Abstract
This paper proposes the method of estimation of error caused by coarseness of finite-differencing in numerical simulation. The results of numerical simulation have solution error which is due to the truncation error of finite difference equation. The solution error can be calculated by using Richardson extrapolation and the estimation of the exact solution can be obtained. The truncation error also calculates from the estimation of the exact solution. In case of 0-equation model (vt=constant) the use of finer mesh dividing leads to decreasing these errors. But in case of k-ε model these errors don't always decrease by use of finer mesh dividing. Truncation error of k-ε model have a tendency to become larger than the value of each terms like flux or diffusion term in k-ε model equations at the region of impinging jets and at surroundings of supply outlets. It is suggested that the investigation of the mesh dividing and the turbulence modeling, including the tuning of the empirical constants, have to be examined still more.
