Computational Methods in Multiphase Flow VI 261

Hydrate formation in natural gas pipelines

M. Naseer & W. Brandstätter
Department of Petroleum Production and Processing,
University of Leoben, Austria

Abstract

Gas hydrates pose a problem to the flow assurance programs in the oil and
gas industry. This paper highlights the mechanisms of hydrate formation in
natural gas pipelines. Computational Fluid Dynamics (CFD) was used to gain
insight in to the mechanisms of water vapour condensation and consequent water
accumulation in the lowered sections of a gas pipeline. The pipeline temperature
profile, condensation of water vapour at walls, hydrate formation and hydrate
slurry rheology were modelled with user defined functions (UDF) integrated into
the CFD-software Fluent. It was found that the uphill sections of gas pipelines are
natural sites for water accumulation and hydrate formation. It is demonstrated that
CFD is a valuable tool to understand complex physical phenomena occurring in
multiphase flow problems.
Keywords: natural gas, pipelines, condensation, hydrate formation, uphill
sections, viscosity, nucleation, sagging sections, kinetics, macroscopic.

1 Introduction
The growth of energy demand is constantly pushing the gas producers to operate
at higher pressures. In the start of the twentieth century, the natural gas industry
started facing the problem of unexpected pipeline blockage. It was shown in 1934
by HammerSchmidt that the blockages appearing to be ice were actually gas
hydrates [1]. The gas hydrate formation in oil and gas pipelines is an undesirable
situation, undermining the flow assurance programs in addition to posing threat to
the personnel and equipment. Clathrate hydrates are non-stoichiometric mixtures
of water and natural gas in which the gas molecules are trapped in polygonal
crystalline structure made of water molecules [2]. The water molecules arrange
themselves in an orderly fashion around the gas molecules, thus entrapping them.

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
doi:10.2495/MPF110221
262 Computational Methods in Multiphase Flow VI

These gas hydrates have the appearance of ice but can form well above the freezing
point of water at high pressures.
Hydrate formation in natural gas pipelines is a complex phenomenon to capture
in its entirety. The sequence of events leading to hydrate formation in gas pipelines
involves water vapor condensation, accumulation of water at lowered sections of
the pipeline, nucleation and growth of hydrate particles ultimately blocking the
pipeline. The site of pipeline blockage may not be the same as that of hydrate
deposition because just at the beginning of summer, the hydrate just released from
the pipewall along the length may migrate downstream and deposit at sagging
sections of the pipeline to cause the blockage [3]. Experiments on natural gas
hydrate under actual operating conditions in pipelines is difficult to emulate
due to scale of the problem and high pressures involved. However, different
experimenters have focused on various aspects of the hydrate formation. Bishoni
et al. [4] presented the formation and decomposition kinetics of gas hydrates at
a conceptual level, highlighting the different phases of hydrate formation and
growth. Riberio et al. [5] provided a critical review of literature regarding hydrate
formation kinetics, highlighting the modeling efforts by different researchers. In
an effort to emulate actual hydrate formation conditions, industrial scale flow loop
experiments were setup by ExxonMobil Upstream Research Company [6] and
Tulsa University [7] and Davies et al. [8] validated their hydrate formation model
against the experimental data. While the hydrate formation model just mentioned
was for the oil dominated flowlines, the hydrate formation may be quite different
in natural gas flowlines.

2 Theory
Natural gas pipelines always have some moisture content as the gas from the
formations reach the well head saturated with water [3]. During the cold season,
the ground temperatures are lower resulting in the temperature drop of the pipeline
content. If the temperature of the gas falls below the saturation temperature
corresponding to the dew point of water, condensation of water vapors will start. In
straighter sections of the pipeline, the condensate starts to accumulate at the base of
the pipe due to gravity. The accumulation may increase at the base of the sagging/
uphill sections of the pipe. If enough water is present and the thermodynamic
conditions are suitable, hydrate nucleation processes are initiated. To analyse
hydrate formation on a theoretical basis in principle two choices exist: one-
dimensional (1-D) or three-dimensional (3-D) models. While 1-D CFD models
may be suitable for very long pipeline sections, the effects of water condensation
and accumulation can only be modelled using the 3-D CFD methods.

3 CFD model
As an application example the pipeline section depicted in figure 1 was selected.
The computational domain consists of approximately 84000 polyhedral control

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
 Computational Methods in Multiphase Flow VI 263

Figure 1: Natural gas pipeline.

volumes. The grid was refined during the computational run time to resolve the
accumulation phenomena.
The “Mixture Model” of Fluent [9] was used to represent the interpenetrating
continua consisting of mixture components.

Table 1: Model parameters.

No. Model parameter Value

1. Pipe diameter 0.25m
2. Inlet velocity 2m/s
3. Water vapor volume fraction 0.2
4. Inlet temperature 315K
5. Operating pressure 6.3Mpa

Table 1 depicts different parameters of the pipeline section. The Pressure

Implicit Split Operator (PISO) algorithm was used to couple the pressure and
velocity fields. The K-Epsilon model was included to model the effects of
turbulence.

3.1 Mathematical framework

The mixture model treats ‘n’ different phases as interpenetrating and models
the fluid flow with conservation equations of mass, momentum and energy. The
continuity, momentum and energy equations for the mixture in differential form [9]
are given below.

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
264 Computational Methods in Multiphase Flow VI

Continuity:
∂
(ρm ) + ∇ · (ρmvm ) = 0 (1)
∂t
Momentum:
∂

(ρmvm ) + ∇ · (ρmvmvm ) = ∇p + ∇ · [µm ∇vm + ∇vm T
+ ρmg,
∂t
 n 

+ F + ∇ · αk ρkvdr,k vdr,k (2)
k=1

Energy:

∂  
n n
(αk ρk Ek ) + ∇ · (αk vk (ρk Ek + p)) = ∇ · [keff ∇T ] + SE (3)
∂t
k=1 k=1

where vm is the mass averaged velocity, ρm and µm are volume fraction averaged
mixture density and viscosity respectively, vdr,k is the drift velocity of individual
phases, Ek is the energy of phase k, αk represents the volume fraction of phase k,
p represents pressure, keff is the effective thermal conductivity, T represents the
temperature and SE represents energy source. In order to model the drag of the
primary phase on the secondary phases, a drift flux model available in the Fluent
was used where, the slip velocity was modelled as provided by Minnan et al. [9]
and the drag function correlation of Schiller and Neumann [9, 10] was used.

3.2 Model setup

For the sake of this study, a certain composition of natural gas was considered at the
inlet to the pipe. The molar concentrations were used as input to the program ‘CSM
Hyd’ [2] and a pressure-temperature equilibrium curve for the hydrate formation
was obtained. Based on the best fit through the equilibrium data, an equation was
obtained representing the threshold temperature for hydrate formation as

T = 8.5274 × ln(P ) + 270.86 (4)

Where T is Temperature in Kelvin and P is the pressure in MPa.

To calculate the saturation temperature according to the partial pressure of
vapor, the following equation was used.

Tsat = 16.335 × ln (P ) + 167.08 (5)

In order to accurately model the condensation in Fluent, a relationship

developed by De Schepper et al. [11, 12] was employed.

Mwater = 0.1 × αvapour × ρvapour |Tsat − Twall |/Tsat (6)

Modelling the temperature profile of the pipeline content along the length of the
buried pipeline is a challenging task because it varies with the seasonal changes

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
 Computational Methods in Multiphase Flow VI 265

and also with the local changes in the conditions in the vicinity of the pipeline. A
simplified expression was used [13] to describe the pipe temperature profile along
the length.  
−kπDx
T (x) = TGr + (To − TGr ) exp (7)
0.28QCp
where TGr is the ground temperature, x is the distance along the pipe from the
inlet, D is the pipe diameter, k is thermal conductivity, Q is the volumetric flow
rate and Cp is the specific heat capacity.
The kinetic modeling of hydrate nucleation and growth requires the definition
of a driving force. While different researchers have presented their versions of
kinetic models, there is a general lack of consensus among them. The applicability
of these models to dynamic flow systems is limited by assumptions regarding
thermodynamics at the molecular level and to the specifics of the experimental
setup. Most notable among them is the work by Kashchiev and Firoozabadi who
considered the difference between chemical potential of the species as the driving
force [14].
An authentic kinetic model can serve as bridge between the microscopic
phenomenon of crystallization and its macroscopic effects. The assumption is
made here that there is an onset of instantaneous nucleation due to high number
of nucleation sites (on the pipe wall due to roughness) and high supersaturation
of the condensed liquid water film on the pipe wall with hydrocarbon gas at
high pressure. Based on the general stoichiometric molar relationship for hydrate
formation, the threshold mass fraction (considering water and hydrocarbon gas)
in a cell was set to be approximately 0.866. This threshold value was imposed
through the UDFs.
When hydrate formation conditions are reached in a computational cell, the
hydrate formation is modelled by mass transfer equations for both the phases i.e.
water and gas by the following simplified relationship:

Mhyd = (ρwater × αwater ) + (ρgas × αgas ) (8)

To incorporate the rheological effects of the increase of suspension in the

solution, the following relationship was used [10, 15]:

µslurry = αwater (1 − αhyd )−2.5 (9)

4 Results and discussion

The effect of water condensation at the pipe walls and accumulation at the bottom
of the pipe section is depicted in figure 2. At the onset of condensation of water
vapor, the condensed water starts to accumulate at the bottom section of the pipe
i.e. at the 6 o’clock position and it flows along the mean flow direction. This effect
can be clearly observed in figure 2. It is asserted that the effect of accumulation
can only be highlighted in three dimensional simulations and can be completely
overlooked in the one dimensional models. Furthermore, the accumulation of free

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
266 Computational Methods in Multiphase Flow VI

Figure 2: Water phase contours showing condensation at wall and accumulation at

the bottom of the pipe.

water at the base of the uphill section of the pipe is displayed. It is concluded that
there are two main mechanisms of water accumulation in pipelines due to gravity
i.e. accumulation at bottom section of the pipe and accumulation at the base of
uphill/ sagging sections of the pipe. The accumulation phenomena is a compound
effect taking place at the base of uphill section of the pipe for aforementioned
reasons and is clearly visible in figure 2.
Presence of free water is a key factor in the hydrate formation initiation as
argued by Sloan [2]. The simulation shows that the base of the sagging sections
of pipelines are most susceptible to hydrate nucleation as these regions serve
as stagnation points/ traps for any condensed water in gas flows. The hydrate
formation at these sections can start in winters. Multiple hydrate deposits along the
length of the pipe result in additional pressure drop. The real problem may start
at the beginning of summers, when the hydrate loosens up at the pipeline walls
(not completely dissociated) and starts traveling [3]. The sloughed off hydrate
then flows along the mean flow direction until it reaches lowered sections and
gets trapped as packed deposits.
The region of high hydrate volume fraction in accumulated water is shown in
figure 3. The high hydrate volume fraction region is marked by the iso-surfaces
of hydrate concentration which can be clearly distinguished from the water iso-
surfaces. It is evident that the water stagnation points are most susceptible to
hydrate formation. The hydrate formation acts as additional resistance to the
flow as shown by the velocity vectors in the cut-plane through x = 0. The
velocity vectors actually drawn at x = 0 iso-plane are offset by some distance
for clearer visibility. Stagnation zone with high hydrate volume fraction regions
can be observed. In real pipelines the difference in velocities between the mean
flow and the hydrate deposits would result in a drag force between phases leading
to pressure drop undermining the flow assurance.

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
 Computational Methods in Multiphase Flow VI 267

Figure 3: Velocity vectors at x=0 plane drawn at an offset showing the effect of
high hydrate concentration.

Figure 4: Line used for the plots.

In order to plot different parameters, a line just at the bottom of the lowered
section of the pipe in between the sections 22m and 28m (measured from the inlet)
was used as shown in figure 4.
Figure 5 shows the variation of viscosity and water volume fraction at the base
of the lowered
n(straight) section. The gradual increase in viscosity mixture defined
as µm = k=1 αk µk goes hand in hand with increase in the water volume
fraction with a sudden spike indicating high hydrate volume fraction influencing
the mixture viscosity. The gradual slope in the curve before 27m highlights the
accumulation phenomena of water (pileup). The sudden increase in the viscosity

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
268 Computational Methods in Multiphase Flow VI

Figure 5: Mixture viscosity and water volume fraction along the length of lowered
section of the pipe.

beyond 27m is indicative of the formation of a new phase drastically effecting the
viscosity of the mixture.
The water volume fraction depicted in figure 5 shows a gradual increase initially
followed by a sharp increase, a sharp decrease and again a sharp increase. The
sharp decrease belongs to the region of high hydrate volume fraction and is acting
as a ‘double sided tidal barrage’ between down stream and upstream water. It is
interesting to note the effect of hydrate deposition on the water film moving in
the mean flow direction. At the upstream side of the deposit, it is acting like a
wall, resulting in increase in water volume as depicted in the figure and at the
downstream side, the result of water accumulation is due to the backflow of water
film adjacent to the deposit. The backflow is assisted by loss in the dynamic
pressure in the region shadowed by the deposit and gravity effects at the uphill
section. 
The mixture velocity vm = nk=1 αk ρk vk /ρm at the bottom of the pipe section
is shown in figure 6. The increase in the velocity between 22–22.5m section is due
to mean flow from the downhill section impinging the wall resulting in increase in
the velocity. Beyond 22.5m the mixture velocity drops due to friction at the walls.
In the section between 23–26m, the mixture velocity oscillates with a mean value
of about 0.5m/s. The “fish-hook” curve beyond 27m shows the drastic drop in the
mixture velocity due to high hydrate volume fraction.
The variation of static and dynamic pressure is depicted in figure 7. The drastic
increase of static pressure at about 27m (and decrease of dynamic pressure)
indicates the effects of high hydrate deposition.

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
 Computational Methods in Multiphase Flow VI 269

Figure 6: Mixture velocity and hydrate volume fraction along the length of lowered
section of the pipe.

Figure 7: Pressure variation along the length of lowered section of the pipe.

5 Conclusions and future work

A 3D-CFD model for hydrate formation in gas pipelines has been developed.
The results achieved so far are in agreement with existing knowledge about
the condensation and accumulation of liquid water regions due to gravity in

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)
270 Computational Methods in Multiphase Flow VI

hilly terrain pipelines. Furthermore critical regions for hydrate formation could
be identified. Overall very plausible theoretical explanations about the complex
physical phenomenon taking place could be found.
Unfortunately no direct comparison with experimental results due to lack of
data so far could be made. However, this will be subject to a forthcoming paper.
Measurements for model validation in a real hilly terrain pipeline are currently
ongoing. Future work is also aimed at integrating a correct kinetic model to
simulate the effect of hydrate nucleation and growth phenomena on a macroscopic
scale.

References
[1] J.J. Carroll, Natural Gas Hydrate: A Guide for Engineers. Elsevier Science,
2003.
[2] E.D. Sloan & K.A. Koh, Clathrate Hydrates of Natural Gases, 2008.
[3] W.M. Deaton & E.M. Frost, Jr., Gas hydrates and their relation to
the operation of natural-gas pipe lines. Technical report, United States
Department of the Interior, Bureau of Mines, July 1946.
[4] P.Raj Bishoni & V. Natarajan, Formation and decomposition of gas hydrates.
Fluid Phase Equilibria, 117, pp. 168–177, 1996.
[5] C.P. Ribeiro Jr. & P.L.C. Lage, Modelling of hydrate formation kinetics:
State-of-the-art and future directions. Chemical Engineering Science, 63(8),
pp. 2007–2034, 2008.
[6] UBC, https://circle.ubc.ca/bitstream/handle/2429/1133/5434.pdf.
[7] UT, http://www.tuhfp.utulsa.edu/.
[8] S.R. Davies, J.A. Boxall, L.E. Dieker, A.K. Sum, C.A. Koh, E.D. Sloan,
J.L. Creek & Z.-G. Xu, Predicting hydrate plug formation in oil-dominated
flowlines. Journal of Petroleum Science and Engineering, 72(3-4), pp. 302–
309, 2010.
[9] Fluent 6.3 Users Guide.
[10] B.V. Balakin, A.C. Hoffmann & P. Kosinki, Experimental study and
computational fluid dynamics modeling of deposition of hydrate particles
in a pipeline with turbulent water flow. Chemical Engineering Science, 22
November 2010.
[11] S.C.K. De Schepper, G.J. Heynderickx & G.B. Marin, Modeling the
evaporation of a hydrocarbon feedstock in the convection section of a steam
cracker. Computers & Chemical Engineering, 33(1), pp. 122–132, 2009.
[12] A. Alizadehdakhel, M. Rahimi & A.A. Alsairafi, Cfd modeling of flow and
heat transfer in a thermosyphon. International Communications in Heat and
Mass Transfer, 37(3), pp. 312–318, 2010.
[13] Y.F. Makogon, Hydrates of Hydrocarbons. PennWell Books, 1997.
[14] D. Kashchiev and A. Firoozabadi, Nucleation of gas hydrates. Journal of
Crystal Growth, 243(3-4), pp. 476–489, 2002.
[15] P. Mills, Non-Newtonian behaviour of flocculated suspensions. J. Phys. Lett.,
46, pp. L301–L309, 1985.

WIT Transactions on Engineering Sciences, Vol 70, © 2011 WIT Press

www.witpress.com, ISSN 1743-3533 (on-line)