In this dissertation, the study of jet in a cross flow was based on computational fluid dynamics. There were two different models for simulating the jet in crossflow. The performance of turbulence model named Reynolds-averaged Navier-Stokes equation, abbreviated as RANS, was studied and compared. The simulations for two configurations were performed by software Star CCM+ whose numerical method used is finite volume. The velocity ratios of mean jet velocity to free-stream velocity for two cases were 1/6 and 5.7, respectively. Corresponding Reynolds numbers which are based on bulk jet velocity and jet-exit diameter are 3333 and 5000. The computational domains were also different. Air was used as the fluid for investigation of simulations. The turbulence models specified in this dissertation were realizable k-epsilon two layer model (k-epsilon), elliptical blending Reynolds stress model (EBRSM), elliptic blending k-epsilon model and V2-f model.For Configuration 1, the simulated conditions were based on the validated data from Zhao Wu. Due to the location of the jet in Zhao Wu's large eddy simulation, the length of the streamline was shorter. The results of streamline velocity and turbulent kinetic energy simulated by kepsilon turbulence model were in good agreement with DNS model. For simulating the velocity and turbulent kinetic energy at different stations vertical to the bottom wall along the symmetry plane, the EBRSM performed well among four turbulence models. The results of EB-k-epsilon model deviated the most from the DNS model.For Configuration 2, simulations were conducted at the same conditions by Muppidi & Mahesh (2006). When simulating the center streamline, three turbulence models under predicted the horizontal velocities. The EBRSM predicted the tendency of trajectory-parallel component of velocity us the best. However, the models matched the tendency of streamline velocities. At the further station parallel to the wall along the symmetry plane, three models overpredicted the vertical velocities, but the trend was same. At the lower stations, they all matched with DNS model well.
|Date of Award||1 Aug 2017|
- The University of Manchester
|Supervisor||Imran Afgan (Supervisor) & Dominique Laurence (Supervisor)|