Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 28

Chem. Eng. Comm.

, 194:9951021, 2007 Copyright#Taylor & Francis Group, LLC ISSN: 0098-6445 print/1563-5201 online DOI: 10.1080/00986440701242808

In Situ Combustion in Enhanced Oil Recovery (EOR): A Review


NADER MAHINPEY, APRAMEYA AMBALAE, AND KOOROSH ASGHARI
Faculty of Engineering, University of Regina, Regina, Canada
Enhanced oil recovery (EOR) refers to the technologies developed to increase extraction of crude oil from reservoirs after primary production. In situ combustion (ISC) is one of the methods developed for EOR. This review examines studies done by researchers worldwide to improve our understanding of the mechanism of oil cracking kinetics, which is one of the fundamental mechanisms of in situ combustion. Good agreement between the laboratory and field results has encouraged further research in this field. Extensive research at the laboratory scale to understand the pyrolysis and oxidation behavior of coke formed from medium and light oil and also to propose more realistic models to mimic the true behavior of in situ combustion has been undertaken in recent years. Apart from the classical Arrhenius model, researchers have come up with other models (two-step oxidation model) based on the type of combustion activity observed from their samples, thus modeling the process more accurately. Research work showing optimization of the parameters of ISC and improving the economic viability of the entire process is been one of the main focuses of this article. The review also explains the nature of the various experiments, sheds light on some of the concepts that remain unexplained, and opens the way for fresh thinking in those areas. It also highlights the possibility of developing global solutions for numerical simulation of this EOR process. KeywordsCombustion; Enhanced oil recovery; In situ combustion; Pyrolysis; Reaction kinetics

Introduction
With the decline in oil reserves and problems in locating new oil fields, emphasis has been given to recovering the remaining oil in old reservoirs. The concept of enhanced oil recovery (EOR) was adopted in order to recover such untapped oil from partially depleted reservoirs. The first use of EOR in Canada was a pilot test of a horizontal hydrocarbon miscible flood by Mobil Oil in the North Pembina Cardium Unit (NPCU) in 195758, (Howes, 1978). With EOR gaining impetus in recent years, it has become essential to understand thoroughly the mechanisms involved in these processes. Careful observation and meticulous design go far in achieving the desired results in EOR. To this end, with respect to the thermal methods of EOR, it is necessary to have a clear knowledge of oil cracking kinetics during oil production. In situ combustion is one of the thermal EOR techniques that is becoming more widely and
Address correspondence to Nader Mahinpey, Faculty of Engineering, University of Regina, 3737 Wascana Parkway, Regina, Sask., Canada, S4S 0A2. E-mail: nader.mahinpey@ uregina.ca

995

996

N. Mahinpey et al.

successfully applied, and our understanding of it, while improving, requires both synthesis and further development of the current research. It is necessary to understand the viability of in situ combustion technically as well as economically. In situ combustion is commercially viable but not being widely adopted at the field level. There is a necessity to understand the cracking kinetics of in situ combustion since it is one of the main mechanisms of the combustion process. Results need to be modeled effectively so that its applicability and economic viability can be predicted for any given oil reserve. Furthermore, development in this field should reach a level such that global solutions can predict in situ combustion during the EOR process. This review demonstrates that the fundamental research on in situ combustion has been achieved through contributions made by various researchers worldwide in kinetics and modeling. It also elaborates on shortcomings in the research and highlights further work needed in order to make in situ combustion an optimal and widely adopted technology for EOR. The key focus identified for future research is to achieve generalizability in the kinetics and modeling so that they can be applied to any given field and commercial operation.

In Situ Combustion: An Overview


In situ combustion is a thermal oil recovery technique in which energy is generated by a combustion front that is propagated along the reservoir. In situ combustion takes advantage of an important physical-chemical phenomenon that can take place in reservoirs. In situ in Latin means in place. Th erefore, in situ combustion is simply the burning of fuel where it sits in a reservoir (Prats, 1982). During the oil recovery process, air or any gas containing oxygen is injected into the reservoir. This reacts with the fuel present in the reservoir, producing heat, which ignites the oil in the surrounding area, forming a combustion front. This combustion propagates through the reservoir, thereby helping in recovery of the unburned crude. However, for heavy oils and cooler reservoirs, the injection well needs to be heated by other means to ignite the reservoir. In either case, during this process, crude oil undergoes a series of physical and chemical changes. The carbon-rich residue produced by the thermal cracking and distillation of the residual crude near the combustion front sustains the in situ combustion. In an in situ combustion process, coke, an important solid fuel that is formed in the reservoir, acts as a fuel for the self-sustaining combustion front. The amount of fuel present per bulk volume of the reservoir is an extremely important parameter in combustion operations. It generally determines the air required to burn a unit bulk volume of the reservoir. It is well known that lowtemperature oxidation enhances the amount of fuel available during high-temperature oxidation. The correlation between fuel burned and air consumed is generally excellent for combustion at high temperatures, as shown in Figure 1. Adewusi and Greaves (1991) have reported that enrichment of air with a 35% volume of oxygen increased the overall oil recovery. However, these experimental results would not be directly applicable to field studies. Experiments have shown that when 100%pure oxygen was injected, oil recovery was reduced. Hence, an optimum amount of pure oxygen must be used. With air being one of the process variables, changes in its volume have shown some positive effects in terms of oil recovery through this EOR technique. Varying this parameter decreased the density of the produced oil and also reduced the viscosity up to 60%.

In Situ Combustion in EOR

997

Figure 1.Effect of fuel burned on total air requirement (Prats, 1982).

In situ combustion can be accomplished in either the forward or reverse direction in reference to the movement of the combustion front in the reservoir. In forward combustion, the air flows in the same direction in which the combustion front advances, whereas in the reverse combustion process, the airflow is in a countercurrent direction to the movement of the combustion front. The most frequently used method is forward combustion; reverse combustion is characterized by spontaneous ignition in the reservoir, which causes instability, making it unsuitable as an EOR technique. Nevertheless, both these methods are operable in two modes dry and wet. In the dry mode, dry air alone is injected. In wet combustion, a mixture of air and water is passed through the combustion front. The water becomes vaporized in the hot, burned-out zone surrounding the combustion front. In situ combustion exhibits reactions in two temperature regimes, low temperature and high temperature. However, the temperature range for both regimes is different for various samples. Detailed experiments conducted in these two temperature regimes will help in arriving at more reliable inputs for numerical analysis of the EOR process. The effect of process variables on in situ combustion was studied as early as 1958 (Martin et al.). More recently, a set of design considerations for the in situ combustion process was proposed by Moore et al. (1997). They indicate that the performance of a combustion process would be successful if important criteria such as fuel requirement, equivalent fuel saturation, equivalent mobile oil saturation, equivalent mobile water saturation (wet combustion), burn volume, oil and water displaced from burn volume, water=oil ratio (average), oil and water displacement ratio (average or per cycle), and injected air=oil ratio are calculated well in advance.

Chemistry of In Situ Combustion


Although extensive research has been done at the application level, it is equally important to understand the underlying chemistry of the phenomenon of in situ combustion so that the kinetics can be delineated and modeled. The combustion front propagation and air requirements are controlled by the exothermic oxidation reactions (Burger and Sahuquet, 1971). Moreover, the peak temperature (the

998

N. Mahinpey et al.

temperature at which the rate of the reaction is maximum) is related to the heat released by oxidation and combustion reactions. This elicits a need to quantitatively estimate the different parameters that relate to the chemistry of the process. The complete oxidation of a hydrocarbon- or oxygen-containing organic compound will produce CO 2 and H 2 O as final products with an appreciable release of heat. Some of the prime parameters to be considered while modeling the process are the stoichiometry of the reactions, heat release, heat conductivity, and heat transfer coefficients, since they correspond so strongly to the chemical rates (Benson, 1982). Kinetics and chemistry of cracking hydrocarbons was described by Rosen (1941). From his experiments, he proposed that during the cracking of olefins, two main reactions, namely polymerization and decomposition, occur. However, he proposed that the kinetics of all reactions could not be studied individually, so the kinetics of the total reaction should be considered. Second-order kinetics was observed for the cracking of olefins under elevated pressure and at a temperature of 500_C. However, at temperatures of 600_C and higher, first-order kinetics was observed. He also indicated that the energy of activation is not affected by the molecular weight of the cracking compound. It should be noted that the kinetics obtained in these reactions were from a vapor state and would be expected to differ significantly from what might occur during the in situ combustion process in a reservoir, where the hydrocarbons exist in a liquid or a solid state. Details of the chemistry of an in situ combustion system, describing the general and theoretical considerations of heat of reactions and kinetics, were investigated by Burger and Sahuquet (1971). They discussed the reaction products that occur as a result of in situ combustion. These products are an outcome of different chemical reactions, which include, but are not limited to, complete combustion, incomplete combustion, and oxidation to carboxylic acids, aldehyde, ketone, alcohol, and hydroperoxide. These reactions contribute to the destruction of hydrocarbon chains during high-temperature oxidation and the linking of oxygen atoms to the hydrocarbon molecule through low-temperature oxidation reactions. An empirical formula to define the stoichiometry of the combustion process was given as follows: CHx f2b=2 1b x=4gO 2! f1=1bgCO 2 fb=1bgCOx=2H 2 O 1 where xH=C ratio of the fuel burned andbvolume ratio of CO=CO 2 in the exhaust gases. This formula is based on the assumption that combustion takes place at higher temperatures. In our group, we use this type of equation but interpret burning characteristics based on the intensity of the combustion parameters, e.g., temperature and pressure. Their empirical reaction was based on the assumptions that only incomplete combustion occurs in the high-temperature zone, that the fuel burned contains only carbon and hydrogen, and that the water produced in the above reaction is condensed. However, they argue that knowledge of the enthalpy of formation of the fuel burned is needed for exact calculations. This parameter is unknown since the chemical nature of the fuel is usually poorly defined. Studies from pyrolysis and oxidation reactions on linear and branched alkanes (Ranzi et al., 1997) showed how the kinetic modeling of oxidation reactions could improve the understanding of hydrocarbon mixture ignition and combustion phenomena. This concept can be extended to in situ combustion. They stated that

In Situ Combustion in EOR the combustion process exhibited a more complex variety of reactions due to the thermal and kinetic interactions of numerous radical and molecular intermediaries through a broad kinetic mechanism. Based on their experiments, they concluded that only two primary classes of propagation reactions needed to be considered in pyrolysis, namely, isomerization of alkyl radicals R andb -decomposition of alkyl radicals R. For oxidation reactions, the same reactions as in pyrolysis were considered coupled with oxidation sub-mechanisms, which primarily depended on the kind of fuel being consumed. The authors have indicated that although the dimension and complexity of such a system is enormous, only a limited set of kinetic parameters and simplifying assumptions could be made in order to describe, in detail, the hydrocarbon pyrolysis, partial oxidation, and combustion.

999

Oxidation and Pyrolysis Reactions in In Situ Combustion


After discussion of the chemistry of in situ combustion, we now examine the two prime reactions that take place in the in situ combustion process, namely oxidation and pyrolysis reactions. Many researchers have carried out experiments to simulate the combustion and pyrolysis reactions occurring in a reservoir. Experiments have been conducted to evaluate the coke (fuel) formation and its decomposition process. Simulated lab-scale tubular reactors, thermogravimetric analyzers (TGA), and other instruments have been used by researchers to evaluate the kinetic parameters and to model the system, thereby providing inputs for the numerical simulation of the EOR process. First, it is important to note that crude oil can be represented as a combination of saturates, aromatics, resins and asphaltenes (SARA), which form the major portion of the oil, and components such as paraffin. Separation of oil into less complex, more chemically representative constituents (SARA fractions) offers a more meaningful kinetic study of oxidation and thermal cracking reactions than does the whole oil. Experiments to evaluate the combustion behavior of coke formed from Neilburg oil were conducted by Ambalae et al. (2006). Isothermal combustion of coke formed from Neilburg oil at different temperatures is plotted in Figure 2. Kinetics of these

Figure 2.Isothermal combustion results for whole oil.

1000

N. Mahinpey et al.

reactions was also evaluated. Experiments were conducted in the temperature range of 375_to 500_C at intervals of 25_C. Experiments on heavy oil cores, oils, and SARA fractions under thermal stress (Verkoczy, 1993) showed that the significant coke=residue contributors are the asphaltene and resin fractions, while small contributions from aromatic fractions and none from saturate fractions were observed. Asphaltenes, one of the important fractions of any crude oil, has high density and molecular weight. Asphaltenes are dark brown friable solids that have no definite melting point and, when heated, usually intumesce and decompose, leaving carbonaceous residue. Asphaltenes are generally precipitated by solvents, and their elemental compositions vary with the type of solvent used. Elemental compositions of asphaltenes from crudes in different parts of the world are given in Table I. A model was proposed (Fassihi et al., 1984) to analyze and differentiate between crude oil=oxygen reactions at different temperatures. The results of this analysis, along with correlations of apparent hydrogen=carbon (H=C) ratio and molar carbon dioxide=carbon monoxide (CO 2=CO) ratio, indicated three major reactions at different temperatures. Low-temperature oxidation (LTO) appears to occur between the gas and liquid phases. Middle-temperature fuel deposition reactions appear to be homogeneous. The latter was found to be the rate-determining step in clean sands. An apparatus was designed to study the reaction kinetics of oil oxidation in porous media. The experimental data were analyzed using a kinetic model. The overall oxidation mechanism of crude oil in porous medium at different temperatures is an overlap of several consecutive reactions, of which three can be distinguished. LTO reactions are heterogeneous (gas=liquid), which may spark spontaneous ignition. Middle-temperature reactions are caused mainly by the oxidation of the products of distillation and pyrolysis. These reactions leave a heavy oil residue on the solid matrix. High-temperature reactions were heterogeneous and are a result of the combustion of the deposited residue. To summarize the results: 1) These reactions were found to be kinetically controlled, and diffusion effects were minimal. 2) The combustion reaction is slower than fuel deposition in a clean sand matrix. Metallic additives and clays have catalytic effects on the reactions. Clays and fine sands enhance deposition of more fuel because of the absorption characteristics on a higher surface area. These additives may shift the rate-determining step from fuel combustion toward middle-temperature reactions. Oxidation reaction of coke obtained from the Neilburg oil was modeled by Ren et al. (submitted). A two-coke model was proposed. Coke samples were prepared at four different temperatures between 425_and 600_C by fully cracking heavy oil from the Neilburg field under a nitrogen atmosphere. Subsequently, the coke samples were subjected to two types of combustion tests. First, non-isothermal oxidation runs were carried out to investigate the general effects of coking temperature and the source of the coke. Second, isothermal oxidation runs were performed for the determination of kinetic parameters for the conventional one-coke oxidation reaction model. Comparison of the two-coke model with conventional Arrhenius model revealed that the two-coke oxidation reaction model fitted the experimental data more accurately.

In Situ Combustion in EOR Table I.Elemental composition of asphaltenes Composition (weight percent) Atomic Ratios Source (Country) C H N O S H=C N=C O=C Canada 79.0 8.0 1.0 3.9 8.1 1.21 0.011 0.037 79.5 8.0 1.2 3.8 7.5 1.21 0.013 0.036 88.5 8.2 1.6 1.4 0.3 1.11 0.015 0.012 86.8 10.2 1.3 1.1 0.6 1.41 0.013 0.010 85.1 11.1 0.7 2.5 0.6 1.56 0.007 0.022 81.9 8.1 1.2 1.0 7.8 1.19 0.012 0.009 82.2 8.2 1.6 0.4 7.6 1.19 0.017 0.004 80.4 7.8 2.6 2.0 7.2 1.17 0.028 0.019 82.7 7.8 2.8 1.0 5.8 1.12 0.029 0.009 88.7 8.5 0.7 0.5 1.7 1.15 0.007 0.004 82.8 7.5 2.1 1.6 6.0 1.09 0.022 0.015 83.3 7.8 1.5 1.9 5.6 1.12 0.015 0.017 82.3 7.7 2.3 1.3 6.4 1.13 0.024 0.012 82.6 8.2 1.4 1.6 6.2 1.19 0.014 0.014 82.9 8.2 1.5 0.5 6.8 1.19 0.015 0.005 84.4 7.9 1.6 1.1 5.0 1.12 0.017 0.009 84.8 6.9 1.8 0.9 5.5 0.98 0.018 0.008 85.3 7.4 1.9 0.8 4.6 1.05 0.019 0.007 87.1 9.5 1.0 1.7 0.8 1.30 0.010 0.015 84.9 8.7 1.4 3.7 1.3 1.23 0.014 0.033 85.7 7.8 1.4 1.3 5.8 1.09 0.014 0.011 84.5 7.5 2.4 1.2 4.5 1.06 0.024 0.011 87.9 7.6 2.2 1.8 0.5 1.04 0.022 0.015 86.5 7.7 1.3 3.1 1.4 1.07 0.013 0.026 86.4 9.0 1.7 2.1 0.8 1.25 0.016 0.018 83.7 7.8 1.7 1.0 5.8 Iran 83.8 7.5 1.4 2.3 5.0 Iraq 80.6 80.9 78.3 81.7 82.2 81.8 81.6 82.1 81.6 82.1 82.4 81.4 81.7 81.3 7.7 7.5 7.9 7.9 8.0 8.1 8.0 8.1 8.1 8.0 7.8 8.0 8.8 8.5 0.8 2.6 0.7 0.8 1.7 0.9 0.8 0.6 1.0 1.7 0.9 0.6 1.5 trace 0.3 2.6 4.5 1.1 0.6 1.7 1.8 1.3 1.5 0.6 1.5 1.7 1.8 4.9 9.7 9.0 8.6 8.5 7.6 7.5 7.8 8.0 7.8 7.6 7.4 8.3 6.3 5.2

1001

S=C 0.038 0.035 0.001 0.003 0.003 0.036 0.035 0.034 0.026 0.007 0.025 0.025 0.029 0.028 0.031 0.022 0.024 0.020 0.003 0.006 0.025 0.020 0.002 0.006 0.003 0.017 0.009 0.026 0.014 0.021 0.022 0.009 0.003 0.045 0.042 0.008 0.043 0.041 0.008 0.010 0.039 0.017 0.005 0.035 0.009 0.016 0.034 0.008 0.017 0.036 0.006 0.012 0.037 0.011 0.014 0.036 0.017 0.005 0.035 0.009 0.014 0.034 0.006 0.016 0.038 0.016 0.017 0.029 0.045 0.024
(Continued)

1.19 1.07 1.15 1.11 1.21 1.16 1.17 1.18 1.18 1.18 1.19 1.17 1.14 1.18 1.29 1.25

Kuwait

Mexico Sicily

1002 Table 1.Continued

N. Mahinpey et al.

Composition (weight percent) Atomic Ratios Source (Country) C H N O S H=C N=C O=C S=C 78.0 8.8 trace 3.0 10.2 1.35 0.033 0.049 78.9 7.8 trace 3.1 10.3 1.19 0.029 0.049 U.S.A. 84.5 7.4 0.8 1.7 5.6 1.05 0.008 0.015 0.025 88.2 8.1 1.7 1.3 0.6 1.10 0.017 0.011 0.003 82.9 8.9 2.3 6.5 1.29 0.024 0.029 88.6 7.4 0.8 2.7 0.5 1.00 0.008 0.023 0.002 84.2 7.6 0.8 1.6 5.8 1.08 0.008 0.014 0.026 84.0 7.9 1.9 4.1 2.1 1.13 0.019 0.037 0.009 83.2 8.3 2.3 4.8 1.4 1.20 0.024 0.043 0.006 85.5 8.1 3.3 1.8 1.3 1.14 0.033 0.016 0.006 84.2 7.9 2.0 1.6 4.5 1.13 0.020 0.014 0.020 83.5 8.3 1.0 1.5 2.7 1.19 0.010 0.013 0.012 84.7 8.0 0.9 1.0 5.5 1.13 0.009 0.009 0.024 84.0 7.9 2.0 1.6 4.5 1.13 0.020 0.014 0.020 81.1 7.8 0.2 4.2 6.7 1.15 0.002 0.039 0.031 81.2 7.9 2.0 2.0 6.9 1.17 0.021 0.018 0.032

Venezuela

Experiments to determine the oxidation of heavy oils and their SARA fractions were conducted using thermogravimetric analysis and combustion tube tests in the LTO region (Verkoczy and Freitag, 1997). The TGA profiles indicated that the oxygen uptake regions of different samples (three test samples) were different. An important observation they made in the temperature range of 350_475_C was that there was no negative coefficient gradient (NTG), which was in contrast to what was reported in the literature. However, more experiments in this area had to be conducted to ascertain the exact reason. Further, saturates and aromatics exhibited combustion in the cool-flame regions of 250_300_C and 300_350_C, respectively. For resins, a slow combustion reaction in the range of 300 _400_C was observed. Combustion of asphaltenes was observed only between 350 _and 500_C. The fuel composition was the largest in this fraction. The LTO tests conducted in the combustion tube showed that the SARA fractions together produced more residue, thus making more fuel available for combustion under a low oxygen concentration. Asphaltene fractions were more reactive than the others in the LTO test. Analysis of the whole oil could not give such details as those obtained from the SARA fractions. Hence, it was evident that the analysis of oil by separating it into SARA fractions was more beneficial. The authors also concluded that if the LTO fractions were treated separately from the un-oxidized fractions, the resulting SARA-based framework could form a good basis for representing oxidation reactions in a chemical or simulation model of in situ combustion. Combustion tube experiments showed the cause and effect of low-temperature oxidation (LTO) and high-temperature oxidation (HTO) for heavy crude oils (Mamora and Brigham, 1995). LTO occurred due to the low fuel concentration, which resulted in (i) a reaction-front temperature of only 350_C, compared to

In Situ Combustion in EOR 500_C for HTO, and (ii) oxygen moving ahead of the reaction front and oxidizing the crude oil. Consequently, an oxygenated hydrocarbon fuel was formed. During HTO, practically all of the oxygen injected was consumed so that the fuel was not oxygenated. As a result of distillation, viscosity and specific gravity of the produced oil during HTO decreased significantly. High combustion temperatures were obtained when clay or sand fines were present. These particles reduced the permeability of the sand pack and also provided a large reaction surface area. Consequently, the residual oil saturation, and therefore the fuel concentration, increased, resulting in high-temperature burns. Practically all the injected oxygen was consumed at the combustion zone. Hence, hydrocarbons ahead of the combustion front did not undergo low-temperature oxidation. This resulted in atomic H=C ratios that were similar to those of the original crude oils. In contrast, low-temperature oxidation occurred when the sample matrix contained only 2030 mesh sand. Ignition was not obtained because of the low fuel concentration, and a significant amount of oxygen moved ahead of the combustion zone. This resulted in low-temperature oxidation of the crude oil to form an oxygenated hydrocarbon. Low-temperature oxidation was found to be a very inefficient process, from the standpoint of both oxygen usage and heat generated. They also suggested that, before starting a combustion project, it is essential to run combustion tube experiments using oil and core samples and fluid saturations representative of the field to ascertain that low-temperature oxidation will not occur. A model for the oxidation process of Australian light crude (Kisler and Shallcross, 1997) and the related gas analysis revealed that oxidation behavior was substantially different from that obtained for heavy crude samples. This prompted the authors to propose an improved mathematical model to allow the modeling of the kinetics obtained from their experiments. This took into consideration three overlapping reactions occurring in the low-temperature, medium- or intermediatetemperature, and high-temperature oxidation phases. The reaction rates of these temperature regimes were assumed to be functions of partial pressure of oxygen and fuel concentration, given by the expression R tK t Po mC tn
2 1

1003

where Rreaction rate (min), Kreaction constant, Po 2partial pressure of oxygen (Pa), Ctfuel concentration (mg=unit volume), and m and nreaction rate with respect to partial pressure of oxygen and fuel concentration, respectively. Experimental values showed that oxygen consumption in the light crude had three peaks, at 190_C, 270_C, and 400_C, while heavy crude had two peaks, at 250_C and 380_C. In the case of light crude, saturates underwent oxidation at a low temperature and, hence, displayed peaks at 190 _C and 270_C. Further, the heavy components burned, producing the third peak. However, in heavy crude, saturates are suppressed by the presence of resins and asphaltenes, and, hence, their oxidation is not seen as a significant peak. As a consequence, only two peaks were observed. With this knowledge, further experiments were conducted to investigate the reproducibility and the accuracy of the modeling techniques used, and these were found to be in excellent agreement over a majority of temperature ranges. Experiments to evaluate the influence of the reservoir rock on the crude oil pyrolysis and combustion were conducted by Ranjbar (1993). Previous studies conducted by the same author suggested that the controlling mechanism of fuel formation is the conversion of crude oil components by radical polymerization. It was

1004

N. Mahinpey et al.

observed that, independent of the oil sample, the amount of fuel decreased with an increase in pyrolysis temperature. However, from the experimental results, it was found that the clay minerals in the matrix enhanced fuel deposition during the pyrolysis process and also catalyzed the oxidation of fuel. With the increasing clay content, the activation energy of the combustion process decreased. The author attributed this to the high surface area of clay fractions in the matrix and their catalytic activity. Oxidation behavior of SARA fractions would at most times answer questions about the feasibility of air injection into oil reservoirs and the extent of reactions in the whole oil samples and with the individual fraction and the differences in their kinetic parameters, allowing modeling of the system. Researchers aimed at replacing the conventional two pseudo-component representations by a SARA representation, with the reactions associated together and the conventional oxidation reactions. Simulation of in situ combustion and the air injection process in North Sea light oil extraction was the main focus of the study reported by Kisler and Shallcross (1997). Reactor experiments were performed at a temperature of 480_C and a pressure of 10 MPa. Oxidation reactions in the crude oil showed that three peaks were observed in the plot and that the oxygen consumption was rapid at the lowtemperature region. This supports the work done on Australian light crude (Kisler and Shallcross, 1997). At higher temperatures, it was observed that the oxygen consumption was stoichiometrically equivalent to the carbon oxides produced, indicating that all the oxygen consumed was utilized in producing carbon oxides. This cannot be the case if any hydrogen exists in the fraction undergoing high-temperature combustion, as water is a product of high-temperature combustion. The production of CO 2 increased the oil displacement and had no negative effects on the oil properties. Consumption of oxygen was 23%of the total oxygen injected in the LTO region and was 17%in the HTO region. This low level of oxygen utility in the HTO was mainly attributed to the limited reactions taking place due to the presence of low fuel deposition or due to the high mobility of the crude. This was the general trend found in all samples. For the isolated SARA fractions, up to around 177_C, no appreciable amount of oxygen was consumed and no carbon oxides were produced. However, at 197_C and above, reactions indicating early oxidation were observed. Resins and aromatic fractions showed reduced oxidation peaks at the lower temperature region and relatively high oxidation peaks at the higher temperature region. This indicated that the oxidation kinetics of those fractions could be grouped together. It was concluded that above 477 _C, the whole oil plot was well represented by the SARA fractions. The major contribution for oxygen consumption and fuel deposition was from saturates, indicating high reactivity at relatively low temperature. Air injection into deep light oil reservoirs is a potential technique for large-scale improved oil recovery. The feasibility of an air injection process relies on complete consumption of oxygen from the injected air to achieve a nitrogen flood. The oxidation kinetics of light crude oil at typical reservoir temperatures (90 _140_C) and high pressures was investigated by Ren et al. (1999). The oxygen consumption rate was measured from the reduction in the oxygen partial pressure using a small batch reactor (SBR). Measured pressure data for different crude oils were used to establish a simple reaction rate model of acceptable accuracy for reservoir simulation. Low-temperature oxidation (LTO) studies conducted on four North Sea light crude oils showed that the main reaction products obtained at reservoir

In Situ Combustion in EOR temperatures and after a sufficiently long reaction time were carbon dioxide and water. The reaction mechanism proposed was such that certain hydrocarbon species, especially saturated paraffinic compounds, were oxidized to form intermediate compounds. Four combustion tube tests were performed at a high initial water saturation using Bath Universitys High Pressure Combustion Tube Facility (Greaves et al., 2000). Two tests were conducted on Clair medium heavy oil (19.8 _API) at 75 and 100 bar pressure, with initial oil saturations of 48%and 60%at 80_C initial bed temperature. Maximum combustion temperatures exceeded 600 _C during the early period, settling down to around 400 _C. The combusted zone extended over about 30%of the sand pack length. Oil recovery was mainly affected by the large steam flood generated ahead of the combustion front, due to vaporization of the original water in place, reducing the oil residual down to 21%. The thermal cracking reactions taking place ahead of the combustion front converted part of the residual oil to lighter components, which were displaced with the gas flow, at the same time producing about 10%coke (fuel) for the combustion process. Their observations were: 1) The medium-heavy Clair oil combustion tube tests achieved an average combustion front temperature greater than 400_C. Sufficient fuel was available to consume the injected oxygen with high oxygen utilization, creating a stable combustion process. The in situ combustion was similar to a normal wet combustion, characterized by the existence of a steam plateau region. A high oil recovery of 83%OIP (oil in place) was achieved in one test for a stable combustion propagation of 64%of the sand pack length. 2) Thermal cracking reactions play a very significant role in medium-heavy (and heavy) oil recovery, not only to provide the fuel (coke) to sustain in situ combustion, but also to achieve a degree of upgrading for the produced oil. Upgrading of the light oil was achieved by distillation and gas stripping. Experiments were conducted to study the effects of various additives on the oxidation kinetics of California and Venezuelan oils (Shallcross et al., 1991). Aqueous solutions of 10 metallic salts were mixed with sand and oil from the Huntington Beach, Calif. The mixtures were subjected to a constant flow of air and a linear heating schedule while the effluent gases were analyzed for composition. The variation in the oxygen consumption was analyzed with a model of three competing oxidation reactions. Values of the important kinetic parameters for the three reactions were obtained for each additive. Iron and tin salts were found to enhance fuel formation, while copper, nickel, and cadmium salts had no significant effects. Other experiments with a heavy Venezuelan oil showed that, contrary to earlier suggestions, the use of a ketal did not decrease fuel formation. To study the reaction kinetics of oil, a cell was charged with a mixture of sand, oil, and water or an aqueous solution containing an additive. The cell was then heated at a constant rate of temperature increase while air was passed through the sample at a controlled flow rate and pressure. The effluent gas was then analyzed for its oxygen, CO, and CO 2 content. Oxygen consumption during the in situ combustion process may be modeled adequately considering three competing oxidation reactions. The following conclusions were drawn:

1005

1006

N. Mahinpey et al.

1) An analytical method was developed to estimate the kinetic parameters for those three reactions. 2) The order of the oxidation rate equations with respect to the hydrocarbon fuel concentration generally is less than unity. 3) The presence of iron, tin, and aluminum enhanced fuel deposition for Huntington Beach oil. The presence of copper, nickel, and cadmium had little or no effect. 4) The presence of a ketal did not reduce the amount of fuel deposited by the Venezuelan oil. One of the main objectives of employing in situ combustion for EOR is to reduce the viscosity of the oil in place by increasing its temperature and enhancing mobility. The geochemical alterations that take place in the reservoir during thermal cracking of the oil were studied by Schaffie and Ranjbar (2000). This would shed light on the mechanism of the oil-cracking phenomenon and provide more insight into the physical and chemical changes that take place during the cracking of oil. This knowledge will further help in choosing the right conditions under which the thermal recovery of oil can be employed. They showed that the rate of reaction was controlled by many factors, such as rock lithology, metal compounds, and clay minerals in reservoir rock, chemical composition of the oil, and process conditions. Experiments revealed that the clay minerals and derivatives of the heavy metals had a significant influence in the LTO region. Their presence accelerated the formation of resins and polymerization of those resins asphaltenes. The authors concluded that the mech anism involved in the thermal recovery process provided better material characterization of various crude oils in numerical models describing pyrolysis and oxidation behavior. Elaborate experiments to study the mechanism involved in oil production by underground combustion showed that the temperature distribution in the combustion tube assumed the shape of a heat wave with a steep front (downstream) and a gradual decline (upstream) (Tadema, 1959). This was further investigated by the Humble Oil and Refining Company at Houston, Texas, and Humbles results were found to be in close agreement with those obtained by Tadema. It was seen that the speed of the heat wave was influenced not only by the oxygen supply rate but also by the amount of fuel present in the combustion zone. Coke formation is one of the basic steps involved during the pyrolysis of fuels. Jess (1996) conducted tubular flow reactor experiments and modeled the thermal conversion of aromatic hydrocarbons in the presence of hydrogen and steam. Naphthalene, toluene, and benzene were used as model compounds. A pressure of 160 kPa and temperatures up to 1400_C were used in the experiment. Although the presence of these components in the crude was limited, their influence was significant. Results showed that steam had little or no influence on the reaction rates. Hydrogen inhibited the conversion of benzene and naphthalene. For toluene, hydrogen increased the rate. It was also found that benzene was the key component of thermal decomposition of aromatic hydrocarbons. A study involving the cracking of asphaltenes in the presence of an external catalyst was conducted by Bayambajav and Ohtsuka (2003). They used iron catalyst with the asphaltene obtained from petroleum, and the experiments were conducted in a fixed bed reactor at 300_C in an inert atmosphere (helium). The asphaltene conversion increased linearly with an increase in the average pore diameter of the catalyst (up to 12 nm) and reached approximately 65%. However, the maltene yield was not

In Situ Combustion in EOR affected by the presence of the catalyst, and the conversion was approximately 15%. Analysis revealed that the chemical structure of the asphaltenes recovered after the conversion was different from that of the feed. Thermal methods of analysis include those techniques in which some physical parameter of the system is measured as a function of temperature. This method is best employed only when the measured property changes significantly with temperature so that it can yield valuable information from the overall chemical reaction. There are many thermal methods of analysis, of which one of the most important is TGA. Thermogravimetric analysis provides weight loss data with respect to time or temperature. These data can be used to determine the kinetics of the reaction, and these kinetics can then be used as inputs for the numerical simulation of the EOR process. Thermogravimetric analysis was performed by Karacan and Kok (1997) on two crude samples (medium and heavy). Investigating the SARA fractions individually helped in estimating the temperature ranges at which phenomena like evaporation, oxidation, and combustion effects for each fraction were taking place. It was observed that LTO reactions, which were heterogeneous in nature, produced partially oxygenated compounds such as aldehydes, ketones, and alcohols. This was followed by fuel deposition in the medium-temperature oxidation (MTO) region. MTO reactions, being homogeneous, were involved in the oxidation of products of pyrolysis. In the HTO region, the fuel formed became oxidized in the range of 490 _580_C. In both studies, the asphaltene fractions were the most resistant element. However, this is not true in all the cases, but occurred mainly because of their high molecular weight compared to other fractions. Almost no weight loss was observed for asphaltenes during distillation and very little during LTO, with a value of 1.2%for the Garzan sample and 2.1%for the B. Raman sample. This small weight loss was attributed to the vis-breaking and pyrolysis of asphaltenes in that temperature region. However, for saturates, weight loss was heavy in the LTO. Due to the low energy requirement for saturates, it can be treated as the readily available fraction for triggering the combustion of the whole oil. Aromatics and resins behaved similarly in terms of temperature interval and weight loss. These experiments also supported the hypothesis that the resin formation was due to the reaction between the aromatics and oxygen. Similar experiments were conducted by Ren et al. (2005) using Neilburg oil. Combustion reactions showed LTO in the range of 270 _370 C. Above this, reactions were observed to be in the HTO region. This agreed well with results obtained by other researchers. The different reactions taking place in different temperature zones were tabulated as shown in Table II.

1007

Table II.Thermo-oxidative behavior of Neilburg oil Behavior Temperature Evaporation, distillation, LTO 40_270_C Distillation, vis-breaking, LTO 270_370_C Thermal cracking, combustion 370_560_C Coking, combustion, mineral decomposition 560_750_C

1008

N. Mahinpey et al.

TGA and gas chromatographic analysis on Saudi Arabian light (AL) and heavy oil (AH) conducted by Ali and Saleem (1991) showed methane was the dominant gas evolved in all their experiments. A similar observation was made by Ekwenchi et al. (1984). However, from the decane asphaltene, Ali and Saleem observed that the dominant gas evolved was from C 2 -C 4 . Thermogravimetric and differential thermal analysis (DTA) curves showed that for all the asphaltene samples studied, decomposition was observed below 400_C and reached a maximum rate at 500_C. With respect to the weight loss, the entire process was complete above 600 _C. Experiments were conducted on samples up to 920_C. The TGA experiments showed that the conversion of asphaltenes into volatile matter was greater for the light crude sample at low temperatures and greater for the heavy crude sample at higher temperatures. In the case of asphaltenes precipitated from higher alkane solvents, ethane was the dominant gas evolved. However, in the entire asphaltene samples, comparison of the gas evolved at 520_C showed that methane and hydrogen constituted the major portions. Once the preliminary chemistry and kinetics had been established and models were developed, researchers began field testing to determine the generalizability of results and to determine the most effective applications for modeling and predicting in situ combustion in the field. Akin et al. (2002) modeled the in situ combustion system by simulating different well configurations. The combustion system was operated at a temperature of 500_C. Raman and Kozluca crudes were used in their experiments with the application of different well configurations. Peak temperatures observed in Raman samples were lower than those in Kozluca, although the former represented a stabilized combustion front. The average fuel consumption in the vertical-vertical configuration was highest. For the Raman samples, the total oil recovery was 55%of the oil in place, whereas for the Kozluca samples, it was 45%. Numerical simulation was performed and equilibriumk-values were established. However, numerical simulation underestimated the experimental oil recovery values, especially in terms of the estimated pressure after ignition. To improve the numerical simulation, a kinetic model obtained from the TGA was incorporated. Although improvements modified the predicted results at lab scales, the model could not be applied directly to field-scale simulations. During processing of bitumen to form coke, researchers have encountered technical difficulties and have, as a result, attempted to understand the mechanism of formation, which would enable them to troubleshoot such problems. Banerjee et al. (1986) used four different types of heavy crude for experiments to understand the kinetics of coke formation. Their results showed that coke was formed at a temperature of 400_C, except for aromatics, where it occurred at 465 _C. A first-order rate constant was observed for all the reactions, with the asphaltenes having the highest rate constant and saturates having the lowest. They concluded that aromaticity plays an important role in determining the rate of coke formation. The values of the Arrhenius parameters were greater for aromatics, which suggest that coke formation varies with temperature more strongly for the aromatics than for the other fractions. It was also observed that the saturated hydrocarbons form coke very slowly. A tentative reaction that might occur during coke formation is shown in Figure 3. A SARA-based model to simulate the pyrolysis reactions occurring at high temperature were conducted (Freitag and Exelby, 2002). Experiments were conducted in a tubular flow reactor. Maximum reactor conditions were set at 600 _C and 2800 kPa.

In Situ Combustion in EOR

1009

Figure 3.Reaction during coke formation (Banerjee et al., 1986).

This experiment took into consideration the pressure factor, thereby incorporating realistic conditions that exist in the reservoirs. Reaction rate constants for Neilburg and Wolf Lake SARA fractions were obtained. Reactions showed a first-order rate. Arrhenius parameters were evaluated from these plots. It was also observed that most plots had almost similar slopes, indicating that these reactions could be well associated with each other and, hence, could be generalized for a certain set of samples from different reservoirs. Analysis showed that the coke formation from asphaltenes was in the range of 340_400_C, and hydrocarbon ratios associated with this reaction varied from 0.79 to 1. Simulation tests of in situ combustion performed in an idealized reservoir condition would go far in achieving accurate end results. Reactor experiments on oil samples, with a focus on establishing the relationship between the colloidal composition of the oil and fuel formation as a function of pyrolysis temperature, were performed by Ranjbar (1992). Fuel concentration comparison of these oils showed that at 350_C, they were dependent on the total content of the colloids in the oil. Maximum pyrolysis for resins and asphaltenes was observed at 420 _C and 480_C, respectively. A general scheme of the formation of coke was given by Burger and Sahuquet (1971):

1010

N. Mahinpey et al.

These authors concluded that the previous assumptions that the asphaltenes are the only precursors of fuel formation are not entirely accurate. Their investigations showed that the total colloidal content of the oil has a pronounced influence over the fuel formation and composition up to a temperature of 450 _C. In their experiments, they also compared CO 2 and steam instead of N 2 as a pyrolysis medium and showed that N 2 was still the best in a temperature range of 350_550_C. Isothermal combustion tests and TGA showed that the activation energies determined were independent of the medium. Up to 450_C, fuel formation from samples rich in resin exhibited a higher reactivity towards oxygen than the fuel formed from those rich in asphaltenes. However at a temperature of 550_C, the trend was reversed. A clear knowledge of oil cracking kinetics is necessary for attempting to model the reactions occurring in an in situ combustion process. Experiments to model the kinetics of crude oil, describing the evolution of heavier components of oil and coke, were performed by Behar et al. (1988). Pyrolysis reactions in a sealed gold tube filled with argon were reported at various temperatures and heating times. Boscan (aromatic) and Pematang (paraffin) oils were used in their studies. Coke formation was attributed mainly to the early poly-condensation reactions of asphaltenes and resins and to conversion of the intermediate fractions formed during the oil cracking. The reaction parameter estimation suggested that the activation energies of these reactions were in the same range for those found for either most types of organic matter or Athabasca oil sand. The kinetic parameters were determined at a temperature of 450_C. Estimations showed that coke formation and cracking rate were higher for Boscan oil than for Pematang oil, which suggested that a coking temperature of 450_C may be applicable for heavy oils, but for light oils like the Pematang oil, coking can occur at lower temperatures. Thermal cracking reaction models as inputs to numerical simulators by simulating the thermal recovery for the Athabasca oil sands were provided by Hayashitani et al. (1978). Reactor experiments were conducted in the temperature range of 303_ 452_C in a stainless steel tube reactor. A total of six fractions were tested. Since numerous reactions were involved, they restricted their studies to only a limited number of important reactions. This was achieved by specifying certain reaction constants as zero. Data points to determine reaction rates were selected at 360 _C, 397_C, and 422_C for bitumen and at 397_C for heavy oils and asphaltenes. Two types of models, a single-asphaltene type and a two-asphaltene type, were developed and tested in their experiments. These models were compared with the ones existing in the literature. They concluded that all the activation energy values found using these models lie in the same range as those reported by other investigators. Thermal instability has been observed in Alberta oil sands for a long time. Kinetic parameters for the asphaltenes from various bituminous sand, heavy oil, and conventional oil reservoirs were provided by Ekwenchi et al. (1984). Experimental data in the temperature range of 600_700_C were used to model these samples. Arrhenius plots for all the samples examined were linear, and the pre-exponential factors and activation energies were evaluated from the slope of these plots. A reaction rate of a first-order reaction was observed. Gas analysis showed that methane was the dominant gaseous hydrocarbon product throughout the temperature range studied. A prominent finding from their experiments was that the alkane and alkene generating portions (C1-C5) make up a substantial part of the asphaltene structure. All the samples were first subjected to heat treatment at 400_C to remove less stable materials and were later flash pyrolyzed, which led to excellent reproducibility of the results.

In Situ Combustion in EOR Another important inference from their work was that the Arrhenius parameters were closely related to the burial depths of the samples. This was further supported by the fact that samples dug from different depths showed randomly fluctuating values when compared with each other. Studies on thermal cracking of the isolated fractions of Athabasca bitumen were conducted by Mazza and Cormack (1988). A series of experiments was carried out to understand the cracking behavior of the isolated SARA fractions. Gas products of these fractions were subjected to gas chromatography (GC), high-performance liquid chromatography (HPLC), and Fourier transform-infrared spectrophotometry (FTIR) to identify chemical alterations due to thermal cracking. Chromatograms showed that the major portions of the aromatic structures resulting from the thermal cracking were diaromatic and triaromatic in nature. Results from these studies also clearly indicated that the aromatization and dealkylation of the polycyclic saturated structures in these fractions occur during thermal treatment and lead directly to the production of aromatic compounds. The process of aromatization and dealkylation would also lead to condensation reactions, resulting in the formation of resins. However, information related to this formation process obtained from chromatograms was limited. Traces of complex aromatics were also observed. GC analysis supported the hypothesis that resins and asphaltenes were chemically similar. The model could adequately predict the experimental results obtained at 408 _C but problems were observed at 362_C. This was attributed to the fact that some of the coke that passed through during the separation process was invariably accounted for as asphaltenes. All reactions had a first-order rate. Further, activation energies for the samples were found to lie in the range reported by other researchers in their respective studies. The only drawback of the experimental design was that the reservoir sand was not mixed with the sample. This would, perhaps, have had a catalytic effect on the overall reaction, yielding results more consistent with the previous studies done by different researchers. Modeling of the Maya asphaltene sample subjected to pyrolysis to understand the thermal decomposition and a quantitative analysis of the coke formation suggested that the H=C ratio for the asphaltene sample was 1.062, slightly less than the normal average, but the sample showed more aromaticity (Douda et al., 2004). Experiments showed that maltenes were obtained at a temperature of 350 _C and had an H=C ratio of 1.562. TGA tests for asphaltenes were performed in an inert atmosphere (nitrogen), and it was found that the reaction was rapid at 400 _C and terminated at 646_C. The maximum weight loss was observed at 524 _C. The residue obtained after 646_C was identified as coke. DTA results for the Maya asphaltene sample were consistent with those obtained for Cold Lake bitumen and Garzan and Raman samples (Khulbe et al., 1984; Karacan and Kok, 1997). Decomposition for maltenes was in the range of 308_686_C with peak decomposition at 478_C. Field desorption mass spectrometry (FDMS) and infrared spectroscopy were performed to study the large fragments (composed mainly of molecular ions) and to understand the chemical makeup of the polar compounds formed from the pyrolysis of the Maya asphaltene. Thermal decomposition of a Maya asphaltene at 350_C, 400_C, and 450_C under the inert nitrogen atmosphere produced coke, gas, unreacted asphaltene, and maltene. The pyrolysis kinetics of saturates was greatly influenced by the presence of the asphaltenes in the reaction (Yasar et al., 2000). Pyrolysis reaction was carried out in a reactor at 400_C, 450_C, and 500_C. The product yields were used to calculate the

1011

1012

N. Mahinpey et al.

reaction rate constants and the activation energies of the pure saturate and the saturate-asphaltene samples. It was found that the reactivity of saturates was strongly affected by the presence of asphaltene in the reacting environment. It was also seen that the activation energies of the mixture were lower than those of the pure saturate. Pyrolysis of the asphaltene fraction from the Cold Lake bitumen in a nitrogen environment was conducted by Khulbe et al. (1984) in a TGA. Experiments were performed under hydrodynamic conditions so that mass transfer through the external gas film would not affect the reaction rate. Pyrolysis of the asphaltene fraction was studied in the temperature range of 20_845_C. Maximum decomposition occurred after 12 minutes at about 480_C. It was suggested that thermal decomposition of asphaltenes at temperatures less than 350_C proceeds by elimination of groups situated at the peripheral sites of asphaltenes. More severe degradation occurred only at elevated temperatures (>350_C). Kinetic parameters were established, and a second-order decomposition was found to explain the data satisfactorily.

Field Tests
Special combustion tube tests were performed using a differential thermal analyzer (DTA), and it was found that the thermogram resulting from a quartz (a type) crude oil system showed several erratic peaks before the fuel was finally consumed at about 980_C (Hardy and Raiford, 1975). In contrast, kaolinite crude oil combustion showed a smooth thermogram with the peak at 980 _C. These peaks were representations of the exothermic and endothermic reactions occurring in the system. Further, eight combustion tube tests with volume ratios of 95:5 percent mixture of quartz and kaolinite were conducted in order to determine the process variables of their project. Extraction was performed in different patterns (field area). Gas analysis showed a rapid drop in methane composition, verifying that the oil in place (OIP) was extremely dead. A graphical representation is given in Figure 4. The project eventually had to be terminated, owing to the problems of anisotropic characteristics of the reservoir, old, unplugged wells (leakage problem), and

Figure 4.Iola fire flood, Allen County, Kansas; analyses of produced gas, Stewart well#30 (Hardyand Raiford, 1975).

In Situ Combustion in EOR production of emulsions. However, they concluded that the combustion parameters determined from the samples suggested that the in situ combustion process could be successfully applied to heavy oil in that region if other problems were dealt with efficiently. The latest field projects involving in situ combustion for oil recovery are reported from India and Romania. The two projects in India are operated in the wet mode and are functioning for the past seven years. These projects are applied successfully in deeper reservoirs, having strong lateral water drive. Each of these projects produces approximately 500 m 3 of crude every day. They operate at a pressure greater than 10.3 MPa. Both reservoirs contain coal and carbonaceous material. In both reservoirs the static pressure has been almost constant up to the present, indicating that there is a strong aquifer and the oil is still above the bubble point pressure. The reservoir located in the northwestern part of Romania operates in the dry mode in typical solution gas drive, shallow heavy-oil reservoirs. Some researchers have claimed that this project has been one of the best instrumented in situ combustion projects in the world. Hundreds of bottom hole temperature (BHT) profiles have been taken in the observation and production wells, and some of them saw very high peak temperatures (around 600_C), located in the upper part of the oil layer, clearly indicating the segregated nature of the in situ combustion process. Horizontal wells in conjunction with an in situ combustion process in the field were drilled in two Canadian projects. In the first project at Eyehill in Saskatchewan, three horizontal wells were drilled with a horizontal leg of 10001200 m. The combustion process was active for about 10 years. At the stoppage time the oil recovery was about 10%. One of these wells produced 55 60 m 3=day. The second project was at Battrum, Saskatchewan where one well was drilled and operated in wet mode. The horizontal leg was 610 m. The peak performance was about 35 75 m 3=day.

1013

Other Studies
A thermo-catalytic in situ combustion process used in Hungary was developed to show the effect of the core mineralogy on the combustion of light oil samples (Racz, 1985). The study was focused on determining whether the reservoir rock would hinder the combustion or promote it by acting as a catalyst. A pilot plant test was conducted on the oil sample extracted from a reservoir in Hungary. In a temperature range of 300_400_C, it was demonstrated that the mineralogy of the reservoir promoted the combustion process, but above 600_C, it was shown to hinder the process. Hence, these reservoir rocks could promote the combustion process only in a predetermined temperature range. One of the important catalysts used was iron pentacarbonyl, which was capable of increasing the rate of oxidation above 170 _C at a pressure of 400 kPa and above 200_C at atmospheric pressure. It was further observed that the formation of fuel in the condensation zone was far greater in the presence of the catalyst than in its absence. Hence, enthalpy reactions were affected by the presence of the catalysts. Laboratory experiments had previously demonstrated that the steam zone influenced the oil displacement and oil recovery. Laboratory experiments under these conditions also showed oil recovery in the range of 8090%, while pilot plant results revealed only 60%recovery, owing to the poor geological conditions. A study of asphaltene samples from a synthetic crude oil in a batch reactor representing the amounts of asphaltenes and yields as two classes of products,

1014

N. Mahinpey et al.

namely oil plus gas and coke, was performed by Martinez et al. (1997). They took into consideration three reaction constants and schematically represented them as:

where k1, k2, and k3 are the relevant reaction rate constants. An assumed secondorder reaction rate could fit the data well at temperatures of 425_C, 435_C, and 450_C without consideration of the formation of coke due to secondary cracking. However, the data no longer fit above 475_C, at which point coke formation was evident. Hence, it was suggested that the secondary formation of coke due to cracking of oil at high temperature was of significance. Studies of the thermal cracking of asphaltene samples suggested that the secondary formation of coke in the high-temperature region is very important, as its behavior seems to change from what was observed in the low-temperature region (Wang and Anthony, 2003). Their results at temperatures of 425 _C and 435_C showed a good linear relationship. However, at 450_C, the results appeared to deviate and at 475_C, there was a clear deviation, showing that the equations that neglected secondary cracking of the sample would not simulate well when dealing with hightemperature reactions. Hence, the investigators arrived at a relation of the form: y A0y A y ck yA where y is the weight fraction in the mixture, y A is the fraction of the asphaltene, y C is the coke yield, andkis a coefficient. This was a modified form of an equation (Wilson, 1997). The experimental and predicted values were in good agreement. They concluded that the prediction of time dependence could be improved if a good correlation of asphaltene cracking with the residence time were established. y C0

Kinetic Modeling
Numerous kinetic models to predict the in situ combustion process have been proposed by authors worldwide. These models have adequately described the mechanisms involved in in situ combustion during the EOR process. Models predicting LTO and HTO phase reactions and fuel formation and decomposition processes have all been successful (Khulbe et al., 1984). The authors used a modified form of the Gauss-Newton method to model the pyrolysis and combustion kinetics of an in situ combustion process. They arrived at second-order reaction rate and have predicted the kinetic parameters using their model. A simple kinetic model for coke combustion during an in situ combustion process was proposed by Ren et al. (2005). Several tests of TGA runs were performed to examine the thermal behavior of the oil during the in situ combustion. A sample of thermogravimetry (TG) and differential thermogravimetry (DTG) curves is shown in Figure 5. They carried out analysis to show how a two-step Arrhenius model can

In Situ Combustion in EOR

1015

Figure 5.TG=DTG curves of Neilburg oil in air.

predict combustion of coke, and they evaluated the kinetics of the reactions. Their analysis was also compared with a one-step Arrhenius model. The plots obtained at the initial and final temperatures of combustion reactions are shown in Figures 6 and 7, respectively. Due to the limited kinetic data available on the rates and nature of partial oxidation reactions, it is still difficult to model the in situ combustion system and perform numerical simulations. An investigation conducted on the in situ combustion process paved the way for modeling such a system in a three-dimensional frame. The model was based on grouping of cracking components into six pseudo-components. Five chemical reactions were considered in order to determine the kinetic parameters of the system. They aimed at modeling the in situ combustion system by simulating different well configurations. However, from their results, they concluded that scaling criteria, grid block designs, and other such pertinent factors evidenced a considerable influence during simulations and needed to be addressed in more detail. Some models, although structured well, have not considered certain parameters that are necessary in describing systems specific to some conditions. In such scenarios, those models would not be applicable. Hence, corrections made or inclusions added

Figure 6.Combustion of coke from Neilburg asphaltenes at 403_C.

1016

N. Mahinpey et al.

Figure 7.Combustion of coke from Neilburg asphaltenes at 494 _C.

to the existing model or deduction of new models became necessary to arrive at accurate and reliable inputs for analysis. A description of such a system was reported by Lin et al. (1987). The generally accepted first-order kinetic model for the pure components presented by them was unsatisfactory for the multicomponents characterized by pseudo-components. Hence, some modifications were applied to the existing model in order to clearly define the thermal cracking of the mixtures. The authors main objective was to incorporate into their model such details as: the apparent reaction order is always greater than one, the reaction order is a decreasing function of temperature, and coke may be formed from intermediate products. The new kinetic models developed by incorporating these details were able to correlate the reaction rate parameters and the stoichiometry needed in the numerical simulation of the in situ combustion process and to accurately predict the product distribution. The models developed were global reaction models and were suitable for both thermal and catalytic cracking reactions. An important inference from their work was that coke was not the only source of fuel that was available for in situ combustion. However, the authors stressed that their models do not stipulate how crude should be fractionated to light, medium, and heavy oil pseudo-components but can only describe the cracking kinetics of the pseudo-components. The authors suggested that the new models can be incorporated into a thermal simulator for simulation of the in situ combustion process. Another such full-fledged model was proposed by Belgrave et al. (1990). They argued that previous studies on modeling the in situ combustion process had not incorporated oil compositional changes due to oxidation and also had unrealistically specified reaction products, in the LTO region, as carbon oxides and water. Their improved model considered these factors. It could also predict the dual-oxidation uptake peaks associated with ramped temperature oxidation experiments. Hydrocarbon bypassing due to quenching of the combustion front was also permitted in their model.

Discussion and Conclusion


Good agreement between the laboratory results and those obtained in the field has been a positive development that encourages investment in research to further

In Situ Combustion in EOR contribute to this field. Research in this area provides clear insight into the different parameters of in situ combustion and their effects on the in situ process. It also provides guidance in how these parameters have to be varied in order to increase the efficiency of the system. For example, the Lloydminster reservoir oil has lower viscosity than the Athabasca bitumen and tends to form more viscous emulsions in the oil-rich region. Hence, parameters such as the combustion front temperature, its movement for easier oil mobility, and air injection rate need to be addressed at different levels for each type of reservoir oil. The data for air and fuel requirements, which are some of the key factors in this type of EOR, are usually synthesized at laboratory and pilot scale before being applied at field scale. These data, when used in the field, have given results that were nearly optimal. This has favored further research in this area. Our research group in the past few years has conducted extensive research at the lab scale to understand the pyrolysis and oxidation behavior of coke formed from medium and light oil and also to propose more realistic models to mimic the true behavior of in situ combustion (Ambalae, 2005; Ren, 2006). Although coke combustion studies have long been conducted, a better understanding of reaction kinetics is still needed. The thermal behavior of various feeds has been studied using TGA. These experimental data have provided reliable input to our modeling efforts. Based on the experiments performed on Neilburg oil, it was confirmed that the coke derived from Neilburg asphaltenes is reasonably representative of the coke derived from whole oil. Assuming that coke derived from Neilburg asphaltenes is the same as coke derived from the whole oil will cause no more than acceptably small errors for most simulation purposes. We have recently extended our studies to compare similar results obtained from oil samples belonging to different reservoirs. This could largely help to generalize the results over a broad range of oil samples. As discussed, it has been recognized that in situ combustion is a potentially effective enhanced oil recovery (EOR) technique particularly suitable for medium and heavy crude oilbearing reservoirs, as well as for steam-depleted reservoirs. However, since the in situ combustion process involves a variety of chemical reactions coupled with simultaneous heat, mass, and momentum transfer, it is considered one of the most difficult EOR methods to simulate either physically or numerically. In addition, the fact that the fundamental reaction kinetics and mechanisms of the in situ combustion process have not been completely understood makes its field performance prediction unreliable. Therefore, a comprehensive modeling study will play a significant role in the commercialization of this viable technique. In the literature, the Arrhenius plot method (Arrhenius model) is the most often used model for coke combustion investigation. Meanwhile, other models have been developed to compensate for the shortcomings of the Arrhenius model. Fassihi et al. (1980) identified three consecutive oxidation classes during the combustion of crude oil in porous media and developed a model based on Weijdemas (1968) kinetic equa tion to calculate oxygen consumption and kinetic parameters for each reaction class. We also have been working in this direction to propose models that could explain experimental data more accurately. One such model was proposed by Ren et al. (submitted). Based on the observations from pyrolysis and non-isothermal oxidation, a two-coke oxidation reaction model was proposed in order to accommodate the influence of coke source and coking temperature on coke combustion. It was assumed that cokes obtained at different temperatures were not the same and that there existed two cokes: coke obtained at low temperature and coke obtained at high

1017

1018

N. Mahinpey et al.

temperature, or light coke and heavy coke. The coke samples obtained between the lower and higher temperature can be regarded as mixtures with different percentages of light and heavy coke. Generally, the two-coke model provided better prediction than the Arrhenius model, especially at low temperatures. The other model, a two-step oxidation reaction model, was proposed to correlate the experimental data for isothermal coke combustion runs (Ren et al., 2005). The chemical reactions were simplified into two oxidations occurring in series. In the first reaction, coke was partially oxidized to form an intermediate product, which was then burned in the second reaction. Based on the TGA data, kinetic parameters were estimated with the aid of custom-written software. For comparison, the onestep oxidation reaction model (the Arrhenius model) was also employed to predict the combustion process. The two-step oxidation reaction model gave a better fit to the experimental data. A comparison of the laboratory and field performance for Canadian oils was reported by Moore et al. (1995). The comparison showed that operating on small patterns contributes to the successful burning of the oil. Amoco proved this concept with an Athabasca reservoir sample. Many other tests confirmed the role of the mobility of the oil in the success of an in situ combustion process. Steam preheating followed by fire-flooding was found to be an effective method for oil recovery in the Cold Lake reservoir. There were failures observed in the Wabasca and Lindberg fields, mainly due to large pattern sizes and lack of heated communication paths within these reservoirs. All these observations were similar to what had happened at the laboratory scale. Hence, the success of in situ combustion for the EOR process has always been an encouragement for researchers to delve into this field and search for more information to accurately understand and manipulate this phenomenon, thereby contributing to the overall efficiency of the system. Several conclusions can be drawn from this review of the studies that have been accomplished in the field of in situ combustion for enhanced oil recovery. First, the field test results are in good compliance with those obtained in the laboratory and pilot scale setup. Hence, investment in this area has been fruitful. Second, it has been observed that results obtained at the lab scale and pilot scale have been effectively used in the field. Third, an understanding of the kinetics has been demonstrated. This understanding has led to the development of kinetic models that can predict the in situ combustion process, but in terms of any particular field, few models are able to predict the process very accurately. However, scope exists to formulate accurate global solutions in order to develop precise and dependable numerical simulation software. Fourth, there has always been room for improving economic viability while incorporating the in situ combustion process for enhanced oil recovery. The costs associated with the recovery process have sometimes overshot the available funds and at other times have given a narrow margin of profit. For example, one study revealed that the oil in place (OIP) was dead in terms of methane content. Hence, the investments in these projects were futile. The industry has always sought to optimize the operating parameters and thereby improve the efficiency of the system. Hence, research and development to generalize the established fundamental research for commercial field applications is critical. While laboratory and pilot applications have, as shown in this review, been well studied, design considerations have been provided by few researchers for the in situ combustion process. The future perspective for in situ combustion is that, in years to come, the efficiency and productivity of the process will improve with the evolution of the

In Situ Combustion in EOR preliminary research presented in this review into general field application studies. Success in this will lead to the tools necessary to optimize commercial application of in situ combustion, making it an attractive EOR alternative.

1019

Acknowledgments
The authors wish to acknowledge the financial support extended by the Petroleum Technology Research Centre (PTRC) and Natural Science and Engineering Research Council (NSERC) to conduct pyrolysis and combustion experiments, a part of which has been included in this review. Thanks to Dr. Gordon Moore, Professor, University of Calgary, for his valuable feedback on this review.

References
Adewusi, V. A. and Greaves, M. (1991). Forward in-situ combustion: Oil recovery and properties,Fuel,70(4), 503508. Akin, S., Bagsi, S., and Kok, M. V. (2002). Experimental and numerical analysis of dry forward combustion with diverse well configuration,Energy Fuels,16(4), 892903. Ali, M. F. and Saleem, M. (1991). Thermal decomposition of Saudi crude oil asphaltenes,Fuel Sci. Technol.,9(4), 461484. Ambalae, A. (2005). Studies on pyrolysis and combustion behaviour of Neilburg oil and derived asphaltenes using a thermogravimetric analyser (TGA), Masters thesis, Univer sity of Regina. Ambalae, A., Mahinpey, N., and Freitag, N. P. (2006). Thermogravimetric studies on pyrolysis and combustion behavior of a heavy oil and its asphaltenes,Energy Fuels,20(2), 560565. Banerjee, D. K., Laidler, K. J., Nandi, B. N., and Patmore, D. J. (1986). Kinetic studies of coke formation in hydrocarbon fractions of heavy crudes,Fuel,65, 480484. Bayambajav, E. and Ohtsuka, Y. (2003). Cracking behaviour of asphaltenes in the presence of iron catalysts supported on mesoporous molecular sieve with different pore diameters, Fuel,82, 15711577. Behar, F., Ungerer, P., Audibert, A., and Villalba, M. (1988). Experimental study and kinetic modelling of crude oil pyrolysis in relation to thermal recovery processes, in4th UNITAR=UNDP International Conference on Heavy Crude and Tar Sands,9797(14), Alberta Oil Sands Technology and Research Authority, Edmonton. Belgrave, J. D. M., Moore, R. G., Ursenbach, M. G., and Bennion, D. W. (1990). A comprehensive approach to in-situ combustion modelling, paper presented at SPE 7th Symposium on Enhanced Oil Recovery, Tulsa, Oklahama. Benson, S. W. (1982).The Foundations of Chemical Kinetics, 45 46, McGraw-Hill, New York. Burger, J. G. and Sahuquet, B. C. (1971). Chemical aspects of in-situ combustion heat of combustion and kinetics, inProceedings of the SPE 46th Annual Fall Meeting, New Orleans, Society of Petroleum Engineers, Richardson, Tex. Douda, J., Llonas, M. E., Alvarez, R., Franco, C. L., and Fuente, J. A. (2004). Pyrolysis applied to the study of Maya asphaltene,J. Anal. Appl. Pyrolysis,71, 601 612. Ekwenchi, M. M., Lown, E. M., Montogomery, D. S., and Straucz, O. P. (1984). High temperature pyrolysis of petroleum asphaltenes,AOSTRA J. Res.,1(2), 127 137. Fassihi, M. R., Brigham, W. E., and Ramey, Jr., H. J. (1980). The reaction kinetics of in-situ combustion, paper SPE 9454 presented at the 55th Annual Fall Technical Conference and Exhibition of the Society of Petroleum Engineers of AIME, Dallas, Texas. Fassihi, M. R., Brigham, W. E., and Ramey, Jr., H. J. (1984). Reaction kinetics of in-situ combustion: Part 2Modeling,Soc. Pet. Eng. J.,24(4), 408416.

1020

N. Mahinpey et al.

Freitag, N. P. and Exelby, D. R. (2002). A SARA-based model for simulating the pyrolysis reactions that occur in high temperature EOR processes, InProceedings of the Canadian International Petroleum Conference, Calgary, Alberta, Petroleum society of CIM, Calgary, Alberta. Greaves, M., Young, T. J., El-Usta, S., Rathbone, R. R., Ren, S. R., and Xia, T. S. (2000). Air injection into light and medium heavy oil reservoirs: Combustion tube studies on West of Shetlands Clair oil and light Australian oil,Chem. Eng. Res. Des.,78(A), 721730. Hardy, W. C. and Raiford, J. D. (1975). In-situ combustion in a Bartlesville sand-Allen County, Kansas, inProceedings of the 1st Annual Tertiary Oil Recovery Conference, Wichita, Kansas, vol. 2, 2438, Kansas University Tertiary Oil Recovery Project, Lawrence, Kansas. Hayashitani, M., Bennion, D. W., Donnelly, J. K., and Moore, R. G. (1978). Thermal cracking models for Athabasca oil sands oil, paper presented at the 53rd Annual Technical Conference and Exhibition of the SPE of AIME, Houston, Texas. Howes, B. J. (1978). Enhanced oil recovery in Canada: Success and progress, inProceedings of the Engineering Centennial Convention, Montreal, Canada, Government of Canada Publications, Alberta. Jess, A. (1996). Mechanisms and kinetics of thermal reactions of aromatic hydrocarbons from pyrolysis of solid fuels,Fuel,75(12), 14411448. Karacan, O. and Kok, M. V. (1997). Pyrolysis analysis of crude oils and their fractions, Energy Fuels,11, 385391. Khulbe, K. C., Sachdev, A. K., Mann, R. S., and Davis, S. (1984). TGA studies of asphaltenes derived from cold-lake (Canada) bitumen,Fuel Process. Technol.,8, 259266. Kisler, J. P. and Shallcross. D. C. (1997). An improved model for the oxidation process of light crude oil,Chem. Eng. Res. Des.,75(A), 392400. Lin, C. Y., Chen, W. H., and Culham, W. E. (1987). New kinetic models for thermal cracking of crude oils in in-situ combustion processes,SPE Reserv. Eng.,2, 54 66. Mamora, D. D. and Brigham, W. E. (1995). The effect of low-temperature oxidation on the fuel and produced oil during in-situ combustion of heavy oil,In Situ,19(4), 341365. Martin, W. L., Alexander, J. D., and Dew, J. N. (1958). Process variables of in-situ combustion,Trans. Am. Inst. Min. Metall. Pet. Eng. Inc.,213, 28 35. Martinez, M. T., Benito, A. M., and Callejas, M. A. (1997). Thermal cracking of coal residues: Kinetics of asphaltene decomposition,Fuel,76, 871877. Mazza, A. G. and Cormack, D. E. (1988). Thermal cracking of the major chemical fractions of Athabasca bitumen,AOSTRA J. Res.,4, 193208. Moore, R. G., Laureshen, C. J., John, D. M., Ursenbach, M. G., and Mehta, S. A. (1995). In-situ combustion in Canadian oil reservoirs,Fuel,74(8), 1169 1175. Moore, R. G., Laureshen, C. J., Mehta, S. A., and Ursenbach, M. G. (1997). Observations and design considerations for in-situ combustion projects, paper presented at the 48th Annual Technical Meeting of the Petroleum Society of CIM, Calgary, Alberta. Prats, M. (1982).Thermal Recovery. Society of Petroleum Engineers of AIME, New York. Racz, D. (1985). Development and application of a thermocatalytic in-situ combustion process in Hungary, in:European Meeting on Improved Oil Recovery, 431440, Rome, Italy, SPE International, Rome, Italy. Ranjbar, M. (1992). Improvement of heavy and light oil recovery with thermocatalytic in-situ combustion, paper CIM 9260 presented at the CIM 1992 Annual Technical Conference, Calgary. Ranjbar, M. (1993). Influence of reservoir rock composition on crude oil pyrolysis and combustion,J. Anal. Appl. Pyrolysis,27, 87 95. Ranzi, E., Faravelli, T., Gaffuri, P., and Goldaniga, A. (1997). Primary pyrolysis and oxidation reactions of linear and branched alkanes,Ind. Eng. Chem. Res.,36, 33363344.

In Situ Combustion in EOR


Ren, Y. (2006). Kinetic models for the combustion of cokes obtained from Neilburg oil and its derived asphaltenes, Masters thesis, University of Regina. Ren, S. R., Greaves, M., and Rathbone, R. R. (1999). Oxidation kinetics of North Sea light crude oils at reservoir temperature,Chem. Eng. Res. Des.,77(A), 385 394. Ren, Y., Freitag, N. P., and Mahinpey, N. (2005). A simple kinetic model for coke combustion during an in-situ combustion (ISC) process, paper 2005110 presented at the 56th Annual Technical Meeting of the Petroleum Society, Calgary, Alberta, June 79. Ren, Y., Freitag, N. P., and Mahinpey, N. (Submitted). A kinetic model for the combustion of coke derived at different coking temperatures,Fuel. Rosen, R. (1941). Kinetics and chemistry of cracking of hydrocarbons,Oil Gas J.,40, 49 50. Schaffie, M. and Ranjbar, M. (2000). Geochemical alterations of crude oils during thermal recovery process,J. Pet. Sci. Eng.,26, 5765. Shallcross, D. C., De los Rios, C. F., Castanier, L. M., and Brigham, W. E. (1991). Modifying in-situ combustion performance by the use of water-soluble additives,SPE Reserv. Eng., 6(3), 287294. Tadema, H. J. (1959). Mechanism of oil production by underground combustion,Proc. World Pet. Congr.,5, 279287. Verkoczy, B. (1993). Factors affecting coking in heavy oil cores, oils and SARA fractions under thermal stress,J. Can. Pet. Technol.,32(7), 2533. Verkoczy, B. and Freitag, N. P. (1997). Oxidation of heavy oils and their SARA fractions: Its role in modelling in-situ combustion, paper 97167 presented at the 7th Petroleum Conference of the South Saskatchewan Section, Petroleum Society of CIM, Regina. Wang, J. and Anthony, E. J. (2003). A study of thermal cracking behaviour of asphaltenes, Chem. Eng. Sci.,58, 157162. Weijdema, J. (1968). Determination of the oxidation kinetics of the in-situ combustion process,Exploratie En Produktic Laboratorium, Koninklijke=Shell Rijswijk, the Netherlands. Wilson, J. W. (1997).Fluid Catalytic Cracking Technology and Operations, PenWell, Tulsa, Okla. Yasar, M., Cerci, F. E., and Gulensoy, H. (2000). Effect of asphaltenes on pyrolysis kinetics of saturates,J. Anal. Appl. Pyrolysis,56, 219228.

1021

You might also like