This action might not be possible to undo. Are you sure you want to continue?
Proceedings of OMAE2010
29th International Conference on Offshore Mechanics and Arctic Engineering
June 6 June 11, 2010, Shanghai, China
OMAE201020509
CFD MODELING OF CORRUGATED FLEXIBLE PIPE
Rajeev K. Jaiman
ACUSIM Software, Inc.
Mountain View, CA 94043
Owen H. Oakley, Jr.
Chevron Energy Technology
Company
J. Dean Adkins
Chevron Energy Technology
Company
ABSTRACT
The objective of this paper is to present Computational Fluid
Dynamics (CFD) modeling of fully developed turbulent flow
through a flexible corrugated pipe and to investigate the
pressure drop reduction potential of liners. This work also
aims to establish a framework to be used in large scale
numerical simulations of the offshore transfer of cryogenic
fluids. A 3D CFD approach is considered more appropriate
than 2D axisymmetric one, since the wavy corrugation
profiles lead to a great deal of internal turbulent structures for
high Reynolds number over Re > 10
6
.
Three geometries of the bellows’ (corrugation) depth are
considered to determine the potential value of a cryogenic
liner, corrugation filler or geometric variations for the 16”
pipe. The reduction in cost and complexity of developing a
robust cryogenic liner or corrugation filler, plus eventual
certifications, would be significant and needs to be worth the
improvement (decrease) in pressure drop. We conduct a
straight pipe corrugation depth study for pressure drop (deep
corrugation, shallow corrugation and liner), and include
suitable mesh convergence and unsteady simulations. We also
attempt to validate the friction coefficient data with the
empirical formulas and recent experimental tests. Operational
cryogenic transfer flow rates ranging from Q=1000 m
3
/h to
Q=5000 m
3
/h are considered.
INTRODUCTION
The flexible metal pipe has been used in smaller diameters for
more than 30 years for all kind of cryogenic Liquid Natural
Gas (LNG) transfer applications (Refs. [1,2]). Today these
LNG loading systems have evolved towards large, complex
industrial systems, which have to respect increasingly
stringent rules and standards while continuing to maintain
high levels of safety and availability. For both design and
operational standpoint, the LNG from ship to ship loading is a
new application of this well known technology (Ref. [2]).
Such flexible LNG pipes are usually provided with corrugated
walls. The basic design of a Nexans vacuum insulated LNG
transfer hose is illustrated in Figure 1.
Relatively minor deviations in corrugation geometry can affect
the flow/pressure drop characteristics that are important from
the design and operational standpoint. The relative pressure
drop per unit length in a pipe differs according to the distance
from the inlet, normally related to a distance of some 50 x pipe
inner diameter to achieve a fully developed flow profile. After
this length the flow is normally fully turbulent and the relative
pressure drop per unit length is constant (and lowers than in
the inlet section).
Armoured metallic corrugated pipes are well known structures
which can withstand tensile and internal pressure loads, as
well as perform better from a fatigue and heat transfer
2
standpoints. However, series of corrugations can induce
complex and undesirable flow behavior in the pipes. The
wavy configuration of the corrugations promotes turbulence,
which is a problem because of the extra work required to
surmount the pressure drop. In a broad sense, corrugations
may lead to the following flow physics phenomenon:
Large pressure headloss
Flow induced pulsations
Multiphase with bubbles and cavitations
Increase heat transfer
Figure 1: Nexans vacuum insulated LNG transfer hose
The purpose of the modeling exercise is to assess the pressure
drop and qualify the behavior of the flow in the various
configurations of corrugation geometry. To estimate the
variation of the pressure in the corrugations, we do not model
the phase change and the bubbles cavitations but accurately
evaluate the pressure drop along the pipe. The pressure drop
estimation can be useful to deduce the upstream pressure
which can be imposed to stay everywhere downstream above
the phase change pressure.
The pressure drop and friction factor are related by the
following relationship (Ref. [3]):
2
1
2
4
cross  section area weighted perimeter
h
h
f
p
u
L
D
A
D
C
A C
ρ
 
  ∆
=
 
\ .
\ .
=
=
(1)
, =
where
p
L
∆
denotes pressure drop per unit length L ,
hydraulic pipe diameter
h
D , the density ρ of fluid, the mean
velocity u of flow; dimensionless coefficient of friction f .
In the case of a circular crosssection the hydraulic diameter is
equal to the diameter of the circle.
In this work, the challenge consists in making CFD modeling
accurately describe the turbulent behavior of liquid flow in a
corrugated hose. The macroscopic features as pressure drop
and friction factor in the corrugated pipe are directly related to
the wall shear stress.
To begin, it is useful to summarize briefly the essential
elements of wall turbulence, mainly to establish notation and
define some basic terms. From simple observations, the effect
of turbulence on the mean flow is to flatten the profile relative
to the parabolic profile that occurs in pipe flows. The total
shear stress is sum of the Reynolds stress
uv ρ −
and the
viscous stress
( ) / du dy u
and it is defined as
( ) / du dy uv τ u ρ = − (2)
where u denotes the dynamic viscosity of the fluid.
In fully developed pipe flow in the absence of streamwise
acceleration (i.e., flow no longer changing in streamwise
direction), the mean equation of turbulent motion can be
defined as:
0
p
x y
τ ∂ ∂
=− +
∂ ∂
(3)
Since
p
x
∂
∂
is a function of streamwise direction x and
y
τ ∂
∂
is a function of y alone, both of them must be
constant and the stress distribution is then linearly varying
from the value at the wall w
τ
to zero at the centerline. The
following addresses the problem of developing a CFD model
to obtain an insight into the wall turbulence phenomena and
obtaining values of the pressure gradient
p
x
∂
∂
in the
corrugated pipe.
In corrugated pipe applications, flow physics (e.g.,
recirculation, separation, mean flow threedimensionality,
streamline curvature, flow acceleration) and geometry play an
important role. In this study, we show that CFD modeling can
offer an accurate and powerful predictive tool for estimating
the macroscopic pressure drop and complex flow phenomenon
in the corrugations.
In the following sections we describe the general method of
solving 3D flow equations with turbulence effects. The
problem of creating an optimal mesh is discussed in which the
objective is to combine acceptable solution accuracy with
good solution economy. We then describe simulations results
and assessment of liner designs, and comparing the predicted
pressure drop with the empirical formula and the water test
experiments.
NUMERICAL METHODOLOGY
All of the solutions shown herein are produced using the
AcuSolve
TM
finite element NavierStokes solver based on the
Galerkin/LeastSquares formulation. Although the corrugated
pipe geometry used here has cylindrical symmetry, Cartesian
coordinates are employed; illustrating the potential of the
solver to model complex geometries with large elbows (i.e.
3
catenary with varying curvatures) using unstructured meshes.
The followings are the basic components of CFD modeling.
Turbulence Modeling
The turbulence level is typically high due to the corrugations
and turbulence modeling is critical to get the accurate
predictions. To model the steady effects of the turbulence on
the mean flow field, we employ the SpalartAllmaras
Reynolds Averaged Navier Stokes (RANS) model. This model
is a general purpose model that provides reasonable results for
a broad array of industrial applications. This model relies on
the fact that the complete turbulence behavior has to be
enclosed within an appropriate eddy viscosity variable which
takes into account all turbulence scales (from the largest
eddies to the Kolmogorov scale). This model solves a single
transport partial differential equation (PDE) for the eddy
viscosity with the appropriate boundary conditions and solver
settings.
For unsteady simulations, we employ Delayed Detached Eddy
Simulation (DDES), a hybrid RANS model with Large Eddy
Simulation (LES). In the LES based on dynamic subgrid scale
estimation, an attempt is made to capture the large scale
unsteady motions which carry the bulk of the mass and
momentum in a flow, but the near wall turbulence behavior is
treated with a wall function. In the DDES model (Refs. [6,9]),
we resolve the large eddies that have the biggest effect on the
wall shear stress and use the RANS equations to describe the
flow near the wall. This was done not only to economize on
mesh size, but also because most pipes have relatively rough
walls. Wall functions reduce mesh size by providing an
integrated relationship between the wall and the logarithmic
region of the boundary layer.
Material Model
In all simulations, the working fluid LNG is viscous and in
liquid phase with constant density, that is to say
incompressible flow. The flow is assumed to be isothermal,
i.e., the energy equation is not solved. The flow rate and
Reynolds number based on the averaged velocity are the input
for the models. Table 1 summarizes the material properties of
LNG.
Table 1: Summary of the properties of LNG
Property Value
Density 450 kg m
3
Dynamic viscosity 1.4 x 10
4
Pa sec
Kinematic viscosity 3.11x 10
7
kg m
1
s
1
Solver Settings
The standard solver settings were specified in AcuSolve for
steady RANS calculations. When performing a steady
simulation, the time integration is automatically set to 1st
order accuracy. In this configuration, the timestep size taken
by AcuSolve is set to infinity (1.0x10
10
s) to convect the errors
through the domain and arrive at the steady solution. The mass
continuity and momentum differential equations are then
converged to 4 decades of the solution accuracy. This residual
reduction ensured iterative convergence in all cases.
For unsteady cases, we try to minimize the additional effects
of finite element stabilization in both space and time
integration so that numerics do not suppress the small scales
of turbulent eddy motions. For the spatial integration, we
switch off the discontinuity capturing operator for the DES
model and we set the lumped mass fraction to zero. In time
domain, we utilize an optimal high frequency damping factor
to be 0.5 for the generalized secondorder time integration
scheme. A global time step of 1 x 10
2
sec is used to capture
the unsteady turbulent motions.
Initial Conditions
For steady state simulations, the flow solution for the velocity
vector is initialized by the entrance averaged flow speed and
pressure field is set to be zero. For unsteady calculations, the
flow solution field is initialized by the steady state solution of
the corresponding flow rate. The flow solution is then
advanced in time to obtain the unsteady turbulent motions
with fixed and small timeincrement to resolve the unsteady
local motions.
Boundary Conditions
In various industrial applications, the flow physics and
geometry can be modeled as repetitive in nature by applying
periodicity. Thereby, a representative building block can be
considered for computational efficiency and simplicity, while
maintaining the desired accuracy of flow physics
corresponding to the experiments. In this purpose, the flow
profiles at exit are iteratively reported to the entrance (Ref.
[7]). In AcuSolve, to simulate the large length of corrugated
hose with fully developed flow, periodic conditions are
applied between the outlet (exit) and inlet (entrance) of the
domain.
The quantity imposed at the entrance is the flow rate Q or
equivalently the averaged velocity as the surface integrated
boundary conditions (bulk BC). This is the only user input to
define the fully developed flow and Reynolds number based
on the averaged velocity. Instead of outlet condition at the exit
plane, the classical periodicity on the mean velocity and the
eddy viscosity are used to couple the entrance and the exit
boundaries. By this way, we attempt to achieve the similar
stabilized profiles of mean velocity and eddy viscosity at the
entrance and the exit. For the pressure, the condition based on
the constant offset is imposed to obtain the fully developed
like condition. In other words, the pressure can change along
the streamwise direction and the pressure can be decomposed
4
into a variable term and a linear varying term in the
streamwise direction as
dp
p p x
dx
= − (4)
where the fluctuating terms, p identically repeat in the
periodic direction. The fully developed velocity profile
typically converges to a turbulent parabolictype profile along
the iterations as it is known, except at the vicinity of the
corrugation. By applying these BCs, the SA turbulence model
is tied up with the fully developed NavierStokes flow solution
and the coupled mass, momentum and eddy equations are
wellposed and complete.
NUMERICAL MODELING
Model Description
The first step in generating the mesh for the corrugations is to
define the domain of interest. The actual corrugated hoses
have a length of several tenths of meters, i.e., L/D ~ O (100)
and consequently, it is not needed to model their full length in
this validation exercise. It is desirable to model the shorter
relevant length to save the calculations and capture the fully
developed turbulent flow. Table 2 summarizes the relevant
dimensions of the corrugation profiles. The assumption of 2D
axisymmetric may not be sufficient for the accurate modeling
of three dimensional turbulence effects at high Reynolds
number. Therefore, we consider a 3D geometry with a
circular shape of corrugation profile (i.e., without any
helical/spiral effects) as shown in Fig. 2.
Table 2: Dimensions of corrugation profiles
Geometry
ID
(m)
Pitch, P
(m)
Amplitude,
A (m)
A*=
A/(ID+A)
Base 0.404 0.041 0.02600 0.06047
Line #1 0.404 0.041 0.00650 0.01583
Liner #2 0.404 0.041 0.00325 0.00798
Mesh Generation
Obtaining an economical solution is then simply finding the
minimum number of degrees of freedom in the flow solution
that captures the salient fluid flow effects and predicts the
pressure drop and wall shear stress accurately. We consider
the length of 3D flow domain with L = 6D matching earlier
work on the direct numerical simulations of fully developed
pipe flow (Ref. [4]). We consider the fully developed turbulent
flow in the corrugated hose with the three configurations with
varying depths A*. With the CAD models shown, we
discretize the geometry to form a grid that is reasonable to
capture the dominant flow features. In particular, we need to
pay attention in the meshing near the wall of corrugations to
resolve turbulent boundary layer and to find the optimized
volumetric mesh distribution.
Figure 2: Three model geometries for corrugated hoses:
Base model with A*=0.06 (top), Liner #1 model with
A*=0.01583 (middle), Liner #2 with A*=0.00798 (bottom)
Figure 3 shows the typical mesh distribution at a cross
sectional plane for the base model. The model has a fine mesh
resolution near the wall and a gradual coarsening of the mesh
away from the wall (larger elements in the core region of the
pipe).
Figure 3: Typical mesh distributions (a) full domain cross
sectional mesh (b) closeup mesh
An important quality control practice while performing
numerical analysis is to determine the influence of the
(a)
(b)
5
discretization parameters and to perform sensitivity analyses.
For CFD, this typically entails a mesh refinement study as
well as a near wall modeling sensitivity analysis. A reasonable
mesh sensitivity analysis was performed for this work.
An important issue in the accurate prediction of industrial
turbulent flows is the formulation and the numerical treatment
of equations in regions close to solid walls. The nearwall
formulation determines the accuracy of the wall shear stress
(friction factor) and has an important influence on the
development of boundary layers and near wall turbulence
structures. Typically the two following approaches are used to
model the flow in the nearwall region: (a) the wall function
method (b) the wall integration (low Reynolds number)
method.
Because of the linear variation of total stress, the wall shear
stress and the Reynolds (turbulent) stresses are related,
justifying the introduction of the friction velocity
w
u
τ
τ
ρ
=
as a scale representative of the turbulent
fluctuations. The viscous length scale alluded to the viscous
boundary layer thickness is given by
v
u
τ
ν
δ = and the
distance above the wall scaled in wall units is denoted by
v
y
y
δ
+
= (5)
where y is the distance from the wall and
u
ν
ρ
=
is
kinematic viscosity.
Table 3: Mesh statistics for the base design model A*=0.06
Mesh Fine mesh (Ref.) Medium mesh
(wall function)
Total no of elements 51,468,929 23,293,025
Number of nodes 8,862,821 4,025,153
The reasonable mesh sensitivity implies that we reduce the
mesh about a factor of two (instead of reducing uniformly in
all the directions which is equivalent to factor of 8). Tables 3
5 show the mesh statistics for the three pipe models with y+ <
10. Two successive refined grids were created for the 3D
corrugated pipe model to perform a volumetric grid
refinement study using the wall function approach. The
results of the mesh generation study showed little sensitivity
(< 2%) to the density of the volume mesh for the high
Reynolds number Re=14.1 x 10
6
. The numerical results
presented for the RANS and DES models in the following
sections are those on the fine reference mesh, unless noted
otherwise.
Table 4: Mesh statistics for the liner #1 design model,
A*=0.01583
Mesh Fine mesh (Ref.) Medium mesh
(wall function)
Total no of elements 35,368,685 19,882,402
Number of nodes 6,113,454 3,429,323
Table 5: Mesh statistics for the liner #2 design model,
A*=0.00798
Mesh Fine mesh (Ref.) Medium mesh
(wall function)
Total no of elements 33,107,421 18,780,090
Number of nodes 5,728,804 3,240,773
SIMULATION RESULTS AND VISUALIZATIONS
We present fullscale 3D simulations using the steady RANS
and transient DDES models for the three configurations for the
range of flow rates.
Base Corrugated Model
We first perform 3D RANS simulations using the wall
function approach with the fine mesh. As mentioned earlier,
the length of modeled domain is 6D as shown Fig. 3.
Figure 4. Streamwise variation of velocity magnitude
contours in the corrugated pipe at Q=3333 m
3
/h for the
base model RANS (top) and DDES (bottom)
6
Figures 4 (top) shows the contours of velocity magnitude (top)
using the RANS model at the Reynolds number of Re=9.38 x
10
6
for the base model of corrugated pipe. The fully developed
and time averaged steady flow behavior can be observed from
the figure. As expected from the RANS model, there are no
physical unsteady motions in the velocity field. Figures 4
(bottom) shows the contours of streamwise velocity at the
cross section of the corrugated pipe with the DDES model.
The 3D turbulence structures and unsteadiness in the flow can
clearly be inferred in the image
Figure 5 shows the contours of crossstream velocity
magnitude at the three cross section planes of the corrugated
pipe. Significant circumferential variations in the velocity
magnitude can be seen in the figure. These local variations are
coupled with vorticity, which is defined as the rotation of the
velocity field.
Figure 5. Instantaneous velocity magnitude contours at
the crosssectional planes for flow rate Q=3333 m
3
/h (Re =
9.38 x 10
6
)
Figure 6 shows complex 3D turbulent structures of lowspeed
streaks and inplane streamwise vortices. These vortex
topologies are identified through the Qcriterion which defines
a vortex as a spatial region where
( ) ( )
2 2 1
0,
2
1 1
: ,
2 2
T T
Q
where
= − >
= ∇ + ∇ = ∇ − ∇
Ω S
S v v Ω v v
The interactions between these structures are essential
ingredients of wallbounded turbulence. These fluctuating
vorticity structures are somewhat organized, often called
vortical flow structures.
Figure 6. Isosurface of vorticity variable (Qcriterion)
colored by velocity magnitude
Figure 7 shows the quantification of instantaneous variation of
velocity field in the core flow. The velocity fluctuations are
normalized by the mean inlet velocity. The velocity
fluctuations suggest the maximum turbulence level up to 20 %
at the probe point in the core flow.
3.00E01
2.00E01
1.00E01
0.00E+00
1.00E01
2.00E01
3.00E01
1.0E+00 1.2E+00 1.4E+00 1.6E+00 1.8E+00 2.0E+00
Normalized time
N
o
r
m
a
l
i
z
e
d
v
e
l
o
c
i
t
y
f
l
u
c
t
u
a
t
i
o
n
s
v'/U
w'/U
u'/U
Figure 7. Instantaneous velocity fluctuations at a sample
location in the domain
Liner Based Corrugated Models
In this section, we conduct a simple straight pipe corrugation
depth study with the DES model. The meshing guidelines and
flow conditions are similar to those used for the base model.
On further flow visualizations, Figs. 89 show comparison of
velocity magnitude at the crosssectional and streamwise
planes. In the images, we can see a greater degree of
turbulence structures (red color zone) in the liner 1 model as
compared to the liner 2 model.
7
1.5E01
1.0E01
5.0E02
0.0E+00
5.0E02
1.0E01
1.5E01
1.00E+00 1.10E+00 1.20E+00 1.30E+00 1.40E+00 1.50E+00
Normalized time
N
o
r
m
a
l
i
z
e
d
v
e
l
o
c
i
t
y
f
l
u
c
t
u
a
t
i
o
n
s
v'/U
w'/U
u'/U
1.5E01
1.0E01
5.0E02
0.0E+00
5.0E02
1.0E01
1.5E01
1.00E+00 1.10E+00 1.20E+00 1.30E+00 1.40E+00 1.50E+00
Normalized time
N
o
r
m
a
l
i
z
e
d
v
e
l
o
c
i
t
y
f
l
u
c
t
u
a
t
i
o
n
s
v'/U
w'/U
u'/U
Figure 8. Streamwise variation of velocity magnitude
contours for the liner 1 (top) and liner 2 (bottom) models
at Q=3333 m
3
/h
Figure 9. Instantaneous velocity magnitude contours to
show turbulence spatial structures at the crosssectional
planes for the two liner models for flow rate Q=3333 m
3
/h
(Re = 9.38 x 10
6
)
Figure 10 shows the quantification of instantaneous variation
of velocity field for the two liner models as a function of the
time in a single point, i.e. liner 1 on top and liner 2 below.
This is equivalent to obtaining the turbulence statistics from
the singlepoint probe. The velocity fluctuations are
normalized by the mean velocity. The liner 2 model decreases
the velocity fluctuations (i.e., turbulence level) by a factor of 5
compared to the liner 1 model.
Figure 10. Instantaneous velocity fluctuations at the same
singlepoint location for the liner 1 (top) and liner 2
(bottom)
ASSESSMENT AND DISCUSSION
In this section, we want to quantify the pressure loss and
friction factor for the three geometries. Since there are no
experimental test data for these geometries, we first attempt to
compare the CFD results with the classical roughness theory
and 10.5” (Ref. 3).
Hydraulically smooth pipe regime: From the universal
law of friction for a smooth pipe, the friction factor can be
expressed as
2.0 log 0.8
h
uD
f
f υ
 
=

\ .
1
 (7)
This is Prandtl’s universal law of friction for smooth
pipes and it has been verified with experiments and the
agreement is seen to be excellent up to the Reynolds
number Re
D
= uD
h
/ν=3.2x10
6
.
8
Completely rough regime: In the present configuration,
the corrugations may be considered equivalent to periodic
roughness (i.e., 2k/D
h
) of the diameter of pipe. A
systematic investigation on the effects of Reynolds
number and relative roughness k on the friction factor was
performed by Nikuradse Ref. [5]). For the higher values
of Reynolds number (ranging from 10
4
to 10
7
), the
friction factor was found to be Reynolds number
independent. The theoretical estimate for the friction
factor for the completely rough regime was estimated as
2
1
1.74 2 log
2
h
f
D
k
=
   
+
 
\ . \ .
(8)
Base Corrugated vs. Liner Models
Figure 11 shows the variation of coefficient of friction for the
range of Reynolds number for the three configurations and the
smooth pipe. The friction factor was determined by evaluating
the pressure gradient along the pipe from the integrated
pressure values. For the baseline case, the friction coefficient
is consistently larger than the liner 1 & liner 2 models. The
wall shear stress of the liner 2 model is converging towards
the stress values corresponding to the smooth pipe. This
implies that, by introducing liner materials, the coefficient of
friction can be reduced by 80% with respect to the deeper
metallic hose configuration. Due to complex flow behavior
and recirculation in the base & liner 1 models, the friction
factor changes significantly with the Reynolds numbers.
For the base and liner 1 geometry at Re~10M, an inflectional
behavior in the pressure drop and wall shear stress have been
observed in the RANS and DDES results. This may be
explained in the context of boundary layer separations along
the corrugation profile and the associated effects into the
pressure gradients. In other words, when the imparted shear
stress dominates eddy dissipative scale, a large amount of
friction occurs along the wall or vice versa. The CFD results
follow an inflectional friction factor relationship rather than
the monotonic relationship given by the roughness theory and
the Moody diagram. Figure 11 also presents the roughness
theory predictions given by the lines. For the smooth pipe, the
CFD results and the theory have an excellent match. However,
for the corrugated shapes the roughness theory seems to differ
up to 24%.
0.00E+00
2.50E02
5.00E02
7.50E02
1.00E01
1.25E01
1.0E+05 1.0E+06 1.0E+07 1.0E+08
Reynolds number, Re
C
o
e
f
f
i
c
i
e
n
t
o
f
F
r
i
c
t
i
o
n
Baseline
Liner 1
Liner 2
Smooth
Base Rough. Theory
Liner1 Rough. Theory
Smooth Pipe Theory
Figure 11. Variation of friction coefficients with Reynolds
number
Figure 12 shows a cross plot of friction factor vs. the depth of
corrugation A*, which provides another view of the impact of
reducing the corrugation depth. For the range of flow rates,
the friction factor increases as we increase the depth of
corrugation. This suggests the potential value of liners for
reducing the pressure loss. The friction factor values for the
liner 1 are converging to that of the smooth pipe. The amount
decrease doesn’t appear to be simply a linear function.
0
0.025
0.05
0.075
0.1
0.125
0.0E+00 2.0E02 4.0E02 6.0E02 8.0E02
Ampititude/Diameter, A*
C
o
e
f
f
i
c
i
e
n
t
o
f
F
r
i
c
t
i
o
n
Flow rate Q=5000 m^3/hr
Flow rate Q=3333 m^3/hr
Flow rate Q=2222 m^3/hr
Figure 12. Variation of friction coefficients with respect to
the depth of corrugation A*
Steady RANS vs. DES Simulations
To assess the results of steady RANS with the DDES, we
further conduct a comparative study on the same grids. Figure
13 shows a summary of the friction factor computed based on
the pressure drop for the two CFD models. A reasonable
consistency in the predictions of integrated pressure drop can
be seen in the figure. By tuning the grid distributions, an
9
improved match between the RANS and DDES may be
obtained.
For the base and liner 1 geometry at Re~10M, an inflectional
behavior in the pressure drop and wall shear stress have been
observed in the RANS and DDES results. As mentioned
earlier, this dip in the frictional drag may be attributed to the
sudden shift in the point of separation for the base and liner 1
geometries. In this range, the laminar viscous sublayer
portion of boundary layer becomes unstable and undergoes
transition to turbulence. As seen in the flow over a cylinder,
the turbulent boundary layer, because of its greater energy, is
able to overcome a large adverse pressure gradient. The
turbulent boundary layer separates at a further downstream
location along the corrugation profile, resulting in a thinner
wake and a pressure distribution more similar to that of
potential flow. For values of Re >10M, the separation point
slowly moves upstream as the Reynolds number is increased,
resulting in an increase of the friction factor. For the liner 2
and smooth pipe, the geometry is streamlined and the point of
separation and the transition of boundary layer remain
somewhat unchanged.
0.00E+00
2.50E02
5.00E02
7.50E02
1.00E01
1.25E01
1.0E+06 1.0E+07 1.0E+08
Reynolds number, Re
C
o
e
f
f
i
c
i
e
n
t
o
f
F
r
i
c
t
i
o
n
Baseline
Liner 1
Liner 2
Smooth
BaselineDES
Liner1DES
Liner2DES
Figure 13. Variation of friction coefficients for the range of
Reynolds number for the RANS and DDES models
We also present a comparison with the experimental test, and
perform further analysis of the results to facilitate
interpretation in the commercial settings. The experiment tests
were performed on a 268 mm (10.5”) ID pipe for a range of
flow rates with fresh water as working fluid. All the pressure
drop readings were average values taken over a period of 1
minute
Figure 14 shows the comparison of CFD values with the
experimental test done with water in 10.5” ID pipe (Ref. [1]).
The friction factors are compared with respect the non
dimensional dynamic similarity parameter, Reynolds number.
The depth and shape of the corrugation profiles are marginally
different between the 16” ID pipe and 10.5” pipe. A
reasonable agreement between the CFD and experimental
values can be seen.
4.00E02
6.00E02
8.00E02
1.00E01
1.20E01
1.0E+05 1.0E+06 1.0E+07 1.0E+08
Reynolds number, Re
C
o
e
f
f
i
c
i
e
n
t
o
f
F
r
i
c
t
i
o
n
Base CFD Model
Liner 1 CFD Model
Water Test (Ref. [1])
Figure 14. Comparison of the CFD results for A*=0.0604
of 16” ID pipe with the water test with A*=0.0513 in 10.5”
ID pipe
Next, we would like to establish the relevance of friction
factor in actual pipe configurations as function of pipe lengths
and corrugation depths. This may be important to estimate the
capacity and limitations of LNG pumps for the corrugated
pipe flow. Typical LNG pumps can develop 6.6 bar
differential pressure (97.3 psi) while maintaining reasonable
flow rates for LNG transfer. Figure 15 shows the comparison
of pressure drop as a function of flow rates for the geometries,
where the pressure drop is computed based on the Equation
(1).
Pressure Required vs. Flow Rate for L=100 [m]
0
10
20
30
40
50
60
‐ 2,000 4,000 6,000 8,000 10,000 12,000
Flow Rate Q [m^3/hr]
P
r
e
s
s
u
r
e
D
r
o
p
[
b
a
r
]
Base
Li ner 1
Li ner 2
Figure 15. Pressure drop as a function of flow rates for the
16” pipe with three corrugation geometries
10
CONCLUSIONS
Cryogenic flexible pipe based LNG transfer system seems to
be a good candidate for CFD modeling, and to qualify the pipe
system for the LNG industry requirements.
Significant 3D turbulence effects were found for the pipe
geometry with circular corrugations suggested by both
qualitative features and quantitative information. The 3D
steady RANS and DDES models provided a consistent
estimate of the pressure drop and friction factor for varying
flow rates. We hoped that we have done a reasonable job in
predicting complex internal turbulent structures with the
unsteady DDES simulations.
The pressure drop results were found to be quite sensitive to
the corrugation depth. By introducing the liner materials, the
coefficient of friction can be reduced by 80% with respect to
the deeper metallic hose configuration. The CFD results
appear to be sufficiently accurate that one might seriously
consider using such models to investigate relative geometric
differences and perform parametric studies of various
corrugation configurations. The modeling can reduce the
uncertainty, offer guidance on design variations, improve the
design of full scale tests and potentially eliminate some or all
of such tests. CFD can add substantially to classical roughness
theory (which relies on known pipe characteristics) as it can
address different geometries and dimensions without
additional empirical data.
REFERENCES
[1] Framo Engineering AS Report, “CFD Calculations of
Corrugated Flexible Pipe,” 45770313D, 2006.
[2] Frohne, C., Harten, F., Schippl, K., Steen, K.E.,
Haakonsen, R., Jorgen, E. and Høvik, J. “Innovative Pipe
System for Offshore LNG Transfer,” OTC 19239, 2008.
[3] Schlichting, H., “Boundary Layer Theory,” 7th Edition,
McGrawHill Book Company, 1975.
[4] Eggles, J. et al.,”Fully devloped turbulent pipe flow: a
comparison between direct numerical simulation and
experiment”, Journal of Fluid Mechanics, 268, 1994, 175207
[5] Nikuradse., ”Laws of flow in rough pipes”, NACA TM
1292, 1933
[6] Spalart, Deck, Shur, Squires, Streletes & Travin, “A New
Version of DetachedEddy Simulation, Resistant to
Ambiguous Grid Densities,” Journal of Theoretical &
Computational Fluid Dynamics, 20, 181195, 2006.
[7] Patankar, S.V., Liu, C.H., and Sparrow, E.M., “Fully
developed flow and heat transfer in ducts having streamwise
periodic variations of crosssectional area”, ASME, J. Heat
Transfer, Vol. 99, pp. 180186
[8] Piomelli U. and Balaras, E. “Annual Review of Fluid
Mechanics”, Vol 34, pp 349374, 2002
[9] Shur, M.K., Spalart, P.R., Strelets, M.K. and Travin,
A.K.,”A hybrid RANSLES approach with delayedDES and
wallmodeled LES capabilities”, International Journal of Heat
and Fluid Flow, Vol 29, pp 16381640, 2008.
[10] Allen, J.J, Shockling, M.A., Kunkel, G.J. and Smits,
“Turbulent flow in smooth and rough pipes”, Philosophical
Transactions of the Royal Society, vol. 365, No. 1852, 2007
The following addresses the problem of developing a CFD model to obtain an insight into the wall turbulence phenomena and obtaining values of the pressure gradient corrugated pipe. it is useful to summarize briefly the essential elements of wall turbulence. mainly to establish notation and define some basic terms.standpoints. The total viscous stress µ shear stress is sum of the Reynolds stress − ρ uv and the τ = µ ( du / dy) − ρuv ( du / dy ) and it is defined as (2) where µ denotes the dynamic viscosity of the fluid. streamline curvature. separation.g. In this work. the mean equation of turbulent motion can be defined as: ∂p ∂x in the where ∆p denotes pressure drop per unit length L .e. In a broad sense. In the case of a circular crosssection the hydraulic diameter is equal to the diameter of the circle. C = weighted perimeter L (1) ∂p ∂τ + (3) ∂x ∂y ∂p is a function of streamwise direction x and Since ∂x ∂τ is a function of y alone. and comparing the predicted pressure drop with the empirical formula and the water test experiments. the effect of turbulence on the mean flow is to flatten the profile relative to the parabolic profile that occurs in pipe flows.. In this study. To estimate the variation of the pressure in the corrugations. flow acceleration) and geometry play an important role. which is a problem because of the extra work required to surmount the pressure drop. The wavy configuration of the corrugations promotes turbulence. However. We then describe simulations results and assessment of liner designs. The pressure drop estimation can be useful to deduce the upstream pressure which can be imposed to stay everywhere downstream above the phase change pressure. dimensionless coefficient of friction f . Although the corrugated pipe geometry used here has cylindrical symmetry. The pressure drop and friction factor are related by the following relationship (Ref. Figure 1: Nexans vacuum insulated LNG transfer hose The purpose of the modeling exercise is to assess the pressure drop and qualify the behavior of the flow in the various configurations of corrugation geometry. From simple observations. we do not model the phase change and the bubbles cavitations but accurately evaluate the pressure drop along the pipe. flow physics (e.e.. 2 . [3]): ∆p f 1 2 = ρu Dh 2 4A Dh = C A = cross . The macroscopic features as pressure drop and friction factor in the corrugated pipe are directly related to the wall shear stress. the mean velocity u of flow. Cartesian coordinates are employed. The problem of creating an optimal mesh is discussed in which the objective is to combine acceptable solution accuracy with good solution economy. the challenge consists in making CFD modeling accurately describe the turbulent behavior of liquid flow in a corrugated hose. recirculation. the density ρ of fluid. NUMERICAL METHODOLOGY All of the solutions shown herein are produced using the AcuSolveTM finite element NavierStokes solver based on the Galerkin/LeastSquares formulation. flow no longer changing in streamwise direction). In the following sections we describe the general method of solving 3D flow equations with turbulence effects. illustrating the potential of the solver to model complex geometries with large elbows (i. In corrugated pipe applications.section area. L hydraulic pipe diameter Dh . mean flow threedimensionality. constant and the stress distribution is then linearly varying In fully developed pipe flow in the absence of streamwise acceleration (i. both of them must be ∂y 0=− from the value at the wall τ w to zero at the centerline. corrugations may lead to the following flow physics phenomenon: Large pressure headloss Flow induced pulsations Multiphase with bubbles and cavitations Increase heat transfer To begin. series of corrugations can induce complex and undesirable flow behavior in the pipes. we show that CFD modeling can offer an accurate and powerful predictive tool for estimating the macroscopic pressure drop and complex flow phenomenon in the corrugations.
Table 1: Summary of the properties of LNG Property Density Dynamic viscosity Kinematic viscosity Solver Settings The standard solver settings were specified in AcuSolve for steady RANS calculations. the working fluid LNG is viscous and in liquid phase with constant density. we switch off the discontinuity capturing operator for the DES model and we set the lumped mass fraction to zero. In AcuSolve. In the LES based on dynamic subgrid scale estimation. we employ the SpalartAllmaras Reynolds Averaged Navier Stokes (RANS) model. A global time step of 1 x 102 sec is used to capture the unsteady turbulent motions. we utilize an optimal high frequency damping factor to be 0. Thereby. the condition based on the constant offset is imposed to obtain the fully developed like condition. Table 1 summarizes the material properties of LNG. For unsteady simulations. the energy equation is not solved. This was done not only to economize on mesh size. To model the steady effects of the turbulence on the mean flow field. In other words. In this configuration. For the pressure.11x 107 kg m1 s1 simulation. Wall functions reduce mesh size by providing an integrated relationship between the wall and the logarithmic region of the boundary layer. a representative building block can be considered for computational efficiency and simplicity. For unsteady cases. For the spatial integration. Boundary Conditions In various industrial applications. [6. we try to minimize the additional effects of finite element stabilization in both space and time integration so that numerics do not suppress the small scales of turbulent eddy motions. In the DDES model (Refs. [7]). Instead of outlet condition at the exit plane.5 for the generalized secondorder time integration scheme. This model relies on the fact that the complete turbulence behavior has to be enclosed within an appropriate eddy viscosity variable which takes into account all turbulence scales (from the largest eddies to the Kolmogorov scale). the classical periodicity on the mean velocity and the eddy viscosity are used to couple the entrance and the exit boundaries. to simulate the large length of corrugated hose with fully developed flow. Material Model In all simulations. The flow rate and Reynolds number based on the averaged velocity are the input for the models. but the near wall turbulence behavior is treated with a wall function. the pressure can change along the streamwise direction and the pressure can be decomposed 3 . In this purpose.catenary with varying curvatures) using unstructured meshes. the flow solution field is initialized by the steady state solution of the corresponding flow rate. the flow solution for the velocity vector is initialized by the entrance averaged flow speed and pressure field is set to be zero. The flow solution is then advanced in time to obtain the unsteady turbulent motions with fixed and small timeincrement to resolve the unsteady local motions.9]). periodic conditions are applied between the outlet (exit) and inlet (entrance) of the domain. In time domain. This is the only user input to define the fully developed flow and Reynolds number based on the averaged velocity. Turbulence Modeling The turbulence level is typically high due to the corrugations and turbulence modeling is critical to get the accurate predictions.. the timestep size taken by AcuSolve is set to infinity (1. we attempt to achieve the similar stabilized profiles of mean velocity and eddy viscosity at the entrance and the exit. This model is a general purpose model that provides reasonable results for a broad array of industrial applications. This residual reduction ensured iterative convergence in all cases. The flow is assumed to be isothermal.e. while maintaining the desired accuracy of flow physics corresponding to the experiments. that is to say incompressible flow. the time integration is automatically set to 1st order accuracy. the flow profiles at exit are iteratively reported to the entrance (Ref. The followings are the basic components of CFD modeling. i. but also because most pipes have relatively rough walls. we employ Delayed Detached Eddy Simulation (DDES). The mass continuity and momentum differential equations are then converged to 4 decades of the solution accuracy. an attempt is made to capture the large scale unsteady motions which carry the bulk of the mass and momentum in a flow. we resolve the large eddies that have the biggest effect on the wall shear stress and use the RANS equations to describe the flow near the wall. a hybrid RANS model with Large Eddy Simulation (LES). the flow physics and geometry can be modeled as repetitive in nature by applying periodicity. Initial Conditions For steady state simulations.4 x 104 Pa sec 3. The quantity imposed at the entrance is the flow rate Q or equivalently the averaged velocity as the surface integrated boundary conditions (bulk BC). This model solves a single transport partial differential equation (PDE) for the eddy viscosity with the appropriate boundary conditions and solver settings. For unsteady calculations. By this way. When performing a steady Value 450 kg m 3 1.0x1010 s) to convect the errors through the domain and arrive at the steady solution.
041 0. p identically repeat in the periodic direction.404 0. With the CAD models shown.. Liner #2 with A*=0.404 Pitch. we need to pay attention in the meshing near the wall of corrugations to Figure 3: Typical mesh distributions (a) full domain crosssectional mesh (b) closeup mesh An important quality control practice while performing numerical analysis is to determine the influence of the 4 .041 Amplitude. without any helical/spiral effects) as shown in Fig.00798 Figure 2: Three model geometries for corrugated hoses: Base model with A*=0.01583 0. (a) (b) Mesh Generation Obtaining an economical solution is then simply finding the minimum number of degrees of freedom in the flow solution that captures the salient fluid flow effects and predicts the pressure drop and wall shear stress accurately. we consider a 3D geometry with a circular shape of corrugation profile (i. The fully developed velocity profile typically converges to a turbulent parabolictype profile along the iterations as it is known. we discretize the geometry to form a grid that is reasonable to capture the dominant flow features.404 0. It is desirable to model the shorter relevant length to save the calculations and capture the fully developed turbulent flow. We consider the fully developed turbulent flow in the corrugated hose with the three configurations with varying depths A*. momentum and eddy equations are wellposed and complete. NUMERICAL MODELING Model Description The first step in generating the mesh for the corrugations is to define the domain of interest. A (m) 0.00798 (bottom) Figure 3 shows the typical mesh distribution at a crosssectional plane for the base model. The actual corrugated hoses have a length of several tenths of meters. The model has a fine mesh resolution near the wall and a gradual coarsening of the mesh away from the wall (larger elements in the core region of the pipe). The assumption of 2D axisymmetric may not be sufficient for the accurate modeling of three dimensional turbulence effects at high Reynolds number.into a variable term and a linear varying term in the streamwise direction as resolve turbulent boundary layer and to find the optimized volumetric mesh distribution. p = p− dp x dx (4) where the fluctuating terms. Liner #1 model with A*=0. 2. except at the vicinity of the corrugation.01583 (middle). L/D ~ O (100) and consequently. Table 2 summarizes the relevant dimensions of the corrugation profiles. it is not needed to model their full length in this validation exercise. [4]).e. i.. the SA turbulence model is tied up with the fully developed NavierStokes flow solution and the coupled mass.e.041 0.00325 A*= A/(ID+A) 0. Table 2: Dimensions of corrugation profiles Geometry Base Line #1 Liner #2 ID (m) 0. Therefore. By applying these BCs. We consider the length of 3D flow domain with L = 6D matching earlier work on the direct numerical simulations of fully developed pipe flow (Ref.06047 0. P (m) 0.02600 0.00650 0.06 (top). In particular.
685 6.) 33.discretization parameters and to perform sensitivity analyses.429.804 Medium mesh (wall function) 18. The nearwall formulation determines the accuracy of the wall shear stress (friction factor) and has an important influence on the development of boundary layers and near wall turbulence structures. unless noted otherwise.421 5. Because of the linear variation of total stress.) 35.780.107. ν =µρ is Table 3: Mesh statistics for the base design model A*=0.454 Medium mesh (wall function) 19. the length of modeled domain is 6D as shown Fig. Tables 35 show the mesh statistics for the three pipe models with y+ < 10.293.882. Base Corrugated Model We first perform 3D RANS simulations using the wallfunction approach with the fine mesh.113.1 x 106. Streamwise variation of velocity magnitude contours in the corrugated pipe at Q=3333 m3/h for the base model RANS (top) and DDES (bottom) 5 .025 4. A*=0.153 The reasonable mesh sensitivity implies that we reduce the mesh about a factor of two (instead of reducing uniformly in all the directions which is equivalent to factor of 8).929 8. The viscous length scale alluded to the viscous boundary layer thickness is given by δv = ν u τ and the distance above the wall scaled in wall units is denoted by y+ = y δv (5) where y is the distance from the wall and kinematic viscosity. 3.402 3. justifying the introduction of the friction velocity Table 4: Mesh statistics for the liner #1 design model. An important issue in the accurate prediction of industrial turbulent flows is the formulation and the numerical treatment of equations in regions close to solid walls.773 uτ = τw ρ as a scale representative of the turbulent SIMULATION RESULTS AND VISUALIZATIONS We present fullscale 3D simulations using the steady RANS and transient DDES models for the three configurations for the range of flow rates. Two successive refined grids were created for the 3D corrugated pipe model to perform a volumetric grid refinement study using the wall function approach. The numerical results presented for the RANS and DES models in the following sections are those on the fine reference mesh.090 3. this typically entails a mesh refinement study as well as a near wall modeling sensitivity analysis. As mentioned earlier.025.368. Typically the two following approaches are used to model the flow in the nearwall region: (a) the wall function method (b) the wall integration (low Reynolds number) method.) 51.468.240.06 Mesh Total no of elements Number of nodes Fine mesh (Ref.821 Medium mesh (wall function) 23. fluctuations. the wall shear stress and the Reynolds (turbulent) stresses are related.01583 Mesh Total no of elements Number of nodes Fine mesh (Ref.728. For CFD.00798 Mesh Total no of elements Number of nodes Fine mesh (Ref. A*=0.323 Table 5: Mesh statistics for the liner #2 design model. Figure 4. A reasonable mesh sensitivity analysis was performed for this work.862. The results of the mesh generation study showed little sensitivity (< 2%) to the density of the volume mesh for the high Reynolds number Re=14.
0E+00 v'/U w'/U u'/U Figure 5.8E+00 2. Figs.00E01 3. we conduct a simple straight pipe corrugation depth study with the DES model. 2 1 1 T T where : S = ∇v + ( ∇v ) . The meshing guidelines and flow conditions are similar to those used for the base model.00E+00 1. 3. often called vortical flow structures. In the images. Instantaneous velocity magnitude contours at the crosssectional planes for flow rate Q=3333 m3/h (Re = 9. Figure 6.00E01 1.0E+00 Figure 7. On further flow visualizations. The 3D turbulence structures and unsteadiness in the flow can clearly be inferred in the image Figure 5 shows the contours of crossstream velocity magnitude at the three cross section planes of the corrugated pipe.2E+00 1. The velocity fluctuations suggest the maximum turbulence level up to 20 % at the probe point in the core flow.00E01 2. These local variations are coupled with vorticity. there are no physical unsteady motions in the velocity field. 6 . Ω = ∇v − ( ∇v ) 2 2 Normalized time 1. The velocity fluctuations are normalized by the mean inlet velocity. Isosurface of vorticity variable (Qcriterion) colored by velocity magnitude Figure 7 shows the quantification of instantaneous variation of velocity field in the core flow. As expected from the RANS model. Significant circumferential variations in the velocity magnitude can be seen in the figure. we can see a greater degree of turbulence structures (red color zone) in the liner 1 model as compared to the liner 2 model.00E01 1. which is defined as the rotation of the velocity field.Figures 4 (top) shows the contours of velocity magnitude (top) using the RANS model at the Reynolds number of Re=9. 89 show comparison of velocity magnitude at the crosssectional and streamwise planes.6E+00 1. These vortex topologies are identified through the Qcriterion which defines a vortex as a spatial region where 1.4E+00 Q= 1 2 2 Ω − S > 0. The interactions between these structures are essential ingredients of wallbounded turbulence. The fully developed and time averaged steady flow behavior can be observed from the figure. Instantaneous velocity fluctuations at a sample location in the domain Liner Based Corrugated Models In this section.38 x 106) Figure 6 shows complex 3D turbulent structures of lowspeed streaks and inplane streamwise vortices. Figures 4 (bottom) shows the contours of streamwise velocity at the cross section of the corrugated pipe with the DDES model.00E01 Normalized velocity fluctuations 2. These fluctuating vorticity structures are somewhat organized.00E01 0.38 x 106 for the base model of corrugated pipe.
This is equivalent to obtaining the turbulence statistics from the singlepoint probe.e.5” (Ref.30E+00 Normalized time v'/U w'/U u'/U 1.5E01 1. 1. 7 .0E01 1. Since there are no experimental test data for these geometries.40E+00 1.. 1 uD = 2. Hydraulically smooth pipe regime: From the universal law of friction for a smooth pipe.0E01 5.38 x 106) Figure 10 shows the quantification of instantaneous variation of velocity field for the two liner models as a function of the time in a single point. Instantaneous velocity magnitude contours to show turbulence spatial structures at the crosssectional planes for the two liner models for flow rate Q=3333 m3/h (Re = 9.5E01 1.0.0E02 0.0E+00 5.30E+00 1.00E+00 1.8 f υ (7) This is Prandtl’s universal law of friction for smooth pipes and it has been verified with experiments and the agreement is seen to be excellent up to the Reynolds number ReD= uDh/ν=3.00E+00 1. the friction factor can be expressed as Figure 9.40E+00 1. we want to quantify the pressure loss and friction factor for the three geometries.0E02 0.5E01 Normalized velocity fluctuations 1.10E+00 1.0E01 5. we first attempt to compare the CFD results with the classical roughness theory and 10.0E+00 5.50E+00 Normalized time Figure 8. Instantaneous velocity fluctuations at the same singlepoint location for the liner 1 (top) and liner 2 (bottom) ASSESSMENT AND DISCUSSION In this section.20E+00 1. i. 3). The liner 2 model decreases the velocity fluctuations (i.0 log h f .20E+00 1. liner 1 on top and liner 2 below.0E02 1.e. Streamwise variation of velocity magnitude contours for the liner 1 (top) and liner 2 (bottom) models at Q=3333 m3/h v'/U w'/U u'/U Figure 10.0E02 1. turbulence level) by a factor of 5 compared to the liner 1 model.2x106. The velocity fluctuations are normalized by the mean velocity.50E+00 1.0E01 1.10E+00 1.5E01 Normalized velocity fluctuations 1.
05 Flow rate Q=5000 m^3/hr Flow rate Q=3333 m^3/hr 0.0E02 Figure 12. The amount decrease doesn’t appear to be simply a linear function. when the imparted shear stress dominates eddy dissipative scale. the coefficient of friction can be reduced by 80% with respect to the deeper metallic hose configuration.0E+06 1. Variation of friction coefficients with Reynolds number Figure 12 shows a cross plot of friction factor vs. By tuning the grid distributions.50E02 Base Rough. an 8 .00E+00 1.74 + 2 log 2k 2 2. the friction factor increases as we increase the depth of corrugation.125 Coefficient of Friction 0. For the smooth pipe. DES Simulations To assess the results of steady RANS with the DDES. the corrugations may be considered equivalent to periodic roughness (i. In other words. [5]). the friction factor was found to be Reynolds number independent..0E02 4. For the base and liner 1 geometry at Re~10M. Variation of friction coefficients with respect to the depth of corrugation A* Steady RANS vs. The wall shear stress of the liner 2 model is converging towards the stress values corresponding to the smooth pipe. a large amount of friction occurs along the wall or vice versa.0E+08 Base Corrugated vs. The friction factor values for the liner 1 are converging to that of the smooth pipe. For the range of flow rates. Due to complex flow behavior and recirculation in the base & liner 1 models. Figure 13 shows a summary of the friction factor computed based on the pressure drop for the two CFD models. the CFD results and the theory have an excellent match.0E+07 Reynolds number. This suggests the potential value of liners for reducing the pressure loss. the friction factor changes significantly with the Reynolds numbers.075 0.00E02 Liner 2 Smooth f= 1 Dh 1. For the higher values of Reynolds number (ranging from 104 to 107). Theory Smooth Pipe Theory (8) 0. Figure 11. by introducing liner materials.Coefficient of Friction Completely rough regime: In the present configuration. Re 1.025 Flow rate Q=2222 m^3/hr 0 0. For the baseline case.0E+05 1. The friction factor was determined by evaluating the pressure gradient along the pipe from the integrated pressure values. However. the friction coefficient is consistently larger than the liner 1 & liner 2 models. A* 8. which provides another view of the impact of reducing the corrugation depth. The theoretical estimate for the friction factor for the completely rough regime was estimated as 1. A reasonable consistency in the predictions of integrated pressure drop can be seen in the figure. Figure 11 also presents the roughness theory predictions given by the lines. the depth of corrugation A*.50E02 Baseline Liner 1 5. an inflectional behavior in the pressure drop and wall shear stress have been observed in the RANS and DDES results. 2k/Dh) of the diameter of pipe. Liner Models Figure 11 shows the variation of coefficient of friction for the range of Reynolds number for the three configurations and the smooth pipe. 0. The CFD results follow an inflectional friction factor relationship rather than the monotonic relationship given by the roughness theory and the Moody diagram.25E01 1. we further conduct a comparative study on the same grids. for the corrugated shapes the roughness theory seems to differ up to 24%.1 0.0E+00 2.0E02 6.e.0E02 Ampititude/Diameter. Theory Liner1 Rough. A systematic investigation on the effects of Reynolds number and relative roughness k on the friction factor was performed by Nikuradse Ref.00E01 7. This implies that. This may be explained in the context of boundary layer separations along the corrugation profile and the associated effects into the pressure gradients.
0E+07 Reynolds number. the separation point slowly moves upstream as the Reynolds number is increased.000 Flow Rate Q [m^3/hr] 10. Reynolds number. Typical LNG pumps can develop 6. [1]). [1]) 8.00E02 4.0E+08 Figure 14.0E+08 50 40 30 20 10 0 ‐ Base Liner 1 Liner 2 Figure 13. 1. For the liner 2 and smooth pipe. The depth and shape of the corrugation profiles are marginally different between the 16” ID pipe and 10.5”) ID pipe for a range of flow rates with fresh water as working fluid.20E01 Base CFD Model Liner 1 CFD Model Coefficient of Friction 1. As mentioned earlier. A 2.5” ID pipe (Ref. is able to overcome a large adverse pressure gradient.improved match between the RANS and DDES may be obtained. the turbulent boundary layer. this dip in the frictional drag may be attributed to the sudden shift in the point of separation for the base and liner 1 geometries. and perform further analysis of the results to facilitate interpretation in the commercial settings.00E02 6. reasonable agreement between the CFD and experimental values can be seen. the geometry is streamlined and the point of separation and the transition of boundary layer remain somewhat unchanged. because of its greater energy.0E+06 1.0E+06 60 Pressure Drop [bar] 1. where the pressure drop is computed based on the Equation (1). Pressure drop as a function of flow rates for the 16” pipe with three corrugation geometries 9 . All the pressure drop readings were average values taken over a period of 1 minute Figure 14 shows the comparison of CFD values with the experimental test done with water in 10.00E02 Baseline Liner 1 Liner 2 Smooth BaselineDES Liner1DES Liner2DES 2.6 bar differential pressure (97.000 6. Figure 15 shows the comparison of pressure drop as a function of flow rates for the geometries. For the base and liner 1 geometry at Re~10M. The experiment tests were performed on a 268 mm (10. For values of Re >10M. The friction factors are compared with respect the nondimensional dynamic similarity parameter.50E02 5.0E+05 1. In this range. Re 1. we would like to establish the relevance of friction factor in actual pipe configurations as function of pipe lengths and corrugation depths.5” ID pipe Next.000 8.0513 in 10.0E+07 Reynolds number.000 12.00E02 1.3 psi) while maintaining reasonable flow rates for LNG transfer. The turbulent boundary layer separates at a further downstream location along the corrugation profile. This may be important to estimate the capacity and limitations of LNG pumps for the corrugated pipe flow. Re 1. resulting in an increase of the friction factor.00E+00 1.00E01 C o e fficien t o f F rictio n 7.000 Figure 15. an inflectional behavior in the pressure drop and wall shear stress have been observed in the RANS and DDES results. Flow Rate for L=100 [m] 1. As seen in the flow over a cylinder. resulting in a thinner wake and a pressure distribution more similar to that of potential flow.00E01 Water Test (Ref. the laminar viscous sublayer portion of boundary layer becomes unstable and undergoes transition to turbulence. Pressure Required vs.000 4. Variation of friction coefficients for the range of Reynolds number for the RANS and DDES models We also present a comparison with the experimental test. Comparison of the CFD results for A*=0.25E01 1.0604 of 16” ID pipe with the water test with A*=0.50E02 0.5” pipe.
Harten. Vol. Streletes & Travin. Heat Transfer. H.. The 3D steady RANS and DDES models provided a consistent estimate of the pressure drop and friction factor for varying flow rates. Spalart. 2002 [9] Shur. vol.H. “Boundary Layer Theory. 1994. Schippl.K. K. E.. E... and Smits. REFERENCES [1] Framo Engineering AS Report. and Sparrow. Deck.J. Shur. 2006. 175207 [5] Nikuradse. F. ”Laws of flow in rough pipes”.” 45770313D.. improve the design of full scale tests and potentially eliminate some or all of such tests. A.. 2006.”Fully devloped turbulent pipe flow: a comparison between direct numerical simulation and experiment”. Journal of Fluid Mechanics. No. J. The pressure drop results were found to be quite sensitive to the corrugation depth. Significant 3D turbulence effects were found for the pipe geometry with circular corrugations suggested by both qualitative features and quantitative information. 1933 [6] Spalart.V. [7] Patankar. Steen. “CFD Calculations of Corrugated Flexible Pipe.periodic variations of crosssectional area”. 1975.E. J. J. 20.” Journal of Theoretical & Computational Fluid Dynamics. “Annual Review of Fluid Mechanics”. The CFD results appear to be sufficiently accurate that one might seriously consider using such models to investigate relative geometric differences and perform parametric studies of various corrugation configurations. 268. pp 16381640. R. 99. 181195.M. and Høvik. Shockling.. pp 349374.”A hybrid RANSLES approach with delayedDES and wallmodeled LES capabilities”. Liu. 1852. “Innovative Pipe System for Offshore LNG Transfer. J. K. Kunkel. Strelets. Resistant to Ambiguous Grid Densities. pp. International Journal of Heat and Fluid Flow. [3] Schlichting. Haakonsen.” OTC 19239. et al.. 365...R. McGrawHill Book Company.” 7th Edition. M. 2007 10 . Philosophical Transactions of the Royal Society. Vol 34. [4] Eggles...A. C. CFD can add substantially to classical roughness theory (which relies on known pipe characteristics) as it can address different geometries and dimensions without additional empirical data. the coefficient of friction can be reduced by 80% with respect to the deeper metallic hose configuration. P. and to qualify the pipe system for the LNG industry requirements. “Turbulent flow in smooth and rough pipes”. “Fully developed flow and heat transfer in ducts having streamwise[8] Piomelli U. 2008. [10] Allen. M. We hoped that we have done a reasonable job in predicting complex internal turbulent structures with the unsteady DDES simulations.K. offer guidance on design variations. and Travin. NACA TM 1292. Squires.J. By introducing the liner materials. M. ASME. C. The modeling can reduce the uncertainty. Jorgen. S.. and Balaras. G. 2008. 180186 CONCLUSIONS Cryogenic flexible pipe based LNG transfer system seems to be a good candidate for CFD modeling. Vol 29.. “A New Version of DetachedEddy Simulation. E. [2] Frohne..K.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.