Scipedia

Carneiro, F.

Bastos, L.

Teixeira, S.

Teixeira, J.

Study of hemodynamic in the iliac bifurcation through Fluid-Structure Interaction simulations

Rosic, B.

Goedde, T.

Towards domain decomposition of large neural networks

PETER, J.

LIPPI, M.

Comparison of data -driven methods for the prediction of flows about a wing

Soppi, R.

Basile, F.

Péron, S.

Lagha, M.

Hammani, I.

Local Immersed Boundary Method on body-conformal grids for the capture of geometrical features in aerodynamic simulations

Dell'Agnello, F.

Vignali, E.

Capellini, K.

Scarpolini, M.

Gasparotti, E.

Cademartiri, F.

Celi, S.

A Dynamic CT Based Pipeline To Assess Hemodynamic Indexes and Wall Stiffness of the Aorta

Pozzi, M.

Cominelli, S.

Braghin, F.

Design of an Acoustic Luneburg Lens via Topology Optimization of Pentamode Lattice

Marussig, B.

Rüberg, T.

Zechner, J.

Kielhorn, L.

Fries, T.

Reducing Meshing Requirements for Electrostatic Problems using a Galerkin Boundary Element Method

Bahramian, L.

Amani, A.

Rigola, J.

Oliet, C.

Pérez-Segarra, C.

The 9th European Congress on Computational Methods in Applied Sciences and Engineering
ECCOMAS Congress 2024
3–7 June 2024, Lisboa, Portugal
NUMERICAL STUDY OF AN INERTIAL PARTICLE
SEPARATOR EFFICIENCY UNDER THE EFFECT OF ICE
ACCRETION CAUSED BY THE WATER IMPINGEMENT
LINDA BAHRAMIAN, AHMAD AMANI, JOAQUIM RIGOLA, CARLES
OLIET AND CARLOS D. PÉREZ-SEGARRA
Heat and Mass Transfer Technological Center (CTTC)
Universidad Politécnica de Catalunya – Barcelona Tech (UPC)
Carrer de Colom 11, 08222 Terrassa (Barcelona), Spain
e-mail: {linda.bahramian, ahmad.amani, joaquim.rigola, carles.oliet,
cdavid.perez.segarra}@upc.edu, www.cttc.upc.edu
Key words: Inertial Particle Separator, Ice Accretion, Water Impingement, Computational
Fluid Dynamics, Numerical Simulation
Summary. The inertial particle separator (IPS) can face ice accretion due to the impingement
of super-cooled water droplets on its inner surface. This work studies a numerical study using
Computational Fluid Dynamics (CFD) for the ice accretion inside an IPS. This study is crucial
in investigating the IPS performance that is affected by ice accretion. First, the numerical
simulation of ice accretion for a simpler geometry, a cylinder, is validated with the experimental
data. Then, the simulation of water impingement and ice accretion is studied for an IPS device.
The ice profiles for different wall sections of the IPS show that ice accumulates on the bent
surface of the shroud, the windward side of the hub, and the leading edge of the splitter. This
accumulated ice changes the geometry of the IPS. Therefore, the separation efficiency of the IPS
is expected to decrease due to the generation of ice in the way of the scavenge outlet.
1 INTRODUCTION
An Inertial particle separator (IPS) can be situated in the aircraft engine’s intake to prevent
the Foreign Object Debris (FOD) from entering the compressor. An IPS consists of two types
of outlets: core and scavenge. The goal is to navigate particles into the scavenge outlet, which
is implemented by a rapid change in flow direction. Therefore, particles separate from the core
outlet by their inertia. IPS performance is measured by the separation efficiency, calculated as
the mass of particles leaving the scavenge outlet over the total mass of particles leaving both
outlets.
Connolly et al. [1] present experimental data for a wide range of flow path designs and test
dust for an IPS device. The geometry was proposed first by Barone et al. [2]. These data can
be used for validation purposes. In the work of Bahramian et al. [3], the validation of numerical
simulation of an IPS device with the experimental benchmark case of Barone et al. [2] is carried
out, and the separation efficiency for spherical particles is examined with the experimental data.
In another work, Bahramian et al. [4] have investigated the effect of different Reynolds numbers
and scavenge mass flow fractions on the separation efficiency of an IPS device through different
turbulent numerical simulations.
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
An IPS also can face the problem of icing. In cold weather, super-cooled water droplets in
the air may impact the inner surface of the IPS and cause icing, which significantly influences
the IPS performance. Therefore, in order to design an effective anti-icing system, understanding
the ice accretion process within the IPS is crucial. The effects of the mean droplet diameter
(MVD), liquid water content (LWC), and incoming velocity on ice accretion of an IPS device of
a helicopter are studied by Qiu et al. [5]. As was shown, the ice thickness on the wall surfaces
changes according to the different MVDs, LWCs, and inlet velocities.
2 MATHEMATICAL MODEL
The Navier-Stokes equations represent the equations of a viscous compressible continuous
fluid. To implement the Navier-Stokes equations in Large Eddy Simulation (LES) and Reynolds
Averaged Navier-Stokes (RANS) turbulence methods, they can reformulated in the filtered form.
In LES, the focus is on resolving the larger scale flow characteristics while the subgrid scales
(SGS) are modeled. The scale separation is achieved by applying a low-pass filter to the transport
equations as detailed in the work of Pope [6]. The continuity, momentum, and energy equations,
which govern the flow of a compressible Newtonian fluid, are detailed in the work of Versteeg et
al. [7] as follows:
∂ρ
∂t
+∇ · (ρu) = 0 (1)
∂(ρu)
∂t
+∇ · (ρuu) = −∇p+∇ · (µ∇u) + Su (2)
∂(ρE)
∂t
+∇ · (ρuH) = −∇ · (pu) +∇ · (k∇T ) + SE (3)
where ρ, u and p are the density, the velocity vector, and the pressure of the fluid. µ, Su are
the dynamic viscosity and the momentum source term. In the energy equation, E, H, k, T
and SE represent the total energy per unit mass, the total enthalpy per unit mass, the thermal
conductivity, the temperature, and the energy source term, respectively.
As described in ANSYS Fluent, the Eulerian formulation is employed to simulate the im-
pingement of water droplets. The general Eulerian-Eulerian model consists of the Euler or
Navier-Stokes equations with additional terms accounting for the droplets:
∂α
∂t
+
−→
∇ .
(
α
−→
V d
)
= 0 (4)
∂
(
α
−→
V d
)
∂t
+
−→
∇
[
α
−→
V d ⊗
−→
V d
]
=
CDRed
24K
α
(−→
V a −
−→
V d
)
+ α
(
1− ρa
ρd
)
1
Fr2
(5)
where α and Vd are mean-field values of the droplet concentration and velocity, respectively. CD,
Re, andK are the drag coefficient, Reynolds number, and droplet inertia parameter, respectively.
The subscripts a stand for the air and d for the droplet, and the Fr represents the Froude number.
The first term on the right-hand side of the momentum equation accounts for the drag force
acting on droplets with a mean diameter of d. The second term accounts for buoyancy and
gravity forces and is proportional to the local Froude number.
2
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
Besides continuity and momentum conservation equations for droplets, the thermal energy
equation can also be considered, which enables heat exchange between droplets, airflow, and
water vapor systems.
For icing, ANSYS Fluent solves a system of two partial differential equations on all solid
surfaces. The first equation represents mass conservation:
ρf
[
∂hf
∂t
+
−→
∇ .
(−→
V fhf
)]
= V∞LWCβ − ṁevap − ṁice (6)
where hf and Vf are the height and velocity of the water film. The three terms on the right-
hand side account for the mass transfer by water droplet impingement, the mass transfer by
evaporation, and ice accretion.
The second partial differential equation is the conservation of energy:
ρf
[
∂hfcfTf
∂t
+
−→
∇ .
(−→
V fhfcfTf
)]
=
(
ed +
1
2
−→
V d
2
)
V∞LWCβ − ṁevap (Levap)
+ṁice (Lfus − ciceTice)− ch (Tf − Trec) + σε
(
T 4∞ − T 4f
)
+Qanti−icing
(7)
where the first three terms represent the energy transfer by the impinging super-cooled water
droplets, evaporation, and ice accretion, and the last three terms correspond, respectively, to
the radiative, convective, and 1D conductive heat fluxes that could be applied by an anti-icing
heater device.
The coefficients ρf , cf , Levap, cice, Lfus, σ, ε are physical properties of water and ice that are
adjustable in the user interface. The reference conditions T∞, V∞, LWC are airflow and droplet
parameters of the freestream.
Droplet simulation provides local values of the collection efficiency β, droplet energy ed, and
droplet impact velocity Vd. The local water collection efficiency can be determined using:
β = − α
−→
Vd.
−→n
(LWC∞)U∞
(8)
where α is the local liquid water content;
−→
Vd and
−→n are the droplet impact velocity and the
surface normal vector; and LWC∞ and U∞ are the freestream liquid water content and velocity
of the droplets, respectively.
The evaporative mass flux mevap is recovered from the convective heat transfer coefficient
using a parametric model. Qanti−icing is the anti-icing heat flux for wet air calculations, hf
and Tf are the film thickness and the equilibrium temperature at the air/water film/ice/wall
interface, and mice is the instantaneous mass accumulation of ice.
3 RESULTS AND DISCUSSIONS
In this section, first, a validation for water impingement and ice accretion is implemented,
and in the following, the ice accretion is studied in an IPS device. For water impingement, the
Eulerian-Eulerian approach is implemented. The numerical simulations were carried out using
the turbulence method of RANS by the Spalart-Allmaras model in ANSYS Fluent.
3
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
3.1 Ice accretion validation
For the validation purpose, a simple case of a cylinder with a diameter of 38.1mm is selected.
The experimental results for this case are presented by Koss [8]. A mesh sensitivity analysis is
performed, and as shown in figure 1, a structured mesh of around 200K control volumes (CVs) is
implemented. The mesh is refined adequately around the cylinder to maintain better precision
in the results. The Reynolds number of inlet flow is in the range of 2 million. The ice condition
is listed in the table 1.
Table 1: Simulation conditions for ice accretion on the cylinder
U∞(m/s) T (K) LWC(g/m
3) MVD (µm) t(s)
30 269.15 0.4 50 600
Figure 2 shows the water collection coefficient on the cylinder. Due to the high inertia of the
water droplets, it is challenging to change their trajectory, resulting in most droplets impacting
the leading edge of the cylinder, where β becomes significantly large. Figure 3 shows the ice
profile comparison of the numerical simulation with the experimental data, which presents a
good agreement between them. The results demonstrate the accuracy and capability of the
current model, which can be implemented for more complex geometry.
Figure 1: Structured mesh used for the simulation of ice accretion on the cylinder.
3.2 IPS water impingement and ice accretion
For IPS, the OSG-1 geometry of Barone et al. [2] with the scavenge mass flow fraction of 0.16
is selected. The width of the geometry is set to 0.2m. Figure 4 represents a general geometric
model of an IPS, showing the inlet, scavenge and core outlets, and three different walls: shroud,
hub, and splitter. As shown in figure 5, a structured mesh refined especially close to the walls and
in the curvatures is used. The simulation condition is listed in table 2. The Reynolds number
of the IPS based on the characteristic length of 0.068m is around 500K. The inlet pressure is
set to atmospheric pressure (101.325 kPa), while the mass flow rates are specified for both the
core and scavenge outlets.
Figure 6 shows the local water collection coefficient distribution, β, for three different walls
of the IPS: shroud, hub, and splitter. The reason behind the water accumulation in the bent
4
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
Figure 2: Contour of the water collection coef-
ficient on the cylinder.
-0.6
-0.4
-0.2
 0
 0.2
 0.4
 0.6
-0.8 -0.6 -0.4 -0.2  0  0.2  0.4  0.6
Y
/D
X/D
Numerical
Clean Cylinder
Experiment
Figure 3: Comparison of the accumulated ice
profile with the experimental data.
Figure 4: Geometric model of the IPS [5].
Table 2: Simulation conditions for ice accretion on the IPS
U∞(m/s) T (K) LWC(g/m
3) MVD (µm) t(s)
90 253.15 2 20 360
surface of the shroud is as super-cooled water droplets enter the IPS, they gradually accelerate to
their peak velocity before passing through the throat, where the curvature is most pronounced.
Due to their high inertia, these high-speed droplets cannot change direction as quickly as the
surrounding airflow. Consequently, they deviate from the air stream and impact the wall inde-
pendently. As a result, the IPS effectively reduces the amount of water entering the core flow
path by capturing it on the inner walls.
The ice profiles for these three walls of the IPS are shown in figure 7 within an icing time of
6 min. The ice generated on these surfaces will change the internal geometry of the IPS, and as
5
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
Figure 5: Structured mesh generated for the simulation of the IPS.
 0
 0.1
 0.2
 0.3
 0.4
 0.5
-0.3 -0.2 -0.1  0  0.1  0.2  0.3  0.4  0.5  0.6
b
X (m)
Shroud
Hub
Splitter
Figure 6: Local water collection coefficient distribution on different walls of the IPS.
a result, the ability of the IPS to navigate the FOD to the scavenge outlet will change. The wall
collision coefficient will change accordingly as the ice material differs from the walls. Due to the
accumulated ice in the shroud, the efficiency of the IPS is expected to decrease. However, the
numerical value of separation efficiency is under implementation for a comprehensive conclusion
of this phenomenon.
4 CONCLUSIONS
This work investigated the numerical simulation of water impingement and ice accretion inside
an IPS device. The 3D RANS turbulence model is implemented through ANSYS Fluent software
for the numerical simulation. The benchmark case of Koss [8] was chosen as a validation test
case of a cylinder for the ice accretion numerical simulation. Numerical results on the ice profile
of the cylinder benchmark case were in good agreement with the experimental data. The results
of the water impingement and the ice accretion of the IPS indicate the critical area to consider
for the separation efficiency. As the accumulated ice in the shroud wall is narrowing the scavenge
outlet entrance, it prevents the particles from erasing correctly from the scavenge outlet, and as
a result, the separation efficiency of the IPS will decrease. The effect of LWC, MVD, incoming
6
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
-0.1
 0
 0.1
 0.2
-0.4 -0.2  0  0.2  0.4  0.6
Y
 (
m
)
X (m)
Ice Surface
Wall of IPS
Figure 7: 2D ice shape on different walls of the IPS.
velocity, and temperature on ice accretion inside the IPS is under implementation. Further
analyses of the effect of ice accretion on the airflow field and separation efficiency value of the
IPS will be investigated.
ACKNOWLEDGEMENTS
Linda Bahramian acknowledges the financial support from the Secretariat of Universities and
Research of the Generalitat de Catalunya and the European Social Fund, FI AGAUR Grant
(2019 FI B 01205), and the UPC-Santander Grant. Carles Oliet, as a Serra Húnter associate
professor, acknowledges the Catalan Government for the support through this Programme.
This project has received funding from the Clean Sky 2 Joint Undertaking under the European
Union’s Horizon 2020 research and innovation programme under grant agreement No 821352
(ANTIFOD project).
Disclaimer: The contents presented in this article reflect only the author’s point of view: Clean
Sky JU are not responsible for any use that may be made of the information it contains.
REFERENCES
[1] Connolly B.J., Loth E., and Smith C.F. 2022. “Unsteady separated flows in an S-Duct and
a bifurcating duct.” J. Aircr. 59(1): 47-57. https://doi.org/10.2514/1.C036383
[2] Barone D., Loth E., and Snyder P. 2017. “Influence of particle size
on inertial particle separator efficiency.” Powder Technol. 318: 177-85.
https://doi.org/10.1016/j.powtec.2017.04.044
7
Linda Bahramian, Ahmad Amani, Joaquim Rigola, Carles Oliet and Carlos D. Pérez-Segarra
[3] Bahramian, L., Amani, A., Rigola, J., Oliet, C., and Pérez-Segarra, C.D. 2024, May. “On
the validation of numerical simulation with experimental results on compressible turbulent
flow in an inertial particle separator device.” In Journal of Physics: Conference Series, vol.
2766, no. 1: 012113. IOP Publishing. http://doi.org/10.1088/1742-6596/2766/1/012113
[4] Bahramian, L., Amani, A., Rigola, J., Oliet, C., and Pérez-Segarra, C.D. 2023. “A numer-
ical study on turbulent flow in an inertial particle separator device.” 10th Int. Symp. on
Turbulence, Heat and Mass Transfer (Rome) ICHMT Digital Library Online (New York:
Begel House Inc.) http://doi.org/10.1615/ICHMT.THMT-23.830
[5] Qiu, C., Chen, N., Hu, Y., Wang, B., and Hu, S. 2022. “Numerical study
of ice accretion inside an inertial particle separator.” AIP Advances 12, no. 5.
https://doi.org/10.1063/5.0093145
[6] Pope S.B. 2001. Turbulent Flows, Measurement Science and Technology vol 12 no. 11:
2020-1
[7] Versteeg H.K., and Malalasekera W. 2007. An Introduction to Computational Fluid Dy-
namics: the Finite Volume Method, (Pearson education)
[8] Koss, H.H., Gjelstrup, H., and Georgakis, C.T. 2012. “Experimental study of ice accretion
on circular cylinders at moderate low temperatures.” Journal of Wind Engineering and
Industrial Aerodynamics, 104: 540-546. https://doi.org/10.1016/j.jweia.2012.03.024
8


Numerical Study of an Inertial Particle Separator Efficiency Under the Effect of Ice Accretion Caused by the Water Impingement

Ribeiro, L.

Beck, A.

Parisi, F.

Kroetz, H.

Comparison of the optimal risk-based design of planar RC frames with different aspect-ratios under progressive collapse

Barbosa, I.

Mendonça, P.

Maduro, A.

Injury Risk Assessment of Microtome Manipulation Using Kinematic Analysis

Not a member yet