Hello,

I am starting a research project in the field of hydraulic machinery in the Nottingham Trent University. I am trying to develop an efficient design of a low head propeller water turbine for small power generations and I plan to use some kind of CFD software. I know of FIDAP, PHOENIX and FLOTRAN that can be used. Can anyone help me decide what should be the way forward for such a work?

Thanks and looking forward to hearing from someone.

(1). Currently, I am checking out this code called "TASCFLOW" for turbomachinery applications. (2). The code can do analysis, so you will have to know how to design the turbomachinery in the first place.

What about the cost of the software and the system requirements?

Drona

(1). I think, the code is now under the name "CFX-Tascflow" by the AEA corporation. (one of the forum sponsor) (2). You should first visit their home website. It will have the information you are looking for. (3). We use Sun, and SGI UNIX workstations. (4). One thing you have to remember is that commercial CFD codes are not performance codes or preliminary design codes.

Dear CFDians,

Any more suggestions on what software would be the best for my work on design of small propeller water turbine?

Please ask me questions if something is not clear. Don't take for granted that I know something on CFD. I am completely novice in this field and want to learn something from a vast knowledge bank of yours.

Thanks

Drona

John, a prof of mine told me TASCFLOW isn't all that great for turbomachinery because it uses first order upwinding which damps the solution too much

perhaps if your first aim is design then you shuld find/develop a simple code to do meanline and simplified radial flow analysis rather than jumping in with both feet into using a full 3d NS code. there are a few preliminary design codes on the market and i'm sure your advisor can turn you on to some developed by gov't agencies (NASA, VKI, DERA, RAE etc) that you may be able to get for free. you can try NASA Lewis as they develop a lot of design tools for turbomachinery including optimisers and other useful things

(1). I am just checking out the code. And we have many codes on the system. I can't tell you more than that,except that it is for axial turbines in my case. (2). I don't mind what is inside the commercial code, and normally there are more than just one option. There is always a first-order up-wind scheme in the options. I normally use it to get the solution started. For very low Mach number flow ,say 0.05, the density based code normally will have trouble in convergence. (3). Your professor's statement is very vague. I would say: For high Reynolds number flows, if one is able to obtain stable and converged solutions using two different schemes for the convection terms ( the central difference, and the upwind difference schemes, or the like) on the identical coarse mesh, then the solution obtained from the upwind method will exhibit higher diffusive behavior than that obtained from the central difference method. This relative behavior becomes more pronounce when the local cells Reynolds number is high, and when the mesh is not aligned with the streamline. So, for the statement to be true, several conditions must be met. (4). In a turbulent flow, the typical Reynolds number is of order O(100). So, it shouldn't be that difficult to create a mesh with the cells Reynolds number less than 2 . In this case, most people would include a switch in the code which automatically change the numerical scheme. (5). So, for real world problems, there are many issues need to be considered. It is a good idea to know the problem itself and the code performance also. And even the first-order method sometimes is quite useful. "too much" is always a relative term.

Dear Ms. Upadhyay

We have been reading your comments and questions you posted on CFD-online about CFD package for small propeller turbine design.

We thought it might be helpful to inform you that NUMECA offers fully integrated and customized CFD tool for turbomachinery design and analysis.

Our integrated environment FINE/Turbo has been dedicated to the 3D CFD simulation of Turbines, Pumps and Compressors, including full automatic meshing tool, rapid and accurate flow solver and automatic macro-based flow visualization system. A turnkey mode allows to run a full 3D analysis in less than an hour on a PC. In addition, a series of module allows to make optimisation, with advanced inverse CFD techniques.

We would be pleased to let you know more about our product and services and we invite you to visit our we site (www.numeca.com) or to contact directly Vanessa Briffaut at vanessa@numeca.be who will be able to provide you with the necessary information.

Very best regards Marc Tombroff

___________________________________________ Marc Tombroff, General Manager NUMECA international Av. Franklin Roosevelt 5 B-1050 Brussels Tel: +32 2 647.83.11 / Fax: +32 2 647.93.98 mailto:marc@numeca.be / http://www.numeca.com ____________________________________________

As John Chien pointed out, most CFD codes have multiple discretization schemes, and this is true of CFX-TASCflow. It is possible to employ a first-order upwinding scheme when using CFX-TASCflow, but we recommend using the higher order schemes in order to obtain solutions which are more accurate than first order.

David Tweddell, Technical Account Manager, AEA Technology Engineering Software Ltd.

Regarding the comment:

"John, a prof of mine told me TASCFLOW isn't all that great for turbomachinery because it uses first order upwinding which damps the solution too much"

What interaction has your professor had with either TASCflow or AEA Technology? His comments are very misleading. You should take a look at the latest issue of CFXUpdate which can be ordered through www.aeat.com/cfx.

Gavin

(1). The life in the real world is " it is easier to get the converged solution based on the upwind method for convection term." You can accomplish something in the first place. (2). The difficulty with the higher-order methods for convection term is " it is harder to get the converged solution." (3). So, the real life paradox is " accuracy and stability do not always coexist." Accuracy means sensitivity, which means less stable, which means harder to get the converged answer.

'One has to bear in mind that a higher-order approximation does not necessarily guarantee a more accurate solution on any single grid; high accuracy is achieved only when the grid is SUFFICIENTLY fine to capture the essential details of the solution; when this happens can be determined only by systematic grid refinement.' by Ferziger and Peric, p77, Computational Methods for Fluid Dynamics 2nd Ed., 1999.

It is the fact that currently, most of industrial applications are still calculated on very coase grids.

(1). Fine mesh is not really a problem with industrial applications. The memory is very cheap. And even the commercial codes have multi-processor options. (2). So, what is the problem with the higher-order methods? It is the convergence problem. (3). This is very difficult with some commercial codes, it is very difficult to get it started with fine mesh, higher-order methods, and a very poor initial flow field. (4). So, a good initial flow field becomes the key issue. If you can't get the solver started and moving, you simply don't have an answer. For simple standard problem, this may not be a problem. For real 3-D case, with several hundred thousands of mesh points, how to create a good initial flow field is the biggest problem.