Introduction Usually hydraulic turbines have to be designed individually according to the local operating conditions of power station such as discharge, head and given geometrical situations. This requires a tailor-made design mainly for the turbine runners. The shapes of the runners are rather complicated and therefore the understanding of the complex geometry and especially of the spatial structures of the flow are very difficult to analyze by two-dimensional display tools.

Many different views and cuts have to be prepared and analyzed, which is quite uneconomical and time consuming. In a virtual reality environment, however, the complex geometry and especially the flow behavior can be controlled much faster and in more detail. This is particularly the case for students, who are not completely familiar with turbine shapes. Consequently a design tool based on VR-techniques is developed at IHS and RUS. The system is used for educational purposes as well as for industrial applications.

The primary design is obtained on the basis of Euler's equation for turbomachinery. The design parameters such as head, discharge, speed, number of blades can be changed by sliders. Other parameters, e. g. blade angle can be manipulated by interaction in VR using 3D interaction devices. Since Euler's equation is relatively simple the geometrical shape can be calculated online. Therefore many sets of parameters can be investigated and optimized in a short time.

After the primary design the blade shape is fixed and can then be analyzed by numerical flow simulation. The flow calculation, based on the Navier-Stokes equations, can be carried out on a supercomputer in order to achieve a short response time. The calculated flow behavior is studied in the VR-environment. Again, it is much easier to understand the complex relations between the flow and the geometry in VR than using two-dimensional display facilities. This is principally of great importance for people not entirely accustomed with turbine runner flows. Consequently the VR-environment is particularly important for educational purposes.

Visualization Software COVISE is a software environment developed by the Computer Center (RUS) which tries to integrate visualization and simulation tasks across heterogeneous hardware platforms in a seamless manner. The user interface is based on the visual programming paradigm. Distributed applications can be built by combining modules (modeled as processes) from different application categories on different hosts to form more or less complex module networks.

At the end of such networks usually the rendering step does the final visualization. A special feature of COVISE is that it allows several users to work in a collaborative way providing online consulting to end users at remote sites or tele-teaching. For the visualization COVISE supports desktop as well as VR oriented renderer modules. Hardware equipment For the VR based visualization two different environments are used, one at IHS and one at RUS. In both environments the visualization runs on local SGI workstations, the flow simulation program runs on the super-computers of HLRS (Höchstleistungsrechenzentrum Stuttgart) either on a NEC SX-4 vector computer or on a CRAY T3E computer in parallel.

ISIMADE, Baden-Baden, 1999 The workstations and the super-computers are connected either by a high speed network (HIPPI, ATM) or by FDDI. The environment is schematically shown in fig. 1.

CRAY T3E 512

RUS

VR

HIPPI

environment

SGI Onyx2

FDDI

IHS

VR

NEC SX-4

environment

SGI Crimson

HLRS

Fig. 1: Working environment

The currently installed hardware at IHS is a single wall back projection system driven by a SGI RealityEngine system, see fig. 2. A four side back projection system called the CUBE is installed at RUS. The CUBE is connected to a Silicon Graphics Onyx2 double rack system with 14 R10000 CPUs and 4GB of main memory. The Onyx2 InfiniteReality pipes each equipped with two raster managers.

Several magnetic tracking systems such as Polhemus Fastrack with Stylus pen and the Ascension Motionstar with 3D mouse are supported for the interaction of the user with the virtual environment. Fig. 2: VR equipment at IHS Design System

It is a parametric design system based on Euler's equation for turbomachinery. Using this equation the flow angles upstream as well as downstream of the runner can be calculated that are needed to produce the runner torque necessary for the required turbine power output. These flow angles are transformed to blade angles taking into account knowledge based assumptions for incidence and deviation angles. In between, from leading to trailing edge of the blade, mean lines are created and profile coordinates are added taken from catalogue. Then, curvature and thickness distributions are optimized using CFD in order to define the final blade shape.

Simulation software The flow simulation is carried out using FENFLOSS, a finite element flow simulation program, developed at IHS. It is based on the Reynolds averaged Navier-Stokes equations -2-

ISIMADE, Baden-Baden, 1999 with various models of turbulence. Usually the k-ε model is used. FENFLOSS runs on various platforms ranging from PC to vector-supercomputers and massively parallel machines. The parallelization is obtained by a domain decomposition algorithm with overlapping meshes. For the online simulation FENFLOSS has been integrated into COVISE as a module.

Application The application shown in this paper is an axial propeller turbine. The geometry of the adjustable guide vanes and of the fixed runner blades is shown in fig. 3. The turbine is composed for the following main data:

• Head: 7.9 m • Discharge: 5.6 m3/s • Speed: 250 rpm • No of runner blades: 6 By using the design system the influence of the different parameters can be easily demonstrated. So the students as well as the industrial customers can get an impression of the blade shape very fast. As an example in fig. 4 the runner blades are shown Fig. 3: Geometry of an axial propeller turbine for the data described above as well as for a reduced discharge rate of 4 m3/s keeping all other parameters constant. The influence of the discharge can be seen clearly. For the high flow rate the runner blades are steeper. By reducing the flow rate the inlet and outlet angles of the blades have to be more flat and consequently the blade channel is becomes narrower.

a) high discharge rate b) lower discharge rate Fig. 4: Runner blades design for two different discharge rates

As already mentioned after the preliminary investigation of the blade shapes the flow in the runner is calculated numerically. For this calculation a computational grid is needed. Usually a periodical flow condition is assumed. Consequently only one channel of the runner has to -3-

ISIMADE, Baden-Baden, 1999 be considered. The grid is obtained by a specialized grid generator. A typical grid around a runner blade is shown in fig. 5. A typical grid consists of approximately 100000-200000 nodes. The flow is calculated in a frame of reference rotating with the runner. In this frame the flow is assumed to be steady state.

During the simulation the results are sent from time to time to COVISE and displayed in the VR environment. The current iteration step as well as the final results can than be analyzed interactively using VR techniques. For example particle traces or streamlines can be started, cutting planes can be located according to the Fig. 5: Computational grid around the runner blade users requirements etc. This allows a fast understanding of the complete three-dimensional flow structure. In fig. 6 the distribution of streamlines at the hub for the shown geometry is presented. It can be seen, that the flow angle corresponds quite well to the blade angle. This is usually the most critical part of the blade. Conclusion

A tool for the design of hydro turbine runner is under development for educational and industrial purposes, respectively. The tool is based on Virtual Reality technique. The application of the tool during the lecture courses allows the student to investigate quickly the influence of the various parameters (head, discharge, speed etc.) and consequently to get a much better understanding of the correlation between the parameters and blade shape.

Contact: Dr.-Ing. A. Ruprecht Institute for Fluid Mechanics and Hydraulic Machinery Pfaffenwaldring 10 70550 Stuttgart, Germany Tel.: +49-711-685-3259 Fax: +49-711-685-3255 Email: [email protected]

Fig. 6: Calculated streamlines