New CFD-based method for erosion prediction in

control valves
Prof. Stefano Malavasi, Dr. Gianandrea V. Messa
Politecnico di Milano

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 1
 The impact erosion issue in control valves

What is the impact erosion?

The impact erosion is the loss of material from a surface
subjected to the impingements of solid particles (even very
small) dragged by a fluid.

Why it is a relevant problem in control valves?

Eroson may cause:
 Change of the valve’s performace
 Valve’s life reduction
 Management issues
 Service downtime

Main engineering concerns:

• Identification of the erosion hotspots locations
• Erosion rate estimation

Gharaibah et al., OTC 2013, Paper No 24271

Proper valve service management
27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 2
 Presentation outline

 Available practical approaches to erosion

 How is erosion estimated via CFD ?
 Our method
Direct Impact Test
 First benchmark cases
 Direct Impact Test
 Needle & Seat choke valve
 Conclusion and future developments

Choke Valve

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 3
 Available practical approaches to erosion

Experimental Approach  Direct evaluation of the material losses:

 Expensive
 Limited testable device sizes
 Not generalizable results

Numerical Approach  Evaluation of particle parameters via CFD

(two steps)
 Empirical erosion model

 Applicable only to dilute flows (concentration <0.1%)

 No quantitative prediction
 No geometry changes

Why these limitations ?

Actually, the excessive computational cost …
 Available practical approaches to erosion

Just an example…
Erosion in a choke valve (Gharaibah et al., 2013)

Exp.
≈ 4-20!!!
CFD

The main deviations (R=20-23) are imputable to the geometry changes

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 5
 How is the erosion estimated via CFD ?

Dilute flows only (C<0.1%) → «one way coupling» assumption (particles don't affect the fluid flow)

Fluid simulation Particle Tracking Erosion model

RANS equations Newton’s 2nd law Algebraic empirical

formula

Loss
dv
U, P, k ,  ,... mp  F of material
dt

Numerical solution of the

Reynolds-Averaged Navier-Stokes
Equations (RANS)
Numerical determination of
all particles' trajectories

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 6
 How is the erosion estimated via CFD ?
eroding
particle
vp
Empirical estimation of material loss
θp

Main input parameters (by the particle tracking): eroded surface

• Particle impact velocity vp
• Particle impact angle θp
• Particle mass mp
Lower wear
• Material properties: density, hardness, …

90°
Normalized erosion rate

Maximum wear
Effect of the impact angle

≈36°

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies
Particle impact angle 7
 Our method

Goal the reduction of computational burden

Idea do heavy calculations only where needed
How with same numerical tricks
(EE simulation + active volume + interface laws)
FULL Eulerian-Lagrangian
Useless (Standard Method)
particle
Particle
data Eulerian-Eulerian simulation
tracking on
the whole NEW METHOD
flow field mean solid
flow on the
whole flow
field

Particle
active volume tracking
Useful particle data only on
active
volume

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 8
 Our method
EE simulation (mean solid flow)

 
    f 1   EE  U EE      f  'EE u 'EE  0

 
    EE  p VEE      p  'EE v 'EE  0

    f 1   EE  U EE U EE     1   EE      EE
t
 U EE  
Interface laws
 
   f  'EE u 'EE U EE   1   EE  PEE  M EE

    p  EE VEE VEE       EE  EE
t
VEE  
1/2
 2k 
v i  VEE  ξ  EE   
   p  'EE v 'EE VEE   EE PEE  M EE
 3 
  EE 
   n
t
AP
mP   
 p EE EE
V   EE  n
VEE   
n
VEE

Particle Tracking active volume definition

dv  by experience
mp  F  by testing particles
dt

Erosion hotspot

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 9
 Specimen
First benchmark case:
direct impact test 12.7 mm
76.2 mm
Nozzle
Preliminary tests concerned the normal
r
impingement of an abrasive submerged jet
z
against a specimen of erodible material.
8 mm
• sand particles with dp=120 μm
• volume fraction = 0.1%
Abrasive jet
• 2D, axisymmetric domain

Compared to the full Eulerian-Lagrangian

approach:
 -60% disk space
 -55% CPU time for particle tracking

r
Specimen

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 10
 Second benchmark case:
Needle & Seat choke valve
Benchmark description

Valve and Flow features:

• valve size = 2 in.
• water flow rate = 14.2 l/s (7 m/s in the inlet pipe)
• sand particles with dp=400 μm
• particle concentration = 0.1%

Inlet
Mesh detail

Simulation details: D
• PHOENICS 2011 CFD code employed
• Turbulent, incompressible RANS model
• Standard k-ε turbulence model 9D
• IPSA EE model of Spalding (1980) Wall (needle)
• GENTRA particle tracker 2010 version
Wall (interior)
• Structured mesh of about 4.7M cells
Outlet

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 11
 Second benchmark case:
Needle & Seat choke valve
Active volume definition…

…by releasing 1000

Wear
testing particles in the ≈36°

flow domain.
Main hotspot θp
(impacts with θp ≈ 36°)

Inlet
Reducer wall

Outlet

Wear
Active volume Needle wall 90°
θp

Secondary hotspot
27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies
Active volume (impacts with θp ≈ 90°)
 Second benchmark case:
Needle & Seat choke valve

Erosion estimation

Active
volume

Integral profile of loss

material within active
volume
θ

Valve
geometry
profile

Active volume

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 13
 Second benchmark case:
Needle & Seat choke valve
Comparison with full Eulerian-Lagrangian approach

Our method
Full EL approach

Valve
geometry
profile

Active volume

 Comparable results
 -77% disk space
 -63% CPU time for particle tracking

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 14
 Conclusion and future developments

 We presented a new CFD-based methodology for erosion prediction in control valves.

 Our method allow results comparable with standard methods but with a significant
reduction in computational burden on dilute flows.
Solid concentration =0.1%
Disk space reduction > 70%
CPU time reduction > 60% (tracking particles)

 This result allows to overcome the actual computational limits and to use the numerical
approach in practical applications

 Increase the solid concentration (computational advantage increase with concentration)

 Quantitative erosion can be provided
 Geometry changes can be considered

 A next important step is the validation of erosion models which became reliable with
our model.
To do this we are designing a slurry test plant for control valve in our hydraulic lab

27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies
 Thank you for the kindly attention

…. any questions?

Prof. Stefano Malavasi, PhD

Politecnico di Milano - D.I.I.A.R.-sez. Ingegneria Idraulica
Piazza Leonardo da Vinci, 32 20133 Milano - Italy
tel.: +39 02 2399 6261 mob.: +39 335 7982622
e-mail: [email protected]
http://intranet.dica.polimi.it/pagine-docenti/malavasi-stefano/
27/05/2015 IVS 2015 • International Conference on Valve and Flow Control Technologies 16