Chinese J. Chem. Bag, 14 (5) 491 ~ 496 (2008) An Experimental Study on the Flow Characteristics of Oil- Water Two-Phase Flow in Horizontal Straight Pipes* LIU Wenhong(#| x #r), GUO Liejin(% AL#)"*, WU Tiejun(% &#) and ZHANG Ximin(# 8 &) State Key Laboratory of Multiphase Flow in Power Bugineering, Xi'an Jiaotong University, Xian 710049, China Abstract The flow patterns and their transitions of oil-water two-phase flow in horizontal pipes were studied. The nts were conducted in two kinds of horigontal bes, made of plexiglas pipe and stainless stel pipe with 0mm ID respectively. No.46 mechanical oll and tap water were used as working Maids. The superficial velocity ranges of oil and water wore: 0.01—L2ms~! and 0.01—2.2ms"*, respectively. he flow patterns were identified by visualization and by transient fluctuation signals of differential pressure drop. Tho flow patterns were defined according tothe relative distribution of oll and water phases inthe pipes. Plow pattern maps were obtained for both pipelines. In addition, somitheoretical transition criteria for the Mow patterns wore proposed, and the proposed {ransitional erteria are in reasonable agreement with available data in liquid-liquid systems, 1 INTRODUCTION In pipelines for oil production, mixtures of oil and water are transported over a long distance. Compared with gas-liquid and liquid-solid two-phase flows, there are few studies concerning cil-water two immiscible liquid Hows. Understanding of the features of oil-water flows is not only extremely important for the design of pipelines and production logging instruments, but also crucial for prediction of the amount of free water in contact with the pipe wall that could eause corrosion problems!~S1. Knowledge of the features of oil-water systems, together with those for gas-liquid systems, can bo used a a basis to understand more complex cases of gasoil-water mixtures in Ube future Liquid-liquid flow in pipes may adopt diferent flow configurations, depending on the situation of the liquid-liquid interface. The flow patterns play an e5- sential role in oi-water flows. Russell et all found three flow patterns in a horizontal pipeline: bubble, stratified and mixed flow. Arirachakaran et all!) pre- sented experimental oil-water flow pattern maps, and found that the fow pattern in oikwater mixture de- pended primarily on mixture velocity, water fraction and oil viscosity when oil is continous phase. Nadler and Mewes!) investigated on the flow pattern of oik water two-phase flow in a horizontal straight pipe with 591mm ID. The flow patterns were distinguished in de- tail and a flow pattern map was presented, a Ute study no intermediate flow was observed at all. Li et all reported an experimental research on the oil ‘water two-phase flow patterns and their transitions in both hotizontal and vertical helically coiled tubes Received 2003-02-18, accepted 2008-04-25. jpported by the Nat The effect of dimension and location of the helically coiled tubes on flow pattern transition and phase in- version point was also examined. Although lots of work on oil-water two-phase flow has been conducted in the past 50 years, some un- certainties still exist. As pointed out by Hewitt! the transition from gas-liquid flow to liquid-liquid two- phase flow study is still a difficult journey. A lot of ‘work has to be done before we could get a clear, deep and generalized understanding in this area. For the purpose of providing more data and deeper understanding of liquid-liquid two-phase flow, we con ducted some new experiments in two kinds of horizon- tal straight tubes, which are made of plexiglas pipe and stainless steel pipe with 40mm ID respectively, and classified flow patterns in the present work. In ad- ition, semi-theoretical transition criteria for the flow patterns were proposed. 2 EXPERIMENTAL APPARATUS AND PROCEDURE, 2.1 Experimental apparatus A new testing facility for air-oil-water three-phase flow was designed, constructed and operated in our laboratory (se Fig.1). It is capable of coping with three phase flows in pipelines at a pressure up to 8x 10°Pa. An air compressor supplied air. A me- ctianical oil of No. 45 with viscosity of (1310.5) mPas at 20°C was used in experiment. Oil and water were pumped by screw pump and centrifugal pump respec tively and mixed in a liquid-liquid mixer and then en- tered into the testing section before through a three nal Natutn! Science Foundation of China (No.59905462 and 10172060). s+ Torwhom corzexpondence should be addressed. E-mail: l-guotmailxitu.edu.cnwn phases fluid mixer. In present experiment, only oil and water were employed. After passing through the testing section, the mixture was led into a separation tank, where oil and water were separated by gravity, Finally the oil and water were pumped back to a stor- age tank for further circulation. bed sky be >» om ball ase chet lve pressure page once plate coo ale ameyahe temperature pwns oil-sir flow facility ‘sil pump: 4-—eilloop orifice plates; S—eil sample; ‘6 water tank; T~ water pump; 6 water stabilzer; 9 —water-loop orifice plate; 10—air compressor; Lair stabilizer; 12—gas-loop orifice paves 13—-nolve eeducer; 14 gusliquid weparaton, 15 water separator; 16—wavy-plate units 1Tothree-phase mixer; 1S—testing section ‘The testing sections include two parallel straight pipes, shown schematically in Fig.2. One of them is ‘made of a plexiglas pipe with internal diameter 40mm, and the other is made of stainless steel pipe with the samme inner diameter. ‘Chinese J. Ch. B. (Vol. 11, No.8) pattern was distinguished visually and complemented with the analysis of the pressure drop signals. All signals representing the flow rates, temperature, pres- sure and differential pressure wore transmitted to a Philips computer and recorded by using IMP(isolated measurement pod) acquisition system. 2.2 Experimental procedure The oil-water flow experiments were conducted both in acrylic and stainless pipelines approximately 20m long, which was installed on an inclinable trestle ‘A detailed discription is given below During these oil-water two-phase flow experiments the water low rate was kept constant and the oil flow rate was increased stepwise. At each combination of cil and water flow rates, the flow pattern was care: fully observed and recorded, and at the same time the other parameters were also sampled and recorded. In order to minimize the random errors, all the val- vues were averaged using 1000 separate samples. The above procedure was repeated until the final oil low rate of interest or the maximum oil flow rate was reached. Then a new water flow rate was chosen and the above procedure was repeated. For these oil- water experiments, the oil superficial velocity ranges from 0.04m-s~! to 1.2m-s~! and the water superficial velocity from 0.04m-s~! to 2.2ms~!. The volume fraction (water-cut) of water in the oil-water mixture changes from 0,05 10 0,976, The system pressure is in the range of 1 x 10°Pa to 2 x 10° Pa (gauge) and the temperature of the working fluids was kept in the range of 5—20°C. No. 46 mechanical oil was used. Be- fore the experiment, the oil properties, such as vis city, deus wud interfacial Gecsion with water were measured, which are shown in Table 1. ‘Table 1 Properties of oll used in present studs tines ste pipe ceric resin pipe Figure 2 Schematic diagram of testing sections The flow rates of oil and water were measured by using orifice plates with different ranges, which were carefully calibrated prior to the experiment. The tem- peratures of the working fluids in the testing section were measured at the inlet by using thermocouples with an accuracy of 0.2°C. The pressure of the working fluids in the testing section was also measured at the inlet. Two DPLL51 differential pressuro transducers ‘were used to measure the pressure drop over the whole testing section. The arrangement and location of the pressure transducers were shown in Fig.2. The flow October, 2003, enaty po = 883001 gm Viscosity jug = 428.518e~0.099 109 Pas Sinverfacial vemion 208 Nm 27.72e00018 10-9 3 RESULTS AND DISCUSSION 3.1 Classification of flow patterns As above mentioned, some previous investigations oon oil-water two-phase flow were carried out and dif. ferent flow patterns were distinguished, however, the ifferences still exit. In present investigation, seven flow patterns were distinguished and observed which are schematically shown in Fig. 3 Stratified flow pattern (ST) occurs at the lower mixture velocities, in which the two fluids flow in sep- arate layers on the top and the bottom of the pipe according to their densities and the location of inter- face [see Fig 2 (a).‘The Plow Characteristics of Oil-Water Two-Phase Flow in Horizontal Straight Pipes 493 om (2) Stati’ wavy flow @ Fx () Thee layer low @ GES (6) Pally dispersed flow (6) Two layer ow @ aa (3) Watercbascd annular Bove Or (b) Stratified wavy drop tow @ Re, (4) Oitdroplet stratified Sow Ors (0 Sluz ow (8) Oinbased aanular flow Mt DD wee AAs the mixture velocity inereased drops appeared in the interfacial region and the flow pattern became stratified wavy with drops (SWD) [see Fig. 3(b)]. This flow pattern exits over a higher velocity range in the stainless steel pipe (Uso = 0.055--0.1002 ms! and Usw = 0.055—0.101 ms~') than in the plexiglas Pipe (Usu = 0.054 0.091mo~! and Usw = 0.051— 0.091 ms"). Three layer flow pattern (3L) exhibits distinct con- tinuous layers of oil and water at the top and bottom ‘of the pipe respectively but in the interfacial region there exists a layer of drops, in which drops of one phase appear within the other phase [see Fig. 3(c)] When R equals 1:1, 1:2, 2:1 and 3:1, for moder- ate superficial velocity, the 3L flow pattern is obtained both in plexiglas and stainless steel pipes. For exam- ple, when R= 1: 1, the velocity range is Uso ~ 0.1— 0.121 ms“! and Usw = 0.1—-0.121 ms! in plexiglas pipe, while in stainless steel pipe, the velocity range is Uso = 0.125~0.357ms"! and Usw = 0.122 0.348 ms“ Cilkdroplets stratahied low (ODST), which is simi lar to the mixed flow pattern defined by Arirachakaran, et all!l, occurs at higher water fractions. Tn this flow pattern, the oil phase is characterized by different drop size existing in the continuous water phase. Owing to the effect of the buoyaney force, all the oil droplets agglomerate at the upper part of the tube, while the water phoce flowe at the lowor past [soe Fig 4(@)] ‘The flow patter is obtained in the two pipes when is less than 5:1, In the plexiglas pipe, oil superficial velocit ranges from 0.8m-s"? to 0.509ms~! and perficial velocity ranges from 0.061 ‘while in the stainless steel pipe oll superficial veloc ranges from 0.4031m-s~! to 0.5 ms"? and water super ficial velocity ranges from 0.081 m's~! to 0,102 ms~* Fully dispersed or mixed flow pattern (M) is shown in Fig. 8), in which one phase is dispersed into the other more or less uniformly and occupies the whole pipe cross section. It is similar to the dispersed flow defined by Arirachakaran et atl"! in horizontal flows and to the oil bubble flow observed by Govior et al!) in vertical oil-water flow. According to experimental observation, all the oil phase was characterized by oil Aroplets and distributed uniformly in the continuous water phase, This flow pattern occurs at higher mix- tuze velocity. For all ratios of oil to water, the flow pattern was obtained in the experiment. Slug flow occurs under the condition of higher wa- ter flow rate and lower oil fiow rate, where large con- tinuous oil pockets similar to Taylor bubbles can be observed. The oil-water interface is not stable enough and the pocket could break up very quickly into small droplets [see Fig. 3(f)]. When R < 1:3, the slug flow appears in the two pipes, and the corresponding su- ficial velocities are Uso = 0.06—0.072m-5~! and Usw = 0.180214 ms“! "Two layer flow pattern (21) oceurs at higher rates of oil to water at lower mixture velocity, where the continuous oil layer is at the top of the pipes, a layer of droplets is at the bottom and the interface of two Layers diminishes (see Pig.3(g))- Tn the plexi- slas pipe, when R equals 5 : 1, the flow pattern ap- pears first. The corresponding superficial velocity is Uso = 0206s"! and Usy = 0.0¢ms-!. When R is larger than 5:1, the 21. flow pattern is easily ob- tained. ‘The flow pattern in the steel pipe has the same trends, and the first spatial distribution appears when Uso = 0.265ms~" and Usy = 0.053 ms il based annular low (OA) occurs under the con- dition of higher ol flow cate and lower water flow rate, ‘where the oil phase along the tube wall is a contina- ‘ous phase, while water in the core of the tube is water droplets and water filaments [see Fig. 3(h)]. When Ris Jargor than 9:1, the oil based annular flow is obtained. in the plexiglas pipe when Uso = 0.469--0.082m-~! and Usw = 0.048—0.068 m-s™ and a little difference in the stainless steel pipe, Uso = 0.469—1.006 ms~t, Us = 0.033—0.187 ms Iv is worthwhile to note that the oil-water flow pat- tern of water-based annular flow [see Fig. 3(i)} was not observed in the present experiment, which is mainly due to the low viscosity oil employed in our exper- iment, Tt war concluded that such viscosity is not high enough to keep a stable oil core in the water anularl 39) 3.2 Plow pattern maps Seven flow patterns were observed and their spa- tial configurations of the oil-water interface are shown in Fig.3. The corresponding flow pattern maps with superficial velocity coordinates are presented in Figs. 4 Chinese J. Ch. B, 11 (5) 492 (2008)494 Chinese J. Ch. B. (Vol: 11, No.5) and 5. As complement, flow pattern maps based on mixture velocity Um = Usw-+Uso and the water frac- tion ew — Usw /Um are also shown. Superficil veloc ity of the oil (or water) is defined as the volumet- rie flow rate (measured by the orifice plates) per unit cross-sectional area of the pipe. In Fig. 4 the solid lines correspond to a constant oil-water ratio, while R= 0 is for water only and R = oo for oil only. Uae! Ugnest OCTET TTT TOF TS OTTO Figure 4 Flow patterns in the plexigias test section In the range of low superficial oil and water ve locities the stratified flow regimes (ST) are observed. For R <1: 3 the oil is dispersed in droplets within the water phase at the top of the pipe. Increasing the velocities of both phases promotes turbulence in the flow and the formation of dispersion, "The experiment indicates that flow pattern and their transitions have similar trends in plexiglas pipes and stainless steel pipes, but the quantities are shifted. 4 MODELING FOR FLOW TRANSITIONS ‘The flow structure of cilwater mixtures in pipes is quit different from that of gas-liquid mixtuzes™!000, The different flow structure is mainly caused by the large liquid-liquid momentum transfer capacity and small buoyancy effects. Also, the lower free energy at the interface allows the formation of shorter inter facial waves and smaller droplet size. Therefore, oik water systems are expected to behave differently from gas-liquid systems? Unfortunately, Tittle work ms been done in de- PATTERN October, 2003 scribing the flow regimes that occur in oil-water two- phase flow!?*l. ‘This study addresses the determina. tion of the oil water flow pattora when the Buia phys- ical properties (density, viscosity and interfacial tea- sion), the operational variables (phase flow rates) and geometrical variables are given. Figure § Flow patte test section 4.1 ST/non-ST (3L, ODST and Slug) bound- ary For low oil and water superficial velocities, the flow is gravitationally dominated and the phases are segre- gated. A further increase in the flow rates causes the appearance of droplete close to the interface. When R is less than 1 : 3 (the water fraction is 75%), the oil superficial velocity is larger than 0.06m-s~!, and the water superficial velocity is larger than 0.18m-s~, the slug flow pattern appears; when R <1: 5, any strat- ifled flow disappears in this experiment and slug flow appears. The reason is the wettabillity of oil to the pipe wall and large water flow velocity. When was larger than 1:3, the breakup of the interface will not occur due to small energy dissipation in the pipe wall region. Thus the 3L- flow pattern is formed. As flow rates increase, the region of droplet layer increases, until the ODST is formed. ‘Taking into account the interaction among oil inertia force, the water gravity ‘and the interfacial tension between oil and water, to- gether with Froude number and Weber number, we‘The Flow Ch can obtain Kutadelaze number Kum (WeFr3)% = Usw 085 /ialow — pu)oowl? «) Regressing the experimental data, the following equa tions are obtained Ku =9.2(Rew/ Reo) @ so (ogo) = 12 pwHoUsw/pouwUso) (3) Standard error of the estimation is 0.18, and coeti- Gent of multiple determination is 0.94. ‘The result is shown in Fig. 6. Uses Figure 6 Comparison between predicted and ‘measured flow pattern transitions in plexiglas stainless steel pipes 4.2 3L/ODST boundary In oik-water two-phase flow at sufficiently low rates ratio, the oil phase may form a swarm of drops (or bubbles) dispersed in water phase, Due to the fioat- ing process and consequent coalescence, the dispersed phase drops may maintain their shape and dispersed pattem provided they are small enough, so that stir face tension overcomes them due to buoyancy. Based on Taitel and Dulker's analysis of gas-liquid ‘two-phase flow in horizontal pipes!" for the transition from slug flow to dispersed low, the transition from SL to ODST is mainly induced by buoyancy force and. water turbulence intensity for oil-water two-phase flow in horizontal stratified pipes. When the effect of wa- ter turbulence intensity prevails over that of buoyancy force, mutual penetration of vortices takes place in cach of the phases. This interfacial turbulence lends to the dispersion of the liquids. According to the effects above, the following criterion for ODST is obtained Wew =0.103(p0 /pw) 1" (nw /H0)-*°h(Us0/Usw) 1" ® where Wew = pwtldwd/oow acteristics of Oll-Water Two-Phase Flow in Horizontal Straight Pipes 495 Standard error of Eq, (4) i 0.011, and coefciont of cnuliple determination is 0.98, ‘The result is ako shown in Fig. 6. 4.3. Mixed flow pattern boundary As discussed above, with the increase of Usw while iaintaining low Uso, the turbulent breakage force tends to destroy large bubbles. According to Hinze’s model", originally developed for liquid-liquid disper- sion, the maximum dispersed phase diameter i elated to turhnlent dissipation eealed by 4 = Ci(eo)lz(oow pw)? (5a) a= 2fU3/d (5b) dex = {60/\(ow ~ po) 9)}°° Ge) And U, co + Uw (Bs) J = 0.046(pdU / ww)-®* (6b) €0 = Uso /(Uso + Usw) (Ge) where the constant C,(éo) may be generally related to in situ hold up and is to be determined by experi- ments. Utilizing Eq.(6) in Eq. (5) yields a fdee = Cale)Weg Print Reg (7) where Wem = cow/poUind (7a) Frm = poUa/ Sead) (7) Rev = poUnd/tve (Fe) Obviously, as the maximum drop sive is sivaller than the critical size, small drops maintaining their identity and a fully dispersed pattern may pre- vail. Thus, a fully dispersed pattern is expected for ya {dee <1. ‘In Eq. (7), letting dya/der = 1, one may obtain C2 according to the experimental data, Utilizing Ea.(5), the following is deduced Cp = Weg Prt? Rez ©) Using transition data in the experiment, we can obiain Cas Cy = 272650? ©) Standard error of Eq.(9) is 0.22, and coefficient of ‘multiple determination is 0.89, From the above discussion, the semi-theoretical transition criterion is obtained 21a 58% = ieee! MPO (10) Given the properties of the fluids and Uso, we can calculate Uy, and the superficial water velocity Usw. ‘Tho vom shown oF Chinese J. Gh. B. 11 (6) 401 (2003)496 ‘Chinese J. Ch. E. (Vol. 11, No.5) 4.4 Oil-based annular boundary ‘Transition criteria for gas-liquid two-phase flow ‘can give some notes to oikwater wwuxphase: flow. Ln ‘gaseliquid flow, gas inertia force and liquid gravity force are Key factors, but the surface tension may be more important for oil-water two-phase flow!™!*l ‘Thus we can use Froude number and Weber number to describe the transition for annular flow in oil-water ‘two-phase flow and obtain the Kutadelaze number Ku= (WeFr3) = Vsoe8* /lalow ~ po)oow? (a1) Here, the Froude number and Kutadelaze number were used forthe transition criterion for oil based an- nla low (Fru? & (12) Standard error of Eq, (12) is 0.023, and coefficient of ‘multiple determination is 0.95. The result is shown in Fig. 6. 5 CONCLUSIONS ‘The following main conclusions can be dravin from cour investigation: (1) The flow patterns observed in experiments are similar to those defined by previous investigators, but there exists substantial differences in the conditions under which they occur. Seven flow patterns are defined and observed in plexiglas and stainless steel pipes. OiL-based annular flow appears at about water fraction below 10% in our experiments. (2) fu the horizontal pipes, the slug flow of oil water flow is different from that in gas-liquid flow Large continuous oil pockets similar to Taylor bubbles can be observed, the oil-water interface is not stable enough and the pocket may break up very quit small droplets. (3) Comprehensive models developed for gas-liquid flow cannot be readily extended to oll-water syste. Semi-theoretical transition criteria for the flow pat- terns were proposed. Comparisons of the proposed transitional criteria with available data in liquid-liquid systems are in reasonable agreement. ly into NOMENCLATURE Cu Ce constant 4 diameter of oil drop, Fe Froude number f Feietion factor 2 ‘gravitational constant, keana™ Ke Kutadelaze number 1 the mechanical power dissipated per unit mass, October, 2003, ratio of ol to water in volume Re Reynolds number £ temperature, °C UY fuid velocity, me" We Weber number © phase fraction liquid viscosity, Pass density of liquid, hem © interfacial tension, Nem=* Subscripts cr ctitical point of oll droplet mi minure veloly of liquid © oil phase SO superficial velocity of oil SW superficial velocity of water W water phase REFERENCES 1 Asirachalaran, S., Oglesby, D., Mainowsk, M.S, Shoham, 6, Brill, P, "An analysis of il water fow pienomens hoviaoatal pipes, SPELSSSE, 155-165 (1989. “raller, JL Intevep, S.A, Bil, LP. "A stay of wae ter flow patterns in horizontal pipes, SPE 36600, 968-875 (1986). 915, 6.,, Guo, LJ, Gao, H., Yu, La, Huang, 1.0, Chen, XJ, “Plow pattern of llwaterlgsidliquid two-phase fow in elcally coiled tube", J. Cherm. Ind. Eng. (China), 81 (2), 299-242 (200). (in Chinese) 4 Nadler, it, Mewes, D, “Flow tue amuisifeation the flow of wo immiscible igulds In horizontal pipes", Za J ‘Malthe Flow, 28 (1), 5508 (1997). 5 Chen, X., Guo, L., "Flow patterns and pressure drop in oilairwater three-phase flow through helally exiled tates Tne J. Multiphase Flow, 28 (6-7), 1083-1072 (2000). 6 Rosell, EWP, Hodgson, G.W., Govier, G.W.,“Horizon- tal pipeline fw of mixtures of oll and water’, Cun. J Shotnng, 37, 917 (1950). 7 Howitt, Gi “From gueligud to liquid-liquid two-phase flow, 2 dificlt journey”, Ins Proceedings of lt. Symp. on LiqhidLiquid Two-Phase Flow and Transport Phencmess, Bogell House Tae Analy, Turkey, 9-7 (1097) § Covi, G.W., Sullivan, GA, Wood, RK, “The upward vertical low of olrater mixtures”, Can. J; Chem. Eve, 50, 67 (1961) 9 Guo, LJ, "The oikgas-vater three plse fw and sand sparatioa theories in hlially coiled tubing separation technology”, Joura of Engineering Thermophyscs, 23 (1), 107-110 (2002). (a. Chinese) 10 Angeli, By Hewit, GF ‘iowster fw" Int (2000). Brauner, N., Maroa) Moaloa D, “Flow pattern transitions tworpbas> igus ow i horton tubs Int J ‘Mattiphase Flow, 18 (1), 123—140 (1992) 12 Charles, MLE, Govier, G.W., Hodgson, G.W., “The h teontal pipeline flow of equal dennty water mutates Can. J.Chem. Eng. 88,2736 (186). 13 Taitel, Y., Dulker, A.E,, °A model for predicting How regime trouiions in horioutal and near horisoal Bow", AICHE J, 22 (18), 47-95 (1970) LM Hinge, 1.0, “Pondamentals of the hydrodynamics mechan iesn of spliting in dispersion proces", AICHE J, 1 (3) 289 (1965). low structure in horizontal J. Multiphase Flow, 26, 11171140,