Image processing based characterisation of coal cleat networks

https://doi.org/10.1016/j.coal.2016.11.010Get rights and content

Highlights

  • An image processing workflow to characterise coal cleat networks is introduced.

  • Workflow is based on the distinction of perpendicular face and butt cleats.

  • Geometric parameters for three different cleat networks are deduced.

Abstract

Characterisation of the cleat network serves as the basis for estimating the hydraulic and mechanical seam properties which in turn are fundamental for flow and geomechanical modelling in the context of underground coal mining. Cleat and cleat network geometry can be described as a function of frequency, aperture, size, orientation relative to in situ stresses, connectivity and porosity, with mineralised and un-mineralised cleats occurring. To describe these properties, CT-scans of core samples of a Bowen Basin coal in central Queensland, Australia, are analysed.

A unique image processing workflow method is introduced to extract the key statistical parameters of perpendicular butt and face cleats present in a two-dimensional image. As face and butt cleats have different characteristics, the presented method distinguishes face cleats and butt cleats by direction and present detailed data for both cleat types. The results comprise cleat length, apertures, sizes, intensities, densities, shape parameter, spacing, orientation and connectivity and are therefore more comprehensive than previous cleat descriptions. Three generally different cleat geometries are considered within this study, one sample shows perpendicular face and butt cleats, the second two sets of face cleats intersected by butt cleats and the third parallel face cleats only.

Introduction

Coal is characterised by its unique microstructure of a low permeable porous matrix intersected by a network of face and butt cleats. Cleats are natural opening-mode fractures within coal beds. Together with fault related and mining induced large-scale fractures, these small scale cleats provide the principal source of permeability for groundwater and gas flow within the seam (Laubach et al., 1998). In contrast to the fractures, the smaller systematic cleats do not cut clastic facies adjacent to the coal layers (Dron, 1925). Cleats are distinguished by their orientation into two types (Fig. 1a). Face cleats are dominant and are directed perpendicular to the bedding plane. The fewer orthogonal butt cleats generally terminate when they encounter face cleats (Laubach et al., 1998). Butt cleats are thought to accommodate relaxation of the stress which originally formed the face cleats. This leads to perpendicularity between both cleat types (Golab et al., 2013). Most research has been done on this orthogonal cleat system. For example, Robertson and Christiansen (2006) described the cleat system as a system consisting of cubic matrix blocks. However, Nick et al., 1995 observed more complex structure systems which need to be characterised differently. A variety of cleat patterns in Permian Queensland coals has been described (Pattison et al., 1996). Turner (2015) identified four groups of cleat morphologies based on CT-scans and core photographs (Fig. 1b).

Regardless of the cleat system, cleat geometry can be described as a function of frequency, aperture, size, the orientation both relative to other cleats and relative to in situ stresses, as well as the degree of their connectivity and porosity (Close, 1993). Further, connectivity and spatial distribution of cleats, their cementation, filling and weathering influence geomechanical and hydraulic properties. In the subsurface, over geologic time, carbonate, quartz, or other cement may precipitate on the cleat walls, often disrupting connected networks. Cleat voids can contain organic materials, resin and authigenic minerals like clays, quartz and carbonates that can occlude or preserve fracture porosity (Laubach et al., 1998).

While for many coal fracture systems the porosity of fractures is found to be dependent on the stress field and therefore the depth (Barton et al., 1995), Laubach et al. (2004) showed that open fractures are not necessarily aligned or more permeable in certain directions relative to the stress field. In fact, the orientation of open natural fractures depends on the relative stiffness of the fracture and host material. Most of the coal measures in the East of Australia have little to no primary porosity. Approximately 75% of eastern Australian coals current porosity is formed by a pore structure in the size range of meso- and micro-pores (< 50 nm) (Faiz and Aziz, 1992).

The term cleat refers to opening-mode fractures in coal beds. Besides cleat systems, the following literature review also covers fracture systems of other geo-materials. As a consequence, the terms cleat and fracture are used synonymously, with cleats referring to research done on the typical small scale structures in coal.

The aperture between fracture walls has been discussed as one of the main influencing parameters for flow in fractures. Fracture apertures cover a wide scale range as their variation is influenced by mechanical and chemical actions in the system (Bonnet et al., 2001). Philip et al. (2002) numerically investigated the effect of diagenesis on the initial flow properties of fracture systems, especially with respect to diagenetic effects on the connectivity of the fracture network. The results indicated that fracture permeability is more sensitive to fracture patterns and connectivity than aperture. The most commonly applied relation between aperture and permeability is the cubic law (Witherspoon et al., 1979). Long and Witherspoon (1985) claimed that fracture length and density are interrelated with flow rate. Networks with longer fracture lengths and lower fracture densities exhibit higher connectivities and therefore higher permeabilities than those formed of shorter fracture lengths with higher densities.

The common approach for the modelling of fluid flow in fractures is to randomly generate artificial sets of fractures based on commonly known distributions (De Dreuzy et al., 2001). Galindo-Torres et al. (2015) examined randomly generated two-dimensional fracture networks and showed that the connectivity and conductivity of each individual fracture are related to the macro-conductivity and hence can be described by universal functions. The prediction of permeability is economically crucial as it determines whether commercial gas production rates can be achieved. Gas stored in coal seams (Clarkson and Bustin, 2011), as well as groundwater in the system is transported through the cleats towards the producing wells (Scott, 2002). Because seam permeability is a crucial factor for mine stability, minability, gas well performance and drainage behaviour, knowledge of cleat characteristics is essential for safe and efficient mining. The need for a better representation of fracture characteristics during permeability measurements has been pointed out by Huy et al. (2010). Further, the structure of cleat networks influences geomechanical behaviour like load bearing capacities of the seams and therefore impacts the mining process. The crucial role of the fracture system on the micro-scale requires, therefore, suitable image analysis tools allowing a streamlined and efficient, and yet comprehensive and differentiated characterisation of the fracture system.

While image processing has been widely used to describe structural properties of porous media (e.g. Vogel and Roth (2001), Khan et al. (2012)), literature on image processing methods for analysing cleat structure of coal samples is comparably sparse. Based on scanning electron microscopy of polished samples, aperture size distributions have been investigated and characterised by Karacan and Okandan (2000). Weniger et al. (2016) used images taken with optical microscopy of scanned polished sections to quantify cleat aperture, spacing, height, and frequency data and to derive permeability related to aperture and spacing based on the cubic law.

X-ray computed tomography (CT) scanning of coal samples as used in the presented study allows a non-destructive insight into a sample through the acquisition of projection images from three different directions. The interior of the sample is represented depending on their X-ray attenuation. By stacking slices of two-dimensional images, a three-dimensional data set is obtained. Coal matrix and pore spaces filled with air or liquid can be distinguished, as the X-ray attenuation is primarily a function of the energy, density and atomic number of the material (Ketcham and Iturrino, 2005). Fractures and/or cleats and their variation of occurrence through the length of a cored coal sample can be detected and parameters like size, aperture and cleat connectivity can be quantified. Mazumder et al. (2006) used CT-scans to describe fracture orientations and cleat aperture and spacing in coal samples. The approach has been extended to deduce coal density and the distribution of inorganic material (Klawitter et al., 2013). Wolf et al. (2008) compared cleat angle distributions from drilling cutting to cleat orientation distributions from CT-scans from coal blocks of the same seams.

The presented paper aims at providing an image processing workflow to obtain geometrical properties of cleats in coal in a systematic way. CT scans are a useful tool for the testing and calibration of image and network-based models (Kumar et al., 2010). An innovative integrated application combining CT-scan based material characterisation and process monitoring is promising to give an insight into the influence of heterogeneous rock properties like porosity, hydraulic conductivity and diffusivity on fluid transport processes and geomechanical properties (Cnudde & Boone, 2013). We are introducing an image processing based workflow to determine geometrical parameters of coal cleats and discuss different cleat geometries. As shown by Laubach et al. (1998), face and butt cleats have different characteristics. We have developed a method that distinguishes face cleats and butt cleats by direction (Busse et al., 2015) and present detailed data for both cleat types. The results comprise cleat length, apertures, sizes, intensities, densities, shape parameter, spacing, orientation and connectivity. Therefore, the presented method provides more comprehensive information than previous cleat description approaches.

Section snippets

Sample origin and preparation

The coal samples used in this study were taken at the extension site of the Hail Creek Mine. The mine is situated on the East Coast of Australia, 120 km south-west of Mackay, Queensland, and is operated by Rio Tinto Coal Australia. Currently, sub-bituminous coal is extracted from two seams of the Rangal coal measures; the Elphinstone Seam, with an average thickness of 6.4 metres (m) and the Hynds Seam, averaging 8.3 m in thickness. Underlying are the Fort Cooper coal measures. They are extended

Image processing

An algorithm to extract geometric and topologic cleat information from a two-dimensional greyscale image has been developed and implemented in the MATLAB® environment. The form definition of a cleat in a coal specimen is based on the widely accepted concept of an orthogonal network of butt and face cleats with both normal to the bedding plane (Fig. 1). For the study of cleat occurrence and geometry in 2D image processing, a cleat has been defined as a feature that is longer than wide and

Results

Characteristics of coal cleat networks have been extracted based on the image processing method presented in the Analysis chapter. Three coal samples have been analysed; the results are summarised in the following. The cleats captured in the images are mineralised, which makes them well visible in the scans. Cleat mineralogy in this set of samples comprises of kaolinite, calcite, and traces of anastase.

Cleat characteristics

Three different network types of coal cleats have been described based on CT scans taken from coal samples. The classic cleat model with face and butt cleats, therefore, can be extended to a more accurate description as has been proposed by Turner (2015) (Fig. 1). Using a MatLab based image processing workflow, we present and characterise three different cleat networks. One sample (E1) shows a geometry with two types of face cleats and one set of butt cleats. Sample E2 reveals a coal network

Conclusion

We present a solid image processing method to extract main shape features of a cleat network in 2D images of coal samples. A simple criterion for clustering is used: if a gap between two cleats that are aligned along the same axis is smaller than the aperture, they are considered as one cleat and grouped together. With the presented method, the distinction between three different types of cleat geometries is possible that are extending the common definition of face and butt cleats in coal

Acknowledgement

The coal samples used in this analysis were provided by Rio Tinto Coal Australia. The contributions and discussions with Patrick Schmidt of the University of Queensland are gratefully acknowledged. The manuscript has been improved thanks to the valuable input of three anonymous reviewers. This research has been funded by the Australian Coal Association Research Program (ACARP C20022).

References (51)

  • S.E. Laubach et al.

    Are open fractures necessarily aligned with maximum horizontal stress?

    Earth Planet. Sci. Lett.

    (2004)
  • S. Mazumder

    Application of X-ray computed tomography for analyzing cleat spacing and cleat aperture in coal samples

    Int. J. Coal Geol.

    (2006)
  • A. Odgaard et al.

    Quantification of connectivity in cancellous bone, with special emphasis on 3-D reconstructions

    Bone

    (1993)
  • S.D. Priest et al.

    Discontinuity spacings in rock

    Int. J. Rock Mech. Min. Sci. Geomech. Abstr.

    (1976)
  • P.L. Rosin

    Unimodal thresholding

    Pattern Recogn.

    (2001)
  • A.R. Scott

    Hydrogeologic factors affecting gas content distribution in coal beds

    Int. J. Coal Geol.

    (2002)
  • H.J. Vogel et al.

    Quantitative morphology and network representation of soil pore structure

    Adv. Water Resour.

    (2001)
  • J.J. Walsh et al.

    Fractal analysis of fracture patterns using the standard box-counting technique: valid and invalid methodologies

    (1993)
  • G.X. Wang et al.

    An improved permeability model of coal for coalbed methane recovery and CO2 geosequestration

    Int. J. Coal Geol.

    (2009)
  • S. Weniger et al.

    Characterizing coal cleats from optical measurements for CBM evaluation

    Int. J. Coal Geol.

    (2016)
  • K.H.A.A. Wolf

    Determination of the cleat angle distribution of the RECOPOL coal seams, using CT-scans and image analysis on drilling cuttings and coal blocks

    Int. J. Coal Geol.

    (2008)
  • C.A. Barton et al.

    Fluid flow along potentially active faults in crystalline rock

    Geology

    (1995)
  • E. Bonnet

    Scaling of fracture systems in geological media

    Rev. Geophys.

    (2001)
  • J. Busse

    Detection and Description of Fracture Networks in Coal Samples for Micro-Scale Flow Simulations

    (2015)
  • J.P. Chilès

    Stochastic modeling of natural fractured media: a review

  • Cited by (45)

    • Universal scaling solution for the connectivity of discrete fracture networks

      2022, Physica A: Statistical Mechanics and its Applications
      Citation Excerpt :

      Geological fractures in rock masses, including cleats, joints, and faults, comprise a complex fracture network system [1].

    • Pore space attributes of nonconventional reservoirs

      2022, Developments in Petroleum Science
    View all citing articles on Scopus
    View full text