant after the first year. Shear wave velocity was observed to increase with depth in all three cases. The rate of increase with depth was more pronounced in the 3∶1 mixture. Only two data points were available for the just-filled 2.4∶1 deposit because the top bender pair was found slightly rotated and measurements were disregarded. Shear wave velocity for that mixture was almost exactly equal to the average of the other two near the bottom and closer to the shear wave velocity profile of the 3∶1 mixture at shallower depths. The mine tailings profile started at 118 m=s and increased to 160 m=s between 7 and 18 m from the surface, whereas the V s range for the 3∶1 mixture was 153–268 m=s between 6 and 18 m. The available range for the 2.4∶1 deposit was 194–208 m=s between 12 and 18 m. The 2∶1 shear wave velocity profile would presumably lie between the mine tailings and 2.4∶1 curves. Fig. 4 confirms the expected increase in stiffness with the addition of waste rock to the tailings. Dynamic Loading The dynamic response of the tailings and mixed deposits to Motions 1 and 2 is explored in terms of soil acceleration, lateral displacement, and pore pressure buildup with depth. Plotted in Fig. 5 are soil accelerations measured at the surface of the deposits during Motion 1. Base excitation is included in the graphs to demonstrate the modification of the acceleration amplitude and frequency content as it traveled through the deposits. Waste rock 04017099-4 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 5. Motion 1: soil acceleration at a small depth from the surface plotted with base excitation for (a) mine tailings; (b) 2∶1 mixture; (c) 2.4∶1 mixture; (d) 3∶1 mixture (prototype scale) content increases from top to bottom, with the uppermost graph of Fig. 5 corresponding to tailings alone and the bottom graph corresponding to the 3∶1 mixture. Mine tailings were highly liquefiable, as is indicated by the decrease in soil acceleration and the period elongation captured in Fig. 5(a). In fact, the slope liquefied before the maximum base acceleration was reached. It is visible that the soil (in black) stopped being able to follow the base movement (in gray) after only a few cycles, both in terms of acceleration amplitude and frequency of oscillation, because loss of strength and stiffness were caused by liquefaction. The response changed drastically with the rise in mixture ratio. The 2∶1 mixture exhibited asymmetrical dilative spikes in the direction toward the top of the slope, as seen in the second graph of Fig. 5. At the same time, it experienced a rapid decrease in acceleration and concurrent period elongation, but only after the maximum base acceleration was applied and amplified by the soil. The 2.4∶1 mixture behaved similarly, but the base excitation amplitude was 86% higher than that applied to the 2∶1 mixture (Table 1). That led to sooner and higher amplification of the base motion, followed by deamplification as the surficial material started liquefying. A slower, more gradual decrease in acceleration amplitude paired with lower simultaneous spikes asymmetrically pointing upslope indicated lower strains than those experienced by the 2∶1 mixture. Signs of dilative behavior were present in the acceleration of the 3∶1 deposit in the © ASCE Fig. 6. Motion 1: soil acceleration at the depth of the first layer plotted with base excitation for (a) mine tailings; (b) 2∶1 mixture; (c) 2.4∶1 mixture; (d) 3∶1 mixture (prototype scale) form of increasingly pronounced spikes (Zeghal and Elgamal 1994). High amplification occurred through the soil profile, and soil acceleration amplitude continued to rise during the entire motion. The behavior was evidently dominated by the rock skeleton once the tailings became less than enough to fill the voids. Plotted in Fig. 6 is soil acceleration deeper within the deposits during Motion 1. Base excitation is included in these graphs as well. Prior to dynamic loading, all deposits had an average height between 18.5 and 19 m, with the exception of the 2∶1 deposit that had an average height of 17.4 m. At a depth of 13–14 m from the surface, which corresponded to Layer 1, all deposits were capable of sustaining the applied base excitation and only presented small amplification of the motion. It should be taken into account that the highest base acceleration was applied to the 2.4∶1 mixture and the lowest to the 2∶1 mixture (Table 1). They are, however, considered comparable for the purposes of this study. Figs. 8 and S2 analogously compare the response of the four deposits when subjected to the stronger Motion 2. The improvement in dynamic response because of comixing with waste rock became more evident when a stronger earthquake was applied. At 1.9 m from the surface, the mine tailings liquefied after four loading cycles, whereas period elongation was noticeable even before the soil acceleration started decreasing (Fig. S2). The material at that depth completely lost its strength and stiffness, as was indicated by its inability to transmit any acceleration. At 3 m from its 04017099-5 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 7. Motion 2: soil acceleration at the depth of the second layer plotted with base excitation for (a) mine tailings; (b) 2∶1 mixture, and at the depth of the first layer for (c) 2.4∶1 mixture; (d) 3∶1 mixture (prototype scale) surface, the 2∶1 mixture started exhibiting dilative spikes associated with simultaneous pore-pressure drops (Zeghal et al. 1996; Zeghal and Elgamal 1994) after approximately five acceleration cycles. The spikes were asymmetrically higher toward the top of the slope, indicating that the main direction of soil deformation was downslope. Only a few cycles later, this material also liquefied completely, and the acceleration reading became zero. Spikes appeared in the 2.4∶1 acceleration record at 1.4 m from the surface at approximately the same time as in the 2∶1, but the soil required more loading cycles to substantially lose strength. Period elongation became visible in both mixtures as their stiffness started to drop. Comparably to Motion 1, the 2.4∶1 mixture exhibited lower asymmetrical spikes, which were present until the end of the base excitation, indicating some residual stiffness as the deposit continued to accumulate shear strain. Conversely to the deterioration in the response of the first three deposits because of the increase in acceleration amplitude, the 3∶1 deposit behaved practically identically to Motion 1 at shallow depths. In Fig. 7, soil acceleration at a larger depth is plotted with Motion 2 for the four deposits. The response of the tailings deposit appears only slightly improved compared to Fig. S2 because it was able to sustain two more loading cycles at this depth. Soil acceleration dropped to zero, suggesting that liquefaction reached a minimum depth of 10.7 m. A small improvement is also visible in the 2∶1 acceleration, during which the material required a few more cycles to completely © ASCE liquefy at 9.6 m from the surface. The closest available accelerometers located in the 2.4∶1 and 3∶1 deposits were somewhat deeper, but that does not affect the conclusions that can be drawn from the comparison. The 2.4∶1 acceleration record at 12 m shows mostly dilative behavior of the material at that depth. Some loss of strength and stiffness occurred near the end of dynamic loading, and the spikes remained asymmetrical toward the top of the slope. Acceleration measured at 7.2 m showed high spikes earlier on, followed by zero positive acceleration with negative spikes near the end of the motion, indicating that liquefaction reached a depth of 7.2 m in the 2.4∶1 mixture. The 3∶1 mixture essentially followed the base movement at 13.2 m from its surface. The depth is not directly comparable to that of the plotted tailings and 2∶1 records, but the next available record was located at a depth of 5 m and looked identical to the bottom graph of Fig. 7, if slightly amplified. It can be concluded that the 3∶1 mixture did not liquefy at any depth, irrespective of the base excitation intensity. Fig. 8 illustrates the performance of the four materials in terms of residual horizontal displacement in the direction of dynamic loading. Data was obtained from horizontal LVDTs glued on the container rings. Final deformed profiles after application of Motion 1 are plotted in Fig. 8(a). Positive displacement denotes movement toward the toe of the slope. Maximum final displacements are quite small, ranging from 15 cm downslope experienced by the tailings deposit to 31 cm upslope experienced by the 2.4∶1 mixture. All mixtures permanently deformed toward the head of the slope, despite the fact that the slope surface moved toward the toe in all cases. Video tracking of surface targets using TEMA software gave an estimate of horizontal slope displacement, which was in the region of 1 m for the mine tailings and the 2∶1 and 2.4∶1 mixtures, and in the region of 10 cm for the 3∶1 mixture. These values are approximate but give a good estimate of the slope surface displacement as opposed to the profile deformation, which was measured by LVDTs along the container rings and did not reflect the displacement of surficial material during slope failure. Four targets were typically used, and average displacement is reported. From the shape of the deformed profiles in Fig. 8, the top 6–7 m of the tailings deposit and the top 2–3 m of the 2∶1 deposit appeared to be liquefied and isolated from the deposit below that depth. A similar observation cannot be made for the 2.4∶1 and 3∶1 deposits, which presented a more uniform distribution of lateral displacement. Deformed profiles only due to application of Motion 2 are plotted in Fig. 8(b). The mine tailings deposit visibly deformed downslope, with a maximum value of 68 cm at the surface. The 2∶1 mixture deformed in the same direction with a maximum value of 40 cm. The top few meters appear isolated from the rest of the deposit in both cases, but displacements are sizeable below as well. The 2.4∶1 mixture only slightly deformed during Motion 2, whereas the 3∶1 mixture exhibited an almost identical response as to Motion 1. Slope surface displacements obtained from video tracking were in the region of 1 m for tailings and the 2∶1 mixture, 50 cm for the 2.4∶1, and 20 cm for the 3∶1. Motions 1 and 2 were applied as a sequence to the mixed deposits, and cumulative displacement profiles are displayed in Fig. 9. Mixture response was comparable in terms of maximum values, but the existence of the 4% slope was only sufficient to bias the 2∶1 deformation direction. The deformed profiles of tailings for Motions 1 and 2 were not combined because they came from two different tests (Tests 5 and 6), but if added, the maximum displacement would be 83 cm as opposed to the mixture range of 24–27 cm. Accumulated soil surface displacements obtained from video tracking were in the range of 2 m for the tailings and 2∶1 deposits, 1.5 m for the 2.4∶1, and 30 cm for the 3∶1 deposit, all toward the toe of the slope. 04017099-6 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 8. Horizontal displacement profiles for mine tailings for 2∶1 mixture, 2.4∶1 mixture, and 3∶1 mixture due to (a) Motion 1; (b) Motion 2 (prototype scale) Fig. 9. Final horizontal displacement profiles for 2∶1 mixture, 2.4∶1 mixture, and 3∶1 mixture (prototype scale) Figs. 10 and S3 illustrate the development of excess pore pressure with depth during application of Motions 1 and 2, respectively. Waste rock content increases from top to bottom in the figure. Figs. 10(a) and S3(a) corresponds to mine tailings and Figs. 10(d) and S3(d) corresponds to the 3∶1 mixture ratio. Markers denote data points obtained from floating pore pressure transducers, and corresponding curves denote continuous pore-pressure buildup profiles that were estimated using the displayed points. The black solid line denotes buildup at the end of dynamic loading. Vertical effective stress at rest is included in the graphs to enable evaluation of liquefied depth. Unit weight and water level varied to some extent between materials as well as with time, and effective stress was assessed separately for every deposit and motion. As shown in Fig. 10(a), pore pressure gradually built up in the tailings deposit, and by the end of Motion 1, the top 6 m or so had © ASCE liquefied. This observation is consistent with the deformed profile of Fig. 8(a). The same is plotted in Fig. 10(b) for the 2∶1 mixture, the top 4 m of which experienced complete liquefaction. The top 2–3 m appeared separated in Fig. 8(a), but the next available LVDT reading was located at a depth of 8 m, which prevented a more accurate estimation of the liquefied depth from the deformed profile. Considerable pore pressure built up below the liquefied depth. In this case, most of the buildup happened within the first 15 s, and the maximum occurred at approximately 50 s. Because of an increased permeability of the mixture compared to tailings, pore pressure started dissipating as soon as acceleration amplitude started decreasing. The shape of the profiles in Fig. 10(c) looks quite different, which can be attributed to the more permeable continuous arrangement of rock particles in the just-filled case of the 2.4∶1 mixture ratio. Most of the buildup transpired within the first 10–15 s, and it appeared almost as if there was double drainage, at the top and the bottom of the deposit. After the first 15 s of Motion 1, the bottom half of the deposit started to catch up, and maximum buildup occurred at 50 s, when dissipation was initiated. The material did not liquefy, although the top half came close to zero effective stress conditions. The surface acceleration depicted in Fig. 5 did exhibit signs of liquefaction. Fig. 10(d) clearly displays that the 3∶1 mixture did not come close to losing its strength at any depth. Pore pressure buildup did transpire but was much lower than in the other deposits. It mostly took place within the first 10–15 s, and the maximum occurred at 15–30 s. The top 14.5 m—at least—of the tailings deposit liquefied during Motion 2, as shown in Fig. S3(a). The top half of the deposit liquefied during the first 15 s, and the overall maximum occurred at 50–60 s. The response of the 2∶1 mixture, as captured in Fig. S3(b), looks very similar. The top 14 m at a minimum liquefied. Most of the buildup took place within the first 30 s, and the maximum occurred at 50 s. As when subjected to Motion 1, pore pressure in the 2∶1 deposit started dissipating slowly when acceleration amplitude started decreasing. Fig. S3(c) depicts pore pressure buildup in the 2.4∶1 deposit during Motion 2. The shape of the profiles was similar to that during Motion 1, resembling double drainage conditions; however, pore pressure built up more steadily until complete liquefaction of the top half took place at approximately 50 s. In this case, 04017099-7 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Fig. 10. Motion 1 (0.10 g): evolution of pore pressure buildup plotted with effective stress profile for (a) mine tailings; (b) 2∶1 mixture; (c) 2.4∶1 mixture; (d) 3∶1 mixture (prototype scale) dissipation did not initiate before the end of dynamic loading, and maximum pore pressure buildup values occurred at 60 s. Fig. S3(d) demonstrates that, consistently with measured accelerations and displacements, the 3∶1 mixture behaved almost identically when subjected to Motions 1 and 2. Some buildup was measured but remained far from liquefaction throughout dynamic loading. The maximum occurred at 50 s, and some dissipation became noticeable by the end of the motion. The four materials are compared with respect to settlement corresponding to each phase of the centrifuge tests in Table 2. Settlement was normalized with deposit height during each phase to allow for meaningful comparisons between deposits. The primary observation is that consolidation settlement decreased when waste rock content increased. The difference was very noticeable between mine tailings and the 2∶1 mixture, from 33 to 15%, respectively, but not as large when more waste rock was added Table 2. Settlement Normalized with Model Height for Every Testing Phase and Material Consolidation Dynamic loading Motion 1 Material Per layer (20 g) (%) Complete (80 g) (%) Total (%) Head (%) Tailings 2.0∶1 2.4∶1 3.0∶1 20 11 7 4 13 4 2 2 33 15 9 5 4 3 5 4 © ASCE 04017099-8 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 Motion 2 Toe (%) 2 −1 0 3 Head (%) 7 3 3 4 Toe (%) −3 −1 0 3 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. to the mix. This remark is consistent with pore pressure dissipation trends previously discussed and shown in Fig. 3. The three mixtures exhibited comparable behavior, whereas the difference with mine tailings alone was more distinct. Most consolidation-induced settlement took place during layer consolidation at 20 g. Specifically, 61% of the total settlement occurred during layer consolidation of mine tailings. That proportion increased to 73% for a 2∶1 mixture ratio, and to 78 and 80% for 2.4∶1 and 3∶1, respectively. That was to be expected because the mixture permeability and consolidation rate rose with an increasing mixture ratio. The bottom half of Table 2 lists the settlements that occurred during dynamic loading at the head and the toe of the slopes. The differences between material performances in terms of earthquake-induced settlement might appear negligible, but they were notable when considering stability of a 4% slope. Negative settlement denotes apparent rise in elevation occurring at the toe of the slope when some of the surficial material moved downslope. During both motions, settlement at the head of the slope was larger than at the toe in all deposits, denoting some reduction of the slope inclination. Tailings settlement at the head was 4% for Motion 1 and 7% for Motion 2. Settlement at the toe of the slope was 2% because of Motion 1, whereas a 3% uplift was caused by Motion 2, resulting in 2.8 and 0.6% final inclinations, respectively. The 2∶1 deposit experienced 3% settlement at the head and 1% uplift at the toe during both motions. Inclination was approximately equal to 2.8% after Motion 1 and 1.4% after Motion 2. The second motion was applied to the same mixed deposits after pore pressure buildup caused by Motion 1 dissipated and was hence applied to a 2.8% slope in this case. The 2.4∶1 deposit settled 5 and 3% at the head of the slope for Motions 1 and 2, correspondingly, whereas the toe did not settle or elevate for either motion. The aforementioned settlement resulted in 3.4 and 2.4% inclination, respectively. The 3∶1 deposit responded similarly to both motions in terms of settlement, experiencing 4% at the head and 3% at the toe and settling to 3.5 and 3.1% inclinations. Slope stability undoubtedly showed considerable improvement as waste rock content increased. Conclusions Comixing of waste rock and tailings was found to be beneficial both in terms of consolidation rate and settlement as well as dynamic response and slope stability. The mixture ratio of waste rock to tailings by dry mass to achieve contact between rock particles by simply filling the voids with tailings was estimated at 2.3∶1. Consolidation time and settlement were reduced by approximately 2.5 times when comixing at a ratio of 2∶1 was applied. At higher ratios, mixtures consolidated even faster, and settlement was further reduced. At 1.25 years, shear wave velocity of the 3∶1 mixture was 1.3–1.6 times higher than that of the tailings within the same depth range (6–18 m), whereas the 2.4∶1 mixture lied in between. Further increase with time was not measurable, even though consolidation continued to occur. The improvement in dynamic response was not pronounced when the tailings proportion was larger than required to fill the rock skeleton. That highlighted the importance of accurately estimating the theoretically optimum mixture ratio (just-filled), particularly when the objectives include both shear strength increase and saturation of the waste rock for AMD reduction. Mine tailings and three mixed deposits were subjected to two levels of harmonic base excitation, with average acceleration amplitudes of 0.10 and 0.24 g. The tailings and 2∶1 mixture liquefied in both cases. The liquefied depth was at least 25–30% of the deposit depth for Motion 1 and roughly 80% for Motion 2. The 2∶1 mixture exhibited some © ASCE signs of dilation in the acceleration data. Only the surface of the 2.4∶1 deposit showed signs of liquefaction for Motion 1, whereas 45% of the profile liquefied during Motion 2. The 3∶1 deposit exhibited strictly dilative behavior regardless of base excitation intensity. Residual horizontal container displacement induced by Motion 1 was low for all deposits, and the improvement with rise in mixture ratio became substantial only when base acceleration was higher. Video tracking showed that all slopes moved toward the toe, but deformation of the 3∶1 slope was practically negligible. Slope performance was only satisfactory for mixtures 2.4∶1 and 3∶1 when subjected to Motion 1 but markedly superior for the 3∶1 deposit when subjected to Motion 2. The 3∶1 ratio, however, does not satisfy saturation of the waste rock skeleton, thus preventing water seepage and AMD. The possibility of partially or completely eliminating the retaining structure when tailings are mixed with waste rock at a just-filled configuration is intriguing because it would be very advantageous to the mining industry. Additional centrifuge tests are planned to investigate the feasibility of free-standing or partially supported mixtures deposited at their angle of repose. Acknowledgments The authors would like to acknowledge Golder Associates Ltd. for partial funding of this research. Supplemental Data Figs. S1–S3 are available online in the ASCE Library (www. ascelibrary.org). References Antonaki, N., Sasanakul, I., Abdoun, T., Sanin, M. V., Puebla, H., and Ubilla, J. (2013). “Centrifuge modelling of the consolidation evolution of fine-grained mine tailings.” Proc., Tailings 2013: 1st Int. Seminar on Tailings Management, Gecamin, Chile. Antonaki, N., Sasanakul, I., Abdoun, T., Sanin, M. V., Puebla, H., and Ubilla, J. (2014). “Centrifuge modeling of deposition and consolidation of fine-grained mine tailings.” Proc., Geo-Congress: Geo-Characterization and Modelling for Sustainability, ASCE, Reston, VA. Blight, G. (2009). Geotechnical engineering for mine waste storage facilities, Taylor & Francis Group, London. Brawner, C. O. (1978). “Concepts and experience for subsurface storage of tailings.” Proc., 2nd Int. Tailings Symp., Tailings Disposal Today, Miller Freeman Publications, San Francisco, 153–177. Bussiere, B. (2007). “Hydro-geotechnical properties of hard rock tailings from metal mines and emerging geo-environmental disposal approaches.” Can. Geotech. J., 44(9), 1019–1052. Engels, J. (2002). “Beach slope prediction methods.” 〈http://www.tailings .info/technical/beach-slope.html〉 (May 18, 2016). Leduc, M., and Smith, M. E. (2003). “Tailings co-disposal: Innovations for cost savings and liability reduction.” Latin America Mining Record, Sedalia, CO. Morris, P. H., and Williams, D. J. (1997). “Co-disposal of washery wastes at Jeebropilly colliery, Queensland, Australia.” Trans. Inst. Min. Metall. Sect. A Min. Ind., 106, A25–A29. Rico, M., Benito, G., Salgueiro, A. R., Díez-Herrero, A., and Pereira, H. G. (2008). “Reported tailings dam failures: A review of the European incidents in the worldwide context.” J. Hazard. Mater., 152(2), 846–852. Sanin, M. V., Humberto, P., and Terry, E. (2012). “Cyclic behaviour of thickened tailings.” Proc., Tailings and Mine Waste 2012, Univ. of British Columbia, Vancouver, BC, Canada. 04017099-9 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng. Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 10/26/17. Copyright ASCE. For personal use only; all rights reserved. Taylor, R. E. (1995). Geotechnical centrifuge technology, Blackie Academic and Professional, London. TEMA version 3.5-016 [Computer software]. Image systems, Linköping, Sweden. Vick, S. G. (1990). Planning, design, and analysis of tailings dams, BiTech Publishers, Vancouver, BC, Canada. Wickland, B. E. (2006). “Volume change and permeability of mixtures of waste rock and fine tailings.” Ph.D. dissertation, Univ. of British Columbia, Vancouver, BC, Canada. Wickland, B. E., and Wilson, G. W. (2005). “Self-weight consolidation of mixtures of mine waste rock and tailings.” Can. Geotech. J., 42(2), 327–339. Wickland, B. E., Wilson, G. W., and Wijewickreme, D. (2010). “Hydraulic conductivity and consolidation response of mixtures of mine waste rock and tailings.” Can. Geotech. J., 47(4), 472–485. © ASCE Wickland, B. E., Wilson, G. W., Wijewickreme, D., and Klein, B. (2006). “Design and evaluation of mixtures of mine waste rock and tailings.” Can. Geotech. J., 43(9), 928–945. Wijewickreme, D., Khalili, A., and Wilson, G. W. (2010). “Mechanical response of highly gap-graded mixtures of waste rock and tailings. II: Undrained cyclic and post-cyclic shear response.” Can. Geotech. J., 47(5), 566–582. Zeghal, M., Elgamal, A. W., and Parra, E. (1996). “Analyses of site liquefaction using downhole array seismic records.” Proc., 11th World Conf. on Earthquake Engineering, Pergamon, Oxford, U.K. Zeghal, M., and Elgamal, A.-W. (1994). “Analysis of site liquefaction using earthquake records.” J. Geotech. Eng., 10.1061/(ASCE)0733-9410 (1994)120:6(996), 996–1017. 04017099-10 J. Geotech. Geoenviron. Eng., 2018, 144(1): 04017099 J. Geotech. Geoenviron. Eng.