SPE Emulsion Formation in a Model Choke-valve P.H. Janssen, and C. Noik, SPE, and C. Dalmazzone, SPE, Institut Français du Pétrole

Transcription

1 SPE 7147 Emulsion Formation in a Model Choke-valve P.H. Janssen, and C. Noik, SPE, and C. Dalmazzone, SPE, Institut Français du Pétrole Copyright 001, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 001 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, 0 September October 001. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 00 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 886, Richardson, TX , U.S.A., fax Abstract A problem in mature oil fields is the large amount of water produced alongside the oil. At the moment that both phases are leaving the wellbore of an oil well, tight water-in-oil emulsions can be formed due to the turbulence in the chokevalve at the wellhead. This flow process was simulated in the laboratory, by pumping crude oil-brine mixtures through a model choke-valve represented by a calibrated orifice. The parameters varied were the water-cut, the production flow rate and the orifice dimensions. The water-in-oil emulsion droplets were encapsulated immediately after formation and the emulsion characteristics, i.e. droplet-size distributions and water fraction were determined by optical microscopy and DSC. The results of the experiments revealed that water-in-oil emulsions could be formed in the choke-valve if a certain critical value of the average energy dissipation rate was exceeded. This critical value depends on the type of crude oil and on the oil-water ratio. The diameter of the emulsion droplets found in the experiments could be explained in the framework of inertial and viscous break-up theories. In experiments with a crude oil with little asphaltenes the droplet diameters were in the order of the Kolmogorov length scale and the average diameter D av of the emulsion droplets decreased with increasing energy dissipation rate following a power-law relation: D av n. The value of the exponent n was found to be 0. and 0.1 at a watercut of 0 vol.% and 50 vol.% respectively. In experiments with an asphaltenic crude oil, the final droplet diameter was determined by viscous forces, probably due to relaminarising flow conditions. This paper gives new information with respect to the characteristics of brine-in-crude-oil emulsions being formed in a model choke-valve and can be used to obtain insight in the emulsion formation problematic at the wellhead of an oil well. Additionally, the paper provides data on the size of droplets formed in turbulent and laminar flow at high dispersed-phase fraction. Introduction A choke-valve that is positioned at the wellhead of an oil well is used to control the production flow rate and to reduce the pressure in the tubing to a value which is favourable for further processing of the produced fluids. The choking of the production flow in the choke-valve implies the conversion of a part of the kinetic energy of the flow into heat by viscous dissipation. The flow regime in the choke-valve is generally completely turbulent. The turbulence phenomenon can be regarded as a collection of a large number of rotating fluid structures ( eddies ) of different dimensions. These eddies will inevitably have a mixing effect and a stable emulsion can be formed if two immiscible fluid phases are present. The stability of the formed emulsion is primary determined by the droplet diameter and the chemical components present at the droplet interface, such as asphaltenes and resins which are considered as the natural surfactants of crude oils. The break-up of oil droplets in water during flow through an orifice has been studied extensively by van der Zande 1. In his experiments he has measured the droplet size of oil droplets leaving from an orifice (figure 1) as a function of pressure drop, orifice dimensions, oil viscosity and diameter of the arriving oil droplets. His experiments were performed with a mineral oil-tap water system at very low dispersed phase fraction (<1 vol.% oil), to exclude the influence of surfactants and droplet coalescence on the break-up process respectively. For low oil viscosity, the droplet diameters measured by van der Zande 1 were found to be equal to the values predicted by the droplet break-up theory of Hinze. In our experimental investigation, we have performed similar experiments, pumping a mixture of crude oil and brine through an orifice. The water fraction, which was the dispersed phase, was rather high, up to 50 vol.%. This causes both theoretically as experimentally major difficulties. The formed water-in-oil emulsions were encapsulated instantaneously to avoid coalescence and their droplet-size distributions were determined subsequently by means of optical microscopy. The results of these droplet diameter measurements as function of several relevant parameters have been compared to droplet

2 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 diameter calculations of various droplet break-up theories available in literature. Theory Orifice flow. The flow regimes, which are prevailing for flow in a pipe in which an orifice is mounted (figure 1), were investigated by Lakshmana et al.. In contrast to flow in a straight round tube, which becomes turbulent for tube Reynolds numbers 4 larger than 100, the flow downstream of an orifice becomes turbulent at significantly smaller orifice Reynolds number Re o. Lakshmana et al. found that the critical orifice Reynolds number above which the flow downstream of an orifice becomes turbulent is constant for small values of (=D o /D t ) and has a value between 00 and 40 for a quadrant-edged orifice. This implies that the approaching flow always will be laminar at the conditions at which the flow downstream of the orifice just becomes turbulent, since Re t = Re o. The permanent pressure drop P perm of flow through an orifice is given by: P 1 1 perm U t c (1) where c 0 is the orifice coefficient, the fluid density, U t the average tube velocity and the ratio of orifice and tube diameter. In case of two-phase flow a volume-averaged density should be used: d d c c, where and are the volume fractions and the densities of the dispersed and continuous phase. The orifice coefficient approaches a value of 0.61 for high Reynolds number 4. The permanent pressure drop is caused by the dissipation of kinetic energy due to turbulence. Under the assumption of homogeneous isotropic turbulence the average amount of energy that is being dissipated per time and mass unit can be estimated by: PpermU o U 1 1 t....() 4 L dis co Ldis where U o is the average fluid velocity in the orifice and L dis is the length of the zone in which most of the kinetic energy is dissipated. The dimensionless length of the dissipation zone has been investigated experimentally by Van der Zande 1 as a function of and by Lakshmana et al. as a function of Re t. Their studies result in the following correlation for the length of the dissipation zone, for Re t <00: Ldis 1. 1 f ( ) g( Ret ) ( ) 1550 Ret Dt () In the Reynolds number range between 50 and 1000 the function g(re t ) increases from a value of 4.1 to 5.1. Inertial break-up. The effect of turbulence on droplet breakup has been investigated first by Hinze, who derived expressions for the disturbing and restoring stresses working on a single droplet in a homogeneous isotropic turbulent flow field. Under the condition that the disturbing inertial stress working on a droplet is significantly larger than the disturbing viscous stress, i.e. Re d >>1, and the drop viscosity is relatively low, an expression for the maximum droplet diameter d max can be derived: d max 5 5 We 5 crit c...(4) where is the interfacial tension between the continuous and dispersed phase and We crit is the Weber number of the maximum stable droplet: We crit 4 d c...(5) Experiments of van der Zande 1 have shown that the relationship for the maximum droplet diameter is correct within a few percent if the dispersed phase viscosity d and dispersed phase fraction d both were sufficiently small, and if the residence time of the droplet in the turbulent zone was sufficient. The break-up of a droplet in a turbulent flow field consists of a Seri of subsequent break-up events till the droplet has reached its steady-state size (eq. 4), or We has reached We crit. Turbulence can be considered as a collection of rotating fluid structures 5, called eddies. The largest eddies in which most of the kinetic energy is stored, transfer their kinetic energy to somewhat smaller vortices. This process continues until the kinetic energy finally is converted into heat in the smallest eddies by viscous dissipation. The diameter of the smallest eddies in isotropic turbulence, the Kolmogorov length-scale l K, is characterised by a droplet Reynoldsnumber of 1: l K ( c c ) (6) For a droplet with diameter d to break up in a turbulent flow field it takes a certain time t wait before the droplet encounters an eddy which is able to deform it 6 : t wait d 0 min ( d ) 11 e d c 6 f 11 c ddf bv (7)

3 SPE 7147 EMULSION FORMATION IN A MODEL CHOKE-VALVE where is the eddy size and min is the smallest eddy size that contributes to the inertial stress. bv is the volume break-up fraction which is defined as the ratio of the largest daughter droplet volume after break-up and the volume of the mother droplet. c is the dimensionless increase of the interfacial area due to the break-up event. At the moment that the droplet has met up with a suitable eddy, the break-up process starts which lasts a time t br 7,8 : t br d d ln c ( d)...(8) ( d ) d c where d is the dispersed phase viscosity. If the lifetime of the eddy is longer than the break-up time, finally the droplet will break up. This process continues until either the droplets have reached their steady-state value (eq. 4 or 6) or they have left the turbulent zone. Viscous break-up. If the droplet Reynolds number Re d becomes significantly smaller than one the break-up process will be governed by the viscous stress v c G, where G is the shear rate or strain rate in case of respectively simple shear flow and plane hyperbolic flow. This can occur in laminar flow or in turbulent flow when the droplet sizes become of the order of the smallest eddies (eq. 6). The viscous stress is counteracted by the Laplace pressure /d and the ratio is the capillary number Ca. A droplet is broken up if its capillary number Ca is larger than a certain critical capillary number Ca crit. Experiments performed by Grace 9 have shown that the critical capillary number is a function of viscosity ratio and type of flow (see figure 9). Material and methods Three series of experiments have been performed with our experimental set-up. In all experiment series a mixture of crude oil and brine were pumped through an orifice at varying flow rate (see figure 1). The objective of experiment serie I was to determine under which circumstances a stable emulsion was generated in the orifice. Experiment series II and III were focused on the determination of the droplet-size distribution if an emulsion was produced. In the last two series, small concentrations of chemicals were added to both the crude oil and brine, which reacted with each other forming a thin polymer layer at the oil-water interface 10. The objective of this interfacial polymerisation reaction was to encapsulate the formed water-in-oil emulsions immediately after their formation, in order to avoid coalescence. The experimental set-up. In figure a schematic overview of the experimental set-up used for the orifice-flow experiments is depicted. The orifice was mounted in a vertical-positioned tube with an inner diameter D t of 4.8 mm. The orifice plate had a thickness h of 1.5 mm and the diameter D o of the circular hole in the orifice plate was respectively 0.6, 0.7, 0.8, 0.9 and 1 mm. The oil and water supply tubes had an inner diameter of 4 mm and they were joining each other 1 cm upstream of the orifice enabling some premixing of the phases. The fluid pressure was measured at two locations, respectively 5.8 cm upstream and downstream of the orifice. A maximum volume of 500 ml of each phase could be pumped by two piston pumps at a fixed flow rate into the supply tubes. This implies that the maximum emulsion volume that could be generated during each experiment was one litre. The produced oil-water mixture was collected in a reservoir which was mounted 10 cm above the orifice and could be observed through a glass window before depressurisation. The entire equipment was placed inside a cupboard and could be heated up to a maximum temperature of 100 C. The experiment could be done at a fixed fluid pressure, up to 50 bars. The experiments. Experiment series I and II have been performed with crude oil A, containing little asphaltenes, whereas experiment serie III was performed with an asphaltenic crude oil B (see table ). In experiment serie I, 50 vol.% brine and 50 vol.% crude oil were pumped through an orifice, without applying encapsulation. After each experiment the volume of separated water, separated oil and emulsion was determined. The emulsions formed were always from the water-in-oil type and the percentage of water in the emulsion was measured by means of Differential Scanning Calorimetry (DSC). Experiment serie II was performed with a water percentage of 0 and 50 vol.%. All orifice diameters were used for the highest water-cut (50 vol.%), whereas only the 0.6, 0.8 and 1 mm-orifices were used in case of a water-cut of 0 vol.%. Experiment serie III was done with a water percentage of 0 and 50 vol.% and an orifice diameter of 0.8 mm. In both series (II and III), the formed emulsions were encapsulated instantaneously. The droplet-diameter distributions of the emulsions were determined by optical microscopy. Experiment serie II (with crude A) was performed under ambient conditions (0 C, 1 bar) and no gas was dissolved in the crude oil. In experiment serie III (with crude B) the influence of high temperature (60 and 80 C), elevated pressure (1, 1 and 4 bars) and dissolved methane gas on the formation of emulsions was investigated. The amount of methane gas in the crude oil was 0.11 mol/l (Gas to Oil Ratio:.7 at 0 C and 1 bar). In all experiments the flow rate was varied. The experimental conditions at which all experiment series were executed are summarised in table 1. Fluids. The composition and physical properties of crude oils A and B are given in Tables, and 4. The brine used in experiment series I has been made artificially by dissolving 7 grams of NaCl per litre demineralised filtered water. The brines used in experiment series II (brine A) and III (brine B) have been recovered together with the respective crude oils in the wells. The composition of the different brines is presented in table 5 and their physical properties in tables 6 and 7. Encapsulation. Encapsulation of the emulsion droplets was achieved by adding a concentration of mol/l of triethylene tetramine (TETRA) and terephthaloyl dichloride

4 4 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 (TDC) to the brine A and crude oil A respectively 10. The TDC was first dissolved in xylene. In case of brine B and crude oil B the concentration of both chemicals was mol/l. With these concentrations, satisfactory results were obtained. Tests have been done as well to encapsulate the formed emulsion after its formation by putting a small quantity of crude oil in which TDC was dissolved in the reservoir (see figure ). In that case, the crude oil in the pump did not contain any TDC. The advantage of this method was that the encapsulation process takes place after the formation of the emulsion, without influencing it. Furthermore, the exact value of the interfacial tension can be determined. This will be discussed later. Interfacial Tension. The interfacial tensions between the crude oils A and B and their respective brines have been measured at ambient temperature (0 C) by means of the Wilhelmy-plate method. The value of the interfacial tension decreased as a function of time and reached a steady minimum value after 15 to 45 minutes. The interfacial tension between brine and crude oil including their encapsulation chemicals, which is in fact the most relevant value, can not be measured per definition. Nevertheless, the interfacial tension between brine, containing TETRA, and pure crude, was measured as well as the interfacial tension between pure brine and TDCcontaining crude oil. The concentrations of encapsulation chemicals were similar to the concentrations used in experiment series II and III. The initial values and the steadystate values of the interfacial tensions of the different systems are presented in table 8. Experimental procedure. Before the start of an experiment, both the oil and water phases are mixed with their respective encapsulation chemicals (in case of experiment series II and III). Both pumps are filled up and the cupboard is heated up to the desired temperature. During the heating process the oil and water in the pumps are mixed gently by a propeller to guarantee a homogeneous fluid composition. If the experiment is performed at elevated pressure, the pressure in the pumps before the experiment is already raised to the desired pressure. At the beginning of an experiment, the pump pistons start to replace the fluids and an automatic valve before the orifice opens when the pressure has reached the desired set value. The pressures in the reservoir and in the tube downstream of the automatic valve have been increased up to the desired pressure by filling it with nitrogen gas. During the experiment, the emulsion replaces the nitrogen which leaves the reservoir through a controllable overpressure-valve on top of the reservoir. After each experiment, the produced oil-water mixture is purged into a bottle by opening a tap which is placed at the bottom side of the reservoir. Differential Scanning Calorimetry. Differential Scanning Calorimetry is a laboratory analysis technique which is usually used to measure physico-chemical transformations or phase changes. In the field of emulsions, it is usually used to determine the amount of emulsified water 11. A known quantity of the sample is put into an aluminium crucible and is sealed by a press. The crucible is cooled down to a temperature of - 60 C in the calorimeter by circulation of nitrogen gas. The heat flow of the sample is measured and recorded as a function of temperature. If one of the components crystallises, heat will be produced due to the phase transition which will be visible as a peak in the thermogram. It has been shown recently that DSC can be used as well to assess the polydispersity of w/o emulsions 11. The cooling velocity will be smaller in general than 5 C/min to exclude dynamic effects. After the cooling cycle, the sample can be heated again. When applying a DSCmeasurement on a water-in-oil emulsion the endothermic phase transition during melting of the water droplets will take place at a temperature of 0 C for pure water and smaller than 0 C for saline water. By integrating the fusion-peak in the thermogram, the amount of water present in the sample can be determined. The DSC-apparatus used in our study was a DSC 90 of TA Instruments. The emulsion mass sampled for DSC-measurements was between 70 and 150 mg. Microscopy. The droplet-size distributions of the emulsions analysed in experiment series II and III were determined by optical microscopy. A small sample of each emulsion was taken from the bottle in which the emulsion was purged and put on a microscope glass plate. The drop was covered by a second glass plate, forming a very thin layer of emulsion. The sample was observed under an Olympus PROVIS AX70 microscope. The best contrast between water droplets and oil could be obtained using a lens with a magnification of 400 and 1000 in case of crude oil B and A respectively. The pictures obtained were recorded and all visible water droplets were marked by hand. Average droplet diameter, standard deviation and the number of measured droplets were calculated by a computer program. Of each emulsion, the diameter of about 400 droplets was measured. All bottles in which the emulsions were purged, were identical. The sampling of the emulsion was performed in the middle of the bottle. The settling velocity of the water droplets was small due to the high viscosity of the emulsions. Experimental results Experiment Serie I. In figure the energy necessary to melt the water droplets in one gram of emulsion E melt is plotted as a function of the average energy dissipation rate (eq. ) in the orifice. The results were obtained for a water-cut of 50 vol.% water and for orifice diameters of 0.6, 0.7, 0.8 and 1 mm respectively. The water percentage present in the emulsions was determined by dividing the melting enthalpy of the emulsion by the melting enthalpy of the NaCl-solution used, which was equal to 98 J/g. At low energy dissipation rate a mixture of separated water, separated oil and emulsion was produced. The water percentage in these emulsions, determined by DSC varied between 0 and 8.5 vol.%. If the energy dissipation rate exceeded a threshold-value of about W/kg the produced mixture only contained a w/oemulsion with a water percentage of 50 vol.%. The sudden increase of the emulsion volume to 50 vol.% above a threshold-value for of W/kg was observed for all orifice diameters.

5 SPE 7147 EMULSION FORMATION IN A MODEL CHOKE-VALVE 5 Experiment Series II and III. The permanent pressure drop P perm along the orifice has been measured in series II and III as a function of flow rate Q for different orifices. In figure 4 the values of P perm have been plotted as a function of 4 1 U t 1 11 for experiment series II and III. Both series of data points can be fitted well by a straight line with slope.78 (experiment serie II) and.61 (serie III). This corresponds to values for the orifice coefficient c o of 0.60 and 0.6 respectively. The oil-water mixtures produced during the experiment series II consisted of free water, free oil and w/o-emulsion at low flow rate, and of w/o-emulsion only at high flow rate. In experiments with 0 vol.% water, the transition to an emulsion only took place at a lower average energy dissipation rate compared to the experiments with 50 vol.% water. The emulsions produced under identical conditions but with different encapsulation methods differed significantly. When the encapsulation chemicals were added to the fluids, the resulting emulsions were very homogeneous and no discrete droplets could be observed. When the emulsion was encapsulated after formation by adding the second chemical to the reservoir, the emulsion was less homogeneous and some free water and oil were present in the bottle. For this reason, the first method was taken in our investigation. For all experiments of experiment series II and III the average diameter, measured by microscopy, was determined and plotted as a function of the average turbulent energy dissipation rate in the orifice if only a w/o-emulsion was produced. In figure 5a to 5d the average droplet diameter has been plotted as a function of average energy dissipation rate on double-logarithmic scale for both crude oil A and B and a watercut of 0 and 50 vol.% respectively. In case of crude oil A the average energy dissipation rate necessary to create a homogeneous emulsion was almost one order of magnitude larger than compared to crude B. In all cases the average droplet diameter decreased as a function of average energy dissipation rate. The straight lines in the graph represent the best-fitting power-law curves: D av n. The exponents n for the four cases are given in table 9. The exponent n is for both crude oils at a water-cut of 0 vol.% equal to -0. and has a value of 0.10 at a water-cut of 50 vol.%. The influence of the orifice diameter D o on the relationship between average emulsion-droplet diameter D av and average energy dissipation rate has been investigated for crude oil A at a water-cut of 0 and 50 vol.%. The results are shown in figure 5a and 6 respectively. The D av - -datapoints obtained for different orifice diameters seem to be fitted well by one power-law relationship in case of a water-cut of 0 vol.% (figure 5a). In case of a water-cut of 50 vol.% the scaling between the average droplet diameter and the average energy dissipation rate is less good, especially for the smallest orifice diameter (D o =0.6 mm). The exponents n and the correlation coefficients for the different orifices in case of a water-cut of 50 vol.% are given in Table 10. The impact of the temperature on the average droplet diameter is depicted in figure 5b. The droplets obtained at a temperature of 80 C are slightly larger than at a temperature of 60 C if all other parameters are the same. The influence of the fluid pressure and gas content of the crude oil on the droplet-diameter results did not seem to be significant but additional experiments are requested to confirm this conclusion. Numerical calculations. To point out the unsteady nature of droplet break-up, calculations have been performed for three different flow rates which corresponded to experiments of serie II (D o =0.8 mm, w =0.). Equation 7 and 8 have been used to calculate the waiting time t wait and the break-up time t br as a function of average energy dissipation rate and droplet diameter d. The initial droplet diameter was assumed to be equal to the orifice diameter D o and the energy dissipation rate is assumed to be constant. Furthermore, it is assumed that f bv =0.8 (see van der Zande 1 ) which implies that the largest daughter droplet has a diameter, which is 0.9 times the diameter of the mother droplet. min has been taken equal to the Kolmogorov length l K scale to solve the integral in equation 7. The interfacial tension was 0 mn/m. Numerical results. In figure 7, the theoretical maximum droplet diameter is plotted as a function of time for low, medium and high values of using a double logarithmic scale. Furthermore the datapoints representing the measured droplet diameter D meas as a function of the residence time t res (=L dis /U o ) are plotted as well. Details of the calculations and the corresponding experimental results are given in table 11. The maximum transient droplet diameter D trans,max is the calculated maximum diameter of the droplets that leave the turbulence zone. Discussion In table 1 the tube and orifice Reynolds numbers at which experiment series II and III were performed, are presented. The crude oil viscosity was taken as the continuous phase viscosity..from these values it can be derived that in both series the flow before the orifice was laminar but became turbulent in the zone after the orifice. Beside, the initial droplet Reynolds number Re d,initial is very large, if we assume that the initial drop diameter is in the order of the orifice diameter. This implies that in first instance, the water will be dispersed in the oil by inertial forces, and the water droplet diameters will be governed by inertial break-up. Later on when the water droplets reach their final size the final droplet Reynolds numbers Re d,final has become significantly smaller than one in the experiments with crude B but stay slightly larger than one in the experiments with crude A (see table 1). This means that theoretically the final droplet diameter in experiment series II (crude A) is determined by the turbulence mechanism whereas in experiment series III (crude B) the final droplet diameter could be governed by viscous break-up. In figure 8, the droplet diameters of both series are plotted as a function of, together with the droplet diameter correlation

6 6 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 according to the theory of Hinze (eq.4) and the Kolmogorov length (eq.6). For both experiment series, the maximum droplet diameters given by the inertial break-up theory of Hinze are significantly smaller than the Kolmogorov length scale. This implies that Hinze s formula for d max will not be valid in this case. Furthermore, it can be seen that the droplet diameters obtained in experiment serie II are of the order of the Kolmogorov length whereas in experiment serie III the droplets are significantly smaller than the Kolmogorov length. This observation enforces our supposition that the final droplet diameter in experiment serie III is determined by viscous forces. An explanation could be that the initially turbulent orifice flow quickly becomes laminar downstream of the orifice due to the increasing effective viscosity of the emulsion. The droplets observed in experiment serie II are broken up by turbulence eddies until their diameter has become of the order of the smallest eddies, of which the size is given by the Kolmogorov length. This would also explain the found value for the exponent n at a water-cut of 0 vol.% which is equal to 0.0. Theoretically, if the droplet diameters are of the order of the Kolmogorov length the exponent would be 0.5. The numerical calculations performed for three energy dissipation rates (see figure 7) show that the droplet diameters converge to their respective final diameter which is equal to the Kolmogorov length, before leaving the turbulent zone, which is in agreement with the experimental results. Nevertheless, because of the large number of experimental parameters and simplifications put into the transient model, the results only should be interpreted qualitatively. The impact of the water-cut on the droplet diameter-energy dissipation relationship is that the exponent n increases significantly (see table 9) from a value of 0.0 to This could be explained by the fact that during the emulsification process in the orifice a higher water-cut will lead to a relatively larger suppression of the turbulence. This causes that increasing the energy dissipation rate will have a smaller effect on the decrease of the droplet diameter if the water-cut is higher. It should be realised that the average distance between the droplets at a water-cut of 50 vol.% is very small. The effect of the orifice diameter on the relationship between the droplet diameter and the energy dissipation rate is depicted in figure 5a and 6. For a water-cut of 0 vol.% (figure 5a) the relationship is independent of the orifice diameter which is according to the theory and shows the good scaling. For a water-cut of 50 vol.% the results do not seem to scale perfectly for different orifice diameters. This is probably caused by the high-dispersed phase fraction and the consequent suppression of the turbulence as mentioned before. The supposition that the final break-up of the emulsion droplets in experiment serie III was governed by viscous forces can be verified by determining the values of the capillary number Ca of the final droplets. This has been done for the different flow rates (40, 60 and 80 l/h) and for the two temperatures investigated. For the shear rate, the average shear rate in the orifice, assuming a parabolic velocity distribution has been taken: 8U o Do. The resulting capillary numbers are depicted in figure 9 as a function of viscosity ratio. The capillary numbers are very close to the critical capillary numbers Ca crit for pure shear flow as determined by Grace 9. This result seems to confirm that the final break-up in experiment series III is viscous-dominated. The value of the interfacial tension which has been used in the calculations of inertial and viscous break-up was 0 mn/m. This value seems to be a good choice considering the measured values in table 8. The real value of the interfacial tension could not be measured since the encapsulation reaction takes place very quickly 10. Nevertheless, by comparing an estimation of the polymerisation reaction time with the residence time of the emulsions in our experiments it could be assumed that the encapsulation process takes place after the emulsion has left the break-up zone. In this study the average droplet-diameter has been measured instead of the maximum droplet diameter, which has to be used generally in the droplet break-up theory of Hinze. Nevertheless, the Kolmogorov length scale with which the experimental results of serie II can be explained gives an estimate of the smallest eddy size, as well as an estimate of the smallest droplets that can occur in turbulent flow. Conclusions In this study, the emulsion characteristics of w/o-emulsions have been investigated which were formed in a model chokevalve, using different crude oil-brine systems. The results show that a minimum threshold amount of power per mass unit has to be dissipated to create a homogeneous stable emulsion. The diameter of the emulsion droplets found in the experiments could be explained in the framework of inertial and viscous break-up theories. In experiments with an asphaltenic crude oil (crude oil B), emulsions were formed at a significantly lower energy dissipation rate compared to experiments with a crude oil with little asphaltenes (crude oil A). In the experiments with the asphaltenic crude oil (crude oil B), the final size of the emulsion droplets was determined by viscous break-up. In the experiments with the crude oil with little asphaltenes (crude oil A) the emulsion droplets were formed by turbulent break-up and their final size was of the order of the smallest eddies of the turbulent flow. Although the present experiments have revealed some of the hydrodynamic aspects of the emulsion formation problem, more experiments have to be done to investigate the impact of the physical-chemical properties of the crude oil, temperature, pressure and dissolved gas on emulsion formation in the choke-valve. Acknowledgement The authors wish to thank TotalFinaELF for their interest in the work. They gratefully acknowledge Aurélie Mouret and Laurence Podesta-Foley for performing the experiments. Furthermore, we are especially grateful to Dr. M.J. van der Zande for his support and the useful discussions.

7 SPE 7147 EMULSION FORMATION IN A MODEL CHOKE-VALVE 7 Nomenclature Ca= capillary number c d D= diameter, L, m E melt = melting energy, J/g L= length, L, m P= pressure, M/T L, kg s - m -1 cu d Re d = droplet Reynolds number c Re o = orifice Reynolds number cu o Do c Re t = tube Reynolds number cu t Dt c U= average flow velocity, L/T, m s -1 We= Weber number cu d c ( d) d c o = orifice coefficient, [-] c = dimensionless increase of the interfacial area, [-] d= droplet diameter, L, m ()= function in eq. bv = volume break-up fraction, [-] g(re t )= function in eq. h= thickness orifice plate, L, m t= time, T, s u= mean turbulent velocity fluctuation, L/T, ms -1 = D o /D t, [-] = energy dissipation rate, L /T, m s - = strain rate, 1/T, s -1 = volume fraction [-] = shear rate, 1/T, s -1 = eddy size, L, m = dynamic viscosity, M/LT, Pa.s = density, M/L, kg/m = interfacial tension, M/T, kg s - = deforming stress, m/(lt ), kg/(ms ) Subscripts br= break-up c= continuous crit= critical d= dispersed phase dis= dissipation K= Kolmogorov max= maximum min= minimum o= orifice perm= permanent t= tube trans= transient v= viscous wait= waiting References 1. Zande, M.J. van der, Droplet break-up in turbulent oilin-water flow through a restriction, Ph.D. thesis, Delft University of Technology, Delft, The Netherlands, Hinze, J.O., Fundamentals of the hydrodynamic mechanism of splitting in dispersion processes, AIChE- Journal, vol.1, no., 89-95, Lakshmana Rao, N.S., Sridharan, K and Alvi, S.H., Critical Reynolds number for orifice and nozzle flows in pipes, J. of Hydr. Res., vol. 15, no., Bird, R.B., Stewart, W.E., and Lightfoot, E.D., Transport Phenomena, John Wiley & Sons, Inc., New York, Davies, J.T., Turbulence phenomena, Academic Press, Inc., New York, Luo, H., Svendsen, H.F., Theoretical model of drop and bubble break-up in turbulent dispersions, AIChE- Journal, vol. 4, no. 5, Arai, K., Konno, M., Matunga, Y, Saito, S., Effect of dispersed-phase viscosity on the maximum stable drop size for break-up in turbulent flow, J. of Chem. Eng. Of Japan, vol. 10, no. 4, 5-0, Das, P.K., Prediction of maximum stable diameter of viscous drops in a turbulent dispersion, Chem. Eng. Technol., vol. 19, 9-4, Grace, H.P., Dispersion phenomena in high viscosity immiscible fluid systems and application of static mixers as dispersion devices in such systems., Chem. Eng. Commun., vol. 14, 5-77, Morgan, P.W., Kwolek, S.L., Interfacial polycondensation. II. Fundamentals of polymer formation at liquid interfaces, J. of Polym. Sc., vol. 60, 99-7, Dalmazzone, C., Clausse, D., Microcalorimetry, Ch. 14 in Encyclopaedic handbook of emulsion technology, ed. by J. Sjöblom, 7-47, 001.

8 8 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 Experiment Oil Water Serie I Crude A Synthetic crude: 7 g NaCl/l Serie II Crude A Brine A Watercut [vol.%] 50 Orifice diameter [mm] 0.6/0.7/ 0.8/ /0.8/ /0.7/0.8/ 0.9/1.0 Flow rate [l/h] Serie III Crude B Brine B 0/ Pressure [bar] Temp. [ C] CH 4 Encapsulation Microscopy ambient 0 no no no ambient 0 no yes yes ambient, 1,4 Table 1: The experimental conditions of the different Experiment Series performed 60/80 no/yes yes yes Oil Composition [%] Asphaltenes Saturates Aromatics Resins Crude A Crude B Table : Composition of crude oil A and B (SARA-method) Oil Dynamic viscosity at 0 C [mpa.s] Dynamic viscosity at 0 C [mpa.s] Density at 0 C Density at 0 C Crude A Table : Physical properties of crude oil A at 0 C and 0 C Oil Dynamic viscosity at Dynamic viscosity at Density at 60 C Density at 80 C 60 C [mpa.s] 80 C [mpa.s] Crude B Table 4: Physical properties of crude oil B at 60 C and 80 C Water HCO CO SO4 Cl NH Na K Ca Mg N Brine A < Brine B < Table 5: Concentrations in milligrams per litter of components in brine A and B Water Dynamic viscosity at 0 C [mpa.s] Dynamic viscosity at 0 C [mpa.s] Density at 0 C Density at 0 C Brine A Table 6: Physical properties of brine A at 0 C and 0 C Water Dynamic viscosity at 60 C [mpa.s] Dynamic viscosity at 80 C [mpa.s] Density at 60 C Density at 80 C Brine B Table 7: Physical properties of brine B at 60 C and 80 C

9 SPE 7147 EMULSION FORMATION IN A MODEL CHOKE-VALVE 9 Organic phase: Brine A Brine A mol/l TETRA Brine A Brine B Brine B mol/l TETRA Brine B Oleic phase: Crude A Crude A Crude A mol/l TDC Crude B Crude B Crude B mol/l TDC After: initial final initial final initial final initial final initial final initial final * Interfacial Tension [mn/m] Table 8: The initial and final steady state values of the interfacial tension between the different brines and crude oils; the value marked with a star was measured after one hour without reaching steady state. Watercut [vol.%] Crude A Crude B Table 9: Values of the exponent n for different crude oils and watercuts. Orifice diameter [mm] n Correlation coefficient Table 10: Values of the exponent n and the correlation coefficient for different orifice diameters and a water-cut of 50 vol.% (crude A). [W/kg] D meas D trans, max [µm] t res [ms] [µm] Table 11: Results of transient droplet diameter calculations. Experiment serie Re t Re o Re d, initial Re d, final II III Table 1: Range of tube and orifice Reynolds numbers, and initial and final droplet Reynolds numbers for Experiment series II and III. Figure 1: Turbulent flow through a tube with an orifice (above) and the corresponding axial pressure.

10 10 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 h=1.5 mm reservoir Do Dt 10 cm orifice 1 cm Emelt [J/g] Do=0,8 mm Do=0,7 mm Do=0,6 mm Do=1 mm [W/kg] Figure : Melting energy per gram of emulsion as a function of average energy dissipation rate for different orifice diameters and a watercut of 50 vol.%. Water 4 mm 4 mm Oil Figure : The experimental set-up Pperm [Pa] Do=0,6 mm Do=0,7 mm Do=0,8 mm Do=0,9 mm Do=1,0 mm Serie III, Do=0,8 mm /U t (1/ 4-1)(1- ) [Pa] Figure 4: Permanent pressure drop along the orifice as a function of 1 1 ρu t 4 11 β for experiment series II and III. β

11 SPE 7147 EMULSION FORMATION IN A MODEL CHOKE-VALVE 11 crude A, w=0% Do=0,6 mm Do=0,8 mm 1,0E-04 crude B, w=0% Temp.=60 C Temp.=80 C Do=1,0 mm Dav [m] Dav [m] 1,0E+06 1,0E+07 1,0E+08 [W/kg] Figure 5a 1,0E+05 1,0E+06 1,0E+07 [W/kg] Figure 5b crude A, w=50% 1,0E-04 crude B, w=50% Dav [m] Dav [m] 1,0E+06 1,0E+07 1,0E+08 [W/kg] Figure 5c crude A, w =50% Do=0,6 mm Do=0,7 mm Do=0,8 mm Do=0,9 mm Do=1,0 mm Dav [m] 1,0E+05 1,0E+06 1,0E+07 [W/kg] Figure 5d Figure 5a-5d: Average droplet diameter as a function of average energy dissipation rate for crude oils A and B and for a water-cut of 0 and 50 vol.% respectively. 1,0E+06 1,0E+07 1,0E+08 [W/kg] Figure 6: Average emulsion droplet diameter as a function of average energy dissipation rate for different orifice diameters, for crude A oil and for a watercut of 50 vol.%.

12 1 P.H. JANSSEN, C. NOIK, C. DALMAZONNE SPE 7147 D [m] 1,0E-04 energy dissipation rate=1.4e6 energy dissipation rate=.e7 energy dissipation rate=4.9e6 D meas 1E-07 1E-06 0, ,0001 0,001 0,01 0,1 1 t [s] t res Figure 7: Theoretical evolution in time of the maximum droplet diameter for three energy dissipation rates (lines) and the corresponding average measured droplet diameter-residence time data points (single symbols). Dav [m] 1,0E-04 Kolm., crude B Hinze, crude B Crude B, 50 vol.% water Crude B, 0 vol.% water Crude A, 0 vol.% water Crude A, 50 vol.% water Kolm., crude A Hinze, crude A 1,0E-07 1,0E+05 1,0E+06 1,0E+07 1,0E+08 [W/kg] Figure 8: Experimental results of average droplet diameter as a function of average energy dissipation rate (symbols) and theoretical values of the droplet diameter according to Hinze (eq. 4) and the Kolmogorov length scale (eq. 6). Cacrit crude B, T=60 C crude B, T=80 C simple shear flow 1 0,1 plane hyperbolic flow 0, ,0001 0, d/ c Figure 9: The data points for the critical capillary number of Experiment Series III in the graph of Grace for different flow rates and temperatures.