monitoring multiphase flow via earth’s - sut · pdf filemonitoring multiphase flow via...
TRANSCRIPT
Monitoring Multiphase Flow via Earth’s
Field Nuclear Magnetic Resonance
Keelan O’Neill University of Western Australia
School of Mechanical and Chemical Engineering
SUT Subsea Controls Down Under 20th October 2016
Fluid Science & Resources
www.fsr.ecm.uwa.edu.au/
Motivation for research
2
The future of oil and gas production
• Development of deeper offshore fields
and marginal fields
• Increased produced water
• Increased subsea processing
Production hub
Satellite field wells
Subsea manifolds
Satellite field wells
Atkinson, I., et al., A New Horizon in Multiphase Flow
Measurement. Oilfield Review, 2004. 16(4): p. 52-63
Subsea production arrangement
Objectives of flow metering
• Measurement of phase fractions
• Measurement of phase velocities
• Flow regime identification
J. Yoder, The many phases of multiphase flowmeters,
Pipeline & Gas Journal, 240 (2013) 40-41
0
100
200
300
400
500
2011 2012 2013 2014 2015 2016
Tota
l sh
ipm
en
ts (
$U
S m
illio
n)
Year
World Multiphase flow meter market
14% growth
Commercial magnetic resonance flowmeter: M-Phase 5000
Multiphase flow meter technologies
3
Advantages of nuclear magnetic resonance
• Non-invasive measurement
• Non-radioactive technology
• Flow regime independent
Common measurement principles
• Gamma ray attenuation
• Electrical impedance measurement
• Ultrasonic measurement systems
Superficial Gas Velocity, USG [m/s]
Su
perf
icia
l L
iqu
id V
elo
cit
y,
US
L [m
/s]
Flow
direction
Upstream transducer
Downstream transducer
Ultrasonic
emission Flow
Gamma-ray
source
Detector
Collimator
Nuclear magnetic resonance
4
Ultra-low field Low field High field
Summary of nuclear magnetic resonance (NMR)
magnetic field strengths
50 µT 50 mT 7 T
Earth’s field
NMR
Well-logging
tool
Magnetic resonance
imaging
Chemical
spectroscopy
Nuclear magnetic resonance
5
tdelay ta
90o RF pulse
Free induction decay
Pulse and Collect Experiment
Basic classical mechanics description
1. Align nuclei (1H atoms) with
magnetic field (polarisation)
B0
M
X
Z
Y
3. Observe the
magnetisation relax
2. Apply radio
frequency pulse
Radio frequency
pulse
The Earth’s field NMR flow meter
6
Pre-polarising
magnet EFNMR
detection coil Flow
direction
Faraday
cage Key system features
• Detected in the Earth’s
magnetic field
• Time of flight measurement
• Remote detection system
EFNMR: Earth’s field nuclear
magnetic resonance
Independent
flowrate
measurement
1. Polarisation
B0
M
X
Z
Y
2. Excitation
Radio frequency
pulse
3. Detection
Model for NMR signal
7
LP
𝐒(𝑣, 𝑡𝑎) = 𝑆0 1 − 𝑒− 𝜏𝑃𝑇1 𝑒
− 𝜏𝑃𝐷𝑇1 1 −
𝑡𝑑𝑒𝑙𝑎𝑦 + 𝑡𝑎
𝜏𝐷𝑒− 𝑡𝑑𝑒𝑙𝑎𝑦+𝑡𝑎
𝑇2∗
τ𝐷 = 𝐿𝐷𝑣
τ𝑃𝐷 = 𝐿𝑃𝐷𝑣
τ𝑃𝐷 = 𝐿𝑃𝐷𝑣
Halbach
array
EFNMR
Detection
Flow
LD
Polarisation Intermediate decay Detection
tdelay ta
90o RF pulse
Free induction decay
Parameters L: length τ: residence time v: velocity S0: initial magnetization T1: spin-lattice relaxation T2
*: observed spin-spin relaxation
LPD
Pulse and Collect Sequence
Tikhonov Regularisation
8
Real image Reconstructed image Image with noise
M. Bertero and P. Boccacci, Introduction to inverse problems in imaging. 1998, CRC Press.
Pipe velocity distribution
r
z
The inverse problem
Model
transfer
matrix
NMR Signal
Velocity probability
distribution
A × P 𝑣 = S P(𝑣) = A−1 × S
Tikhonov regularisation
• Least squares fitting
method
• Allows complex models to
be fit to experimental data
Single phase velocity analysis
9
0
1
2
3
4
5
6
0 0.5 1 1.5 2P
(v)
Velocity [m/s]
0.18 m/s
0.35 m/s
0.53 m/s
0.71 m/s
0.88 m/s
1.06 m/s
0
5
10
15
20
25
30
0 0.1 0.2 0.3 0.4 0.5 0.6
NM
R S
ign
al [μ
V]
Time since pulse [s]
0.18 m/s
0.35 m/s
0.53 m/s
0.71 m/s
0.88 m/s
1.06 m/s
Expt.
data
Model
fit
Experimental NMR signals fit using
regularisation
Corresponding velocity probability
distributions
Experimental conditions
Single phase water flows at 4 – 52 L/min (0.08 – 1.15 m/s)
Velocity comparison
10
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Pre
dic
ted
me
an
ve
loc
ity f
rom
NM
R
an
aly
sis
[m
/s]
Measured mean velocity from in-line rotameter [m/s]
-5.0%
-2.5%
0.0%
2.5%
5.0%
0 0.2 0.4 0.6 0.8 1 1.2
NM
R p
red
icte
d v
elo
cit
y d
evia
tio
n
Measured mean velocity from in-line rotameter [m/s]
Comparison of measured mean
velocities
Mean absolute error = 1.9 %
Deviation plot
NMR measurement
section
Independent rotameter
measurement (Uncertainty of ± 5%)
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2N
MR
pre
dic
ted
ve
loc
itie
s [
m/s
]
Rotameter measured velocities [m/s]
Two pipe analysis
11
Experimental conditions • Overall flow rates:
15 – 50 L/min
• Separation distance (LPD):
0.68, 0.78. 0.88 and 0.98 m
Velocity (pipe
A) ~ 33% of
velocity (pipe B)
Flow Pipe A
Pipe B
Halbach array
EFNMR detection
coil
0
1
2
3
4
0 0.4 0.8 1.2 1.6
P(v
)
Velocity [m/s]
30 l/min
40 l/min
50 l/min
LPD
Two pipe system velocity probability
distributions
Comparison of measured mean
velocities
B 0.68 m
0.78 m
0.88 m
0.98 m
Pipe
A
LP
D
Gas/liquid analysis
12
Flow regime identification
0
20
40
60
80
0 10 20 30 40 50
NM
R S
ign
al [µ
V]
Time [s]
Stratifiedflow
Slug flow
Video analysis Video analysis
region
EFNMR
Detection
Coil
Halbach
Array
ID:
31 mm
Liq
uid
Ho
ldu
p
Time [s]
USL = 0.13 m/s
USG = 0.88 m/s
Velocity and holdup determination
13
0
2
4
6
8
0 0.1 0.2 0.3 0.4
NM
R s
ign
al [µ
V]
Time since pulse [s]
Experimental data
Model fit
0
1
2
3
4
0 0.25 0.5 0.75 1
P(v
)
Velocity [m/s]
Processed signal & model fit Velocity probability
distribution
0
0.25
0.5
0.75
1
0 10 20 30 40
Ve
loci
ty [
m/s
]
Overall time [s]
𝑥𝐿 = 𝑆0,2𝑃𝑆0,𝐹𝑢𝑙𝑙
𝑆 0,2𝑃
𝑆 0,𝐹𝑢𝑙𝑙
Liquid holdup
determination
0
0.25
0.5
0.75
1
0 10 20 30 40Li
qu
id H
old
up
Overall time [s]
Stratifiedflow
Slugflow
0.00
0.25
0.50
0.75
1.00
0 10 20 30 40 50 60
Liq
uid
Ho
ldu
p
Video holdup
NMR holdup
0.00
0.25
0.50
0.75
1.00
0 10 20 30 40 50 60
Liq
uid
Ho
ldu
p
Experimental Time [s]
Video holdup
NMR holdup
Holdup analysis comparison
14
Liquid: 4 L/min
Gas: 40 L/min
Liquid: 8 L/min
Gas: 40 L/min
Liquid: 12 L/min
Gas: 20 L/min
Higher frequency
slug flow
Low frequency
slug flow
Stratified flow
0.00
0.25
0.50
0.75
1.00
0 10 20 30 40 50 60
Liq
uid
Ho
ldu
p
Video holdupNMR holdup
Background stratified
under-prediction
• Gas bubbles
• Meniscus
• Increased relaxation
Slug unit
Background stratified
component
Partial slug capture
• Slug residence
time ~0.2 s
• Scan time ~0.7 s
• Slug not fully
captured
0
0.1
0.2
0.3
0.4
0 0.1 0.2 0.3 0.4
NM
R p
red
icte
d s
up
erf
icia
l velo
cit
y [
m/s
]
Rotameter measured superficial velocity [m/s]
10 L/min
20 L/min
40 L/min
Gas Flow rate
Two phase velocity analysis
15
0
0.25
0.5
0.75
1
0 10 20 30 40
Ve
loci
ty [
m/s
]
Overall time [s]
0
0.25
0.5
0.75
1
0 10 20 30 40
Liq
uid
Ho
ldu
p
Overall time [s]
𝑈𝑆𝐿 = 𝑥𝑖𝑣𝑖𝑁𝑖=1
𝑁
0.01
0.1
1
0 0.1 0.2 0.3N
MR
ve
loc
ity s
tan
da
rd d
evia
tio
n
[m/s
] Rotameter measured superficial velocity [m/s]
10 L/min
20 L/min
40 L/min
Gas Flow rate
Stratified Flow
Slug Flow
Conclusions
• Can accurately measure and model the FID
signal of a moving fluid through the flow
metering system
16
• Can determine the velocity probability
distribution of liquid moving in the system
via Tikhonov regularisation
• Able estimate the liquid velocity and holdup
over time for stratified and slug flow
0
1
2
3
4
5
0 0.5 1 1.5 2
P(v
)
Velocity [m/s]
0.22 m/s
0.66 m/s
1.10 m/s
0
0.25
0.5
0.75
1
0 10 20 30 40 50
Ve
loci
ty [
m/s
]
Overall time [s]
Stratified flow
Slug flow
Future work
• Analyse fluid flow in further
air/water flow regimes
17
Superficial Gas Velocity, USG [m/s]
Su
perf
icia
l L
iqu
id V
elo
cit
y,
US
L [m
/s]
• Fully develop analysis techniques
to be able to interpret three phase
flow (oil/gas/water)
• Apply dynamic nuclear
polarisation for signal
enhancement
Hogendoorn, J., et al., Magnetic resonance multiphase flowmeter: measuring principle
and multiple test results, in Upstream Production Measurement. 2015: Houston.
Gas
Oil
Water
• Incorporate oil into flow
metering system
Acknowledgements
Supervisors
Michael Johns, Einar Fridjonsson and Paul Stanwix
Final Year Project Students Adeline Klotz and Jason Collis
18
Thank you for your attention!
Questions?
Tikhonov regularisation
19
The inverse problem;
Model
transfer
matrix
NMR
Signal
Velocity
probability
distribution
Pipe velocity distribution
r
z
Residual norm Second
moment of P(v)
Generalised cross validation
method is used to optimise the
smoothing parameter (λ)
min ( A × 𝑃 𝑣 − 𝑆 2 + λ 𝑃 𝑣 2)
A × 𝑃 𝑣 = 𝑆 𝑃(𝑣) = A−1 × 𝑆
0
2
4
6
8
0 0.5 1 1.5
P(v
) Velocity [m/s]
λ = 1×10-3
λ = 75 (optimal)
λ = 1×104
Apply Tikhonov regularisation;
Comparison to a theoretical model
Theoretical turbulent power law distributions
𝑈 𝑟 = 𝑉𝑀𝑛 + 1
𝑛1 −
𝑟
𝑅
1𝑛
20
0
2
4
6
8
0 0.2 0.4 0.6 0.8 1
P(v
)
Velocity [m/s]
n = 4
n = 5.37
n = 7
Experimental
Experimental and theoretical distributions at v = 0.44 m/s
𝑛 = 𝑓 𝑅𝑒
𝑛 0.44𝑚
𝑠= 5.37
(Zagarola et al. 1997)