ARC Forum, Orlando, Florida February 8-11, 2016

Is it possible to develop an intelligent (and accurate) multiphase flow meter?

University of Pisa, Italy

About us

• The University of Pisa was established in 1343 and has been the university •

of Galileo Galilei and Enrico Fermi. With this background, it is still one of the leading European Universities in the area of Physics. TEA Sistemi is a research company founded in 1990 as a spin-off of the University of Pisa in the field of Applied Fluid Mechanics. The company now counts on more than 50 employees and established a close cooperation with major International Companies such as ABB and ENI.

• The company’s Multiphase Flow Laboratory is used for research and

•

equipment testing in areas such as gas-liquid-liquid flow and separation, multiphase flow metering, down-hole liquid-liquid separation. In the Oil & Gas Sector, our services cover the fields of Flow Assurance, Risk and Environmental Studies, Subsea and Topside Process Design. Our main technical achievement in 2015 has been the open sea, successful test of CUBE, the ENI system for the containment of subsea blowouts. VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 2

Multiphase Metering

• Falcone & Harrison [1] report that in 2010 the total number of Multiphase •

Flow Meters (MFM) installed worldwide was 3500, with an average growth rate of about 400 meters per year in the period 2011-2014. An extrapolation of these figures indicates that, up to now, more than 5000 MFMs have been installed all over the world. This number roughly corresponds only to 0.5% of producing wells and suggests that MFMS may represent a very interesting business.

• The obvious questions are:  Is it possible to continue with such growth rates in a time of low oil price?  Should we look for low cost, possibly intelligent and more accurate MFMs? [1] G. Falcone, B. Harrison (2011) Forecast expects continued multiphase flowmeter growth Oil&Gas Journal

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 3

Multiphase Metering

• Most of the MFMs on the market consists of a combination of different •

instruments, typically, a gamma-densitometer, a water-cut meter and a Venturi meter. These instruments do not directly measure the flow rates. Flow rates are predicted by means of empirical equations of limited validity. The combined effects of measuring errors of individual instruments and poor correlations explains the low accuracy of conventional MFMs.

• Why are used? A possible explanation is that, when a test separator is

available, you can calibrate the MFM. This greatly enhances its accuracy. If it is not available there is no evidence that the readings of the MFM are wrong.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 4

Multiphase Metering

• There are MFMs on the market which are based on the concept of isokinetic

sampling, do not require calibration and directly measure flow rates by means of standard field instrumentation. In principle, direct measurements are much more accurate than indirect ones, but of course, the accuracy of these MFMs depends on the sampling method.

• Multiphase flow metering can also be based on the analysis of process

signals. In this case the flow meter is only a software tool. These systems are known as “Virtual Flow Meters, VFM”. Their cost can be very low and their accuracy very poor, unless they are calibrated with a test separator. This is often the case also with conventional MFMs.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 5

Technical background

• The Multiphase Flow group at the University of Pisa has been active in the field

• • •

of experimental fluid mechanics and model development since mid 70’s. This background has been transferred to TEA Sistemi and used, over the last 25 years, in a number of projects focused on the development of different types of MFMs. The first MFM developed by TEA Sistemi was based on the use of a gamma– densitometer coupled with a water-cut and a Venturi meter (tested in Trecate field, in 1993). In the same years (1993-1995), TEA Sistemi developed metering systems based on the analysis of pressure signals (VFM). Since then this approach has been adopted in a number of applications.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 6

Relevant publications (academic) • Andreussi P, Pitton E, Ciandri P, Scozzari A, (2016). Measurement of liquid film distribution in near-horizontal pipes with an array of wire probes. Flow Meas Inst, vol. 47.

• Bonizzi M, Andreussi P, Banerjee S, (2009). Flow regime independent, high resolution, multifield modelling of gas-liquid flows in pipelines. Int J Multiphase Flow, vol. 35.

• Belsito S, Lombardi P, Andreussi P, Banerjee S. (1998). Leak detection in liquified gas pipeline by artificial neural networks. AICHE J, vol. 44.

• Nydal O J, Pintus S, Andreussi P (1992). Statistical characterization of slug flow in horizontal pipes. Int J Multiphase Flow, vol. 18,

• Andreussi P, Didonfrancesco A, Messia M (1988). An impedance method for the measurement of liquid hold-up in 2-phase flow. Int J Multiphase Flow vol. 14,

• Asali J C, Hanratty T J, Andreussi P (1985). Interfacial drag and film height for vertical annular-flow. AICHE J, vol. 31.

• Brown R C, Andreussi P, Zanelli S (1978). The use of wire probes for the measurement of liquid film thickness in annular gas-liquid flows. Can J Chem Eng, vol. 56.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 7

Technical background

• The Multiphase Flow group at the University of Pisa has been active in the field

• • •

of experimental fluid mechanics and model development since mid 70’s. This background has been transferred to TEA Sistemi and used, over the last 25 years, in a number of projects focused on the development of different types of MFMs. The first MFM developed by TEA Sistemi was based on the use of a gamma– densitometer coupled with a water-cut and a Venturi meter (tested in Trecate field, in 1993). In the same years (1993-1995), TEA Sistemi developed metering systems based on the analysis of pressure signals (VFM). Since then this approach has been adopted in a number of applications.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 8

Deep Panuke (ENCANA) o Real-time visualisation of DCS signals (valve opening, pressures, temperatures, levels). o Production is measured with a VFM approach from pressure signals.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 9

Production Allocation in Multi-layer Wells

• The analysis of available P&T signals is the basis of the FACT software, developed in cooperation with Well Dynamics (Halliburton) for production allocation in smart wells.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 10

The case of wet gas

• MFMs based on gamma sources have also been deployed to measure natural

• •

gas production in presence of condensate and water. These applications are a real challenge because gamma sources are not able to measure liquid holdups below 1-2% by volume with an acceptable accuracy. In 1998 TEA Sistemi started the development of a wet gas flow meter based on the isokinetic sampling concept. Sampling is followed by phase separation and metering of individual phase flow rates. In these MFMs, when sampling is isokinetic, the gas and liquid flow rates are immediately obtained from the measurements, without field calibration or recourse to empirical equations.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 11

The isokinetic sampling method (VEGA) Qs= 0

 Sampling off DP0= aQ0

2

Q1=Q0

Q0 DP

Qs= 0.1Q0

 Sampling on DP1=

a(0.9)2Q02

Q1=0.9 Q0

Q0 DP

 For a 10% sample, sampling is isokinetic when the sample flow rate is such that DP1 /DP0 = 0.81 TEA

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 12

New Multiphase Flow Meter Main Flow

∆P2

∆P3

Control Panel

∆P1

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 13

King Kong VEGA (GOM), 2002

• • • • • • •

Pressure 1000-2000 psig Temperature 40-50 °F Gas Flow 25-100 MMscfd Liquid Flow  Water 25-100 bbl/d  Methanol 25-100 bbl/d  Condensate 125-800 bbl/d Gas Velocity (actual) 7-60 ft/s Pipe Diameter 8” Dimensions (main vessel)  height 7 ft  diameter 18 in

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 14

King Kong VEGA (GOM), 2002

• This wet gas meter has been able to detect with a very good accuracy overall VEGA as vs Separator Measurement liquid volume fractions low as Liquids 0.2% and water Comparison fractions(BBLD) of 0.02%. Allocated Acqueous Phase Production 0

10

20

30

40

50

60

70

80

90

500

100 100

Total Liquid Mean Difference: 1.8% 90

VEGA Condensate Measured Production

Condensate Mean Difference: 1.9% 400

80 Acqueous Phase Mean Difference: 3.5%

350

70

300

60

250

50

200

40

150

30

100

20

50

10

0 0

50

100

150

200

250

300

350

400

450

VEGA Acqueous Phase Measured Production

450

Total Liquid Condensate Acqueous Phase

0 500

Allocated Condensate Production

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 15

New Ideas

• In 2015 the low price of crude oil induced TEA Sistemi to study alternative •

• •

solutions to the expensive equipment deployed for the exploitation of subsea fields. In the case of multiphase metering the obvious choice would be some type of VFM, considering that in this metering system the hardware cost is close to zero. Unfortunately, also its accuracy is very low. There are other problems with VFMs, for instance it can be extremely difficult to derive the water-cut from process signals. These limitations of virtual flow meters led us to develop a more complex configuration, still based on the VFM approach.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 16

New Ideas

• In a conventional MFM a gamma-source is used to measure the local liquid • • •

hold-up. This is a cumbersome and expensive method. In a vertical tube, with some modeling, the liquid hold-up can be derived from the pressure gradient. In practice, a single differential pressure transmitter may replace a gamma-densitometer. In both cases it is necessary to use a set of empirical equations to derive useful quantities from these measurements. The quality of the obtained results strictly depends on the quality of these equations. A careful analysis of the flow structure led the development of a physical model of upward gas-liquid flow based on 5 dimensionless numbers.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 17

New Multiphase Flow Meter

Orifice

Static Head

Outlet Flow Inlet Flow VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 18

Static head meter, raw data

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 19

Correlation of Static Head data

•

The pressure losses in a vertical pipe Δ𝑃𝑆𝐻 = 𝜌𝑀 𝑔Δ𝐻 ∙ 1 + can be expressed as

Δ𝑃𝑆𝐻 𝜌𝐿 𝑔Δ𝐻

=𝑓

2𝑓𝑀 2 𝑈 𝑔𝐷 𝑀

𝜌𝐺 𝑈𝑆𝐺 , , 𝑅𝑒𝐿 , 𝐹𝑟𝐿 𝜌𝐿 𝑈𝑆𝐿

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 20

New Ideas

• In a conventional MFM a Venturi meter is often used to measure the •

• • •

momentum of the multiphase mixture. In a VFM the same job can be done with a valve. In practice, any flow obstruction can be used, such as a calibrated orifice. When dealing with multiphase flow, there are no good reasons to use a Venturi rather than an orifice and the savings related to the lower cost of an orifice are significant. In both cases empirical equations are required to obtain useful data, but the development of these equations for a Venturi is as difficult as it is for an orifice. The physical model developed for the orifice is based on the same set of dimensionless numbers as the static head equation.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 21

Orifice – raw data

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 22

Correlation of orifice data

•

The pressure losses through the orifice,

Δ𝑃𝑂𝑅 = 𝐾𝐿𝑜𝑠𝑠 𝐹𝑟𝐿 , 𝑅𝑒 ∙ 𝜌𝐿 𝑈𝑆𝐿 + 𝜌𝐺 𝑈𝑆𝐺 ∙ 𝑈𝑆𝐿 + 𝑈𝑆𝐺 can be expressed as

Low Froude number region (=0.7)

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 23

Model results – Prediction of gas and liquid flow rates • Static head and orifice meter provide gas and liquid flow rates with mean errors equal to 2% for the liquid and 4% for the gas.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 24

How can we measure the water-cut?

• Water-cut measurements in a three-phase mixture are a challenge.

•

Various methods have been adopted in conventional MFMs and it is difficult to understand if these methods can be reliable at varying flow conditions. On the other hand, it is quite obvious that the water-cut measurement after gas separation is a much better option than trying to measure the water-cut in presence of the gas phase.

• This measurement can be easily accomplished in a MFM based on

isokinetic sampling, considering that the meter control system can guarantee the time required for liquid-liquid separation.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 25

New Multiphase Flow Meter

Orifice Static Head

Outlet Flow Inlet Flow

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 26

New Multiphase Flow Meter

Orifice

Water Cut Measurement

Static Head

Outlet Flow Inlet Flow VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 27

Why this MFM is “intelligent” (and eventually accurate)?

• MFMs produce a great amount of data, but often only part of these data is • • •

used. In the development of correlations, we did not use only time-averaged values of differential pressure signals. As a matter of fact, a careful analysis of transient signals allowed a better understanding of the flow structure and the development of physically based equations. There is also a trick: the measurements are redundant. Gas and liquid flow rates are also derived from the measured value of the sampled liquid flowrate. In conclusion, we may assume that “intelligent” means knowledge based and able to learn from extensive data. In the present application it also means lowcost and accurate.

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 28

Conclusions

• MFMs can be made of five/six conventional pressure transmitters, rather than

• •

•

expensive and cumbersome instruments, such as a gamma-densitometer. With this essential set-up, the quality of single phase flow rate measurements does not necessarily deteriorate: this strictly depends on the quality of the data (field and/or laboratory calibrations) and the correlations adopted. This MFM presents numerous advantages in terms of easy of maintenance, commissioning, shipping, no radioactive sources and, of course, in terms of cost. In its simpler gas-liquid configuration, it does not need much more than a flow computer, like a VFM. As a matter of fact, it is a VFM. A final question regards the entire field of subsea processing: Is it better to invest in the development of expensive equipment or in a deeper knowledge of multiphase flow?

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 29

Questions?

Thank you For enquiries please e-mail to [email protected]

VISION, EXPERIENCE, ANSWERS FOR INDUSTRY

© ARC Advisory Group • 30