Professional Documents
Culture Documents
net/publication/267754955
CFD STUDY OF SURFACE FLOW AND GAS DISPERSION FROM A SUBSEA GAS
RELEASE
CITATIONS READS
5 1,549
5 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Lars Sætran on 05 November 2014.
OMAE2014-24707
INTRODUCTION
Increasing subsea oil and gas activity and critically long
operation time on many existing pipe lines necessitates
improved methods for risk assessment concerning subsea gas
release. A gas release is caused by well blowouts, pipeline
failure (rupture or leak) or malfunction of subsea processing
equipment. Engebretsen et al. [2] reported several accidental
gas releases in the past years. Figure 1: Schematic of sub-sea gas release; from the release
If a release occurs, gas bubbles will travel to the surface point, through the fully developed plume, to the degassing zone
due to buoyancy. The transport of gas and gas bubbles in the with induced surface flow [1].
F
dt
D Fvirtual mass dtb
Here b b b is the bubble bulk density. The relaxation time mb k 1 t f
F c
rel is controlled by the speed of the breakup or the coalescence Vcell
process, and the dominating process is determined by the
Here M b is the total mass flow rate of one parcel trajectory and
comparison between the local diameter and equilibrium
diameters. k represents each trajectory in the control volume. Vcell is the
if d b d beq control volume. FD and Fvirtual mass are drag and virtual mass
rel B (6)
C if d b d beq forces on a single bubble. By integrating them with respect to
The relaxation time is restricted by the turbulent microscale the bubble time step tb within the flow time step t f , F c is
given by k 6
obtained.
, where is the kinematic viscosity of Through the Boussinesq hypothesis, the turbulent stress is
fluid. We have rel rel , k ; B is the breakup time scale represented by:
max
2 u u
2 1 f ui' u'j f kij t i j
which is modeled as B db 3 and C is the coalescence
3
3 x j xi
db
time scale which is modeled as C . Details
0.2 6 b k 1.3. Free surface damping
There is an increase in turbulent dissipation at the free
can be found in Laux and Johansen [6].
surface in bubble plumes as reported by Soga and Rehmann [9].
1.2. VOF This is the same effect which is observed at walls. Walls and
The VOF method which is a single phase Eulerian-Eulerian internal interfaces cause damping of turbulence. However, most
mixture model, employs an advanced interface tracking scheme implementations of the standard k-ε turbulence model only
known as Youngs’ VOF [8] to track the interface through the consider walls. Internal interfaces such as a liquid-gas
Eulerian mesh. If a cell is found to consist of two or more interphase in a VOF-model is normally ignored. Thus the
phases, an interface must be interpolated through it. The volume standard implementation of the k-ε model over-predict the TKE
fraction mass equations for water (f) and atmosphere (g) are at the free surface when compared to experiments as reported
solved. Note that liquid displacement by bubbles is neglected by Sheng and Irons [10]. Cloete et al. [1] also pointed out that
since coupling to the bubbles is only accounted for in the the standard k-ε model under predicts the surface velocity at
momentum equations. higher flow rate. “This is due to a phenomenon of increased
The transport equation for volume fractions of read: turbulence kinetic energy dissipation in the region of the free
surface which is not captured by the standard turbulence model.
k u j k 0 (7) When turbulent eddies approach and locally lift a free surface,
t x j
there is an increase in the rate of the turbulent energy cascade,
with the constraint g f 1 . which ultimately leads to increased TKE dissipation rates.”
This is solved together with a single set of Reynolds averaged Due to this we have to improve the standard
Navier- Stokes equations expressing conservation of momentum implementation of the k-ε model to account for the proximity of
a free surface which is seen as an internal interphase in the VOF
concept. Being aware that the epsilon transport equation is
actually the eddy length scale equation, a model is required to
supply the correct characteristic length to treat the near surface
u u j ui
u j C1 t ( i )
t x j k x j xi x j Figure 3. Predicted gas density [kg/m3], changing with height,
(12)
2 for a gas release of 170 Nl/s.
t C2 Sdamping
x j x j k
source
where k k and
k k The source terms in
(12) is implemented through a UDF in FLUENT.
1.5
0.5
0
0 0.5 1 1.5 2
radial distance [m] Figure 7: Comparison of contours of TKE [m2/s2]. Left is the
Figure 5: Comparison of predicted vertical velocities and standard k-epsilon model and the right is with damping effects.
experiments at 1.75 m, 3.8 m and 5.88m in height for a gas
release of 170 Nl/s.
6.2 exp
Height [m]
standard
6 damping
5.8
5.6
5.4
5.2
5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
velocity magnitude [m/s]
ACKNOWLEDGMENTS
The authors would like to thank the funding from SURE
project (Advanced Modeling Tool for Subsea Gas Release) in
SINTEF for this research.
DownloadedViewFrom:
publicationhttp://proceedings.asmedigitalcollection.asme.org/
stats on 11/04/2014 Terms of Use: http://asme.org/terms