Quarter O-grid topology for propeller blade
 

Hi,

I am trying to mesh a propeller blade using ICEM CFD. I tried O block topology to mesh my geometry but the quality was not good. So I was recommended to try Quarter O-grid (aka Y block). The periodic angle is 32.72727 (11 blades). The flow is along X direction. I drawn Quarter O-grid using four faces. I have attached few pic, Figure1 represents the geomentry, Figure2 - initial block round the peopeller domain. Figure3- quater O-grid by using four faces (inlet face, outlet face, left side periodic face, bottem face), Figure4-after asssociation of vertex with points. face, two periodic faces and outlet face). I don't known how to fix the quarter O-grid to capture the propeller blade geometry. I have seen some post about Meshing wind turbine but I didn't get an idea to mesh my geometry. It will be great if will be great if someone help me to fix my problem. I have attached my geometry and blk file in the following link.

Geometry file
http://jyraphe.isae.fr/file.php?h=R5...44f48e5c361015

Block file
http://jyraphe.isae.fr/file.php?h=R7...9c816d30ae511e

project file
http://jyraphe.isae.fr/file.php?h=R0...6e5ea47498ce4d

Thanks in advance for your information. please help me to mesh this geometry. I am trying for the past 1month to mesh this geometry.

do you need meshing like this? http://www.cfse.ch/cfse/site/openfoam.php

yes, I need a similar mesh. But I was told to use ICEM CFD for meshing. Could you please explain me how to use quarter O-grid for my geometry. I have gone through wind turbine meshing and dough meshing but i could get an idea how to start with quarter O-grid.

Here is my first try. I think it is successful. Also sharp trailing is handled in nice way as you can see and this is already discussed in thread structured hexa mesh on wing. Still I am associating the intermediate edges to curves on the constant span locations to capture the blade shape properly. Edges, curves and points are already created.

Also some working is needed in the tip region. This time I did not extended the o-grid into far field, because it will not suit the cylindrical far field (some one correct me, if it is wrong). Attaching the files for my intial try so that you can understand the way I am working.

Also I expect more sophisticated topologies to come latter in this discussion.


https://dl.dropbox.com/u/68746918/propeller_blade.zip


http://img35.imageshack.us/img35/1725/48393663.png

This is my idea, I will use this blocking as building block and then adopt the blocking given in previous post. This post will explain the use of Quarter O-grid http://www.cfd-online.com/Forums/ans...ic-trough.html

https://dl.dropbox.com/u/68746918/pr...rter-ogrid.zip

http://img209.imageshack.us/img209/6519/23984784.png

http://img24.imageshack.us/img24/6081/91571257.png

http://img820.imageshack.us/img820/7919/79825390.png

Change the shape of periodic boundaries in the region of blade. We shall follow the meshing strategy as discussed above and also some features from following Fig.

http://img801.imageshack.us/img801/2...olumesmall.png


http://img856.imageshack.us/img856/2...olumesmall.png

Quarter O-grid and periodic surfaces
 

Here is method for selection of faces to create the quarter o-grid...

http://img715.imageshack.us/img715/5729/84637680.png

http://img802.imageshack.us/img802/3879/27607197.png

Change periodic boundaries shape.

http://img43.imageshack.us/img43/8987/87460780.png

Hi Far,
Thanks a lot for your kind help. I am trying to understand that you did. I am going to explain you that I understood from your meshing topology (post number 4). Please correct if it there is any mistake.

Initially, you created a block around the domain and made the vertex periodic then associated Edges to curves. Then you made two horizontal (at suction and pressure side) and two vertical split (at LE and TE side) Then you made Two horizontal split ( suction and pressure side) at each section of the airfoil. then associated point to vertex then edges to curves. then you created a O-grid around the blade. then you made a split (horizontal) across the blade such that it cuts the small Lading edge bock and Trailing edge block. then two more split perpendicular to the blade (Vertical split). Till now it is clear for me because I learnt this from your video tutorial.

Actually you said that you associated the intermediate edges to curves on the constant span locations to capture the blade shape properly. With my little knowledge in ICEM, I could not understand the point clearly. Could please tell me, Why cant we associate all the curves to edges as you did for the two airfoil near the hub section and finish the blocking?.

In the post number 6, you said "we are going to change the shape of periodic boundaries in the region of blade and follow the mesh strategy as we discussed". I couldn't understand how to change the shape of periodic boundaries in the region of the blade. does it mean allingment of edges properly.

In the post number 7, you told me to change the periodic boundary shape.
do you want me to change the entire shape of the domain as you proposed.

I am trying to understand the adaption of Quarter O-grid with the above discussed statergy (like Meshing of a parabolic trough).

Once again thanks your information and help. It is really great.

Thats right

First I associated edges at hub and tip section to corresponding curves. When generated premesh, I got a mess. I had two solutions:
1. Edit edge and either make them surface projected or associate them to some curve (for which i need vertical curves, which I dont have in this model) and then project them to curve with the help of automatic linear option OR
2. I create four to five curves and then make the splits for each curve and associate them.
Although I use to prefer first approach, but this time I thought 2nd approach would do the better work. Although I did not compare both for your model. So you can try both methods.

Either you change this in CAD software by cutting in such way that it approximately follow the average blade contour or do it in ICEM.

See above. I just want you to change to shape of periodic boundaries in region of blade. Say half chord upstream, region parallel to blade and region one chord downstream. (downstream = left of blade and upstream = right of blade according to my convention). May be after changing the periodic boundary shape we don't require quarter o-grid (just a guess, but we need to try).

I will work my self on this geometry tomorrow evening as tomorrow I have job interview.

Thanks far. I will work on the model and post the doubts and results. Best wishes for your job interview.

Hi Far,

Is this the periodic boundary you want me to change near the region of the blade (figure1). I thought there will be a periodicity problem if I don't change the periodic shape at top of the far field (figure2). But there are some problem to create surface at the top of the far field. I could not get a proper surface. The blade also interact with the domain (figure 3).

Or, should I just want to copy and translate the boundary surface near the blade to far field.

here is the tin file.
http://jyraphe.isae.fr/file.php?h=R1...dd05940808ec49

First concentrate on getting the mesh for the available model and then we modify it. I tell you that how to handle the annular models.

I am trying to say, we must the master the process and when applied to better geometry topology will produce the better results.

nice job far ;)

energy382 is back. :D

Since energy382 is here with us again. He will tell us how to block this model with shifted periodic and other things from his experience of mixed flow turbine. Ralen, diamondx and BrolY are also going to share their experiences. I hope we get the very nice mesh.

I am also waiting for ultimate topology and tips from Simon.

I created quater O-grid and tried to adapt block as suggested in post 4. but could not get a fine mesh. I don't know where I am going wrong.

Firstly I made a horizontal and vertical split (Figure1) and then to create quarter O-grid, I selected 4 faces as shown in figure2. I thought of associating the edges parallel to the two curves (figure3). I made horizontal split at each section of the blade. then associated edges to curves. I tried to make a O-grid but ICEM should me an error message saying that O-grid not successful.

The mesh around the pressure side of the blade flows the profile of the blade (figure 4) but where as near the suction side, mesh does not follows the blade's profile (figure5). mesh near the trailing edge are clustered. The quality at the tip of the blade is also bad. should we want to do a special treatment
near the tip region.

Here are the files
https://www.dropbox.com/s/k28abvd4b3...quater%20O.prj
https://www.dropbox.com/s/indjfxjj4b...quater%20O.blk
https://www.dropbox.com/s/8vcbc6uod3...quater%20O.tin


waiting for your suggestions.

Hi far,

Can you please help me to adapt the quater O-grid with the method that you described. I tried a lot but I am getting bad quality elements in the tip region.
Don't know how to proceed. please give me some suggestions.

Sorry. I was out of station for few days, just came back today. Lets start it again.

Thanks for coming back. I have tried to mesh the geometry. Got a nice mesh on the blade. This time I created quater O-grid as you said and collapsed the trailling edge block. I inserted Y grid on the tip of the blade. I extend all the blocks till the far field. But there is cluster of mesh near the leading edge of the blade (Figure1) and Far Field (Figure2). Mesh in the the upstream of the blade follows the shape of blade. Waiting for your suggestions. website to upload files is not working in my lab so I am sending all the files to your mail.

I forgot to mention one thing. I aligned vertex (XY direction) in the far field (domain top ) with reference to tip vertex of the blade.

Is there any way to avoid the clustering of mesh near the leading edge of the blade. I tried to split the block near the leading edge of the blade but no change in the mesh distribution. It will be great if you give some suggestion.

Hi Far
Please, can you help me? I wont to mesh a horizontal cylinder of 5.83m length and 25.4mm diameter with (O Grid) by using Gambit 2.3 thanks.

low angle quality at Y-block
 

Hi far,
I have changed the domain shape. I had inserted Y-block at the tip of the propeller blade. In the region of Y-block, i have low angle quality as shown in figure1. Is there any way to improve the angle at the tip of the propeller.

Files can be downloaded form the following link.
https://www.dropbox.com/sh/c6e3l1hh0o1f32a/pxjVod6pAW

Hi Far,
I did a simulation in fluent. The residuals oscillates. i think it is due to low angle quality. Can you suggest me idea to increase the minimum angle quality.

Congratulation for your work and thumsup for persistence

http://img526.imageshack.us/img526/6...mbsupgreen.png
http://img818.imageshack.us/img818/1472/27187417.jpg

unfortunately it is due to geometry and blocking is perfect. So it is impossible to improve the quality (angle is 9 deg) with Y-block.

You have three options

1. Use wedge block (I prefer this). For this just leave the block as it is i.e. don't insert Yblock.

2. Inset swept block

3. Change the trailing edge shape by trimming the sharp corner.

Thanks for your support.

1. Use wedge block (I prefer this). For this just leave the block as it is i.e. don't insert Yblock.
I think, Because of sharp corner I will have low quality if I leave the bock without inserting Y block.

2.Insert swept block.
Could you please explain a bit more about inserting swept block.

3. Change the trailing edge shape by trimming the sharp corner.
Will trimming the sharp corner affects the flow behavior at the trailing edge.

1. Use wedge block (I prefer this). For this just leave the block as it is i.e. don't insert Yblock.
I think, Because of sharp corner I will have low quality if I leave the bock without inserting Y block.

No. Wedge blocks are blocks with prism elements with good quality. If one recall the prism is the shape with 3 rectangles on three sides with base and top of triangle shaped.

2.Insert swept block.
Could you please explain a bit more about inserting swept block.
In the menu, where you chose the Y-block option, you can see the swept block option there. It just change the elements from structured to unstructured and you have option either use tetra or quadrilateral.


3. Change the trailing edge shape by trimming the sharp corner.
Will trimming the sharp corner affects the flow behavior at the trailing edge.

It is common practice. Moreover from manufacturing point of view it is not possible to manufacture the wing this feature. Therefore may be (or may not be ) trimming the sharp corner affects the flow, but you will be close to practical design.

We should call it Triangular prism

thanks far. final question. should i want to insert the swept block in the y-grid or i should directly insert swept block without creating y-grid.

Directly. Check also in ICEM help

From ICEM Help

Swept


Free

Sorry I mixed up the swept block with free block. Anyhow please check both options and educate us as I am also not clear about swept block.

I converted the Y-block to a free block. The minimum angle has improved to 19. I will also try the wedge block and post the mesh.

Hi Far,
I tried wedge block without inserting Y-block at the trailing edge. There is a clustering of elements at the trailling edge as you can see in the figure1. the Determinant (3x3x3) is very low and the minimum angle is just 9.
https://www.dropbox.com/sh/c6e3l1hh0o1f32a/pxjVod6pAW

methode2: free block

I converted the Y-block to a free block. I was able to improve the minimum angle to 24 with a Determinant (3x3x3) quality 5.5. Have a look in Figure3. You can see the unstructured elements at the tip block.

Any update please

Hi far,
I had a problem in the simulation in fluent. FLUENT gave me an error saying that mixing-plane is not in the acceptable tolerance limit. I think the periodic domain boundary that i changed has created some problem.

I think the periodic domain boundary that i changed is not correct. Can you please tell me how to change the periodic boundary that follow the shape of the blade near the hub.

could you please post some pics

Actually it is a contra rotating rotor. I have meshed each rotor separately and merged in ICEM. In the figure (top view), you can see the misalignment of interface between the two rotor domain.

I have just offset the hub curve to 15.5 in the farfield and then rotated the farfield curve to a periodic angle of 32.72727.

Could you please tell me where I am going wrong.

Actually I am able to launch the calculation in fluent, but the residuals oscillates due to the problem in mixing plane (interface between the two rotor domain)

https://www.dropbox.com/sh/7r01gwggwzd78ig/veHVVDfrVc

are you giving both rotors motion with +ve and -ve rpms? Is it possible to give the one rotor rpm1+rpm2 and other one as stationary? or this problem is related to meshing? can you post more pics with different views?

why are you merging meshes in ICEM?

By definition, mixing plane should have same radial extent on both sides. But there is no limit on rotational / circumferential offset. For example rotor1 may be placed at 0-90 deg and rotor 2 may be placed within 90-270. And these values are arbitrary and make no difference into computation of mixing plane.

Fig1. View of rotor 1 and rotor 2 at hub
Fig2. View of rotor 1 and rotor 2 at shroud
Fig3. Overall view of rotor 1 and rotor 2



http://img40.imageshack.us/img40/507/propeller1.png

http://img40.imageshack.us/img40/820/propeller2.png

http://img690.imageshack.us/img690/9029/propeller3.png

yes I am giving both rotors motion with +ve and -ve rpms. It is not possible to give the one rotor rpm1+rpm2 and other one as stationary because in contra rotaing rotor, the front rotor rotates in clockwise and rear in anti clockwise direction. I am sure it is not because of meshing problem because I ran a simulation with unstructured mesh which worked perfectly fine. I also ran a simulation with structured mesh without changing the boundary shape of the domain near the hub ( with poor quality mesh). it also worked fine. So I am sure it is not because of mesh. It is due to change in the boundary shape. In the figure you can see the interaction of interface between front rotor domain and rear rotor domain. this interaction has to be avoided. I think I went somewhere wrong in changing the domain boundary shape.

"why are you merging meshes in ICEM?"

Actually the front rotor consist of 11 bades with peridic angle 32.72727 and the rear rotor with 9 blades (peridic angle 40). So meshes them seperately and merged it.

Could please tell me how to change the domain geomentry shape.

Far you are too quick :)

I seen your reply after posting my last message. so do you mean that I have to keep same shroud height for both the rotors.