SPE 28993 Society of Petcloun Ergooars Thesmodynamic Modelling for Prediction of Asphaltene Deposition in Live Oils Roberto Cimino, Sebastiano Correra, and PA. Sacomani, Eniricerche SpA, and Carlo Camiani, Ag SPR rors, ayo Becta ghee, ‘vower ne pdr anemone Enns Sms on OW Chr ed Ba Al TE USA 47 ey 868, “epg me cn pata 9 FE pce bra bn raed Soe of ue Ego finite oes eeu tagece ton erate Powe Shaman Enns ances my oe ee ee ere ABSTRACT Original phase equilibria data on asphaltenic mixtures show that the principal modeltistic, agproaches described in the fiterature do not accurately reproduce asphaltene phase behaviour. A new, more reliable modellistie approach is proposed in this paper, and 2 procedure for predicting the stability of asphaltene solutions is outlined. Application of the new model to two oils gives satisfactory results when compared with field observations. INTRODUCTION Asphaltene deposition in wells and in production facilities is more and more frequently encountered." The in situ formation of asphaltene deposits can very seriously damage the reservoir formation, leading t0 2 dectine in the oil productivity. The occurrence of such a phenomenon in the early stage of production, in a major italian field has been described elsewere? and led to the organization of an ambitious, multdisciplinary study of the problem. One of the goals of the study was the development of a reliable, predictive model for asphaltene destabilization, order to assist the petroleum engineer in optimizing the production strategy. ‘References and illustrations at ead of paper. In this paper a modelling approach based on polymer solution thermodynamics has been followed. The ‘model has been developed using experimental phase behaviour data from laboratory tests. Accurate data for asphaltene phase stability have therefore been produced: on the basis of such data a new thermodynamic interpretation for asphaltene stability is proposed. The mode! allows the prediction of asphaltene stability from a limited amount of experimental data, The asphaltene stability predictions relative to two oils from italian fields suffering from asphaltenic deposition are presented and compared with the observed field behaviour. BACKGROUND Asphaltene stability in crude oils has been studied since the 30's’, many fundamental aspects of the problem, however, have yet to be solved, Though recent studies" address asphaltene aggregatio ails, the details of asphaltene stabilization mechanism have not yet clarified. Attempts to model asphaltene solution behaviour, therefore, suffer from this lack of fundamental insight. Consequently, of the modellistic approaches proposed in the fast ten years, some have relied on a colloidal mechanism to describe the asphaltene phase stability, some have adopted a fully thermodynamic approach,2 ‘Thermodynamic modelling of asphaltene deposition ‘Among the models based on a colloidal view of the problem, the work of Leontartis and Mansoorif can be cited: asphaltene stabilization is attributed to steric. stabilization peptization by means of the so-called resin fraction, and asphaltene destabilization is described as an irreversible flocculation of the insoluble asphaltene particles. On the other side of the spectrum, several models have adopted the idea of # solubility mechanism" The work of Hirschberg and co-workers’ is perhaps the best-known and most widely used."™" On the basis of a simplified Flory-Huggins theory for polymer solutions, Hirschberg derived a model describing asphaltene stability in terms of reversible solution equilibria. All of the proposed models have the common feature of using a fiquid-solid (colloidal approach) or a Tiquid-pseudotiquid —modet (thermodynamic approach) to describe asphaltene stability in stock tank oil, coupled with a vapourliquid model to represent the influence of the gaseous components present in the live oil. First the vapour liquid equibibrium (VLE) is usually solved, to determine liquid phase properties; then the liquid-pseudoliquid equilibrium is calculated, assuming no influence of the precipitated asphaltene phase on the previously calculated VLE. VLE is calculated by means of an ‘equation of state (EOS) : widely used are the SRK (Soave-Redlich-Kwong), PR (Peng-Robinson) and BWR (Benedict-Webb-Rubin) EOS. The lack of fundamental experimental data on asphaltene phase stability”, however, does not allow a fully-satisfactory comparative assessment of the proposed models: published asphaltene precipitation data are few, and the experiments have been carried out on different oils for which the analytical characterisation available are not always complete, Accurate experimental studies on asphaltene phase stability were therefore carried out and used to ‘evaluate the principal proposed models EXPERIMENTAL Cloud point measurements" have been proposed for polymer solutions as techniques to investigate the phase behaviour of large, polydisperse molecules. Such techniques are usually carried out by visually observing the phase separations, induced for example by temperature change or antisolvent addition 500 SPE 28993 AAs far as asphaltenic solutions are concerned, visual observation is not feasible because of their dark colour even at high dilution. Spectrophotometric techniques, however, have been used”! for asphaltene stability detection in order to assess the efficiency of asphaltene solvents and precipitation inhibitors. A spectrophotometric technique has been employed in the present study to detect the cloud point of asphaltene solutions. The experimental apparatus used is shown in fig. 1, By means of a peristaltic pump, an asphaltene solution (or an oil) is recirculated through a UV spectrophotometer (mod. HP 152A), where the solution absorbance at 800 fm is recorded. An antisolvent is added by an automatic titrator to the asphaltene solution, leading to phase separation (cloud point or onset of flocculation") which is detected as a minimum in the recorded absorbance. This is related, as indicated previously by Hotier", to the scattering of ‘transmitted light due to phase nucleation, The solntion composition at the cloud point is determined in terms of the antisolvent, solvent and asphaltene weight fractions. In order to investigate the influence of asphaltene concentration, asphaltene-solvent solutions st various asphaltene compositions were prepared All the experiments were carried out at atmospheric pressure and 25 °C. In order to avoid any uncertainty in the interpretation of the data obtained, model mixtures were initially investigated. The “inode!” mixtures were solutions of an asphaltene fraction (n-pentane and n-heptane precipitate from stock tank cil), @ solvent (toluene or tetraine), playing the role of the "oil", and an antisolvent (a-pentane, n-heptane, n-decane), added in order to simulate a decrease of the oil's solvating power. ‘Two different asphaltenic fractions (n-pentane and heptane) were selected for measurement, The CS asphaltenes are a less polar asphaltenic fraction than the n-C7 asphaltenes. They can be considered as the sum of the n-heptane asphaltene fraction plus a fraction (C5- C7) which is close, as far as polarity is concened, to the resin molecules. Asphaltenes were obtained according to a modified TP 143 procedure. The C, H, N content, determined by elemental analysis and number molecular weight (obtained by GPC) are shown in Table 1 for n-C5 and n-C7 asphaltenes from a Villafortuna-Trecate (VE-TR) if, Cloud points (the onset of flocculation) for toluene solutions of the n-C$ and n-CT VE-TR asphaltenes are given in Table 2. Some3 of the n-C7-asphaltene data are shown in Fig, 2 , in terms of solvent weight per asphaltene unit mass (x axis) and antisolvent weight per asphaltene unit mass (y axis); the solvents used were toluene and tetrain, while the precipitant was n-pentane. ‘As can be seen from the Figure, the trend of y (antisolvent weight/asphaltene weight) with x (olvent weightasphaltenc weight) is linear: a straight line passing through the axis origin correlates the experimental points. This behaviour indicates that the cloud independent on the asphaltene concentration. in ‘other words, the ratio y/x is constant. These results are in agreement with the data reported in the literature for the onset of flocculation'*'* of other oils. Another observation concerns the stability of n-C5 asphaltenes in relation to the n-C7 fraction (Fig. 3): when cloud points for these two fractions are compared in the same solvent, the n-CS asphaltenes show an enhanced stability. This result shows that the CS5-C7 fraction enhances n-C7 asphaltenes stability. Evidently, therefore, the asphaltene molecules less polar than the n-C7 fraction play 2 role in their stabilization. This observation is in agreement with reported behaviour of the so-called resins. Finally, cloud point measurements were carried out on the VE-TR stock tank oil; detections were performed at atmospheric pressure and 25 C, Also in this case a solvent (toluene) was used prior to antisolvent (nC7) addition: although not necessary from a theoretical standpoint, solvent was used t0 redissolve the spontaneous asphaltenic deposit present in the stock tank oil. A straight line connecting the experimental data was again obcained; in this case the straight line does not pass through the origin, but shows an intercept a= -0.467 for (extrapolated) x-0. This is tobe expected, since at x=0 (no solvent addition to oil) y represents the volume of precipitant needed to destabilize asphaltenes from oil. Noteworthy is the %, negative vaiue, VF-TR stock tank oil, in fact, shows spontaneous aphaltenic deposition. A “negative” mass of precipitant is to be added to destabilize the ot 501 R.Cimino, $.Correra, P.Sacomani, C.Camiani SPE 28993 ASSESSMENT OF THE LITERATURE MODELS The two principal approaches for modelling asphaltene stability -Hirschbere's thermodynamic and Leontaritis and Mansoor's colloidal model- have been evaluated by their fitting of the experimental data obtained Hirschberg’s model is represented by the equation: = 1) + 46.-8)"=0...0) where 6, is the volumetric asphaltene fraction at the onset of flocculation, v, and v,are the molar volumes of asphaltenes and oil (considered as the solvent for asphaltenes), 5, and 8, are their solubility parameters, R is the universal gas constant, and T is the temperature. Leontaritis and Mansoori's approach is summarized by (2) . Au RT = In()+ 1 BAB 8.) 2) where Au.’ is the chemicat potential of the resins in the solvent (oil) phase, 6, is their volumetric fraction. in the solvent at the onset of flocculation, v, is the resin molar volume, and 5, and 5, ate the resin and solvent (oil) solubility parameters. At the onset of flocculation Au = An", ‘The parameter Au", value, to be determined from ‘onset of flocculation experiments, is defined as the resin critical chemical potential, and corresponds to the minimum amount of resins adsorbed onto the asphaltene particle surface capable of peptizing the asphaltenes, When Awe, < A”, @ the system is no fonger stable and extensive floceulation takes place. Hirschberg's model has been tested by fitting both n-C5 and n-C7 VE-TR asphaltene model solution data; results, expressed as AAD {absolute average deviation, see Tab. 3 for definition), are shown in Table 3.‘The solubility parameters and molar volumes of pure solvents have been taken from the literature."*"” 4, is given by the asphaltene concentration at the ‘onset of flocculation and v, is determinated from the ‘molecular weight of the n-C7 asphaltenes, hypotising asphaltene density p, equal to 1.22 glee. 8, is used as fitting parameter. It clearly makes no sense to test Leontaritis! ‘approach on asphaltene model mixtures: according tothe mathematical description underlying the colloidal model, a resin fraction must be present in the system in order to stabilize the asphaltenes, Since no resins are present in the model mixtures, the model forecasts complete insolubility in any given soivent; this is neither the case for the studied systems nor for many observations reported in literature (asphaltenes do dissolve, provided a proper solvent is chosen). Such a simple observation raises some doubt about the physicat basis of this model, Leontaritis and Mansoor’s model, therefore, tas been tested with stock-tank oil cloud point data, in ‘which resins are present. Hirschberg’s model has also been fitted to the same data; 6, and v, for Hirschberg model have been estimated as previously described 4, and v, for Leontaritis model have been estimated from experimental values: 6, is given by the concentration of resins at the onset of flocculation and v, is determined from the resin molecular weight, Ihypotising a resin density equal to 1 g/cc. In both cases, however, oil enters the picture, and its solubility parameter at 25°C has to be calculated, ‘One possibility would be to evaluate it from an EOS provided oil characterization is available; such a calculation however is strongly dependent on oil characterisation and, in particular, on assumptions on its heavy ends." To avoid any uncertainty, therefore, the oil solubility parameter has been taken as a fitting parameter for both models. As can be seen from Tab. 3, the ability of both theoretical approaches to reproduce the observed behaviour is poor: the reason can be readily understood by inspection of eq. 1 and 2: both equations, in fact, foresee a dependence of the cloud poinvflocculation threshold on concentration (asphaltene concentration for Hirschberg model and resin concentration for Leontaritis model), while the observed behaviour show an independence of the cloud point on concentration. ‘Thermodynamic modelling of asphaltene deposition SPE 28993 A correct application, however, of palymer solution thermodynamics” allows a good representation of the experimental phase behaviour (see appendix A for theorical derivation), by means of the following equation: His fR6.-o0'o? (5) ‘where the symbols are as defined above, except for $e which is asphaltene volumetric fraction in the nucleating phase at the cloud poinvonset of flocculation: such an equation is derived under the hypothesis that, on phase separation, not a pure asphaltene phase mucleates but a phase concentrated in asphaltenes containing also a fraction of solvent. In Table 3, AAD for calculated (from eq. 5) and experimental points is given: the correlation is, in this case, quite reasonable, Figure 4 also reports a comparison between the models as far as their ability to ft VE-TR stock tank oil data is concerned. The poor ability of Hirschberg's model to reproduce experimental behaviour is related to a simplifying assumption made in deriving the final form of eq. 1 from the original Flory's theory (see app. A). in particular, his assumption that the ‘precipitated phase is pure asphaltene" leads, from a mathematical standpoint, to the prediction of a dependence of the cloud point on the asphaltene concentration ‘The same simplifying assumption has been also made by other researchers’, who emploied a more complicated polydistributed polymer solution theory, ‘once again, this leads to a predicted dependence of flocculation threshold on asphaltene concentration USE OF THE MODEL PROPOSED ‘As can be seen from eq, 5, in order to use the model it is necessary to estimate solubility parameters and molar volumes for the oil and the asphaltenes and 4", Experimental data on VF-TR n-C5 and n-C7 asphaltenes show that the C5 - C7 fraction has an influence on n-C7 asphaltenes: it is therefore reasonable to obtain experimental value of asphaltene solubility parameter by cloud points measurements carried out on the stock tank oil Tn stock tank oil, in fact, the whole distribution of polycondensated, high molecular weight molecules involved in the equilibrium is present, Performing the experiment on stock tank oil thus avoids the5 R.Cimino, $.Conrera, P.Sacomani, C.Camiani necessity to identify and separate the whole distribution of such molecules playing a role in phase separation. Data on stock tank oil, moreover, can supply an experimental value for oit solubil parameter at the temperature of the experiment: this value can be used to tune the EOS as far as the heavy ends are concerned, Any EOS, in fact, requires a pseudoization procedure of the oil: oil has to be split into several distillation cuts, of which normal boiling temperature and density (or molecular weight) must be known in order to calculate the pscudocomponent critical temperature, critical pressure and acentrie factor, One of the major problems of such a procedure concems the heavy oil ends, for which normal boiling point temperature and, in some case, even molecular weight are usually not reliable, The experimental oil solubility parameter, therefore, can be used to tune the normal boiling temperature of heavy oil ends, thereby assuring a more reliable calculation of this parameter at other temperatures and pressures, The proposed predictive approach is summarized in Fig. 5, and it is essentially made up by the following steps: 1. Experimental determination of stock tank oil cloud points: an antiso\vent (n-heptane, for instance) is added, at a certain temperature, to stock tank oil, diluted with different solvents. 2.8 8 and", determination: by means of eq. 5. experimental cloud points are regressed to find the three parameters at the temperature of the experiment, v, and v, must be known in order to use eq. 5: oll molar volume can be calcufated from stock tank density and molecular weight. v, can be approximated from n-C7 asphaltenes molecular weight (determined, for instance, by GPC) and @ value for asphaltene density (p,). In all our calculations, p, was set equal to 1.22 glee. 3 Caleulation of asphaltene phase stability at reservoir conditions : eq.5 allows the determination ‘of asphaltene phase stability provided v,, 8, and 6°, are known at reservoir conditions (reservoir temperature ,T., and pressure, P.). As far as v, is ‘concerned, one can assume that temperature and pressure do not influence asphaltene molar volume: v, value calculated from GPC molecular weight and py can be retained . As far as 8, is concerned, an SPE 24993 independence of this parameter on pressure can be assumed’: only the dependence of 8, on temperature has to be determined, therefore. If the experimental measure of cloud points is carried out at reservoir conditions, 5, at T, is obtained fiom eq. 5 by regression of the experimental data. When the reservoir temperature is, very high, such measurements are not convenient; 8, at T, can, in this case, be calculated from low ‘temperature 8, value by using the correlation” Be(T) = Sa(To)exp [-9.1 * 10“ -To)}....(6) 4°, was found by numerical simulation not to be very much sensitive to solvent power variation: it js,therefore, assumed constant and equal to the value obtained from stock tank oil cloud point measurements Once 8, and v, at T, have been calculated, it is possible to calculate the foreseen asphaltene phase stability curve at reservoir conditions: eq. 5, in fact, becomes a function of & and v,once temperature has been fixed. The infinite couples of 8,0, values satisfying eq. 5 describe a curve, on the 8,,v, plane, which divides the plane into two regions (see Fig. 6). One region corresponds to solvent phases with stabilizing characteristics for asphaltenes, the other to solvent phases which separate out asphaltenes. Once the solubility parameter and molar volume are calculated for the live oil, itis possible to assess whether the live oil representative point (6,, v,) falls in the stability region (one phase) or in the two-phase area of the above mentioned plane. 4. Live oll molar volume and solubility parameter caleulation: A cubic equation of state can be used to calculate these parameters: in particular, we used a modified SRK EOS ™. Standard PVT analysis are used to obtain the pseudocomponent critical properties and acentric factors, the C,,. boiling ‘temperature is tuned on experimental stock tank oil solubility parameter. Interaction constants of the EOS are then tuned to reproduce reservoir bubble pressure and live oil density. The tuned EOS is used to calculate both 6, and v, 6036 ‘Thermodynamic modelling of asphaltene deposition MODEL APPLICATION ‘The proposed approach has been used 10 predict asphaltene stability for two oils (VF-TR and oil A) suffering from asphaltenic deposition. In Table 4 are reported PVT data and main analytical determinations for these two oils. For the VE-TR case, formation damage, attributed to asphaltenic deposition, has been observed since the earliest stages of oil production: the live oil is underseturated in the reservoir, since the reservoir static pressure is 1055 bar and the bubble point at the reservoir temperature is 177.8 bar. Asphaltene destabilization is, in this case, due to natural reservoir depletion and the formation damage occurs at pressures much higher than the ‘bubble poi. In the second case, oil A, asphaltene deposition has, heen observed in the producing string: a destabilization pressure has been evaluated around 260 bar. Again, at a pressure higher than the live oil bubble pressure, asphaltene destabilization occurs. Following the outlined procedure for prediction, calculations for the two cases have been carried out For the VF-TR case, owing to the high reservoir temperature, cloud points for stock tank oil at room temperature have been used: 8, has been therefore extrapolated to reservoir temperature by means of eq. 6. For oil A, cloud point experiments have been catried out at reservoir temperature: 5, obtained from these data has been used without extrapolation In Fig. 7 and 8, asphaltene phase stability curves for the two cases are shown: &, and v,, are reported for the two oils at reservoir temperature and different pressure conditions, spanning from reservoir pressure to oil bubble point. For the VE-TR case, a destabilization pressure of around 800 bar is calculated, while for oil A the phase separation pressure is 210 bar. The calculated values show that, in both cases, asphaltene deposition is to be expected by natural oil depletion when pressure is ‘well above bubble point: the phenomenon, of course, increases as pressure drops down to oil bubble point. ‘The predictions for the stability of these oils compare reasonably well with the observed field behaviour, although experimental high pressure ‘onset of flocculation data should be produced in order to allow a quantitative assessment of the predictive power of the proposed model. Work is currently in progress to produce such data. 504 SPE 28993 CONCLUSIONS Phase behaviour data carried out on several asphaltene-solvent-antisolvent mixtures. show how the principal approaches to modelling, asphaltene stability do not satisfactorily reproduce the physical behaviour of the system. A more correct application of polymer solution theory leads to a better representation of asphaltene phase behaviour and to a new model for asphaltene stability. Predictions of asphaltene stability in two live oils suffering from asphattenic deposition have been carried out by means of the model coupled with a properly tuned EOS on the basis of a limited amoum of experimental work. The predictions are confirmed by the field observations, showimg. in particular, that asphaltene deposition is to be expected for pressures higher than live oil bubble pressure for the two crude oils under considera NOMENCLATURE 11= number of experimental points v= molar volume of asphaltenes ¥,, = live oil molar volume v,= molar volume of oil v,= resin molar volume x= diluent weight to solute weight ratio at the onset of flocculation Y= precipitant weight to solute weight ratio at the onset of flocculation AAD = absolute average deviation R= universal gas constant ‘T= temperature T= reservoir temperature T,= reference temperature = estimated value of Z variable Z* = measured value of Z variable tercept of the y vs. x plot @, = slope of the y vs. x plot 8, = solubility parameter of asphaltenes &, = live oil solubility parameter solubility parameter of il 8, = resin solubility parameters ‘Aji, = resin chemical potential in the solventAu, = resin critical chemical potential ‘hu, = solvent chemical potential ‘AuL, = polymer chemical potential (b, = volumetric asphaltene fraction at the onset of flocculation '.= asphaltene volumetric fraction in the hucleating phase at the cloud point/onset of floceulation 4, = resin volumetric fraction in the solvent at the onset of flocculation 1. Flon's interaction parameter ,= asphaltene density ACKNOWLEDGMENT ‘The authors thank Mr. B. Radaelli for valuable experimental support and Dr. G. Soave for helpful discussions on thermodynamic modelling ‘The permission of Eniricerche and Agip to publish ‘this paper is gratefully acknowledged. REFERENCES. 1. Thayer, R. Nicoli, D.C. and Dick, G.: “Asphaltene Deposition in Production Facilities” SPE 18473. Houston (1989) 2. Galoppini, G. and Tambini, M. : "Asphaltene Deposition Monitoring and Removal Treatments: An Experience in Ultra Deep Wells"SPE 27622. Aberdeen (1994) 3. Nellenstein, F.I. : "The Colloidal Structure of Bitumens" The Science of Petroleum. A. E, Dunstan (ed). Oxford U. Press, London (1938) 4 4, Espinat, D. and Ravey, J.C. : "Colloidal Structure of Asphaltente Solutions and Heavy Oil Fractions Studied by Small-Angle Neutron and X-Ray Scattering " SPE 25187 . New Orleans (1993) 5. Camahan, NF, Quintero, L.. Pfund, DM, Fulton, JL, Smith, R.D., Capel, M. and Leontaritis, K. :*A Small Angle X-Ray Scattering Study of the Effect of Pressure on the Aggregation of Asphaltene Fractions in Petroleum fluids under Near-Critical Solvent Conditions" Langmuir (1993), 9, 2035-2044 6, Leontaritis, KJ, and Mansoori, G.A.: "Asphaltene Flocculation During Oil Production and Processing: A Thermodynamic-Coiloidal Model". Proc. 1987 R.Cimino, $.Correra, PSacomani, C.Camiani SPE 28993 SPE Inter. Symp. on Oilfield Chem, San Antonio, Texas (Jan.), SPE 16258, 149-158. 7. Hirschberg, A, de Jong, L.NJ., Schipper, B.A. and Mejiers, J.G.: "Influence of Temperature and Pressure on Asphaltene Flocculation", SPE J., June 1984, 283-293 8. Kanawaka, S., Park, S.J. and Mansoori, G.A. "Organic Deposition From Reservoir Fluids: A Thermodynamic Predictive Technique”, SPE. Res Eng., May 1991, 185-192. 9. Chung, F.T-H.: "Thermodynamic modeling of organic solid precipitation", SPE 24851, oct.92, 10. Kokal, S.L.. Najman, J, SayeghS.G, and George, AE.: “Measurement and Correlation of Asphaltene Precipitation From Heavy Oils by Gas Injection”, J. of Can, Petr, Techn,, April 1992, Vol. 31, No. 4, 24-36, 11. Burke, N.E., Hobbs, R. E, and Kashou, S.P. "Measurement and Modeling of Asphaltene Precipitation", JPT, Nov. 1990, 1440-1446. 12. Leontaritis, KJ. : "Asphaltene Deposition: A Comprehensive Description of Problem Manifestations and Modeling Approaches", SPE Production Operations Symp. SPE 18892, Oklahoma City (1989). 13. Kawate, K, Imagawa, I. and Nakata, M "Cloud-Points "of Temary System Nitroethane+CyclohexanetPolystyrene Determined by a Novel Method", Polymer J, (1991), Vol. 23, No3, 233-240 14, Hotier, G. and Robin, M.: "Action de divers iluants sur es produits pétroliers lurds: mesure, interprétation et prévision de la floculation des asphaltenes", Revue de ILFP. (1983), 38 (1), 101-119 15, Reichert, C., Fur, BJ. and Klein, LL, 1. Can Petr, Eng. (1986), Sept -Oct, 33-37. 16. Barton, AF-M.: "CRC Handbook of Solubility Parameters and Other Cohesion Parameters", CRC Press, Boca Raton (1991) 17, Perry LHL: "Chemical Engineers’ Handbook", McGraw-Hill, New York (1963) 18. Pedersen, K.S., Fredenslund, Aa, Thomassen, P.: "Properties of Oils and Natural Gases" (1989), Gulf Publishing Company, Houston 19, Tompa, H.:"Polymer Solutions", Butterworths, Scientific Publications, London (1956). 20. Soave, G.“Application of Equations of State and the Theory of Group Solutions to Phase Equilibrium Prediction", Fluid Phase Equilibria, 87 (1993), 23-25.8 ‘Thermodynamic modelling of asphaltene deposition 21 Prausnitz, J.M., Lichtenthaler, RLN., de Azevedo, E.G.: "Molecular Thermodynamics of Fluid-Phase Equilibria", Prentice-Hall Inc., Englewood Cliffs (1986). APPENDIX Polymers phase equilibrium modelisation can be faced in different ways: in particular, complex polymer physical reality (molecular weight and, sometimes, chemical characteristics distribution) may be taken into account by a number of complicated models, The easiest way to describe polymer phase behaviour, however, is classic Flory's pseudobinary approach”, Such an approach assumes that the studied system is made up by two ensembles of molecules: “solvent” (s), which clusters “low' molecular weight species and “polymer", which represents “high molecular weight 5 homogeneous macromolecule characterized by a single value of molecular weight On phase splitting, thermodynamic equilibrium of such a system is described by means of the equality of chemical potentials relative to the two species (solvent: "s" and polymer: "p") in the two phases: seve AM,” AH, AH, = BH, Az Some simplification of the system of these two equations can be done by considering the physics of phase splitting in this particular case. ‘The set of the above equations represents a physical reality that can be also depicted in a versus polymer volume fraction diagram, % parameter describing the solvent power of the solvent with respect to the polymer (fg. 9). In such a diagram, a tipical miscibility gap between ‘s" and "p" is reported: being such a gap fiction of the ratio r (polymer molar volume versus solvent molar volume), in fig. 9 a curve for r10 is shown, This ¢ value is typical for an asphaltene-oil solution, as far as asphaltenes and oils molar volumes are concemed. When we consider typical asphaltene volume fraction in oils, it can be estimated in the range of 10° to 10° for oils with asphaltene weight fraction from 1 to 10. An oil stable with respect to asphaltene flocculation is therefore represented by SPE 28993 point A in fig, 9: as oil (or mode! solution) loses. for any reason, its solvent power, x, parameter decreases Until the point reptesenting the oil crosses the miscibility gap curve. On onset of flocculation, two phases are formed: point B represents the majority phase (polymer-or asphaltene-tean) while point C epresent the minority phase (polymer-or asphattene-rich), ‘As can be seen from fig. 9, while minority phase is concentrated in polymer (but by no means pure in polymer), majority phase has a very low asphaltene concentration: it can safely be assumed to be pure solvent, that is 9, = 0. Such an assumption leads to the final equation Aw’ ai=0 AS ‘Noteworthy is the consideration that Hirschberg simplifies as well the same set of equations, but assuming 2 pure asphaltenic phase nucleation; such an hypothesis leads to derive his model, which shows. an exponential asphaltenic concentration dependence By using Flory's theory to express Au", we obtain: tnc1—o5)+(1-¥2 Jor ex0%=0...a4 Expressing x as!! =24(8,—8) w= PEG. 8)?.. AS the final form (eq. 5) is obtained. By solving equation A4 for different values of and rit is possible to see that *,, in the dilution region of interest for oils, it is not very sensitive: it generally ‘can be assumed constant.Table 1: VF-TR asphaltenes properties nC5-Asphaltenes | nC7-Asphaltenes IMn 1000 1300 Ic [Yow] 89.50 89.50 IH [%w] 669 6.30 [%w] 073 0.75 ‘Tle 2 weigh facto aloud pat | sao sons eas Os Sao son Sa ase ines tao at ore ois; tise ; war fot Ge kar : a 2 oss Ose oan seer fam at coos at Osea Sant —— ti os con cm a bat oo sr ka die som ses aroma oar Seto cTOR sat | caso ooo oes | ote cots oe on ou oa ‘cimeaias | tr i TOR oat —tar j cas (807 | cas tows sss ___ toesTable 3: Absolute Average Deviation (AAD) of models." ‘Model Hirschber “This model [C5 asphaltenes mixtures 14 = 62 C7 asphaltenes mbxtures | 18.7 = 12.4 ‘Stock tankoil | 29.7 $3.7 69 * AAD is defined by AaD= 1S ‘Table 4: characterisation data of oils, [Stock tank Oil VF-TR OWA "APL 12 35.6 [Reservoir pressure [bar] 1055, 398 [Reservoir temperature [°C] 172 56 [Live oil bubble pressure [bar] 177.8 135.7 Mw 4 158. {c02 tw] 0.00 0.03 IN2 [Yow] 0.00 310° fmethane [24w] 0.00 0.33 ethane [w] 5.24 10° 0.86 ropane_(%w] ou 1.86 j-butane [%v] 0.27 1.02, fn-butane (%%w] 0.58 218 i-pentane [Your] 0.61 1.62 foepentane [%w] fhexanes (26w] lheptanes [24w] Joctanes [2%W] Joonanes ["4w] [deoanes [76w] z fundecanes [26w] 6.50 374 | {dodecanes [Yaw] 3.79 3.30 [c13+ (ow 61.32 59.76 IResins [Yow] 1.30 4.80 IC5-C7 Asphaltenes [%w] 0.10 0.50 C7 Asphaltenes (%w] 0.30 1.40Figae 1: Experimental appari, pee o “ s 0 @ © © © —— ne 3 .Riaonbac "a penadne pnp, Fo contr 250 p= « toluene 4 tetraline Figure 3: solvent tetraline/antisol vent n-CS 300 - a «= CS-Asphaltenes , C7-Asphaltenes 609Figure 4: comparison between models 5 oo ‘stock tank oll VF-TR Solvent: toluene Hirschber 4 | pntisoivent: n-C7 . ‘a Experimental values 4 Leontaritis Figure 5: Proposed predictive approach Onset of Mloceutation a ata ‘nC7 asphaltenes Ma. and Density data Live oil predictions PVT report Joe ata 51020 19 17 BIMPx“(1/2)] oil solubility parameter [MPa*1/2] 50 fo 100 Figure 6 Stability Asphaltene flocculation 150 200 250 300 Vg [ecfmole} 2 phases 140 160 180 200 oil molar volume [ml/mol] BitFlory interaction parameter oil solubility parameter [MPa’1/2] gE = 2 phases onset of asphaltene flocculation 135 1 145 oi! molar volume {ml/mol] Figure 9: polymer-solvent miscibility gap 1 ey SEP | A 0.95 g 09 1 | ret 085 | | os | : : re 0 Of 02 03 04 OS 06 Polymer volumetric fraction 512