You are on page 1of 12

Paper No.

3964

Sand Fines Erosion in Gas Pipelines – Experiments and CFD Modeling

Praveen Kumar G, Byron Smith R J*, Damodaran Vedapuri


Tridiagonal Solutions Inc
12703 Spectrum Drive, Suite 101
San Antonio, Texas, 78249

Hariprasad J Subramani, Lee D Rhyne


Chevron Energy Technology Company
1400 Smith Street, Houston, TX 77002

ABSTRACT

Solid particle erosion is an important flow assurance concern encountered in existing as well as new oil
and gas pipelines. In spite of the sand control/management techniques implemented down-hole, fine
sand (less than 50-75 microns) often may find its way into the piping components of onshore, offshore,
and subsea facilities causing erosion/wear and subsequent pipeline integrity issues. Suitable design of
the piping components with capability to monitor wear at hotspots and/or flow manipulations could be
made if the erosive impact of this fine sand could be quantified by predictive methods using CFD
framework. In this work, experimental results and CFD erosion prediction in a 50.8 mm (2 in) erosion
test loop of elbows in series have been reported. The test has been carried out using air/sand system
with air flow rate of 32 m/s and 28 micron sand of 0.0025 volume percentage concentration. The
thickness loss in the elbows measured using ultrasonic erosion probes has been reported. CFD
modeling of the elbows in series has been carried out with the single particle erosion models reported
by Oka et al. (Wear, 2005), DNV (DNV RP o501), Grant and Tabakoff (J. Aircraft, 1975) and E/CRC
(Zhang et al., J. Fluids Engg., 2009) and predictive capability of these models has been compared to
experimental results.

Key words: Sand Fines Erosion, CFD, Erosion Test Loop

INTRODUCTION

Erosion of pipe fittings and associated equipment in oil and gas pipelines is a persistent problem
encountered by pipeline engineers. Many sand management systems have been implemented over a
period of time to try to exclude sand at its source down hole of the well. These sand exclusion
techniques involve gravel packing at the well head and/or using screens to exclude sand from entering
the pipeline. These sand exclusion methods along with continuous sand monitoring and control have
been successful in cutting down the production of sand in the pipelines lines to a large extent and are

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
1
extensively used in oil and gas production wells. In spite of these measures, based on the geography of
the well location, fine sand (less than 50 – 75 microns) finds its way through the sand exclusion system
and travels along with the production fluid through the pipelines causing erosive wear and damage to
the piping components. The damage is more pronounced in sections where there is an abrupt change
in direction of flow like chokes, choke valves, elbows and tees which are integral part of the piping
system. This sand will also cause damage to the separation equipment connected to these pipelines.
Erosion can also cause problems in subsea and cross country pipelines, as vast tracks of the pipeline
are in inaccessible locations and they can be only managed with proper design of the pipes and piping
components with adequate offset for erosive loss of material with time. This requires the designers of
pipelines to have good erosion prediction tools.

Erosion prediction is typically attempted using empirical and semi empirical modeling of the material
removal from the wall of the piping components due to it being impacted by the flowing sand. The
common parameters included in these modeling approaches include particle impact velocity, angle of
impact and physical properties of particle and wall1.Qualitative and quantitative erosion prediction
equations have been attempted since the times of Finnie2 in 1960, who developed an erosion model
based on the cutting of material of construction by sand particles. This model could not predict the
erosion occurring at higher impact angles and hence Bitter3,4 modified Finnie’s equation by adding
deformation wear to Finnie’s cutting action. The Finnie and Bitter equations have been modified by
several authors later to fit to their experimental data by changing the exponents and other empirical
constants. Lyczkowski and Bouillard5 has grouped all the erosion prediction models under three
categories namely single particle models, dense phase fluidized bed models and energy dissipation
models. The single particle models are for dilute flows where particle – particle interactions are
neglected and only particle – wall interactions are considered. All the pipeline erosion models come
under this category. The dense phase models incorporate interactions between particle – particle – wall
and are used for fluidized beds. The energy dissipation models track the energy lost by particles to the
wall and relate it to the erosion and these are computationally intensive models as they require
resolving of all individual particles and their impact on the wall of the pipeline.

In spite of the number of erosion prediction models being published, few of these models have their
applicability over a wide range of operating conditions. The API* RP 14E6 and DNV* RP O5017 are the
widely used recommended practice for erosion prediction in the oil and gas industry. API RP 14E
provides the maximum fluid velocity to minimize erosion whereas DNV RP O501 allows the designer to
calculate the erosion rate. Apart from these recommended practices, other models published by Grant
and Tabakoff8, Oka et al9, ECRC-University of Tulsa10 etc. are also frequently used for erosion rate
prediction. Considering the fact that the sand particles impact location and velocity are based on the
fluid flow phenomena, Computational Fluid Dynamics (CFD) is being increasingly used to predict
erosion. In this approach, the fluid flow with turbulence is modeled by the CFD and the particle velocity
and its angle of impact are used to compute erosion rate based on the erosion models. Most of the
commercial CFD packages have the option of predicting erosion in pipelines using different models.
The implementation of erosion models into CFD to predict erosion rate has been carried out by several
authors11 – 15.

Even though several erosion models are available, most of the studies on using these models have
been tested on experimental data gathered for particle sizes greater than 100 – 150 microns. Moreover,
most of the studies were carried out on single elbows in experimental loops. But in real time
applications, there are locations where there is more than one elbow available in quick succession. The
literature search did not yield studies reporting the erosion caused by fine sand in elbows fitted in series
in a pipe line. To overcome these lacunae in data, this study concentrated on the erosive behavior of
fine sand in three elbows in series, both experimentally and through modeling. Selected erosion
prediction equations were analyzed for their prediction capability of fine sand erosion using CFD.
___________________________________________________
*API – American Petroleum Institute Recommended Practice
DNV – Det Norske Veritas Recommended Practice

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
2
EXPERIMENTAL PROCEDURE

To carry out the erosion studies, a carbon steel (ASTM A 106 Gr. B)* erosion loop with Nominal Pipe
Size (NPS) of 50.8 mm (2 in) of schedule 80 [ID= 49.25 mm (1.939 in)] was constructed. A screw
compressor is used to supply air at 6.9 atm to an air receiving tank. The air flow to the test loop is
controlled by a combination of pressure regulator and flow control valve. The pressure regulator
reduces the pressure to the test section and maintains it slightly above the atmospheric pressure. A
constant air flow rate of 130 acfm was maintained in the loop using a flow meter. The solid injection
system consists of a pressurized dip stick mechanism. This works on the principle of local fluidization of
the sand which is then sucked in and delivered into the pipeline using compressed air with pressure
slightly higher than the pipeline pressure. The air with sand then flows through a choke which was 56
inches upstream of a series of three elbows. The choke is made of 38.1 mm (1.5 in) NPS and 80
schedule carbon steel pipe [ID= 38.1 mm (1.5 in), 1016 mm (40 in) long]. The first two elbows are short
radius elbows (R/D = 1) and are kept apart horizontally at a distance of 609.6 mm (24 in). The third
elbow is 1422.4 mm (56 in) from the second elbow. Elbow three is a standard elbow (R/D=1.5) and it
points up vertically. The pipeline from elbow three is connected to a cyclone separator followed by a
cartridge filter to remove the fine sand and the air is vented out. The vent is open to the atmosphere
and the series of pressure gauges reported a total pressure drop of 0.07 atm across the test section.
The sand is passed through the system twice and then discarded. The schematic of the erosion testing
loop is given in Figure 1.

Air Out Filter

Cyclone

Elbow 3 Elbow 2

Sand Collection
Sand Injection System
Compressor

Choke Elbow 1
Flow Meter

Pressure Gauges
Air Receiving Tank

Figure 1: Schematic Diagram of the Experimental Setup

AGSCO Quartz #325 (AGSCO Corporation, Illinois, USA) with a mean diameter of 28 microns, density
of 2650 kg/m3 and Mohs hardness of 7 was used in the experiments. The sand is loaded in the solid
injection system and air is allowed to flow at the required rate of 32.2 m/s through the loop. After a
steady flow is established in the loop, sand injection is started. The experiment was carried out for a
total of 14 hours during which 234 kg of sand was passed through the loop. After the experiment was
over, the elbows were dismantled and subjected to erosion testing using an ultrasonic thickness
measurement device. These elbows were also visually checked for erosion by cutting them open.
__________________
* UNS: K03006
©2014 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
3
The ultrasonic thickness measurement device consists of a pulse transmitter/receiver, oscilloscope and
a contact transducer. It works on the principle of equating the time of ultrasound propagation through
the material to the thickness of the material. The elbows are marked with a grid of horizontal lines along
the center line and 10 degrees above and below the centerline. The vertical lines are marked every 5
degrees along the length of the elbows too. The intersection of the grid lines are the places where the
measurements are made. To measure the thickness of the material, the transducer is placed on the
measurement point. The pulse transmitter/receiver sends and receives the ultrasonic sound through the
transducer and the signal is processed and stored in the oscilloscope. Software is developed in-house
to filter out the noise and relate the signal to the thickness in microns. All measurements were
triplicated to ensure reproducibility. The error among triplicates was approximately 50 microns.

CFD SIMULATION FORMULATION

The CFD simulations for erosion prediction were carried out using the commercial solver Ansys
Fluent. The geometry was meshed with 3.38 million cells with the first layer thickness (48 microns) fixed
approximately twice the diameter of the particle. The grid used for the simulation (Figure 2) reflects the
imposition of the choke on the gridding. The boundary layer used was able to provide an average y+
value of approximately 7.5 for the first grid point from the wall. The Realizable k – ε turbulence model
with enhanced wall function was found to be suitable for capturing the flow path based on previous
studies carried out by the authors. The fluid flow uses the Eulerian frame work and the particles are
tracked in the flow using the Lagrangian framework through the Discrete Particle Model (DPM). The
turbulence – particle interactions are captured by the Discrete Random Walk (DRW) method. DRW is a
stochastic model used to predict the trajectories of particles in a turbulent flow using the mean fluid
phase velocity16 and velocity fluctuations based on local turbulence levels. Mass and velocity residuals
converged to less than 1E-05 and turbulence residuals converged to less than 1E-3. Over 99% of the
particles tracked exited the system. The particle – wall interactions are characterized by the coefficient
of restitution. The normal and tangential coefficient of restitution is set at the wall of the geometry which
determines the rebound nature of the particles impacting the wall. The stochastic restitution coefficient
model proposed by Forder et al11 for carbon steel - sand is used in the simulations. Considering the
criteria of Elghobashi reported by Tian17, the particle - particle interaction can be neglected as the ratio
of mass of particles per unit volume to the particle density is less than 10-3.

Figure 2: Mesh used in the Simulation

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
4
The Erosion rates are predicted using the following erosion prediction models.
1. DNV Model7
2. ECRC – University of Tulsa Model10
3. Oka et al Model9
4. Grant and Tabakoff Model8
The variables that are required for computing erosion in these models include the particle impact
velocity, the angle of impact and the solid particle mass flow rate. The Oka et al. model also uses the
material hardness in the prediction whereas the Grant and Tabakoff model has an inbuilt coefficient of
restitution term inside the model. The - particle velocity exponent vary in all the models and so do the
units for erosion rate. DNV model reports erosion rate in mm/yr, ECRC in kg/kgsand, Oka et al in
mm3/kgsand and Grant and Tabakoff in kg/kgsand .These models were coded as User Defined Function
(UDF) into the commercial solver Ansys Fluent where the predicted erosion rates on the wall surfaces
were converted into thickness loss in microns for comparison with experimental data.

RESULTS

The signals obtained from the ultrasonic erosion measurement device were related to the thickness
loss in microns. It is observed in all the three elbows that the fine sand erodes the entire concave part
of the elbows. There are a few comparatively higher erosion spots in the elbows. These results are
different from the sand erosion data reported for larger sand particles (particles with mean diameter
above 100 microns) where one high erosion spot is seen with less erosion throughout the remaining
areas10,12,14. Thus the fine sand creates generalized erosion across a large portion of the wall surface.
The comparatively higher erosion spots were not exactly located on the centerline of the elbows, but
more dependent on the secondary flows of the carrier fluids path in elbows in series. The three elbows
in series are numbered 1 – 3 for convenience from the inlet. To show the broad regions eroded by fine
sand, the measured values along the centerline of the elbows and those corresponding to 200 above
and below the centerline are also reported and Figure 3 shows the measurement region. Plots of the
experimental values of wall thickness loss on the three elbows are depicted in Figures 4 - 6.

Figure 3: Erosion measurement Plane

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
5
Figure 4: Experimental erosion damage measured for Elbow 1 (along the centerline, 200 above
centerline and 200 below centerline)

Figure 5: Experimental erosion damage measured for Elbow 2(along the centerline, 200 above
centerline and 200 below centerline)

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
6
Figure 6: Experimental erosion damage measured for Elbow 3 (along the centerline, 200 above
centerline and 200 below centerline)

For elbow 1, the experimental results showed that the thickness loss increases as the angle from the
inlet increases from 300 to 600 and then decreases toward the outlet. This pattern is similar to one
observed for larger particles, greater than 100-150 microns, except that the angle where peak erosion
is observed has shifted to the right for fine particles. A thickness loss of 540 microns was observed
near 550 from the inlet on the centerline. A thickness loss patch of 400+ microns extended from 500 to
700 from the inlet and 00 to 200 from the centerline. Elbow 2 showed erosion throughout the elbow
concave face. The experiments showed an erosion zone, with a thickness loss ranging from 250 to 350
microns, from an angle of 500 to the exit of the elbow and extended 200 both below and above the
centerline. Elbow 3 is 56 inches (28D) away from elbow 2, but diverts the flow from the horizontal
direction vertically upwards. Even though the magnitude of maximum erosion is smaller than elbows 1
and 2, more regions underwent erosion in elbow 3. The maximum erosion extended from 200 from
entrance till the exit of the elbow. The elbow had four large patches of erosion below the centerline with
an average thickness loss of 300 microns. The erosion pattern in elbow 1 indicates localized impact. In
elbow 2 and 3 such a pattern is not observed. An erosion patch is observed. Based on the experimental
data one can speculate that the fine particles are redistributed due to secondary flows and impact the
walls more uniformly across a patch in elbows 2 and 3. In elbow 1 the presence of a choke upstream
may result in a localized impact. The erosion levels are higher for elbow 1 than elbow 2 and 3 and this
is likely from the higher jet velocity.

The erosion rate computed using CFD and the four selected erosion prediction models along the
centerline for elbow 1 are presented in Figure 7. All the four models predicted erosion throughout the
concave part of the elbow. The erosion rates increase from the entrance up to a sweep angle of 800
and then decreases towards the exit of the elbow. This trend is reasonable considering the fact that
elbow 1 is a short radius elbow with R/D=1. Moreover, due to the R/D being 1, there is a high velocity
created close to the intrados of the bend which creates a recirculation region downstream. The
recirculation region effectively removes flow cross-sectional area which further promotes the maximum
erosion rates being experienced on the extrados exit of the elbow. The erosion profile is however not
predicted correctly in comparison to the experimental values. All the models predicted the maximum
thickness loss at 750 whereas the experimental peak was obtained at 550. Except for the ECRC model
which over predicted the thickness loss by 2 times the reported thickness loss, all the other three

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
7
models predicted the thickness loss within +50% of the experimental values. The Oka et al., model
predicted the thickness loss more accurately although the location of that loss is displaced. The error
may be a result of jet dynamics still present from the choke that is about 28D upstream. The jet would
drive a more peaked velocity profile whose momentum would drive the high speed flows into the wall
sooner than expected. Since all the erosion models have similar shape and error, the error could be in
the fluid flow model turbulence not being properly propagated downstream or it could also be in the
particle – turbulence coupling.

Figure 7: Experimental and CFD Predicted erosion damage for Elbow # 1 along centerline

Figure 8: Experimental and CFD Predicted erosion damage for Elbow # 2 plotted along
centerline

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
8
The prediction of the erosion damage using CFD for elbow 2 is plotted in Figure 8. All the models
showed the maximum peaks towards the exit of the elbow. But the experimental data showed uniform
erosion throughout the centerline with no distinguished peaks. For elbow 2, the DNV model provided
best prediction of the thickness loss even though the errors of the prediction are +80% to -30% at any
particular location. The other models over predict the maximum erosion by 100-400% and they under-
predict erosion in the actual locations of observed maximal erosion. Figure 9 shows the particle path
lines as predicted by CFD. When visualizing the particle path lines in CFD, it is observed that elbow 1
induces a swirl on the particles. A large fraction of the particle concentrates on the circumference at the
inlet to elbow 2 and the particles entering the elbow on the convex side have a direct impact on the
concave surface of the elbow at an angle of impact of 450. The particle concentrations as predicted by
the model are also high at that location resulting in a higher erosion rate prediction at this location. So if
the elbows are close enough the particle-turbulence interaction is not being appropriately modeled for
fine particles resulting in the erosion profile too being off the mark. More studies are needed to see as
to how the predictions could be improved for elbows close to each other.

Figure 9: Particle path lines between elbows 1 and 2 as estimated by CFD analysis. The image
on right shows a zoomed view of particle tracks for elbow 2

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
9
Figure 10: Experimental and CFD Predicted erosion damage for Elbow # 3 plotted along
centerline

The CFD prediction of the thickness loss using the erosion models for elbow 3 are depicted in Figure
10. Experimentally, elbow 3 also showed a constant erosion profile whereas all the erosion models
predicted a peak thickness loss at 750. The erosion regions in CFD were located below the centerline
and similar pattern was seen in the experiment. Since the distance between the elbow 2 and 3 is large,
the CFD study predicts the peak erosion similar to Elbow 1 which is not seen in the experimental data.
The magnitude of thickness loss was well predicted by the Oka et al model in comparison to the other
models.

In the experiments carried out, the peak thickness loss was seen only in Elbow 1 which being the first
impact and then the other two elbows show uniform erosion. But in CFD, due to the upstream length,
Elbow 1 and Elbow 3 show maximum thickness loss peaks which shows that for fine sand erosion
prediction, better erosion models are required.

CONCLUSIONS

The experiments detailed herein provide a good benchmark dataset for erosion of carbon steel by fine
sand with a mean particle diameter of 28 microns in high speed gas flowing through a choke and three
elbows in series. The experiments showed that the erosion of the elbows is wide spread along the
concave face of the elbows and with multiple highly eroded spots. Certain maximum erosion spots were
also found. This is in contrast to the focused erosion experienced in sand particles with diameter
greater than 100 microns. For elbow 2 which is close to elbow 1, the predicted erosion profile was poor.

CFD with four widely-used erosion models was used to predict the observed erosion. The models
tended to over-predict the maximum erosion by as much as 400% compared to experiment. The TU-

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
10
ECRC model always over-predicted this data by at least 100%. Almost all the models gave a peaked
focused erosion pattern as compared to the more general erosion experimentally observed. The
predicted erosion pattern in the downstream elbows displaced the location of maximum erosion. This
difference in patterns caused under-prediction of erosion through much of the elbow and over-
prediction of the maximum erosion at a displaced location.

Such large errors in the CFD predicted erosion indicate that there is a deficit in some of the basic
physics of particle-turbulence interactions or turbulence transport that warrants further investigation.
This deficit in erosion predictions may be the result of different mechanisms occurring with small
particles as compared to big particles. The experimental results also indicate that caution is needed
when predicting erosion in systems with pipe fittings (chokes, valves, elbows, tees) within close
proximity to each other such as in manifolds or wellheads.

REFERENCES

1. H.C. Meng and K.C. Ludema, "Wear models and predictive equations: their form and content",
Wear 181-183 (1995) pp. 443-457

2. I. Finnie, "Erosion of Surfaces by Solid Particles", Wear 3 (1960) pp. 87 - 103

3. J. G. A. Bitter, “A study of erosion phenomena – Part I”, Wear 6 (1963) pp. 5 – 21

4. J. G. A. Bitter, “A study of erosion phenomena – Part II”, Wear 6 (1963) pp. 169 – 190

5. Robert W. Lyczkowski and Jacques X. Bouillard, "State-of-the-art review of erosion modeling in


fluid/solids systems", Progress in Energy and Combustion Science 28 (2002) pp. 543–602

6. API RP 14E (1991, reaffirmed 2000), “Recommended practice for design and installation of
offshore production platform piping systems”, (Washington, American Petroleum Institute)

7. DNV RP O501 (Revision 4.2 - 2007), "DNV Recommended Practice RP o501 - Erosive Wear in
Piping Systems", (DNV)
8. G. Grant and W. Tabakoff, “Erosion prediction in turbo machinery resulting from environmental
solid particles”, Journal of Aircraft 12 (1975) pp. 471 – 478

9. Y.I. Oka., K. Okamura and T. Yoshida, “Practical estimation of erosion damage caused by solid
particle impact – Part 1: Effects of impact parameters on a predictive equation”, Wear 259
(2005) PP. 95–101

10. Yongli Zhang, Brenton S. McLaury and Siamack A. Shirazi, “Improvements of particle near-wall
velocity and erosion predictions using a commercial CFD code”, Journal of Fluids Engineering
131 (2009)

11. Alister Forder, Martin Thew and David Harrison, " A numerical investigation of solid particle
erosion experienced within oilfield control valves", Wear 216 (1998) pp. 184 – 193

12. Q. Chen and D.Y. Li, "Computer simulation of solid particle erosion", Wear 254 (2003) pp. 203–
210

13. Xianghui Chen, Brenton S. McLaury and Siamack A. Shirazi, "Application and experimental
validation of a computational fluid dynamics (CFD)-based erosion prediction model in elbows
and plugged tees", Computers & Fluids 33 (2004) pp. 1251–1272

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
11
14. Derrick O. Njobuenwu, Michael Fairweather, "Modelling of pipe bend erosion by dilute particle
suspensions", Computers and Chemical Engineering 42 (2012) pp. 235– 247

15. F N Fard, B McLaury and S Shirazi, " Effect of Cell Size on Particle - Eddy Interaction and
Erosion Predictions Using Commercial CFD Software (FLUENT)", Proceedings of the ASME
2012 Fluids Engineering Summer Meeting, FEDSM2012-72294 (Rio Grande, Puerti Rico July 8-
12, 2012)

16. Ansys Fluent 13.0 Theory Guide, Ansys Inc

17. Zhaofeng Tian, "Numerical Modelling of Turbulent Gas-Particle Flow and Its Applications ", PhD
Thesis, RMIT University (2006), Australia

©2014 by NACE International.


Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
12

You might also like