Journal of Natural Gas Science and Engineering 28 (2016) 241e253

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

A One-D approach for modeling transport and deposition of Black

Powder particles in gas network
A. Filali, L. Khezzar*, M. Alshehhi, N. Kharoua
Department of Mechanical Engineering, Petroleum Institute, Abu Dhabi, United Arab Emirates

a r t i c l e i n f o a b s t r a c t

Article history: Black Powder (BP) is a universal issue in sales gas transmission pipelines. It can cause serious problems in
Received 5 May 2015 pipelines operations and instruments and contaminate customer supply. The objective of the present
Received in revised form study is to develop a novel methodology for tracking the dispersion of Black Powder within gas trans-
15 November 2015
mission pipelines using a 1D approach based on the dusty gas assumption and the usage of analytical
Accepted 16 November 2015
solutions of one-dimensional scalar advection/reaction equation. The study takes into account the
Available online 23 November 2015
deposition of Black Powder particles under different flow conditions, different particle diameters and
different surface roughness. The proposed approach is applied to particle-laden flow in pipe segments
Keywords:
Black Powder
with and without junctions and contrasted against CFD simulations based on the Discrete Phase Model
Dusty gas (DPM). The results show that the finer particles with diameters dp < 1 (mm) can be transported easily to
Deposition rate models the downstream of the tree shaped network, while the larger particles dp > 1 (mm) are likely to settle
Gas transmission network rapidly near to the source location where Black Powder is generated and, consequently, forming beds. It
is shown also that surface roughness increases the deposition rate of small particles dp
1 (mm)
controlled by the Brownian forces. However, the deposition of larger particles in the inertial regime is not
affected by the change in the surface roughness. These reported results are of significant practical in-
terest, due to the lack of available data of Black Powder concentration in natural gas networks that have
been reported in the literature to date.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
The transport of solid particles by turbulent flow occurs in a
variety of natural and industrial processes. Among the later, the
movement of solid particles such as Black Powder (BP) is becoming
a prominent problem in major gas transmission networks over
several countries around the globe (Baldwin, 1998; Tsochatzidis,
2008). The presence of BP in Gas pipelines has presented a range
of challenges to the gas industry and can have a large number of
detrimental effects in pipeline operations such as reducing flow
efficiencies, fouling of compressors, blocking of orifice meters,
contamination of instrumentation and control valves.
Evidence collected so far, points to the fact that BP particles are
generated when hydrogen sulfide (H2S), carbon dioxide (CO2) or
oxygen (O2) are present in the gas or by bacterial corrosion of the
internal walls of the pipelines (Sherik, 2007, 2008; Sherik et al.,
2008; Cattanach et al., 2010). BP is known to contain primarily
* Corresponding author.
E-mail address: lkhezzar@pi.ac.ae (L. Khezzar).
81% iron oxides (Godoy et al., 2005; Tsochatzidis and Maroulis,
2007) and, in other cases, a mixture of iron oxides and iron sul-
fides (Baldwin, 1998; Godoy et al., 2005; Sherik, 2007; Tsochatzidis
and Maroulis, 2007; Sherik, 2008; Sherik et al., 2008; Cattanach
et al., 2010).
The Black Powder deposits, which are mainly a set of very fine
particles, can be readily transported through a transmission pipe-
line system by the pressurized gas flow. It is therefore important to
be able to understand and predict the movement, deposition and
re-entrainment of BP in a pipeline. The present paper is an effort in
that direction which presents a novel methodology based on one-
dimensional approach to analyze the Black Powder movement
and shed some new light related to this topic.
Several theoretical and experimental research studies have been
directed at understanding BP movement in pipelines (Smart,
2007a, 2007b; Tsochatzidis and Maroulis, 2007; Smart and
Winters, 2008). All these works focused only on the notion of
critical velocity required to keep solid particles in movement. BP
movement in gas pipelines belongs to solid laden gas flows that are
encountered in several engineering applications. Several studies

http://dx.doi.org/10.1016/j.jngse.2015.11.034
1875-5100/© 2015 Elsevier B.V. All rights reserved.
242 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253
considered solideliquid flow transport in pipes such as the recent
study reported by Morten and Gustavo (2015). Extensive theoret-
ical, numerical and experimental studies were devoted to under-
standing diverse aspects of solidegas flows such as aerosol
dynamics, effect of the particle diameter, particle density, particle
volume fraction, gas velocity on the gasesolid flow (Boothroyd,
1971; Friedlander, 1977; Hetsroni, 1982; Elghobashi, 1991;
Hidayat and Rasmuson, 2005; Balachandar and Eaton, 2010;
Xinxiang et al., 2011; Tenneti and Subramaniam, 2013), pneu-
matic conveying in both horizontal and vertical pipes (Tsuji et al.,
1992; Mezhericher et al., 2011; Li et al., 2013, 2014), bends (Chu
and Yu, 2008; Elsaadawy and Sherik, 2012) and simulations
involving junctions (Li and Shen, 1996; Schneider et al., 2002) can
be found. The results of these studies were limited to very simple
systems and geometries which consisted of single or several pipes
of limited dimensions and were not conducted to simulate a long-
distance natural gas transmission network.
A full 3D computational fluid dynamics simulation (CFD)
approach for large transmission networks with long pipelines is
beyond present day CFD capabilities especially for multiphase
flows. The use of one-dimensional (1D) approach, where the sys-
tem geometry is formulated conceptually as a linear network of
segments or volume sections (Ben-Avraham and Havlin, 1982;
Sahimi et al., 1983; Bunde and Havlin, 1991; Rossman and Boulos,
1996; Makse et al., 2000; Fernandes and Karney, 2001;
Kirkpatrick et al., 2003; Andrei-Mugur and Sanda-Carmen, 2012;
Chahibi et al., 2013), presents a viable alternative. In this approach,
variation of flow parameters, such as concentration, occurs longi-
tudinally (along the pipeline length) as the gas/liquid is transported
out of one segment and into the next. The concentration of solid
particles within such a system is continually modified by the
physical processes of advection, dispersion and source or sinks.
These influence the concentration along the pipe whereby sources
represent addition, through mechanisms like pickup of particles
and sinks represent settling or deposition of particles on the pipe
wall. The limitation of the 1D approach is naturally the limited
information extracted compared with a full three-dimensional
simulation and the application of such an approach rests on the
assumption of a dusty gas behavior.
To model deposition and pick-up in the source term of the 1D
approach, usage is made of semi-empirical knowledge developed
so far on deposition and pick up rates. Previous relevant work on
deposition rate of solid particle in turbulent flow can be found in
Wood (1981); Papavergos and Hedley (1984); Fan and Ahmadi
(1993); Lai (2002); Bouilly et al. (2005); Guha (2008) whereas
that relevant to the prediction of the pickup rate from surfaces or
stationary beds can be found in Sehmel (1980); Reeks et al. (1988);
Nicholson (1988); Ziskind et al. (1995); Reeks and Hall (2001);
Alloul-Marmor (2002). In addition phase split at pipe junctions
remains an unknown in this approach and has to be modeled.
The present work attempts to model the behavior of BP move-
ment in a pipe network using a 1D approach based on the dusty gas
assumption and the usage of analytical solutions of steady
oneedimensional scalar advection/reaction equation. In the dusty
gas approach the gas-particles mixture is treated as a continuum
(Davies, 1966; Crowe, 1982; Van Genuchten et al., 1982) and thus
for solidegas flow through a straight pipe, the particles follow the
fluid (Balachandar and Eaton, 2010). For these reasons the flow can
be treated as a pseudo-single phase flow with the fluid density
depending on the local mass fraction of the particles. Such models
require definitions of the volume concentration and the bulk den-
sity of the two phases. The definition of the bulk density leads to
the notion of loading which is the ratio of the solid phase bulk
density and the f1uid-phase density. The loading is an important
parameter that characterizes the coupling between the phases. At
low loading, the coupling between particles and background phase
can he assumed to be one-way (Friedlander and Johnstone, 1957)
and particles can be modeled as tracers (Balachandar and Eaton,
2010). For BP, present available evidence suggests that when flow
is taking place, the volume fraction is less than 106 (Trifilieff and
Thomas, 2009) and hence the condition of dilute gasesolid flow
regime can be assumed.
The other important parameter is the Stokes number St
(Friedlander and Johnstone, 1957). Stokes number is defined as the
relation between the particle response time and the system
response time:
tp
St ¼ ; (1)
ts
rp d2p
wheretp ¼ 18mg , rp is the particle density, dp is the particle diam-
eter and mg is the gas dynamic viscosity. The system response time ts
is based on the characteristic length considered in the present
study as the pipe diameter D and the characteristic velocity
considered as the fluid average velocity inside the pipe U.
U
ts ¼ (2)
D
In the present cases of BP considered and believed to take place,
the typical Stokes number is considered to be less and/or near the
unity value which together with the low loading justifies the use of
the dusty gas approach.
The objective of the present work is to develop a novel meth-
odology for tracking the dispersion of Black Powder within gas
transmission pipelines taking into account its deposition within
smooth and rough surfaces, while pick up of particles may take
place it is ignored in this work. The analytical solution of the
advection-reaction equation and the approach is tested by
considering particle-laden flow in pipe segments with and without
junctions which were simulated using CFD based on the Discrete
Phase Model (DPM). Three-dimensional CFD simulations were
conducted as a support tool to the 1D approach developed. The 3D
simulations were limited to small segments of the piping network
and addressed the problem of phase split at junctions and particle
depositions in straight pipes.
2. Governing equations
In the present study, the flow is assumed to be one-dimensional,
along the x-direction, and to take place in a pipe of cross-sectional
area A. The one-dimensional advection-diffusion-reaction equation
can be written for the concentration C expressed in mass per unit
volume as:
vC vC v2 C
þU ¼ Ddiff 2 þ S (3)
vt vx vx
where Ddiff is the diffusion coefficient. For fully developed turbulent
2
flow, the axial diffusion term Ddiff vvxC2 is negligible compared to the
dominant advection term U vC vx and the radial diffusion which is
responsible for deposition or pickup of solid particle is lumped in
the source term S. The “source” or “sink” of the quantity C is
formulated as S ¼ ± b C.
In the case of settling of particles with a deposition rate bdep, S is
negative and means that a concentration quantity is subtracted
from the upstream concentration. Significant quantities of Black
Powder accumulate during long term (days, months and/or years).
The analytical solution of the Eq. (3) was derived first by Van-
Genuchten et al. (1982). However, for steady-state solution, Eq.
 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253 243

(3) can be simplified to the one-dimensional advection-reaction based on the eddy life time concept of the Fluent code.
equation and can be written as

vC 3. Deposition rate models

U ¼ bdep C (4)
vx
The deposition rate bdep, is defined as The modeling of the particle deposition rates on solid surfaces
has been the subject of several theoretical and experimental in-
4uþ u* vestigations. The published experimental studies (Wood, 1981;
bdep ¼ (5) Papavergos and Hedley, 1984; Fan and Ahmadi, 1993, Lai, 2002;
D
Bouilly et al., 2005; Guha, 2008), show that the measured deposi-
where D is the pipe diameter, uþ is the non-dimensional deposition tion rates vary with particle sizes in a similar trend; but they can
velocity calculated for each deposition model and u* is the friction differ an experiment to another due to the influence of several
velocity defined as parameters such as airflow pattern, turbulence level, and proper-
rffiffiffiffiffiffi ties of internal surfaces. The modeling of particle deposition is
tw
u* ¼ (6) usually presented by plotting non-dimensional deposition velocity
rg Vþ þ
d against dimensionless relaxation time tp . These are defined by

where tw is the wall shear stress and rg is the fluid density. Jw

Vdþ ¼ (12)
The steady-state solution of Eq. (4) can be obtained (Bear, 1972) rp;mean u*
and written as
  where Jw is the deposition mass flux, rp,mean is the mean particle
bdep $x density in the pipe. The non-dimensional particle relaxation time,
CðxÞ ¼ C0 exp (7)
U tþ
p , is defined as

For a tree shaped network which consists of several connected

tp u*2 Sratio d2p u*2
pipes having same or different diameters and lengths, the discrete tþ
p ¼ ¼ Cc (13)
form of equation (7) can be written as
Vg 18vg

  where ng is the gas kinematic viscosity, dp is the particle diameter,

bdep ðiÞ$xði; jÞ
C ði; jÞ ¼ C ði; 1Þexp (8) Sratio ¼ rp/rg is the particle-to-fluid density ratio, and the Cun-
U ðiÞ
ningham correction factor Cc is determined according to Eq. (14):
where j represents the location at each length step Dx of the pipe i
Cc ¼ 1 þ Kn½2:514 þ 0:8 expð  0:55=KnÞ (14)
where Dx is the distance between node j and jþ1, U(i) ¼ U is the
fluid average velocity assumed to be constant for each pipe i and in which the Knudsen number, Kn, is the quotient of mean free path
bdep(i) and C(i,1) are the deposition rate and inlet concentration of the carrier gas.
calculated in each pipe i. The wall shear stress has to be calculated using the Blasius
The CFD simulations, used for validation purposes here, were correlations for a fully developed turbulent flow in the pipe,
conducted using the Discrete Phase Model (DPM) subsequent to a (Schlichting, 1987). Reynolds number based on the hydraulic di-
single phase fully developed pipe flow simulation using the Rey- ameters Dh ¼ D is defined as
nolds Stress model. The CFD simulations started with an initial fluid
flow field calculated based on the assumption of fully developed rg UD
flow at the inlet of the computational domain. Once converged, the Re ¼ (15)
mg
resulting flow field was used as an initial background phase for the
injected particles. The DPM consists of solving the force balance where U is the mean pipe velocity and D is the pipe diameter.
equation for each particle. The shear velocity u* is obtained as a function of the friction
  factor f and the mean pipe velocity U (Pope, 2000) defined as
dup;i   gi rp  rg
¼ FD ug;i  up;i þ þ Fothers (9) rffiffiffi
dt rp u* f
¼ (16)
U 8
3 mg CD Rer
FD ¼ (10) where the friction factor f is calculated using an explicit equivalent
4 rp d2p
version of the classical implicit ColebrookeWhite equation
(Colebrook and White, 1937). Winning and Coole (2013) reviewed
where Rer is the relative Reynolds number, which is defined as
28 explicit equations for approximating the friction factor to inte-
  grate both the accuracy to the implicit ColebrookeWhite equation
rg dp up;i  ug 
Rer ¼ (11) and the relative computational efficiency of the explicit equations.
mg In the present study we have considered the more accurate cor-
The drag coefficient for spherical particles is taken from Morsi relation developed by Buzzelli (2008) to calculate the friction factor
and Alexander (1972) while that of non-spherical particles from f explicitly, defined as
Haider and Levenspiel (1989). The one-way coupling was applied  
2 B2 3
for the cases with Black Powder particles. This is a reasonable B1 þ 2 log Re
1
assumption because of the very low volume fraction of particles pffiffiffi ¼ B1  4 5 (17)
encountered (Balachandar and Eaton, 2010). Brownian force was f 1 þ 2:18
B2
not included because sub-micron particles were not considered and
turbulent dispersion was modeled using the Random Walk model where
244 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253

8 " #
> ½0:774 lnðReÞ  1:41
> B1 ¼
> rffiffiffiffi 1
2
ffi ffi
ln ð1þfÞ 2 þp13tan1 2f1
p 1 ln
ð1þfk Þ2 p1 ffi 2f 1
þ 3tan1 pk3 ffi
>
< k
2=3 6 1fþf 3 6 1fk þf2
1 þ 1:32 Is ¼ 14:5Sc
k (25)
D (18)
>
> for 0:45kþ < 5
>
>
: k
B2 ¼ Re þ 2:51B1
3:7D where
were k is the roughness coefficient, k ¼ 0 for smooth surface, D is
u* 1 kþ
the pipe diameter and Re is Reynolds number defined in Eq. (15). kþ ¼ k; f ¼ Sc1=3 ; fk ¼ Sc1=3 (26)
In the present study two different deposition models will be ng 2:9 32:2
considered and the results will be compared with CFD The non-dimensional surface roughness kþ is assumed to vary
computations. within the hydraulically smooth regime in which kþ
5 (Wood,
The first model considered is the one developed by Wood 1981) for all the pipes of the tree shaped network case.
(1981), where in the case of a smooth surface, the non- The second model was developed by Fan and Ahmadi (1993)
dimensional velocity is given by where the effect of the external forces was included in their
work. They extended the turbulent sub-layer model of Cleaver
Vpþ ¼ 0:057Sc2=3 þ 4:5  104 tþ þ
p þ ut (19) and Yates (1976) by including gravity and lift forces in their
calculation. They also accounted for the effect of Brownian
where Sc ¼ ng/Ddiff is the Schmidt number, and Ddiff being the diffusion and surface roughness. The developed empirical equa-
particle mass diffusivity given as tion for deposition of particles in vertical duct including the ef-
fect of gravity along the flow direction and the surface roughness
kb T
Ddiff ¼ Cc (20) is given by:
3pmg dp

8
>
>
>
>  
>
>  2
>
> 2 31 1þtþp 2 Lþ1
>
> þ 1 þ tþ2 þ þ
g L

>
> 0:64k þ 2dp þ p 1
>
> 16
6 0:01085 1þtþ
2 þ
p L1
7
>
>
> 0:084Sc 2=3
þ 4     7
5
>
> 2 3:42 þ t g þ L = 0:01085 1 þ t L
þ 2 þ þ 2 þ
>
< p 1 p 1
Vdþ ¼

(27)
>
>
> 2 0:037
>  1 þ 8eðtp 10Þ =32
þ

>
>    if Vdþ < 0:14
>
>
>
> 1  tp L1 1 þ g þ =0:037
þ2 þ
>
>
>
>
>
>
>
> 0:14 otherwise
>
>
:

where Lþ þ *
þ u
kb ¼ 1.38065  1023 (J/K) is the Boltzmann constant and T is the 1 ¼ 3.08/(Sdp ) and dp ¼ ng dp

temperature. uþ For the validation case with CFD which consists of a single pipe
t in Eq. (19), represents the contribution to particle
deposition velocity by gravitational sedimentation in horizontal geometry, the CFD simulations of Black Powder deposition used the
duct, which is defined as DNS mean fluctuating velocity profile of Wu and Moin (2008) in
which a smooth pipe assumption was considered, hence, the sur-
uþ þ þ face roughness for this case is equal to zero (k ¼ 0). For the hori-
t ¼ tp g (21)
zontal pipe with rough surface (k # 0) as considered in the present
where g þ is expressed in the following form study, gþ ¼ 0 in Eq. (22) and the gravitational sedimentation ve-
locity uþ
t of Eq. (21) must be added to the expression. Thus, Eq. (27)
ng can be simplified as
gþ ¼ 3
g (22)
u*

and g is the gravitational acceleration.

For a rough surface, Wood (1981) modified the theoretical re- 2 2 3  2 þ

1 1þtþ
p L1
sults of Davies (1966) and recommended the following analytical 0:64kþ þ 12dþ
p
16 7
solution for deposition of solid particles in the diffusion regime by Vdþ ¼0:084Sc2=3 þ 4 5
2 3:42 (28)
considering the effect of the surface roughness

2 0:037
 1 þ 8eðtp 10Þ =32
þ

Vþ ¼
1
(23) þ uþ
t
1  tþ þ2
IB þ Is p L1

In the case of a smooth pipe, a zero value of the roughness co-

IB ¼ 24:2 for 0:45kþ < 5 (24)
efficient is considered (k ¼ 0).
 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253
Fig. 1. Program algorithm.
4. Methodology
The methodology to simulate the transport and deposition of BP
is incorporated in a calculation program. The program algorithm is
summarized in the flowchart shown in Fig. 1. After determining the
connectivity of the network, the mass flow rate is calculated at each
branch, then the analytical solution for steady state Advection-
Reaction equation is used to calculate the concentration distribu-
tion along each length step of the pipeline for the whole network
taking into account the phase split at the existing junctions.
The following boundary conditions for the mass flow rate and
the concentration hold at the network nodes:
a) At the source node: A constant mass flow rate and a constant
concentration C(1,1) ¼ C0 are imposed at the inlet of the first pipe
assumed to represent the total mass flow rate and the total quantity
of Black Powder obtained after cleaning operation during a long
term period, respectively.
b) At the junction nodes: Conservation of mass must be pre-
served at these locations to calculate the mass flow rates at each
pipe. For the solid particles, the phase split at the junctions is ob-
tained from CFD calculations (Kharoua et al., 2015) and imple-
mented in the 1D code where the boundary condition of the
concentration at each junction is defined as.
For i > 1
Cði; jÞ ¼ Split*Cði  1; Nði  1ÞÞ
(29)
Cði; jÞ ¼ 1  Split*Cði  1; Nði  1ÞÞ
where Split is the phase split coefficient and N(i-1) is the last in-
ternal node of the pipe (i-1).
c) At the demand nodes: A constant fluid mass flow rate is
defined at each demand node.
245
The estimated deposited bed height is calculated as follow.
- Calculate the deposited concentration.
- Calculate the mass of the deposit.
- Calculate the number of particles deposited in each length step
dx of the pipes by knowing the mass of one particle.
- Calculate the volume of the deposit in each length step dx of the
pipes by knowing the volume of one particle and assuming that
the packed bed voidage, ε ¼ 0.37. Several authors ranged the
voidage 0.37 to 0.39 for dense packed bed of mono-sized
spherical particles (Yang, 2003).
- Calculate the bed height.
5. CFD simulation of phase split and particle deposition
A CFD analysis cannot be applied with the required resolution to
very long pipelines as those considered in this study. On the other
hand, phenomena, such as phase split cannot be explicitly deter-
mined from the proposed 1D approach and there is a lack of data in
the literature. CFD is therefore used as a support tool to provide the
1D approach with the necessary data on phase split. More details
on the CFD results on phase split can be found in Kharoua et al.
(2015).
As mentioned above, deposition is accounted for, in the con-
centration equation, through the source term (Eq. 4) using the
empirical models of Wood (1981) and Fan and Ahmadi (1993) and
is contrasted with the CFD results using the DPM. For particle
deposition in pipes, the DPM model was combined with the tur-
bulence model RSM (Reynolds stress model) to account for
anisotropy of near wall fluctuating velocities. The flow was char-
acterized by a Reynolds number equal to 44000 and the experi-
ments of Liu and Agarwal (1974) were considered for validation
246 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253
Fig. 2. Normalized deposition velocity vs Normalized relaxation time (CFD simulation of Black Powder deposition in horizontal pipe flow).
purposes. The fluid turbulent field was corrected by imposing DNS
profiles of Wu and Moin (2008) for a more accurate prediction of
the deposition velocity. Assuming symmetry, a horizontally-
directed half pipe, with a diameter equal to 0.06 m and length
equal to 0.9 m, was used to simulate particle deposition. Trap
boundary condition was imposed at the walls assuming no re-
entrainment of particles. The computational grid consisted of
about 12 million cells. The mesh, near the walls, was refined to
resolve the viscous sub-layer explicitly (yþ < 1). Simulations with
mono-dispersed and poly-dispersed particles (dp ¼ 1e50 mm) gave
identical results in terms of deposition velocity. It is worth to
mention that previous numerical studies used 2D geometries due
to the excessive computational cost required by full 3D geometries
for such problems. The deposition velocity curve for Black Powder
is presented in Fig. 2. The curve obtained using CFD is meant to
quantify the deposition velocity for Black Powder and to assess the
existing empirical models developed for different types of particles
which are implemented in the proposed 1D approach to model the
appropriate deposition rate in the source term of Eq. 4. CFD results
of the concentration decay along the pipe length were extracted
and compared with those predicted using the 1D code for the same
case. These are presented in section 6.
6. Model validation using CFD
In the present study, two different geometries are used to test
the approach. A single pipe geometry and a simple tree shaped
network schematically shown in Figs. 3 and 4 respectively. The first
test case for the single pipe geometry which consists of 2 nodes is
used to validate the 1D approach with the 3D simulation results
obtained using FLUENT code. Node S1 is the source (where Black
Powder is assumed to be generated), while D1 is a demand node.
The second test case considered in the present study is a tree
shaped network which consists of 6 nodes and 5 branches. Node S1
is the source (where Black Powder is assumed to be generated),
while nodes D1, D2, D3, are the point of demand, and nodes J1, J2
are internal junctions. The phase split data is extracted from
Kharoua et al. (2015) and the dimensions and the flow conditions
for the considered case are shown in Table 1.
Fig. 3. Schematic of single pipe geometry.
7. Results and discussion
In this section, numerical calculations using the 1D approach
will be presented. The first test case is used as benchmark for the
1D approach by considering a single pipe geometry. Results are first
compared with the 3D CFD simulations for different particle di-
ameters. The next step is to further examine the validity and extent
of the current models using a tree shaped network. Results are
presented and discussed in terms of the concentration distribution
and the developed bed height trough the different gas pipelines for
different particle diameters and flow conditions.
7.1. Application 1 e single pipe geometry for validation
The first application of the 1D approach consists of a single pipe
supplied by a contaminated reservoir with different inlet concen-
trations as shown in Fig. 3. This test is a simple application that
serves to validate the 1D approach with the CFD results obtained
using the DPM model for three values of the particle diameter,
dp ¼ 1, 10 and 20 (mm). Researchers reported that particle size of
Black Powder is typically in the range of submicron to up to
approximately 100 (mm) but around 95% of the cumulative particle
size distribution of Black Powder particles in gas transmission
network is less than 20 (mm), (Trifilieff et al., 2009). Furthermore,
for larger particle diameters the inertial effect becomes important
and the validity of the two developed models of Wood (1981) and
Fan and Ahmadi (1993) becomes questionable due to the lack of
experimental data in the inertial-moderated region. The flow in the
pipe is considered to be turbulent and fully developed and details of
the corresponding data provided by CFD calculations are summa-
rized in Table 2. The Black Powder properties used in this study can
be found in Khan and Al-Shehhi (2015).
Fig. 4. Schematic of tree shaped gas distribution pipeline.
 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253
Table 1
Dimensions and the flow operating conditions for the gas network.
Pipe N Start node End node
1 S1 J1
2 J1 D1
3 J1 J2
4 J2 D2
5 J2 D3
Table 2
Flow and boundary conditions provided by CFD calculations.
Particle diameter (mm) m_ p (kg/s) m_ gas (kg/s)
1 1.09E-06 0.0221
10 4.37E-05 0.0221
20 0.000393 0.0221
Fig. 5 represents the concentration distribution through the pipe
and the comparison of analytical solution implemented in the 1D
code with the results obtained from the CFD computations for three
values of the particle diameter. The calculated deposition rate (bdep)
obtained from CFD was included in the analytical solution, Eq. (7)
and the concentration distribution is plotted in Fig. 5. A good
agreement is obtained between the 1D code and the CFD calcula-
tions. This agreement confirms the validity of the analytical solu-
tion of the Advection-Reaction equation for solid particles
movement in turbulent pipe flow and the validity of using the 1D
approach.
For the same geometry and boundary conditions, a comparison
between the CFD results and the 1D code results using the two
selected deposition models, is conducted for different particle di-
ameters as shown in Fig. 6. The comparison between the 1D code
results using (Wood, 1981) and (Fan and Ahmadi, 1993) models
with the CFD calculation shows a good agreement for particle di-
ameters less than or equal to 10 (mm), while for larger solid parti-
cles, dp ¼ 20 (mm), a remarkable difference is obtained for the
concentration distribution. For a particle diameter dp ¼ 20 (mm), the
deposition rates obtained from the 1D code using both models are
higher compared to the CFD values, hence the concentration of BP
particles obtained from the 1D code decreases rapidly along the
pipe.
The difference between the two deposition models and the CFD
results for larger particle diameters in the upper limit of the
diffusion-impaction regime comes from the fact that the models
are known to perform in a less accurate way for large particles
Fig. 5. Comparison of analytical solution implemented in the 1D code with the results
provided by the CFD computations for three values of the particle diameter, dp ¼ 1, 10
and 20 (mm).
247
Pipe length, L, (km) Pipe diameter, D, (in) Mass flow rate, m_ (kg/s)
15 24 43.364
3 4 0.04475
27 24 42.9165
10.53 24 39.6
15.3 36 3.9195
rp (kg/m3) rgas (kg/m3) Мgas (m3/s) Deposition rate bdep
3169.3 4.18 1.14e05 0.2331
3169.3 4.18 1.14e05 0.5512
3169.3 4.18 1.14e05 2.2743
belonging to the beginning of the inertial regime. For example, the
prediction of Fan and Ahmadi (1993) model are in good agreement
with the general shape of the experimental data, although for the
deposition to a smooth surface in the range of 1 < tpþ < 10, the
model predicts an increase in deposition velocity that is sharper
than observed experimentally (Sippola and William, 2002). In the
diffusion-impaction regime and for tpþ < 4, the model for the
smooth surface under-predicts deposition velocities and for tpþ >
4, the model over-predicts deposition velocities (Sippola and
William, 2002). This inaccurate prediction of the models may
derive from the fact that most of these models are developed based
on mean turbulence properties, rather than variable with a distri-
bution of properties. If deposition is more strongly influenced by
turbulence parameters that are more intense than the average,
models that are based on mean turbulence properties will give less
accurate predictions.
7.2. Application 2 e tree shaped network
In this section, the Advection-Reaction equation is applied to the
tree shaped network presented in Fig. 2 with the appropriate di-
mensions and flow conditions presented in Table 1. The different
segments of the network are analyzed by considering the deposi-
tion of Black Powder particles using the two selected models of
Wood (1981) and Fan and Ahmadi (1993) for smooth and rough
pipe surfaces. The objective is to calculate the concentration dis-
tribution and the estimated bed height of the accumulated Black
Powder in each length step dx along the pipelines for different
particle diameters 0.5, 1, 8 and 15 (mm) respectively.
Fig. 7 shows the concentration distribution and the corre-
sponding estimated bed height of Black Powder particles of 0.5
(mm) diameters within the different segments of the network. A
constant value of the surface roughness k ¼ 0.01 (mm) is consid-
ered, however different values of the non-dimensional surface
roughness kþ (kþ ¼ 0.075, 1, 1.925, 1.94 and 1.795) were obtained
due to the change in the friction velocity as a consequence of the
change in the flow rate for each pipe.
For a smooth surface (kþ ¼ 0), it is shown that the concentration
is decreasing along the pipes and that around 60% of Black Powder
is deposited in the first pipe while only 40% is transported to the
next segments and smaller amounts of Black Powder reached pipes
4 and 5. The split at the first junction shows that 90% of the solid
particles follow the gas flow transported into pipe 3 while only 10%
is transported to pipe 2. This is due to the higher flow rate in pipe 3
when compared to the flow rate in pipe 2. The maximum estimated
248 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253
Fig. 6. Comparison of analytical solution implemented in the 1D code using the two deposition models selected with the results provided by the CFD computations for three values
of the particle diameter.
bed height reaches a value of 7.4 (mm) in the first pipe, this value
can be higher when increasing the assumed inlet concentration.
For rough pipe (kþ s 0), it is shown that the deposition of
submicron particles is highly affected by the surface roughness and
the solid particles are likely to settle rapidly due to the increase of
the deposition rate when compared to the smooth surface case
using the same flow conditions. This behavior is expected and has
been reported in the literature, (Wood, 1981) and (Fan and Ahmadi,
1993). Furthermore, the large change in the deposition velocity is
valid for small particles away from the inertial regime (tþ< 5) and it
was explained by the dominance of the inertia-interception
mechanisms. It is shown also that Fan and Ahmadi (1993) model
predicts a higher deposition velocity when compared to Wood
(1981) model and results into a faster decay and accumulation of
solid particles with higher bed heights.
Fig. 8 shows the concentration distribution and the corre-
sponding estimated bed height of Black Powder particles of 1 (mm)
diameter within the first three segments of the network (pipe 1, 2
and 3). The results for pipe 4 and 5, are not presented here because
the Black Powder particles cannot reach these pipes and the con-
centration is equal to zero.
The results for the two deposition models for the smooth sur-
face case (kþ ¼ 0) are almost identical with minor differences. It is
shown that the concentration is decreasing along the pipe and that
around 95% of Black Powder is deposited in the first pipe while only
5% is transported to the second and the third pipes and almost no
solid particle reached pipes 4 and 5. For a rough surface, the
deposition rate is increased substantially and the obtained results
using Fan and Ahmadi (1993) show that all particles are deposited
in the first pipe forming a higher bed height at the first 3 (km) of the
pipe. However, Wood model (1981), predicts the transport of a
smaller quantity Black Powder to the downstream segments.
For particle diameters of 8 and 15 (mm), results presented in
Fig. 9, and 10 respectively show that all the quantity of Black
Powder deposited rapidly and accumulated within the first 500 (m)
of the first pipe for the particle diameter of 8 (mm) and within 200
(m) only for particles of 15 (mm). Figures for the other pipelines are
not presented here, but again no solid particles can be transported
to reach these pipelines. It is shown also that both results for
smooth and rough surfaces for the larger particles sizes are
identical and that increasing the surface roughness does not affect
the deposition velocity. In the development of Wood (1981) and Fan
and Ahamdi (1993) models, plots of the non-dimensional deposi-
tion velocity against the non-dimensional relaxation time show
that the S-curves corresponding to different values of the rough-
ness converge towards the S-curve of the smooth surface case in
the inertial regime. In this case, the deposition of solid particles is
dominated by the gravitational forces and the Brownian forces are
negligible. Furthermore, the effect of the surface roughness was
shown to be higher for small particles away from the inertial
regime (tþ < 5), in which the particles are assumed to follow the
fluid streamlines which are too close to the rough walls. For smaller
particles, the Brownian diffusion dominates and the effect of
gravity is negligible.
It can be concluded that, for the same flow conditions, the
largest particles deposit faster, the greatest accumulation of Black
Powder particles lay in the first segments of the network and only
small quantities of very fine particles of around 1 (mm) diameter or
less can be transported to the next segments under the assumption
of smooth surfaces and the conditions provided in this example. It
is worth mentioning that the level of the estimated bed height
depends on the quantity of Black Powder generated which is
related to the assumed inlet concentration. Increasing the inlet
concentration will lead to a higher bed height.
8. Conclusion
A novel methodology for tracking the dispersion of Black Pow-
der within gas transmission pipelines is developed and the depo-
sition of Black Powder particles was investigated. The analytical
solution of the advection-reaction equation is used to calculate the
particle concentration under different working conditions,
different particle diameters and pipe wall roughness. The estimated
bed height is calculated based on the number of particles deposited
in the pipelines. The 1D approach was validated against CFD
simulation based on the Discrete Phase Model (DPM). The reported
results are of significant practical interest and can contribute to a
better understanding of the movement of Black Powder particles in
the gas network.
The presented results showed that for the investigated flow
Fig. 7. Concentration distribution and the estimated bed height of 0.5 (mm) particles within the different segment of the network using (Wood, 1981) and (Fan and Ahmadi, 1993)
models.
250 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253

Fig. 8. Concentration distribution and the estimated bed height of 1 (mm) particles within the different segment of the network using (Wood, 1981) and (Fan and Ahmadi, 1993)
models.

conditions which have been assumed to be close to real gas flow
conditions, very fine particles of 1 (mm) diameter or less, once
generated, can be transported easily towards the downstream
segments of the network and find their way towards the far supply
points to customers. In contrast, larger Black Powder particles with
dp > 1 (mm), are likely to settle rapidly near to the regions where
they were generated.
The results showed also the effect of the surface roughness on
the Black Powder concentration. As reported in the literature, these
effects are limited to very fine particles of 1 (mm) diameter or less.
Hence, the exponential decay of the Black Powder concentration is
increased due to the increase of the surface roughness leading to
faster accumulation of particles over shorter distances when
compared to the case of a smooth surface. However, the larger
particles are not affected and the changes in the concentration
distribution are identical for both smooth and rough surface cases.
For practical application of this modeling approach, additional
work is needed to account for the pickup case where the Black
Powder particles can be picked up from a stationary bed and a
continued development of methods for evaluating more general
and accurate correlation to calculate the deposition rate is also
needed. To further validate the model, it would also be of great
benefit to have the results of carefully conducted experiments in
which particle deposition or pickup as a function of size and near-
surface air flow conditions are well characterized.
 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253 251

Fig. 9. Concentration distribution and the estimated bed height of 8 (mm) particles within the first segment of the network using (Wood, 1981) and (Fan and Ahmadi, 1993) models.

Fig. 10. Concentration distribution and the estimated bed height of 15 (mm) particles within the first segment of the network using (Wood, 1981) and (Fan and Ahmadi, 1993)
models.
252 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253

Acknowledgments granular materials in pipe branches. Part. Sci. Technol. 14, 213e219.
Li, K., Kuang, S.B., Zou, R.P., Pan, R.H., Yu, A.B., 2014. Numerical study of horizontal
pneumatic conveying: effect of material properties. Powder Technol. 251,
The financial support of this research study was granted by the 15e24.
Gas Processing and Material Science Research Centre GRC at the Li, K., Kuang, S.B., Zou, R.P., Pan, R.H., Yu, A.B., 2013. Numerical study of vertical
Petroleum Institute of Abu Dhabi in the framework of the research pneumatic conveying: effect of friction coefficient. In: powders and grains 2013.
In: Proceedings of the 7th International Conference on Micromechanics of
project GRC-002. Granular Media. AIP Conference Proceedings, vol. 1542. July 8-12, 2013, Uni-
versity of New South Wales (UNSW), Australia.
Liu, B.Y.H., Agarwal, J.K., 1974. Experimental observation of aerosol deposition in
References turbulent flow j. Aerosol. Sci. 5, 145e155.
Makse, H.A., Andrade, J.S., Stanley, H.E., 2000. Tracer dispersion in a percolation
network with spatial correlations. Phys. Rev. 61, 583e586.
Alloul-Marmor, L., 2002. Re entrainement par e coulement d’air d’une contamination
Mezhericher, M., Brosh, T., Levy, A., 2011. Modeling of particle pneumatic conveying
particulaire depose sur une surface: application au cas d’un “tas” de poudre.
using DEM and DPM methods particulate. Sci. Technol. 29, 197e208.
PhD thesis (Universite  Paris XII), report ISRN IRSN e 2002/28 e FR.
Morsi, S.A., Alexander, A.J., 1972. An investigation of particle trajectories in two-
Andrei-Mugur, G., Sanda-Carmen, G., 2012. Chlorine concentration decay in the
phase flow systems. J. Fluid Mech. 55, 193e208.
water distribution system of a town with 50000 inhabitants. U.P.B. Sci. Bull. Ser.
Morten, L., Gustavo, A.Z., 2015. Particle transport in semi-dilute turbulent pipe flow.
D. 74, 1454e2358.
J. Dispers. Sci. Technol. 36, 1513e1526.
Baldwin, R., 1998. Here are procedures for handling persistent black-powder
Nicholson, K.W., 1988. A review of particle resuspension. Atmos. Environ. 22,
contamination. Oil Gas J. 96, 51e58.
2639e2651.
Balachandar, S., Eaton, J.K., 2010. Turbulent dispersed multiphase flow. Annu. Rev.
Papavergos, P.G., Hedley, A.B., 1984. Particle deposition behavior from turbulent
Fluid Mech. 42, 111e133.
flows. Chem. Eng. Des. 62, 275e295.
Bear, J., 1972. Dynamics of Fluids in Porous Media. Elsevier, New York.
Pope, S.B., 2000. Turbulent Flows. Cambridge University Press, New York.
Ben-Avraham, D., Havlin, S., 1982. Diffusion on percolation clusters at criticality.
Reeks, M.W., Hall, D., 2001. Kinetic models for particle resuspension in turbulent
J. Phys. A Math. General 15, 691e697.
flows: theory and measurement. J. Aerosol. Sci. 32, 1e31.
Boothroyd, R.G., 1971. Flowing Gas-Solids Suspensions. Chapman & Hall, London.
Reeks, M.W., Reed, J., Hall, D., 1988. On the resuspension of small particles by a
Bouilly, J., Limam, K., Beghein, D., Allard, F., 2005. Effect of ventilation strategies on
turbulent flow. J. Phys. D. Appl. Phys. 21, 574e589.
particle decay rates indoors: an experimental and modelling study. Atmos.
Rossman, L.A., Boulos, P.F., 1996. Numerical methods for modeling water quality in
Environ. 39, 4885e4892.
distribution systems: a comparison. J. Water Resour. Plan. Manag. ASCE 122,
Bunde, A., Havlin, S., 1991. Fractals and Disordered Systems. Springer-Verlag, New
137e146.
York.
Sahimi, M., Davis, H.T., Scriven, L.E., 1983. Dispersion in disordered porous media.
Buzzelli, D., 2008. Calculating friction in one step. Mach. Des. 80, 54e55.
Chem. Eng. Commun. 23, 329e341.
Cattanach, K., Ramachandran, S., Jovancicevic, V., Sherik, A.M., 2010. Novel test
Schlichting, H., 1987. Boundary Layer Theory, seventh ed. McGraw-Hall, NY.
procedures to understand and mitigate black powder formation in sales gas
Schneider, H., Frank, T., Pachler, K., Bernert, K., 2002. A numerical study of the gas-
pipelines. In: In: 13th Middle East Corrosion Conference and Exhibition, 14-17
particle flow in pipework and flow splitting devices of coal-fired power plant.
February, 2010, Bahrain.
In: In: 10th Workshop on Two-phase Fow Predictions, April 9-12, 2002. Martin-
Chahibi, Y., Pierobon, M., Song, S.O., Akyildiz, I.F.A., 2013. Molecular communication €t Halle-Wittenberg, Halle (Saale), Germany.
Luther-Universita
system model for particulate drug delivery systems. IEEE Trans. Biomed. Eng.
Sehmel, G.A., 1980. Particle resuspension: a review. Envirn. Int. 4, 107e127.
60, 3468e3483.
Sherik, A.M., 2007. Black powder in sales gas transmission lines. Saudi Aramco J.
Chu, K.W., Yu, A.B., 2008. Numerical simulation of the gassolid flow in three-
Technol. 2e10 fall 2007.
dimensional pneumatic conveying bends. Ind. Eng. Chem. Res. 47, 7058e7071.
Sherik, A.M., 2008. Black powder-conclusion: management requires multiple ap-
Cleaver, J.W., Yates, B., 1976. The effect of re-entrainment of particle deposition.
proaches. Oil Gas J. 106, 66e68.
Chem. Eng. Sci. 31, 147e151.
Sherik, A.M., Lewis, A., Duval, L.S., 2008. . Managing black powder in sales gas
Colebrook, C.F., White, C.M., 1937. Experiments with fluid friction in roughened
transmission pipelines. Saudi Aramco J. Technol. 42e48. Fall 2008.
pipes. In: Proceedings of the Royal Society of London. Series a - Mathematical
Sippola, M.R., William, W.N., 2002. Particle Deposition from Turbulent Flow: Review
and Physical Sciences, 161, pp. 367e381.
of Published Research and its Applicability to Ventilation Ducts in Commercial
Crowe, J.C.T., 1982. Review - numerical models for dilute gas-particle flows. J. Fluids
Buildings. LBNL Report 51432. Lawrence Berkeley National Laboratory, Berkeley,
Eng. 104, 297e303.
California.
Davies, C.N., 1966. Deposition of aerosol from turbulent flow through pipes. Proc. R.
Smart, J., 2007a. Determining the velocity required to keep solids moving in
Soc. A 289, 235e246.
pipeline. J. Pipeline Eng. 6, 5e13.
Elghobashi, S., 1991. Particle-laden turbulent flows: direct simulation and closure
Smart, J., 2007b. Movement of black powder in gas pipelines. J. Pipeline Eng. 234,
models. Appl. Sci. Res. 48, 301e314.
82e85.
Elsaadawy, E.A., Sherik, A.M., 2012. Pipe diameter does not affect black powder
Smart, J., Winters, R., 2008. Black powder migration in gas pipelines and associated
distribution, erosion. Oil Gas J. 110, 104e110.
problems. In: In: Proceedings of the 20th International Pipeline Pigging and
Fan, F.G., Ahmadi, G., 1993. A sublayer model for turbulent deposition of particles in
Integrity Management Conference, 13e14 February 2008 (Houston, Texas, USA).
vertical ducts with smooth and rough surfaces. J. Aerosol. Sci. 24, 45e64.
Tenneti, S., Subramaniam, S., 2013. Particle-resolved Direct Numerical Simulation
Fernandes, C., Karney, B., 2001. Numerical solution of the advection-dispersion-
for Gas-Solid Flow Model Development Annual Review of Fluid Mechanics, vol.
reaction equation under transient hydraulic conditions. In: In: Proceedings of
46, pp. 199e230.
the 29th International Association of Hydraulic Engineering and Research
Trifilieff, O., Thomas, H.W., 2009. Black Powder Removal from Transmission Pipe-
(IAHR) Congress, September 16-21, 2001, Beijing, China.
lines: Diagnostics and Solutions. Scientific and Technical Report. Presented at
Friedlander, S.K., Johnstone, H.F., 1957. Deposition of suspended particles from
the Pipeline Rehabilitation & Maintenance Conference. Gulf International
turbulent gas streams. Industrial Eng. Chem. 49, 1151e1156.
Convention Center, Bahrain. January 19-21, 2009.
Friedlander, S.K., 1977. Smoke, Dust and Haze: Fundamentals of Aerosol Behavior.
Tsochatzidis, N.A., Maroulis, K.E., 2007. Methods help remove black powder from
Wiley, New York.
gas pipelines. Oil Gas J. 105, 52e58.
Godoy, J.M., Carvalho, F., Cordilha, A., Matta, L.E., Gody, M.L., 2005. 210 Pb content in
Tsochatzidis, N.A., 2008. Study addresses black powder's effects on metering
natural gas pipeline residues “ black-powder ” and its correlation with the
equipment. Oil and Gas Journal 106, 56e61.
chemical composition. J. Environ. Radioact. 83, 101e111.
Tsuji, Y., Tanaka, T., Ishida, T., 1992. Lagrangian numerical-simulation of plug flow of
Guha, A., 2008. Transport and deposition of particles in turbulent and laminar flow.
cohesionless particles in a horizontal pipe. Powder Technol. 71, 239e250.
Annu. Rev. Fluid Mech. 40, 311e341.
Van Genuchten, M.T., Alves, W.J., 1982. Analytical Solutions of the One-dimensional
Haider, A., Levenspiel, O., 1989. Drag coefficient and terminal velocity of spherical
Convective-dispersive Solute Transport Equation. Technical Bulletin, No. 1661,
and non-spherical particles. Powder Tech. 58, 63e70.
second ed. U.S. Department of Agriculture, Washington, DC.
Hetsroni, G., 1982. In: Handbook of Multiphase Systems. McGraw-Hill, New York.
Winning, H.K., Coole, T., 2013. Explicit friction factor accuracy and computational
Hidayat, M., Rasmuson, A., 2005. Some aspects on gasesolid flow in a U-bend:
efficiency for turbulent flow in pipes. Flow. Turbul. Combust. 90, 1e27.
numerical investigation. Powder Tech. 153, 1e12.
Wood, N.B., 1981. A simple method for the calculation of turbulent deposition to
Khan, S.T., Al-Shehhi, M.S., 2015. Review of black powder in gas pipelines e an
smooth and rough surfaces. J. Aerosol. Sci. 12, 275e290.
industrial perspective. J. Nat. Gas Sci. Eng 25, 66e76.
Wu, X., Moin, P., 2008. A direct numerical simulation study on the mean velocity
Kharoua, N., AlShehhi, M., Khezzar, L., 2015. Prediction of black powder distribution
characteristics in turbulent pipe flow. J. Fluid Mech. 608, 81e112.
in junctions using the discrete phase model. Powder Tech. 286, 202e211.
Xinxiang, P., Xue, L., Guohui, Li, Tao, Li, 2011. Numerical investigation on
Kirkpatrick, J.P., Brizel, D.M., Dewhirst, M.W.A., 2003. Mathematical model of tumor
gaseparticle flows in horizontal channel under the reduced gravity environ-
oxygen and glucose mass transport and metabolism with complex reaction
ments. Acta Astronaut. 68, 133e140.
kinetics. Radiat. Res. 159, 336e344.
Yang, W., 2003. Handbook of Fluidization and Fluid-particle Systems. Marcel Dek-
Lai, A.C.K., 2002. Particle deposition indoors: a review. Indoor Air 12, 211e214.
ker, Inc, New York. ISBN: 0-8247-0259-X.
Li, X., Shen, H.H., 1996. A computer simulation of dilute pneumatic conveying of
 A. Filali et al. / Journal of Natural Gas Science and Engineering 28 (2016) 241e253 253

Ziskind, G., Fichman, M., Gutfinger, C., 1995. Resuspension of particulates from Sratio: Particle-to-fluid density ratio
surfaces to turbulent flowsdreview and analysis. J. Aerosol. Sci. 26, 613e644. St: Stoke Number
Split: Phase split coefficient
t: time (s)
Nomenclature T: Temperature (K)
U: Mean gas velocity in the (x) direction (m/s)
C0: Inlet Particle Concentration uþt : Contribution to particle deposition velocity by gravitational sedimentation
C: Particle Concentration u*: Friction Velocity (m/s)
Cc: Cunningham correction factor up: Particle Velocity (m/s)
CD: Drag coefficient Vþd : Non-dimensional deposition velocity
dp: Particle diameter (m) x: Distance along the flow direction (m)
D: Pipe diameter (m)
Ddiff: Diffusion coefficient (m2/s) Greek Symbols
F: Friction Coefficient
g: Gravitational acceleration (m/s2)
bdep: Deposition rate (s1)
gþ: Non-dimensional gravitational acceleration
rg: Gas density (kg/m3)
Jw: deposition mass flux (particles/m2/s)
rp: Particle density (kg/m3)
k: Roughness coefficient (m)
rp,mean: Mean particle density (kg/m3)
kþ: Non-dimensional surface roughness
mg: Gas dynamic viscosity (N.s/m2)
kb: Boltzmann Constant (J/K)
ng: Gas Kinematics Viscosity (m2/s)
Kn: Quotient of mean free path of the carrier gas (m)
tw: Wall shear stress
L: Pipe length (m)
tþ
p : Non-dimensional particle relaxation time
m_ : Mass flow rate (kg/s)
N(i): Last internal node of the pipe (i).
Subscripts
R: Pipe radius (m)
Re: Reynolds Number
Rer: Relative Reynolds Number g: Gas
S: Source term p: Particle
Sc: Schmidt Number