Investigation On The Turbulence Characteristics In The Near Field Of Round Jet Flow With Dns, Rans And Les

Turbulent jet flow is very common in our life and engineering, therefore experiments and numerical simulations about turbulent jet flow are one of the hottest topics in hydrodynamics. Since 1920, thousands of scientists and scholars have deeply investigated the jet flow by experiments, theoretical analysis and numerical simulations. At present, there are there numerical simulation methods including Direct Numerical Simulation (DNS), Large Eddy Simulation (LES) and Reynolds Averaged Navier-Stocks Simulation. DNS directly solves the Navier-Stoks equations without any assumption, which can get transient velocity involving with time, and then count the sample velocity filed to obtain the average turbulent characteristic. DNS have plenty of advantages, for example, it can provide all the information about flow variable including pressure fluctuation and the distribution of the vorticity which are not obtained by experiments in the velocity field at any time, so we can check the turbulence models compared with DNS, and then modifies the models. In other words, DNS can build a database which can be used to check and modify the turbulence models. This paper takes advantage of DNS to overcome the difficulty of experiment to provide reliable turbulence database which can be used to analyze the rules of diffusion on Turbulence Kinetic Energy transport equations. The paper detailed the program on the three dimensional compressed jet flow and the arithmetic of large scale simulation in parallel, compared the DNS results and the classical data of experiment. It proved that the DNS program is reliable by compared the Reynolds stress to modify RANS models and LES models. A round jet flow was computed withκ-εmodel by FLUENT software, in the jet flow Reynolds number is about 5000, and the number of grid is about twenty thousand. It is found that the peak value inκ-εmodel is commonly less than the result of DNS. So it indicated that there are much inaccuracies of turbulent characteristics in the near field. And then some modifications were made in theκ-εmodel, and a new function about turbulent viscosity was proposed, which can be as a reference to compute the near field of jet flow in RANS simulation. LES can predict a good result of Reynolds stress, which matches very well with the result of DNS, so LES can be used to compute the round jet flow in engineering. Comparing with the curve of velocity attenuation, we can find that the length of potential core in LES is obviously longer than that in DNS and RANS, the dissipativeness of the LES model maybe the reason. But at the same time the length of potential core in RANS is close with that in DNS, so it indicates that RANS can predict the first order statistics well.