The data in this paper focus on logging interpretation, seismic interpretation, fracturing operation and dynamic monitoring data of 161 horizontal wells in the 1^st Member of Longmaxi Formation in Changning shale gas field. The logging interpretation data includes well path, reservoir property, rock mechanics, in-situ stress, the seismic interpretation data include the occurrence of natural fracture zone, the fracturing operation data include single well fracturing stage length, liquid and proppant strength, pumping rate, the dynamic monitoring data include micro-seismic monitoring results, test and cumulative production, EUR, which provide basic data for the study of the main controlling factors of shale gas well stimulation potential.

The study first extracted the required source factors (SF) and their derived factors (DF) as much as possible through logging and seismic interpretation data, which can be divided into matrix brittleness (MB), fracture propagation (FP) and reservoir properties (RP) (Table 1).

Derived factors refer to all kinds of parameters that are combined with source factors or represent reservoir geological characteristics indirectly. They include dynamic modulus of elasticity (E _dyn), Poisson’s ratio (ν), the total brittleness mineral mass fraction (V _Total), type I fracture toughness (K _I) and type II fracture toughness (K _II), natural fracture belt approaching angle (θ), in-situ stress difference (Δσ), micro natural fracture development index (W _NF), micro bedding development index (W _LM), the fracture pressure (P _p), critical net pressure (P _net), free gas volume (Gas _F), absorbed gas volume (Gas _A), vertical depth at the starting point (D _A) and ending point (D _B) of horizontal well fracturing stage, total gas content (Gas _T), vertical depth difference of horizontal well (ΔD, D _A is always subtracted from D _B to obtain the difference, which can distinguish updip well from downdip well).

We also give the calculation formulas of the above parameters to make the evaluation factors as complete as possible (Table 1). It is worth noting that the better the matrix brittleness factor is, the greater the brittleness is, the more beneficial to improve the complexity of the fracture network (Rickman et al., 2008). The better the fracture propagation factors are, the wider the hydraulic fracture propagation may be. The better the reservoir resource factor, the more resources can be recovered within the transformed volume. In more detail, we introduce a large number of parameters not involved in conventional evaluation: ①Generally, the azimuth angle β of natural fracture zone is of great significance for well trajectory design, but for drilled and fractured wells, the approach angle θ is more valuable for analysis; ②W _NF and W _LM reflect the development scale of micro-fractures and micro-bedding inside the matrix rock, which is controlled by sedimentary facies and is positively correlated with the content of mineral components, and can be obtained by multiple regression of V _Si and V _Ca in the source factors; ③P _p is a newly added parameter in this study that comprehensively considers three-dimensional stress and natural fracture zone angle to evaluate whether fractures can be opened. The smaller the value is, the easier the natural fracture zone is to be opened, and the more conducive it is to hydraulic fracturing; ④P _net is introduced as a newly added parameter, the smaller the value is, it indicates that the penetration and steering of hydraulic fractures are less difficult under the same construction intensity, and the fracture complexity is higher; ⑤Horizontal well path also has a crucial influence on long-term production effect. Both updip and downdip wells have large dip angles, which will lead to fluid accumulation at heel and toe, respectively, which will affect long-term productivity.

The study also collected source targets (ST) and derived targets (DT) of dynamic monitoring, which can be divided into two categories: microseismic monitoring and production dynamic monitoring. Microseismic monitoring includes SRV (SRV), spread length (L), spread width (W), spread height (H) and derived parameters spread length-width ratio (R). Production dynamic monitoring includes daily test yeild (Q _Test), 30/60/90/180/330 days’ cumulative yeild (Q ₃₀, Q ₆₀, Q ₉₀, Q ₁₈₀, Q ₃₃₀), EUR (EUR) and derivative parameters equivalent construction intensity production data (Q _Test+, Q ₃₀₊, Q ₆₀₊, Q ₉₀₊, Q ₁₈₀₊, Q ₃₃₀₊, EUR ₊).

Micro-seismic monitoring can be divided into two modes: downhole micro-seismic monitoring and surface micro-seismic monitoring. To eliminate the influence of different monitoring modes, this study selected 554 stages of 25 downhole micro-seismic wells with similar fracturing operation parameters in the Changning gasfield as the target, using the seismic and logging data to read the matrix brittleness and fracture propagation factors of each stage of horizontal well, and the correlation analysis is conducted with the target parameters of each stage monitored by micro-seismic.

Univariate analysis was carried out based on the calculation results of 554 stages of 25 wells in Changning gas field and SRV or R of each stage (Table 2). Due to the influence of the distance between the downhole micro-seismic monitoring sensor and the monitoring well, the monitoring data of each well is susceptible to the influence of the signal strength in the processing and has the difference in the value size of SRV. Therefore, the collected 554 segments of data are not completely linear. Taking 40 stages in three wells on platform H6 as an example, the factors with the most significant correlation with SRV were obtained, including P _net, Δσ, E _dyn, σ _v, σ _h, σ _H (Figure 2). The factors with the most significant correlation with R mainly included E _dyn, ν, W _NF, W _LM, K _II, K _I (Figure 3). The ranking of other relevant factors can be seen in Table 2.

The evaluation of the stimulation effect in Sichuan Basin usually takes the test yield as the main goal and tends to ignore the long-term stable yield ability. On the other hand, reservoir quality also determines the upper limit of production. Since it is difficult to evaluate productivity by stages as a unit, this study selected 161 wells in Changning gasfield as the analysis target, and took into account the factors that affect SRV and R of micro-seismic monitoring. At the same time, several factors affecting the reservoir quality and production effect, such as POR, TOC, Gas _F, Gas _A, Gas _T, V _Total, ΔD, were considered to analyze the influence on the production parameters of the same stimulated intensity.

Univariate analysis was conducted by using the calculation results of 161 wells in Changning gasfield and Q _Test+ , Q ₃₀₊ , Q ₆₀₊ , Q ₉₀₊ , Q ₁₈₀₊ , Q ₃₃₀₊ , EUR ₊ of a single well (Table 3). The analysis shows that there are significant differences in the main controlling factors of gas well production in different production stages.

For SRV, P _net is a parameter derived from Δσ and θ, and its influence on SRV is much higher than that of Δσ. The smaller P _net is, the greater the probability of hydraulic fractures penetration and diversion towards natural fractures is, and the wider the spread hydraulic fractures is. Therefore, the development direction of natural fracture belt and the comprehensive influence of horizontal stress difference play a key role in hydraulic fracture propagation. We also can see that P _net can basically embody the characteristics of Δσ, σ _H, σ _h. Therefore, P _net and E _dyn are selected as the main analysis objects. it can be seen from the statistical figure that the fracturing stages with smaller P _net and larger E _dyn in a single well generally have higher SRV (Figures 4A,B), which can still show strong negative correlation and positive correlation, respectively, on the whole. According to ΔP _net = 1 MPa as the change step size, the mean value of P _net and SRV in the interval corresponding to the segment of the line (Figure 4C) was calculated. Similarly, according to ΔE _dyn = 2.5 MPa as the change step size, the mean value of E _dyn and SRV in the corresponding interval of E _dyn was calculated (Figure 4D). It can be seen that both P _net and E _dyn have a good power relationship with SRV, and the accuracy of the trend line is high. Therefore, E _dyn is obtained based on the comprehensive calculation of mineral and saturated fluid matrix, and also has the significance of rock ore itself. Therefore, P _net and E _dyn are selected as the main factors to evaluate the formation of large SRV in the reservoir under the precondition of not repeated consideration.

For R, it can be seen that the influence of E _dyn and ν on R is much higher than that of other factors, indicating that the reservoir with higher matrix brittleness is more conducive to be uniformly stimulated. W _NF and W _LM also have a large influence, indicating that the development of natural weak plane may aggravate the non-uniform propagation of hydraulic fractures, resulting in the imbalance of spread length-width ratio. The increase of K _II also leads to the imbalance of fracture uniformity, indicating that the stronger the fracture resistance of rock is, the smaller the possibility of hydraulic fracture bending is. The influence of mineral composition on fracture uniformity is small, which may be related to the difficulty of mineral composition to directly reflect the rock structure and mechanical properties. Compared with SRV, the influence of in-situ stress and its derived parameters on R is small. E _dyn and ν are extracted as the main analysis objects according to the correlation, and 554 stages of data are also analyzed statistically. It can be seen from the statistical figure that the lower the ν and the greater the E _dyn in a single well, the smaller R is generally (Figures 5A,B). According to ΔE _dyn = 2 GPa as the change step size, the mean E _dyn and R in the interval corresponding to E _dyn was calculated (Figure 5C). In the same way, according to Δν = 0.02 as the changing step size, the mean ν and R in the interval corresponding to the ν is calculated (Figure 5D). It can be seen that both E _dyn and ν have a good power relationship with R, and the precision of the trend line is high. Therefore, E _dyn and ν are selected as the main controlling factors to evaluate the uniform expansion of the reservoir.

For Q _Test+, Gas_T, TOC, Gas _A, POR, Gas _F are the main factors affecting the test daily production, which indicates that the reservoir quality plays a decisive role in the initial productivity. The obvious negative correlation between V _Total and Q _Test+ is not because the high brittleness of the reservoir affects the production, but more likely because the gas wells with high brittleness do not have high quality reservoir quality under the influence of deposition and diagenesis (Shen et al., 2020b). It also shows that it is not sufficient to evaluate the reservoir stimulation potential only by using brittle minerals as an engineering factor. The influence of ΔD on the daily test yield is different from other factors. When it is negative, it means that the heel of the horizontal well is higher than the toe, indicating downdip well; otherwise, it means updip well; and its absolute value represents the inclination of the horizontal well. Influenced by the structure of the Sichuan Basin, the dip angle of 0–10° is always present in the main gas producing layers, and the larger dip angle has an obvious risk of fluid accumulation, thus affecting the production (Wei et al., 2019). Therefore, there is a binomial relationship between the dip Angle and the measured daily production.

But for EUR+, it shows that the main controlling factors have inherited the changing trend from the daily test yield to the EUR of the period mentioned above. The main controlling factors are changed, and changed in order. From the results of univariate analysis, the dominant factors have four characteristics. The first is the instability of the main control factors. The main control factors always change, with the passage of production time, the overall change is great. Second, and most important, with the passing of production time, the weight of P _net and E _dyn, which influence the effect of fracture network stimulation, is gradually increased, and the primary factor is gradually changed from “geological parameters” to “engineering parameters.” Third, reservoir property parameters could not to be ignored to productivity effect, which is the basic guarantee for the production after the formation of complex fracture network. Fourth, ΔD can not be ignored. The analysis shows that the influence of the horizontal well trajectory inclination on the long-term stable production is gradually increasing, which may be related to the fluid accumulation at the heel and toe of the upward dip well after a long production time. Comprehensive analysis shows that long-term stable and high production of gas well is not only affected by static geological parameters of reservoir, but also by rock mechanics, fracture mechanics, geomechanics and well path. Thus, the influence model of shale gas well geology-engineering-production model is established (Figure 6).

In the early stage of production, under the precondition of smooth fracturing operation, gas well productivity is mainly controlled by RP, that is, the stronger the reservoir properties, the higher the gas content, and the overall high production of the gas well. Factors as production time, MB and FP gradually highlight the influence of factors on gas well productivity, show the resources of enrichment of reservoir formed by hydraulic fracturing complex network, get bigger the SRV, is the basis of the gas wells to maintain long-term stable and high yield eventually safeguard, also shows that the reservoir itself have the ability to form complex joint network and the larger the SRV, under the same construction intensity, relatively high yields can be obtained. The main controlling factors of productivity in different stages also show that the main controlling factors of gas well are RP factor, MB factor and FP factor successively: Regardless of the technology, the near-wellbore zone can be fully stimulated. Therefore, in the early stage of production, production mainly comes from the near-wellbore zone where complex fracture networks are formed by fracturing, and the reservoir and gas content of the reservoir itself absolutely control the productivity (Figures 6A,B). As production time goes by, the fracture network formed by hydraulic fracturing connects the reservoir matrix and communicates with natural fractures, which becomes the main factor affecting production replacement. The more channels gas flows from the matrix to the wellbore, the more effectively the production of the gas well can be maintained (Figures 6C,D). At the end of production, gas well productivity is more reflected in the initiative of stimulated reservoir volume, emphasizing that the wider control area of hydraulic fracture in three-dimensional space, the more conducive to obtain the final cumulative high production (Figures 6E,F).

Matrix brittleness is fused with E _dyn and ν, considering the difference of the two orders of magnitude, the two parameters are firstly dimensionless processing: E ’ = E dyn − E dyn   min E dyn   max − E dyn   min ,    ν ’ = ν − ν min ν max − ν min ,   K ’ I = K I − K I   min K I max − K I   min ,   K ’ II = K I I − K I I   min K I I   max − K I I   min (3)

Then, the dimensionless elastic modulus and the dimensionless Poisson’s ratio were dimensionless again to establish the matrix brittleness potential index: B N = ( E ’ / ν ’ ) − ( E ’ / ν ’ ) min ( E ’ / ν ’ ) max − ( E ’ / ν ’ ) min , K N = ( K ’ I × K ’ II ) max − ( K ’ I × K ’ II ) ( K ’ I × K ’ II ) max − ( K ’ I × K ’ II ) min ,   2 C F = 1 B N + 1 K N (4)

P _net is dimensionless to obtain the fracture propagation potential index: P F = P net   max − P net P net   max − P net   min (5)

Since it is difficult to combine POR and TOC, they should be combined by considering the weight. Therefore, the corresponding correlations of 0.1768 and 0.1734 in single factor analysis of POR and TOC were selected to evaluate EUR of gas wells. ΔD is introduced, and 1000 m vertical depth difference is taken as the gas well limit, and enrichment and development potential index is established to reflect the impact of gas well inclination on production: POR ’ = POR − POR min POR max − POR min , TOC ’ = TOC − TOC min TOC max − TOC min (6) 0.1768 + 0.1734 R F ( 1 − | Δ D / 1000 | ) = 0.1734 POR ’ + 0.1768 T O C ’ (7)

Finally, the three factors are combined to form the stimulation potential index: 3 M F = 1 C F + 1 P F + 1 R F (8)

The study collected EUR ₊ data from 301 wells, including 161 analysis wells above and 140 validation wells. A total of 301 wells were used for the analysis, and once the majority of wells met the evaluation requirements, the method proved to be of widespread value. With EUR ₊ setting 0.5 × 10⁸ m³ as step length, the productivity interval was divided. C _F and P _F were used as horizontal and vertical coordinates respectively to analyze the distribution interval of high-producing and low-producing Wells (Figure 7A). Wells with EUR ₊ at 1.5 × 10⁸ m³ are mainly distributed in the dominant zones (C _F and P _F greater than 0.5), but the red dominant zones also have a significant number of low-producing wells, indicating that effective fracturing potential alone is not sufficient to reflect stimulation potential.

For convenience analysis, C _F and P _F are equivalent combined as Q _F (Eq. 9) to comprehensively evaluate stimulated effect and form a Cartesian coordinate system with R _F (Figure 7B). In terms of the distribution of gas Wells, the high-producing wells are always distributed in the dominant area (Q _F and R _F are greater than 0.5), while the low-producing wells are “eliminated” to other areas, with a more obvious differentiation, which confirms that the factors selected in this study and the indicators formed are effective in the evaluation of the stimulated potential of gas wells. 2 Q F = 1 C F + 1 P F (9)

A geological model was established based on 301 logging data to reflect the burial depth of the reservoir (Figure 8A) and the comprehensive index M _F (Figure 8B). The higher the M _F, the better the stimulated potential of the reservoir, and the more conducive to achieving high yield. It can be seen that the high part of the structure and the slope area (MF > 0.5) are the main positions to obtain high yield (Liu et al., 2020, 2021), while the axial part of the tectonic syncline (MF < 0.5) is more complicated due to stress concentration.

Moreover, in the fracturing process of horizontal wells in Changning gasfield, the liquid strength, proppant strength and pumping rate in a single well are within the range of 20–45 m³/m, 1∼5 t/m and 8–18 m³/min respectively. A total of 181 wells in high part of structure and slope areas saw a significant increase in EUR as the fracturing parameters increased, and it seems that the fracturing parameters could be further increased to achieve the purpose of production increase (Figures 9A,C,E). However, from the parameter changes of 120 wells in the synclinal axis, the increase rate of EUR per well is low, and the increase of production is no longer obvious after fracturing operation parameters increase to a certain scale, that is, reservoir conditions control the upper limit of gas well productivity in this area (Figures 9B,D,F).