Automated runner blade design optimization process

based on CFD verification

By Jacek Swiderski, P. Eng. Swiderski Engineering, Ottawa, Canada
Joseph N. Martin, Norcan Hydraulic Turbine Inc., Carleton Place, Canada
Richard Norrena, Norcan Hydraulic Turbine Inc., Carleton Place, Canada

ABSTRACT

Norcan Hydraulic Turbine Inc. has introduced Computational Fluid Dynamics (CFD), as
a foundation for the development of new hydraulic technologies. Recently, research
and development (R&D) efforts have been focused on the enhancement of turbine
runner design methods, resulting in the creation of runner blade shape optimization
algorithms, in an attempt to maximize the hydraulic efficiency of the entire turbine. The
optimization procedure links a series of geometry manipulation programs with a CFD
module, which provides verifications of each iteration step. The CFD module has
become a central object of this algorithm. This paper presents a general outline of a
blade shape optimization algorithm, theoretical basis for experimentation and selected
examples of applications.

Preface

This article presents the theory and methodology behind an automated runner blade
design process, focusing primarily on the initial stages of development conducted by the
authors. The goal of this project is to produce runner blade designs, which when
analysed will create desired flow patterns. This process is also referred as “Results
Orientated Solution” (ROS). Utilizing previous experience in water turbine design,
through a process, which relies greatly on the use of Computational Fluid Dynamics
(CFD) results, the various steps and procedures within this complicated process were
itemized, allowing the entire process to be mapped out, and in the end combined
through numerous algorithms.
Historically, the prevailing approaches used in blade design (i.e. [1], [2]), appeared to be
focused primarily on the verification of existing blade design methods, which do
inherently appreciate the importance of CFD techniques and results.

The "Results Orientated Solution "

The primary focus in hydraulic turbine design has been water flow passage design.
The geometry of the water flow passage has the greatest impact on the efficiency and
the power produced by the runner.
The numerous published design algorithms (i.e.: [1], [2]) establish a proof of the
importance of this task. However the solvability of the available algorithms is highly
dependant on certain assumptions, which greatly simplify the values associated with the
natural physical phenomenon involved with this process. In previous methods, these
assumptions were integral to the solving of the above-mentioned algorithms known to
the authors.

Assuming that the CFD models a real flow in a highly satisfactory way [5], [6], [7], [8],
the CFD results should be accepted as a final verification of the performances of
hydraulic flow passages. Based on the foregoing assumptions, the authors have
attempted to extrapolate and perfect the design method, based on the back-loop
principle of CFD results and shape correction, which by making alterations based on
pressure and velocity distribution, gives the desirable change of the flow.

The First Approach

The main objective of this presentation is to demonstrate the ability of solving the
following problem: to find the shape of the flow passage geometry (de facto: the "blade
shape") which enables the user to achieve the desirable parameters of the flow pattern
at the runner exit, with minimal levels of energy losses in the blade-to-blade space.
For clarity of this presentation, we will demonstrate the method, which allows the user to
automatically achieve the following objectives:
a) the desired vortex at the draft tube inlet (runner exit).
b) no-shock entry to the runner
Initial guess of Geometry data:
the blade shape a) casing
b) stay vanes and wicket gates
c) skirt and hub outlines
d) blade: leading edge and trailing
revolving outline
e) draft tube

Geometry data:
a) casing
b) stay vanes and wicket gates
c) skirt and hub outlines
d) blade: leading edge and trailing
revolving outline
Draft tube inflow: e) draft tube
desired vortex Grid

Rated design Blade modification

parameters:
Q11
n11
Flow Analysis Flow data
CFD module at the runner inflow

Evaluation
End
Efficiency

Figure 1.The general scheme of the method based on the back-loop principle -
CFD-shape
A Blade Kaplan’s Turbine Designing In a Classical Configuration

Four bladed Kaplan runner should be designed for the following rated operating
parameters:

Q11 = 2.1
n11 = 125

The number of the runner blades Z=4

The following are the input geometry data sets required for analysis
a) The hydraulic outline of the runner (Figure 2)
b) Geometry of the distributor, including the wicket gate angle (αWG)
c) Draft tube shape

The desired draft tube inlet vortex was assumed as per Figure 3.

Leading edge outline

Trailing edge outline

Draft tube inlet

Figure 2. Hydraulic outline of the runner

 Draft tube inlet
- velocity angle distribution -
6

5
ALFA
4
Velocity angle
(ALFA) [deg]

3 C

2 Cm

-1
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
Radius [m]

Figure 3. Distribution of the velocity angles at the draft tube inlet (draft tube inlet vortex)

Wicked gate passage

Inlet

Gate - blade boundary

Runner blade

Draft tube inlet

Figure 4. Grid used to test the blade design optimization algorithm

The calculations involved in this iteration process, were conducted for the single blade-
to-blade passage modeled by the 47x29x25 grid. The presented here results of the
iteration steps are dealing with a very simple Initial Guess geometry: flat plate blade.
 Flat plane blade
η = 67%
σcr = 2.5

Modification 1
η = 78%
σcr = 1.75

Modification 3
η = 82%
σcr = 1.62
 Modification 8
η = 86%
σcr = 1.1

Modification 14
η = 91%
σcr = 0.9

Figure 5. Results of the flow analysis for selected iteration steps of the design
optimization
Observations and Conclusions

In general, the proposed iteration process (Figure 1.) demonstrates good convergence
for the relaxation coefficient ζ =0.75.

At the nominal operating parameters, the blades created by a presented algorithm

achieve:

a) expected distribution of static pressure on both (pressure and suction) blade

surfaces,
b) expected draft tube inlet vortex (as per Figure 3.)

A presented method, as was shown, is based on relatively large amount of primary

assumptions - mostly dealing with the geometry.

The Authors, when analyzed their own design methods (CFD based), were trying to
determine sub-processes, which can be solved without a control of the designer - in an
automatic way by a computer program. One of them is the presented process of a blade
shape optimization, which in Authors’ opinion is a valuable component of optimization
the whole process of the water turbine design.

The problems that are still left to be automat zed still remain and need to be
algorymized:

a) the outline of the leading edge ( in a described method they are assumed )
(Figure 1)
b) the shape of the blade section profiles or in other words: the blade thickness
distribution

Where turbine geometry data are known (and for practical purposes should not be
considered for modification), the design methodology is very consistent with the
presented method.
Recently completed work on the design of Francis runner blades, using the method
described above as a component in the system of projecting turbine performances,
show very promising results.
References::

[1] Krzemianowski Z.: Two-dimensional and three-dimensional models of the pump-

turbine, Hydroforum 2000 (in Polish)
[2] Rogalsky T., S. Kocabiyik, R.W. Derksen: Differential Evolution in Aerodynamic
Optimization, Canadian Aeronautics and Space Journal, vol. 46, No. 4, December,
2000.
[3] Krzyzanowski W.: Water Turbines – Design and Regulation, WNT, Warsaw, 1977.
(in Polish)
[4] Nechleba M.: Hydraulic Turbines, Their Design and Equipment, ARTIA, Prague,
1957.
[5] CFX - TASCflow Computational Fluid Dynamics Software, Version 2.10 – Manuals,
AEA Technology, Engineering Software, 2000.
[6] Swiderski J. , J. Martin : Practical implementations of Computational Fluid Dynamics
in a design practice - virtual hydraulic laboratory, Hydroforum 2000, Poland
[7] Swiderski J, J. Martin : High Power Francis runner – upgrade with a new design
runner, Norcan Hydraulic Turbine Inc. report, 1999
[8] Bennett K., J. Swiderski : Application of CFD Turbine Design for Small Hydro Elliott
Falls, A Case Study, Waterpower XII, Salt lake City, 2001