You are on page 1of 9

SPE 93008

The Key To Predicting Emulsion Stability: Solid Content


Michael K. Poindexter, SPE, Nalco Energy Services; Shaokun Chuai, Memorial Sloan-Kettering Cancer Center; Robert A.
Marble and Samuel C. Marsh, Nalco Energy Services

Copyright 2005, Society of Petroleum Engineers Inc.


assessing the economic and technical risks associated with
This paper was prepared for presentation at the 2005 SPE International Symposium on Oilfield new projects or modifications to existing systems.
Chemistry held in Houston, Texas, U.S.A., 2 – 4 February 2005.
As related by Hopf, crude oil characterization played a
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
strong role in the development of organic chemistry.5 For
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to example, the cage hydrocarbon adamantane, a classic in
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at skeletal construct, was first isolated in minute amounts
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
(0.0004%) from petroleum in 1933, eight years before a
for commercial purposes without the written consent of the Society of Petroleum Engineers is synthetic confirmation. The characterization of crude oil has
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous thus developed both one molecule at a time and by groups of
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
molecules. Fractionation into saturates, aromatics, resins and
asphaltenes (SARA) remains the most common method for
Abstract classifying crude oils by groups. A variety of techniques exist
Chemical demulsifiers are routinely added in the oilfield to for classifying crude oils by SARA analysis. Two recent
effectively resolve water-in-crude oil emulsions. As used in advances in SARA characterization center on the use of high-
the common bottle test, demulsifiers in effect probe or performance liquid chromatography6 and infrared and near-
interrogate emulsion stability strength. Emulsion stability in infrared spectroscopy.7 The most studied petroleum group is
turn is defined by no less than three parameters – water drop, the asphaltene fraction which is an ensemble of molecules
oil dryness and interface quality. All three parameters are defined by both solubility and insolubility properties8 and
direct outputs of the bottle test, and collectively, all three certain general features regarding structure and atomic
provide a more complete picture of emulsion stability as composition.9 Recently, a process for quantifying the
opposed to the use of any singular parameter. overwhelming molecular complexity of crude has been
By selecting a wide variety of demulsifiers and illustrated by high resolution mass spectroscopy.10 Such
performing a standardized bottle test (as introduced in SPE quantifications are providing a fingerprint with which to
84610), emulsion stability from a variety of sites can be compare and contrast crude oils. This technique is quite an
quantified and compared. By coupling bottle test results with advancement that holds unprecedented potential. While
corresponding crude oil analytical data, fundamental questions certain atomic arrangement information is still not possible
concerning factors governing emulsion stability can be with mass spectroscopy, the ability to rapidly amass and store
quantified. The results show that solid content, not asphaltene crude oil information for any molecular parameter holds great
content or any other crude oil parameter investigated, is by far promise.
the best single predictor for gauging emulsion stability. Additionally, Buckley pointed out that the
Furthermore, statistical analysis via partition trees shows that characterization of crude oil and asphaltenes in particular is
emulsion stability is most aptly described using several input based not just on the output of laboratory methods but by
parameters as opposed to a single factor. This statistical process conditions which influence the solubility parameter
technique produces emulsion stability descriptions with and ultimate fate of the asphaltenes.11,12 Such an operational
Rsquare values on the order of 0.9. definition clarifies the need to couple crude oil properties (i.e.
characterizations) with crude oil behavior in the field. The
Introduction industrial practice of establishing and managing such critical
Crude oil characterization remains a challenging proposition. relationships ultimately falls under the realm of flow
Many crude oil characterizations are based on a separation assurance. Flow assurance can be divided into the deposition
scheme (e.g. distillation,1 chromatography,2,3 precipitation,4 of solids (e.g. hydrates, waxes, asphaltenes and scales) and
etcetera), a bulk property of the fluid (e.g. viscosity, API disturbed fluid behavior (e.g. foaming and emulsification).13
gravity, surface tension, etcetera) or a combination of The latter of these areas is the focus of this study.
properties. The intent of most characterizations is to relate One of the main requirements in upstream production
some property or group of properties back to the fluid’s involves the dehydration of oil to meet pipeline water
behavior in production or refining. Establishing a valid cause- specifications. Numerous parameters have been cited as
and-effect relationship can lead to greater confidence when playing a role in emulsion stability.14 Sorting through the
influence of some of the parameters can be accomplished via
2 SPE 93008

the field bottle test. In reality, the bottle test does not technique for investigating the type of multivariate problem
prescribe to a set procedure that is maintained for all emulsion posed,22 partition tree analysis does not create new latent
studies. There is no standard method for conducting a bottle variables that are a linear combination of the original input
test in the literature nor should there be.14 The bottle test is as variables. Latent variables (sometimes called eigenvectors)
much method as procedure. Some of the variables which are can serve as excellent predictors, but their exact underlying
initially examined in the field include the influence of meaning can be difficult to ascertain as they are in effect each
demulsifier chemistry, demulsifier dosage, test length, a mixture of variables. Instead, analysis via partition trees
temperature, degree of agitation, etcetera. Experimental uses input variable ranges most apt in describing the variance
design can help in establishing a bottle test procedure that is of the target variable. Partition tree results are very
representative of field conditions (i.e. one that ultimately interpretable regarding what independent variables are
produces dry oil and accurately simulates field conditions).15,16 selected in model construction.
Linking bottle test data to crude oil properties, which are
determined through extensive laboratory analyses, is a way to Experimental Procedures
further examine the factors that govern emulsion stability. As Chemicals. Thirty-six different demulsifiers from Nalco
introduced in SPE 84610, a select group of demulsifiers can be Energy Services, spanning six chemical families (see Table 1),
used to probe emulsion strength.17 These chemicals act to were used in each emulsion study. Demulsifier dosages for all
thoroughly interrogate emulsion stability. Emulsion stability tests were held at 150 ppm (by vol) using five percent
is quite a broad term. Oil-water resolution is often defined as dilutions (by wt) of chemical in heavy aromatic solvent.
water drop (i.e. bulk separation of water from oil measured Stoddard solvent or Varsol (Registry No. 8052-41-3) was used
visually) over time yet this description by itself is incomplete. as the diluent in determining oil dryness values.
The amount of water remaining in the oil also describes a
further aspect of emulsion stability, and this aspect is not Field Emulsions. Crude oil emulsions obtained from the field
adequately defined as a mere subtraction of the water drop were free of chemical demulsifier, and any free water found
from the total water cut (i.e. water content). As detailed later was drained from each sample before the start of each test.
in the section on Emulsion Stability Parameters, this residual The thirteen emulsions under investigation are listed in Table
water can be quantified using several different parameters. 2 along with a basic chemical description of each oil along
Some of the residual water remaining in the oil phase remains with their abbreviations. To conform with production
in an emulsified state. Quantifying this type of water provides conditions, the two Canadian emulsions (SGD-1 and SGD-2)
a more complete description of emulsion stability and helps and two Venezuelan emulsions (EV and WV) were diluted
illustrate the differences observed among different oilfield with field condensate after collection from the wellhead. This
emulsions. was necessary to keep the bottle test emulsion identical to the
It is generally recognized that emulsions have a time field process fluid.
dependent nature such that aged crude emulsions are more
difficult to resolve than their fresh counterparts.18,19 For this Bottle Test Procedure. Methods used to determine the
reason, all bottle test data reported in this study was generated theoretical oil-water interface level (commonly called a
in field laboratories using fresh emulsion samples. There is grindout), water drop, oil dryness and interface quality were
reasonable debate whether the bottle test accurately represents followed according to SPE 84610.17 Additionally, for
field conditions.20 Once a sample is removed from the extensive descriptions of the field bottle test, see the work of
process, sample denaturizing commences. Laboratory studies Leopold23 as well as Manning and Thompson.24
face the same dilemma unless conducted on live-oil samples.
Conducting such studies for a large number of bottle test Crude Oil Characterizations. Field emulsions were sent to a
conditions would be cost prohibitive in most instances. Thus, central laboratory where oil-water separation was performed
aside from performing a field evaluation, the on-site bottle test using the demulsifiers listed in Table 1. Crude oil collected
is probably the method most able to closely mimic the for the SARA analyses involved removal via syringe of the
conditions of a production facility. top portion of resolved crude oil (i.e. ca. 15% of the total oil
To determine the crude oil properties influencing the phase) from each lab bottle test. A literature procedure, based
bottle test results, numerous analyses were conducted on on n-heptane as the asphaltene precipitant, was used to
emulsions shipped back to the lab for characterization. In separate the crude oils into their SARA fractions.25 Crude oil
effect, the crude oil characteristics serve as the independent or viscosities were measured at both 25 °C and the bottle test
input variables used to describe emulsion stability. Emulsion temperature using a Brookfield Programmable DV-III+
stability in turn is described by the bottle test results (i.e. the Rheometer. Total acid numbers were determined according to
dependent or target variables). As pointed out by McLean and ASTM D66426 while naphthenic acid numbers involved solid
Kilpatrick, it is unlikely that one independent parameter can phase extraction of the maltenes (i.e. the n-heptane soluble
adequately predict emulsion stability.21 More than likely, a fraction) followed by titration in an isopropyl alcohol/toluene
combination of factors acting in tandem is responsible. A solution using potassium hydroxide. Both acid numbers are
variety of statistical techniques were used to search for cause- reported in milligrams KOH/gram of crude oil. Solid content
and-effect relationships. In this work, the partition tree determinations followed ASTM D480727 and results are
method was found to be very useful in providing descriptions reported in pounds per thousand barrels (PTB).28 Iron, nickel
of emulsion stability with Rsquare values on the order of 0.9. and vanadium contents were determined on a Jarrell-Ash 61E
Unlike principal component regression, another useful inductively-coupled argon plasma emission spectrometer and
SPE 93008 3

are reported in parts per million (ppm). Elemental analyses a parameter has an overwhelming influence, then this
were conducted by Galbraith Laboratories located in parameter will stand out in the analysis.
Knoxville, TN. To compare and contrast the bottle test data from the
different emulsions, thirty-six different demulsifiers were
Statistical Analyses. All statistical analyses (spec. linear selected to probe emulsion stability (see Table 1). The
regressions and the multivariate model partition tree) were demulsifiers span six distinctly different chemical classes, and
conducted using JMP Statistical Discovery Software Version the dosage for all chemicals used in each study was 150 ppm
5.0.1 from the SAS Institute, Inc., Cary, NC. by volume. To some degree, the choice of demulsifiers used
is arbitrary. Almost any group of chemicals having interfacial
Lab versus Field Emulsions activity could work in this methodology. All field bottle test
The study of water-in-crude oil emulsions is often conducted studies were conducted using fresh samples taken after the
in the laboratory using model systems. With such a controlled wellhead. After distributing 100 mL of emulsion into bottles
arrangement, given amounts of a crude oil component (often having graduation markings, each sample was dosed with a
asphaltenes, resins, naphthenic acids, solids or combinations single demulsifier (36 demulsifiers in all). A blank and the
of these components) are added to a solvent system which is incumbent chemical used in the field were also included in the
generally a mixture of heptane and toluene.21,29-33 This solvent field work, but this data was not included in the analyses of
system acts as a substitute for the lighter, less polar crude oil this report. The main reason for running the blank and
components (i.e. the saturate and aromatic fractions) and incumbent was to ensure the test results were representative of
permits adjustment of the aliphatic-aromatic balance that the field system. If the blank showed good performance or the
contributes largely to the solubility and aggregation state of incumbent showed poor performance, then the test results
the more polar components. Water is combined with the were suspect and discarded.
synthetic organic phase followed by agitation to create the The rationale for using a range of demulsifiers was to
emulsion. Generating emulsions in this fashion allows thoroughly probe emulsion stability. Some chemicals excel in
exacting control over the factors governing emulsion stability. producing more complete water drop, others have a propensity
The approach presented in this study uses the crude oil to produce dry oil, while some chemicals have no effect at all.
emulsions as obtained from the field. Each fluid brings its By subjecting each emulsion to the same set of chemical
own unique inventory of material to bear on the problem, probes at the same concentration, comparisons between
namely emulsion stability. Comparing emulsion stability emulsions become possible. Such a protocol does not imply
where numerous input or crude oil parameters are changed in that other factors, like test temperature or amount of agitation,
basically an uncontrolled manner would seem not helpful in are not also important. However, some of these other
unraveling the complexity. However, examining emulsion parameters are partly incorporated into the parameters under
stability in this manner becomes a multivariate problem that investigation (see Table 2 for the crude oil parameters). For
can be addressed using various statistical techniques. example, bottle test temperature would influence the viscosity
In field separators, all crude oil emulsion components are of the crude oil, and this parameter is one of the independent
present simultaneously and act in tandem, e.g. asphaltenes are variables under study.
never devoid of resins. The bottle test in essence maintains
the complex interplay of crude oil components as encountered Statistical Parameters
in the field. Similar to lab studies, this multivariate approach Crude Oil Variables. From each field site, emulsions were
permits the same basic questions to be addressed regarding sent to a central laboratory where the emulsions were resolved
factors contributing to emulsion stability. With all the crude using the demulsifiers from Table 1. The top portion of
oil constituents simultaneously engaged, if a given parameter resolved oil (ca. upper 15%) was collected and used for
or possibly group of parameters aptly describes an aspect of analysis. A variety of crude oil parameters were ascertained
emulsion stability across a wide variety of emulsions, then the regarding their possible contribution to emulsion stabilization.
identified parameter is highly relevant. In essence, this Most of the variables listed in Table 2 have been used to either
alternative methodology is a different yet complementary study emulsion stability in the lab or classify crude oils.
approach to the laboratory model studies. Crude oil groups, determined from fractionation, include
Furthermore, the crude oil parameters chosen for this asphaltenes, resins, aromatics, saturates, total acid content,
investigation were based in part on prior laboratory studies naphthenic acid content and solids. Bulk crude oil properties
that have shown how specific crude oil components or or parameters investigated include viscosity (measured at both
properties either influence emulsion stability or act as major room temperature and the bottle test temperature), aromaticity
crude oil classifiers. The bottle test approach coupled with (defined as the H/C ratio of the aromatic fraction)32 and the
statistical analysis of the data allows both the influence of content of three commonly reported transition metals– iron,34
each factor to be examined using linear regression analysis nickel and vanadium.35
(e.g. determining whether more or less of a certain component
is detrimental to emulsion stability) and the ranking of critical Emulsion Stability Variables. Bottle test results provide a
factors determined to be important using partition tree wealth of information. Water drop versus time, oil dryness
analysis. Clearly, examining the influence of each individual and quality of the interface are all aspects open to
crude oil parameter is partly confounded in this approach as investigation using the method. As mentioned, studies
the parameters, as stated earlier, act collectively. However, if reporting only one of these parameters can obscure the overall
process of water-oil separation. Good water drop does not
4 SPE 93008

necessarily imply dry oil.17 Only determining the water more water resides closer to the oil-water interface. Both thief
content in the oil can determine whether such a relationship and composite values are reported as percentages since both
holds. In this study, five of the most important bottle test values measure the amount of residual emulsion or water
parameters were selected to define emulsion stability. Mean remaining in the oil. To compare and contrast crude oil
values of the five bottle test parameters were calculated for emulsion stability, the mean values of the five bottle test
each emulsion and then the results were examined against the parameters for each emulsion were used as the dependent
crude oil parameters determined in the lab. In essence, variables in the statistical analyses (see Table 3).
emulsions showing minimal water drop or wet oil are likely
stabilized by components not present in oils showing high Statistical Analyses
levels of resolution. Linear Regressions. Linear regression is the most common
The five bottle test parameters consist of final water drop, way to establish whether a relationship exists between two
unresolved emulsion and total water remaining in a sample of variables. When regressing the independent crude oil
oil removed from the top oil and similarly unresolved parameters of Table 2 with the dependent bottle test variables
emulsion and total water remaining from the bulk oil phase. of Table 3, solid content had the highest Rsquare value for
Final water drop is expressed as a percentage where the last or each bottle test result. For comparison, the Rsquare values for
final water drop reading taken during the test is divided by the asphaltene content, the most commonly studied crude oil
water cut (or water content) of the emulsion. The procedure fraction, are also included (see Table 4). For both asphaltene
for determining the water cut is often called a “grindout” and and solid contents, higher concentrations favored emulsion
is described in SPE 84610 as well as other references.23,24 stabilization for all five bottle test variables. Regarding the
Chemicals promoting coalescence of water droplets show high solid content results, four of the five Rsquare values are
water drop values. around 0.8. This is remarkably high as the solids are vying
The second and third bottle test parameters describe with all the other crude components in stabilizing the
different aspects of oil dryness. Both parameters start by emulsion. Figure 1 illustrates these trends for the oil dryness
taking a small aliquot (sometimes called a thief or centrifuge parameter having the highest Rsquare value for asphaltene
cut23) of oil via syringe at a specified level above the content, namely slug(thief). Final water drop gave a
theoretical oil-water interface which is another term for water noticeably lower Rsquare value (0.22) for solid content than
cut. In these studies, the thief was taken 15 mL above the for the other four descriptions of emulsion stability. This
interface. One part of thief (about 6 mL) is added to a result indicates that of the crude oil properties selected no
graduated API centrifuge tube containing one part of cutter single variable adequately describes water drop.
solvent (typically Varsol, Stoddard solvent or some other low Upon further examination of Figure 1A, it might be worth
aromatic, high boiling solvent) and the mixture shaken. The considering a non-linear fit to describe the influence of
cutter solvent acts to dilute the emulsion and enhance oil- asphaltene content on slug(thief). Until about 8% asphaltenes,
water separation. Following centrifugation of the sample, this oil dryness parameter remains small in value. After this
unresolved emulsion is determined and reported as basic point, some of the data rises abruptly while part continues to
sediment in the thief, BS(thief). To resolve the emulsion show little asphaltenic influence. In fact, the three most
completely and separate any undetected residual water still asphaltenic crude oils (CAN, EV and WV) would seem to
held in the oil phase, a large dose of chemical (often called a have sufficient asphaltenes to stabilize an emulsion yet all
slug with treat levels typically at a few thousand parts per have relatively low slug(thief) values. Regarding the solid
million) is added. Most slugging chemicals have low content of these three crudes, CAN, WV and EV finish in
molecular weights with the sulfonate functionality. The sixth, ninth and last place, respectively. Conversely, for the
slugging agent is mixed into the solution, and the contents four crudes veering upward in Figure 1A (MT-1, MT-2,
centrifuged. As before, the water content is recorded and SGD-1 and SGD-2), all have lower asphaltenic contents than
reported as slug(thief). Demulsifiers showing low thief values CAN, EV and WV but are the most abundant in solid content.
are often viewed as good flocculants. While additional asphaltenic properties not accounted for,
The fourth and fifth bottle test parameters further describe such as molecular weight or heteroatom content, may play a
oil-water resolution. Following the two thief measurements, role in defining emulsion stability, it is clear that solid content
the bulk resolved water in the bottles is removed via syringe is playing a very critical role in predicting emulsion stability.
and the remaining oil, which is often wet, is shaken
thoroughly. One part of this wet oil is added to an API Partition Tree Analyses. Partition tree analysis is sometimes
centrifuge tube containing one part of cutter solvent and referred to as decision or regression tree analysis. This
processed using the same steps used in the thief technique is based on the earlier techniques called AID
measurements. These values are often called composites, as (Automatic Interaction Detection) and CART (classification
they quantify the water content in the final composite, and are and regression trees).36 Analysis via partition trees permits
reported as BS(composite) and slug(composite). Composite construction of highly informative multivariate models using
readings provide additional information regarding oil dryness more than one input variable. This statistical technique
and interface quality. For example, the closer the composite determines input variables over their most pertinent or
readings are to the thief values, the more evenly distributed is descriptive data range. In most regression techniques, the full
the residual water throughout the oil phase. Conversely, if range of the input variables is regressed against the full range
thief measurements are much lower than the composite values, of a given target variable. There are cases where a portion of
then the unresolved water in the oil exists as a gradient where the dependent variable is not well defined by the full range of
SPE 93008 5

the independent variable. Partition tree analysis allows such uses heptane-toluene mixtures in place of the crude oil derived
descriptions to be explored and identified. In effect, a virtue saturate-aromatic fractions. Returning to Table 5, higher
of the technique is that partition trees can account for non- aromatic content favors greater coalescence as seen for the
linear associations which are complementary to linear water drop parameter. However, higher aromatic content
regression analysis. leads to higher slug(composite) values. Thus, water drop
Partition tree models were used to examine the collective appears not to be related to oil dryness regarding aromatic
influences of the crude oil parameters on emulsion stability. content. A similar observation was noted in SPE 84610.
Partition tree analysis starts by selecting a target variable (i.e. Viscosity determined at room temperature received a
one of the five bottle test parameters from Table 3) and secondary split for BS(thief). As expected, higher viscosity
finding the independent variable that optimally splits the target hindered dehydration. Concerning the remaining Table 2
variable. In this study, the crude oil parameters serve as the entrants, crude oil viscosity measured at the bottle test
partitions for each of the objects (i.e. the emulsions). Groups temperature, both acid numbers, two of the metals (nickel and
resulting from the initial split can be partitioned further using vanadium) and the aromaticity of the aromatic fraction
the first split independent variable or another independent received no splits in any of the models.
variable. If the same independent variable is used in Of the five bottle test variables, water drop had the lowest
successive splits then a strong predictor has been identified. Rsquare value (0.70). The same trend was found in the linear
In each instance, the partition is dictated by finding the regression analyses (see Table 4). A possible explanation for
independent variable that best divides the data cluster under this result might be the beginning of demulsifier
consideration for split. overtreatment. In a related study using 50 ppm of
Figure 2 is representative of the technique and uses the demulsifier38 (versus the 150 ppm used in this study), solid
same output variable discussed in Figure 1, namely slug(thief). content was used to make the first split of all five bottle test
Solid content was selected as the first split with a value of 661 parameters, and the Rsquare value for the partition tree model
PTB where more solids gave higher slug(thief) values (i.e. describing water drop was 0.85. Thus, the drop in Rsquare
more wet oil). For the second level splits, asphaltene and solid value for the higher demulsifier dosage study could indicate
contents were used for the left- and right-side splits, that the demulsifiers are not just probing emulsion strength but
respectively. In summary, the bottle test parameter slug(thief) possibly contributing to emulsification.
is best described by a combination of solid and asphaltene
contents where high solid and asphaltene contents both Solid Content
increase the slug(thief) values. The solid content reported in Table 2 is defined by ASTM
Splits describing all five bottle test variables are D4807. Crude oil used for the analyses was taken from the
summarized in Table 5 along with the numerical value of the upper or top portion of the bottles. Extremely large solid
splits. Solid content clearly dominates the results. This crude particles are not likely to be found in this section of the bottle.
oil parameter received at least one split in each analysis, four The ASTM procedure uses 0.45 µm pore size filter paper and
of the five first level splits and nine of a possible fifteen splits. includes a hot toluene wash of the solids. Washing with a hot
In fact, solid content received all three splits used to describe aromatic solvent will remove most, if not all, of the organic
the BS(composite) parameter. No other crude oil property material associated with the solids. This would include the
was able to predict emulsion stability to this degree. For each heavier crude oil components such as the resins, asphaltenes
split involving solid content, higher levels always indicated and naphthenic acids. Thus, determining the interactions
greater emulsion stability. Similarly, asphaltene and resin between the isolated solids and the more polar crude oil
contents always partitioned such that higher levels components is not possible using the ASTM procedure.
corresponded to greater emulsion stability (see slug(thief) and Having such an organic-inorganic description well established
slug(composite) models). This agrees with much of the would ultimately be beneficial for defining the exact ensemble
literature as these more polar components are known to of constituents responsible for emulsion stabilization. This
contribute to emulsion stabilization. However, model studies avenue of investigation was not undertaken here. For oil sand
have also done much to shed light on the balance of bitumen, Kotlyar et al. demonstrated that solids and
asphaltenes and resins (typically reported as the asphaltenes are strongly associated.39
resin/asphaltene ratio, R/A) regarding emulsion stability. From this work, solid content proves to be an excellent
Generally, lower R/A ratios favor more stable emulsions.37 measure for describing emulsion strength. Further
While the R/A ratios are not listed in Table 2, this quotient descriptions using size distribution and elemental composition
was also included in all the partition tree analyses but was were not needed. When reviewing numerous past analytical
never selected as a partition. records involving crude oils not used in this study, it was
Iron content served as the first split variable for the final determined that all the crude oils had some solid content as
water drop parameter where crudes containing higher iron defined by ASTM 4807. There were no crude oils with zero
levels gave less water drop. Iron (especially ferric iron, Fe3+) solid content. Assuming this holds true for crude oils
is known to have an influential role in determining asphaltene throughout the world, then solid content should be reported in
polarity and ultimately asphaltene solubility properties. all crude oil assays.
Higher iron levels appear to promote asphaltene aggregation.34 In related work, aliquots of emulsion were taken from the
Aromatic content received two splits in Table 5. It is difficult oil-water interface for each of the emulsions studied. In other
to judge how this crude oil fraction influences emulsion words, these samples were taken from the bottom portion of
stability as it is rarely studied in the lab. Most model work the oil phase, not the top. This region of the bottle likely holds
6 SPE 93008

other valuable information regarding emulsion stability as the 2. ASTM D4124: “Standard Test Methods for Separation of
water content is generally higher nearer the interface. Solids Asphalt into Four Fractions”.
from this region were also isolated using ASTM D4807 and 3. McLean, J. D.; Kilpatrick, P. K. “Comparison of Precipitation
further examined using scanning electron microscopy. In each and Extrography in the Fractionation of Crude Oil Residua,”
Energy Fuels 1997, 11, 570-585.
case, the dimensions of isolated inorganic solids would be 4. Akbarzadeh, K.; Dhillon, A.; Svrcek, W. Y.; Yarranton, H. W.
classified as clay (0-3.9 µm) or silt (3.9-62.5 µm) particles and “Methodology for the Characterization and Modeling of
not the various larger sand particles.40 The importance of this Asphaltene Precipitation from Heavy Oils Diluted with n-
smaller size range has been discussed41 as well as other Alkanes,” Energy Fuels 2004, 18, 1434-1441.
important features of solids in contributing to emulsion 5. Hopf, H. “Adamantane and Other Cage Hydrocarbons,” in
stability.42,43 Classics in Hydrocarbon Chemistry; Wiley-VCH: Weinheim,
The ability of solids alone in stabilizing emulsions was 2000; Chapter 3.
firmly established years ago by Pickering.44 This work, 6. Fan, T.; Wang, J.; Buckley, J. S. “Evaluating Crude Oils by
SARA Analysis,” SPE paper 75228 presented at the SPE/DOE
however, does not intend to suggest that solid content is the
Thirteenth Improved Oil Recovery Symposium, Tulsa, 13-17
only factor contributing to crude oil emulsion stability. April 2002.
Numerous other factors have been shown to influence 7. Aske, N.; Kallevik, H.; Sjöblom, J. “Determination of Saturate,
emulsion strength. Nonetheless, solid content appears to be Aromatic, Resin, and Asphaltenic (SARA) Components in
the best predictive variable for describing emulsion strength. Crude Oils by Means of Infrared and Near-Infrared
Spectroscopy,” Energy Fuels 2001, 15, 1304-1312.
Conclusions 8. Pfeiffer. J. P.; Saal, R. N. J. “Asphaltic Bitumen as Colloid
Thirteen water-in-crude oil emulsions were investigated using System,” J. Phys. Chem. 1940, 44, 139-149.
the field bottle test followed by crude oil characterization and 9. Koots, J. A.; Speight, J. G. “Relation of Petroleum Resins to
Asphaltenes,” Fuel 1975, 54, 179-184.
fractionation into SARA and other commonly analyzed
10. Marshall, A. G.; Rodgers, R. P. “Petroleomics: The Next Grand
components (spec. solid, naphthenic acid and total acid Challenge for Chemical Analysis,” Acc. Chem. Res. 2004, 37,
content). A variety of demulsifiers was used on each emulsion 53-59.
to quantify and compare emulsion strength. Crude oil 11. Buckley, J. S. “Predicting the Onset of Asphaltene Precipitation
properties commonly cited as influencing emulsion stability from Refractive Index Measurements,” Energy Fuels 1999, 13,
were used to examine if one or more of the attributes could 328-332.
describe emulsion stability as defined by five bottle test 12. Wang, J. X.; Buckley, J. S.; Burke, N. E.; Creek, J. L. “A
parameters. The primary technique used to elucidate cause- Practical Method for Anticipating Asphaltene Problems,”
and-effect relationships was partition tree analysis. From the SPEPF 2004, 19, 152-160.
13. Fu, B. “Flow Assurance – A Technological Review of
five bottle test variables, solid content showed by far the
Managing Fluid Behavior and Solid Deposition to Ensure
greatest ability to predict emulsion strength. In comparison to Optimum Flow,” Deeptec 2000 7th Annual International Forum
the other crude oil variables used in the partition tree analysis, on Deepwater Technologies, Aberdeen, U.K.; IIR Limited, 26-
solid content received at least one split to describe each of the 28 January 2000.
bottle test parameters, four of the five first splits and nine of a 14. Kokal, S. “Crude Oil Emulsions: A State-of-the-Art Review,”
possible fifteen splits. No other input parameter came close to SPE paper 77497 presented at the SPE Annual Technical
matching this level of importance. In each case where solid Conference and Exhibition, San Antonio, 29 September - 2
content was found to be influential, higher solid content was October 2002.
shown to be detrimental. 15. Bowman, R. W.; Burton, W. D.; Pryor, J. A. “Statistically
Designed Oil Dehydration Tests,” SPE paper 6529 presented at
When considering projects involving emulsion resolution,
the Society of Petroleum Engineers of AIME, Bakersfield, 13-15
this study indicates that the reporting of crude oil properties April 1977.
should include solid content as well as the other commonly 16. Poindexter, M. K.; Lindemuth, P. M. “Applied Statistics: Crude
reported values. Without consideration of this key variable, Oil Emulsions and Demulsifiers,” J. Disp. Sci. Tech. 2004, 25,
predictions of emulsion stability strength will likely be 311-320.
lacking. Results from this study do not imply that the crude 17. Poindexter, M. K.; Chuai, S.; Marble, R. A.; Marsh, S. C.
oil parameters used herein are the only ones responsible for “Classifying Crude Oil Emulsions Using Chemical Demulsifiers
emulsion stability. Beside crude oil properties, other and Statistical Analyses,” SPE paper 84610 presented at the
parameters influencing emulsion stability likely include water Annual Technical Conference and Exhibition, Denver, 5-8
October 2003.
chemistry as well as process conditions.
18. Mikula, R. J.; Munoz, V. A. “Characterization of Demulsifiers,”
in Surfactants: Fundamentals and Applications in the Petroleum
Acknowledgments Industry; Schramm, L. L., Ed.; Cambridge University Press:
The authors thank Nalco Energy Services for permission to Cambridge, 2000; Chapter 2, p 51-77.
publish this work as well as Becky Ramsey, Bob Pultz and 19. Jones, T. J.; Neustadter, E. L.; Whittingham, K. P. “Water-in-
Terry Street for determining the acid numbers and the metal Crude Oil Emulsion Stability and Emulsion Destabilization by
and solid contents. Chemical Demulsifiers,” J. Can. Petrol. Tech. 1978, 17, 100-
108.
References 20. Gramme, P. E. “The Behavior of Crude Oil Fluids in Oil and
1. Speight, J. G. The Chemistry and Technology of Petroleum, 3 rd Gas Production and Processing with Focus on Multiphase
ed.; Marcel Dekker: New York, 1998; Chapter 8, p 324. Separation and Flow,” presented at the AIChE 2nd International
SPE 93008 7

Symposium on the Chemistry and Physics of Petroleum 37. Schorling, P. -C.; Kessel, D. G.; Rahimian, I. “Influence of the
Oil/Water Emulsions, Houston, 18 March 1999. Crude Oil Resin/Asphaltene Ratio on the Stability of Oil/Water
21. McLean, J. D.; Kilpatrick, P. K. “Effects of Asphaltene Emulsions,” Colloids Surfaces A: Physicochem. Eng. Aspects
Solvency on Stability of Water-in-Crude Oil Emulsions,” J. 1999, 152, 95-102.
Colloid Interface Sci. 1997, 189, 242-253. 38. Poindexter, M. K.; Chuai, S.; Marble, R. A.; Marsh, S. C. “Solid
22. Esbensen, K. H. Multivariate Data Analysis – In Practice; 5th Content Dominates Emulsion Stability Predictions,” accepted
ed.; CAMO Process AS: Oslo, 2002, p 128. for publication in Energy Fuels.
23. Leopold, G. “Breaking Produced-Fluid and Process-Stream 39. Kotlyar, L. S.; Sparks, B. D.; Woods, J. R.; Chung, K. H.
Emulsions,” in Emulsions – Fundamentals and Applications in “Solids Associated with the Asphaltene Fraction of Oil Sands
the Petroleum Industry; Schramm, L. L. Ed.; American Bitumen,” Energy Fuels 1999, 13, 346-350.
Chemical Society: Washington, 1992; Chapter 10, p 341-383. 40. Schramm, L. L. Dictionary of Colloid and Interface Science,
24. Manning, F. S.; Thompson, R. E. “Dehydration of Crude Oil,” Wiley-Interscience: New York, 2001, p 197.
in Oilfield Processing Volume 2: Crude Oil; PennWell: Tulsa, 41. Menon, V. B.; Wasan, D. T. “A Review of the Factors Affecting
1995; Chapter 7, p 113-143. the Stability of Solids-Stabilized Emulsions,” Separation Sci.
25. Poindexter, M. K.; Zaki, N. N.; Kilpatrick, P. K.; Marsh, S. C.; Tech. 1988, 23, 2131-2142.
Emmons, D. H. “Factors Contributing to Petroleum Foaming. 1. 42. Menon, V. B.; Wasan, D. T. “Characterization of Oil-Water
Crude Oil Systems,” Energy Fuels 2002, 16, 700-710. Interfaces Containing Finely Divided Solids with Applications
26. ASTM D664: “Standard Test Method for Acid Number of to the Coalescence of Water-in-Oil Emulsions: A Review,”
Petroleum Products by Potentiometric Titration.” Colloids Surf. 1988, 29, 7-27.
27. ASTM D4807: “Standard Test Method for Sediment in Crude 43. Aveyard, R.; Clint, J. H. “Solid Particles at Liquid Interfaces,
Oil by Membrane Filtration.” Including Their Effects on Emulsion and Foam Stability,” in
28. Hyne, J. M. Dictionary of Petroleum Exploration, Drilling, and Adsorption and Aggregation of Surfactants in Solution;
Production: Tulsa, 1991. Surfactant Science Series, Marcel Dekker: New York, 2003;
29. McLean, J. D.; Kilpatrick, P. K. “Effects of Asphaltene Vol. 109, Chapter 3, p 61-90.
Aggregation in Model Heptane-Toluene Mixtures on Stability of 44. Pickering, S. U. “Emulsions,” J. Chem. Soc. 1907, 91, 2001-
Water-in-Oil Emulsions,” J. Colloid Interface Sci. 1997, 196, 2021.
23-34.
30. Kim, Y. H.; Wasan, D. T. “Effect of Demulsifier Partitioning on
the Destabilization of Water-in-Oil Emulsions,” Ind. Eng. Chem.
Res. 1996, 35, 1141-1149.
31. Gafonova, O.V.; Yarranton, H. W. “The Stabilization of Water-
in-Hydrocarbon Emulsions by Asphaltenes and Resins,” J. Table 1. Chemical Demulsifier Families
Colloid Interface Sci. 2001, 241, 469-478. Number of Demulsifiers Chemical Family
32. Aske, N.; Orr, R.; Sjöblom, J. “Dilatational Elasticity Moduli of 8 Resins with nonyl substituent
Water-Crude Oil Interfaces Using the Oscillating Pendant
5 Resins with butyl substituent
Drop,” J. Disp. Sci. Tech. 2002, 23, 809-825.
33. Sztukowski, D. M.; Yarranton, H. W. “Characterization and 8 Propylene glycol backbone
Interfacial Behavior of Oil Sands Solids Implicated in Emulsion 5 Triol backbone
Stability,” J. Disp. Sci. Tech. 2004, 25, 299-310. 4 Hexol backbone
34. Nalwaya, V.; Tangtayakom, V.; Piumsomboon, P.; Fogler, S. 6 Crosslinked resins
“Studies on Asphaltenes through Analysis of Polar Fractions,”
Ind. Eng. Chem. Res. 1999, 38, 964-972.
35. Barwise, A. J. G. “Role of Nickel and Vanadium in Petroleum
Classification,” Energy Fuels 1990, 4, 647-652.
36. Breiman, L.; Friedman, J. H.; Olshen, R. A.; Stone, C. J.
Classification and Regression Trees; Chapman and Hall/CRC:
Boca Raton, 1998.

Table 2. Summary of Crude Oil Properties


Bottle Test
Viscosity, Viscosity, % % % % Acid Naphthenic Fe, Ni, V, Aromatic Temperature
Crude Oil (source) Designation cP (25 °C) cP (test) Saturates Aromatics Resins Asphaltenes Numbera Acidsa Solidsb ppm ppm ppm H/Cc (°F)
Alberta CAN 47,150 360 18.8 51.9 14.6 14.3 0.9 0.4 401 6.8 64 160 1.54 194
Alberta SGD-1 8,850 144 18.6 48.9 13.7 13.1 1.0 1.0 661 3.6 66 150 1.52 195
Alberta SGD-2 7,618 168 15.9 54.7 13.6 12.3 1.8 0.6 985 14 62 150 1.52 199
California (offshore) COf 235 24 14.8 43.9 16.7 11.4 1.1 0.4 257 120 69 175 1.71 180
California Midway-Sunset MS-1 98,979 480 10.7 57.4 24.1 7.9 3.3 1.8 156 32 67 75 1.51 190
California Midway-Sunset MS-2 52,200 330 14.6 53.1 25.0 8.3 5.3 2.6 634 19 53 62 1.56 190
Eastern Venezuela EV 4,385 114 13.8 49.0 14.1 14.5 4.0 1.7 96 6.8 76 335 1.55 194
Gulf of Mexico GM-1 17 8 65.2 26.8 2.6 0 1.1 0.3 176 13 0.9 0.7 1.54 120
Gulf of Mexico GM-2 214 29 17.0 59.9 11.9 3.0 1.9 1.3 286 22 21 60 1.54 165
Montana MT-1 38 18 38.6 28.6 4.3 8.1 1.1 0.2 664 5.0 2.9 2.7 1.46 125
Montana MT-2 218 57 34.5 30.8 4.2 11.6 0.3 0.1 1,090 13 4.6 1.0 1.46 125
Western Venezuela WV 90,800 630 11.6 49.5 16.9 17.0 2.1 0.4 208 4.8 86 980 1.52 199
Wyoming WY 179 44 24.2 49.3 9.9 7.6 0.3 0.2 157 2.7 34 130 1.54 132

a
Reported as mg KOH/g of crude. b Reported as pounds/1,000 barrels of crude (PTB). c
Atomic ratio for aromatic fraction.
8 SPE 93008

Table 3. Mean Bottle Test Parameters


Water Drop Oil Dryness Oil Dryness Interface Interface
Crude Oil (Final), % BS(thief), % slug(thief), % BS(composite), % slug(composite), %
CAN 50.2 2.0 5.7 3.0 13.4
SGD-1 60.7 9.6 14.4 20.4 25.5
SGD-2 25.6 22.0 30.0 22.9 42.3
COf 62.1 3.1 3.8 4.1 13.6
MS-1 48.3 2.1 2.7 2.5 14.4
MS-2 64.9 0.9 2.6 7.9 21.6
EV 54.8 1.1 3.1 2.4 6.0
GM-1 54.9 0.4 1.5 2.6 10.8
GM-2 56.9 0.2 1.7 3.8 10.8
MT-1 57.3 15.2 16.5 15.4 19.6
MT-2 53.9 28.5 24.0 27.5 27.7
WV 75.5 1.3 7.4 4.0 10.7
WY 80.3 0.6 1.1 1.0 5.5

Table 4. Rsquare Values for Two Crude Oil


Properties versus Bottle Test Parameters
Bottle Test Parameter Asphaltenes, % Solids, PTB
Water Drop (final), % 0 0.22
BS(thief), % 0.05 0.81
Slug(thief), % 0.12 0.79
BS(composite), % 0.05 0.89
Slug(composite), % 0.04 0.81

Table 5. Summary of Partition Tree Results


Bottle Test Parameter R2 First Split Variable Second Split Variables (left - right)

iron solids - aromatics


Water Drop (final), % 0.704 ≥5/<5 ppm ≥985/<985 PTB - <49.3/≥49.3 %

solids solids - viscosity (25 °C)


BS(thief), % 0.973 <664/≥664 PTB <661/≥661 PTB - <218/≥218 cP

solids asphaltenes - solids


Slug(thief), % 0.969 <661/≥661 PTB <11.4/≥11.4 % - <985/≥985 PTB

solids solids - solids


BS(composite), % 0.970 <661/≥661 PTB <634/≥634 PTB - <985/≥985 PTB

solids resins - aromatics


Slug(composite), % 0.941 <634/≥634 PTB <14.6/≥14.6 % - <54.7/≥54.7 %
SPE 93008 9

35 35
Figure 1A Figure 1B
30 SGD-2 30 SGD-2

25 25
MT-2 MT-2
Slug(thief), %

Slug(thief), %
20 20
MT-1 MT-1
15 SGD-1 15 SGD-1

10 10
WV WV
5 CAN 5 CAN
COf EV COf
EVMS-1
GM-1 GM-2 MS-1
MS-2
WY WY
GM-1GM-2 MS-2
0 0
-5 0 5 10 15 20 0 250 500 750 1000 1250
Asphaltenes (%) Solids, PTB

Figure 1A and 1B. Plots of asphaltenes and solids versus the oil dryness parameter, slug(thief).
Each point on the plots represents a crude oil emulsion.

35
SGD-2
30
Solids, PTB>=985
25
MT-2
Slug(thief), %

Solids, PTB>=661
20
Solids, PTB<985 MT-1
15 SGD-1
10 All Rows
WV
5 Asphaltenes
CAN (%)>=11.4
Solids, PTB<661 COf
Asphaltenes
MS-2 (%)<11.4 MS-1 EV
GM-2 GM-1 WY
0
Asphaltenes (%)<11.4 Asphaltenes (%)>=11.4 Solids, PTB<985 Solids, PTB>=985
Solids, PTB<661 Solids, PTB>=661
All Rows

Figure 2. Partition tree of slug(thief) using the crude oil parameters from Table 2.

You might also like