Rapid flow slides of coal-mine waste in British Columbia, Canada
Published:January 01, 2002
- PDF LinkChapter PDF
O. Hungr, R.F. Dawson, A. Kent, D. Campbell, N.R. Morgenstern, 2002. "Rapid flow slides of coal-mine waste in British Columbia, Canada", Catastrophic Landslides, Stephen G. Evans, Jerome V. Degraff
Download citation file:
Failures of waste piles in open-pit coal mines of mountainous southeastern British Columbia produce highly mobile flow slides. This chapter is a review of recent research by several groups into the causes and behavior of the flow slides. The failures are preceded by slow initial displacements of several tens of meters at the pile crests. Observations of prefailure displacement rates have been used successfully to predict the onset of rapid motion. Volumes of individual flow slide events are typically in the hundreds of thousands of cubic meters, but range to several million cubic meters. The flow slides travel distances of as much as several kilometers at speeds exceeding 10 m/s. The mechanisms of failure appear to involve an undrained collapse of fine, saturated waste, or loose saturated foundation soils. The collapse process has been observed in undrained triaxial tests in the laboratory and simulated by a finite-element model of spontaneous liquefaction. The long runout distances appear to result from rapid undrained loading and entrainment of loose saturated soil from the slide path. A dynamic model of flow sliding based on frictional rheology with pore pressure has been calibrated empirically by back-analysis of a number of actual events. It is now possible to predict approximately the velocity and runout distance of typical waste flow slides.
Failures of coal-mine waste piles have long been known to produce rapid, hazardous flow slides (e.g., Bishop, 1973). The coalfields of southeastern British Columbia are particularly prone to this type of mass movement, and the region offers a unique opportunity to observe many large, highly mobile landslides, which are normally rare. Several research groups have recently completed studies of various aspects of the flow slides, but few of their observations have been published.
The first purpose of this chapter is to present a general description of the flow slides and the circumstances under which they occur. Second, we review recent research concerning the mechanics of initial failure and subsequent long-displacement motion. We believe that such a review may be relevant not only to readers concerned with mine waste management, but also to those interested in rapid landslides in general. The behavior of coal waste flow slides appears similar to that of natural destructive mass movements such as rock avalanches.
The coal-producing regions are found near the center of the southern Continental Ranges of the Rocky Mountains in British Columbia, Canada. These are fold mountains with summit elevations of 2000–2500 m above sea level. U-shaped glacier-scoured valleys separate the mountain ranges, forming steep slopes with as much as 1500 m of local relief. High-grade coal seams crop out on the steep mountain flanks and ridges of this rugged terrain.
Seven large open pit coal mines in the Province produce ~25 × 106 t of metallurgical and thermal coal annually, mainly for the export market (Fig. 1). To expose the coal seams, it is necessary to remove 125–150 × 106 m3 of waste rock each year (volume as excavated, before bulking). The waste rock is blasted in benches, loaded using large electric shovels, and hauled to the spoil piles in dump trucks with a capacity of as much as 240 t.
The economics of the hauling operations demand that the waste be spoiled at elevations close to the current mining levels. The haul roads therefore follow the contours out of the pits and across the mountain slopes to platforms, from which the waste rock is end dumped down the slope. This process results in waste piles that drape the mountain flanks and fill natural gullies (Fig. 2). As each mine pit deepens, the spoil pile platforms are constructed at lower elevations and younger piles skirt the lower slopes of existing piles. Individual pile faces can be as high as 400 m and 700 m in slope length.
On average, 6–12 failures of the mountain spoil piles occur each year in British Columbia. Some turn into highly mobile flow slides (for definition see Hutchinson, 1988), traveling over distances of several kilometers. Most of the slides have caused relatively modest direct damage, due largely to the remoteness of the mining areas. Despite their high mobility, most flow slides remain within the boundaries of the mine properties. Designed protective dikes are increasingly used to limit the runout distance. Full-time monitoring procedures described here protect mine personnel working on the dumps. Over ~40 yr of operations there have been eight fatalities in four flow slides, involving persons trapped in the runout zones. Most of the fatal accidents occurred on spoil piles that were not being monitored at the time.
Over the past decade the waste piles have been the subject of intensive research sponsored by the Governments of Canada and British Columbia and by the mining companies. The purpose of the research was to assess the environmental consequences of the slides (Golder Associates Ltd., 1993; Smith et al., 1995), determine the causes of rapid failures (Dawson et al., 1998), document case histories (Golder Associates Ltd., 1992; Broughton, 1992), improve methods of prediction of flow slide runout (Kent and Hungr, 1995), and suggest possible means of reducing the frequency of occurrence (Dawson et al., 1998). Copies of some of these cited documents can be obtained from Bi-Tech Publishers Ltd., Vancouver.
PHYSIOGRAPHIC AND GEOLOGIC SETTING
The Lower Cretaceous and Jurassic sedimentary rocks of the Continental Ranges occur in sequences of northwest-trending broad folds and thrusts, paralleling the axis of the Cordillera. Wide glacial trough valleys are eroded in the same dominant structural direction, separating long linear ridges of nearly constant height. The main rivers occupy braided or meandering channels in wide flood plains on the valley floors. Tributary streams, flowing perpendicular to the structural trend, are usually small and steep, except where they drain hanging valleys left by Pleistocene alpine glaciers. Present-day glaciers are sparse.
The bedrock in the mining areas consists primarily of sandstone and siltstone, and lesser proportions of mudstone and coal. The rock ranges from weak to medium strong (uniaxial compressive strength of 5–50 MPa) and is moderately durable. The various rock types are randomly interbedded in layers generally <0.5 m thick. The mountain ridges most often coincide with synclines; scarp slopes are common and dip slopes are rare. The sedimentary beds have long, uniform dips of <30° and gentle curvature. Longitudinal and cross joints are moderately to widely spaced (typical spacing of 0.1–1 m).
The upper mountain slopes and ridge tops are formed of bare rock, with thin discontinuous veneers of granular colluvium and scree. Lower slopes and hanging valleys are mantled by terraced or undulating deposits of dense silty or sandy glacial drift derived from local rock. The main valley floors are underlain by loose sandy or gravelly fluvial deposits, or rare shallow lacustrine clays, and less often by peat.
The climate of the region is continental with cold winters, warm summers, and moderate precipitation (typically <1000 mm/yr). The vegetation ranges from dense coniferous forests in the valleys to alpine tundra on the ridges.
PROPERTIES OF THE COAL-MINE WASTE
The mine waste consists of angular fragments of the local sedimentary rocks, broken by blasting and mechanical handling. Due to the random interbedding of the source rock, the waste varies widely in character. Sandstone zones produce strong, coarse waste rock with a modal grain size of several hundred millimeters and only a small fraction of sand. Zones rich in mudstone produce finer waste rock, sometimes with as much as 40% sand content and a small fraction of silt. The fines content increases by weathering and mechanical breakdown during primary and secondary handling.
For the purposes of flow slide analysis, it is convenient to distinguish two classes of waste rock (Dawson et al., 1998). Soillike waste has a sand content (<2 mm) >20% and its geotechnical properties are largely determined by the silty sand matrix (Figs. 3 and 4A). Rock-like waste is dominated by coarse angular fragments ranging in size from gravel to boulders (Fig. 4B). Soil-like material is found in the segregated crest zones of end-dumped piles, in layers parallel with the pile slope (see following), and in the runout zones of flow slides. Rock-like material is characteristic of segregated basal layers immediately above the pile foundations (Campbell and Kent, 1993).
Dawson et al. (1998) examined the properties of soil-like waste sampled from finer layers of the dumps and from debris in the runout zones of three flow slides. The grain-size distribution of this material is shown in Figure 3. The samples contained 50%–70% carbonaceous shale, 20%–50% sandstone and siltstone, and minor percentages of coal. The field void ratios ranged between 0.2 and 0.6. A saturated permeability of between 10−4 and 10−2 cm/s was measured. Field saturation was 10%–60%, while the degree of saturation corresponding to the field capacity to absorb and retain interstitial water ranged to 80%.
Consolidated undrained triaxial tests on saturated samples of this waste showed collapsive behavior, as shown in Figure 5. Effective stress paths shown in the figure indicate gradual contractive increase in pore pressure with increasing deviatoric stress up to a certain maximum level. Once this collapse stress level is reached, further increase in pore pressure continues spontaneously and the shear stress falls to a residual level. The percentage difference between the residual undrained shear strength and that corresponding to the onset of collapse is the undrained brittleness index (Bishop, 1973). For void ratios exceeding 0.3 and average effective stress levels <~500 kPa, it was found to range to 45%. This identifies the loose mine waste as a brittle material, capable of rapid undrained failure (Bishop, 1973). The material is dilatant and ductile at void ratios >0.3. It is also ductile at effective stress levels >500 kPa, due to the increase of void ratio by consolidation (Dawson et al., 1998).
STUCTURE OF HIGH SPOIL PILES
The spoil piles are formed by end dumping of the waste from large dump trucks onto slopes as high as 400 m (Fig. 6). Typically, finer soil-like material accumulating near the pile crest may cause steepening to 40°. There is also evidence that lower slopes immediately above the toe may steepen to 40° due to toe bulging, the mid-slopes being flatter. The overall slope is typically in the range of 36°–38°.
Each load issuing from the bed of the truck moves down-slope over the surface of the pile as a dry grain flow <1 m thick. The larger fragments gradually begin to rotate and roll ahead, eventually overtaking the mass of finer material, rolling and bouncing to the lower slopes or even beyond the toe of the pile. The finer material continues to flow as a sheet-like mass and spread longitudinally until it comes to rest on the upper to middle surface of the slope face.
The deposition process described here forms characteristic layering and sorting of the pile. The layers are parallel with the pile face and result from textural differences between individual sheets of finer debris, deposited as a mass. This layering tends to be best developed in the upper part of the pile, although it often extends as far as the toe. Figure 7 shows the layering exposed by a large-scale excavation of a waste pile in a metal mine in Montana. Similar layering has been observed in the coal-mine waste piles.
Sorting results from the mobility of the larger rolling fragments, which accumulate in a coarse layer at the toe of the pile (Fig. 8). The structure of the piles varies depending on the texture of the waste; finer waste produces more pronounced and extensive downslope layering and coarser waste forms a thicker basal accumulation.
The complex internal structure of the waste piles creates complex groundwater seepage conditions (Smith et al., 1995). The compacted pile platform surfaces probably retard infiltration into the top of the dumps. The coarse basal segregated zone represents a pervious layer, and frequently allows streams to seep underneath the piles. For these reasons, conventional stability analyses are usually carried out assuming fully drained conditions (Campbell and Kent, 1993).
It is likely, however, that perched water tables and zones of retained moisture exist in association with the fine layers of the pile stratigraphy. As mentioned here, fine soil-like layers have been observed in the field to retain a fairly high degree of saturation in the form of vadose water (Dawson et al., 1998). Moisture contents may increase due to consolidation. No groundwater discharge is generally observed on the pile faces, but this is consistent with the orientation of layers parallel with the face. Unfortunately, there are few direct observational data regarding the seepage regime in the piles (Smith et al., 1995).
SPOIL PILE FOUNDATIONS
Many of the spoil piles are built on steeply sloping foundations. A review of 46 large failures compiled by Golder Associates Ltd. (1992) recorded an average foundation slope angle of 25.4° (standard deviation 6.4, range 10°–35°). There are very few records of slides occurring in piles formed on gently sloping, horizontal or adversely sloping foundations. One large recent failure occurred by the liquefaction of underlying sensitive lacustrine clay that formed part of the spoil pile foundation on horizontal ground, but this is not typical.
The foundation materials are bedrock, or bedrock covered by a sandy colluvial veneer commonly <0.5 m thick in the upper parts of natural slopes. The materials may include talus or silty and sandy glacial drift on the terraced lower slopes. The colluvium and till are granular and nonplastic, with effective friction angles in the range of 29°–41° (Campbell and Kent, 1993).
Whereas the parent drift material tends to be dense at depth (often it is lodgement till), a loose or compact to dense weathered surficial layer as thick as ~2 m is usually found, consistent with an average frost penetration of 2–3 m. The soils are frozen for several months of the year and are susceptible to loosening and weakening by frost heave. Their drained friction angle may be reduced by 3°–5° and their shear behavior may be changed from dilatant to contractant.
It is possible that some of the looser shallow weathered soil horizons are collapsive when saturated, in the same way as the soil-like waste material; if so, it is only under the extraordinary loading imposed by the waste piles, because natural rapid debris slides are not common in the area. The undrained strength behavior of the foundation soils has not yet been adequately studied.
Usually, no special preparation of the foundation areas is done, other than logging of the tree cover using skidder equipment. The groundwater conditions in the sloping foundation areas are not well known, although a high degree of saturation coinciding with rainy periods and the spring thaw can be assumed.
The spoil piles settle and deform as they are being constructed, as a result of consolidation and plastic deformation of the loose waste material. Crest displacements of spoil piles form series of normal scarps and tension cracks along the edges of the spoiling platforms (Fig. 9). The displacement vectors in the crest areas are usually steeper than the face of the pile, i.e., inclined at 50°–60° to the horizontal. A few measurements on the pile faces indicate that the movement vectors become flatter below the crest. Displacement rates of up to 1 m per day are normal on the larger piles and cumulative displacements of several tens of meters often precede the onset of rapid failure. The scarps and cracks are periodically leveled by earth-moving machinery to maintain trafficability near the platform edges.
Displacements are monitored at regular intervals by wire extensometers strung between pins anchored 5–10 m downslope of the pile crest and reference tripods located on the platforms ~20–30 m behind the crest (Fig. 9). The extensometers are typically read at two hour intervals on active piles, or more frequently where acceleration occurs. Some are monitored telemetrically from the mine office.
Impending major failures of active spoil piles are first indicated by an acceleration of the crest displacements to velocities in excess of ~ 1 m and up to 5 m per day. All mines have developed criteria for closure of the spoil piles depending on observed movement velocity. Prior to the onset of rapid failure, the movement rates gradually increase and may exceed 1 m/h on large piles. Figure 10 shows the prefailure movement rates recorded prior to a 3 × 106 m3 failure illustrated in Figure 11.
The accelerated sagging of the crest leading to a large failure occurs over a period of a few hours and up to a few days. On active piles it is often accompanied by visible bulging of the pile face at midslope and steepening and possible outward displacement of the pile toe. The ultimate rapid stage of pile failure has been reported by eyewitnesses to begin by an “explosive” movement in the toe area, followed instantly by a retrogression toward the crest (Broughton, 1992). The failure movements are flow-like and extremely rapid, sometimes covering hundreds of meters in a few tens of seconds (Fig. 12).
A typical flow slide forms a bowl-shaped scar with a steep back scarp, 40°–55° (segment 1 in Fig. 13) and a basal failure surface that often appears parallel to the pile foundation. The presence of the coarse toe segregation layer makes it unlikely that the basal plane passes immediately above the foundation. Therefore, the basal plane passes through the foundation material beneath the waste, if the foundation material is sufficiently weak (Fig. 13A) or through finer waste above the segregated layer (Fig. 13B).
A connecting middle segment of the rupture surface may form in a layer of weaker waste parallel with the pile face. However, the central part of the rupture surface invariably remains covered by slide debris and the precise extent of segment 3 is unknown.
The failures are often analyzed for stability using a double wedge model, consisting only of a steep back scarp and a basal plane following the foundation contact (Campbell and Kent, 1993). However, the central part of the rupture surface may follow fine-grained layers in the pile interior, as shown in Figure 13. It is in this central portion of the surface of rupture that collapse of the waste material is thought to occur (Dawson et al., 1998). The relative extent of the central segment of the rupture surface probably varies depending on the distribution of finer layers within the pile. The central segment must be fairly short in some cases, as indicated by the considerable height of the back scarp. Some of the slide scars are deep, narrow and elongate (Fig. 14).
Records kept by the Energy and Minerals Division of the British Columbia Ministry of Energy, Mines and Petroleum Resources indicate that more than 100 waste pile failures occurred in the period 1980–1997 in British Columbia. Failure volumes range from ~50 × 103 to 8 × 106 m3, with a median value of ~0.5 × 106 m3 (Golder Associates Ltd., 1992). Runout distances (toe of pile to distal edge of debris deposits) range to 3 km: the median is 700 m. Most of the failures occur on active piles during spoiling operations, or a few weeks after operations ceased. However, there are some cases where failure occurred on inactive spoils as long as one year after the end of construction (Dawson and Morgenstern, 1995). This shows that high loading rate is not always a decisive factor.
The deep-scated nature of the failure scars shows clearly that the failures are controlled by weak zones at depth within the interior of the pile and/or within the foundation.
Research carried out at the University of Alberta shows that sloping fine-grained layers within the internal structure of the piles can be subject to collapse, leading to the creation of excess pore pressure or spontaneous liquefaction. The presence of collapse and excess pore pressures during and following failure is indicated by the sudden change of behavior from stable, ductile deformation to rapid brittle failure. The process was described by Dawson et al. (1998) as follows.
During the construction of the pile, loose layers of soil-like material form. The layers may reach a high degree of saturation by vadose water or localized perched water tables. As construction continues, an element of waste in the embankment, the void ratio of which is initially described by point A in Figure 15A, follows a consolidation path from A to B. Significant shear stress will be generated due to the steep angle of repose slopes.
At point B the soil element is potentially collapsible, because it is looser than the steady-state void ratio and its stress state places it close to a collapse line (Fig. 15B). The collapse line, unique for a given void ratio, is defined as the locus of peaks on all constant volume stress paths, as shown in Figure 5. A “trigger” loading, due perhaps to a small increase in local pore pressure, shear stress, or both, brings the soil element to the collapse point C. At this point the soil skeleton structure collapses, excess pore pressures are generated spontaneously, and the stress path follows from C to D in Figure 15B. The resulting strain weakening initiates the load-transfer process that causes collapse of adjacent elements and eventually results in overall progressive failure. The collapse process is rapid enough so that drainage of the excess pore pressure is retarded even in the soil-like waste with permeability of the order of 10−4 to 10−2 cm/s.
Dawson et al. (1998) carried out stability analyses of three pile failures examined in detail. Fully drained analyses using the effective strength parameters of coarse and fine waste and the foundation soil gave factors of safety of ~1.2. Thus, without excess pore pressure, the slopes should have had reasonable stability margins. The analyses were repeated using collapse strength parameters, similar to those proposed by Sladen et al. (1985). These are lower than the conventional Mohr-Coulomb parameters as they relate not to the failure envelope, but to the collapse line as shown in Figure 15B. Using the collapse strength in the fine waste, factors of safety of ~1.0 were obtained, consistent with the occurrence of failures. Subsequent brittle acceleration can likewise be explained by substituting the still lower steady-state strength parameters into the analysis, as discussed in the following.
A finite element model has been developed at the University of Alberta, incorporating a numerical representation the stress path pattern shown in Figure 5 (Gu et al., 1993). The model was used by Dawson et al. (1998) to simulate the development of the collapse process in several actual cases. The material characteristics for the analyses were determined from tests similar to that in Figure 5. It was found that, even if the collapse process is confined to a small domain within the pile, catastrophic displacements and accelerations result. An example displacement field obtained from such an analysis is shown in Figure 16.
Shear collapse may also occur in the foundation beneath the pile, where this consists of loose, saturated, and relatively fine grained colluvial soil or glacial drift (Campbell and Kent, 1993). In addition to the collapse process already described, excess pore pressure may also be generated in the foundation soils by rapid loading, given that the pile faces advance as rapidly as 1 m/day. Once large-scale shearing displacements begin in the toe area, the overriding of colluvial soils in front of the original toe location leads to further rapid loading and to the possible generation of excess pore pressures in a manner similar to that proposed by Sassa(1988).
Thus, collapse of loose saturated material appears to provide a plausible explanation of both the occurrence and the brittle nature of the observed pile failures. There is still a degree of debate among workers in this field, concerning the most likely location of the initial collapse. Dawson et al. (1998) showed that collapse is possible in the fine-grained waste layers in the pile interior (segment 3 in Fig. 13). Campbell and Kent (1993), however, speculated that excess pore pressures generated in wet foundation material in segment 2 in Figure 13A are the decisive factor. Both groups agree, however, that fine-grained poor-quality waste and steep or weak foundations are necessary prerequisites for the occurrence of flow slides. It is possible that in some of the failures, collapse or liquefaction zones form both in the foundation and in the waste material.
Foundation failure, even if essentially ductile, may serve as a trigger for brittle sliding of the overlying waste pile. For example, one recently examined pile was founded on a slope of 26°, underlain by sandstone layers dipping slightly steeper than the slope angle. The surcharge imposed by the waste pile appears to have triggered sliding on bedding planes, with a buckling release zone surrounding the pile perimeter (Fig. 14). While such a mechanism of rock foundation instability is likely ductile, the waste failure was sudden and extremely rapid, leading to a flow over a distance of more than of 2 km (Fig. 12). A possible interpretation of this behavior is that the ductile yielding of the rock foundation caused strains in the waste pile, sufficient to trigger collapse at some location within it.
The great mobility of flow slides of coal waste spoil piles is illustrated by the range of runout path profiles compiled by Golder Associates Ltd. (1992) from 44 reported cases (Fig. 17). Fahrböschung (or “travel angle”), the slope of a line drawn between the crown of the slide scar and the toe of runout, is an index of landslide mobility. As shown in Figure 18, the waste flow slides exhibit greater mobility within a given magnitude range than many large rock avalanches. However, there is a large scatter in the behavior of the waste flow slides, and some are considerably more mobile than others.
In an attempt to provide a means of predicting flow slide runout, the 44 cases described by Golder Associates Ltd. (1992) were subjected to dynamic runout analyses using the model DAN (Hungr, 1995). This is a solution to the equations of motion of a slide mass traveling along a path of predetermined width. Both the type of rheological constitutive relationship and the value of parameters can be varied in this model. The rheology and parameters for typical coal waste flow slides were determined empirically by back-analysis of actual cases, matching observed behavior (Kent and Hungr, 1995).
In the first instance, all the cases were back-analyzed using the frictional rheology, where the resisting shear stress on the base of the flow is proportional to the effective normal stress and independent of flow velocity. The pore pressure at the base of the flow was assumed equal to the total normal stress, multiplied by a constant pore-pressure ratio, ru. With this assumption, a constant bulk friction angle, ϕb, can be related to the effective dynamic friction angle, ϕ′, of the flowing material as follows:
The concept of a bulk friction angle is useful for back analysis, because it combines both the unknowns (pore pressure and dynamic friction angle) into a single parameter. For each flow slide profile, a bulk friction angle was found so as to produce the required displacement along the slide path (Fig. 19). The distribution of the resulting values of the angle is given in Figure 20: ~70% of the cases were successfully simulated with angles ranging between 18° and 24°, averaging 21.2°. These cases are referred to as “normal” runout cases. In a few cases it was possible to roughly estimate movement velocities from superelevation of the flow surface in bends (Fig. 21) and trajectories where the debris became airborne. The calculated velocities matched the observations for the normal runout cases (see Fig. 19). In addition, the model produced reasonable longitudinal distributions of the debris deposits. Thus, the validity of the frictional model has been confirmed and the magnitude of the controlling parameter, ϕb, bracketed within fairly narrow limits for these cases. This allows the runout behavior and distance of potential flow slides to be estimated beforehand.
The results of the back-analyses produced a smaller group of events for which the friction angle needed to be significantly lower than for the main data set. All of these “mobile” runouts occurred on paths that followed confined valleys or gullies containing loose colluvium and organic soils with high water tables (e.g., Fig. 12). The frictional analyses of these cases not only required low values of bulk friction, but also greatly overestimated the velocity of the flow front compared to the data from superelevation in bends. Generally, these results indicate that the mobile flow slides assume the character of flows, moving at moderate but steady velocity down their confined paths. It is considered that this behavior of the mobile cases is due to entrainment of saturated loose soil and organics from the flow path, which corresponds to field observations. Such cases cannot be satisfactorily analyzed with the frictional model and a constant bulk friction angle.
An improvement in the numerical simulation of the mobile group of flow slides was achieved by introducing a different rheology at the point in the path where the flow entered the floor of a channel or a gully and where entrainment of saturated soil was likely to occur. The rheology chosen for the distal part of the path was that proposed by Voellmy (1955) for snow avalanches, consisting of a frictional term and a turbulent resistance term proportional to the square of the mean flow velocity (Hungr, 1995). Figure 22 shows an analysis of a mobile flow slide using the combined rheologics.
The preceding discussion suggests that the initial motion of the flow slides is controlled by frictional effects, as could be expected given the granular character of both the waste material and most of the foundation soils. The mean frictional resistance is, however, reduced by excess pore pressure, which is presumably connected with undrained collapse and/or rapid loading on large parts of the rupture surface. The average bulk friction angle, ϕb, is 21.4°. Estimating 32° as a typical effective dynamic friction angle of the waste material, the mean pore-pressure ratio can be calculated from equation 1 as 0.37, which corresponds to more than 50% of full saturation pore pressure, or 37% of full liquefaction (i.e., zero effective stress). This value likely results from certain averaging of resistances between fully liquefied, partially liquefied, and dry portions of the rupture surface.
The velocity-dependent flow resistance along the distal parts of the mobile flow slides is presumably caused by a greater proportion of fully liquefied material, including loose soil overridden in the flow path, mobilized by rapid loading and entrained by the flow slide. It is suggested that this entrained saturated material imparts a rate-dependent rheology to the flow, possibly due to dilatancy at high rates of shearing. The Voellmy model with a friction coefficient of 0.05–0.1 and a turbulence coefficient of 200–500 m/s2 produces a reasonable simulation of this flow for the events which were investigated.
Some evidence for the processes described here has been observed in test pits excavated into the deposit of the flow slide shown in Figures 12 and 21 and analyzed in Figure 22. One test pit was located in the middle reaches of the path and the other near the distal end, past a segment where the flow was confined in a gully filled with colluvium and organic soils (see Fig. 22).
The test pit excavated in the middle reaches showed that the base of the waste deposit is in sharp contact with the underlying till-like substrate (Fig. 23). The dark colored waste rock above the contact shows no contamination by the beige till. The contact is neither slickensided nor smooth; it is undulating with an amplitude of ~0.2 m and a wavelength of 1 m. An indistinct zone of fine-grained material forms the lowermost 0.2 m of the waste sheet. The topsoil that must have existed at this location prior to the slide is missing. The till is relatively loose within 0.2 m of the contact and dense below, and appears undisturbed.
The preceding evidence is somewhat contradictory. On one hand, the absence of topsoil indicates that a certain thickness of the substrate has been removed and incorporated in the slide. On the other hand, the sharpness and undulating topography of the contact and the lack of mixing suggest that the bulk of the shearing took place within the waste rock, perhaps in the fine-grained basal layer. If this was not so, one would expect the waste rock near the contact to be contaminated by till particles. Based on temperature records preceding the April 29 slide, the ground was frozen, except for a thin surficial unfrozen layer. It is possible that entrainment of soil from the path occurred initially, but was arrested by the frozen horizon.
From the preceding, it is apparent that various parts the flow slide can be both erosive and capable of flowing over the substrate. The flow resistance therefore reflects the undrained strength of both the waste and substrate materials, which may further have a large variety of water contents, depending on position within the flow path.
The second test pit, excavated in the distal portion of the path of the same slide after it had passed through a confined gully, showed that the basal part of the waste sheet was mixed over a depth of nearly 1 m with loose substrate soil and organics. The same materials were seen in many places extruded to the top of the deposit from localized vertical channels in the waste flow sheet. The extruded soil spilled into patches that cover ~10% of the deposit surface area. This is considered as evidence that liquefaction of loose saturated substrate soil due to rapid undrained loading by the flow slide front is occurring and plays a role in the flow mechanism of the more mobile events.
The study of flow slides in coal waste has until now been driven mainly by practical interests. However, the work reviewed herein also provides an opportunity to comment on some general aspects of the behavior of rapid mass movements.
The sudden onset of rapid failure following a long period of ductile displacements is well explained by undrained collapse of loose saturated fine waste or foundation soil. Such a mechanism is consistent with the material properties and the structure of the waste piles. The collapse process can be duplicated in laboratory tests on fine, soil-like waste material. Similar collapsive behavior can be expected in natural accumulations of granular soils, which are loose and contain fine-grained saturated zones. As shown by the field observations and theoretical analyses described herein, the collapsive zones with sufficiently high degree of saturation may represent only a small fraction of the total unstable volume and may not be visible during surface or shallow investigations.
Coal waste flow slides are surprisingly mobile, given that the bulk of the waste material involved in them is coarse, granular, and essentially dry. In this, they are similar to several types of natural landslides, including major natural rock avalanches, rockslides, and falls involving porous rock and rockslides or debris avalanches that hit or override loose saturated soils in their path (e.g., Hutchinson, 1988).
Various mechanisms have been advanced to explain the reduction of shear resistance, which is apparent from the high mobility of such landslides. The main hypotheses include the following (e.g., Hungr, 1990): (1) sliding on an air cushion trapped beneath the slide sheet; (2) fluidization by trapped air; (3) fluidization by vapor generated by frictional heating of pore water; (4) rock melting or dissociation by frictional heat; (5) fluidization by dust dispersions; (6) mechanical fluidization, defined as a reduction of friction angle at very high rates of shearing in dry material; (7) acoustic fluidization due to vibrations generated by the movement of the flowing sheet; and (8) lubrication by partially or fully liquefied saturated soil at the base of the slide.
Observations reviewed in this article point to hypothesis 8 as the most likely mechanism responsible for the high mobility of coal waste flow slides. Liquefied or partially liquefied zones may be generated on the rupture surface during the initial collapse. More excess pore pressure is generated locally by overriding and entrainment of loose saturated soil material from the path of the slide. Perhaps the strongest evidence for this is that the slide mobility tends to be strongly influenced by the character of the substrate in the movement path. Normal mobility, which nevertheless requires an average pore pressure of more than one-third of the total stress, results on substrates of thin colluvium and surficially loosened soil. Much higher mobility is observed when the flow slides cross the floors of gullies, or other sites where deeper accumulations of loose saturated soils and organics exist, creating an increased potential for pore-pressure generation. Direct evidence of entrainment of liquefied substrate material has been found in the distal parts of the flow slide deposits under these conditions.
In analogy to the coal waste slides, the dynamic behavior of natural rock and debris avalanches is probably also influenced by the character of the substrate over which they travel. This should be taken into consideration in developing predictive models.
These observations also indicate that the resisting stresses at the base of a flow slide result from an averaging of shear resistance in different materials and under different conditions. The flow slides are erosive in some parts of the path and flow in a laminar mode in others. They incorporate substrate soil, but not everywhere. The mean pore-pressure ratios estimated from the back-analysis of normal runout events probably represent average conditions. The rheology of the flowing mass can also change with distance in the more mobile cases, as a greater proportion of liquefied material is incorporated. In view of these considerations, it is unlikely that satisfactory predictions of the dynamic behavior can be achieved based on laboratory-derived constitutive relationships. Empirical determination of the bulk rheology, based on back-analysis of actual events, is more realistic. This approach has produced satisfactory results for coal waste flow slides.
The Energy and Minerals Division of the British Columbia Ministry of Energy, Mines and Petroleum Resources, represented by T. Eaton, the Canada Centre for Mineral and Energy Technology (A. Stokes), and several mining companies contributed significantly to research on flow slides. T. Eaton also provided a number of valuable comments on the manuscript. We are grateful to the editors and reviewers of this volume for many constructive suggestions.
Figures & Tables
>This volume documents further advances in our knowledge of catastrophic landslides since the pioneering compilations of the late 1970s by Barry Voight. It provides a worldwide survey of catastrophic landslide events written by leading authorities. Catastrophic Landslides begins by drawing upon South America to dramatically illustrate the impact of these phenomena on human populations. The occurrence of catastrophic landslides, including site-specific insights, is shown through six events of the past 20 years. Several other chapters focus on the mechanisms involved with catastrophic landsides both in relation to geologic factors in a particular geographic area as well as to specific geologic processes.