Numerical Modelling of Stress-Strain Analysis in Underground Thick Coal Mining

Thick coal mine operations have various methods. If the thick coal seam cannot be mined by single pass longwall (SPL), then Longwall top coal caving (LTCC) method or multi slice longwall (MSL) can be employed. In Turkey, caving methods are commonly used in mining of thick coal seams as long as the roof strata are suitable for their use. Longwall with caving is always preferred to filling methods because of its simplicity, favorable economics. In the literature, underground production methods are studied oftenly on the contrary rock mechanics and roof conditions. Rock mechanics and roof strata conditions examining with Finite Element Methods (FEM) are very important in terms of efficiency and continuous production. In this study, FEM analysis are made for thick coal mining methods. Numerical modelling from the field for laboratory tests on samples obtained after the Mohr-Coulomb and/or Hoek-Brown rock failure criteria determining the rock mass properties modeling has been performed. Similar strain values are computed in front of the face in all three methods considered. However, when stress-strain values on the gob area are measured, stress-strain distribution in lower face method is found to be more stabilized and proper.


I. INTRODUCTION
In many of the countries on the world, lignite coal is still one of the major energy resources. World lignite coal production is 905 Mt in 2012 according to the International Energy Agency (IEA, 2013c). In addition, lignite coal production of Turkey is 68.1 Mt in the same year. 33.3 Mt of this production is mined by Turkish Coal Institution (TKI), 28.5 Mt is mined by EUAS (Electricity Generation Company) and the rest by private companies. Total lignite reserves of Turkey amount to approximately 14 Gt in recent years which constitutes 1.52 % of total world reserves. In Turkey, caving methods are mostly employed in mining of thick coal seams as long as the roof strata are suitable for their use. Longwall with caving is always preferred to stowing faces because of its simplicity, favorable economics, and high productivity. It is assumed that the upper bound of applying SPL method as a mechanized system in thick coal seams is about 6m. If the thick coal seam cannot be mined by SPL method, then MSL method can be employed (Peng and Chiang, 1984;Singh and Singh, 1999;Hartman and Mutmansky, 2002;Hebblewhite and Cai, 2004; Simsir and Ozfirat, 2008). LTCC and MSL methods are given in Figure 1. In this study, stress analysis is investigated for production methods of thick coal seams using Phase 2D which is a twodimensional stress analysis software. The production method used in the current system, which is LTCC from bottom face method, is modelled and the stress values are found. Then alternative production methods which are lower-upper face and sliced methods are modelled. In all three methods handled, minimum vertical stress is found on the face. But, LTCC from bottom face method is more advantageous than other methods according to the stress analysis results. Stress values on the face, in front of the face and rear face in this method lower than others. LTCC from bottom face method (current method in mine) is found to be preferable method compared to others.

II. STUDY FIELD
Omerler Underground Mine is in the inner Aegean District of Turkey near Tuncbilek-Tavsanlı, Kutahya area and belongs to Turkish Coal Enterprises ( Figure 2). The total proven lignite reserve in the district is about 330 million tons. The average depth is approximately 240 m below surface. The thickness of the coal seam is 8m with a slope of 10o (Taskin, 1999;Destanoglu et al., 2000;Yasitli, 2002;Yasitli andUnver, 2005, Ozfirat, 2007). A generalized lithologic column showing the coal seam together with roof and floor strata is given in Fig. 3. There are three main geological layers in the mine area which are claystone, clayey marl and marl (Destanoglu et al., 2000). Physical and mechanical characteristics of coal and surrounding rock are presented in Table 1 (Kose et al., 1994;Taskin, 1999, Destanoglu et al., 2000, Yasitli, 2002, Yasitli and Unver, 2005, Ozfirat, 2007. Coal has been produced by means of longwall retreat with the top-coal-caving production method where a 3 m high longwall face was operated at the floor of the coal seam (Fig. 4). Topslice coal having a thickness of 5 m was caved and produced through windows located at the top of the shields. Fig. 4 gives the plan (a) and the cross-sectional views (b) of the longwall. In addition roof support properties are given in Table 1

III. IN-SITU ROCK-MASS STRENGTH PROPERTIES
The first task in applying these models is to make an initial estimate of the range of potential strength and stiffness properties for the various major rock units present. This is done by assuming a failure criterion for the rock and by estimating the strength properties using the geotechnical characterization and available laboratory-testing data (Gönen and Köse, 2011; Malli et al, 2017).

A. Failure Criteria
The Hoek-Brown failure criterion is a commonly accepted method for estimating the relation of the principal stresses at failure for a rock mass. Hoek-Brown (1997) studied a relation failure conditions for rocks under stressing forces. They used trial and error methodology and found that the relation between the major principal stress and the minor principal stress is curve linear. The failure criterion relates the major principal stress (σı) to the minor principal stress (σ3) at failure. The Equation (1) describing the criteria is given below.
where is the Hoek-Brown constant for the particular rock type, and s depends on the characteristics of the rock mass. The value σc: is the uniaxial compressive strength of the intact rock. The calculation of the mb, and s parameters is based on the degree of jointing and the alteration of joint surfaces reflected in the value of the RMR in Equation (2) and Equation (3).

B. Goaf Materials
Modeling of the gob area is another important step that affects the accuracy of the obtained results. It is rather difficult to model gob material by numerical analyses. Since gob is mainly made of broken rock pieces, its deformational properties are complex due to an ongoing consolidation process with an increase in the amount of load. Xie et al. (1999) suggested the Equation (4) (2004) may be used to describe the stress-strain behavior of goaf material. Determination of the two parameters, Ɛm and Eo; is essential to describe the complete stress-strain curve for a site-specific caved rock material. Ɛm merely depends on the initial bulking factor, b; and it can be determined as given in Equation (5) and Equation (6).
where σ: is the uniaxial stress applied to the material, Ɛ: the strain occurring under the applied stress, Eo: is the initial tangent modulus and Ɛm: is the maximum possible strain of the bulked rock material.

IV. NUMERICAL MODELS
LTCC method has many advantageous in production of thick coal seams. However due to production losses in caving and dilution other thick coal seam production methods should also be examined. During this examination mostly production characteristics are considered and rock mechanics characteristics are examined very little. In this study, for Omerler region, three different production methods are compared in the terms of rock mechanics using numerical models. Average thickness of the seam is 8 m. Bottom face, lower-upper face and three slice production are evaluated in terms of rock mass.
LTCC method has many advantageous in production of thick coal seams. However due to production losses in caving and dilution other thick coal seam production methods should also be examined (Yavuz, 2003 (Figure 8b). On the other hand, in lower face method, these forces reach up to gravitational strain value at 150 m behind the face (Figure 8a). This difference is to the additional strain forces caused by lower face.
In multi slice method, maximum vertical strain measured on the top longwall, middle longwall and bottom longwall are 7 MPa, 6.5 MPa and 6 MPa respectively (Figure 8c). Strain values decrease gradually from top to bottom. According to these measures, strain decreases by 7.14% from the first slice to the second slice. In addition, strain decreases by 14.29% from the first slice to the third slice. This is because, 2nd and 3rd slice are on the zone of relaxation of the first slice.
VI. CONCLUSION Numerical modeling is an important and useful method in order to determine roof support dimensions. It is also beneficial in order to see work safety of production methods, to examine whether, the roof supports are sufficient and to be more careful in points with maximum stress values.
Production methods used in thick coal seams are analyzed according to strain forces using Phase2D modelling. Similar strain values are computed in front of the face in all three methods considered. However, when strain values on the gob area are measured, strain distribution in lower face method is found to be more stabilized and proper.