Elsevier

Powder Technology

Volume 367, 1 May 2020, Pages 336-346
Powder Technology

Evolution of gas transport pattern with the variation of coal particle size: Kinetic model and experiments

https://doi.org/10.1016/j.powtec.2020.03.061Get rights and content

Highlights

  • Summarize internal links between commonly used gas transport models in coal.

  • Propose a model that can describe multi gas transport in multi shape pores.

  • Obtain controlling transport pattern in different gas desorption curves.

Abstract

Gas desorption laws varies with coal particle size. Their proper description is of great importance to natural gas engineering. This paper tries to summarize the internal links in commonly used models and combine them into a more general form. The combined general model is capable of describing fracture flow, matrix diffusion and surface sorption processes in pores with shapes like flat plates, cylinders and spheres. It can also be simplified into a power function which can be readily implemented into simulation of sorption and calculation of dynamic Fickian diffusion coefficients. Experiments including low-temperature liquid N2 adsorption, proximate analysis, isothermal methane sorption and desorption experiments with different particle sizes were conducted to validate the model. Other sorption data from literature were also collected for validation. The fitting results can help explain the size dependence of desorption characteristics and show the flow type evolutions during the damage of pore system.

Introduction

Gas transport through coal or shale is presently an extensively investigated issue worldwide. Appropriate mathematic description to gas transport is beneficial to solving problems like the estimation of gas production [1,2], the prevention of gas-involved disasters [[3], [4], [5]] and the enhancement of greenhouse gas storage [6]. Generally, common gas transport models can be divided into two categories: mathematical models and empirical models [[7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20]] (as listed in Table 1). The derived models are usually in the form of spherical diffusion equations or first-order relaxation models which are termed the unipore Fickian diffusion model (UFDM), the bidisperse Fickian diffusion model (BFDM), the first-order relaxation model (FRM), the dual first-order relaxation model (DFRM) and the Fickian diffusion and first-order relaxation model (FDFRM).

Crank [21] was the first to solve the spherical Fickian diffusion equation which was widely known as the UFDM (Eq. 1). In this model, the porous media are assumed to be homogeneous and the concentration on the surface and in the center stays constant. Then, Nandi [19] used the model to describe gas desorption from coal and obtain the diffusion coefficient. The simplified form, termed the t model, is also widely applied in the field of gas engineering. However, because of heterogeneity of media, there is a common issue that the unipore model can hardly fit well with experimental data over the entire desorption period.MtM=16π2n=11n2expDn2π2a2twhere Mt and M are the accumulated desorption volume at time t and the total desorption volume, respectively. D is the diffusion coefficient; t is the time; a is the diffusion path length.

Then, Ruckenstein et al. [10] proposed a more complicated model, the BFDM (Eq. 2), which assumed two constant diffusion coefficients in a dual-porosity medium: One is for macropores and the other is for micropores. This model can successfully fit desorption data for a long time. In 1999, Clarkson and Bustin [13] developed this model by replacing the linear Henry law (only appropriate for the low gas pressures) with the nonlinear Langmuir isotherm law (also appropriate for the high gas pressures). However, it is impossible to fit the model manually due to the complicated forms. Numerical computer calculation has to be adopted.MtM=λ16π2n=11n2expD1n2π2a12t+1λ16π2n=11n2expD2n2π2a22twhere λ is the contribution of desorption from macropores or in the fast stage; D1 and D2 are diffusion coefficients of gas in macropores and micropores, respectively; t is the time; a1 and a2 are the diffusion path lengths for macropores and micropores, respectively.

The FRM is a classic sorption kinetic model that can be used to describe the chemical reaction process, the surface sorption process [22] and the dynamic particle penetration process [23]. It assumes that the overall sorption is dominated by surface interaction, chemical or physical interaction. The dynamic coefficient K is positively correlated to the diffusion coefficient D.MtM=1eKtwhere K, the first-order dynamic coefficient, depends upon the diffusion coefficient and is inversely proportional to the square of diffusion length.

The DFRM, which shared a similar combination form with the BFDM, was used by Busch et al. [14] to describe the gas adsorption rate in a space with a limited volume. Its simple form allows it to be easily applied to fitting, compared with the infinite series form of the diffusion model.1MtM=λexpk1t+1λexpk2twhere k1 and k2 are the first-order dynamic coefficients of gas in macropores and micropores, respectively.

The FDFRM was used by Staib et al. [17] to describe Fickian diffusion and pure relaxation:MtM=λ16π2n=11n2expD1n2π2a12t+1λ1expk2t

However, for empirical models, there are various equation forms. Models such as the power function, the exponential function, the logarithmic function, the Langmuir function are all simple and easy to be applied in engineering. Among them, some models have similar forms with the mathematically derived models. For example, the Airey [9] model and the Bolt [8] model are similar with the FRM, whereas the Yang [24] model and the Nie [25] model are the approximate fitting of the UFDM. These models are usually used to infer the gas volume at certain time point, which is beneficial for the assessment of gas content and disaster risk in engineering.

Here, only the mathematically derived models are discussed. The unipore model and the bidisperse model share a mathematical transformation relation which isMtM=M1+M2M1+M2=M1/M11+M2/M1+M2/M2M1/M2+1=ζ1+ζM1M1+11+ζM2M2=λM1M1+1λM2M2where ζ is the ratio of final desorption volume from macropores and micropores, ζ = M1∞/M2∞.

Therefore, for a dual-porosity medium, the gas transport model can be expressed as the following form:MtM=λM+1λNwhere M = M1/M1∞ and N = M2/M2∞.

This equation means that the total amount of gas desorption is determined by the contribution of two pore systems. The forms of M and N, or the flow type in each system, will not change the additive relationship or the parallel relationship in Eq. 7. M and N are totally independent.

As the first type of bidisperse model proposed, BFDM only considers the influence of Fickian diffusion. Afterwards, several other types of bidisperse models in which sorption or surface diffusion began to be considered were proposed in recent decades. The comparison between Eq. 6 and Eqs. (1), (2), (3), (4) indicates that the abovementioned mathematic models basically comprise either or both of the following parts: One is the diffusion part or the unipore model, and the other is the sorption part or the first-order relaxation model, as shown in Fig. 1. Different combinations represent different processes. If the basic unipore model is assumed to be A1 (macropores) and A2 (micropores), while similarly, the basic first-order relaxation model is taken as B1 (macropores) and B2 (micropores), then five combinations of the above four items can be found in literature:

  • (1)

    UFDM: A1 or A2. Under this situation, the overall process is dominated by diffusion, and only one kind of pore exists in coal.

  • (2)

    BFDM: A1A2. Under this situation, the overall process is dominated first by macropore diffusion and then by micropore diffusion, and coal is considered to be a dual-porosity medium.

  • (3)

    FRM: B1 or B2. Under this situation, the overall process is dominated by sorption rate, and only one kind of pore exists in coal.

  • (4)

    DFRM: B1B2. Under this situation, the overall process is dominated first by macropore sorption rate and then by micropore sorption rate, and coal is considered to be a dual-porosity medium.

  • (5)

    FDFRM: A1B2. Under this situation, the overall process is dominated first by diffusion and then by sorption, and coal is considered to be a dual-porosity medium.

Coal is commonly considered to be a dual-porosity medium with a fracture system and a matrix system [29,30]. Differing from the abovementioned common dual-porosity structures, the structure the BFDM describes is hypothetically spherical macropore and micropore systems, as exhibited in Fig. 2. The macropore system controls the stage of rapid diffusion, whereas the micropore system controls the stage of slow diffusion. In permeability modeling, the generally used classical cubic model can be changed into other matrix shapes by considering the shape factor [[31], [32], [33]]. However, the shape of BFDM is unchangeable because of the original assumption of homogeneous spheres. If the diffusion equation and the permeability equation are linked with apparent diffusion coefficient or apparent permeability [34], it could be inferred that the BFDM is a specific case for the common dual-porosity permeability model. More shapes should be considered in the diffusion model.

Besides, the BFDM may not describe the laminar flow in fractures. Linear desorption curves are similar to the shape of Darcy's flow, but they fail to fit well with the spherical diffusion model. It may induce errors especially for the initial stage of desorption because linear facture flow generally releases first. In addition, the DFRM and the FDFRM cannot represent any coal pore structures since they are based on a surface relaxation process. They are more suitable for describing the surface sorption phenomenon. Thus, the exploration of a universal form that can describe the processes of fracture laminar flow, matrix diffusion flow and sorption in dual-porosity media is worthy of effort.

Section snippets

Basic terms for gas transport model

Commonly, there are three gas flow patterns in pore and fracture systems. In factures, gas will transport in laminar flow driven by pressure difference, whereas in the matrix, gas will transport in diffusion flow driven by concentration difference. When the transport ends, gas molecules will finally be adsorbed on the pore surface [16]. Based on the parallel relationship used in the bidisperse model, the equations for gas transport in the three patterns can be first written, and their

Experiments and results

In this section, actual experimental desorption data were fitted by the general gas transport model. Based on the standards of ISO 17246:2010 (Coal—Proximate Analysis) and MT/T752–1997 (Determine Method of Methane Adsorption Capacity in Coal), a proximate analysis and isothermal adsorption experiments were conducted to obtain compositions and adsorption ability of coal, as listed in Table 3. Then, the pore shape of sample was analyzed with the method of Liquid N2 adsorption in light of ISO

Application of SGGTM in gas transport pattern identifications

With the N2 adsorption results that the pore shape is mainly slit-shaped pores, the specific expression of Eq. 17 for the coal sample can be written as:MtM=β1n=082n+12π2expD2n+12π2l2t+γ2k0cxlt+1βγ1eKt

The fitting results of Eq. 26 are given in Table 4. All the correlation coefficients are found to be larger than 0.999, which validates correctness of the model. Besides, the contributions made by the matrix diffusion (β) and the surface sorption (1 − β − γ) increase with the decrease of

Conclusion

In previous studies, several mathematic models have been established to describe the gas transport through porous media by considering different transport mechanisms. This paper tries to summarize their internal links and obtain a general combination form that can describe different flow types in different shaped systems. The conclusions are drawn as follows:

  • Previous used gas transport models are generally different combinations of the original unipore Fickian diffusion model and the basic

Nomenclature

    a

    diffusion path length

    a1

    diffusion path length of gas in macropores

    a2

    diffusion path length of gas in micropores

    ac

    characteristic length for different geometries

    aL, bL

    Langmuir constants

    A, B, ω,ς,k

    fitting coefficients

    cx

    gas concentration in the region that gas is not desorbed

    D

    diffusion coefficient

    D1

    diffusion coefficients of gas in macropores

    D2

    diffusion coefficients of gas in micropores

    Dt

    dynamic diffusion coefficient

    f(t),g(t),φ(t)

    functions

    K

    first-order dynamic coefficient

    k1

    first-order dynamic

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

The authors are grateful to the financial support from the Beijing Municipal Natural Science Foundation (8194072), the National Science Foundation of China (Nos. 51904311, 51874314), the Fundamental Research Funds for the Central Universities (2019QY02), the State Key Laboratory Cultivation Base for Gas Geology and Gas Control (Henan Polytechnic University), China (WS2019A04). Comments by all anonymous reviewers are highly appreciated.

References (42)

Cited by (20)

  • Modeling of multi-field gas desorption-diffusion in coal: A new insight into the bidisperse model

    2023, Energy
    Citation Excerpt :

    Although the initial gas diffusion coefficient of macropores increases with the increase of coal particle size, the gas diffusion path increases with the increase of coal particle size. The effective diffusion coefficient can be written as D/rc2, which comprehensively considers the effects of gas diffusion velocity and coal particle size on gas diffusion process [41]. It can be seen in Fig. 7 that the initial effective diffusion coefficient of macropores decreases with the increase of coal particle sizes, which is consistent with the macroscopic trend of gas diffusion.

  • Thermokinetic analysis of low-rank bituminous coal during low-temperature oxidation: A case study of the Jurassic coal in Shendong coalfield, Ordos Basin, China

    2022, Energy
    Citation Excerpt :

    Zhao et al. [30] used a programmed experimental system to determine the variation of apparent active energies, index gases, and thermal energy release during coal high-temperature oxidization. Furthermore, some scholars concluded that the particle size, water content, oxygen supply, ambient temperature, and pore distribution are important factors to affect the predicting accuracy of coal spontaneous combustion [31,32]. Crossing-point and TG experiments are normally methods to determine the characteristic temperatures.

  • Effect of stress, concentration and temperature on gas diffusion coefficient of coal measured through a direct method and its model application

    2022, Fuel
    Citation Excerpt :

    Diffusion coefficient exhibits a asymmetric “U” shape trend with the metamorphic degree increase of coal [24,27], and the diffusion property of low rank coal sample is higher than that of high rank coal sample [28]. Tectonic deformation increases accessible mesopores and causes micropores collapse [29], resulting in higher diffusion coefficient of tectonic coal than intact coal [30–32]. The influence of gas pressure on diffusion coefficient is complex [33].

View all citing articles on Scopus
View full text