Volumetric fraction measurement in oil-water-gas multiphase flow with dual energy gamma-ray system

Li et al. /J Zhejiang Univ SCI 2005 64(12):1405-14111405Journal of Zhejiang University SCIENCEISSN 1009-3095htp://ww.zju.edu.cn/jzusJzUSE-mail: jzus@zju.edu.cnV olumetric fraction measurement in oil-water-gas multiphase flowwith dual energy gamma-ray systemLI Dong-hui (李东晖), WU Ying xiang (吴应湘), LI Zhi-biao (李志彪), ZHONG Xing-fu (钟兴福)(Instiute of Mechanics, Chinese Academy of Sciences, Beijing 10080, China)tE-mail: donghui_ li@ 126.com; yxwu@imech.ac.cnReceived Sept. 7, 2005; revision accepted Sept. 25, 2005Abstract: V olumetric fraction distribution measurement is a constituent part of process tomography system in oil-water-gasmutiphase flow. With the technological development of nuclear radial inspection, dual-energy γ-ray techniques make it possibleto investigate the concentration of the different components on the cross-section of oil-water-gas multiphase pipe-flow. Thedual-energy gamma-ray technique is based on materials attenuation coefficients measurement comprised of two radioactiveisotopes of 24.Am and 24 Cs which have emission energies at 59.5 keV and 662 keV in this project. Nuclear instruments and dataacquisition system were designed to measure the material's attenuation dose rate and a number of static tests were conducted at theMultiphase Laboratory, Institute of Mechanics, Chinese Academy of Sciences. Three phases of oil-water-gas media were inves-tigated for their possible use to simulate different media volumetric fraction distributions in experimental vessels. Attenuationintensities were measured, and the arithmetic of linear attenuation cofficients and the equations of volumetric fractions werestudied. Investigation of an unexpected measurement error from attenuation equations revealed that a modified arithmetic wasinvolved and finally the system achieved acceptable accuracy in experimental research.Key words: Volumetric fraction, Multiphase flow, Dual-energy y-ray, Process tomographydoi: 10.1631/jzus.2005.A1405Document code: ACLC number: TB126INTRODUCTIONdia in the oil pipe are used to deduce information onthe components volume fraction. Dual-energy γ-rayOil-water-gas multiphase mixed transport in oil technique has been applied to a difficult case withpipe is extensively used in ocean oil industries. How three basic materials having obviously varying dis-to measure the volumetric fractions of oil-water-gas tribution of linear attenuation coefficients, and provedmultiphase flow is an important subject of study in the to be a very promising technique for simple and fastoil industry. Related research work was started in the estimation of the volumetric fraction of oil-water-gas1980s. The measurement of component ratios in multiphase flow, and has become a constituent part ofmultiphase flows using γ-ray attenuation was first radial base process tomography (Grassler and Wirth,suggested by Abouelwafa and Kendall (1980), and2001).the technique has been used in many current com-mercial multiphase metering systems. Single energyγ-ray technique working as a densimeter is satisfac-DUAL ENERGIES THEORIEStory for two components measurement of gas/ liquidor oil/water pipe-flow if we do not care about the"he following absorption law describes theconcentration distribution of the different components mathematical connection between the γ-ray intensityin the cross-section. Dual-energy γ-ray techniques Io radiated by the y-ray source and the remaininghave developed rapidly in the last decade. The dif- intensity I after trans| 中国煤化工objetferent attenuation properties of the three phases' me- of given length L ariechti,TYHCNMH G1406 .Li et al. /J Zhejiang Univ SCI 2005 64(12):1405-14111955; Mormeburg, 1995): .(ρ=1.00 g/cm' ) has higher density than most mineraloils (typically ρ=0.80~0.90 g/cm).I= I, exp(-n(ZE)pL)2.5The absorption coefficient n is a function of the γY-ray2.0energy E and the atom number Z. Under the y-ray1.s-Waterenergy E, then: .- Oil1.I(E)= I。(E)exp(- -u(E)L)0.0.0For oil-water-gas three-phase flow, the attenuation20 4(0 8(100coefficient of the mixture pu(E) is represented by:Photon energy (keV)Fig.1 Water and oil attenuation coefficientsμ(E)= au(E)。+ βu(E)w + yu(E)。Since the photoelecric interaction is a strongerwhere μ(E)o, u(E)w and u(E)g are the linear attenua-function of photon energy and atomic number thantion coffcients of the oil, water and gas and a,B,y the Compton interaction, there is greater contrast inare the respective volumetric fractions. The trans- the linear atenuation cofficents of oil and water inmitted intensity I through a thickness L of anthe region where the photoelectric effect dominatesoil/water/gas mixture is therefore:the interaction cross-section of both materials (Key,1999). This is ilustrated in Fig.2, where the relative1。exp[- -(au(E)。+ βu(E)w + nu(E)4]difference in the linear attenuation coefficients isplotted as a function of photon energy for the sameif the build-up factor is eliminated through good col-energy interval. It indicates that the γ-ray measure-limation at the source and detector. If the transmitted ment system relying on photon atenuation in oil andfluxesL| and 12 at two energies E and E2 are measured, water to distinguish the two materials would yieldthe volume fractions can be calculated if the linearmaximum discrimination in the energy region belowattenuation coefficients of the flow components at E40 keV.and Ez are known since:80In(/I)1 =-[au(E)。+ βu(E). + yu(E)]L (1)60(uwuer Tua)/wmwerIn(I/0)2 =-aqu(E2)。+ Bu(E2)w +yu(E2)]L (2)4(a+β+y=13)200There are three equations with three unknown)40608010volumetric fractions a, β and γ. In practice, μ(Ei)g andμ(E2)g can be taken as zero without appreciable errorin the volume fraction measurements because of theFig.2 Radiographic contrast between oil and watergas phase' s low density.The linear attenruation coefficients of water andIn this project, the γ-ray system is comprised ofmineral oil (kerosene) over the same energy range two radioactive isotopes of 2 " Am and 13 'Cs whichreveals that the differences in the photon absorption have emission energies at 59.5 keV and 662 keV. Themay be used to distinguish the two materials. Fig.1 lower Y-ray energy is higher than 40 keV, so that theshows that the photon attenuation in water is greater linear attenuation coefficients of two energies in oilthan that in oil. This is because oxygen has a higher and water are som-d. Theatomic number than carbon, and also because water necessary condition中国煤化工inearlyTYHCNMHGLi et al. /J Zhejiang Univ SCI 2005 64(12):1405-14111407independent must be:the detector.Nuclear instrumentμ(E)w大u(E,)wThe nuclear instrument used in this project wasμ(E)。u(E)。designed as a multi-channels instrument. The systemis operated in count mode. It is comprised of highBut the too close linear attenuation coefficients of thevoltage source, amplifier, shaping amplifier and pro-two energies may translate small intensity measure-grammable data acquisition system (Fig.4).ment errors into large errors in thickness estimation;this situation is avoided only by proper selection ofthe energies.EXPERIMENTAL SETUPγ-ray sourceFig.3 Scintillator detectorIn this project, the γy-ray system is comprised oftwo radioactive isotopes of 24 Am and with emissionenergies of 59.5 keV and 662 keV. Both radioactive-DetectorData acquisitionAmplifierisotopes were assembled and shielded in a thick leadpot together to prevent harmful high energy emissionof 135 'Cs. The radiation intensity of the two isotopesvas 100 mCi and 20 mCi respectively. The reason forShape amplifierchoosing the radiation intensity of 4Am was it haslower photon energy than 1 ”Cs and so, weakens after. High voltage sourcepenetration into the measurement pipe. A collimatedsingle γy-ray beam (diameter 20 mm) comes out fromFig.4 Nuclear instructionthe bottom of the source pot and can be turned on/offby a switch mechanism which can ensure operationsafety.Experimental stack and vesselAs the experiment was static test, an experi-Scintillator detectormental stack and 4 square vessels were designed as anAn important step in dual energies rY-ray design unclear static test stack which stood apart from theis the selection of scintillation detectors. Two pa- multiphase flow loop, and could also serve as a staticrameters are important in selecting the scintillation calibration stack in further work. The stack was one :detectors: detection efficiency and decay constant of meter high and supported the source pot on its top.the scinillator. High dection efficiency is required Plexiglass vessel [100 mm (I)x 100 mm (WV)x600 mmto reduce the source strength and short decay constant (H)] with 5 mm thick wall was designed to accom-is required for high count rate to avoid pulse pile up or modate oil-water-gas three phase media. The vesselsaturation. NaI (T) crystal is the most commonly worked in horizontal direction and had total thicknessused scintillator due to its high detection efficiency;it of 110 mm and valid space of 100 mm. Its positionis very important for 137Cs due to its high emission (height) between Y-ray source and detector could beenergy (662 keV) and strong penetration ability.adjusted by two jacks depending on the experimentA 40 mm (H)x40 mm (D) column crystal of requirements (Fig.5). .scintillator was made and assembled with a photomultiplier tube (PMT). Fig.3 shows that the totaldiameter of the detector is OS55 mm and the length is EXPERIMENTAL ARRANGEMENT220 mm. In addition, a collimation 30 mm (W)x50mm (L)x 150 mm (H) hole was installed on the top ofIn order to中国煤化工:uationMHCNMHG1408Li et al.1J Zhejiang Univ SCI 2005 64(12):1405-1411character, an experimental arrangement was designed RESULTS AND DISCUSSIONSon Table 1, total 22 measurements included in theexperimental arrangement. In the valid space of 100The static measurement of items 1~11 in Table 1mm of the vessel, a minimum step of 10 mm media simulated the gas-water two phase flow. Although itwas changed during the experiment.was the simplest state in multiphase flow, the resultsshowed that it had unexpected measurement error(Fig.6). The error curve is shown in Fig.6b. As shownin Fig.6a, measured only by 24lAm, the experimentalvalue was obviously higher than the theoretical value,especially at the medium thickness of 2~6 cm, wherethe errors reduce gradually to the ends of the curve.The maximum value of the error was nearly +2.6%.However, the circumstance was better in results of157 Cs; the maximum error was not more than +1.8%(Fig.7).-o- Exp.-0- The. .50Air-water two phase (Am)Fig.5 Experimental stack令0Table 1 Experimental arrangementThickness (mm)Items -0102030405060708090100G1012345678910Thickness (water) (cm)(awW.6Ai-water two phase (Am)4-W, G11 F1o-02823456789100Thicknes (watr) (cm)(b)1:CFig.6 Gas-water two phase attenuation measured by24Am (a) and its error curve (b),0W,GCareful study of the measurement values,showing that the count rates of all the experiment2(O,Gpoints were higher than theoretical rates, indicated2O, Wthat there were extra counts added to the measurementW, CO=oil, W=water, G-gaschannel so that posi中国煤化工intheexperiment value. C:d to beMYHCNMH GLi et al. /J Zhejiang Univ SCI 2005 64(12):1405-1411140917-0- Exp.50-o- The.-0- The.1:Air-water two phase (C)Air-oil two phase (m)多40三13-莫123 3081111(25|2(8δ→τ2345678910012Thickness (oil) (cm)Thickness (water) (cm)a)(a1.0.20 r1.(Air-oil two phase (4m)0181Air-water two phase (C8)0. 16-0.9-0.8-0.70.080.000.4- -o-y--0.0479xr+0.4168x+0.1170.02 I.00[- 00712+0.0621x+0.024103234567801234678910Thickness (oi) (cm)b)Fig.8 Gas-oil two phases attenuation measured byAm (a) and its error curve (b)Fig.7 Gas-water two phases attenuation measured by37Cs (a) and its error curve (b)16the main explanation of this phenomenon: the higher__ Exp.energies Y-ray emitted by 5 'Cs interact with the me-Air-water two plhase (Cx)dia cross-section to generate the Compton scatter. AsCompton scatter has a widely distributed continuousspectrum, a part of the lower energies emission of12-Compton scatter enters the-4 'Am measurement11channel thus accounting for the occurrence of extracount. In general, this error is very harmful for the .012345678910measurement system that means the measurementaccuracy is flow regime dependent. It seems that theinfluence of scatter is less in the”Cs measurement0.12channel; this is because the Compton scatter always0.10Ai-water two phase (Cs)emits lower energies to all the directions of space butrarely enters the5 Cs itself in the measurement0.06channel. The main way to solve this problem is well0.04collimated on scintillation detector or modifies the0.02experiment values with the calibration error curve;-o-. -0.0082 +.1003 0.155subtracts the extra count according to the error curvefrom the measurement values.i一.23456789The static measurement of items 12~ 16 in Table(b)1 simulated the gas-oil two phase flow, the results andconclusions are the same for gas-water two phaseFig.9 Gas-oil two| 中国煤化工°d byflow (Figs.8 and 9). .isCs (a) and its errYHCNM HG1410 .Li et al.1J Zhejiang Univ SCI 2005 64(12):1405-1411The static measurement of items 17~22 in Table three phase pipeflow should bemixed in different1 simulated the oil-water-gas three phase flow. In the distributions. But are the results of dynamic tests thethree phase volumetric fraction measurement, assame as those of static tests? Perhaps further dynamicmentioned in Section 2 of this paper, the γ-ray studies should be done to answer the questions, but asmeasurement system relying on photon attenuation in we estimate the two results are close because theoil and water to distinguish the two materials yieldeddual-energy gamma-ray technique is based on mate-maximum discrimination in the energy region below rials attenuation coefficients measurement, the mate-40 keV, but the lower energy isotopes of 24 Am used rial's attenuation dose rate is not influenced by thein this project has emission energies at 59.5 keV，medium distribution in same radioactive area.therefore the linear attenuation coefficients of theenergy in oil and in water is closer than expected.The too close linear attenuation coefficients may0 Actual water0 Actual oil0 Calculated water 0 Calculated oilmagnify small errors in intensities measurements into0 Modifed waten_ nl I Modifed oillarge errors by Eqs.(1), (2) and (3), thus a modifica-tion arithmetic was developed to improve the threephase measurement accuracy with the calibrationerror curve. The results are shown in Fig.10.Actual thicknessCalculated value directly3467Calculated value by modificationGroupFig.11 Histogram of oil-water-gas three phase meas-stAir-oil-water three phase (4m+Cs)urementCONCLUSIONDual-energy γy-ray techniques are based on ma-Thickness (water) (cm)terials attenuation coefficients measurement. In oil-Fig.10 Oil-water- gas three phase volumetric fractionwater-gas three phase system, if the lower γ-ray en-measurementergy above 40 keV (59 keV 4Am in this project), thelinear attenuation coefficients of the energy in oil andAs shown in Fig.10, corresponding to the sixin water are somewhat close to each other; smallgiven points, the six calculated values obtained di-errors in the intensities measurements maybe ampli-rectly from Eqs.(1), (2) and (3) have significant bias,fied into large errors in the thickness estimation ob-the error distribution is seriously flawed. Anothertained by Eqs.(1), (2) and (3).group of six points calculated by modification arith-As an acceptable measurement technique, it is .metic have reasonable accuracy and acceptable error.required that the system has high accuracy of materi-The histogram of the results of measurement is shown als atnuation coffcients. The modification arith-in Fig.11.metic is helpful for removing the extraneous Comp-The histogram of oil-water-gas three phaseton scatter from the measurement values.measurements revealed that the modified results haveFinally, a well designed dual-energy γy-ray sys-acceptable accuracy maximum error of not more thantem has the potential to yield accuracy better than95% in average in oil-water-gas three phase system.6% (full scale) for every phase.All the above results and discussions were basedon static experiments, the oil-water-gas three phase Referencesmedia distributed in three clear layers in the vessel,Abouelwafa, M.S.A.,中国煤化工-urementbut the oil-water-gas three phase medium in actualof component ratYHCNMHGg Y-rayLi et al. /J Zhejiang Univ SCI 2005 64(12):1405-14111411attenuation. J. Phy. E: Sci. Instrum, 13:341.mographic Multiphase Flow Measurerment. Ph.D Thesis,Grassler, T, Wirth, K E., 2001. Dual-Energy X-Ray Tomo-University of Surrey.graphy in Process Engineering A Non-Intrusive Tech- Minder, W., Liechti, A., 1955. Rontgenphysik. Springer Ver-nique to Characterize Vertical Multiphase Flows. 2ndlag, Wien. .World Congress on Industrial Process Tomography， Morneburg, H.(Ed.)， 1995. Bilgebende Systeme fir dieHannover, Germany .medizinische Diagnostik.Publicis MCD Verlag,Key, M.J, 1999. Gas Microstructure X-Ray Detectors and To-Minchen.Editors-in-Chief: Pan Yun-he(ISSN 1009-3095, Monthly)Journal of Zhejiang UniversitySCIENCE Ahttp://ww. zju.edu.cn/jzusJZUS-A focuses on“Applied Physics & Engineering”➢Welcome Your Contributions to JZUS-AJournal of Zhejiang University SCIENCE A warmly and sincerely welcomes scientists all overthe world to contribute to JZUS-A in the form of Review, Article and Science Letters focused onApplied Physics & Engineering areas. Especially, Science Letters (3- -4 pages) would be publishedas soon as about 30 days (Note: detailed research articles can still be published in the professionaljourmals in the future after Science Letters is published by JZUS-A.Contributions requests(1) Electronic manuscript should be sent to jzus@zju.edu.cn only. If you have any questions,please feel free to visit our website: http://www.zju.edu.cn/jzus, and hit“For Authors'".(2) English abstract should include Objective, Method, Result and Conclusion.(3) Tables and figures could be used to prove your research result.(4) Full text of the Science Letters should be in 3-4 pages. The length of articles and reviews isnot limited.(5) Please visit our website (ttp://www .zju.edu.cn/jzus/pformat.htm) to see paper format.中国煤化工MHCNM HG