[StefanG]

Dear CFD-Online community,

I experienced some convergence problems with the simulation of a flow channel with a small slit for the cross-flow-injection of a vaporized kerosene jet (see figures 1 and 2). The figures display just half of the geometry to be meshed, because of my restricted computer resources. The symmetry plane is the black surface in figure 1.
The aim of my simulation is to investigate the mixing of the kerosene with the air. By now I just made simulations with ANSYS CFX without any injection of kerosene through the slit. One of the possible reasons for the lack of convergence might be my mesh. Due to the inlet conditions (total pressure: 2.63 bars; total temperature: 600 Kelvin) and the conditions at the outlet (static pressure: 1 bar; total temperature: 300 Kelvin) just a part of the divergent part of the nozzle will be supersonic. As I am not a very experienced ICEM and ANSYS user I now have some questions concerning the improvement of my mesh in order to get better convergence in ANSYS CFX.

1. As I am using a hexa mesh for my entire geometry the shape of the convergent part of the nozzle is quite steep and so the min angle criterion in this region is quite bad (see figure 3). To improve this, I tried to generate some prism layers in this region, but I failed generating the with the part mesh setup. As I want to model also the near wall behavior of my flow with the SST model, I generated a node distribution near the walls of: spacing near the wall: 0.0001; ratio: 1.1. This ratio I wanted to keep with the prism layers for the nozzle part (height 0.0001: height ratio 1.1, layers 10), but I failed to make the connection between the nozzle part and the rest of the flow channel. Has anyone an idea how to solve this?

2. Furthermore I have problems with the meshing at the end of the flow channel (figure 4). Can anyone tell me whether this meshing at the end of the channel is ok or should be improved and if so, how could I improve it?

3. Figure 5 displays the meshing of the connection of the kerosene injector and the bottom of the flow channel. Here I have a problem with the node distribution on the parallel edges of my blocking (see figure 6, blocking strategy in this region and figure 7 for general blocking strategy). On the bottom of the channel we see two edges with 50 nodes distributed on the edges of the flow channel's breadth. Due to my meshing strategy there are there are two other edges (which consist of two edges each) on the level of the kerosene slit between these two edges. My question is now, how it is possible to have the same node distribution on these two edges, as on all other parallel edges?

4. As mentioned above, only in a smart part of the divergent part of the nozzle the flow will be supersonic. According to Q1D theory there must be a straight shock in the divergent part under the conditions at the inlet and the outlet. Might it be the reason, that my mesh isn't fine enough in this region to obtain convergence in my simulations? Does Ansys CFX generally have problems in solving these kind of simulation where a straight shock in a nozzle is expected?

I now I wrote a lot of text, but I would be very grateful for any ideas regarding my questions. I also would be very glad for every general hint on my meshing and blocking strategy, for example possible simplifaction, other meshing type and so on.


Thank you very much in advance for your support,

Stefan

[StefanG]

Here the two missing figures.

[Far]

Could you attach the blk and tin files so that we can help you.

[StefanG]

Thanks for your quick reply Far!

I am sorry but my zipped tin file is 102.1 kb big so due to the max. file size of 97.7kb I am unable to upload it. Any suggestions how I could manage it?

[Far]

I use 4shared.com for uploading larger files. http://www.4shared.com/

[StefanG]

Thanks. Here the url: http://www.4shared.com/folder/NXU4gmU-/_online.html

[Far]

Here is the blocking strategy for the divergent section, you can extend blocking to full geometry using the option "extrude along curve" in blocking panel


http://www.4shared.com/zip/aim2UwHx/...annel_Far.html

[Far]

Applied Y-blocking for both corner and detailed procedure is discussed here.

http://www.cfd-online.com/Forums/ans...tices-why.html

http://www.youtube.com/watch?v=92vScop7b8Q

[StefanG]

Thank you very much for your support. I will now try to redo the same meshing by my self. Really good blocking strategy! Thanks alot!

[Far]

First I have made block only for the convergent section (earlier I refereed as divergent mistakenly) then extruded blocks for remaining geometry and as well for small inlet. Now I am attaching files for complete geometry. Boundary conditions are specified and mesh output was checked in Fluent and every thing is fine. Quality is above 0.6.

http://www.4shared.com/rar/APdi_-ZB/..._Far_full.html

[StefanG]

Thank you very much for your help Far! Finally I have a ggod mesh, but still I haven't yet achieved any convergence for my flow problem, so I will go on in the CFX Forum. Maybe you wanna have a look on my problem there, too?

[Far]

Finally I have a ggod mesh, but still I haven't yet achieved any convergence for my flow problem, so I will go on in the CFX Forum. Maybe you wanna have a look on my problem there, too?

Ya sure. I would be happy if I can help you there as well.

[StefanG]

Thank you! Right now I am running some more simulations with the geometry and as soon that I will have the results, I will post here the link to the thread I will create in the ANSYS CFX Forum.

Best regards,
Stefan

[Far]

Could you please post few pics of your final blocking along with quality metrics e.g. angle, quality etc.

[StefanG]

Sorry that it took so long, but this the version of my geometry and meshing I will now use for my simulations (see attached pictures).

I moved the inlet farer away from the convergent part of the nozzle and also changed the vertical wall at the end of the flow channel, as this caused a strange recirculation zone in this zone. The maximum node spaching is less than 3.5mm.

I integrated Far's idea with the y-block, which helped me a lot in improving the meshing quality (min angle criteria). But in addition I used an O-Grid to be able to obtain a good near-wall meshing for the SST-model I intend to use for my simulations.

The min. angle is slightly bigger than 40 degrees and the max. aspect ratio is 57.3 and the min. quality is bigger than 0.64.

If anyone has some more hints for me, I would be glad to read them here. and if someone is interested in the blocking, I can upload it here.


Best regards,

Stefan

[Far]

where did you put the O-grid?

[StefanG]

First I initialized a block around th nozzle. and then I created an O-Grid in this block, selecting the faces all the faces except of the faces on the upper and down side of the nozzle.

[Far]

would you like to show some pics with o-grid?

[Far]

Ok got it .

[Far]

If anyone has some more hints for me, I would be glad to read them here. and if someone is interested in the blocking, I can upload it here.

Please upload .tin and .blk.