New Insights into the Understanding of High-Pressure Air Injection (HPAI): The Role of the Different Chemical Reactions

Abstract

High-pressure air injection (HPAI) is an enhanced oil recovery process in which compressed air is injected into deep, light oil reservoirs, with the expectation that the oxygen in the injected air will react with a fraction of the reservoir oil at an elevated temperature to produce carbon dioxide. The different chemical reactions taking place can be grouped into oxygen addition, thermal cracking, oxygen-induced cracking, and bond scission reactions. The latter reactions involve the combustion of a flammable vapor as well as the combustion of solid fuel, commonly known as “coke”. Since stable peak temperatures observed during HPAI experiments are typically below 300 °C, it has been suggested that thermal cracking and combustion of solid fuel may not be important reaction mechanisms for the process. The objective of this work is to assess the validity of that hypothesis. Therefore, this study makes use of different oxidation and combustion HPAI experiments, which were performed on two different light oil reservoir samples. Modeling of those tests indicate that thermal cracking is not an important reaction mechanism during HPAI and can potentially be ignored. The work also suggests that the main fuel consumed by the process is a flammable vapor generated by the chemical reactions. This represents a shift from the original in situ combustion paradigm, which is based on the combustion of coke.

1. Introduction

High-pressure air injection (HPAI) is a technology for the recovery of light oils [1,2] that has proven to be a valuable enhanced oil recovery (EOR) process, especially in deep reservoirs [3,4,5]. However, the chemical reactions that occur during the in situ combustion (ISC) process that takes place are not completely understood, which limits the predictive capabilities of simulation models. This seems to have limited the widespread application of HPAI within the oil and gas industry.

Historically, the air injection literature has stated that the main fuel for the ISC process is the carbon-rich, solid-like residue resulting from distillation, oxidation, and thermal cracking of the residual oil near the combustion front, commonly referred to as “coke” [6,7,8,9]. However, combustion tube tests performed on light oils at high pressures rarely display any significant sign of coke deposition [10,11]. Furthermore, stable peak temperatures during those experiments are typically below 300 °C, which is beneath the necessary temperature required for the combustion of coke.

In previous recent papers [10,11], a change to the original ISC paradigm was suggested regarding the nature of the fuel, especially for the case of HPAI. Based on analysis of experimental data, those studies highlighted the occurrence and importance of combustion in the gas phase, especially in the case of light oils. Furthermore, a recent reservoir simulation study supported such proposed theories by modeling high-pressure ramped temperature oxidation (HPRTO) and combustion tube (CT) tests on three different crude oils, which included two light oil samples [12,13]. That study developed a kinetic model for each reservoir by history matching an HPRTO experiment for each case. Such reaction models were subsequently validated by history matching a CT test for each laboratory sample. An important feature of all those experiments is that they were performed at representative reservoir pressure conditions.

This study leverages those simulation results from the two light oils by performing additional sensitivity runs using their numerical models. This work supports the previously proposed theories and confirms the idea that the main fuel consumed by the HPAI process is likely a flammable vapor generated by chemical reactions, which burns in the gas phase. This represents a shift from the original in situ combustion paradigm, which is based on the generation and consumption of semi-solid residue.

On the other hand, since stable peak temperatures observed during combustion experiments performed on light oils at high pressures are typically below 300 °C, it is expected that thermal cracking reactions and combustion of solid fuel are not important reaction mechanisms during the HPAI process. Numerical results presented in this work support those observations and suggest that such reactions can potentially be ignored. This would simplify kinetic models used to model the process, which would result in higher computational efficiencies when performing simulations at the field scale.

2. Materials and Methods

The analysis presented in this work uses kinetic models developed previously for two light oil reservoirs, based on the modeling of thermal cracking, oxidation, and combustion experiments [12,13]. Such models are evaluated in order to gain further insight into the HPAI process and understand the role played by the different reactions.

2.1. Conceptual Basis of the Modeling Approach

The concepts and modeling approach used here can be described in terms of conventional combustion theory, which is referred to as “aboveground combustion” [11].

It is well known that liquid and solid fuels do not ignite. It is the vapors given off from the surface of the materials that ignite, combust, and drive the consumption of the fuel. A good practical example of this is the lighting of a candle, which is made of wax (i.e., a hydrocarbon mixture). In order to light the candle, one does not place a flame on the solid wax but on the wick, and it is the emitted vapors (another hydrocarbon mixture) from the heating of the wax that ignite, combust, and drive the consumption of the candle. In fact, immediately after being put out, a candle can be relit by simply using a flame on the emitted vapors, in proximity to it, without the flame physically touching the wick. Moreover, a lit candle has a “pool” of liquid wax (also a hydrocarbon mixture) at the top of it, which is continuously oxidized (not burned) by the surrounding oxygen present in the air. This example also illustrates the difference between oxidation and combustion of hydrocarbon mixtures, which are known to occur “aboveground”.

In this context, an oxidation reaction “aboveground” consists of adding oxygen to a hydrocarbon, which results in the generation of heat. During “aboveground” combustion, on the contrary, hydrocarbons are ignited, completely burned, and produce carbon oxides and water. The combustion process as it occurs in situ, within porous media, appears to be similar to “aboveground” combustion and can be generally described by the following types of reactions: oxygen addition, bond scission, thermal cracking, and oxygen-induced cracking reactions.

The simulation models used by this work have been presented previously [12,13]. Such models were used to develop a reaction framework for a variety of oils, which included the two light oils of interest to this study. The modeling approach included the simulation of high-pressure ramped temperature cracking (HPRTC) and high-pressure ramped temperature oxidation (HPRTO) experiments, which were used to derive kinetic models, and the simulation of high-pressure combustion tube (HPCT) tests that served to validate the chemical reaction frameworks developed previously.

2.2. Overview of the Crude Oils

The crude oils used during the experiments of interest to this study came from two different reservoirs and exhibit different phase behavior at reservoir conditions. Although all laboratory tests used dead oil samples, for reference it is worth mentioning the differences in the phase behavior of the crude oils at reservoir conditions.

Oil I is a conventional oil that comes from a sandstone reservoir encountered at 98 °C. The oil is reported to behave as a “black oil” and separator samples from the field indicate an API gravity of 36 [14]. On the other hand, Oil K comes from a carbonate reservoir encountered at 149 °C. The oil is reported to behave as a “volatile oil” and separator samples from the field indicate an API gravity of 44 [15].

Regarding the dead oil samples actually used during the experiments, the API gravities of Oils I and K are 33.1 and 38.8, respectively. These are some of the samples also described by Gutiérrez et al. [10,11,12].

One important characteristic of these two crude oils lies on their low asphaltene content, which has a direct impact on the viscosity and density of the sample and also has an effect on the chemical reactions taking place. Oxidation of such a fraction tends to lead to the generation of even heavier hydrocarbons and, eventually, coke, which influences the fluid flow and performance of the ISC process. While the mass of asphaltenes in a heavy oil sample is likely higher than 10 percent, light oils tend to exhibit low asphaltene content, likely lower than 5 percent in mass [10]. In the cases of interest to this study, the mass of asphaltene in Oil K is basically non-existent (only 0.05 percent mass), and the mass content of asphaltenes in Oil I is also very low (approximately 0.5 percent mass).

2.3. Overview of the Laboratory Experiments

The experiments of interest to this study are HPRTC and HPRTO tests, as well as HPCT experiments.

2.3.1. High-Pressure Ramped Temperature Experiments

A ramped temperature (oxidation or cracking) test involves the controlled heating of a recombined oil-saturated core, which is packed into a one-dimensional plug flow reactor and mounted into a high-pressure annular jacket. A cracking experiment involves the injection of nitrogen, whereas air is injected during an oxidation experiment. This reactor is subjected to a heating ramp rate, and comparisons of the temperatures of the seven reactor zones (into which the core matrix is packed) provide an accurate means of identification of the temperature at which significant changes in energy generation or consumption rates occur. Analysis of the product gas stream provides information as to the oxygen uptake and fuel consumption rates, the appearance of distillation/cracking reactions, and the subsequent generation of light components.

Produced liquid analysis provides the changes in the oil (upgrading/downgrading). Post-test core analysis provides information about the composition and concentrations of the residual hydrocarbons in the post-test core.

The high-pressure ramped temperature system was first described by Barzin et al. [16] to study the oxidation behavior of lighter oils, and it is capable of operating at pressures up to 42 MPag. Figure 1 depicts a schematic of the system. Typically, a mixture of reservoir core, oil, and brine is packed in the reactor and it is pressurized with air [or ultra-high purity (UHP) nitrogen], while the annular space in the enclosing pressure jacket is simultaneously pressurized with helium to bring the whole system to reservoir pressure. Once the heating schedule is defined for the system, air (or UHP nitrogen) is fed into the reactor at a pre-determined injection flux. The typical slow heating rate used (40 to 60 °C/h) for the reactor allows for the collection of the requisite amount of produced gas compositional data to enable the detailed analyses of each test. Liquid production is collected with a trap system capable of isolating liquid samples at design pressure without the loss of vapor associated with sampling by venting the trap to the atmosphere.

The test is generally operated in the vertical position with injection at the top and production at the bottom. The tubular reactor is constructed from nominal one inch (25.4 mm) Inconel tubing with 477 mm of inside length. The reactor is equipped with seven internal thermocouples, equivalently spaced across the length of the reactor. A proprietary heater design is used to supply energy to the tubular reactor and a custom-designed data acquisition system is used to record reactor and heater temperatures and pressure data. A mass flow controller is used to control the rate of flow of gas into the reactor while a back pressure regulator is used to maintain the desired test pressure. The composition of the produced gas is analyzed using a gas chromatograph and stored on a computer for subsequent analysis, and a wet test meter provides a cumulative reading of the volumetric flow of the exit gas stream. Some of the experimental HPRTO results have been presented and discussed previously [10,11,12,13]. In particular, experimental details and results of the experiments performed on Oils I and K have been reported by Apolinar Morales [14] and Hernández Hernández [15], respectively.

These tests are quantitative kinetic experiments that can be history-matched in order to derive a reaction scheme that can be used to model the HPAI process.

Table 1. HPRTC core pack details and operating conditions.

Properties	Oil I	Oil K
Type of core	Sandstone	Dolomite
Brine, g	13.7	12.6
Oil, g	44.6	28.8
Porosity, %	44.1	36.8
Permeability, mD 1	11,000	15,000
Oil density at 25 °C, g/cm3	0.8528	0.8238
Oil API gravity	33.1	38.8
Oil viscosity at 25 °C, cP	11.3	4.1
Oil molecular weight, g/gmol	219	190
Asphaltene mass percent, %	0.44	0.05
Injection gas	Nitrogen	Nitrogen
Operating pressure, MPa	11.9	15.3
Initial temperature, °C	25	28
Injection flux, std m3/m2 h	30.0	30.0
Setpoint temperature, °C	450	450
Ramp rate, °C/h	40	40

¹ Assumed parameter.

summarizes the properties of the reactor core packs and operating conditions for the two HPRTC experiments. Similarly,

Table 2. HPRTO core pack details and operating conditions.

Properties	Oil I	Oil K
Type of core	Sandstone	Dolomite
Brine, g	15.2	15.8
Oil, g	44.8	27.7
Porosity, %	44.3	36.5
Permeability, mD 1	11,000	15,000
Injection gas	Normal air	Normal air
Operating pressure, MPa	11.9	15.3
Initial temperature, °C	24	29
Injection flux, std m3/m2 h	30.0	29.1
Setpoint temperature, °C	450	450
Ramp rate, °C/h	40	40

¹ Assumed parameter.

illustrates the test conditions and core sample properties corresponding to the HPRTO experiments. The same crude oil sample was used for both experiments, HPRTC and HPRTO. Notice that the properties of the core packs are very similar and the operating conditions were almost identical. It is also noted that the core used is not a consolidated rock but a pack that is obtained after crushing the original rock sample until a sand-like consistency is achieved. This typically results in higher porosities and permeabilities than those encountered in the reservoir. Additionally, permeability is typically not measured during this experiment. The values listed in Table 1 and Table 2 are assumed based on experience from other experiments, such as combustion tube tests.

2.3.2. High-Pressure Combustion Tube Test

The overall purpose of a combustion tube test is to assess the burning characteristics of a particular reservoir material (core and fluids) under conditions that would be encountered in the field. The combustion tube tests under study were performed in a high-pressure combustion tube system.

The core holder used for this test is a 4 inch (10 cm) outer diameter, 5.5 ft (1.69 m) long, thin walled (2 mm) type 600 Inconel tube. The tube is equipped with 33 clamshell-type heaters, forming 33 heating zones, each 2 inches (5 cm) in length, thus allowing for near adiabatic operation of the combustion test. Associated with each heating zone is a pair of fixed location thermocouples, one that extends radially into the centerline of the tube and one that is fastened to the tube wall at the same axial position. As the centerline thermocouple senses the approach of the combustion front by the increased difference with the corresponding wall thermocouple, the heater is activated in order to bring the wall temperature within 5 °C to 10 °C of the centerline value in order to minimize heat losses to the surroundings. A schematic drawing of the combustion tube is given in Figure 2. TC refers to the core thermocouples, TW refers to the wall thermocouples, and HR represents the different heaters.

This combustion tube is capable of operation at pressures up to 6000 psig (42 MPag). It is generally operated in the vertical position with injection at the top and production at the bottom. This HPCT system is optimized for testing conventional and light oil reservoirs for air injection-based recovery processes. Some of the combustion tube test results corresponding to Oils I and K have been discussed previously [10,11,12,13]. In particular, a complete description of this experiment, as well as details corresponding to the test performed on Oil K, have been presented by Ruteaga Romero [17].

These experiments are combustion performance tests, which can be modeled for the purpose of validating a kinetic model developed previously by history matching HPRTO tests.

Table 3. HPCT core pack details and operating conditions.

Properties	Oil I	Oil K
Type of core	Sandstone	Dolomite
Porosity, %	51.0	41.6
Core pack volume, cm3	6465	5013
Permeability, mD	11,300	15,000
Start time (time zero)	Start of air injection	Start of air injection
Oil, g	3265.6	2555.1
Brine, g	2587.5	1557.2
Initial oil saturation, %	65.4	69.1
Initial water saturation, %	34.6	30.9
Initial gas saturation, %	0.0	0.0
Operating pressure, MPa	11.9	15.3
Core temperature, °C	91	149
Ignition temperature, °C	175	175
Injection flux, std m3/m2 h	29.9	30.0
Injection gas	Normal air	Normal air

summarizes the properties of the reactor core packs and operating conditions for the two HPCT experiments. These tests used the same crude oil samples as the ramped temperature experiments, the properties of which are listed in Table 1.

2.4. Overview of the Reservoir Simulation Models

All numerical models were built using simulation software developed by the Computer Modelling Group (CMG) [18]. An overview of the two models is presented below; additional details have been reported elsewhere [12,13].

2.4.1. HPRTC and HPRTO Numerical Model

The simulation model uses a two-dimensional (2D) cylindrical coordinate system that resembles the geometry and dimensions of the reactor. Figure 3 illustrates a photo of the experimental core holder along with the numerical representation of the reactor.

The simulation grid consists of five grid blocks in the radial direction, one grid block in the azimuthal direction, and 50 grid blocks in the vertical direction, which is the direction of fluid flow (top–down). The injection and production ends of the reactor are packed with frac sand (no liquid saturation) so the thickness of the top and bottom layers is calculated based on the reported density and mass of the frac sand used. This typically gives a grid block thickness of approximately 1 cm for the layers at the injection and production ends. The remaining layers, where the actual core pack is located, uses a constant grid block thickness of 0.95 cm.

The five grid blocks in the radial direction are designed to represent the core pack, reactor wall, fiberglass tape insulation, copper barstock, and the heaters of the equipment. The intention is to capture heat transfer from the heaters to the core as well as the heat losses experienced by the core.

The equipment is assumed to be adiabatic, with no heat losses allowed from the model to the environment, which is reasonable since the outer boundary of the equipment is modeled by the heating block.

The injection end is controlled by the injection rate and injection temperature to honor experimental conditions. The production end is controlled by a constant production pressure (backpressure) to honor the experimental measurements.

2.4.2. HPCT Numerical Model

The simulation model uses a one-dimensional (1D) cylindrical coordinate system that resembles the geometry and dimensions of the combustion tube. Figure 4 illustrates a photo of the experimental HPCT core holder along with a numerical representation of the combustion tube.

The simulation grid consists of one grid block in the radial direction and one grid block in the azimuthal direction. This is used to simulate the core pack. In the vertical direction, the HPCT model has 164 grid blocks. This gives a grid block thickness of approximately 1.02 cm, which is similar to the value used for the HPRTC and HPRTO experiments. Since the main purpose of this step is to validate the kinetic models developed previously, the level of rigor regarding radial heat transfer and heat losses was not considered so important [12,13]. Also, the model was assumed to be adiabatic, with no heat losses allowed from the core to the surroundings.

The injection end is controlled by the injection rate and injection temperature to honor the experimental conditions. The production end is controlled by a constant production pressure (backpressure) to honor the experimental measurements. Heaters are used only in the first few grid blocks, corresponding to the ignition zone, in order to model the ignition process that occurs at time zero, right at the start of air injection. Once ignition temperature is achieved, the heaters in the ignition zone—at the top of the core pack—are turned off and the ISC process is self-sustained by the injection of air, consumption of oxygen and fuel, and generation of heat.

2.5. Reaction Models

This study uses the same simulation models developed previously for other investigations [12,13], which are an extension of the kinetic model originally proposed by Belgrave et al. [19]. The intention of this work is to explain the role played by the different reactions, especially thermal cracking and combustion of solid fuel, during HPAI.

In these models, the original oil was described in terms of maltenes and asphaltenes, and additional pseudo-components were added in order to represent other species.

2.5.1. Thermal Cracking Model

The reactions used to model the HPRTC experiments are of the form:

Maltenes → Asphaltenes + Upgrade + Gas

(1)

Asphaltenes → Residue + Upgrade + Gas

(2)

The gas pseudo-component is present only in the gas phase (non-condensable gas) and it is meant to represent all of the generated gases (i.e., hydrocarbons and hydrogen) produced during the experiments. The residue pseudo-component represents the solid hydrocarbon fraction commonly known as coke. The upgrade pseudo-component is meant to represent the modified produced oil fraction observed during the HPRTC experiments.

All liquid hydrocarbon pseudo-components are considered to be dead and are only present in the oleic liquid phase. For the thermal cracking experiments, the only component allowed to vaporize, which is present in more than one phase, is water.

The stoichiometry and kinetic parameters of the reactions are different depending on each crude oil system, which has been discussed in detail elsewhere [13].

2.5.2. HPAI Reaction Model

The reactions used to model the HPRTO and HPCT experiments are similar to those proposed by Belgrave et al. [19], and are of the form:

• Thermal Cracking Reactions

Maltenes → Asphaltenes + Gas

(3)

Asphaltenes → Residue + Flammable _{(liquid + gas)} + Gas

(4)

• Oxygen Addition Reactions

Maltenes + Oxygen → Pseudo-Asphaltenes + Energy

(5)

• Oxygen-induced Cracking Reactions

Pseudo-Asphaltenes → Residue + Flammable_{(liquid + gas)}

(6)

• Bond Scission Reactions

Residue + Oxygen → CO₂ + CO + H₂O + Energy

(7)

Flammable_(gas) + Oxygen → CO₂ + CO + H₂O + Energy

(8)

Two additional pseudo-components are added to the HPAI reaction model. One is called “pseudo-asphaltenes”, which is a product of oxygen addition reactions and represents an oxidized heavy fraction. The other one is named “flammable”, and it represents a flammable hydrocarbon mixture that is a product of oxygen-induced cracking reactions. This pseudo-component results from the cracking of the pseudo-asphaltenes and may exist in the liquid and gas phases (i.e., it is a flammable vapor).

This is a relatively long list of reactions, which involve several different kinetic parameters that affect the ability to understand the physics and chemistry of the process. Furthermore, such reactions also have an impact on computing time, which becomes critical for field-scale simulation models. This is why an optimum number of chemical reactions is desired.

3. Results

The simulation results of the different experiments have already been presented [12,13]. Such models make use of all of the reactions described above. In this study, the purpose is to evaluate the significance of the reactions by performing additional sensitivity runs. In particular, the interest is to understand the role played by thermal cracking reactions as well as the combustion of solid fuel.

3.1. Observations from HPRTC Experiments

Based on experimental HPRTC results, Gutiérrez et al. [10] showed that light oils did not generate much solid residue at the temperatures used in their experiments (400–450 °C). This is significant, as stable peak temperatures in most light oil HPCT tests are typically below 300 °C, which led them to suggest that thermal cracking was likely not an important source of fuel during HPAI.

Those HPRTC experiments were history-matched and yielded important conclusions. The final simulation model was capable of accurately representing variables such as core temperatures, fluid production, pressure drop, composition of the produced gas, properties of the produced oil (i.e., density and viscosity), and the amount of the residual phases in the post-test core [13].

Figure 5 and Figure 6 illustrate the simulated core temperatures of the HPRTC tests performed on Oil I and Oil K, respectively. These figures highlight endothermic events, which are associated to the evaporation of water. Since the oil is only allowed to be in the liquid phase, it is clear that such endothermic responses are not related to the distillation of the liquid hydrocarbon. This confirms that the oil can be modeled as a dead liquid.

More importantly, Figure 7 and Figure 8 show the simulation results of the residual phases in the post-test core as well as the mole fractions of generated gases, which are products of thermal cracking. For these experiments, such residual phases mostly refer to the amount of solid residue (residue pseudo-component) since the amounts of residual oil and water are very small. Also, the operation of the test only allows measurement of the phases within the core at the end of the experiment (i.e., one data point). Nevertheless, it is interesting to see the predicted variations of the different phases within the core with time.

The most relevant aspect of Figure 7 and Figure 8 is that the generation of solid residue appears to take off above 300 °C. Similarly, the generation of gases also occurs at temperatures above that level. This suggests that thermal cracking reactions are likely not an important mechanism for fuel generation during the HPAI process, wherein stable peak temperatures are typically below 300 °C.

The practical implication of this discussion is that such reactions could be ignored during the simulation of HPAI, which would have a positive effect on the computing time for HPAI field models. In order to assess this, HPRTO and HPCT models were run in the absence of such reactions.

3.2. Observations from HPRTO Experiments

As mentioned previously, since peak temperatures typically observed during HPAI experiments are below 300 °C, thermal cracking is not expected to be a major reaction mechanism during the process. In order to assess this, the HPRTO models were run in the absence of thermal cracking reactions (TCR) for the two light oils.

The simulation results obtained for the two oils are, for practical purposes, equivalent to those obtained when thermal cracking reactions were active in the models, which were previously reported [12]. The main difference, for the HPRTO experiments, is associated with the production of the gas pseudo-component, which is absent when thermal cracking reactions are de-activated. This occurs because, in these models, the gas pseudo-component is assumed only to be a product of thermal cracking reactions (Equations (3) and (4)). Nevertheless, the amount is so small that the impact can be ignored.

Figures illustrating all of the simulation results of the HPRTO experiments for the two oils, when ignoring thermal cracking reactions, have been presented by Gutiérrez [13]. Clearly, in the absence of thermal cracking, the gas pseudo-component is not produced. In the interest of brevity, such figures are not shown here.

Perhaps the most interesting aspect to study is the role played by the combustion of solid residue (CSR), which historically has been assumed to be the main source of energy during the ISC process, including during the application to light oils.

3.2.1. HPRTO Oil I

Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14 and Figure 15 illustrate the simulation results of the HPRTO experiment with Oil I after ignoring thermal cracking reactions (Equations (3) and (4)) and the combustion of solid residue (Equation (7)). They include the results of injection pressure (Figure 9), core temperatures (Figure 10), fluid production (Figure 11), gas composition (Figure 12 and Figure 13), properties of the produced oil (Figure 14), and residual phases in the post-test core (Figure 15).

Clearly, the simulation results mimic the experimental data well. As expected, mismatches begin to occur at temperatures higher than approximately 300 °C, which can be observed from the core temperatures (Figure 10). Also, the absence of the combustion of solid fuel results in the underestimation of the water produced (Figure 11), which also occurs after approximately 300 °C. Another difference can be observed from the produced gas composition. The absence of thermal cracking reactions results in missing the production of the gas pseudo-component (Figure 13). Similarly, the absence of CSR results in unreacted oxygen and lack of production of carbon oxides (Figure 12). In other words, oxygen is not reacting with solid fuel and therefore is not generating carbon oxides. This effect also occurs at temperatures higher than approximately 300 °C. Finally, Figure 15 shows that the model overestimates the amount of residue pseudo-component left in the post-test core (3.05 vs. 0.05 g) due to the absence of the reaction responsible for its consumption (CSR).

The discrepancies observed between the experimental data and simulation results, when ignoring TCR and CSR, do not seem significant enough to require the inclusion of such reactions in the kinetic model. However, this will be assessed during the simulation of HPCT, which is more representative of the HPAI process as it occurs in the field.

3.2.2. HPRTO Oil K

Figure 16, Figure 17, Figure 18, Figure 19, Figure 20, Figure 21 and Figure 22 illustrate the simulation results of the HPRTO experiment with Oil K after ignoring thermal cracking reactions (Equations (3) and (4)) and the combustion of solid residue (Equation (7)). They include the results of injection pressure (Figure 16), core temperatures (Figure 17), fluid production (Figure 18), gas composition (Figure 19 and Figure 20), properties of the produced oil (Figure 21), and residual phases in the post-test core (Figure 22).

Clearly, the simulation results mimic the experimental data very well. As expected, mismatches begin to occur at temperatures higher than approximately 300 °C. Since peak temperatures in all thermocouple occurred before the heater reached 300 °C, the simulated temperatures match very well for all thermocouples (Figure 17). However, the absence of the combustion of solid fuel results in the underestimation of the water produced (Figure 18), which occurs after approximately 300 °C. Another difference can be observed from the produced gas composition. The absence of thermal cracking reactions results in missing the production of the gas pseudo-component (Figure 20). Similarly, the absence of CSR results in unreacted oxygen and lack of production of carbon oxides (Figure 19). In other words, oxygen is not reacting with solid fuel and therefore is not generating carbon oxides. This effect also occurs at temperatures higher than approximately 300 °C. Finally, Figure 22 shows that the model overestimates the amount of residue left in the post-test core (3.21 vs. 0.48 g) due to the absence of the reaction responsible for its consumption (CSR).

3.3. Observations from HPCT Tests

The impact of ignoring thermal cracking reactions (TCR) and combustion of solid residue (CSR) was also assessed by simulating the HPCT experiments.

3.3.1. HPCT Oil I

Figure 23, Figure 24, Figure 25, Figure 26, Figure 27, Figure 28, Figure 29, Figure 30 and Figure 31 illustrate the simulation results of the HPCT experiment with Oil I after ignoring thermal cracking reactions (Equations (3) and (4)) and the combustion of solid residue (Equation (7)). They include the results of injection pressure (Figure 23), core temperatures (Figure 24, Figure 25 and Figure 26), fluid production (Figure 27), gas composition (Figure 28 and Figure 29), properties of the produced oil (Figure 30), and residual phases in the post-test core (Figure 31).

The simulation results match the experimental data well. They are very similar to those reported by Gutiérrez et al. [12] when using all six reactions (Equations (3)–(8)).

Regarding core temperatures, the main mismatch occurs after approximately 7 h during the HPCT test (Figure 26), wherein the simulated combustion front moves slightly faster than during the actual experiment. During the test, it appears that a liquid bank (oil and water) builds up ahead of the combustion front (See Figure 27), which results in the reduction of gas saturation and hence in reduced gas relative permeability in that region. This results in a reduction in the velocity of the thermal front. This effect is not captured by the model, which exhibits a relatively constant combustion front velocity.

Regarding the composition of the produced gases (Figure 28 and Figure 29), it is noted that the concentration of the gas pseudo-component (Figure 29), as indicated by the model, is zero. This is because such a pseudo-component is assumed to be a product only of thermal cracking reactions, which are currently being ignored.

3.3.2. HPCT Oil K

Figure 32, Figure 33, Figure 34, Figure 35, Figure 36, Figure 37, Figure 38, Figure 39 and Figure 40 illustrate the simulation results of the HPCT test with Oil K after ignoring thermal cracking reactions (Equations (3) and (4)) and the combustion of solid residue (Equation (7)). They include the results of injection pressure (Figure 32), core temperatures (Figure 33, Figure 34 and Figure 35), fluid production (Figure 36), gas composition (Figure 37 and Figure 38), properties of the produced oil (Figure 39), and residual phases in the post-test core (Figure 40).

The simulation results match the experimental data well. In fact, they are very similar to those reported by Gutiérrez et al. [12] when using all six reactions (Equations (3)–(8)). This suggests that a similarly good match can been achieved without including such reactions in the first place, which would simplify the work to derive the kinetic model.

Regarding the core temperatures, and similar to the case of Oil I, the main mismatch occurs after approximately 7 h during the HPCT test (Figure 35), wherein the simulated combustion front moves slightly faster than during the actual experiment. During the test, it appears that a liquid bank (oil and water) builds up ahead of the combustion front (see Figure 36), which results in the reduction of gas saturation and hence in reduced gas relative permeability in that region. This results in a reduction in the velocity of the thermal front. This effect is not captured by the model, which exhibits a relatively constant combustion front velocity.

Regarding the composition of the produced gases (Figure 37 and Figure 38), it is noted that the concentration of the gas pseudo-component (Figure 38), as indicated by the model, is zero. This is because such a pseudo-component is assumed to be a product only of thermal cracking reactions, which are currently being ignored.

4. Discussion

The HPAI process had previously been described and modeled using six reactions (Equations (3)–(8)), which included thermal cracking reactions, oxygen addition reactions, oxygen-induced cracking reactions, and bond scission reactions [12]. This last group includes the combustion of a flammable vapor, as well as the combustion of solid fuel. However, the results from this work suggest that it might be possible to simplify the model significantly by omitting thermal cracking reactions (Equations (3) and (4)) as well as the combustion of solid fuel (Equation (7)), which is not unexpected based on observations from a variety of experiments [10].

4.1. The Role of Thermal Cracking Reactions

High-pressure ramped temperature cracking experiments indicate that the amounts of vapor and solid fuel created by thermal cracking of a light oil sample tend to be very small. Furthermore, modeling of those tests indicate that such gas and solid fractions begin to appear at temperatures higher than approximately 300 °C (see Figure 7 and Figure 8), which is the stable peak temperature commonly observed during the HPAI process. Therefore, that reaction mechanism is not expected be an important source of fuel for the process and does not seem to play an important role during HPAI.

The above discussion implies that for HPAI cases in which stable peak temperatures are below approximately 300 °C, a new kinetic model could be developed by ignoring such a reaction mechanism. Moreover, if the amount of the gas pseudo-component is small (as in the cases discussed), this also means that such a pseudo-component could be completely removed from the model. However, if the presence of such gases is important, they could be included as a product of oxygen-induced cracking reactions (Equation (6)), which are also a source of those light gases and tend to occur at lower temperatures than those of thermal cracking reactions.

The fact that removing thermal cracking reactions from the kinetic model did not have much impact on the history matching of the HPRTO and HPCT experiments appear to confirm the above conclusion.

Additionally, if the amount of asphaltenes in the original oil sample is small, as in the cases considered in this study, the asphaltene pseudo-component could also be ignored and removed from the fluid model.

4.2. The Role of Combustion of Solid Fuel

Different HPAI oxidation and combustion experiments result in stable peak temperatures below approximately 300 °C, which is beneath the temperature required for the combustion of solid fuel (Equation (7)), commonly known as coke. This suggests that the reaction mechanism may not be important for the process, which was confirmed by the numerical simulation results presented previously.

As with thermal cracking reactions, it appears that as long as the HPAI stable peak temperatures are below approximately 300 °C, the combustion of solid fuel is not an important mechanism during HPAI.

This implies that, for such cases, a new kinetic model could also be developed by ignoring such a reaction mechanism. Moreover, if the amount of solid residue encountered during the tests is sufficiently small, the residue pseudo-component could be completely ignored and removed from the model. This would also bring an added benefit to the numerical simulation of field-scale models.

4.3. The Role of Oxygen Addition Reactions

These reactions (Equation (5)) correspond to the oxidation process mentioned previously. They occur mostly within the liquid oil phase and can occur over a wide temperature range. Nevertheless, they tend to dominate at lower temperatures. In general, these reactions involve oxygen binding with hydrocarbon molecules to produce oxygenated species such as aldehydes, alcohols, ketones, and hydroperoxides, and they also generate heat. These new species tend to further react and polymerize with each other, forming heavier, less desirable compounds such as asphaltenes and, eventually, a solid residue.

In the context of this study, such reactions result in a heavy pseudo-component that was called pseudo-asphaltenes. This oil fraction is important because it has the ability to capture potential increases in the density and viscosity of the oil, in the case that the process is not operated under appropriate conditions (e.g., under suboptimum air injection rates) [11]. Additionally, cracking of the pseudo-component is believed to be the source of lighter hydrocarbon fractions, which are considered to be the main fuel used by the HPAI process. These reactions cannot be ignored.

4.4. The Role of Oxygen-Induced Cracking Reactions

These reactions, also known as oxidative cracking reactions (Equation (6)), are similar to thermal cracking reactions but occur in the presence of oxygen. In the presence of oxygen, the same thermal cracking reactions become more thermodynamically favorable and occur at lower temperatures than in cases where oxygen is absent, resulting in the generation of light hydrocarbon fractions [20,21].

In the context of this study, this reaction is represented by the thermal cracking of pseudo-asphaltenes, which results in the generation of lighter hydrocarbon components as well as solid residue. The main hydrocarbon product of interest to this reaction is the flammable pseudo-component, which is believed to be the main fuel during the HPAI process. The other two potential products of the reaction, gas and residue pseudo-components, are considered to be optional, depending on the characteristics and needs of the particular case under study. These reactions cannot be ignored.

4.5. The Role of Combustion of Vapor Fuel

Combustion of the flammable pseudo-component is the main mechanism of heat generation during the HPAI process and cannot be ignored. This is a traditional combustion reaction that involves the destructive oxidation of the hydrocarbons to produce carbon oxides, water, and heat (Equation (8)).

This flammable pseudo-component is actually a complex mixture of hydrocarbons, which is modeled in a rather simplistic way in this study. For further research, it is recommended to investigate the nature and composition of such fuel in order to represent the process in more detail, which may enhance the predictive capabilities of such models.

4.6. The Dynamics of the Reactions during the HPAI Process

As oxygen is injected into the reservoir, it has the option to react with either the hydrocarbon gases or the liquid oil. However, the composition of the original reservoir gas (typically rich in methane) is such that the oxygen–gas mixture is likely outside of the flammable range (i.e., fuel rich) and below the ignition temperature, at reservoir conditions. Therefore, the injected oxygen is more likely to react with liquid oil first.

As oxygen enters into solution in the liquid oil, it will react mostly through oxygen addition reactions, which generate heat. At lower temperatures, this heat can trigger oxygen-induced cracking reactions that generate flammable vapors, which may become fuel for the combustion process. This is why oxygen addition reactions—commonly known as low-temperature oxidation (LTO) reactions—are so important for ignition.

In the case of light oils originating from reservoirs that are initially at a high temperature and pressure, oxygen addition reactions result in a moderate increase in the light oil viscosity. However, oxygen-induced cracking reactions generate flammable vapors that can ignite at moderately higher temperatures than the initial reservoir temperature. Also, the amount of heavier fractions (such as asphaltenes) is not as high when compared to heavier oils, and the properties of the oil are such that they can remain in solution and flow downstream even in the event of an increase of oxygen addition reactions. Therefore, for light oils at lower temperatures, the “competition” for oxygen is typically dominated by bond scission reactions over oxygen addition reactions. This may explain why the HPAI process in lighter oils can be successfully operated at relatively low temperatures (i.e., below approximately 300 °C) as compared to heavier oils.

Given the analysis presented above, the HPAI process can potentially be described by the following three reactions, instead of six.

• Oxygen Addition Reactions

Maltenes + Oxygen → Pseudo-Asphaltenes + Energy

(9)

• Oxygen-induced Cracking Reactions

Pseudo-Asphaltenes → Residue + Flammable_{(liquid + gas)} + Gas

(10)

• Bond Scission Reactions

Flammable_(gas) + Oxygen → CO₂ + CO + H₂O + Energy

(11)

The above analysis has been performed on laboratory experiments and models, which need to be validated at the field scale. Based on experimental combustion tube data performed at representative reservoir pressures, it is expected that peak combustion temperatures in the field will also be of the same order (i.e., around 300 °C). Nevertheless, these experiments do not properly capture the impact of time on the dynamics of the process, which is very important at the field scale, especially regarding oxygen addition reactions. Such reactions play an important role during ignition and may be important ahead of the combustion front should oxygen bypass the main reaction zone. Therefore, it is recommended to validate these results at the reservoir scale during a field HPAI operation.

5. Conclusions

This work presents the simulation results of a variety of thermal cracking, oxidation, and combustion experiments performed on two light oils at representative reservoir pressure conditions. One reservoir sample comes from a sandstone reservoir and the other comes from a carbonate formation. Such experiments are meant to study the performance of the high-pressure air injection (HPAI) process, and their modeling allows for the identification of the most relevant types of reactions taking place. The study reveals that if stable peak combustion temperatures are below approximately 300 °C, thermal cracking and combustion of solid fuel are not important reaction mechanisms for the HPAI process and can potentially be ignored, which would have a positive impact on the computing time during field-scale reservoir simulations. Most importantly, the simulation results suggest that the main fuel for the HPAI process is a flammable vapor generated by oxidation reactions. This represents a shift from the original in situ combustion paradigm, which is based on the combustion of solid fuel. Nevertheless, these results appear to be valid at the laboratory scale and should be validated at the reservoir scale during a field HPAI operation.