SPE-189727-MS Phase Behavior Modelling of Oils in Terms of ...

SPE-189727-MS

Phase Behavior Modelling of Oils in Terms of SARA Fractions

D. Gutiérrez, AnBound Energy Inc.; R. G. Moore, S. A. Mehta, and M. Ursenbach, University of Calgary; A. Bernal,AnBound Energy Inc

Copyright 2018, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Canada Heavy Oil Technical Conference held in Calgary, Alberta, Canada, 13-14 March 2018.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contentsof the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflectany position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the writtenconsent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations maynot be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

AbstractOne of the key steps towards improving the predictability of air-injection-based processes relies on thedevelopment of accurate phase behavior models of the oil.

Historically, for in-situ combustion (ISC) in heavy oils and bitumens, phase behavior was often ignored,as the physical aspects of the process (e.g. distillation) were not considered to be as significant as theoxidation reactions. However, this step is important for several reasons. First, the compositional modelshould reflect the phase behaviour of the original fluids. Second, reaction rates are dependent on theconcentration of the reactants, which in turn are affected by the volatility of the components. This isparticularly important for lighter oils (but not unimportant for heavier oils) where the phase equilibriumbetween the liquid and vapour can have a significant impact on the flammability range for vapour phasecombustion at given temperature and pressure conditions. Finally, for the case of lighter oils, a good phasebehaviour model is required to capture the compositional effects of the resulting flue-gas drive.

This study presents a practical workflow to develop a phase behavior model in terms of SARA fractions(saturates, aromatics, resins and asphaltenes), which is aligned with the reaction modelling approachused in most kinetic models. The methodology requires conventional oil characterization (i.e. based ondistillation cuts) and conventional phase behavior experiments (e.g. differential liberation), as well as oilcharacterization in terms of SARA fractions.

The first step of the method consists of splitting of the heaviest oil fraction (i.e. plus fraction), followedby the lumping of all single-carbon-number components, in such a way that the new oil characterizationhonours the SARA data available, such as composition, and physical properties of each fraction (e.g.molecular weight). In addition, the gas components (e.g. Methane) would be treated as additionalcomponents as necessary. The second step is to tune an equation of state (EoS), in terms of the SARA-basedmodel, to match the relevant laboratory experiments. Finally, the tuned EoS would be used to export theequilibrium constants (K-value tables) to the thermal numerical simulator.

Different examples on the application of the phase behavior modelling workflow are presented anddiscussed in detail, for heavy and light oils. This work opens up opportunities to model the ISC process forany oil (i.e. light or heavy) by utilizing the currently available kinetic models, which in turn is an importantstep towards improving the predictability of ISC processes using reservoir simulation.

2 SPE-189727-MS

IntroductionThe traditional approach to characterize the oil is based on distillation cuts, which has proven to be aneffective way to model the phase behaviour of complex hydrocarbon systems and allows capturing themain compositional effects present during many EOR processes by using a properly tuned equation of state.However, this approach implicitly assumes that all single-carbon-number (SCN) components included ina particular distillation cut will behave similarly during the chemical reactions occurring in air-injection-based processes, which is not necessarily the case.1

Fig. 1 illustrates heat flow traces of two pure "C16" hydrocarbons of different nature (one paraffinicand one aromatic) at different pressures, which were obtained using a pressurized differential scanningcalorimeter (PDSC) in an air environment2. Since they have the same carbon number, a traditional oilcharacterization procedure would likely place them into the same pseudo-component; however, from theoxidation point of view, their behaviours are different from each other. While the paraffin tends to burn in thelow temperature range, the aromatic burns in the high temperature region. As discussed previously1,3, thisdifference is very important to the performance of the process and should be considered when modelling it.

In this sense, the choice of the pseudo-components should allow for proper modelling of the phasebehavior of the oil as well as the different oxidation behaviors exhibited by the different oil fractions.

Figure 1—Comparison of the Heat Flow Curves of n-hexadecaneand 2-ethylanthracene in an Air Environment, at Different Pressures2

Figure 2—Heat Flow Traces of Each SARA Fraction of a Light Oil, in and Air Environment at 100 kPa Pressure2

SPE-189727-MS 3

When it comes to ISC modeling, the oil has traditionally been fractionated according to its solubilityin solvents and their affinity for absorption on solid granular packs. This usually results in the oil beingcharacterized in terms of maltenes and asphaltenes, or in terms of SARA fractions. In the latter approach,the maltenes are further separated into saturates, aromatics and resins to complete the SARA fractions. Thistype of characterization comes from the different oxidation behaviours exhibited by those fractions4-10. Forinstance, Fig. 2 shows the oxidation behaviour of the SARA fractions of a light oil (37 °API) which illustratesthat while saturates tend to burn (i.e. undergo bond-scission reactions) readily in the low temperature region,aromatics, resins and asphaltenes tend to burn in the high temperature range1.

Using SARA fractions allows for capturing the main oxidation behaviors of the oil but, most importantly,it also has the potential to allow for proper modelling of its phase behavior, which is the main subject ofthis paper. This has been an important area of research11-14 but, to the authors’ knowledge, this is the firsttime a comprehensive pseudoization technique is presented to take into account both reactivity and phasebehavior of a reactive-compositional process such as in-situ combustion. This could potentially open upnew opportunities to try and use the available SARA-based kinetic models15-17 to any oil (i.e. light or heavy),and improve the predictability of the ISC process.

Phase Behavior Modelling WorkflowThe methodology proposed in this paper requires conventional oil characterization (i.e. based ondistillation cuts) and conventional phase behavior experiments (e.g. differential liberation), as well as oilcharacterization in terms of SARA fractions. The former is data which usually is readily available for mostreservoirs as it is gathered during the early stages of field appraisal. The latter, on the other hand, is lesscommon as SARA fractionation is not performed routinely but only carried out for specific studies such asasphaltene precipitation or EOR projects.

The concept behind the workflow is very simple. SARA fractionation is performed on samples ofdead oil (i.e. live oil flashed to standard conditions), so the idea is simply to take the conventional oilcharacterization (i.e. in terms of SCN fractions) of the flashed oil (Fig. 3), and describe the oil in termsof 4 pseudo-components, which will represent the saturates, aromatics, resins and asphaltenes (Fig. 4).This representation should honour the measured SARA composition (i.e. mass fractions) as well as anymeasured physical properties such as molecular weight and specific gravity. This exercise provides a"direct" relationship between the two oil characterization approaches as it will indicate the ranges ofSCN components that will belong to each of the SARA fractions. In other words, it provides the pseudo-component lumping scheme that needs to be applied when performing the phase behavior modelling workon the live oil sample.

4 SPE-189727-MS

Figure 3—Mass Percent Composition of Flashed Oil

Figure 4—Flashed Oil Characterization in SARA Fractions

It is worth noting that the "direct" relationship between the two oil characterization approaches is not strictfrom the physical point of view, but simply a way to reconcile the two methods. For instance, saturate andaromatic components span over a large number of SCN fractions and can have the same carbon number, sowhen lumping the SCN fractions they will fall into the same SARA fraction due to their similar distillationcharacteristics. Nevertheless, this SARA approach honours the basic molecular weight (MW) relationship:Saturates’ molecular weight (MW) < Aromatics’ MW < Resins’ MW < Asphaltenes’ MW, so Fig. 4 is betterinterpreted as a plot of composition vs molecular weight.

The challenge with this method is that some of the SARA pseudo-components can be very heavy (e.g.asphaltenes) and conventional oil characterization may not span to that degree of detail. In fact, traditionalcompositional analyses are seldom performed beyond a carbon plus fraction of C36+, which can representa large fraction of the oil, particularly in heavy oils. To overcome this, the plus fraction needs to be splittedby using one of the plus fraction splitting methods18-20. Once that is complete, SCN fractions of the dead oilsample can be lumped into SARA fractions, and define the lumping scheme required for the phase behaviormodelling work.

Clearly, defining the splitting and final lumping scheme is a trial and error procedure which needs tobe performed in order to honour all the SARA-based data available. However, this process is facilitated

SPE-189727-MS 5

by the methods available in the literature for plus fraction splitting/lumping18-22, some of which have beenimplemented in commercial phase behavior software.

Following is a description of the steps required to carry out the phase behavior modeling workflow. It isassumed that the oil sample is representative, the experimental data have passed a quality control, and phasebehavior modelling of the oil can be successfully performed using the distillation-based data. Otherwise,the chances of success of this workflow are minimal

Definition of the Oil Characterization SchemeThe goal of the methodology proposed in this paper is to model the phase behavior of the oil in termsof SARA fractions. Hence, the definition of the pseudoization scheme should be based on the dead oilsample used for the SARA characterization, which is equivalent to the flashed oil sample obtained duringthe conventional distillation-based characterization.

To achieve this, the carbon plus fraction needs to be characterized, which is done in two steps: splittingof the plus fraction, and lumping into SARA fractions.

Splitting of the Plus Fraction of the Flashed OilThis step consists in the estimation of the molar distribution of the plus fraction (i.e. mole fraction vs carbonnumber), and results in the plus fraction being splitted into a larger number of SCN fractions. Normallythe critical properties of the resulting SCN fractions would be estimated as well but, since the purpose ofthis step is just to define the characterization scheme, it is not yet necessary. That will be done during thecharacterization of the actual live oil sample which is the one of interest.

This splitting of the plus fraction can be done in several different ways (Fig. 5), by using some ofthe methods/distributions available in the literature such as the exponential distribution18,20 or the Gammadistribution19(Fig. 6). In this paper we make use of the Gamma distribution (Fig. 6), mostly because we cantake advantage of the extended compositional analysis data (i.e. say up to C20+ or more) to estimate themolar distribution of the whole sample. And this can be done to any carbon number we desire.

Figure 5—Sample Plus Fraction Splitting Distributions

6 SPE-189727-MS

Figure 6—Gamma Function23

In absence of an extended compositional analysis (say less than C12+), any method could be used.However, it would possibly be a longer trial and error process, as not any molar distribution would honorthe available SARA data, and may not allow a reasonable match of the phase behavior experiments.

Lumping Flashed Oil into SARA FractionsThis step, in this context, consists in the grouping of the different SCN fractions into different pseudo-components. Since the purpose here is just to define the lumping scheme, it is not yet necessary to estimatethe critical properties of the fractions. That will be done during the characterization of the actual live oilsample which is the one of interest.

This grouping is typically done based on the measured mass fractions of the SARA pseudo-components,which is the only data usually available. For instance, when doing the splitting, the carbon number of thenew plus fraction is chosen in such a way that it honours the mass fraction of the asphaltenes. Similarly, therest of the pseudo-components (i.e. SAR) are grouped following the same criteria.

Nevertheless, if there are additional SARA data available such as molecular weight and/or specific gravityof each of the fractions, they would need to be honoured as well. This means that the splitting distributionfunction might need to be revisited several times until the desired SARA data is fully honoured. Althoughsomewhat tedious, this procedure can be done and has been reported in the literature24.

Oil Characterization of the Live SampleOnce the oil characterization scheme has been chosen, the next step is to apply it to the live oil (singlephase) sample. This includes the splitting of the plus fraction, and further lumping, as well as the estimationof the critical properties of each of the resulting SCN fractions, and SARA pseudo-components. The maindifference with the previous step is that the composition of the live oil sample contains volatile componentswhich are not present in the dead oil sample. And these need to be treated/lumped separately, in additionto the SARA fractions.

One of the main challenges of this step is the estimation of the critical properties of the heaviest pseudo-components, as the traditional correlations available in the literature25-27, and implemented in commercialsoftware, are not usually suited for that. In fact, researchers have acknowledged that they have preferredto use the regular approach for characterizing the oil, as attributing critical properties to SARA fractionsis indeed difficult14.

Fortunately there are some correlations which can be easily used for this purpose28-29. Also, estimatingthe critical properties of the SCN fractions may not be necessary, and that could be done after lumping intoSARA fractions. The main aspect one needs to pay attention to is ensuring that the properties of the pseudo-

SPE-189727-MS 7

components (i.e. critical pressure, critical temperature and acentric factor) do follow the right trend as themolecular weight increases, which could be accidentally violated during the splitting/lumping process.

During the splitting of the plus fraction, molecular weight and specific gravity of each of the SCNfractions is estimated based on the molar distribution applied. After lumping, each of the resulting pseudo-components (i.e. SARA), will also have an estimated molecular weight and specific gravity, which can beused to obtain a first estimate of the critical properties of the SARA fractions28-30. These will be used a firstestimate during the matching of the different phase behavior experiments.

Match of Phase Behavior ExperimentsOnce the characterization of the live oil sample has been completed, the next step is to match the phasebehavior experiments, by tuning a cubic equation of state. The main tuning parameters are the criticalproperties of the heaviest fractions as well as the binary interaction coefficients, and volume shifts of allpseudo-components.

Although this is the last step of the workflow, by no means should it be considered independent fromthe previous steps. In fact, just like the regular modelling approach, this workflow has higher chances ofsuccess when the matching is performed along with the splitting/lumping steps. This helps to keep track ofthe moment when the splitting or lumping process is changing the character of the fluid. For instance, it isa common practice to create a P-T diagram of the original fluid and continue to monitor it along the wayto make sure it is not distorted. The same can be done for basic properties such as the saturation pressure,API gravity, and gas-oil ratio.

The choice of which experiments to match depend on the data available but, most importantly, itdepends on the EOR process of interest. For heavier oils modelling of the basic PVT experiments (constantcomposition expansion, differential liberation and separator tests) is likely sufficient, and an additional focusmight be done on the viscosity data as a function of temperature. Light oils, on the other hand, might requiremodelling of swelling tests or multi-contact experiments to capture possible multi-contact miscibility of theoil with the flue gases resulting from the combustion process.

Also, as will be seen from the light oil case study, lumping into just four –SARA-pseudo-components(plus volatile ones) may not be enough and further splitting of a pseudo-component may be required to beable to model the phase behavior of the oil properly.

Finally, as a last quality check step, we should always perform a flash of the single-phase (live) sampleto standard conditions to make sure the resulting composition of the flashed liquid does indeed match theSARA mass fractions of the original dead oil sample.

Once a tuned EoS has been developed, it can be used to export the equilibrium constants (K-value tables)to the thermal numerical simulator. Or, it can simply be used directly in a fully compositional thermalsimulator, if available.

Case StudiesThe proposed workflow has been successfully applied to different oils, and three of the case studies arediscussed below. The software used is CMG’s phase behavior modelling package, Winprop.

Table 1 presents the flashed oil and single-phase compositions (as mass fractions) of the three oils. OilA is a heavy oil with an API gravity of 11. Oil B is heavier than water, having and API gravity of 9, andOil C, is a light oil, with an API gravity of 34.

All oil samples were characterized up to C30+, which facilitated the estimation of the molar distribution.The Gamma function was used for this purpose, and the correlations of Nji et al.28-29 were used to estimatethe critical properties of the pseudo-components.

8 SPE-189727-MS

Oil A - 11 °APIThe plus fraction (C30+) of the flashed oil sample was splitted to C71+ in order to honour the mass fractionof the asphaltenes, which was 10.7 percent. The extended analysis from C12 to C30+ was used to estimatethe alpha parameter that best fitted the data, which was 1.34. The molar distributions of the sample beforeand after the split are shown in Fig. 7. Furthermore, Fig. 8 illustrates the lumping scheme derived, whichwas used to characterize the live oil in terms of SARA fractions.

Figure 7—Oil A Molar Distribution of Flashed Oil Before and After Splitting

Figure 8—Oil A Characterization Scheme in Terms of SARA Fractions

SPE-189727-MS 9

Table 1—Reservoir Fluid Analyses of Different Oils

10 SPE-189727-MS

Having determined the lumping scheme to represent Oil A in terms of SARA fractions, the next step wasto apply it to the actual live oil sample of interest and perform the match of the laboratory experiments.At a reservoir temperature of 145 °F, the saturation pressure was measured as 55 psia, and the followingexperiments were available: a constant composition expansion (CCE) and a separator test. The resultingphase envelope is illustrated in Fig. 9, which looks as expected from the original characterization. Theresults of the match are presented in Figs. 10 and 11. A second EoS for surface conditions was used in orderto match the separator API of 11 and the GOR of 3 scf/stb; however, only the volume shifts were variedas compared to the original EoS.

The resulting characterization of the oil is presented in Table 2. Notice the addition of components CO2and N2 and pseudo-component C1-C7.

Figure 9—Oil A Pressure-Temperature Diagram

Figure 10—Oil A CCE Results

SPE-189727-MS 11

Figure 11—Oil A Density and Viscosity

Figure 12—Oil B Characterization Scheme in Terms of SARA Fractions

Table 2—Oil A Characterization in Terms of SARA Fractions

Oil B – 9 °APIThe plus fraction (C30+) of the flashed oil sample was splitted to C135+ in order to honour the mass fractionof asphaltenes, which was 14.4 percent. The extended analysis from C12 to C30+ was used to estimate thealpha parameter that best fitted the data, which was 0.9355. Fig. 12 illustrates the lumping scheme derived,which was used to characterize the live oil in terms of SARA fractions.

The lumping scheme shown in Fig. 12 was applied to the actual live oil sample and the laboratoryexperiments were matched with an EoS. At a reservoir temperature of 114 °F, the saturation pressure wasmeasured as 562 psia, and the following experiments were available: a CCE, a differential liberation (Diff.

12 SPE-189727-MS

Lib.), and a separator test. The resulting phase envelope is illustrated in Fig. 13, which looks as expectedfrom the original characterization. Figs. 14 to 16 show the match of the different experiments. A secondEoS for surface conditions was used in order to match the separator API of 9, the GOR of 53 scf/stb andthe formation volume factor of 1.052; however, only the volume shifts were varied as compared to theoriginal EoS.

Due to its heavy nature, the main challenge encountered with this sample was the estimation of the criticalproperties of the pseudo-components.

The resulting characterization of the oil is presented in Table 3. Notice the addition of components CO2and N2 and pseudo-component C1-C2.

Figure 13—Oil B Pressure-Temperature Diagram

Figure 14—Oil B CCE Results

SPE-189727-MS 13

Figure 15—Oil B Differential Liberation Results

Figure 16—Oil B Density and Viscosity

Table 3—Oil B Characterization in Terms of SARA Fractions

Oil C – 34 °APIThe plus fraction (C30+) of the flashed oil sample was splitted to C92+ in order to honour the mass fractionof asphaltenes, which was 2.7 percent. The extended analysis from C12 to C30+ was used to estimate the

14 SPE-189727-MS

alpha parameter that best fitted the data, which was 0.8808. Fig. 17 illustrates the lumping scheme derived,which was used to characterize the live oil in terms of SARA fractions.

The lumping scheme shown in Fig. 17 was applied to the live oil sample and the laboratory experimentswere matched with an EoS. At a reservoir temperature of 192 °F, the measured saturation pressure was1,580 psia, and the following experiments were available: a CCE, a differential liberation, and a multi-stageseparator test.

The main challenge encountered for the phase behavior modelling of this sample was that it could not bemodelled with only the 4 SARA pseudo-components. Due to the light nature of the fluid, it was necessaryto subdivide the Saturates fraction into two fractions which we called "Lighter Saturates" and "HeavierSaturates".

The resulting phase envelope is illustrated in Fig. 18, which looks as expected from the originalcharacterization. Figs. 19 to 24 show the match of the different experiments. One single EoS was enoughto match the separator API of 34, the GOR of 380 scf/stb, and the formation volume factor of 1.2568.

The resulting characterization of the oil is presented in Table 4. Notice the addition of components CO2and N2 and pseudo-component C1-C2.

Figure 17—Oil C Characterization Scheme in Terms of SARA Fractions

SPE-189727-MS 15

Figure 18—Oil C Pressure-Temperature Diagram

Figure 19—Oil C CCE Results

Figure 20—Oil C Density from CCE

16 SPE-189727-MS

Figure 21—Oil C Diff. Lib. Results

Figure 22—Oil C Density from Differential Liberation

Figure 23—Oil C Viscosity

SPE-189727-MS 17

Figure 24—Oil C Gas Volume Factor

Table 4—Oil C Characterization in Terms of SARA Fractions

Possible Variations of the Workflow1. In all the cases discussed, the Gamma function was used to estimate the molar distribution of the plus

fraction. This was to take advantage of the extended compositional analysis (up to C30+). However,one can use any of the available plus fraction splitting methods18-20 and apply it directly to the plusfraction. The challenge with this approach is that it could result in a discontinuous distribution (i.e. asharp change at the original plus fraction composition), and one would need to select the appropriatemethod and parameters to promote a smooth distribution.

2. In the case studies presented, all the oil samples were characterized up to C30+, and nowadays it iscommon to perform compositional analysis up to C36+ or even more, which would make the workfloweven more robust. However, it is not uncommon to find older reports with oil characterizations upto C6+, C7+ or C12+. For these cases, one needs to perform a trial and error process where the plusfraction splitting/lumping methods18-21 are used until the obtained distribution matched the availableSARA data.

3. If both the SARA composition and physical properties of each SARA fraction (e.g. specific gravityand molecular weight) are available, one could try to perform the phase behavior modelling directlyby using that information. First estimates of the critical properties can be obtained by usingcorrelations28-30.

18 SPE-189727-MS

4. In the odd case the physical properties of the SARA fractions are available (i.e. MW and SG) but notthe composition, one can still perform a trial-and-error process by using different molar distributionsof the dead oil, until the physical properties of each fraction are honoured. The end product would bethe SARA composition which would be used going forward. An example of this approach has beenpresented in the literature24.

Advantages and Disadvantages of the WorkflowTraditionally, kinetic modelling for ISC has been performed in terms of maltenes and asphaltenes31 which,from the phase behaviour point of view, may not always be sufficient for proper EoS modeling. And thatis the advantage this method provides. It allows capturing phase behavior as well as reactive behavior.However, if that approach was enough, the SARA model could be simplified to maltenes and asphalteneswith little difficulty.

Nevertheless, the advantages of using SARA fractions do not come without challenges1. The first obviousone is time and costs associated to SARA fractionation. Depending on the oil, a SARA analysis could takeseveral days or weeks to perform. Moreover, there are different methods for SARA analyses which normallylead to different results, even within the same laboratory. This could cause confusion if one is not familiarwith the experimental procedures32-33.

Another challenge is related to the measurement of the physical properties such as molecular weightand specific gravity of each of the fractions. These properties are required to estimate critical temperature,critical pressure, and acentric factor (of each of the fractions) using correlations available in most PVTpackages, and are not always requested to the laboratory. One of the challenges here is that, given thecomplex nature of the heaviest fractions, the estimation of their physical properties involves a fair degree ofuncertainty and should be taken with caution. Furthermore, most of the correlations for critical propertiesavailable in commercial PVT packages are not suited for heavy fractions and another method might berequired to obtain a first estimate of them28-30. Nevertheless, this is not really a problem as the criticalproperties of those pseudo-components are likely to be used as regression parameters for the tuning of theEoS so only a first estimate of them is required.

ConclusionsA workflow for phase behavior modelling of oils in terms of SARA-based oil characterization has beenpresented, which is fully compatible with traditional cubic equations of state. It is a good alternative to builda fluid model which can capture the basic characteristics of both the phase behaviour and oxidation reactionstaking place during the in-situ combustion process. Furthermore, the case studies presented demonstratethat the workflow is fully applicable to light oils and heavy oils.

This work opens up opportunities to model the ISC process for any oil by utilizing the currently availablekinetic models, which in turn is an important step towards improving the predictability of air-injection-based processes using reservoir simulation.

AcknowledgementsThe authors want to thank the Computer Modelling Group (CMG) for allowing us to use their phase behaviormodelling software (Winprop) for this study. Also, the support of the In Situ Combustion Research Groupand industrial sponsors and partners is gratefully acknowledged.

SPE-189727-MS 19

References1. Gutiérrez, D., Moore, R.G., Ursenbach, M.G., and Mehta, S.A. 2012. The ABCs of In-Situ

Combustion Simulations: From Laboratory Experiments to Field Scale. J Can Pet Technol 51 (4):256-267. SPE-148754-PA. http://doi.org/10.2118/148754-PA.

2. Li, J. 2006. New Insights of Oxidation in Crude Oils. PhD dissertation. University of Calgary,Calgary, Alberta.

3. Gutiérrez, D., Moore, R.G., Mehta, S.A., Ursenbach, M.G., and Skoreyko, F. 2009. TheChallenge of Predicting Field Performance of Air Injection Projects Based on Laboratoryand Numerical Modelling. J Can Pet Technol 48 (4): 23-34. PETSOC-09-04-23-DA. http://doi.org/10.2118/09-04-23-DA.

4. Adegbesan, K.O., Donnelly, J.K., Moore, R.G., and Bennion, D.W. 1987. Low TemperatureOxidation Kinetic Parameters for In-Situ Combustion Numerical Simulation. SPE Res Eng 2 (4):573-582. SPE-12004-PA. http://dx.doi.org/10.2118/12004-PA.

5. Millour, J.P., Moore, R.G., Bennion, D.W., Ursenbach, M.G., and Gie, D.N. 1985. A SimpleImplicit Model for Thermal Cracking of Crude Oils. Paper SPE 14226 presented at the SPEAnnual Technical Conference and Exhibition, Las Vegas, Nevada, USA, 22–26 September. http://dx.doi.org/10.2118/14226-MS.

6. Millour, J.P., Moore, R.G., Bennion, D.W., Ursenbach, M.G., and Gie, D.N. 1987. An ExpandedCompositional Model for Low Temperature Oxidation of Athabasca Bitumen. J Can Pet Technol26 (3): 24–32. JCPT Paper No. 887-03-02. http://dx.doi.org/10.2118/87-03-02.

7. Hayashitani, M., Bennion, D.W., Donnelly, J.K., and Moore, R.G. 1978. Thermal CrackingModels for Athabasca Oil Sands Oil. Paper SPE 7549 presented at the SPE Annual Fall TechnicalConference and Exhibition, Houston, 1–3 October. http://dx.doi.org/10.2118/7549-MS.

8. Freitag, N.P. and Verkoczy, B. 2005. Low-Temperature Oxidation of Oils in Terms of SARAFractions: Why Simple Reaction Models Don’t Work. J Can Pet Technol 44 (2): 54–61.PETSOC-05-03-05. http://doi.org/10.2118/05-03-05.

9. Kok, M.V. and Karacan, C.O. 2000. Behavior and Effect of SARA Fractions of OilDuring Combustion. SPE Res Eval & Eng 3 (5): 380–385. SPE-66021-PA. http://dx.doi.org/10.2118/66021-PA.

10. Li, J., Mehta, S.A., Moore, R.G., Zalewski, E., Ursenbach, M.G., and Van Fraassen, K. 2006.Investigation of the Oxidation Behaviour of Pure Hydrocarbon Components and Crude OilsUtilizing PDSC Thermal Technique. J Can Pet Technol 45 (1): 48–53. PETSOC-06-01-04. http://dx.doi.org/10.2118/06-01-04.

11. Kristensen, M.R., Gerritsen, M.G., Thomsen, P.G., Michelsen, M.L., and Stenby, E.H. 2007.Efficient integration of stiff kinetics with phase change detection for reactive reservoir processes.Transport Porous Media 69 (3): 383–409. http://dx.doi.org/10.1007/s11242-006-9079-y.

12. Kristensen, M., Gerritsen, M., Thomsen, P., Michelsen, M., and Stenby, E. 2009. An Equation-of-State Compositional In-Situ Combustion Model: A Study of Phase Behavior Sensitivity.Transport Porous Media 76 (2): 219–246. http://dx.doi.org/10.1007/s11242-008-9244-6.

13. Lapene, A., Debenest, G., Quintard, M., Castanier, L.M., Gerritsen, M.G., Kovscek, A.R. 2011.Kinetics Oxidation of Heavy Oil. 1. Compositional and Full Equation of State Model. Energ.Fuel. 25 (11): 4886-4895. http://dx.doi.org/10.1021/ef200365y.

14. Lapene, A., Debenest, G., Quintard, M., Castanier, L.M., Gerritsen, M.G., Kovscek, A.R 2015.Kinetics Oxidation of Heavy Oil. 2. Application of Genetic Algorithm for Evaluation of KineticParameters. Energ. Fuel. 29 (2): 1119-1129. http://dx.doi.org/10.1021/ef501392k.

20 SPE-189727-MS

15. Freitag, N.P. and Exelby, D.R. 2006. A SARA Based Model for Simulating PyrolysisReactions That Occur in High Temperature EOR Processes. J Can Pet Technol 45 (3): 38–44.PETSOC-06-03-02. http://doi.org/10.2118/06-03-02.

16. Ren, Y., Freitag, N.P., and Mahinpey, N. 2007. A Simple Kinetic Model for CokeCombustion During an In-Situ Combustion (ISC) Process. J Can Pet Technol 46 (4): 47–53.PETSOC-07-04-05. http://dx.doi.org/10.2118/07-04-05.

17. Sequera, B., Moore, R.G., Mehta, S.A., and Ursenbach, M.G. 2010. Numerical Simulation of In-Situ Combustion Experiments Operated Under Low Temperature Conditions. J Can Pet Technol49 (1): 55–64. SPE-132486-PA. http://dx.doi.org/10.2118/132486-PA.

18. Pedersen, K.S., Thomassen, P., and Fredenslund, Aa. 1983. SRK-EOS calculation for crude oils.Fluid Phase Equilibria 14 : 209-218. https://doi.org/10.1016/0378-3812(83)80127-7

19. Whitson, C.H. 1983. Characterizing Hydrocarbon Plus Fractions. SPE J 23 (4): 683-694. https://doi.org/10.2118/12233-PA

20. Pedersen, K.S., Thomassen, P., and Fredenslund, Aa. 1984. Thermodynamics of petroleummixtures containing heavy hydrocarbons. 1. Phase envelope calculations by use of the Soave-Redlich-Kwong equation of state. Ind. Eng. Chem. Process Des. Dev. 23 (1): 163–170. http://pubs.acs.org/doi/abs/10.1021/i200024a027

21. Danesh, A., Xu, D., Todd, A.C. 1992. A Grouping Method to Optimize Oil Description forCompositional Simulation of Gas-Injection Processes. SPE Res Eng 7 (3): 343-348. SPE-20745-PA. https://doi.org/10.2118/20745-PA

22. Whitson, C.H., Andersen, T.F., Soreide, I. 1989. C7+ Characterization of Related EquilibriumFluids Using the Gamma Distribution. Advances in Thermodynamics, Taylor & Francis, NewYork 1: 35-56.

23. Brulé, M.R. Kumar, K.H., Watansiri, S. 1985. Characterization Methods Improve Phase-BehaviorPredictions. Oil & Gas J 83 (6): 87.

24. Anaya, I., Cruz, R.E.L., Alvarez, A.J., Gutierrez, D., Skoreyko, F.A., and Card, C. 2010.Simulation Study for Designing an In-Situ Combustion Pilot in the Orinoco Belt of Venezuela:From Laboratory Studies to the Field Scale. Paper SPE 137491 presented at the CanadianUnconventional Resources and International Petroleum Conference, Calgary, 19–21 October.http://dx.doi.org/10.2118/137491-MS.

25. Twu, C.H. 1984. An Internally Consistent Correlation for Predicting the Critical Properties andMolecular Weights of Petroleum and Coal-Tar Liquids. Fluid Phase Equilibria 16: 137-150.https://doi.org/10.1016/0378-3812(84)85027-X

26. Riazi, M.R., Daubert, T.E. 1980. Simplify Property Predictions. Hydrocarbon Processing (59) 3:115-116.

27. Kesler, M.G., Lee, B.I. 1976. Improve Predictions of Enthalpy of Fractions. Hydro. Proc. 55:153-158.

28. Nji, G.N., Svrcek, W.Y., Yarranton, H., and Satyro, M.A. 2008. Characterization of HeavyOils and Bitumens. 2. Improving the Prediction of Vapor Pressures for Heavy Hydrocarbons atLow Reduced Temperatures Using the Peng–Robinson Equation of State. Energy Fuels 23 (1):366–373. http://dx.doi.org/10.1021/ef8006855.

29. Nji, G.N., Svrcek, W.Y., Yarranton, H.W., and Satyro, M.A. 2007. Characterization of HeavyOils and Bitumens. 1. Vapor Pressure and Critical Constant Prediction Method for HeavyHydrocarbons. Energy & Fuels 22 (1): 455–462. http://dx.doi.org/10.1021/ef700488b

30. Sabbagh, O., Akbarzadeh, K., Badamchi-Zadeh, A., Svrcek, W.Y., and Yarranton, H.W. 2006.Applying the PR-EoS to Asphaltene Precipitation from n-Alkane Diluted Heavy Oils andBitumens. Energy & Fuels 20 (2): 625–634. http://dx.doi.org/10.1021/ef0502709

SPE-189727-MS 21

31. Belgrave, J.D.M., Moore, R.G., Ursenbach, M.G., and Bennion, D.W. 1993. A ComprehensiveApproach to In-Situ Combustion Modeling. SPE Advanced Technology Series 1 (1): 98–107.SPE-20250-PA. http://dx.doi.org/10.2118/20250-PA.

32. Fan, T. and Buckley, J.S. 2002. Rapid and Accurate SARA Analysis of Medium Gravity CrudeOils. Energy & Fuels 16 (6): 1571–1575. http://dx.doi.org/10.1021/ef0201228.

33. Kharrat, A.M., Zacharia, J., Cherian, V.J., and Anyatonwu, A. 2007. Issues with ComparingSARA Methodologies. Energy & Fuels 21 (6): 3618–3621. http://dx.doi.org/10.1021/ef700393a.