|
发表于 2003-2-26 00:43:37
|
显示全部楼层
请教版主一个问题
First question in mymailyh: In EURANUS, the flow solver in Fine/Turbo, there are basically two methods to treat quasi-steady computation in multi-blade rows: Mixing plane and Frozen rotor.IRSNEW =1, means that the interface is FULL-NON-Macthed (FMG), IRSNEW= 2 means Frozen rotor method is used. When IRSNEW=2, the whole rotor and/or stator has to be meshed. That is why the blades equal to 1.
Second question in mymailyh: Are you computing centrifugal impeller with volute? Is your ROT BC the interface between the impeller and the volute?
Please check if the flow path between these two part are well fitted in the meridional view.
In addition, is the computation fully converged? This error of mass was got from coarser mesh or fine mesh computation? Do you have more than one STRIPs in the interface (Fig.9.2.1-1)?
Third question in mymailyh: You have to realize that in the Full-MultiGrid computation which is only used for creating a better initial flow field, first order scheme is used. Hence the convergence is generally fast. After the Full-Multi-grid cycle is finished, second order scheme is used. So, the convergence problem could come from following different reasons:
1) Quality of the mesh.Please check the mesh quality from IGG. A better convergence and high accuracy may be got with Aspect Ratio <1000; Extension<3.
2) Operation point. You have to realize that different mesh level may give different working points even though the boundary condition is the same, and a converged solution can only be obtained in the machine working range. A suggestion is: if the mass flow is too low in the FMG cycle, or lower than what you expected, you should reduce the exit pressure.
3) In addition, computation with finer mesh will predict more detailed vortex or separation flow, which may also make the convergence difficult. Hence, you may reduce CFL when the finer mesh is computing.
Question in Batman: Please check if the two different inlet boundary conditions give the same computed working point. If it is different, you should then modifying values which make the divergence.
Hope my reply give you some help!
|
|