Research Papers

An Investigation of Nonlinear Flow Oscillations in a High-Pressure Centrifugal Pump

[+] Author and Article Information
Claudio Lettieri

MIT Gas Turbine Laboratory,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: lettieri@mit.edu

Jeff Defoe

MIT Gas Turbine Laboratory,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: jdefoe@uwindsor.ca

Zoltán S. Spakovszky

MIT Gas Turbine Laboratory,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: zolti@mit.edu

1Corresponding author.

2Current position: Assistant Professor of Mechanical Engineering, University of Windsor, Windsor, ON N9B 3P4, Canada.

Contributed by the International Gas Turbine Institute (IGTI) of ASME for publication in the JOURNAL OF TURBOMACHINERY. Manuscript received November 12, 2014; final manuscript received August 4, 2015; published online September 2, 2015. Assoc. Editor: Stephen W. T. Spence.

J. Turbomach 137(11), 111004 (Sep 02, 2015) (9 pages) Paper No: TURBO-14-1294; doi: 10.1115/1.4031250 History: Received November 12, 2014; Revised August 04, 2015

High-pressure multistage pumps and their coupled piping systems, typically used in the process and power generation industry, can experience dangerous system-level instabilities. This can occur at flow coefficients well away from the surge limit and in the absence of cavitation. Such a pumping system and a related new kind of instability are the focus of this paper. A system-wide instability was observed at 0.05 times rotor frequency for flow coefficients near maximum head rise but at negative slope, thus on the stable side of the head rise characteristic. A previous study based on system-level experiments concluded that this instability differs from classical surge, cavitation surge, rotating stall, and rotating cavitation, but the underlying mechanism and necessary flow conditions remain unknown. This paper investigates the root cause of the system-wide pump instability, employing a systematic analysis of the impact of geometry changes on pump stability and performance. It is found that the upstream influence of the unsteady flow separation in the return channel leads to a time-varying incidence angle change on the volute tongue which causes periodic ingestion of low-stagnation pressure fluid into the diffuser passages. This sets up a limit cycle, promoting the system-wide instability. With the instability mechanism determined, the pump is redesigned to remove the flow separation while maintaining performance at design conditions. Unsteady numerical simulations demonstrate improved efficiency and pressure recovery at low flow coefficients. A time accurate calculation also indicates stable operation at all relevant flow conditions. The paper resolves a long-standing pump stability problem and provides design guidelines for reliable and improved performance, important to the chemical processing and power generation industry.

Copyright © 2015 by ASME
Your Session has timed out. Please sign back in to continue.


Kawata, Y. , Ebara, K. , Uehara, S. , and Takata, T. , 1987, “ System Instability Caused by the Dynamic Behavior of a Centrifugal Pump at Partial Operation,” JSME Int. J., 30(260), pp. 271–278. [CrossRef]
Brennen, C. , and Acosta, A. , 1976, “ The Dynamic Transfer Function for a Cavitating Inducer,” ASME J. Fluids Eng., 98(2), pp. 182–191. [CrossRef]
Greitzer, E. , 1981, “ The Stability of Pumping Systems: The 1980 Freeman Scholar Lecture,” ASME J. Fluids Eng., 103(2), pp. 193–248. [CrossRef]
Tsujimoto, Y. , Kamijo, K. , and Brennen, C. , 2001, “ Unified Treatment of Flow Instabilities of Turbomachines,” AIAA J. Propul. Power, 17(3), pp. 636–643. [CrossRef]
Brennen, C ., 2007, “ Multifrequency Instability of Cavitating Inducers,” ASME J. Fluids Eng., 129(6), pp. 732–736. [CrossRef]
Makay, E. , 1978, “ Survey of Feed Pump Outages, Research Project 641,” EPRI, Report No. FP-754 4-2.
Gulich, J. F. , 2010, Centrifugal Pumps, Springer, London.
Casey, M. , Dalbert, P. , and Schurter, E. , 1990, “ Radial Compressor Stages for Low Flow Coefficients,” 4th European Congress on Fluid Machinery for Oil Petrochemical and Related Industries, The Hague, May 21–23, Vol. C403/004, pp. 117–126.
Lettieri, C. , Baltadjiev, N. , Casey, M. , and Spakovszky, Z. , 2014, “ Low-Flow-Coefficient Centrifugal Compressor Design for Supercritical CO2,” ASME J. Turbomach., 136(8), p. 081009. [CrossRef]
ANSYS, 2012, “Academic Research Release 14.5, Theory Manual,” ANSYS, Irvine CA.
Defoe, J. , Narkaj, A. , and Spakovszky, Z. , 2009, “ A Novel MPT Noise Methodology for Highly-Integrated Propulsion Systems With Inlet Flow Distortion,” AIAA Paper No. 2009-3366.
Rumsey, C. , Biedron, R. , and Farassat, F. , 1998, “ Ducted-Fan Engine Acoustic Predictions Using a Navier–Stokes Code,” J. Sound Vib., 213(4), pp. 643–664. [CrossRef]
Brennen, C. , 1994, Hydrodynamics of Pumps, Oxford University Press, Oxford, UK.
Wiesner, F. J. , 1967, “ Review of Slip Factors for Centrifugal Impellers,” ASME J. Eng. Gas Turbines Power, 89(4), pp. 558–572. [CrossRef]


Grahic Jump Location
Fig. 1

Measured head rise characteristic of two-stage pump

Grahic Jump Location
Fig. 2

Grid sensitivity analysis—velocity profiles at diffuser inlet for four meshes with increasing refinement levels

Grahic Jump Location
Fig. 3

Datum stage geometry

Grahic Jump Location
Fig. 4

Geometries used for datum stage calculations: (a) full stage, (b) single-passage, static components only, (c) single-passage, static components with return channel removed, and (d) single-passage, static components with vaneless space and volute tongue removed

Grahic Jump Location
Fig. 5

Time-averaged flow field at midspan in static component calculation. Left: nondimensional velocity field and right: vorticity field. In each image, the portion to the left of the discontinuity is at midspan in the volute; the portion at right is at midspan in the deswirler.

Grahic Jump Location
Fig. 6

Low-stagnation pressure fluid (red arrow) shed from volute tongue due to incidence swings enters adjacent diffusing passage

Grahic Jump Location
Fig. 7

Tongue boundary layer pathline, showing long residence time in the vaneless space. Pathline shown both upstream and downstream of the point indicated by the cross. Flow direction is counterclockwise as shown.

Grahic Jump Location
Fig. 8

Low-stagnation pressure fluid (red arrow) interacts with boundary layer in the diffusing passage, leading to vortex shedding (purple arrow). Shed vortices interact with the return vane, leading to bluff-body separation with upstream influence.

Grahic Jump Location
Fig. 9

Time trace of loss coefficient at stage exit, ω = (pt1−p¯t2M)/0.5ρUT2

Grahic Jump Location
Fig. 10

(a) Single passage of static components, (b) static components with return channel removed, and (c) static components with vaneless space and volute tongue removed

Grahic Jump Location
Fig. 11

Midspan velocity field: (a) all static components present, (b) return channel removed, and (c) vaneless space and volute tongue removed. Vortex shedding in the diffusing passage is inhibited when the return channel is removed. Vortex shedding is eliminated when the vaneless space and tongue are removed.

Grahic Jump Location
Fig. 12

Conventional flow path design with volute (left) versus new design with constant-width flow path and vaned diffuser (right)

Grahic Jump Location
Fig. 13

To maintain fixed work input coefficient (and thus cuT), the redesign with an increased, constant-width flow path leads to increased trailing edge blade angle and increase absolute flow angle at impeller exit

Grahic Jump Location
Fig. 14

Effect of b2* on mass-averaged diffuser inlet flow angle and total-to-total head rise coefficient

Grahic Jump Location
Fig. 15

Computed stage total-to-total head rise coefficient without leakage flow for original design and stage redesign. Experimental data (with leakage flows) are also shown.

Grahic Jump Location
Fig. 16

Polytropic efficiency versus flow coefficient

Grahic Jump Location
Fig. 17

Contour of normalized velocity at 50% span at the design flow coefficient of pump redesign. Velocities are relative in the rotating frame of the impeller and absolute for the stationary components. (a) Impeller and diffuser and (b) return channel.

Grahic Jump Location
Fig. 18

Time trace of loss coefficient at stage exit, ω = pt1−p¯t2M/0.5ρUT2: black—redesign stage and gray—datum stage



Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In