Field Nuclear Magnetic Resonance Keelan ONeill University of Western - PowerPoint PPT Presentation

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 20 th October 2016 Fluid Science & Resources www.fsr.ecm.uwa.edu.au/

Motivation for research Objectives of flow metering The future of oil and gas production • Development of deeper offshore fields • Measurement of phase fractions and marginal fields • Measurement of phase velocities • Increased produced water • Flow regime identification • Increased subsea processing World Multiphase flow meter market Subsea production arrangement 500 Total shipments ($US million) 14% growth 400 Subsea manifolds 300 200 100 Production 0 hub 2011 2012 2013 2014 2015 2016 Satellite field Year Satellite field wells wells J. Yoder, The many phases of multiphase flowmeters, Pipeline & Gas Journal, 240 (2013) 40-41 Atkinson, I., et al., A New Horizon in Multiphase Flow Measurement. Oilfield Review, 2004. 16(4): p. 52-63 2

Multiphase flow meter technologies Common measurement principles Downstream transducer Detector Flow Gamma-ray • Gamma ray attenuation source direction • Electrical impedance measurement Ultrasonic Flow emission • Ultrasonic measurement systems Collimator Upstream transducer Advantages of nuclear magnetic resonance • Non-invasive measurement • Non-radioactive technology Commercial magnetic resonance flowmeter: M-Phase 5000 • Flow regime independent Superficial Liquid Velocity, U SL [m/s] 3 Superficial Gas Velocity, U SG [m/s]

Nuclear magnetic resonance Summary of nuclear magnetic resonance (NMR) magnetic field strengths Earth’s field Well-logging Magnetic resonance Chemical NMR tool imaging spectroscopy Ultra-low field Low field High field 50 µT 50 mT 7 T 4

Nuclear magnetic resonance Basic classical mechanics description 1. Align nuclei ( 1 H atoms) with 2. Apply radio magnetic field (polarisation) frequency pulse Radio frequency Pulse and Collect Experiment pulse B 0 Free induction decay 90 o RF pulse Z t a t delay 3. Observe the magnetisation relax X M Y 5

The Earth’s field NMR flow meter Faraday Pre-polarising EFNMR Key system features cage Flow magnet detection coil direction • Detected in the Earth’s 2. Excitation magnetic field • Time of flight measurement • Remote detection system 1. Polarisation Radio frequency pulse B 0 3. Detection Z X M Y Independent flowrate measurement EFNMR: Earth’s field nuclear magnetic resonance 6

Model for NMR signal Polarisation Intermediate decay Detection Pulse and Collect Sequence 90 o RF pulse Free induction decay Flow Halbach EFNMR array Detection t a t delay L PD L D L P Par arameters rs L: length τ 𝑄𝐸 = 𝑀 𝑄𝐸 τ 𝑄𝐸 = 𝑀 𝑄𝐸 τ 𝐸 = 𝑀 𝐸 τ : residence time 𝑤 𝑤 𝑤 v: velocity S 0 : initial magnetization T 1 : spin-lattice relaxation − 𝑢 𝑒𝑓𝑚𝑏𝑧 +𝑢 𝑏 T 2 * : observed spin-spin − 𝜐 𝑄 − 𝜐 𝑄𝐸 1 − 𝑢 𝑒𝑓𝑚𝑏𝑧 + 𝑢 𝑏 ∗ 1 𝑓 𝑈 𝑈 𝑈 𝐓(𝑤, 𝑢 𝑏 ) = 𝑇 0 1 − 𝑓 𝑓 relaxation 1 2 𝜐 𝐸 7

Tikhonov Regularisation Tikhonov regularisation The inverse problem A × P 𝑤 = S  P(𝑤) = A −1 × S • Pipe velocity distribution Least squares fitting method Model r NMR Signal transfer • Allows complex models to Velocity probability matrix z be fit to experimental data distribution Real image Image with noise Reconstructed image M. Bertero and P. Boccacci, Introduction to inverse problems in imaging. 1998, CRC Press. 8

Single phase velocity analysis Experimental conditions Single phase water flows at 4 – 52 L/min (0.08 – 1.15 m/s) Experimental NMR signals fit using Corresponding velocity probability regularisation distributions 6 30 Expt. Model 0.18 m/s data fit 0.35 m/s 25 5 0.18 m/s 0.35 m/s 0.53 m/s 0.53 m/s NMR Signal [ μ V] 0.71 m/s 20 4 0.71 m/s 0.88 m/s 0.88 m/s 1.06 m/s P(v) 1.06 m/s 15 3 10 2 5 1 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5 2 Time since pulse [s] Velocity [m/s] 9

Velocity comparison Independent rotameter NMR measurement measurement (Uncertainty of ± 5%) section Comparison of measured mean Deviation plot velocities 1.2 5.0% Predicted mean velocity from NMR NMR predicted velocity deviation 1 2.5% analysis [m/s] 0.8 0.6 0.0% 0.4 -2.5% 0.2 0 -5.0% 0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2 Measured mean velocity from in-line rotameter Measured mean velocity from in-line rotameter [m/s] [m/s] Mean absolute error = 1.9 % 10

Two pipe analysis L PD Experimental conditions Pipe A • Overall flow rates: EFNMR Velocity (pipe Flow Halbach 15 – 50 L/min detection A) ~ 33% of array coil • Separation distance ( L PD ): velocity (pipe B) Pipe B 0.68, 0.78. 0.88 and 0.98 m Comparison of measured mean Two pipe system velocity probability velocities distributions 4 1.2 NMR predicted velocities [m/s] 30 l/min 1 40 l/min 3 50 l/min 0.8 P(v) 0.6 2 Pipe 0.4 A B 0.68 m 1 0.78 m L PD 0.2 0.88 m 0.98 m 0 0 0 0.2 0.4 0.6 0.8 1 1.2 0 0.4 0.8 1.2 1.6 Rotameter measured velocities [m/s] 11 Velocity [m/s]

Gas/liquid analysis Flow regime identification 80 NMR Signal [µV] 60 Stratified flow 40 Slug flow 20 0 0 10 20 30 40 50 Time [s] Video analysis Video analysis region ID: 31 mm EFNMR Halbach Liquid Holdup Detection Array Coil U SL = 0.13 m/s U SG = 0.88 m/s Time [s] 12

Velocity and holdup determination Processed signal & model fit Velocity probability Liquid holdup distribution 8 determination Experimental data 4 Model fit NMR signal [µV] 6 𝑇 0,𝐺𝑣𝑚𝑚 3 P(v) 𝑇 0,2𝑄 2 4 1 0 2 0 0.25 0.5 0.75 1 𝑦 𝑀 = 𝑇 0,2𝑄 Velocity [m/s] 0 𝑇 0,𝐺𝑣𝑚𝑚 0 0.1 0.2 0.3 0.4 Time since pulse [s] 1 1 Liquid Holdup Velocity [m/s] 0.75 0.75 Stratified flow 0.5 0.5 Slug 0.25 0.25 flow 0 0 0 10 20 30 40 0 10 20 30 40 Overall time [s] Overall time [s] 13

Holdup analysis comparison 1.00 Liquid Holdup Slug unit Stratified flow 0.75 Liquid: 4 L/min 0.50 Gas: 40 L/min 0.25 Video holdup NMR holdup Background stratified 0.00 component 0 10 20 30 40 50 60 1.00 Background stratified Low frequency Liquid Holdup under-prediction 0.75 slug flow • Gas bubbles Liquid: 8 L/min 0.50 • Meniscus Gas: 40 L/min • Increased relaxation 0.25 Video holdup NMR holdup 0.00 0 10 20 30 40 50 60 1.00 Liquid Holdup Higher frequency Partial slug capture 0.75 • slug flow Slug residence time ~0.2 s 0.50 Liquid: 12 L/min • Scan time ~0.7 s Gas: 20 L/min 0.25 Video holdup • Slug not fully NMR holdup 0.00 captured 0 10 20 30 40 50 60 14 Experimental Time [s]

Two phase velocity analysis 1 1 Liquid Holdup Velocity [m/s] 0.75 0.75 0.5 0.5 0.25 0.25 0 0 0 10 20 30 40 0 10 20 30 40 Overall time [s] Overall time [s] 0.4 𝑂 𝑉 𝑇𝑀 = 𝑦 𝑗 𝑤 𝑗 1 𝑗=1 NMR velocity standard deviation NMR predicted superficial Slug Flow Gas Flow rate 𝑂 10 L/min 0.3 20 L/min velocity [m/s] 40 L/min 0.2 [m/s] 0.1 Gas Flow rate 0.1 10 L/min Stratified Flow 20 L/min 40 L/min 0 0 0.1 0.2 0.3 0.4 0.01 Rotameter measured superficial velocity 0 0.1 0.2 0.3 [m/s] Rotameter measured superficial velocity [m/s] 15

Conclusions • Can accurately measure and model the FID signal of a moving fluid through the flow metering system 5 • Can determine the velocity probability 0.22 m/s 4 0.66 m/s distribution of liquid moving in the system 1.10 m/s 3 P(v) via Tikhonov regularisation 2 1 0 0 0.5 1 1.5 2 Velocity [m/s] 1 Velocity [m/s] Stratified flow 0.75 • Able estimate the liquid velocity and holdup Slug flow 0.5 over time for stratified and slug flow 0.25 0 0 10 20 30 40 50 Overall time [s] 16

Future work Superficial Liquid Velocity, U SL [m/s] • Analyse fluid flow in further air/water flow regimes • Incorporate oil into flow metering system • Apply dynamic nuclear polarisation for signal Superficial Gas Velocity, U SG [m/s] enhancement Gas • Fully develop analysis techniques Oil to be able to interpret three phase Water flow (oil/gas/water) Hogendoorn, J., et al., Magnetic resonance multiphase flowmeter: measuring principle 17 and multiple test results, in Upstream Production Measurement. 2015: Houston.

Acknowledgements Supervisors Michael Johns, Einar Fridjonsson and Paul Stanwix Final Year Project Students Adeline Klotz and Jason Collis Thank you for your attention! Questions? 18

Tikhonov regularisation The inverse problem; Pipe velocity distribution  𝑄(𝑤) = A −1 × 𝑇 r A × 𝑄 𝑤 = 𝑇 z Model Velocity NMR transfer probability Signal matrix distribution 8 λ = 1 × 10 -3 Apply Tikhonov regularisation; λ = 75 (optimal) ( A × 𝑄 𝑤 − 𝑇 2 + λ 𝑄 𝑤 6 2 ) λ = 1 × 10 4 min P(v) 4 Second Residual norm moment of P(v) 2 Generalised cross validation 0 method is used to optimise the 0 0.5 1 1.5 smoothing parameter ( λ ) Velocity [m/s] 19