I am doing a research project on the design of a centrifugal single vane wastewater pump. I have only just begun, and I am still very much on the low end of the learning curve. I am going to be using Gambit and Fluent as my main design tools. I need to purchase a computer which I can use to run these two packages efficiently and effectively. Any help as to a good specification of a pc would be appreciated. I mean this in terms of RAM, Hard disk space etc.... Also, it would be nice if anyone who may have information or know of a good place to find information which would be relevant to my project could let me know. Thanks a million, Jack.

Dear Jack

It dependes on your problem and your decesion for grid generation, not depends on which code you are going to be using. For my case, in general, personal computer with 160M RAM, 4GB HARD and 400MHz CPU was fairly good for my 2D steady simulation. I am using high speed workstation for 2D unsteady or 3D(steady or unsteady) simulation.

It seems that your application can be solved by fluid flow calculation only(continuity and momentum). If your problem is 2D, above mentioned or lower technical spec. would be enough for your simulation. If your problem is 3D and you are going to be using personal computer, I recommend higher speed CPU, more RAM and big HARDDISK. That is to say, best spec. PC within your budjet.

Followings are my personal point of view for your reference. Do not attempt to obtain everything by CFD. Instead, clearly define your objective function, or design goal, then you can save calculation time and your efforts very much.

SIncerely, Jin-Wook

Have you got any idea how large cases you intend to run? Will you base your gambit model on large cad drawings? Which CAD system will you use?

Gambit is still a bit difficult to use for turbomachinery applications where you want to resolve your boundaries well, especially if you have complex blade shapes with double-curved surfaces - the boundary layer handling in gambit is not that stable.

My initial CAD drawing will be from AutoCAD. This will be a 3d drawing. Also, what do you mean when you say double-curved surfaces? Anyway, all comments which may be useful to a complete novice like myself will be much appreciated.

Dear Jin, Well I will be running a 3d model of the problem. So I think a pretty big pc is in order. My objective is to reduce the amount of power used to run the pump. i.e., increase effiency. I have no idea at present what the causes of ineffiecy may be, but I think close CFD analysis of the current pumping system should give me enough information to find out what parts of the impeller, casing etc. are causing losses. Any comments are truly welcome! Jack.

AutoCAD is good - you'll be able to import it directly as real geometry in Gambit via the ASIC format. If you intend to mesh large CAD-geometries in Gambit you need a fast computer with plenty of memory - Fluent should be able to help you with recommendations about suitable hardware.

Double curved surfave - a surface which is curved in two directions - a cylinder is "single curved", a part of a sphere is the simplest example of a "double curved" surface. Any surface where you don't have "straight lines" in one direction is "double curved". Hope I'm making sence.

Thanks....yes I understand...I guess a cylinder is a double curve with one curve infinite in size.?? Anyway, what do you recommend in terms of RAM, CPU, Hard drive space etc.? The impeller consists of a single blade. The blade is mounted directly onto a base. The blade is essentially spiral in shape and sweeps throught approximately 300 degrees of a revolution....getting closer ( obviously!!) to the edge (volute) as the angle increases. The problem with this geometry is I cannot find any information in journals or books which may be of use to me. Can you or anyone help me? Is it a typical sewage/wastewater pump?? Thanks!

Dear Jack

I am not a specialist in turbo-machineey area, so that I can not give you detail information. Sorry.

Within my knowledge, in general, secondary flow, recirculation and flow with preponderant regin may cause inefficiency. That is to say, uniform flow along the main direction would be very good in general.

If you don't have insight about your application area, at first, try some simulations with very coarse grid net(you are going to use PC). Then you can have some knowledge about your system. After coarse grid simulation, try detail simualation with more grid, especially near your interesting area in the flow field.

It might be poor answer, but hope that it will help you.

Sincerely, Jin-Wook LEE

Dear Jin, Thank you for your comments. I have one question. What does preponderant regin mean? Excuse my ignorance on this issue! Thank you for your advise on using a very coarse grid for initial simulations. My major problem with regard to meshing is trying to mesh the shape of the geometry. A single vane impeller where the blade is a spiral filling 300 degrees of a circle. Anyway, thank you again, Jack.

Dear Jack

<Preponderant region> I guess that it is my poor English, not your fault. I mean that more flow in one region(say, with 10 m/sec),less flow in onother region(say 2 m/sec) and velocity not toward side boundary and so on .......

If it is still poor for you to understand, give me e-mail, then I will explain you with graphical tool.

Sincerely, Jin-Wook LEE