© 2011 ANSYS, Inc. June 21, 2012 1 Erosion Modeling and Sand Management with ANSYS CFD Madhusuden Agrawal ANSYS Houston
© 2011 ANSYS, Inc. June 21, 2012 1
Erosion Modeling and Sand Management with ANSYS CFD
Madhusuden Agrawal
ANSYS Houston
© 2011 ANSYS, Inc. June 21, 2012 2
Particulate modeling in ANSYS CFD
Sand Control and Sand Management
• Sand Filtration
• Sand Transport in pipelines
• Proppant Placement
Erosion Modeling
• Challenges in Erosion Modeling
• Key components of erosion modeling
• ANSYS solution for erosion modeling
• Erosion Module
• Examples
OUTLINE
© 2011 ANSYS, Inc. June 21, 2012 3
Recap: Particulate Modeling
© 2011 ANSYS, Inc. June 21, 2012 4
Spans wide range of
• Length scales
• Time scales
Physics
• Particulate physics
• Fluid particle interaction
• Particle size distribution
• Homogenous and heterogeneous reaction
• Particle structure interaction
Challenges in Particulate Modeling
From: Fundamental of Multiphase Flow, C. E. Brennen
© 2011 ANSYS, Inc. June 21, 2012 5
Particulate Flows Regimes Diluted vs. Dense Flow
es
t12/tcol
10-3 10-1 10-5 10-7
104
102
100
10-2
dilute dense
101 102 100 (x1-x2)/dp
4-way coupling
2-way coupling
1-way coupling
Particles reduce
turbulence
Particles enhance
turbulence negligible effect on
turbulence
102
100
10-2
t12/teddy
Dilute Dense Relative motion between particles Large Small
Particle-particle interaction Weak Strong
Apparent viscosity of the solid phase
Particle-fluid
interactions
Particle-particle
interaction
© 2011 ANSYS, Inc. June 21, 2012 6
Modeling Particulate Flow
Particle Phase
Particle size
P-P Interaction
Fluid-P Interaction
Eulerian
Lagrangian
Sub grid scale
Super grid scale
Hybrid
Resolved
Modeled
Resolved
Modeled
© 2011 ANSYS, Inc. June 21, 2012 7
Platform for Simulating Particulate Systems
ANSYS CFD provides a platform which can adapt to the multi-physics, multi-components and multi-scale configurations of particulate flows and their industrial applications
Eulerian Granular
MPM DDPM-DEM DPM
© 2011 ANSYS, Inc. June 21, 2012 9
Models for Particulate Flows Model Numerical
approach Particle fluid interaction
Particle-Particle interaction
Particle size distribution
DPM Fluid – Eulerian Particles – Lagrangian
Empirical models for sub-grid particles
Particles are treated as points
Easy to include PSD because of Lagrangian description
DDPM - KTGF Fluid – Eulerian Particles – Lagrangian
Empirical; sub-grid particles
Approximate P-P interactions determined by granular models
Easy to include PSD because of Lagrangian description
DDPM - DEM Fluid – Eulerian Particles – Lagrangian
Empirical; sub-grid particles
Accurate determination of P-P interactions.
Can account for all PSD physics accurately including geometric effects
Euler Granular model
Fluid – Eulerian Particles – Eulerian
Empirical; sub-grid particles
P-P interactions modeled by fluid properties, such as granular pressure, viscosity, drag etc.
Different phases to account for a PSD; when size change operations happen use population balance models
Macroscopic Particle Model
Fluid – Eulerian Particles – Lagrangian
Interactions determined as part of solution; particles span many fluid cells
Accurate determination of P-P interactions.
Easy to include PSD; if particles become smaller than the mesh, uses an empiricial model
© 2011 ANSYS, Inc. June 21, 2012 10
Sand Control
© 2011 ANSYS, Inc. June 21, 2012 11
• Sand is often produced in both onshore and offshore production systems • Sand production may be continuous, or sudden
• The sediment consists mud, sand and scale picked up during the transport of the oil
• Sand Management is important in oil production to ensure system integrity and efficiency
• Excessive sand leads to • Partial or complete blockage of flowlines
• Enhanced pipe bottom corrosion and erosion
• Trapping of pigs
• Reduced production time and increased
maintenance and operating costs
Sedimentation in Oil & Gas
Internal flow of natural gas containing sand particles. particle trajectories are colored in grey. The erosive wear hotspots on the piping is colored out in red.
© 2011 ANSYS, Inc. June 21, 2012 12
Sand control strategies
• Preventing formation failure
• Sand exclusion techniques
• Sand management
Sand Control
Key areas to understand fundamental nature of sand in the reservoir and the wellbore
Hydraulic fracturing (Proppant transport)
© 2011 ANSYS, Inc. June 21, 2012 13
Sand Exclusion Techniques
© 2011 ANSYS, Inc. June 21, 2012 14
Sand control screen systems
• Screens
• Gravel and frac packing
Example: Sand Filtering Systems in O&G
Bulk process Surface process
© 2011 ANSYS, Inc. June 21, 2012 15
Euler Granular Model
• Porous media model with physical velocity formulation
• Low permeability for the particulate phase
• May not be able to simulate particle size dependent filtering
Particulate Models
• DDPM model with DEM closure for particle-particle interaction
• Particles can be stopped by reflect or trap boundary conditions
• Can model particle size effects.
• Macro Particle Model will physically filter particles through pores
Modeling Filtration with ANSYS
© 2011 ANSYS, Inc. June 21, 2012 16
Euler Granular Model for Filtration
t = 16 sec.
t = 100 sec.
t = 135 sec.
t = 60 sec.
Solid Phase Volume Faction Contours Velocity Vectors of Solid Phase
© 2011 ANSYS, Inc. June 21, 2012 17
Filter Cake Formation in Vertical Wells… Journal of Petroleum and Gas Engineering Vol. 2(7), pp. 146-164, November 2011 Mohd. A. Kabir and Isaac K. Gamwo
© 2011 ANSYS, Inc. June 21, 2012 18
Filtration Modeling Using DDPM/DEM
Filter: Allows particles below a threshold to pass through, Filter represented by a internal boundary condition.
Inlet Outlet
© 2011 ANSYS, Inc. June 21, 2012 19
Particle separation through a filter element at three instances in time. The flow is from left to right. The small particles flow through the holes in the perforated plate and exit the pipe on the right. The plate blocks the bigger particles.
Filtration Modeling using MPM
© 2011 ANSYS, Inc. June 21, 2012 20
Particulate Migration in Gravel Pack
• Micro scale Simulation for fine particles transport through pores in gravel pack
• Study Permeability alterations in the gravel pack due to fine particles entrainments, transport and deposition
• Filtration of fine particles
© 2011 ANSYS, Inc. June 21, 2012 21
Sand Transport
© 2011 ANSYS, Inc. June 21, 2012 22
Example: Sand Transport in Pipelines
• Sand-Water slurry flow in horizontal pipe • Pipe diameter D = 0.0505m
• Pipe length L = 4m
• 30% volume loading
• Four Different Slurry Flow Rates
• DDPM with DEM Collision
• Particle staggering for surface injection
• Low value of Spring Constant as buoyancy force is important.
• Almost 3 millions parcels
Gravity
Slurry Velocity (m/s)
dp/dx (Pa/m)
SRC: Saskatchewan Research Council
Expected Results
To be published in collaboration with Shell
© 2011 ANSYS, Inc. June 21, 2012 23
• Mean Static Pressure is plotted on the line coinciding with the axis of the pipe.
• dp/dx is calculated between z=3m to z=4m as it varies linearly in this range for all the cases.
Results: Pressure Gradient
Slurry Velocity (m/s)
dp/dx (Pa/m)
dp/dx pipe length dp/dx slurry velocity
© 2011 ANSYS, Inc. June 21, 2012 24
• Reduced particle time step size to more accurately model collisions.
– Little difference in predicted pressure gradient.
– Considerable increase in simulation time.
Effect of particle time step size Mixture Velocity (m/s) Baseline Particle Time
Step Size (s) Smaller Particle Time Step Size (s)
0.7 2.50E-04 1.0E-04
1.42 1.00E-04 4.00E-05
3 5.00E-05 2.50E-05
Slurry Velocity (m/s)
dp/dx (Pa/m)
dp/dx slurry velocity
© 2011 ANSYS, Inc. June 21, 2012 25
• It is important to keep particles suspended
• Critical flow velocity which keeps sand particles moving along the pipe depends on • Liquid holdup and flow rates, Pipe diameter, Fluids properties, Sand
properties, Pipe inclination angle
• Many correlations exists for solids transportation in multiphase flow
• Based on experiments for
single phase flow on small pipes
• Lot of variability in measurements
Sand Transport in Pipelines
Hjulstrom Diagram
© 2011 ANSYS, Inc. June 21, 2012 26
Transport paths • Traction or full contact
– sand rolling or sliding across bottom
• Saltation – sand hop/ bounce along bottom
• Bedload – combined traction and saltation
• Suspended load – sand carried without settling
– upward forces > downwarde
Sand Transport in Pipelines
All these paths for sand transport can be addressed by Particulate modeling in ANSYS CFD.
© 2011 ANSYS, Inc. June 21, 2012 27
Demonstrate Case of Particle Lift Off using MPM • Geometry of a long narrow channel
• Steady state periodic flow profile applied at Inlet
• A 200 microns diameter particle was placed on bottom of the channel
• Advanced Turbulence Model
Particle Transport – MPM Simulation
Fine mesh (about 4 fluid cells across particle diameter)
Initial Location of the Particle
Flow Direction
© 2011 ANSYS, Inc. June 21, 2012 28
Lift Force Validation in MPM
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Axial Distance (microns)
Distan
ce fro
m Wa
ll (mi
crons
) Particle Trajectory
Axial Distance (in microns)
Distan
ce from
Wall
MPM is a DNS technique which calculates particle forces directly from pressure and flow field
MPM automatically predicts particle lift force without including any lift force correlation (Saffman etc)
Particle Transport – MPM Simulation
© 2011 ANSYS, Inc. June 21, 2012 29
Proppant Placement
© 2011 ANSYS, Inc. June 21, 2012 30
• Complex multiphase flow problem
• Proppant settles to the bottom – Mound develops –Reaches an equilibrium height
• Until the equilibrium height – Proppant bed gets higher and then it spreads laterally
Example: Proppant Transport
Reference: Patankar, N.A., Joseph, D.D., Wang, J., Barree, R.D., Conway, M., Asadi, M., 2002. Power law correlations for sediment transport in pressure driven channel flows. International Journal of Multiphase Flow. 28. 1269–1292.
© 2011 ANSYS, Inc. June 21, 2012 31
• Drag Force Modified for Dense system
– Single particle drag + Concentration effect + Hindered settling effect
• Collisional and frictional effects (becomes important near packing limit) are considered
Proppant Transport: Granular Model
300 ft
40 ft
Fracture Width = 0.5 cm
Full 3D – Wall Effects and Leak Off – Modeled Slurry flow: Mixture of Frac-Fluid and Proppant
© 2011 ANSYS, Inc. June 21, 2012 32
500 µm 300 µm
time
Proppant Transport: Granular Model
More settling was observed in 500 micron
© 2011 ANSYS, Inc. June 21, 2012 33
Proppant Transport: Wash Out Process
300 µm– Proppant 100 µm - Proppant
The mound started loosing proppant and the height decreased
The mound created a re-circulating zone upstream and allowed settling in this zone The mound grew over a period of time
© 2011 ANSYS, Inc. June 21, 2012 34
• The proppant transport process using DDPM-DEM
• Lagrangian tracking process
• Collision and frictional terms are modeled discretely
• Problem description
• Domain with dimensions: 3 X 0.3 X 0.01 m
• Proppants of 0.5mm size particles, 1 kg/s – 1 Parcel = 10 particles
– 1.8 million parcels at pseudo-steady state
• Water at 4.5 kg/s
Proppant Transport: DEM Analysis
© 2011 ANSYS, Inc. June 21, 2012 35
Proppant Transport – DDPM DEM
Volume fraction of proppant
Velocity of proppant
© 2011 ANSYS, Inc. June 21, 2012 36
• Sand Management is critical in oil production to ensure system integrity and efficiency
• It is important to predict various phenomena involved in sand transport and sedimentation
• ANSYS CFD provides a platform for comprehensive particulate modeling
• Few examples of sand filtration, sand transport and proppant placement were demonstrated
Summary: Sand Management
© 2011 ANSYS, Inc. June 21, 2012 37
Erosion Modeling
© 2011 ANSYS, Inc. June 21, 2012 38
Sand Erosion
• Sand Erosion of pipelines and equipment is a major problem
• Solids entrained in the fluid impinge the walls of piping and equipment causing in removal of wall material, reducing the service life.
• Erosion limits the expected life time of piping details, and is vital in risk management studies
• It is critical to predict the erosion damages in a flow system accurately
© 2011 ANSYS, Inc. June 21, 2012 39
• Erosion is Complex Phenomena, depends on
– Particle properties and particle tracks
– Local Flow and turbulence field
– Surface conditioning
– Multiphase effects • Erosion shield due to solid accumulation
• Damping effect due to liquid film
– Effect of local cavities due to material removal
• Nearly imposition to have a universal erosion model
– Different models for different flow regimes
– Always need experimental data to tune model parameters
Challenges in Erosion Modeling
© 2011 ANSYS, Inc. June 21, 2012 40
• Physical testing of new prototype designs
– Time consuming
– Degree of trial and error
• Semi-empirical models and correlations of erosive wear
– Limited to predicting peak values of wear
– Usually exist only for simple standard geometries
– API RP 14E • Ad-hoc methods that are independent of the sand production rate
• “erosional velocity” – Based on an empirical constant
(C-factor) and the fluid mixture density
Erosion Modeling – Traditional approach
© 2011 ANSYS, Inc. June 21, 2012 41
• CFD modeling provides the user with detailed information on the exact location and magnitude of the erosive wear.
• Single phase Computational Fluid Dynamics simulations
– Applicable for dilute particle phase
– Based on Eulerian-Lagrangian methodology
• Single phase simulation + DPM
– Lots of literature and many erosion models
– Provides detailed information on the exact location and magnitude of the erosive wear
– Potential to allow design to be optimized prior to testing
Erosion Modeling – CFD approach
© 2011 ANSYS, Inc. June 21, 2012 42
• Multiphase CFD Simulations
– More realistic for full particle loading from low, medium to high range
– Based on Eulerian-Granular multi-fluid approach
– Captures four-way couplings including fluid-particle, particle-fluid, particle-particle, and turbulence interactions
– Capture particle shielding and liquid damping effects
– Lacks proper erosion models for abrasive erosion
Erosion Modeling – CFD approach
CFD Modeling Complement Experimental testing for Erosion Predictions
© 2011 ANSYS, Inc. June 21, 2012 43
• Different particulate modeling options • DPM, DDPM, DEM, Eulerian-Granular
• Wide Varieties of Erosion Models are available in ANSYS FLUENT • FLUENT’s Default Erosion Model
• Mclaury et. Al Erosion Model
• Salama & Venkatesh Erosion Model
• Tulsa Erosion Model
• DNV Erosion Model
• Erosion Model based on Wall Shear Stress
• Flexibility to incorporate any erosion model
• Erosion pattern in complex flows and geometries can be predicted with a good accuracy
ANSYS Solution for Erosion Modeling
Contours of Erosion Rate
© 2011 ANSYS, Inc. June 21, 2012 44
• Typical variables affecting Erosion rate • Angle of impingement
• Impact velocity
• Particle diameter
• Particle mass
• Collision frequency between particles and solid walls
• Material properties for particle and solid surface
• Coefficients of restitution for particle-wall collision
m : Mass flow rate of the particles f(a) : Impingement angle function V : Particle impact velocity b : Velocity exponent C(Dp): Particle diameter function
Erosion caused by particle impact
Incoming particulate
ANSYS Solution for Erosion Modeling
© 2011 ANSYS, Inc. June 21, 2012 45
Erosion Model Based on Wall Shear Stress
SSVAER n
w A = Constant (diameter function) n = Velocity Exponent SS = Wall Shear Stress
Erosion Model for Dense System
wsp ERERER Overall Erosion Rate
Dense DPM accounts for particle-particle interaction and solid volume effect on fluid phase
ABRASIVE EROSION: Erosive due to relative motion of solid particles moving nearly parallel to a solid surface
© 2011 ANSYS, Inc. June 21, 2012 46
• Removal of solid surface material due to Erosion creates localized cavities which affect the flow field, particle tracking and hence the erosion.
• Such dynamically changing eroded curvature effect needs to be incorporated for more accurate erosion calculation
• ANSYS FLUENT has developed Erosion-MDM connectivity using a User-defined Function (UDF) to dynamically deform the solid wall surface based on local erosion rate
• Similar workflow has been developed for ANSYS CFX
Coupling Erosion with MDM
© 2011 ANSYS, Inc. June 21, 2012 47
Ei
ri
R
Rri
n
i
Rri
n
i
i
node
i
i
r
L
r
LE
ER
,
,
Ei = Erosion rate for ith face ri = Distance of ith face center from the node L = Minimum cell length connected to the node R = Radius of region considered for averaging (user input is R/L) n = Rate of decay (user input) f = Maximum mesh move limit (user input)
fLdensityWall
ERL node
_
Erosion Modeling – Coupled Simulation
• The erosion rate is averaged and smoothed according to the equation:
A value of 0 for “n” will result in equal weighting for all nodes within “R”. A very large value of “n” will render the smoothing algorithm negligible.
© 2011 ANSYS, Inc. June 21, 2012 48 48
Erosion Module
• Easy to use template to perform an Erosion Simulation through a single GUI Panel
• User inputs drive the UDF and journal file in the background
• Varieties of Erosion Models to choose
• Built-in Smart defaults for DPM settings
• Customized post processing for erosion rate
• Complete Automation of Erosion-MDM coupled simulation
• Including postprocessing and animation
• Ability to allow multiphase erosion simulations
• Choose secondary phase for particle tracking
© 2011 ANSYS, Inc. June 21, 2012 49
Option to start a new Erosion-MDM simulation or restart from the existing data file at previous time interval
Opens Fluent’s panel to read the case file for the flow field
Opens Fluent’s DPM injection panel to define particle injections
Opens Fluent’s boundary condition panel to set DPM BCs for wall zones
Opens Fluent’s DPM panel to set parameters for particle tracking
Opens Fluent’s panel to read the data file for the flow field
Display erosion rate on all wall zones
Display cumulative eroded distance at wall zones
Opens Fluent’s panel to start iterating for erosion-only analysis
Option to run erosion-only or erosion-MDM coupled simulation
Various erosion models to choose from
Option to choose secondary phase flow velocities for DPM particle tracking
Opens panel to define required parameters for Erosion-MDM coupling
Opens panel to start erosion-MDM simulation
Erosion Module
© 2011 ANSYS, Inc. June 21, 2012 50
Example – Control and Delay Erosion
Courtesy of DNV
Area of high erosion
Particle trajectories
colored by velocity and
associated erosion area
for two chokes
Flow Inlet
Problem • Particle impact at the small area with
high velocity causing excessive erosion
Solution • Modify exit flow from chock without
causing additional pressure drop. • ANSYS multiphase flow solutions to
understand and change particulate flow patterns
Result • Modified chock geometry leads to flow
streamlines parallel to exit pipe. • Increase particle impact area while
reducing particle impact velocity • Reduce chock maintenance and
replacement cost
© 2011 ANSYS, Inc. June 21, 2012 51
Double Elbow Geometry
Relatively Low solid loading (~8% volume loading)
DPM vs DDPM Simulation and same Erosion Settings
Particle shielding effect captured in multiphase simulation
Single phase predicts conservative erosion
Example: Single Phase vs Multiphase Erosion
Single Phase Erosion Multiphase Erosion Sand Volume Fraction
© 2011 ANSYS, Inc. June 21, 2012 52
Example: Erosion in Gas-Liquid-Solid System
Liquid Volume Fraction Contours
Solid Volume Fraction Contours
Vapor Velocity Contours Contours of Erosion Rate
Erosion in a Pipe Assembly Courtesy of Suncor
Low Erosion due to liquid cushion and particle shielding
© 2011 ANSYS, Inc. June 21, 2012 53
CFD Simulation to analyze flow field and erosion pattern in frac pack tools
Calibration of erosion model based on lab tests and Erosion pattern compared with large scale tests.
Tool Erosion in Gravel Pack: (OTC 17452 – Halliburton)
Erosion pattern on the inside surface of upper extension sleeve
Fluid Velocity Proppant VOF Turbulent Slurry flow with high proppant concentrations Non-newtonian fluids Calibration of Impact angle function
© 2011 ANSYS, Inc. June 21, 2012 54
Erosion - MDM
Contour of Erosion Rate
Contour of Total Eroded Distance
© 2011 ANSYS, Inc. June 21, 2012 55
Erosion - MDM
Plots of erosion contours in a 4 inch test case
FLOW
Larger ID After 42 hr
Eroded Material is Removed -> Better Material Thickness Prediction
© 2011 ANSYS, Inc. June 21, 2012 56
• It is important to predict erosion rate accurately
• Erosion is a complex phenomena
• Semi-empirical models and correlations are not enough
• Need for CFD in erosion modeling
• CFD can provide valuable information for erosion predictions
• Multiphase flow modeling for dense slurry
• Erosion-MDM coupling
• ANSYS CFD equipped with all required modeling needs
• ANSYS CFD - Proven approach for many erosion studies for oil & gas industries
Summary: Erosion Modeling
© 2011 ANSYS, Inc. June 21, 2012 57
THANK YOU