ttGOAosMSM

Inge Harald Auflem Influence of Asphaltene Aggregation and Pressure on Crude Oil Emulsion Stability Influence of Asphaltene Aggregation and Pressure on Crude Oil Emulsion Stability

by

Inge Harald Auflem

Thesis Submitted in Partial Fulfilment of the Requirements for the Degree of

DOKTOR INGENI0R

Department of Chemical Engineering Norwegian University of Science and Technology Trondheim, June 2002 Preface

Preface

This thesis, submitted in partial fulfilment of the requirements for the degree of dr.ing. at the Norwegian University of Science and Technology, consists of five articles, one patent and one book chapter. The thesis is based on work performed at Statoil Research Centre, the Institute for Surface Chemistry in Stockholm, and Chalmers University of Technology in Gothenburg in the period from August 1999 to June 2002.

My supervisor introduced me to the field of surface and colloid chemistry in 1997, while studying physical chemistry as an undergraduate student at the University of Bergen. Little did I then know that the work would lead me into the petroleum industry, and a struggle with "the good and the bad asphaltenes". As an experimentalist, the work in the laboratory has, more often than not, provided results difficult to interpret. Nevertheless, the work has continued, and in-between the total failures, there has been some near successes, which have resulted in the thesis you are now paging through.

During my time as a dr.ing. student I have had the fortune to participate in a project with the acronym FLUCHA II, which stands for Fluid Characterisation at elevated pressures and temperatures. The group has consisted of 3 dr.ing. students and 1 post doc, under the guidance of the enthusiastic and demanding Prof. Sjdblom. The work tasks of the group have covered a number of flow assurance related problems, i.e. asphaltene precipitation, emulsion formation and stabilisation, naphthenate formation, crude oil characterisation, etc.

i Acknowledgements

Acknowledgements

First of all, I would like to express my sincere gratitude to my academic advisor Prof. Johan Sjoblom, for his years of guidance and invaluable encouragement, along with generous hosting when teambuilding throughout this research work.

During my three years of work, I have enjoyed the opportunity of working with several fellow graduate students and postdoctoral associates, which have provided intellectual assistance and an enjoyable working environment. I am also grateful to all my co ­ authors, whom I have had the fruitful pleasure of collaborating with.

I would also like to acknowledge the FLUCHA II program financed by the Research Council of Norway and the oil industry. Statoil ASA is especially thanked for providing office space and access to laboratory equipment.

Finally, I wholeheartedly thank Helene and my parents for their love and support, without which this work would never have been completed.

n Abstract

Abstract

Water-in-crude oil emulsions stabilised by various surface-active components are one of the major problems in relation to petroleum production. This thesis presents results from high-pressure separation experiments on "live" crude oil and model oil emulsions, as well as studies of interactions between various indigenous stabilising materials in crude oil. A high-pressure separation rig was used to study the influence of gas and gas bubbles on the separation of water-in-crude oil emulsions. The results were interpreted as a flotation effect from rising gas bubbles, which led to increased separation efficiency. The separation properties of a "live" crude oil were compared to crude oil samples recombined with various gases. The results showed that water-in-oil emulsions produced from the "live" crude oil samples, generally separated faster and more complete, than emulsions based on recombined samples of the same crude oil.

Adsorption of asphaltenes and resins onto a hydrophilic surface from solutions with varying aromatic/aliphatic character was investigated by a quarts crystal microbalance. The results showed that asphaltenes adsorbed to a larger degree than the resins. The resins were unable to desorb pre-adsorbed asphaltenes from the surface, and neither did they adsorb onto the asphaltene-coated surface. In solutions of both of resins and asphaltenes the two constituents associated in bulk liquid and adsorbed to the surface in the form of mixed aggregates. Near infrared spectroscopy and pulsed field gradient spin echo nuclear magnetic resonance were used to study asphaltene aggregation and the influence of various amphiphiles on the asphaltene aggregate size. The results showed interactions between the asphaltenes and various chemicals, which were proposed to be due to acid-base interactions. Among the chemicals used were various naphthenic acids. Synthesised monodisperse acids gave a reduction of size of the asphaltene aggregates, whereas polydisperse naphthenic acids seemed to affect the state of the asphaltenes only to a minor extent. The effect of the naphthenic acids on the asphaltenes appeared however, to depend on the asphaltene type. Other amphiphiles such as amines and alcohols, showed a varying effect on the dispersion of the asphaltenes into smaller aggregates. Furthermore, measurements of diffusion coefficients upon increased concentration of asphaltenes, implied that the asphaltenes began to self-associate at concentrations above 0.1 wt-% in toluene-dg. Table of Content

Table of Content

Preface ...... i Acknowledgements ...... II Abstract ...... ill Table of Content ...... iv List of Publications...... v Complimentary work...... vi 1 Introduction ...... 1 2 Theory ...... 3 2.1 Crude Oil Composition...... 3 2.2 Asphaltene Chemistry...... 4 2.3 Emulsions and Emulsion Stability...... 9 2.4 Stabilisation of Water-in-Crude Oil Emulsions...... 10 2.5 Destabilisation of Crude Oil Emulsions...... 12 3 Methodology and Theory ...... 16 3.1 High-pressure High Temperature Separation Rig (HPHT-rig)...... 16 3.2 Quartz Crystal Microbalance (QCM)...... 18 3.3 Near Infrared Spectroscopy (NIR)...... 20 3.4 Nuclear Magnetic Resonance (NMR)...... 22 4 Main Results ...... 26 4.1 Paper 1...... 26 4.2 Paper II...... 28 4.3 Paper III...... 33 4.4 Paper IV...... 36 4.5 Paper V...... 37 4.6 Paper VI...... 40 5 Summary and Conclusions ...... 44 References ...... 46 Papers I-VII

IV List of Publications

List of Publications

1. Auflem, I.H., Kallevik, H., Westvik, A. and Sjoblom, J., Influence of Pressure and Solvency on the Separation of Water-in-Oil Emulsions from the North Sea. Journal of Petroleum Science and Engineering, 2001. 31(1): p. 1-12.

2. Kallevik, H., Sjoblom, J., Westvik, A., Auflem, I.H., Process for separation of water and oil in a separator by breaking water-in-oil emulsions, P 4202-1, PCT- application, 23. February 2002

3. Auflem, I.H., Westvik, A. and Sjoblom, J., Destabilisation of water-in-oil emulsions based on recombined oil samples at various pressures. Journal of Dispersion Science and Technology, Submitted

4. Ekholm, P., Blomberg, E., Claesson, P., Auflem, I.H., Sjoblom, J. and Kornfeldt, A., A Quartz Crystal Microbalance Study of the Adsorption of Asphaltenes and Resins onto a Hydrophilic Surface. Journal of Colloid and Interface Science, 2002. 247(2): p. 342-350

5. Auflem, I.H., Havre, T.E. and Sjoblom, J., Near Infrared Study on the Dispersive Effects of Amphiphiles and Naphthenic Acids on Asphaltenes in Model Heptane- Toluene Mixtures. Colloid and Polymer Science, In Press, 2002

6. Ostlund, J.-A., Nyden, M., Auflem, I.H. and Sjoblom, J., Interactions between asphaltenes and naphthenic acids. Energy & Fuel, Submitted

7. Sjoblom, J., Johnsen, E.E., Westvik, A., Ese, M.H., Djuve, J., Auflem, I.H. and Kallevik, H., Demulsifiers in Oil Industry, in Encyclopedic Handbook of Emulsion Technology, J. Sjoblom, Editor. 2001, Marcel Dekker, Inc.: New York. p. 595-619

v List of Publications

Complimentary work

1. Johan Sjoblom, Einar Eng Johnson, Arild Westvik, Linn Bergflodt, Inge H Auflem, Trond E Havre and Harald Kallevik: Colloid Chemistry in Sub Sea Petroleum and Gas Processing, Presented at: "The 2nd International Conference on Petroleum and Gas Phase Behaviour and Fouling", Copenhagen, Denmark, August 27-31, 2000

2. Sjoblom, J., Kallevik, H., Aske, N., Auflem, I. H., Havre, T. E., Saether, 0. and Orr, R.: Recent Development in the Understanding of the Stability and Destabilization of Water-in-Crude OH Emulsions, Presented at: "The 3rd International Conference on Petroleum Phase Behavior and Fouling", New Orleans, USA, March 10-14, 2002

3. Johan Sjoblom, Narve Aske, Inge Harald Auflem, 0ystein Brandal, Trond Erik Havre, 0ystein Saether, Arild Westvik, Einar Eng Johnsen, Harald Kallevik, Our Current Understanding of Water-in-Crude Oil Emulsions. Recent Characterization Techniques and High Pressure Performance, A Collection of Invited Papers in Honour of Professor J.Th.G. Overbeek on the occasion of his 90th Birthday, Advances in Colloid and Interface Science, In press, 2002

VI Chapter 1 - Introduction

1 Introduction

In the North Sea, the hydrocarbon reserves include several marginal fields, which are either small or deep-water fields. A feasible economical exploitation of these reserves requires the introduction of subsea developments and multiphase fluid transport over long distances. This has made it extremely important to reliably predict and control fluid behaviour, in order to minimise the need for additional process installations or costly well interventions. Typical problems that may occur are deposition of organic matter in reservoir and process equipment, and the formation of stable water-in-crude oil emulsions when co-produced water and oil are mixed through chokes and pipelines.

The heaviest and most polar fraction of the crude oil is named asphaltenes, and gives rise to a variety of nuisances during crude oil production. It is widely recognized that flocculation and deposition of asphaltenes may occur when the thermodynamic equilibrium is disturbed. This may come as a result of changes in pressure and temperature [1-3], as a result of compositional alterations when blending fluid streams [4], or due to injection of gas during improved oil recovery (IOR) operations. The most serious precipitation problem is the creation of a formation damage [5], i.e. partial or complete blockage of the inflow zone around a well, and thereby loss of productivity. Another possible problem is adsorption of asphaltenes on to the reservoirs mineral surfaces, whereby the wettability of the reservoir is changed from water-wet to oil-wet [6] and thereby reducing the potential oil recovery. In addition, the asphaltenes may deposit on the steel walls in the production line, or be transported along in the pipeline only to accumulate in separators or other fluid processing units. Clean up of deposited asphaltenes in the field may necessitate well shut-in and loss of oil production. Hence, preventing asphaltene flocculation is preferable from both an operational and economical viewpoint.

During oil production and transportation, the water and oil phases are co-produced, and thereby exposed to sufficient mixing energy to form dispersions of water droplets in oil and, conversely, oil droplets in the water. Unfortunately, the crude oil contains a number of components, which are interfacially active in nature, i.e. asphaltenes, resins and naphthenic acids. These components may accumulate at the water-oil interface and hinder the droplets from re-forming a separate phase. Among these components, asphaltenes are believed to be the major material involved in emulsion stabilisation.

1 Chapter 1 - Introduction

Asphaltenes tend to adsorb at water-in-crude oil interfaces to form a rigid film surrounding the water droplet, thereby protecting the interfacial film from rupturing during droplet-droplet collisions [7-11]. Hence, the formation of extremely stable water- in-crude oil emulsions is facilitated. This results in a demand for expensive emulsion separation equipment such as separators, water treaters and coalescers. However, with reliable information concerning the crude oil and its emulsifying properties, steps can be taken in pre-treating the crude oils with destabilising chemicals, or by installing equipment specifically designed for each field, to avoid emulsion problems.

2 Chapter 2 - Theory

2 Theory

This chapter contains theory and references related to the topics discussed in this thesis. The theoretical consideration is primarily concerned with the formation, stabilisation, and destabilisation of crude oil emulsions, as well as the chemistry behind the natural surfactants responsible for stabilising the emulsions.

2.1 Crude Oil Composition

Crude oil is a complex mixture of hydrocarbons, with small amounts of sulphur, oxygen and nitrogen, as well as various metallic constituents, particularly vanadium, nickel, iron and copper [12]. A typical North Sea Crude Oil consists of 84.5 % carbon, 13 % hydrogen, 0.5 % nitrogen, 1.5 % sulphur and 0.5 % oxygen. The number of single components that exist in a crude oil is unknown. To determine the exact structure and composition of the various components is thus a daunting task, and the selection of fractionation procedure depends on the information desired.

The asphaltene content of petroleum is an important aspect of fluid processability. The SARA method, where the asphaltenes are separated as a group, is therefore often used to conveniently separate the crude oil into four major fractions: saturates (including waxes), aromatics, resins and asphaltenes (SARA), based on their solubility and polarity as shown in Figure 2-1.

Crude Oil I dilute with n-alkane ...... J

Maltenes precipitate

Adsorb on silica elute with

n-alkane toluene toluene/methanol ______I______Saturates Aromatics Resins Asphaltenes

Figure 2-1 Typical scheme for separating crude oil into saturate, aromatic, resin and asphaltene (SARA) components.

3 Chapter 2 - Theory

The basis for the method is that asphaltenes are removed by precipitation in a paraffinic solvent, and the deasphalted oil is separated into saturates, aromatics and resins by chromatographic fractionation [13-17], Of the four classes of compounds, only the saturates are easily distinguishable from the rest of the hydrocarbons in the mixture. The absence of it-bonds allows them to be readily differentiated from the aromatic components by virtue of the difference in their polarities. The remainder of the oil is composed of aromatics and heteroatomic compounds of varying degree of condensation, alkyl substitution and functionalism, which constitute a compositional continuum with respect to molecular weight and polarity [18]. The many variations in the recommended procedures, may all have some influence upon yield and chemical nature of the fractions. The properties of asphaltenes, for example, have shown to be affected by temperature, precipitating solvent, solvent-to-oil ratio and separation time [19].

2.2 Asphaltene Chemistry

The word asphaltene was coined in France by Boussingault [20] in 1837. Boussingault described the constituents of some bitumens found at that time in eastern France and in Peru. He named the fraction of distillation residue, which was insoluble in alcohol and soluble in essence of turpentine, "asphaltene", since it resembled the original asphalt. The strong interest in developing a better understanding of the solution behaviour of asphaltenes, has been motivated by their impact on production, transportation, refining and utilization of petroleum. The asphaltene fraction is composed of the heaviest and components in crude oils. Separated solid asphaltenes usually appears brown to black in colour and has no definite melting point but decomposes when the temperature exceeds 300-400 °C. It has been shown that changes in temperature [21, 22], pressure [1, 23- 25] and oil composition [26] can cause asphaltene precipitation.

Asphaltenes are operationally defined as the non-volatile and polar fraction of petroleum that is insoluble in n-alkanes (i.e. pentane or heptane). As a result, asphaltenes constitute a solubility class of crude oil components, rather than a chemical class. The molecular weight, polarity and aromaticity of precipitated asphaltenes generally increase with increasing carbon number of n-alkane precipitant. A schematic diagram representing the range of heavy compounds precipitated by mixing crude oil with n-pentane and n- heptane is shown in Figure 2-2.

4 Chapter 2 - Theory

n~C7 asphaltenes

Polarity & aromaticity

Figure 2-2 Hypothetical diagram representing the molecular characteristics of the asphaltenes precipitated from petroleum by n-alkane addition [27, 28].

A number of investigators have constructed model structures for asphaltenes, resins, and other heavy fractions based on physical and chemical methods. Physical methods include IR, NMR, ESR, mass spectrometry, X-ray, ultra-centrifugation, electron microscopy, small angle neutron scattering, small angle X-ray scattering, quasi-elastic light scattering spectroscopy, VPO, GPC, etc. Chemical methods involve oxidation, hydrogenation, etc.

While asphaltenes are recognised to be remarkably polydisperse in heteroatomic functionality, molecular weight, and carbon backbone structure, some common features have been established. Asphaltenes are characterized by fused ring aromaticity, small aliphatic side chains, and other elements including sulphur, oxygen, nitrogen, and metals such as vanadium and nickel. The heteroatoms accounts for a variety of polar groups such as aldehyde, carbonyl, carboxylic acid and amide, which are found in the asphaltene molecules. The aromatic carbon content of asphaltenes is typically in the range of 40 to 60 %, with a corresponding H/C atomic ratio of 1.0-1.2. A large percentage of these aromatic carbon rings are interconnected in the molecular structure and, consequently, the asphaltene molecule appears flat or planar. Figure 2-3 shows a suggested asphaltene structure. Yen and co-workers [29, 30] proposed a macrostructure model, where the asphaltenes was depicted as stacks of flat sheets of condensed aromatic systems, which was interconnected by sulphide, ether, of aliphatic chains. Espinat et al. [31] suggested the asphaltene molecules to be disc-like with polyaromatic fused ring cores containing polar functional groups. It is currently accepted that asphaltenes consist of aromatic compounds with n-n interactions, which undergo acid-base interactions and self associate through hydrogen bonding [32, 33].

5 Chapter 2 - Theory

Figure 2-3 Hypothetical molecular structure of the asphaltenes. By courtesy of the Statoil DART (Downhole Asphaltene Remediation Technology) program.

Several major problems associated with the recovery and refining of petroleum [34-39], are related to the aggregation and precipitation of asphaltenes. Investigations have shown that asphaltene particles may self-associate, and form aggregates in the presence of aromatic hydrocarbons [26]. The degree of association is largely dependent upon the aliphatic/aromatic ratio of the solvent. Due to the aggregation phenomena, measurements of the true molecular weight and the aggregate size are inherently difficult and have resulted in numerous research efforts. The size of the aggregate structure has been suggested to lie between 2 and 25 nm in diameter [40-42]. The molecular weight obtained have ranged from a few hundred to several million gmol" 1, however, the most recent values from several different types of asphaltenes tend to suggest values from 600-1500 gmol" 1 [43-49],

The aggregation is thought to occur through hydrogen bonding, however there is some inconsistency in the description of asphaltene self-association [30, 50-52], and both micelles and colloids are used in reference to asphaltenes. An asphaltene colloid is

defined as a submicron particle consisting of several asphaltene molecules bound by 71- bond interactions between polyaromatic clusters. Asphaltene micelles are considered analogous to a surfactant micelle, where the association of molecules is driven by hydrophobic-hydrophilic interactions. The terms "aggregate" and "micelle" are often used interchangeably in the literature. It has been shown that resins are essential in dissolving the asphaltenes in the crude oil. They are thought to attach to the asphaltene micelles/aggregates with their polar groups, and stretch their aliphatic groups outward to form a steric-stabilisation layer around asphaltenes [53, 54]. However, there still remains

6 Chapter 2 - Theory

the debate about whether the micelle in petroleum is homogeneous insofar as it is composed only from asphaltenes, or if both asphaltene and resin molecules constitute a mixed micelle [30, 54].

Resins are defined as the non-volatile and polar fraction of crude oil that is soluble in n- alkanes (i.e., pentane) and aromatic solvents (i.e., toluene) and insoluble in liquid propane. They are structurally similar to asphaltenes, on the other hand, molar mass is lower, hydrogen/carbon ratio higher, and the heteroatom content lower. Long et al. [28] showed that once resins were removed from the crude by adsorption chromatography, the remaining oil phase could no longer stabilise the asphaltenes.

Asphaltenes are also known to self-associate due to pressure depletion [1-3, 55]. At high pressures in the reservoir, the asphaltenes are dissolved in the monophasic crude oil. When the pressure is reduced the molar volume and the solubility parameter difference between asphaltenes and the crude oil increases towards a maximum at the bubble point of the crude oil. As a result of the reduced solvating power, the asphaltenes may start to precipitate at some onset pressure higher than the bubble point. Prior to the precipitation a stepwise association of the asphaltene molecules will take place. The final precipitation is due to a strong attraction between the colloidal particles and the formation of agglomerates. Once gas evolves, the light alkane fraction of the liquid phase is reduced, and thereby the solvating power for asphaltene molecules increases. The relative change in asphaltene solubility has been shown to be highest for light crude oils that are undersaturated with gas, and which usually contain only a small amount of asphaltenes. This gives the surprising result that light reservoir oils, which are low in asphaltenes are considered to be more likely to experience asphaltene related field problems than heavier, less undersaturated, asphaltenic oils.

A possible way of avoiding asphaltene precipitation is by adding chemicals that act in a way similar to resins by dispersing the asphaltenes in solution. Gonzales et al. [56] investigated the peptization of asphaltenes in aliphatic solvent by various oil-soluble amphiphiles including long-chain alkylbenzene, alkyl alcohol, alkylamine and p- alkylphenol. They found that the head group of the amphiphile influenced the effectiveness of the amphiphiles. Chang and Fogler [32, 33], using a series of alkylbenzene-derived amphiphiles as the asphaltene stabilisers, investigated the influence of the chemical structure on the asphaltene solubilisation and the strength of

7 Chapter 2 - Theory

the amphiphile-asphaltene interactions. The results showed that the polarity of the amphiphile head group and the length of the alkyl tail controlled the amphiphile effectiveness. Increasing the acidity of the amphiphile head group could promote the amphiphile ability to stabilise asphaltenes, probably through acid - base interactions between the asphaltene and the amphiphiles. Leon et al. [57] showed results from adsorption studies on asphaltene particles, where the adsorption isotherms of two amphiphiles (nonylphenol and nonylphenolic resin) were compared to a native resin. The adsorption isotherm for the natural resins was characterised by the continuous increase in the amount of adsorbed resins, and there was no indication of a plateau similar to the ones shown by the amphiphiles. This type of isotherm was explained by the penetration of substrate micropores by resin molecules, which lead to the partial breakdown of the asphaltene macrostructure.

In addition to the resins, other molecules in the petroleum mixture have also shown a tendency to stabilise the asphaltene particles/aggregates. Auflem et al. [58] showed that natural and synthetic naphthenic acids have a tendency to disperse the asphaltenes, and reduce the asphaltene particle size. This was proposed to occur through acid-base interactions between the naphthenic acids and asphaltenes, whereby the naphthenic acid would disperse the asphaltenes in solution in a similar way as the resins.

Naphthenic acids are classified as monobasic carboxylic acids of the general formula RCOOH, where R represents a cycloaliphatic structure. The classification contains a wide variety of structures with carbon number from C10 to C50, and from 0 to 6 saturated rings [59]. In crude oil production, the problems related to naphthenic acids arise from the processing conditions. As the pressure drops during production and carbon dioxide is lost from solution, the pH of the brine increases, which in turn leads to dissociation of the naphthenic acid (RCOOH -» RCOO" + H+). As a result, the following may occur: i) deposition of naphthenates [60] in oil/water separators, de-salters, tubing or pipelines following complexation of naphthenic acids with metal cations present in the aqueous phase and, ii) formation of stabilised emulsions due to naphthenic acids and naphthenates accumulating at the w/o interface [61] and thereby stabilising colloidal structures.

8 Chapter 2 - Theory

2.3 Emulsions and Emulsion Stability

Emulsions have long been of great practical interest due to their widespread occurrence in everyday life. They may be found in important areas such as food, cosmetics, pulp & paper, pharmaceutical and agricultural industry. Emulsions are also found in the petroleum industry, where they are typically undesirable and can result in high pumping costs, reduced throughput and special handling equipment. An emulsion is usually defined as a system consisting of a liquid dispersed in another immiscible liquid, as droplets of colloidal sizes (~ 0.1-10 urn) or larger. If the oil is the dispersed phase, the emulsion is termed oil-in-water (o/w) emulsion, conversely, if the aqueous medium is the dispersed phase, it is termed a water-in-oil (w/o) emulsion. This classification is not always appropriate and other types as, for instance, multiple emulsions of the type o/w/o, may also be found. In the emulsified state, the interfacial area between the dispersed droplets and the bulk phase represents an increase in the systems free energy. Consequently, the emulsions are not thermodynamically stable, and will seek to minimise the surface area by separating into the different phases. For an emulsion to separate, the droplets must merge with each other, or with the homophase continuum that gradually forms.

Processes that facilitate the separation are sedimentation/creaming, flocculation and coalescence [62-64], as shown in Figure 2-4. Creaming and sedimentation create a droplet concentration gradient due to a density difference between the two liquid phases, which result in a close packing of the droplets. Aggregation of droplets may be said to occur when they stay very close to one another for a far longer time than if there were no attractive forces acting between them. The size and shape of the individual droplets are for the most part retained. The mechanism of coalescence occurs in two stages; film drainage and film rupture. In order to have film drainage there must be a flow of fluid in the film, and a pressure gradient present. However, when the interfacial film between the droplets has thinned to below some critical thickness, it ruptures, and the capillary pressure difference causes the droplets to rapidly fuse into one droplet. Hence, the properties of the thin film are of uttermost importance for the separation. If the droplets deform, the area of the interface increases and consequently the drainage path in the film also increases, resulting in lower drainage rates.

Electrical double layer repulsion, or steric stabilisation by polymers and surfactants with protruding molecular chains, may prevent the droplets to come into contact with each

9 Chapter 2 - Theory

other. Also, polymers, surfactants or adsorbed particles can create a mechanically strong and elastic interfacial film that act as a barrier against aggregation and coalescence. A film of closed packed particles has considerable mechanical strength, and the most stable emulsions occur when the contact angle is close to 90°, so that the particles will collect at the interface. Particles, which are oil-wet, tend to stabilise w/o emulsions while those that are water-wet tend to stabilise o/w emulsions. In order to stabilise the emulsions the particles should be least one order of magnitude smaller in size than the emulsion droplets and in sufficiently high concentration.

Other factors that usually favour emulsion stability are low interfacial tension, high viscosity of the bulk phase and relatively small volumes of dispersed phase. A narrow droplet distribution of droplets with small sizes is also advantageous, since polydisperse dispersions will result in a growth of large droplets on the expense of smaller ones, an effect termed Ostwald ripening [65]. Special features of surfactant association into liquid crystalline phases with lamellar geometries that facilitates the stabilisation may also occur [66].

f z- \ / OO t * * Flocculation oo & & CO £& % o / \

Figure 2-4 Processes taking place in an emulsion leading to emulsion breakdown and separation.

2.4 Stabilisation of Water-in-Crude Oil Emulsions

The oil industry has an interest in crude oil emulsions for two main reasons: i) Water-in ­ crude oil emulsions can form in the processing of fluids from hydrocarbon reservoirs to

10 Chapter 2 - Theory

the refinery or in production facilities during extraction and cleaning. The emulsified water adds significant volume to the crude oil, causes corrosion in the pipelines and increases the cost of transportation and refining, ii) Water-in-crude oil emulsions can form in oceanic spills. These emulsions are very stable and the oil phase is difficult to recover, leading to great environmental damage. Due to their colour and semisolid consistency, they are often named chocolate mousse.

In order to devise optimum treatment for water-in-oil emulsions, it is vital to understand how they are stabilised. The predominant mechanism whereby petroleum emulsions are stabilised, is through the formation of a film with elastic or viscous properties. This film is thought to consist of a physical, cross-linked network of asphaltenic molecules, which aggregate through lateral intermolecular forces to form primary aggregates or micelles at the oil-water interface [8, 9, 67-73]. In addition, adsorption of solid particles from wax, clays, inorganic material or naphthenates may contribute to the film strength. Hence, the emulsion stability arises from a physical barrier that hinders the film to break when insufficient energies are involved in collisions between droplets.

Asphaltenes are thought to be peptised in the oil phase by the resinous components, and are hence prevented from precipitation. However, when water is introduced to the crude oil, the asphaltenic aggregates in the oil phase adsorbs to the new oil-water interface. The resins are likely shed and do not participate in the stabilising film [74], Figure 2-5. Eley et al. [75] showed that the stability of water-in-crude oil emulsions was related to the asphaltene precipitation point. The most stable emulsions occurred when the asphaltenes were on the verge of precipitation or above.

Kilpatrick et. al [74] have shown that the resins are unnecessary in the stabilisation of the asphaltenic film. The exact conformation in which asphaltenes organize at oil-water interfaces and the corresponding intermolecular interactions have yet to be agreed upon. The often suggested explanations are either H-bonding between acidic functional groups (such as carboxyl, pyrrolic and sulfoxide), electron donor-acceptor bonding between transition metal atoms and electron-rich polar functional groups, or some other type of force such as n-bonding between delocalised n electrons in fused aromatic rings. The relative strength and importance of each in forming the viscoelastic film and their consequent roles in stabilising water-in-oil emulsions have still not been fully explained.

11 Chapter 2 - Theory

Primary asphaltene-resin aggregates in oleic phase

Aayi lanci ic ayyi cym-ca adsorb to oil-water interface

Wau. ------Oil-Water Interface

Figure 2-5 Proposed stabilising mechanisms for asphaltenes in petroleum by resin molecules. Asphaltene aggregates shed solvating resins and adsorb to oil-water interface through polar interactions and hydrogen bonding [76].

2.5 Destabilisation of Crude Oil Emulsions

The destabilisation of crude oil emulsions forms an integral part of crude oil production. Stable emulsions are typically broken using gravity or centrifugal settling, application of high electric fields and addition of destabilising chemicals (demulsifiers). Other methods such as pH adjustment, filtration, membrane separation and heat treatment techniques, may also be used.

Gravity settling tanks, cyclones, centrifugal separators and other kinds of mechanical separation tools are typical equipment used in the destabilisation of crude oil emulsions. However, this hardware is of considerable volume as well as expensive to install on offshore platforms typical for North Sea conditions. It is therefore of great economical benefit whenever the installations can be kept at a minimum in size and number. Chemical destabilisation is therefore a very common method for destabilising emulsions. Also, the capital cost of implementing or changing a chemical emulsion-breaking program is relatively small and can be accomplished without a shutdown. The separation rate of a

12 Chapter 2 - Theory

w/o emulsion depends upon the matching of the demulsifier with the process residence time, the concentration and the stability of the emulsion, the temperature, the process vessel, the mixing energy and the type of stabilising mechanisms. Through building up more fundamental knowledge concerning the processes involved in stabilising and breaking the emulsions, the development and use of environmentally friendlier chemicals is facilitated. Also, the optimisation of type and amount of chemicals employed, contributes to reducing the oil content in the produced water offshore.

Commercial demulsifiers are typically mixtures of several components, which have various chemical structures and cover a wide molecular weight distribution. Some typical chemical structures used as demulsifiers are listed by Jones et al. [77] and Djuve et al. [78]. Each component of the demulsifier typically possesses a different partitioning ability and a different interfacial activity, and thus should provide a range of properties such as: i) Strong attraction to the oil/water interface, with the ability to destabilise the protective film around the droplet, ii) The ability to function as a wetting agent, changing the contact angel of solids, iii) The ability to act as flocculants and, iv) promotion of film drainage and thinning of the interdroplet lamella by inducing changes to the interfacial rheological properties such as decreased interfacial viscosity and increased compressibility [73, 79, 80]. Krawczyk [81] showed that demulsifiers with equal partitioning between the aqueous and oil phase, gave the best destabilising efficiency. This balance would lead to a maximum in the surface adsorption of demulsifier and a minimum in interfacial tension. However, partitioning would not be a dominant factor when other effects such as dissolution of the interfacial material or their flocculation by the demulsifier occur.

When two water droplets approach each other, the capillary pressure acting normal to the interface causes liquid to be squeezed out of the film into the bulk. This liquid flow results in a viscous drag on the surfactants in the sublayer, and the adsorbed emulsifier are carried away towards the film periphery, thereby creating a nonuniform concentration distribution. Demulsifier molecules may then occupy the empty spaces available for adsorption, and due to the high interfacial activity of the demulsifier, the interfacial tension gradient is reduced. This leads to a strong increase in the rate of film thinning, and ultimately, when the film thickness decreases below some critical value, the film ruptures and the droplets coalesce.

13 Chapter 2 - Theory

Strong attraction to the oil/water interface is often dependent on diffusibility and interfacial activity of the demulsifier. For fast diffusion to the interface, the molecular weight of the demulsifier becomes important. The demulsifiers relative solubility in oil is also important for mass transport to the interface, and where this is inadequate, carrier solvents (e.g. alcohols or benzene derivatives) are often used. At the interface, the demulsifier may influence the droplet interfacial film material by displacement, complexation, changing the solubility in the continuous phase, changing the viscosity of the interfacial film, or through quick diffusivity and adsorption, thus inhibiting the Gibbs- Marangoni effect, which counteracts film drainage.

In residual emulsions, the droplets are finely dispersed and widely distributed, and the flocculating ability of the demulsifier is required to gather up the droplets. Then, high molecular weights highly branched demulsifiers, with an affinity for the water droplet, are necessary. For emulsions with particle-stabilised films, demulsifiers, which act as wetting agents, may prove effective. The demulsifier may adsorb on to the solids, causing them to be more oil or water wettable, and thereby more easily transported into the continuous phase away from the interface. In some situations the demulsifiers have been used as inhibitors, i.e. injected before the emulsification process has taken place. This gives the demulsifier the chance to compete with the emulsifying agent in the process of covering the interface as the emulsifying process occurs, and thereby hinder the formation of a stabilising film. One should however, not forget to clarify the effect of concentration of the injected chemicals on the emulsion stability, as too much chemicals injected may result in an overtreat where the emulsion is actually stabilised, or a new emulsion type is created. Also, the injected demulsifiers should be checked to be compatible with other chemicals (corrosion inhibitors, scale inhibitors and flow enhancers) used in the stream as well as the components in the produced stream itself.

The effect of increased temperature is the sum of changes in several parameters. For instance, changes in the solubility of the crude oil surfactants or injected treating chemicals may occur as a result of increasing temperature. The density of the oil is reduced faster than the density of water as temperature increases, thereby accelerating the settling. Bulk viscosity of the crude oil decreases with increasing temperature, hence facilitating an increased collision frequency between water droplets, in addition to increasing the settling rate. Essential for the coalescence, especially in flocculated systems, is the influence of the interfacial viscosity. Depending on the type of interface the interfacial viscosity may decrease, increase or remain unchanged with increased

14 Chapter 2 - Theory

temperature [77]. With highly paraffinic crudes found in the North Sea, waxes are strongly correlated to the stability of emulsions. The wax may contribute to the stability through particle stabilisation, or from increasing the viscosity of the crude oil. Therefore, melting and crystallisation sequence of wax is of importance for the stabilising properties of these compounds [82]. High operational temperatures may however result in high losses of light end molecules, and consequently an increased potential for asphaltene deposition.

Electrical resolution of crude oil emulsions is possible since the systems are relatively non-conducting. In 1965 Waterman [83] summarised the main behaviours of a drop, or a pair of drops under an electric field. The mechanism promoting separation are the result of either forces between particles resulting from induced dipoles charges (dipole coalescence), or forces that result from interactions between unidirectional field and particles having a net charge (electrofining). The principle behind the electrically induced coalescence is often divided into: i) non interacting droplets approaching each other, ii) deformation of droplets and formation of plane-parallel films, and ill) thinning of the films to a critical thickness at which the film becomes unstable, ruptures and the two drops unify and form a single large droplet. Important features of a typical electrocoalescer are: The electric field (AC or DC), frequency, and set up for electrodes. The electrocoalescers in the oil and petroleum industry uses both AC and DC electric fields for the separation of water-in-oil emulsions [84]. One problem is that most of the equipment in the marked today is big and bulky, and it would therefore be of interest to develop small portable devices, incorporating features such as an optimum applied field strength combined with centrifugal force, to further enhance the separation.

15 Chapter 3 - Methodology and Theory

3 Methodology and Theory

In the following section, a summary of the various methods utilised in this thesis, is presented. A high-pressure separation rig is used in paper I-III, the quartz crystal microbalance technique is found in paper IV. In Paper V and VI, respectively, near infrared spectroscopy and nuclear magnetic resonance is employed.

3.1 High-pressure High Temperature Separation Rig (HPHT-rig)

In order to study separation of emulsions under realistic conditions, a high-pressure separation rig has been constructed at Statoil's R&D Centre. The rig can be used to prepare emulsions and monitor the separation of oil and water, as well as any stable foam formation, in a vertical batch separation cell. The separator cell is made from sapphire assuring full visibility of the separation processes, it has a volume of 0.5 litres and tolerates pressures up to 200 bar. A schematic drawing of the rig is presented in Figure 3.1. The rig includes four 600 ml high-pressure sample cylinders. With the aid of four motor driven high capacity piston pumps, water and oil are pumped from the sample cylinders, through the choke valves and into the separator. The four pumps can be controlled independently, however, the total flow rate is usually kept constant. Pressure drop through the choke valves are back-pressure controlled. In order to control the separation pressure, the separation cell is pressurised with gas, inert or natural gas, and the pressure is regulated by another back-pressure controlled valve. To ensure temperature control of the system, a thermostated cabinet encloses the separation cell and provides temperatures in the range of -7 °C to 175 °C.

The principle of the rig is that flows of two pressurized fluids meet and stream through a choke valve (VD2 or VD3). The streams from VD2 and VD3 meet in a third choke valve, VD1, before entering a vertical batch separator. As the fluid mixture passes through the choke valves it undergoes pressure drops, which provides the shear force necessary to create more interface between the oil and water phases, thereby forming water droplets in the oil phase.

16 Chapter 3 - Methodology and Theory

Figure 3.1 Schematic view of the high-pressure separation rig.

If the pressure drops below the bubble point pressure of the oil, a gaseous phase appears. The gas evolved may form a foam layer, as well as influence the settling and coalescence of water droplets. The quantity of the different phases, foam, oil, emulsion layer and water, can thereafter be recorded as a function of time. To aid in the monitoring process, video cameras have been connect to the rig. Samples of oil, water and emulsion layer may be sampled by connecting a pipeline to the bottom of the separator cell. To study the effect of emulsion and foam inhibitors or demulsifiers, the rig is equipped with two independent high precision pumps (5 and 6), which deliver volumes down to 0.03 mlh"1. Low concentration chemicals can thus be injected to any of the flow lines. Injections can also be made in the bottom of the cell, where a stirrer can be used to distribute the chemicals.

The high-pressure separation rig is a batch separator, and the results will therefore not apply exactly for a field separation process. Nevertheless, the results will show trends for temperature, pressure, pressure-drop, mixing with other oils, etc. It will also indicate whether there is a need for chemical treatments (demulsifier, foam inhibitor etc). Comparisons of field tests and laboratory studies of the separation of oils have shown that the high-pressure separation rig give the same ranking of the oils, the same ranking of the demulsifiers efficiency and the same optimum demulsifier concentration as in the field tests. It should also be noted that bottle tests did not give the same ranking of the chemicals as the field tests and the separation rig experiments.

17 Chapter 3 - Methodology and Theory

3.2 Quartz Crystal Microbalance (QCM)

The first thorough investigation of the piezoelectric effect is often attributed to Jaques and Pierre Curie, as early as 1880 [85]. However, it was not until 1917 when Langevin [86] showed that quartz crystals could be used as transducers and receivers of ultrasound in water, that a more detailed study of piezoelectricity started. In 1959, Sauerbrey [87] published a Paper showing that the frequency shift of the quarts crystal was proportional to the added mass. This signified the birth of a new quantitative method for measuring very small masses, i.e. the quartz crystal microbalance. Another important step was a Paper by Nomura and Okuhara [88], where QCM was proven reliable for measurements in the liquid phase. Preferably, one side of the sensor should be exposed to the liquid, and the other to the gas phase. At present, the QCM technique is in rapid expansion, and has found a wide range of applications in areas such as food, environmental and clinical analyses [89].

The conventional quartz crystal microbalance, QCM, consists of a thin disk of piezoelectric quartz crystal, which can be used to measure very small masses. The crystal is sandwiched between a pair of electrodes, which are hooked up to an electronic oscillator. When an AC voltage is applied over the electrodes, the crystal can be made to oscillate at its resonance frequency, f, via the piezoelectric effect. However, the oscillatory motion is damped due to i) energy losses in the crystal, ii) energy losses due to deposited material on the sensor, and iii) energy losses to the surrounding medium. The magnitude of these losses can be measured by suddenly switching off the driving field to the sensor crystal and monitor the oscillation which will rapidly decay in amplitude in the form of a damped sinusoidal wave, characterised by the frequency, f, and the time constant, t, for the damping. The latter factor is inversely proportional to the sum of dissipative mechanisms, termed the dissipative factor, D.

The QCM-D™ technique is based on simultaneously measurements of both f and D. Changes in the conditions of the sensor crystal due to adsorption on the crystal surface induces a corresponding change in both frequency and dissipation factor. By continuous measurements of Af and AD during adsorption, information is obtained about the adsorption kinetics and the amount of adsorbed matter, as well as viscoelastic properties of the overlayer. The adsorption of matter onto the crystal is treated as an equivalent mass change of the crystal itself, and the increase in mass, Am, induces a proportional shift in frequency, Af, which was demonstrated by Sauerbrey [87] in 1959:

18 Chapter 3 - Methodology and Theory

oy (3-1) f0n 2/0 n n where pq and vq are the specific density and the shear wave velocity in quartz respectively, tq is the thickness of the quartz crystal, and fo the fundamental resonance frequency (when n=l). However, for this relation to be considered valid, the following conditions must be fulfilled: i) the adsorbed mass is distributed evenly over the crystal. ii) Am is much smaller than the mass of the crystal itself, and iii) the adsorbed mass is rigidly attached, with no slip or inelastic deformation in the added mass due to the oscillatory motion.

The shift in dissipation factor in a liquid environment may be calculated from Stockbridge relation [90]

AD = (3-2)

where 7, and pi are the viscosity and density of the fluid, respectively tq and pq are the thickness and the density of the quartz plate.

Figure 3.2 Schematic views of the quartz crystal microbalance cell and quartz crystal.

A schematic drawing of the measuring cell is shown in Figure 3.2. The key components are i) the sensor crystal mounted in a measurement chamber with facilities for batch of flow measurements in liquid or gas, ii) the drive electronics (relay and signal generator), and iii) the recording electronics (probe, reference frequency, filter) including data-

19 Chapter 3 - Methodology and Theory

handling and software (analogue-to-digital converter and computer). The apparatus used in this study is a QCM-D™ device from Q-Sense, Gothenburg, Sweden. AT-cut quartz crystal oscillators were used, with approximately 100 nm thick gold electrodes evaporated onto the crystal surface. To minimise the temperature flux from the system, the surrounding room was temperature controlled, and all solutions were stored in the same room. The temperature variations in the room were monitored to be in the range of ± 0.5°C, and the temperature in the chamber was assumed to be the same.

3.3 Near Infrared Spectroscopy (NIR)

William Herschel is credited as the father of near-IR techniques, for his discovery of the near infrared region as early as 1800. Today, over 200 years later, near infrared spectroscopy is one of the fastest growing analytical techniques, particularly in the food and agricultural industries [91, 92]. With recent advances in instrumentation and multivariate data analysis, the technique has also awoken the attention of the pharmaceutical industry. The reason for this massive interest is probably a direct result of its advantages as an analytical tool for quality control. The molar absorptivity of NIR bands permits operations in the reflectance mode, and hence the measurements can be made directly on the material itself. The measurements are thus rapid and non-invasive [93], and there is usually no need for extensive sample preparation. Also, the NIR spectra contain information on both chemical composition and physical properties of the sample [94]. This permits not only the identification of compounds, but also total characterisation of samples and determination of non-chemical parameters.

The near infrared region is found between the visible and middle infrared regions (MIR) of the electromagnetic spectrum. According to the American Society for Testing and Materials (ASTM), it is defined as the spectral region spanning 780 - 2526 nm (12820 - 3959 cm"1). Light absorption in this region is primarily due to overtones and combinations of fundamental vibration bands occurring in the MIR region. This makes NIR an excellent choice for hydrocarbon analysis, where functional groups such as methylenic, olefinic and aromatic C-H give rise to various C-H stretching vibrations that are mainly independent of the rest of the molecule.

In addition to molecular absorption, the NIR spectra are dependent upon several physical parameters, where the most prominent is scattering from particles. As the particle size

20 Chapter 3 - Methodology and Theory

changes it causes a change in the amount of radiation scattered by the sample [95], and this is reflected in the NIR spectra as a shift of the baseline. A typical representation of the baseline shift in a system as a consequence of change in particle sizes are shown in Figure 3.3.

VlQO 1200 1300 1400 1S00 1600 1700 1800 1900 2000 2100 2200 Wavelength [nm]

Figure 3.3 Optical density, a sum of scattering and absorption of transmitted light, plotted against wavelength for several NIR spectra. The system consists of asphaltene particles in model oil (heptane/toluene 70/30 vol. %) and is measure at defined time intervals after injection of a chemical. The lowering of the baseline is a measure of decreased scattering as the chemical disperses the particles.

For slightly lossy dielectric spheres in the Rayleigh limit (rA s 0.05), the scattering and absorption processes contribute separately to the extinction coefficient [96, 97]. That is

= er + £7, (3-3)

where a to t, a sc and o abs are the total, scattering and absorption cross-sections, respectively. The ratio of scattering to absorption scales with r3, indicating the importance of particle size on the total light extinction. The relation between optical

21 Chapter 3 - Methodology and Theory

density (OD), light intensity (I), particle diameter (N) and particle cross section (o t01) is given as

(3-4) where I0 and I are the intensities of incident and transmitted light, respectively. The effect of multiple scattering is not accounted for in this equation. Details on light scattering in the near infrared region can be found in the literature [98-100].

The NIR-measurements were performed with a Brimrose AOTF Luminar 2000 spectrometer, equipped with a fibre optic sampling probe for transflectance measurements (see Figure 3.4). In the study of various chemicals influence on asphaltene aggregates sizes, the optical density at 1600 nm wavelength was utilised. In this region the hydrocarbon absorption is minimal, and it is the near-infrared region with the least noise in the measurements.

Figure 3.4 The near infrared spectrometer setup.

3.4 Nuclear Magnetic Resonance (NMR)

The property of Nuclear Magnetic Resonance (NMR) was first described by Purcell [101] and Bloch [102] in 1946, work for which they received the Nobel Prize in 1952. Since then NMR has become a powerful tool in the analysis of chemical composition and structure [103, 104]. The NMR experiments are performed by immersing atoms in a static magnetic field (B0), which polarises the sample such that it has a bulk magnetisation aligned with the direction of the field. In order for this to occur the nuclei

22 Chapter 3 - Methodology and Theory

must possess a non-zero spin (e.g., lH, 2H, 13C, 19F and 31P). An oscillating magnetic field, in form of a radio frequency (r.f.) pulse, is then applied for a short time orthogonal to S0 and causes the longitudinal magnetisation to be tipped into the transverse plane. The absorption or emission of electromagnetic radiation by the nuclear spins causes transitions between the two energy states, spin-up and spin-down. The specific frequency at which a given type of nuclei absorbs is given by the Larmor equation:

co0 = y ■ Bo (3-5)

where co0 is the Larmor angular frequency, y the gyromagnetic ratio of the nuclei and B0 the strength of the magnetic field. Since the application of a resonant r.f. pulse disturbs the spin system, there must subsequently be a process of returning to equilibrium. This involves exchange of energy between the spin system and its surroundings. Such a process is called spin-lattice relaxation, and the rate at which equilibrium is restored is characterised by the spin-lattice or longitudinal relaxation time, 7). The spins do not only exchange energy with the surrounding lattice, but also among themselves. This is generally a faster process than spin-lattice relaxation, and is characterised by the spin- spin relaxation time, T2. The relaxation processes induce a voltage that can be detected by a suitably tuned coil of wire, amplified, and the signal displayed as the free induction decay (FID). This gives rise to characteristic spectra, which are functions of several factors i) the type of nucleus, ii) the chemical environment of the nucleus, and iii) on the spatial location in the magnetic field if that field is not uniform everywhere.

Pulsed Field Gradient Spin Echo NMR (PFG-SE NMR)

The principle of measuring molecular self-diffusion by NMR is based on the possibility to label molecules according to their position in the sample by applying a magnetic field gradient with position-dependent strength. The applied field will change the refocusing in the spin-echo experiments, which will lead to a reduction in signal intensity if the labelled molecules have diffused to new positions during the experiment. Self-diffusion measurements by NMR have been utilised in numerous studies ever since the discovery of spin echo by Hahn in 1950 [105]. Several new effects on spin echoes were presented, one of which was the diffusional effect on echo amplitudes in an inhomogeneous magnetic field. Carr and Purcell [106] provided a more precise theoretical description four years later, where they also modified the experiment by employing different magnetic field gradients. This made it possible to measure diffusion.

23 Chapter 3 - Methodology and Theory

The spin echo method was significantly improved in the mid sixties with the pulsed field gradient spin-echo (PFG-SE) technique. McCall et al. [107] are usually credited for the basic idea published in 1963, while the methodology and theory were presented later on by Stejskal and Tanner [108]. Several modifications to the technique have been made, and presently, PFG-SE NMR has evolved into a very useful approach in the studies of surface and colloid chemistry. The method is non-evasive, relatively fast, and measures the true molecular self-diffusion coefficients. It provides component resolved information concerning structural changes, bindings and associated phenomena, as well as sizes and shapes, from complex mixtures. Hence, the PFG-SE NMR technique offers an alternative way of obtaining information from, for instance, hydrocarbon mixtures, where typical light transmission techniques are difficult to use due to the opaqueness.

The Basic 90 °-180° Experiment

In its simplest form the PFG-SE NMR method consists of two radio frequency pulse spin- echo experiments, with identical magnetic field gradient pulses of magnitude G and duration S and time delay a applied, respectively. An initial 90° r.f. pulse produces an oscillating field Bx perpendicular to B0, while the gradient pulse causes a rapid precessional phase shift depending on the position of each nucleus in the sample. After a time t after the 90° pulse, a 180° pulse is applied, which inverts the phase shift. The succeeding gradient pulse produces phase compensation, i.e. refocuses the spins. If nuclei have changed position during A due to diffusion, the refocusing will be incomplete and consequently the attenuation of the spin echo will decrease. Spins having completed a change of location, due to Brownian motion during the time period A between both gradient pulses, will however experience different phase shifts by the two gradient pulses. As a consequence they are incompletely refocussed and lead to echo decay.

The Stimulated Echo Method (90° - 90° - 90°)

Diffusion experiments are usually facilitated by long spin-lattice (T2) relaxation times and high gyromagnetic ratios. However, for slow motional processes, chemical exchange or spin relaxation effects, T2 may in some cases be much smaller than Tlt e.g. for large and/or rigid molecules. When T? is small, parts of the signal may be lost due to natural

7> relaxation during A. By minimising the time period in which the spins are projected onto the xy plane, this effect may be limited. This is usually achieved by the stimulated echo pulse sequence. The experiment utilizes three 90° pulses, where the first pulse rotates the magnetisation into the xy-plane, after which the spins in various volume

24 Chapter 3 - Methodology and Theory

elements lose coherence, and acquire various angles in the rotating frame. The second pulse stores the memory of the current phase angles in the z-direction, where they are unaffected by field gradients and relax in the longitudinal direction. The third pulse restores the phase angles with reversed signs, so that they now precess to form an echo.

The PFG-SE NMR experiments were performed on a Unity Inova 500 MHz spectrometer and an Oxford magnet equipped with a diffusion probe from DOTY Sci. Inc., USA. The pulse sequence used for the diffusion measurements was a stimulated echo where the gradient pulse duration (6) and the experimental observation time (a ) were kept constant at 4 and 70 ms, respectively. A sine-shaped gradient was used to minimise the effect of eddy-currents. The gradient strength (g) was varied in 41 or, in the case when naphthenic acid had been added, 51 linear steps from 0 to a maximum value chosen so as to obtain a hundredfold decrease of the signal attenuation.

25 Chapter 4 - Main Results

4 Main Results

This chapter presents a summary of results from the Papers included in this thesis. In Paper I, a new mechanism involved in the breaking of crude oil emulsions is proposed. Further investigations of the influence of gas bubbles on the stability of crude oil emulsions, have resulted in Paper II. This describes a patent on a new method for breaking of particle stabilised crude oil emulsions by injection of polar gases. The objective of Paper III is whether or not a "dead" crude oil sample, by recombination with a gas phase, can recover the emulsion separation properties of the original "live" crude oil. In the first three Papers the experiments are conducted in a high-pressure separation rig. In Paper IV the focus is slightly shifted to the surface-active agents involved in stabilising the water-in-oil emulsions, and the interactions between asphaltenes and resins are studied using a quartz crystal microbalance. The Paper deals with the adsorption of these indigenous surfactants on hydrophilic surfaces, individually, through co-adsorption, or in the form of competing adsorption. Paper V further explores the interactions between asphaltenes and various surfactants, including synthetic and natural naphthenic acids, and how they influence the asphaltene aggregate sizes. These studies are done by near infrared spectroscopy. In the final Paper, Paper VI, the interactions between asphaltenes and naphthenic acids are studied by nuclear magnetic resonance and near infrared spectroscopy, and information about size and shape of the asphaltene aggregates is obtained. Included in this thesis is also the chapter "Demulsifiers in Oil Industry" from "Encyclopedic Handbook of Emulsion Technology.

4.1 Paper I

One of the largest problems in oil production is the formation of emulsions stabilised by heavy crude oil components like asphaltenes, resins and waxes. Such problems may in some cases be solved by means of injections of chemicals or introduction of mechanical separation facilities. However, the costs of these solutions are normally high and the search for new and efficient separation tools is important.

The objective of the first Paper was to investigate the effects of separation pressure, pressure drop and solvency on the stability of crude oil emulsions. A North Sea crude oil

26 Chapter 4 - Main Results

was recombined with dry natural gas to a separator pressure of 11 bar. Thereafter the sample was mechanically pressurised further to 100 bar, or in some cases 182 bar, by a piston pump connected to the sample cylinder. In some of the experiments the crude oil was diluted with various amounts of toluene to modify the aromaticity of the oil phase. The experiments were performed in a high-pressure separation rig, which is further described in chapter 3.1. In the rig, the fluids are mixed by a pressure drop through choke valves into a vertical batch separator cell, where the separation of the different phases are monitored visually.

In the experiments several effects on the separation were observed: i) increased separation with increasing pressure drop below the separation pressure. This was argued as owing to gas bubbles that propagate through the emulsion, tearing away stabilising material from the water/oil interface, ii) Increased pressure drop gave more stable emulsions for separation above the bubble pressure, and iii) toluene dilution of the crude oil resulted in less stable emulsions. A higher energy input, due to increased pressure drop, obviously resulted in smaller water droplets and consequently a slower separation process. The relation between energy input and droplet size has been shown before by several authors [109-113]. Also the destabilising effect from diluting the crude oil with toluene, was as expected. McLean and Kilpatrick [70] showed that as the aromaticity of the oil phase increased, the asphaltene aggregates were dissolved and the stability of the emulsions was reduced. Further, the foamability was also affected by the toluene addition. For increasing content of toluene in the oil phase, the capacity of the system to form foam decreased as a result of dissolution of stabilising material.

More interesting was the comparison of experiments performed on a recombined oil phase, with a recombined oil phase that had been degassed (I.e. the recombined oil phase was depressurised to atmospheric pressure, while allowing the gas phase to evaporate). The degassed sample was then repressurised mechanically by use of a piston pump to the original pressure (100 bar). The two types of oil samples were put through an identical emulsification procedure. The oil was mixed with pressurised formation water, 35 volume %, by pressure drops through two succeeding choke valves: From 100 to 11 bar and from 11 to the separation pressure in the separator (7 or 1 bar). For the recombined samples there were a significant foam formation and relatively fast separation for both separation at 7 and 1 bar. It was interesting to notice that the experiment with the largest pressure drop over the second choke valve, separated

27 Chapter 4 - Main Results

fastest. For the degassed samples there were no foam formation, and both the separation at 7 and 1 bar were equally slow, as shown in Figure 4-1.

11->1 bar (degassed)

1l->7 bar (degassed)

11->1 bar

Time [min]

Figure 4-1 Resolution of water vs. time, for water-in-oil emulsions made from recombined samples and degassed recombined samples at 7 and 1 bar separation pressure.

These results were accounted for by the flotation effect of gas bubbles on the stabilising material. As the oil phase was depressurised, the solubility of light end molecules decreased, and a gas phase evolved. The gas phase would then rise through the solution in the form of bubbles, which could rip off surface-active materials from the water-oil interface. Paper II deals with this mechanism in further detail.

4.2 Paper II

In order to investigate the influence of flotation upon separation of particle stabilised water-in-oil emulsions, a series of experiments on different crude oil and model oil systems were performed. The results led to the development of a patent for the use of a polar gas as a separation promoter for breaking water-in-oil emulsions. The basic idea is that the gas phase should be mixed with the water phase at an early stage in the separation process. In this way the emulsification takes place with an aqueous phase enriched with dissolved gas. When lowering the pressure in a separator tank, there will be a release of gas in the form of bubbles, which can enhance the breaking of oil- continuous emulsions. In the experiments described in this Paper, C02 was used as the gas phase.

28 Chapter 4 - Main Results

Results from two different North Sea crudes, termed A and B, are shown. Both of these have been known to give stable water in crude oil emulsions although the stabilising mechanisms can be different. Crude A is a heavy crude with a high content of asphaltenes, while crude Bis a acidic crude with a high amount of naphthenic acids. In addition a model system consisting of crude A (1% of A in Exxsol D-80) was tested. Essential for the discussion is that these samples were run through pressure reductions where the initial pressure (100 bar) was reduced to the separator pressure 65 bar. A schematic drawing of the experimental setup is shown in Figure 4-2, and a more extensive overview of the high-pressure high temperature rig is given in chapter 3.1.

P13: P12: PU: P10: 100 bar 85 or 70 bar 65 bar 100 bar

Water Water Model oil Empty with CO or crude

Figure 4-2 Experimental setup. The separator pressure was reduced to 1 bar after 5 minutes.

The emulsions were kept in the vertical separator for 5 minutes, before the final pressure was adjusted as a gas release from 65 bar to 1 bar. During this period of time the emulsion undergoes a settling process. The effect of propagating gas bubbles should be increased if the major part of the water droplets is assembled in a dense packed region. Figure 4-3 and 4-5 show the separation of water as a function of time for emulsions made up from crude oil A.

29 Chapter 4 - Main Results

»-oq, AP20 3— Water AP 20 t-cq, a?; APS

Figure 4-3 Resolution of water vs. time for the dead crude oil A system. The pressure in the separation cell was reduced from 65 to 1 bar after 5 minutes.

The dispersed aqueous phase was either pure water or water saturated with C02. For crude oil A, Figure 4-3 reveals the effect of the pressure gradient over the choke and the addition of C02. The effect of C02 seemed to increase with increased pressure gradient (5 or 20 bar). However, in most of the cases the separation was accelerated by the release of C02 after 5 minutes. With a 5 bar pressure gradient there was no significant difference to the samples containing only pure water. However, the large effect was seen for the emulsion with a AP = 20 bar and an aqueous phase saturated with C02. For the first 5 minutes, the separator pressure was kept at 65 bar, and the level of separation was low or almost negligible. Then, as the pressure reduction took place, between 50 and 60 % of the water phase would separate within 1-2 minutes. After 15 minutes 90 % of the water had separated. This was a significant result for a crude oil, which has proven to give very stable emulsions that are resistant to both chemical and mechanical treatment.

30 Chapter 4 - Main Results

100 i

Water AP 5 Water A P 20 CO:, AP 5 CO, AP20

Figure 4-4 Resolution of water vs. time for crude oil B. The pressure in the separation cell was reduced from 65 to 1 bar after 5 minutes.

Figure 4-4 shows the separation sequence for emulsions based on crude oil B. Characteristic for this system was that some separation, approximately 10 %, would take place already at 65 bar. However, when the gas was released after 5 minutes the separation profile changed dramatically. All curves, independent of pressure drops over the chokes and content of C02 in the water phase, showed a faster resolution of water. Hence, the selectivity between the different emulsions was lost. Large effects were seen both with and without C02 in the aqueous phase, and with small and larger pressure gradients over the choke. In these cases, one could not with certainty relate the increased separability to carbon dioxide release. The model system, where 1 % of crude oil A was diluted into a paraffinic fluid (Exxsol-D 80) and combined with 40 % water with and without C02, is presented in Figure 4-5. The separation level of the model emulsions was much lower, but also in this case an acceleration of the gas release upon the separation of water, was clearly seen. Based on the results obtained in the experiments, two processes are thought to commence upon pressure reduction in a separator tank.

31 Chapter 4 - Main Results

o $ 10 15 20 25 Time [mm]

Figure 4-5 Resolution of water vs. time for a model system consisting of 1 % crude oil A in Exxsol D-80. The pressure in the separation cell was reduced from 65 to 1 bar after 5 minutes.

Proposed mechanisms for breakingof oil-continuous emulsions:

The droplet rupturing effect:

The C02 dissolved into the aqueous phase (the droplets) will rapidly form small bubbles upon a pressure reduction. Due to the gravity difference these bubbles propagates through the emulsified system. When the bubbles leave the water droplet they have to pass an interface built up by indigenous polar surfactants (asphaltenes and resins). As a consequence the interfacial film will be ruptured. If the C02 bubbles carry with them surface active material from the interface (flotation effect) the time for the interface to reform will, most likely, be longer than the coalescence time. Hence the system will break and water and oil phases should appear. Application pressures could be about 60 bar depending on the chemical system and the whole process design.

The film drainage effect:

The C02 dissolved in the oil phase (the continuous phase) will also rapidly coalesce and form bubbles upon a pressure reduction. The buoyancy forces cause the bubbles to propagate through the emulsified system. In doing so they may tear off surface-active material from the o/w interface described as a flotation effect. This effect should not be

32 Chapter 4 - Main Results

specific for C02 but expected to be common for all oil soluble gases below the bubble point.

Bubbles inside droplets Bubbles rupture interface oo=> Pressure <=> CO drop Coalescence

oo Flotation Film drainage

Figure 4-6 Illustration of the proposed effects from C02 bubbles on water droplets in an oil- continuous phase: i) The droplet rupturing effect, ii) The film drainage effect.

It was experimentally shown that a polar gas, such as C02, could accelerate the breaking of crude oil based emulsions. However, this was not possible for all types of crude oil emulsions. Presumably it is feasible only for those types of crude oil emulsions, which are particle stabilised. Also, the use of C02 will be effective in a gravitation separator and most effective in a separator of batch type. The emulsion will be held a few minutes in the separator to settle before the gas pressure in the separator is reduced. However, in a continuous process, the effect will be much less.

It is well known that C02 forms gas hydrates at low temperatures and high pressures. Therefore, it should be pointed out that separation and injection conditions should be far from the thermodynamic conditions for gas hydrate formation. Dissolved C02 may also constitute a danger for corrosion and low pH's. These conditions must be taken into account in designing a future process and in the choice of the materials.

4.3 Paper III

The emulsion stability for a "live" crude oil was compared to the emulsions stability of the same crude oil recombined with, l\l2, C02, CH4r C2H6 or a natural gas mixture. Emulsion stability experiments, where varying amounts of the lighter molecules in the "live" crude oil had been removed, were also performed. The experiments were thus comprised from

33 Chapter 4 - Main Results

the following three types of oil samples: I) "Live" crude oil samples with a bubble point of 15 bar, II) Samples where the gas phase had been removed from "live" samples by depressurising to 1 bar, while allowing the gas to evaporate (degassing), and thereafter repressurised to 15 bar with either N2, C02, CH4, C2H6 or the natural gas mixture, iii) Samples that were degassed in the recombination cell to 10 or 1 bar, respectively, and thereafter mechanically repressurised to 15 bar by use of a piston pump and no addition of gas.

The oil samples were transferred into the sample cylinder on the high-pressure high temperature rig, described in chapter 3.1, and further pressurised to 20 bar by use of a piston pump. To create emulsions, the oil samples were streamed together with synthetic formation water through a choke valve, while varying the pressure drop and separation pressure. The decomposition of the resulting emulsion and foam layer could thereafter be monitored visually in the vertical high-pressure separation cell. As a result, the separation properties of the water-in-crude oil emulsions from the different recombined samples and the "live" crude oil could be compared.

The results from experiments performed at oil type i) and ii) showed the following trends: Water-in-oil emulsions produced from "live" North Sea crude oil, generally separated faster and more complete than emulsions based on recombined samples of the same crude oil. An example of such a water resolution chart is shown in Figure 4-7. Increased water content or smaller pressure drop into the separator, resulted in faster and more complete separation of the emulsions for both "live" and recombined samples. It was also noted that the height of the foam layer increased when reducing the water content from 60 to 40 volume %.

34 Chapter 4 Main Results

Methane

Figure 4-7 Resolution of water vs. time for "live" and recombined oil samples. Separation pressure, pressure drop and water content were 15 bar, 5 bar and 60 volume %, respectively.

Experiments performed on oil type 3, i.e. oil samples with varying content of light molecules, can be concluded as follows: As expected the "live" (15 bar) sample gave the highest amount of foam for water content of both 60 and 40 volume %. A smaller amount was obtained for the 10 bar sample and none for the sample degassed to 1 bar. The emulsion stability for the mechanically recombined crude oil samples seemed to depend on the degassing pressure of the "live" sample, i.e. the content of gas present in the oil phase. Samples with the lowest gas content gave, probably as a result of higher viscosity, a less complete emulsification. This would in turn create a higher number of relatively large droplets, which separated within the first few minutes, while the rest emulsion maintained the same stability as for the samples with higher amount of gas remaining. As for the other oil samples, the amount of water influenced the emulsion stability together with the pressure gradient over the choke. Smaller water content (40 %) and large AP over the choke (19 bar) gave higher emulsion stability in comparison with 60 % of water and 5 bar pressure drop. Most likely the drop size distribution was quite different for these samples with much smaller droplets for 40 volume % of water and high AP.

35 Chapter 4 - Main Results

4.4 Paper IV

The adsorption of stabilising material onto hydrophilic surfaces was investigated in Paper IV. To do this, the adsorption of asphaltenes and resins onto a hydrophilic gold surface, was measured as a function of bulk concentration. The measurements were performed by a quarts crystal microbalance with dissipation measurements (QCM-D™), which is described in chapter 3.2. This device allows for simultaneous measurements of frequency, f, and energy dissipation factor, D. The change in frequency is related to the mass adsorbed onto the surface of the sensor crystal, and from the change in dissipation factor, information about the interfacial processes can be resolved.

2.5 -

1.5 -

Concentration (ppm)

Figure 4-8. Adsorption isotherms for resin adsorption onto a hydrophilic gold surface as a function of resin concentration in pure n-heptane.

The results showed that the resins in pure heptane adsorb onto a gold surface, and pack into a compact monolayer (Figure 4-8). However, the resins showed no tendency to aggregate on the surface. With increasing amount of aromaticity in the solvent, the adsorbed quantity decreased, and was practically zero in pure toluene. This was related to an increased solvency of the resins. The asphaltenes in heptane/toluene mixtures, or pure toluene, adsorbed to a larger extent (Figure 4-9). The adsorption was higher than observed for typical non-associating polymers, which indicated adsorption of aggregates. At lower concentrations the asphaltenes formed a rigid layer. When higher concentrations were injected it was possible to obtain further adsorption, which was related to the strong tendency of aggregation of asphaltenes in bulk solution. Supposing this multilayer adsorption also occurs onto water droplets dispersed in oil, it may result in a thick barrier that stabilise the droplets from coalescence.

36 Chapter 4 - Main Results

Desorption studies showed that resins were not able to desorb pre-adsorbed asphaltenes from the surface. Neither did they adsorb onto the asphaltene-coated surface. On the other hand, resins and asphaltenes associated in bulk liquid, and the adsorption from mixtures containing both resins and asphaltenes was markedly different to that observed for the pure components. It was therefore concluded that preformed resin/asphaltene- aggregates adsorb to the surface.

Concentration (ppm)

Figure 4-9 Adsorption isotherm for asphaltene adsorption onto a hydrophilic gold surface as a function of asphaltene concentration in pure toluene.

The irreversibly adsorbed amount for a crude oil solution was smaller than for the asphaltene and resin mixture but quite similar to that of the separate fractions. When effects from other constituents like paraffin and wax were absent, the resin and asphaltene molecules arranged in a different way in the adsorbed layer. When paraffin and wax was present they could be incorporated in the adsorbed layer, or affected the interaction forces in the bulk of the crude.

4.5 Paper V

Resins are usually thought to function as a dispersant of asphaltenes in crude oil. In order to hinder asphaltene deposition, the petroleum industry injects large volumes of chemicals into reservoirs and pipelines. These chemicals are supposed to imitate the

37 Chapter 4 - Main Results

resin function, by dispersing the asphaltenes in the hydrocarbon mixture. The size of the asphaltene aggregates also influence the capacity to form emulsions, where the optimum size for stabilising, depends upon the size of the water drops. That is, changing the size of the asphaltene aggregates beyond the optimum size region, may also prevent emulsion formation. In Paper V, near infrared (NIR) spectroscopy was introduced as a potent tool for studying the effect of chemicals in dissolution of asphaltene aggregates. As described in chapter 3.3, the NIR technique is sensitive to the size of scattering particles. Thus, the change in size of the asphaltene aggregates could be probed as a function of time and additive concentration.

Different chemicals with various functional groups were employed in the experiments; fatty alcohols and fatty amines, which are typical ingredients in asphaltene inhibitors, as well as a commercial inhibitor designed to inhibit asphaltene precipitation. In addition, another group of indigenous components found in crude oil, namely naphthenic acids were incorporated in the experimental matrix.

The experiments were performed by continuously measuring the change in scattering at 1600 nm wavelength, upon addition of chemicals, in a solution of asphaltenes in heptane/toluene (70/30 by volume). At that aromatic/paraffinic ratio, the asphaltenes were expected to form rather large aggregates, and any effect on the size should be easy detectable. In Figure 4-11, the effect of increasing concentration of a polydisperse naphthenic acid on the aggregate size is shown. The results showed a clear decrease in scattering as a function of time after the acid was introduced, i.e. the aggregate sizes decreased. As more concentrated solutions of acid was injected, the scattering decreased more rapidly as a function of time.

38 Chapter 4 - Main Results

Time [min]

0.000 No additive

0.125 wt% -0.005

1.25 wt%

“ -0.010

5 -0.015 3.25 wt%

6.25 wt°/o -0.020

12.5 wt% -0.025

-0.030

Figure 4-10 NIR scattering measurements at 1600 nm for 0.125 wt% asphaltenes in a 70/30 by volume n-heptane/toluene mixture with crude naphthenic acid (CNA) added in various concentrations.

When comparing the various naphthenic acids, the synthesised monodisperse acids showed the largest influence upon the asphaltene aggregates, the polydisperse naphthenic acids seemed to affect the state of the asphaltenes only to a minor extent. The other amphiphiles, showed a varying effect on the disintegration of the asphaltenes. In all cases inhibitor A, the commercial mixture, gave the most efficient treatment.

The oil phases consisting of mixtures of heptane, toluene, asphaltenes and various chemicals, were also subjected to an emulsification with tap water, 80/20 by volume. The mixing was done with an Ultra Turrax T25 rotor emulsifier at 22500 rpm for 30 seconds. The stability of the resulting emulsions was thereafter measured with a critical electric field emulsion stability devise (ECnt), which measures the necessary electric field one must apply in order to break the emulsion. A further description of the specific device and related theory is given by Aske et al. [114]. The results from the Ecrit experiments showed that all the emulsions based on oil phases containing additives such as naphthenic acids or other amphiphiles, gave unstable emulsions. Reference samples containing only asphaltenes in heptane and toluene were noticeably more stable. This may indicate that the additives dispersed the asphaltene aggregates to such a degree that they were unable to reach the water-oil interface and facilitate stable emulsions.

39 Chapter 4 - Main Results

4.6 Paper VI

In Paper V it was shown that addition of naphthenic acids to solutions with asphaltene aggregates, appeared to disperse the asphaltene aggregates into smaller sizes. This was thought to occur as a consequence of acid-base interactions between naphthenic acids and asphaltene. PFG-SE NMR (pulsed field gradient-spin echo nuclear magnetic resonance) measurements were combined with NIR (near infrared) spectroscopy to further evaluate potential interactions between asphaltenes and naphthenic acids. The experiments were run with to types of asphaltenes, one extracted from an acidic crude (asphaltene 1) and one from a neutral crude (asphaltene 2). The naphthenic acids employed in the experiments were synthetic monodisperse acids. A concentration series with asphaltenes in pure toluene was also prepared and studied, in order to obtain information about self-association of the asphaltene molecules.

PFG-SE NMR (described in chapter 3.4) measurements of the concentration series of asphaltenes 1 dissolved in toluene-d 8 are presented in Figure 4-12. The median diffusion coefficient of the asphaltenes decreased as a function of increased asphaltene concentration. Ostlund et al. [115] have shown that the obstruction effect in asphaltenic systems is large, due to the asphaltenes having a disc-like structure. However, the decrease observed in this system was significantly larger than previously reported. It was thus likely that the asphaltenes investigated were not only subjected to obstruction, but also to self-association with an onset of flocculation at 0.1 wt-% asphaltenes.

40 Chapter 4 - Main Results

E O

e (wt-%)

3.4 10'

3.2 10

3 3.010

2.6 10

2.4 10

2.2 10

Figure 4-11 The median diffusion coefficients are displayed as a function of the asphaltene concentration (O). Also included are the calculated diffusion coefficients of the asphaltene aggregates (■). The calculated values of the diffusion of the asphaltene aggregates (■) are also shown in this Figure. The full line illustrates the decrease in the diffusion coefficients that was expected only due to obstruction (under the assumption that the micelles are monodisperse and oblate shaped with an axial ratio of 1:20).

In order to study systems containing both asphaltenes and naphthenic acids, both NIR and PFG-SE NMR were employed. The NIR experiments were performed upon systems where the asphaltenes were slightly above the precipitation point, as opposed to the PFG-SE NMR experiments where the systems were below this point. When samples containing both asphaltenes and naphthenic acid were studied by PFG-SE NMR, it was observed that the entire signal from 5- p(H)-cholanoic acid (CHOL) appeared at the same frequency (0.7-2.1 ppm) as the signal from the asphaltenes. The complete overlap of the signals complicated the evaluation of the samples containing CHOL. 1-

41 Chapter 4 - Main Results

naphthalenepentanoic acid, decahydro- (2C4), fortunately had an additional peak at 4 ppm, which made it possible to study the diffusion of 2C4 without any contribution from the asphaltenes. It was observed that the echo decay of 2C4 was biexponential in the presence of both asphaltene 1 and asphaltene 2, which indicated that there were monomeric acid as well as associated acid in the samples. When evaluating the echo decay arising from combined signals, the fit of the experimental data using a Levenberg- Marquardt algorithm was seen to give reasonable results. The fitted results were verified by Monte Carlo simulations [116] and the program CORE [117, 118].

It was interesting to note from the results shown in Figure 4-13 that the diffusion of asphaltene 2 decreased in all cases independently of which naphthenic acid that had been added. This indicated that both CHOL and 2C4 interacted with asphaltene 2. The diffusion coefficient of asphaltene 1, on the other hand did not change and. Thus, it appeared as if there were no or only weak interactions between the naphthenic acids and asphaltenes of type 1.

to Mr -

e.o to

6.010

10 10

2.0 10

S d d

Figure 4-12 The results from samples containing naphthenic acid and asphaltenes. (■) corresponds to the diffusion of the naphthenic acid (0.5 or 2.4 wt-% of CHOL alternatively 2C4) while (•) corresponds to the diffusion of asphaltene 1 (Al) or asphaltene 2 (A2). The diffusion of only asphaltene 1 or 2 in toluene-d8 (reference samples) has been included. Frames have been put around the diffusion coefficients from asphaltenes of the same kind (Al or A2).

42 Chapter 4 - Main Results

The NIR measurements (Figure 4-13), where the asphaltene particle size was followed as a function of time after addition of naphthenic acid, supported these results. It was shown that the particle size was reduced significantly more for asphaltene 2 than for asphaltene 1, upon addition of both CHOL and 2C4.

Time [min] 0 20 40 60 80 ! 00

d -4-E-03 -

Figure 4-13 The change in optical density (scattering) of asphaltene 1 (Al) and asphaltene 2 (A2) is displayed as function of time after addition of naphthenic acid (CHOL or 2C4). The spectrum at time = 0 was used as reference and has been subtracted from the subsequent spectra, thus eliminating the contribution from absorption to the optical density.

43 Chapter 5 - Summary and Conclusions

5 Summary and Conclusions

Studies of "live" crude oil emulsions in a high-pressure separation rig have led to the proposal of two new mechanisms for destabilisation of water-in-oil emulsions. Increased separation efficiency was observed for water-in-crude oil emulsions, when gas bubbles propagated through the mixture due to a pressure drop below the bubble point of the oil phase into a vertical separator. These results were accounted for by a flotation effect from gas bubbles on the stabilising material at the water-oil interface. Separation experiments were also performed on pressurised dead crude and model oil systems, where the flotation effect was produced through gas release from a water phase saturated with C02 gas. The effect is thought to occur for particle stabilised water-in-oil emulsions. The use of flotation of stabilising material as a separation tool is thought to be most effective in gravitational separators of batch type.

The separation properties of a "live" crude oil were compared to crude oil samples recombined with various gases to pressures equal to the "live" samples. The results showed that water-in-oil emulsions produced from the "live" North Sea crude oil samples, generally separated faster and more complete, than emulsions based on recombined samples of the same crude oil.

The adsorption of asphaltenes and resins onto a hydrophilic gold surface was investigated by a quarts crystal microbalance with dissipation measurements (QCM-D™). The results showed that resins in pure heptane adsorb onto a hydrophilic surface, and pack into a compact monolayer. Asphaltenes in heptane/toluene mixtures, or pure toluene, adsorbed to a larger degree than the resins. The adsorption was higher than observed for typical non-associating polymers indicating aggregate adsorption. Desorption studies showed that resins were unable to desorb pre-adsorbed asphaltenes from the surface. Neither did they adsorb onto the asphaltene-coated surface. However, mixtures of resins and asphaltenes associated in bulk liquid and preformed resin/asphaltene-aggregates adsorbed to the surface.

Near infrared (NIR) spectroscopy was introduced as a potent tool for studying the effect of different chemicals in dissolving asphaltene aggregates. When comparing the effect from various naphthenic acids, synthesised monodisperse acids showed a reduction of

44 Chapter 5 - Summary and Conclusions

size of the asphaltene aggregates, whereas the more polydisperse naphthenic acids seemed to affect the state of the asphaltenes only to a minor extent. Other amphiphiles such as amines and alcohols, showed a varying effect on the disintegration of the asphaltenes.

Asphaltenes from two different oil types were studied upon addition of two kinds of naphthenic acids by employing PFG-SE NMR (pulsed field gradient spin echo nuclear magnetic resonance) and NIR (near infrared) spectroscopy. The results implied that there were interactions between the asphaltenes and the acids. The dispersing effect of the naphthenic acids on the asphaltenes was also evaluated, and it appeared as if the effectiveness of the acids depended on the asphaltene type. Furthermore, a concentration series of one of the asphaltenes was prepared, and a dramatic decrease in diffusion coefficients upon increased concentration implied that the asphaltenes began to self-associate at concentrations above 0.1 wt-% of asphaltenes in toluene-dg. When comparing the decrease of the diffusion coefficients with theory, it appeared likely that the asphaltenes were oblate shaped aggregates with an axial ratio of approximately

1:20.

45 References

References

1. Hirschberg, A., L.N.J. DeJong, B.A. Schipper, and J.G. Meijer, Influence of Temperature and Pressure on Asphaltene Flocculation. Soc. Pet. Eng. 3., 1984. 24(3): p. 283-293. 2. Hammami, A., C.P. Phelps, T. Monger-McClure, and T.M. Little, Asphaltene Precipitation from Live Oils: An Experimental Investigation of Onset Conditions and Reversibility. Energy & Fuels, 2000. 14: p. 14-18. 3. De Boer, R.B., K. Leerlooyer, M.R.P. Eigner, and A.R.D. Van Bergen, Screening of Crude Oils for Asphaltene Precipitation: Theory, Practice, and the Selection of Inhibitors. Society of Petroleum Engineers, 1992: p. 259-270. 4. Wiehe, I.A., R.J. Kennedy, and G. Dickakian, Fouling of Nearly Incompatible Oils, in Symposium on Heavy Oil and Resid Compatibility and Stability. Presented before the Division of Petroleum Chemistry, Inc. 221st National Meeting, American Chemical Society. 2001. San Diego, CA. 5. All, M.A. and M.R. Islam, The effect of asphaltene precipitation on carbonate rock permeability: An experimental and numerical approach, in SPE Annual Technical Meeting and Exhibition. 1997. San Antonio, Texas. 6. Collins, S.H. and 3.G. Melrose, Adsorption of asphaltenes and water on reservoir rock minerals, in SPE Oilfield and Geothermal Chemistry International Symposium. 1983. Denver, CO. 7. Berridge, S.A., M.T. Thew, and A.G. Loriston-Clarke, The Formation and Stability of Emulsions of Water in Crude Petroleum and Similar Stocks. Journal of the Institute of Petroleum, 1968. 54(539): p. 333-357. 8. Fprdedal, H., Y. Schildberg, 3. Sjoblom, and 3.L. Voile, Crude oil emulsions in high electric fields as studied by dielectric spectroscopy. Influence of interaction between commercial and indigenous surfactants. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 1996. 106(1): p. 33-47. 9. Strassner, 3.E., Effect of pH on Interfacial Films and Stability of Crude Oil-Water Emulsions. Journal of Petroleum Technology, 1968: p. 303-312. 10. Sjoblom, 3., Stabilization and Destabilization of Water-in-Crude Oil Emulsions from the Norwegian Continental Shelf. Correlation with Model Systems. Advances in Colloid and Interface Science, 1992. 41: p. 241-271. 11. McLean, J.D., P.M. Spiecker, A.P. Sullivan, and P.K. Kilpatrick, The Role of Petroleum Asphaltenes in the Stabilization of Water-in-Oil Emulsion, in Structures and Dynamics of Asphaltene, O.C. Mullins and E.Y. Sheu, Editors. 1998, Plenum Press: New York. p. 377- 422. 12. Speight, J.G., The chemistry and technology of petroleum. 1999, New York. Marcel Dekker. 13. Vasquez, D. and G.A. Mansoori, Identification and measurement of petroleum precipitates. Journal of Petroleum Science and Engineering, 2000. 26: p. 49-55. 14. Speight, 3.G., Fuel Science and Technology Handbook. 1990, New York: Marcel Dekker. 15. Speight, J.G., Asphaltic Constituents, in The Chemistry and Technology of Petroleum. 1991, Marcel Dekker, Inc.: New York. 16. Miller, R., Hydrocarbon class fractionation with bonded phase liquid chromatography. Anal. Chem, 1982. 54: p. 1742. 17. Aske, N., H. Kallevik, and 3. Sjoblom, Determination of Saturate, Aromatic, Resin, and Asphaltenic (SARA) Components in Crude Oils by Means of Infrared and Near-Infrared Spectroscopy. Energy & Fuels, 2001. 15(5): p. 1304-1312. 18. Altgelt, K.H. and M.M. Boduszynski, Composition and Analysis of Heavy Petroleum Fractions. 1994, New York: Marcel Dekker Inc.

46 References

19. Speight, J.G., The Chemistry and Technology of Petroleum. 1991, New York: Marcel Dekker. 20. Boussingault, 3.B., Memoire sur la composition des bitumens. Ann. Chim. Phys., 1837. 64(141). 21. Ruhr, B.3., C. Cathrea, L. Coates, H. Kalra, and A.I. Majeed, Properties of asphaltenes from a waxy crude. Fuel, 1991. 70: p. 1293-1297. 22. All, L.H. and K.A. Al-Ghannam, Investigations into asphaltenes in heavy crude oils. I. Effect of temperature on precipitation by alkane solvents. Fuel, 1981. 60(November): p. 1043- 1046. 23. Joshi, N.B., O.C. Mullins, 3. Abdul, 3. Creek, and 3. McFadden, Asphaltene Precipitation from Live Crude Oils. Energy & Fuels, 2001. 15: p. 979-986. 24. Fotland, P., Precipitation of Asphaltenes at high Pressures Experimental Techniques and Results. Fuel and Science Technology, 1996. 14((1&2)): p. 313-325. 25. Laux, H., I. Rahimian, and D. Browarzik, Flocculation of asphaltenes at high pressure. I. Experimental determination of the onset of flocculation. Petroleum Science and Technology, 2001. 19(9&10): p. 1155-1166. 26. Sheu, E.Y., Physics of asphaltene micelles and microemulsions - theory and experiment. 3. Phys. Condens. Matter, 1996. 8(25A): p. A125-A141. 27. Cimino, R., S. Correra, A. Del Bianco, T.P. Lockheart, and S.p.A. Einricerche, Solubility and phase behavior of asphaltenes in hydrocarbon media, in Asphaltenes: Fundamentals and Applications, E.Y. Sheu and O.C. Mullins, Editors. 1995, Plenum Press: New York. p. 7-130. 28. Long, B.R., The Concept of Asphaltenes, in Chemistry of Asphaltenes, 3.W. Bunger and N.C. Li, Editors. 1981, American Chemical Society: Washington, DC. p. 17-27. 29. Yen, T.F., 3.G. Erdman, and S.S. Pollack, Anal. Chem. 1961. 33(11): p. 1587. 30. Dickie, 3.P. and T.F. Yen, Macrostructures of the Asphaltic Fractions by Various Instrumental Methods. Anal. Chem, 1967. 39(14): p. 1847-1852. 31. Espinat, D., 3.C. Ravey, V. Guille, 3. Lambard, T. Zemb, and 3.P. Cotton, Colloidal macrostructure of crude oil studied by neutron and X-ray small angle scattering techniques. 3ournal de Physique IV, 1993. 3(Colloque C8, supplement au 3ournal de Physique I): p. 181-184. 32. Chang, C.-L. and S.H. Fogler, Stabilization of Asphaltenes in Aliphatic Solvents Using Alkylbenzene-Derived Amphiphiles. 2. Study of the Asphaltene-Amphiphile Interactions and Structures Using Fourier Transform Infrared Spectroscopy and Small-Angel X-ray Scattering Techniques. Langmuir, 1994. 10: p. 1758-1766. 33. Chang, C.-L. and S.H. Fogler, Stabilization of Asphaltenes in Aliphatic Solvents Using Alkylbenzene-Derived Amphiphiles. 1. Effect of the Chemical Structure of Amphiphiles on the Asphaltenes Stabilization. Langmuir, 1994. 10: p. 1749-1757. 34. Mullins, O.C. and E.Y. Sheu, Structures and Dynamics of Asphaltenes. 1998: Plenum Press, New York. 35. Chilingarian, G.V. and T.F. Yen, Bitumens, Asphalts, and Tar Sands. 1978, New York: Elsevier Scientific Publishing Co. 36. Speight, 3.G., The Chemistry and Technology of Petroleum. 1980, New York: Marcel Dekker. 37. Tissot, B.P. and D.H. Welte, Petroleum Formation and Occurence. 1984, Berlin: Springer Verlag. 38. Bunger, 3.W. and N.C. Li, Chemistry of Asphaltenes, in Advances in Chemistry Series 195. 1981, American Chemical Society: Washington. 39. Sheu, E.Y. and O.C. Mullins, Asphaltenes: Fundamentals and Applications. 1995, New York: Plenum Press.

47 References

40. Sheu, E.Y., in Asphaltene-Fundamentals and Applications, E.Y. Sheu and O.C. Mullins, Editors. 1995, Plenum: New York. 41. Ravey, J.C., G. Ducouret, and D. Espinat, Asphaltene macrostructure by small angle neutron scattering. Fuel, 1988. 67(November): p. 1560-1567. 42. Burya, Y.G., I.K. Yudin, V.A. Dechabo, and M.A. Anisimov, Colloidal Properties of Crude Oil Studied by Dynamic Light-Scattering. International Journal of Thermophysics, 2001. 22(5): p. 1397-1410. 43. Boduszynski, M.W., Asphaltenes in petroleum asphalts: composition and formation, in Chemistry of Asphaltenes, J.W. Bunger and N.C. Li, Editors. 1981 Washington, DC. 44. Groenzin, H. and O.C. Mullins, Molecular Size and Structure of Asphaltenes from Various Sources. Energy & Fuels, 2000. 14(3): p. 677-684. 45. Groezin, H. and O. Mulling, Molecular size and structure of asphaltenes. Petroleum Science and Technology, 2001. 19(1&2): p. 219-230. 46. Buenrostro-Gonzalez, E., H. Groenzin, C. Lira-Galeana, and O.C. Mullins, The Overriding Chemical Principles that Define Asphaltenes. Energy Fuels, 2001. 15(4): p. 972-978. 47. Storm, D.A., S.J. DeCanion, and M.M. DeTar, Fuel, 1990. 69(735). 48. Acevedo, S., G. Escobar, L. Gutierrez, and H. Rivas, Isolation and characterization of natural surfactants from extra heavy crude oils, asphaltenes and maltenes. Interpretation of their interfacial tension - pH behaviour in terms of ion pair formation. Fuel, 1992. 71(June): p. 619-623. 49. Yarranton, H.W., H. Hussein, and J.H. Masliyah, Water-in-Hydrocarbon Emulsions Stabilized by Asphaltenes at Low Concentrations. Journal of Colloid and Interface Science, 2000. 228: p. 52-63. 50. Nellensteyn, F.J., The constitution of asphalt. Journal of the Institute of Petroleum Technology, 1924. 10: p. 311. 51. Sheu, E.Y., M.M. De Tar, D.A. Storm, and S.J. DeCanio, Aggregation and self-association kinetics of asphaltenes in organic solvents. Fuel, 1992. 71: p. 299-302. 52. Pfeiffer, J.P. and R.N. Saal, Asphaltic Bitumen as Colloid System. Journal of Physical Chemistry, 1940. 44(2): p. 139. 53. Leontaritis, K.J. and G.A. Mansoori. Asphaltene Flocculation During Oil recovery and processing: A Thermodynamic-Colloidal Model, in SPE Int. Symposium on Oil Field Chemistry. 1987. San Antonio, TX. 54. Koots, J.A. and J.G. Speight, Relation of petroleum resins to asphaltenes. Fuel, 1975. 54: p. 179-184. 55. Peramanu, S., C. Singh, M. Agrawala, and H.W. Yarranton, Investigation on the Reversibility of Asphaltene Precipitation. Energy & Fuels, 2001. 15: p. 910-917. 56. Gonzalez, G. and A. Middea, Peptization of asphaltene by various oil soluble amphiphiles. Colloids and Surfaces, 1991. 52: p. 207-217. 57. Leon, O., E. Contreras, E. Rogel, G. Dambakli, J. Espidel, and S. Acevedo, The Influence of the Adsorption of Amphiphiles and Resins in Controlling Asphaltene Flocculation. Fuel, 2001. 15: p. 1028-1032. 58. Auflem, I.H., T.E. Havre, and J. Sjoblom, Near Infrared Study on the Dispersive Effects of Amphiphiles and Naphthenic Acids on Asphaltenes in Model Heptane-Toluene Mixtures. Colloid and Polymer Science, In Press, 2002 (DPI 10.1007/s00396-002-0660-9). 59. Sjoblom, J., N. Aske, I.H. Auflem, 0. Brandal, T.E. Havre, 0. Saether, A. Westvik, E.E. Johnsen, and H. Kallevik, Our Current Understanding of Water-in-Crude Oil Emulsions. Recent Characterization Techniques and High Pressure Performance. A Collection of Invited Papers in Honour of Professor J.Th.C. Overbeek on the occasion of his 90th Birthday. Advances in Colloid and Interface Science, In press, 2002.

48 References

60. Rousseau, G., H. Zhou, and C. Hurtevent. Calcium Carbonate and Naphthenate Mixed Scale in Deep-offshore Fields, in SPE Oilfield Scale Symposium. 2000. Aberdeen, UK: Society of Petroleum Engineers Inc. 61. Havre, T.E., M.-H. Ese, J. Sjoblom, and A.M. Blokhus, Langmuir Films of Naphthenic Adds at Different pHs and Electrolyte Concentrations. Colloid Polymer Science, In Press. 62. Becher, P., ed. Encyclopedia of Emulsion Technology. Vol. 1. 1983, Marcel Dekker: New York. 63. Sjoblom, J., ed. Emulsions- A fundamental and practical approach. NATO ASI Series. 1992, Kluwert Academic Publishers: Dordrect. 64. Lissant, L, ed. Emulsion and emulsion technology. . Vol. 2. 1976, Marcel Dekker: New York.

65. Sjoblom, J., ed. Emulsions and Emulsion Stability. . 1996, Marcel Dekker: New York. 66. Sjoblom, J., R. Lindberg, and S.E. Friberg, Microemulsions - Phase equilibria characterization, structures, applications and chemical reactions. Advances in Colloid and Interface Science, 1996. 65: p. 125-287. 67. Sjoblom, ]., H. Soderlund, S. Lindblad, EJ. Johansen, and I.M. Skjarvo, Water-in-crude oil emulsions from the Norwegian continental shelf. Part II. Chemical destabilization and interfacial tensions. Colloid & Polymer Science, 1990. 268: p. 389-398. 68. Sjoblom, J., O. Urdahl, H. Holland, A.A. Christy, and E.J. Johansen, Water-in-crude oil emulsions. Formation, characterization, and destabilization. Progress in Colloid & Polymer Science, 1990. 82: p. 131-139. 69. McLean, J.D. and P.K. Kilpatrick, Effects of asphaltene solvency on stability of water-in- crude-oil emulsions. Journal of Colloid and Interface Science, 1997. 189(2): p. 242-253. 70. McLean, J.D. and P.K. Kilpatrick, Effects of asphaltene aggregation in model heptane- toluene mixtures on stability of water-in-oil emulsions. Journal of Colloid and Interface Science, 1997. 196(1): p. 23-34. 71. Singh, B.P., Correlation Between Surface Film Pressure and Stability of Emulsions. Energy Sources, 1997. 19: p. 783-788. 72. Fingas, M., Water-in-Oil Emulsion Formation: A Review of Physics and Mathematical Modelling. Spill Science & technology Bulletin, 1995. 2(1): p. 55-59. 73. Mohammed, R.A., A.I. Bailey, P.F. Luckham, and S.E. Taylor, Dewatering of crude oil emulsions 2. Interfacial properties of the asphaltic constituents of crude oil. Colloids Surfaces A: Physicochem. Eng. Aspects, 1993. 80: p. 237-242. 74. Kilpatrick, P.K. and M.P. Spiecker, Asphaltene Emulsions, in Encyclopedic Handbook of Emulsion Technology, J. Sjoblom, Editor. 2001, Marcel Dekker, Inc.: New York. 75. Eley, D.D., M.J. Hey, and J.D. Symonds, Emulsions of Water in Asphaltene-Containing Oils. 1. Droplet Size Distribution and Emulsification Rates. Colloids and Surfaces, 1988. 32: p. 87-101. 76. Spicker, P.M., The Impact of Apshaltene Chemistry and Solvation on Emulsion and Interfacial Film Formation, in Department of Chemical Engineering. 2001, North Carolina State University: Raleigh, p. 287. 77. Jones, T.J., E.L. Neustadter, and K.P. Whittingham, Water-in-crude oil emulsion stability and emulsion destabilization by chemical demulsifiers. The Journal of Canadian Petroleum Technolog, 1978. 17: p. 100-108. 78. Djuve, J., X. Yang, I.J. Fjellanger, J. Sjoblom, and E. Pelizzetti, Chemical destabilization of crude oil based emulsions and asphaltene stabilized emulsions. Colloid Polymer Science, 2001. 279: p. 232-239. 79. Eley, D.D., M.J. Hey, and M.A. Lee, Rheological Studies of Asphaltene Films Adsorbed at the Oil/Water Interface. J. Colloid. Interface. Sci., 1987. 24: p. 173-182.

49 Paper I Journalof Petroleum Science & Engineering ELSEVIER Journal of Petroleum Science and Engineering 31 (2001) 1-12 www.elsevier.com/locate/jpetscieng

Influence of pressure and solvency on the separation of water-in-crude-oil emulsions from the North Sea

Inge Harald Auflema, Harald Kallevik \ Arild Westvikb , Johan Sjoblom b ’ *

" Department of Chemical Engineering, The Norwegian University of Science and Technology (NTNU), N-7491 Trondheim, Norway b StatoilR&D Centre, Arkitekt Ebbellsvei 10, Rotvoll, N-7005 Trondheim, Norway

Received 15 August 2000; accepted 4 July 2001

Abstract

The effect of pressure on the stability of emulsions has been studied for a recombined crude oil from the North Sea. The experiments have been conducted in a combined high pressure and temperature rig at Statoil R&D Centre. Through addition of the solvent toluene, the composition of the crude has been altered. This addition results in a decreased stability of the emulsions at the corresponding pressures, which has also been confirmed by bottle tests at 1 bar and ambient temperatures. The crude oil has been recombined with separator gas giving a bubble point of 11 bars (100 °C). Experiments show a significant increase in separation efficiency, as the separation pressure is lowered from the bubble point and downward. The experiments conducted at lower separation pressures all show a degassing, which creates a flotation effect where the rising gas bubbles transport surface active material away from the water-oil interface, resulting in less stable emulsions. The formation of a foam in the separator verifies this proposed mechanism. ©2001 Elsevier Science B.V. All rights reserved.

Keywords: Pressure; Separation; Flotation; Oil-in-water emulsion; Asphaltene

1. Introduction future. This means that the challenges are focused on small and marginal fields in combination with a Conventional offshore and onshore petroleum future exploration at deep-water fields, usually in production has focused on major reservoirs, large combination with high pressure and high tempera ­ enough to have a production unit with all necessary ture. In the struggle to make these findings worth­ production facilities. In the offshore petroleum pro ­ while to explore, new technology must be developed. duction on the Norwegian Continental Shelf, the This places severe requirements on technology solu­ possibilities of preserving such a profile are less tions as well as a better understanding of fluid probable for the years to come. Most likely, such properties in flow assurance, separation and trans ­ optimal production conditions with a single large port. Without mastering these challenges, successful oil-bearing field, which could be built out with single explorations might be questionable. A step in this topside solutions, are less probable in the immediate direction are the huge sub-sea solutions recently taken into production in the North Sea. A direct

* Corresponding author, Tel.: +47-73-584-134; fax; +47-73- advantage behind the sub-sea separation is to benefit 584-890. from the crude oil chemistry at higher pressures. It is E-mail address. [email protected] (J. Sjoblom). well known that the crude oils undergo composi-

0920-4105/01/$ - see from matter ©2001 Elsevier Science B.V. All rights reserved. Pit: S0920-4105(01)00134-6 2 I.H Auflem et a!./Journal of Petroleum Science and Engineering 31 (2001) 1-12 tional changes as a function of pressure and tempera ­ if the solution conditions or changes herein so favor. ture. Two major trends can be distinguished, i.e. Obviously, the final particle size or the flocculation compositional changes due to asphaltene precipita ­ of these nanosized particles is critical with regard to tion (Hirschberg et al., 1984; Mansoori et al., 1988; the stabilizing capacity of these entities (Sjoblom et Mansoori, 1996, 1997; Park and Mansoori, 1988; al., 1998). Park et al., 1994; Andersen and Birdi. 1990; De Boer It is well recognized that such emulsions are et al., 1992; Escobedo and Mansoori. 1995, 1997) stabilized by means of an interplay between different (Fig. 10) and due to degassing. heavy components, organic and inorganic particles, respectively. Heavy components cover asphaltenes, resins, etc. In a depressurized anhydrous crude oil, 2. Theory the asphaltenes are normally in a particulate form. The role of the resins (and lighter polar components) 2.1. Stability mechanisms of crude oil emulsions is to stabilize the asphaltene dispersion (suspension) by adsorption. Due to the strong interaction, the The lifetime of the emulsion (and the retention asphaltene particles are prevented from concomitant time in a full-scale separator) is depending on the coagulation and precipitation. The stability will also kind of stability mechanisms involved. There exist put some restriction with regard to particle sizes several possibilities of finding stabilizing agents in since the largest particles are supposed to show the either the crude oil itself or in added production highest rate of individual sedimentation. When water chemicals. Among the indigenous stabilizers, as­ is combined with the crude oil, the situation will phaltenes /resins /porphyrines are mentioned as pos ­ drastically change. The system will reach an energet ­ sible candidates for the stabilization of the w/o ically higher level, where the energy difference is emulsions. In most cases it is, however, not possible proportional to the interfacial area created during the to blink out one single component responsible for the mixing process. This fresh interfacial area will attract stability of the dispersed droplets, but several com­ components in the system. The molecules possessing ponents /fractions are in parallel the sources for the the highest interfacial activity will try to cover the stability of the emulsion. The components/fractions fresh w/o interface and hence minimize the energy in the crude oil show a large range of molecular level of the system. This category of indigenous weights (Yen, 1974; Speight et al., 1985). Lighter components is normally covered by the lighter polar components like the resins can act as individual fraction, i.e. the resins. As a consequence, a competi ­ monomers in a similar manner as traditional oil-solu ­ tion situation between resin molecules at the w/o ble surfactants. The driving force for their action is interface and on the solid asphaltene particles will the presence of water (and the existence of a w/o occur. Decisive factors determining the final position interface). Usually, the low molecular weight resins of the resins are the hydrophilic/lipophilic balance have a tendency to be the most interfacially active, of these molecules and the corresponding properties i.e. to first reach and cover a fresh w/o interface. of the solid surface. One could imagine that a very However, this is mostly a necessary requirement but hydrophobic particle surface and a very polar w/o not a sufficient one for the formation of stable w/o interface would extract different types of resins for emulsions. The next step in the stabilization process the adsorption. However, as pointed out, highly in ­ involves the interaction with the heavier crude oil terfacially active resins will show preference for the components, i.e. the asphaltenes. Depending on the w/o interface over not only less hydrophobic resin production history and the fluid properties, these molecules but also over asphaltenes. As a conse ­ molecules can be either in a monomeric or associ­ quence, the solubility conditions of the asphaltenes ated state. In the latter case, small particles have will drastically change and a particulate precipitation formed. The formation of these is normally due to will take place. With aqueous droplets coated by an the stacking tendency of the individual asphaltene interfacial resin film as closest neighbors, the asphal ­ molecules. These nanosized particles will have a tene particles will precipitate and accumulate at the strong tendency to accumulate at the w/o interfaces, droplet surface. The resulting interfacial properties I.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12 3

Table 1 sandwich-like structures as a consequence of the Measured physical properties for North Sea crude oil molecular association. The presence of other fines Physical properties Measured like organic wax particles and inorganic clay parti ­ value cles will further enhance the stability level. Molecular weight of crude 226 The interfacial conditions are reflected in the level (g/mol) (Measured by of the interfacial pressure (ir). Sjoblom et al. (1992) freezing point depression) showed that there is a correlation between the level Density at 60 °C (g/cm3) 0.8209 Asphaltene content (wt.%) 1.4 of 77 and the macroscopic emulsion stability. Prefer­ (Precipitated in 1:10 ably, the interfacial pressure should be above 10-14 crude/ n-pentane) mN/m for stable emulsions. Aromatic molecules Phase inversion point for 60-65 like benzene will substantially lower the level of tt. water-in-crude oil emulsion (vol.% water) With increasing content of aromatic molecules, the Sulphur content (wt.%) 0.26 interfacial activity of the indigenous surfactants will Water content (wt.%) 0.0456 be canceled and hence the emulsion stability will vanish. The dilution with aromatic solvents is in practical use in many places in the world where heavy crude oils create transport and emulsion prob ­ will be much more rigid!fied and the stability of the lems. corresponding emulsions is profoundly improved. Coalescence is defined as the combination of two Central mechanisms involved in the stabilization or more droplets to form a larger drop. When these process will hence be both steric and particle stabi ­ droplets approach each other, a thin film of the lization. continuous phase will therefore be trapped between The mechanical properties of the protecting inter ­ the droplets, and it is obvious that the properties of facial film are essential for the final stability level of this film will determine the stability of the emulsion the w/o emulsions. Concentrated polymeric interfa ­ (Brown and Hanson, 1968). The mechanism of coa­ cial films may display either elastic or viscous prop ­ lescence occurs in two stages; film thinning and film erties that make the destabilization process difficult rupture. In order to have film thinning, there must be and time consuming. The aromatic asphaltene a flow of fluid in the film, and a pressure gradient molecules will normally undergo a stacking into present. It is obvious that the rate of film thinning is 4 I.H. Auflem et al./Journal of Petroleum Science and Engineering SI (2001) 1-12 affected by the properties of the colloidal system. fluids for each experiment was 100 ml, with a con ­ Some of the most important parameters are defined stant water content of 35 vol.%. For each experi ­ as (Liem and Woods, 1974) viscosity and density of ment, the content of toluene in the crude oil phase the two phases present, interfacial tension and its was varied. The separation was carried out in 100-ml gradient, interfacial shear and dilatational viscosities capped measuring cylinders, and was monitored vi­ and elasticities, drop size, concentration and type of sually as volumes of water separated from the emul­ surfactant present at the interface and forces acting sion as a function of time. between the interfaces. Considerable effort has been made to develop models for prediction of the rate of 3.3. High pressure and temperature separation rig film thinning and critical film thickness. (HPHT rig) In this study of emulsion stability at high pres ­ sures, we intend to view new destabilization mecha ­ The high pressure and temperature separation rig nisms. For this purpose, we use a high-pressure and (Sjoblom et al., 2001), as illustrated in Fig. 1 was high-temperature rig for the emulsion formation used both for preparation of the water-in-oil emul­ through a choke. sions and for monitoring of the eventual resolution

3. Experimental Table 2 Overview over experiments presented in the figures 3.1. Chemicals Data presented Pressure in sample Toluene in figure no. cylinders -* pressure content The crude oil is from a Statoil field in the North after PI3 -» pressure in oil Sea. The separator gas, which was used for recom­ after P12 = pressure in phase (%) bining the crude oil, was collected from the same oil separator cell (bar) well. Recombining the crude gave a gas saturation 2 100—* 11—^7 0 point (bubble point) at 11 bars and 100 °C. The oil gas cap removed 2 100 -»11 -»1 0 and gas mixture was stored in a pressurized sample gas cap removed bottle at 12 bars. Using a high-pressure pump, the 2 and 3 100 — 11 — 1 0 recombined crude oil was transferred into the sample 3 100 —»11 —»1 20 cylinders, while keeping the pressure above 12 bars. 3 100 -»11 ->1 40 Known data of the crude are presented in Table 1. 3 100-11 -1 40 2 and 4 100-11-7 0 Toluene of p.a. quality from Lab-Scan was used as 4 100-11-7 5 supplied for the experiments. The water used in the 4 100-> 11 -*7 45 experiments was prepared according to a normal ion 4 100 —»11 ->7 50 composition of North Sea formation water, i.e. 111,11 4 100 -> 11 ->7 60 g/1 NaCl, 3,80 g/1 KC1, 50,11 g/1 CaCl-, and 10,40 4 100 -»11 -» 7 80 2 and 5 100-17-11 0 g/1 MgCl, all of p.a. quality. The salts were mixed 5 100-17-11 20 in de-ionized water and the pH for the mixture was 5 100-17-11 40 measured to 3.70 at 22.9 °C. 5 100 — 17 - 11 60 5 100-17-11 80 3.2. Bottle tests 6 182- 122 -11 20 6 182 - 122 - 11 30 6 182 - 122 -11 40 The emulsions used in the bottle tests were mixed 8 Bottle test 0 using a Silverson L4RT emulsifier at 2000 rpm for 8 Bottle test 20 1.0 min. Before the aqueous phase was added, the 8 Bottle test 30 toluene and crude oil were homogenized for 1.0 min. 8 Bottle test 40 8 Bottle test 50 All fluids and emulsions were kept at 60 °C, using a 8 Bottle test 60 temperature-controlled water bath. Total volume of I.H. Auflem et at./ Journal of Petroleum Science and Engineering 31 (2001) 1-12 5 of phases. The rig has four 600 cm3 high-pressure in the cell, pressure drop over choke valves P12 and sample cylinders, which are controlled by four mo­ P13, and toluene content of oil phase. Table 2 con ­ tor-driven high capacity piston pumps, which are tains an overview of the experiments included in this independent of each other. In the pressure experi ­ paper, with a reference to the figures they are pre ­ ments, two sample cylinders, B3 and B4 were filled sented in. The pressures referred to for each experi ­ with toluene and crude, respectively, and kept at a ment, as for instance 100 —* 11 —» 1, can be ex­ constant pressure and 60 °C. The saline water was plained as the following. The first number is the contained on the right side of the rig in sample pressure in the sample cylinders, the second is the cylinder B2, at the same temperature and pressure. In pressure downstream choke valve P13, and the last each experiment, toluene and crude were mixed number is the pressure downstream choke valve P12 through a choke valve (P13). Downstream of PI3, and in the separator cell. the oil phase was mixed with the water phase through another choke valve (PI2), before it entered the separator cell. The pressure drops over the choke 4. Results valves and the pressure in the separator cell, are all back-pressure controlled. The 450-cm3 separator cell is made out of sapphire, assuring full visibility of the 4.1. Effect of separation pressure on emulsion stabil­ separation process, and has an upper pressure limita­ ity, without toluene addition tion of 200 bars. Before each experiment, the separa ­ tor cell was pressurized with separator gas to the Fig. 2 shows results where the separation pressure desired pressure. After the separator cell was filled, has been varied. The resolution curves present the the amount of the different contents of foam, oil, amount of water separated as a function of time. emulsion/dispersion and water was recorded as a With a bubble point pressure of 11 bars for the function of time. Filling time for the separator cell gas-oil mixture, the final separation pressures are for each experiment is 30 s. All parts of the rig are designed to be at or below the bubble point pressure. thermostated, and temperature, cell pressure, valves, A separator pressure of 11 bars gave no separation pumps and video camera are computer controlled. within the duration of the experiment, i.e. 20 min. As in the bottle tests, the water content was 35 Seven bars gave up to 30% water separation, while vol.% in all experiments. Parameters that were var­ lowest pressure under study, i.e. 1 bar gave a signifi ­ ied in the experiments included separation pressure cantly larger resolution, i.e. close to 80%.

E 70.0 —bar (gets cap removed) 0% —100->11->7 baitgas cap removed) 0% -*~100->11->1 bar 0% -*-100->11->7barO% —1QO->17->1 1 b«r0%

Time [mini Fig. 2. Amount of separated water [vol.%] from crude oil emulsions at different separation pressures presented as a function of time. 6 I.H. Auflem et at. / Journal of Petroleum Science and Engineering 31 (2001) 1-12

To further investigate the influence of separation resulting water-drop diameters are of the same order pressure on the stability of emulsions, two experi ­ of magnitudes, and the dominating factor for emul­ ments were performed, where the gas cap on the sion instability seems to be the separation pressure. crude oil-gas mixture was removed. Hence, the lighter volatile components in the mixture were per ­ 4.2. Effect of toluene addition on emulsion stability manently removed when depressurizing the oil sam­ in HPHT rig ple to ambient pressure and venting. Thereafter, the oil sample was once more repressurized to 100 bars Figs. 3-6 show the effect of toluene addition to by using a piston as a mechanical device. The pres ­ the oil phase prior to emulsification at various sepa ­ sures in the separation cylinder were 1 and 7 bars, ration pressures. In all figures, the separation effi­ and in both experiments the maximum separation of ciency is displayed as amount of separated water as a water was below 8%. This was significantly lower function of time. At 1 bar, the assumed flotation than in the similar experiments with the gas cap effect is dominant creating unstable emulsions at all intact. The reason for the difference in stability in the toluene amounts. As the separation pressure is in ­ experiments is believed to be a flotation effect cre­ creased, the flotation effect is less pronounced, and ated by gas bubbles. When the pressure is lowered the impact of toluene addition to the oil phase prior below the bubble point pressure of gas in crude, the to emulsification is easier to differentiate. With in ­ lighter components undergo a phase transition from creasing percentage of toluene added, the stability of liquid to gas. These rising gas bubbles can transport the emulsion is clearly declining. This effect is seen the surface-active material away from the emulsion for separation pressures of both 7 and 11 bars. By water-oil interface, thereby destabilizing the emul­ increasing the amount of toluene in the oil phase, the sions. composition is gradually changed, and both the den ­ Measurements of the drop sizes on the stable sity and viscosity of the oil phase are somewhat emulsions, where the gas cap on the crude has been influenced. This will to a certain degree affect the removed, showed a drop diameter of 2.9 and 3.4 p,m stability of the emulsions. However, another major for pressure drops (P12) of 10 and 4 bars, respec ­ effect of the toluene addition, and in this context tively. For such small differences in pressure, the more interesting, is the influence on the aggregate

—&-100>H->l bar 40% -B-100->11->1 bar 40%

100->11->1 bar0%

0,0 5,0 10,0 15,0 20,0 Time [min] Fig. 3. Amount of separated water [%] at 1-bar separation pressures, with different amounts of toluene added to the crude oil phase, presented as a function of time. (AP = 10 bars over manifold). l.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12 7

100,0

90.0

z 80 0 | 70,0

| 80,0 -B~i0O->H->7bar6O% -*-100->11->7bw50%

1 », 100->11->7bar5% £ 100->11->7bar0% | 40,0 1 30.0

20.0

10.0

0.0 0,0,0 5,0 10,0 15,0 20,0 Time [min] Fig. 4. Amount of separated water [%] at 7-bar separation pressures, with different amounts of toluene added to the crude oil phase, presented as a function of time. (AP = 4 bars over manifold). state of the stabilizing asphaltene molecules. Through different pressure drop over the choke valve where toluene addition the aromaticity of the oil phase is the oil and water phases are mixed (6 and 111 bars). increased, creating a dissolution or de-aggregation of It is expected that an increase in pressure gradient asphaltenic aggregates as shown by McLean and over the choke should result in smaller dispersed Kilpatrick (1997a,b) and Singh et al. (1999). droplets (Hinze, 1955; Karabelas, 1978). The emul­ The experiments in Figs. 5 and 6 are carried out sion stability with 20% toluene added is somewhat at the same separation pressure (11 bars), but with larger for A P = 111 bars (Fig. 6), than for A P = 6

—100->17->11 bar 80% -e~100->17->11 bar 80% -*-100->17->11 bar 40% -e- 100*17*11 bar 20% -*~100*17->11 bffl-0%

Fig. 5. Amount of separated water [%] at 11-bar separation pressures, with different amounts of toluene added to the crude oil phase, presented as a function of time. (AP — 6 bars over manifold). 8 I.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12

182->122-»1140% 182->122->11 30%

Time [min] Fig. 6. Amount of separated water [%] at 11-bar separation pressures, with different amounts of toluene added to the crude oil phase, presented as a function of time. (A/* = 111 bars over manifold). bars (in Fig. 5). This can be interpreted as a droplet height [ml] in the separator cell for a number of size effect. experiments at different separation pressures and amounts of toluene added. The foam-height was 4.3. Foamability in the separator cell during demul- measured immediately at filling the cell. Foaming sification occurred when the separation pressure was lower The foamability, i.e. the ability of the emulsified than the bubble point pressure. Qualitatively, a posi ­ system to foam, is displayed in Fig. 7 as the foam- tive correlation was found between foaming and fast

250.00 I £ 200.00-

E 150.00- 1

100 00-

50.00

0.00 * aS # * * # 8 I 2 2 2 2 cS eS c£ s

T- ^ T" N S |V 5 & & l l l l l l l 1 | I I s I a a Fig. 7. The foamability for the system is given as the foam height [ml] in the separator cell for different samples. I.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12 9

Time [min] Fig. 8. Bottle tests for different amounts of toluene added to the crude oil phase, presented as amount of water [%] separated as a function of time.

destabilization of the w/o emulsion, which strongly phase giving a w/o emulsion with a certain droplet supports the idea of a flotation effect. Another obser ­ size distribution. In the case a w/o emulsion is vation was that a high amount of toluene added injected through a choke, the emulsion will be re­ seems to inhibit the foamability but to promote the mixed, and fresh interface will be established (Be­ demulsiftcation, which is in line with a monomeriza- efier, 1983; Larsson and Friberg, 1990; Sjoblom, tion of the stabilizing asphaltene particles. 1996; Mullins and Sheu, 1998). Under such condi ­ tions, it is essential which components will first 4.4. Bottle tests on toluene addition to oil phase reach the w/o interface (diffusion controlled) and how long it will take to obtain the final equilibrium. The results from the bench experiments are sum­ It is normally considered, and in many cases proved, marized in Fig. 8. The graph shows the correspond ­ that the lower molecular weight resins (rather polar) ing curves to the rig experiments but now based on will have the fastest diffusion to the fresh w/o bottle tests at 1 bar. The effect of toluene addition is interface. After this, higher molecular weight species clearly seen. When the toluene content in the oil (and small particles) will be located at the oil-water phase increases the stability of the emulsion de­ interface. In our experiments, we have shown that creases completely in line with the observations from the time after the choke is long enough to produce the pressure study above. very stable w/o emulsions. Hence, the different pressure gradients over the choke should mainly 5. Discussion dictate the droplet sizes and droplet size distributions (Hinze, 1955; Karabelas, 1978). In this session, we will mainly focus on identify ­ The experimental observation with a correlation ing destabilization mechanisms at elevated pressures between foamability and emulsion destabilization is and in presence of toluene. very important. In our experiments with a natural In the choice of experimental parameters, we have gas, the major gas compounds are low molecular, focused on different pressure gradients over chokes, more-or-less non-polar species (CH4, C2H6, etc.) different separator pressures and finally different that should be dissolved mainly in the continuous chemical conditions for the bulk phase of the w/o hydrocarbon phase. The main reason for destabiliza ­ emulsions. The pressure gradient over the chokes tion of the w/o emulsion, is the degassing of a will represent a mixing of the aqueous and the oil non-polar gas and removal of particulate matter at 10 I.H. Auflem et al./ Journal of Petroleum Science and Engineering 31 (2001) 1-12 the w/o interface, or a so-called flotation effect (Fig. only minor water amounts separated ( < 8%) during 9) (Verlon, 1975; Grimsley, 1981; Stefan and Szeri, the time of observation. 1999). The impact of this effect must be highly The addition of toluene accelerates the separation dependent on the nature of the asphaltene particles. of water from the emulsions. In many cases, the Normally, one would expect that these particles are separation of water is complete for the higher toluene located at the interface in such a manner that the additions. The reason for the efficiency of toluene polar groups are towards the aqueous phase and the addition may be manifold. The physical properties less polar backbone towards the oil. In order to like density and viscosity are changed and this will remove the particles from the interface, the gas alter the separation picture. However, the most sig­ bubbles and the asphaltenes must interact. This inter ­ nificant change is in the chemistry of stabilizing action takes place as the gas bubbles rises through asphaltene units. At increasing amounts of toluene, the mixture in the separator (Fig. 10). there will be a disintegration of the association struc­ However, the rapid decline in the stability of the tures formed, and a less pronounced tendency to non polar foam shows that the stability mechanisms form new nanosized particles. This means that the of such a colloidal state are not optimal. The trends tendency to form w/o emulsions will most likely be discussed above are confirmed with the experiments the same or even higher, but the stability of the with “gas cap removed ” (Fig. 2). In these experi ­ dispersion will be significantly lower. This should be ments, the equilibrium gas phase has been removed the governing tendency after toluene contents in and the start pressure has been achieved by using a excess of 20 vol.%. The foamability is also affected piston (and no additional gas), i.e. by a volume by the disintegration of the asphaltene particles. Fig. reduction. Under these conditions (with no gas re­ 7 shows that the capacity of the system to form lease), the emulsions show a very high stability with oil-based foam is decreasing, most remarkably so for

Degassing Coated gas bubbles Foam formation

! I

Fig. 9. Schematic overview of removal of stabilizing surface active material from the surface of water-droplets by adsorption on rising gas bubbles. I.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12 11

— Bubble point 9 Production pathway — Hydrates (H) ♦ Wax (W) — Asphaltenes (A)

Temperature, °C Fig. 10. PT-diagram for the phase envelope of a North Sea crude oil. The onset of asphaltene precipitation at depressurizing is drawn in. As the properties of the overall petroleum phase approaches the asphaltenes, the asphaltene aggregates are assumed to redissolve.

the toluene content > 40%. In this case, the flotation phase, together with a dissolution or de-aggregation effect is absent due to the lack of particles at the of the stabilizing asphaltenic aggregates. w/o interface.

Acknowledgements

6. Conclusions Inge Harald Auflem would like to acknowledge the technology program Flucha II, financed by the We have shown a dependency between the stabil ­ Norwegian Research Council (NFR) and the industry ity of crude-oil emulsions and the separation pres ­ for a PhD grant. Statoil R&D Centre is especially sure for the emulsion and correlated this phe ­ acknowledged for supplying experimental facilities nomenon with the bubble point pressure for the and willingly sharing expertise and competence in crude oil separator gas mixture. When lowering the the field of high pressure technology. separation pressure below the bubble point pressure for the recombined crude oil, the lighter components will evaporate. These gas bubbles will rise through References the emulsion due to the density difference between the two phases creating a flotation effect, where the Andersen, S.I., Birdi, K.S., 1990. Influence of temperature and solvent on the precipitation of asphaltenes. Fuel Science & surface-active components adsorb on to the gas bub ­ Technology International 8 (6), 593-615. bles and are removed from the oil-water interface. Becher, P. (Ed.), 1983. Encyclopedia of Emulsion Technology. As the stabilizing surface-active components are Marcel Dekker, New York, pp. 57-127. flotated, the emulsion stability decreases. Brown, A.H., Hanson, C., 1968. The effect of oscillating electric Another effect shown was that by diluting the fields on the coalescence of liquid droplets. Chemical Engi ­ neering Science 23, 841-848. crude oil phase with toluene, the stability of the De Boer, R.B., Leerlooyer, K. et al., 1992. Screening of crude oils emulsions decreased. This was explained by a com­ for asphaltene precipitation: theory, practice, and the selection bination of reducing viscosity and density of the oil of inhibitors. Society of Petroleum Engineers, 259-270. 12 l.H. Auflem et al./Journal of Petroleum Science and Engineering 31 (2001) 1-12

Escobedo, J., Mansoori, G.A., 1995. Viscometric determination of Mullins, O.C.. Sheu,' E.Y., 1998. Structures and Dynamics of the onset of asphaltene flocculation —a novel method. SPE Asphaltenes. Plenum, New York. Production & Facilities 10 (2), 115-118. Park. S.J., Mansoori, G.A., 1988. Aggregation and deposition of Escobedo, J., Mansoori, G.A., 1997. Viscometric principles of heavy organics in petroleum crudes. Energy Sources 10 (2), onsets of colloidal asphaltene flocculation in paraffinic oils 109-125. and asphaltene micellization in aromatics. SPE Production & Park, S.J., Escobedo, J. et al., 1994. Asphaltene and other heavy Facilities 12 (2), 116-122. organic depositions. In: Yen, T.F., Chilingarian, G.V. (Eds.), Grimsley, R.L., 1981 Hybrid Gas Flotation Separator. United Asphaltenes and Asphalts, 1. Developments in Petroleum Sci­ States Patent. USA, Conoco, Ponca City, OK. ence, vol. 40, Elsevier, pp. 179-205. Hinze, J.O., 1955. Fundamentals of the hydrodynamic mechanism Singh. S., McLean, J.D. et al., 1999. Fused ring aromatic solvency of splitting in dispersion processes. AIChE Journal 1 (3), in destabilizing water-in-asphaltene-heptane-toluene emul­ 289-295. sions. Journal of Dispersion Science and Technology 20 (1 Hirschberg, A., De Jong, L.N.J. et al., 1984. Influence of tempera ­ and 2), 279-293. ture and pressure on asphaltene flocculation. Society of Sjoblom, J. (Ed.), 1996. Emulsions and Emulsion Stability. Mar­ Petroleum Engineers of AIME (11202)., 283-293. cel Dekker, New York, pp. 237-285. Karabelas, A.J., 1978. Droplet size spectra generated in turbulent Sjoblom, J., Mingyuan, L.I., Gu, T.R., Christy, A.A., 1992. pipe flow of dilute liquid/liquid dispersions. AIChE Journal Water-in-crude-oil emulsions from the Norwegian Continen ­ 24 (2), 170-180. tal-Shelf: 7. Interfacial pressure and emulsion stability. Col­ Larsson, K„ Friberg, S. (Eds.), 1990. Food Emulsions. Marcel loids and Surfaces 66 (l), 55-62. Dekker. New York, pp. 1-57. Sjoblom. J„ Saether, O. et al., 1998. Asphaltene and resin stabi ­ Liem, A.J.S., Woods, D.R., 1974. Review of coalescence phe ­ lized crude oil emulsions. In: Mullins, O.C., Sheu, E.Y. (Eds.), nomena. AIChE Symposium 70 (8). Structures and Dynamics of Asphaltenes. Plenum, New York, Mansoori, G.A., 1996. Asphaltene, resin, and wax deposition from p. 337. petroleum fluids: mechanisms and modeling. Arabian Journal Sjoblom, J., Johnsen. E.E., Westvik, A., Esc, M.H., Djuve, J., for Science and Engineering 21 (4B), 707-723. Auflem, I.H., Kallevik, H., 2001. Demulsifiers in oil industry. Mansoori, G.A., 1997. Modeling of asphaltene and other heavy In: Sjoblom, J. (Ed.), Encyclopedia Handbook of Emulsion organic depositions. Journal of Petroleum Science and Engi ­ Technology. Marcel Dekker, New York, pp. 595-619. neering 17 (1-2), 101-111. Speight, J.G., Wemick, D.L. et al., 1985. Rev. 40, 51 Mansoori. G.A., Jiang, T.S. et al., 1988. Asphaltene deposition Stefan, R.L., Szeri, A.J., 1999. Surfactant scavenging and surface and its role in petroleum production and processing. Arabian deposition by rising bubbles. Journal of Colloid and Interface Journal for Science and Engineering 13 (1), 17-34. Science 212, 1-13. McLean, J.D., Kilpatrick. P.K.. 1997a. Effects of asphaltene Verlon, L., 1975. Process for Separating Low API Gravity Oil aggregation in model heptane-toluene mixtures on stability of from Water. United States Patent. USA, Union Oil Company water-in-oil emulsions. Journal of Colloid and Interface Sci­ of California, Brea, CA. ence 196 (1), 23-34. Yen, T.F., 1974. Structure of petroleum asphaltene and its signifi ­ McLean, J.D., Kilpatrick, P.K., 1997b. Effects of asphaltene cance. Energy Sources 1 (4), 447-463. solvency on stability of water-in-crude-oil emulsions. Journal of Colloid and Interface Science 189 (2), 242-253. Paper II PCX (Patent Cooperation Treaty) - application

Applicant: STATOIL A.S. 4035 STAVANGER

Agency: ONSAGERS AS Postboks 265 Sentrum N-0103 OSLO

Inventors: Johan Sjoblom, Harald Kallevik, Arild Westvik, Inge Harald Auflem

Title: Process for separation of oil and water in a separator by breaking water-in-oil emulsions 1

FIELD OF INVENTION

This invention relates to a process for separation of oil, water and gas in a separator by breaking of water-in-oil emulsions in a composition comprising water, oil and optionally a gaseous phase.

5 BACKGROUND OF THE INVENTION

The crude oil production on the Norwegian Continental Shelf is facing new challenges. Many of the oil fields discovered the last years are economically marginal due to their size and location. Thus, the demand for lowering the production cost is obvious. The separation of oil and water and optionally gas from 10 such fields is likely to experience the same type of problems as observed on the larger fields in production today. One of the largest production problems is the formation of emulsions stabilized by polar heavy crude oil components like asphaltenes, resins and waxes. Such problems can be solved by means of addition of chemicals or by use of mechanical separation facilities. However, the costs of these 15 solutions are normally high and the search for new and efficient separation tools is of highest priority. The environmental issue regarding the use of chemicals is an additional disadvantage. Per today there is an increasing challenge for the use of greener and environmentally more friendly chemicals as normally imposed by the state authorities.

20 The use of a hydrophilic gas, such as for example CO?, dissolved in water as a possibility to enhance the separability of water and oil is known per se. Usually, the CO? is mixed with the water phase at an early stage in the separation process and the emulsification takes place with an aqueous phase rich in dissolved CO?. By lowering the pressure in the separator, there will be a release of the CO? and the 25 concomitant break of the oil-continuous emulsion.

It is well known that CO? will form gas hydrates at low temperatures and high pressures. Separation and injection conditions should therefore be far from the thermodynamic conditions for gas hydrate formation. Further, dissolved CO? can constitute a danger for corrosion and low pH's. These conditions must be taken into 30 account in designing a future process and in the choice of the materials.

It would be most beneficial to apply the CO? into a two-phase stream with water and oil, i.e. before the emulsification of the phases. The injected polar CO? is partitioning between the two phases in the oil-continuous emulsion. Upon pressure reduction (for instance in the separator) two processes will commence:

35 i) the CO? dissolved into the aqueous phase (the droplets) will rapidly coalesce and form small bubbles upon a pressure reduction. Due to gravity reasons these bubbles will propagate through the emulsified system. When the droplet leaves the water

1 2

droplet it has to pass a phase boundary built up by indigenous polar surfactants (asphaltenes, resins and waxes). As a consequence the interface will be ruptured.

If the CO2 bubbles carry with them surface active material from the interface (flotation effect) the time for the interface to reform will be most likely much longer 5 than the coalescence time. Hence the system will break and water and oil phase should appear.

ii) the CO2 dissolved in the oil phase will also rapidly coalesce and form bubbles upon a pressure reduction. Due to buoyancy forces the bubbles will propagate through the emulsified system. In doing so they will rip off surface-active material 10 from the o/w interface described as a flotation effect. This effect is not specific for

the CO2 but common for all oil soluble gases below the bubble point.

NO B 171 096 describes a process and apparatus for separating a dispersed phase from a continuous phase by means of dissolved gas flotation. The gas is, preferably before the pressure reduction, dissolved in the liquid system. As the pressure is 15 reduced, the dissolved gas is released and thus forming gas bubbles which improves the flotation and hence separation.

US A 4 251 361 describes a gas flotation separator for separating oil and water by

means of gas, for example CO2, is dissolved in the liquid and is released as the pressure is reduced. The gas bubbles thus formed in the liquid improves the 20 flotation.

EP A2 298 610 describes a method of oil removal from oil-water emulsions by means of volatile hydrocarbons forming a two-phase system when in contact with the emulsion under pressure to effect the replacement of at least some of the oil in the emulsion phase and dissolved in the volatile hydrocarbon phase. When the 25 pressure is reduced, the volatile hydrocarbon is vaporized and the emulsion separates into a water phase and an oil phase.

The prior art documents referred to above relates to the use of C02 as separation promoter which all are based upon flotation of oil drops, when the purpose is to remove relatively small amounts of oil in a continuous water phase.

30 Further, representative but not exhaustive documents of the present art are GB A 2 323 048, US A 5 580 464, and US A 3 884 803.

2 3

SUMMARY OF THE INVENTION

It is desirable to provide a process for separation of oil, water and gas in a separator by breaking of water-in-oil emulsions in a composition comprising water, oil and optionally a gaseous phase.

5 In accordance with this invention, there is provided a process for separation of oil, water and gas in a separator by breaking of water-in-oil emulsions in a composition comprising a water, oil and optionally a gaseous phase, comprising the following steps: a) dissolving a gas comprising one or more components into said composition prior 10 to said separator, wherein the amount of said water phase is at least about 1 weight- % based on total composition, b) introducing said composition into said separator, wherein pressure in said separator is of at least about 2 bar, and c) reducing the pressure in said separator in order to facilitate the separation of oil, 15 water and the optional gaseous phase.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a graph showing water separated versus time for a 1 % Crude A in Exxsol D-80, water content 40%. The pressure in the separator was reduced from 65 bar to 1 bar after 5 minutes.

20 Figure 2 is a graph showing % water separated versus time for a Crude A, water content 40%. The pressure in the separator was reduced from 65 bar to 1 bar after 5 minutes.

Figure 3 is a graph showing % water separated versus time for a Crude B, water content 40%. The pressure in the separator was reduced from 65 bar to 1 bar after 5 25 minutes.

Figure 4 is a graph showing % water separated versus time for a Crude C, water content 35%. There is no pressure drop in the separator.

Figure 5 is a graph showing % water separated versus time for a Crude C at several

levels of water content - with and without CO2. The pressure in the separator was 30 reduced from 65 bar to 1 bar after 2 minutes.

Figure 6 is a graph showing % water separated versus time for a Crude A at several

levels of water content - with and without CO2. The pressure in the separator was reduced from 65 bar to 1 bar after 2 minutes.

3 4

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to the use of a polar gas as separation promoter for breaking of water-in-oil emulsions in a composition comprising a water, oil and optionally a gaseous phase, wherein the purpose is to separate the water phase (the 5 dispersed phase) from the oil phase (the continuous phase) in order to break the emulsion and separate into the original components.

The polar gas is selected from polar components, which are water-soluble,

preferably wherein the gas is selected from CO2, N2, O2, and mixtures thereof, and more preferably wherein the gas is C02. Further, the polar gas is selected from non ­ 10 polar components, which are oil-soluble, preferably wherein the gas is selected

from H2, CH4, paraffin ’s, and mixtures thereof.

The polar gas is dissolved directly into the water phase before the phases are mixed, or the gas is dissolved countercurrently into the composition before the composition is introduced into the separator.

15 In fig. 1 -6 there is experimentally shown that a polar gas, such as CO2, can accelerate the breaking of crude oil based emulsions. However, this is not possible for all types of crude oil emulsions, but presumably it is possible only for those types of crude oil emulsions which are particle stabilized. In accordance with the present invention, it is shown that merely some types of crude oil emulsions will be 20 effective.

The composition comprising a water, oil and optionally a gaseous phase contains inorganic or organic particulate solids, wherein the amount of inorganic or organic particulate solids is at least about 0,01 weight- % based on total composition, preferably wherein the amount is in the range of 5-15 weight-%.

25 The organic particulate solids are constituted by nanosized asphaltene, metal organic acid, like calcium naphthenate particles, or wax particles, like heavy paraffin particles, or mixtures thereof.

The inorganic particulate solids are constituted by clay, precipitated metal salt or scale particles like calcium carbonate, barium sulphate, or mixtures thereof.

30 In accordance with the present invention the use of C02 will be effective in a gravitational separator and most effective in a separator of batch type (batch process), and the emulsion will be hold a few minutes in the separator and settle before the gas pressure in the separator is reduced. In a continuous process, the

effect of the content of CO2 is much less.

35 The pressure in the separator is reduced only after the composition is maintained in the separator at a specific period of time. The period of time is at least about 2

4 5

minutes, preferably wherein the period of time is in the range of about 2 to about 5 minutes, and more preferably wherein the period of time is about 4 minutes.

Before pressure reduction in the separator, the composition will essentially be steady, and then the released gas effects breaking of the water-in-oil emulsions and 5 thus improves separation. The pressure in the separator is reduced by releasing the separator gas, which can be selected from natural gas, nitrogen, and combinations thereof.

The pressure in the separator is of at least about 60 bar, wherein the pressure in the separator is reduced to a value below the separation pressure of the gas in the 10 composition, and the purpose is to reduce the pressure in the separator to a pressure

under about 60 bar as CO2 gas is released as a gas. The pressure in the separator is eventually reduced to 1 bar.

The more the emulsions are stabilized by particulate solids (asphaltenes), the better the method of pressure drop works. Especially, small nanosized asphaltene particles 15 will be most effective. If there are no such particulate solids present in the

emulsions, the effect of adding CO2 is not ambiguously positive.

The amount of the water phase is at least about 10 weight-% based on total composition. Preferably, the amount is at least about 30 weight-%, and more preferably the amount is up to about 70 weight-%.

20 Finally, the pressure of the composition entering the separator is at least about 10 bar. Preferably, the pressure is at least about 30 bar, but can be at least about 70 bar.

The following examples are provided for purposes of illustrating the present invention, and are not intended to be limiting of the broadest concepts of the present 25 invention. Unless otherwise stated, all percentages are by weight.

Data of the series of experiments are listed in table 1.P1 denotes the pressure of a composition comprising a water phase and an oil phase from a production site, P2 denotes that the pressure of the composition is reduced somewhat before introducing into the separator, and P3 denotes the pressure over the inlet choke to 30 the separator which also applies for the pressure in the separator.

Crude oil characteristics of three types of crude oil emulsions, indicated as Crude A, Crude B and Crude C, are shown in tables 2, 3 and 4, respectively.

5 6

Table 1. Experimental design

AP P Pressure release in wc Separator Exp No Oil Phase Water Phase (mixing) (PI ->P2->P3) separator % gas 1 % Crude A in 441 De ionised water 20 100 -> 85 -> 65 Yes - 5 min 40 n2 Exxsol D-80 1 % Crude A in 442 De-ionised water 5 100 -> 70 > 65 Yes - 5 min 40 n2 Exxsol D-80 1 % Crude A in De ionised water with 443 5 100 -> 70 -> 65 Yes 5 min 40 n2 Exxsol D-80 co 2 1 % Crude A in De-ionised water with 444 20 100 -> 85 -> 65 Yes 5 min 40 n2 Exxsol D-80 co 2 445 Crude A De-ionised water 20 100-> 85 -> 65 Yes 5 min 40 n2 446 Crude A De-ionised water 5 100 > 70 -> 65 Yes - 5 min 40 n2 De-ionised water with 447 Crude A 5 100 -> 70 -> 65 Yes 5 min 40 n2 co 2 De-ionised water with 448 Crude A 20 100 -> 85 -> 65 Yes 5 min 40 n2 co 2 De-ionised water with 449 Crude A 20 100 -> 85 -> 65 Yes 5 min 40 n2 co 2 De-ionised water with 450 Crude A 5 100 -> 70 -> 65 Yes 5 min 40 n2 CO] 451 CrudeB De-ionised water 5 100 -> 70 -> 65 Yes - 5 min 40 n2 452 Crude B De-ionised water 20 100 -> 85 -> 65 Yes - 5 min 40 n2 De-ionised water with 453 Crude B 5 100 -> 70 -> 65 Yes - 5 min 40 n2 C02 De-ionised water with 454 Crude B 20 100 -> 85 -> 65 Yes - 5 min 40 n2 CO] Synthetic formation 512 Crude C 84 85 -> 85 -> 1 No 35 Natural gas water Synthetic formation 513 Crude C 10 85 -> 85 -> 75 No 35 Natural gas water Synthetic formation 514 CrudeC 45 85 > 85 > 40 No 35 Natural gas water Synthetic formation 515 Crude C 10 85 -> 85 -> 75 No 35 Natural gas water Synthetic formation 516 Crude C 45 85 -> 85 -> 40 No 35 Natural gas water Synthetic formation 517 Crude C 84 85 -> 85 -> 1 No 35 Natural gas water Synthetic formation 518 Crude C 84 85 -> 85 > 1 No 35 Natural gas water with C02 Synthetic formation 519 CrudeC 10 85 -> 85 -> 75 No 35 Natural gas water with C02 Synthetic formation 520 Crude C 45 85 -> 85 -> 40 No 35 Natural gas water with C02 Synthetic formation 521 Crude C 10 85 -> 85 -> 75 No 35 Natural gas water with C02 Synthetic formation 522 Crude C 45 85 -> 85 -> 40 No 35 Natural gas water with C02 Synthetic formation 523 Crude C 84 85 -> 85 -> 1 No 35 Natural gas water with C02 698 Crude C De-ionised water 10 75 -> 75 -> 65 Yes 2 min 20 Natural gas

6 7

699 Crude C De-ionised water 10 75 _> 75 -> 65 Yes - 2 min 30 Natural gas 700 Crude C De-ionised water 10 75 -> 75 -> 65 Yes - 2 min 40 Natural gas 701 Crude C De-ionised water 10 75 -> 75 -> 65 Yes 2 min 50 Natural gas 702 Crude C De-ionised water 10 75 -> 75 -> 65 Yes - 2 min 60 Natural gas 703 Crude C De-ionised water 10 75 -> 75 > 65 Yes - 2 min 70 Natural gas De-ionised water with 704 Crude C 10 75 -> 75 -> 65 Yes 2 min 20 Natural gas C02 De-ionised water with 705 Crude C 10 75 -> 75 -> 65 Yes - 2 min 30 Natural gas C02 De-ionised water with 706 Crude C co 2 10 75 -> 75 -> 65 Yes - 2 min 40 Natural gas De-ionised water with 707 Crude C 10 75 -> 75 -> 65 Yes - 2 min 50 Natural gas C02 De-ionised water with 708 Crude C 10 75 -> 75 -> 65 Yes - 2 min 60 Natural gas C02 De-ionised water with 709 Crude C co 2 10 75 -> 75 -> 65 Yes - 2 min 70 Natural gas De-ionised water with 710 Crude A 10 75 -> 75 -> 65 Yes 2 min 20 Natural gas C02 De-ionised water with 711 Crude A co 2 10 75 -> 75 -> 65 Yes - 2 min 30 Natural gas De-ionised water with 712 Crude A 10 75 -> 75 -> 65 Yes - 2 min 40 Natural gas C02 De-ionised water with 713 Crude A co 2 10 75 -> 75 -> 65 Yes 2 min 50 Natural gas De-ionised water with 714 Crude A 10 75 -> 75 -> 65 Yes 2 min 60 Natural gas CO, 715 Crude A De-ionised water 10 75 > 75 -> 65 Yes - 2 min 40 Natural gas 716 Crude A De-ionised water 10 75 -> 75 -> 65 Yes - 2 min 50 Natural gas 717 Crude A De-ionised water 10 75 > 75 -> 65 Yes - 2 min 60 Natural gas

7 8

Crude oil characteristics of Crudes A, B and C

Table 2: Basic characterisation of "dead” crude oil A 5 Density, g/cm3 0.94 Interfacial tension, mN/m 27.4 Water content, % 0.04 Molweight, g/mol 333 Asphaltene content, % 4.7 Viscosity, mPa-s, Shear 100 1 Is, 25/40 °C 389/ 141 Total Acid Number (TAN) 2.3

Table 3: Basic characterisation of "dead" crude oil B

Density, g/cm3 0.92 Interfacial tension, mN/m 24.8 Water content, % 0.06 Molweight, g/mol 279 Asphaltene content, % 0.9 Viscosity, mPa-s, Shear 100 1/s, 25/40 °C 58/27 Total Acid Number (TAN), % 3.1

10 Table 4: Basic characterisation of "dead" crude oil C

Density, g/cm3 0.84 Interfacial tension, mN/m 27.4 Water content, % 0.05 Molweight, g/mol 298 Asphaltene content, % 4.2 Viscosity, mPa-s, Shear 100 1/s, 25/40 °C 64/ 14 Total Acid Number (TAN), % 0.36

15

8 9

Examples

The high pressure separation process is governed by several sub-processes among them both mechanical and compositional effects which seem to be decisive for a successful result. Mechanical effects involve both pressure gradients over inlet

5 chokes (ap„„) and gas release in the separator (AP,ep ). The former parameter should determine the level of the energy dissipitation for the dispersive system and hence the droplet size (and distribution) for the dispersed droplets. A large gradient over the inlet choke should render small droplets. The level of APsep = (Pinl( ,t - Pfmai) refers to the gas release intensity in the separator, and the final separation will take 10 place below the bubble point of the gas mixture. The compositional state of the crude oil is also influenced by several external parameters among them the start pressure of the mixture comprising the water phase and the oil phase and the final separator pressure at which the separation takes place. It is obvious that the start pressure and the amount of light components being mixed with the original crude 15 components will influence the solubility of heavy components and hence their level of association. The direct conditions in the separator, such as the composition of the separator gas and residence time, will influence the separation process. If the separator is run as a batch separator, i.e. where the separator is filled, the emulsion is allowed to separate. Some of the experiments in this series are performed in a 20 way that the emulsion was kept at an inlet pressure for 2 or 5 minutes after which the pressure was reduced by a degassing process. Under this period of time, the emulsion will undergo a sedimentation process if the droplet size is large enough. The effect of propagating gas bubbles should be larger if the major part of the water droplets is compressed to a dense packed region. Finally, the original composition 25 of the crude oils themselves will of course influence how a gas release can improve the destabilization process. Small nanosized asphaltene particles will give a high level of stability since the protecting w/o interface will show a high rigidity under these conditions.

In the first series of experiments, two different candidates of North Sea crudes are 30 tested, crude oils named Crude A and Crude B. They have both provided very stable water in crude oil emulsions, although the stabilizing mechanisms can be different as revealed by the composition shown in Tables 2 and 3. Crude A is a heavy crude with a high content of asphaltenes, while Crude B is a very acidic crude with a high amount of naphthenic acids. Thus, a value of TAN (Total Acid Number) of Crude B 35 is higher than for Crude A. In addition a model system consisting of Crude A (1 % of Crude A in Exxsol D-80) is tested. Essential for the discussion is that these compositions are run through pressure reductions where the initial pressure (100 bar) is reduced to the separator pressure 65 bar. The separator was run as a batch separator in the sense that the emulsions were placed in the separator for 5 minutes 40 before the final pressure was adjusted as a gas release from 65 bar to 1 bar. Under

9 10

these experimental conditions, Figures 1 and 2 summarize the separation of water from the corresponding emulsions. The dispersed aqueous phase is either pure water

or water saturated with CO2.

The system where 1 % of Crude A is diluted into Exxsol-D 80 and combined with

5 40 % water with and without CO2 is presented in Figure 1 The separation level of the model emulsions is much lower, but in this case an acceleration of the gas

release on the separation of water is clearly seen. The effect of CO2 is clearly positive and selective.

For Crude A, Figure 2 reveals the effect of the pressure gradient over the inlet

10 choke to the separator and the addition of CO2. Without CO2, the pressure gradient (20 or 5 bar) seems to have a minor influence on the separation process. The emulsions are stable, and only 20-25 % of water is separated after 20 minutes.

However, in most of the cases the separation is accelerated by the release of CO2 after 5 minutes. With a pressure gradient of 5 bar, there is no significant difference

15 to the samples without CO2. However, the large effect is seen for the emulsion with

a AP = 20 bar and an aqueous phase saturated with CO2. For these emulsions one observes that as long as the separator pressure is kept at 65 bar (i.e. for 5 minutes), the level of separation is low or almost negligible. After 5 minutes in the separator, i.e. when the pressure reduction takes place, between 50 and 60 % of the water 20 phase will separate within 1 -2 minutes, and after 15 minutes, 90 % of the emulsion has broken and separated into the original components. This is a remarkable result for a crude oil, which has proven to give very stable emulsions that are resistant to both chemical and mechanical treatment.

Figure 3 gives the separation sequence for emulsions based on Crude B. 25 Characteristically; some separation will for this system take place already at 65 bar. The water separation is at a level of at least about 10 %. However, when the gas is released after 5 minutes, the separation profile changes dramatically. All the graphs (with different pressure drops over the inlet choke to the separator and with and

without CO2 in the water phase) show an acceleration of the resolution of water. 30 However, the selectivity between the different emulsions is lost. Large effects are seen both with and without C02 in the aqueous phase and with small and larger pressure gradients over the inlet choke to the separator. One cannot with certainty relate the increased separability to carbon dioxide release.

In all the experiments above, N2 is used as separator gas.

35 The next series of experiments (Figure 4) refers to Crude C as the crude oil and a natural gas as the separator gas phase. The composition of Crude C is shown in Table 4. These series of experiments are performed according to varying separator pressures. In other words, the pressure after the inlet choke to the separator is also

10 11

the final separator pressure. Hence the separator pressure was kept at levels of 1,40 and 75 bar, and most of the pressure related effects will take place over the inlet choke to the separator. For instance when the separator pressure is below 60 bar,

considerable amounts of the CO2 will be released already over the inlet choke to the 5 separator. The separation picture as a whole is rather «patchy», but after a residence time of 4 minutes, the best result is accomplished for a mild mixing (AP =10 bar) and a corresponding high separator pressure. This picture seems to hold both for

systems with and without CO2. Obviously, the droplet formation conditions seem to be the decisive factor if a batch separator is run according to these specifications. It 10 should be noted that no pressure release is conducted in the separator.

The final Figures 5 and 6 view the influence of the water content for Crude C and

Crude A on the separation profile at 1 bar and with and without CO2. Figure 5 presents the separation profile of Crude C for the water contents between 20 and 70 volume %. It can be seen that the induction time is high, i.e. after 4 minutes there is 15 only one significant solution of water at the level of 15 %. This accounts for the composition with 70 % of water saturated with COz. The highest overall separation

(approximately 45 %) is accomplished for 70 % of water without CO2. In general, one can conclude that crude oil-based emulsions (Crude C) respond very badly to the separation conditions disclosed in Table 1 For the emulsions based on the 20 Crude A, the situation seems to be more favorable with regard to the solution of water and oil. For higher water contents (i.e. > 30 %), the influence of the released

CO2 gas on the separation efficiency seems to be clear. Within a normal residence time of 4 minutes, almost 80 % of the aqueous phase has separated if the total water content is as high as 60 %.

25

11 12

PATENT CLAIM

1 Process for separation of oil, water and gas in a separator by breaking of water-in-oil emulsions in a composition comprising a water, oil and optionally a gaseous phase, characterized by the following steps: 5 a) dissolving a gas comprising one or more components into said composition prior to said separator, wherein the amount of said water phase is at least about 1 weight- % based on total composition, b) introducing said composition into said separator, wherein pressure in said separator is of at least about 2 bar, and 10 c) reducing the pressure in said separator in order to facilitate the separation of oil, water and the optional gaseous phase.

2. Process according to claim 1, wherein said gas dissolved is selected from polar components which are water-soluble, preferably wherein said gas dissolved is

selected from CO2, N2, O2, and mixtures thereof, and more preferably wherein said 15 gas dissolved is C02.

3. Process according to claim 1, wherein said gas dissolved is selected from non-polar components which are oil-soluble, preferably wherein said gas dissolved is selected from H2, CH4, paraffin ’s, and mixtures thereof.

4. Process according to any of the preceding claims, wherein said gas is 20 dissolved directly into said water phase before said phases are mixed, or said gas is dissolved countercurrently into said composition before said composition is introduced into said separator.

5. Process according to any of the preceding claims, wherein said composition contains inorganic or organic particulate solids.

25 6. Process according to claim 5, wherein the amount of said inorganic or organic particulate solids is at least about 0,01 weight-% based on total composition, preferably wherein said amount is in the range of 5-15 weight-%.

7. Process according to claim 5 or 6, wherein said organic particulate solids are constituted by nanosized asphaltene, metal organic acid, like calcium naphthenate 30 particles, or wax particles, like heavy paraffin particles, or mixtures thereof.

8. Process according to claim 5 or 6, wherein said inorganic particulate solids are constituted by clay, precipitated metal salt or scale particles like calcium carbonate, barium sulphate, or mixtures thereof.

9. Process according to any of the preceding claims, wherein the pressure in 35 said separator is of at least about 60 bar.

12 13

10. Process according to claim 1, wherein said pressure in said separator is reduced to a value below the separation pressure of said gas in said composition.

11 Process according to claim 1, wherein said pressure in said separator is reduced to 1 bar.

5 12. Process according to claim 11, wherein said pressure in said separator is reduced only after said composition is maintained in said separator at a specific period of time.

13. Process according to claim 12, wherein the period of time is at least about 2 minutes, preferably wherein said period of time is in the range of about 2 to about 5 10 minutes, and more preferably wherein said period of time is about 4 minutes.

14. Process according to any of the preceding claims, wherein said separator is a gravitational separator.

15. Process according to any of the preceding claims, wherein said separator is a separator of batch type.

15 16. Process according to any of the preceding claims, wherein the amount of said water phase is at least about 10 weight-% based on total composition, preferably wherein said amount is at least about 30 weight-%, and more preferably wherein said amount is up to about 70 weight-%.

17. Process according to any of the preceding claims, wherein the pressure of 20 said composition entering said separator is at least about 10 bar, preferably wherein said pressure is at least about 30 bar, and more preferably wherein said pressure is at least about 70 bar.

13 14

SUMMARY

This invention relates to a process for separation of oil, water and gas in a separator by breaking of water-in-oil emulsions in a composition comprising water, oil and 5 optionally a gaseous phase.

14 15

Figure 1.

-B~ 444 Water w/C02, AP=20 —■—441 Water, AP=20 -9- 443 Water w/C02, AP=5

-#-442 Water a P=5

Gas release P -» 1 bar

Time [min]

15 Figure % water separated

2. Gas -*-446 -e- -41-445 -B- -A- -*-449

release

447 450 448

Water. Water Water Water. Water Water

I

w/CO,. w/C0 WCO%. w/C0 AP=5 AP=20 2 2 , ,

AP=5 AP=5 AP=20 AP=20

16 16

17

Figure 3.

-e- 454 Water w/OO^, AP=20 | - 452 Water. AP=20 | -e_ 453 Water w/CO;, AP-5 I 451 Water. AP“5

Gas release P —» 1 bar

0 5 10 15 20 Time [min]

17 Vi Figure

O 4.

tn

HI 3' (6

o

i M 19

Figure 5.

"698 Water 100 WC=20 % ~ 699 Water WC~30 % 90 -700 Water WO40% 80 "701 Water woso% - 702 Water I 70 WC=60% -703 Water 60 4 WC=70 % - 704 Water wCO, s 50 WC=20% -70S Water WC% B WC=30 % m 40 ' 706 Water vaC O, WC=40% a? - 707 Water w/CO, 30 WC=50 % "70S Water WCO, 20 WC=60 %

10

0■ 0 10 15 Time [min]

19 Figure % water separated

o 3 8 8 S sags 6. i------1------1------1------1

K> o O

S ^ S ^ O ~4 O O) O ui sg s y <5 y s y % Ul Paper III DESTABILISATION OF WATER-IN-CRUDE OIL

EMULSIONS BASED ON RECOMBINED OIL SAMPLES

AT VARIOUS PRESSURES

Inge Harald Auflem1, Arild Westvik2 and Johan Sjdblom 1,2,

1 Ugelstad Laboratory, Department of Chemical Engineering,

Norwegian University of Science and Technology,

N-7491 Trondheim, Norway

2 Statoil ASA R&D Centre, Arkitekt Ebbelsvei 10,

N-7005 Trondheim, Norway

Corresponding author (e-mail: [email protected] ABSTRACT

This paper describes how recombination of a crude oil with various gases (N2, CO2,

CH4, C2H6 and dry natural gas) influences the stability for the concomitant water-in ­ oil emulsions. The resulting separation behaviour for recombined samples is compared with the live crude oil sample for separation pressures equal to the bubble point of the live sample (15 bar) and below (10 and 1 bar). The experiments were performed by mixing the recombined oil samples with synthetic formation water by a pressure drop through a choke valve and into a vertical high-pressure separation cell.

Experimental parameters were separation pressure, pressure drop and water content.

The stability of the resulting water-in-oil emulsions was monitored visually as a function of time, and is discussed in terms of residual gas content, droplet sizes and emulsification conditions.

KEYWORDS: Recombination, water-in-crude oil emulsion, destabilisation, pressure, separation

INTRODUCTION

Destabilisation of crude oil based emulsions is a topic of continuous interest. At the

Norwegian Continental Shelf, subsea processing units will become increasingly important when exploring new fields and in "tail end" production of existing fields. In planning and designing such processes, the fluid properties are crucial for a correct choice of design and dimensioning of the process units. Test procedures that are performed in order to evaluate the emulsion stability have traditionally been based on

2 atmospheric bottle tests. If the crude oil has been exposed to air, oxidation will significantly change the crude oil properties as shown earlier by Ranningsen et. al.

(1995) [1] and Sjoblom et. al. (1995) [2], It is motivated to question how representative such samples are as candidates for the fluid streams at subsea conditions. One severe drawback in carrying out the bottle tests is the lack of pressure dependence. As an alternative, field tests at real conditions can be used to evaluate the crude oil behaviour. In these cases the oil will obviously be highly representative.

Unfortunately, high expenses in connection with production shutdown, limited possibilities for parameter variation and lack of experimental instrumentation prevent this from being a practical test method. At Statoil R&D Centre they have tried to solve the problem with relevant test samples by constructing a separation rig that can operate from 1-200 bar and -7 to 180 °C. This rig has been presented in earlier papers by Johnsen et. al. [3], Sjoblom et. al. [4] and Auflem et. al. [5], It is normal that the crude oils are delivered as dead (degassed) samples, since true bottom hole samples are expensive and difficult to obtain. As a consequence one could, in order to study the emulsion properties at elevated pressures, undertake a recombination with a gas mixture to the pressure and temperature in question. However, the problem with a relevant recombination of dead oil samples still remains to be solved. In this paper we present demulsification data on a North Sea Crude oil that has been recombined with a variety of gases (methane, ethane, carbon dioxide, nitrogen) and a natural gas mixture from Statoil Kars to. The separation behaviour for recombined samples is compared with the live crude oil sample for separation pressures equal to the bubble point of the live sample (15 bar) and below (10 and 1 bar).

3 EXPERIMENTAL

Chemicals

A crude oil sample from the North Sea with a bubble point of 15 bar at 50 °C was used in the experiments. Characterisation data for the degassed crude oil are given in table 1. The water phase used was synthetic formation water based on data from the oil reservoir, consisting mainly of chlorides dissolved in distilled water. All fluids were kept at 50 °C at all times during the experiments. N2, C02, CH4, and C2H6 used for recombination were delivered by Hydrogas with a purity of 99.999, 99.999, 99.5 and 99.95 mol %, respectively. Dry natural gas (NG) with the following composition:

1.12 mol % C02, 1.30 mol % N2, 82.07 mol % CH4, 14.66 mol % C2H6, 0.76 mol %

C3H8 and 0.09 mol % C4+, was obtained from the Statoil Karsto NG plant.

Table 1. Summary of characterisation data for North Sea crude oil.

Crude oil properties Values

Density [g/ml] 0.913

Molecular weight [g/mol] 273.93

Asphaltenes [wt %] 0.6

Interfacial tension [mN/m] 24.2

Total Acid Number fmg KOH/11 2.95

Recombination of oil phases

The following three types of oil samples were used in the separation experiments: 1)

“Live” crude oil samples with a bubble point of 15 bar, 2) samples where the light end

4 molecules were removed from "live" samples by depressurisation to 1 bar, while allowing the evolving gas to evaporate (degassing), thereafter the oil was recombined to 15 bar with either N2, C02, CH4, C2H6 or dry natural gas, 3) samples that were degassed in the recombination cell to 10 or 1 bar, and thereafter mechanically repressurised to 15 bar by use of a piston pump and no addition of gas. All the recombination work upon the crude oil was performed in a standard 2-litre oxygen- free recombination cell at 50 °C with constant mixing. The recombined oil phases (15 bar) were allowed to equilibrate for another 24 hours, before being transferred into the sample cylinders on the high pressure high temperature rig (HPHT-rig. The oil and water phases were kept in sample cylinders B1 and B2, respectively, both at 50 °C and 20 bar, The separator cell was pressurised to the separation pressure with the recombination gas in use. For the case of “live” crude oil, the natural gas mixture was employed. A schematic overview of the HPHT-rig, is shown in figure 1.

In the experiments, the flows of water and oil undergo a pressure drop through a choke valve (VD1). The energy release creates more interfacial area between the oil and water phases, and thus a water-in-oil emulsion is formed. The flow passes into a transparent separator cell, where the separation of the various phases (water, oil, foam and emulsion) can be recorded visually as a function of time. The HPHT-rig is detailed described in “Encyclopedic Handbook of Emulsion Technology ” by Sjoblom

[4], The variables in the experiments performed were: repressurising gas, pressure drop over choke valve (AP), separator pressure (Psep ) and water content.

5 PI 3 P12 P11 PI 0

Figure 1. Schematic drawing showing the high-pressure high temperature separation rig at the Statoil

R&D Centre.

RESULTS AND DISCUSSION

Comparison of “live ” vs. recombined samples

Three different series of experiments with recombined and “live” oil samples are reported. The first one includes a pressure drop through the choke from 20 to 15 bar

(AP = 5 bar) as the emulsification step. In the second, AP = 10 bar, i.e. a pressure drop of 20 to 10 bar, and finally a series with a mixing step of 20 to 1 bar (AP = 19 bar) is discussed. The experiments are performed with a water content of 60 and 40 volume

%.

60 % water

Figure 2 shows that the “live” oil sample gave the best separation profile, with an almost complete water separation after approximately 2 minutes. The recombined samples (with CH4, C^Hg, CO2, N% and NG) started at a lower water separation and

6 reached approximately 90 % separation level at 8 minutes. Closest to the separation profile of the original “live” oil was the oil sample where the recombination was performed with ethane. It should be noted that the bubble point for the investigated

crude was 15 bar, and as the separation was performed at this pressure there was no

foam formation. Further, the pressure drop of AP = 5 bar (20 to 15 bar) provided a rather poor emulsification, and as a consequence the drops should be rather large.

> 60 Nitrogen Natural gm

Methane Ethane

Carbon dioxide

Time [min]

Figure 2. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 15 bar, 50 °C and

60 volume %, respectively.

The second series (figure 3) shows the separation profile at 10 bar separation pressure, i.e. substantially below the bubble point of the "live" oil. Also in this case the “live” oil gave the best separation of water with over 90 % separated at 4 minutes. The sequence from the previous figure was more-or-less reproduced, and most of the recombined systems reached “complete” separation after 8 minutes. The pressure

7 drop was now increased to AP = 10 bar, resulting in smaller droplets and more stable emulsions.

"Wo: - Nitrogen - Natural gas

- Methane • Ethane

Carbon dioxide 20 1H'

0 2 4 6 8 10 12

Time [man]

Figure 3. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 10 bar, 50 °C and

60 volume %, respectively.

The behaviour at 1 bar differed from the other two systems reported. As seen before, the “live” oil gave the best separation of water closely followed by systems recombined with N2, CTL, and natural gas. Emulsions with 60 % of water, where the oil phase was recombined with ethane or carbon dioxide, were very stable under these degassing experiments (see figure 4). No indications of water separation were seen during the first 10 minutes. Due to the large pressure drop from 20 to 1 bar, rather small droplets and stable emulsions were expected. PvT flash simulation showed that, at this pressure, most of the gas from the recombination was evaporated. This was visually confirmed as shown in figure 5, by the formation of a foam layer with a lifetime of a few minutes.

8 too 1 J,------——------*

"BwToa Nitrogen

Natural gas

Methane Ethane

Carbon dioxide

0 ...-#---- ■ -* — •------*=------*- 0 Z 4 6 * 10 n Time [min]

Figure 4. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 1 bar, 50 °C and

60 volume %, respectively.

"* "live" oil Nitrogen

-B~ Natural gas

Methane Ethane 200 ' Carbon dioxide

Time [mm1

Figure 5. Foam stability with water-in-oil emulsions based on recombined oil samples displayed as foam height vs. time. Separation pressure, temperature and total water content are 1 bar, 50 °C and 60 volume %, respectively.

9 40 % water

Generally the separation process was slowed down at lower water content, and a rest emulsion with up to 30 % of water was observed. The “live” oil gave the fastest separation. Samples recombined with ethane reproduced the “live” conditions rather well, in experiments where the separation pressure was equal to the bubble point pressure (figure 6). All recombined systems showed a separation of water of about 70

%, however, 10 minutes or more was required to achieve this.

"live" Oil Nitrogen Natural gm.

Methane Ethane

Carbon dioxide

0 2 4 6 8 10 12

Time [min]

Figure 6. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 15 bar, 50 °C and

40 volume %, respectively.

The same trend was reproduced with separation at 10 bar (figure 7). The time needed for separation of 70 % of water, “live” crude and “ethane” recombined crude, was now in the region of 8-10 minutes, and the amount of rest emulsion was higher than at

15 bar separation pressure.

10 "live" oil Nitrogen ~JB~ Natural ga& Methane- Ethane

Carbon dioxide

Time [nun

Figure 7. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 10 bar, 50 °C and

40 volume %, respectively.

As for the 60 % water case the emulsions were very stable for degassing to 1 bar separation pressure (figure 8). Obviously the emulsification conditions (AP =19 bar) produced relatively small droplets. An exception was the “live” system that provided a separation between 30-40 % of water. Another observation for the 40 % case (figure

9) was that the foam formation was increased in comparison with 60 % case. This was probably due to the increased oil to water ratio of the fluid stream.

11 "live” oil ™*' “ Nitrogen

~B~" Natural gas Methane Ethane

“*"* Carbon dioxide

Time [min]

Figure 8. Separation curves for water-in-oil emulsions based on recombined oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 1 bar, 50 °C and

40 volume %, respectively.

( —4* "live" oil " ♦" Nitrogen

“B~ Natural gas

Met hane Ethane

Carbon dioxide

Figure 9. Foam stability with water-in-oil emulsions based on recombined oil samples displayed as foam height vs. timeSeparation pressure, temperature and total water content are 1 bar, 50 °C and 40 volume %, respectively.

12 Mechanical repressuring of degassed samples

For the mechanical recombined samples, i.e. samples depressurised to 10 and 1 bar, before mechanically being repressurised to 15 bar, the following trends could be observed:

60 % water

At a separation pressure of 15 bar (AP = 5 bar) it was difficult to distinguish between the different samples. The separation was rapid, and after 2 minutes almost complete separation had occurred (around 90 % water resolved). When the separation pressure was reduced to 1 bar (AP =19 bar), the bulk of the water still separated. However, it took a longer separation time, between 2 to 6 minutes, to reach the same water resolution as for the lower energy mixing case.

Pb = IS bar f live”)

Pb * i 0 bar (degassed)

Pb - 1 bar (dega&aed) 30 '

20 "

10

0 2 4 * S 10 12 Time [min]

Figure 10. Separation curves for water-in-oil emulsions based on mechanically repressurised oil samples displayed as water resolved vs. time. Separation pressure, temperature and total water content are 1 bar, 50 °C and 60 volume %, respectively.

13 The sample degassed to 1 bar displayed the fastest separation behaviour, trailed by the

sample degassed to 10 bar and then the "live" (15 bar) oil sample as shown in figure

10.

- Pb = 15 bar ('live'*)

Ph - 10 bar (degassed)

Pb — 1 bar (degassed)

Time [min

Figure 11. Separation curves for water-in-oil emulsions based on mechanically repressurised oil

samples displayed as water resolved vs. time. Separation pressure, temperature and total water content

are 15 bar, 50 °C and 40 volume %, respectively.

40% water

At a separation pressure of 15 bar (AP = 5 bar) the separation of water was considerably less than for the equivalent 60 % water case, and a rest emulsion containing 20-30 % water was remaining after 6 minutes. The sample degassed to 1 bar gave the fastest separation, followed by the sample degassed to 10 bar and then the "live" (15 bar) oil sample, as shown in figure 11 When the separation pressure was decreased to 1 bar (AP = 19 bar) the separation was further reduced. The resulting

14 rest emulsion contained between 30-60 % of water, and once again the sample

degassed to 1 bar showed the fastest separation, while the sample degassed to 10 bar

and the "live" (15 bar) sample displayed slower separation profiles.

The emulsion stability for the mechanically recombined samples seemed, to depend on the degassing pressure of the "live" sample, i.e. the content of gas present in the oil phase. The reason for this could be that the samples with the lowest gas content, as a result of the higher viscosities, experienced a less complete emulsification. This could result in a higher number of relatively large droplets, which would separate within the first few minutes, while the rest emulsion maintained the same stability as for the samples with higher amount of gas remaining. The amount of water influences the emulsion stability together with the pressure gradient over the choke. Smaller amounts of water (40 %) and large AP (19 bar) gave remarkably higher emulsion stability in comparison with 60 % of water and 5 bar pressure drop. Most likely the drop size distribution was quite different for these samples with much smaller droplets for 40 % of water and high AP.

CONCLUSIONS

Water-in-oil emulsions based on a "live" North Sea crude oil, were shown to separate faster and more complete, than emulsions based on recombined samples of the same crude oil. Increased water content, or reduced mixing through smaller pressure drop into the separator, resulted in faster and more complete separation of the emulsions for both "live" and recombined samples.

15 For the mechanically recombined samples, the "live" (15 bar) crude oil sample gave the largest amount of foam for both the 60 % and 40 % cases, a smaller amount was obtained for the sample degassed to 10 bar and none for the 1 bar sample. The emulsion stability was shown to depend on the degassing pressure of the sample, where a lower residual gas content gave a lower stability of the emulsions.

ACKNOWLEDGEMENTS

Inge Harald Auflem would like to acknowledge the technology program FLUCHA II financed by The Research Council of Norway and the oil industry for a PhD grant.

Statoil is acknowledged for the kind permission to publish the results.

REFERENCES

1. Renningsen, H P., Sjoblom, J.; Mingyuan, L. Water-in-crude oil emulsions from the Norwegian continental shelf 11. Ageing of crude oils and its influence on the emulsion stability. Colloid and Surfaces. 1995, 97,119-128.

2. Sjoblom, J., Mingyuan, L.; Christy, A.A.; Ronningsen, H.P. Water-in-crude oil emulsions from the Norwegian continental shelf 10. Ageing of the interfacially active components and the influence on the emulsion stability. Colloid and Surfaces

A: Physicochemical and Engineering Aspects. 1995, 96, 261-272.

3. Johnsen, E.E., Westvik, A. Separation of live oil and brine at high pressure and temperature in laboratory. Effects of pressure drop and separation pressure, in The

16 Second International Conference on Petroleum Phase Behavior and Fouling, AIChE

Copenhagen, Denmark, 2000.

4. Sjoblom, J.; Johnsen, E.E., Westvik, A.; Ese, M.H.; Djuve, J.; Auflem, I.H.;

Kallevik, H. Demulsifiers in Oil Industry, in Encyclopedic Handbook of Emulsion

Technology, J. Sjoblom, Editor., Marcel Dekker: New York, 2001; 595-619.

5. Auflem, I.H.; Kallevik, H.; Westvik, A., Sjoblom, J. Influence of Pressure and

Solvency on the Separation of Water-in-Oil Emulsions form the North Sea. Journal of

Petroleum Science and Engineering, 2001, 31(1), 1-12.

17 Paper IV Journal of Colloid and Interface Science 247, 342-350 (2002) 0 doi: 10.1006/jcis.2002.8122, available online at http://www.idealibrary.com on IB E

A Quartz Crystal Microbalance Study of the Adsorption of Asphaltenes and Resins onto a Hydrophilic Surface

Pontus Ekholm,* Eva Blomberg,* *1 Per Claesson,* Inge Harald Auflem.f t Johan Sjdblom.f i and Anna Komfeldt§

* Department of Chemistry, Surface Chemistry, Royal Institute of Technology, Drottning Kristinas vdg 51, SE-100 44 Stockholm, Sweden, and Institute for Surface Chemistry, Box 5607, SC-114 86 Stockholm, Sweden ; fDepartment of Chemical Engineering, The Norwegian University of Science and Technology (NTNU). N-7491 Trondheim. Norway; \Statoil R&D, N-7005 Trondheim, Norway: and %ABB Corporate Research, SE-72J 78 Vdsteras, Sweden

Received July 17, 2001: accepted November 24, 2001; published online February 5,2002

certain pressure and temperature referred to as “the onset point, ” The adsorption of extracted and purified samples of asphaltenes where a precipitation of asphaltenes is observed. Before the pre ­ and resins onto gold surfaces has been studied as a function of cipitation, a stepwise association of the asphaltene molecules bulk concentration using a quarts crystal microbalance with dis ­ will take place. The final precipitation is due to a strong attraction sipation measurements (QCM-D). With this device, which works between the colloidal particles and the formation of agglomer ­ equally well in transparent, opaque, and nontransparent samples, ates. The presence of hydrophobic dispersants can stabilize the the adsorbed amount is measured through a change in resonant fre ­ quency of the quartz oscillator. The measured change in dissipation colloidal asphaltene aggregates and prevent large-scale agglom ­ reports on changes in layer viscoelasticity and slip of the solvent at eration (7). Hence, the final point of precipitation will depend on the surface. The results show that the adsorbed amount for resins the presence of oil-soluble dispersants like resins and the nature from heptane corresponds to a rigidly attached monolayer. The ad ­ and structure of the asphaltene molecules (8). sorbed amount decreases with increasing amount of toluene in the This precipitation might occur in the formation and cause solvent and is virtually zero in pure toluene. Asphaltenes, on the a plugging of the pore systems giving rise to process prob ­ other hand, adsorb in large quantities and the mass and dissipation lems. Obviously, the interplay between resins, napthenic acids data demonstrate the presence of aggregates on the surface. The ag ­ and asphaltenes is of crucial importance for the stability of the gregates are firmly attached and cannot be removed by addition of whole system. Acidic crudes (with a high amount of naphtenic resins. On the other hand, resins and asphaltenes associate in bulk acids) are often a nightmare for refinery people due to corro ­ liquid and the adsorption from mixtures containing both resins and sion problems. The actual mechanism of the corrosion in an asphaltenes is markedly different from that obtained from the pure components. Hence, we conclude that preformed resin aggregates oil environment is not clarified in detail. Another problem with adsorb to the surface. These results are compared and discussed the napthenic crudes is their reactivity with multivalent ions in relation to adsorption from crude oil diluted in heptane/toluene (Ca 2+ Ba 2+ Mg 2+, etc.) and the concomitant formation of acid mixtures. e Z0«2 Bmur Science (USA) soaps (9). These are normally formed at w/o interfaces as small Key Words: quartz crystal microbalance (QCM); asphaltene; nuclei that undergo a growth to form particles. A large-scale resin; adsorption. deposition of such particles in a separator will cause an opera ­ tional shutdown for a mechanical or chemical cleaning. Further, together with an aqueous phase these interfacially active com ­ INTRODUCTION ponents will form and stabilize water-in-oil emulsions. These emulsions normally cause severe production and quality prob ­ The presence of indigenous oil-soluble surfactants like as ­ lems. In order to optimise the destabilisation process, it is of phaltenes, resins, and naphtenic acids, will cause problems along crucial importance to understand the details of the stabiliza ­ the whole production chain from the reservoir to the refinery tion of the w/o emulsions. This question is intimately correlated (1-3). For asphaltenes, this is often due to the strong depen ­ with the formation of a rigidly attached interfacial film con ­ dence of the state of the asphaltenes at different pressures (4- taining components like asphaltenes and resins (10-18). In the 6). Normally at high pressures in the reservoir, the asphaltenes buildup of such a film, problems like adsorption, competitive ad ­ are in a monomeric state and dissolved in the monophasic sorption, deposition, of particles, etc., will all be of imperative crude oil. With a reduction in pressure, the density of the oil importance. In this paper, we have utilised a relatively new tech­ phase is reduced, and the oil phase becomes a poorer solvent for nique, i.e., dissipative quartz crystal microbalance (QCM-D) to the asphaltenes. As the pressure is further reduced, there exists a reveal the steps in the formation of these interfacial structures, and their relevance for the stabilisation of w/o emulsions. To our

1 To whom correspondence should be sent. Fax: +46 8 20 89 98. E-mail: knowledge, this technique has not previously been employed for [email protected] . studying similar systems.

0021-9797/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved. ADSORPTION OF ASPHALTENES ONTO A HYDROPHILIC SURFACE 343

However, previously, Acevedo et al. studied the adsorption attached, with no slip or inelastic deformation in the added mass of asphaltene and resin fraction onto silica particles and as ­ due to the oscillatory motion. The last condition is valid when phaltene particle by UV spectroscopy (solute-solid adsorption the frequency decreases in proportion to the true mass of the isotherms) (19, 20), as well as the adsorption of asphaltene on adsorbate with no change in energy dissipation. Variations in silica and glass plates using photothermal surface deformation the energy dissipation upon adsorption thus reflect the energy method (20, 21). From the studies, they could conclude that dissipation in the adlayer or at its interfaces. For some data re ­ the adsorption of some types of asphaltene resulted in multi­ ported in this investigation the change in energy dissipation is layer formation and that the adsorption increased slowly with significant and the mass change, deduced from Eq. [1], underes ­ time (ranging from 10 to 100 h). For other types of asphal­ timates the real adsorbed mass. We note that the mass detected tene, a Langmuir type of adsorption was obtained. A possible with the QCM-D device includes any change in the amount of correlation between these differences and the tendency for the solvent that oscillates with the surface. This quantity may be asphaltene to precipitate from the crude oil was suggested. For significant for extended layers and thus one expects to obtain a the resins, they observed that multilayer formation was appar­ higher value for the adsorbed amount from QCM-D studies than ent on an asphaltene surface in heptane and that the adsorption from, e g., ellipsometry. However, ellipsometry cannot be used decreases with the toluene content in the solvent (19). for the systems studied here due to the nontransparent nature of the solutions, which, on the other hand, is no problem when a QCM-D device is used. METHOD

Quartz Crystal Microbalance The Dissipation Factor

The dissipative quartz crystal microbalance, QCM-D, consists A film that is “soft ” (viscoelastic) will not fully couple to the of a thin plate of a piezoelectric quartz crystal, sandwiched be ­ oscillation of the crystal. The dissipation factor is proportional tween a pair of electrodes, it measures simultaneously changes to the power dissipation in the oscillatory system and can give in resonance frequency, /, and dissipation, D (the frictional and valuable information about the rigidity of the film: viscoelastic energy losses in the system), due to adsorption on a crystal surface. / is measured before disconnecting the driv ­ ing oscillator, and D is obtained by suddenly disconnecting the ^dissipated D = [2] driving field and recording the damped oscillating signal as its 2rr/t stored vibration amplitude decays exponentially. Mechanical stress causes electric polarization in a piezoelec ­ tric material. The converse effect refers to the deformation of Here £dissipated is the energy dissipated during one oscillation the same material by applying an electric field. Therefore, when and £st0red is the energy stored in the oscillating system. Hence, an AC voltage is applied over the electrodes the crystal can be the measured change in D is due to changes in contributions made to oscillate. Resonance is obtained when the thickness of from, e.g., slip and viscous losses. For QCM measurements in the plate is an odd integer, n, of half wavelengths of the in­ liquids, the major contribution to D comes from frictional (vis­ duced wave, n being an integer since the applied potential over cous) losses within the liquid contacting the crystal. According the electrodes is always in anti-phase. If something is adsorbed to Stockbridge (23), the shift in dissipation factor in a liquid onto the crystal, it can be treated as an equivalent mass change environment is of the crystal itself The increase in mass, Am, induces a propor ­ tional shift in frequency, A/. This linear relationship between Am and A f was for the first time demonstrated by Sauerbrey JL /AH [3] (22) in 1959: Pqtq V 2tt/ ’

where i?i and pi are the viscosity and density of the fluid, re ­ fo n 2/02 n n spectively, and tq and pq are the thickness and the density of the quartz plate. When the adsorbed film slips on the electrode, where pq and vq are the specific density and the shear wave veloc ­ frictional energy is created that increases the dissipation. Fur­ ity in quartz respectively; tq is the thickness of the quartz crystal, thermore, if the film is viscous, energy is also dissipated due and /o the fundamental resonance frequency (when n= 1). For to the oscillatory motion induced in the film (internal friction the crystal used m these measurements the constant C has a value in the film). Hence, a rigid adsorbed layer gives no change in of 17.8 ng cm-2 Hz"1 The relation is valid when the follow ­ dissipation while a loose layer gives a dissipation increase. ing conditions are fulfilled: (i) the adsorbed mass is distributed For a more thorough guide through the theory behind the evenly over the crystal, (ii) Am is much smaller than the mass quartz crystal microbalance technique, see Michael Rodahl ’s of the crystal itself (< 1%), and (iii) the adsorbed mass is rigidly (24) and Fredrik Hook ’s (25) doctoral theses. 344 EKHOLM ET AL.

EXPERIMENTAL vent baseline was tested and accepted as stable when the signal fluctuations was less than ±5 Hz for a time period of half an hour. Instrumentation When this was achieved, the sampling could proceed and the frequency shifts caused by adsorption were measured relative The apparatus used in this study is a QCM-D device from to this baseline reference. Normally the solutions are injected Q-Sense (Gothenburg, Sweden). Figure 1 shows a schematic il­ lustration of the whole measurement system. The quartz crystal into the measurement cell by a syringe in the following way: oscillators were of AT-cut family with approximately 100-nm- (i) first by filling the temperature loop (T loop) to achieve the thick gold electrodes evaporated onto the crystal surface. They same temperature as in the measuring chamber, (ii) next, by had a diameter of 14 mm and a shear oscillation with a funda ­ injecting the sample into the cell after ~5 min. In the rebuilt mental frequency of 5 MHz. Originally (26), the cell is mounted device used here, T-loop thermostating is not relevant due to on a Peltier element that gives an accurate temperature con ­ the uncovered chamber and consequently the signals often take trol ( ±0.02°C) in the measurement chamber. In this study, the some minutes to stabilize. The measurements were made using instrument was rebuilt by removing the lid and the flow stop ­ both the fundamental frequency (5 MHz) and the third overtone pers. This was done because of difficulties in keeping the pres ­ frequency (15 MHz). As the third overtone (n = 3) shear wave is sure constant when working with volatile organic solvents. The not reaching as far from the surface as the fundamental (n = 1) solution to the pressure problem was to enclose the whole sys­ shear wave, it is more sensitive to the surface region. tem with closed bottles at the outflow (one from the temperature loop and one from the cell) and control the flow by a three-way Interpretation of Graphs valve. However, by solving the pressure problem we created a In QCM-D measurements one obtains information about the temperature problem as a result of the uncovered measurement shifts in the frequency and the dissipation factor. In some figures, chamber. To minimize the temperature flux from the system, the there are two separate y-axes and one t axis. The X-axis is the surrounding room was temperature controlled and all solutions time, the left-sided y-axis is the frequency shift, and the right­ were stored in the same room. The temperature variations in sided y-axis is the change in dissipation factor. A frequency shift the room were monitored to be in the range of ±0.5°C, and the of -10 Hz for the overtone frequency (15 MHz) corresponds to temperature in the chamber was assumed the same. an adsorption of 0.59 mg/m 2 as calculated from Sauerbrey rela ­ tion. When working with the crude oil fraction there is a temper ­ Sample Preparation ature and pressure stabilization sequence before the new level is obtained. This is seen in the graph as nanrow noisy vibrations Before compiling the measurement chamber, every compo ­ before a stable line is obtained. nent was cleaned in ethanol (Kemetyl, 99.5%). In addition, the surface of the crystals was washed in ethanol and 2% Heilman Extraction Procedure for Asphaltenes and Resins from Ex™ before measurements. The stock solutions (around Crude Oil 20 000 ppm) were prepared by dissolving the appropriate amount of dried asphaltenes or resins in toluene (Merck, 99.5%), The fractions used in this study were extracted from a North n-heptane (Merck, 99.5%) or mixtures of the two at room tem ­ Sea crude oil according to the following procedure: The as ­ perature. To ensure dissolution the solutions were left in an ul­ phaltenes were precipitated by gently mixing crude oil and trasonic bath until the aggregates were dispersed. Thereafter the n-pentane (1:10 g/mL) at room temperature for 24 h. This solutions were diluted to the appropriate concentration. Before mixture was then filtrated through a microfilter (pore diameter studying the adsorption, one has to flush the cell with pure sol ­ 0.45 p,m) and the precipitated asphaltenes were dried at 60°C vent and check for a stable baseline. For each experiment the sol- under No atmosphere until constant weight (3 h). Two grams of Silanol particles (size 55-105 /im in diameter) per milliliter of crude oil was then added to the supernatant to adsorb the resin fraction (24 h). In order to wash the particles the mixture was stirred with n-pentane and centrifuged at 3000 rpm for 3 min and the supernatant was removed. This was repeated until the supernatant was colorless. To desorb the resins the particles were mixed in a blend of 7% methanol in dichloromethane: this was repeated until the supernatant was colorless, and the supernatant was collected and filtrated. The bulk of the solvent was removed in a rotary evaporator at 70°C, and the last part was removed in plate under a water bath (70°C) under No atmosphere. The asphaltene and resin fractions as well as the crude oil itself FIG. 1. A schematic overview of the QCM-D cell together with an illustra­ went through a series of standard characterization methods. The tion of the quartz crystal covered with gold electrodes. obtained data is summarized in Table 1. ADSORPTION OF ASPHALTENES ONTO A HYDROPHILIC SURFACE 345

TABLE 1 An Overview of the Characterization Data for the North Sea Crude Oil, Obtained from Various Measuring Techniques

Property Measuring method Values 2.5

Molecular weight Crude oil (g/mol) Cryometer 226 Asphaltenes (g/mol) Cryometer (pentane precipitation 1:10) 865 Resins (g/mol) Cryometer (pentane precipitation 1:10) 319 Viscosity (mPa s) 20° C Rheometer (shear 1000 s 1) 10.5-13 arc Rheometer (shear 1000 s-!) 3.5-4.0 Density 20°C Densiometer 0.8545 arc Densiometer 0.8387 Asphaltenes (wt%) SARA (hexane) 2.9 Resins (wt%) SARA (trichloromethane) 13.8 Sulfur (wt%) IR 0.26 Concentration (ppm) Interfacial tension (mN/m) Surface tensiometer 27.4 Water content (wt%) Karl Fischer 0.0456 FIG. 3. Adsorption isotherm for resins onto a hydrophilic gold surface as a function of resin concentration in an n-heptane solution. The solid and dashed lines represent two different sets of measurements. For more details, see the text below. RESULTS AND DISCUSSION

Adsorption Study—Resins This suggests that the adsorbed resins form a rigidly attached monolayer on the surface and that individual resin molecule Resins redissolved in n-heptane. The adsorption of redis ­ rather than aggregates are present on the surface. The small solved resins in n-heptane onto a hydrophilic gold surface was increase in dissipation at the last injection (5000 ppm) might studied for concentrations between 100 and 5000 ppm. Figure 2 indicate that a second outer layer is forming, though this layer is shows that the adsorption increases with concentration, and at easily desorbed. The observation that it is impossible to desorb 5000 ppm, an adsorbed amount of 2.3 mg/m 2 (Fig. 3) is obtained. most of the layer within a reasonable time scale (<1 h), indi ­ Moreover, the signal stabilizes after a few minutes, which indi ­ cates that the binding of the inner layer is strong. In fact, this is cates that the resins adsorb, not deposit, onto the gold surface. observed for most polymer-surface systems and is a result of the By rinsing the cell with pure heptane, the adsorbed amount de ­ many binding sites per polymer and slow mass transport away creases to 1.9 mg/m 2. Further, the shift in dissipation factor is from the surface (27). close to zero at all concentrations except at 5000 ppm, where a The adsorption isotherm is shown in Fig. 3, where the ad ­ slight increase in the dissipation shift is observed. The change in sorbed amount is plotted as a function of resin concentration. dissipation factor decreases again after rinsing with pure solvent. Results from two resin-in-heptane studies, where the concen ­ tration of resins in solution has been increased stepwise, are presented. The difference in adsorbed amount of the two se ­ ries is probably related to how the concentration change was made, with a higher adsorbed amount being obtained when the 5 000 ppm 500 ppm bulk concentration was increased in smaller steps. The fact that

300 ppm the details of the adsorption process is affected by the details 200 ppm 100 ppm of the experimental procedure is not surprising considering that resins is a solubility class of compounds and not one well-defined chemical entity. Resins redissolved in 50.50 heptane!toluene mixture and in pure toluene, respectively. The adsorption of resins redissolved in heptane/toluene and toluene onto a hydrophilic gold surface was determined for concentrations between 25 and 2500 ppm in the mixture and between 100 and 5000 ppm in pure toluene. No significant adsorption was found in pure toluene. For resins in heptane/toluene, a frequency shift of approximately -10 Hz was Time (min) observed. This corresponds to an adsorbed amount of 0.59 mg/ FIG. 2. Frequency and dissipation shift as a function of time for the adsorp ­ m2; hence, the adsorbed amount is small. Further, the nonshift­ tion of resins onto a hydrophilic gold surface in an w-heptane solution. ing dissipation factor tells us that the adsorbing molecules are 346 EKHOLM ET AL.

firmly attached and presumably adopt a flat conformation. The fact that the resins adsorbed significantly in heptane (2.3 mg/m 2), less in heptane/toluene (0.6 mg/m 2), and hardly at all in toluene is suggested to be related to the solubility of resins in the differ ­ ent solvents; i.e., the solubility increases with increasing toluene content and the nature of the resin-surface and solvent-surface interaction. The aromatic toluene binds stronger to the gold sur­ face than heptane. All these factors combine to make the net driving force for resin adsorption lower in toluene than in hep­ tane. Studies have also been made for adsorption of resins re ­ dissolved in heptane/toluene and toluene onto a hydrophobic surface and the same result, i.e., very limited or no adsorption, was achieved. This indicates that the solubility of the resin, i.e., the resin-solvent interaction, is a key factor in determining the 25 50 100 250 500 1 000 wash extent of adsorption The same effect has also been observed by Concentration (ppm) Acevedo et al. (19). They studied the adsorption of resins onto FIG. 5. Adsorbed amount of asphaltenes onto a hydrophilic gold surface heated asphaltene particles and noticed that the adsorption of as a function of asphaltene concentration, 25-1000 ppm, in a heptane/toluene resins decreases with the content of toluene in the solution. (50/50) solution.

Adsorption Study—Asphaltenes

Studies of asphaltenes redissolved in n-heptane. Asphal­ frequency (4.2 mg/m 2) and for the dissipation factor (+5). Fur­ tenes precipitate in heptane, though they are not completely in­ ther, the stabilization of the signal after a few minutes illustrates soluble in this solvent. Adsorption of asphaltenes redissolved in that the asphaltenes adsorb as small aggregates onto the sur­ heptane to a hydrophilic gold surface has been studied but due face, no slow deposition of large asphaltene aggregates occurs. to difficulties with precipitation in the injection syringe even at Higher concentrations than 50 ppm give only a slight increase in very low concentrations, these measurements are of no signifi ­ the adsorption to 5.3 mg/m 2 at 1000 ppm (Fig. 5). By rinsing the cance, except for the observation of the strong aggregation and measuring chamber with pure solvent, the frequency shift de ­ precipitation. creases to a level of —80 Hz, which corresponds to an adsorbed Asphaltenes redissolved in 50:50 heptane!toluene mixture. amount of 4.8 mg/m 2 (Figs. 4 and 5). The rather small dissi ­ The adsorption of asphaltenes redissolved in a heptane/toluene pation shift, despite the high-adsorbed amount, indicates that a mixture onto a hydrophilic gold surface was studied for concen ­ rigid layer of asphaltenes is adsorbed onto the surface. The shift trations in the range of 25-1000 ppm (Fig. 4). The first thing in dissipation factor increases slightly for concentrations higher to observe is a large shift when injecting 50 ppm—both for the than 50 ppm. This suggests that the asphaltenes to some extent adsorb onto other asphaltenes, and the desorption by dilution indicates that they only are weakly adsorbed. Even though we are very close to the flocculation threshold for the asphaltene 500 ppm in heptane/toluene mixtures, about 50%, we do not observe any ^ 1000 ppm 50 ppm ppm 250 ppm adsorption of large aggregates on the surface. We also note that the adsorbed amount is higher than normally found for nonag ­ gregating polymers (1-3 mg/m 2). Hence, it seems likely that asphaltene aggregates are attached to the surface. To get more information about the large shift at the 50 ppm in­ jection, a solution, with a concentration of 25 ppm. was replaced six times in the measuring chamber (Fig. 6). Surprisingly, the ad­ sorbed amount increased with each injection. This demonstrates that the adsorption is not only dependent on the bulk concentra ­ tion, but presumably, also on the number and size of aggregates Frequency shift Dissipation shift present. After these injections and two additional injections of 35 ppm and rinsing with heptane/toluene, an adsorbed amount of 4.4 mg/m 2 was observed. The adsorbed amounts for the two Time (min) studies can be compared in Figs. 5 and 6. The dissipation shift in the 25 ppm case was smaller; it increased only to approximately FIG. 4. Frequency and dissipation shift as a function of time for the adsorp­ tion of asphaltene onto a hydrophilic gold surface in a (50/50) heptane/toluene +2, compared to +5 at higher asphaltene concentrations. This solution. indicates that a more compact and rigid layer is obtained when a ADSORPTION OF ASPHALTENES ONTO A HYDROPHILIC SURFACE 347

10 000 ppm 5 000 ppm

150 ppm 1 000 ppm V V 400 ppm 50 ppm

-150 - Frequency shift Dissipation shift

Concentration (ppm) Time (min) FIG. 6. Adsorbed amount of asphaltenes onto a hydrophilic gold surface in a heptane/toluene (50/50) solution with repeated replacement of 25 and 35 ppm FIG. 8 . Frequency and dissipation shift as a function of time for the adsorp ­ asphaltene solutions. tion of asphaltene onto a hydrophilic gold surface in a toluene solution. smaller concentration of asphaltenes is injected repeatedly into bulk concentration determines the size of the aggregates. An­ the measuring cell. This is particularly clearly seen when the other more likely explanation may be that the adsorption from change in dissipation is plotted as a function of the adsorbed a more concentrated solution occurs more rapidly and thus less amount for the two sets of experiments (Fig. 7). The reason for time is allowed for the molecules to rearrange on the surface. this difference may be that injections of higher concentrations re ­ Even so, almost the same adsorbed amount is reached after rins­ sult in an adsorption of larger and more voluminous aggregates ing with pure solvent for the two cases. (larger increase in dissipation shift), while for lower concen ­ Asphaltenes redissolved in toluene. In toluene, the adsorp ­ trations smaller aggregates adsorb. This could be explained by tion of asphaltenes onto a gold surface was studied for concen ­ the fact that we are close to the flocculation threshold for the trations between 50 and 10,000 ppm (Fig. 8). Again, the stabi ­ asphaltene in heptane/toluene mixtures, which means that the lization of the signal after a few minutes indicates adsorption of asphaltene rather than deposition of large aggregates. From Fig. 9, it is seen that the adsorption increases with increasing con ­ 7 centration. However, no real leveling off in the adsorbed amount with increasing concentration is observed, which may indicate 6

5

| 4 I ' 1= I 2

1

0 < 0 1 2 3 4 5 6 Adsorbed amount (mg/m2)

FIG. 7. The dissipation factor plotted as a function of adsorbed amount for the two series of asphaltenes in heptane/toluene (50/50). Filled circles represent 10000 the series with concentrations between 25 and 1000ppm; open triangles the series Concentration (ppm) where 25 ppm solutions were replaced repeatedly in the measuring chamber. The adsorbed amount and dissipation value for the two series after rinsing with the FIG. 9. Adsorption isotherm for asphaltenes onto a hydrophilic gold surface solvent is shown as an open circle and filled triangle, respectively. as a function of asphaltene concentration in a toluene solution. 348 EKHOLM ET AL.

resin concentration was carried out. Now, a 1000 ppm solution 12 of resins was injected; however, no effect was observed, and the adsorbed amount remained the same. This demonstrates that the 10 interactions between resin and preadsorbed asphaltene are very weak and that the presence of the asphaltenes on the surface o ~ s prevents resin adsorption. Under such circumstances, there will, £ of course, be no desorption of asphaltenes.

Asphaltenes!Resins Mixture Adsorption Study .9- S 4 Adsorption of a asphaltenes!resins mixture re-dissolved in 50.50 heptaneltoluene mixture. In this set of experiments, asphaltenes and resins were mixed and redissolved in 50:50 heptane/toluene with the aim of obtaining a model of the crude without paraffins and waxes. Below, these results will be com ­ 0 pared with the adsorption results from crude oil, as well as the 0 2 4 6 8 10 Adsorbed amount (mg/m2) data about adsorption from the pure asphaltene and resin frac ­ tions described above. According to the amount and the pro ­ FIG. 10. The dissipation factor plotted as a function of adsorbed amount portions in the crude oil, the following concentrations of as ­ for asphaltenes dissolved in toluene. The value after rinsing with solvent is phaltenes and resins were studied: The first mixture (i) models represented as a triangle. a 10,000 ppm crude oil. This solution contained 290 ppm as ­ phaltenes and 1380 ppm resins. The second mixture (ii) models a 50,000 ppm crude oil. This solution contained 1450 ppm as ­ that the asphaltene dissolved in toluene to some extent adsorb in phaltenes and 6900 ppm resins. First mixture (i) was injected multilayers onto the surface. It is also seen that the shift in dissi ­ twice and then mixture (ii) was injected twice. Finally, the sur­ pation increases with increasing adsorbed amount (Fig. 10). The face was washed with 50:50 heptane/toluene (see Fig. 12). The adsorbed amount at 10,(XX) ppm is about 8-9 mg/m 2 and when final adsorbed amount corresponds to 6.8 mg/m 2 (—120 Hz shift) the cell is rinsed with toluene, the adsorbed amount decreases to and the adsorbed layer seems to be quite rigid according to 7.1 mg/m 2 Moreover, a decrease in the dissipation shift is also the small dissipation shift. The study shows that the adsorbed observed upon rinsing, which shows that desorption also leads amount in the mixture is larger than the adsorbed amount for to formation of a more compact asphaltene layer. the pure fractions. Hence, it seems likely that mixed aggregates are formed in bulk solution and then these aggregates adsorb to A Competitive Adsorption and Desorption Study the surface. After the cell was rinsed with solvent a slight de ­ Adsorption of resins and asphaltenes redissolved in SO: 50 crease in the adsorbed amount was observed, while the change heptaneltoluene. In crude oils, resins disperse asphaltenes as in dissipation factor decreased significantly, indicating that a resin-solvated aggregates and prevent precipitation of asphal- tenic particles. Seen from the fraction studies in the 50:50 hep­ tane/toluene mixture, which is supposed to model the aromatic- aliphatic solvency in the crude, the asphaltenes adsorb strongly while the resins adsorb only to a very limited degree. Here our purpose has been to investigate if the resins adsorb onto the preadsorbed asphaltenes or redisperse them into the solution. In the first step a solution of 25 ppm asphaltenes redissolved in hep­ tane/toluene was injected twice. A layer of asphaltenes with an adsorbed amount of about 3.9 mg/m 2 was obtained. In the next step, a 25 ppm resin solution was injected twice and no notice ­ able change in adsorbed amount was observed; i.e., hardly any resin adsorbed to the asphaltene-coated surface and hardly any asphaltenes were desorbed (Fig. 11). Another 25 ppm asphal ­ tene solution was injected thereafter to see if the layer was com ­ plete. This resulted in a small increase in adsorbed amount of as ­ 25 ppm 1 000 ppm phaltenes, and a frequency shift of -80 Hz (4.8 mg/m 2); i.e., the Asph Asph Resin Resin Asph Resin same amount as in the previous asphaltene-in-heptane/toluene FIG. 11. Competing adsorption of asphaltenes and resins —adsorbed study. As the resins, at a concentration of 25 ppm, do not show amount of matter into a hydrophilic gold surface as a function of various in ­ any adsorption onto the asphaltenes, an experiment with a higher jections of asphaltene and resins in a heptane/toluene (50/50) solution. ADSORPTION OF ASPHALTENES ONTO A HYDROPHILIC SURFACE 349

leading to an adsorbed amount of 8.9 mg/m 2. Furthermore, the change in dissipation factor increases to approximately +11 which tells us that the surface layer has become looser in nature due to the increased adsorption. The adsorbed amount and the dissipation shift at a concentration of 50,000 ppm in the crude oil are significantly larger than the values for the corresponding mixture of resins and asphaltenes. This demonstrates that the aggregates adsorbed from the crude oil are larger and more loosely packed compared to the aggregates formed in the mixture of pure fractions of resins and asphaltenes. Rinsing the surface with 50:50 heptane/toluene gives a rigid layer at —75 Hz, corresponding to an adsorbed amount of 4.8 mg/m 2 Hence, the adsorbed layer structure before rinsing is that of an inner firmly bound fraction and on top of this layer a less 290 ppm Asph 290 ppm AspEi 1450 ppm Asp};. 1450 ppm Aspii wash strongly bound fraction is obtained. Compared to the study of i380 ppm Resin 1380 ppm Resin 6900 ppm Resia 6960 ppm Resin the purified fractions of asphaltenes and resins, and the mixture FIG. 12. The adsorbed amount (mg/m2) and change in dissipation factor of asphaltenes and resins redissolved in 50:50 n -heptane/ for mixtures of resins/asphaltenes in various ratios in heptane/toluene (50/50), toluene there is a significant difference in adsorption. The when adsorbed onto a hydrophilic gold surface. adsorbed amount for the 10,000 ppm crude is smaller than for the corresponding asphaltene and resin mixture (290 ppm as ­ very compact layer remains on the surface. However, this layer phaltene + 1380 ppm resin), but quite similar to that obtained for asphaltene in 50:50 heptane/toluene. Further, for diluted crude was not as compact as the layer achieved when first injecting oil the adsorbed amount remaining on the surface, after rinsing asphaltenes and then resins. with the pure solvent, corresponds to 4.2 mg/m 2 (—75 Hz shift). Adsorption Study—Crude Oil This should be compared to 6.8 mg/m 2 (—120 Hz shift) for the mixture. These differences as well as differences in dissipation Studies of crude oil diluted in heptane/toluene. The adsorp ­ shifts suggest that the resin and asphaltene molecules (or just tion of crude oil diluted in a 50:50 heptane/toluene mixture the asphaltenes) arrange in a different way in the adsorbed onto a gold surface was studied for the concentrations 10,000 layer when the effect from other constituents like paraffins and and 50,000 ppm. First, the 10,000 ppm solution was replaced waxes are absent. These constituents may be incorporated in three times in the measuring chamber. The frequency shift is the adsorbed layer or affect the interaction forces in the bulk. approximately —70 Hz after these injections, which corresponds to an adsorbed amount of 4.2 mg/m 2 (Fig. 13). When injecting the 50,000 ppm solution, a shift to —150 Hz is observed, CONCLUSIONS

The adsorption of resins redissolved in heptane onto a gold surface shows that the adsorbed resins pack into a compact $ Adsorbed amount monolayer and that there is no tendency of resin aggregation on the surface. The adsorption of resins decreases with increas ­ ing toluene content in the solvent and it is argued that this mainly is due to an increased solvency of resins even though a favor ­ able toluene-surface interaction may contribute to the observed effect. The asphaltenes redissolved in heptane/toluene and toluene adsorb to a larger extent. The adsorption is higher than observed for typical nonassociating polymers, indicating aggregate ad ­ sorption. Asphaltenes form a rigid layer at lower concentrations. When higher concentrations are injected, it is possible to obtain further adsorption, which is related to the strong tendency of aggregation of asphaltenes in bulk solution. Desorption studies show that resins are not able to desorb Concentration of crude oil preadsorbed asphaltenes from the surface. Neither do they ad ­ FIG. 13. The adsorbed amount (mg/m2) and change in dissipation factor sorb onto the asphaltene-coated surface. as a function of concentration for the injections of crude oil redissolved in The irreversibly adsorbed amount for the crude oil solution is heptane/toluene (50/50). smaller than for the asphaltene and resin mixture but quite similar 350 EKHOLM ET AL. to that of the asphaltene solutions. The resin and asphaltene 9. Rousseau, G., Zhou, H., Hurtevent, C., and Production, T. E„ in “SPE molecules (or just the asphaltenes) arrange in a different way Oilfield Scale Symposium,” Society of Petroleum Engineers, Aberdeen, in the adsorbed layer when the effect from other constituents UK, 2000. 10. McLean, J. D., and Kilpatrick, P. K., J. Colloid Interface Sci. 189, 242 like paraffins and waxes are absent. These constituents may be (1997). incorporated in the adsorbed layer or affect the interaction forces 11. McLean, J. D., and Kilpatrick, P. K., Energy Fuels 11,570 (1997). in the bulk of the crude. 12. Yang, X., Lu, W., Ese, M.-H., and Sjbblom, J., in “Encyclopedic Handbook of Emulsion Technology ” (J. Sjbblom, Ed.), p. 525, Dekker, New York, 2001. ACKNOWLEDGMENTS 13. Sjbblom, J., Saether, 0., Midttun, 0., Ese, M.-H., Urdahl, O., and Fprdedal, H., in “Structures and Dynamics of Asphaltenes ” (O. C. Mullins and E. Y. The study was financed by ABB, Norsk Hydro and Statoil, who are all thanked Shew, Eds.), p. 337, Plenum Press, New York, 1998. for their kind permission to publish the results. Inge Harald Auflem thanks the 14. Ese, M.-H., Yang, X., and Sjbblom, J., ColloidPolym. Sci. 276,800 (1998). technology program Flucha II for a Ph.D. grant financed by oil industry and 15. Ese, M.-H., Galet, L., Clausse, D., and Sjbblom, J.,7. Colloid Interface Sci. Norwegian Research Council (NFR). 220,293 (1999). 16. Singh, B. P., and Pandey, B. R, Ind. J. Technol. 29,443 (1991). REFERENCES 17. Nordli, K. G., Sjbblom, J., Kizling, J., and Stenius, P., Colloids Surf. 57,83 (1991). 1. Leontaritis, K. J., “Asphaltene Deposition: A Comprehensive Description of 18. Leblanc, R. M., and Thyrion, F. C., Fuel 68 , 260 (1989). Problem Manifestations and Modelling Approaches, ” Society of Petroleum 19. Acevedo, S., Ranaudo, M. A., Escobar, G„ Gutierrez, L., and Ortega, R, Engineers, Aberdeen, UK, 1989. Fuel 74, 595 (1995). 2. Mansoori, G. A., Jiang, T. S., and Kawanaka, S., Arabian J. Sci. Eng. 13, 20. Acevedo, S., Ranaudo, M. A., Garcia, C„ Castillo, J., Fernandez, A., 17 (1988). Gaetano, M., and Goncalvez, S., Colloids Surf. A 166,145 (2000). 3. Speight, J. G., “The Chemistry and Technology of Petroleum,” Dekker, 21. Acevedo, S„ Castillo, J., Fernandez, A., Goncalvez, S., and Ranaudo, New York, 1999. M. A., Energy Fuels 12,386 (1998). 4. Hirschberg, A., deJong, L. N. J„ Schipper, B. A., and Meijer, J. G., “Influ­ 22. Sauerbrey, G., Z. Phys 155,206 (1959). ence of Temperature and Pressure on Asphaltene Flocculation,” Society of 23. Stockbridge, C. D,, in “Vacuum Microbalance Techniques, ” p. 147 Plenum Petroleum Engineers of AIME, Aberdeen, UK, 1984. Press, New York, 1966. 5. Hammami, A., Phelps, C. P., Monger-McClure, T., and Little, T. M., Energy 24. Rodahl, M., Ph.D. thesis, Chalmers University of Technology and Fuels 14,14 (2000). Gothenburg University, Gothenburg, Sweden, 1995. 6. Fotland, P., Fuel Sci. Technol. 14, 313 (1996). 25. Hook, F., Ph.D. thesis, Chalmers University of Technology and Gothenburg 7. Pfeiffer, J. P., and Saal, R. N., J. Phys. Chem. 44,139 (1940). University, Gothenburg, Sweden, 1997. 8 . Sjbblom, J., John sen, E. E., Westvik, A., Ese, M.-H., Djuve, J., Auflem, 26. Stdlgren, J. J. R., Claesson, R M., and Wamheim, T., Adv. Colloid Interface I. H., and Kallevik, H., in “Encyclopedic Handbook of Emulsion Technol­ Sci. 89-90, 383 (2001). ogy ” (J. Sjbblom, Ed.), p. 595, Dekker, New York, 2001. 27. Cohen Stuart, M. A., and Fleer, G. J,,Annu. Rev. Mater. Sci. 26,463 (1996). Paper V Colloid Polym Sci (2002) 280: 695-700 DOI 10.1007/S00396-002-0660-9 ORIGINAL CONTRIBUTION

Inge Harald Auflem Near-IR study on the dispersive effects Trond Erik Havre Johan Sjoblom of amphiphiles and naphthenic acids on asphaltenes in model heptane-toluene mixtures

Abstract Near-IR spectroscopy is Keywords Asphaltenes Received: 11 May 2001 Accepted: 8 January 2002 used to follow the disintegration of Naphthenic acids Near-IR Published online: 23 May 2002 asphaltene aggregates at 1,600 nm. spectroscopy Inhibitors © Springer-Verlag 2002 It is shown that the technique is applicable to determine the efficiency I.H. Auflem (E) T.E. Havre J. Sjoblom Ugelstad Laboratory, Department of of various additives as disaggrega ­ Chemical Engineering, Norwegian tion chemicals. University of Science and Technology (NTNU), 7491 Trondheim, Norway E-mail: inhatinstatoil.com J Sjoblom Statoil ASA, R & D Centre. 7005 Trondheim, Norway

aggregates are kept dispersed in the solution by resins Introduction [18, 19, 20, 21, 22], which are molecules similar to the Asphaltenes are defined by solubility characteristics: asphaltenes containing various polar groups as the as ­ they are insoluble in light alkanes such as »-heptane and phaltenes do, yet are soluble in light alkanes and insol ­ are soluble in toluene [1], They are generally composed uble in toluene. of polyaromatic nuclei carrying aliphatic chains and Gonzalez and Middea [23] studied the peptization of rings and a number of heteroatoms, including sulphur, asphaltenes in aliphatic solvents by various oil-soluble oxygen, nitrogen and metals such as vanadium, nickel amphiphiles. They showed that the effectiveness of and iron. These heteroelements account for a variety of amphiphiles on asphaltene stabilization was influenced polar groups, such as aldehyde, carbonyl, carboxylic by the interactions between the polar headgroups of the acid, amine and amide [2, 3, 4], As one of the primary amphiphiles and polar groups on the asphaltene mole ­ components of crude oil, asphaltenes are known to cause cules; however, their results also indicated that other a number of problems in connection with production, interactions could be of importance, for instance, the transport and processing of crude oils [5, 6, 7, 8], The n electrons of the aromatic portions of the asphaltenes state of the asphaltenes is dependent upon pressure and may act as electron donors for hydrogen bonds with temperature as well as the composition of the crude oil hydroxyl groups of the amphiphiles. In 1994, Chang [9, 10, 11 12, 13, 14]. It is probable that they will pre ­ and Fogler [24, 25] discussed the stabilization of as ­ cipitate and cause plugging of reservoirs or production phaltenes in aliphatic solvents using alkylbenzene-de- equipment if they are exposed to large changes in these rived amphiphiles. The results supported earlier parameters. The general view is that at high pressures, suggestions of a hydrogen-bonding effect or possible the asphaltenes are in a condition of monomeric entities acid-base interactions between the amphiphile head- dissolved in the crude oil. At lower pressures, the as ­ groups and polar groups on the asphaltenes, and also phaltenes exists as aggregates (micelles) [15, 16] of enti ­ showed that the length of the amphiphiles alkyl tail was ties with an altogether elliptical form [17]. These of significance. 696

In this study the disintegration of asphaltene ag ­ NIR measurements. The physical properties of the chemical addi­ gregates is studied by means of near-IR (NIR) spec ­ tives and solvents are summarized in Table 1. troscopy. The NIR technique is introduced as a powerful tool for screening the efficiency of chemicals NIR sampling for this purpose. The NIR measurements were performed with a Brimrose AOTF Luminar 2000 spectrometer equipped with a fiber optic sampling probe for transflectance measurements. The wavelength region was Experimental set to 1,100-2,200 nm, and the total number of scans per spectrum to 32. The total path length was 1 mm, and the experiments were performed at room temperature (22 °C). In the comparison of the Preparation of asphaltenes effectiveness of the additives, we utilized the results from 1,600 nm. The asphaltenes were precipitated from a North Sea crude oil by The reason for this choice is that hydrocarbon absorption is min ­ gently mixing crude oil and /7-heptane (1:20 gml'1) at room tem­ imal in this region, and that this is the NIR region with the least perature for 12 h without exposure to light. The mixture was then noise in the measurements. filtrated through a microfilter (pore diameter 0.45 pm) to separate the asphaltenes from the diluted crude oil. The filtrate was then washed in /7-heptane with constant stirring for 1 h before being filtrated and dried under a N2 atmosphere to constant weight. Theory

Light scattering by asphaltene aggregates Preparation of asphaltene stock solution in the NIR region The dried asphaltenes were dissolved in toluene before /7 -heptane was added. The stock solution then consisted of 0.25 wt% as­ The general theory behind optical scattering and ab ­ phaltenes in a mixture of n-heptane and toluene with a ratio of 70/ sorption is a well-explored field and is described in detail 30 v/v. The stock solution was allowed to equilibrate with constant stirring for 48 h before the measurements started. in the literature [26], In this work, our interest centers on the light extinction in the IRspectral range, which can be accounted for by two distinct mechanisms: electronic Preparation of samples with additives absorption by organic molecules and scattering from

The additives were dissolved in /7 -heptane/toluene, or in some cases particles or aggregates. The absorption spectra in the where this was difficult, they were first dissolved in toluene and then NIR range from 780 to 2,500 nm (12,820 to 4,000 cm'1) the /7 -heptane was added. The solutions with additives were then consist of overtones and combinations of the funda ­ mixed with an equal amount by weight of asphaltene stock solu­ tion, before the NIR measurements started. Each sample was then mental molecular vibration bands, which are primarily made up of 0.125 wt% asphaltenes in 70/30 //-heptane/toluene with due to hydrogenic stretches of C H, N H. S-H and O various concentrations of additives. Spectra were taken every H bonds. In crude oils, the scattering depends strongly minute (Fig. 1) with constant mixing throughout the experiment. upon the shapes and sizes of asphaltene, wax and hy­ The first spectrum in each series was set as a background and drate particles. A common way to create light spectra is subtracted from the rest of the spectra. Since the asphaltenes were partly precipitated in the //-heptane/toluene mixture, it was crucial to measure the relative amount of light transmitted with constant stirring to ensure a homogeneous mixture during the through a solution and to convert the measured trans ­ mittance, T, into optical density (OD) by nonlinear transformation. The OD of the sample expresses the amount of light that has been lost in the sample and is linearly related to the total cross section by

OD = log(/ 0//) = 0.434A

where /0 and / are the intensities of incident and trans ­ mitted light and N is the number of particles in the total cross section aiot. For slightly lossy dielectric spheres, electronic absorption, cabs , and optical scattering, usca , contribute separately to the light extinction.

'hot — (Tsca T Albs (2)

The scattering can be divided into two groups: wave ­ length-independent scattering, where the size of the Fig. I. Near-IR (NIR) spectra of 0.125 wt.% asphaltenes in a 70/30 //-heptane/toluene mixture with 1.25 wt% inhibitor A added. A scatterer is very large compared to the wavelength of the total of 1,200 spectra were measured and the time between each light, and wavelength-dependent scattering, where spectrum was 1 mm the particles are of comparable sizes to or smaller than 697

Table 1. A summary of Molecular Purity the amphiphiles and naphthenic Name Chemical structure Source acids used as additives in this weight (wt%) study 5-p(II)-cholanoie acid 361 c#' |CHOL| Chiron AS >95

1 -Napblhalcncpcnianoiu 238 Chiron AS >90 acid, dccahydro [2C4] 0

1-Naphthalenoic acid, 238 Chiron AS >90 decahydro-2-buthyl [C42j I N ■' ■

Crude naphthenic acid Com van 250 Mixture - [CNA] der Locke

Fluka naphthenic acid 240 Mixture Fluka - [FlukaJ

Naphthenic acid from 400 Mixture Extracted - North Sea [North Sea]

2-Ethyl-1 -hcxanol 13023 Merck > 99

1 -Octanol 130.23 Merck >99

Benzyl alcohol 108.14 Fluka >98 H 1

FIcxylaminc 101.94 Merck >98

n-Alkylbenzenesulfonic acid. 385.5 ' Alla Aesar 96+ (n - Chi-Cu) [ABSA]

Inhibitor A Mixture - Tros / Dyno -

aUp to 30 wt% unsaturated bonds in cyclic part of molecular structure b Up to 10 wt% unsaturated bonds in cyclic part of molecular structure 698

ihe wavelength of the light. The latter group contains the In setting up the matrix for the chemicals some fun­ case of r/7l < 0.05 (the Rayleigh condition), where X is damental interaction patterns were considered. The the wavelength of light and r the particle radius. Then, starting point is that the dominating attractive forces for the scattering is purely dipolar with high-order scatter ­ asphaltene aggregation are hydrogen bonding, acid-base ing moments vanishing, and rrsua is given by interactions and charge transfer between aromatics. The main strategy for the disintegration of asphaltene ag ­ fiV gregates should hence be to break these bonds and sta ­ (3) bilize the smaller asphaltenes aggregates. The supposed interactions for the individual chemicals are further where e is the dielectric constant and k — 2njX.

The results obtained can be summarised as follows. H-heptane/asphaitene mixtures with various naphthenic The relative OD at 1,600 nm versus time for toluene/ acids is plotted in Fig. 3. The commercial Fluka naph­ H-heptane/asphaltene mixtures with different amounts of thenic acid and the naphthenic acid extracted from a CNA is shown in Fig. 2. The influence of additive upon North Sea crude seem to affect the state of the as ­ aggregate size is depicted as the decrease of scattering as phaltenes only to a minor extent. CNA is most efficient a function of time. The relative OD vs. time for toluene/ of these polydisperse naphthenic acids.

Time [min] 0 200 400 600 800 1000 1200 Time [min] 0 200 400 600 800 1000 1200

0.000 No additive

0.000 No additive 0.125 wt% -0.005 -0.005 1.25 wt% V- -0.010 2-ethyl- 1-hexanol Z -0.010

3 -0.015 I 3.25 wt% —... l-octanol 5 -0.015 1 \ benzyl alcohol 6.25 wt% -0.020 8

-0.020 hexylamine 12.5 wt% -0.025

-0.025 Inhibitor A -0.030

Fig. 2. NIR scattering measurements al 1,600 nm for 0.125 wt% -0.030 asphaltenes in a 70/30 n-heptane/toluene mixture with crude naphthenic acid (CNA) added in various concentrations Fig. 4. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 n-heptane/toluene mixture with 1.25 wt% various amphiphiles added Time [min] 0 200 400 600 800 1000 1200

0.000

-0.005 North Sea

-0.010

CHOL -0.015

-0.020

-0.025 Inhibitor A

-0.030

Fig. 3. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 n-heptane/toluene mixture with 1.25 wt% of various naphthenic acids added: 5-/j!-(//)-choianoic acid (Choi), 1- napthalenepentanoic acid, decahydro (2C4), 1 -naphthalenoic acid, Fig. 5. NIR scattering measurements at 1,600 nm for 0.125 wt% decahydro-2-butyl (C42), CNA, Fluka naphthenic acid (Fluka) and asphaltenes in a 70/30 n-heptane/toluene mixture with 1.25 wt% North Sea naphthenic acid (North Sea) 77 -alkyibcnzencsulfonic acid (ABSA) added 700

A comparison between 2C4, C42 and Choi shows Conclusions these species to be somewhat more efficient than the previous group. Especially the 2C4 molecule has a very It was shown that NIR spectroscopy is a powerful efficient breakdown to start with and also attains a low method to follow the disintegration of asphaltene ag ­ final value. It is thrilling to see that the molecular gregates upon addition of chemicals. The method is structure affects the results to this extent. based on the scattering from preferentially large aggre ­ The other amphiphiles, presented in Fig. 4, have a gates. The NIR technique, which is very fast and accu ­ varying effect on the disintegration of the asphaltenes. rate. is a good choice for the initial screening of large The most efficient one is 1-hexylamine, which is very numbers of chemicals for asphaltene inhibition. The reminiscent of 2C4 with regard to short-term efficiency results show that additives which are efficient in re ­ and the final state. The most efficient treatment is due to placing hydrogen bonds possess dispersive power and inhibitor A, the commercial mixture. can serve as inhibitors. Commercial blends of active Alkylbenzenesulfonic acid is shown to associate with molecules gave the best results. the asphaltenes and create aggregates of increased sizes (Fig. 5). This is in accordance with the results obtained by Chang and Fogler [25] in a UV/vis spectroscopic Acknowledgements T.E.H. and l.H.A. acknowledge the technolo ­ gy program FLUCHA II, financed by the Research Council of study They suggested that asphaltenes and dodecyl- Norway and the oil industry, for doctoral grants. We also express benzenesulfonic acid could associate into large electronic our appreciation to Statoil R&D Centre for the use of all necessary conjugated complexes. instrumentation in this research.

References

1 Speight JG (1991) The chemistry and 13. McLean JD, Kilpatrick PK (1997) 24. Chang C-L, Fogler SH (1994) technology of petroleum. Dekker. New J Colloid Interface Sci 189:242-253 Langmuir 10:1749-1757 York 14. Andersen SI, Birdi KS (1990) Fuel Sci 25. Chang C-L, Fogler SH (1994) 2. Dickie JP, Yen TF (1967) Anal Chem Technol Int 8:593 615 Langmuir 10:1758-1766 39:1847-1852 15. Ncllensteyn FJ (1924) J Inst Pel Tech ­ 26. Kerker M (1969) In: Loebl EM (ed) 3. Mitchell DL, Speight JG (1973) Fuel nol 10:311 The scattering of light and other elec­ 52:149-152 16. Yen TF (1981) In: Hunger JW. Li NC tromagnetic radiation. Physical chem ­ 4. Strausz OP, Mojelsky TW, Town EM (eds) Chemistry of asphaltenes. Ad­ istry. A series of monographs, vol 16. (1992) Fuel 71:1355 1363 vances in chemistry series 195. Ameri­ Academic, New York, p 666 5. Park SJ, Mansoori GA (1988) Energy can Chemical Society, Washington, 27 Heller W (1964) J Chem Phys 40:2700 Sources 10:109-125 DC, p 39 28. Stevenson AF (1953) J Appl Phys 6. Speight JR, Long RG, Trawbridge TD 17. Sheu EY, De Tar MM, Storm DA 24:1134 (1984) Fuel 63:616-620 (1994) In: Sharma MK, Yen TF (eds) 29. Stevenson AF (1953) J Appl Phys 7. Mansoori GA (1996) Arab J Sci Eng Asphaltene particles in fossil fuel ex­ 24:1143 21:707-723 ploration, recovery, refining, and pro ­ 30. Mullins OC (1990) Anal Chem 62:508 8. Sjoblom J, Skodvin T. Holt O, Nilscn duction processes. Plenum, New York, 514 FP (1997) Colloids Surf A 123:593-607 p 155-165 31. Kallevik H (1999) Characterisation of 9. Hunger JW, Li NC (1981) Chemistry of 18. Pfeiffer JP. Saal RN (1940) J Phys crude oil and model oil emulsions by asphaltenes. Advances in chemistry se­ Chem 44:139 means of near infrared spectroscopy ries 195. American Chemical Society, 19. Swanson J (1942) J Phys Chem 46: and multivariate analysis. Department Washington, DC 141 150 of Chemistry, University of Bergen 10. Hirschberg A, de Jong LNJ. Schippcr 20. Koots JA, Speight JG (1975) Fuel 32. Joshi NB, Mullins OC, Abdul J, Creek BA, Mcijcr JG (1984) SPEJ, Soc Pet 54:179-184 J. McFadden J (2001) Energy Fuels Eng 24(3):283-293 21. Hammami A, Fern worn KA, Nighs- 15:979-986 11 Mansoori GA. Jiang TS, Kawanaka S wander JA (1998) Pet Sci Technol 33. Newberry ME, Barker KM (1983) US (1988) Arab J Sci Eng 13:17 34 16:227 249 Patent 4414035 12. De Boer RB, Leerlooyer K, Eigner 22. Andersen SI, Speight JG (2001) Pet Sci MRP, Van Bergen ARD (1995) SPE Technol 19:1-34 Production and Facilities 2:55-61 23. Gonzalez G, Moreira MBC (1991) Colloids Surf 58:293-302 (Colloid and Polymer Science I© Springer-Verlag 2002 IX )1 10.1007 s00396-0<)2-()660 -9

Original Contribution Near-IR study on the dispersive effects of amphiphiles and naphthenic acids on asphaltenes in model heptane-toluene mixtures

Inge Harald AuflembQ, Trend Erik Havre1 and Johan SjobloinO 2

(1) Ugelstad Laboratory, Department of Chemical Engineering, Norwegian University of Science and Technology (NTNU), 7491 Trondheim, Norway (2) Statoil ASA, R & D Centre, 7005 Trondheim, Norway ^E-mail: [email protected]

Received: 11 May 2001 / Accepted: 8 January 2002 / Published online: 23 May 2002

Abstract. Near-IR spectroscopy is used to follow the disintegration of asphaltene aggregates at 1.600 nm. It is shown that the technique is applicable to determine the efficiency of various additives as disaggregation chemicals.

Keywords. Asphaltenes - Naphthenic acids - Near-IR spectroscopy - Inhibitors

Introduction

Asphaltenes are defined by solubility characteristics: they are insoluble in light alkanes such as n-heptane and are soluble in toluene [7 ]. They are generally composed of polyaromatic nuclei carrying aliphatic chains and rings and a number of heteroatoms, including sulphur, oxygen, nitrogen and metals such as vanadium, nickel and iron. These heteroelements account for a variety of polar groups, such as aldehyde, carbonyl, carboxylic acid, amine and amide [2, 3, 4 ]. As one of the primary components of crude oil, asphaltenes are known to cause a number of problems in connection with production, transport and processing of crude oils [5, 6, 7, 8], The state of the asphaltenes is dependent upon pressure and temperature as well as the composition of the crude oil [9, 70, 77, L2, I_3, 14 ]. It is probable that they will precipitate and cause plugging of reservoirs or production equipment if they are exposed to large changes in these parameters. The general view is that at high pressures, the asphaltenes are in a condition of monomeric entities dissolved in the crude oil. At lower pressures, the asphaltenes exists as aggregates (micelles) [75, 16\ of entities with an altogether elliptical form [77]. These aggregates are kept dispersed in the solution by resins [78, 79, 20, 27, 22 ], which are molecules similar to the asphaltenes containing various polar groups as the asphaltenes do, yet are soluble in light alkanes and insoluble in toluene.

Gonzalez and Middea [23] studied the peptization of asphaltenes in aliphatic solvents by various oil-soluble amphiphiles. They showed that the effectiveness of amphiphiles on asphaltene stabilization was influenced by the interactions between the polar headgroups of the amphiphiles and polar groups on the asphaltene molecules; however, their results also indicated that other interactions could be of importance, for instance, the it electrons of the aromatic portions of the asphaltenes may act as electron donors for hydrogen bonds with hydroxyl groups of the amphiphiles. In 1994, Chang and Fogler [24, 25 ] discussed the stabilization of asphaltenes in aliphatic solvents using alkylbenzene-derived amphiphiles. The results supported earlier suggestions of a hydrogen-bonding effect or possible acid-base interactions between the amphiphile headgroups and polar groups on the asphaltenes, and also showed that the length of the amphiphiles alkyl tail was of significance.

In this study the disintegration of asphaltene aggregates is studied by means of near- IR (NIR) spectroscopy. The NIR technique is introduced as a powerful tool for screening the efficiency of chemicals for this purpose.

Experimental

Preparation of asphaltenes

The asphaltenes were precipitated from a North Sea crude oil by gently mixing crude oil and n-heptane (1:20 gmH ) at room temperature for 12 h without exposure to light. The mixture was then filtrated through a microfilter (pore diameter 0.45 pm) to separate the asphaltenes from the diluted crude oil. The filtrate was then washed in n- heptane with constant stirring for 1 h before being filtrated and dried under a N2 atmosphere to constant weight.

Preparation of asphaltene stock solution

The dried asphaltenes were dissolved in toluene before n-heptane was added. The stock solution then consisted of 0.25 wt% asphaltenes in a mixture of n -heptane and toluene with a ratio of 70/30 v/v. The stock solution was allowed to equilibrate with constant stirring for 48 h before the measurements started.

Preparation of samples with additives

The additives were dissolved in n-heptane/toluene, or in some cases where this was difficult, they were first dissolved in toluene and then the n -heptane was added. The solutions with additives were then mixed with an equal amount by weight of asphaltene stock solution, before the NIR measurements started. Each sample was then made up of 0.125 wt% asphaltenes in 70/30 n-heptane/toluene with various concentrations of additives. Spectra were taken every minute (Fig. 1) with constant mixing throughout the experiment. The first spectrum in each series was set as a background and subtracted from the rest of the spectra. Since the asphaltenes were partly precipitated in the n-heptane/toluene mixture, it was crucial with constant stirring to ensure a homogeneous mixture during the NIR measurements. The physical properties of the chemical additives and solvents are summarized in Table E Time [rrmj 0 200 400 600 800 1000 1200

0.000 No additive

-0.005

2-ethyl- i-hexartol S' -o.oio I \ ...... —l-octanol 3 -0.015 X berayl alcohol 8

-0.020 hexylamne

-0.025 inhibitor A

-0.030

Fig. 1. Near-IR (NIR) spectra of 0.125 wt% asphaltenes in a 70/30 n -heptane/toluene mixture with 1.25 wt% inhibitor A added. A total of 1,200 spectra were measured and the time between each spectrum was 1 min

Table 1. A summary of the amphiphiles and naphthenic acids used as additives in this study

Molecular Purity Name Chemical structure Source weight (wt%)

5-P(H)-cholanoic acid 361 [CHOL] Chiron AS >95

1 -NafAthalenepentanoic 238 Chiron AS >90 acid, decahvdro [2C4j i

1-Naphthalenoic acid, 238 Chiron AS >90 decahydro-2-buthyl [C42] LU ' ' "

Crude naphthenic acid Com van 250 Mixture - (CNA| der Locke

Fluka naphthenic acid 240 Mixture Fluka IFhikaJ

Naphthenic acid from 400 Mixture Extracted - North Sea [North Sea] 2-Ethyl-1-hexanol 130.23 "1 Merck >99

1 -Octanol 130.23 Merck >99

-X Benzyl alcohol 108.14 Fluka >98

Hexylamine 101.94 Merck >98

n-Alkylbenzenesulfonic acid, 385.5 Alfa Aesar 96+ (n = C„rC,i) IABSA]

Inhibitor A Mixture - Tros / Dyno - aUp to 30 wt% unsaturated bonds in cyclic part of molecular structure b Up to 10 wt% unsaturated bonds in cyclic part of molecular structure

NIR sampling

The NIR measurements were performed with a Brimrose AOTF Luminar 2000 spectrometer equipped with a fiber optic sampling probe for transfiectance measurements. The wavelength region was set to 1,100-2,200 nm, and the total number of scans per spectrum to 32. The total path length was 1 mm, and the experiments were performed at room temperature (22 °C). In the comparison of the effectiveness of the additives, we utilized the results from 1,600 nm. The reason for this choice is that hydrocarbon absorption is minimal in this region, and that this is the NIR region with the least noise in the measurements. Theory

Light scattering by asphaltene aggregates in the NIR region

The general theory behind optical scattering and absorption is a well-explored field and is described in detail in the literature [26 ]. In this work, our interest centers on the light extinction in the IRspectral range, which can be accounted for by two distinct mechanisms: electronic absorption by organic molecules and scattering from particles or aggregates. The absorption spectra in the NIR range from 780 to 2,500 nm (12,820 to 4,000 cm-1 ) consist of overtones and combinations of the fundamental molecular vibration bands, which are primarily due to hydrogenic stretches of C-H, N-H, S-H and O-H bonds. In crude oils, the scattering depends strongly upon the shapes and sizes of asphaltene, wax and hydrate particles. A common way to create light spectra is to measure the relative amount of light transmitted through a solution and to convert the measured transmittance, T, into optical density (OD) by nonlinear transformation. The OD of the sample expresses the amount of light that has been lost in the sample and is linearly related to the total cross section by

OD = k>g(7o//) = 0.434i\riJtot, (1) where 70 and I are the intensities of incident and transmitted light and N is the number of particles in the total cross section

C^tot — ^sca. H~ (2)

The scattering can be divided into two groups: wavelength-independent scattering, where the size of the scatterer is very large compared to the wavelength of the light, and wavelength-dependent scattering, where the particles are of comparable sizes to or smaller than the wavelength of the light. The latter group contains the case of r/ A 5s0.05 (the Rayleigh condition), where Ais the wavelength of light and r the particle radius. Then, the scattering is purely dipolar with high-order scattering moments vanishing, and

Stt , £ — 1 (3) £+2 where cis the dielectric constant and k=2n! A.

(7 a .hs - 127rfcr 3 (4) where r'and t'are the real and imaginary parts of the dielectric constant, respectively. The ratio of scattering to absorption cross-sections scales with r3; hence, for a given value of e', the particle size is very important for determining the magnitude of the particle scattering. This produces the result that, under the assumption of no multiple scattering, a larger number of smaller spheres are less efficient scatterers than a smaller number of larger spheres, for a given mass of material.

For particles approaching the wavelength of light, the scattering is still dependent on the wavelength, but with a power less than 4. Heller [27] used the Stevenson [2& 29 ] extension of the Rayleigh equation to develop a relation for sizes just larger than Rayleigh scatterers. Here the wavelength exponent, g, i.e. the power of r/ A in the scattering efficiency, can be represented in closed form by

12(n 2 — 2) a2 + 5(t7 2 + 2) + 6(r2 - 2)a- (5) where

27trn a = Q < 1 (6) A and n is the ratio of the discrete phase to the continuous index of refraction.

For a more extensive deduction of the theory behind NIR measurements on asphaltene particles, the works of Mullins [30], Kallevik [3T\ and Joshi et al. [J2] are recommended for further details.

Results and discussion

In this study NIR spectroscopy was applied as a tool to probe the effect of various chemicals on asphaltene aggregates as a function of time and additive concentration.

In setting up the matrix for the chemicals some fundamental interaction patterns were considered. The starting point is that the dominating attractive forces for asphaltene aggregation are hydrogen bonding, acid-base interactions and charge transfer between aromatics. The main strategy for the disintegration of asphaltene aggregates should hence be to break these bonds and stabilize the smaller asphaltenes aggregates. The supposed interactions for the individual chemicals are further specified as follows.

• u-Alkylbenzenesulfonic acid: Dodecylbenzenesulfonic acid is known to efficiently disperse asphaltenes and form stable suspensions [33 ]. The mechanism behind the efficiency is assumed to be a strong acid-base interaction between the sulfonic acid headgroup and basic material in the asphaltene molecule. The alkyl chain is long enough (n-C]0„,3) to disperse the asphaltene molecules and to give steric stabilisation.

• Naphthenic acids: In crude oils the naphthenic acids are normally incorporated in the group of resin molecules. It is generally believed that these acids (a mixture of condensed ring structures with various numbers of rings and alkyl moieties, various positions of the COOH groups, various hydrophile-lipophile balance values, etc.) will possess various properties with regard to the asphaltenes. The strong interaction is between the acid groups and basic components in the asphaltene molecule. The dispersing power is determined by the molecular structure. Most probably various naphthenic acid structures will possess different dispersing powers. In this study we investigated the following naphthenic acids: 5-0-(H )-cholanoic acid (Choi), 1-napthalenepentanoic acid, decahydro (2C4), 1-naphthalenoic acid, decahydro-2-butyl (C42), crude naphthenic acid (CNA), Fluka naphthenic acid, and North Sea naphthenic acid. The structures and molecular weights are summarized in Table L

• Fatty alcohols: Fatty alcohols are normally efficient solvent molecules owing to their efficiency in breaking existing bonds and in forming new more favorable ones. Short-chain alcohols are believed to break down existing intermolecular hydrogen bonds between different asphaltene molecules and to replace them with alcohol-asphaltene hydrogen bonds. The alcohols used in this study were limited to 1-octanol, which is a normal paraffinic alcohol, 2- ethyl-l-hexanol, which is a branched paraffinic alcohol with the same number of carbon atoms, and benzyl alcohol, which is an aromatic derivative. •

• Fatty amines: The functionality of these is very much the same as for the fatty alcohols, i.e. the capability of replacing the hydrogen bonds with the amine group and to disperse with the alkyl moiety.

• Inhibitor A. This is a commercial blend consisting of fatty amines and acids in polar solvents.

The results obtained can be summarised as follows. The relative OD at 1,600 nm versus time for toluene/n-heptane/asphaltene mixtures with different amounts of CNA is shown in Fig. 2 The influence of additive upon aggregate size is depicted as the decrease of scattering as a function of time. The relative OD vs. time for toluene/n-heptane/asphaltene mixtures with various naphthenic acids is plotted in Fig. 3 . The commercial Fluka naphthenic acid and the naphthenic acid extracted from a North Sea crude seem to affect the state of the asphaltenes onlyto a minor extent. CNA is most efficient of these polydisperse naphthenic acids.

Time [min] 0 200 400 600 800 1000 1200

0.000 No additive: x

-0.005

2-ethyl- 1-hexanol £ -o.oio I 1-octanol 8 -0.015 X benzyl alcohol §

-0.020 hexylamme

-0.025 Inhibitor A

-0.030

Fig. 2. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 m-heptane/toluene mixture with crude naphthenic acid {CNA) added in various concentrations Time [min] 0 200 400 600 800 moo 1200

Fig. 3. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 n-heptane/toluene mixture with 1.25 wt% of various naphthenic acids added: 5- 3-(//)-cholanoic acid (Choi), 1-napthalenepentanoic acid, decahydro (2C4), 1- naphthalenoic acid, decahydro-2-butyl (C42), CNA, Fluka naphthenic acid (Fluka) and North Sea naphthenic acid {North Sea)

A comparison between 2C4, C42 and Choi shows these species to be somewhat more efficient than the previous group. Especially the 2C4 molecule has a very efficient breakdown to start with and also attains a low final value. It is thrilling to see that the molecular structure affects the results to this extent.

The other amphiphiles, presented in Fig. 4 , have a varying effect on the disintegration of the asphaltenes. The most efficient one is 1-hexylamine, which is very reminiscent of 2C4 with regard to short-term efficiency and the final state. The most efficient treatment is due to inhibitor A, the commercial mixture. Time [mn] 0 200 400 600 800 1000 1200

0,000

•0,005

2-ethyl-1-hexanol

-0.020

•0.025

Fig. 4. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 M-heptane/toIuene mixture with 1.25 wt% various amphiphiles added

Alkylbenzenesulfonic acid is shown to associate with the asphaltenes and create aggregates of increased sizes (Fig. 5). This is in accordance with the results obtained by Chang and Fogler [25 ] in a UV/vis spectroscopic study. They suggested that asphaltenes and dodecylbenzenesulfonic acid could associate into large electronic conjugated complexes.

Time [min] 0 200 400 600 800 1000 1200

-0.005

2-ethyl-1-hexanol

-0.025

-0.030

Fig. 5. NIR scattering measurements at 1,600 nm for 0.125 wt% asphaltenes in a 70/30 n-heptane/toluene mixture with 1.25 wt% w-alkylbenzenesulfonic acid (ABSA) added Conclusions

It was shown that NIR spectroscopy is a powerful method to follow the disintegration of asphaltene aggregates upon addition of chemicals. The method is based on the scattering from preferentially large aggregates. The NIR technique, which is very fast and accurate, is a good choice for the initial screening of large numbers of chemicals for asphaltene inhibition. The results show that additives which are efficient in replacing hydrogen bonds possess dispersive power and can serve as inhibitors. Commercial blends of active molecules gave the best results.

Acknowledgements. T.E.H, and I.H.A. acknowledge the technology program FLUCHAII, financed by the Research Council of Norway and the oil industry, for doctoral grants. We also express our appreciation to Statoil R&D Centre for the use of all necessary instrumentation in this research.

References

I .Speight JG (1991) The chemistry and technology of petroleum. Dekker, New York

2. Dickie JP, Yen TF (1967) Anal Chem 39:1847-1852

3. Mitchell DL, Speight JG (1973) Fuel 52:149-152

4.Strausz OP, Mojelsky TW, Town EM (1992) Fuel 71:1355-1363

5.Park SJ, Mansoori GA (1988) Energy Sources 10:109-125

6.Speight JR, Long RG, Trawbridge TD (1984) Fuel 63:616-620

7.Mansoori GA (1996) Arab J Sci Eng 21:707-723

S.Sjdblom J, Skodvin T, Holt O, Nilsen FP (1997) Colloids Surf A 123:593-607

9. Hunger JW, Li NC (1981) Chemistry of asphaltenes. Advances in chemistry series 195. American Chemical Society, Washington, DC

10. Hirschberg A, de Jong LNJ, Schipper BA, Meijer JG (1984) SPEJ, Soc Pet Eng 24(3):283-293

II .Mansoori GA, Jiang TS, Kawanaka S (1988) Arab J Sci Eng 13:17-34

12. De Boer RB, Leerlooyer K, Eigner MRP, Van Bergen ARD (1995) SPE Production and Facilities 2:55-61

13. McLean JD, Kilpatrick PK (1997) J Colloid Interface Sci 189:242-253

14. Andersen SI, Birdi KS (1990) Fuel Sci Technol Int 8:593-615

15. Nellensteyn FJ (1924) J Inst Pet Technol 10:311

16. Yen TF (1981) In: Hunger JW, Li NC (eds) Chemistry of asphaltenes. Advances in chemistry series 195. American Chemical Society, Washington, DC, p 39 17. Sheu EY, De Tar MM, Storm DA (1994) In: Sharma MK, Yen TF (eds) Asphaltene particles in fossil fuel exploration, recovery, refining, and production processes. Plenum, New York, p 155-165

lS.Pfeiffer JP, Saal RN (1940) J Phys Chem 44:139

19.Swanson J (1942) J Phys Chem 46:141-150

20. Koots JA, Speight JG (1975) Fuel 54:179-184

21. Hammami A, Femwom KA, Nighswander JA (1998) Pet Sci Technol 16:227-249

22. Andersen SI, Speight JG (2001) Pet Sci Technol 19:1-34

23. Gonzalez G, Moreira MBC (1991) Colloids Surf 58:293-302

24. Chang C-L, Fogler SH (1994) Langmuir 10:1749-1757

25. Chang C-L, Fogler SH (1994) Langmuir 10:1758-1766

26. Kerker M (1969) In: Loebl EM (ed) The scattering of light and other electromagnetic radiation. Physical chemistry. A series of monographs, vol 16. Academic, New York, p 666

27. Heller W (1964) J Chem Phys 40:2700

28. Stevenson AF (1953) J Appl Phys 24:1134

29.Stevenson AF (1953) J Appl Phys 24:1143

30.Mullins OC (1990) Anal Chem 62:508-514

31 .Kallevik H (1999) Characterisation of crude oil and model oil emulsions by means of near infrared spectroscopy and multivariate analysis. Department of Chemistry, University of Bergen

32. Joshi NB, Mullins OC, Abdul J, Creek J, McFadden J (2001) Energy Fuels 15:979- 986

33. Newberry ME, Barker KM (1983) US Patent 4414035 Paper VI INTERACTIONS BETWEEN ASPHALTENES AND

NAPHTHENIC ACIDS

Jenny-Ann Ostlund 1’3, Magnus Nyden 1

Inge Harald Auflem2’*, Johan Sjoblom 2,4

1 Department of Applied Surface Chemistry,

Chalmers University of Technology, 412 96 Goteborg, Sweden

2 Ugelstad Laboratory, Department of Chemical Engineering,

Norwegian University of Science and Technology, 7491 Trondheim, Norway

3 Bycosin AB,

Box 627, 651 14 Karlstad, Sweden

4 Statoil ASA,

Arkitekt Ebbelsvei 10, 7005 Trondheim, Norway

Corresponding author. E-mail: [email protected] Abstract

Asphaltenes from two different oil types were studied upon addition of two kinds of naphthenic acids by employing PFG-SE NMR (pulsed field gradient spin echo nuclear magnetic resonance) and NIR (near infrared) spectroscopy. The results implied that there were interactions between the asphaltenes and the acids. The dispersing effect of the naphthenic acids on the asphaltenes was also evaluated, and it appeared as if the effectiveness of the acids depended on the asphaltene type. Furthermore, a concentration series of one of the asphaltenes was prepared, and a dramatic decrease in diffusion coefficients upon increased concentration implied that the asphaltenes began to self-associate at concentrations above 0.1 wt-% of asphaltenes in toluene-d*.

Keywords: Asphaltenes; naphthenic acids; pulsed field gradient spin echo nuclear magnetic resonance; near infrared spectroscopy; diffusion; scattering; self-association; aggregation; interactions

Introduction

The need to characterize crude oils arises from the fact that they might cause severe operational problems. With a thorough characterization one can at an early stage predict the behavior of the crude oil with regard to emulsion formation, foaming and deposition of organic and inorganic precipitates. Such a prediction facilitates the planning of the whole exploration process with regard to sizing of equipment, use of chemicals and transport distances. Traditional characterization has involved a determination of central physical chemical properties of the crude oils. In this respect one often measures the content of

2 saturates, aromatics, resins and asphaltenes by means of the so-called SARA technique, which is normally based on high-pressure liquid chromatography (HPLC) combined with refractive index (RI) detection. The asphaltenes, which are the compounds in oil that have the highest molecular weight as well as the highest C/H ratio, are normally precipitated from solution before the SARA analysis.

Another important property to be measured is the particle size (and the size distribution) in these systems. The reason for this is that the asphaltene molecules self-assemble into aggregates of nanosized or larger dimensions. This association is facilitated by changes in for example pressure, 1"4 temperature 5,6 or solvency power of the surrounding medium.7,8 There is a clear need to follow these processes since the asphaltenes will stabilize emulsions and foams, and also cause operational problems as a result of deposition.

Traditional techniques to determine the particle sizes are normally based on light transmission or scattering. These methods might suffer from problems related to the dark color of the crude oils and to high amounts of scattering objects. However, the PFG-SE NMR (pulsed field gradient spin echo nuclear magnetic resonance) technique offers a rapid and accurate way to solve these problems. The method provides component resolved information concerning structural and dynamic aspects in complex mixtures9'11 and it has been proven suitable for studies of the diffusion of asphaltenes. 12

The objective of this study was to investigate mixtures of asphaltenes and naphthenic acids in a solvent phase of deuterated toluene and heptane. Naphthenic acids are among the naturally occurring components in most crude oils and the oil industry is usually concerned about problems with precipitation in the form of naphthenates. 13 One should, however, not forget

3 that naphthenic acids are polar molecules that are likely to interact with asphaltene molecules.

While asphaltene interactions with resins have received a lot of attention during the last decades14"16 possible interactions with naphthenic acids17 have largely been neglected. Thus, in this study PFG-SE NMR measurements were combined with NIR (near infrared spectroscopy) studies with the aim to evaluate potential interactions between asphaltenes and naphthenic acids. Furthermore, a concentration series with asphaltenes in toluene was prepared and studied, and information about self-association of the asphaltene molecules was obtained.

Experimental Section

Chemicals

The toluene and n-heptane that were used for extraction of asphaltenes from crude oil and as solvent in the NIR experiments were used as supplied from Merck. Both chemicals had a purity of >99%. Deuterated toluene (99.8 atom % D) and heptane (99 atom % D) used for the samples for the NMR measurements were supplied from Dr. Glaser AG, Basel. 5-p(H)- cholanoic acid [CHOL] and 1-naphthalenepentanoic acid, decahydro-, [2C4], had a purity of

>95% and >90%, respectively. The naphthenic acids were used as supplied from Chiron AS and their chemical structures are depicted in table 1. Characterization data from the two North

Sea crude oils and corresponding asphaltenes are summarized in tables 2 and 3, respectively.

The amounts of saturates, aromatics, resins and asphaltenes were measured with the SARA technique as described by Aske et al.18 .

4 Table 1 Schematics of the molecular structures of naphthenic acids utilized in this study

Molecular Purity Name Chemical structure Source weight (wt%)

5-(3(H)-cholanoic acid / 'm Chiron 361 >95 AS [CHOL] L 1' *

1 -Naphthalenepentanoic Chiron acid, decahydro 238 >90 AS [2C4] I

1 ..—Up to— :—30- wt % unsaturated bonds in cyclic part of molecular structure

Table 2. A summary of data obtained by the SARA technique and potentiometric titration

(TAN = Total Acid Number) on the two crude oils

Name Saturates Aromatics Resins Asphaltenes TAN

[wt%] [wt%] [wt%] [wt%] [mg KOH/lj

Oil 1 35.3 36.8 24.5 3.5 1.8

Oil 2 24.4 43.4 19.9 12.4 0.0

Table 3. Elemental composition and base number of the asphaltenes studied

Name C H N O Base number

[%] [%] [%] [%] [mg KOH/g]

Asphaltene 1 85.0 8.6 1.2 1.2 6

Asphaltene 2 78.8 8.0 0.6 1.3 3

5 Total acid number (TAN) of the crude oils was measured by a standard titration with

potassium hydroxide (KOH)19 The elemental composition and base number of the two

asphaltenes were obtained through atomic absorption and potentiometric titration with acid20,

respectively.

Extraction of asphaltenes

The asphaltenes were extracted from two North Sea crude oils by n-heptane precipitation as

follows: The mixture of crude oil and n-heptane (1:20 gram per ml) was left in a dark room

for 24 hours at 22 °C under gentle stirring. The precipitated asphaltenes were filtered off on a microfilter (pore diameter 0.45 pm) and then redispersed in n-heptane under stirring for two hours to remove remaining impurities from the crude oil. The asphaltenes were then once more filtered through a microfilter and dried under N% atmosphere at 115 °C until constant weight.

Preparation of samples for PFG-SE NMR experiments

The experimental matrix consisted of a concentration series of asphaltenes 1 in toluene-dg with concentrations between 0.040 and 4.0 wt %. In the rest of the experiments naphthenic acids [CHOL] or [2C4] were added to asphaltene 1 or, alternatively, asphaltene 2. The amount of acid was either 0.5 or 2.4 wt-% yielding 8 samples in total. The heptane-di6 to toluene-dg ratio was chosen as 30/70 (wt / wt %) with the aim of getting as close to the precipitation point as possible, while still producing stable samples for the PFG-SE NMR measurements.

The heptane/toluene ratio of 30/70 (wt / wt %) was considered stable after n-heptane titration tests had been performed using NIR.

6 The weighted amounts of asphaltenes and naphthenic acids were left to dissolve for 8 hours in toluene-dg. Heptane-d,6 was then added and the mixtures were heated and left at 70 °C for one hour to ensure that the acids were properly dissolved. The resulting solutions were placed in an ultrasonic water bath for 15 minutes for homogenization. Finally, the mixtures were added to 5 mm NMR-tubes that were flame-sealed. PFG-SE NMR measurements were performed during the next couple of weeks.

Preparation of samples for NIR experiments

The weighted amount of asphaltenes was dissolved in toluene for 12 hours and then placed in an ultrasonic water bath for 30 minutes before the n-heptane was added. The solutions were then allowed to equilibrate for a week with continuous stirring before the addition of naphthenic acid and subsequent measurements. In the experiments we used a ratio of 50/50

(wt / wt %) «-heptane/toluene was used with the intention to stay at or slightly above the precipitation point. The concentration of asphaltenes and naphthenic acids in each sample were 0.5 wt % and 0.5 wt %, respectively. The results were logged continuously with one- minute intervals for a period of 100 minutes, with continuous stirring to prevent any possible sedimentation to influence the results.

PFG-SE NMR measurements

The PFG-SE NMR experiments were performed on a Unity Inova 500 MHz spectrometer and an Oxford magnet equipped with a diffusion probe from DOTY Sci. Inc., USA. The pulse sequence used for the diffusion measurements was a stimulated echo where the gradient pulse duration (<5) and the experimental observation time (A) were kept constant at 4 and 70 ms,

7 respectively. A sine-shaped gradient was used to minimize the effect of eddy-currents. The gradient strength (g) was varied in 41 or, in the case when naphthenic acid had been added, 51 linear steps from 0 to a maximum value chosen so as to obtain a hundredfold decrease of the signal attenuation. All experiments were performed at 23 °C.

NIR measurements

The NIR measurements utilized a Brimrose AOTF Luminar 2000 spectrometer with a fiber optic sampling probe for transflectance measurements. The experiments were performed at a temperature of 22 °C and a total path length of 1 mm. Each spectrum was an average of 32 scans. The wavelength used for monitoring the variations in aggregate size was 1600 nm. The spectrum at time = 0 for each time series was used as a background spectrum and subtracted from the succeeding spectra. This was done to ensure that the different samples could be compared even if there was a small inaccuracy in weighting of the asphaltenes in each sample.

Theoretical considerations

NIR theory

The absorption spectrum in the near infrared region covers the wavelengths from 780 to 2500 nm (12820 to 4000 cm'1). It consists of overtones and combinations of the fundamental molecular vibration bands that originate from hydrogen bond stretches in C-H, N-H, S-H and

O-H. By measuring the relative amount of light transmitted through a solution, the light extinction profile in the near infrared spectral range can be recorded. This diminution of light

8 can be accounted for by two separate mechanisms; electronic absorption by organic molecules

and optical scattering from particles or aggregates.21 The scattering can be divided into two

groups: wavelength-independent, where the size of the scatterer is very large compared to the

wavelength of the light, and wavelength-dependent, where the particles are of comparable

sizes or smaller than the wavelength of the light. The latter group contains the case of r/X <

0.05 (the Rayleigh condition) where X is the wavelength of light and r the particle radius. The

scattering cross section

S — 1

section is dependent upon the radius of the particle raised to the sixth power. For the absorption cross-section, %», the dependency is proportional to the third power of the particle radius:

= \ 2nkr (g' + 2y+f" (2)

Here e' and e is the real and imaginary part of the dielectric constant, respectively. Hence, for a given value of e", the particle size is very important for determining the magnitude of the particle scattering. The total cross section, atol, is then the sum of the contribution from the scattering cross-section and the absorption cross-section:

(3)

Further, the total cross section is linearly related to optical density (OD), which expresses the amount of light lost in the sample due to absorption and scattering:

OD = log(/ 0 /1) = 0.434/VX, (4)

9 lo and / are the intensities of incident and transmitted light and N is the number of particles in

the total cross section atot.

PFG-SENMR theory

The PFG-SE NMR method 9,10 measures molecular diffusion. For single diffusing species the

attenuation of the spin echo is given by:

I - Iq exp(-£Ds) (5)

where / is the integrated area of the peak of interest, /«the intensity of the signal at g = 0 and,

in the case of sine-shaped pulses:

k = Y1gS 1 (4A -8)/ n 2 (6) where y is the gyromagnetic ratio of the nuclei (in this study *H), 8 the gradient pulse

duration and A the effective diffusion time. By applying a non-linear least-square fit of Eq. 5 to the experimental data the self-diffusion coefficient, Ds, can be evaluated. However, in

many cases the diffusing species is polydisperse and the echo-decay is then better represented

by a function containing a distribution, P(D), of diffusion coefficients:

I = 7 0 \P(D)exp(~kD)dD (7)

The functional form of the distribution function may vary, but in this case it was noted that a

log-normal distribution described the data well:

1 ' fln(D)-ln(D.)^ exp (8) Dajln V2<(7 where Dm is the mass weighted median self-diffusion coefficient and cris the standard deviation of the logarithm of the diffusion coefficient.

10 If, in addition to a continuous distribution of diffusion coefficients, a signal from one monodisperse species (such as for example a solvent) is contained in the NMR signal Eq. 5 and Eq. 7 may be combined into:

( I = h /, \P(D) exp( - kD)dD + /„ exp (~kDn ) (9) where ft is the fraction that is described by a distribution in diffusion coefficients and f„ the fraction that diffuse with single diffusion coefficients D„. It was noted that Eq. 9 was sufficient to describe the echo decay in the case of asphaltenes in toluene-d*. When studying the samples containing both naphthenic acid and asphaltenes Eq. 9 was modified to contain one more single exponential function.

It is well known that in order to extract the relevant parameters in Eq. 9 from a fit, the experimental data must be of a very high quality. In this case the relevance of the outcome of the fit was verified with a Monte Carlo calculation 22 of the error in each fitted parameter. The results from such calculations are displayed as error bars in the figures. It is shown that the obtained parameters are indeed relevant since the error bars are very small due to the high quality of the data. In addition to this type of data evaluation, the program CORE written by

Stilbs 23 24 and designed to use the information contained in the full NMR spectrum, was used as a complementary evaluation. CORE is also model dependent when it comes to describing functional forms of the echo-decay. For example, a distribution in diffusion coefficients is conveniently described by a stretched exponential. One of the very nice features of CORE is that it calculates the bandshape for the individual diffusing species that has been decided on in the model. Thus, a stretched exponential plus two single exponentials then gives three different bandshapes. By starting with a simple model such as s single exponential function,

11 the model is refined until CORE calculates the bandshapes that correspond well to the NMR spectrum from the pure components.

Results and discussion

Concentration series studied by PFG-SE NMR

Results from PFG-SE NMR measurements of the concentration series of asphaltenes 1 dissolved in toluene-dg are presented in Fig. la. As can be seen, the median diffusion coefficient of the asphaltenes decreased as a function of increased asphaltene concentration.

The decrease was fast and the rationale for these results may be that the asphaltenes self- associated and/or that the obstruction effect was considerable. Earlier observations showed that the obstruction effect in asphaltenic systems is significant due to the asphaltenes having a oblate shaped structure.12 However, the decrease observed in this system was even more drastic than that previously reported. It was thus likely that the asphaltenes investigated were not only subjected to obstruction but also to self-association.

It is common in surfactant systems, where the exchange of monomers between the bulk and the micelles is fast on the timescale of the experiment, to apply a two-site model for evaluation of the diffusion of micelles Dmic:25 ’26

Q'obs ~ Pbound ^mic Pjree^jree 0^)

where Dobs is the observed average diffusion coefficient of free and bound asphaltene molecules in the system, Dfree is the diffusion coefficient of free asphaltene molecules in the

12 system and p is the fraction of unimers either free or bound. The following relationships holds

for the two fractions: Pfree-i-pbound=CMC/Ctomi. In this investigation the CMC was regarded

as the onset of flocculation, which appeared to be 0.1 wt-%. A CMC of 0.4 wt-% asphaltenes

in toluene has been reported previously in the literature.27 From the experimental data Do was

extrapolated at infinite dilution to 5.7- 10"10 m2/s. The calculated values of Dmlc, at infinite dilution, using Eq. 10 have been included in Fig. la and b.

Figure 1.

The decrease in diffusion that can be expected due to obstruction, if the particles have a disc­ like geometry, is displayed as a full line in Fig. lb. The theoretical decrease in diffusion was calculated employing a model proposed by Jonstromer et al.28 Calculations of the diffusion was based on the assumption of an axial ratio of 1:20 and Do,miceiie~3.5-10" 10 m2/s, as determined by extrapolation. As can be seen in Fig. lb, a good agreement between the theoretical curve for monodisperse discs and the values of Dmtc of the asphaltenes is obtained.

The North Sea asphaltenes appeared to be more inclined to self-associate than did the asphaltenes of Venezuelan origin investigated previously. 12 The reason for this is most likely the complex and diverse chemical structure of the asphaltenes, which most certainly will influence the tendency of the asphaltenes to self-associate. The properties of the crude oil, different extraction procedures 29 and aging will also influence the association behavior. Sheu et al.30 stated that the self-association of asphaltene aggregates is limited while the association between monomers is high. This indicates that a possible driving force behind asphaltene association could be electron donor-electron acceptor interactions.

13 a. 7 10'10

6 10'°

5 10 10

E, Q 4 IQ'10

3 10"'°

2 10 10 0.01 0.1 1 10 c (wt-%) b.

c (wt-%)

Figure 1. (a) The median diffusion coefficients are displayed as a function of the asphaltene concentration (O).

Also included are the diffusion coefficients of the asphaltene micelles (■) as calculated by Eq. 10. (b) The calculated values of the diffusion of the asphaltene micelles (■) are also shown in this figure. The full line illustrates the decrease in the diffusion coefficients that was expected only due to obstruction (under the assumption that the micelles are monodisperse and oblate shaped with an axial ratio of 20).

14 Another mechanism behind the association suggested in the literature is hydrogen

bonding, 31,32 where the polarity / H-bond acceptor capacity will determinate the level of self­

association.

The diffusion of naphthenic acids and asphaltenes studied by PFG-SE NMR

When samples that contained both asphaltenes and naphthenic acid were studied, it was

observed that the entire signal from 5-(3(H)-cholanoic acid [CHOL] appeared at the same

frequency (0.7-2.1 ppm) as did the signal from the asphaltenes. The complete overlap of the

signals complicated the evaluation of the samples containing [CHOL]. However, as

previously stated, the parameters obtained from the fit were relevant and the values were

verified by the use of CORE. The situation was different when 1 -naphthalenepentanoic acid,

decahydro-, [2C4] had been added to the asphaltenes. Although the majority of its peaks

appeared between 0.7 and 2.1 ppm, this acid had a peak at 4 ppm. There was no overlap with the signal from the asphaltenes in this region. Hence, it was possible to study the diffusion of

[2C4] without any contribution from the asphaltenes and it was observed that the echo decay

of [2C4] was biexponential in the presence of both asphaltene 1 and asphaltene 2. The

diffusion coefficients of [2C4] are shown in Fig. 2.

Figure 2.

The signal intensity of naphthenic acids in pure toluene was monoexponetial. It was therefore concluded that the acids were monodisperse and not very prone to self-associate. An interesting observation was that the slow diffusion of [2C4] was close to the diffusion of the asphaltenes (2-1 O'10 m2/s), while the fast diffusion was the same as the diffusion of pure acid

15 in solvent (l-10"9m2/s). The fact that two different diffusion coefficients were detected when

asphaltenes were present strongly indicated that there were monomeric acid as well as

associated acid in the samples. Furthermore it was noted that the exchange dynamics between

the two sites was slow compared to the experimental diffusion time of 70 ms. Thus, it

appeared as if a fraction of [2C4] interacted with both kinds of asphaltenes.

1.80 10"9

1.35 10 9

£ 9.00 10'° o

4.50 IQ10

Tf in 'f. in CM B CL o. TT T o O U O C\i_ CN CN CM T-" CN cm" < < < <

Figure 2. The diffusion coefficients of [2C4] at different concentrations (0.5 or 2.4 wt-%) in samples containing asphaltene 1 (Al) or asphaltene 2 (A2). A function containing two exponentials was fitted to the data and a low

(■) as well as a high (•) diffusion coefficient were found for [2C4] in the different samples, probably arising from bound and free acid, respectively.

16 We were also interested in extracting information from the signals that were overlapping.

However, we faced a delicate problem when evaluating the echo decay arising from these

combined signals, since extracting quantitative values by fitting complex functions to

experimental data is not easy. Nevertheless, the fit of Eq. 9 to the experimental data using a

Levenberg-Marquardt algorithm was seen to give reasonable results. A typically example is

shown in Fig. 3.

0.0006

0.0004 I •»

0.001 5.0 10 1.0 10 1.5 10 2.0 10 2.5 10 k (rad2s/m2)

Figure 3. The echo attenuation of a typical sample containing both naphthenic acid and asphaltenes. The full line represents the fit of Eq. 9 to the experimental data. Inserted in the top comer are the residuals from the fit. It can be seen that the error are small and random, indicating that the agreement between theory and data is adequate.

17 Figure 3.

The results from the fitting of Eq. 9 to the data are displayed in Fig. 4. As previously

mentioned the fitted results were also analyzed by Monte Carlo simulations. 22

Figure 4.

Although the error bars were small a second evaluation of the data, this time with CORE, was

performed. The diffusion coefficients obtained with this program were the same as the ones

evaluated by Eq. 9. CORE presents the bandshape of each diffusing species (the number of

diffusing species is determined manually by the experimentalist). The bandshape for the pure

components is known and hence it is relatively simple to decide if the model for the different

diffusing species and the functional form of their echo-decay as determined by CORE is

correct. That is, if CORE presents a bandshape that corresponds well to what is known the

model chosen should be correct.

It is interesting to note from the results shown in Fig. 4 that the diffusion of asphaltene 2

decreased in all cases independently of what kind of naphthenic acid that had been added.

This indicated that both [CHOL] and [2C4] interacted with asphaltene 2. The diffusion coefficient of asphaltene 1, on the other hand did not change and thus, it appeared as if there were no or only weak interactions between the naphthenic acids and asphaltenes of type 1.

18 1.2 10

1.0 10 s

8.0 10"'° ? &

6.0 10"'°

4.0 10"'° * 1 M H

2.0 10"'° X U) 2 S' B < B B B B B B B x o u o o o 3 o o X X CM X X q q cm" CM o o < < v- < < CM CM < < < <

Figure 4. The results from samples containing naphthenic acid and asphaltenes. (■) corresponds to the diffusion of the naphthenic acid (0.5 or 2.4 wt-% of [CHOL] alternatively [2C4]) while (•) corresponds to the diffusion of asphaltene 1 (Al) or asphaltene 2 (A2). With the aim of facilitating interpretation, the diffusion of only asphaltene 1 or 2 in toluene-dg (reference samples) has been included. Frames have been put around the diffusion coefficients from asphaltenes of the same kind (Al or A2).

The influence of naphthenic acids on aggregate size of asphaltenes as studied by NIR

Systems containing both asphaltenes and naphthenic acids were also studied by NIR. The NIR experiments were performed upon systems where the asphaltenes were slightly above the precipitation point, as opposed to the PFG-SE NMR experiments where the systems were below this point. Fig. 5 shows the NIR measurements on the two types of asphaltenes as a function of time after the addition of the naphthenic acids [CHOL] and [2C4]. If the naphthenic acids have a dispersing effect on the asphaltene particles, the particle size should decrease, as would also the optical density. The basis for this reasoning is that large particles

19 scatter light more strongly than small particles. Assuming that no chemical reactions occur

that would change the concentration of asphaltenes, the only parameters left to consider are

size and number of the asphaltene particles. In other words: a decrease in optical density must

be due to a change from a smaller number of large particles to a larger number of small ones.

Figure 5.

The results displayed in Fig. 5 show that when naphthenic acids [CHOL] or [2C4] were added

to solutions containing asphaltene 1, the reduction in particle size was minimal for both acids.

Time [min] 0 20 40 60 80 I0(

-5.E-03

Figure 5. The change in optical density (scattering) of asphaltene 1 (Al) and asphaltene 2 (A2) is displayed as function of time after addition of naphthenic acid ([CHOL] or [2C4]). The spectrum at time = 0 was used as reference and has been subtracted from the subsequent spectra, thus eliminating the contribution from absorption to the optical density.

20 When the same acids were added to solutions of asphaltene 2 the reduction in optical density and, consequently, in particle size was much more distinct. It therefore seemed as if naphthenic acids had a stronger dispersive effect on asphaltene 2 than they had on asphaltene

1. Rogel et al. showed that the effectiveness of an inhibitor to hinder asphaltene precipitation, decreased as the base number of the crude oil increased. 33 It is interesting to note that asphaltenes 1 have a higher base number and also appeared to be less affected by the naphthenic acids than did asphaltenes 2.

Conclusions

The diffusion coefficients of asphaltene 1 were observed to decrease rapidly as a function of increased concentration of asphaltenes in pure toluene-dg. The drastic decrease could not be explained solely by a non-spherical structure of the asphaltenes. Hence, it was proposed that there was an onset of flocculation at 0.1 wt-% asphaltenes. When comparing the decrease of the diffusion coefficients with theory, it appeared likely that the asphaltenes were oblate shaped aggregates with an axial ratio of approximately 1:20.

From the PFG-SE NMR measurements of the samples containing both naphthenic acid and asphaltenes it appeared as if [2C4] interacted with the asphaltenes, possibly through acid-base interactions. The evaluation of the diffusion of the asphaltenes also suggested that the degree of interactions between asphaltenes of type 2 and [CHOL] and [2C4] was higher than that of asphaltenes 1 with [CHOL] and [2C4].

The NIR measurements, where the asphaltene particle size was followed as a function of time after addition of naphthenic acid, supported these results. It was shown that the particle size

21 was reduced significantly more for asphaltene 2 than for asphaltene 1, upon addition of both

[CHOL] and [2C4],

Acknowledgments

Inge Harald Auflem would like to acknowledge the technology program FLUCHAII, financed by The Research Council of Norway and the oil industry, for a PhD grant.

Jenny-Aim Ostlund acknowledges Bycosin AB (subsidiary of Octel Corp.) and the

Foundation for Knowledge and Competence Development in Sweden for financial support.

The authors would like to thank Statoil ASA for the kind permission to publish the results.

The Swedish NMR centre in Goteborg is thanked for granting spectrometer time. Professor

Peter Stilbs is thanked for being helpful and explanatory about CORE. Professor Krister

Holmberg is acknowledged for valuable discussions and comments on this manuscript.

REFERENCES

(1) Joshi, N.B.; Mullins, O.C.; Abdul, J.; Creek, J.; McFadden, J. Energy Fuels 2001,15,

979-986.

(2) Hirschberg, A., De Jong, L.N.J., Schipper, B.A., Meijer, J.G. Soc. Pet. Eng. AIME,

1984, 283-293.

(3) Park, S.J.; Mansoori, G.A. Energy Sources 1988, 10, 109-125.

(4) Mansoori, G.A.; Jiang, T.S.; Kawanaka, S. Arabian J. Sci. Eng. 1988, 13, 17-34.

(5) Andersen, S.I., Birdi, K.S. Fuel Sci. Technol. Int. 1990, 8,593-615.

(6) Speight, J.G. The chemistry and technology of petroleum, Marcel Dekker: New York,

1999.

22 (7) McLean, J.D.; Kilpatrick, P.K. J. Colloid Interface Sci. 1997,196, 23-34.

(8) McLean, J.D., Kilpatrick, P.K. J. Colloid Interface Sci. 1997,189, 242-253.

(9) Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1987,19,1-45.

(10) Callaghan, P.T. Aust. J. Phys. 1984, 37, 359-397.

(11) Price, W.S. Concepts. Magn. Reson. 1997, 9,299-336.

(12) Ostlund, J.-A.; Andersson, S.-I.; Nyden, M. Fuel 2001, 80, 1529-1533.

(13) Rousseau, G.; Zhou, H.; Hurtevent, C. 2000. Aberdeen, UK: Society of Petroleum

Engineers Inc.

(14) Leon, O.; Rogel, E.; Espidel, J.; Torres, G. Energy Fuels 2000,14, 6-10.

(15) Roots, A.J.; Speight, LG. Fuel 1975,54, 179-184.

(16) Schabron, J.F.; Speight, J.G. Pet. Sci. Technol. 1998,16, 361-375.

(17) Auflem, I.H.; Havre, T.E., Sjoblom, J. ColloidPolym. Sci. Accepted.

(18) Aske, N.; Kallevik, H.; Sjoblom, J. Energy Fuels 2001,15, 1304-1312.

(19) ASTMD-974

(20) ASTM D2896-74

(21) Mullins, O.C. Anal. Chem. 1990, 62, 508-514.

(22) Alper, J.S., Gelb, R.I. J. Phys. Chem. 1990, 94,4747-4751.

(23) Stilbs, P.; Paulsen, K.; Griffiths, P C. J. Phys. Chem. 1996,100, 8180-8189.

(24) Stilbs, P. J. Magn. Reson. 1998,135, 236-241.

(25) Soderman, O., Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1994, 26,445-482.

(26) Drakenberg, T., Lindman, B. J. Colloid Interface Sci. 1973, 44,184-186.

(27) Andersen, S.I„ Birdi, K.S. J. Colloid Interface Sci. 1991,142, 497-502.

(28) Jonstromer, M.; Jonsson, B.; Lindman, B. J. Phys. Chem. 1991, 95, 3293-3300

(29) Alboudwarej, H.; Beck, H.; Svrcek, W.Y., Yarranton, H.W. Energy Fuels 2002,16,

462-469

23 (30) Sheu, E.Y.; Storm, D.A.; DeTar, M.M. J. Non-Cryst. Solids 1991,131-133, 341-347.

(31) Mohamed, R.S.; Ramos, A.C.S., Loh, W. Energy Fuels 1999,13, 323-327.

(32) Chang, C.-L.; Fogler, S.H. Langmuir 1994,10, 1749-1757.

(33) Rogel, E.; Leon, O.; Espidel, Y.; Gonzalez, Y. SPE 2001, 5, 84-88.

24 Paper VII 25

Demulsifiers in the Oil Industry

Johan Sjoblom, Einar Eng Johnson, and Arild Westvik Statoil A/S, Trondheim, Norway Marit-Helen Ese and Jostein Djuve University of Bergen, Bergen, Norway Inge H. Auflem and Harald Kallevik Norwegian University of Science and Technology, Trondheim, Norway

I. INTRODUCTION the stabilization of W/O emulsions. In most cases it is, however, not possible to pick out one single component According to our traditional understanding of emul ­ responsible for the stability of the dispersed droplets, sion formation and stabilization, there is a need to but several components/fractions in parallel are the introduce mechanical energy and stabilizing agents to source for the stability of the emulsion. The compo ­ a water/oil mixture in order to create stable emulsions. nents/fractions in the crude oil show a large range of In bench experiments the energy input is typically con ­ molecular weights. Lighter components like the resins trolled by means of different kinds of rotors and can act as individual monomers in a similar manner to homogenizers. Under real oil-production conditions, traditional surfactants. The driving force for their pressure gradients over chokes and valves will guaran ­ action is the presence of water (and the existence of a tee that there will be a sufficiently high mechanical W/O interface). Usually the low molecular weight resins energy input in order to rupture original solution struc­ have a tendency to be the most interfacially active, i.e., tures and to form new fresh W/O interfaces. Obviously to reach first and cover a fresh W/O interface. However, the magnitude of the pressure gradient over the choke/ this is mostly a necessary requirement but not a suffi­ valve will be decisive for the droplet size distribution in cient one for the formation of stable W/O emulsions. the fresh emulsions. Since the transport over a choke/ The next step in the stabilization process involves inter ­ valve will mean the creation of a new W/O interface action with the heavier crude oil components, i.e., the the nature of the emulsion before and after the valve/ asphaltenes. Depending on the production history and choke may differ significantly (1-5). the fluid properties these molecules can be either in a The lifetime of the emulsion (and the retention time monomeric or associated state. In the latter case small in the full-scale separator) depends on the kind of sta ­ particles are formed. The formation of these is normally bility mechanisms involved. There exist several possibi ­ a result of the stacking tendency of the individual lities of finding stabilizing agents (or solid fines) in either asphaltene molecules. These nanosized particles will the crude oil itself or in added production chemicals. have a strong tendency to accumulate at the W/O inter ­ Among the indigenous stabilizers, asphaltenes/resins/ faces, if the solution conditions or changes herein so porphyrins are mentioned as possible candidates for favor. Obviously the final particle size or the fioccula-

595 596 Sjoblom et al.

tion of these nanosized particles is critical with regard to of the Wilhelmy method (8-10). Modification of the the stabilizing capacity of these entities (6, 7). trough design has made it possible to carry out the It is of crucial importance to understand the stabi ­ same kind of experiments on a liquid/liquid interface, lizing mechanisms when discussing demulsifiers and i.e., the oil/water interface, instead of on the liquid the efficiency of these. In this chapter we are going to surface. A prototype trough has been designed by discuss different types of demulsifiers, i.e., from simple KSV Chemicals in collaboration with the solvents to intriguing macromolecules. It is also our University of Compiegne, France (11). The trough, intention to view how new instrumentation can reveal entirely made of Delrin, is a “double ” trough (Fig. important and to some extent unexpected properties of 1) where the barriers contain holes to allow the flow these chemicals. In this chapter we introduce, in addi ­ of the light phase as the compression of the interface tion to conventional techniques, the use of Langmuir proceeds. The Wilhelmy plate is first placed in the and Langmuir-Blodgett techniques, atomic-force aqueous phase, then the oil phase is added until the microscopy (AFM) and near-infrared spectroscopy plate is totally immersed. (NIR), when analyzing the effects of the demulsifying The most common and adequate way of presenting chemicals. We can also for the first time report on the results obtained from the Langmuir technique is a destabilization experiments (with demulsifiers) at ele ­ plot of surface pressure as a function of the area of vated pressures. These experiments have been carried surface available to each molecule, i.e., the mean out in a special separation rig constructed for Statoil. molecular area. The measurements are carried out at a constant temperature and are known as surface pressure/area isotherms (Fig. 2). The film is com ­ pressed at a constant rate by the moving barriers II. EXPERIMENTAL TECHNIQUES while the surface pressure is continuously monitored. A. The Langmuir Technique Generally, a number of distinct regions are apparent on examining the isotherm. As the surface area is The Langmuir technique is used in order to charac ­ reduced from its initial high value, there is a gradual terize monolayer properties of surface-active materi ­ onset of surface pressure until an approximately hor­ als. The instrumentation consists of a shallow izontal region is reached. In this region the hydropho ­ rectangular container (trough) in which a liquid sub­ bic parts of the molecules, originally distributed near phase is added until a meniscus appears above the the water surface, are being lifted away. However, rim, whereupon the film is spread. The barrier for this part of the isotherm is often not resolved by manipulation of the film rests-across the edges of the the apparatus, because the surface pressure at which container. For a more thorough description of the this occurs is usually quite small (< 1 mN/m) due to experimental setup, Petty and Barlow (8) are recom ­ the weakness of interaction between water and the mended. The surface pressure is measured by means tailgroups. This region is followed by a second abrupt

Cross-section

100

) 8 64 101- 48

Q-Q'Q'Q m glptae ,,, I % 1 mK^jiqueous phase ^ 'A aqueous phase

Figure 1 Schematic drawing of the liquid-liquid interfacial trough (size in mm). Demulsifiers in the Oil Industry 597

Surface Pressure movement of the molecules in the film to form a new /mN m" phase. In order to investigate the stability of a monolayer, the area loss at constant surface pressure or the decrease in surface pressure at constant area is mea ­ sured. Different destabilization mechanisms are illu­ strated by the shape of the relative area relaxation isotherms in Fig. 3. The compression is stopped at a predetermined surface pressure and the relative area loss is plotted as a function of time. Curve (a) in Fig. 3 shows the behavior of a totally stable film, with no 30 - area reduction. Isotherm (b) shows an initial area loss, which is attributed to structural rearrangements in the monolayer to form a coherent close-packed film (16- 19). This process depends on the rate of compression. Fast barrier movement creates a higher degree of dis ­ order in the monolayer, and the initial area loss is increased. A continuous decrease in area, as illustrated by curve (c), is the result of a slow dissolution of the film-forming material into the subphase or evaporation of the monolayer This film loss increases with surface pressure, and is a consequence of the huge volume difference between the material in the film and the Area per moiecule/nm ’ subphase liquid and gas phase to which the monolayer Figure 2 Typical surface pressure/area isotherm of stearic is exposed. Even a low solubility or vapor pressure acid on an acidified water subphase. may lead to destabilization of the film due to solution or evaporation (20). Isotherm (d) in Fig. 3 is character ­ transition to a steeply linear region, with an approxi ­ ized by a relaxation rate that increases with time. This mately constant compressibility. kind of behavior is observed for systems that undergo Further reduction in surface area results in an nucleation and further growth of the film material into abrupt increase of slope, and hence reduced compres ­ bulk fragments (19, 21). sibility. All these different regions indicate different states of the monolayer, and analogously to bulk mat ­ ter these are characterized as gas-, liquid- and solid-like (12-14). Additional techniques (15) such as X-ray scat ­ tering, electron diffraction, fluorescence, polarized fluorescence, atomic-force microscopy, and Brewster angle microscopy have proved the existence of meso- phases in Langmuir films. The complexity of behavior is a result of the differerent intermolecular interactions in the film (alkyl group/alkyl group and polar group/ polar group interactions) and between the film and the subphase (polar group/subphase interactions). Hence, the interaction forces would undergo certain changes, which would be related to the packing of the molecules in the two-dimensional plane. More details may be Time (min) found in the books by Gains (9) and Birdi (10). Figure 3 Relative relaxation curves at constant surface A sharp break at small areas in the fl-yf isotherm is pressure for monolayers showing different stabilities: attributed to the collapse of the monolayer under the (a) stable monolayer; (b) rearrangements of the film mole­ given experimental conditions. In general, the collapse cules; (c) dissolution of film molecules into the subphase; pressure is the highest surface pressure to which a (d) collapse by nucleation and growth of bulk solid frag­ monolayer can be compressed without a detectable ments. 598 Sjoblom et al.

B. Langmuir-Blodgett Deposition The AFM technique exploits the forces that exist between atoms and molecules. The force exerted The most commonly used process of transferring a upon a tip mounted on to a cantilever (with a floating insoluble monolayer to a solid surface is known spring constant, weaker than the equivalent Langmuir's original method (22). A clean wettable spring constant between atoms) is monitored as the solid is placed in the subphase before a monolayer is tip passes over the surface. Measurements of the spread, and then drawn up through the surface after cantilever deflection, which is proportional to the formation of the film. The transfer process is critically magnitude of the force, during the scan make it pos ­ dependent on the surface pressure, so it is desirable to sible to obtain images of the surface topography. All maintain a constant pressure as the film is removed types of materials exert these forces, so there is no from the surface. Using this kind of deposition techni­ restriction regarding composition of the analyzed que, both the film and a thin layer of water are trans ­ components. Another important advantage with ferred to the solid substrate. The water is later removed AFM is that it can operate in a variety of environ ­ by drainage or evaporation, leaving a monolayer on ments. It is especially convenient that the measure ­ the solid surface. ments may be performed in air at atmospheric The rate of deposition is an important factor. pressure. In addition, AFM may also be operated Optimal values of this parameter depend partly on in liquids. the rate of drainage of the intervening liquid film There are three scanning modes for AFM, i.e., from the monolayer/slide interface and partly on the contact mode, noncontact mode, and tapping mode dynamic properties of the monolayer on the liquid sur­ (24). In the contact mode the tip is touching the face, i.e., the film viscosity. sample surface where the repulsive forces dominate, When conducting structural studies of Langmuir- while the attractive forces dominate in the noncon ­ Blodgett (LB) films, careful consideration has to be tact mode. The tapping mode represents a compro ­ taken of possible effects that might arise during film mise between these two, giving better resolution deposition. Irregularity in the dipping motion may than the noncontact mode and is not as damaging result in formation of striations in the deposited for the sample as the contact mode. A tip in contact layer. Trapped water droplets between the film and with the surface may generate extremely high pres ­ the solid surface is another possible reason for imper ­ sures on the small contact area between the tip and fections in the monolayer. Circumstances such as low the sample, which may result in indentation of the surface pressure or weak interactions between the film- tip in soft materials. In tapping mode the cantilever forming molecules, i.e., the monolayer is not coherent, is oscillating near its resonance frequency, with a may result in deposition of irregular films. This is prob ­ high enough amplitude to allow the tip to dip per ­ ably a result of expansion, contraction, or flow during iodically in the contamination layer. Measuring the the transfer process. change in amplitude or phase for the oscillating can ­ A solid surface, which is not smooth on a molecular tilever provides images of the surface (Fig. 4). scale, may lead to problems when transferring a mono- For more information regarding the use of AFM molecular film on to it. At the moment of deposition and related probe techniques for imaging LB films, the film may bridge over the surface roughness, espe­ review articles by Zasadzinski et al. (25) and DeRose cially if the film is closely packed and under high sur­ and Leblanc (26) are recommended. face pressure. This kind of bridging is often supported by the intervening water layer, so when this layer is D. Test Procedures for Demulsification removed the film may collapse. It is commonly known that the administration of che­ C. Atomic-force Microscopy micals is very essential. Depending on the administra ­ tion procedure one can expect different efficiencies of Atomic-force microscopy (AFM) (23) is a nearly ideal, the demulsifiers. Different administration procedures high-resolution method in providing a molecular-scale are reviewed in the following. topographic view of a variety of solid surfaces, The traditional testing of demulsifiers is to under ­ organic, inorganic, or biomolecular. Under optimal take bottle-shake tests. In these tests one has a pre ­ conditions this microscopic technique is capable of mixed emulsion and the chemical under study is producing images showing details of molecular resolu ­ applied. After this the bottle is gently shaken in tion in LB films. order to distribute the chemical evenly into the emul- Demulsifiers in the Oil Industry 599

position sensitive detector

cantilever with probe tip

Figure 4 Schematic representation of an atomic-force microscope.

sifted system. The efficiency of the chemical applied is E. Separation Rig read from the resolution of the dispersed phase in volume as a function of time. Normally the size and A separation rig, as illustrated in Fig. 5, is used to shape are such that area effects can be neglected The prepare W/O emulsions and monitor the separation demulsification normally progresses under stagnant of them. The principle is that two pressurized fluids conditions with no mechanical energy input. When meet just before a choke valve (VD1) and flow the coalescence is in progress there will be a resolu ­ through the valve into the separation cell. The two tion of the dispersed phase as a function of time, i.e., fluids can either be oil and water or for instance two so-called volume-time plots. If the administration of premixed oil/water dispersions. Through the choke the chemical only results in an accelerated creaming/ valve the fluid mixture undergoes a pressure drop. sedimentation, there will be an increased concentra ­ The low-pressure side of the choke valve equals the tion of droplets on the top (or the bottom) of the separator pressure. The pressure drop through the otherwise clear bulk phase. This situation should choke leads to the creation of more interface between not be misinterpreted as a breaking of the emulsion, the oil and water, i.e., water droplets are formed and since gentle shaking will redistribute the droplets dispersed into the oil. At the same time the light end again. of the oil undergoes a phase change from liquid to The weakness of the bottle tests is that a true pro ­ gas. (The gas evolved may form a foam and may cess is not reproduced. This fact has been accounted influence the sedimentation and coalescence of the for in a variety of test rigs simulating true flow con ­ water droplets.) After the cell has been filled, the ditions, process kits, and separation conditions. The amount of the different phases (foam, oil, emulsion/ Statoil R&D Center has recently constructed a special dispersion, and water) is recorded as a function of rig for simulation of high-pressure processes. In this time. Upstream of the choke valve VD1 there are rig (Fig. 5) emulsions can be formed at different pres ­ two other choke valves; one on each line (VD2 and sures before being brought into the separation cham­ VD3). Through these choke valves the same processes ber. The final droplet sizes and size distributions are take place as described for VD1. Through VD2 and determined by the pressure drop over the chokes and VD3 dispersions of oil and water can be made. In this valves. The separation can be performed at elevated way the dispersion which enters the separation cell pressures if so wanted. The rig is described in detail can be a mixture of the two dispersions made through below. VD2 and VD3. 600 Sjoblom et at.

Figure S The high-pressure separation rig.

Chemicals can be injected into any of the flow of the separation process. Video cameras are installed lines; at the high pressure upstream of VD2 and to follow the separation process. VD3, at the medium pressure upstream of VD1 or All parts of the rig are thermostated. Temperature, at the lowest pressure downstream of VD1 just cell pressures, valves, pumps, and video cameras are before the fluid enters the separation cell. computer controlled. Chemicals can also be injected into the bottom of The above-mentioned techniques all have in com ­ the cell where a stirrer can be used to distribute mon that they are of macroscopic scale, often enabling the chemicals. The injection of for instance demulsi ­ diagrams of separated volume as a function of time to fiers takes place through 1/16-inch tubes with very be displayed. small inner diameters. The chemical injection pumps deliver volumes down to 0.03 ml/h. The rig consists of four 600-ml high-pressure sample III. STABILITY OF WATER-IN-CRUDE OIL cylinders. Usually, two are filled with water (brine), EMULSIONS and two are filled with oil. With the aid of four motor-driven high-capacity piston pumps, water and It is well recognized that these emulsions are stabilized oil are pumped through the choke valves. The four by means of an interplay between different heavy com ­ pumps are independent of each other, but in most of ponents, organic and inorganic particles, respectively. the experimental series the total flow has been kept Heavy components cover asphaltenes, resins, etc. In a constant. The pressure drops through the choke valves depressurized anhydrous crude oil the asphaltenes are are back-pressure controlled. The pressure in the normally in a particulate form. The role of the resins separation cell is regulated by a back-pressure con ­ (and lighter polar components) is to stabilize the trolled valve. The maximum pressure in the cell is asphaltene dispersion (suspension) by adsorption 200 bar. The cell is filled with gas, inert or natural mechanisms. Owing to this strong interaction the gas, to the desired pressure before the filling of water asphaltene particles are prevented from concomitant and oil into the separation cell starts. The separation coagulation and precipitation. The stability will also cell (450 ml) is made of sapphire, assuring full visibility put some restriction with regard to particle sizes since Demulsifiers in the Oil Industry 601 the largest particles are supposed to show the highest stable emulsions. Aromatic molecules such as benzene rate of sedimentation. When water is mixed with the will substantially lower the level of tt. With increasing crude oil, the situation will drastically change. The content of aromatic molecules the intcrfacial activity of system will reach an energetically higher level, where the indigenous surfactants will be canceled and hence the energy difference is proportional to the intcrfacial the emulsion stability will vanish. The dilution with area created during the mixing process. This fresh aromatic solvents is in practical use in many places interfacial area will attract components in the system. in the world where heavy crude oils create transport The molecules possessing the highest interfacial activ ­ and emulsion problems. ity will try to cover the fresh W/O interface and hence minimize the energy level of the system. This category A. Coalescence of indigenous components is normally covered by the lighter polar fraction, i.e., the resins. As a consequence Coalescence is defined as the combination of two or a competition situation between resin molecules at the more droplets to form a larger drop. When these W/O interface and on the solid asphaltene particles will droplets approach each other, a thin film of the con ­ occur. Decisive factors determining the final position tinuous phase will therefore be trapped between the of the resins are the hydrophilic/lipophilic balance of droplets, and it is obvious that the properties of this these molecules and the corresponding properties of film will determine the stability of the emulsion (28). the solid surface. One could imagine that a very hydro- The mechanism of coalescence occurs in two stages: phobic particle surface and a very polar W/O interface film thinning and film rupture. In order to have film would extract different types of resins for the different thinning there must be a flow of fluid in the film, and activities. However, as pointed out, highly interfacially a pressure gradient present. It is obvious that the rate active resins will show preference for the W/O interface of film thinning is affected by the properties of the over not only less hydrophobic resin molecules but also colloidal system. Some of the most important para ­ over asphaltenes. As a consequence the solubility con ­ meters (29) are defined as viscosity and density of the ditions for the asphaltenes will drastically change and a two phases present, interfacial tension and its gradi ­ particulate precipitation will take place. With aqueous ent, intcrfacial shear and dilational viscosities and droplets coated by an interfacial resin film as closest elasticities, drop size, concentration and type of sur­ neighbors the asphaltene particles will precipitate and factant present at the interface, and forces acting accumulate at the droplet surface. The resulting inter- between the interfaces. Considerable effort has been facial properties will be much more rigidified and the made to develop models for prediction of the rate of stability of the corresponding emulsions profoundly film thinning and critical film thickness. Reynolds (30) improved. Central mechanisms involved in the stabili ­ made the first mathematical analysis of parallel disks. zation process will hence be both steric and particle He assumed the bounding interfaces to be solid and stabilization. the film to be of uniform thickness. Frank and Mysels The mechanical properties of the protecting intcr­ (31) investigated dimple formation and drainage facial film arc essential for the final stability level of the through the dimple. Later models of film thinning W/O emulsions. Concentrated polymeric interfacial are those of Zapryanov el al (32) and Lin and films may display either clastic or viscous properties Slattery (33, 34). Zapryanov investigated surfactant that make the destabilization process difficult and partitioning at the interface using the parallel-disk time consuming. The aromatic asphaltene molecules model. This model has later been extended to account will normally undergo a stacking into sandwich-like for the adsorption/desorption kinetics of surfactants structures as a consequence of the molecular associa ­ (35). Film rupture is a nonequilibrium process that tion. The presence of other nanosized-particles like may occur as a result of flow instabilities, temperature organic wax particles and inorganic clay particles will fluctuations, electric fields, or Marangoni effects (36). further enhance the stability level. However, these Investigations by de Vries (37) and Lang (38) showed compounds are not further dealt with in the present that there exists a critical film thickness. Above this chapter. thickness the probability of rupture is zero, and below The interfacial conditions are reflected in the level of it the probability of rupture increases with decreasing the interfacial pressure (rr). Sjoblom et al. (27) showed film thickness. Scheludko and Manner (39) investi ­ that there is a correlation between the level of n and gated the rupture of thin liquid films between two the macroscopic emulsion stability. Preferably the droplets in relation to fluctuations at the interface. interfacial pressure should be above 10-14 mN/m for He also developed an expression for the cntical film 602 Sjoblom et al.

thickness with only van der Waals forces acting: finic hydrocarbons < aromatic hydrocarbons < alco ­ d c = [An/32K 2y0f 25 where A is the Hamaker con ­ hols < diols, etc. stant, y0 is the interfacial tension between the contin ­ uous and dispersed phase, and K is the wavenumber A. Low Molecular Weight (LMW) Demulsifiers of the surface fluctuations. Vrij (40) has derived an alternative expression for d c', for larger thicknesses: Basically, the functionality of this category of demul ­ d c = 0.268[.-f 2/?2/y07r/] 0,14 where R is the droplet sifiers is based on two specific mechanisms, i.e., radius, and / is dependent on d. For small thick­ increased interfacial activity and changed wettability nesses: d c = 0.22[AR2/y(/]0'2S According to the first of stabilizing components, respectively. The increased equation d Q -» oo when y -* 0, i.e., the film should interfacial activity results in a suppression of the inter- spontaneously rupture at large d values. However, facial tension. Hence, these molecules tend to replace this is not the case since emulsion droplets become other, already existing molecules at the interface. This highly stable when y -*■ 0. Also, the first equation is a thermodynamical result, but in practice it can be predicts that as R -* 0, d c -» 0, i.e., small emulsion difficult for these surfactant-like molecules to reach the droplets would never rupture. Sonntag and Strenge W/O interface. A common retention mechanism is the (41) showed that d c will not change when the contact adsorption on to solid material, primarily asphaltenes, area is varied. This is due to the fact that the lamella but also inorganic oxides and organic waxes. The formed between two droplets, at nonequilibrium adsorption process can change the wettability of the separations, does not have an idealized planar inter ­ solid particles. In order to complete the adsorption face between them. Sonntag and Strenge (41) also process the LMW adsorbent must complete with natu ­ showed that emulsion films of octane/water droplets rally occurring dispersants like resins. The final equili ­ stabilized by a nonylphenol ethoxylated surfactant brium conditions on the surface of the solid particles plus an oil-soluble surfactant had a d c independent will hence reflect a balance between attraction to the of Ho- surface, interaction with resin-like molecules on the surface, and retention mechanisms in the bulk phase.

B. High Molecular Weight (HMW) Demulsifiers IV. GENERAL THEORY OF DEMULSIFIER ACTION These molecules are actually supposed to penetrate the interfacial film surrounding the water droplets and Commercially available demulsifiers can generally be hereby to alter the rheological properties of the film described as "chemical cocktails." The terminology is material. From the low dosage levels used, i.e., 5-20 introduced in order to describe the fact that one ppm, one can conclude that these molecules are extre ­ expects to find a synergistic effect of one or two (or mely efficient as film modifiers. A critical and decisive more) active components that are dissolved in an active step for the HMW demulsifiers to perform optimally is solvent. The solvent should be so hydrophobic that the the time requirement for the diffusion to the interfacial active components can be readily dissolved in the crude membrane and for the reorientation movement inside oil. From this one can see that here are some general the film until local equilibrium is attained. rules of thumb for the demulsifiers. Below, we sum­ marize some of the most pertinent features. We can C. Solvents briefly classify the demulsifiers according to their mole ­ cular weight, as high molecular weight (HMW) and The action of the solvent can be manifold. However, low molecular weight (LMW) demulsifiers and pure the commonly used aromatics efficiently dissolve the solvents. aromatic particles in a swelling process leaving behind The HMW molecules include different kinds of oligomeric and monomeric asphaltenes. As shown polymers and macromolecules (block copolymers, before, an increasing aromatic content will gradually etc.), together with polyelectrolytes. Typical HMWs decrease and finally eliminate the interfacial activity should be > 5000 g/mol. The LMW demulsifiers are of the indigenous crude oil components. By experi ­ in most cases some types of oil-soluble surfactants with mentally following the interfacial pressure as a func­ co-operativity between the molecules. The solvents in tion of the aromatic concentration one can conclude use can be classified according to the polarity. Simple qualitatively if the level of emulsion stability is high examples on increasing polarity would be pure paraf ­ or low. Demulsifiers in the Oil Industry 603

Most chemical agents used for demulsification are Krawczyk (44) investigated the influence of different preferentially oil-soluble blends consisting of HMW demulsifiers on the stability of water-in-crude oil emul ­ polymers. These blends commonly consist of: (1) floc- sions. He defined a partitioning coefficient, KP = culants (large, slow acting polymers); (2) coalescers ca /cG, where ca refers to the demulsifier concentration (LMW polyethers); (3) wetting agents; and (4) sol ­ in the aqueous phase, and c0 to the concentration in vents/cosolvents. Some chemical structures of demulsi ­ the oil phase. He concluded that demulsifiers with K = fiers used for breaking crude oil emulsions have been I gives the best results. He also concluded that the listed by Jones et al. (42). Much work has been carried interfacial activity and adsorption kinetics of the out in order to identify and understand the mechan ­ demulsifier are important parameters. The interfacial isms behind chemical demulsification. Fiocco (43) con ­ region can be expected to be more dynamic, and con ­ cluded that the intcrfacial viscosity was kept at a low siderable interfacial fluctuations may occur in the level when demulsifiers were present. Later on it was presence of medium-chain alcohols. realized that the interfacial shear viscosity of crude oil The mechanism behind a destabilization with sur­ emulsions does not have to be very low in order to factants is probably an interfacial competition. In ensure accelerated water separation (44). this situation the indigenous crude oil film will be Wasan and coworkers (45, 46) investigated the totally or partially replaced by a surfactant layer coalescence of systems containing petroleum sulfo ­ which cannot stabilize the crude oil emulsion. nates. They concluded that the coalescence rates cor ­ When comparing two different hydrophobic surfac ­ related well with the interfacial shear viscosity, while tants, tetraoxyethylene nonylphenol ether (Triton N- no correlation was observed with the interfacial ten ­ 42) and sodium bis-(2-elhylhexyl)sulfosuccinate sion. Aveyard and coworkers (47, 48) investigated the (AOT), it was found that the ionic surfactant, correlation between surfactant interfacial behavior, AOT, was more efficient than the nonionic analogue. surfactant association, and the destabilization effi ­ Three different hydrophilic, fluorinated surfactants ciency. They observed a clear correlation between were also investigated in Ref. 52. They were all the demulsifier concentration at optimal demulsifica ­ very efficient as destabilizes, probably because of tion efficiency and the critical micellization concentra ­ their high interfacial activity (53, 54). As mentioned tion (CMC) of the demulsifier in the crude oil system earlier, Wasan and coworkers (49, 50) have analyzed as long as simple surfactants were used. This means in detail the processes taking place at the O/W inter ­ that the monomer activity of the surfactants is crucial face during destabilization. The results for the hydro- the for destabilization of the emulsion system. Wasan phobic surfactants are in direct agreement with their and coworkers (49, 50) investigated in detail the pro ­ conclusions. Also, in the case of common solvents cesses taking place at the O/W interface during a where we found medium-chain alcohols to be effi ­ destabilization process with a LMW amphiphilic cient as destabilizers, the results also correspond compound. From studies of different additives they with their conclusions. The medium-chain alcohols concluded that oil-soluble destabilizers should be are soluble in all three pseudophases and will there ­ able to partition into the aqueous droplets in order fore partition between these. The hydrophobic sur­ to act as destabilizers. The concentration of the factant AOT is soluble in water up to a few per cent, demulsifier inside the droplets should be high enough and will therefore also be present in the aqueous to ensure a diffusion flux to the O/W interface. In phase, whereas Triton N-42 is completely water inso ­ order to be efficient as destabilizers the additives luble. This will most likely contribute to the differ ­ must show a high rate of adsorption to the interface. ences between the surfactants. Wasan and coworkers also emphasized the impor ­ Aveyard and coworkers (47, 48) have stressed the tance of sufficiently high interfacial activity of the importance of monomer activity when simple surfac ­ demulsifier to suppress the interfacial tension gradi ­ tants are used as demulsifiers. For a commercial ent. In this way the film drainage will be accelerated demulsifier the interfacial tension between oil/water and droplet coalescence will be promoted. Little (51) seems to pass through a minimum for NaCl concen ­ suggested that the sequence of steps leading to demul ­ trations between zero and 1 M. According to Menon sification of peteroleum emulsions involves the displa ­ and Wasan (55), AOT has been found to have a cmc cement of asphaltic material from the interface by the at approximately 300 ppm in a water/oil system with demulsifier followed by the formation of demulsifier asphaltenes present. This means that in our destabili ­ micelles which solubilize and/or stabilize the asphal ­ zation tests, where the concentration of AOT is up to tene compounds in the oil. 100 ppm. the results correspond with the conclusions 604 Sjoblom et at. from Aveyard (47, 48). Fluorinated surfactants have emulsification) and their performance in Langmuir been investigated for systems containing both distilled and Langmuir- Blodgett films. We have also per ­ water and synthetic formation water (52). The results formed an AFM study on the demulsifiers in order showed that the resolution of water was faster when to visualize the interactions taking place between indi ­ synthetic formation water was used as the dispersed genous crude oil surfactants and the LMW/HMW phase. The explanation of this might be in accordance demulsifiers. with Aveyard ’s conclusions (47, 48). In the case of Triton N-42 the influence of salt is not believed to A. Crude Oil Matrix be significant since this is a nonionic surfactant, and its phase behavior is not so sensitive to the addition The crudes span geographically over large areas: North of salt. Sea, European continent, Africa, Asia, etc. This is a The mechanism behind destabilization with necessity since if the crude oils in the test matrix are macromolecules is very dependent on the size of interrelated one cannot universalize the results. Table 1 the molecule. Polymers of lower molecular mass lists the crude oils and their origin. To start with we can show a strong affinity to the oil/water interface, determined the inversion point (or alternatively, the adsorb irreversibly and destabilize in this way. maximum content of water that can be introduced Another route of destabilization is flocculation. into the oil without a phase separation). We have Flocculation is an aggregation process in which dro ­ chosen to study emulsions that are 10% below the plets form three-dimensional clusters, each droplet inversion point. Exceptions in this respect are the retaining its individual identity. In order to model two European crudes with 5% water stabilized. The the importance of flocculation in the destabilization crude oils were characterized by means of density, sur­ of model systems, one can investigate a-alumina face tension, and viscosity measurements. The results dispersions (52). V. are summarized in Table 2. All experiments involving emulsions were carried out at 50°C. The reason for working at elevated temperature is to melt the wax in V. EXPERIMENTAL DEMULSIFICATION the oils and thereby prevent the influence of the wax on emulsion stability. The elevated temperature is also In this section we compile information about demul ­ more closely related to the real working temperature sifiers active in W/O emulsions (or added prior to the used in the processes in the field.

Table I Crude Oils and Their Origin. Types 1-3 Refer to Increasing Emulsion Stability; Model Emulsions are Based on Extracted Asphaltenes

Ratio of Crude Model oil W/O Inversion based based Emulsion Crude used point emulsion emulsion

i Venezuela 40/60 50/50 Type 3 Type 2 (3) 2 Nigeria 50/50 60/40 Type 2 Type 2 3 Nigeria 30/70 40/60 Type 2 Type 2 4 North Sea 50/50 60/40 Type 2 Type 2 5 North Sea 40/60 50/50 Type 2 (3) Type 3 6 European continent 40/60 50/50 Type 2 Type 2 7 North Sea 5/95 5/95 Type 1 3 8 North Sea 5/95 5/95 Type 1 3 9 North Sea 40/60 50/50 Type 2 (3) Type 2 (3) 10 North Sea 30/70 30/70 Type 1 3 11 North Sea 20/80 30/70 Type 1 3 12 North Sea 20/80 30/70 Type 2 3 13 North Sea 20/80 30/70 Type 2(1) Type 2

‘'Model oils based on asphaltene that did not stabilize any water. Type 1 - least stable; Type 3 - most stable. Demulsifiers in the Oil Industry 605

Table 2 Physical Characteristics of Crude Oils

Surface Interfacial tension. Density tension with Viscosity Viscosity Crude yao(mN/m) (g/ml), 3.5% NaCl (cP), (cP), Asphaltene nr. at 25°C <5 25°C (mN/m) 25°C 50°C content (%)

i 27.0 0.88 23.95 35.7 13.8 7.82 2 26.5 0.84 23.06 8.95 4.34 2.58 3 27.8 0.87 16.20 70.8 20.05 5.16 4 27.6 0.87 27.79 14.3 2.12 3.63 5 25.9 0.86 27.66 15.6 6.97 11.69 6 26.9 0.83 28.00 12.0 5.96 1.86 7 24.8 0.84 19.75 2.93 1.93 Negligible 8 24.7 0.84 21.31 3.05 2.35 Negligible 9 29.9 0.92 25.08 Above 50.95 2.59 critical viscosity 10 24.5 0.80 23.5 3.95 2.37 Negligible 11 25.4 0.82 23.86 5.38 2.69 0.5 12 26.8 0.84 31.80 12.7 5.43 1.99 13 24.8 0.81 25.45 3.27 2.09 2.19

Table 3 Interfacial Tensions of Low Molecular Weight Chemical Additives

Interfacial tension Interfacial tension Interfacial tension (25 ppm), (50 ppm), (100 ppm). Nr. Chemical Kwo(mN/m) XwofmN/m) ywo(mN/m)

1 Fatty alcohol ethoxylate 27.8 25.4 23.6 2 Vinylidene alcohol 40.2 40.0 39.4 3 Sulfosuccmate 12.4 9.3 5.3 4 Linear 5.3 2.6 1.3 monoisopropylamine alkylbenzene sulfonate 5 Emulsifier for hydrophobic 14.9 11.7 7.5 system 6 Coconut fatty acid 27.5 23.7 19.8 diethanolamide 7 Plasticizer 36.5 36.2 36.5 8 Rape oil fatty acid 26.5 23.2 19.3 diethanol 9 Anionic/nonionic blend 4.5 2.6 1.6 10 Emulsifier for 39.3 39.4 38.6 polymerization processes 11 Fatty alcohol 39.2 38.5 38.4 12 Long-chain fatty alcohol 38.3 35.4 31.5 13 long-chain a-olefin 39.4 38.8 38.6

Oil phase = 30/70 toluene/decane. Aqueous phase = 3.5% NaCt. 606 Sjoblom et al.

Table 4 Interfacial Tensions of High Molecular Weight Chemical Additives

Interfacial tension Interfacial tension Interfacial tension (25 ppm), (50 ppm), (100 ppm), Nr. Chemical Kwo(mN/m) ywo(mN/m) y Wo(mN/m)

A Complex block polymer 3.4 0.7 0.6 B Complex block polymer 10.5 7.7 7.3 C Phenolic resin alkoxylate 1.1 0.6 0.4 D Phenolic resin alkoxylate 0.2 0.2 0.2 E Phenolic resin alkoxylate 5.0 3.6 2.4 F Complex block polymer 17.4 16.2 15.1 G Complex block polymer 10.4 9.2 7.7 H Mildly cationic; high 15.6 15.5 15 8 molecular weight block polymer I Polyester 9.2 8.0 6.3 J Polyol dispoxide reaction 16.2 14.6 12.5 product

Oil phase = 30/70 toluene/decane. Brine = 3.5% NaCl.

B. Chemical Additives The effect of adding a demulsifier is presented in Tables 5 and 6. Table 5 contains the test results for The chemicals added represent two of the categories LMW additives, and Table 6 gives the test results when mentioned above, i.e., the LMW and BMW demulsi ­ the BMW additives have been used. Both tables refer fiers. The definition of these additives (see Tables 3 and to the percentage of water separated after 30 min. 4) is very general. However, from a functional point of From the tables, some interesting features are revealed. view their interfacial activity under operational condi ­ Naturally, the addition of demulsifiers affect the rate tions is of interest. The interfacial tension between an of separation. If we compare the separation rate of organic phase (30/70 toluene/decane) and an aqueous water from crude oil emulsions with the addition of phase (with 3.5% NaCl) was determined upon addi ­ chemicals compared with blank samples, it is clear tion of 25, 50, and 100 ppm of additives; y0 without that the addition can enhance separation. This is not additives is 36.3 mN/m. true for all chemicals added. Some demulsifiers have no apparent effect on separation and some demulsifiers C. Destabilization even make the emulsions more stable. For the last case the separation of water after addition of demulsifier is Experimental conditions are found in the paper by lower than without demulsifier. The average separation Djuve et al. (56). without addition of any chemicals is 21.5% of water The stability of the different crude oil-based emul ­ after 30 min and particularly for the LMW chemicals sions varies a lot. The water cuts range from 5 to 60%. a reduced separation is observed. The BMW chemicals These values can be compared with the stability for the on the other hand seem to enhance separation. model emulsions containing only dissolved asphaltene The difference found in the performance between residues (see Table 1). Thislarge difference in the water the BMW and LMW chemical activites should be cut dispersed cannot be explained by the differences in traced back to the interfacial film and the added spe­ asphaltene content alone (Table 2). The large differ ­ cies. It should be noted that the interfacial tension ence in stabilization ability reflects the wide and differ ­ measurements for the demulsifiers refer to a pure W/ ent distribution in size, state, structure, polarity, and O interface, while the demulsifier action actually refers mass that exists between the asphaltenes found in each to a W/O interface covered by indigenous components, oil. It is generally believed that it is mostly the state of like asphaltenes, resins, waxes, etc. With a weak rever ­ the asphaltene and not the amount that controls the sible adsorption on to the film material, the destabiliz ­ stability in an W/O emulsion, i.e., whether the asphal ­ ing effect of the LMW additives (e.g., oil-soluble tenes are in a particulate form or not. surfactants) will be rather limited, if penetration into Demulsifiers in the Oil Industry 607

Table 5 Effect of Low Molecular Weight Chemicals on Crude Oil Based W/O Emulsions 8

Emulsion sample i 2 3 4 5 6 7 8 9 10 11 12 13 ESeP/"oil

1 0 0 0 0 0 0 50 20 0 40 0 0 20 10 2 0 0 0 0 0 0 40 0 0 73 0 0 0 8.6 3 0 0 0 100 0 20 100 60 0 56 0 0 0 25.8 4 0 0 0 16 0 0 100 40 0 65 0 0 0 17 5 0 0 0 4 0 0 80 60 0 70 0 0 0 16.5 6 0 0 0 0 0 0 60 40 0 83 0 0 0 14 7 0 0 0 0 0 0 40 0 0 93 0 0 0 10.3 8 0 0 0 0 0 0 60 10 0 100 0 0 0 13 9 0 8 0 0 0 5 80 60 0 100 0 0 0 19.5 10 0 0 0 0 0 0 0 80 0 100 0 0 0 13.8 11 0 0 0 0 0 0 0 80 0 67 0 0 0 11.3 12 0 0 0 0 0 0 80 75 0 67 0 0 0 17.0 13 0 0 0 0 0 0 80 55 0 83 0 0 0 16.5 Blank 0 0 0 0 0 0 80 100 0 100 0 0 0 21.5

‘Separation is given in percentage of water separated after 30 min.

the film material is obstructed. From Langmuir studies D. Model Emulsions Based on Asphaltene of asphaltene films it is known that polymeric demul ­ sifiers can penetrate the films and strongly modify the Generally, for water-in-crude oil emulsions the indi ­ film properties (57). genous component thought to have the largest effect Each demulsifier also behaves differently, depending on stability is the asphaltenes. Therefore, model sys­ on which oil the emulsion is based on (Tables 5 and 6). tems based on asphaltenes were prepared for selected That was expected since most of the crude oils are not oils. From Table 1 one can observe that not all pre ­ interrelated. cipitated asphaltene fractions could stabilize a model Based on the results from the bottle tests outlined emulsion. However, many model emulsions had a simi­ above a selection of crudes for the next experiments lar stability as the original crude oil-based emulsion, was made, omitting the crude oils that caused either indicating that the fraction extracted from the oil plays spontaneous separation or a complete separation a central role in the stabilization. within 30 min.

Table 6 Effect of High Molecular Weight Chemicals on Crude Oil-based W/O Emulsions 8

Emulsion sample 1 2 3 4 5 6 7 8 9 10 11 12 13 Ese P/"oii

A 0 64 0 100 0 15 60 84 0 67 98 98 58 49.5 B 0 20 0 0 0 0 80 100 0 100 85 5 50 33 C 0 80 0 38 0 47.5 80 80 0 56.7 93 100 95 51.6 D 0 63 0 100 0 45 80 82 0 80 87 95 40 51.7 E 0 63.8 0 18 0 65 100 100 0 100 100 100 85 49.7 F 0 10 0 0 0 62.5 100 100 0 100 100 100 0 44.0 G 0 67 0 84 0 54 60 94 0 93 85 100 68 54.2 H 0 10 0 100 0 70 80 60 0 70 100 0 93 44.8 I 0 61 0 0 0 18 60 70 0 93 85 0 48 33 5 J 0 0 0 0 0 54 40 60 0 88 100 0 48 30 Blank 0 0 0 0 0 0 80 100 0 100 0 0 0 21.5

'Separation is given in percentage of water separated after 30 mm. 608 Sjoblom et al.

The representability of such a “model emulsion ” of units. In the case of G, ywzo is about 10 times higher vis-a-vis the original emulsions has been debated. and reaches a value of ~ 10 mN nT 1 Obviously, pure This brings forward the question of stabilization displacement processes, where demulsifiers A and C mechanisms and the state of asphaltenes. Basically (owing to a lower interfacial tension toward water) we try to mimick interfacial conditions from true can create a new interface and in this way destabilize crude oil-based emulsions to model emulsions. In an emulsion, play a role in the destabilization process. order to do so it is essential that the model oil used Also, G has a yw/o value lower than that of pure (heptol) can promote particle formation in the asphal ­ asphaltenes at the interface. In the latter case yw/„ is tenes. In this way there should be a similarity in around 20-25 mN/m. asphaltene-based nanoparticles located at the W/O interface in both types of emulsions under study. E, Demulsifiers Used as Inhibitors Since the asphaltene particles will hinder an efficient coalescence of the aqueous droplets, one can expect We have also added the destabilizing agents directly approximately the same level of stability against co ­ to the oil before the emulsification. The result as alescence and similar actions of the demulsifiers. revealed from Tables 8 and 9 is very encouraging. Addition of demulsifiers to the asphaltene-stabilized In most cases there is a substantial enhancement in model emulsions accelerated in some cases the resolu ­ the efficiency of the action of the added chemicals, tion of water. In particular three demulsifiers seemed also for the low molecular species. The reason for to be most efficient, i.e., A, C, and G. It is obvious this can be two-fold, i.e., either an interfacial compe ­ when comparing Tables 6 and 7 that the same demul ­ tition or a strong bulk interacation between the sifiers are effective in both model emulsion systems and demulsifier added and the stabilizing crude oil species. true crude oil-based W/O emulsions. This means that The interfacial competition can be traced back to the we can trace back the destabilization effect to an inter ­ yw/0 values. Although the concentrations of the action between the demulsifying agent and the asphal ­ demulsifier added are small (% 50 ppm) and there is tene fraction in the crude, and that this interaction is most likely not enough molecules to create a stable the most significant one. Other possibilities for inter ­ emulsion with all the water molecules dispersed, some actions leading to destabilization would be demulsifier/ molecules of A, B, and G present at the interface can wax particles, demulsifier/resins, demulsifier/solid inor ­ cancel the stabilizing properties of asphaltene parti ­ ganic particles, etc. An investigation of asphaltene- cles at the W/O interface. A strong bulk interaction stabilized W/O emulsions has obviously shed light on between the demulsifiers and the asphaltenes must fundamental destabilization mechanisms in the crude change the state of the asphaltene particles in order oil-based emulsions. to cancel their stabilizing effects. Obviously, one Table 4 shows that the demulsifiers A and C exhibit should anticipate the demulsifying agents dissolving a low interfacial tension, i.e., of the order of a couple the asphaltene particles to substantially smaller units

Table 7 Effect of High Molecular Weight Chemicals on Asphaltene-based Model Emulsions 8

Emulsion chemical 1 2 3 4 5 6 9 13 £ sep/zt oi |

A 0 92.5 0 80 0 92.5 0 0 33.1 B 0 0 3 43 0 43 0 0 11 C 58 88 77 46 0 75 0 0 43 D 0 0 0 0 0 62.5 0 0 7.8 E 15 87.5 6.6 5 0 40 0 0 19.3 F 0 0 0 0 0 27.5 0 0 3.4 G 90 90 3.3 12.5 0 90 0 75 45.1 H 0 0 3 0 0 10 0 0 2 1 0 0 0 0 0 80 0 0 10 J 0 0 3 0 0 18 0 0 3 Blank 0 0 0 0 0 0 0 0 0

8 Separation is given in percentage water separated after 30 min. Demulsifiers in the Oil Industry 609

Table 8 Effect of High Molecular Weight Chemicals as Inhibitors on Crude Oil Emulsions 3'11

Emulsion chemical 1 2 3 4 5 6 9 13 22seP/n oii

A 75 92 7 100 55 80 100 35 68.0 B 40 44 77 80 85 78 85 25 64.3 C 0 94 53 78 65 93 63 100 68.3 D 35 94 0 100 60 83 45 65 60.3 E 43 76 67 78 0 60 88 85 62.9 F 50 84 0 80 63 73 53 5 51.0 G 0 98 73 100 75 83 75 95 74.9 H 65 98 70 100 78 93 83 100 85.9 I 100 92 0 76 60 88 80 100 74.5 J 78 92 47 82 83 80 70 100 79.0 Blank 0 0 0 0 0 0 0 0 0

‘Chemicals added before emulsification. b Separation in percentage water separated after 30 min.

not possessing stabilizing effects. There are some indi ­ Another essential deviation in sample representability cations that this might be the case (58). is due to the time delay in sampling of the crude oil samples to be compared. It is well known that the VI. DEMULSIFICATION UNDER crude oil characteristic from a field consisting of LABORATORY AND FIELD CONDITIONS several wells (up to 30-40 wells) will change over time. The best way to overcome the classical difficul ­ Normally, the bottle-shake tests with depressurized ties with representative samples is to work with pres ­ crude oils are upscaled to real separation conditions surized samples. The separation rig presented in Fig. topside. However, it has been constantly pointed out 5 has the great advantage of permitting this and that the samples in use at the laboratory are not preventing the crude oils under study to contact representative of the samples from the same field. air. In addition to this the mixing conditions (the The main reason for this is that the laboratory sam ­ magnitude of AP over the chokes) can be adjusted ples have undergone oxidation upon storage. to real process conditions.

Table 9 Effect of Low Molecular Weight Chemicals as Inhibitors on Crude Oil Emulsions 3 (Chemicals added before emulsification)

Emulsion chemical 1 2 3 4 5 6 9 13 Z sep/«oii

1 1 46 0 44 0 25 0 0 13 2 0 72 0 0 0 0 0 0 8 3 0 0 0 100 73 73 0 0 27 4 0 52 0 70 40 80 0 0 27 5 0 80 0 90 53 70 23 45 40 6 65 7 0 0 4 0 0 10 22 7 0 80 0 0 0 0 0 0 9 8 0 70 0 0 83 0 3 5 21 9 0 72 0 0 0 48 58 20 22 10 0 66 0 0 0 0 0 10 8 11 0 40 0 0 0 38 0 8 10 12 0 0 0 0 0 0 0 0 0 13 0 4 0 0 0 0 0 0 5 Blank 0 0 0 0 0 0 0 0 0

‘Separation in percent water separated after 30 min. 610 Sjoblom et al.

A. Tests of Demulsifiers - Comparison with Field Tests

The laboratory tests were conducted to qualify the separation rig by performing tests as tsimilar as possi ­ m 60 ble to the field tests done previously. An important © 0 ppm difference between the tests performed offshore and in the laboratory is the type of separator. The field ♦ 50 ppm tests were performed in a horizontal continuous grav ­ -* 100 ppm — ity separator whereas the separation in the laboratory

° 20 rig took place in a vertical batch separator. The oils and brine used in the laboratory were sampled offshore and kept under pressure until the tests were performed. In the laboratory tests only one module of the 01 23456789 10 separation rig was used. Only one oil was tested at a Separation time (min) time, and there was a pressure drop through only one of the choke valves (VDl in Figure 5). Oil and water Figure 6 Examples of separation as function of time. Oil l with 20% water cut; Demulsifier B. were mixed upstream of VDl. There was no pressure drop through VDl. The demulsifier was mixed into the flow line just downstream of VDl. The pressure drop in the system was through VDl just ahead of the separation cell. The oils tested were at their bubble

100 r 90

« 70 100 ppm ® 50 -O-OiM, Demulsifier B -O— Oil 2, Demulsifier B o 30 #- 011 1, Demulsifier A § 20 Oil 2, Demulsifier A

Water cut (%)

Figure 7 Amount of water separated after 1.5 min for the two oils 1 and 2 at various water cuts and for the two demulsifiers A and B at various concentrations (1.5 min separation time corresponds to 2 min separation time in Fig. 6. In Fig. 6 the filling time of 30 s is included in the separation time). Demulsifiers in the Oil Industry 611 points at 11 bar and 60°C. The experiments were per ­ 5. Demulsifier B increased the separation effi ­ formed at 60°C and with a pressure drop through VD1 ciency with increasing concentrations up to from 11 to 7 bar. The separation took place under 7 100 ppm. bar pressure. They were typical North Sea crude oils 6. Foam was never any problem (stable for a max ­ with density and viscosity values for stabilized oils at imum of 30 s). 60°C at ~ 0.8 g/ml and ~ 3.5 mPa. There were small All these results confirmed the offshore field-test differences in the characteristics of the oils. The more results. In addition, one could observe visually in the dense oil was also the more viscous oil. The composi ­ laboratory tests how the demulsifiers affected the sys­ tions of the two tested demulsifiers were totally differ ­ tem. Without demulsifier in the system an emulsion ent from each other. Three concentrations of the layer always formed between the oil and the water demulsifiers were tested: 5, 50, and 100 ppm. In addi ­ phase. When the demulsifier was added, no such sepa­ tion, tests without demulsifier were performed. The rate emulsion layer was observed (except for 5 ppm water cut values were 5, 20, and 35 vol. %. demulsifier in the system with 5% water cut in the Some of the results are shown in Figs 6 and 7 The more viscous oil). main results in the laboratory tests were: Results from the laboratory separation rig have also been verified with results from other field tests. 1 Oil 1 had better separation characteristics than As a conclusion to this section one can say that a Oil 2. (Oil 1 was the lighter of the two oils.) laboratory test kit has been constructed which can be 2. Demulsifier A performed better than used to test oils and chemicals in pressurized systems. Demulsifier B at concentrations of 5 and 50 The results are consistent with results achieved under ppm (for water cuts of 20 and 35%). offshore field conditions. The results obtained in the 3. Demulsifier B performed better than laboratory are based on correct sampling and handling Demulsifier A at 100 ppm (for water cuts of of the fluids. The oil samples are kept under pressure 20 and 35%). and are never exposed to air during storing. Further, 4. No increase in separation efficiency was the results show that we can dose with chemicals down observed when the concentration of to concentrations as low as 5 ppm. The advantages of Demulsifier A was increased from 50 to 100 laboratory studies are smaller volumes, cheaper tests, ppm. more parameter variations can be performed within

100% asph. 80% asph.

60% asph. - - - 50% asph.

100% resin

Figure 8 fl-A is isotherms of asphaltene/resin mixtures spread from pure toluene on pure water (bulk concentration — 4 mg/ ml). 612 Sjoblom et at. short time limits, and access to more advanced char­ resins start to dominate the film properties when the acterization systems (e.g., drop size measurements) is amount of this lighter fraction exceeds 40 wt% (Fig. available. 8). The more hydrophilic resin fraction starts to domi ­ The separation rig has also been used to show the nate the film properties owing to the higher affinity influence of an internal separator pressure up to 180 towards the surface. bar on the separation characteristics and efficiency. The influence of chemical additives on asphaltene films on the water surface and at the oil/water interface B. Langmuir Films have also been studied by means of the Langmuir tech­ nique. This was done in order to view the interaction In order to obtain a better understanding of the between demulsifiers added and asphaltenes, and to mechanisms behind the effect of asphaltenes and resins show the importance of this on emulsion stability. on emulsion stability, we chose to investigate the film The film properties of pure demulsifiers of high properties of these components. Such studies provide molecular weight are shown by the isotherms in Fig. information on the rigidity and stability of films con ­ 9. The shape of some of these isotherms, especially that sisting of indigenous surface-active material. The rigid ­ of Demulsifier G and to some extent those of H and I, ity of the interfacial film is important for the stability resembles pure resin films. The others, especially of emulsions, in as much as a rigid film on the emulsion Compound A, give more rigid films, characteristic of droplets prevents coalescence, while a highly compres ­ the pure asphaltene film. sible film is more easily ruptured, leaving the droplets Compressible resin films will not alone stabilize a free to coalesce. crude oil emulsion. Related to this, demulsifiers, By means of the Langmuir technique, asphaltenes are which form films of low rigidity and high compressi ­ found to build up close-packed rigid films, which give bility, should be the most efficient. When used as rise to quite high surface pressures. Resin films, on the demulsifiers, the efficiency depends on the ability of other hand, are considerably more compressible (Fig. 8). the chemicals to interact with and modify the film This may explain the experimental observations show­ built up by asphaltene particles. ing that asphaltenes are able to stabilize crude oil-based Addition of demulsifiers of high molecular weight to emulsions, while resins alone fail to do so. Singh and the asphaltene film gave the isotherms in Fig. 10. The Pandey (59) also concluded that a high interfacial pres ­ influence of the chemicals G, H, and I is most pro ­ sure correlated with high W/O emulsion stability. On nounced with respect to an increased compressibility, adding asphaltenes and resins together to a mixed film, together with a reduced rigidity. The effect of this kind the properties gradually change from a rigid to a com ­ of manipulation of the asphaltene film is similar to the pressible structure as the resin content is increased. The effects observed when resins are mixed together with

---- G V \ X - — • — c

Figure 9 n-A isotherms of high molecular weight demulsifiers on pure water. 613

Figure 10 fl-/t isotherms of mixed monolayers of asphaltenes and varying concentrations of different demulsifiers on pure water. asphaltenes (Fig. 8). However, the concentration within the film, but also on the ability of demulsifiers needed to achieve the same effects is considerably to reach the W/O interface in an emulsion (diffusion lower when demulsifiers are used instead of resins. through the fluid). This is a critical step regarding the Demulsifier A has a quite small influence on a film of effective concentration of demulsifiers at the interface. asphaltenes. A comparison with Fig. 9 shows that che­ These aspects make it difficult to undertake a direct mical A is the component with the most rigid and comparison between the influence of demulsifier on asphaltene-like film behavior of all the tested HMW Langmuir surface films, where all demulsifier mole ­ demulsifiers. cules are implanted in the film, and on real emulsions. From the film studies outlined above one can con ­ In order to represent more realistic emulsion condi ­ clude that the best candidates for emulsion breaking tions, Langmuir interfacial films adsorbed at the O/W should be G, H, and I. However, the efficiency depends interface were analyzed. The isotherms depicted in Fig. not only on the direct influence of chemical additives 11 illustrate some of the film properties of naturally 614 Sjoblom et al.

Pm(arf)

—0.01% asph.(VB5) — +5 ppm G — - +20 ppm G + 50 ppm G

Area (cm2)

Figure 11 Interfacial pressure isotherms of films formed between water and oil containing different ratios of asphaltenes and resins or different amounts of added chemicals.

occurring crude oil components adsorbed at the W/O ity of these chemicals to prevent formation of rela ­ interface. tively rigid asphaltene films at the O/W interface. For The oil phase containing only 0.01 wt % asphaltene concentrations higher than 20 ppm of chemical A gives rise to a less rigid interfacial film than observed at there is no pressure increase during the compression. the water surface (Fig. 8). This is most likely due to the Hence, the film that is formed at the interface is possibility of the hydrocarbon tails of the asphaltenes highly compressible. So instead of increasing the pres ­ to orient toward the highly aliphatic oil phase, making sure. the components will build up a multilayer, or the interactions between the film material and, hence, the film may dissolve under the influence of compres ­ the pressure increase during film compression, less sion An increased inhibitor concentration reduces the extensive. In general, interactions between the bulk interfacial pressure, but has no influence on the film phase and interfacial components are different from behavior. The reduced pressure is probably as a result the water/air case. of a more complete cover of inhibitor at the interface. Addition of resins to 0.01 wt % asphaltene solutions That is, fewer components from the asphaltene frac ­ further reduces the adsorption of interfacially active tion are adsorbed together with the chemical additive components on to the O/W interface, even if the when the inhibitor concentration becomes high total amount of naturally occurring surfactants is con ­ enough. siderably higher in these oil phases. The reduction is The results obtained upon addition of seen as reduced pressure at constant interfacial area. Demulsifier G are similar to those of A. However, These changes may be attributed to the ability of resins G clearly increases the compressibility of the film to disperse asphaltenes in the bulk oil phase, and thus even at low concentration. The difference between prevent this heavy fraction from building up a stabiliz ­ 20 and 50 ppm is quite small, so it is reasonable to ing film between oil and water. believe that maximum efficiency, resulting from the Introducing chemical additives together with competing adsorption in a system like this, is asphaltenes into the oil phase may highlight the abil ­ already reached at a concentration of 20 ppm in Demulsifiers in the Oil Industry 615

100% asphaltenes 80% asph./20% res. 20.0

10.0 20.0

Figure 12 AFM images (20 x 20 /im) of monolayers with increasing resin-to-asphaltene (R/A) ratio; LB film deposited onto mica substrates. The fractions are extracted from a crude from a production field in France (crude F). the oil phase. With 20 ppm or more of G present, investigated film properties, AFM was used to examine only small amounts of asphaltene will reach the the topography of these deposited layers. interface. The images shown in Figs 12 and 13 show the struc­ The results obtained from the Langmuir interfacial tural change in the monolayer at a surface pressure of film studies are important in explaining why certain 10 mN/m, when the composition of the film was gra ­ chemicals are more effective as inhibitors than as dually changed from pure asphaltenes to pure resins. demulsifiers. Obviously, the inhibitor/asphaltene Images of pure asphaltene show a closed-packed struc­ interaction is so strong in the bulk oil phase that ture of nanosized particles. Addition of resins modifies the interfacial structures being gradually built up this rigid structure toward an open structure with will no longer possess properties required to stabilize regions completely uncovered by film material. Pure W/O emulsions. resins build up a layer with an open fractal network. The individual film units increase in size upon addi ­ C. Langmuir-Blodgett Films Studied by tion of resins. This indicates interactions between Means of AFM asphaltenes and resins, providing aggregates of larger dimensions than observed for the pure fractions. Small Monolayers of asphaltenes and resins on the water and moderate amounts of resins give rise to a more surface were transferred at a surface pressure of 10 polydisperse distribution of the film material, while a mN/m on to mica substrates by using the Langmuir- further increase in the resin content (i.e., 60 wt % Blodgett technique. In order to visualize the earlier resins) reduces the polydispersity, i.e., the monolayer 616 Sjoblom et at.

Figure 13 AFM images (20 x 20 /im) of monolayers of pure components from a crude from a production field in the North Sea; LB film deposited on to mica substrates.

pure asphaltenes

asphaltenes + 100 ppm A

Figure 14 AFM images (20 x 20 /rm) of monolayers consisting of asphaltenes from crude F and 100 ppm high molecular weight demulsifiers/inhibitors; LB film deposited on to mica substrates. Demulsifiers in the Oil Industry 617

becomes more uniform in component size when one of quite similar to the effect on structural changes brought the pure fractions dominates the film properties. about by the resins. These results indicate that the The AFM images visualize why asphaltenes alone observed structural changes in the film are qualitatively can stabilize emulsions while films dominated by the essential in order to reduce the emulsion stability. resin fraction do not. Hence, when the amount of It is important to keep in mind that the AFM resins present in the film is so large that the structure images visualize conditions in Langmuir films at the in the film changes toward a more open fractal net ­ aqueous surface. Once again all interactions between work, the efficiency of film components as emulsifier an oil phase and interfacial components are lacking. In is reduced. a real W/O emulsion there are no guaranteees that all The AFM images of asphaltene films containing 100 these components will be present at the W/O interface ppm of different HMW demulsifiers/inhibitors (Fig. 14) due to solubility in the oil phase. Hence, results from show that the effect of these components on the film is an AFM study of LB films should not be too far-reach-

Table 10 Composition of the Samples

Inhibitor concentration Toluene concentration Sample name (ppm) (%)

Crude 0 0 Crude + Toluene 0 1.25 Inhibitor G (300 ppm) 300 0.75 Inhibitor A (300 ppm) 300 0.75

Crude

Crude + Toluene

■300 ppm inhibitor A 300 ppm inhibitor G

1600 181 Wavelength (nm)

Figure 15 Near infrared spectra of the Grane crude oil with no additives, and with the addition of 500 ppm toluene, 300 ppm inhibitor G, and 300 ppm inhibitor A. 618 Sjoblom el al. ing when considering real conditions in W/O emul ­ H. Auflem. Statoil A/S is acknowledged for permission sions. However, the effect of demulsifiers on the film to publish results from the high-pressure separation material remains indisputable. rig-

D. Near-infrared (NIR) Characterization of the Effect of Emulsion Inhibitors REFERENCES

Aggregates of colloidal size scatter near-infrared radia ­ 1 P Beefier, ed. Encyclopedia of Emulsion Technology. tion (X = 700-2500 nm) in accordance with Rayleigh New York: Marcel Dekker, 1983. theory (60). This is observed in the spectrum as a rise 2. K Larsson, S Friberg. Food Emulsions. New York: of the spectral baseline. The extinction of radiation Marcel Dekker, 1990. 3 J Sjoblom, ed. Emulsions - A Fundamental and increases with increasing radius of the scattering par ­ Practical Approach. NATO ASI Series, Vol 363. ticles, and thus the spectrum yields information about Dordrecht: Kluwer, 1991. the size of the aggregates. The effect of adding two 4. J Sjoblom, ed. Emulsions and Emulsion Stability. New different emulsion inhibitors to a crude oil was deter ­ York: Marcel Dekker, 1996. mined by means of near-infrared spectroscopy. In pre ­ 5. OC Mullins, E Sheu, eds. Structures and Dynamics of vious work (57) by our group it was stated that these Asphaltenes. New York: Plenum Press, 1998. inhibitors have a resin-like influence on the aggrega ­ 6. J Sjoblom, 0 Szether, 0 Midttun, M-H Ese, O Urdahl, tion of asphaltenes, i.e., a solvating effect. Near-infra- H Fordedal. In: OC Mullins, EY Sheu, eds. Structures red spectroscopy should thus be able to detect the and Dynamics of Asphaltenes. New York: Plenum changes in the aggregation state by direct measure ­ Press. 1998, p 337. ments on the crude oil. The near-infrared sampling 7. JD McLean, PM Spiecker, AP Sullivan, PK Kilpatrick. In: OC Mullins, EY Sheu, eds. Structures and of the crude oil was performed on a NirSystems 6500 Dynamics of Asphaltenes. New York: Plenum Press, spectrophotometer, equipped with a fiber-optic sam ­ 1998. p 377. pling probe for transflectance sampling. The wave ­ 8. MC Petty, WA Barlow. In: G Roberts, ed. Langmuir- length region was set to 1100-2250 nm. The total Blodgett Films. New York: Plenum Press, 1990. pathlength was 2.5 mm. The total number of scans 9. GL Gaines, Jr. Insoluble Monolayers at Liquid-Gas per spectra was set to 32 and the sampling was carried Interfaces. New York: John Wiley, 1966. out at 25°C. The compositions of the four samples 10. KS Birdi. Lipid and Biopolymer Monolayers at Liquid investigated are listed in Table 10. Interfaces. New York: Plenum Press, 1989. The inhibitors were diluted in toluene because of 11. L Galet, I Pezron, W Kunz, C Larpent J Zhu, C their high viscosity. The effect of toluene alone was Lheveder. Colloids Surfaces 151: 85, 1999. tested on one of the samples. Figure 15 shows the 12. DG Dervichian. J Chem Phys 7: 931, 1939. 13. WD Harkins, D Boyd. J Phys Chem 45: 20, 1941 near-infrared spectra of the four samples. 14. NK Adam. Physics and Chemistry of Surfaces. 3rd ed. The interpretation of Fig. 15 is that inhibitors have London: Oxford University Press, 1941. a solvating effect on the asphaltene aggregates. The 15. GA Overbeck, D Mobius. J Phys Chem 97- 7999, 1993. reduction in aggregation size is observed as a decrease 16. RD Smith, JC Berg. J Colloid Interface Sci 74: 273, in the extinction of radiation due to scattering. It is 1980. shown that the effect of the inhibitors is more promi ­ 17. E Pezron, PM Claesson, D Vollard. J Colloid Interface nent than the effect of toluene alone. The findings sug ­ Sci. 138: 245, 1990. gest that near-infrared spectroscopy could be used for 18. M Tomoaia-Cotisel, J Zsako, E Chifu, DA Cadenhead characterization of the effects of inhibitors on crude Langmuir 6:191, 1990. oils. 19. EP Honig, JHT Hengst, D den Engelsen. J Colloid Interface Sci 45: 92, 1973. 20. JH Brooks, AE Alexander. Proceedings of the Third International Congress of Surface Activity, Vol II, p 196. ACKNOWLEDGMENTS 21. ES Nikomaro. Langmuir 6: 410, 1990. 22. I Langmuir. Trans Faraday Soc 15:62, 1920. The technology programs Flucha I and II, financed by 23. G Binnig, CF Quate, C Gerber. Phys Rev Lett. 56: 930, the oil industry and the Norwegian Research Council 1986. (NFR), are acknowledged for PhD grants to Marit- 24. D Parrat, F Sommer, JM Soiled, Trans Minh Due. J Helen Ese, Jostein Djuve, Harald Kallevik, and Inge Trace Microprobe Tech 13: 343, 1995. Demulsifiers in the Oil Industry 619

25. JA Zasadzinski, R Viswanathan, DK Schwartz. J 46. DT Wasan, NF Djabbarah, MK Vora, ST Shah. Lect Garnaes, L Madsen, S Chiruvolu. JT Woodward, ML Notes Phys 105: 205, 1979. Longo. Colloids Surfaces 93: 305, 1994. 47. R Aveyard, PB Binks, PDI Fleischer, JR Lu. J Colloid 26. JA DeRose, RM Leblanc. Surfactant Sci Rep 22: 73, Interface Sci 139: 128, 1990. 1995. 48. R Aveyard, PB Binks, PDI Fleischer, RYe, JR Lu. In: J 27. J Sjoblom, L Mingyuan, T Gu, AA Christy. Colloids Sjoblom, ed. Emulsions - A Fundamental and Practical Surfaces 66: 55, 1992. Approach. NATO ASI Series, Dordrecht: Kluwer 28. AH Brown. Chem Ind 30:990. 1968. Academic 1992, p 97. 29. AJS Liem, DR Woods. Review of Coalescence 49. MA Krawczyk, DT Wasan, SS Chandrashekar. Ind Phenomena. AIChE Symp Ser 70, 8, 1974. Eng Chem Res 30: 367, 1991. 30. O Reynolds. Phil Trans R Soc (London) A177:157, 50. DT Wasan. In: J Sjoblom, ed. Emulsions - A 1886. Fundamental and Practical Approach. NATO ASI 31. SP Frank, KJ Mysels. J Phys Chem 66: 190, 1960. Series, Dordrecht: Kluwer Academic 1992, p 283. 32. Z Zapryanov, AK Malhorta, N Aderangi, DT Wasan. 51 RC Little. Environ Sci Technol 15: 1184, 1981. Int J Multiphase Flow 9: 105, 1983. 52. O Urdahl, AE Movik, J Sjoblom. Colloid Surface A 74: 33. CY Lin, JC Slatter. AIChEJ 28: 147, 1987. 293, 1993. 34. CY Lin, JC Slatter. AIChEJ 28: 786, 1982. 53. HB Clark, MT Pike, GL Rengel. Petrol Technol 7: 35. AK Malhorta. PhD thesis, Illinois Institute of 1565, 1982. Technology, Chicago, 1984. 54. HB Clark, MT Pike, GL Rengel. Proceedings of the 36. CV Stemling, LE Scriven. AIChE J 5: 514, 1959. AIME International Symposium on Oilfield and 37. AJ de Vries, Reel Trav Chim 77: 383, 441, 1958. Geothermal Chemistry, Houston, TX, 1979, SPE 38. SS Lang. PhD thesis. University of California, 1962. Paper 7894. 39. A Scheludko, E Manner. Trans Faraday Soc 64: 1123, 55. VB Menon, DT Wasan. Colloids Surfaces 19: 89, 1986. 1968. 56. J Djuve, X Yang, IJ Fjellanger, J Sjoblom, E Pelizzetti 40. A Vrij. Disc Faraday Soc 42:23, 1966. Colloid Polym Sci. In press, 2000. 41. H Sonntag, K Strenge. Coagulation and Stability of 57. M-H Esc. Langmuir film properties of indigenous crude Disperse Systems. New York: Halsteady-Wiley, 1969. oil components: influence of demulsifiers, PhD thesis, 42. TJ Jones, EL Neustadter, KP Whittingham. J Can University of Bergen, 1999. Petrol Technol 17- 100, 1978. 58. H Kallevik. Characterisation of crude oil and model oil 43. R Fiocco. US Patent 3 536 529, 1970. emulsions by means of near infrared spectroscopy and 44. MA Krawczyk. PhD thesis, Illinois Institute of multivariate analysis. PhD thesis, University of Bergen, Technology, Chicago, 1990. 1999. 45. DT Wasan, K Sampath, N Aderangi. AIChE Symp Ser 59. BP Singh, BP Pandey. Indian J Technol 29: 443, 1991. 76: 93, 1980. 60 'OC Mullins. Analyy Chem 62: 508-514, 1990.