Greece has been traditionally placed among the leading lignite (brown coal) producers in Europe. In fact, lignite is the most significant fossil fuel energy source in the country and it has contributed to its electrification and overall economic growth since the 1950s (an era when intense and systematic exploitation commenced). In 2017, 37.7 Mt of lignite were mined, the vast majority of which were from the deep surface mines of the state-owned Public Power Corporation (PPC) (Euracoal, 2018). Despite the recent decrease of lignite share in the total energy mix of the country (in 2017 it amounted to 33% of electricity generation, while just a few years ago it was more than 50%), lignite exploitations may still be a viable operation in the future, as long as mining procedures are optimized and effectively managed (Roumpos, Pavloudakis, Liakoura, Nalmpanti, & Arampatzis, 2018). A critical aspect for the effective management of lignite mines relates to geotechnical engineering and particularly to slope stability of the corresponding deep surface excavations (sometimes as deep as 200 m below the original ground surface). Actually, in the case of the Greek lignite, which is a low grade coal, exploitation has to be performed cheaply (Kavvadas, Papadopoulos, & Kalteziotis, 1994), so optimal excavations are necessary. For instance, Karas (1988) had shown that lignite pit slopes that are flatter even by one degree than the optimum angle, may lead – depending on the height of the pit – to conservative and uneconomical exploitation. Similar examples have been shown by Leonardos (2004a). Zevgolis, Deliveris, and Koukouzas (2018), based on a probabilistic geotechnical design optimization framework, illustrated the connection between the pit slope and the total excavation cost for a typical surface lignite mine.
On the other hand, over the last decades, excavations for further exploitation of the Greek lignite has unavoidably become deeper and, as such, more difficult to manage. Leonardos and Terezopoulos (2003) mentioned that the maximum pit depth in the Greek lignite mines increased from 70 m in 1980 to 170 m in 2000. In addition, with reference to PPC mines Roumpos et al. (2018) distinguished four periods of relatively constant stripping ratios (defined as the volume in m3 of waste material to be removed in order to produce 1 ton of lignite): 1–2 m3/t during 1959–1967, 2–3 m3/t during 1968–1991, 3–4 m3/t during 1992–2003, and 4–5 m3/t from 2004 until today. In other words, stripping ratios steadily increase with time, which means that, in general, the excavations become larger and deeper and as a result, more challenging from a geotechnical point of view. In fact, over the last ten years many incidents of excessive displacements and/or instability phenomena in the Greek lignite mines have been reported in the literature (Kavvadas, Agioutantis, Schilizzi, & Steiakakis, 2013; Marinos et al., 2015; Prountzopoulos, Fortsakis, Marinos, & Kavaddas, 2010; Prountzopoulos, Fortsakis, Marinos, & Marinos, 2017; Steiakakis, Agioutantis, Apostolou, & Papavgeri, 2017; Steiakakis, Apostolou, Papavgeri, Agioutantis, & Schilizzi, 2016; Zevgolis, Koukouzas, & Schilizzi, 2015). In addition, other incidents became known to the public through the media, but have not yet been reported in the literature. In some cases, on top of serious operational and production difficulties, some of the above incidents have also created serious concerns to the local societies, given the proximity of certain mines to inhabited areas.
Taking into consideration the above, it is clear that robust practice of geotechnical engineering is very important for the future of the Greek lignite mines. In this context, the scope of the present work is to provide a critical and comprehensive review of the relative literature over the last 40 years, and also to briefly present some cases (of geotechnical interest) for which no information has been published in the literature until now. The paper is structured as following: Following the introduction, a brief description of the mining methods that are being used and of the typical excavation profiles that are being encountered is provided. The above are followed by an inclusive review of the earlier studies and investigations. Then, a more extensive and elaborative review of recent cases (i.e. events of the last decade) is presented. Then, a critical discussion on commonly anticipated failure mechanisms in the Greek lignite mines is presented. Based on all the above, in the last section of the paper relevant conclusions are drawn and discussed. It shall be noted that this work discusses stability problems related to excavations, rather than to spoil dumps and waste embankments. Therefore, such works (Masoudian et al., 2019; Steiakakis, Kavouridis, & Monopolis, 2009; Zevgolis, Koukouzas, Roumpos, Deliveris, & Marshall, 2018) fall beyond the scope of the present study and are omitted.
Mining methods and typical excavation profiles
In Greece, lignite deposits under exploitation are primarily located in the Region of Western Macedonia in northern Greece (mostly in Ptolemais-Amyntaio basin, but also in Florina and Kozani-Servia basins), and secondarily in the Megalopolis area (Region of Peloponnese in southern Greece) (Fig. 1). The vast majority of the lignite is mined by the state-owned PPC. For instance, out of the 37.7 Mt of lignite that were mined in 2017, 35.4 Mt (94%) came from the surface mines of PPC and 2.3 Mt (6%) from a few private mines (Euracoal, 2018). The most common exploitation method in PPC’s surface lignite mines is the continuous surface mining method (sometimes referred to as the German method, or in older references, the terrace mining method) using bucket wheel excavators. Lignite and waste material are usually transported by conveyor belts, while spreaders are used for dumping. Occasionally, when required for operational needs, conventional mining methods (hydraulic excavators, loaders, and dumpers) are also used on a smaller scale. Drill and blast is being used in certain cases that hard strata are encountered (such as, for instance, in the South Field mine in the Ptolemais area). Conventional methods (hydraulic excavators, etc.) are also used in the few small private mines (for example, in the Lava and the Prosilio mines) that are located in the Region of Western Macedonia (Kozani-Servia basin). It should be noted that over the last three decades lignite mining in Greece has been taking place exclusively via surface mining. The only exception is a private underground exploitation that has been recently initiated, combined with surface operations, in the Prosilio mine (Kozani-Servia basin) (Tzalamarias, Tzalamarias, Benardos, & Marinos, 2018). In the deep lignite mines of PPC, particularly in the Ptolemais-Amyntaio basin, exploitable deposits usually lie at an average depth of 150–200 m below original ground surface. So, the total depth of the mining pit slopes in this area might be up to 200 m. A general view of such a deep excavation (the Mavropigi mine, in Ptolemais) is shown in Fig. 2. In general, the PPC excavations in Ptolemais-Amyntaio are deeper than the company’s excavations in Megalopolis. Nonetheless, according to Kavvadas, Roumpos, and Schilizzi (2018), although the later ones are more shallow, they present similar stability problems because of the lower shear strength of the excavated material. Furthermore, the excavations of the few private lignite mines in northern Greece are of a smaller scale (but are usually steeper) than the PPC mines. In all cases, the excavations are separated in several vertical or sub-vertical consecutive benches. In the case of the PPC mines, the height of these benches usually varies between 15 and 25 m (with an average of approximately 20 m), or even between 10 and 30 m, depending on several parameters, such as the height of the excavators and possible stability concerns. The width of the benches usually ranges between 30 and 120 m, depending on stability, functionality issues, and the stage at which the excavation has proceeded. A characteristic close-up view of five benches of the Mavropigi mine, with several bucket wheel excavators in place, is shown in Fig. 3. According to Kavvadas et al. (2018), nowadays, the overall inclination in the PPC mines usually varies between 1 V:4H (14°) to 1 V:7H (8°). Nonetheless, elder references also refer to steeper mines. For instance, Steiakakis and Agioutantis (2010) refer to a case study from the mid 1990s in which the overall inclination of a 140 m-deep excavation (Tomeas 6 mine, in Ptolemais) was about 1 V:2.5H (∼22°). Also, Leonardos (2004a) provides certain cross sections (of failure incidents) from mid to late 1990s, in which overall inclinations varied between 12° and 20°. In any case, an indicative and simplified section of a typical deep lignite excavation is shown in Fig. 4.
In terms of typical excavation profiles and stratigraphy, ground profiles in the mines consist of a thick overburden zone of sterile materials (Quaternary deposits of several tens of meters average thickness) that overly the exploitable lignite deposits (Neogene deposits). The later ones most often alternate with intermediate thin layers of sterile material. A characteristic multi-layer stratigraphy from PPC mines is shown in Fig. 5. Also, a general geological stratigraphy column of the Western Macedonia lignite basin is provided in Fig. 6. The sterile material (of both Quaternary and Neogene origin) mostly consists of marls and clays, often stiff to hard, and occasionally, of weak conglomerates and water bearing sands. So, in general, when it comes to slope stability studies in the mines, soil mechanics (rather than rock mechanics) principles are applied. It should be noted that due to mild tectonic conditions, the orientation of the layers is relatively constant and the dipping relatively gentle (0–6°), resulting in horizontal or sub-horizontal strata. In any case, as stated by Steiakakis and Agioutantis (2010), local tectonics (presence of faults, etc.) should be taken into account during planning and exploitation of every mine, given that they often play a very critical role in the kinetic behaviour of excavation pits.
Elaborate descriptions of typical excavation profiles in the PPC mines have been provided by several authors in the past. For example, from top to bottom, Kavvadas et al. (2018) refer to the following formations: (i) recent Quaternary (Holocene) deposits (average thickness 5–10 m), (ii) upper horizon of Pleistocene Quaternary sediments that includes sterile material, such as marls, stiff clays, and occasionally, weak conglomerates and water bearing sandy layers (average thickness 70–120 m), (iii) lower horizon of Lower Pleistocene (Quaternary) – Upper Miocene (Neogene) deposits (average thickness 50–80 m), that includes most of the lignite seams and very stiff to hard marls and clays. The main body of the lignite seam is located quite closely to the bottom of the excavated pit. The upper zone of the horizon consists of sterile material (marls/clays), with thickness varying between millimeters and several tens of centimeters, that are alternated with thinner lignite layers, and last (iv) hard Basal Marl that is encountered at the bottom of the lignite layers and indicates the bottom of the exploitable deposits. Monopolis, Stiakakis, Agioutantis, and Kavouridis (1999) and Steiakakis and Agioutantis (2010) present a similar stratigraphy (on a profile based on the Tomeas 6 mine). Particularly with respect to the Neogene deposits (lignite bearing layers), they refer to the presence of the following three series: (i) the upper series (maximum thickness of 100 m) with clays and marls, (ii) the lignite bearing series (average thickness 135 m), and (iii) the lower series, whose uppermost 40 m consist of clayey (or in places sandy) marls, with intercalations of marly limestones. As far as the lignite bearing series, they refer to a lower (80 m) and an upper (40 m) multi-stratified lignite sequence and an intermediate waste sequence (15 m). The intermediate waste sequence and the interbedded waste material (in the upper and lower lignite sequences) mostly consist of marls, clays, and calcareous mudstone.
Early studies and investigations
To the authors’ knowledge, the first research work referring to the Greek lignite mines from a geotechnical perspective was the work of Anagnostopoulos (1979, 1982). In particular, investigating the compressibility behaviour of the Megalopolis’ lignite, he concluded that lignite behaves like a typical overconsolidated clay, exhibiting a significant secondary compression. Nonetheless, Anagnostopoulos did not extend his research into stability considerations. One of the earliest published works that dealt with slope stability analysis (but not failure incidents) was the work of Stamatopoulos and Kotzias (1981). They analyzed the stability of a 60 m deep surface mine of PPC, on top of which a levee (of maximum 15 m height and about 4 km length) would be constructed, alongside the Alfios River in the Megalopolis area. Based on their investigation they concluded that the lignite and the inter-layered clay differed mostly in terms of plasticity, pure lignite being non plastic. Another published study in the 1980s was the work of Karas (1988). He provided a short and general discussion on geotechnical aspects and design issues of excavated slopes of the Greek lignite mines (referring to the Main Field mine in Ptolemais and the Amyntaio mine which at the time was relatively new). He also made a very short note about two failure incidents from the 1970s, without providing though many details. These were the 1974 and 1976 landslides in the Thoknia mine, which was the first mine ever excavated in Megalopolis. In addition, Karas referred to the abundance of stiff overconsolidated clays both in Megalopolis and Ptolemais, and in particular to the critical role of water infiltration in their joints and fissures. Also, he mentioned the sporadic presence of sensitive (thixotropic) clays in the Ptolemais area. The vital role of groundwater infiltration was also highlighted by Kavvadas, Marinos, and Anagnostopoulos (1992), who showed that water infiltrations in joints and open faults exert hydraulic pressures and, as a result, they play a considerable role (in addition to the material properties) in the stability of the excavated slopes. Furthermore, based on an elaborative investigation on the geotechnical properties of the Ptolemais lignite, Kavvadas, Anagnostopoulos, Leonardos, and Karras (1993), Kavvadas, Papadopoulos, Kalteziotis (1994) showed that it has a high and variable void ratio incompatible with its preconsolidation pressure, which may be attributed to the existence of structure (chemical bonding). Based on triaxial CU and CD tests, they showed that the shear strength is variable but strongly dependent on the in-situ moisture content. In fact, they considered that the magnitude of bond strength (expressed by the peak shear strength) is inversely proportional to the moisture content and they suggested that the peak shear strength of the material can be estimated from the in-situ moisture content with reasonable accuracy.
The first case of significant displacements in a lignite mine that was extensively reported in the literature was the case of PPC’s Tomeas 6 mine in Ptolemais (Monopolis, 1999). In 1994 and 1995 cracks were noted on the surface behind some slopes of the mine and significant heaving (about 0.6 m) was recorded on a production bench. Because of these events, PPC conducted a field monitoring program, composed of topographic surveying and some inclinometers. The major concern was whether the displacements were due to stress release phenomena (given the size of the 120 m deep excavation and the significant ground masses removal), or if it was a forerunner of an imminent catastrophic slope failure. Monopolis et al. (1999) presented the geotechnical investigation that had taken place, including laboratory testing, field measurements, and limit equilibrium stability analyses. They concluded that the under examination slope cracks and floor heaving could be attributed to stress relief due to the mining process (i.e. rebound of the formations due to the excavation), rather than on the development of a general slide. The work of Monopolis et al. was further extended by Steiakakis in his doctoral dissertation (Steiakakis, 2003). He thoroughly studied the behaviour of the deep excavation of the Tomeas 6 mine, by investigating both the geotechnical parameters (through an extensive laboratory testing investigation) and the kinetic behaviour of the mine (through analytical, limit equilibrium, and numerical models). His work was also discussed in Steiakakis and Agioutantis (2010). Overall, Steiakakis and Agioutantis concluded that the marls decisively influenced the kinetic behaviour of the mine excavation. Furthermore, using a finite elements program (Plaxis), they modeled the mobility that was evident during the excavation of the mine and their results compared very well with monitoring data. Ultimately, the authors concluded that the developing shear stresses were not high enough to justify the occurrence of a failure mechanism. Even though local overstresses and instabilities were disclosed, these were not linked to a global pit slope failure. In terms of the large magnitude of deformations (i.e. estimated horizontal displacements of about 0.8 m and floor heaving of about 0.5 m), the authors stated that such deformations should be expected in excavations with a depth of around 150 m in the Ptolemais area. They also claimed that if the site is not locally influenced by adverse tectonics and stratigraphy, the danger of large-scale landslides was limited. To summarise, Steiakakis and Agioutantis (2010) concluded that the kinetic behaviour of the Tomeas 6 mine was attributed mainly to heaving due to unloading as excavation proceeds combined with the elasto-plastic behaviour of the geological formations.
Another interesting work that tackled several geotechnical aspects of the deep Greek lignite mines is the doctoral dissertation of Leonardos (2004a) and the relevant publications (Leonardos, 2004b; Leonardos & Terezopoulos, 2002, 2003). Based on data and observations from his own professional experience in the Greek lignite mines, Leonardos made a number of remarks about the geotechnical behaviour of the deep Greek lignite mines. First of all, he identified what he called a “compound failure mechanism” as the most critical failure mechanism in the Greek lignite mines. He described it as being composed of a horizontal or sub-horizontal, practically linear, sliding surface followed by a rising curve at the rear end (somewhere behind the crest of the slope). The linear surface is usually found along a weak stratum, which is often an interface between lignite and underlying clay. Fig. 7 shows five examples of slope stability failures (during the period 1996–2001) whose mechanism (together with other issues) was discussed by Leonardos (2004a). All cases refer to failure surfaces, on which a more or less linear (horizontal or sub-horizontal) part was distinct. Excavations’ geometries varied, with heights ranging between 75 and 165 m and overall inclinations between 11.5° and 19.5°. A summary of these events is provided in Table 1. Further discussion on the subject of failure mechanisms is provided on a following section of the present paper. In addition to the above cases, which Leonardos characterized as progressive-type failures, he also provided some data from incidents of regressive displacements. His distinction was based on the concepts described by Zavodni (2001), who introduced the concept of regressive and progressive stages of instabilities: in the first case, the pit slopes demonstrate cycles of decelerating displacements, and eventually they reach ultimate stability, while in the second case, accelerating cycles of movements occur and an ultimate collapse is imminent (unless effective remedial measures are applied). In the context of regressive displacements, Table 2 refers to four cases, on which noteworthy (but regressive) displacements were recorded. In all four cases, the pit eventually reached an ultimate stability stage. It is interesting to note that in three out of the four cases, extensive and intense rainfall had preceded the displacement incidents. In addition to the above, according to Leonardos, the most important factor for stability is the shear strength of the aforementioned clay layer. In this context, he also discussed the critical effect of the residual friction angle (of the clay layer) on stability. Furthermore, through simple analyses, he briefly highlighted the role of adverse inclination (towards the excavation) of this layer, as well as the role of water infiltration into ground fissures and cracks. In addition, he outlined the importance of field monitoring in deep lignite mines, which was not usual practice until the mid 1990s, in order to ensure effective slope stability evaluation and safety. In fact, he mentioned that the very first time that inclinometers had been used in the Greek lignite mines was in 1991, in order to monitor a 1 V:3H 95 m high slope that was moving at a rate of 8–20 mm/month. Overall, in Leonardos’ opinion, impending slope failures can be detected and analyzed long before any crack formation at the slope crest becomes visible (Leonardos & Terezopoulos, 2003). He actually suggested that the velocity of recorded displacements (rate of movement) is a reliable criterion for imminent failure conditions. Specifically, based on his data and observations, he stated that when the rate of movement steadily increases above 20 mm/day then slope failure is expected to take place within a period of 6–12 days after the velocity has exceeded this limit (Leonardos & Terezopoulos, 2002). However, this conclusion was seriously questioned by Kavvadas et al. (2013; 2018), as will be discussed in the following section. Last but not least, and in addition to all the above, Leonardos critically discussed remedial measures for slopes’ failures in the Greek deep lignite mines and he concluded that often the most appropriate method of stabilization is the internal (i.e. within the mine) deposition of sterile material (Leonardos, 2004b).
Recent studies and cases of instability incidents
In an interesting recent study, Kavvadas et al. (2018) investigated the stability of slopes in the PPC lignite mines and they stated that slope instabilities are usually governed by sliding along a sub-horizontal, unfavorably sloping (towards the excavation), interface between lignite and an underlying stiff, high plasticity clay or marl layer, very close to the bottom of the slope, where shear stresses are greatest. As described by the authors, the typical mechanism of such instabilities is triggered by the sharp stiffness contrast between adjacent lignite and clay/marl layers that causes different elastic rebound upon the removal of the horizontal confinement during excavation. Based on a series of parametric analyses, they investigated the influence (on stability) of parameters such as the pit’s geometry (total height and overall inclination), the groundwater conditions, and the shear strength of critical interfaces between clays and lignite. In addition, based on their analyses, they concluded that even a small but adverse inclination (i.e. towards the mine) of the lignite/sterile interface plays a decisive role in the stability. The paper also reviews slopes in the PPC lignite mines that remained stable despite movements with relatively constant velocities reaching up to 100 mm/day, while others failed when velocities accelerated abruptly although they were much smaller. According to the authors, these cases show that the absolute magnitude of slope velocity is not always relevant in predicting slope instability, while slope acceleration (plotted as the inverse of velocity versus time) is a better indicator. In other words, they questioned the conclusions that were made in the past by Leonardos and Terezopoulos (2002) and Leonardos (2004a). In any case, Kavvadas et al. (2018) believe that any attempt to correlate slope safety factors with velocities or accelerations should be made with great caution. Even more caution should be paid in circumstances when attempting to estimate the time until an imminent slope collapse.
In addition to the above study that addressed several geotechnical issues of Greek lignite mines, many studies in the recent past focused on certain cases of slope failure incidents and other geotechnical concerns. These incidents, summarized in Table 3, are discussed in the following paragraphs.
Mavropigi mine (Ptolemais)
A very interesting case study has been the Mavropigi mine in Ptolemais. In 2010, noteworthy surface tension cracks appeared at the crest of the southeast slopes of the excavation (northwest of the namesake village of Mavropigi). The phenomenon raised major concerns not only about the stability of the pit, but also about the effect of the exploitation on the surrounding area (Kalogirou, Tsapanos, Karakostas, Marinos, & Chatzipetros, 2014; Marinos et al., 2015). Since then, significant monitoring has been taking place and counter-measures have been applied in order to ensure the stability of the mine slopes and the surrounding areas. According to Steiakakis et al. (2013), the reasons behind the unstable wedge that appeared in Mavropigi in 2010 were: (i) the adverse steeply convex excavation geometry, (ii) the complicated tectonic structure (presence of steep inclined faults), and (iii) the presence of a sub-horizontal unfavorably dipping thin clay of very low residual shear strength. It should be noted that the unconventional and adverse steep excavation was created mainly due to land acquisition issues and expropriation limitations at that time. Fig. 8 shows the unconventional progress of the excavation (from left to right) due to the expropriation limitations, which eventually led to a steeply convex geometry (shown clearly in the right figure) and provoked adverse geotechnical effects.
Two very interesting studies related to the Mavropigi case were presented by Kavvadas et al. (2013) and Steiakakis et al. (2017). Kavvadas et al. (2013) investigated the stability of the moving southeast slope of the mine and presented information from an extensive monitoring campaign, with survey prisms, inclinometers and piezometers. Based on this data, they concluded that horizontal displacements were developing upon a sub-horizontal thin layer of high plasticity clay at an average magnitude of 2 cm/day (reaching 5 cm/day during periods of heavy rainfalls). The use of the investigation data to evaluate the type of movement, the geometry of sliding surface and the effectiveness of remediation measures, as well the procedure of assessing the stability and safe slope operation during production, were analyzed in detail. It should be noted that the high plasticity clay layer dictating the sliding mechanism was found (via limit equilibrium back-analyses) to be at residual stress state, with a residual friction angle of approximately 7° (and zero residual cohesion). Furthermore, based on information gathered from a thorough geotechnical investigation (including boreholes, inclinometers, survey measurements, geophysics, and DInSAR), Steiakakis et al. (2017) proposed a detailed geological, geotechnical, and kinematic model of the sliding mass and they validated it via limit equilibrium back analyses (which indicated a safety factor slightly smaller than unity). This model is shown in Fig. 9. Based on this model, Zevgolis et al. (2015) performed preliminary finite elements analyses using the shear strength reduction technique. Overall, the studies of Kavvadas et al. (2013) and Steiakakis et al. (2017) indicated that, if properly performed, mining can actually take place even within a landslide. In fact, it was shown that there are situations where mine slopes can move several meters and still be operational without catastrophic failures. A view of significant tension cracks and subsidence southeast of the excavation face three years after the initiation of the phenomenon is shown in Fig. 10. The aforementioned conclusions come in accordance with what is often the case generally in mining engineering, where the term “slope stability failure” is defined differently than in civil engineering. From a mining perspective, most open pit lignite mines have sufficient operational flexibility to withstand considerable slope displacements, as long as these are effectively monitored and modeled, so that: (i) sliding mechanisms are well interpreted, (ii) predictions regarding their magnitude and kinetic behaviour can be made, and (iii) operational safety is guaranteed. As a result, the design and construction of pit slopes at economically optimum angles, even if frequently accompanied by appreciable movements, is of great importance. In fact, mining operations may take place even in moving slopes, as long as safety of personnel and equipment is satisfied (Kavvadas et al., 2013). Zavodni (2001) stated that mining can progress with safety and minimum disruption, provided that evolution of displacements with time is constantly recorded, and the developing sliding mechanism is well understood. He even noted that instability in mining can be advantageous from an economic point of view, if the costs incurred (including costs to ensure safe working conditions) are exceeded by the benefits gained due to mining at a steep slope angle.
Lava mine (Kozani-Servia)
Prountzopoulos et al. (2010) presented a complex and relatively large slide that occurred in 2007 in the Lava mine. This is a private mine which is quite small compared to the large PPC excavations and is located a few kilometers south-west of the town of Servia, in the Prefecture of Kozani (Region of Western Macedonia). In this case, the slide was not related to the production benches (in spite of the steep overall inclination of the excavation), yet it took place on the very upper part of the production slope. This part consisted mainly of plastic clayey marls, of which the authors made a distinction between intact (φ’ = 40°, c’ = 200 kPa, PI = 2–24) and weathered (φ’ = 23°, c’ = 75 kPa, PI = 13–25) material. They concluded that the slide was controlled mostly by the very low residual shear strength parameters along a sub-horizontal weak plane of weathered marl that was followed by a rising curve up through the marls and the quaternary deposits. This is shown in Fig. 11. Based on parametric back analyses and in the absence of any piezometer data, they suggested that the residual friction angle of the critical surface must have been around 9–11° (in the absence of a groundwater table) or 15–17° (in the case of groundwater table near the ground surface). The first author of this paper was recently informed that new problems had arisen in the mine, with significant slides of the quaternary deposits upon the marly formations (GMMSA LARCO, personal communication, August 2018). These slides are clearly shown in the upper left part of Fig. 12.
Prosilio mine (Kozani-Servia)
Prountzopoulos et al. (2017) presented another noteworthy landslide in the Prosilio mine. This is again a small private mine that is located a few kilometers south of the town of Servia (Prefecture of Kozani, Region of Western Macedonia), i.e. very close to the Lava mine. Problems were first reported in 2008, when cracks at the ground surface NW of the mine were observed. Based on survey monitoring, which indicated a slow but constant kinematic activity, the mine owner decided to construct a lateral counterweight from sterile material, in order to cease the displacements. However, the counterweight was not completed due to adjudications. At the end of 2010, an increase in the magnitude and the velocity of displacements was observed, and eventually a large-scale landslide took place in February 2011. In general, the slide went through marly lacustrine sediments. Based on the studies of Prountzopoulos et al. (2017), the factors that triggered the phenomenon were the non-completion of the counterweight, the large height of the slope (more than 60 m), the presence of a specific weak surface between intact and weathered marls (which had caused smaller scale problems on several occasions in the past), the clayey and weak nature of the marl (and its sensitivity to environmental agents), and last but not least the dual action of water that not only deteriorated the quality of the upper marly layers, but also contributed to gradually increasing pore pressures within the layers. Regarding the Prosilio mine, it should be noted that eventually part of the open pit excavation turned into an underground excavation using the room and pillar mining method. This is the first time that an underground lignite exploitation has been pursued in Greece since the 1980s (after the closure of the underground Aliveri mine in Evia, central Greece). Given that the present work focuses on surface excavations, further elaboration on the topic falls beyond the scope of the study. Nonetheless, interested readers might find more information about the underground pursue in the literature (Deliveris & Benardos, 2017; Tzalamarias et al., 2018).
Choremi mine (Megalopolis)
The Choremi mine is operated by PPC and it is a large lignite mine in southern Greece (Megalopolis area) with an annual production (2013) of about 8 million tonnes of lignite. In 2011, surface cracks appeared at a distance of 150 m southwest from the crest (and parallel to it) of the SW part of the mine, towards the village of Tripotamos. A systematic follow up of the phenomenon took place, including stability studies and monitoring, and no further instability phenomena appeared until 2013 (PPC, personal communication, 2015). However, on September 12th and 13th, 2013, surface cracks were detected, this time in the eastern section of the southern part of the mine. Eventually, on September 14th, 2013 a large landslide took place in two phases. First, the slope in the SE part of the excavation was horizontally displaced by about 36 m and at the same time a subsidence of about 20 m occurred at the area of the crest of the excavation. This was the major landslide and took place over approximately 1 h. Before the landslide, the average inclination of the pit in that area was about 1 V:5.5H, i.e. a very smooth inclination of about 10.5°. In the second phase, during the same day, a secondary landslide took place in the SW part of the excavation (towards the Tripotamos village, i.e. close to the cracks of 2011). In this part, the horizontal displacement was smaller, about 17 m, and the settlement at the crest of the excavation was about 10–15 m (on which the average inclination was about 9.5°, i.e. 1 V:6H). An illustrative picture showing the horizontal displacement of the bench (as one block) is shown in Fig. 13. Based on back analyses, the failure surface was primarily defined by a thin (less than 1 m) horizontal clay layer underlying the bottom of the lignite deposit, whose residual friction angle was estimated to be approximately 6–7° and residual cohesion to be about zero (PPC, personal communication, 2015). In addition to the above information, in the context of the present study the landslide and the corresponding surface ground deformations were mapped using remote sensing techniques, based on Landsat 8 satellite images that were processed using the Cosi-Corr software. The results of the analysis are shown in Fig. 14. In the upper part of the figure the displacement vectors are plotted upon the mine’s satellite image. The vectors indicate the direction and magnitude of displacements. The results of the analysis are even clearer in the second and third part of the Figure, which correspond to the horizontal displacements along the North-South and East-West direction, respectively. The two different (by a few hours) slides, at the SE and SW of the excavation, are evidently identified in the figure: maximum displacements along the south-to-north direction are about 40 m at the SE part of the excavation (the first event), and about 15 m at the SW part of the excavation (the second event). Positive values refer to displacements from south to north. Maximum monitored displacements along the east-to-west direction are about 10 m at the SW part (the first event) and very small at the SW part of the excavation (the second event).
Amyntaio mine – the 2016 landslide
The Amyntaio mine in Western Macedonia is in operation by PPC since the mid 1980s and is considered a highly important mine for energy generation in Greece. It is a typical example of a deep surface excavation with depths of up to 180 m. Due to unfavourable geological, geotechnical and hydrogeological conditions a significant number of landslides have occurred in the past. Recently, Steiakakis et al. (2016) presented a case where the combination of thin clay layers (with back-calculated residual friction angles between 9° and 10°), overlaying deep sandy layers with an unfavourable inclination and complex tectonic formation triggered a slow moving landslide (covering an area of about 1.2 km × 0.5 km) inside the mine pit. In addition, Tzampoglou and Loupasakis (2017, 2018) have studied land subsidence phenomena at the perimeter of the Amyntaio line, based on the evaluation of geological and geotechnical data and with a focus on the overexploitation aquifer phenomena. Nonetheless, this type of phenomena fall beyond the scope of the present study and no further discussion is provided.
Amyntaio mine – the enormous landslide of 2017
The landslide that took place in June 12, 2017 was probably an unprecedented event in the history of the Greek lignite mines and it is probably one of the largest lignite mine landslides worldwide (Fig. 15). As shown in Fig. 16, the sliding mass moved from south (very close to the Anargyri settlement) to north and covered a huge area, which based on satellite images was estimated to be between 2.98 and 3.56 km2 (Valkaniotis, Ganas, & Papathanassiou, 2017). Given the devastating consequences of the landslide, two independent committees (one by PPC and the other by the Ministry of Environment & Energy) were ordered to investigate the phenomenon. The reports of these two committees have not yet been made available to the public. Nonetheless, a report that was recently published by the Independent Authority of the Greek Ombudsman summarizes the major conclusions of the two committees’ reports. To be more specific, according to the committee delegated by PPC, the main reasons behind the landslide were the following ones (Pottakis, 2019):
- The combination of the rotating-pit operations at the western part of the SW slope and the dip of the bedding planes of the lignite deposits’ base towards the NW, combined with an exploitation depth increase due to deep excavations at the western edge of the foot of the SW slope (where the deepest part of the lignite deposit is located).
- The activation of two nearby faults, namely the Vegoritis and particularly the Anargyri faults, which took place due to a decrease in the normal stress on the faults planes and slow shear sliding mechanisms. The reason behind the activation was the unloading due to the gradual expansion of the mine and the removal of significant ground masses.
- The filling with groundwater of the significant surface crack that appeared on May 15th, 2017 (i.e. four weeks before the event) at the crest of the western part of the southwest slopes. This crack had expanded rapidly by the end of May along the entire crest of the slope. After the event of June 12th, the main scarp of the landslide revealed itself where the crack had been located before the event.
It must be mentioned that, as reported by Pottakis (2019), according to the PPC study the landslide did not take place due to the usual reasons that cause stability problems in the PPC lignite mines. For example, the slope’s inclination could not have been the reason behind the landslide, given that the average inclination of the southwest slope was much milder than the usual recommendations for the PPC mines (1 V:5H). In addition, the committee concluded that the landslide was provoked neither by significantly high water pressures build up within the slope mass, nor by the absence of support of the slope’s foot through internal deposition of sterile material and ash. This is the case because although the approach of internal deposition at the foot of the southwest slope was not practically applicable, the slope has been operating for at least five years without any stability problems. Overall, the PPC study stated that while – generally speaking – the above factors (steep inclination, built up water pressures, and lack of internal deposition) are always unfavourable to the stability of the slopes of the company’s mines, in the case of Amyntaio these factors had remained virtually unchanged for at least five years before the event, so they were not considered the cause of the landslide. On the other hand, according to the study of the Ministry of Environment and Energy, the landslide took place due to a series of reasons that were adversely and simultaneously found to be taking place. It was the result of unfavourable and complex geological, tectonic, hydrogeological and geotechnical conditions in combination with the mining activity. In particular, the committee classified the factors that triggered the catastrophic event into two major categories (Pottakis, 2019): endogenous and anthropogenic factors. Critical endogenous factors were considered to be the presence of surfaces with low shear strength, the presence of groundwater aquifer under the mine floor, and the presence of the Anargyri and Vegoritis faults. In particular the faults, according to the committee, affected the geometry of the sliding mass, but they were not the landslide’s triggering factor, given that according to the committee the faults’ activation was not documented. In addition to the above, the committee reported two anthropogenic factors that played a role in the landslide. These were the locally increased inclinations of the slopes of the mine (a contradictory conclusion to the conclusion of the first committee), and the water pressures that had been developing for months, both within the tectonic surfaces and the ground fissures, due to the infiltration of groundwater and/or storm-water. The ministry’s committee also stated that the stability conditions of the slopes on the outskirts of the Anargyri settlement were worse than the rest of the slopes of the mine, because of the high groundwater level due to the interruption of PPC pumping from 2014.
Commonly encountered failure mechanisms
From a general point of view, the factors that typically affect instability mechanisms in surface coal and lignite mines can be summarized as follows (Steiakis & Agioutantis, 2007 (i) initial stress field and subsequent changes (e.g. decrease of in-situ horizontal stresses): to it due to the excavation (ii) geological structure (e.g. tectonics), of the area, (iii) geotechnical properties of the formations involved, (iv) groundwater conditions, (v) size of the excavation, and (vi) intensity of vibrations, if any (whether due to natural or mining-induced seismicity). Therefore, it is clear that instability mechanisms in deep lignite excavations depend on many different reasons, the combination of which may lead to highly complex and complicated failure mechanisms. As a result, sometimes the corresponding failure surfaces may be difficult to explain.
In any case, particularly when referring to the deep surface lignite mines in Greece, it can be stated that failure surfaces are most often associated with sliding upon weak zones, either along the lignite-marls/clays interface, along thin clay layers of high plasticity and low shear strength, or along interfaces of marl layers with different shear strengths (and different deformation moduli). This sliding surface typically follows the orientation of soil stratigraphy and, as such, it is practically linear and horizontal or sub-horizontal. It should be emphasized that the interface’s inclination has a great effect on the stability of the pit. Small differences of a few degrees (of adverse dip towards the excavation, which is not rare in the case of the Greek mines) may result in noteworthily different safety factors in terms of slope stability. This was briefly illustrated by Leonardos (2004a) and thoroughly discussed by Kavvadas et al. (2018). Eventually, the failure surface reaches the ground level at a certain distance behind the slope’s crest in the form of a rear (back) scarp surface that might be curved and rising or approximately planar. Such an indicative cross section is shown in Fig. 17. Across the literature, such type of failure mechanisms have been described by the term compound (Bromhead, 1992) or composite (Chowdhury, Flentje, & Bhattacharya, 2009; Duncan, Wright, & Brandon, 2014) failure mechanism. Leonardos, who mostly cited curved rising back scarp surfaces, used the first term (Leonardos, 2004a; Leonardos & Terezopoulos, 2003) for the Greek lignite mines. The same term (compound failure mechanism) was also used by Steiakakis et al. (2016) to describe the 2016 Amyntaio landslide. On the other hand, Kavvadas et al. (2018) avoid referring to either of the two terms (compound or composite). Nonetheless, they provided an excellent and robust explanation (from a stress-strain perspective) of the reasons behind sliding surfaces along certain lignite/sterile interfaces that reach the ground surface through a planar back scarp surface. Overall, the (simplistic but meaningful) geometry suggested by Kavvadas has a trapezoidal (or block) shape, such as the one in Fig. 17. Trapezoidal cross sections were also discussed and proposed for analyzing stability of excavations due to surface mining by Huang (1983). Regardless of the term that might be assigned to them, the most significant feature of the failure mechanisms commonly encountered in the Greek lignite mines is that they are not circular. In fact, the critical aspect of the mechanism is undoubtedly the horizontal/sub-horizontal sliding surface that was mentioned above. Overall, the development of the failure mechanism can be described as follows:
When slopes are formed by a deep excavation they tend to displace towards the excavation due to the significant reduction of confinement stresses (in-situ horizontal stresses), which is induced by the removal of the excavated soil. This unloading process naturally results in the development of considerable horizontal tensile strains in the slope (εh ≠ 0). However, induced strains are different in the lignite and the sterile material above or below it (e.g. marls and/or clays), due to the difference in the materials’ unloading modulus of deformation (as stated by Kavvadas et al. (2018), the brittle lignite has a lower deformation modulus during unloading, than the ductile marls/clays). Eventually, due to the different horizontal tensile strains, noteworthy plastic shearing takes place along the lignite/sterile interfaces. If the resulting shear stresses exceed the peak shear strength (which is often the case), then the interface’s mechanical behaviour shall be described by post-peak or even residual shear strength parameters. Consequently, in such cases, residual shear strength parameters shall be used for the evaluation of pit slope stability. Actually, this has often been the design practice in the PPC lignite mines.
As far as the back-scarp surface is concerned, the previously described conditions also lead to the creation of vertical or sub-vertical tensile cracks in the ground mass at a distance behind the face of the excavation. As the excavation becomes deeper, and the confinement stresses further decrease (due to the excavation), tensile strains increase and, therefore, the (tensile) cracks increase as well. Eventually, if the conditions lead to a catastrophic failure (landslide), the back-scarp surface is more of a curved rising surface. More unfavourable conditions exist when a tectonic fault (with an adverse dip, i.e. towards the excavation) exists behind the crest of a pit. In this case, its mobilization might be triggered by the horizontal strains developed. To be more specific, the mobilization of the fault is caused by the decrease of confinement stresses (which are due to the excavation) that in turn leads to the reduction of normal stresses acting on the fault and, as a result, to the reduction of the shear strength along the fault’s plane. In such cases, the back-scarp surface is often a (sub-vertical) planar surface, so the overall failure surface is more like a block or a trapezoid. It shall be noted that in the case of intense rainfall, fault zones (or tensile cracks) fill with water and water pressures acting on the fault plane increase, leading to the reduction of effective stresses and, as a result, to the reduction of shear strength. The decrease of the fault’s plane shear strength provokes sliding upon it, which in turn induces translational movements on the lignite/sterile interface sub-horizontal plane.
It is important to note that the previously described mechanisms (of linear sliding along the interfaces) have been identified and understood based on knowledge that was gained in the profession over the last 25–30 years. In fact, it was (i) the development of field instrumentation and monitoring techniques during the 1990s, (ii) the development and common use of advanced slope stability computer programs, and (iii) the meticulous observations on certain failure cases, which eventually led to the conclusions above. In addition, according to Leonardos (2004a), all slope stability design studies for Greek lignite mines before 1990 (and quite many during the 1990s) were based on the assumption of rotational failure surfaces (whether circular or non-circular). For instance, Karas (1988) stated that the state of practice for slope design was (at the time) the simplified Bishop method. However, as stated by several authors, this is an inappropriate and inaccurate method for composite noncircular slip surfaces (Chowdhury et al., 2009; Duncan et al., 2014). So, perhaps, this is the reason that according to Leonardos (2004a), before the 1990s there were incidents where, despite the fact that slopes were designed based on proper lab testing and adequate safety factors, these slopes failed.
Anticipated failure surfaces from old incidents of instability are shown in Fig. 7. In the incident of the Tomeas 6 mine (failure surface along a thin clay layer) and in the two incidents of the Choremi mine (failure surface at the interface between lignite and an underlying clay), the linear surface crossed the excavation at the toe of the pit. On the other hand, in the case of the South Field mine (failure surface along a thin zone of marls in between lignite layers), the linear surface crossed the excavation at a distance of 35 m above the toe (floor) of the pit, while in the case of Marathousa at a distance of 35 m below the toe (floor) of the pit (deep failure). Among the recent cases, a typical example is the complex case of Mavropigi where determining the kinematics of the model was a difficult task due to complicated site-specific conditions. Nonetheless, based on extensive monitoring and limit equilibrium back analyses, the failure surface was identified as a combination of a near horizontal surface along a thin clay layer (of about 7° residual friction angle and zero residual cohesion) with a sub-vertical back scarp (Fig. 9). In the case of the Lava mine, the major slip surface also consisted of a sub-horizontal surface (although it was considerably above the lignite seams and the production faces) followed by a curve that went through the marls and the tertiary (Fig. 11). Additionally, in the case of the Prosilio mine, the linear part of the failure surface was about 20 m above and parallel to the top of the exploitable lignite layer, along the contact surface between a weak and an underlying competent marl (Prountzopoulos et al., 2017). It ought to be noted that in the case of Prosilio the second part of the failure surface (that reached the original ground surface) seemed to vary between different parts of the landslide widthwise (i.e. both sub-vertical and much gentler slopes were indicated). In the 2016 Amyntaio case, the sliding mass presented a deep failure surface with a major linear part passing through a low strength clay layer and a curved part in the toe and the crest (Steiakakis et al., 2016). Therefore, based on all the above cases, both recent and old, it can be concluded that the horizontal/sub-horizontal failure surface may cross either at the toe of the pit slope, at a distance above it or even at a depth below it. Nonetheless, for interfaces of similar characteristics, Kavvadas et al. (2018) consider that an interface close to the bottom of the slope is more critical than an interface at higher or much deeper elevations. In the first case (interfaces at higher elevation vs. interfaces at the toe of the pit), this is so because the mass of the sliding body (and therefore the driving forces) increases with depth. In the second case (interfaces at much deeper elevation vs. interfaces at the toe of the pit), this is so mostly due to kinematic reasons and due to the absence of any passive earth pressure at the toe of the pit.
Summary and conclusions
A thorough literature review, covering a period of approximately 40 years, on slope failure incidents and other stability concerns in Greek surface lignite mines has been presented and discussed in the present work. The major conclusions from this study can be summarized as follows:
-As indicated by the steadily increasing stripping ratios over the years, excavations in Greek lignite mines unavoidably become larger and deeper (in several cases up to 150–200 m deep). As a result, they become more difficult and challenging to manage from a geotechnical point of view. So, robust and thorough practice of geotechnics is becoming increasingly crucial for the effective and sustainable management of the Greek lignite mines.
-The majority of the mines are deep excavations over a thick overburden zone of sterile material that lay above the exploitable lignite deposits. In addition, lignite seams often alternate with intermediate layers of sterile material. In general, sterile mostly consists of marls and clays (often stiff to hard), and occasionally of weak conglomerates and water bearing sands. In any case, slope stability aspects are addressed by soil mechanics principles (rather than rock mechanics).
-The most important characteristic of the failure mechanisms is that they are not circular. In fact, most often, failure modes are associated with a horizontal or sub-horizontal (practically linear) sliding surface along a weak zone at the interface between lignite and sterile material (or even between sterile materials of different engineering characteristics). This sliding is due to the different stress-strain response of lignite and sterile in unloading conditions (decrease of confinement stresses) that take place during an excavation. In any case, since sliding along this surface corresponds to large plastic shear strains, residual shear strength parameters along this surface are applicable for the evaluation of pit slope stability. In fact, this has been the state of practice in most Greek lignite mines. It is interesting to note that in some cases residual friction angles of as low as 6–7° (with zero residual cohesion) have been reported.
-Failure mechanisms are sometimes too complex and the local tectonic conditions frequently have a decisive role on failure incidents. For example, when faults with an adverse dip (towards the excavation) are present behind the pits, they can strongly influence their kinetic behaviour. Therefore, they should be taken into serious account, as long as their presence is known, during planning and exploitation.
-In addition to the above parameters, the inclination of the lignite/sterile interface has an important effect on the stability of the pit. Even a few degrees of adverse dipping (towards the excavation), can drastically affect stability conditions. Furthermore, water infiltration in joints and open faults also plays a significant role in the stability of the excavated slopes because of the hydraulic pressures provoked. Besides, many failure incidents have been connected to intense rainfall events that preceded stability failures.
-Given the size of the excavations (nowadays up to 200 m), large deformations often take place regardless of the degree of conservatism adapted during design. In fact, instability incidents have been reported even at very mild overall inclinations of about 10°. Nonetheless, these deformations do not necessarily lead to catastrophic failures. Again, tectonics and stratigraphy must also be taken into consideration. In any case, certain studies have indicated that, if properly performed, mining can actually take place even in situations where mine slopes are moving by several meters. These are cases where shear stresses are not high enough to provoke a catastrophic failure.
-The significant role of slope monitoring during mining has been emphasized over the last couple of decades. Elder works had suggested that the velocity of recorded displacements (rate of movement) is a reliable criterion for imminent failure conditions. Nonetheless, this conclusion has been recently questioned and the acceleration of pit slope velocities over their absolute values is seen to be a better indicator of slope stability during the development of progressive failures.
Source: Slope failure incidents and other stability concerns in surface lignite mines in Greece
Authors: Ioannis E.Zevgolis, Alexandros V.Deliveris, Nikolaos C.Koukouzas