TY - JOUR
T1 - Experimental validation of RANS and DES modelling of pipe flow mixing
AU - Gotfredsen, Erik
AU - Kunoy, Jens Dahl
AU - Mayer, Stefan
AU - Meyer, Knud Erik
PY - 2020
Y1 - 2020
N2 - Velocity and concentration profiles in a pipe flow was measured downstream of injection of a tracer gas at the pipe centerline. The pipe had diameter D = 0.2 m and two Reynolds numbers, Re = 50000 and Re = 100000, were used. The profiles were measured at positions 5D and 10D downstream of the injection point. Three different industrial relevant geometrical configurations were used upstream of the injection point: a 10D straight pipe, two 10D pipes connected with a 90° bend or a straight 10D pipe with a mixer plate mounted 2D upstream the injection point. In all cases, air entered the pipe from the surroundings through a sharp-edged inlet. This represents many practical flow applications and is also a well-defined inlet condition that generates turbulence in the vena contracta in the inlet. The measurements were compared to predictions from three different computational models: two with Reynolds Averaged Navier-Stokes (RANS) and one with high-resolution Detached Eddy Simulation (DES). For RANS, the k-ω SST model had difficulty in predicting the turbulence created by the vena contracta. The k-ε model performed better, but gave completely wrong results for the inlet with a pipe bend. The DES was successful for all cases with only minor deviations from measurements.
AB - Velocity and concentration profiles in a pipe flow was measured downstream of injection of a tracer gas at the pipe centerline. The pipe had diameter D = 0.2 m and two Reynolds numbers, Re = 50000 and Re = 100000, were used. The profiles were measured at positions 5D and 10D downstream of the injection point. Three different industrial relevant geometrical configurations were used upstream of the injection point: a 10D straight pipe, two 10D pipes connected with a 90° bend or a straight 10D pipe with a mixer plate mounted 2D upstream the injection point. In all cases, air entered the pipe from the surroundings through a sharp-edged inlet. This represents many practical flow applications and is also a well-defined inlet condition that generates turbulence in the vena contracta in the inlet. The measurements were compared to predictions from three different computational models: two with Reynolds Averaged Navier-Stokes (RANS) and one with high-resolution Detached Eddy Simulation (DES). For RANS, the k-ω SST model had difficulty in predicting the turbulence created by the vena contracta. The k-ε model performed better, but gave completely wrong results for the inlet with a pipe bend. The DES was successful for all cases with only minor deviations from measurements.
U2 - 10.1007/s00231-020-02835-8
DO - 10.1007/s00231-020-02835-8
M3 - Journal article
SN - 0947-7411
VL - 56
SP - 2211
EP - 2224
JO - Heat and Mass Transfer
JF - Heat and Mass Transfer
ER -