Steam assisted gravity drainage (SAGD) has become the main technology for the development of super-heavy oil reservoirs and has been applied industrially at a large scale in China and abroad^[1]. In order to apply SAGD successfully, it is necessary not only to ensure that the steam quality at the bottomhole conditions is greater than 70%, but also to ensure the maximum steam sweeping volume. At the same time, in terms of technology, the production rate of production well should match with the natural oil drainage of steam chamber, ensuring the highest oil displacement efficiency by steam and thermal efficiency^[2]. Under the SAGD operation mode, on the one hand, the high quality steam overlap makes the steam chamber develop rapidly upward, and the overburden layer will be heated when it reaches the top of the oil layer, resulting in serious waste of the injected heat^[3]. On the other hand, due to the heterogeneity of the reservoir, the expansion of the steam chamber is extremely nonuniform, which reduces the swept volume and thus decreases the oil recovery^[4]. In view of the above problems, the author started to study the gas-assisted SAGD technology for super-heavy oil development in 2006, including nitrogen^[5,6], carbon dioxide^[7], flue gas and air.

In 2011, Oskouei et al.^[8] carried out the study on the feasibility of air injected into the steam chamber at the end of SAGD using two-dimensional physical model and the result shows that the recovery of air injection after SAGD can increase by 10%. In 2011, Rahnema et al.^[9] conducted the combustion drive assisted gravity drainage (CAGD) experiment. The result indicates that the air injection would cause coking in the horizontal wellbore due to the intense well spacing of the dual horizontal wells, which would lead to various problems such as wellbore plugging and affect stability of the air injection process. Rahnema et al.^[10] continued to carry out the three-dimensional physical experiment of air injection in the end of SAGD, and finally observed the stable high temperature front of 617 °C. In addition, they proved that the recovery of air injection after SAGD can be improved by 12%. In 2013, Oskouei et al.^[11] conducted the three-dimensional physical experiment of air injection in the end of the dual horizontal well SAGD, and proved that under the condition of 245 °C, high temperature combustion of heavy oil can be realized, and the ultimate recovery increases by 10%. Many Chinese scholars also studied the technology of air injection enhanced oil recovery. In 2014, Ma DS et al.^[12] carried out a study on air injection in the mid and late period of SAGD with combination of dual horizontal wells, and the result manifests that 50% of the reserves can be produced continually by air injection. Jiang Youwei, Li Songlin, Wang Zhengmao et al. proved that low temperature oxidation of air injection can improve the development performance of low permeability reservoirs^[13], light oil reservoirs^[14,15] and heavy oil reservoirs after cold production. In addition to conventional displacement, Liang Jinzhong^[16], Wang Y et al.^[17] explored the application of air injection in huff and puff. In this paper, the feasibility of low temperature oxidation (LTO) or high temperature oxidation (HTO) of air injection in the combination of vertical and horizontal wells at the end of SAGD is investigated, taking the SAGD test area of vertical and horizontal wells combination in Xing Ⅵ Formation, Du 84 block, Liaohe oilfield as an example.

The oxidation characteristics of super-heavy oil under different heating rates and gas injection rates were studied using TGA/DSC simultaneous thermal analyzer according to the particularity of oxygen existing in the air. Meanwhile, the kinetic parameters in the reaction process of crude oil and air were calculated based on the heat released and the lost mass of crude oil in the oxidation process^[18]. Experimental parameters are as follows: (1) The heating rates in the experiment are 2 °C/min and 5 °C/min. (2) The air injection rates are 30 mL/min and 50 mL/min. (3) Temperature range is 25-600 °C and 25-350 °C, which can meet the needs of research on low temperature oxidation characteristics of the crude oil. (4) The quality of crude oil sample put into the crucible is controlled at about 20 mg. Start the test program and the thermal gravity analysis curve (TG), the differential thermal gravity curve (DTG) and the differential scanning calorimetry curve (DSC), which are collectively called TGA/DSC curves, in the given temperature range can be obtained at the end of the experiment.

1.1.1. Oxidation of super-heavy oil at different temperature ranges

Fig. 1 shows the TGA/DSC curves of the super-heavy oil sample at different heating rates. It can be seen that the trends of TGA and DSC curves are similar at different heating rates. The heat flux of the sample is stable under 200 °C, showing that the low temperature oxidation reaction rate in this temperature range is very slow, and there is no exothermic display in the DSC curve. With temperature of 200 to 350 °C, a exothermic peak appears in the DSC curve, indicating that the reaction rate of LTO speeds up^[19]. In the temperature range from 350 to 480 °C, steps and fluctuations occur simultaneously in the DSC and TGA curves, which indicates that it is mainly the cracking reaction that happens in the super- heavy oil within the temperature range which is generally called the negative temperature gradient range^[20]. When the temperature is between 480 and 540 °C, there is obvious exothermic peak appearing in the DSC curve and the peak value is very high, showing that the intense oxidation reaction takes place in this range and releases a lot of heat. The released heat rises with the increase of heating rate. And this temperature range corresponds to the combustion process of crude oil^[21].

1.1.2. Oxidation of super-heavy oil at different gas injection rates

Fig. 2 shows the TGA/DSC curves of the super-heavy oil sample at different gas injection rates. The trends of TGA curves under the two given gas injection rates in the experiment are similar, and the weight loss steps are basically the same, which manifests that gas injection rate has little impact on the super-heavy oil characteristics of low temperature oxidation below 350 °C. Parameters of the activation energy and pre-exponential factor are determined by the experiment, and they characterize the oxidation characteristics of crude oil and are used for the subsequent numerical simulation. The activation energy can be read as 23.11-27.71 kJ/mol from the experimental instrument, and the pre-exponential factor is 0.092-0.280^[22].

1.2.1. Viscosity

The purpose of the experiment is to determine the variation of sample viscosity after oxidation of super-heavy oil sample at different temperatures and air to oil ratios (AOR) under reservoir pressure (4.0 MPa). The viscosities of super-heavy oil under the conditions of three oxidation temperatures (150, 200, 250°C), and three air to oil ratios (2.5, 6.0, 12.0 cm³/g) are tested by StessTech rheometer (Table 1). It can be seen that under the same air to oil ratio, with the increase of oxidation temperature, the oxidation degree of the super-heavy oil intensifies, and the viscosity of the super-heavy oil grows up greatly. At the same temperature, when the air to oil ratio rises, the oxidation degree of the super-heavy oil increases, indicating that the oxidation degree of super-heavy oil is also related to the air to oil ratio.

1.2.2. SARA component

The SARA components of super-heavy oil samples under different oxidation temperatures are tested (Table 2). It is observed that oxidation reaction of the super-heavy oil sample is able to occur at low temperature (150-250 °C) and the property changes greatly, showing that in terms of SARA components, the saturates and aromatics decrease while the asphaltene increases. The super-heavy oil deposits on the rock after oxidation reaction, with poor viscosity and high content of asphaltene. Fig. 3 shows the comparison of appearance of the oil sample before and after the low temperature oxidation. Thereinto, the appearance of the oxidized sample is the experiment result under the condition of temperature of 250 °C and air to oil ratio of 12.0 cm³/g.

1.2.3. Core permeability

Oxidation experiments at different temperatures are carried out for the actual core samples. Core preparation method: samples are selected before and after oil flooding, and the parts near the exterior are selected to grind the core to observe the characteristic of pore structure. Selection principle: considering the difference between the inside and outside of the sample in the oxidation experiment and oil washing extraction, the part near the exterior of the sample is selected. The experiment temperature is 150 °C and 250 °C respectively and the pressure is 10 MPa.

Air injection can change the properties of the filling instead of the pore structure. Air injection tends to age the super- heavy oil samples, resulting in the increase of asphaltene and non-hydrocarbon components that are not easily extracted, and the darker color of residue in the slice. Meanwhile, the asphaltene is easy to block the originally connected pore throat (Fig. 4).

The test results of core permeability (Table 3) indicate that core permeability is basically unchanged after steam injection and core permeability is reduced by 5%-8% after the reaction of steam and air injection.

For the core and super-heavy oil in the target layer, the experiment of displacement efficiency of steam flooding with the oxidized super-heavy oil under various temperatures is carried out under the constant temperature condition by using the one-dimensional long core flooding device. The length of core is 30.0 cm with a diameter of 3.0 cm. The porosity is 36.8% and the water phase permeability is 1.91 μm². First, the super-heavy oil is oxidized under the condition of AOR of 12.0 cm³/g to get the experimental oil samples of different oxidation temperatures. Then, steam flooding experiment is conducted at the temperature of 250 °C. The experimental results are as shown in Table 4. It can be seen that different oxidation temperatures have significant influence on the oil displacement efficiency of steam flooding of super-heavy oil. In the experimental temperature range, oil displacement efficiency of steam flooding decreases by 16.54%-23.89% and residual oil saturation increases by 12.50%-18.17% under temperature of 250 °C with oxidized super-heavy oil. The main reason is that the SARA components varied greatly (Table 2) after the oxidation reaction of the super-heavy oil. On the one hand, the viscosity of super-heavy oil increases (Table 1). On the other hand, the content of asphaltene in the super-heavy oil greatly adds, resulting in the enhanced adhesion of the super-heavy oil on the rock, and consequently making more driving force be needed when the steam drives the residual oil. At the same time, the saturated hydrocarbon (light component) in the super-heavy oil decreases significantly, indicating that the steam distillment is weakened during steam flooding.

Based on well logging data and seismic interpretation data, a three-dimensional heterogeneous geological model (Fig. 5) is built with Petrel software. There are 4 horizontal wells and 28 vertical wells in the model, and the main reservoir parameters are shown in Table 5.

According to the kinetic parameters of oxidation reaction (reaction activation energy, pre-exponential factor, etc.) of the oil obtained by experiment and considering the low temperature oxidation and high temperature combustion models commonly used in combustion simulation^[21,22], the air injection oxidation reaction model suitable for the reservoir is established, as shown below.

(1) Thermal cracking reaction:

(2) Low temperature oxidation:

(3) High temperature oxidation:

According to the characteristics of temperature distribution in the end of SAGD (Fig. 6), and considering that the remaining oil mainly distributes in the upper part of the reservoir, a scheme that re-perforate in the upper section of the vertical steam injection well is designed, using the STARS multi-component multi-phase numerical simulation software. In the scheme, air is continuously injected in the vertical well, and high temperature combustion is achieved by heating the air through the igniter. The well pattern in the air injection period is shown in Fig. 7. Three development schemes are simulated, namely low temperature oxidation of steam and air, low temperature oxidation of air and high temperature combustion of air. The grid size of both I and J directions in the model is 5 m, while that of K direction is 2-3 m. The total number of grids is 17 952.

2.2.1. Comparison of low temperature oxidation of steam and air co-injection with that of air injection

The simulation parameters are as follows: steam injection rate is 100 m³/d per well; the steam quality at the well bottom is 70%; air injection rate is 10 000 m³/d; the production rate of horizontal well is 420 m³/d and the production-injection ratio is 1.2. The simulated dynamic production curves of LTO of steam and air and LTO of air are shown in Fig. 8, and the Table 6 is the statistics of production effect. It can be seen that the scheme of steam and air has better production results than that of air and the recovery is higher, while the oil and steam ratio of the former is low and the economic benefit is poor.

By comparing the distribution of remaining oil after the production of the two development schemes (Fig. 9), it can be found that the oil displacement efficiency in the lower part of the steam + air model is higher, because the steam condenses into hot water flowing near the horizontal well, which maintains the temperature in the area, thus improving the oil mobility. However, a large amount of residual oil still exists in the area of the air model.

Comparing the gas streamlines of the same section of the two development schemes (Fig. 10), it is observed that in the steam + air model, the steam is injected in the middle and lower part of the reservoir, which maintains the high pressure of this area. So the air is forced to flow into the horizontal well through the remaining oil belt from above, which increases the swept volume of air. However, in the air model, air flows around the remaining oil area, resulting in inefficient circulation, which affects the oil displacement efficiency of the whole process.

2.2.2. Comparison between low temperature oxidation and high temperature combustion

The simulated air injection rate is 20 000 m³/d per well. The simulated production curves of low temperature oxidation and high temperature combustion are shown in Fig. 11. Table 7 is the statistics of production performance. It can be seen that the effect of high temperature combustion is far better than that of low temperature oxidation. Not only is the air to oil ratio of the high temperature combustion lower than that of low temperature oxidation, but also the oil recovery rate of the high temperature combustion is faster and the final recovery factor of the high temperature combustion is higher.

Through comparison of the remaining oil distributions of after production of low temperature oxidation and high temperature combustion (Fig. 12), it can be seen that in the high temperature combustion mode, the oil saturation in the area displaced by combustion front is extremely low, and the remaining oil mainly is located on the top of the reservoir and the inter-well area which is not affected by air^[23]. While in the low temperature oxidation mode, the oil saturation near the injection well is relatively low due to enough air injection and sufficient oxidation. Although the area about 20 m away from the injection well is also displaced by air, the oil saturation in this area is high because of low oil displacement efficiency.

From the comparison of temperature distribution of low temperature oxidation and that of high temperature combustion after production (Fig. 13), it is observed that in high temperature combustion mode, swept region can be heated to above 400 °C and the temperature of combustion front can reach 550 °C. Meanwhile, the temperature of the whole reservoir is very high. While in the low temperature oxidation mode, the high temperature regions mainly distribute near the injection well, and the temperature of most of the swept reservoir is 100-150 °C.

It can be seen from the coking zone distributions of low temperature oxidation and high temperature combustion after production (Fig. 14) that the coking zone in the low temperature oxidation mode is large, and mainly distributes near the oxidation front. While the coking zone of high temperature combustion mode is very little and mainly distributes at the bottom of the reservoir, indicating that the combustion front has spread near the production well.

The temperature of the steam chamber formed in the super-heavy oil development process by SAGD of Xing VI Formation in Du 84 block of Liaohe Oilfield is 150-250 °C, during which low temperature oxidation and coking can occur when air is injected, shown by experimental study. The coking crude oil deposits on the rock causing the decrease of the permeability. After low temperature oxidation, the properties of oil become worse with great increase of viscosity, decrease of content of saturated hydrocarbon and aromatic hydrocarbon and increase of the content of asphaltene. Consequently, the oil displacement efficiency decreases by 16.5%-23.9%. The numerical simulation shows that when air is injected during steam injection, coking zone is formed around the steam chamber due to low temperature oxidation, which prevents expansion of the steam chamber, thus resulting in the decrease of production, low oil steam ratio and poor economic benefit. When the high temperature combustion technology is adopted in the end of SAGD, the oil production is higher, with lower air to oil ratio and better economic performance. Therefore, it is suggested that for the Xing VI Formation of Du 84 block, the high temperature combustion is more favorable in the end of SAGD to further improve oil recovery.