EROSION EVALUATION ON OFFSHORE PRODUCTION FACILITY BY

DIFFERENT CFD APPROACHES: LANGRANGIAN/EULERIAN AND

EULERIAN/EULERIAN

L. Cadei, A. Bennardo, F. Podenzani, N. Buongiorno, A. Bianco, S. Masi, Eni Spa

This paper was presented at the 13th Offshore Mediterranean Conference and Exhibition in Ravenna, Italy, March 29-31, 2017. It was
selected for presentation by OMC 2017 Programme Committee following review of information contained in the abstract submitted by
the author(s). The Paper as presented at OMC 2017 has not been reviewed by the Programme Committee.

ABSTRACT

The current industrial standards adopted to simulate the liquid particle erosion phenomena are
based on 1-D equations prescribed by the API114E, which computes the erosion limit velocity
related to different materials considered. Therefore, this method relies on simplified models and
empirical correlations, validated by experimental data, but characterized by the requirements of an
over-design margin factor, to face the large uncertainty intrinsic with the physical phenomenon.
In this paper, two CFD multiphase procedures able to simulate the liquid erosion effect in
continuous gas phase are presented and applied to support the start-up operations of a new
gas/condensate field located offshore in Central America. Indeed, the insurgence of dangerous
liquid/gas erosions phenomena, can affect significantly the platform production manifold before the
sea line.
The procedures are based on two different approaches and tools: the Eulerian/Eulerian
implemented on OpenFOAM and the Lagrangian/Eulerian investigated with Fluent. In both cases,
the flux is considered fully dispersed, and a new algorithm that evaluates and sets the liquid droplet
diameter as a function of Webber number, flow regime and geometry variations has been included.
In the first case both liquid and gas are modelled using the Eulerian framework, applying an
innovative application devolved in OpenFOAM for two-phase dynamics. The critical erosion
velocity is computed along with the superficial pressure and consequently the shear stress on the
wall. This allows to apply an experimental model (from literature) able to estimate the rate of
erosion to the Eulerian solver. On the other hand, the second approach implements a coupled
Lagrangian (liquid) Eulerian (gas) framework. In this case the dispersed phase is not able to
penetrate the continuous one, but it is absorbed by the wall after the erosive impact. The same
reference model for erosion rate evaluation has been adopted to tune the tools already offered by
Fluent. The comparison of the results obtained by both CFD analyseis and industrial standards is
reported and within this work. Final recommendations to be applied on field for start-up operations
have been summarized.
The CFD procedures developed seems to offer the possibility to gain a deeper understanding of
erosional physical phenomenon effects: erosion rate, critical velocity, shear stress and phases
distribution. This gives the capability to reduce the uncertainty related to the industrial standards,
maintaining the computational efforts for the Oil and Gas time scale competitive and optimizing the
operating conditions during the entire life of the asset.

INTRODUCTION

The global energy demand has grown exponentially in the last century and it will continue in the
next twenty years simply because the world economy carries on to expand, requiring always more
energy. The fuel mix continues to shift towards sustainable and renewable sources, however fossil
fuels remain the dominant source of energy powering the world economy, supplying 60% of the

1
energy increase up to 2035, with gas that is going to assume a central role due to the new
discoveries and to the advantageous policies.
The relentless depletion of the traditional and easily exploitable reservoir has led the major Oil
Companies to focus their attention on more complex reservoirs, implementing innovative
production techniques and facing always more frequently critical phenomena, such as the
insurgence of multiphase flow. In particular, this last operating condition is increasingly related to
the exploitation of gas reservoirs and can lead to the loss of production capacity of the entire
production system, decreasing eventually the company profit. The attention has been pointed on
the improvement of the production system efficiencies through the development of new
methodologies able to handle the multiphase flow problems.
The current paper is focused on the development and application of advanced CFD procedures in
order to manage one of the main issues concerning multiphase flow: the erosion phenomenon. In
particular, the CFD analysis has been applied to a real case study, implementing two different
methodologies:
1. Eulerian-Eulerian approach. In this case, the multiphase solver used is an innovative in-
house CFD procedure developed specifically for two-phase flow. The numerical calculation
are performed using OpenFOAM, an advanced and free CFD Toolbox, able to customize
and extend software solutions to the simulations process. The most important characteristic
of OpenFOAM is the possibility to create solvers ad hoc in order to study several physical
problems such as the continuum mechanic;
2. Eulerian-Lagrangian approach. The algorithm used is a proven and reliable procedure
already applied within the oil and gas business, in a wide range of two and three-phase
problems. The computations occurred by using Ansys Fluent, a proprietary software, which
includes large spectrum of already developed tools for fluid dynamic analysis.
The erosion evaluation is based on the provisions given by the American Petroleum Institute (API
RP 14E), which represents the industrial standard currently applied in mono-dimensional models.
The calculation methodologies applied for the erosion in oil and gas applications, due to water
droplets, has been modified and tuned starting from the standard model usually applied for erosion
caused by sand particles. New parameters from literatures have been implemented in order to
simulate the erosion due to liquid particle impact.
The current work can be seen as a further development in the erosion evaluation analysis,
implementing inside a 3D multiphase CFD framework a tuned tool for erosion evaluation, based on
both the API RP 14E and the concept of Erosion Rate.
In this paper, a real case study has been analysed, applying both CFD procedures and validating
the results thanks to the industrial standards usually used. The work has been included into a full
field de-bottlenecking project, which aims at maximizing the hydrocarbon production of the wells.
The case study deals with an offshore gas and condensate field, where the hydrocarbon fluids
produced are collected on an offshore platform that works as a manifold, gathering the multiphase
flow to an on-shore treatment plant. The erosion problems have been analysed on the piping
installed on the platform, just before the sea line manifold and could arise due to the production
increase that has to be achieved through the debottlenecking project.

EROSION EVALUATION BACKGROUND

Nowadays, there is a much better understanding of the droplet erosion mechanism than particulate
erosion one. Droplet erosion is obviously related to wet gas and multiphase flows in which droplets
can form inside the stream. The erosion rate is dependent on a wide number of factors including
the droplet size, the impact velocity, the impact frequency, and the liquid and gas density and
viscosity. As many of these parameters are unknown due to the field complex conditions, it is very
difficult to predict the rate of droplet erosion. The analysis of this factor should be problematical
also in laboratory tests and the simulation of real operating conditions is frequently very difficult.

2
The equation proposed by the American Petroleum Institute and reported inside the
Recommended Practice 14E (API RP 14E) for the calculation of the erosional velocity ve for two-
phase flow (liquid/gas system) can be written as follow:

(1)

where c is an empirical constant and ρmix is the two-phase density of the mixture expressed in
American Unit of measurement (lb/ft3). Typical values are provided for the empirical constant. In
particular, c should be equal to 100 for continuous service and c equal to 125 for intermittent
service, considering solid-free fluids. Some authors have proposed an empirical constant
increased up to c = 250 when corrosion is not anticipated and that the basis for API RP 14E may
be due to liquid impact erosion.
The erosional velocity calculated, thanks to the API model, has been demonstrated to be very
conservative compared to the experimental data from literature. Shirazi and McLaury [9, 12] have
carried out extensive theoretical and experimental research on sand erosion but not proper
investigations on liquid/gas erosion has not been done for oil and gas industry applications. The
current paper starts the implementation of the erosion model inside CFD procedures from this
experimental work and from the work on erosion of solid particle proposed by Edwards [9].
Some researchers extended the velocity limit by increasing the empirical constant as for example
Salama and Venkatesh [12], who have proposed an increment in c up to 300 for liquid
impingement erosional velocity, state that solids-free erosion only occurs at very high velocities.
Obviously, high velocities can cause a high-pressure drops along the production system, therefore
the conditions required for droplet erosion are unlikely to occur in correctly designed production
pipework systems. Salama and Venkatesh [12] also collate published droplet impingement erosion
threshold velocities for different steels. These values range from 26 to 118 m/s. This approach is
unfortunately based on semi-experimental correlation and usually is valid only in a limited range of
operating conditions. Experimental and CFD studies related to liquid droplet erosion have been
conducted in other fields such as in aerospace engineering and in power plant industries. It has to
be considered that the impingement velocity for these applications is much higher than the
operational velocities in the oil and gas industry not only because of erosion risk but also due to
pressure drop and other production limitations. Therefore, the semi-empirical relationship derived
from these studies could not be applied to the oil and gas industry without further investigation.
However, the CFD approach to the physical phenomenon could be implemented to develop
models and calculation procedures to predict erosion failures in oil and gas typical system.
A more comprehensive approach to calculate erosion due to liquid impacts is to estimate the
erosion ratio of the material using impact variables and calculate the thickness loss rate using the
rate of water droplets that impinge a specific area. In this work, the standard model applied in CFD
codes for solid particle erosion has been used, tuning it for gas/liquid system. This approach is
based on the work of R. Li et al. [7, 8] that considers the multiphase flow conditions in nuclear
power plant. Here the CFD has been used to predict and analyse the erosion main characteristics,
gaining a deep insight into the physical phenomenon, which is particularly challenging to obtain
replicating real cases with experiments. The application of this approach to two different CFD
methodologies has allowed to refine the tool developed, achieving a high accuracy description of
the real field erosion problem and monitoring a wide range of critical parameters.

CFD MODEL DESCRIPTION

The Computational fluid dynamic approaches used for the multiphase flow modelling will be briefly
described in the next paragraphs. Moreover, the further models developed and implemented within
the CFD solvers in order to evaluate the erosion process will also be presented in this chapter.

3
Eulerian-Eulerian multiphase modeling
A brief description CFD in-house procedure developed is presented in this paragraph. The CFD
final algorithm represents the synthesis of two main different models implemented in series:
1. Simplified Single-Phase approach.
This solver has been developed to overcome the high demand of computation resources
when the simulation of a real geometry is addressed. The Navier-Sokes equation is solved
only for the continues phase considered as incompressible, and the segregation process of
the liquid droplet is modelled with the transport of the scalar variable α, which represents
the volumetric fraction of liquid in the total fluid flow. The idea is to describe the effect of the
dispersed phase with the help of a simple scalar transport equation, reducing in this way
the number of the equations to be solved to simulate the multiphase phenomenon. The
simplified approach presented here allows solving only the continuity equation and one
momentum equation for the continuous phase. A source term Sp is mathematically
introduced as a scalar quantity inside the formulation of continuity equation and should
model the dispersed phase effect on the mixture flow. The source term represents the
following forces: drag, buoyancy and gravity. Considering the physics of this process, it is
possible to state that these forces produce a flow of liquid particles in the resultant direction
(usually mainly determined by the gravity force direction), which can be conveniently
represented by the divergence operator. This formulation allows rearranging the continuity
equation. As follow:

(2)

The volumetric flux of this new divergence term has been represented by the terminal
velocity defined as follow (after a tuning process of the algorithm):

(3)

The Drag coefficient is computed with the Schiller-Naumann model, assuming the relative
velocity equal to the settling velocity. The value of Vt is determined by means of an iterative
procedure. The coalescence mechanism has been simulated by a new model developed
for this scope. The idea in this case is to find a threshold value of the liquid volume fraction
αlim, above which the droplets start to aggregate, generating particles with larger diameter.
The coalescence model has been opportunely tuned using real equipment operating the
separation process. An advantage of this simplified approach is that the system of
equations is now decoupled. There is no more interaction between the momentum and
continuity equations of the two phases, but there is a single-phase system of equation
together with a decoupled transport equation which exploits the convective term. This
involves a dramatic reduction of the computational effort. The applied numerical schemes
and the solution algorithm have been tuned in order to grant the highest accuracy and
stability of the solver. Finally, reported below the momentum transfer equation:
(4)

2. Two-Phase approach.
The Two-Fluid model has been adopted in this procedure in order to simulate with the best
accuracy the two-phase dynamic. This methodology requires less computational efforts
than the Direct Numerical Simulation (DNS) and it is more suitable for every flow regime
than Dispersed Phase Element (DPE), thanks to its two-way coupling granted by the
Eulerian-Eularian approach. The governing equations are written in the averaged form for
both the fluids, considering each phase as continuous, allowing in this way the

4
 interpenetration of one phase into the other one. The momentum transfer between the two
different phases is taken into account through the term Mφ that considers the forces acting
on the interface between the fluids: drag, lift, virtual mass and turbulent effect. For each
phase it is possible to write the mass and momentum conservation respectively:

(5)

(6)

Where Rφ is the combined turbulent and viscous stress Reynolds number, Mφ is the
averaged interphase momentum transfer term. For the clarity sake, it is reported the
volume phase fraction αφ, defined as:

(7)

Combining the two continuity equations for the two phases φ1 and φ2, it rearrange the
continuity equation in order to obtain a pressure equation to be solved. Regarding the
interphase momentum, the lift contribution has been neglected on droplets. The turbulence
has been described by the standard κ-ε model, which is suitable for the flow regime
conditions under study. A source term has been included to incorporate the dispersed
phase on turbulence. This approach can provide the best approximation of a real two-
phase physical phenomenon such as the real case considered in the current paper.

The final procedure developed is able to combine the computational lightness achieved with the
single-phase solver and the high level of accuracy typical of the two-phase application. In this way,
the perfect trade-off, between precision and reduction of required time, is achieved, making
feasible the application of the CFD approach in the Oil&Gas industry.

Lagrangian-Eulerian multiphase modeling

The Lagrangian discrete phase model in Fluent follows the Euler-Lagrange consolidated approach.
The fluid phase is treated as a continuum by solving the time-averaged Navier-Stokes equations,
while the dispersed phase is solved by tracking a large number of particles, bubbles, or droplets
through the calculated flow field. The dispersed phase can exchange momentum, mass, and
energy with the fluid phase. A fundamental assumption made in this model is that the dispersed
second phase occupies a low volume fraction, even though high mass loading (mparticle ≥ mfluid) is
acceptable. The particle or droplet trajectories are computed individually at specified intervals
during the fluid phase calculation. This makes the model appropriate for the modeling of spray
dryers, coal and liquid fuel combustion, and some particle-laden flows, but inappropriate for the
modeling of liquid-liquid mixtures, fluidized beds, or any application where the volume fraction of
the second phase is not negligible.
Fluent predicts the trajectory of a discrete phase particle (or droplet or bubble) by integrating the
force balance on the particle, written in a Lagrangian reference frame. This force balance equates
the particle inertia with the forces acting on the particle, and can be written (for the direction in
Cartesian coordinates) as:

(8)

Where Fx is an additional acceleration (force/unit particle mass) term and takes into consideration
virtual mass and the pressure gradient, FD(u- up) is the drag force per unit particle mass and finally
FD can be written as follow:

(9)

5
The drag model used is the k-ε, implementing the Schiller-Naumann drag coefficient. The lift effect
has been neglected due to the large dimension and shape characteristic of the liquid droplet used
for simulation. The continuous phase flow is modelled using the Eulerian framework, which has
been briefly presented previously, and it is coupled with the particle transport equation via the
called two-way coupling.

Droplet size modeling

The dimension of the liquid droplet assumes a central role in this particular case study, due to the
significant influence of droplet diameter on the erosion evaluation model. The Webber model has
been chosen, in order to model the liquid particle size. This model is able to capture quite
accurately the relationship between turbulent and surface forces, which act in the mechanism of
continuous coalescence and break-up of the droplet inside the multiphase flow (Figure 1).

Figure 1: Mechanism of droplet coalescence due to surface tension and ad break-up due to
turbulence
Below the equations taken into consideration:

(10)

(11)

(12)

where (13)

Where We is the Webber number, ε is the dissipation energy per mass unit and u 2 is the turbulent
velocity fluctuation term. Using conveniently the equations 10, 11, 12 and 13 allows defining two
different droplet diameters, dependent on the operating conditions selected. In particular, in this
case the droplet dimensions are deeply influenced by the different pipeline diameters considered in
the simulations. An interpolation function was created between the two main diameter calculated,
in order to reduce the computational effort, related to the implementation of a new equation for the
droplet diameter inside the calculation loop, maintaining at the same time a high level of accuracy.

6
The interpolation function adopted has been generated considering the physical behaviour of the
growth and break-up of liquid particle. Figure 2 shows the mechanism previously described.

Figure 2: Interpolation function for growth and break-up of liquid droplet

Erosion CFD model Applied

Few equations able to predict liquid droplet erosion have been developed nowadays, although
there are many equations to evaluate solid particle erosion in literature. The erosion equations are
generally based on: particle impact speed, angle and particle/material properties. The model
implemented in this work is based on the consolidated on solid particle erosion model proposed by
Edwards [9], and then tuned with parameters that take into account the liquid/gas erosion
mechanism.
The model used for erosion calculation due to liquid particle impact on pipe wall has been
rearranged from the model already implemented on Ansys Fluent for solid particle erosion, using a
Lagrangian-Eulerian approach. This formulation has been implemented also within the Eulerian-
Eulerian application previously presented. In particular, the particle erosion has to be monitored at
wall boundaries. The erosion rate is defined using the Edwards model as follow:

(14)

Where N is the number of particles that impacts onto the wall, C(dp) is a function of particle
diameter and impinged wall density, is the impact angle of the particle path with the wall
face, f( ) is a function of impact angle, v is the relative particle velocity, b(v) is a function of
relative particle velocity, and Aface is the area of the cell face at the wall. Appropriate values must
be specified for a specific particle eroding a defined material. The erosion rate as calculated above
is displayed in units of removed material per area and time. In this case, proper values for the
variables described above, proposed by R. Li et al. [7, 8], have been implemented in order to
simulate erosion due to liquid droplets impingement in spite of solid particle impact.
In the Edward’s model the sand particles are assumed as impinged materials. However, in case of
liquid droplets due to their characteristics, they behave in a different way compared to the solid
particles, thus some reasonable modification should be done. The solid particle (sand)
impingement angle function proposed is written as follows:
(15)
Unlike solid particle erosion, the scratch stress caused by tangential momentum might be
negligibly small, in case of liquid particle erosion. Therefore, the scratch stress is much less
important due to the liquid mobility and deformation. The normal component of impact velocity is a
major factor for the liquid droplet erosion damage whilst the tangential component has little effect.

7
Li and Ninokata [7, 8], using the volume of fluid (VOF) computational approach, investigated the
effect of droplet impingement angle on liquid droplets erosion. Based on this work available in
literature, the impingement angle function can be tuned and fitted with a trend line equation
(eq.16), which is also implemented into both codes to support the calculation of the erosion rate.
(16)

CASE STUDY

The debottlenecking project applied to the gas and condensate field under study have introduced
the need to analyse more in detail the off-shore platform fluid dynamic, considering the new
operating conditions and the increased production of fluids from the wells. A multiphase CFD study
has been set up in order to analyse the fluid-dynamic inside a real the pipeline connecting the
wellhead manifold to the sealine manifold. In particular, the attention has been focused on the
erosion effects caused by the liquid droplets transported by the multiphase flow coming from the
wells and impacting on the walls. The aim of this study is the verification of the possible insurgence
of erosion phenomena within this offshore structure, in order to ensure the achievement of the
optimum production, maintaining the safety of the exploitation operations.

CAD model and meshing strategy

Errore. L'origine riferimento non è stata trovata. shows the 3D model generated with strict
bend, that represents the high erosion risk segment of the pipe.

Figure 3: 3D CAD model considering strict bend

8
In order to reduce the computational effort related to both CFD approaches presented, the most
critical section of the piping system for erosion problem has been identified and considered. In
particular, the attention has been focused on the low-pressure zone, with higher velocity and
geometry variation due to the presence of “tee” junction. Hence, the 3D model built, represents
pipeline segment just before the flow-meter, and includes: one “tee” junction 8” pipe, one diameter
reduction from 8” to 4” and one “tee” junction 4” pipe. Also a different geometrical configuration
with 90° elbow has been taken into consideration.
The mesh has been generated using Gambit, which is compatible for both software used in the
current work for CFD analysis. This approach allows to compare more in detail the results obtained
by the different CFD frameworks.

Boundary and initial conditions

Two different operating conditions scenario have been analysed. The main data are reported in
Table 1 and Table 2 respectively.

Table 1: Production scenario 1, main operating data

Gas Condensate
Qgas ρg_ac ρg_st Qliq ρl_ac ρl_st µl σ
µg [cp] Z
[MMScf/d] [kg/m3] [kg/sm3] [Stb/d] [kg/m3] [kg/sm3] [cp] [dyne/cm]
157 121 0.79 0.01792 0.78 5716 630 750 0.3762 10.12

Pin Pout Ddroplet_8” Ddroplet_4”

[psig] [psig] [µm] [µm]
2073.5 1812 287 113

Table 2: Production scenario 2, main operating data

Gas Condensate
Qgas ρg_ac ρg_st Qliq ρl_ac ρl_st µl σ
µg [cp] Z
[MMScf/d] [kg/m3] [kg/sm3] [Stb/d] [kg/m3] [kg/sm3] [cp] [dyne/cm]
210 135 0.79 0.01792 0.78 7646 636 750 0.3762 10.12

Pin Pout Ddroplet_8” Ddroplet_4”

[psig] [psig] [µm] [µm]
2291 1812 209 85

The boundary conditions type selected in both cases are tuned in order to maximize the CFD
application stability and accuracy. Note that the liquid particle size has been considered constant
within the pipeline segment with the same internal diameter. The liquid droplet distribution has
been considered completely dispersed within the continuous gas phase at the inlet of the
geometrical domain simulated. Finally, the liquid droplet, responsible for the erosion phenomenon
due to the impingement on the wall, have been considered captured by the wall itself after the
impact in both CFD approaches.

SIMULATION RESULTS ANALYSIS

The erosion effect is computed thanks to Edwards model in kg/(m2s). It is necessary to divide the
erosion rate obtained in this way by the density of the material that is exposed to the impingement
of liquid droplets, in order to obtain the erosion velocity in terms of mm/year, which is the unit of
measurement used in this work for the classification of the erosion intensity. Moreover, proper
coefficient able to convert the m/s to mm/year has been applied to the results of the erosion rate
equation (3.1536E10). The erosion phenomenon described thanks to both the CFD applications

9
considered seems to be highly variable along the spatial domain of the simulation. Hence, the
results have been normalized to the most frequent value along the 8” segment in the strict bend
model considering the actual operating condition of
the field:
 Eulerian-Eulerian approach: 6.1E-8 kg/(m2s), equal to 0.240 mm/y considering a stainless
steel density of 8000 kg/m3;
 Lagrangian-Eulerian approach: 5.7E-8 kg/(m2s), equal to 0.225 mm/y considering a
stainless steel density of 8000 kg/m3.

The normalization applied to the results allows reducing the propagation of the significant
uncertainty related to the erosion model. The results are analyzed and compared by the
normalization values in spite of the absolute one, maintaining a good level of conservativeness
inside the outcomes. It is possible to state that the velocity, pressure and erosion computed by the
different approaches are almost equal in term of distributions and values, highlighting the same
main concerns related to the different producing scenario. Regarding the dispersed phase fraction
distribution, the accordance seems to be respected in term of spatial positioning of the liquid
droplets inside the simulation domain. Otherwise, the absolute value of volumetric fraction of liquid
are significantly different for the two solvers applied. This is mainly due to the different capability of
the CFD methodologies considered to analyze the dispersed phase fluid dynamics. In particular,
the results obtained using the Lagrangian framework can be considered accurate in a smaller
range of low liquid fraction distribution compared to the Euler one. On the other hand, the Euler-
Lagrangian CFD approach allows to reduce significantly the computational effort required to
simulate, thanks to its consolidated and stable algorithm, included in Fluent.
Figure 4 and Figure 5 report the results obtained with Fluent (Lagrangian-Eulerian framework),
considering scenario 2 (maximum flow rate from the wells) and the strict bend geometry.

Figure 4: Scenario 2 operating conditions: Erosion rate distribution along the 8” segment
and the convergent junction for two different views a) and b).

10
 Figure 5: Scenario 2 operating conditions: Erosion rate distribution along the 4” segment
for two different views a) and b).

Figure 6 shows the regions mainly affected by the liquid erosion effect within the two segment: 8”
and 4”. In particular, it is possible to note that the convergent junction and the external section of
the 4” bend are the most critical parts that could be interested by the erosion. Both the CFD tools
have pointed out this situation. The maximum level of erosion rate identified are in both cases
respectively 4 and 5 times the values chosen for the normalization of the results, previously
reported. This outcome has been obtained with both the CFD procedure applied.

Figure 6: The regions mainly affected by the erosion have been highlighted for: a) 8”
segment and b) 4” segment. Fluent results

The velocity vector distribution of the continuous gas phase have been reported in Figure 7 and
Figure 8. As it is possible to see, the highest module of the vector velocity are achieved within the
zone most interested in erosion effect.

11
 Figure 7: The velocity vector distribution for Scenario 2 obtained. The scale is related to
colour (module [m/s]) and no to the dimension of the vector. a) 8” bend and b) convergent
conical junction

Figure 8: The velocity vector distribution for Scenario 2. Section view a) out of 8” bend, b)
inlet of conical junction and c) out of 4”

Analyzing more in detail the results obtained, the velocity distribution of gas phase is similar for
both CFD approaches, scaled for different production scenario. The bends included in the spatial
domain generate a spread recirculation zone, which interests also the convergent junction in the 8”
segment and is widespread up to 10 hydraulic diameter within the 4” segment. In fact, the
convergent junction is able to handle effectively the multiphase flow, generating a uniform velocity
distribution at its outlet.
Comparing the results obtained for the strict bend geometry and the 90° elbow 3D model, the
erosion rate is much higher in the second one. For every geometry and CFD approach used, the 4”
segment is the most critical zone for erosion phenomena. On the other hand, also within the 8”
segment just after the bend the erosion effect has to be considered carefully. The highest peaks of
erosion are achieved in the lateral and external part of the bends, where the gas recirculation
impose a rigid liquid droplets impingement on the wall of the pipe. As stated previously, the
convergent conical junction is highly affected by erosion, in particular, in the opposite zone of the
upstream 8” bend.
Both CFD approaches highlight that Scenario 1 involves a higher erosion rate compared to
Scenario 2 (maximum flow rate) within the 8” segment. This is mainly related to the larger droplets
diameter in case 1 (286 µm) compared to case 2. Hence, liquid particles brings a higher inertial
force and relative velocity, which induced a higher number of impacts on the wall and consequently
a higher erosion rate.

12
CONCLUSIONS

The work should be divided into two different main sections:

 The implementation of a complete model of liquid particle erosion evaluation within a
Computational Fluid Dynamic procedure. The liquid droplet erosion has been simulated
using the solid particle framework, opportunely tuned to take into account the liquid physical
characteristics. Two different approaches to the multiphase flow CFD modeling have been
selected: the Eulerian-Eulerian framework and the Eulerian-Lagrangian one. In the first
case an in-house procedure, previously developed by the authors have been selected,
using OpenFOAM as software. The implementation and discretization of the liquid erosion
model has been performed maintaining the solver stability and efficiency, without an
increase in the total simulation time. The Dispersed Phase Element (Lagrangian-Eulerian)
approach used is already implemented and consolidated in Ansys Fluent, while the erosion
model has been included considering the complete absorption of liquid droplet at the body
wall after the impact;
 The application of the two different approaches developed to a real case study. A
gas/condensate field has been considered within a wide debottlenecking project. The
increase of fluids production leads to the possible insurgence of liquid erosion inside the
pipelie between the wellhead manifold and the sealine manifold. This possibility have been
evidenced by the traditional industrial standards, without specific results on the erosion rate
distribution along the pipeline. Both CFD methodologies allow to analyze the 3D multiphase
fluid dynamics gaining a deep insight into the liquid droplets erosion mechanism.
The results given by the CFD approaches highlight the possible insurgence of liquid particle
erosion in a specific segment of the pipeline analyzed. The Eulerian-Eulerian and the Lagrangian-
Eulerina framework shows a high level of accordance in term of velocity, pressure and erosion rate
distribution along the spatial domain. The main difference between the solvers used, is related to
the dispersed phase distribution numerical values and on the computational effort required. In fact,
the Eulerian-Eulerian shows a higher accuracy in terms of dispersed phase distribution evaluation,
in particular in a wider range of liquid phase fraction. On the other hand, it requires a larger time of
computation. The CFD approaches have allowed evaluating the operating conditions provided by
the debottlenecking project, avoiding future loss of production and safety problems.
Further development will be carried on with a focus on the tuning parameters to match the liquid
erosion model with the real system physical behavior.

REFERENCES

[1] OpenCFD Ltd., “OpenFOAM Programmer’s Guide”. Online available: http://www.openfoam.org

[2] OpenCFD Ltd., “OpenFOAM User’s Guide”. Online available: http://www.openfoam.org
[3] H.Rusche, “Computational Fluid Dynamics of Dispersed Two-Phase Flows at High Phase
Fractions”, Imperial College of London, London, 2002
[4] Malalasekera, H.K. Versteeg and W., An introduction to computational fluid dynamics,
PEARSON Prentice Hall, 2007
[5] J.H.Ferziger, M. Peric, “Computational Methods for Fluid Dynamics”, Springer, third edition
[6] L. Cadei, G. Montenegro et al., “Slug Catcher Multiphase CFD modeling: Optimization and
Comparison with Industrial Standards”, Journal of Applied Fluid Mechanics, Vol. 9, Special Issue 1,
pp. 1-9, 2016.
[7] Li, R., Ninokata, H., 2011. “A numerical study of impact force caused by liquid droplet
impingement onto a rigid wall.” Progress in Nuclear Energy 53, 881– 885.
[8] Li, R., Ninokata, H., M. Mori, 2012. “A calculation methodology proposed for liquid droplet
impingement erosion.” Nuclear Engineering and desingn 242, 157– 163.

13
[9] Edwards, J.K., McLaury, B.S., Shirazi, S.A., 2000. “Evaluation of alternative pipe bend fittings in
erosive service.” In: Proceedings of ASME FEDSM‘00: ASME 2000 Fluids Engineering Division
Summer Meeting, Boston, June 2000.
[10] FLUENT Version 12.0, Theory Guide. ANSYS Group.
[11] Springer, G.S., 1976. “Erosion by Liquid Impact.” John Wiley and Sons, New York, NY.
[12] H. Arabnejad, S.A. Shirazi, B.S. McLaury and J.R. Shadley, “A guideline to Calculate
Erosional Velocity due to Liquid Droplets for Oil and Gas Industry”, SPE-170951-MS
[13] BP, “Energy Outlook 2035”, 2016. Online available: http://www.bp.com
[14] L. Cadei, S. Ritondale, P. Ciccorossi, A. Bianco, S. Masi, 2016, “cfd multiphase tool: new
application on a real debottlenecking (DBN) project”, OMC-2017

14