ITCMG master

1

Technical Investigation Department
2
METHOD FOR 3-D MODELLINGOF A MIXED FLOW
PUMPUSING PHOENICSD Radosavljevic
3
Introduction

• Background information on the investigation
• CFD and PHOENICS role
• Modelling with PHOENICS
• Results and analysis
• Conclusions (simulation, project)

4
The Situation

• A major water supply project located in North
  Africa (484 pumps).
• Pumps reported to have been unused when first put
  into service on this project.
• The type of pump is defined as a wellpump of the
  vertical submersible turbine type (7 stages).
• Pumps were specified to meet a range of duties
  for 25 years in relation to the envisaged
  drawdown schedule.

5
The Situation
6
The Problem

• During the course of approximately 3 years of
  operation, pump performance problems were
  encountered in a number of wells.
• Upon withdrawal from the well, a pump was
  observed to exhibit severe cracking and
  corrosion, in particular in the region of the
  upper pump bowl.
• Cracking was also observed in the corresponding
  corroded impeller.

7
The Problem
8
The Problem
9
Approach

• Identifying the nature of the processes involved.
  The primary ones may be categorised as
• - physical (clogging and abrasion)
• - chemical (clogging and electro-chemical
  corrosion)
• - microbial (clogging and microbially-induced
  corrosion)

10
Approach

• Identifying the nature of the processes involved.
  Important subsidiary factors
• - operational (steady loads (static water head),
  unsteady loads (water hammer), intermittent
  pumping and over-abstraction
• - structural and mechanical (design/construction
  and materials).

11
Approach

• A number of separate studies defined
  including objectives to
• determine quasi-steady hydrodynamically-induced
  loadings, using CFD analysis (PHOENICS)
• determine other loadings from specification, such
  as self-weight, torque and centrifugal
• apply all loadings to finite element analysis
  model and determine individual and combined
  stresses

12
Geometry

• Not supplied (proprietary vane design)
• Perform sectioning of impeller and the bowl in
  order to take measurements.

13
Modelling in PHOENICS

• Model one full stage of the pump as a single
  device
• Apply sliding grid with Multiblock. Rotating
  block - impeller and stationary block - bowl
• Advantage
• Capture of full transient effects and (true)
  dynamic loading

14
Modelling in PHOENICS

• Problems (constraints of sliding MB)
• no surface porosities allowed (vanes?)
• only uniform grid in circumferential direction
  allowed
• only clock-wise rotation is allowed (pump rotates
  anti-clockwise).

15
Modelling in PHOENICS

• Despite all the Problems !

16
Modelling in PHOENICS

• Compromise approach
• Treat impeller and the bowl as separate
  components
• Steady simulation of the impeller
• Transient simulation of the diffuser with the
  correct input flow field (impeller exit). (More
  accurate rotor-stator interaction)

17
Modelling in PHOENICS

• Impeller modelling
• Steady
• BFC grid
• Single passage (1/6 of the flow volume)
• Cyclic boundary at the exit (vaneless space)
• 2900 rpm (ROTA patch for rotational forces)
• Wall friction, k-e model
• Outlet flow field data saved in a file.

18
Modelling in PHOENICS

• Impeller modelling - velocity field

19
Modelling in PHOENICS

• Stator modelling
• Transient
• Inlet flow field cycles through impeller exit
  data
• BFC grid
• Single passage (1/7 of the flow volume)
• Cyclic boundary at the exit and inlet (vaneless
  space)
• Wall friction, k-e model.

20
Modelling in PHOENICS

• Stator modelling - Ground

21
Modelling in PHOENICS

• Stator modelling - Numerics
• Convergence generally within 500sw(/tstep)
• Stator - start from steady solution in aligned
  position
• 10 hours CPU for the transient run

22
Modelling in PHOENICS

• Stator modelling - velocity field

23
Modelling in PHOENICS

• Stator modelling - Assumptions
• Impeller flow calculated in isolation - no
  interaction with the stator
• Cyclic condition ahead of stator inlet
• Assessment of accuracy
• Pressure increase within impeller 3.6 bar
• Pressure increase within stator 2-3 bar
• 5.71 bar per stage
• Torque 98.7 kW vs. 103kW (GXDRAG).

24
Modelling in PHOENICSTransient pressure field in
vaneless space

• Effect of the impeller blade passing
• NOTE
• Contour scaling at plane values

25
CONCLUSIONS

• PHOENICS
• GROUND proved extremely valuable
• Allowed extensive modification to the calculation
  procedure
• Pump
• Obtained estimates of the hydrodynamic loading
  within the pump
• Results do not identify any pronounced local
  peaks in pressure

26
CONCLUSIONS