The concept of residual in fluent / starccm / commercial CFD software

The residual is the sum of the flux of each face of the cell. When converged, in theory, when there is no source term in the cell, the flux flowing into each face, that is, the sum of the transport of physical quantities should be zero.

The maximum residual or RSM residual reflects the residual of the flow field and the flow field to be simulated (only the flow field that should be obtained after convergence, of course, there is still a certain gap between the flow field obtained after convergence and the real flow field). The smaller the better, because there is a numerical accuracy problem, it is impossible to get a residual of 0. For single-precision calculations, it should generally be lower than the initial residual of less than 1e-03. Continuous terms are not easy to converge and energy terms are easy). 

Generally, the inlet and outlet flow can be monitored in FLUENT. When the residual converges to a certain degree, it is necessary to see whether the inlet and outlet flow is stable and balanced before determining whether to converge or not (the balance of lift resistance must be monitored during airfoil calculation).
If the residual oscillates at a higher level, it is necessary to check whether the boundary conditions are reasonable, and then check whether the initial conditions are reasonable, such as the shock wave flow field. The improper initial conditions will cause the flow field to oscillate. Sometimes the flow field may have separation or backflow, which is an unsteady phenomenon in itself, and the residual error will oscillate to a certain extent during calculation. At this time, if the inlet and outlet flow rates reach a stable balance, the flow field can also be considered to be convergent. In addition, fluent uses a multi-grid by default. Setting the multi-slave grid to 0 in the later stage of the calculation can prevent some wavelength residuals from oscillating on the fine grid.

Guess you like

Origin www.cnblogs.com/2016-11-13/p/12726627.html