An experimental and numerical study of two-way splits and junctions in mine airways

Although several studies in combining and bifurcating flows in two-way junctions and splits are reported, deviations persist in the estimated shock loss coefficients (SLCs). Due to their complex nature, more work has to be done to be able to correctly predict flow losses at these locations. In two-way splits, the flow at bifurcation is known to be characterized mainly by two recirculating zones in the downstream branches, whereas, in a junction one recirculating zone is observed immediately downstream of the main branch. The relative flow rate in the branches (quantity ratios) is treated as the only parameter defining the losses at these locations. However, the effect of wall roughness, if any, on SLCs becomes important for mine ventilation. This paper presents the critical determinants of shock losses at splits and junctions at varying quantity ratios. The flow at 90º split and junction was examined both experimentally and numerically for Reynolds number in the range of 1.0×104 to 1.6×105. With a well designed laboratory setup, using precision instruments, velocity and pressure measurements were carried out, by using two scales of wall roughness. 3D numerical simulations for these experiments were conducted by using ANSYS CFX modeling tool. CFD simulations yielded results that are reasonably close to experiments, providing high degree of satisfaction in numerically predicting losses across the splits and junctions. However, comparisons of these numerical simulation results with existing literature standards showed significant under-estimation of shock losses by the standard literature by as much as 50%. A clear rise in the magnitude of shock losses was noted in both the approaches, with increased wall roughness. Based on the current studies, predictor equations incorporating quantity ratios are developed for accurate estimation of shock losses. Further investigations on roughness are needed, before one is able to incorporate wall roughness into the prediction of SLC.