|
|
|
|
|
|
|
|
|
|
Description: Wind Fields - Convergence | Version: 4.9.0 | Updated: 10.03.09 | |
||||
|
||||||||
The solution to the wind field is found using an iterative procedure. The initial conditions are used as a start field. The solution is corrected in each iteration, presumably a converged solution is found after a certain number of iterations. A graphical window displaying the field values of the variable used for convergence monitoring is displayed. The field values are displayed at the cells adjacent to the ground. When the field values don't change anymore then a converged solution has been found. In order to limit the size of the files containing the field values, the maximum steps in the animations is set to 200. Hence, if a simulation is started with more than 200 iterations, then there is no direct correspondence between the steps of the animation and the iteration, and skipping will be performed.
|
|
|
|
||
| Figure 1. | Graphical monitoring of the convergence. Convergence is reached when the field values become constant. | |
By default a "Spot value" node is defined in the middle of the terrain at ground level (nx/2,ny/2,1). The values of all flow variables and derived variables are given for each iteration at the spot value node. A converged solution is obtained when the flow variables become constant, i.e. when the changes given in the fourth column become negligible. The "Spot values" are scaled according the Min and Max values found in the two first columns in the monitoring window.
The residual expresses the error in the numerical solution. It is the quantity obtained when all terms are put on one side of the equation; Left side - Right side,Residual. For a fully converged solution the residual will be zero. The % Error column gives a weighted residual
Both "Spot values" and "% Error" are saved in the files [project folder]\windfield\nnn.result, where nnn is the sector angle.
Models with abrupt terrain changes might fail to converge. Likewise if unstable situations occurs, typically in areas with recirculations, and where time-dependent solutions should be considered. In general it is easier to get converged solutions in small models than in large models in terms of number of cells. If the solution diverges the nesting procedure could be used, starting with a coarse model and by doing successive refinements in a specific area. One should also consider to activate the "Terrain smoothing" and "Orthogonalize 3-D grid" in the module Terrain.