Numerical Simulation Study on the Main Controlling Factors of Water Cut Rise in Thick Carbonate Reservoirs Based on Multi-Scale Hierarchical Analysis

Abstract

Based on the waterflooding development practice of thick carbonate reservoirs in the Middle East, aiming at the practical problems such as complex water invasion types, rapid water breakthrough of oil wells and poor development performance in such reservoirs, this study takes the MB1 reservoir of H Oilfield as the research object and establishes a multi-scale hierarchical screening scheme for the main controlling factors of water cut rise covering the reservoir-block-well group levels. Firstly, the target reservoir is divided into several typical development blocks by means of numerical simulation technology. On this basis, the dynamic development characteristics of the reservoir, typical blocks and well groups are analyzed respectively. The multi-sequence grey correlation method is adopted to screen the common influencing factors of water cut rise in typical blocks, and then the multi-factor sensitivity analysis of the screened key factors is carried out by numerical simulation. Finally, it is determined that the main controlling factors affecting the water cut rise in the reservoir are the development degree of high-permeability layers, the rationality of well pattern layout, and the injection–production intensity, and the corresponding development adjustment strategies are proposed accordingly. Guided by the multi-scale hierarchical screening of main controlling factors for water cut rise, this study improves the traditional grey correlation method and proposes a multi-sequence grey correlation analysis method. This method for determining the controlling factors, which combines mathematical approaches with reservoir numerical simulation techniques, gives full play to the advantages of both. It reduces the range of variables in numerical simulation analysis, avoids the sharp increase in simulation complexity caused by multi-factor coupling, and greatly improves work efficiency while ensuring research depth.

1. Introduction

As a core component of the global petroleum resource system, carbonate reservoirs present significant strategic value and development potential. These reservoirs are distributed across 43 countries and regions, with concentrated occurrences in the Middle East, and propose a multi-sequence grey correlation analysis method. This method combines mathematical approaches with reservoir numerical simulation to determine the dominant controlling factors. North America and Central Asia. According to the Statistical Review of World Energy (Energy Institute, 2024) [1,2], carbonate reservoirs account for approximately 56% of global hydrocarbon reserves, highlighting their critical role in the global petroleum resource system. Among them, the remaining recoverable hydrocarbon reserves in the Middle East account for 50.7% of the world’s total, and 80% of the oil-bearing formations are carbonate reservoirs, indicating enormous development potential, the global significance of carbonate reservoirs is summarized in Figure 1.

In the Middle East, where carbonate reservoirs are widely distributed, there exists a special type of carbonate reservoir—thick carbonate reservoirs. Such reservoirs are usually rich in reserves and characterized by large thickness, multiple development stages, and strong reservoir heterogeneity. Thick carbonate reservoirs are typically characterized by strong heterogeneity, complex internal architecture, and well-developed high-permeability zones and interbeds. For instance, Al-Jawad et al. (2020) [3,4] and Kargarpour (2020) [5] demonstrated that carbonate reservoirs in the Middle East exhibit strong facies-controlled heterogeneity and complex flow unit distribution, which significantly affect reservoir performance. This feature leads to a rapid water breakthrough rate during reservoir development, and at the same time, the reservoir exhibits complex water invasion mechanisms, including edge-water invasion, bottom water coning, and injected water channelling [6,7].

To timely optimize and adjust reservoir development, it is necessary to clarify the main controlling factors of reservoir water cut rise. In the research on the main controlling factors of reservoir water cut rise, previous studies commonly classify the influencing factors of water cut evolution into geological factors and development factors, and then apply statistical or mathematical methods to identify dominant controls. For example, Li (2022) [8] proposed a fuzzy comprehensive evaluation method for waterflood performance, while Feng et al. (2004) [9] developed an evaluation approach based on development indicators. However, for thick carbonate reservoirs, due to their large vertical thickness and strong reservoir heterogeneity, the complex internal structure leads to serious inter-well interference during water flooding development. The study of main controlling factors at a single scale will mask the differences between blocks and ignore the influencing factors at the well group scale. In terms of research methods, existing research on dominant controlling factors of water cut rise mainly includes experimental analysis and numerical simulation approaches. Li et al. (2009) [10] investigated production behaviour in carbonate reservoirs through physical experiments, while Li et al. (2021) [11] analyzed waterflooding performance using numerical simulation methods. These approaches provide valuable insights but still have limitations in representing complex large-scale reservoirs. Nevertheless, due to the large thickness of thick carbonate reservoirs, the fluid under the action of gravity has significant vertical displacement. However, due to the limitation of experimental objects, physical experiment analysis cannot realize the physical simulation of thick reservoirs, and many theoretical and experimental studies neglect the effect of gravity and vertical heterogeneity. Zhong et al. (2018) [12] and Tadayoni et al. (2020) [13] pointed out that reservoir heterogeneity and structural complexity significantly influence fluid flow behaviour, which cannot be fully captured in simplified models. Although numerical simulation methods can realize the simulation of large-thickness models, when facing multiple blocks, multiple well groups and multiple influencing factors, the simulation process is cumbersome, with many simulation times and a heavy workload. At the same time, both physical simulation and numerical simulation only correlate the change in reservoir water cut in the current research on the main controlling factors of water cut [14], which includes the conventional high water cut development phenomenon caused by high production and high recovery degree, making it impossible to accurately identify the main controlling factors of abnormal reservoir water cut rise, difficult to provide a precise direction for reservoir development optimization, and thus lacking practical value. Meanwhile, recent studies have provided important insights into the evolution of permeability and fracture structures in geomaterials under complex physical conditions. For example, Teng et al. (2026) [15] revealed the coupled relationship between permeability and energy evolution in coal under unloading confining pressure, while Wu et al. (2026) [16] demonstrated the progressive development of fracture networks and mechanical deterioration in granite under cyclic thermal and cryogenic conditions. These studies highlight the intrinsic linkage between structural evolution, permeability variation, and fluid flow behaviour, which provides an important physical basis for understanding the formation of high-permeability pathways in reservoirs. However, these studies are mainly conducted at the laboratory scale and focus on single-material systems, lacking direct applicability to reservoir-scale development dynamics.

To accurately identify the main controlling factors of water cut rise in thick carbonate reservoirs, this study innovatively proposes a multi-scale hierarchical main controlling factor screening method. This method uses numerical simulation to segment the target reservoir and obtain multiple typical blocks. Through dynamic analysis of the three levels (reservoir, block, and well group), the law of reservoir water cut rise, water flooding sweep path, and well group water cut rise type are clarified [17]. Subsequently, a multi-sequence grey correlation method is proposed by optimizing the grey correlation method, which correlates block water cut, production, and recovery degree to screen out the common main controlling factors of abnormal high water cut in multiple blocks. On this basis, numerical simulation is used to conduct multi-factor sensitivity analysis for a single block, and relying on the hierarchical screening idea of reservoir-block-well group, the main controlling factors of water cut rise at different levels are accurately identified [18].

1.1. Geological Characteristics of the Target Reservoir

The target block selected in this study is the Middle East H Oilfield, which is located in a province in southeastern Iraq and develops a thick carbonate reservoir. This reservoir has experienced five phases of water level rise and fall, forming a multi-stage gentle slope sedimentary model. Controlled by the carbonate sedimentation–diagenesis coupling effect, it has a complex sedimentary structure and strong reservoir heterogeneity, serving as a typical representative of thick carbonate reservoirs in the Middle East.

Among them, the MB1 reservoir is a layered edge-water reservoir developed in the M Formation of the H Oilfield, with three small layers developed vertically, a burial depth range of 3000–3100 m, and an average reservoir thickness of 97 m, which belongs to the category of typical thick carbonate reservoirs. Both interbeds and high-permeability zones are well-developed inside the reservoir. The high-permeability zones are mainly concentrated in the channel and shoal sedimentary facies, distributed in strip, dot or sheet shapes. The overall reservoir is dominated by poor reservoirs, intercalated with layered medium-good reservoirs composed of shoal-channel sediments, and the connectivity of reservoir units is complex. The core part of the reservoir is a medium-thick layer-sheet medium-good reservoir formed by the superposition of channels and shoals, with relatively strong dissolution and transformation, and the scale of reservoir connectivity units is relatively large; the flank part is interbedded with thin layer-sheet medium-good reservoirs and lagoon facies poor reservoirs, with weak dissolution and transformation, and both reservoir connectivity and reservoir performance are weaker than those of the core part, the oil–water relationship profile of the M reservoir is shown in Figure 2.

In addition, controlled by sequence stratigraphic cycles and sedimentary landforms, the MB1 reservoir is developed in a semi-restricted platform environment, with intra-platform shoals formed within a lagoon-dominated depositional environment. The sedimentary microfacies are characterized by thin interbeds vertically and rapid lateral changes, which further intensify reservoir heterogeneity.

The reservoir of the MB1 oilfield is mainly of pore type, with complex pore structure and multi-modal characteristics. The porosity and permeability vary greatly, with a maximum difference of four orders of magnitude; the crude oil viscosity is generally low, ranging from 1.6 to 3.29 cp; the original formation pressure is 5027 psi, the saturation pressure is 2675 psi, the formation temperature is between 84 and 89 °C, and the pressure coefficient is 1.16,the pore-throat radius distribution characteristics of the MB1 reservoir are shown in Figure 3.

1.2. Development Characteristics of the Target Reservoir

From the perspective of development performance, the production of the MB1 reservoir has gone through a “rapid growth-fluctuation and stability” stage, showing strong development stability; the current recovery degree is low, and the water cut has steadily increased to 33% with the increase in recovery degree; injected water is observed in the structural high of the reservoir, formation water or mixed water is observed in the edge part, and 97 wells have seen water so far, accounting for 54% of the total number of wells, with a water cut rise rate of 5.2%,The production performance curves of the MB1 reservoir are shown in Figure 4, and the relationship between water cut and recovery percent is presented in Figure 5.

1.3. Development Challenges of the Target Reservoir

During the development of the MB1 reservoir, multiple modes including edge-water invasion, bottom water coning and injected water channelling coexist, leading to a relatively fast water cut rise rate of production wells. At present, it is urgent for the reservoir to find out the water flooding channel and water cut rise types, and clarify the main controlling factors of various water cut rise types and multiple water breakthrough modes, so as to provide support for the optimization and adjustment of development, The three main water invasion types and the production curves of typical wells in the MB1 reservoir are shown in Figure 6 and Figure 7, respectively.

2. Methodology

2.1. Research Method of Multi-Scale Hierarchical Dominant Controlling Factors

This study adopts a multi-scale hierarchical analysis framework of “reservoir–block–well group” to systematically identify the dominant controlling factors of water cut rise. At the reservoir scale, the primary types of factors affecting waterflooding performance are identified from an overall perspective, enabling preliminary screening of dominant factors.

At the block scale, the influence of these factors is further analyzed across different reservoir architecture types, highlighting their variability under distinct structural conditions. At the well-group scale, the analysis is refined from the perspective of local dynamic response to characterize the specific influence patterns and mechanisms of dominant factors on water cut evolution.

These three scales are not independent but constitute a progressive analytical framework from overall screening to differential analysis and finally to mechanism characterization, enabling a systematic refinement of dominant factor identification from macro to micro scales.

Based on this framework, a stepwise screening strategy is implemented. At each scale, candidate influencing factors are first identified and their contributions are quantitatively evaluated, after which the key dominant factors are determined. The results from different scales are then integrated to establish a unified dominant factor system under the combined control of reservoir, block, and well-group levels.

Compared with conventional approaches that simply classify influencing factors into geological and development categories, this method introduces a multi-scale perspective, allowing dominant factors to be identified through hierarchical screening. This not only reveals the differences in controlling mechanisms at different scales but also provides a clearer basis for subsequent reservoir development optimization.

2.2. Reservoir Division and Modelling Study of Typical Blocks

The MB1 reservoir presents a saddle-shaped structural form from northwest to southeast. Combined with the lateral changes in sedimentary facies belts, it results in significant differences in sedimentary facies superposition, reservoir combination, reservoir type and well pattern adaptability in different regions. The traditional overall research method makes it difficult to accurately characterize such differences. Therefore, carrying out zonal research based on structural position, sedimentary facies superposition and development performance is the key path to deepen reservoir understanding, optimize development decisions and realize efficient development.

Based on this, the entire reservoir is divided into three regions from the northwest flank to the southeast flank according to the sedimentary facies superposition characteristics, reservoir type and high-permeability zone distribution mode: The basis for the zonal study of the reservoir is summarized in

Table 1. Basis for zonal study.

Architecture Zoning	Northwestern Part	Core Part	Southeastern Part
Reservoir type
Well pattern layout
HPS distribution pattern	Continuous distribution of HPS in the shoal	Banded distribution of HPS in channels	Mosaic distribution of HPS in shoals and channels
BBS distribution pattern
Production dynamics of each region

. The reservoir in the northwest flank is an edge-water reservoir, with shoals deposited under lagoonal conditions, and the shoals are distributed in a clamped shape; the well pattern layout is mainly a row well pattern composed of vertical wells and horizontal wells; the interlayer at the top of the structure is fully developed, and the internal interbeds are crisscrossed, accounting for 22.7%; the high-permeability zones are distributed in patches, and the dominant flow units account for 10.8%. The reservoir in the structural core has no edge and bottom water, the strip-shaped channels and shoals are superimposed and grown, forming the characteristics of interbedded development of medium-good reservoirs; the well pattern layout is mainly a row well pattern composed of vertical wells; the interlayer at the top of the structure is fully developed, and the internal interbeds are crisscrossed, accounting for 19.0%; the high-permeability zones are distributed in strips, and the dominant flow units account for 19.5%. The reservoir in the southeast flank is an edge-water reservoir; the sedimentary microfacies are mainly composed of lagoons, shoal flanks and channels, showing an overall discontinuous and alternating distribution state; the well pattern layout is mainly an irregular well pattern composed of vertical wells and horizontal branch wells; the interlayer at the top of the structure is fully developed, and the internal interbeds are crisscrossed, accounting for 20.3%; the high-permeability zones are distributed in patches, and the dominant flow units account for 14.8%.

According to the above reservoir regional division, inside each structural region, blocks with typical characteristics are further divided to carry out targeted research on the main controlling factors of reservoir water cut rise.

Specifically, the northwest structural region is divided to obtain a typical block—the lagoon-shoal type, which is vertically composed of lagoons in the upper and lower parts and shoal flanks in the middle part, with uniform distribution of lagoons and shoal flanks on the plane; the central structural region is divided to obtain a typical block—the channel composite type, which is generally developed with strip-shaped channels superimposed with shoals, forming the state of interbedded development of medium-good reservoirs; the southeast structural region is divided to obtain a typical block—the composite composition type, which is generally composed of multiple composite sedimentary microfacies, showing an alternating distribution state; at the same time, to support subsequent production performance analysis, numerical models of each typical block are established. The basic information of the typical block models is summarized in

Table 2. Basic information on typical block models.

Architecture Zoning	Lagoon-Shoal Type		Channel Composite Type		Composite Composition Type
Basic model information	Number of grids	29,676	Number of grids	39,334	Number of grids	40,684
	Average porosity	14.7%	Average porosity	13.1%	Average porosity	14.6%
	Average Permeability	37.8 mD	Average Permeability	55.4 mD	Average Permeability	48.7 mD
	Oil saturation	57.1%	Oil saturation	48.7%	Oil saturation	59.2%
Sedimentary facies stacking pattern
Plane distribution pattern
Reservoir model
Sedimentary rhythm	Fluctuating composite rhythm		“K-Type” composite rhythm		“>-Type” composite rhythm
Figure legend

.

2.3. Multi-Sequence Grey Relational Analysis

Guided by the multi-scale hierarchical screening method for dominant controlling factors, the main factors affecting water cut rising at different scales can be preliminarily identified. If numerical simulation is directly adopted to conduct multi-scale and multi-factor simulations for the three blocks respectively at this stage, and taking three variable levels designed for each influencing factor as an example, the workload of numerical simulation will be extremely heavy and the research process will become extraordinarily complicated.

In this study, a multi-sequence grey relational analysis method is proposed for the common screening of dominant factors controlling water cut in multiple blocks. Grey relational analysis (GRA), originally proposed by Deng (1982) [19], is a widely used method for analyzing systems with incomplete information and small sample sizes. It has been extensively applied in engineering systems, energy evaluation, and reservoir performance analysis due to its ability to quantify the degree of influence between multiple factors and system responses (Liu et al., 2013; Deng, 1989) [20,21].

Combined with the core problem of identifying the dominant factors affecting water cut rising in this study, if only a single water cut is used as the relational sequence, it will cover the conventional phenomenon of high water cut caused by high production and high recovery degree in the block, and fail to accurately identify the dominant factors of abnormal water cut rise under the background of low production and low recovery degree. Therefore, based on the conventional grey relational analysis method, this study proposes a multi-sequence grey relational analysis method, expanding the relational sequence from a single water cut to water cut, production, and recovery degree. Among them, water cut has a positive correlation effect on the calculation of relational degree, while production and recovery degree have a negative correlation effect and participate in the calculation of relational degree.

At this stage, the introduction of the multi-sequence grey relational analysis method in this study is extremely appropriate. For the three blocks, after the preliminary screening of the dominant factors at three scales, if numerical simulation analysis is adopted, although it can compare the influence degrees of the dominant factors at different scales, this method is only applicable to the analysis of a single block. Due to the distinct characteristics of each of the three blocks, when the research objects are different, numerical simulation cannot realize the horizontal comparison of common influencing factors among blocks. In contrast, the grey relational analysis (GRA) method shows extremely strong applicability: it can realize cross-regional comparison, combine the water cut, production and recovery degree of different blocks to form the reference sequence of the multi-sequence grey relational analysis method. By calculating the relational degree, the common dominant factors affecting water cut rising among different blocks can be accurately identified. This process is the rough screening of the dominant factors controlling water cut, which greatly reduces the workload of subsequent numerical simulation. The respective roles of grey relational analysis and numerical simulation in this study are compared in

Table 3. Comparison of research methods for dominant controlling factors.

		Grey Relational Analysis	Numerical Simulation Analysis
Research object		Horizontal comparison of the three blocks	In-depth analysis of a single block
Research goal		Searching for cross-block common factors	Discovering Intra-block Individual Factors
Advantages and disadvantages		Fast and efficient, suitable for preliminary diagnosis; only reveals correlation with weak prediction ability	Reliable analysis and reveals causal relationships; multi-factor analysis requires massive simulations

.

Meanwhile, after the rough screening of dominant factors, the numerical simulation method is adopted to conduct detailed analysis and research on individual blocks. In this study, the multi-sequence grey relational analysis method is used to roughly screen the common dominant factors, and then numerical simulation is employed to perform sensitivity analysis on the dominant factors of each block. The grey relational analysis (GRA) method and numerical simulation method are closely combined, giving full play to the advantages of the two research methods. This combined method not only simplifies the entire research workflow but also does not lose the accuracy of the dominant factor research, balancing work efficiency and research depth.

2.4. Method Comparison and Justification

In the analysis of dominant controlling factors in reservoir development, traditional methods typically employ a single indicator for correlation analysis. For example, water cut is often used as the sole reference sequence to evaluate the influence of various factors on water cut evolution. Although such methods can reflect the relationship between influencing factors and a single development indicator, they exhibit certain limitations in multi-index evaluation and cross-block comparative analysis.

First, a single reference sequence cannot comprehensively capture the dynamic characteristics of reservoir development. Water cut, production rate, and recovery factor represent different aspects of the waterflooding process. Relying solely on water cut may overlook the influence of production variation and reservoir utilization degree, leading to a certain degree of bias in the identification of dominant controlling factors.

Second, in multi-block comparative analysis, single-indicator methods face difficulties in establishing a unified evaluation standard across different blocks. Due to variations in reservoir architecture and development characteristics, it is challenging to identify common controlling factors across different blocks when only a single indicator is used.

To address these limitations, this study introduces a multi-sequence grey relational analysis method, in which water cut, production rate, and recovery factor are jointly used as reference sequences to construct a multi-index evaluation framework. This approach enables a more comprehensive characterization of reservoir development behaviour and provides a more reliable assessment of the influence of various factors.

Furthermore, the multi-sequence grey relational method demonstrates clear advantages in cross-block comparison. By establishing a unified multi-sequence reference system, it allows for consistent evaluation of influencing factors under different reservoir architecture conditions, thereby effectively identifying common dominant factors across different blocks. This feature is particularly important for dominant factor screening in complex reservoirs and provides a reliable basis for subsequent numerical simulation and sensitivity analysis. The differences between the traditional single-sequence method and the multi-sequence grey relational analysis used in this study are summarized in

Table 4. Comparison of dominant controlling factor identification methods.

Comparison Aspect	Traditional Single-Sequence Grey Relational Analysis	Multi-Sequence Grey Relational Analysis (This Study)
Research object	Single block or single development stage	Comparative analysis across multiple blocks
Reference sequence	Single indicator (e.g., water cut)	Multiple indicators (e.g., water cut, production rate, recovery factor)
Analysis capability	Reflects partial relationships with certain limitations	Integrates multiple development dynamics with more comprehensive results
Cross-block comparability	Difficult to achieve a unified comparison across different blocks	Enables unified evaluation and comparison across different blocks
Applicability	Suitable for single-block analysis	Suitable for complex reservoirs and multi-block comparative analysis

.

2.5. Application of Multi-Sequence Grey Relational Analysis in Typical Blocks

To implement the multi-sequence grey relational analysis method in this study, the typical blocks identified in the previous section were selected as the analysis objects, namely the lagoon-shoal block, channel composite block, and composite composition block. A comparative analysis of the dominant controlling factors under different reservoir architecture conditions was then conducted.

Based on the numerical simulation results of each typical block, water cut (WCT), production rate, and recovery factor (RF) were selected as the reference sequences to construct a multi-sequence relational analysis framework. Among them, water cut reflects the dynamic behaviour of waterflooding, production rate characterizes the development response capacity, and recovery factor represents the degree of reserve utilization. These three indicators comprehensively describe reservoir development performance from different perspectives, thereby avoiding the limitations associated with single-index analysis.

On this basis, combined with production performance analysis and numerical simulation results, potential influencing factors at different scales were selected as comparison sequences. These include key parameters such as HPS permeability, well pattern layout, and injection–production intensity. All factor data were derived from the dynamic responses and simulation outputs of the corresponding blocks, ensuring consistency between the relational analysis and actual reservoir development characteristics.

Through dimensionless processing of the reference and comparison sequences, the grey relational degree was calculated to quantitatively evaluate the influence of each factor on the multi-sequence reference system. Based on this, the correlation results of the three types of typical blocks were compared to identify the dominant controlling factors under different reservoir architecture conditions, and further determine the common controlling factors shared among different blocks.

It should be noted that the application of grey relational analysis in this study focuses on cross-block comparative analysis rather than ranking within a single block. Through cross-block comparison, key factors that exert a significant influence under different reservoir architecture conditions can be effectively identified. This provides a basis for subsequent sensitivity analysis and reduces the complexity of variable selection in numerical simulation.

2.6. Innovations and Methodological Contributions

(1) From factor-type classification to multi-scale hierarchical analysis: enhancing the structural identification of dominant factors.

Previous studies typically classify the factors influencing water cut rise into geological and development factors and analyze them within a single scale, which makes it difficult to capture the variability of controlling mechanisms across different spatial scales.

In this study, a multi-scale framework of “reservoir–block–well group” is introduced to enable hierarchical identification and comparative analysis of influencing factors at different scales. This approach not only reveals the differences in dominant factors across scales, but also clarifies their hierarchical relationships, thereby upgrading dominant factor identification from single-level evaluation to a structured, progressive analysis.

(2) From single-sequence to multi-sequence grey relational analysis: enabling cross-block comparison.

Conventional grey relational analysis usually employs water cut as the sole reference sequence, which may confuse normal high water cut under high production and recovery conditions with abnormal water breakthrough, and makes it difficult to achieve consistent comparison across different blocks.

In this study, the reference sequences are extended to a multi-sequence combination of water cut, production rate, and recovery factor. This enables characterization of reservoir development from multiple dimensions and supports a more comprehensive relational analysis.

Moreover, the proposed method establishes a unified evaluation framework across different blocks, allowing horizontal comparison and integrated screening of dominant factors, thereby improving both the accuracy and applicability of the results.

(3) From single-method analysis to a coupled workflow of grey relational analysis and numerical simulation: improving reliability and efficiency.

In previous studies, dominant factor identification is typically based on a single method: either grey relational analysis is directly used to determine influencing factors, or numerical simulation is employed to conduct multi-factor comparisons. The former lacks physical mechanism support, while the latter is computationally intensive and inefficient.

In this study, these two approaches are integrated into a coupled workflow. Grey relational analysis is first applied to preliminarily screen key influencing factors, followed by numerical simulation-based sensitivity analysis to quantitatively evaluate their mechanisms and impact.

This coupled approach reduces computational complexity while significantly improving the reliability and robustness of the analysis results, The overall technical workflow of the study is shown in Figure 8.

3. Case Study

3.1. Block Dynamic Analysis

The lagoon-shoal block is characterized by obvious stable production with noticeable production fluctuation. It is currently in a stable production stage, with a slow water cut increase and a stable gas-oil ratio. The channel composite block shows a significant production decline, accompanied by a rapid increase in water cut. The composite composition block exhibits an obvious production increase, with an extremely slow rising trend of water cut. The production performance of the three typical blocks is compared in Figure 9, and their relationships between water cut and recovery degree are presented in Figure 10.

In the lagoon-shoal block, the shoal bodies are continuously distributed. The injected water advances slowly along the shoals, resulting in uniform and efficient water flooding. The relationship curve between water cut and recovery degree shows a J-shaped increase.

In the channel composite block, the channels are banded and embedded in a lagoon background. The injected water rushes rapidly along the tidal channels. The relationship curve between water cut and recovery degree shows an S-shaped increase with a fast water cut rise.

In the composite composition block, shoals, shoal flanks, channels and lagoons are discontinuously distributed, leading to complex fluid flow. The relationship curve between water cut and recovery degree shows an L-shaped increase.

3.2. Water Cut Rising Types of Well Groups

Representative WCT versus cumulative oil production curves for fast water-breakthrough, slow-rising, and non-water-cut well groups are shown in Figure 11, Figure 12 and Figure 13, respectively. Fast water-breakthrough well groups: These well groups mostly occur in lagoon-shoal and channel composite blocks. Water breaks through early in the production process, and the water cut rises rapidly after the breakthrough. The curve of water cut versus cumulative oil production shows a rapid S-shaped increase. The water cut characteristics of such well groups are mainly affected by high-permeability layers. The injection–production wells are located in high-permeability channels and continuous shoal bodies. Meanwhile, affected by edge-water reservoirs, the injection–production wells are close to the edge water, resulting in early water breakthrough and a fast water cut rising in the well groups.

Slow-rising well groups: These well groups feature late water breakthrough, and water cut rises slowly after breakthrough. The curve of water cut versus cumulative oil production shows a slow J-shaped increase with no obvious inflexion point. The water cut characteristics of such well groups are mainly affected by injection–production intensity and injection–production connectivity. One scenario is good inter-well connectivity but overall low injection–production intensity. The other is that injection and production wells are located in different sedimentary facies, resulting in poor injection–production connectivity and slow water cut rise.

Non-water-cut well groups: These well groups are usually non-responsive well groups with almost no water breakthrough inside. The curve of water cut versus cumulative oil production is approximately linear. The water cut rising characteristics of such well groups are mainly affected by injection–production connectivity and well pattern arrangement. The core factor is that the well groups are not effectively stimulated. Due to the distribution of interlayers, injection barriers are formed, resulting in non-responsive well groups and almost no change in water cut. Another situation is the unreasonable arrangement of injection–production well patterns, which also leads to non-responsive well groups and nearly unchanged water cut.

3.3. Study on Water Flooding Sweep Channels

Reservoir models under different reservoir architectures were established through numerical simulation, and water-flooding front tracers were added to the reservoirs. The numerical simulation results clearly show that the lagoon-shoal block has a large sweep efficiency in both areal and vertical directions, with uniform overall sweep. This result is mainly attributed to the continuous distribution of beach bodies. The large-area continuous distribution of good-quality reservoirs provides gentle flow channels during water flooding, leading to a wide sweep range with a water-flooding sweep efficiency of 64.5%. However, the overall injection–production intensity is relatively low, and the sweep efficiency is low in the facies-change zones between shoal bodies and lagoons, showing an overall areal sweep pattern. The overlay of streamlines and water saturation, together with the areal and vertical water-flooding front tracer responses of the lagoon-shoal block, are shown in Figure 14, Figure 15 and Figure 16, respectively.

The overlay plot of streamlines and water saturation for the channel composite block reveals its unique water-flooding sweep channels. Two streamline-concentrated zones appear in the block, showing distinct flow channels. Vertically, the water-flooding sweep channels are banded along the channels, while perpendicular to the channels, water flooding exhibits a spotted sweep. The overall pattern is characterized by a banded sweep. The overlay of streamlines and water saturation, together with the areal and vertical water-flooding front tracer responses of the channel composite block, are shown in Figure 17, Figure 18 and Figure 19, respectively.

The composite composition block is dominated by a lagoon background, with shoals, shoal flanks and channels distributed within it. Medium-good reservoirs are relatively continuous on the plane, but their continuity is poorer than that in the lagoon-shoal block. Meanwhile, due to the irregular well pattern in the southeastern structural area, the outer part of the composite structural block is injected by vertical wells, and the central part is produced by horizontal wells. This arrangement results in an unswept central area, forming an edge-type sweep pattern.

Figure 20 shows the well pattern layout and well trajectories in the composite superimposed reservoir. It can be observed that injection wells are mainly distributed along the outer boundary of the region, while production wells are predominantly horizontal wells located in the central area.

Based on the superimposed map of streamlines and water saturation (Figure 21), it can be seen that under this well pattern configuration, the central part of the reservoir is difficult to be effectively swept. The areal and vertical water-flooding front tracer responses of the composite composition block are further shown in Figure 22 and Figure 23, respectively.

4. Study on the Main Controlling Factors of Water Cut Rising

Based on the above dynamic analyses of reservoirs, blocks and well groups, nine main geological and development factors influencing water cut rising were initially screened at the reservoir, block and well group scales. The screening idea of these main controlling factors is fully integrated with actual development practices. Through the progressive and step-by-step identification method, the main controlling factors affecting water cut rising can be accurately identified. The preliminary screening results of the main controlling factors of water cut rise at different scales are listed in

Table 5. Primary selection of the main controlling factors for water cut rising.

		Lagoon-Shoal Block—Channel Composite Block—Composite Composition Block
Reservoir scale		Heterogeneity	Stacked architecture	HPS permeability
Block scale		Well pattern layout	Oil production rate	Injection–production corresponding relation
Well group scale		Injection–production intensity	Injection–production connectivity	Well spacing

.

By using the multi-sequence grey correlation method, the correlation sequence was extended from a single water cut to water cut, production rate, and recovery degree. Water cut has a positive correlation effect on the calculation of correlation degree, while production rate and recovery degree are treated as negative correlations in the calculation.

Terminology Definition:

1. Injection–production correspondence relationship: The injection–production correspondence relationship refers to the configuration matching between injection wells and production wells in terms of vertical reservoir layers or perforated intervals, which is used to characterize the layer-wise matching relationship between injection positions and production positions.

2. Injection–production connectivity:

Injection–production connectivity refers to the ability of effective seepage flow channels to be established between injection wells and production wells through the reservoir, which is used to characterize the strength of interwell flow communication.

Application in this study:

1. In this study, injection–production correspondence is considered as a key controlling factor at the block scale. It is used to characterize the balance of waterflood sweep among different layers under various injection schemes and is quantitatively described by the layer-wise sweep efficiency (upper, middle, and lower layers) and its variability.

2. In this study, injection–production connectivity is mainly used to characterize the reservoir communication condition between injection wells and production wells, that is, whether injected water can reach production wells through effective seepage pathways and how strong the interwell flow communication is. In the dominant-factor analysis at the well-group scale, this parameter is used to identify whether effective waterflood response conditions exist between injection and production wells.

4.1. Calculation Steps of the Grey Correlation Method

1. Determine the reference sequence and comparison sequences.

Reference Sequences:

$$X_0=\{x_0(k)/k=1,2,\dots ,n\}$$

(1)

Comparison Sequences:

$$X_i=\{x_i(k)/k=1,2,\dots ,n\}, i=1,2,\dots ,m$$

(2)

where $X_0$ is the reference sequence representing the variation characteristics of the evaluation object, and $x_0(k)$ is the reference value at the $k-th$ sample point. $X_i$ is the $i-th$ comparison sequence representing the value series of the $i-th$ influencing factor, and $x_i(k)$ is its value at the $k-th$ sample point. $m$ and $n$ denote the number of comparison sequences and sample points, respectively.

2. Dimensionless processing of data.

To eliminate the differences in dimensions and orders of magnitude, the raw data are standardized.

$$x_i^{\prime }(k)={\displaystyle \frac{x_i(k)}{{\overline{x}}_i}}, {\overline{x}}_i={\displaystyle \frac{1}{n}}{\displaystyle \sum _{k=1}^n{x}_i}(k)$$

(3)

where $x_i^{\prime }(k)$ is the dimensionless value of the $i-th$ comparison sequence at the $k-th$ sample point after normalization, $X_i$ is the mean value of the $i-th$ comparison sequence, and $x_i(k)$ is the original data before normalization.

3. Calculate the absolute difference sequence.

$${\Delta }_i(k)=|x_0^{\prime }(k)-x_i^{\prime }(k)|$$

(4)

where ${\Delta }_i(k)$ denotes the absolute difference between the reference sequence and the $i-th$ comparison sequence at the $k-th$ sample point, and $x_0^{\prime }(k)$ and $x_i^{\prime }(k)$ are the normalized values of the reference sequence and comparison sequence, respectively.

4. Calculate the range.

$${\Delta }_{\min }={\min }_i{\min }_k{\Delta }_i(k), {\Delta }_{\max }={\max }_i{\max }_k{\Delta }_i(k)$$

(5)

where ${\Delta }_{\min }$ and ${\Delta }_{\max }$ denote the minimum and maximum values of the absolute differences between the reference sequence and all comparison sequences, respectively.

5. Calculate the correlation coefficient.

$${\xi }_i\left(k\right)={\displaystyle \frac{{\Delta }_{\min }+\rho {\Delta }_{\max }}{{\Delta }_i\left(k\right)+\rho {\Delta }_{\max }}}, \mathrm{where} \rho \mathrm{is} \mathrm{the} \mathrm{resolution} \mathrm{coefficient}, \rho = 0.5.$$

(6)

where ${\xi }_i\left(k\right)$ denotes the grey relational coefficient of the $i-th$ comparison sequence at the $k-th$ sample point, reflecting the similarity between the comparison sequence and the reference sequence.

6. Calculate the grey correlation degree.

The average value of the correlation coefficients is taken to obtain the grey correlation degree:

$$r_i={\displaystyle \frac{1}{n}}{\displaystyle \sum _{k=1}^n{\xi }_i}\left(k\right)$$

(7)

where $r_i$ is the grey relational degree between the $i-th$ comparison sequence and the reference sequence, representing the overall correlation level of the influencing factor with respect to the reference indicator, and $n$ is the number of sample points. A larger $r_i$ indicates a higher similarity in variation trends between the comparison sequence and the reference sequence, implying a stronger influence of the corresponding factor on the research object.

7. Ranking and Analysis of Correlation Degrees.

The sequences are sorted in descending order of the correlation degree $r_i$, a larger correlation degree indicates a more significant influence of the comparison sequence on the reference sequence.

4.2. Study on Main Controlling Factors at Reservoir Scale

The main controlling factors at the reservoir scale include reservoir heterogeneity, Stacked architecture, and HPS permeability. Reservoir heterogeneity is characterized by the permeability mutation coefficient in the X-direction of the reservoir. The vertical stacking characteristics of reservoirs in different blocks are represented by the permeability mutation coefficient in the Z-direction. The development degree of high-permeability zones is quantified by the ratio of high-permeability zone thickness to total reservoir thickness. The correlation calculation results for the dominant factors at the reservoir scale are presented in

Table 6. Correlation calculation of dominant factors at the reservoir scale.

Sample Number	Reference Sequences X1 (WCT)	Reference Sequences X2 (Cum. Prod.)	Reference Sequences X3 (RF)	Comparison Sequences 1	Comparison Sequences 2	Comparison Sequences 3
				Heterogeneity	Stacked Architecture	HPS Permeability
1- Lagoon-shoal block	11%	15.36 MMstb	6.80%	1.73	2.18	9%
2- Channel composite block	54%	45.80 MMstb	9.15%	2.9	3.54	15%
3- Composite composition block	8.2%	13.50 MMstb	4.55%	2.36	3.08	8%

.

4.3. Study on Main Controlling Factors at Block Scale

At the block scale, the main controlling factors are well pattern layout, oil production rate, and the injection–production corresponding relation. The perfection of well pattern layout is characterized by the sweep efficiency and stimulation degree of production wells. The oil production rate of the block is represented by the ratio of annual oil production to geological reserves. The equilibrium degree of injected water advance under different injection modes is quantified by the mutation coefficient of sweep degree data for each sublayer. The correlation calculation results for the dominant factors at the block scale are presented in

Table 7. Correlation calculation of dominant factors at the block scale.

Sample Number		Lagoon-Shoal Block	2- Channel Composite Block	3- Composite Composition Block
Reference sequences X1 (WCT)		11%	54%	8.2%
Reference sequences X2 (Cum. Prod.)		15.36 MMstb	45.80 MMstb	13.50 MMstb
Reference sequences X3 (RF)		6.80%	9.15%	4.55%
Comparison sequences 1—well pattern layout	Receiving efficiency	70%	58.3%	62.5%
	Sweep degree	64.5%	49.6%	39.1%
	Comprehensive calculation	67.25%	53.9%	50.8%
Comparison sequences 2—oil production rate		1.06%	0.71%	0.65%
Comparison sequences 3—injection–production corresponding relation	Upper-layer sweep degree	68.3%	54.6%	11.8%
	Middle-layer sweep degree	77.5%	57.1%	42.5%
	Lower-layer sweep degree	50.8%	54.3%	51.8%
	Comprehensive Calculation	20.7%	2.8%	59.2%

.

4.4. Study on Main Controlling Factors at Well Group Scale

At the well Group scale, the dominant controlling factors include injection–production intensity, injection–production connectivity, and well spacing. The injection–production intensity of the well group is characterized by the injection–production ratio. The injection–production connectivity is represented by the ratio of the number of actually responding production wells to the theoretically responding production wells within the well pattern. The well spacing is characterized by the average distance between injection wells and production wells. The correlation calculation results for the dominant factors at the well-group scale are presented in

Table 8. Correlation calculation of dominant factors at the well-group scale.

Well Group	Reference Sequences X1 (WCT)	Reference Sequences X2 (Cum. Prod.)	Reference Sequences X3 (RF)	Comparison Sequences 1	Comparison Sequences 2	Comparison Sequences 3
				Injection–Production Intensity	Injection–Production Connectivity	Well Spacing
Group-1	22%	6.71 MMstb	0.11%	1.02	75%	1873.5 ft
Group-2	65%	5.12 MMstb	2.6%	2.6	50%	1869.2 ft
Group-3	14%	4.74 MMstb	0.66%	0.66	75%	2615.7 ft

.

4.5. Results of Correlation Degree Calculation

Through preliminary screening of dominant factors using the multi-sequence grey correlation method with water cut, production and recovery degree as reference sequences, HPS permeability, well pattern layout and injection–production intensity are initially identified as the core dominant factors controlling water cut rise. These results not only narrow the variable range for subsequent numerical simulation analysis and avoid the sharp increase in simulation complexity caused by multi-factor coupling, but also clarify the research focus and provide a clear direction for reservoir development adjustment and water cut rise control. The ranking results of the dominant factors at different scales are summarized in

Table 9. Ranking of dominant factors at different scales.

Reservoir Scale	Correlation Degree	Rank	Block Scale	Correlation Degree	Rank	Well Group Scale	Correlation Degree	Rank
HPS distribution	0.9055	1	Well pattern layout	0.9795	1	Injection–production intensity	1.1775	1
Heterogeneity	0.8565	2	Injection–production corresponding relation	0.9245	2	Injection–production connectivity	0.849	2
Stacked architecture	0.8025	3	Oil production rate	0.8695	3	Well spacing	0.826	3

. The calculated correlation degrees and ranking results are further illustrated in Figure 24.

4.6. Integrated Analysis of Dominant Controlling Factors at Multiple Scales

By integrating the results from the reservoir, block, and well-group scales, it can be observed that the dominant controlling factors identified at different scales are consistent in type. These factors mainly include high-permeability layer permeability, well pattern layout, and injection–production intensity. From different perspectives, these factors reflect reservoir heterogeneity, development strategy, and displacement driving conditions, respectively, and together constitute the primary factor system controlling water cut rise.

Although the types of dominant factors are consistent, their manifestations and influence degrees vary across different scales. At the reservoir scale, the analysis focuses on the overall identification of dominant factor types. At the block scale, it highlights the differences in the influence of dominant factors under various reservoir architecture conditions. At the well-group scale, the analysis further reveals the specific mechanisms and dynamic response characteristics of these factors. Therefore, the results at different scales are not independent or contradictory, but rather represent the same controlling mechanisms from different perspectives.

From the perspective of reservoir architecture, the lagoon-shoal type is mainly controlled by injection–production intensity and well pattern layout. The channel composite type is jointly controlled by high-permeability layer permeability and well pattern layout. In contrast, the composite superimposed type is primarily dominated by well pattern layout, with injection–production intensity playing a secondary role. These results indicate that the combinations and relative importance of dominant factors vary under different reservoir architecture conditions.

Overall, high-permeability layer permeability, well pattern layout, and injection–production intensity do not act independently but jointly form a coupled controlling system governing water cut evolution. Under different reservoir architectures, these factors interact in different combinations, resulting in distinct water cut rise characteristics.

5. Sensitivity Analysis of Dominant Factors Based on Numerical Simulation

5.1. Establishment of a Simplified Model

During the study of dominant factors controlling water cut in the target reservoir, the actual geological and development conditions are characterized by strong complexity and heterogeneity. The reservoir involves numerous complex influencing factors, including differences in pore structure, uneven distribution of sedimentary microfacies, and complex fluid seepage laws. These factors are interrelated and mutually restricted, forming a multi-dimensional and nonlinear complex system. If the simulation is directly conducted based on the original actual reservoir model, the excessive model parameters and overly complex structure will lead to an exponential increase in computational load, greatly raising the computational cost and time consumption. Furthermore, the interference of minor factors may mask the core mechanism of the dominant controlling factors of water cut, making it difficult to accurately identify and quantify the degree of influence of each dominant factor on water cut variation, thus affecting the scientificity and reliability of the research conclusions.

Therefore, to achieve accurate identification and in-depth investigation of the dominant factors controlling water cut in the reservoir, a reasonable simplification of the actual reservoir model is of great necessity and scientific significance. The core objective of model simplification is to retain the key geological characteristics of the reservoir, the fundamental laws of fluid flow, and the dominant factors governing water cut variation, while eliminating minor factors with extremely low correlation to the main controlling factors of water cut, and simplifying the complex parameter system and structural configuration, so as to establish a simplified simulation model with rationality and operability.

5.1.1. Principles and Contents of Model Simplification

In this study, simplified models are constructed to reveal the controlling effects of different reservoir stacking patterns on waterflooding development behaviour. These models correspond to the typical blocks identified in the previous sections, with emphasis on highlighting the differences among various reservoir architecture types.

The simplified models are mechanism-oriented numerical models established using the tNavigator (2023) simulation software. The key objective is to preserve essential geological features, including reservoir stacking relationships, interlayer petrophysical contrasts, and sedimentary rhythms, in order to analyze the dominant controlling factors of water cut rise. Based on this objective, the model simplification follows the principle of “preserving key geological characteristics while weakening secondary heterogeneity,” with targeted simplifications applied to geometric structure, interlayer relationships, and property distribution.

(1) Geometric dimensions unchanged

During the simplification process, the overall geometric dimensions of the original numerical model in the X, Y, and Z directions are kept unchanged. On this basis, the planar grids are regularized, and the number of grid cells in the X and Y directions is adjusted to integer values. The actual physical dimensions in each direction remain unchanged, ensuring model scale consistency while reducing grid numbers.

(2) Vertical layer merging

Vertically, the original model is simplified into 20 layers. The merging process is guided by sedimentary rhythms and interlayer petrophysical differences, whereby adjacent layers with similar depositional characteristics and permeability levels are combined. This treatment increases the thickness of individual layers while preserving the total thickness and interlayer configuration, thereby maintaining the vertical stacking structure of the reservoir.

(3) Averaging of intra-layer permeability

The permeability within each layer is averaged, and the mean permeability is used to represent the overall flow capacity of that layer. This approach reduces the influence of intra-layer local heterogeneity and small-scale preferential flow channels, while highlighting interlayer differences.

(4) Simplification of relative permeability and capillary pressure curves

Based on reservoir petrophysical characteristics, the reservoir is classified into four types: high-permeability layers, high-quality reservoir, medium-quality reservoir, and poor reservoir. Corresponding relative permeability curves and capillary pressure curves from the original model are assigned to each reservoir type, ensuring that differences in flow capacity among different reservoir types are preserved.

5.1.2. Core Parameter Settings

(1) Relative permeability curves

In this study, the relative permeability curves are directly inherited from the original numerical model, without re-fitting or manual modification. To adapt to the simplified reservoir representation, the original curves are categorized according to reservoir petrophysical properties into four types: high-permeability layers, high-quality reservoir, medium-quality reservoir, and poor-quality reservoir. Representative relative permeability curves are then selected for each reservoir type and applied in the simplified model (

Table 10. Basic information on the simplified model.

	1- Lagoon-Shoal Block		2- Channel Composite Block		3- Composite Composition Block
Basic information	Number of grids	30 × 18 × 20	Number of grids	30 × 30 × 20	Number of grids	30 × 30 × 20
	Model size	4500 × 2700 × 328	Model size	2250 × 2250 × 328	Model size	5292 × 6900 × 328
	Average porosity	16.8%	Average porosity	16.4%	Average porosity	16.2%
	Average permeability	37.6 mD	Average permeability	59.2 mD	Average permeability	47.4 mD
	Oil saturation	60%	Oil saturation	66.1%	Oil saturation	58%
Permeability distribution
Water saturation distribution
Relative permeability characteristics

).

(2) Capillary pressure model

The capillary pressure curves are also derived from the original numerical model and are categorized according to reservoir types during the simplification process. Specifically, based on reservoir petrophysical classification, the capillary pressure curves are assigned correspondingly to the relative permeability curves, forming four representative categories: high-permeability layers, high-quality reservoir, medium-quality reservoir, and poor-quality reservoir.

This approach preserves the fundamental shape and variation trends of the original capillary pressure curves while converting the original layer-by-layer assignment into a reservoir-type-based assignment, ensuring consistency with the simplified reservoir representation.

(3) Fluid property parameters

Fluid property parameters, including crude oil viscosity and formation water properties, are directly inherited from the original model without simplification or adjustment. By maintaining consistent fluid properties, the differences in flow behaviour between models are primarily attributed to reservoir structure and petrophysical distribution rather than fluid property variations.

(4) Initial water saturation

The initial water saturation distribution is retained from the original model without manual adjustment or reassignment. During the simplification process, the initial water saturation of merged layers is obtained through weighted averaging of adjacent layers, while maintaining the average water saturation level of each layer. This ensures the continuity of vertical water saturation distribution characteristics.

5.1.3. Criteria for Eliminating Secondary Factors

In the construction of the simplified model, the elimination of secondary factors is not based on empirical simplification, but on a targeted screening of influencing factors by integrating the results of dynamic analysis and numerical simulation.

First, according to the dynamic analysis results, the waterflooding process under different reservoir architecture types is mainly controlled by high-permeability layer permeability, well pattern layout, and injection–production intensity. These factors directly determine the waterflood propagation pathways and the characteristics of the water cut rise. In contrast, some factors only affect local flow details without altering the overall flow pattern or the evolution trend of water cut. Therefore, such factors are weakened or eliminated in the simplified model. The specific treatments are as follows:

(1) Neglect of minor structural factors

The original model includes a certain degree of formation dip; however, its magnitude is relatively small. At the scale of this study, such structural inclination has a limited impact on gravity segregation and the overall advancement direction of the waterflood front, and does not play a dominant role in controlling water cut evolution. Therefore, this factor is not explicitly represented in the simplified model.

(2) Merging of non-critical stratified structures

The original model contains finely subdivided vertical layers, among which some layers exhibit similar depositional characteristics and petrophysical properties. These layers are primarily used to describe local variations but have limited influence on interlayer contrasts and overall reservoir architecture. Therefore, layers with similar sedimentary rhythms and permeability levels are merged during simplification, preserving interlayer differences while reducing intra-layer variability.

(3) Averaging of secondary petrophysical fluctuations

For small-scale fluctuations in porosity and permeability, an averaging approach is applied under the premise of maintaining the average property level of each layer and the interlayer contrast relationships. This treatment reduces the influence of local randomness on simulation results.

The elimination of secondary factors follows the fundamental criteria of “not altering the overall flow pattern and not changing the direction of influence of dominant controlling factors.” Through this approach, while preserving reservoir architecture and interlayer petrophysical contrasts, the interference caused by multi-factor coupling is reduced, thereby improving the reliability of dominant factor identification and sensitivity analysis results.

5.1.4. Validation of Simplification

The above simplification processes inevitably weaken the influence of intra-layer heterogeneity and local high-permeability channels on flow behaviour, leading to some differences between the simplified model and the original model in terms of water breakthrough time and local dynamic responses. It should be emphasized that these differences are mainly reflected in the development stage division and local response details.

However, the simplified model preserves key characteristics such as geometric scale, interlayer petrophysical contrasts, and reservoir architecture. Therefore, from a structural perspective, the simplified model is capable of reproducing the fundamental flow behaviour and dominant controlling mechanisms under different reservoir architecture types, and can serve as a reliable base model for subsequent sensitivity analysis and dominant factor identification.

To further verify the applicability of the simplified model in dominant factor analysis, three types of typical blocks—lagoon-shoal, channel composite, and composite superimposed reservoirs—are selected. A comparative analysis of the water cut versus recovery factor curves between the original model and the simplified model is then conducted. The comparison between the original models and the simplified models is shown in Figure 25.

The simplified model established in this study is a mechanism-oriented model, whose primary objective is to highlight the controlling effects of different reservoir architectures on waterflooding development behaviour, rather than to achieve an exact match with the dynamic performance of the original model. Therefore, the simulation results of the simplified model are not expected to exhibit point-by-point agreement with those of the original model in terms of water breakthrough time and curve values. Meanwhile, the curves of the original model reflect the numerical simulation results at the current development stage, whereas the curves of the simplified model represent continued production under simplified conditions. As a result, differences exist in development stages and the recovery factor ranges between the two models. Under this premise, the focus of this validation is not on strict numerical correspondence, but on whether the simplified model can preserve the fundamental patterns of water cut evolution and curve characteristics under different reservoir architectures, as well as reasonably reflect the differences among various regions.

From the comparison results, the three types of reservoir architectures in the simplified model all retain water cut rise patterns and overall trends similar to those of the original model. The lagoon-shoal type exhibits a relatively slow increase in water cut at the early stage, followed by gradual acceleration in the later stage. In the simplified model, the corresponding curve maintains a low water cut level over a relatively long recovery period and then rises progressively, showing an overall shape consistent with that of the original model.

The channel composite type is characterized by a relatively rapid increase in water cut. In the original model, the curve rises sharply in the middle stage, while the simplified model similarly shows an early transition into the rising phase followed by continuous growth. The overall trend and curve morphology remain consistent between the two models.

For the composite superimposed reservoir, the simplified model exhibits a relatively high water cut at the early stage, which differs significantly from the original model. This discrepancy is mainly attributed to the weakening of intra-layer heterogeneity and local flow barriers during the simplification process. Nevertheless, as development proceeds, the water cut curve of the simplified model gradually enters a stable growth stage, and its late-stage behaviour is generally consistent with the water cut evolution trend reflected by the original model.

Overall, although differences exist between the simplified model and the original model in terms of water breakthrough time and local dynamic details, they are consistent in terms of water cut rise patterns, overall trends, and late-stage curve morphology across different reservoir architectures. This indicates that the simplified model can effectively capture the overall flow behaviour and dominant controlling mechanisms under different reservoir stacking patterns, and is therefore suitable for subsequent sensitivity analysis and dominant factor identification.

5.2. Sensitivity Analysis of Dominant Factors in the Lagoon-Shoal Block

5.2.1. Effect of HPS Permeability

The permeability of high-permeability layers in the block was set to five levels from 200 mD to 1000 mD. Variations in water cut, sweep efficiency, recovery degree, and water-free oil production period were compared. The numerical simulation results show that the presence of high-permeability layers leads to a rapid increase in water cut rise rate and a sharp decrease in sweep efficiency. Compared with the other two blocks, the lagoon beach-bar type has relatively homogeneous reservoir distribution. When the permeability of high-permeability layers reaches 1000 mD, the recovery degree decreases by 6%, the sweep efficiency drops by approximately 15%, and the water-free oil production period is shortened by 25 months. The WCT–RF curves and the corresponding variation trends of related parameters under different HPS permeabilities in the lagoon-shoal block are shown in Figure 26 and Figure 27, respectively.

5.2.2. Effect of Well Pattern Layout

Four well patterns were designed, including four-spot, seven-spot, inverted nine-spot, and liner well patterns. Variations in water cut, sweep efficiency, recovery degree, and water-free oil production period were compared. Numerical simulation results show that water cut and sweep efficiency are improved to varying degrees under different well patterns. The inverted nine-spot and line drive patterns have the highest well density, resulting in higher water cut. Among them, the inverted nine-spot pattern has the greatest impact on reservoir development performance: water cut increases by 8.2%, recovery degree rises by two percentage points, and the water-free oil production period is shortened by 28 months. The WCT–RF curves and the corresponding variation trends of related parameters under different well pattern layouts in the lagoon-shoal block are shown in Figure 28 and Figure 29, respectively.

5.2.3. Effect of Injection–Production Intensity

Five levels of injection–production intensity from 0.6 to 1.4 were set, and variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. Numerical simulation results show that with the increase in injection–production intensity, both water cut and sweep efficiency of the reservoir increase continuously. Compared with the original model with an injection–production ratio of 1.0, the water cut rising curve shows a similar shape with a slightly higher increasing rate. When the injection–production ratio reaches 1.4, the water cut rise slows down obviously. Compared with the injection–production ratio of 1.0, water cut increases by 10% and sweep efficiency improves by seven percentage points, while recovery degree only rises by two percentage points. Meanwhile, with the increase in injection–production ratio, the water-free oil production period gradually decreases, shortened by 13 months. The WCT–RF curves and the corresponding variation trends of related parameters under different injection–production intensities in the lagoon-shoal block are shown in Figure 30 and Figure 31, respectively.

5.3. Sensitivity Analysis of Dominant Factors in the Channel Composite Block

5.3.1. Effect of HPS Permeability

The permeability of high-permeability layers in the block was set to five levels from 200 mD to 1000 mD. Variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. The reservoir is a composite positive–negative rhythm reservoir with strong heterogeneity. Numerical simulation results show that the existence of high-permeability layers generally leads to an increase in water cut. The change in HPS permeability increases the water cut by approximately 10%. When the permeability of high-permeability layers reaches 1000 mD, the sweep efficiency decreases by 11%. The WCT–RF curves and the corresponding variation trends of related parameters under different HPS permeabilities in the channel composite block are shown in Figure 32 and Figure 33, respectively.

5.3.2. Effect of Well Pattern Layout

Four well patterns were designed, including four-spot, seven-spot, inverted nine-spot and liner well patterns. Variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. Numerical simulation results show that water cut and sweep efficiency of the reservoir are improved to varying degrees under different well patterns. Among them, the inverted nine-spot pattern has the greatest impact on reservoir development performance: water cut increases by 11.7%, recovery degree improves by nearly seven percentage points, and sweep efficiency increases by 16.7%. Different well patterns exert a significant influence on the fluctuation degree of various reservoir development indicators. he WCT–RF curves and the corresponding variation trends of related parameters under different well pattern layouts in the channel composite block are shown in Figure 34 and Figure 35, respectively.

5.3.3. Effect of Injection–Production Intensity

Five levels of injection–production intensity from 0.6 to 1.4 were set, and variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. Numerical simulation results show that with the increase in injection–production intensity, water cut and sweep efficiency of the reservoir increase continuously, and the water cut variation curves are similar. When the injection–production ratio exceeds 1, the water cut rise slows down obviously. Compared with the case at injection–production ratio of 1.0, when the ratio reaches 1.4, water cut increases by 5%, sweep efficiency improves by 9 percentage points, while recovery degree only rises by 2.4 percentage points. Meanwhile, due to the strong heterogeneity of the block, water breakthrough occurs immediately after water injection, so there is no water-free oil production period under different injection–production intensities. The WCT–RF curves and the corresponding variation trends of related parameters under different injection–production intensities in the channel composite block are shown in Figure 36 and Figure 37, respectively.

5.4. Sensitivity Analysis of Dominant Factors in the Composite Composition Block

5.4.1. Effect of HPS Permeability

The permeability of high-permeability layers in the block was set to five levels from 200 mD to 1000 mD. Variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. The reservoir is of a positive–negative composite rhythm with strong heterogeneity. Numerical results show that the water cut of the reservoir generally increases with the increase in permeability of high-permeability layers. When the permeability of high-permeability layers reaches 1000 mD, water cut rises by 10%. Meanwhile, the oil saturation is the highest in the upper part of the block, and the maximum recovery degree is obtained when the high-permeability layer is located at the high position. With the increase in permeability of high-permeability layers, sweep efficiency and recovery degree decrease rapidly, while water cut rises gradually. The WCT–RF curves and the corresponding variation trends of related parameters under different HPS permeabilities in the composite composition block are shown in Figure 38 and Figure 39, respectively.

5.4.2. Effect of Well Pattern Layout

Four well patterns were designed, including four-spot, seven-spot, inverted nine-spot and liner well patterns. Variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. Numerical simulation results show that the water cut of the reservoir is improved to varying degrees under different well patterns, and sweep efficiency gradually increases with the increase in well pattern density. Among them, the inverted nine-spot pattern exerts the greatest impact on the development performance of the block: water cut and sweep efficiency are increased by 10% and 8%, respectively. Different well patterns have a significant influence on the fluctuation degree of various reservoir development indicators. The WCT–RF curves and the corresponding variation trends of related parameters under different well pattern layouts in the composite composition block are shown in Figure 40 and Figure 41, respectively.

5.4.3. Effect of Injection–Production Intensity

Five levels of injection–production intensity from 0.6 to 1.4 were set, and variations in water cut, sweep efficiency, recovery degree and water-free oil production period were compared. The block is a positive–negative composite rhythm reservoir with the highest permeability in the middle zone. The injection–production layers in the mechanism model mainly cover the middle interval. Therefore, numerical simulation results show that the injection–production performance of the block reaches its limit at an injection–production ratio of 1. Further increasing the injection–production ratio only results in ineffective water circulation without improving the development performance. Thus, the influence of different injection–production intensities on water cut, sweep efficiency and recovery factor of the block is negligible. The WCT–RF curves and the corresponding variation trends of related parameters under different injection–production intensities in the composite composition block are shown in Figure 42 and Figure 43, respectively.

5.5. Analysis Results and Adjustment Suggestions

Based on the results of the aforementioned numerical simulation sensitivity analysis, the effects of different influencing factors in various types of regions were systematically summarized. Using water cut, sweep efficiency, and recovery factor as evaluation indicators, the response magnitude induced by variations in each factor was comparatively analyzed. The influence degrees of the dominant controlling factors on different indicators in the lagoon-shoal block, channel composite block, and composite composition block are shown in Figure 44, Figure 45 and Figure 46, respectively.

In the lagoon-shoal block, the influence of different factors can be ranked as follows: injection–production intensity exhibits the strongest overall impact, followed by well pattern layout, while the influence of HPS permeability is the weakest. Specifically, injection–production intensity mainly controls sweep efficiency and recovery factor, whereas well pattern layout has a more pronounced effect on water cut variation. In contrast, the impact of HPS permeability on all indicators is relatively limited. Overall, the development performance of this type of reservoir is primarily governed by injection–production intensity and well pattern layout, with a comparatively minor influence from the permeability of HPS.

In the channel composite block, the influence of different factors can be summarized as follows: HPS permeability and well pattern layout exhibit strong overall impacts, whereas injection–production intensity plays a relatively minor role. Specifically, HPS permeability has a significant effect on sweep efficiency and recovery factor, while well pattern layout shows a pronounced influence on both water cut variation and recovery factor. In contrast, the effect of injection–production intensity on all indicators is relatively limited. Overall, the development performance of this type of reservoir is primarily controlled by the combined effects of HPS permeability and well pattern layout, with injection–production intensity playing a secondary role.

In the composite composition block, the influence of different factors can be summarized as follows: well pattern layout exhibits the strongest overall impact, followed by injection–production intensity, while the influence of HPS permeability is relatively weak. Specifically, well pattern layout plays a dominant role in water cut and recovery factor, whereas injection–production intensity shows a more pronounced effect on water cut variation and sweep efficiency. In contrast, the influence of HPS permeability on all indicators is relatively limited. Overall, the development performance of this type of reservoir is mainly controlled by well pattern configuration and injection–production conditions, while the effect of HPS permeability is comparatively minor.

5.6. Development Recommendations Based on Dominant Controlling Factors

Although the three blocks have their own characteristics in structural location and sedimentary superposition pattern, the sensitivity analysis of main controlling factors based on numerical simulation shows that high-permeability layers and well pattern arrangement have the most significant impact on the water cut and development performance of the blocks, followed by injection–production intensity.

Based on the research results of the main controlling factors of water cut and combined with the current status of each block, corresponding optimization directions are proposed for the future development of each block. For the lagoon-shoal block, water cut rises steadily but the recovery factor is at a medium level; the reservoir is relatively evenly produced but there is local remaining oil. It is feasible to adjust injection–production parameters and deploy new wells to maintain the slow rising trend of water cut and improve the recovery factor to a high level.

For the channel composite block, the recovery factor is high but water cut rises rapidly; the reservoir has high-permeability dominant channels, which are prone to water channelling. The rising speed of water cut can be controlled by profile control and water plugging, optimizing perforation intervals and other methods, so as to avoid entering the high water cut stage prematurely and extend the stable production period with a high recovery factor.

For the composite composition block, water cut rises slowly but recovery factor is low, and the reservoir has a relatively low recovery factor. The reservoir recovery factor can be improved by adjusting well pattern arrangement and stratified water injection, so as to promote the recovery factor from a low level to a medium level and extend the stable production period with low water cut.

6. Conclusions

This study focuses on the MB1 reservoir, a typical ultra-thick carbonate reservoir in the Middle East. Aiming at the problems of rapid water cut increase and uneven displacement during waterflood development, a comprehensive investigation of water cut evolution and its main controlling factors was conducted by integrating production performance analysis with numerical simulation. The main conclusions are summarized as follows:

(1) In response to the strong heterogeneity and significant inter-block differences in ultra-thick carbonate reservoirs, a multi-scale analytical framework of “reservoir–block–well group” was established. A multi-sequence grey relational analysis method was introduced, in which water cut, production rate, and recovery factor were jointly used as evaluation criteria to achieve unified screening of dominant factors across different blocks. Compared with the conventional single water cut-based approach, this method constructs a multi-index evaluation system, enabling more effective identification of controlling factors for abnormal high water cut. Meanwhile, it facilitates cross-block comparison and improves the consistency in identifying dominant factors among different blocks.

(2) The patterns of water cut increase are primarily controlled by reservoir stacking architecture, with the essential mechanisms being the development of high-permeability layers and reservoir connectivity. Reservoir architecture governs the flow pathways and sweep efficiency of waterflooding, thereby determining the evolution of water cut and the distribution characteristics of remaining oil in different blocks.

(3) By integrating multi-sequence grey relational analysis with numerical simulation, the dominant controlling factors of water cut at different scales and in different blocks were systematically identified. Furthermore, sensitivity analysis was conducted to quantify the relative influence of these factors within each block.

(4) Based on the identified dominant factors, differentiated development strategies are proposed. For zones with well-developed high-permeability channels, profile control and water shutoff measures should be strengthened; for underproduced zones, optimization of well pattern and zonal injection–production schemes is recommended to improve sweep efficiency and enhance recovery.

(5) This study still has certain limitations. First, the multi-sequence grey relational analysis is essentially a correlation-based method, which has limited capability in revealing causal relationships among factors. Second, the numerical simulation model was simplified and homogenized during sensitivity analysis, and thus, the influence of local strong heterogeneity may not be fully captured. In addition, the study mainly focuses on typical blocks, and the applicability of the conclusions under more complex reservoir conditions requires further validation. Future work may involve more refined characterization of reservoir heterogeneity, incorporation of dynamic production data for model calibration, and exploration of more advanced data-driven approaches for dominant factor identification, in order to further improve the reliability and accuracy of the analysis.