A s Bingham Canyon Mine Pit Wall S des

Proceedings Tailings and Mine Waste 2016 Keystone, Colorado, USA October 2-5, 2016 A s Bingham Canyon Mine Pit Wall S des M.A. Llano-Serna The University of Queensland, Brisbane, QLD, Australia D.. Williams The University of Queensland, Brisbane, QLD, Australia M.R. Ruest The University of Queensland, Brisbane, QLD, Australia Bingham Canyon Mine is the largest open pit worldwide, measuring 1 km deep by 4 k!"1# $% 11&' # (% 1& #)* +&1", - # /#)%0 ( 2*2 $*(%3 a geotechnical fault on its %()th-eastern wall. Monitore $** (d% 2 -$ $00*)$ % he preceding weeks, from 1 to 5 mm/day, and ()k)2 ) d$0w$ -) ) %( 0$2w$* 2, although the extent of the slide was much larger than anticipated and considerable haulage equipment was buried. About 15& Mt of material moved and debris covered two- -)2 (t - pit base. The mine was op)$ (%$* $3$% -% $e(w 14 months. T- ( 2*2 ) $00(mpanied by seismic events (t $3% w 5.1 and magnitude 4.9, respectively. A further seis0 event was recorde after the second slide, plus 1' additional seismic events. A landslide analysis was applied to simulate the two slides, and the results matched the reported extent of the slides )$2(%$e*r ** 1 INRODUCTION This paper presents a preliminary landslide analysis of Kennecott Utah Copper's Bingham Canyon Mine pit wall slides, which occurred along a geotechnical fault on its north-eastern wall at 9.31pm and 11:06pm on 10 April 2013. Monitored wall movements had accelerated in the preceding weeks from 1 to 5 mm/day, and workers were evacuated to avoid any casualties. However, considerable haulage equipment was buried below about 150 Mt of material that moved. These slides highlight the importance of improving predictive tools. The significant difference between the expected and observed performance is also important. This paper focuses on the numerical simulation of landslides such as the one described therein. Catastrophic events such as open pit failures have huge economic consequences, and the potential for loss of life. Roughly two open pit failures occur worldwide each year. This failure rate is more than two orders of magnitude higher than the failure rate of conventional civil engineering excavations. Even when a potential landslide can be predicted, the question of how far the debris may travel remains. The answer to this question is critical to preventing further losses and mitigating the loss of life hazard (Llano-Serna et al. 2015). The modelling of large-scale run-out processes may be classified numerically as a finite deformation problem, making them very complex and difficult to solve. During run-out, compaction, and localised failures can occur, where stresses reach threshold levels. Popular modelling methods include the discrete element method (DEM), smoothed particle hydrodynamics (SPH), and the material point method (MPM). In the DEM, each particle is considered discretely; hence, the macroscopic behaviour cannot reasonably be directly modelled. On the other hand, the SPH and MPM are derived from continuum mechanics, which allows the use of conventional geotechnical constitutive models. It also means that they may be implemented in current finite element methods. Advantages and disadvantages of methods used to model runout are described by Soga et al. (2015). 787

Failures 52# - * $ (%2 $ %w)0$* 2w*$ (% $r #)2%, - w2 (t $ -$ 0$* 0-6 niques is invaluable in understanding the kinematics of a landslide. It can provide, for example, information for use in risk assessments, such as the delineation of the area that would potentially be affected by a landslide run-out. It enables the quantification of potential losses of human life, infrastructure, and environmental assets. The delineation of the extent and run-out of an open pit slope failure also enables planning for the safe evacuation of the pit and rapid recovery post-failure. 2 BINGHAM CANYON MINE The Bingham Canyon Mine is located 30 km south-west of Salt Lake City, Utah, USA, and measures 1 km deep by 4 km wide. It is the largest man-made excavation in the world. The two distinct slides, separated by approximately 1.5 hours, were detected by a comprehensive seismic and infrasound network installed in the region. The surface wave-detection magnitudes were 5.1 and 4.9 for the first and second events, respectively (Hibert et al. 2014). A further seismic event was recorded 1 min after the end of the second slide, and 16 additional seismic events were detected and located in the mine area. This incident may be the only large-scale event to have triggered several small earthquakes (Pankow et al. 2013). Using seismic observations of durations, peak amplitudes of the seismogram envelopes, and the area underneath each envelope, Pankow et al. (2013) also determined that both slides had roughly the same volume. Later, Hibert et al. (2014) used the inversion method to determine the slide force history (LFH analysis; Ekstrom & Stark 2013). The method is based on the assumption that a slide seismic source can be described as a time-varying force taking into account the long-period signals. Based on the displacement history presented by Hibert et al. (2014), it is possible to draw some interesting conclusions. Firstly, the first slide caused the onset of mobilisation of the second slide at a higher elevation. Essentially, the first slide constituted the passive wedge in the whole slide mechanism. As a consequence, the second slide, comprising the active wedge was triggered. Secondly, the LFH analysis enables the progression of the centre of mass to be determined. These observations are depicted in Figure 1. 3 NUMERICAL MODEL The numerical framework used therein is based on the approach proposed by Llano-Serna et al. (2015). The methodology has been successfully used to model run-out processes in natural slopes, highlighting the applicability of the MPM. Note that this framework focuses on the sequential ground flow after initiation of the slope failure. One of the main advantages of the method relies on its capacity to take advantage of results of traditional slope stability models or field observations. It enables results to be obtained quickly from a computational point of view. Usually, the cross-section for the simulation is composed of the sliding slope. The part that is not expected to slide during the run-out is considered as a stiff body. In this study, an initial two-dimensional idealisation was made, based on the trajectory obtained by Hibert et al. (2014), overlapped on the topographic survey available in Google Earth Pro from 2007. It is noted in advance that this time lag accounts for a vertical difference of about 40 m between the cross-section based on Google Earth Pro from 2007 and the pit geometry at the time in the incident. Although it can be argued that significant geometrical changes occurred between 2007 and 2013, the main aim of the paper is to show how a very simple MPM model is useful for simulating the complex mechanics of run-out progression. The adopted cross-section is shown in Figure 2. The failure lines were defined based on photographic interpretation of Figure 1. The MPM size was defined as 5 m according to Llano-Serna et al. (2015) for the model presented therein. 788

Proceedings Tailings and Mine Waste 2016 Keystone, Colorado, USA October 2-5, 2016 F789:; <= Photograph of two Bingham Canyon Mine slides. 789

Failures H809 m G E C Second Slide RIJIK LNKO First Slide >??@ B Figure 2. MPM numerical model of cross-section of Bingham Canyon in slide area, based on Google Earth Pro from 2007. 3.1 Material Parameters As demonstrated by Llano-Serna et al. (2015) a total stress analysis is appropriate for the postfailure analysis of landslides. As described by Pankow et al. (2013), the Bingham Canyon Mine slides lasted roughly 90 s. From a mechanical point of view, this means that the cross-section of the true failure envelope on a deviatoric plane in stress space is considered circular. This means that the shear strength resistance may readily be described by a single shear strength parameter adopted in a total stresses analysis. The material parameters given in Table 1 were adopted for the analyses, which were derived from the results presented by Styles et al. (2011) and a backanalysis performed by the authors. Table 1. Material parameters used in analysis of Bingham Canyon slides. Parameter Symbol Value Unit Weight (kn/m 3 ) γ 23.0! Elastic Modulus (MPa) E 600 Shear Strength (MPa) τ 250 Poisson Ratio ν 0.33 The material parameters summarised in Table 1 were used for the analyses of both slides. It is not possible to build a more detailed model, including different material layers with different parameters, since the information available is rather limited. Nevertheless, despite the simplifications, the results obtained are satisfactory, as described in the following paragraphs. The simulation was carried out in two stages. The first slide was activated by increasing gravity up to g = 9.8 m/s 2. In this stage, the mass equivalent to the second slide is also considered as a rigid material; unable to flow down the slope. Once the first slide achieves a state of repose, the position of the material points corresponding to the first slide are imported into a second analysis in which the second slide is allowed to run-out under the action of gravity and interact with the deposited debris from the first slide. 3.2 Simulation of First Slide In Figure 3 it is possible to see how the first slide progresses down the slope. After the first 20 s, the slide achieves its maximum velocity (see also Fig. 4). The sliding mass reaches an average peak kinetic energy of about 50 kj and a maximum speed of about 35 m/s. After that, at approximately t = 30 s, the first particles reach the base of the pit, and a reduction in energy may be observed. The speed decreases and the movement stops at t = 60 s. At this po0int, the velocity of some of the particles decreases, while they are accommodated in the new topography and the movement gradually stops after t = 85 s. 790

Proceedings Tailings and Mine Waste 2016 Keystone, Colorado, USA October 2-5, 2016 Figure 3. Progression and changes of kinetic energy of first landslide. According to Pankow et al. (2013), the seismic records indicate that the slide duration was about 90 s, which is close to the value obtained in the simulations presented therein. The resulting velocities are also compared via computations using the LFH with the inversion method developed by Ekstrom & Stark (2013) (see Fig. 4). The inverse model of seismic data allows the interpretation of long- and short-period seismic signals generated by landslides, shedding light on the dynamics of a slope failure. From Figure 4, it can be seen that both techniques match the principal features and shape of the slide. For example, the maximum velocity, and s 791

Failures )$ 2 (t %0)$2%3 $% 0)$2%3 d*(0 r 2-(% % P3w) Q $ 0- )$2(%$e*r ** S(d), some differences are apparent, probably caused by constraints imposed by the two-dimensional model presented herein. Figure 4. Change in velocity with time during first slide. 3.3 Simulation of Second Slide The simulated progression and kinematic energy released in the second slide are presented in Figure 5. It can be observed that the kinetic energy increases until t = 20 s, when the advancing front reaches the accumulation of debris left after the first slide. At this stage, the maximum velocity also reaches its peak of about 30 m/s (see Fig. 6). At t = 40 s, the second slide reaches the depositional area from the first slide; then the velocity starts to decrease and after t = 75 s most of the kinetic energy is dissipated. From Figure 6, it can be noted that the estimates given by the MPM and LFH analyses are similar. The first 20 s are comparable both in shape and increasing rate of velocity. However, there is a difference of about 5 m/s in the maximum velocity. The initial descending slopes from both approaches are very similar, but are dislocated by approximately 10 s. The MPM shows a smoother transition to repose, while the LFH shows a smaller increase by the end of the slide, similar to Figure 4. 4 CONCLUSIONS This paper shows how the MPM can deal with large-strain problems such as open pit slides. Despite the limitations in the available input information and the simplicity of some of the numerical assumptions, the two-dimensional analyses presented herein were able to describe the principal features observed in Bingham Canyon Mine pit wall slides. The numerical simulations presented provide the details of the interaction between the two slides with the surrounding topography. They also make possible the calculation of the energy released in such events and the estimation of the extent of the slides. The accurate prediction of 792

Proceedings Tailings and Mine Waste 2016 Keystone, Colorado, USA October 2-5, 2016 )w%6(w 2 $%02 % *$%2*2 )$%2 $% 22w - $%$*r22 #)2% -)% $r -*# ( elineate the extent of a run-out in open pits. This feature would enable an improved risk assessment to assist in the planning of the safe evacuation of personnel and equipment from the pit and its efficient recovery post-failure. Figure 5. Progression and changes of kinetic energy of second slide. 793

Failures Figure 6. Change in velocity with time during second slide. 5 ACKNOWLEDGMENT The authors thank Dr Clément Hibert for his assistance in retrieving the LFH data used to validate the landslide model described and applied herein. 6 REFERENCES Ekstrom, G. & Stark, C.P. 2013. Simple scaling of catastrophic landslide dynamics. Science, 339(6126), 1416 1419. Hibert, C., Ekström, G. & Stark, C.P. 2014. Dynamics of the Bingham Canyon Mine landslides from seismic signal analysis. Geophysical Research Letters, 41(13), 4535 4541. Llano-Serna, M.A., Farias, M.M. & Pedroso, D.M., 2015. An assessment of the material point method for modelling large scale run-out processes in landslides. Landslides, 1 10. Pankow, K.L., Moore, J.R., Mark Hale, J., Koper, K.D., Kubacki, T., Whidden, K.M. & McCarter, M.K. 2013. Massive landslide at Utah copper mine generates wealth of geophysical data. GSA Today, 24(1), 4 9. Soga, K., Alonso, E., Yerro, A. & Bandara, S. 2015. Trends in large-deformation analysis of landslide mass movements with particular emphasis on the material point method. Géotechnique, 1 26. Styles, T., Stead, D., Eberhardt, E., Rabus, B. Gaida, M. & Bloom, J. 2011. Integrated numerical modelling and Insar monitoring of a slow moving slope instability at Bingham Canyon Mine. In: Proceedings of International Symposium on Rock Slope Stability in Open Pit Mining and Civil Engineering. Vancouver, Canada, 18-21 September 2011, 14 p. 794