There are many influence factors about fire flooding to be considered. They can be summarized in three aspects as follows: reservoir geological conditions, reservoir development conditions and engineering. These three aspects cover almost all influence factors of fire flooding. Most of them are not controllable and some need to be adjusted and reformed. All of them can impact the effect of fire flooding more or less.

Reservoir geological factors are uncontrollable factors in the fire flooding process, which need to be overcome by strict screening criteria. Generally, the reservoir screening quantitative geological factors of fire flooding need to be considered are buried depth, thickness, dip, permeability, and fluid properties, etc.

In heavy oil reservoirs at a later stage of steam stimulation, long time of steam stimulation will lead to the change of reservoir parameters, so it is necessary to discuss the effect of reservoir development on fire flooding. Reservoir pressure, oil saturation and formation temperature can beused to characterize the reservoir development status after steam stimulation or other recovery methods [5, 6]. Lower reservoir pressure is beneficial for fire flooding. After years of steam flooding, a large amount of thermal energy has been added to the formation and the water saturation has increased at the same time.

Operation process of fire flooding mainly through the mediation of fire flooding engineering parameters [7] to realize the control of the underground combustion, air injection rate, water/gas ratio and injector to producer well ratio are the main operation parameters of conventional fire flooding.

Correlation analysis is the significant content in the grey theory created by professor Deng Julong [8], is a kind of uncertain analysis method, the essence of which is convey the diffidence among the geometry of the curve, based on the difference size to determine the degree of close [9]. Various influence factors in the process of fire flooding have different degrees on effect and the relationship between the factors are not entirely clear. So it can be regarded as an information incomplete system, namely grey system.

From the published literature [2, 10, 11], collected 34 cases of fire flooding both at home and abroad, including primary development, fire flooding development after steam stimulation, dry and wet fire flooding, and relevant parameters involved in three categories of geology, exploitation and engineering, and it can sum up to 14 items, the sample has universality and representativeness.

In the grey system of fire flooding, the relationship between a variety of factors (comparison array) and fire flooding effect (reference array) need to be determined ultimately, there are many factors affecting the fire flooding effect, through the analysis, buried depth (Z), thickness (h), dip (θ), porosity (ϕ) and permeability (K), reservoir temperature (T), reservoir pressure (P_r), oil saturation (S_o), crude oil density (D), crude oil viscosity (μ), mobility (Kh/μ), gas injection velocity (v), water/gas ratio (WOR) and storability (ϕS_o), a total of 14 representative factors are selected as comparison sequence. Air/oil ratio (m³/m³) which represents the fire flooding effect is taken as reference sequence of grey relation analysis.

According to close degree obtained from sample data, influence factors are classified, and the results are shown in Table 1. Analyze factors of different classes, and then screen out main factors, which are taken as key consideration in the next step of block selection and engineering operation process.

According to the results of correlation analysis, select close and close level parameters as main influence factors of in-situ combustion. The geological influence factors of in-situ combustion include depth, mobility, permeability, porosity and oil saturation; developing factors mainly consist of reservoir temperature and storability; engineering factor is mainly the average gas injection rate.

The above discussions are based on the analysis conclusion of influence factors, and all analysis is for single factor that lack of a lateral comprehensive comparison of various factors. For a specific reservoir, the final fire flooding effect will be decided by one or a combination of several factors, so it is necessary to establish a fire flooding block selection mode based on the level of influence factors.

According to characteristics of heavy oil reservoir, we classify the list heavy oil reservoirs, the block diagram which is suitable for the in-situ combustion is shown in Fig. (1), and it was divided into three types, preferred reservoir, second reservoir, unsuitable reservoir.

Fig. (1). Screening reservoir type diagram.

The preferred reservoir types include the dip reservoir, the edge water reservoir and thin layer reservoir. Structure dip is the most important factor for combustion wells location selection, and the migration rate of gas injection and combustion front is faster in updip direction than downdip, so gas injectors are usually located in the structure of high position. Moderate reservoir thickness can avoid gas overlap and excessive gas heat loss of top and bottom, generally between 1.5 and 15 m. The only requirement for permeability is that air can be injected to the formation under the pressure of cap rock condition allowing, and the minimum requirement for oil property is the mobility in the underground.

The second reservoir include top and bottom water reservoir, edge and bottom water reservoir, the layered oil reservoir of medium thickness and the massive reservoir. For the second reservoir, risk will increase in the process of in-situ combustion, but successful examples [12-17] at home and abroad are still existed. Massive thick and layer thick reservoirs can improve the development effect in horizontal well by in-situ combustion, which should be considered in the economic evaluation. Top and bottom water reservoir as well as edge and bottom water reservoir must use determinant well pattern. The screening criteria should be improved for the next round of screening of the second reservoir.

The inappropriate reservoirs for in-situ combustion mainly are those with gas cap, active bottom water and fractures, and successful examples at home and abroad are very few, even field operation is very difficult and those are not recommended as candidate reservoir blocks.

In-situ combustion can be applied to natural extraction and steam stimulation. And strong waterflooding reservoirs are unsuitable for in-situ combustion, but if just a short period of time, in-situ combustion can go on. Details are shown in Fig. (2).

Fig. (2). In-situ combustion adaptability analysis diagram of reservoir development history.

Steps of screening block according to the reservoir and fluid parameters are follows;

1. Fuel Content: The fuel content of the candidate reservoir should be within the scope of 13~ 45 kg/m³.
2. Reservoir Connectivity: Reservoir has good lateral connectivity, namely the objective layer is relatively stable in work area, lithofacies and petrophysics change little.
3. Reservoir Heterogeneity: There are no thief zones in vertical, permeability variation coefficient is less than 0.75.
4. Reservoir and Fluid Physical Properties: The properties can be compared obtained based on successful examples of in-situ combustion at home and abroad, also can refer to data from Table 1, which is on the basis of formers' research results. According to the actual production and Support Vector Machine (SVM) [4] to establish the preferred and second reservoir model, then calculate the reservoir screening parameters in Table 2.

In order to reflect the interaction effects between various factors and the comprehensive effect, Chu [10] in 1977, researched the regression function of Y, Y is a function of reservoir buried depth, porosity, permeability, initial oil saturation and oil viscosity. Because Y function is convenient and the result of calculation is reliable. So it is often taken as a necessary reference in fire flooding reservoir screening. But Y function does not consider the engineering factors. There will be differences from the actual situations. Therefore, a new function is proposed. In order to distinguish it from Y function, it is called J function in this function regression. According to the correlation analysis, reservoir temperature and injection rate are added to the regression independent variable as close parameters, J is multi-variable linear function of reservoir buried depth, temperature, permeability, mobility, gas injection rate and coefficient of storage .

Giving regression analysis of J function using Marquardt method of Matlab software, and J value is defined as follows: (1) J = 1, scheme is feasible in technology and economy; (2) J = 0, technically feasible, economically may fail; (3) J = -1, completely infeasible. The final equation is as following:

(1)

Where,

Z-Reservoir Buried Depth, m;

T-Reservoir Temperature,°C;

K-Average Reservoir Permeability,10^-3μm²;

h-Reservoir Thickness, m;

μ-Viscosity of Crude Oil, mPa.s;

v-Average Gas Injection Rate,×10³m³/d;

ϕ-Reservoir Porosity, Decimal;

So-Oil Saturation,%;

By simulation, multiple correlation coefficient of the formula is 0.9231. When using function J, selected parameters are plugged into the function J to calculate firstly, and screened fire flooding blocks according the J value, then analyzed rationality of fire flooding parameters, and finally evaluated fire flooding effect. Successful projects of statistical technique and economy, calculated the value of J is between 0.3 ~ 1.5. For J value that less than 0.3 is not successful or unsuccessful project economically, and the bigger value of J is more close to economic success as a whole.

In terms of reservoir types, D66, D48, M5 blocks with extra-low permeability and thin heavy oil reservoirs, J35, W82 medium-thick alternating layers heavy oil reservoirs are suitable for in-situ combustion, while micro-fracture and bottom water of CT reservoirs are not conducive to conduct in-situ reservoir combustion, however, as long as the crack growth is not strong, bottom water is inactive, it also can conduct in-situ combustion, and only better conditions of other factors can make sure the success of in-situ combustion. When the optional blocks structure exist high permeability channel, many reservoir fractures and large gas caps, they are not applicable for in-situ combustion, so filtered out directly and not entering the next round of screening.

In terms of reservoir types, applicability of the optional six in-situ combustion reservoirs from strong to weak order is: D66 and D48, M5, J35 and W82, CT.

Reservoir development history of each block are shown in Table 3. Reservoir development history of optional block mainly includes steam stimulation and waterflooding. From the research, reservoirs after steam stimulation or waterflooding are feasible for in-situ combustion, and the shorter production history, the better the effect of in-situ combustion. However the risk of in-situ combustion will increase in reservoirs after strong water flooding.

According to reservoir parameters of successful fire flooding reservoirs, the six blocks M5, D66, D48, W82, J35 were screened. And D66, D48 and M5 blocks are selected because their parameters mainly within the reservoir parameters of fire flooding succeed.

By using the J function to calculate the screening parameters of these three blocks, the results are in Table 4 and the J value of D48 block is 0.92. It means that fire flooding is more likely to succeed in this block and this is consistent with the results of support vector machine(SVM) method [18]. So, D48 block is chosen as the preferred block for in-situ combustion.