SPE Copyright 1999, Society of Petroleum Engineers Inc.

SPE 664 Effect of Flow Through a Choke Valve on Emulsion Stability M.J. van er Zane, SPE, K.R. van Heuven, J.H. Muntinga, SPE, an W.M.G.T. van en Broek, SPE, Delft University of Technology Copyright 1999, Society of Petroleum Engineers Inc. This paper was prepare for presentation at the 1999 SPE Annual Technical Conference an Exhibition hel in Houston, Texas, 6 October 1999. This paper was selecte for presentation by an SPE Program Committee following review of information containe in an abstract submitte by the author(s). Contents of the paper, as presente, have not been reviewe by the Society of Petroleum Engineers an are subject to correction by the author(s). The material, as presente, oes not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presente at SPE meetings are subject to publication review by Eitorial Committees of the Society of Petroleum Engineers. Electronic reprouction, istribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibite. Permission to reprouce in print is restricte to an abstract of not more than wors; illustrations may not be copie. The abstract must contain conspicuous acknowlegment of where an by whom the paper was presente. Write Librarian, SPE, P.O. Box 886, Richarson, TX 78-86, U.S.A., fax 1-972-92-94. Abstract When oil is prouce uner high water-cut conitions, oil in water emulsions can be forme. The break-up of oil roplets preominantly takes place in the choke valve. We have conucte laboratory experiments to investigate the effect of flow through a choke valve on the oil-roplet-size istribution in the emulsion. In these experiments the choke is moele as a circular orifice in a pipe. The roplet sizes after break-up can be correlate to the mean energy issipation rate per unit mass in the orifice. The experiments have been conucte with two set-ups on a ifferent scale. The relation, which we have erive for the maximum stable roplet iameter ownstream of the orifice can be applie to both scales. Furthermore the effect of oil viscosity on the roplet sizes after break-up has been investigate. Introuction A problem, which is of great concern to the oil inustry nowaays, is the prouction of water simultaneously with oil. In the North Sea, wells proucing at a water cut of 9 % are not uncommon. For these high water cuts the fluis are prouce as oil in water emulsions. In these cases, large efforts are neee for the hanling of the prouction fluis. A serious point of attention is the oil concentration in the waste water after separation. In case the oil concentration is above a certain value (of the orer of tens of ppm), environmental legislation prescribes that isposal of this waste into the sea is not allowe. In orer to reuce the oil concentration in the waste water, the efficiency of the separation process has to be increase. Separators that are commonly use in oil prouction are the plate separator, the hyrocyclone an the centrifuge. The efficiency of these separators is a function of the flow rate, imensions of the separator an the roplet-size istribution of the prouce emulsion that enters the separator 1. For a given separator an flow rate, the separation efficiency is % for sufficiently large roplets. For roplets smaller than a certain critical iameter the efficiency ecreases with ecreasing roplet size. Uner normal prouction conitions this critical iameter varies from approximately µm for a plate separator own to about µm for a centrifuge. Oil-roplet sizes at the bottom of the well can be of the orer of tens of micrometers up to millimeters 2, which is typically larger than the critical roplet iameter of a separator. At the surface facilities, however, cases are known where roplets smaller than µm enter the separator. These small roplets are forme uring turbulent break-up in the choke valve, an stabilize by the various chemical components in crue oil. In this paper laboratory experiments are escribe to investigate the effect of flow through the choke valve on the oil-roplet size. The emulsification of prouction fluis in the choke valve has been previously investigate with a smallscale set-up 4, where the pipe iameter was 4. mm. In the current paper, aitional experiments on a larger scale are escribe (pipe iameter of 1. cm). These experiments have been conucte in orer to investigate the effect of choke size on the break-up process. In this way a relation between choke conitions, oil properties an roplet size is escribe, which is inepenent of the scale. Furthermore the effect of oil viscosity is investigate. Theory The choke valve is a control valve that is place in the wellhea of a prouction system for several reasons. The main purpose of this valve is to control the flow rate. Insie the choke the fluis are force to flow through a small opening. The fluis accelerate an consequently the turbulent intensity increases, which results in a permanent pressure rop. This pressure rop is use to ajust the prouction rate. Another purpose of the valve is to create a low pressure at the surface facilities, which is beneficial in connection with the safety emans. The last reason is that in the case of prouction of

2 M.J. VAN DER ZANDE, K.R. VAN HEUVEN, J.H. MUNTINGA, W.M.G.T. VAN DEN BROEK SPE 664 Figure 1: Schematic representation of flow through a circular orifice in a pipe. gas, the velocity insie the choke can be larger than the velocity of soun in the flui, resulting in critical flow. Consequently, uner these conitions it is prevente that the pressure fluctuations at the surface facilities propagate in an upstream irection, own to the reservoir. In this way formation amage can be omitte. Orifice. In practice many ifferent choke geometries are being use: fixe bean valve, neele an seat valve, multiple orifice valve, an plug an cage valve. All valves have one thing in common, which is that the flui is force to flowthrough a reuce flow area. Although there is an effect of the internal shape of a choke on roplet break-up, we have not looke in this paper at the various shapes, but moele the choke as a circular orifice in a pipe. In Fig. 1 this geometry is sketche. Morisson 6 investigate the flow of air through an orifice meter, which has a similar geometry. From his measurements it can be conclue that most of the energy is issipate in the jet zone ownstream of the orifice (the shae volume in Fig. 1). Base on this conclusion we have erive 4 an expression for the mean energy issipation rate per unit mass in the shae volume, : ρ c () 2/, where ρ c is the ensity of the continuous phase an the roplet iameter. The eforming stress is counteracte by two stabilizing stresses. Firstly, the effect of interfacial tension, σ, is to minimize the interfacial area of the roplet. Basically this means that the interfacial tension opposes eformation an tries to keep the roplet in its original spherical shape. The stress inuce by this effect is proportional to σ/. The secon stabilizing stress is a result of the isperse phase viscosity, µ. Before a roplet breaks up it first has to eform. This results in internal flow insie the roplet, which is hinere by µ. The resulting stress is assume to be proportional to µ Γ, where Γ is the velocity graient insie the isperse phase. Various expressions have been propose for this velocity graient 8-1, all resulting in a ifferent relation for the maximum stable roplet iameter in a turbulent flow. Oil viscosity. As iscusse above, many authors escribe the effect of isperse phase viscosity ifferently. In the invisci limit (µ ), however, most authors 8-1 come up with an ientical relation for the maximum stable roplet iameter, max : pu = o ρ x... (1) max σ ρ c 2.....(2) where p is the pressure rop across the orifice, U o the mean flui velocity in the orifice, ρ the ensity of the flui an x the length of the issipation zone, which Morisson measure to be approximately 2. pipe iameters. For the calculation of the energy issipation rate in our experiments we will use this value of 2.. Break-up. When a ispersion of oil in water flows through an orifice, roplets can break up. For sufficiently large flow rates, the ominant mechanism is turbulent break-up in the jet zone, an not break-up ue to the acceleration in the entrance zone 7. A roplet present in a turbulent flow will be eforme by inertial forces. These forces are inuce by the turbulent velocity fluctuations aroun the roplet in the continuous phase. An alternative way to escribe this it is that these fluctuations inuce a ynamic pressure ifference across the roplet iameter. This eforming stress is proportional to This relation can be erive by assuming that for the maximum stable roplet iameter in the invisci case, the inertial stress is just balance by the restoring stress inuce by the interfacial tension. Hinze 8 argues that Eq. 2 only has to be moifie slightly to incorporate the effect of isperse phase viscosity: ma x 1 + ϕ µ ρ σ σ ρ c 2....() where ϕ is an arbitrary function, which ecreases to zero as its argument nears zero. Note that in the invisci limit Eq. becomes equal to Eq. 2.

SPE 664 EFFECT OF FLOW THROUGH A CHOKE VALVE ON EMULSION STABILITY Davies 9 approaches the problem in a slightly ifferent way. His suggestion is to a the stress inuce by viscosity to the interfacial stress. In this way the relation for max becomes: 4 oil max µ σ + ρ c u 2 4...(4) 2 6 p 7 where u is the magnitue of the turbulent velocity fluctuation aroun the roplet. From turbulence theory it can be erive that u () 1/. Davies, however, argues that ucan be assume to be approximately constant. Uner this assumption, max increases with µ / for a high isperse phase viscosity. Again in the invisci limit Eq. 4 becomes equal to Eq. 2. In his review article, Walstra 1 presents ata on the effect of isperse phase viscosity. For a fixe µ, the ata follows Eq. 2. Furthermore, it can be erive from his ata that for a given energy issipation rate, max is proportional to µ k, with.<k<.9. Arai et al. 11 an Das 12 use the Voigt moel to escribe the turbulent break-up process. In the invisci limit this results in Eq. 2. For the very viscous limit, when the effect of µ is ominant over the effect of σ, they erive that: max µ ρ c 4 1 4..... () It shoul be note that in the approach of Arai et al. an Das, the power of has change from -2/ to -1/4. Calabrese et al. 1 escribe the break-up process in terms of energy. For the maximum stable roplet iameter, the interfacial energy plus the viscous energy of the roplet is equal to the turbulent energy transferre to the roplet by the continuous phase. In the invisci limit this yiels Eq. 2 for max. In the very viscous limit they obtain a relation similar to Eq., only ρ c has to be substitute by ρ c ρ. Calabrese et al. conucte numerous experiments on turbulent break-up in stirre-tank contactors. For very viscous oils ( Pa s<µ <1 Pa s), they obtaine the following empirical relation: max µ µ c 2 ρ c 8 1 4.... (6) It is note that the measure relation between max an is equal to the one preicte by theory (power -1/4). The relation with µ, however, is slightly ifferent, the power is /8 instea of /4. water 1 Figure 2: Schematic representation of the large set-up. An oilwater mixture flows through a restriction (6) an the roplet-size istribution is measure ownstream (7). The etails of the equipment are escribe in the text. Experimental Set-up To investigate the break-up of oil roplets in flow through a restriction we have use two set-ups. The main ifference between these set-ups is the size of the pipes an orifices. Large Set-up. A schematic representation of the large set-up is shown in Fig. 2. An eccentric screw pump (1) pumps tap water out of a large container of 1 m volume. Downstream of the pump the flow rate is measure (2). The pump has a maximum flow rate of 11 l/min, an is able to overcome a pressure of 12 bar. Part of the water is sucke out of the main flow into a sie-track, by means of a gear pump (). The flow rate in the sie-track is measure (4). Oil is pumpe by a piston pump into a neele valve (), where the oil is isperse in the water. The roplet-size istribution epens on the water flow rate through the neele valve: with increasing flow rate, the roplet size ecreases. After the ispersion has been forme, the flow in the sietrack is recombine with the main flow. Here the oil concentration is typically of the orer of.1 vol.%. The pipe iameter in the main circuit is 1.2 mm. To ensure that no roplets break up uring the combination of the two flows, the mean velocity in the main flow is ecrease by increasing the iameter of the pipe by a factor. After recombination, the total flow is gently accelerate by flowing through a slowly converging pipe. Subsequently, the oil-water mixture is irecte through an orifice (6) after flowing through a straight pipe of 7 cm. The orifice iameter can be varie between an 11 mm in steps of 2 mm. The thickness of the orifice plate is mm. The pressure rop across the orifice is measure using 16 pressure taps. In this way the pressure istribution from 2. pipe iameters upstream to 1 pipe iameters ownstream of the orifice can be measure. For the analysis of the ata escribe in this paper we have only use the permanent pressure rop. Downstream of the orifice the 8

4 M.J. VAN DER ZANDE, K.R. VAN HEUVEN, J.H. MUNTINGA, W.M.G.T. VAN DEN BROEK SPE 664 mixture flows through a Malvern particle sizer (7) where the roplet-size istribution is measure by means of laser-light iffraction. At the en of the system a control valve is place (8). The purpose of this valve is to prevent cavitation to take place in the orifice. In the jet zone of the orifice the pressure can ecrease to such a low value that the air, which is issolve in the water uner normal conitions, comes free. By gently closing the control valve, the pressure in the system, an consequently in the orifice, is increase to a value where no cavitation takes place. Finally the mixture is separate in large settling tanks. Small Set-up. The schematic representation of the small set-up is similar to the one shown in Fig. 2. The main ifference is that instea of a pipe iameter of 1.2 mm, a pipe iameter of 4. mm has been use. Furthermore, the maximum flow rate through the system is 1.8 l/min. A etaile escription of this small set-up is given in Refs. 4 an. Disperse Phase Properties. The experiments have been carrie out with various isperse phases. In the large set-up we use Shell Vitrea 9, 46 an 68, which are all three mineral oils. In the small set-up n-heptane an Vitrea 9 an 46 have been use. The properties of these isperse phases are given in Table 1. The interfacial tensions have been measure with the Du- Nouy ring metho. The experiments in the large set-up have been carrie out with tap water as the continuous phase. For most of the experiments in the small set-up this was emineralize water. In the secon column of Table 1, the interfacial tension between the isperse phase an emineralize water is given, in the thir that between the isperse phase an tap water. The ifference in the values of these two columns is not only ue to the ifference in emineralize water an tap water, but also to the fact that the isperse phase has been taken from ifferent batches of oil. The isperse phase viscosity is strongly temperatureepenent. Since the size of the injecte roplets is rather small (they are in the orer of µm) it is assume that the roplets immeiately aapt the temperature of the continuous phase. The temperature of the water in the experiments with the large set-up was 11 C, in the small set-up it was 18 C. The reason for this ifference in temperature is that the small scale experiments have been carrie out in the summer, whereas the large scale experiments were performe uring 9, own (µm) 2 2 1 9, stable 2 4 6 9, inj (µm) Figure : Results of the measurements with Vitrea 46. The experiments have been carrie out with the large set-up. The conitions for this set of experiments are a flow rate of l/min an an orifice iameter of 9 mm. winter. To check the effect of the scale of the set-up one set of experiments has been conucte with the small set-up with Vitrea 46 oil as the isperse phase an tap water at 11 C as the continuous phase. Experimental Proceure. The experiments have been carrie out in the following way. First an orifice size an main flow rate through the orifice are selecte. Then, for a given flow rate through the sietrack, the injecte roplet size istribution is measure. This is the size istribution of the mixture that flows through the Malvern particle sizer when no orifice is present. Subsequently the istribution is measure with the orifice in place. For both istributions the 9 is etermine, which is the roplet iameter below which 9 vol.% of the isperse phase is present. The next step is to vary the flow rate through the sietrack, while leaving the orifice size an main flow rate unaltere. In this way the injecte roplet-size istribution is change. Again the 9,inj of the injecte istribution an the 9,own of the istribution ownstream of the orifice are etermine. For a given set of conitions, all these pairs of iameters result in a graph as shown in Fig.. The proceure as escribe above, is repeate for various orifice sizes an main flow rates. σ, emi water (mn/m) σ, tap water (mn/m) µ at 11 C (mpa s) µ at 18 C (mpa s) ρ (kg/m ) n-heptane 4.7 - - 4. 1-1 684 Vitrea 9 4..4 2. 1 1 1.7 1 1 867 Vitrea 46 4.2 7.8 2. 1 2 1. 1 2 877 Vitrea 68-42.2 4.1 1 2-882 Table 1: Properties of the various isperse phases, which have been use in our experiments.

SPE 664 EFFECT OF FLOW THROUGH A CHOKE VALVE ON EMULSION STABILITY Results an Discussion In Fig. it can be observe that there is an effect of the roplet size upstream of the orifice on the size after break-up. In Ref. a theoretical moel is escribe, which can be use to simulate this effect. In this break-up moel the process is escribe in terms of time scales. The explanation of the observe effect is that roplets o not remain long enough in the turbulent zone of the orifice to break up to their maximum stable roplet iameter. Hypothetically, when a roplet woul re-enter the orifice it woul break up even more. The stable istribution, for which is vali that no roplets break up uring passage through the orifice, can be characterize by 9,stable. In the example given in Fig. the proceure for the estimation of this stable iameter is shown. In this specific case this results for Vitrea 46 in a 9,stable of approximately µm for a flow rate of l/min an an orifice iameter of 9 mm in the large set-up. For all measure conitions the 9,stable can be etermine. In Fig. 4 the measure values of 9,stable at various conitions are plotte versus the mean energy issipation rate per unit mass. In orer to etermine the power of in the expression for the maximum stable roplet iameter, the ata is plotte on a log-log scale. In the small set-up the power of is.41, -.6 an.8 for n-heptane, Vitrea 9 an Vitrea 46, respectively. For the large set-up this power is.41 for Vitrea 9 an 46 an.42 for Vitrea 68. From Fig. 4 it can be conclue that in the measure range of isperse phase viscosities, the power of oes not epen on µ. Base on the relation that Das 11 erive an which he fitte to the ata of Calabrese et al. 1, it is expecte that the power of starts to increase from.4 for low isperse phase viscosities to approximately.2 for a isperse phase viscosity of 2 mpa s (Vitrea 46). In the large set-up we measure that the power of remains approximately.4 up to isperse viscosities as high as 41 mpa s. The ata obtaine with the small set-up show more scatter than the ata obtaine with the large set-up. We believe that this sprea is mainly ue to the limite valiness of the assumptions, which have been mae to calculate the mean energy issipation rate per unit mass. With increasing pipe iameter an flow rate the flow in our experiments behaves more like the flow escribe in Ref. 6, which we use to erive the expression in Eq. 1. We conclue that the ata of our experiments is best escribe with an expression, which incorporates the following relation, inepenently of the isperse phase viscosity: max.4 A consequence of this is that the relations escribe in Eqs., an 6 are not suitable for the escription of turbulent breakup of roplets in our set-up, since these relations result in a power of.2 for high values of the isperse phase viscosity. At first sight, Eq. 4 provies a goo relation for the escription of our ata, in the sense that the power of is.4, inepenently of µ. A closer examination of this expression shows that u in Eq. 4 is not a constant, but proportional to () 1/. When this relation is substitute in Eq. 4, the high viscosity limit becomes equal to the expression in Eq.. In summary, we conclue that non of the relations, which we foun in literature, is able to preict our ata. Base on the iscussion given above, we have ecie to fit our ata with the expression shown in Eq. 2. Since we have not varie the ensity of the continuous phase, an we i not vary the interfacial tension inepenently of the isperse phase viscosity, we assume that the relation between max an σ an ρ c is as erive in Eq. 2. The constant of proportionality in this equation is assume to be a function of the isperse phase viscosity. n-heptane Vitrea 9 Vitrea 46 Vitrea 9 Vitrea 46 Vitrea 68 9, stable (µm) -.8 9, stable (µm) -.42 -.6 -.41 1 1.E+ 1.E+4 1.E+ 1.E+6 (W/kg) -.41 -.41 1 1.E+ 1.E+4 1.E+ 1.E+6 (W/kg) Figure 4: On the left the ata on the stable 9 for the small set-up; on the right the ata for the large set-up. The iameters have been plotte versus the energy issipation rates on a log-log plot. The slopes of the tren lines are given in the graphs.

6 M.J. VAN DER ZANDE, K.R. VAN HEUVEN, J.H. MUNTINGA, W.M.G.T. VAN DEN BROEK SPE 664 9, stable (µm) 2 2 1.8 n-heptane Vitrea 9 Vitrea 46 1.2.88 9, stable (µm) 2 2 1 Vitrea 9 Vitrea 46 Vitrea 68..1 1.7 2 4 6 8 12 14 2 4 6 8 (σ/ρ c ).6 -.4 (µm) (σ/ρ c ).6 -.4 (µm) Figure : On the left the ata on the stable 9 for the small set-up; on the right the ata for the large set-up. The iameters are plotte versus the expression given in Eq. 2. The slopes of the tren lines are given in the graphs. In Fig. the ata is plotte as function of the relation in Eq. 2. The ata is fitte with a straight line through the origin. For the small set-up the slope of this line is.88, 1.2 an.8 for n-heptane, Vitrea 9 an Vitrea 46, respectively. For the large set-up these values are 1.7 for Vitrea 9,.1 for Vitrea 46 an. for Vitrea 68. It can be observe that the slope of the tren line increases with increasing isperse phase viscosity in each set-up. Now we have foun six constants of proportionality for Eq. 2, for six ifferent values of the isperse phase viscosity. Before we can etermine the relation between this constant an µ, it is important to etermine whether the ata obtaine with the small set-up is in agreement with the ata of the large one. In Fig. 6 three series of ata on the stable 9 of Vitrea 46 have been plotte. Two series are measure with the small setup. In one case e-mineralize water at a temperature of 18 C has been use as the continuous phase, in the other tap water at 9, stable (µm) 2 2 1 Small, 18 o C Large, 11 o C Small, 11 o C 1 2 4 6 7 (σ/ρ).6 -.4 (µm) Figure 6: Stable 9 for Vitrea 46 for three ifferent conitions. The experiments have been carrie out with the small an the large set-up at two ifferent temperatures. a temperature of 11 C. The experiments with the large set-up have been carrie out with tap water at 11 C. It can be seen in Fig. 6 that all three ata sets follow a straight line through the origin fairly well. At first sight, for the ata obtaine with the small set-up, there is harly any effect of the temperature of the water on the roplet size. Furthermore, for tap water at 11 C, the small an the large set-up give approximately ientical values for 9. This inicates that the erive expression for the energy issipation rate (Eq. 1) is applicable on the small as well as on the large set-up. When the iniviual sets of ata in Fig. 6 are fitte with a straight line through the origin, the following slopes are foun:.8 for the ata obtaine with the small set-up an water of 18 C,. for that obtaine with the same set-up an water of 11 C, an.1 for the ata obtaine with the large set-up an water of 11 C. Although the ata presumably follows the same tren line, there is a large sprea in slopes for the iniviual series. As iscusse earlier, none of the relations, which we foun in literature is able to escribe our experimental ata. We have fitte our ata to the relation escribe in Eq. 2, an we fin that the slope of the tren line, which fits our ata on 9,stable, increases with increasing isperse phase viscosity. In Table 2 these slopes are given for the various values of µ. In Fig. 7 this ata is plotte. It can be seen that the slope for the Vitrea 46 ata, obtaine with the small set-up an water at 18 C, oes not follow the tren of the other isperse phases. We o not have an explanation for this eviation. The value of the slope is base upon many measurements, an for all measurements we have use the ientical proceure as escribe in the preceing section. In Fig. 7 a fit through the ata has been plotte. For the fit we have assume that the n-heptane ata represents the ieal case of an invisci isperse phase. Furthermore, we assume that the effect of µ can be escribe by a power law. The

SPE 664 EFFECT OF FLOW THROUGH A CHOKE VALVE ON EMULSION STABILITY 7 slope (-) 4 2 1.1.2..4 µ (Pas) Figure 7: The slopes of the tren lines in Fig. as function of the isperse phase viscosity. µ (Pa s) slope 4 1-4.88 1.7 1-2 1.2 2. 1-2 1.7 1. 1-1.8 2. 1-1.1 2. 1-1. 4.1 1-1. Table 2: The slopes of the tren lines in Fig.. result of these assumptions is the following expression for the slope: µ A 1+ B D where A=.88, B=4. 1-2 Pa s an D=.6. We emphasize that there is no physical basis for this expression. More theoretical research is neee to unerstan an escribe the effect of the isperse phase viscosity in our experiments better. Conclusions From our experiments on break-up of oil roplets in flow through an orifice, the following conclusions can be rawn. 1. Droplet sizes ownstream of the orifice are correlate to the mean energy issipation rate per unit mass in the orifice zone. 2. The relation given in Eq. 1 can be use for the calculation of this energy issipation rate. The relation is applicable for orifices in various sizes of pipes.. The expression given in Eq. 2 is a suitable relation to preict the roplet sizes ownstream of an orifice. 4. For a given flow rate an orifice size, the stable roplet iameter increases with increasing isperse phase viscosity.. For high values of the isperse phase viscosity, the relations, which we foun in literature o not escribe our experimental ata. To be able to apply the results of our experiments to real choke valves, it is necessary that a goo expression for the mean energy issipation rate per unit mass is obtaine. For this purpose the pressure rop over the choke valve an the flow rate have to be measure. Furthermore etaile measurements of the flow insie the choke valve, or flow simulations using computational flui ynamics have to be carrie out to estimate the volume of the zone in which most of the energy is issipate. Nomenclature U =velocity, m/s =iameter, m p =pressure, Pa u =turbulent velocity fluctuation, m/s x =istance, m = ifference Γ =velocity graient, s -1 =energy issipation rate per unit mass, W/kg ϕ =arbitrary function µ =viscosity, Pa s ρ = ensity, kg/m σ =interfacial tension, N/m Subscript c = continuous phase =isperse phase own = ownstream inj =injecte max = maximum stable References 1 Van en Broek, W.M.G.T., Plat, R., an Van er Zane, M.J.: Comparison of Plate Separator, Centrifuge an Hyrocyclone, paper SPE 4887, proceeings of the 1998 SPE International Conference an Exhibition in China, Beijing, November 2-6, 1, p. 91-98. 2 Janssen, P.H., an Harris, C.K.: Emulsion Characteristics of High Water-Cut Oil Wells, paper SPE 4977, proceeings Σ, of the 1998 SPE Annual Technical Conference an Exhibition, New Orleans, USA, September 27-, p. 4-414. Sarbar, M.A., an Wingrove, M.D.: Physical an Chemical Characterization of Saui Arabian Crue Oil Emulsions, paper

8 M.J. VAN DER ZANDE, K.R. VAN HEUVEN, J.H. MUNTINGA, W.M.G.T. VAN DEN BROEK SPE 664 SPE 8817, proceeings Π, of the 1997 SPE Annual Technical Conference an Exhibition, San Antonio, USA, October -8, p. 67-68. 4 Van er Zane, M.J., Muntinga, J.H., an Van en Broek, W.M.G.T.: Emulsification of Prouction Fluis in the Choke Valve, paper SPE 4917, proceeings Π, of the 1998 SPE Annual Technical Conference an Exhibition, New Orleans, USA, September 27-, p. 2-2. Van er Zane, M.J., Muntinga, J.H., an Van en Broek, W.M.G.T.: The Effects of Prouction Rate an Choke Size on Emulsion Stability, paper EXPL-6-MZ, to be presente at the r International Seminar in Practices of Oil an Gas Exploitation, INGEPET 99, October 27-29, Lima, Peru. 6 Morrison, G.L., DeOtte, R.E., Nail, G.H., an Panak, D.L.: Mean Velocity an Turbulence Fiels Insie a β=. Orifice Flowmeter, AIChE-Journal, 9, No., 199, p. 74-76. 7 Van er Zane, M.J., an Van en Broek, W.M.G.T.: The Effect of Tubing an Choke Valve on Oil-Droplet Break-up, proceeings of the 1 st North American Conference on Multiphase Technology, Banff, Canaa, June 1-11, 1998, p. 89-. 8 Hinze, J.O.: Funamentals of the Hyroynamic Mechanism of Splitting in Dispersion Processes, AIChE-Journal, 1, No., September 19, p. 289-29. 9 Davies, J.T.: Drop Sizes of Emulsions Relate to Turbulent Energy Dissipation Rates, Chem. Eng. Sci., 4, No., 198, p. 89-842. 1 Walstra, P.: Principles of Emulsion Formation, Chem. Eng. Sci., 48, No. 2, 199, p. -49. 11 Das, P.K.: Preiction of Maximum Stable Diameter of Viscous Drops in a Turbulent Dispersion, Chem. Eng. Technol., 19, 1996, p. 9-42. 12 Arai, K., Konno, M., Matunga, Y., an Saito, S.: Effect of Disperse-Phase Viscosity on the Maximum Stable Drop Size for Breakup in Turbulent Flow, Journal of Chemical Engineering of Japan, 1, No. 4, 1977, p. 2-. 1 Calabrese, R.V., Chang, T.P.K., an Dang, P.T.: Drop Breakup in Turbulent Stirre-Tank Contactors, Part I: Effect of Disperse-Phase Viscosity, AIChE-Journal, 2, No. 4, 1986, p. 67-666.