Genesis Mechanism and Logging Evaluation Methods for Low-Resistivity Contrast Gas-Bearing Layers in Shallow Gas Reservoirs

Abstract

Shallow gas reservoirs exhibit low formation pressure and gas injection levels, leading to low-resistivity contrast between gas-bearing reservoirs and fully water-saturated layers. Gas-bearing formation identification and water saturation estimation face great challenges. To improve the accuracy of shallow gas reservoir identification and logging evaluation, it is essential to analyze the genesis mechanisms underlying the low-resistivity contrast. This study used the HJ Formation, a typical shallow gas reservoir located in the BY Sag of the eastern South China Sea Basin as an example. Combining the results of nuclear magnetic resonance (NMR), full rock mineral analysis and X-ray diffraction of clay minerals in the laboratory, it was determined that the genesis mechanism for the low-resistivity contrast in the gas-bearing reservoir was due to the high irreducible water saturation (S_wi) and the cation-induced supplementary conductivity. Afterwards, we integrated three methods, density–neutron correlation, calculation of the apparent formation water resistivity, and cross-plots of conventional and gas-logging curves, to identify shallow gas reservoirs. In addition, we also established a Waxman–Smits-based model to estimate water saturation. Compared with the typical Archie’s equation, the predicted water saturation curve using the Waxman–Smits-based model was more reasonable. The established methods and models can be used in target shallow gas reservoir evaluations, and it also has reference value for other types of oilfields with similar physical characteristics.

1. Introduction

Shallow gas resources (<1500 m) represented both a critical drilling hazard and underexploited energy asset, and are characterized by a low formation pressure, unconsolidated formation, poor cementation and compaction [1,2,3]. In the current global context, the demand for hydrocarbons has intensified the exploration of such marginal reserves. However, a systematic understanding of the formation mechanism of and exploration methodologies for these reserves remain underdeveloped. In particular, conventional petrophysical models for low-resistivity contrast gas-bearing reservoirs (LRGRs) often fail.

Within the central southern Pearl River Mouth Basin, the BY Sag exemplifies these challenges. It is one of the largest Cenozoic sedimentary depressions, with a total area exceeding 25,000 km² and a sedimentation thickness of more than 12 km [4,5]. Its shallow HJ formation hosts pervasive LRGRs with a fine lithology and complex microporous structure. These characteristics create challenges for effective reservoir identification, fluid property evaluation and sweet spot prediction. It is difficult to perform pore fluid identification in these absolute LRGRs, which are defined by an ultra-low resistivity (<5 Ω·m) rather than a low gas-to-water contrast [6,7,8].

Numerous studies have made significant efforts to study the genesis of LRGRs [9,10,11,12]. However, there is relatively little systematic research on the characteristic and genesis mechanism of shallow buried layers. The genesis of LRGRs involves multi-scale factors. The macroscale factors affecting low-resistivity contrast are engineering factors, such as the deep invasion of high-salinity mud filtrate, which reduce formation resistivity [13]. At the microscale, some studies have shown that the presence of high-conductivity minerals or fluids (e.g., pyrite, clay minerals with a high cation exchange ability or high-salinity formation water) reduces resistivity [14,15,16,17]. Besides these, the complex pore structure, the phenomena of an enhanced water retention capacity of the rock and increased capillary force, and the large surface area of the particles can all result in high irreducible water saturation (S_wi) [18,19]. The current identification methods primarily use traditional cross-plots and curve overlays. Advanced techniques (e.g., nuclear magnetic resonance (NMR) and array acoustic logging) can enable the identification of multi-parameter fluid sensitivity factors [20]. However, ambiguous fluid signals and undiagnosed genesis mechanisms in uncompacted formations can decrease the evaluation accuracy.

This study focused on the genesis mechanism and log evaluation of shallow LRGRs in the HJ Formation in the northern BY Sag in the South China Sea Basin. By combining petrophysical experiment data with advanced logging data from the L19 Region, the genesis mechanism of our target formation was determined and the effect of a high S_wi and high cation exchange capacity on reducing formation resistivity were elucidated. By analyzing the interplay between lithology, pore structure and fluid properties, we aimed to establish region-specific identification criteria. This study provides an integrated workflow for clay-rich shallow reservoirs.

2. Methodologies and Results

2.1. Genesis Mechanism of Low-Resistivity Contrast Gas-Bearing Formation

2.1.1. High S_wi Decreases Resistivity

Irreducible water comprises three distinct types: surface-bound water on non-clay materials, clay-associated water within clay materials, and capillary retention water in microporous networks [21,22,23]. Unlike capillary retention water, clay-bound water remains invariant under reservoir perturbations. This immobile phrase significantly elevates the conductivity due to the expansion of the surface area of clay minerals.

The measured results of X-ray diffraction experiments in the laboratory clearly illustrated that our target formation features a high clay content and its lithology is dominated by clay and argillaceous siltstone, while core samples recovered from conventional reservoirs exhibited lower clay contents (

Table 1. X-ray diffraction results for five core samples drilled from two types of formations in well L19-2.

Well	Type	Depth (m)	Clay Content (%)	Pyrite (%)	Clay Mineral Content (%)
					Kaolin	Chlorite	Illite	I/S
L19-2	LRGR	x68.25	29.2	0.6	7.3	7.0	7.1	7.8
		x77	30.5	0.4	6.6	7.0	8.5	8.4
		x77.5	30.3	1.9	6.9	6.8	7.5	9.1
		x85.75	31.2	/	7.0	7.1	8.5	8.6
	Conventional reservoir	x66.75	12.9	0.4	3.0	2.3	2.6	5.0
		x86.5	21.2	1.8	6.0	4.8	5.0	5.4

). Generally, a high clay content always increases S_wi, thus decreasing the resistivity [24]. Figure 1 clearly indicates that the deep induction resistivity (R_t) was inversely related to clay content. The other factor, the low pyrite content (with an average value of 0.77%), was too low to reduce resistivity. From Table 1 and Figure 1, we can conclude that a high clay content decreases the resistivity in gas-bearing reservoirs.

As shown in Figure 2, the HJ Formation in well K18-1 exhibited similar low-resistivity characteristics (1–5 Ω.m). The drill stem test (DST) data illustrated that the interval of xx50-xx60 m is a pure gas-bearing formation without any water. The array induction resistivity indicated that this interval could be defined as a low-resistivity contrast gas-bearing reservoir. The measured NMR T₂ spectra shown in the sixth track in Figure 2 and the core-derived results (Figure 3a) indicated that the HJ Formation is dominated by micro-pores. Crucially, the negative correlation between S_wi and R_t indicated that an increase in S_wi would lead to a significant reduction in resistivity in gas-bearing zones (Figure 3b). In water-saturated zones, this reduction was attenuated due to the presence of movable water.

2.1.2. High Cation-Induced Supplementary Conductivity Decreases Resistivity

Clay conductivity is a well-documented contributor to low-resistivity contrast reservoirs [8] due to the prevalence of mixed layers of illite and smectite (I/S), which generate conductive pathways. Such clays exhibit characteristically high cation exchange capacities (CECs). When the CEC value exceeds 3 mmol/100 g, the additional conductivity must be considered [25].

The measured results for 14 core samples revealed high I/S contents (the average value was 28.13% in Figure 4a). Meanwhile, high I/S contents greatly increased the CEC (Figure 4b). On the contrary, core samples recovered from conventional sandstones exhibited relative low total clay and I/S contents and CECs. The integration of mineralogical data with petrophysical logging data confirmed that cation-induced supplementary conductivity is another key factor influencing the resistivity in gas-bearing reservoirs.

2.2. Methods of Identifying Pore Fluids

2.2.1. Typical Log Response Differences Between Gas-Bearing and Water-Saturated Reservoirs

Figure 5 compares the log responses of a typical gas-bearing reservoir with high resistivity and a low-resistivity contrast reservoir in well L19-1. The typical gas-bearing reservoir exhibited a pronounce resistivity increase, with a clear intersection of the density and neutron curves, while the LRGR showed a moderate resistivity increase and weak density–neutron crossover. However, if we ignore the absolute values of the density and neutron curves and analyze the trends of these two curves, we can observe that their positively correlated trends were similar. The main difference between these two types of formations was the value of their density curves. Typical gas-bearing reservoirs have a relatively lower density, even if their neutron values are nearly identical. Figure 6 shows the log responses in a water-saturated reservoir, with the accompanying cross-plot showing a negative relationship between the density and neutron curves. This result was completely opposite to the one shown in Figure 5, indicating effective pore fluid identification.

2.2.2. Pore Fluid Identification Based on Correlation Analysis Method

An obstacle in low-resistivity contrast gas-bearing formation identification is their insignificant responses; the correlation between the density and neutron curves provides an effective way to amplify the response [26,27]. We calculated the correlation coefficient between the density and neutron porosities:

$$R({\phi }_N,{\phi }_D)={\displaystyle \frac{n{\sum }_{i=1}^n{{\phi }_N}_i\times {\phi }_{D_i}-{\sum }_{i=1}^n{\phi }_{N_i}\times {\sum }_{i=1}^n{\phi }_{D_i}}{\sqrt{n{\sum }_{i=1}^n{\phi }_{N_i}^2-({\sum }_{i=1}^n{\phi }_{N_i}{)}^2}\times \sqrt{n{\sum }_{i=1}^n{\phi }_{D_i}^2-({\sum }_{i=1}^n{\phi }_{D_i}{)}^2}}}$$

(1)

In this equation, ${\phi }_N$ represents the neutron porosity in % and ${\phi }_D$ denotes the density porosity in %. They were calculated using the following equations [26]:

$${\phi }_D={\displaystyle \frac{{\rho }_b-{\rho }_{ma}}{{\rho }_f-{\rho }_{ma}}}\times 100$$

(2)

$${\phi }_N=CNL+1.5$$

(3)

In these equations, ${\rho }_b$ represents the formation’s bulk density in g/cm³, ${\rho }_{ma}$ denotes the matrix density in g/cm³, ${\rho }_f$ is the density of the pore fluid in g/cm³ and $CNL$ is the neutron logging data in %.

It should be noted that density is negatively related to neutrons in gas-bearing reservoirs and positively related to neutrons in water-saturated reservoirs. However, if the density and neutron curves are translated into density and neutron porosities using Equations (2) and (3), the correlation between them would be the opposite due to the negative relationship between formation bulk density and porosity.

Figure 7 shows an example of using the correlation analysis method to identify pore fluids. The calculated coefficient ranged from −1 to 1. When the pore space was occupied by natural gas, the coefficient was mainly negative. On the contrary, in the water-saturated reservoir, the calculated coefficient was mainly positive. By using this method, pore fluids can be identified.

2.2.3. Identifying Pore Fluids Using Apparent Formation Water Resistivity

Based on Archie’s equation, the relationship between the formation factor (F), formation water resistivity (R_w) and rock resistivity under complete water saturation (R₀) can be expressed as

$$F={\displaystyle \frac{R_0}{R_w}}={\displaystyle \frac{a}{{\varphi }^m}}$$

(4)

In this equation, R₀ represents the rock resistivity under complete water saturation in Ω.m, R_w is the formation water resistivity in Ω.m, $\varphi$ denotes the rock porosity in v/v, $a$ is the lithology coefficient and $m$ is the cementation factor.

If we replace R₀ with R_t, the true formation water resistivity cannot be calculated from Equation (4), but an apparent formation water resistivity ($R_{wa}$) could be calculated:

$$R_{wa}={\displaystyle \frac{R_t{\varphi }^m}{a}}$$

(5)

In this equation, $R_t$ represents the true formation resistivity in Ω.m and R_wa is the apparent formation water resistivity in Ω.m.

Research on apparent formation water resistivity commonly uses its square root (P^1/2) to establish pore fluid identification criteria. Theoretically, R_wa should be a single value in a water-saturated formation due to the stability of R_t. However, lithological heterogeneities—particularly the presence of calcite, salinity variations and thermal disequilibria—can induce stochastic R_wa fluctuations around a certain value. This systematic deviation necessitated a distribution analysis, where a normal distribution can be effectively used to resolve fluid types using R_wa.

For the standard water-saturated reservoir, the data points were relatively concentrated, with a narrower distribution, presenting a typical normal distribution. In contrast, for the typical gas-bearing reservoir, the range of R_wa was significantly wider and the normal distribution characteristic was not obvious. Similar to the typical gas-bearing reservoir, the LRGR also showed a wide R_wa range and the distribution pattern had a wider normal distribution. The main difference between these two types of gas-bearing reservoirs was the variation in their normal distribution curves (Figure 8).

A standardized template was developed to quantify pore fluid identification criteria through P^1/2 (Figure 9a). The distributions of the different types of reservoirs varied. The conventional gas-bearing layer showed the widest P^1/2 span (0.4–0.6), while the LRGR exhibited a compressed distribution like that of a water-saturated layer. The template showed a good result for pore fluid identification. To further verify the validity of this method, data acquired from well L19-3 was used (Figure 10). A representative gas-bearing reservoir and LRGR were identified in the intervals of x70–x72 m and x92–x96 m. The conventional gas-bearing reservoir exhibited an elevated R_wa (0.2–0.35 Ω.m), while the LRGR showed a significant reduction (0.08–0.17 Ω.m). When these reservoirs are plotted in this special template (Figure 9b), the different reservoirs appeared to be well separated.

2.2.4. Identifying Pore Fluids by Combining Conventional and Gas-Logging Data

During drilling, when encountering a gas-bearing formation, the gas within the pore space might infiltrate into the drilling fluid due to rock fragmentation and seepage. Consequently, gas-logging data can be used to directly identify hydrocarbon-bearing formations. Meanwhile, resistivity is crucial in identifying pore fluids. In this study, we combined the peak-to-base ratio of total gas (TG_YS) with conventional logging data and significantly enhanced the accuracy of pore fluid identification. Figure 11a,b shows the cross-plots of deep induction resistivity and density versus TG_YS. These two cross-plots effectively differentiated gas-bearing formations from water-saturated intervals.

By integrating all three types of methods mentioned, we established pore fluid identification criteria, which are summarized in

Table 2. Criteria for identifying pore fluids based on logging data that were established using three types of methods in HJ Formation of BY Sag.

Formation	Deep Induction Resistivity	TG_YS	Correlation	Slope of P1/2
Conventional gas-bearing formation	≥2	≥2.16	Negative	≥0.083
Low-resistivity contrast gas-bearing formation	1.3~2	≥2.16	Negative	0.0415~0.083
Water-saturated layer	<1.3	<2.16	Positive or unrelated	<0.0415

. It is important to note that a positive identification of pore fluids required the formation to fulfill at least two-thirds of the criteria listed in Table 2.

2.3. Formation Parameter Evaluation

2.3.1. Shaly Content and Porosity Evaluation

Due to the ineffectiveness of using natural gamma curves to identify sandy reservoirs in this region, the shaly content was estimated using density–neutron cross-plots [28,29]. Figure 12 shows the triangle density–neutron cross-plot. Using mathematic transformation, the shaly content could be determined as follows:

$$V_{sh}={\displaystyle \frac{{\varnothing }_N-{\varnothing }_D}{{\varnothing }_{Nclay}-{\varnothing }_{Dclay}}}$$

(6)

In this equation, $V_{sh}$ represents the shaly content in v/v and ${\varnothing }_{Nclay}$ and ${\varnothing }_{Dclay}$ are the neutron and density porosities of shale in v/v.

Apart from the shaly content, the formation’s effective porosity ${\varnothing }_{ND}$ can also be calculated from the triangle density–neutron cross-plot:

$${\varnothing }_{ND}={\displaystyle \frac{\left|{\varnothing }_N\cdot {\varnothing }_{Dclay}-{\varnothing }_D\cdot {\varnothing }_{Ncaly}\right|}{\left|{\varnothing }_{Dclay}-{\varnothing }_{Ncaly}\right|}}$$

(7)

In the equation, ${\varnothing }_{ND}$ is the effective porosity in v/v.

Figure 13 shows comparisons of the calculated effective porosity and shaly content from the triangle density–neutron cross-plot of the core-derived results. Figure 13a clearly indicates that the shaly content derived from the two batches of core samples showed strong consistency with the calculated results from the well L19-2 samples. Meanwhile, Figure 13b also exhibits strong consistency between the predicted effective porosity (POR) and core-derived porosity (CPOR) in well B13-1.

2.3.2. Water Saturation Calculation

As a semiempirical extension of Archie’s equation, the Waxman–Smits-based model takes into account the conductivity due to clay particles [30]. Considering this region’s high CEC, the Waxman–Smits-based model was introduced for regional water saturation evaluation. This model can be expressed as follows:

$$R_t={\phi }^{-m\ast }{S}_w^{-n^{\ast }}\left({\displaystyle \frac{R_w}{1+R_{w}B\frac{Q_v}{S_w}}}\right)$$

(8)

In this equation, $S_w$ represents the water saturation in v/v, $B$ denotes the equivalent conductance of equilibrated cations in mL/(ohm·m·meq), $Q_v$ is the cation exchange concentration in meq/cm³, and $m^{*}$ and $n^{*}$ are for the cementation exponent and saturation exponent.

To quantify the difference between the water saturation evaluations using Archie’s equation and the Waxman–Smits-based model, these two models were used to analyze all our target formations. Figure 14 shows a comparison of the calculated water saturation levels using Archie’s equation (SWAR) and the Waxman–Smits-based model (SWWS). It can be clearly observed that, in conventional gas-bearing reservoirs, the predicted water saturation levels were closed to each other. However, in the LRGR intervals (R_t < 2 Ω.m), the Waxman–Smits-based model yielded a 3% lower value than the Archie’s equation due to including the cation-induced supplementary conductivity. Figure 14 highlights that the Waxman–Smits-based model provides more accurate water saturation estimations in low-resistivity contrast shallow gas reservoirs with a rich clay content.

2.3.3. Permeability Prediction

In 1968, Timur proposed an equation relating permeability to porosity and S_wi, which has been verified to be effective over the last few decades [31]. In this study, the Timur equation was employed to establish a permeability prediction model based on the measured porosity and S_wi of core samples. Five core samples, which were drilled from well K18-1, were subjected to routine and NMR measurements, which were used to calibrate the parameters in the Timur equation. The suitable Timur equation for our target region was obtained and can be expressed as follows:

$$K={\displaystyle \frac{0.0019*{\varphi }^{4.4}}{S_{wi}^2}}$$

(9)

In the equation, $K$ represents the permeability in mD.

Accurately determining S_wi is critical for permeability evaluations via the Timur equation. In the HJ Formation of our target region, the DST data from four intervals were obtained, all of which indicated no water production. Hence, the S_wi values within these target intervals were hypothesized to be equivalent to the calculated water saturation levels based on the Waxman–Smits-based model. By inputting the predicted porosity and water saturation values into Equation (9), a consecutive permeability curve was obtained. To validate the accuracy of the calculated permeability, effective permeabilities were calculated from the pressure test data in well L19-1, and they were compared with the predicted results based on Equation (9) (Figure 15). CPERM is the acquired permeability from the pressure test data, and PERM is the predicted permeability from the Timur model. The comparisons indicated strong concordance between the results. Further quantification via cross-plots showed good consistency, and the statistical average relative error was only 16.77% (Figure 16).

3. Discussion

In the Timur model, S_wi is a crucial factor as it directly determines the accuracy of the predicted permeability. From our target formation, only five core samples were subjected to NMR experiments in the laboratory, and the results were used to calibrate the model. This limited the applicability of the Timur model. It was impossible to establish a reasonable S_wi prediction model using the limited core samples. In addition, NMR logging data was only acquired for well K18-1. Out of necessity, we used the calculated water saturation levels based on the Waxman–Smits-based model to replace S_wi to calculate the permeability. This is feasible for pure gas-bearing formations. However, in water-saturated layers and in the intervals with gas and water production, the S_wi was much lower than the value from the SWWS; thus, the predicted permeability would be underestimated. To improve permeability prediction accuracy, NMR logging data should be acquired in many more wells or many core samples should subjected to NMR measurements to establish a reasonable model to predict S_wi from conventional well logging data.

4. Conclusions

This study established a genesis mechanism–evaluation integrated workflow for LRGR characterization in shallow gas reservoirs of the BY Sag. The three key conclusions are as follows.

1.

Genesis Factors: High clay (Vsh > 20%) and smectite-dominated clay mineral (CEC > 15 mmol/100 g) contents are the primary drivers of low-resistivity contrast, with microporous networks exacerbating the water retention.

2.

Identification Framework: The integration of (a) cross-plot analysis, (b) Pearson correlation thresholds (gas: r < −0.5; water: r > 0.7) and (c) apparent formation water resistivity distributions enables pore fluid identification.

3.

Model Optimization: Proper evaluation models were established, and the field applications showed high consistency with the core-derived results.

This method could enable economically viable exploitation of shallow gas resources by improving sweet-spot prediction. Future work should focus on extending the methodology to regional analogs and incorporating machine learning to automate pore fluid identification.