Comprehensive analysis of the saddle curve of pump turbines in pump mode using PIV and CFD

Christian Widmer, Nikolaus Spyrou, Helmut Jaberg, Florian Senn, Mark Guggenberger

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review


Pump turbines are mainly designed in pump mode since its retarding flow is more
complex and more sensitive to flow detachment. In order to reduce the size of the machine and
to gain a better turbine performance, the layout of the pump targets at pressure coefficients as
high as possible. At high pressure coefficients it comes into conflict with the unstable branch
of the pump characteristic curve. Due to the sudden drop of the pump characteristic, the
unstable branch is called saddle curve. The operation in the saddle curve must be avoided at all
since there is no possibility to leave the dropped curve. Even a safety margin must be
maintained. Herewith the layout of the pump to highest possible pressure coefficients is
delimited. Pump turbine designers are faced with the difficulty that the proper prediction of the
pump saddle curve with standard simulation tools is very demanding. Therefor it is essential to
better understand the detailed flow features during this unstable regime. ANDRITZ HYDRO
and the Graz University of Technology investigate in a cooperative research project the key
mechanisms leading to the saddle curve in pump mode of pump turbines by means of 2D laser
particle image velocimetry (PIV) measurement and advanced CFD simulations. A pump
turbine model has been modified in order to gain visual access to the flow on the low and the
high pressure side of the runner while keeping the same hydraulic behavior. The PIV laser
measurements have been performed at the test rig premises of the Graz University of
Technology in the draft tube cone on the one hand to deeply investigate the pre-rotation of the
flow during the unstable operation. On the other hand, instability phenomena on the high
pressure side of the model such as rotating stall in the guide vanes could be visualized with
PIV. Together with HD video recordings, detailed flow features could be detected and
allocated to the unstable flow field. Additionally, highly dynamic pressure sensors have been
mounted in a large area of the model in order to detect the transient pressure pulsations during
the unstable regime. CFD simulations accompanied the test rig measurements. The results of
the simulations have been compared and calibrated with the measurements in order to even
better understand the flow pattern during the instability in the draft tube cone, the guide and
stay vanes but also in the runner where no visual access for the laser beam was possible. The
flow features of the simulations and the measurements are compared and similarities as well as
differences are outlined.
Original languageEnglish
Title of host publicationIAHR Conference on Hydraulic machines and systems, Grenoble, 4-8 July 2016
Subtitle of host publicationPublication in IOP Conference Series
Publication statusPublished - 4 Jul 2016

Fields of Expertise

  • Sustainable Systems

Cite this