Inverse-design and optimization methods for centrifugal pumps and fans (TSF.6157)
Project nummer:
tsf6157
Omschrijving van het onderzoek
Research
The main objective of turbomachine design is to achieve efficient, lightweight and compact machines, while reducing the cost and duration of the design cycle in order to meet the rapidly changing demands of the market.
The design of turbomachine components, such as rotor (impeller) and stator (diffusor) blades is a complicated task due to the high and often contradictory performance requirements. The performance of a turbomachine is determined by the flow that occurs inside the machine, and the pressure distribution on the blades in particular.
Up to a few years ago the designs usually were based on the experience of the designer in conjunction with largely empirical models and a trial-and-error procedure. Disadvantages of these design methods are high design costs, long design cycles and, at times, insufficient performance of the resulting designs.
Presently, these empirical design methods are complemented with more rational computational fluid dynamics analyses based on Reynolds-averaged Navier-Stokes flow models. These models have their limitations, such as: (i) very large requirements in terms of computer resources, (ii) large influence of mesh size and mesh quality near solid walls on the predictions, (iii) the choice of an appropriate turbulence model that realistically accounts for the effects of three-dimensional boundary layers, curvature and system rotation is far from trivial and (iv) accurate modelling of boundary layer transition and separation is of great importance for the accuracy of the off-design predictions.
We propose to use a quite different flow model, the augmented potential-flow model. We have already investigated the feasibility of this model for analysing the performance of centrifugal pumps and fans. For flow analyses around the Best Efficiency Point the augmented potential-flow model, with its modest computing requirements, is far more suitable for design optimizations than methods based on Reynolds-averaged Navier-Stokes flow models.
Using the augmented potential-flow model, two types of design methods will be investigated: (i) inverse-design methods and (ii) optimization methods. We will focus our attention on centrifugal pumps and fans, since this type of turbomachine is of great industrial importance and it has not received sufficient attention.
With inverse-design methods, the designer can specify a desirable pressure distribution on the blades; the computations will provide the geometry of the blades that results in this pressure distribution. Such an inverse-design method is far more effective for the designer than the cumbersome iterative process of: (i) computing the pressure distribution on the blades for a given geometry, (ii) checking whether the desirable pressure distribution on the blades is obtained and (iii) updating the geometry (if necessary).
In optimization methods, the design is optimized iteratively in terms of a cost function specified by the designer. This cost function involves desirable pressure distributions on the blades, mechanical and economical constraints, etc.
We will develop three-dimensional inverse-design techniques, novel numerical methods, three-dimensional direct optimization methods and multi-point optimization methods. We will also investigate the use of a hybrid method called inverse-design optimization.
The developed methods will be validated by testing a scale model of a centrifugal pump in the test-rig of the Department of Mechanical Engineering of the University of Twente.
Utilisation
The project will provide the designer of centrifugal pumps and fans with next generation tools that will allow him to make better designs in a shorter time span, and thus improve the competitive edge of the industries involved. These tools will closely match the designers way of dealing with the design problem. Various industries have already expressed interest in this project. They have also committed themselves to provide financial support for this project directly to STW.
The inverse-design and optimization methods to be developed will be implemented in computer programs. These programs can be used to develop completely new designs, or to improve existing designs.
New design methodologies will be formulated that can be used by industry to complement their existing design methodologies. Together with the user group test cases for the inverse-design and optimization methods will be identified.
To ensure that the new methods will be used and accepted in practice, requirements for the computer programs to be developed are: (i) ease of use, (ii) limited requirements on user input, (iii) easy visualisation of results and (iv) availability of interfaces to CAD-CAM packages. Furthermore, courses will be organized on the scientific basis, capabilities and use of the programs.
Resultaten van het onderzoek
More information can be found on the website of the research group.
Gebruikers
Er zijn 7 bedrijven bij dit project betrokken.
Projectleider
| Prof.dr.ir. H.W.M. Hoeijmakers |
Universiteit Twente
Construerende Technische Wetenschappen
Technische Stromingsleer |
Postbus 217
7500 AE Enschede
|
Status van het project
| Gestart | : 01-11-2003 |
| Einddatum | : 01-11-2006 |
Trefwoorden
Aerodynamica, Computational fluid dynamics, Inverse modelling, Ontwerpsoftware, Optimalisatie, Stromingsonderzoek, Turbine, Turbine blade.
.