A new gas identification chart is developed based on the simulation results. A new parameter △NG₁ is defined as △NG₁ = △NG/Φ, which means the ratio between the difference in neutron–gamma count rate per unit porosity between target layers and mudstone layers. △NG is defined as △NG = NG-NG₀, that is, the difference in the neutron–gamma count rate between target layers (NG) and mudstone layers (NG₀). △NG₁ is set as the X-axis of the gas identification chart and formation water salinity is set as the Y-axis. The points of gas layers appear in different parts of this chart depending on the levels of gas saturation and formation fluid salinity.

The data of 51 well test layers in 22 old wells from neutron–gamma logging and formation water analysis were used to make the gas identification chart (Figure 5). The points of gas layers in the study area, where the gamma-ray count rate is higher and which are mostly concentrated in the top section of the cross-plot, are obviously different from those of non-gas layers. The dividing line rises upward as formation fluid salinity increases, which is consistent with the result of MCNP simulation.

The new gas identification chart is used to evaluate reservoirs based on well logs in Langgu Depression. One of the results is presented in Figure 6. Well A1 is one of the wells used for old well reexamination. Neutron–gamma logging is an important method that can be used because the available logging curves from old wells are limited. The reservoirs are sandstone formations with high density and low porosity. Layer #125 is a tight sandstone reservoir with porosity of about 10%. The value of △NG₁ in layer #125 is about 2.4, and the formation fluid salinity is 4650 ppm. Using the new gas identification chart, we interpret this layer as a gas layer, and this interpretation result is identical with the well test result, indicating that neutron–gamma logging is effective in the discovery and evaluation of such gas layers.

Biot and Gassmann identified the relationship between the Lame’s coefficients and bulk modulus for saturated and dry rocks (Biot, 1941; Gassmann, 1951) λ s a t = λ d r y + β 2 M (1) K s a t = K d r y + β 2 M (2) where λ s a t is the Lame’s coefficient for the fluid-saturated rock, λ d r y is the Lame’s coefficient for the dry rock, β is the Biot coefficient, and M is the modulus. Under low-frequency conditions, Gassmann gave the following equations β = 1 − K d r y K m (3) 1 M = β − ϕ K m + ϕ K f l (4) where K m and K f l are the bulk modulus of the matrix material and that of the formation fluid, respectively.

By substituting Eqs 3, 4 into Eq. 2, the following equation can be obtained (Mavko et al., 2009): K s a t = K d r y + ( 1 − K d r y K m ) 2 ϕ K f l + 1 − ϕ K m − K d r y K m 2 (5)

The term β 2 M is defined as pore space modulus K _p (Murphy et al., 1993), which is used to reflect the properties of pore fluid and distinguish the dry rock from the saturated rock. By combining the term β 2 M with Biot and Gassmann equations, we can obtain the equation for P-wave velocity V _p , which is written as follows: V p = λ + 2 μ ρ s a t = λ d r y + 2 μ + β 2 M ρ s a t (6) V p = K + 4 3 μ ρ s a t = K d r y + 4 3 μ + β 2 M ρ s a t (7)

It can be simplified into the following: V p = s + f ρ s a t (8) where ρ s a t is the total density of the saturated rock, s = K d r y + 4 3 μ or s = λ d r y + 2 μ is the dry-skeleton term, and f = β 2 M is a fluid/porosity term. Assuming that μ = μ s a t = μ d r y , the equation for S-wave velocity V _s can be written as follows: V s = μ ρ s a t (9)

As discussed earlier, the equation for P-wave velocity V _p of the rock can be combined with dry-skeleton term s and fluid/porosity term f. Using these two terms, the equation P- and S-wave impedances, Z _P and Z _S , can be rewritten as follows: Z p = ρ V p = ρ ( f + s ) (10) Z s = ρ V s = ρ μ (11)

Assuming that c μ = s , by extracting factor c, we can obtain the following equations c = s μ = λ d r y μ + 2 = K d r y μ + 4 3 = [ V p V s ] d r y 2 (12) ρ f = Z p 2 − c Z s 2 = ( ρ V p ) 2 − c ( ρ V S ) 2 (13) where ρ is the total density of the rock, which can be obtained from density logging, and V _p and V _s can be obtained from array acoustic logging. ρ f is a fluid component that reflects the contribution of the fluid in the pores. Therefore, it was defined as a fluid factor by Russell et al. (2003). The fluid factor can be calculated and used for gas-bearing reservoir identification after the value of c has been found.

Many research studies have been conducted to get an estimate of c. Murphy et al. (1993) found that the average value of K _dry /μ ratio for clean quartz sandstones with different porosities was 0.9, in which the value of c is 2.233. Zhang et al. (2015) measured the P- and S-wave velocity for 75 dry sandstones with different porosities and found that the value of c was 2.324. The P- and S-wave velocities of 44 rock core samples taken from Langgu Depression were measured under saturated and dry conditions to find the suitable value of c in this area. The measurement results are shown in Figure 7. From Figure 7, it can be seen that the ratio of P- and S-wave velocity of dry rock cores is mainly within the range of 1.4 to 1.6, and the average value is 1.54. The ratio of V _p /V _s of saturated rock cores is affected by the porosity, and its value increases as porosity increases. Table 1 shows the average value of V _p /V _s and other elastic parameters of the 44 rock core samples. It can be seen from Table 2 that the value of c for samples taken from Langgu Depression is 2.375.

The data of array acoustic logging and density logging of 53 well test layers in 19 new wells were used to calculate fluid factor and V _p /V _s . The results are shown in Figure 8. Figure 8 shows that the points of gas layers are mostly concentrated in the lower left corner of the cross-plot and are obviously different from those of non-gas layers, and the values of V _p /V _s and fluid factor of gas layers are smaller than those of non-gas layers. Most of the fluid factors of gas layers are less than 15, and those of non-gas layers are greater than 15. Therefore, the fluid factor’s value of 15 is taken as the dividing line between gas and non-gas layers.

The fluid factor was used to identify gas layers in the new wells in which array acoustic logging was carried out. Taking well B1 (Figure 9) in the study area as an example, the first track includes the caliper (CAL), self-potential (SP), and natural gamma (GR), which can reflect the wellbore conditions and the reservoir’s lithology. Track 2 displays the well depth of logs. Track 7 contains the compensated neutron (CNL), density (DEN), and acoustic log (AC), which could provide estimates of the reservoir’s physical properties, such as porosity (track 5) and permeability (track 6). Track 8 shows high-definition array induction logs, which are commonly used to evaluate the properties of reservoir fluid. We can obtain the value of fluid factor ρ f from array acoustic logging and density logging, as shown in track 9, and the fluid factor’s value of 15 is taken as the dividing line.

The target tight reservoir in Figure 9 can be divided into five sandstone layers, represented by different icons in track 3. The three layers, 72, 74, and 75, have high porosity and permeability and are characterized by low density, high neutron, and high resistivity. In particular, the value of ρ f for these three layers is about 10, which is much lower than the value of the dividing line. In summary, these three layers can be interpreted as good gas layers. The logging data of layer 73 have the same trend, but they are not as clear as those of the three layers, and the value of ρ f for this layer is about 12, which is still smaller than 15. Therefore, we interpret this layer as a bad gas layer. The porosity and permeability of layer 76 are lower than those of other layers, no sign of gas in this layer is shown in the traditional logging data, and the value of ρ f for this layer is beyond the dividing line. This layer is interpreted to be a dry layer. Layers 72 to 75 are tested together, and the results show they are gas layers. These results are consistent with the conclusion drawn from well log interpretation.