A New Method of Geological Modeling for the Hydrocarbon Secondary Migration Research

Abstract

Reservoir geological modeling plays a crucial role in characterizing the spatial distribution and heterogeneity of subsurface reservoirs. The exploration of deep oil and gas resources is not only a global trend in the oil industry but also an inevitable choice for China to ensure energy security and achieve sustainable development in the oil and gas industry. Oil and gas exploration and development technologies have also made continuous breakthroughs, providing strong support for the sustained increase in China’s deep and ultra-deep oil and gas production. Deep and ultra-deep oil and gas reservoirs exhibit high levels of heterogeneity, which are governed by the original sedimentation processes and have a significant impact on oil and gas migration and accumulation. However, traditional pixel-based stochastic reservoir modeling encounters challenges when attempting to effectively simulate multiple facies simultaneously or objects with intricate internal hierarchical architectures. To address the characterization of highly heterogeneous deep and ultra-deep oil and gas reservoirs, this study defines unit architecture bodies, such as point bars, braided rivers, and mouth bars, incorporating internal nested hierarchies. Furthermore, a novel object-based stochastic modeling method is proposed, which leverages seismic and well logging interpretation data to construct and simulate reservoir bodies. The methodology is rooted in the unit element theory. In this approach, sedimentary facies models are stochastically constructed by selecting appropriate unit elements from a database of different sedimentary environments using Sequential Indicator Simulation. The modeling process is constrained by time sequence, event, and sedimentary microfacies distributions. Additionally, the porosity and permeability of each microfacies in the reservoir model are quantitatively characterized based on statistics derived from porosity and permeability data of different strata, sedimentary microfacies, and rock facies in the study area. To demonstrate the superiority and reliability of this novel modeling method, a modeling case is presented. The case utilizes braided river unit elements as objects for the stochastic simulation of the target reservoir. The results of the case study highlight the advantages and robustness of the proposed modeling approach.

1. Introduction

Since the 1970s, the field of computer science has undergone rapid development, leading to an increased use of numerical modeling for various purposes related to underground oil and gas resources. These include the quantitative assessment of resources, the prediction of exploration targets, and the design of recovery schemes [1,2,3,4,5,6,7]. A comprehensive geological model encompasses important elements such as tectonic structures, sedimentary facies distribution, and petrophysical properties like porosity and permeability. The accuracy of geological modeling, particularly in three dimensions, forms the foundation for the subsequent numerical modeling used in decision-making processes within the field of oil and gas exploration and production. An unreliable geological model can result in unfeasible simulations of reservoir/hydrocarbon migration and hinder effective decision making [8,9,10,11,12].

Geological modeling work is based on various data sources, including well logs, seismic interpretations, net-to-gross data, and property measurements, among others. The primary objective is to interpolate missing data, and interpolation methods can generally be categorized as deterministic or stochastic [8,13]. The deterministic method widely known for interpolation in geological modeling is the Kriging method, initially employed in ore body modeling [14]. However, it tends to produce smooth results due to its reliance on the variogram. Since geological bodies are typically heterogeneous and possess non-smooth properties, the predictions made by the Kriging method are often considered unreliable [15]. On the other hand, stochastic methods such as Sequential Gaussian Simulation and Sequential Indicator Simulation generate multiple results that follow the same probability distribution, thereby accommodating uncertainties by producing realizations that align better with the available data [16]. These methods discretize the model space into hexahedral cells or pixels in three dimensions. Nonetheless, pixel-based methods may fail to capture the geometries and internal structures of sand bodies and sedimentary facies. To address this limitation, object-based methods were introduced [2], which involve creating geological objects with defined geometries and generating them statistically using stochastic methods within the geological model space [17].

In 1978, Allen introduced the term “architecture” to the field of river sedimentation, establishing the concept of “fluvial architecture” [18]. Expanding on Allen’s work, Miall [19] presented analytical methods for studying the architectural elements of fluvial facies within reservoirs. The notion of architecture elements and interfaces was introduced, with reservoir architecture being defined as “the geometry, size, orientation, and relationships of the reservoir and its internal components”. When diagenesis and fracture formation exert minimal influence, the sedimentary architecture becomes the dominant factor in determining the hierarchical structure of the reservoir, primarily guided by the architecture interface [19,20]. During the development phase of an oil field, fluid movement within the reservoir encounters various flow barriers that significantly influence the formation and distribution of residual oil. Therefore, comprehending the architecture of underground reservoirs is crucial for enhancing hydrocarbon recovery and maximizing the development of oil and gas resources [21].

With the advancement of oil and gas exploration research, there is an increasing emphasis on understanding the heterogeneity of carrier beds and their impact on hydrocarbon accumulation [22]. The distribution, internal structure, and geometry of sand bodies have a significant influence on the migration pathway of hydrocarbons and the amount accumulated in traps [23,24]. Hence, it is imperative for geological models to accurately capture sedimentary and tectonic heterogeneity. To construct a comprehensive three-dimensional subsurface reservoir model for oil and gas exploration, available data such as lithological well logs, seismic horizon or fault interpretations, and net-to-gross interpretations should be incorporated. Presently, reservoir modeling primarily focuses on the facies or subfacies scale of architecture elements, while smaller-scale architectures play a crucial role in controlling the distribution and development of residual oil within reservoirs [25]. In previous research, a novel geological modeling software called DMatlas was developed [26]. This software enables modelers to manually create diverse sand bodies based on sedimentary facies, adhering to the depositional process and fundamental geological principles. It allows for the creation of more than three types of microfacies within a sedimentary environment. However, given the existing data conditions, the previous methodology proposed by Li et al. [26] falls short in effectively characterizing the spatial structural characteristics of strongly heterogeneous reservoirs.

This study centers around the fundamental characterization theory of significant reservoir architecture unit elements within diverse depositional environments. In particular, it provides detailed insights into three types of unit architecture elements, emphasizing their formation process and the principles governing sedimentary evolution. Subsequently, the study introduces the methodology and process of 3D reservoir modeling using unit architecture elements. To validate the approach, a case study is presented, employing a novel object-based stochastic modeling technique. The resulting reservoir model is illustrated through cross-sections showcasing the distribution of sand bodies, accompanied by statistical analysis to evaluate the modeling outcomes.

2. Characterization of ‘Unit Element’ for Reservoir Modeling

Traditional sedimentary facies analysis and sedimentary architecture analysis share the common goal of studying the origin and spatial distribution of sedimentary units. However, they differ in terms of scale and the focus on interfaces. Traditional sedimentary facies analysis typically categorizes facies into the following three levels: facies, subfacies, and microfacies [27]. It should be noted that planar facies distribution often does not reach the scale of individual microfacies. In such cases, multiple distributary channels are often considered as a distributary channel microfacies [28]. In contrast, sedimentary architecture studies concentrate on the distribution within a single unit, with the minimum scale being the accretion body in it. For fluvial deposits, sedimentary microfacies maps usually depict the distribution of channel sand bodies, while architecture analysis further subdivides composite channel sand bodies into different single-channel sand bodies. Additionally, it breaks down individual microfacies (e.g., point bars) within a single-channel sand body and further divides lateral deposits within a single point bar [19]. Regarding facies boundaries, sedimentary facies analysis typically employs statistical boundaries in planar sedimentary facies maps, where one dominant facies encompasses the interior while others are included. On the other hand, architecture analysis necessitates determining the isochronous physical interfaces of a single facies unit in space, such as the top and bottom of channel sand bodies [28].

Miall [19] introduced a subdivision order that aligns with the scale hierarchy in the study of fluvial facies architecture, where the first order represents the lowest level. In this study, we have adopted the latter scheme to describe the characterization and modeling methodology of architecture elements, specifically focusing on the third to fifth interfaces, which define the actual reservoir architecture units. In fluvial deposits, architecture elements bounded by the fifth-order interface generally correspond to the scale of sedimentary microfacies combinations, such as river channels. Architecture elements defined by the fourth-order interface typically represent a single microfacies, such as an individual point bar or crevasse splay. Lastly, architecture elements, defined by the third-order interface, are roughly equivalent to constituent units within a single microfacies, such as the lateral accretionary body within a point bar.

The primary modeling approach employed in this study revolves around assigning each unit reservoir architecture element, hereafter referred to as a ‘unit element’, which is primarily bounded by the fourth or fifth interface. Each unit element is equipped with a default external geometry and associated size parameters that can be modified by the modeler, instead of relying on stereotyped solid geometry created by pixels. Stochastic methods are utilized to generate internal architectural elements bounded by lower hierarchy interfaces (third or fourth interface). The default geometry of distinct unit elements deposited in various sedimentary environments is determined based on the understanding of the formation process and evolutionary laws of sediments.

In this study, three widely recognized unit reservoir bodies predominant in three sedimentary facies environments are comprehensively summarized and characterized. These findings serve as the foundational components of the 3D subsurface reservoir modeling methodology.

2.1. Meandering Fluvial Environment

The architecture and facies heterogeneity of fluvial point bars and braided bars, as significant reservoir bodies within classical channel patterns, have garnered considerable attention and collaboration among scientists [18,29,30,31]. A point bar represents a bank-attached element that forms as a result of lateral accretion or downstream-oriented accretion during the growth of a fluvial meander. Geologists have studied the migration of channel meander bends extensively for several decades [31,32,33,34,35]. During individual flood events, erosion takes place along the outer bank of the channel, leading to the retreat of the cutbank, while deposition occurs along the inner bank of the bend. This process results in the deposition of a new layer of sediment on the point bar, forming a gently dipping bed towards the outer bend, often referred to as a “lateral accretion bed” [9]. These laterally accreted sediments, also known as scroll bars, provide a record of the incremental growth of the meander over time. A point bar typically consists of multiple groups of genetically related scroll bars. In Figure 1A, the channel (in yellow) and the scroll bar (in orange) show the incremental growth procedure. The lateral accretion sand bodies developed in the growth procedure are separated by shale, which is called shale drape. The cross section is shown in Figure 1B.

The deposits of many point bars are typically rich in sand, making them favorable reservoirs. However, the significant architectural heterogeneity of point bars is evident, both in the planform and in cross-section view (Figure 1). In the 1980s, meandering facies models were notably recognized as “Inclined Heterolithic Stratification” (IHS) [36,37,38]. The IHS model comprises inclined packages of alternating sand-prone and mud-prone beds or laminations. This lithologic heterogeneity is mainly driven by the varying proportions of mud- and sand-prone sediment. In actual deposits, the depositional beds exhibit internal complexity, recording the deepening of thalweg areas and the development of bar tops during rising flood stages, followed by the accumulation of sands in channels and mud deposition on the bar tops during falling flood stages.

Meander shape refers to the morphology of a river bend, which exhibits high variability and poses challenges when attempting to quantify it using simple shape parameters. The assessment of meander shape variations is crucial for predicting the distribution of small-scale heterogeneity [20]. The fundamental modes of meander bend transformations, including expansion, translation, rotation, and their combinations, are widely acknowledged [14,39,40]. Understanding these modes helps in comprehending how the variations in bed topography and sedimentary textures vary as a river develops (Figure 2A). However, there remains a relatively limited understanding of the relationship between the migratory behavior of a meander bend and the resulting accumulated sedimentary architecture [41]. Conceptual or qualitative process-based descriptions have inherent limitations when used as guidance for object modeling methods.

Russell et al. [42] introduced an approach for comparing fluvial geometries by examining both the meander shape and the surface expression of the associated scroll-bar accretion direction. This methodology involves subdividing the upstream and downstream limbs at the meander apex. However, to prevent bias or preference, this study will not cover all of the parameters, but instead focus on understanding the potential range of meander shape variability. Through this research, 25 individual meander shapes were identified, and classified into the following sub-groups within 4 parent groups: open asymmetric, open symmetric, open angular, and bulbous (Figure 2B). Among these, open symmetric shapes are the most commonly observed, while angular and bulbous shapes occur less frequently.

Each meander shape exhibits different patterns of scroll bars, and the resulting preserved shape represents the path and evolution of the fluvial point bar, along with its growth characteristics. This study incorporates the variability in the meander shape and scroll bar pattern to expand the database of point bar unit elements, which can be used in subsequent object-based reservoir modeling.

2.2. Sandy Braided Environment

While significant research has been conducted on meandering fluvial systems, there has been less consensus and recognition regarding the geomorphological elements, stratification, and depositional facies sequences in braided rivers [31,43,44,45]. Braided rivers are characterized by wide and shallow channels and bars, with key sedimentary processes such as bar formation, channel-floor dune migration, low-water accretion, and overbank sedimentation [28]. To address these challenges, Miall [28] proposed a categorization of lithofacies types and facies associations using six vertical profile models, which have undergone revision over time. However, these lithofacies codes are descriptive and interpretive, and many facies types can occur in multiple environments within a river. As the interest in 3D reservoir modeling grows, there is a pressing need for a more definitive subdivision and unified nomenclature for the elements of the braided reservoir architecture (

Table 1. Variables Considered for the Model.

Reservoir Architecture Elements	Scale	Miall [28]	Cant [46]	Smith [47]	Geomorphological Features
Fall-siltseams	3rd-order	low-water accretion	none	none	top of braided channel
Fine-grained sediment	4th-order	overbank sedimentation	floodplains	floodplains	top and inside of unit bar
Unit bar	4th-order	linquoid or transverse bar	bar, cross-channel bar, unit bar	unit bar	long, low, slipface-bounded
Compound bar	4th-order	none	sand flat	compound bar	large areas of sand accumulation
Channel fills	4th-order	channel	major channel, minor channel	cross-bar channels	convex downward and separated by bars
Braided unit element	5th-order	river	river	river	actual topographic depressions

).

Cant [46] developed a comprehensive facies model for the sandy braided South Saskatchewan River, incorporating the following four key geomorphological elements: channels, slipface-bounded bars, sand flats, (vegetated) islands, and floodplains. Furthermore, based on morphological classification, Cant [46] also identified ripples, sand waves, and dunes as bedforms in lower flow regimes. Smith et al. [47] further refined the architectural subdivision of sandy braided rivers, delineating unit bars, compound bars, cross-bar channels, and dunes. Compound bars consist of amalgamated unit bar deposits, often intersected by cross-bar channels filled with dunes and rippled sands.

In this study, we introduce the concept of a “braided unit element” as the fundamental unit for stochastically filling braided reservoirs. We address concerns regarding the consistent preservation of fluvial deposits in one dominant direction and the tendency to substitute lenticular reservoir bodies. Based on our understanding of braided river sand bodies, the braided unit element exhibits a lenticular cross-section, resembling the geomorphology of river depressions, with an elliptical planform (Figure 3). Within-channel deposits in this braided unit element are characterized by both singular “unit bars” and individual channel fills. The braided unit element possesses a flat bottom and a convex top, and is superimposed within the convex downward lenticular braided channel sand under varying hydrodynamic and topographic conditions. We do not model complex “compound bars” or “sand flats” as independent elements that can be stochastically simulated through the amalgamation of unit bars. Additionally, the recognition of compound bar deposits poses challenges [48], making dimensional estimation problematic. Bedforms such as dunes and ripples are all upscaled and represented as channel fills within our modeling approach.

The interbeds within the braided unit element consist of fall-siltseams situated at the top of the unit bars, along with floodplain fine-grained sediments at the top of the entire braided unit element. The third-order architecture of fall-siltseams refers to the deposition of large-scale fine-grained suspended sediments at the end of flood events, resulting from sedimentary differentiation. During each flood period of the braided river, fall-siltseams form on the re-emerging unit bar’s upper surface, with their distribution area and shape being influenced by those of the unit bar during that period. To avoid placing excessive emphasis on lower-scale heterogeneity, fall-siltseams are modeled around the circumference of the unit bars

2.3. Delta Front Environment

A delta is defined as a “discrete shoreline protrusion formed where rivers enter oceans, semi-enclosed seas, lakes or lagoons and supply sediments faster than they can be redistributed by basinal processes” [49]. Scruton [50] illustrated a vertical sequence of deltaic facies that become coarser and sandier upward, corresponding to the progradation of bottomset, foreset, and topset strata (Figure 4A). The pro-delta subfacies predominantly consists of claystone with significant sediment thickness, wide distribution, and high organic content, indicating favorable conditions for oil generation [51]. Delta-front subfacies deposits include mouth bars, distal bars, and sheet sands composed of clean and well-sorted sands, indicative of favorable reservoir properties. The clays that overlay the deltaic sand body during the transgression process, as well as the alluvial swamp deposits formed as the delta advances towards the sea, can serve as effective cap rocks.

Distributary mouth bars play a significant role in the active sand deposition within modern deltas, representing a fundamental architectural element [53]. These bars form as sediment transport rates decrease basinward, resulting in deposition at the river mouth and the amalgamation of sand bodies with good lateral continuity. The size and shape of a mouth bar are influenced by factors such as the angle of plume dispersion, flow conditions, and forces acting on the river plume. In river-dominated deltas, the core of the mouth bar primarily consists of relatively pure coarse-grained sand, exhibiting a narrow band of pronounced directionality perpendicular to the shoreline (Figure 4B). Mouth bars typically have width ranging from several hundred meters to a few kilometers in length. Conversely, in wave- or tide-dominated deltas, the mouth bar sand body displays an elongated geometry, elongating perpendicular to the ancient river and parallel to the shoreline.

Architectural element analysis of deltas has not progressed as extensively as that of fluvial systems, but there is a growing body of research focusing on mouth bars deposited in river-dominated delta fronts [54,55,56]. Mouth bars can serve as distinct physical units closely associated with terminal distributary channels, typically characterized by depths ranging from a few meters to tens of meters, and widths spanning tens to hundreds of meters. The foresets, representing inclined sediment layers, can be observed in the cross-sectional dip profile of the delta clinoform (segment C in Figure 4A). These foresets are primarily associated with the delta front and exhibit dips ranging from a few degrees up to 13° (Figure 4C). Along the strike direction, overlapping mouth bars contribute to the formation of lenticular or mounded delta lobes (Figure 4B).

The grain size of sandstones exhibits significant variation among different sedimentary microfacies and different parts of the same microfacies due to differing hydrodynamic forces [57]. Within the middle and upper sections of the mouth bar, the dominant sandstone compositions include medium-coarse sandstone, medium-grained sandstone, and medium-grained fine sandstone, characterized by a coarse grain size and strong hydrodynamic forces. As a result, these sandstones generally have a lower content of plastic debris and matrix, and possess favorable physical properties. In contrast, the sediments at the bottom of the mouth bar primarily consist of fine- to very-fine-grained sandstone, indicating weaker hydrodynamic conditions and a higher content of matrix and plastic particles (heterolithics). Furthermore, the fine-grained sediments contain relatively high levels of carbonate and other cements, resulting in poorer reservoir physical properties. These plastic-rich grain lithic sandstones or calcareous cemented sandstones found at the base of the mouth bar serve as internal architectural elements within the unit mouth bar proposed in this study, acting as flow barriers.

The top surface of an individual mouth bar corresponds to a fourth-order interface, while the calcareous cementation at the base represents a third-order interface [58,59,60]. Leveraging extensive morphology data derived from numerous outcrops and existing literature [57,61,62], a preliminary dataset of mouth bar unit elements has been developed.

3. A Novel Object Stochastic Modeling Method

Stochastic reservoir modeling is an essential technique utilized to simulate the behavior of subsurface hydrocarbon reservoirs by incorporating random variations in geological properties [15]. This method involves generating multiple realizations of reservoir properties through statistical techniques like conditional simulation or geostatistics. Subsequently, numerical reservoir simulation models are employed to simulate fluid flow within the reservoir. The outcomes of these simulations enable the estimation of reserve uncertainty, the optimization of field development plans, and the assessment of exploration and production risks [63]. Geological data from diverse sources, including drill logs, geological maps, and geophysical data, are processed using geostatistical techniques to construct a comprehensive three-dimensional model of the subsurface. This model facilitates the visualization of the distribution of geological features and properties such as the lithology, porosity, and permeability [64].

Traditional geological modeling methods often suffer from assigning facies types or physical properties randomly to gridded structural frameworks based on well and seismic data in pixels. This can result in over-characterization or fuzzy boundaries of reservoir heterogeneity [65]. Object-based modeling has emerged as a popular alternative, defining and modeling geological objects or entities, such as faults and lithological units, as discrete 3D volumes or objects [66]. However, this approach has limitations, particularly when dealing with more than two microfacies or object internal heterogeneity, due to the complexity of algorithms and implicitly generated fractions within pixel-based random modeling.

To overcome these limitations, this study introduces a novel object geological modeling approach that enables a more detailed and accurate representation of reservoir architecture and heterogeneity, as well as the better integration of various types of geological data. This approach directly generates reservoir bodies with definitive geometry, including the length, width, thickness, and their quantitative relationships. By setting geometric parameters of modeling units and adhering to sedimentary laws, this method restores the understanding of geologists, allowing for the reconstruction of spatial geometric characteristics, internal property features, and the inter-relationships of geological elements, such as strata, facies, and structures, under geological sedimentary conditions to the greatest extent possible. To capture the heterogeneity within sand bodies, a new geological modeling method defines unit sand bodies for different sedimentary systems. Internal heterogeneity is identified and incorporated within each unit sand body, while their distribution in the strata reflects three-dimensional spatial heterogeneity. As a result, the heterogeneity at various levels of the migration system can be reasonably expressed.

The development of a conceptual reservoir geological model involves establishing the primary framework for the distribution of reservoir sand bodies in three-dimensional space. This framework, also known as a skeleton model or sand body distribution model, forms the foundation for constructing the reservoir’s physical property model or static model. The physical property model, encompassing parameters such as porosity and permeability, relies on the skeleton model as a starting point and incorporates rock physical property data to quantify the model and determine the spatial distribution of various reservoir physical property zones. This represents the crucial second step in the reservoir modeling process.

3.1. Unit Architecture Element Database and Distribution Model

Stochastic modeling offers a means of characterizing fluvial heterogeneity across different scales with reduced uncertainty by defining the statistical dimensions (length, width, and height) of geobodies. Sequential Indicator Simulation, a geostatistical method, is employed in this modeling approach [64]. Instead of initially constructing a gridded framework and assigning facies or lithology codes in a stereotyped solid geometry using pixels, reservoir models are built through the simulation of geological objects with hierarchical heterogeneity. These objects are extracted from a database of reservoir bodies that can be flexibly customized, allowing for a more flexible and accurate representation of the subsurface architecture. Experimental variograms capture spatial continuity using facies indicators. Limited well data in geostatistical simulations increase uncertainty, mitigated by seismic surveys or geological proportion maps. Seismic data offer broader but lower-resolution coverage than well logs, while refined proportion maps enhance the simulations. Horizontal variograms, often inconsistent due to sparse data, can be inferred from anisotropy ratios or seismic data. Vertical variograms remain stable with dense borehole sampling. Defining variograms within a geological framework is crucial, with modeler expertise guiding reliable formulations when direct data are insufficient.

Modeling each channel bed individually for every flood event would be computationally intensive and may not provide definitive conclusions. Instead, by considering the process of point bar formation and the laws of sedimentary evolution, it is understood that lateral accretion within the river system leads to the development of composite point bar patterns. These patterns emerge through the formation, erosion, and stacking of multi-stage lateral accretion bodies. In this modeling and simulation approach, hierarchical meandering bends are defined as larger-scale channel bar deposits formed during flood stages, which are interspersed with finer-grained deposits generated during falling stages and abandonment channel fills. A point bar unit element consists of stochastically generated drapes, abandonment channel fills, and lateral accretion bodies as a cohesive entity. Among these components, the finer-grained sediments known as low-flow drapes, deposited on the top of bars during falling stages, exert the greatest influence on predicting the corresponding scale of heterogeneity.

To construct a unit architecture element such as a point bar, braided bar, or mouth bar, it is necessary to establish geometric parameters and define the planar or longitudinal profile morphology of the unit element. This customized unit element can then be incorporated into a new or existing database (Figure 5), which will be utilized in the subsequent stochastic modeling process. Single bars within a braided unit element are generated stochastically, taking reference from a customized non-banded reservoir body database. The geometry of the braided unit elements is carefully considered, ensuring accurate representation of the heterogeneity within hierarchical fluvial segments. Additionally, physical properties are assigned to each deposit texture. Multiple reservoir models have been constructed using this object-based approach, which effectively captures the heterogeneity and exhibits strong compatibility with subsequent migration simulations. Figure 6 provides an overview of the river mouth bar reservoir, which has been modeled with sand ratio control. In order to optimize the computational efficiency and avoid potential issues, the bottom calcareous cementation is not shown in this particular workflow step (Figure 6B), considering the size of the project area and the geometry of the reservoir body. Following the gridding process, the petrophysical probability distribution is assigned to each facies, and the flow barrier boundary is depicted in the porosity (Figure 6C) or permeability model.

To generate a reservoir model for different sedimentary environments, additional unit elements with internal architectures can be customized to enhance the dataset and align with the stochastic modeling approach. Probability distributions can be assigned to each geometry parameter, and other controlling data, such as the sand ratio, can be set. This process ensures a comprehensive representation of the reservoir characteristics and facilitates the accurate simulation of various sedimentary environments. Core, thin-section, and logging data are utilized to ascertain the upper and lower boundaries of unit elements, as well as the sand ratio. By integrating the structural model, sedimentary conceptual model, and sand ratio data, suitable unit elements are selected from the reservoir architecture unit model library, taking into account constraints such as the sand body superposition mode and geometric parameters like the length, width, and width–thickness ratio of various sedimentary microfacies sand bodies. These chosen unit elements are subsequently stochastically simulated to construct the reservoir model.

3.2. Reservoir Physical Property Modeling

Once the geological bodies have been properly modeled, the entire layer, comprising both the low-permeability background and the hierarchical reservoir bodies, undergoes thickness adjustments using practical structural data. The key structural deformation elements are extracted, taking into consideration well–seismic stratification, in order to capture the structural deformation characteristics specific to the targeted modeling area.

There are two approaches to modeling faults: customization and automatic generation. In the customization method, fault lines or planes are customized to implement multi-fault modeling. Users can define the spatial distribution of fault traces as two-dimensional fault planes based on the structure contour map. These faults can then be incorporated as fluid flow passages or barriers in the property modeling process. Additionally, the customization method allows for fault modeling by physically separating the layers based on the fault polygon boundary derived from the structural contour. On the other hand, the automatic generation method involves importing seismic fault interpretation data to automatically generate faults in the model. This approach streamlines the fault modeling process and ensures consistency with the seismic data.

To facilitate random fracture modeling based on regional fracture distribution, fracture templates and fracture model libraries are employed. This innovative approach to fracture gridding significantly enhances the efficiency and stability of numerical simulations conducted on fractured reservoirs. It enables the comprehensive characterization of the structural elements within the reservoir system, ensuring the accurate representation of fracture patterns and their impact on the fluid flow.

The final step in constructing a 3D reservoir model is the extraction of the area of interest. For work areas with irregular polygon boundaries, it is possible to rotate them by specifying angles to reorient the I/J directions away from the traditional north–south or east–west orientations. The rotation angle can be adjusted by the user while viewing the real-time red-dotted boundary, ensuring the polygon fits appropriately based on the desired degree of rotation.

Once sediment bodies are modeled in their respective locations, the distribution of sediment types is gridded, and the associations between the depositional texture and petrophysical properties are established. For numerical reservoir simulation and hydrocarbon migration, corner point gridding is employed, with vertical grids evenly dividing each zone based on the stratigraphic relief. When determining the grid resolution, it is essential to account for the size distribution of the reservoir bodies in order to capture the multiple scales of heterogeneity effectively.

Porosity and permeability modeling entails assigning porosity values and defining key parameters such as maximum and minimum values, along with establishing an indirect permeability model. It is assumed that the logarithm of permeability for the same microfacies type is linearly correlated with the porosity, and appropriate parameter adjustments are made to achieve a reasonable distribution of permeability intervals. For overbank deposits in the floodplain, uniform impermeable values are assigned, as the focus lies on defining variations within the channel belt deposits. The resulting porosity and permeability model can be visualized along the I/J/K direction and represented as a fence diagram, respectively. The property model is utilized for the numerical reservoir simulation and hydrocarbon migration.

3.3. Geological Modeling Steps

The new geological modeling methodology comprises facies modeling (using unit element theory), structural modeling, and property modeling. These three modeling procedures are executed sequentially within the workflow, detailed below as follows:

(1)

Facies Modeling:

Modelers must first determine the depositional environment of the target reservoir. Well data and seismic data are imported and used as constraints for the modeling process. Sand bodies, treated as unit elements, are then stochastically distributed within the 3D space, constrained by the imported well and seismic data. The specific sand bodies used are selected from a database based on the identified depositional environments. It is important to note that the initial facies modeling assumes that the geological layers are flat. Subsequently, the layer depths and the sand body positions are deformed to match the actual depth data.

(2)

Structural Modeling:

Structural modeling encompasses the depth variation in the different layers and the modeling of faults and fractures. The depth model for each layer is constructed based on seismic and/or well data. The fault and fracture models are derived from seismic data or “soft data” interpreted from various sources.

(3)

Property Modeling:

This step involves creating a grid with a resolution in each dimension determined by both the computational capacity available and the characteristic size of each facies. The model includes porosity and permeability. Porosity data, obtained from well logs or experimental measurements, are classified according to the facies type. Porosity values are then stochastically assigned to each grid cell within a specific facies type. Finally, porosity–permeability relationships are established for each facies, and the permeability for each grid cell is calculated based on these relationships.

4. Reservoir Modeling of the Sangonghe Formation in the Central Junggar Basin

In this study, the sandstone reservoir of the Sangonghe Formation in the central area of the Junggar Basin, which represents a braided river delta sedimentary system, was selected as a case study to apply the new reservoir modeling method. The main objective was to establish a geological model and accurately characterize the reservoir’s heterogeneity. To assess the reliability of the modeling results, a comparison was made between the drilled sand thickness and the modeled sand thickness. This analysis showcased the flexibility of the new method and highlighted its ability to effectively capture the multi-scale heterogeneity of the reservoir.

4.1. Petroleum Geological Setting

The study area, located in the MXZ area in the central part of the Junggar Basin, is southeast of the Well Pen-1 Western Sag (Figure 7). The dominant structural feature in the study area is a dipping monoclinal structure, with the burial depth gradually increasing from the northeast to the southwest. The target reservoir is generally found at depths ranging from 4000 m to 4500 m. In terms of trap types, the MXZ area does not exhibit typical structural traps, but, instead, a small number of intralayer faults are present in the main oil-bearing formations. The predominant types of oil reservoirs in this area are structural–lithological reservoirs [67].

The central Junggar Basin primarily focuses on the Jurassic formations for oil and gas exploration and development. These formations include J₁b, J₁s, J₂x, and J₁t, from bottom to top (Figure 7C). After the Yanshan Movement, the region experienced uplift, leading to the erosion of the Upper Jurassic strata. Among the Jurassic formations, J₁s is the primary oil-bearing reservoir and can be further subdivided into three members from bottom to top. The J₁s₁ and J₁s₃ consist mainly of semi-deep and deep lacustrine dark mudstone, with thin siltstone and fine sandstone layers. The J₁s₂ member, which belongs to the delta front deposit, serves as the main reservoir. It exhibits longitudinal superposition and transverse composite contiguous distribution of multi-stage sand bodies.

The characteristics of the lower sub-member (J₁s₂¹) and upper sub-member (J₁s₂²) of J₁s₂ differ significantly. J₁s₂¹ was deposited at the front of a braided river delta, showcasing highly developed distribution channel sand bodies and estuarine bar sand bodies. The multi-stage channel sand bodies vertically stack with a cumulative thickness of 60 to 80 m. Notably, the lower part of J₁s₂¹ exhibits pan-connected massive thick sand bodies, accompanied by an increase in the mudstone content and enhanced heterogeneity. On the other hand, J₁s₂² was deposited at the front of the meander delta, featuring isolated or laterally overlapping sand bodies. The section shows a lithological combination known as “mud-coated sand”. Distribution channel sand bodies are found at the base of J₁s₂², generally with a thickness of less than 15 m, and they exhibit poor horizontal connectivity.

The main reservoir within J₁s₂ consists of feldspathic lithic sandstone and lithic sandstone with low compositional maturity. The reservoir is characterized by a significant burial depth, strong diagenesis, a wide range of sandstone physical properties, and porosity ranging from 1.9% to 18% (with an average of 8.3%). The permeability varies from 0.01 × 10⁻³ to 78 × 10⁻³ μm² (with a mean of 0.92 × 10⁻³ μm²) [68]. Due to the influence of the original rock composition and differential diagenesis, the diagenetic evolution within the reservoir shows significant variations. The sandstone can be classified into different rock facies types, including rich plastic grain sandstone, strong carbonate cement sandstone, and poor plastic grain sandstone. The rich plastic grain sandstone and the strong carbonate cement sandstone undergo densification during early diagenesis. The poor plastic grain sandstone, however, experiences multi-stage cementation and dissolution during deep burial, reflecting a complex interaction between organic and inorganic fluids [69,70].

4.2. Structure and Facies Modeling

Based on the data and knowledge of the sedimentary facies, structure and reservoir heterogeneity in the MXZ area of the Junggar Basin, the geological model of the heterogeneous reservoir in the Sangonghe Formation, is constructed by using the above-mentioned reservoir modeling method. Furthermore, the favorable target area is evaluated and optimized by using the relevant parameter.

On the basis of single-well sedimentary microfacies, combining sedimentary facies, logging facies, and seismic facies, it is determined that a set of braided river delta sedimentary system is developed in the Sangonghe Formation. The sandstone reservoirs are mainly distributed in the delta-front subfacies in the lower part of J₁S₂², and the source is mainly from the north. The submarine distributary channels are widely developed and have good continuity. A large number of front-sheet sands are developed outside the distributary channels, with fewer mouth bars developed. The upper and lower parts of J₁S₂¹ developed submarine distributary channels of the delta front and shore–shallow lake facies, which have narrow width, short length, and thin sand bodies. The middle mainly developed braided channel microfacies of delta plain subfacies, with great thickness and good continuity. The bars within the braided channels are widely developed and the flood plain partially deposits the outer part of the braided channels. In this modeling practice, an area of 25.45 km × 18.65 km was targeted, and the sedimentary reservoirs of J₁S₂², J₁S₂¹², and J₁S₂¹¹, from top to bottom, were modeled using this ‘unit element’ stochastic modeling method, which collectively deposited predominantly braided delta plain and front environment with an average thickness of 121 m.

The top structure map of each layer of the Sangonghe Formation in the MXZ area is obtained through seismic data interpretation after well–seismic calibration (Figure 8A). The structure of the study area is relatively stable, showing a monoclinic form with high north and low south. The south is influenced by the Mosuowan uplift, slightly higher in the southeast and lowest in the southwest (Figure 8B).

The braided unit element was chosen to represent the sediments in the target reservoirs as a whole. Unit elements and single-well sequence interface are identified by core, thin-section, and logging data. The sand ratio distribution characteristics of each layer can be obtained by using wave impedance inversion technology under the well–seismic calibration (Figure 9).

According to the field outcrops and previous research results, the thickness and width–thickness ratio of the submarine distributary channel, river mouth bar, and braided bar of the Sangonghe Formation were obtained (Figure 10). The distributary channels have a thickness ranging from 3.5 to 7.8 m and a width between 180 and 535 m. The width-to-thickness ratio falls within the range of 52 to 90. The thickness of the mouth bars ranges from 6 to 14.5 m, the width of which spans from 450 to 815 m, and the width–thickness ratio is in the range of 45 to 95. The braided bars have a width spanning from 6.5 to 100 m and a length between 20 and 240 m. The aspect ratio is between 2.3 and 6.8 m. Based on the geometry size statistics of the sediments, the scale of the braided unit elements was given distributions of widths ranging from 200 to 600 m, lengths between 300 and 1000 m, and width–thickness ratios between 50 and 100. The single braided unit element has a thickness between 2 m and 12 m.

Under the constraints of the upper and lower surface of the single-well unit elements, structure, sedimentary microfacies, sand ratio data, unit element geometric parameters (geometric shape, size, direction, and mutual relationship), and spatial superposition relationship, appropriate unit elements are selected from the unit element model library to construct the three-dimensional heterogeneous reservoir model (Figure 11).

4.3. Property Modeling

Once the structural framework has been established, the next step is to grid the model. To enhance the accuracy of the geological model and achieve a finer characterization of reservoir heterogeneity, the model is divided into a grid with a horizontal resolution of 20 m by 20 m and a vertical resolution of 0.5 m. This grid resolution takes into account the size of the target area and the distribution of wells, ensuring that the model captures the geological features and variations in a more detailed manner.

The physical properties of the reservoirs vary among different sediment microfacies and even within the same microfacies. The hydrodynamic force is the strongest during the deposition of the middle and lower parts of submarine distributary channels, as well as the middle and upper parts of the mouth bars. This results in relatively coarse particle sizes with good sorting and roundness, leading to favorable reservoir physical properties. Conversely, the sand bodies located between channels exhibit the poorest physical properties. To establish the property model of the reservoir, the porosity and permeability of unit sand bodies in different sedimentary microfacies are randomly assigned based on the statistical analysis results of their physical properties. This property model provides a representation of the reservoir’s characteristics, allowing for the simulation and analysis of fluid flow behavior within the reservoir (see Figure 12).

The heterogeneity of the braided reservoir can be observed in Figure 13, which showcases local amplification sections of the unit elements filling model. The sections of the porosity model demonstrate that the internal architecture of the reservoir is well-preserved after property modeling. In the porosity model, it can be observed that single bars exhibit the highest porosity values, ranging from 0.2 to 0.26, represented by a yellowish to orangish color. Channel sands follow, with porosities ranging from 0.1 to 0.18, indicated by a greenish color. The other two fine sediments display a bluish color and exhibit lower porosities ranging from 0.01 to 0.08. The background shale, on the other hand, is assigned the lowest porosity values, ranging from 0.005 to 0.01.

4.4. Model Verification

The reliability of the reservoir modeling results was assessed by comparing the drilled sand thickness with the predicted thickness using the following two wells: Z111, which participated in the modeling process, and Z12, which did not. For the Z111 well, the cumulative thickness of the sand bodies drilled by J₁^S22, J₁^S212, and J₁^S211 in the Sangonghe Formation were 13.4 m, 39.1 m, and 38.1 m, respectively. The model-predicted thicknesses were 10 m, 26 m, and 33 m, respectively, resulting in an overall agreement rate of 81.6%.

Similarly, for the Z12 well, the cumulative thickness of the sand bodies drilled by J₁S₂², J₁S₂¹², and J₁S₂¹¹ in the Sangonghe Formation were 29.8 m, 34.9 m, and 42.5 m, respectively. The model-predicted thicknesses were 19 m, 42 m, and 38 m, respectively, resulting in an overall agreement rate of 92.4% (Figure 14 and

Table 2. Comparison table of the drilled sand body thickness and predicted sandstone thickness of the second member of the Sangonghe Formation in the Z12 and Z111 wells.

Well	Thickness (m)	J1S22	J1S212	J1S211	Agreement Percentage
Z12	Drilled	29.8	34.9	42.5	92.4%
	Modeled	19	42	38
Z111	Drilled	13.4	39.1	38.1	81.6%
	Modeled	10	26	33

). These results indicate that the reservoir modeling results are credible.

In this real-world project application, compared to traditional geological modeling technologies, the fundamental unit element is defined based on bar development within braided fluvial systems. Within this “unit element” (Figure 11), channel fill and individual bars are defined according to the depositional process. The internal heterogeneity of these elements can significantly influence hydrocarbon migration patterns, as discussed in our previous work [71]. However, this new methodology has limitations that stem from the definition of the “unit elements” themselves. In real-world exploration projects, particularly when well log data interpretation is limited, it can be difficult to determine the size and parameters of internal bodies, even when the “unit element” can be observed in different depositional environments. Nevertheless, this new method provides a framework that allows modelers to generate multiple models spanning the uncertainty space to better match the observed data.

5. Discussion

Under the guidance of geological knowledge, reservoir geological modeling aims to achieve a comprehensive three-dimensional quantification of oil and gas reservoirs by interpolating and extrapolating reservoir characteristics based on known control points. The reliability and accuracy of the model are not only heavily reliant on the chosen geological modeling method but also closely tied to the accuracy and precision of reservoir characterization through geological knowledge, well logging, and seismic data. Reservoir geological modeling has evolved from traditional deterministic modeling, two-point geostatistical modeling, multi-point geostatistical modeling, and depositional process modeling, and has now entered the realm of intelligent geological modeling [72].

Deterministic modeling relies on geological knowledge obtained from sedimentology, reservoir geology, and seismic stratigraphy, using methods such as Kriging interpolation to calculate reservoir parameter distributions between wells [13,73]. However, the deterministic approach only provides certain results between control points, which may contradict the actual reservoir conditions, leading to limitations in the accuracy. Two-point geostatistical modeling overcomes this limitation by capturing the uncertainty of reservoir parameter predictions resulting from random variations. It enables multiple solutions for reservoir modeling outcomes and allows for the comprehensive analysis of result uncertainties, meeting the needs of risk decision making in oil and gas exploration and development [18,74]. However, this method may struggle to accurately characterize complex spatial structures and target geometries [75]. In comparison, the multi-point statistical method synthesizes more morphological information, analyzes the statistical characteristics of geological variables using multiple points, and employs a “training image” to represent the spatial structure attributes of geological variables. It can effectively depict the shape and spatial distribution characteristics of target bodies, demonstrating advantages in geometric shape characterization and well data conditioning [3]. However, challenges remain, including the requirement for suitable training images, the limited reproduction ability of modes, and slow computation speed. Further research is needed in areas such as simulation target continuity and non-stationary data processing [76]. Geological process-based modeling techniques consider real geological processes as control functions and simulate the spatial distribution of reservoir parameters under these processes. By adjusting control parameters, different simulation results can be obtained to replicate different geological conditions and optimize a scheme that aligns with geological reality, ensuring consistency with geological laws [77,78]. The integration of artificial intelligence with geological modeling technology, known as intelligent geological modeling, has gained attention. This approach aims to reduce the uncertainties stemming from human subjectivity, enhance the model accuracy, and achieve highly efficient, scientific, and intelligent modeling [79,80,81]. However, deep learning-based geological modeling methods require large training samples with a high information content and low redundancy. Thus, acquiring suitable training samples and enhancing the self-learning capabilities are crucial challenges for deep learning-based geological modeling techniques [82,83].

The heterogeneity of deep and ultra-deep clastic rock reservoirs exhibits distinct structural characteristics, primarily controlled by sedimentary structures. These differences are evident in the rock and mineral composition, structural arrangement, and fluid permeability [25,84,85]. While reservoirs with different sedimentary facies in various regions exhibit notable variations in petrology and sedimentary structure, a classification of rock facies types can be applied to each microfacies [86,87]. During the process of deep burial, the structural heterogeneity of clastic rock reservoirs influences rock formation, and the variations in diagenesis further contribute to reservoir heterogeneity [86,88]. Consequently, the migration and accumulation of oil and gas in structurally heterogeneous reservoirs differ from traditional understandings, emphasizing the importance of considering such heterogeneity in the transport layer.

The traditional reservoir geological modeling methods fall short in providing a detailed description of heterogeneity in deep to ultra-deep reservoirs. In this study, we propose a unit sand body filling modeling method, guided by sedimentary geology (Figure 15), to establish a comprehensive three-dimensional geological model of structurally heterogeneous reservoirs. This model enables the characterization and evaluation of the spatial distribution of effective reservoir sandstones and their inter-relationships. Incorporating the concept of artificial intelligence, we construct various types of cellular sand body models and combine them to create a library of cellular sand body models. By considering reservoir sedimentary characteristics, diagenetic processes, and oil and gas migration and charging in the study area, we select the appropriate unit sand body model from the library and directly incorporate it into the reservoir geological model. Drawing on previous sedimentological knowledge, we establish independent unit sand body models that account for different sedimentary environments, sedimentary characteristics, sand body types, and diagenetic features. These models encompass effective reservoir rocks as well as potential interlayers of mudstone and tight sandstone. It is important to note that the same type of unit sand body exhibits varying structural characteristics under different sedimentary conditions, which necessitates the construction of different unit sand body models. For instance, the unit sand body of the fluvial facies is defined based on the fourth-order configuration interface. In the case of meandering rivers, the point bar system is considered as a complete unit, encompassing the point bar sand body, the abandoned channel, and mud drapes during falling flood stages. This unit allows for the description of lateral deposits within the point bar sand body as well, as the presence of mudstone or plastic granular sandstone interlayers between them. On the other hand, the unit sand body of a braided river consists of a multi-stage interbedding of the braided bars within the small channel, separated by the fourth-order configuration interface. This unit includes the rich plastic sandstone interlayer positioned between the single-braided bars and the overlying floodplain deposits.

In the unit sand body filling modeling method constrained by sedimentary geology, the appropriate variation function difference is carefully selected, taking into account the sequential, event-based, and sedimentary microfacies constrained sand body superposition mode. The Sequential Indicator Algorithm is then employed to effectively fill the unit sand bodies. Drawing from our understanding of the inheritance and transformation of effective rock facies within the reservoir, the porosity and permeability characteristics within each unit sand body are determined based on the specific conditions observed in the study area. This enables the assessment of reservoir availability during different oil–gas accumulation periods and facilitates quantitative characterization of the reservoir properties.

The reliability of geological modeling in deep and ultra-deep highly heterogeneous reservoirs is contingent upon the accuracy of our geological understanding [89,90]. This understanding encompasses several key aspects that need to be comprehensively mastered during the geological modeling process. Firstly, it involves grasping the stratigraphic framework characteristics of oil and gas reservoirs, including fault and stratigraphic structures, as well as their respective inter-relationships. Additionally, it entails a thorough understanding of sedimentary characteristics such as sedimentary facies, subfacies, and microfacies. Furthermore, it encompasses reservoir characteristics such as the lithology, reservoir type, physical properties, oil and gas properties, and the relationships between the porosity and permeability. Understanding the reservoir’s developmental laws, including the controlling factors, diagenesis, qualitative characteristics of reservoir heterogeneity, and spatial distribution patterns, is also crucial. Lastly, knowledge of the reservoir’s spatial distribution characteristics, including the reservoir thickness, range, and the 3D prediction of reservoir parameters, is essential.

In the case of mid-shallow oil and gas reservoirs, the uncertainty in reservoir parameter calculation and spatial distribution prediction based on well logging and seismic data is relatively low due to the accessibility and high quality of the data. However, challenges arise when dealing with deep and ultra-deep layers due to limited data availability and the lower quality of seismic data. This comprehensive geological understanding serves as the foundation for geological modeling, influencing the strategies employed throughout the process. It also plays a role in selecting interpretation or prediction methods for reservoir parameters using geophysical techniques like logging and seismic analysis. Ultimately, the reliability of the geological modeling results is determined by the accuracy and precision of the prior geological understanding. While geological understanding is often subjective, this new workflow provides a flexible framework allowing modelers to define their own “unit elements” based on their individual interpretations of the depositional environment. However, the potential for multiple valid “unit element” definitions highlights the need for further research. Specifically, integrating additional field observations and well data will be crucial for refining these definitions and minimizing uncertainty in the resulting models.

6. Conclusions

Understanding the architecture of a reservoir is essential for predicting fluid flow behavior and optimizing the development of oil and gas resources. It is essential for the sustainable oil and gas exploration and recovery, especially in deep and ultra-deep underground zones. In this study, we proposed the following three reservoir architecture characterizations: point bar, braided channel, and river mouth bar. Each of these characterizations captures lower-scale heterogeneity, drawing upon sedimentary processes and evolution laws. These reservoir architecture characterizations act as prototypes for the object-based modeling method that is the focal point of this research.

A point bar unit element is composed of inclined alternating packages of lateral accretion sand bodies, mud drapes, and abandonment channel fills. The multi-stage lateral accretion bodies exhibit a diverse planform morphology, categorized into 25 shapes and 4 parent groups, separated by stochastically distributed mud drapes. For the braided unit element, it corresponds to a braided river belt defined by a fifth-order boundary. In order to establish a standardized nomenclature for architecture elements within braided channels, a comparative analysis was conducted, leading to the proposal of a new braided architectural unit. This new unit integrates the key elements necessary for constructing braided reservoirs. The braided unit element exhibits a lenticular cross-section, resembling the geomorphology of river depressions, and an elliptical or lobate planform, representing the inconsistent preservation of river channels. Within this braided unit element, intra-channel deposits include individual ‘unit bars’, individual channel fills, and interbeds comprising fall-siltseams and upper-floodplain fine-grained sediments. The unit element has a flat bottom and a convex top, superimposing within a convex downward lenticular braided channel sand. Fall-siltseams are extended to cover the perimeter of unit bars. On the other hand, a mouth bar unit element exhibits a dipping longitudinal section and a lenticular cross-section. The base deposits of this unit consist of fine–very fine grains, characterized by poor plastic lithic grains or calcareous cementation.

The bodies of these architecture units act as “objects” to be filled and fitted into the geological model using stochastic modeling methods under the control of well and seismic data. Internal architecture elements such as mud drapes of meander bends, unit bars deposited in braided channels, and calcareous cementation at the base of mouth bars are randomly generated, controlled by the geometry and size of the unit elements. Physical properties can be assigned to all of the architecture elements, indicating the path or barrier of the fluid flow. A modeling case was explained in detail, aiming to construct a braided environment reservoir and evaluate the target area using this novel object modeling method. The reliability of the reservoir modeling results was tested by comparing the drilled sand thickness with the modeled sand thickness. The overall agreement rate was 92.4%, indicating the credibility of the reservoir modeling results.

To illustrate the effectiveness of this novel object-modelling method, a detailed modelling case was presented, focusing on constructing a reservoir in a braided environment and evaluating the target area. The reliability of the reservoir modelling results was assessed by comparing the drilled sand thickness with the modeled sand thickness, yielding an overall agreement rate of 92.4%. This high agreement rate demonstrates the credibility of the reservoir modelling outcomes.

In short, the main conclusions from the works can be summarized as follows:

• The method uses specific reservoir architecture characterizations (point bar, braided channel, and river mouth bar) as “unit elements”, built upon sedimentary processes and evolution, to represent the reservoir heterogeneity.
• These unit elements are treated as “objects” and stochastically placed within the geological model, with internal features also generated stochastically based on the unit element’s geometry.
• The method allows for the assignment of physical properties to the architectural elements, enabling the good prediction of the fluid flow and demonstrating a high level of accuracy in the validation tests.