CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   CFD pump design. (http://www.cfd-online.com/Forums/main/1977-cfd-pump-design.html)

Jack Keays March 28, 2000 07:32

CFD pump design.
 
Hi! I am a research student involved in using Fluent software to redesign a single blade sewage pump. I am currently modelling a 2d unsteady turbulent model of the pump and getting reasonable reslts. However, apart from looking at the flow field results and determining losses, I am finding it difficlut to find any "theory" or empirical work which would help me in my redesign. Any help would be appreciated, Regards, Jack.

John C. Chien March 29, 2000 10:11

Re: CFD pump design.
 
(1). What is your definition of "redesign"? (2). Pumps are normally covered in Fluid Mechanics, and you can read text books, handbooks, and related technical papers.

Jack Keays March 29, 2000 10:54

Re: CFD pump design.
 
Thanks John,

What I mean is that it is difficult to find a text book, handbook or even technical papers which cover of pump of this type, namely a single spiral blade impeller geometry. I hope this helps you to help me!

John C. Chien March 29, 2000 11:25

Re: CFD pump design.
 
(1). I think, the principle is the same regardless of the number of blades. (2). In the 2-D theory, basically you are dealing with the momentum balance between the inlet and the exit, with a large number of blades. That is all you will find in the books or handbooks. (3). Beyond that point, empirical approach has been used in the pump design. It depends heavily on experience and testing. The new trend is to use CFD to simulate the flow field, which will provide more information than the test approach. (4). But then you need to learn how to read the CFD results in order to re-shape the blade. And there are other issues about other parts of the pump, the entrance area, the exit volute area, etc. (5). You can also try to get all the pump catelogs from the pump manufactures, and study the pictures and geometry. Or buy some real pumps and do the reverse-engineering. (make sure that you don't copy their design unless it is purely for your own research purpose) (6). In a way, it is a wide open field, so there are a lot of interesting things for you to do. And "CFD design" is one thing you can try. That is 3-D design, instead of 2-D or 1-D design. Your one-blade pump is on the 3-D side, it is hard to create a simple 1-D theory. (the 2-D theory can be found in the text book, just momentum balance only)

Alton Reich, PE March 29, 2000 11:47

Re: CFD pump design.
 
If I understand your design correctly, it is a single screw pump. The linage of this type of pump goes all the way back to my friend Archamedies (even though I can't spell his name) who used an inclined screw to pump water up out of rivers and lakes thousands of years ago. The "impeller" is essentially a big screw (i.e.: 1/4 - 32 pan head machine) inside a close fitting cylinder that is rotated by a motor. The rotating screw imparts energy to the fluid as it is forced to move up the screw threads.

The pump equations in a good ME handbook like Mark's Standard Handbook for Mechanical Engineers should be enough to get you started. These equations are based on fluid velocities entering and leaving the impeller. For a screw pump, the axial and radial velocities are going to be based on rotational speed and screw pitch. You should be able to make some estimates of pump performance based on the handbook equations and then compare those predictions to the results of a CFD analysis on your "baseline" configuration. That can be used to develop a fudge factor for correcting the simple equations to better match something that better approximates reality. You can then use the handbook equations to play with some fundamental input parameters like rotational speed and screw pitch to assess the effects before doing another CFD analysis on promising design changes. (Aside: The pureists out there might be offended by the concept of a fudge factor, but I've actually used this technique and it works. While handbook equations are often not particularly exact, they tend to predict trends well.)

You also might check your local large university library to see if they have any pump-oriented handbooks. Mark's is OK, but it devotes less than 100 pages to pumps. The Hydraulic institute has entire books on pumps, and may have one dedicated to screw pumps.

Regards, Alton

Jack Keays March 29, 2000 12:53

Re: CFD pump design.
 
The impellerbasically consists of a flat cylinder on to which an almost spiral line has been drawn. This line is 10mm thick. This line is then extended upwards about 80mm. This is my blade. This fits into a standard single volute. The fluid comes in axially and exits radially. A plan view of the impeller would look like a circle with a 10mm thick spiral upon it. I hope this helps you visualize my situation. Their is no pitch as such. A simple design....very empirical in nature.

Dan Hinch March 29, 2000 14:59

Re: CFD pump design.
 
Jack,

'Centrifugal Pump Design and Performance' by Japikse covers how to design or redesign pump blading using CFD results. There are other books that do the same - this book is published by my company.

Single blade sewage pump design can be accomplished analytically, but historically the design has been very emperical. This is partly due to the fact the medium being pumped is not continuous.

Joerg Weidenfeller March 30, 2000 04:09

Re: CFD pump design.
 
Hi Jack!

There are some papers dealing with screw and labyrinth pumps, but they are unfortunately in german. Maybe you will find a translation. The author's name is GRABOW, GERD and he published some papers in 1996 and 1975. In this papers there is some english literature referenced like WEINIG (1955): Analysis of the Traction Pumps. ASME Paper. That are really old papers and my impression is that the resaerch into that field of pumps isn't continously. The character of this pumps is a bad efficiency (in contrast to radial pumps) because the fluid motion is caused by generating strong vortices through the rotor and the (screwed) casing. The calculation is based on the turbulent shearing stres theory and is really empirical. Some data of experimenatl investigated pumps are: rotary speed: 1500 - 15000 rpm, circumferential speed 4 - 40 m/s, angle of the screw oriented to the diameter: 8 - 15 deg, tip clearance between rotor and casing: 0.03 - 0.1 mm.

Hope it helps

Joerg

Jack Keays March 30, 2000 06:12

Re: CFD pump design.
 
Hi! I don't think people fully know what I am describing. This pump is not a screw pump in the conventional sense. It is a blade who's plan is spiral. It is about 80mm high. It is as simple as that. Their is not angle of the screw oriented to the diameter. The bottom of thr blade is simple a flat circular plate. (half shrouded). Do you still feel that this pump can be analysed using screw pump theory? thanks!

Joerg Weidenfeller March 30, 2000 07:25

Re: CFD pump design.
 
Hi! If the flow motion in your pump is based also on friction, vortices and turbulence the principle is the same as in screw pumps and peripheral pumps.

John C. Chien March 30, 2000 12:16

Re: CFD pump design.
 
(1). There are several books on the radial pumps, including a very thick handbook. I used to have the books with me, those are in English, Japanese, Chinese, and German , mainly dealing with 2-D theory. But I think you should be able to locate one pump handbook in the library. (2). My suggestion is : try the 2-D theory in the book (mainly for infinite number of blades), find some correction factors for the finite number of blades (the flow angle at the exit will be different), and apply it to 5, 3, 1 blades configurations. You have the inlet diameter, inlet blade angle, rpm, mass flow rate, outlet diameter, outlet blade angle, blade total turning angle (length) and the height distribution (constant or variable) to play with.

Jack Ryan March 30, 2000 12:34

Re: CFD pump design.
 
Is it feasible to apply 2d theory for infinite blades with a correction factor to a system with only one blade. As you say the flow angle at the exit is different.......the fluid is effectively turned through 180 degrees as it passes through the blade. Thanks for the help.

John C. Chien March 30, 2000 14:12

Re: CFD pump design.
 
(1). Yes, for 3 blades, the pitch is 120 degrees, for 2 blades, the pitch is 180 degrees, for the 1 blade case, the pitch is 360 degrees. (2). So, I don't think the single blade case is singular. But then one can always try to compute the 3-D flow using 3-D CFD codes, for 3, 2, 1 blades cases and come up with some empirical factors in terms of the number of blades.

Des Aubery March 30, 2000 14:35

Re: CFD pump design.
 
Hi John,

I list a comment from your last email:

"for 3 blades, the pitch is 120 degrees, for 2 blades, the pitch is 180 degrees, for the 1 blade case, the pitch is 360 degrees. "

Something worries me about this extrapolation - since from what I understand the blade is actually a half blade. My guess is that the 360 degrees would really represent a solid blade. My gut feeling is more for 1 blade, at 180 degrees.

I haven't had a chance to dig up my pump-design eqns, but I would honestly try the latter approach first - unless it violates an assumption.

Best regards,

Des Aubery... (adTherm Technologies)

John C. Chien March 30, 2000 17:30

Re: CFD pump design.
 
(1). What I was trying to say is when you have a radial pump with two blades in it, they are 180 degrees apart, looking at the blade leading edge on the inlet circle. (2). When you remove one of the blade, there will be only one blade left to cover the 360 degrees circle. That's all. I was not talking about any equation or formula. (3). It is similar to a single-nylon-wire weed-eater when it is spining. I hope we are talking about the same thing, it is basically a 2-D centrifugal pump. (4). In theory, the 2-D pump with infinite number of blades requires only the simple inviscid theory, the total number of blade can be used as a parameter to modify the theory and empirical formula. For one-blade pump, this may no be available yet. I personally have not seen such formula before for single blade pump.

Des Aubery March 30, 2000 18:05

Re: CFD pump design.
 
Hi John,

I agree with your points here.

If I remember correctly, the pump is essentially treated as a "shear disc", and the greater the number of blades, the more closely the impeller approximates this shear disc.

With one blade, the effects should be really quite interesting. It could be very interesting to model this...

Best regards,

Des Aubery...


All times are GMT -4. The time now is 01:41.