hi everybody, I am writing a comprehensive based computer document on this huge topic. I would appreciate theoretical references/technical papers on the CFD applied in the design of centrifugal,mixed flow and axial flow pumps.

There is something on my home page about CFD and centrifugal fans but it is in german.

There is also a short paper in english about a presentation at the star-cd users meeting 1996

http://www.beilke-cfd.de/handout.ps.gz

It describes my first steps on using transient calculation with moving meshes for centrifugal fans.

Joern

Sorry, can you show me how to open and read your papaer in PS file?

At first you have to use gzip which you will find on any unix system. Some people told me that win-zip also works. Then you can print it on a postscript printer. Otherwise you have to use a postscript interpreter like Ghostscript/ Ghostview to either print or view or convert it.

Thankyou for your interest and response.Iam not covering the compressible fluids but rather liquids etc.There are sites such as Fluent /Cranfield-CFD[TURBOMACHINERY]/AEAT/ASME and no doubt others.The response otherwise has been very poor. I am totally new to the new design processes cfd/fea. I am not linguistic.How much have we come from PASTERNAKS book -CENTRIFUGAL AND AXIAL FLOW PUMPS. KEN

There are some unsteady CFD results for various kinds of turbomachinery by finite element method with sliding mesh technique in my homepage:

http://www.mrl.itri.org.tw/yflow/english.htm

I hope they are interesting and helpful.

Jyi-tyan Yeh

Dr Yeh ,jyi-tyan

Thankyou,for your reply and helpful information.I found your world site interesting and well presented.Iam not able to save the volute pump animation-my home computer/modem is not presently fast enough.Very best wishes to your research programme.

Is there a more efficient geometry than the standard volute/diffusers in pump construction and design.?

In cfd work -how easy is it to cater for solids pumping ie-sewage.?

>>Is there a more efficient geometry than the standard volute/diffusers in pump construction and design.?

I don't know what you mean the "standard" volute.

Based on the assumption of uniform out-flow of impeller, the volute is designed to have linearly increased cross section area w.r.t. angle (from the tongue to the throat). Due to the viscosity of the fluid and the effect of impeller-volute interaction, some modifications of the area and shape of the volute according to the impeller are necessary (that's why CFD).

>>In cfd work -how easy is it to cater for solids pumping ie-sewage.?

It depends on whether there is a suitable reology model (effective viscosity) for the two phase flow or not. Other CFD techniques for pumping are basically the same.

Jyi-tyan Yeh

Hi Ken,

the methodologie for non compressible fluids is not very different. You just have to set the density to a constant value. This works without problems for steady state calculations. For transient cases you might have to introduce some artificial compressibility to keep the solution stable.

Dr Jui-tyan YEH,

Many thanks for your replies on volute geometry assumptions and also Phased flow.Indeed, CFD together with CAD makes for exciting research and practical uses.

How are surface finishes of impeller and casings catered for in the accuracy of results[friction/power absorbed] hence efficiencies etc of scaled up pump models by applying CFD analysis.

regards,ken

To prevent from very fine meshes, a wall function rather than no-slip boundary condition is adopted in my CFD code. The surface roughness on the wall is accounted in the wall function and, therefore, it is possible to evaluate the effect of surface finishes on the performance and efficiency of a pump by CFD. However, I am still trying to figure out an universal model considering the characteristics of surface roughness (for example, cavities on a golf ball). Could someone give me suggestions?

Jyi-tyan Yeh

Dr Jyiityan Yeh,

I would suggest you try Dr. A. Tourlidakis at the Cranfield University siterA Tourlidakis @CRANFIELD.AC.UK. on this pump housing and impeller surface finishes cfd coding aspect i.e[ roughnesses on a universal model basis.Be as specific as possible]. regards ken

hi....my name is bernard and a beginner to CFD software. Currently i'm using FLUENT code with applications in centrifugal pump. Would very appreciate it if anyone could assist me.....

I was thinking of starting off with a 2d model of the impeller and the volute b4 going onto 3d. I'm not to sure which model to choose such laminar or k-e or the RSM. Also the boundary conditions settings? Another problem is should i use structured grids or unstructured grids to model?? Is there any papers which have used FLUENT to model and predict flow field in a centrifugal pump???

thanks you bernard

there is an (2d) example in fluent5/uns for a centrifugal blower (example 16 in FLUENT5 tutorial guide volume 3)

my suggestion: start with good old ka epsilon so you know whats going wrong.

but before all that start reading in the excellent textbook of lakshminarayana on cfd for turbomachinery(Wiley?McGraw?)

the question of what type of mesh you should use is a question of what you want to have and how many hours/weeks/months you can spent with the problem. for a qualitative view on the fundamentals of the flow a quick gerenated unstructured mesh is enough. but if you want to have good numbers you have to spend a lot of time for generating a very fine and ballanced structured mesh.

to my knowledge TASKFLOW is more appropriate for problems with turbo-machins. ask dr scheuerer and/or dr menter (info.ascg.de) for furhter details.

best wished form kiel: peter

Hello:

Thought I might add a few thoughts to your questions. Have been doing axial/cent flow with Tascflow for a few years with some success.

Models: have been using Keps,turbulent in 3d analysis.

Also the boundary conditions settings? Depends on robustness of your code (Fluent)

I usually use 2 steps: first, use mass specified inlet and static pressure outlet(this gives total inlet pressure). secondly, use the solution from the first to run with Inlet total pressure and outlet static pressure.

Another problem is should i use structured grids or unstructured grids to model??

Can only speak from exp. w/structured - hex resolve boundary layers better. sliding mesh in unstructured?

Is there any papers which have used FLUENT to model and predict flow field in a centrifugal pump???

Can only recommend the advise of Dr. Brad Hutchinson at AEA Technology in waterloo canada.

good luck