Experimental investigation of oilwater two-phase flow in horizontalpipes: Pressure losses, liquid holdup and flow patterns

Ahmad Shamsul Izwan Ismail a, Issham Ismail a,n, Mansoor Zoveidavianpoor b,Rahmat Mohsin a, Ali Piroozian a, Mohd Shahir Misnan a, Mior Zaiga Sariman c

a Malaysia Petroleum Resources Corporation (UTM-MPRC) Institute for Oil and Gas, Universiti Teknologi Malaysia, Johor 81310, Malaysiab Department of Petroleum Engineering, Faculty of Petroleum and Renewable Energy Engineering, Universiti Teknologi Malaysia, Johor 81310, Malaysiac Department of Process Engineering, Development Division, PETRONAS Carigali Sdn. Bhd., Kuala Lumpur 50088, Malaysia

a r t i c l e i n f o

Article history:Received 8 April 2014Accepted 31 January 2015Available online 7 February 2015

Keywords:Waxy crudeOil/water flowPressure dropHoldupFlow pattern

a b s t r a c t

Flow experiments have been conducted for oilwater two-phase flow in a horizontal 5.08 cm ID flowloop at a length to diameter ratio of 1311. The fluids were light Malaysian waxy crude oil from theoffshore Terengganu (o818 kg/m3, mo1.75 mPa s and wax content16.15 wt%) and syntheticformation water. The water-cut was varied between 10 to 90% at nine mixture flow rates of 2.0 to16.2 cm3/s. Measuring the changes in pressure drop and liquid holdup at different flow rates of oilwatertwo-phase flow, a new flow pattern was identified. Strong dependence of the oilwater slippage on theminimum flow rate was observed. The highest pressure drop of 11.58 kPa was obtained at maximumflow rate of 16.21 cm3/s and oil fraction of 0.9; while the lowest pressure drop of 1.31 kPa was recordedat the lowest flow rate of 2.03 cm3/s and water fraction of 0.9. The experimental results could be used asa platform to understand better a more complex case of gas/oil/water concurrent flow in a pipeline.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

The need for reliable experimental studies on many engineeringapplications of flow assurance has been the driving force behindextensive research efforts in the area of multiphase flow. Liquidliquid flow could be defined as the simultaneous flow of twoimmiscible liquids in a pipe. Previously, multiphase flow researchworks were mainly focused on gasliquid flow; among the earlieststudies in the gasliquid field were Beggs and Brill (1973), Wicks andDukler (1960), Hagedorn and Brown (1964), Gregory and Aziz (1975)and Cornish (1976). Nevertheless, the industry attention has shiftedtowards the understanding of the simultaneous flow of gasoilwater mixtures (Trallero et al., 1997). Despite the extensive studies ongasliquid two phase flow, liquidliquid flow has received inade-quate research attention (Atmaca et al., 2009). In the oil and gasindustry, simultaneous transport of water and oil in pipelines occursfrequently. For oil fields operating at high water-cuts and lowwellhead pressures, the effect of the water phase with respect topressure drop is of particular importance. Lack of knowledge of theflow patterns, pressure drop and in-situ distributions of the liquidscould be hampered the safe and economic transport of these fluids.

The gained knowledge via experimental analysis can contribute toaccurate modelling and prediction of oilwater flow in pipes.

Due to the dwindling of conventional light crude oil or easy oilreserves and the existence of lots of mature oilfields around theglobe, especially in the Malaysian oilfields, the phenomenon ofconcurrent flow of oil and water in pipelines has been the mainsubject of research studies in petroleum production and enhancedoil recovery with water injection. Furthermore, there are manycases where high water cut is present but the wells are stillconsidered economically viable to operate. Understanding thebehaviour of oilwater flow in pipelines, such as flow pattern,pressure drop, and liquid holdup is crucial for many engineeringapplications such as design and monitoring of the separationprocess, interpretation of production logs, and operation of flowlines and wells (Atmaca et al., 2009).

Some of the oilfields around the world are producing waxycrude oil. This phenomenon is due to the presence of paraffin (C18C36) and/or naphthenic (C30C60) hydrocarbons in the crude oil(Mansoori, 1993). When a crude oil contains waxes, the propertiesof the oil, especially the viscosity, will greatly change. There werenumerous two phase flow experimental studies on the significanceof viscosity, such as Russell et al. (1959), Arirachakaran et al. (1989),Oglesby (1979), Trallero (1995), Alkaya (2000) and Mckibben et al.(2000). All of these prominent researchers have found that theviscosity was greatly affecting the flow pattern, pressure drop, andliquid holdup.

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/petrol

Journal of Petroleum Science and Engineering

http://dx.doi.org/10.1016/j.petrol.2015.01.0380920-4105/& 2015 Elsevier B.V. All rights reserved.

n Corresponding author. Tel.: 60 7 553 5506; fax: 60 7 5581463.E-mail address: issham@petroleum.utm.my (I. Ismail).

Journal of Petroleum Science and Engineering 127 (2015) 409420

www.sciencedirect.com/science/journal/09204105www.elsevier.com/locate/petrolhttp://dx.doi.org/10.1016/j.petrol.2015.01.038http://dx.doi.org/10.1016/j.petrol.2015.01.038http://dx.doi.org/10.1016/j.petrol.2015.01.038http://crossmark.crossref.org/dialog/?doi=10.1016/j.petrol.2015.01.038&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.petrol.2015.01.038&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.petrol.2015.01.038&domain=pdfmailto:issham@petroleum.utm.myhttp://dx.doi.org/10.1016/j.petrol.2015.01.038

Aside, dealing with an oilwater mixture in a pipeline leads tounique and complex problems in the oil and gas industry due to itscomplicated rheological behaviour, and vast difference in pressuregradient encountered for different flow patterns (Arirachakaranet al., 1989). Although two phase flow of oil and water is normallyoccurred in pipes during production or transportation of petroleumfluids, its hydrodynamics behaviour under a wide range of flowconditions and inclination angles still creates a relevant unresolvedissue for the oil industry (Flores et al., 1999). Actually, the mainreported laboratory works on liquidliquid two-phase flow wereaccomplished using gas oil, mineral oil, or refined oil, and limitedexperimental studies were performed on waxy crude oil. Therefore,an unexplored territory arises in terms of the flow behaviour in apipeline when a waxy crude oil is introduced in a two-phase flowsystem. Accordingly, this crude that contains waxes would affect theflow behaviour due to its viscosity changes, complex interfacialchemistry, and natural emulsion effect.

Flow pattern is a particular type of geometric distribution of thecomponents in a pipe and many of the names given to these flowpatterns are now quite standard (Brennen, 2005). As emphasized byTrallero et al. (1997), the subject of oilwater flow cant be addressedin a unified way. Thats because of the diversity of oil properties (e.g.,viscosity, density, rheological behaviour, etc.), which makes theirinvestigation not only too broad and contentious but also importantand worthwhile.

Generally, we need an accurate prediction of waxy crude oilmultiphase flow behaviour to produce and transport the waxy crudeoil safely and economically. Waxy crude oils have complex flowproperties; although considerable research has gone into the solutionof specific industrial pipelining problems, a study devoted to theunderstanding of the behaviour of this material has not beenappeared in the literature. Thus, an experimental investigation hasbeen conducted to study the flow behaviour (i.e., flow pattern,pressure drop, and water holdup) of Malaysian waxy crude oilwaterflow in horizontal pipes. This study addresses the determination ofoilwater flow pattern for Newtonian and low-viscosity Malaysianwaxy crude oil above wax appearance temperature (WAT).

2. Literature review

Crude oil from reservoirs is pushed to the surface by the highunderground pressure (natural drive) and is flowed through theirrespective wellheads and pipelines for further processing. Normally,crude oil pipelines contain a fraction of water due to waterencroachment from an aquifer and among others; and waterpercentage tends to increase in pipes over time. The situation isworsened when the wells are still operating even though theproduction stream is producing at high water-cut (Ngan, 2010). Inliquidliquid flow studies, the necessity to understand the natureand flow behaviour of this type of multiphase flow is crucial due tothe existence of different mechanisms governing them and variousflow patterns configuration. Russell and Charles (1959), Russell et al.(1959) and Charles et al. (1961) were among the earliest researcherswho conducted studies on liquidliquid flows. Most of their resultsbecame a reference for the subsequence studies and also provided a

basic knowledge in understanding better the behaviour of a liquidliquid flow. This scenario has attracted numerous extensive researchworks on this area after a decade, such as Guzhov and Medredev(1971), Guzhov et al. (1973) and Hughmark (1971). Brauner (2002)found that a liquidliquid system is characterized by a low densitydifference between phases and this finding was supported byAtmaca et al. (2009). They explained that the oilwater systemusually has similar densities, a large difference in viscosities, andmore complex interfacial chemistry compared to gasliquid sys-tems (Fig. 1). However, a small density difference in terms of oilproperties (e.g., API 45 to API 10) implies tremendous differences incomposition, viscosity, etc.

Nadler and Mewes (1997) explained that the flow behaviour of oiland water in pipes is heavily relied on the droplet distribution of thedispersed phase and volume fraction of the phases. This dependency isdue to the effect of finite density difference between the oil and waterphases that is contrary to gasliquid flow system that possesses agreat density difference. A simultaneous flow of oil and water willcreate an oilwater emulsion since they mix together when flowing inpipes. This phenomenon completely changes the physical properties ofthe liquids. An emulsion which is formed in a dispersed systemconsists of two immiscible liquids. An unstable emulsion formedduring a dispersed flow could be separated into its original phaseswhen it was left in stationary at a reasonable amount of time(Arirachakaran et al., 1989). Besides, these emulsions may appear tobe a non-Newtonian or Newtonian rheological behaviour (Brauner,2002). The differences in characteristics are triggered mainly by thesmall buoyancy effect, lower free energy at interface, smaller dispersedphase droplet size, and high momentum transfer capacity in liquidliquid flows (Vielma et al., 2007).

An accurate prediction of oilwater flow behaviours, such aspressure drop, flow pattern, and water holdup are imperative inmany advanced engineering applications (Brauner, 2002); such asdesigning and monitoring downhole metering, water-lubricatedpipelines, production optimization, artificial lift design and model-ling, optimum string selection, and production-logging interpretation(Flores et al., 1999). Russell and Charles (1959) extensively studiedthe flow behaviour of oilwater system by considering the flowpattern, pressure drop, and liquid holdups. Russell et al. (1959) havesuccessfully observed the flow characteristics of oilwater in ahorizontal condition using a 2.05 cm ID pipe. They found three typesof flow pattern, namely bubble, stratified, and mixed flows. They alsoobserved that water holdup was greatly influenced by liquid inputratio and viscosity. In principle, co-current flows of liquidliquidmixtures in pipes are stable by considering the flow parameters (i.e.,superficial velocity of each phase, the mixture flowrate, the pipediameter, the surface tension, the finite density difference, pipewettability, and the ratio of viscosity of fluids, as well as the shearstress between the liquid phases). In spite of the parametersmentioned above, pipe plane inclinations also affect the flow pattern.This flow pattern includes horizontal and vertical flow conditionswhich has significant differences in terms of flow pattern identifica-tions (Oddie et al., 2003). Charles et al. (1961) conducted a study onoilwater flow in horizontal pipelines, and they encountered fourtypes of flow pattern namely; water droplets in oil, concentric waterwith oil flowing in the core, oil slugs in water and oil bubble in water.They explained that the viscosity and low density difference betweenoil and water were affecting the flow patterns significantly. Generally,in an experimental study, there are many possible flow patterns thatcan be observed in horizontal conditions apart from those mentionedby Charles et al. (1961) as found by other researchers like Brauner(2002) and Trallero et al. (1997).

Researchers like Vuong et al. (2009), Vielma et al. (2007) andTrallero et al. (1997) found that the pressure drop was stronglydepended on the flow patterns and flow rates. On the other hand,Atmaca et al. (2009) and Sridhar et al. (2011) stated that pressure

Oil

Water

Fig. 1. Examples of oilwater flow in a horizontal pipe.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420410

gradient was mainly influenced by the oil viscosity as well as theeffect of flow patterns transition. Nonetheless, Vielma et al. (2007) intheir experimental study of liquidliquid flow behaviour horizontallybelieved that the dominant effect for pressure losses was the flowregime which associated with the mixing fluid distribution in thecross sectional area of the pipe. They found that in a turbulent flow,the mixing fluids shifted from separated phases to dispersed flow,which has resulted in a greater pressure drop. Vielma et al. (2007)also stated that there was no significant effect of pressure dropduring high water superficial velocity at any oil superficial velocitybecause of the droplets that present at all flowing conditions.Rodriguez and Oliemans (2005), Poesio et al. (2008), Grassi et al.(2008), Angeli and Hewit (1998) and Stapelberg and Mewes (1994)have studied thoroughly pressure drop across horizontal, vertical,and inclined pipes under various operating conditions using oilsamples of low to high viscosity. The change of pressure gradientwas greatly related to phase inversion or flow pattern transitions(Arirachakaran et al., 1989). This phenomenon was supported bymany researchers in their studies on oilwater flow in pipes, such asTrallero et al. (1997), Flores et al. (1999), Poesio et al. (2008), Vuonget al. (2009), Wang and Gong (2010) and Ngan (2010). Trallero et al.(1997) highlighted that an insignificant pressure drop was obtainedafter the transition period from a stratified region to a dispersedregion. This was attributable to the loss of oil continuity where thewater has fully wetted the pipe wall. Consequently, the drag of theflow, which is caused by shear stress, is reduced and eventuallyreduced the pressure loss. Generally, when high viscous oil flowstogether with water, the mixture of high oil viscosity tends to form athin oil film on the pipe wall, thus it increases the shear stress andsubstantially elevates the pressure drop. Other elements that con-tribute to pressure drop are the pipe wettability and roughness. Astainless steel pipe can be water wetted, while flowing oilwatertogether in pipes will cause lower pressure drop. Soleimani et al.(2000) in their study on spatial distribution of oil and water in ahorizontal pipe flow have explained succinctly the pipe wettabilityseffect on pressure drop.

Water holdup (Hw) is known as in-situ volume fraction of waterover a total mixing liquids in a specific length of a test section and itis measured using quick closing valves. A value for liquid holdup orspecifically water holdup cannot be obtained analytically; it has to bedetermined using an empirical technique. This is a function ofvariables such as liquid properties, flow pattern, pipe diameter, andpipe inclination. Via an experimental study, Vielma et al. (2007)revealed that the Hw was found to have increased with the increaseof the superficial water velocity, vsw, and decrease of superficial oilvelocity, vso. However, Soleimani et al. (2000) stated that in a two-phase flow system, the arithmetic mean velocity of a mixture foreach phase does not fully represent the total superficial velocity.Thus, the in-situ fractions of oil and water were differing from thecorresponding input fractions. They found that the holdup wasaltered from the input volume fraction at a minimum flow rateand increased slip ratio. This trend, according to Charles et al. (1961)and Martinez et al. (1988), was due to the liquid that was in contactwith the pipe. It was likely to accumulate in the tubes at a slowervelocity. This also occurred in high viscous oilwater flow in ahorizontal system (Zhang et al., 2010). They discovered that the oilflows at the top of the pipe wall at lower velocity values. The totalliquid holdup can be calculated using equation of, HLHwHo.

There are many researchers such as Trallero et al. (1997), Zhangand Sarica (2006), McKibben et al. (2000), Rodriguez and Oliemans(2005) and Keskin et al. (2007) who have conducted experimentaland modelling studies focused on low-medium oil viscosity-waterflow systems of the multiphase flow. However, the study of the highviscous oilwater flow was limited in numbers. Even though otherresearchers such as Vuong et al. (2009) and Sridhar et al. (2011)came out with a comprehensive and practical classification of flow

patterns based on high viscous liquidliquid relationship, they didnot encompass the effect of using actual waxy crude oil in theirrespective research studies. In spite of being a permanent interestfor the petroleum industry, the issue of waxy crude oilwater flowsystem in pipes has barely been addressed in the technical lite-ratures. This issue is worsened when coupled with the oilwatermixture flow behaviour, which presents a complex problem forpipeline transportation and production of waxy crude oil. Thereason for this complexity is due to the influence of crude oilcompositions that makes a vast difference in the pressure gradientencountered for different flow patterns (Wang and Gong, 2010).

In early experimental research, Oglesby (1979) reported there were14 flow patterns, while others described only three to four (Russellet al., 1959; Malinowsky, 1975). Since the 1990s, with the advancedinstruments and techniques, different flow pattern parameters havebeen measured more accurately, and flow patterns of oilwater flowhave been analyzed objectively (Trallero, 1995; Trallero et al., 1997;Nadler and Mewes, 1997; Angeli and Hewit, 1998; Shi et al., 1999; Shi,2001). Wang and Gong (2010) have thoroughly studied the differencesand similarities of flow patterns and transition characters betweencrude oilwater and high viscosity oilwater flowed in horizontalpipes. They found that the mineral oilwater flow has been widelyused in the indoor experiments to study the two phase flowcharacteristics. Even though its density and viscosity were approxi-mately similar to the actual crude oil, they observed that the flowpattern and transition characters between mineral oilwater flow andactual crude oilwater flow were behaving differently. The similaritiesin terms of flow pattern between both pairs could be classified aswater-dominated, oil-dominated, intermittent, and stratified regions.However, the differences were largely contributed by the transitioncharacter between crude oilwater flow and mineral oilwater flow.There was a delay in transition and switch of transition region in crudeoilwater flow, both of these phenomena were greatly influenced bythe presence of surfactants or some molecules in the crude oil, such asasphaltenes, resin, etc. (Gafanova and Yarranton, 2001).

The design of subsea pipelines carrying waxy crude oil, espe-cially in Offshore Malaysia, is largely controlled by flow patterns andpressure drop requirements. The correct measurement of waxycrude fluid behaviour in pipelines is accordingly important forproper design of pumping capacity and pipeline dimensions. In thispaper, the behaviour of waxy offshore Terengganu crude oil wasstudied above its WAT in terms of pressure drop, water holdup andflow patterns for nine flow rates in a 5.08 cm ID horizontal flowloop at a length to diameter ratio of 1311.

3. Experiments

3.1. Experimental set-up

The experimental set-up where all the tests and measurementsconducted for the Malaysian waxy crude oilwater flow in pipeswas located at the Malaysia Petroleum Resources CorporationInstitute for Oil and Gas (UTM-MPRC for Oil and Gas), UniversitiTeknologi Malaysia (UTM). Fig. 2 shows the complete experimen-tal set-up.

The experimental investigations began with characterization of thetest fluids as shown in Table 1. The test fluids used throughout theexperimental works were waxy crude oil from offshore Terengganu ofMalaysia and synthetic formation water. The crude oil used for theresearch works was considered as mild waxy crude oil since it has lowWAT and pour point of 26 1C and 18 1C, respectively. The pour pointwas measured by ASTM D97/93 standard test method. WAT and waxcontent (%) were determined using the Differential Scanning Calori-metry (DSC) (Visintin et al., 2005). The DSC measurements of originalwax were performed with Perkin- Elmer Pyris 7 DSC thermal analyser

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420 411

system, which was calibrated with an indium standard before use. Thecrude components were provided from Gas ChromatographyMassSpectrometry (GCMS) analysis (Chen et al., 2006).

The facility comprised oil and water tanks of capacity of 209 m3

each. The flow of oil and water from their respective tanks to thetest section were driven by two three-phase induction motors of1.49 kW that could generate 2900 rpm with maximum flowrate of16.2 cm3/s. Meanwhile, a positive displacement mixture pumpwas installed in the closed-loop system to circulate the mixturefluids in the pipeline. The test loop, as depicted in Fig. 3, comprisedtwo 33.5 m long straight stainless steel pipes, connected by a 1.5 mlong U-shape stainless steel pipe to reduce disturbances on theflow pattern due to the sharp turn.

The pipe has a 5.08 cm internal diameter (ID) and has beeninstrumented to permit continuous monitoring of temperatureand pressure drop. Thermocouples were installed along the testsection to continuously monitor the thermal condition inside thepipeline. The outer pipeline up to the settlement tank of testsection was wrapped with fibre cloth to mitigate heat loss. Theclosed-loop was also connected to the Y mixing section at theentrance, as depicted in Fig. 4.

3.2. Flow pattern observation

In the experiments of oil/water flow patterns, investigators havealmost exclusively used visual observation and photography relatedtechniques (Russell et al., 1959; Charles and Lilleleht, 1966;Arirachakaran et al., 1989). The most common way to identify thedifferent flow patterns is to observe the flow in a transparentchannel or through a transparent window on the pipe wall. As anextension to visual observation, photographic or video techniqueshave also been widely used. For very rapid phenomena, high-speedphotography or video is necessary. In this experiment, flow patternswere identified by observing the flow behaviour through theobservation window (i.e., the transparent acrylic pipe) using TheSony cyber-shot DSC-H9 high definition video camera. The camerahas the capability of producing 100 shots at 2.2 frames per secondsat 19201080 pixel. The camera was used together with the videoat HDTV (1080i) video output to capture the flow of the waxy crudeoil and water through the 3.4 m section of observation window.

Since waxy crude oil has a tendency of opaqueness, a fluoresceinpowder was introduced into the water phase to enhance theobservation of flow behaviour between phases. The mixture offluorescein powder with water produced a luminous colour (i.e.,for free water identification) and it was transmitted with ultraviolet(UV) light. In this experiment, 0.1 g of fluorescein powder per1500 ml of water was used and observations were made by UV light.

3.3. Pressure drop measurement

Pressure drop was measured differentially using two pressuretransducers of type Wika Model A-10. The model A-10 includes thefollowing features; output signal420mA, power supply1430 V;pressure ranges0103.4 kPa and accuracy r71% of span. Twopressure transmitters were located over a length of 3 m after theobservation window and 12m from the pump. The National

Fig. 2. Schematic diagram of the oilwater test facility.

Table 1General properties of the test fluids.

Properties Crude oil Water

API gravity (1API) 41.4 10Density (kg/m3) 818 1000Dynamic viscosity @ 30 1C (mPa s) 1.75 1.0Flash point (1C) o19 Pour point (1C) 18 WAT (1C) 26 Wax content (wt%) 16.15 Asphaltenes (wt%) 0.06

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420412

Instrument data acquisition system equipped with Labview Softwarewas used to record the pressure drops.

3.4. Water holdup measurement

Holdups were measured using a 5000 ml graduated beaker.The test fluids flowing in the test section were trapped in theremovable spool using two quick closing valves (also calledsolenoid valves). The quick closing valves were specially designedto close and open instantly using an electronic switch. The trappedliquids were drained into a container and were left to settle for aday prior to taking the measurements.

3.5. Experimental procedures

A total of 72 runs were accomplished in the experimentalworks of waxy crude oilwater flow using a closed-loop pipelineat horizontal condition. The oil percentages were varied from 10 to

90%, while flow rates were ranged from 2.03 to 16.21 m3/s. Theexperimental work was carried out as follows:

(1) Oil and water were stored in their 209 m3 respective storagetanks. Water mixed with fluorescein powder was pumped intothe system until it completely filled up the closed-loop section.

(2) Once the closed-loop section was fully occupied by water, oilwas then injected gradually into the system according to thestipulated percentage (i.e., 10 to 90% of oil) and the displacedwater was then collected at the discharge line of the testsection. The amount of oil injected must be equivalent withthe amount of water collected at the discharge point.

(3) Both the oil and water were kept on circulated in the closed-loop system at the predetermined flow rate (i.e., 2.03 m3/s) for15 min until a stabilized flow was achieved.

(4) The flow patterns were recorded via the observation windowusing a high definition video camera and simultaneously thepressure drops were recorded using pressure transducersduring the 15 min circulations. Lastly, the mixing pump was

Fig. 3. Schematic diagram of multiphase flow loop test section.

Fig. 4. Y mixing section

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420 413

switched off after the solenoid valves were closed to trap themixture fluids in the removing spool for water holdupmeasurement.

(5) Steps (1) to (4) were repeated for different flow rates (i.e., 2.03to 16.21 m3/s) and oil water ratios (i.e., 10 to 90%).

3.6. Crude oil characteristic

Contextually crude oil is known as a cocktail since varioussubstances are existed in the liquid. The behaviour of the crude oilitself is considered relatively complex due to the presence ofparaffin or asphaltenes and other elements in the oil which maygreatly influence its characteristic. Table 2 shows the compositionof the waxy crude oil used in this research work.

According to their research work on oilwater flow in hori-zontal conditions, Wang and Gong (2010) discussed the differ-ences between mineral oilwater flow and crude oilwater flow ina two phase flow system. They found that there were greatalterations in terms of two phase flow characteristics when usingmineral oil in comparison to crude oil due to the crude oilcompositions. Since natural surfactants, such as asphaltenes andresin, are present in crude oil, the oil and water phases have astrong propensity to form water-in-oil emulsion and this kind ofemulsion is usually found to be quite stable, such that it con-tributes to a complex oilwater two phase flow in a pipelinesystem. Nevertheless, Wang and Gong (2010) did not detail out thecrude oils composition which depending on its composition,might pose a substantial effect on the crudes behaviour. Thedetails of crude oil composition are imperative since they influ-ence the oils characteristics. According to Table 2, there are fewcomplex elements contained in the crude oil which can be foundat number of components of 18, 24, 26, 27, and 28.

The waxy crude oil used in this research work was found tobehave as a Newtonian fluid at 30 1C as shown in viscosity graph inFig. 5. From the graph of shear stress versus shear rate curve, itshows that the crude oil exhibits a linear curve. This showed the

crude oil was found to obey the Newtonian fluid characteristic.From the graph, an empirical equation was established to deter-mine the viscosity of the specific crude oil at 30 1C from the shearstress and shear rate relationship as follows:

0:0172 _ 1

The viscosity of oil (mo can be calculated as follows:

mo _

2

where and _ are stand for shear stress and shear rate,respectively.

It should be noted that this study did not investigate the effectof each chemical component of the crude oil towards theirbehaviour. We confined ourselves to the investigation of the flowpattern, pressure drop, and water holdup of the waxy crude oilwater flow in a horizontal pipe.

Table 2Crude oil composition detected using GCMS equipment.

Number of components Components Carbon, C Percentage (%) by weight

1 Nonane C9H20 2.402 Decane C10H22 3.943 Undecane CH3(CH2)9CH3 1.644 Hexacosane CH3(CH2)24CH3 3.315 Pentacosane CH3(CH2)23CH3 3.096 Tetradecane CH3(CH2)12CH3 3.687 Pentadecane CH3(CH2)13CH3 4.208 Hexadecane CH3(CH2)14CH3 5.079 Heptadecane CH3(CH2)15CH3 4.38

10 Tetracosane CH3(CH2)22CH3 4.7111 Octacosane CH3(CH2)26CH3 3.8912 Nonadecane CH3(CH2)17CH3 4.3313 9,10-Dihydrophenanthracene C14H12 4.0014 Icosane C20H42 4.5615 Heneicosane CH3(CH2)19CH3 4.6016 Benzeneacetic acid C8H8O2 3.9617 Triacontane CH3(CH2)28CH3 4.0818 Allylpentaspiro[3.0.3.0.3.0.3.0.3.1] henicosan-21 C21H30O 3.3819 Tricosane CH3(CH2)21CH3 4.6420 Tetracosane H(CH2)24H 3.9421 Docosane CH3(CH2)20CH3 2.7222 Pentacosane CH3(CH2)23CH3 4.2823 20 ,40-Dimethyloxanilic acid C8H16O2 3.3224 2-Methyl-3-phenyl-1H-indole C15H13N 3.1925 Dodecahydropyrido[1,2-b]isoquinolin-6-one C13H21NO 5.9726 1,3-Dimethyl-4-azaphenanthrene C15H13N 0.5527 2-(Acetoxymethyl)-3-(methoxycarbonyl)biphenylene C17H14O4 0.9528 3,3-Diisopropoxy-1,1,1,5,5,5-hexamethyltrisiloxane C12H32O4Si3 1.23

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 20 40 60 80 100 120 140 160 180 200 220

Shea

r str

ess

(D/c

m2 )

Shear rate (1/sec)

Fig. 5. Dynamic viscosity of the crude oil at 30 1C.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420414

4. Results and discussion

4.1. Flow pattern analysis

Flow pattern identification comprised the static image capturingand video recording throughout the respective mixture velocitiesand water cuts. Since the crude oil was black once dispersed, itaffected visual discrimination. Thus, to enhance the visual observa-tion of free water and images, a predetermined quantity offluorescein powder (i.e., 0.1 g per 1500 ml of water) was added to

the water. After the water was mixed with fluorescein powder, itbecame luminous once introduced with the UV light. This coulddistinguish the flow patterns better when oil and water flowedtogether in the pipeline.

Various types of flow pattern have been observed in the case ofoilwater flow in horizontal pipes in the literature (Arirachakaran etal., 1989; Trallero et al., 1997; Atmaca et al., 2009; Vuong et al., 2009).The basic flow patterns observed were stratified flow, large slugelongated or spherical, dispersed flow, annular flow and combinationof these four basic flow patterns (Brauner, 2002). Many of the

Table 3Photographic examples of observed flow pattern.

Types of flow pattern Examples of observed flow pattern

Stratified wavy flow (STW)

Stratified wavy with semi dispersed flow at interface and oil film (STSD&O)

Semi dispersed flow with semi emulsion at interface and thin oil film(SDSE&TO)

Dispersion of water in oil and emulsion (DWE)

Dispersion of oil in water with water continuous (DO)

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420 415

prominent researchers like Trallero et al. (1997), Flores et al. (1999),Lovick and Angeli (2004), Atmaca et al. (2009) and Angeli and Hewit(1998, 1999) have discussed the flow patterns by referring to mineraloil or refined oil as the test fluids in an oilwater system. However,there was limited discussion about the waxy crude oilwater twophase flow patterns in horizontal pipes. It should be noted thatdifferent types of fluid behave differently in a liquidliquid systemand consequently, this issue should be given a serious consideration. Inthis study, when a waxy crude oil was introduced into the oilwatertwo phase flow system, the basic flow patterns have been expectedlyobserved. This research has revealed a new flow pattern during thecontinuous flow; that may be due to the complexity of the Malaysianwaxy crude oil. This phenomenonwas known as oil film and emulsionand such a flow pattern was never discussed in open literature. Eventhough Vuong et al. (2009) and Sridhar et al. (2011) have discussedabout the presence of oil film but, the oil film discovered by themwasjust a very thin layer without the presence of emulsion. Different oilproperties, water ratios, flow rates and the complexity of waxyMalaysian crude may be involved in the presence of the new flowpattern. The oil film and emulsion was found to have formed duringthe dispersed flow of oil in water.

This research has identified five types of flow pattern formedduring the waxy crude oilwater flow in a horizontal 5.08 cm IDpipe at different oilwater ratios and mixture flow rates. The fiveflow patterns were: (1) stratified wavy flow (STW), (2) stratifiedwavy with semi dispersed flow at interface and oil film (STSD&O),(3) a newly found semi dispersed flow with semi emulsion atinterface and thin oil film (SDSE&TO), (4) dispersion of water in oiland emulsion (DWE) and (5) dispersion of oil in water with watercontinuous (DO). Table 3 shows the flow patterns observed duringthe waxy crude oilwater two phase flow in a horizontal pipe atambient condition (30 1C). It was found that the waxy crude oilused in the experimental work has the tendency to stick on thewall of pipe due to its viscosity and pipe wettability effect. Thiseffect is due to the oil-wet nature of the pipe (i.e., oil is adsorbed tothe pipes inner surface) and might not translated to the waxformation in the flow experiment. As shown in Table 1, the WAT is26 1C and the experiment was conducted at ambient condition (i.e., 30 1C). That is to say the washing pipe with some additivesshould convert the pipes inner surface from oil-wet into water-wet state. As the oil fraction and flow rate increased, the oil filmwas found to have encapsulated the water phase by forming acircular layer around the pipe with an emulsified layer at theinterface of oil and water, as depicted in SDSE&TO.

The STW commonly occurs when two layers of oil and waterare completely separated with the oil flows at the top and waterflows at the bottom of the pipe (Sridhar et al., 2011). This type offlow pattern was clearly seen in this study during the low andintermediate mixture velocities ranging from 2.03 to 8.10 m3/s forCw40.5. While waviness was observed at the interface of crude oiland water layers due to their density and viscosity difference atmoderate velocities. Trallero (1995) explained that the oilwaterflow was fully segregated at low velocities due to the effect ofgravity. This behaviour could be clearly seen when the lighterliquid occupied the upper part of the test section while the heavierdensity liquid is in the lower part of the test section.

The STSD&O was found to have occurred as the flow rateincreased the amplitude of the wave at the interface; thus creatinga semi dispersed flow (i.e., oil droplets in water phase) at theinterface while water and oil phase were flowing in the pipeline.Elseth (2001) also found that as the velocity was increased fromslow to moderate, there was a large number of droplets emergedand the interface became even wavier. As sufficient energy wassupplied, the droplets broke away from the waves. The dropletsbecame bigger and eventually formed a semi dispersed flow at theinterface. Considering the crude oil consisted of natural emulsifiers

as discussed by Wang and Gong (2010), therefore the tendency foremulsion (water in oil) to develop was higher. It could as wellaffect the energy breakage of the wave and droplets at theinterface between oil and water.

The oil film started to appear as the water cut and the mixtureflow rate decreased. The oil film was formed on the inner surface ofthe pipeline due to its viscosity and pipe wettability. The acrylicpipe is an oil wet material thus it influences the oil to adhere to thesurface of the pipe. This behaviour could be observed clearly atmixture velocities ranging from 2.01 to 14.18 m3/s for all water cuts.This phenomenon was more favourable in the water dominatedregion. The Malaysianwaxy crude oil behaves like a gluey liquid dueto its viscosity, wax contents and complex natural chemical com-positions. Therefore, oil tends to stick on the surface of the pipe in acontinuous oil flow and eventually creating an oil film. This phen-omenon happened while the water cut decreased and simulta-neously the flow rate increased. The static images of flow rate at8.0 m3/s, as water cut was increased from 50% to 70% show thepresence of oil film on the inner surface of the acrylic pipe.

Above moderate flow rates (i.e., 10.0 m3/s), a SDSE&TO flow wasencountered because of sufficient energy which has been distrib-uted to the oil and water phases. This phenomenon has createdpartial droplets at the interface and is distributed across the pipe.The mixing rate between oil and water layers was found to haveincreased as the flow rates were increased, which results in creatinga stabilized semi dispersed flow. This phenomenon encouraged theformation of oilwater emulsion. Semi dispersed flow with partialemulsion occurred at water cuts ranging from 10% to 80% with flowrates from 6.80 to 14.18 m3/s. The SDSE&TO was considered a newlyfound flow pattern for oilwater flow in this experiment becausenone of the flow researchers have ever explained or mentionedsuch a flow pattern in their published works. This pattern waslargely contributed by the heavier paraffin components, asphal-tenes, resin, and natural emulsifiers in the Malaysian waxy crude(see Table 2).

As the mixture velocities further increased up to vm412.15, afull dispersed flow was found to have formed due to the breakup ofsemi dispersed flow and increased mixing rate between phases.This phenomenon produced either oil in water dispersion (DO) orwater in oil dispersion (DWE) depending on the dominant phase.Since the flow rate was relatively high, all the droplets weredistributed across the pipe and this has created a full dispersedflow. Such a situation definitely induced the formation of oilwateremulsion. Generally, waxy crude oil contains a large amount ofhigher molecular weight hydrocarbons such as paraffin and asphal-tenes. Therefore, the physical interaction between the crude oil andwater above WAT tends to develop an emulsion of water in oil(DWE). The physical interaction refers to contact between twoimmiscible liquids that mix spontaneously during a liquidliquidflow in which one is present as small droplets or microscopic sizeglobules that have distributed in the other phase. Since the utilizedwaxy crude oil contains some polar molecules (see Table 2) andshows Newtonian characteristics at above WAT thus the interactionbecomes apparent and eventually emulsion has been created. Therewas no chemical reaction supposed to take place because deadcrude and the experimental works were conducted at ambientcondition. But the waxy crude oil starts to crystallize below WATand this phenomenon will become more complex and hard to formemulsions; this was beyond the scope of the study.

At high water cuts, it was expected to have a full dispersed flow ofoil in water with water continuous (DO) since the continuous phasewas dominating the flow at this condition. This behaviour wasobserved at higher water fraction (i.e., 90%) and higher flow rate (i.e., 16.21 m3/s). Trallero (1995), Nadler and Mewes (1997) andArirachakaran et al. (1989) also encountered such a phenomenonin their research work.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420416

The two phase flow experimental work conducted on theMalaysian waxy crude oilwater in a horizontal pipeline hasrevealed a new flow pattern called SDSE&TO. The flow patternwas found at water fraction, Cwo0.8, and flow rates ranging from6.08 to 14.18 m3/s. The cross sections of all the flow patterns aredepicted in Fig. 6. For SDSE&TO flow, there was a third phasepresent; an emulsion between oil and water phases. The emulsionphase grew thicker as the oil fraction and mixture velocitiesincreased. In the stage of intermittent flow, water dispersed inoil as discrete droplets and was distributed over the entire pipescross section. The shape of the dispersed water droplets was fromsmall spherical to irregularly globules. Nevertheless, the size anddistribution depended on the flow rate.

4.1.1. Flow pattern mapA flow pattern map, as shown in Fig. 7, has been successfully

generated using an actual waxy crude oil in this study. Otherresearchers who had experienced the same procedure were Wangand Gong (2010) in their comparative studies. However, theirexperiment was limited to non-waxy crude oilwater flow andmineral oilwater flow in horizontal conditions. They decided touse the mineral oilwater flow as the point of comparisonwith crudeoilwater flow behaviour because mineral oil was widely used in thearea of liquidliquid studies. The use of crude oil is of utmostimportance because it represents the actual behaviour of the oilduring oil production and transportation (Vielma et al., 2007).

The boundary layers generated on the flow pattern map wereto show the phase dominated region. In comparison of thegenerated map shown in Fig. 8 with the boundary layer producedby Wang and Gong (2010), it was found that different fluids usedfor the two phase flow system gave different flow patterns atdifferent conditions. The discrepancy between the generated flowpattern map and Wang and Gong flow pattern map proved thatdifferent fluids behaved differently. The boundaries were mainlyaffected by viscosity of the crude oil as well as the existence ofsurface active (Yarranton et al., 2000) such as asphaltenes, resin,etc. which could decrease significantly the surface energy betweenoil and water phases. This phenomenon would enhance the oildomination region and phase inversion from oil in water to waterin oil emulsion.

The crude oil also has a different hydrodynamic flow when itflows together with water in a horizontal pipe. It is obviouslyshown in Fig. 8 that those parameters have greatly influenced theformation of boundaries for water dominated and oil dominatedregions. The experimental results demonstrated that the oildominated region began at moderate flow rates while intermittentwas found to have formed from as low as 2.0 m3/s flow rate for lowwater cuts. On the other hand, Wang and Gong (2010) found thatthe dispersed flow of water in oil started as early as 2.0 m3/s flowrate and 50% water cuts. The reasons why actual crude experi-ments were limited are due to its hazardous nature, presence ofemulsion and difficulties in flow pattern identification because ofits opaqueness. The challenges faced by the oil and gas industry inproducing and transporting waxy crude oil is another comple-mented factor that deters researchers from using waxy crude oil intheir studies.

4.2. Water holdup analysis

The in-situ volume fraction of water over total mixing liquids ina specific length of a test section is commonly known as waterholdup, Hw. The Hw obtained from the research work is shown inFig. 9. Generally, water holdup cannot be calculated analytically,but can be determined empirically because it is a function ofvariables such as liquid properties, flow pattern, pipe diameter,and pipe inclination. Two quick closing valves were installed ateach end of the transparent section to trap the water holdup. Foraccurate measurement of liquid holdup, type of quick closing valveused has to be chosen carefully. This is the aspect that manyresearchers overlooked or unnoticed during the liquid holdupdetermination. A solenoid valve was found to be the best option.Varseveld and Bone (1997) highlighted that a solenoid valve hasless than 1% error of control accuracy and is highly recommendedfor a quick closing valve system. It is fast and directly actuated asthe switches turned on and therefore could avoid human error ascompared to the use of a typical ball valve which might experiencea slip during opening and closing of both valves at two separatepoints during an oilwater flow in a pipe.

Fig. 6. Cross-sectional view of flow patterns chronological behaviour: (a) STW (b) STSD&O (c) SDSE&TO (d) DWE (e) DO.

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Flow

rate

(m3 /s

)

Water fraction

STW

STSD&O

SDSE&TO

DWE

DO

Fig. 7. Flow pattern map generated for the waxy crude oilwater system at 30 1C.

Fig. 8. The sketch map of flow regions.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420 417

The water holdup, as shown in Fig. 9, was found to haveincreased with water fraction. It was also found that no significanteffect of viscosity on water holdup can be inferred from theexperimental results; these findings are in good agreement withthose found by Vuong et al. (2009), Atmaca et al. (2009) and Sridharet al. (2011). For the stratified flow, the slippage effect was found tobe significant since oil was lighter than water in terms of densitythus it moved faster than water in a horizontal condition. Howeverin a dispersed flow, the slippage was insignificant. Soleimani et al.(2000) found that the holdup was influenced by the input volumefraction at the minimum flow rate as well as the increase in slipratio. This trend, according to Charles et al. (1961) and Martinezet al. (1988), was due to the contact between the liquid and thepipe, and the liquid was likely to be accumulating in the pipe at aslower velocity. When oil and water travelled concurrently in apipe, oil was found to travel faster than water causing a slippagebetween the phases. Because of the slippage, the in-situ watervolume fraction at any particular section in the pipe could not becomputed directly from the input conditions.

The water holdup reduced at a lower flow rate due to the dragcaused by the pipes wall. Generally, the drag force slows down theflow of the water with respect to oil and water flows as a thin layerat the bottom of the pipe at a very slow superficial velocity (Vielmaet al., 2007). The water holdup was found to have increased withthe flow rate and water cut because the water layer became thickerand flowed faster due to its low viscosity. As for transition fromwater dominated to oil dominated regions, the holdup experienceda gradual change due to the presence of emulsion at moderate tohigher flow rate. Whereby at the stratified flow, the water holdupwas found to have increased but upon transition to oil dominatedregions, the water holdup reduced considerably. Table 4 tabulatesthe statistics of Hw values for each flow rate from 10% to 90% waterholdup.

4.3. Pressure drop analysis

Flow patterns obtained from this research were found to bestrongly depended on flow rates and flow pattern, as supported byVuong et al. (2009), Vielma et al. (2007) and Trallero (1995). Asdepicted in Fig. 10, the pressure drop increased as oil fraction andflow rates were increased. Pressure drop started to increasegradually at a flow transition from stratified flow to dispersedflow (intermittent region) specifically from STW to STSD&O withflow rate 46.08 m3/s and water fraction, Cwo0.8. As approachingto the dispersed flow in the oil dominated region known as semidispersed flow with SDSE&TO, there was a marginal increase inpressure drop because the waxy crude oil at this condition hasformed a thin layer on the inner surface of the transparent pipe

and thus increasing the drag force as well as the viscosity due tothe presence of non-stabilized emulsion. The existence of non-stabilized emulsion of water in oil has altered the hydrodynamicbehaviour of the mixture fluids that has eventually increased thepressure drop.

As the water fraction reduced from 80% to 10% and flow rateincreased to 12.15 m3/s, there was an increase in pressure drop. Thisincrease was due to the dispersion of water in oil and loss of watercontinuity, during the formation of dispersion of water in oil and oilcontinuous with emulsion (DWE). At water fraction0.9 and flowrate0.8 m3/s, the pressure drop decreased marginally because thewater phase flowed and wetted continuously the pipes wall whiledominating over a lighter phase (i.e., oil). Water was found to havefully wetted the transparent pipes wall, thus this phenomenon hasreduced the drag force and subsequently mitigated the pressuredrop compared to high fraction of oil flow (Co40.2). A similar casewas observed in the water dominated region. Trallero et al. (1997)explained that at water dominated region (i.e., water in continuousflow), the drag force of the flow caused by the shear stress betweenthe water phase and pipes wall was insignificant, and eventuallyreduced the pressure drop. It was also found that the flow rateswith a thin oil film would cause a higher pressure drop due to theexistence of shear stress near the pipes wall which was contributedby paraffin and other natural complex compounds of the waxycrude oil. The water holdup at this type of flow pattern (with thepresence of wax) was found to have reduced because a thin layer ofwax has formed on the pipes wall and subsequently reduced theeffective flow area of the water.

Apart from flow rate and flow pattern which were found to bethe main factors affecting the pressure drop, Vielma et al. (2007)highlighted that flow regime was another factor which influencedthe pressure drop. They found that in the turbulent flow, the mixingfluids shifted from separated phases to dispersed flow that hasresulted in a greater pressure drop. The research finding was foundto be in good agreement with Vielma et al. (2007). It increased as

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0

Wat

er h

oldu

p, H

w

Flow rate (m3/s)

10% oil20% oil30% oil40% oil50% oil60% oil70% Oil80% oil90% oil

Fig. 9. Water holdup versus flow rate at various oil fractions.

Table 4Statistics of Hw values for different flow rate and oilwater fractions.

Flow rate (m3/s) Min Max Average Std. dev

2.03 0.22 0.76 0.49 0.184.05 0.26 0.85 0.55 0.186.08 0.24 0.89 0.58 0.208.10 0.19 0.91 0.57 0.23

10.13 0.17 0.91 0.54 0.2412.15 0.16 0.92 0.51 0.2514.18 0.15 0.92 0.48 0.2516.21 0.14 0.95 0.46 0.25

0.00

2.00

4.00

6.00

8.00

10.00

12.00

0.0 5.0 10.0 15.0 20.0

Pre

ssur

e dr

op (k

Pa)

Flow rate (m3/s)

90% water

80% water

70% water

60% water

50% water

40% water

30% water

20% water

10% water

Fig. 10. Pressure drop versus flow rate at various water fractions.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420418

the flow changed from stratified to dispersed flow. Meanwhile, thepressure drop changes were found to be closely related to phaseinversion or flow pattern transitions (Arirachakaran et al., 1989;Trallero et al., 1997; Flores et al., 1999; Poesio et al., 2008; Vuonget al., 2009). Wang and Gong (2010) and Ngan (2010) also haveencountered the same phenomenon. Trallero et al. (1997) whohighlighted that insignificant pressure drop was experimentedduring the transition from stratified to a dispersed region. Thiswas due to the loss of oil continuity where the water was found tohave fully wetted the pipe wall and thus reducing the drag of theflow that caused by the shear stress. On the other hand, whenwaxycrude oil (emulsion) flowed together with water, the crude oilformed a thin layer of oil film on the pipes wall, thus increasing theshear stress and substantially caused an increase in pressure drop.In the semi dispersed flow with semi emulsion at interface and thinoil film (SDSE&TO), when the oil became more dominant, itencapsulated the water phase and subsequently formed a thin layerof oil film on the pipes wall. This phenomenon would increase thepressure drop due to the shear stress between the oil and pipeswall. Strazza et al., 2011, Brauner (1991), Bannwart (1998), Vuonget al. (2009) and Sridhar et al. (2011) also have explained thor-oughly this scenario. Table 5 tabulates the statistics (maximum,minimum, average and standard deviation) of pressure drop valuesfor each flow rate (2.03 to 16.21 m3/s) with different oilwaterconcentration (i.e., 10% to 90%).

5. Conclusions

The reported experiment is the first stage study to evaluate thebehaviour of crude oil in the subsea flow lines (horizontal pipes)carrying waxy fluid from the offshore Terengganu to onshorefacilities above WAT. A set of experiments with light Malaysianwaxy crude and synthetic formation water have been conductedand the following conclusions can be drawn:

(1) A new flow pattern map (i.e., SDSE&TO) was established forthe Malaysian waxy crude oilwater flow in a horizontal pipeat ambient condition (30 1C). In general, five flow patternswere observed during the experimental works, namely STW,STSD&O, SDSE&TO, DWE and DO.

(2) The water holdup was found to be highest (95.1%) at disper-sion of oil in water with water continuous (DO) and lowest(1.90%) at water dispersed in oil and oil continuous withemulsion (DWE). A higher pressure drop was obtained at thehighest flow rates (i.e., 16.21 m3/s) and oil fraction, Co0.9where P11.58 kPa, while the lowest pressure drop wasrecorded at the lowest flow rates (i.e., 2.03 m3/s) and athighest water fraction, Cw0.9 where P1.31 kPa.

As the present study represents a first attempt to study liquidliquid flow pattern in a light Malaysian waxy crude, while theavailable data is limited, wide experimental data for various two-

fluid pairs is still required in order to further validate and extendthe proposed models. The experimental results could be used todevelop a theoretical model as there was limited published workson waxy crude oilwater flow in open literature. On top of that, theflow behaviour of a liquidliquid system which uses waxy crude isstill obscure and incomprehensible due to the substantial effect ofemulsion and wax deposition. Despite that, prediction of the flowbehaviour, such as flow pattern, pressure drop, and water holdup, iscrucial in economic consideration because an accurate predictioncan lead to production optimization for transport of these fluids. Inthe current experimental study, no significant effect of viscosity onwater holdup was inferred. In order to obtain an in-situ phasedistribution and consequently provide more information on thephase holdups, instrumentation is defiantly required to obtain anin-situ phase distribution of waxy crude oilwater flow system. Inthe future, we expect even more relevant studies, such as investiga-tion of waxy crude oil behaviour at or below WAT, to evaluate thegel up condition and restart pressures needed after long-term flowline shutdown. Accordingly, accurate models will be possible to bedeveloped based on the actual crude oil representation to addresscurrent and future challenges, and also to support the understand-ing of a more complex case of gas/oil/water system.

Acknowledgments

The authors would like to express their sincere gratitude to TheMinistry of Higher Education (MoHE) of Malaysia and UniversitiTeknologi Malaysia (UTM) for funding this research project via theFundamental Research Grant Scheme (FRGS) Vote 4F136 andResearch University Grant (RUG) Vote 01H68, respectively.

References

Alkaya, B., 2000. Oilwater Flow Patterns and Pressure Gradient in Slightly InclinedPipes (Master Thesis). University of Tulsa, Oklahoma.

Angeli, P., Hewit, G.F., 1998. Pressure gradient in horizontal liquidliquid flows. Int.J. Multiphase Flow 24, 11831203.

Angeli, P., Hewit, G.F., 1999. Flow structure in horizontal oilwater flow. Int.J. Multiphase Flow 26, 11171140.

Arirachakaran, S., Oglesby, K.D., Malinowsky, M.S., Shoham, O., Brill, J.P., 1989. Ananalysis of oil/water flow phenomena in horizontal pipes. In: SPE-18836Presented at the SPE Production Operation Symposium, Oklahoma, U.S.A.

Atmaca, S., Sarica, C., Zhang, H.Q., Al-Sarkhi, A.S., 2009. Characterization of oil/waterflows in inclined pipes. In: SPE-115485-PA-P Presented at the SPE AnnualTechnical Conference and Exhibition, Denver, U.S.A.

Bannwart, A.C., 1998. Wavespeed and volumetric fraction in core annular flow. Int.J. Multiphase Flow 24, 961974.

Beggs, H.D., Brill, J.P., 1973. A study of two-phase flow in inclined pipes. SPE 4007.J. Pet. Technol. 255, 607617.

Brauner, N., 1991. Two phase liquidliquid annular flow. Int. J. Multiphase Flow 17,5976.

Brauner, N., 2002. Modeling and Control of Two Phase Phenomena: LiquidLiquidTwo Phase Flow System. CISM Center. School of Engineering, Tel-Aviv Uni-versity, Udine, Italy http://www.eng.tau.ac.il/brauner/publication.html.

Brennen, C.E., 2005. Fundamentals of Multiphase Flow. Cambridge University Pressp. 2005.

Charles, M.E., Govier, G.W., Hodgson, G.W., 1961. The horizontal pipeline flow ofequal density oilwater mixture. Can. J. Chem. Eng. 39, 2729.

Charles, M.E., Lilleleht, L.V., 1966. Correlation of pressure gradients for the stratifiedlaminar-turbulent pipeline flow of two immiscible liquids. Can. J. Chem. Eng.44, 4749.

Chen, S., Oye, G., Sjoblom, J., 2006. Characterization and rheological properties ofwaxy oils. Annu. Trans. Nordic Rheol. Soc., 14.

Cornish, R.E., 1976. The vertical multiphase flow of oil and gas at high rates. SPE5791. J. Pet. Technol. 28, 825826.

Elseth, G., 2001. An Experimental Study of OilWater Flow in Horizontal Pipes (Ph.D. Thesis). Norwegian University of Science and Technology.

Flores, J.G., Chen, X.T., Sarica, C., Brill, J.P., 1999. Characterization of oil/water flowpattern in vertical and deviated wells. SPE-56108. J. SPE Prod. Facil. 14 (2),102109.

Gafanova, O.V., Yarranton, H.W., 2001. The stabilization of water-in-hydrocarbonemulsions by asphaltenes and resins. J. Colloid Interface Sci. 241, 469478.

Table 5pressure drop statistics for different flow rate and oilwater fractions.

Flow rate (m3/s) Min Max Average Std. dev

2.03 0.44 1.36 0.85 0.274.05 0.64 1.59 1.07 0.296.08 0.76 1.86 1.25 0.348.10 0.87 2.02 1.42 0.38

10.13 1.01 2.16 1.56 0.3912.15 1.17 2.39 1.73 0.4014.18 1.33 2.87 2.05 0.5016.21 1.56 3.86 2.71 0.71

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420 419

http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref1http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref1http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref2http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref2http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref3http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref3http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref4http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref4http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref5http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref5http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref6http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref6http://www.eng.tau.ac.il/~brauner/publication.htmlhttp://www.eng.tau.ac.il/~brauner/publication.htmlhttp://refhub.elsevier.com/S0920-4105(15)00047-9/sbref8http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref8http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref9http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref9http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref555555http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref555555http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref555555http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref10http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref10http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref11http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref11http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref12http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref12http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref13http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref13http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref13http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref14http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref14

Grassi, B., Strazza, D., Poesio, P., 2008. Experimental validation of theoretical modelsin two-phase high-viscosity ratio liquidliquid flows in horizontal and slightlyinclined pipes. Int. J. Multiphase Flow 34, 950965.

Gregory, G.A., Aziz, K., 1975. Design of pipelines for multiphase (gas-condensate)flow. JCPT 750302 J. Can. Pet. Technol. 14, 2833.

Guzhov, A.I., Grishin, A.D., Medredev, V.F., Medreva, O.P., 1973. Emulsion formationduring the flow of two immiscible liquids. Int. J. Multiphase Flow 8, 5861.

Guzhov, A.I., Medredev, O.P., 1971. Pressure losses in flow of two mutuallyimmiscible liquids. Int. Chem. Eng. 11, 104106.

Hagedorn, A.R., Brown, K.E., 1964. Experimental study of pressure gradientsoccurring during continuous two-phase flow in small-diameter vertical con-duits. SPE 940 J. Pet. Technol. 17, 475484.

Hughmark, G.A., 1971. Drop breakup in turbulent pipe flow. J. AIChE 17, 10001009.Keskin, C., Zhang, H.Q., Sarica, C., 2007. Identification and classification of new three

gas/oil/water flows patterns. In: SPE 110221 Presented at the SPE AnnualTechnical Conference, California, U.S.A.

Lovick, J., Angeli, P., 2004. Experimental studies on the dual continuous flowpattern in oilwater flows. Int. J. Multiphase Flow 30, 139157.

Malinowsky, M.S., 1975. An experimental study on oil-water and air-oil-waterflowing mixtures in horizontal pipes. (Master Thesis). University of Tulsa,Oklahoma, U.S.A.

Mansoori, G.A., 1993. Paraffin/wax and Waxy Crude Oil: Principles ofNanotechnology-Molecular Based Study of Condensed Matter in Small Systems.World Scientific, pp. 14.

Martinez, A.E., Intevep, S.A., Arirachakaran, S., Shoham, O., Brill, J.P., 1988. Predictionof dispersion viscosity of oil/water mixture flow in horizontal pipes. In: SPE18221 Presented at the SPE Annual Technical Conference, Houston, U.S.A.

McKibben, M.J., Gillies, R.G., Shook, C.A., 2000. A laboratory investigation ofhorizontal well heavy oilwater flows. Can. J. Chem. Eng. 78, 743751.

Nadler, M., Mewes, D., 1997. Flow induced emulsification in the flow of twoimmiscible liquids in horizontal pipes. Int. J. Multiphase Flow 23, 5568.

Ngan, K.H., 2010. Phase Inversion in Dispersed LiquidLiquid Pipe Flow (Ph.D.Thesis). University College London, London.

Oddie, G., Shi, H., Durlofsky, L.J., Aziz, K., Pfeffer, B., Holmes, J.A., 2003. Experimentalstudy of two and three phase flows in large diameter inclined pipes. Int.J. Multiphase Flow 29, 527558.

Oglesby, K.D., 1979. An Experimental Study on the Effect of Oil Viscosity, MixtureVelocity and Water Fraction on Horizontal OilWater Flow (Master Thesis).University of Tulsa, Oklahoma.

Poesio, P., Strazza, D., Sotgia, G., 2008. Very-viscous-oil/water/air flow throughhorizontal pipes: pressure drop measurements and prediction. J. Chem. Eng.Sci. 64, 11361142.

Rodriguez, O.M.H., Oliemans, R.V.A., 2005. Experimental study on oilwater flow inhorizontal and slightly inclined pipes. Int. J. Multiphase Flow 29, 527558.

Russell, T.W.F., Charles, M.E., 1959. The effect of the less viscous liquid in thelaminar flow of two immiscible liquids. Can. J. Chem. Eng. 37, 1824.

Russell, T.W.F., Hodgson, G.W., Govier, G.W., 1959. Horizontal pipeline flow ofmixture of oil and water. Can. J. Chem. Eng. 37, 917.

Shi, H., Cai, J., Jepson, W.P., 1999, Oil-water distributions in large diameterhorizontal pipelines.Presented in 4thMultiphase Flow and Heat TransferInter-national Symposium, Xian, China. pp. 73-80.

Shi, H., 2001. A study of oil-water flows in large diameter horizontal pipeline. (Ph.D.Dissertation). University of Ohio.

Soleimani, A., Lawrence, C.J., Hewitt, G.F., 2000. Spatial distribution of oil and waterin horizontal pipe flow. In: SPE 66906, SPE Annual Technical Conference andExhibition, Houston, U.S.A.

Sridhar, S., Zhang, H.Q., Sarica, C., Pareyra, E., 2011. Experiments and modelassessment on high-viscosity oil/water inclined pipe flows. In: SPE 146448,SPE Annual Technical Conference and Exhibition, Colorado, U.S.A.

Stapelberg, H.H., Mewes, D., 1994. The flow regimes transitions, pressure loss andslug frequency of liquidliquidgas slug flow in horizontal pipes. Int. J. Multi-phase Flow 20, 285303.

Strazza, D, Grassi, B., Demori, M., Ferrari, V., Poesio, P., 2011. Core-annular flow inhorizontal and slightly inclined pipes: existence, pressure drops, and hold-up.J. Chem. Eng. Sci. 66 (2853-10).

Trallero, J.L., 1995. Oilwater Flow Patterns in Horizontal Pipes (Ph.D. Thesis).University of Tulsa. Oklahoma.

Trallero, J.L., Sarica, C., Brill, J.P., 1997. A study of oil/water flow patterns inhorizontal pipes. SPE 36609J. SPE Prod. Facil., 165172.

Varseveld, R.B., Bone, G.M., 1997. Accurate position control of a pneumatic actuatorusing on/off solenoid valves. J. IEEE/ASME Trans. Mechatron. 2, 195204.

Vielma, M., Atmaca, S., Sarica, C., Zhang, H.Q., 2007. Characterization of oil/waterflows in horizontal pipes. In: SPE 109591, SPE Annual Technical Conference andExhibition, Anaheim, U.S.A.

Visintin, R.F., Lapasin, R., Vignati, E., DAntona, P., Lockhart, T.P., 2005. Rheologicalbehavior and structural interpretation of waxy crude oil gels. Langmuir 21,62406249.

Vuong, D.H., Zhang, H.Q., Li, M., 2009. Experimental study on high viscosity oil/water flow in horizontal and vertical pipes. In: SPE-124542-MS-P, SPE AnnualTechnical Conference and Exhibition held in New Orleans, LA, U.S.A.

Wang, W., Gong, J., 2010. Flow regimes and transition characters of the highviscosity oilwater two phase flow. In: SPE-130850-MS-P, CPS/SPE Interna-tional Oil & Gas Conference and Exhibition held in Beijing, China.

Wicks, M., Dukler, A.E., 1960. Entrainment and pressure drop in concurrent gasliquid flow: I. Air-water in horizontal flow. J. AIChE 6, 463468.

Yarranton, H.W., Hussien, H., Masliyah, J.H., 2000. Water-in-hydrocarbon emulsionstabilized by asphaltenes at low concentration. J.Colloid & Interface Science228, 5263.

Zhang, H.Q., Sarica, C., 2006. Unified modeling of gas/oil/water pipe flow-basicapproaches and preliminary validation. In: SPE 95749-PA-P, SPE AnnualTechnical Conference and Exhibition held in Dallas, U.S.A.

Zhang, H.Q., Vuong, D.H., Sarica, C., 2010. Modeling high viscosity oil/water flows inhorizontal and vertical pipes. In: SPE-135099-MS-P, SPE Annual TechnicalConference and Exhibition held in Florence, Italy.

A.S. Izwan Ismail et al. / Journal of Petroleum Science and Engineering 127 (2015) 409420420

http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref15http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref15http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref15http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref16http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref16http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref17http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref17http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref18http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref18http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref19http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref19http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref19http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref20http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref21http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref21http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref22http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref22http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref22http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref23http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref23http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref24http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref24http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref25http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref25http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref26http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref26http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref26http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref27http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref27http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref27http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref28http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref28http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref28http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref29http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref29http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref30http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref30http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref31http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref31http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref32http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref32http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref32http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref33http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref33http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref33http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref34http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref34http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref35http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref35http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref36http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref36http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref37http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref37http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref37http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref38http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref38http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref45454458http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref45454458http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref45454458http://refhub.elsevier.com/S0920-4105(15)00047-9/sbref45454458Experimental investigation of oilwater two-phase flow in horizontal pipes: Pressure losses, liquid holdup and flow...IntroductionLiterature reviewExperimentsExperimental set-upFlow pattern observationPressure drop measurementWater holdup measurementExperimental proceduresCrude oil characteristicResults and discussionFlow pattern analysisFlow pattern mapWater holdup analysisPressure drop analysisConclusionsAcknowledgmentsReferences