Construction and validation of models for a study of the circulation of gas in the waste of French longwall retreat coalfaces

This study was part of a programme of work to develop aerodynamically scaled physical models (of French retreat longwall coalfaces) for the purpose of investigating the flow and dispersion of methane gas and the nitrogen gas which is injected to quell spontaneous combustion. The study follows on from a British Coal / CEC project on the modelling of the dispersion of methane gas in longwall retreat coalfaces (Jones and Lowrie, 1993) which derived scaling relationships for obtaining aerodynamically scaled flow (with respect to turbulent dispersion, dispersion of gas layers in turbulent flow, dispersion in the waste, pressure gradients due to ventilation and gas density differences, and the balance of flows between face and waste). These relationships defined conditions for operating a l/20th scale model (of a 2 m high longwall coalface plus waste) which was designed, constructed and used. This original model of the face and waste was also suitable for representing, at l/30th scale, a typical 3 m high French coalface. Some modifications to this model were needed to make the profile of the intersection between face and waste correspond to the geometry of a French coalface.The scaling relationships showed that 1/30th is the maximum possible reduction in scale for a model with aerodynamic scaling of the dispersion of gas layers by turbulent flow. However, a further reduction in scale is possible for the flows in the waste where the flows are not turbulent. This report describes the design, construction, calibration and validation of a new (l/70th scale) model representing the flows through a waste extending (at the equivalent full scale) 200 m along the face and up to 280 m from the face start line. This 1/70th scale model measured 3 m along the face by 4 m from the start line. This model was mounted on a table which could be tilted by hydraulic rams to give seam slopes of up to ��35? along the face and �� 10? in the transverse direction.The two models are complementary; the role of the 1/7Oth scale model is to provide information on the flow and dispersion of gases within the waste. The role of the l/30th scale model, which includes aerodynamic scaling of the turbulent flow on the face, is to examine the interchange of gas between face and waste and the dispersion on the face. The results obtained in the l/70th scale model should guide the simulation of gas emission into the waste of the 30th scale model.Data describing a real underground longwall retreat face (DORA 1 SUD) at the mine Reumaux are described. These data provide the basis for calibrating the model by preparing the distribution of permeability and gas emission within the model waste to match the distribution in the computer simulation of the same coalface. The ventilation, drainage and gas emission flows were set up to represent those measured underground. The pressure gradients along the full scale face were estimated by calculation of the friction factor using data from another coalface (ERNA 3 NORD).Validation consisted of checking the consistency between the full scale and the model for: the distribution of flow between the face and the waste; the concentration of gas in the return airway; the concentration of gas in the drainage from a chamber at the intersection of the face start line and the abandoned return roadway; the magnitude of the pressure gradient of the model face; the transit times for pulses of tracer gas to cross the waste from the intake side to the return side and from the intake side to the drainage. These transit times were measured in the full scale and in the model by the Institut National de I’Environnment Industriel et des Risques (INERIS). Initial results from these measurements indicated where modifications were needed to improve the realism of the model. These improvements led to reasonably satisfactory agreement between the model and the full scale. There was also a qualitative consistency between die distribution of concentrations in the physical model and those in the computer simulations, although there were differences in the nature of the data. For example, the computer simulation treats the waste as two dimensional whereas the concentrations measured in the physical model were mid-seam concentrations; consequently the concentrations in the computer simulation include the effect of all upstream sources, whereas the finite mixing distance in the physical model meant that the mid-seam concentrations corresponded to an average concentration for gas from sources more than the equivalent of 10 m upstream.In summary, the physical characteristics such as the permeability and the rate of gas emission vary enormously across the waste but, compared with the uncertainty of defining exactly what these are in the full scale and the practical difficulties hi matching the distributions in the model, the level of agreement between the target specification and the measurements in the model was reasonably good. This provides a basis for using the model to investigate the consequences of various changes in parameters such as the drainage flow, drainage location, location and flow of nitrogen injection, changes in ventilation, operating similar faces with different seam slopes, changes in the distribution of the gas emission, etc.