Resumo:
The development of a mathematical model and a computational code that simulate
a two-dimensional, incompressible and isotherm of turbulent jets and secondary
inductions are presented in this work.
The computational program uses a method called Finite Volumes Method, [Patankar,
1980] in order to solve the spatial equations.
The model of turbulence k - e is tested and compared.
The development of the physic domain is done by means of Staggered Grid
strategy and the SOLA (Solution Algorithm) Method [Hirt et al., 1975].
Two convective schemes are presented, the UPWIND scheme and the QUICK
scheme.
The linear equation system solution is obtained by applying the Choleski
Method [ Brebbia, 1978].
The checking of results obtained from the program, in FORTRAN language, is
presented.
It obtains for laminar flow, inside square cavity, results from the program which
fitted satisfactory with the ones from the literature. Comparing the components profiles
of mean velocities, in the middle of cavity, for different Reynolds numbers, with the
standard results from Burggraf (1966), they converged up to more than 10.000 iterations
for all cases presented.
It determines with turbulent flows, inside circular ducts, the components profiles
of axial mean velocities, the kinetic energy of turbulence, and the square and mixed
Reynolds tensors. The results are compared to available experimental data.
Tests were performed for a circular pipe with a length-to-diameter ratio of 57,4.
The flows are atmosphere air getting with flat profiles of velocities, in order to
obtain velocities profiles when a steady-state is establish inside the duct.
To compare the results obtained with experimental data gotten from Razinsky
and Brighton (1971) and also with simulated from Jen (1989), it uses a Pentium III
computer, 850 MHz, 128 Ram.
It has been found that results obtained with the method developing in this work
are in better agreement with the experimental data than those obtained from other
simulation method.
The computational time is approximately 5 minutes when compare the results
simulated using a same grid (20 X 15).
A grid test is realized in order to define the converged grid.
The converged grid is 60 X 30 and the computational time, approximately 28
minutes.