Applications of aerodynamic scale model simulations to predicting the accumulation of methane gas and airborne pollutants in modern coal mines. Final report on CEC research contract 7262/32/229/08

The purpose of this study was the design, construction and application of an aerodynamic scale model of a modern retreat longwall coal face. The objective of the modelling was to simulate the effects of various parameters on the accumulation of methane gas, and hence to provide information which would be of practical relevance to the control of gas in underground mines.This involved (i) the development of specifications for how these simulations could be achieved in the small scale and the essential features to be included in the model; (ii) development and testing of ideas for scaling the effects of turbulent dispersion, the formation of gas layers, and flow through the waste; (iii) development of experimental techniques; (iv) construction of models suitable for the experiments and simulations; and (v) application of the model to the investigation of the consequences of various changes in mining conditions, gas emission, ventilation system, etc.The project was primarily aimed at British mines, but it also included (with support from Charbonnage de France) a study of the feasibility of extending the application of the model and associated techniques to the conditions and issues relevant to the French coal industry.This report describes (i) the principles underlying the scaling of the aerodynamic properties of the flow; (ii) validation of the conditions for turbulent flow; (iii) design and construction of a model of a retreat coal face and waste; (iv) checking, characterising and calibration of the model; and (v) application of the model in simulations.The scaling relationships between the flows in small scale (1710th to 1730th) models of a coal face were formulated in order to define conditions for operating models which would reproduce all the important flow characteristics of the full scale. These included the requirement that the Reynolds number (vdh/v) for the flow on the face be greater than 6000 in order to produce fully developed turbulent flow and a constant non-dimensionalised coefficient of longitudinal dispersion. This meant that with a large reduction in geometric viascale, the flow velocity could not be scaled down by the same amount. Similitude for the dispersion of layers of gas required that the Richardson number be preserved (i.e., that the energy required to overcome the density gradient be kept in proportion with the energy available from the turbulent velocity gradient); this was achieved by using a surrogate gas in place of methane in order to increase the difference in density between the ventilation air and the gas. By maximising the difference in density, it was possible to achieve similitude for the maximum possible reduction in geometric scale; therefore, sulphur hexafluoride (SF6) was chosen as the surrogate gas, thus increasing the density difference by x9.3. However, SF6 is heavier than air, and therefore the model had to be inverted so that the layer of heavy gas would flow along the “”roof which was actually the base of the model. The waste behind the coal face was represented by a granular permeable bed, with the permeability selected so as to produce the proportionate balance of flows between the face and the waste as occur in the real mine. The pressure gradients due to ventilation flow and gas density were also kept in proportion and this placed a constraint on the scaling of the permeability, which effectively selected the scaling of the nominal grain size of the permeable bed. This scaling of the grain size of the waste gave an estimate of the Reynolds number (udg/V for the flow through the granular bed which indicated that, although this did not perfectly match that for the full scale, the difference was negligible compared to the various uncertainties in characterising the properties of the real waste. Consequently, it could be assumed that the relative locations where dispersion would be dependent on inertial dispersion or molecular diffusion would be approximately the same in model and full scale wastes.The minimum requirements for achieving satisfactory turbulent dispersion in the model were verified firstly by measuring the friction factor (which is related to the dispersion coefficient) and then by measuring the dispersion coefficient directly using the experimental technique developed in the preparatory study by Aitken et al (1988). This involved measuring the changing signal detected from forward scattered light from smoke (a neutral density tracer) injected into the model face. A sharp cut-off of the injected smoke produced a gradually decaying signal due to the longitudinal turbulent dispersion of the smoke in the flow between injection point and the detection point. The data curves were used to obtain best fit values for the dispersion coefficient. These tests were conducted first in a model with a 13.5 cm high face equipped with models of the old style hydraulic supports (chocks), and then secondly in a new model set up as a 1720th scale model of a 2 m high coal face. This provided the necessary confirmation that satisfactory flows could be obtained in a model at this scale before proceeding to build a more complex and complete model which included a granular bed to represent the waste and a system for producing a full range of seam slopes.Experimental techniques were developed for the measurement of the concentration of the surrogate gas SF6 both in the waste and in the airways. For this, samples were collected in 10 ml gas tight syringes and injected into a closed-loop flow through a gas analyzer (MIRAN 1 A) which measured gas by absorption of an infra-red beam at a single, tunable wavelength.The feasibility of extending the application of the modelling to address issues relevant to French coal mines was examined. These issues included the modelling of the flow and dispersion of three gases (the ventilation air, the methane, and nitrogen injected to quell spontaneous combustion), steep seam slopes of up to �35? along the face, and the flow and dispersion of gas over relatively large areas of the waste. The first of these involved both the selection of suitable gases to fulfil the scaling requirements and also the development and testing of an experimental method (ethane as a tracer gas in nitrogen or argon) of measuring the gas concentrations. The second was incorporated in the specification for the construction of the model. The third led to definition of the scaling conditions for the flow in the waste which can be further reduced in scale (to 1770th) as these flows in the waste are not turbulent. This feasibility study has led to a further project which includes implementation of these proposals.The complete model of face and waste was constructed to represent (at 1720th scale) 100 m length of a 2 m high face and 60 m of abandoned roadways, i.e., a model 5 m by 3 m with the 5×3 m zone between face and roadways filled with the granular bed representing the waste. As this model was a substantial and heavy construction (when filled with granular material), it required a strong support frame with the facility to produce the required slopes of +10? from inbye to outbye and ��35? from intake to return. The design and construction of a support frame equipped with hydraulic rams to lift the model to any combination of these slopes is described.In the assembly of the model itself, some practical problems were encountered, and the solutions found for these are described. The assembled model was then tested to check the consistency and soundness of the components. The results of these tests provide a description of the degree of consistency in the construction of the waste.Finally the model was used in a range of simulations. These showed firstly that with various sites and rates of injection into the waste, the resultant concentrations spanned data obtained in real mines, which indicated that the model was producing sensible results. Then simulations were used to investigate the consequences and relative importance of various factors including the location of gas emission (injection) sources, the rates of injection, the ventilation flow, the geometry of the face end including the comparison of a back return and a conventional open face end, and seam slopes. These also produced qualitatively sensible results which should provide guidance for ensuring effective control of gas at the return end of retreat longwall coal faces.The simulations included a comparison of the effects on gas concentrations near the return end of the face, of a change in seam slope (of the order of 5?) with that of a change in ventilation (from the equivalent of the minimum to the maximum of the ventilation flows normally encountered in British mines). This showed that seam slope has a significant effect, but the magnitude of the ventilation flow is the more important.Simulations with different locations of the gas source demonstrated that the proximity of the gas sources (in the waste) to the face has a strong effect on the gas concentrations close to the back of the chocks.The performance of a back return at the extremes of its normal cycle of operation (i.e. a new snicket at the face, compared to snickets 6 m behind the chocks) was simulated. These results showed that the back return remained effective in controlling the concentration on the face at all stages. However, it was clear that the gas was kept further back from the face when all the ventilation air passed down the back return.Simulations with gas emission (into the waste) at various rates ranging from 0.2 to 1.6% of the ventilation flow rate, showed that the layering of gas has an important effect in producing higher concentrations near the face.Simulations of the situation where there is a source of gas emission (in the waste) close to the face, examined the interacting effects of variables which would increase the difficulty of maintaining control at the face end. In this situation, a change of approximately 5? in the seam slope to outbye brought the location of significant gas concentrations forward, towards the face, by approximately two snickets. This change also increased the concentration at the back of the chocks. A 10P change in the slope, from intake to return, also shifted the gas concentration by approximately two snickets.Simulations with three different arrangements of the geometry of the face end were used to compare a conventional (open)/ace end with the back return in controlling the accumulation of gas from a source close to the face end. In these tests, the back return was effective in controlling gas in the open channel (of the back return) near the face end. However, when a new snicket was cut, the concentration in the waste adjacent to the face became substantially higher. By contrast (and in accord with expectations), when the face had a conventional open face end, then the gas concentrations were markedly higher in the open air channels near the face end.The benefit of the simulations in the physical scale model is that once the effect of various parameters, or the interaction of various parameters, has been demonstrated then the improved understanding may guide the mining engineer to make the best possible use of the resources available for the control of gas in any given mining situation.Now that the principles underlying the model have been defined, the techniques developed, a substantial model constructed, tested and applied, the natural recommendation is that the full benefit should be harvested by applying the model to a wider range of simulations. The model could be used to further investigate the general effects of various parameters in a wider range of situations or to examine specific practical problems. “”