Fractures in the Dutch Rotliegend—An Overview
Published:January 01, 2011
- PDF LinkChapter PDF
Herald Ligtenberg, Jos Okkerman, Martin De Keijzer, 2011. "Fractures in the Dutch Rotliegend—An Overview", The Permian Rotliegend of the Netherlands, Jürgen Grötsch, Reinhard Gaupp
Download citation file:
The Permian Upper Rotliegend sandstones from the Upper and Lower Slochteren formations form the main gas reservoir of the Netherlands and are the host for the giant Groningen gas field. Fractures in the Rotliegend are quite rare in the Dutch subsurface, based on the very large core and log database that has been acquired over half a century. The fractures therefore are not considered to be a major control on the reservoir response and gas recovery in most of these areas. The Rotliegend can have tight reservoir properties due to the combined effects of more pronounced mechanical compaction and diagenesis. In these situations the presence of natural fractures and creation of hydraulic fractures can assist in improving the vertical and lateral connectivity. The impact of fracture presence depends on the types of natural fractures and their orientation, which define preferred well paths and steer the selection of the optimal hydraulic fraccing methods to deploy in the tight gas reservoirs.
Many Rotliegend fields are compartmentalised by faults, creating baffles and barriers to fluid flow on a production timescale and more rarely on a geological timescale. Hydrocarbon columns in certain Rotliegend fields are longer than can be expected based on mapped structural closure, because of fault seal. Understanding fractures—their origin, orientation, and properties—assisted in improving the understanding of the properties of faults and the different potential fault-seal mechanisms that are observed in the Rotliegend. Fractures are in a way smaller representations of the larger faults in the subsurface.
The various Upper Rotliegend fracture types include cataclastic, cemented, shale smear, phyllosilicate framework, and open fractures. The most common types in the Upper Rotliegend are cataclastic and cemented fractures. Detailed laboratory analysis of fractures in cores has shown that both cemented and cataclastic fractures can hold significant pressure differences. However, cemented fractures and faults are not continuous and therefore likely leak through weaker, poorly cemented windows along the fault and fracture surfaces. Cataclasis may be continuous along the entire fault and fracture surfaces and may form a sealing mechanism in the high-net-to-gross Upper Rotliegend sandstones.
The Rotliegend fracture types and their properties are defined by the conditions of temperature, pressure, and stress during the deformation phase during which these fractures developed and by possible additional effects of conditions at subsequent tectonic phases. Knowledge of the geotectonic history of the Rotliegend is therefore of relevance to improve the understanding of the presence and properties of fractures.
This paper provides an overview of the tectonic history of the Rotliegend, describes fracture origin and propagation, addresses the various Rotliegend fractures in detail, and discusses implications for Rotliegend field development and exploration prospectivity. The purpose of this overview paper is to contribute to Rotliegend structural characterisation, primarily by providing a catalogue of Rotliegend fractures. Reference is made in some places to fields and pressure and fluid data, but the intent is not to provide detailed field cases or fault-seal examples.
A first glance at the fault patterns in the Permian Rotliegend interval shows a quite regular, simple blocky fault pattern in many places, in between zones of more complex faulting (Fig. 1). Although the fault pattern appears simple, most fault systems formed at a very early stage in the structural tectonic history of the region and have been reactivated during most, if not all, subsequent tectonic phases.
Three major tectonic events have affected the region: the Carboniferous Variscan Orogeny, the Mesozoic breakup of Pangaea and related opening of the Atlantic, and the Late Cretaceous to Early Tertiary deformation. Deformation style and intensity varies between the various regions of the Netherlands onshore and offshore. For example, the Late Cretaceous to Early Tertiary inversion was pronounced in the Broad Fourteens Basin (Fig. 1), in contrast to the northern part of the Netherlands onshore, where compressional features are less pronounced.
Observing the Rotliegend fault patterns in more detail, a clear distinction can be made between different fault types and regions with (slightly) different fault-system geometries, suggesting the presence of different structural domains. The variations between the different structural domains with respect to the evolution of magnitudes and orientations of stress and strain, and local diagenesis, may have an effect on fault-zone properties. This may lead to incorrect assumptions and causes inconsistencies when comparing faults from different structural domains when one is not aware of their existence.
This special volume focuses on the exploration activities after the discovery of the major Groningen gas field. In exploration the predominant interest is in the prospect-bounding faults and their hydrocarbon flow properties at a geological timescale: are these faults sealing or non-sealing? The fault-zone properties depend on many factors, including the host rock, the amount of shale in the system (net-to-gross), the dip and strike of faults, and the tectonic history (e.g., fault diagenesis and fault reactivation). In the Rotliegend we observe different potential fault-seal types. The dominant types are cataclastic faults and shale gouge faults. A brief overview and examples are provided of the various fault deformation types and their elements. Unfortunately, drilling through faults has occurred only in few fields and in many cases unintentionally, and characterisation of fault properties is therefore mostly by indirect means. However, studying fault zones in outcrops and their rare occurrence in cores assists in improving our understanding of Rotliegend faults.
Faults as can be imaged and interpreted on seismic are not single fault surfaces but are actually complex zones consisting of a fault core (with sharp fault surfaces, breccia, fault gouge) and a fault damage zone (with fractures and deformation bands). In this article we focus on fractures and faults in Dutch Rotliegend cores, providing a significantly improved insight into the fault deformation mechanisms. Although both fractures and faults occur at various scales, for simplicity we refer to both of these in this paper as fractures (see more on nomenclature below). In many cases, the fractures sampled by core experienced multiple deformation phases as deduced as well seismically for the larger faults. For example, the presence of dominantly cataclastic fractures suggests that the rock has been under the right conditions to generate larger-scale cataclastic faults. In most of the Rotliegend cores studied, we have often observed the presence of different fracture types in the same core, or sets of fractures with very different orientation (Appendix B6). In general, fractures develop in rocks at various times and due to various processes. The various fracture types observed in the Rotliegend cores are categorised into cataclastic fractures, clay-rich fractures, synsedi-mentary fractures, open fractures, cemented fractures, and reactivated fractures. It is important to understand the fracture style, which is partly dependent on lithology.
Capturing the style and occurrence of fractures remains difficult, especially their 3D distribution and connectivity (as inferred from limited well and core data), but these are crucial aspects in unlocking the tight gas areas in the Netherlands onshore and offshore areas.
Regional Tectonic History
In order to understand the faults and fractures in the Rotliegend and their rock properties, it is fundamental to obtain a good understanding of the regional tectonic setting and history of the area. The tectonic history is restricted to a high-level overview. For details, the reader is recommended to study the publications mentioned and the references therein.
Displays of regional fault attributes like that shown in Figure 1 and fault maps in publications like Duin et al. (2006) provide a first view on the various fault trends in the Rotliegend. In general, the fault pattern looks quite blocky, a pavement type of fault system. However, at certain places this fairly regular pattern is segmented by large-scale trending fault systems with different internal fault character. At a regional scale, these particular fault systems strike NW– SE and continue for tens of kilometres through the Rotliegend. In fact, these fault systems characterise basement fault trends associated with the Silurian to Early Devonian Caledonian Orogeny (Ziegler, 1975, 1978; Van Wijhe, 1987; De Jager, 2003, 2007). Major grabens in the Netherlands subsurface follow these trends, like the Roer Valley Graben, the Broad Fourteens Basin, and the Lauwerszee Trough (Fig. 1). The development of the named grabens is largely defined by the pre-existing Silurian to Devonian basement faults.
The Variscan Orogeny started in the Early Carboniferous and ended during the Late Westphalian (Ziegler, 1990; Coward, 1993). This tectonic event is associated with the closure of the southern Proto–Tethys Ocean and the development of the supercontinent Pangaea (Glennie and Underhill, 1998). It had an approximate north–south compressional direction, but it only marginally affected the Netherlands, according to several publications (e.g., Glennie and Underhill, 1998). The area of the Southern North Sea was part of the foreland basin of the Variscan fold-and-thrust belt. Several fault trends are often interpreted to be associated with the Variscan compressional event, such as east–west-striking faults in the deeper subsurface.
At the end of the Carboniferous the area was subjected to post-orogenic tectonism. Oblique-slip faulting and thermal uplift are associated with this tectonic phase. The oblique-slip faulting resulted in the development of large NE–SW and NW–SE conjugate fault systems. The long period of thermal uplift for 10 to 20 million years caused significant erosion of the top Carboniferous (Van Wijhe, 1987; Coward, 1993; Glennie and Underhill, 1998).
It should be noted that these fault patterns already existed before the Rotliegend Interval was deposited. The presence of these faults at the Rotliegend level clearly indicates that many of these faults have been reactivated during different, if not all, subsequent tectonic events.
Regional subsidence due to the relaxation of litho-spheric thermal anomalies from the previous tectonic and volcanic phase resulted in the development of the east– west-trending Southern Permian Basin (Van Wees et al., 2000). The Rotliegend clastic sediments were deposited in this basin.
An early rifting phase took place during the Early Permian. This was strong predominantly in the Viking and Central grabens, and locally in Germany. In the Netherlands, rapid deposition and loading of anhydrite upon a heterogeneous basement, in combination with mild E–W extension, resulted in a series of small Late Permian pull-apart basins and tilted fault blocks that formed and crosscut the Variscan Front (Tubantian phase; Geluk, 2005).
The next tectonic phase was extensional and took place in pulses over a long period of time starting in the Early Triassic up to the Early Cretaceous. The initial phase occurred during the Early Triassic to Early Jurassic. The tensional stresses were related to the opening of the Norwegian–Greenland Sea, and therewith the start of the breakup of the Pangaea supercontinent (Fig. 2). This opening of the Norwegian–Greenland Sea later continued southward and formed part of the Mid-Atlantic Ridge (Ziegler, 1990). This Early Triassic to Early Jurassic phase is also referred to as the Early Kimmerian tectonic phase (e.g., Van Adrichem Boogaert and Kouwe, 1997; De Jager, 2007). A branch of this extensional faulting extended to the east-southeast into the North Sea. The intensity of extension decreased towards the south, although several large grabens are related to this Early Triassic to Early Jurassic extension, including the Dutch Central Graben and the Broad Fourteens Basin. Furthermore, local, smaller-scale Early Triassic to Early Jurassic grabens occur with unknown infill. It is clear from fault maps in the Triassic and Permian (Rotliegend) that the Early Triassic to Early Jurassic extensional grabens formed along the dominant pre-existing fault trends developed at an earlier stage in the structural evolution of the basin.
The Middle Jurassic is characterised by major uplift in relation to thermal doming in the Central North Sea and associated with another pulse of the Mesozoic breakup of the Pangaean supercontinent (Ziegler, 1990; Underhill and Partington, 1993; Glennie and Underhill, 1998) referred to as the Mid-Kimmerian uplift. This tectonic phase is further characterised by volcanism and significant erosion in parts of the North Sea, especially in the Central North Sea.
The Late Jurassic to Early Cretaceous period is the important later phase of the Mesozoic extension that took place in northwest Europe in which the main tectonic structures visible at present were developed in the subsurface of the Netherlands. Extensional tectonic activity increased as a result of the breakup of Laurasia (Ziegler, 1990) and led to the development of the Atlantic Ocean and the spreading between North America and Eurasia. This exten-sional phase had significant influence in the Dutch subsurface and further amplified the development of large-scale grabens like the Broad Fourteens Basin, the Vlieland Gra-ben, and the Lauwerszee Trough.
Extension in this later phase commenced in the Late Jurassic and is related to the rifting stage and break-up of the North Atlantic, the area between Greenland and Norway, splitting North America from Eurasia (Fig. 3; Ziegler, 1990; Glennie and Underhill, 1998). This major tectonic event is referred to as the Late Kimmerian rifting event, starting in the Late Jurassic and continuing into the Early Cretaceous. The tectonic stress orientation started with an approximate ENE–WSW to E–W extension during the Middle to Late Jurassic but rotated towards NE–SW and NNE–SSW in the Late Jurassic (Nalpas et al., 1995).
The approximate east–west extension developed strong continuous north–south-trending faults in the northern part of the North Sea. In the southern part of the North Sea, the main existing structural grain was reactivated. This resulted in significant oblique extension of the northwest– southeast-trending grabens, such as the Broad Fourteens Basin, the West Netherlands Basin, the Roer Valley Graben, and the Lauwerszee Trough. Although this would suggest quite some dextral transtensional movement along NW– SE-trending faults, this is not observed when studied in detail. Only minor dextral movement can be supported locally along some NW–SE-trending faults.
In the Early Cretaceous, rifting was concentrated in the North Atlantic and started the development of oceanic crust. Consequently, the extension in the North Sea decreased rapidly (Ziegler, 1981, 1990).
The final major stage of deformation in the Dutch subsurface took place from the Late Cretaceous to the Early Tertiary, often referred to as the Alpine inversion. The Netherlands was affected by compressional deformation (Ziegler, 1990), resulting in inversion of the existing basins (De Jager, 2003). Although this inversion event is often referred to as one inversion phase, it actually consists of multiple phases, and varying in intensity from basin to basin (Oakman and Partington, 1998; De Jager, 2003, 2007). Obvious expressions of the inversion event are the local presence of overthrusts and the occurrence of pop-ups along major NW–SE fault trends. The pop-ups occur predominantly north of the Broad Fourteens Basin, in the West Netherlands Basin, and in the Groningen Field (Fig. 4A, 4B).
The development of the compressional, inverted features may be related to a locked fault system and the regional transition into another structural domain, resulting in restricted transpressional movement, rotation, and uplift of the lens-shaped fault blocks (Fig. 4A, 4B). The direction of compressional stress was approximately north– south (Nalpas et al., 1995). This Late Cretaceous to Early Tertiary inversion is important in relation to the quality of fault seal because many of the prospect-bounding faults have been reactivated. The conditions at the time of the last reactivation of a fault define the final fault rock properties. These conditions are temperature, pressure, the amount of fault movement, and the (local) stress direction and magnitude.
The Rotliegend underwent many phases of deformation, under different stress orientations (Fig. 5), pressures (burial depth), and temperatures, all of which play their role in the final rock properties of faults and fractures.
Faults and Fault Damage Zones
The prospect-bounding faults and their sealing behaviour at the geological timescale are of prime interest to exploration. Exploration targets in the Rotliegend of the North Sea are often fault-bounded structural closures, i.e., local structural highs enclosed by an adequate top seal, typically the Ten Boer Claystone and/or the Zechstein evaporites (Fig. 6).
Many factors define the quality of a prospect and the possibility of encountering hydrocarbons in Rotliegend reservoirs. When charged with hydrocarbons, these Rotliegend fault structures are commonly filled to a spill point, either to a mapped synclinal spill or a fault juxtaposition spill, whichever is shallower (e.g., Corona, 2005). However, some of these Rotliegend fault structures may contain gas below the fault juxtaposition spill point where the sand-on-sand juxtaposition has been sealing over a geological timescale. In such a case, the structure would have been charged with hydrocarbons in the past and would have kept it in place for millions of years. This requires an effective sealing mechanism for a relatively thin layer in between two sandstone fault blocks, not easily susceptible to small cracks, dissolution, or other means to create holes in this sealing layer.
The key question for exploration regarding these faults is: are these prospect-bounding faults sealing or not sealing? After fifty years of exploration since the Groningen discovery, one would expect there to be a full understanding of the sealing mechanisms of the Rotliegend faults, but this is not the case. Evaluation of fault-zone sealing properties can be difficult, especially when considering the very limited subsurface data set typically available. The main problem lies in the complexity of the structural tectonic history, which largely defines the ultimate quality of the fault seal (see more below).
Fault Damage Zones
Although the hydrocarbon industry has been blessed with the developments in seismic acquisition and high-quality imaging of faults in the subsurface, we may forget about the actual character of faults. We would like to emphasise that faults should not be seen as individual continuous surfaces, as we typically interpret these on seismic, but as a broader zone of deformation with much lateral and vertical variation in style and properties.
Regular outcrop visits assist in obtaining an improved insight into the fault-zone architecture, and with that a better understanding of flow properties across and along the fault zones. In general, fault zones consist of two mechanical units, a fault core and a fault damage zone (Fig. 7). The fault core represents the centre of the fault and contains cataclasites, fault gouge, breccia, and sharp contacts. The fault damage zone contains fractures, folds, veins, and deformation bands. Most of the fault displacement is accommodated in the fault core, which includes the principal slip surfaces of the fault zone.
The definition of a fault damage zone is the area around the fault core in which the density of deformation features exceeds the average regional background level of deformation in the surrounding host rock.
The width of the fault damage zone is defined by the fault throw, the total displacement, and the host-rock mechanical properties. Extensive core studies, distance of wells to the fault zone interpreted on seismic, and their measured petrophysical properties lead to an overall average of 50–70 metres for the width of the prospect-bounding fault damage zones in the Rotliegend (in-house Nederlandse Aardolie Maatschappij (NAM) database). However, as mentioned above, the width of fault damage zones depends on a number of factors, including fault throw. This also implies that when faults die out, it is likely that the damage zone associated with this fault becomes narrower. A strong logarithmically increasing deformation gradient is observable when approaching the core of the fault, becoming increasingly more complex and more intensely fractured, with stronger grain-size reduction, and so on (Du Bernard et al., 2002).
Although relatively narrow with respect to the size of many Rotliegend fields, the fault damage zone plays an important role in well positioning. Production wells targeting the crest of tilted Rotliegend horst should be drilled outside the fault damage zone to avoid significant reservoir impairment due to fracturing and cementation (Fig. 8). On the other hand, in tight Rotliegend reservoirs a higher-density fracture zone may be encountered in these fault damage zones, with possible partially open fractures, enabling improved fluid flow in the otherwise very low-permeability rocks.
Fault Seal Mechanisms
The variation in fault styles, fault trends, and their extent is obvious from fault mapping work on the Rotliegend (Fig. 1). Although these faults may have a similar appearance, their properties may be very different. An extensive list of Dutch field reports and company internal publications and published literature describes the effects faults can have on the production efficiency of fields. Production profiles, differences in geochemical composition of gas samples, and pressure differences across faults suggest that compartmentalisation can occur in the high-net-to-gross Rotliegend sandstones. For example, the Groningen gas field is commonly seen as one large tank, but accurate measurements over the years confirm that many faults behave as production baffles or as strong pressure boundaries over the production timescale. Offshore, in the Broad Fourteens Basin, a number of gas fields are compartmentalised by sand-on-sand juxtaposed sealing faults that can hold pressure differences of more than 200 bar. Also, free-water-level differences of 120–130 m across a fault are encountered, up to roughly 20 bar pressure difference over the geological timescale.
Understanding the sealing behaviour of faults and its quality is very relevant. The number of remaining prospects that could be made economically viable but are fully dependent on the structural closure defined by its spill point is quickly diminishing. Structures that are bounded by well-developed sealing faults could contain a larger hydrocarbon column, and could therefore make small structural closures more attractive for development knowing that a larger column may be encountered. Furthermore, knowing in advance that a field is likely to be compartmentalised would require a very different development approach than in the situation where the field is likely to behave as a structure in full communication.
How could our insight into the fault-seal mechanisms and the related fault rock properties be improved? We do not have many fault calibration points to determine rock properties, such that a database could be built up for predicting the sealing quality of individual faults. Firstly, faults are often avoided when drilling for a prospective target. Secondly, a large number of calibration points are required to build consistency and confidence, covering all fault types in the various settings and Rotliegend lithologies.
From other papers in this special volume it is clear that pronounced differences in Upper Rotliegend reservoir-rock composition can be encountered (Grötsch et al., 2010; Gaupp and Okkerman, this volume). In some areas the Upper Rotliegend can have a high net-to-gross value (90+%), whereas laterally the net-to-gross quickly diminishes, to values of 50% or less, within distances of 5–15 km. Furthermore, postdepositional effects like burial depth and its related temperature, reactivation, diagenesis, and/or fluid circulation along faults are all mechanisms that have an impact on the final properties of the fault rock. Additionally the fault-zone properties are dependent on the amount of shale in the system (net-to-gross) and the dip and strike (relative to the principal stresses).
Although only few faults, or parts thereof, have been dynamically calibrated, fractures and faults in core do provide some insight into the sealing conditions of faults in the Rotliegend. The mechanisms observed at a small scale may indicate that similar mechanisms occur at larger scales. For example, the presence of well-developed cataclastic fractures in core is positive evidence that the temperature–stress conditions at certain stages of burial and structuration were favourable for crushing of quartz grains and recrystallisation of silica (Fisher and Knipe, 1998, 2001; Fisher et al., 2003).
Potential fault-seal mechanisms in the Upper Rotliegend sandstones are cataclastic and shale gouge. Other fault-seal mechanisms, such as cementation and other forms of di-agenesis, also occur in the Rotliegend as observed in core. However, these mechanisms mostly have a local effect and are more likely to form baffles at the production timescale. The various core fracture types are described in more detail in the section on Phyllosilicate Framework Fractures on page 241 and in Appendix B6.
FRACTURES, MORE FRACTURES, AND FRACTURED RESERVOIRS
In this paper we focus on the description of the different styles of natural fractures observed in Rotliegend cores (Appendix B6) and how these can aid, at a relatively generic level, reservoir and fault characterisation. Although many Rotliegend reservoirs are tight, having low porosity and permeability, and rely to varying degrees on the presence and producibility of natural fractures (e.g., Sole Pit area; offshore license blocks K17 and part of L13), it is emphasised here that the presence of fractures in core, for example, does not directly imply a “naturally fractured reservoir”. A naturally fractured reservoir is a reservoir that requires fractures to explain its production behaviour. Aspects of naturally fractured reservoirs not dealt with in detail in this paper include the mechanisms of fracture initiation and propagation, the associated fracture nomenclature (e.g., Ramsay, 1967; Pollard and Aydin, 1988; Knipe, 1989; Kulander et al., 1990; Twiss and Moores, 1992; Mollema and Antonellini, 1998; Myers and Aydin, 2004) and the analysis, calibration, and modelling of fractures for reservoir characterisation (e.g., Aguilera, 1995; Nelson, 2001; Mäkel, 2007; Lonergan et al., 2007; and references in these sources).
However, to set the scene for the description of Rotliegend fractures, it is considered useful to provide some high-level background to natural fractures, covering (1) fracture development and propagation, (2) fracture geological classes, (3) fault damage zones in particular, (4) some key fracture data types, and (5) some aspects of fracture modelling. This is then followed in Section 4 on the description of the various Rotliegend fracture types as observed in core. No assignment is made to geological fracture classes (see section on Fracture Geological Classes, on this page), inasmuch as a field-specific description of the structural setting is outside the scope of this paper. Cataclastic deformation bands dominate, albeit with complex diagenetic histories and common reactivation. These are interpreted as being primarily fault-related.
Fracture Development and Propagation
To understand fractures and the 3D distribution and flow properties of fractures, it is key to understand their origin and subsequent development. Our experience is that there is often confusion around the term “fracture”. First, besides natural fractures in the subsurface, fractures can also be, for example, drilling-induced or coring-induced (e.g., Kulander et al., 1990). Secondly, in the realm of natural or subsurface fractures, nonspecialists working on subsurface studies often have a quite different perception of fractures and their influence. For example, some geologists and geophysicists relate fractures primarily to folding and the concept of curvature (i.e., the more bent a rock layer, the more likely it is to contain brittle fractures). Others relate fractures conceptually more to being located near faults and/or in fault damage zones. Also, there exists a diverse and overlapping fracture terminology, such as joints, tensile fractures, shear fractures, hybrid fractures, wing cracks, veins; Mode I, II, and III fractures, et cetera.
Kulander et al. (1990) provide diagnostic criteria to determine the mechanical origin of fractures from core, such as “tensile”, “shear”, and “coring-induced” fractures. The tensile and shear fractures are further subdivided into different modes. The tensile, or extensional, fracture is classified as a mode I fracture. In this fracture type the relative motion is perpendicular to the fracture. The shear fracture has a sliding motion parallel to the fracture surface and is classified as a mode II or mode III fracture. In mode II fractures the sliding motion is perpendicular to the edge of the fracture, i.e., dip slip. In mode III fractures the sliding motion is parallel to the fracture edge, i.e., strike-slip (Twiss and Moores, 1992).
Fracture Geological Classes
Broadly speaking, natural fractures can be subdivided into four “geological” classes: (1) fault-related fractures, (2) fold-related fractures, (3) strata-bounded fractures, and (4) fracture corridors (Fig. 9).
(1) Fold-Related Fractures.—
Development of these fractures occurs during folding and is the direct consequence of folding through an outer-arc extension or layer-parallel slip. The major dilational fractures related to folds tend to be perpendicular to the bedding, thus forming some kind of fan shape across the fold. Stearns (1964) was one of the first to provide a comprehensive study relating fracturing and folding from well-exposed anticlines in the western U.S. More recent studies with a large outcrop component include Bazalgette (2004), Bazalgette and Petit (2007), and Stephenson et al. (2007). Fold-related fractures are most commonly predicted using curvature analysis of geological surfaces (e.g., Lisle, 1994; Bergbauer and Pollard, 2003, 2004). Reactivation during folding of pre-existing structural features (e.g., Bergbauer and Pollard, 2004; De Keijzer et al., 2007; Rawnsley et al., 2007; Ferrill et al., 2007) instead of formation of new fractures is one aspect that can complicate fracture predictions from curvature analysis. It is difficult, if not impossible, to make any inferences on a possible relationship between folding and fracturing for the fracture examples in Appendix B6, as no other data are considered in this paper.
(2) Fault-Related Fractures.—
As described in more detail below, the damage zone of subseismic-scale and seismic-scale faults may contain fractures alongside other deformational features such as folds, veins, and deformation bands. These zones can be up to 100– 150 m wide, especially in areas with complex fault geometries such as fault intersections, relay zones, and fault bends. A small-scale example of a damage zone in the form of splay fractures around a brecciated fault zone is provided in Appendix B6p. Many papers address the geometry and characteristics of fault damage zones in sandstones (e.g., Edwards et al., 1993; McGrath and Davison, 1995; Beach et al., 1999; Shipton and Cowie, 2003; Berg and Skar 2005).
Edwards et al. (1993) address faults and their damage zones in aeolian Hopeman sandstones along the southern margin of the Moray Firth, Scotland. These sandstones are the Permo-Triassic equivalent of the Dutch Upper Rotliegend reservoirs and as such represent a very good analogue. These authors describe a variety of structural features from solitary deformation bands to compound deformation bands (Fig. 10) to intense deformation along the major fault slip surfaces, all of which form significant flow barriers or baffles. A core equivalent of “compound” deformation bands is provided in Appendix B6l.
(3) Strata-Bounded Fractures.—
Fractures start to develop in the more brittle layers as the differential stress first reaches the failure envelope (Welch et al., 2009a; Welch et al., 2009b) (Appendices B6k and B6o). Where reservoir rocks are mechanically layered, fracture growth is expected to be controlled mechanically as well (e.g., by bedding planes or lithology contrasts). This influences fracture orientations and results in limited vertical fracture persistence. In nature, mechanical layering occurs at various scales, and therefore fracture systems are expected to comprise a multi-scaled, nested system of strata-bounded fractures. Significant efforts have been made in recent years to advance the understanding of mechanical stratigraphy in relationship to fracture development (e.g., Cooke and Underwood, 2001; Lorenz et al., 2002; Guitton et al., 2003; De Keijzer et al., 2007; Stephenson et al., 2007) and fault development (e.g., Wilkins and Gross, 2002; Soliva and Benedicto, 2005). In contrast to localised stress conditions controlling fracture distributions in folds, strata-bounded fracture systems can be governed by regional stresses (e.g. Lorenz et al., 1991). Rawnsley et al. (1998) interpret part of the joint system exposed along the Bristol Channel as being associated with the regional Alpine stress regime extending into northwestern Europe. Similar regional stress-related fractures could also be envisaged in the Dutch subsurface. Examples of mechanically controlled, strata-bounded fractures are shown in Appendices B6e, B6h, B6n, and B6u. An example of more pronounced strata-bounded fracturing observed on core from the overlying Zechstein ZEZ3C interval is shown in Appendix B6o.
(4) Fracture Corridors.—
Fracture corridors comprise clustered, essentially tensile fractures with little to no offset of layering. They occur in carbonate and sandstone rocks alike and can be tens of metres wide. Their formation, distribution, and structural controls are still relatively poorly understood and form a topic of continued research. They may well be intricately related to the (multi-scaled) strata-bounded fractures described above. A key paper on the topic is by Bai and Pollard (2000). Other examples are described by, for example, Sagy and Reches (2006) and De Keijzer et al. (2007). The available cores do not sample evident fracture corridors, although it is difficult to judge given the limitations of core sampling.
Although the above subdivision of fractures is by no means all-encompassing (e.g., another class could be polygonal faults and fractures), it emphasises that one natural fracture is not the other. Different rules apply to the characterisation and modelling of fractures from the various fracture classes, in addition to differences imposed by lithology, reservoir properties, diagenesis, reactivation, etc. Also, the above-mentioned “geological” classes are not necessarily mutually exclusive. For example, the strain associated with long-wavelength folding may be accommodated at smaller scales by small-scale faulting and/or the reactivation of pre-existing tensile fractures in shear mode (e.g., Ferrill et al., 2007; De Keijzer et al., 2007). Finally, kinematic interpretation of, for example, a shear fracture at ca. 60° observed in core need not be straightforward. What was the orientation at the time of formation? Has it seen multiple stages of shear? Is it part of a conjugate fracture/fault set? Is it part of a fracture jump across lithologies with a different angle on average? To what larger-scale structure does it belong?
Fracture Data—Image Logs and Production Data
Strong indications that a reservoir is fractured, i.e., where flow is affected, include a dual permeability response from a well test, gas peaks during drilling, bit drops or jams during drilling, anomalously high permeability measurements, and production logging tool (PLT) measurements. Secondary indications where flow may be affected include electrically conductive fractures from borehole images, poor core recovery (weak fracture or fault surfaces that preferentially lose cohesion), and partly cemented fractures in the core. Although we fully realise that a proper subsurface characterisation of fractures and fractured reservoirs requires full integration of all fracture-related data, this is far beyond the scope of this paper.
In this paper, we focus on core examples. In fact, a core provides the only direct observation of fractures in the subsurface, and core information is therefore crucial to obtain improved insights. One example would be the observation in core of partially open fractures in which crystals have grown in between the fracture walls. This provides one of the most definitive indications that fractures are open in the subsurface and thus potentially provide a preferential pathway for fluids. Other kinds of fracture information that can be derived from core include (or at least provide some constraints) are mechanical layering (i.e., relationships between fractures and lithology), fracture types (see Phyllosilicate Framework Fractures section on page 241), length and size distributions, spacing, relative age (e.g., using abutting and offset relationships), absolute fracture properties (e.g., relative permeability and entry pressures of cataclasites; (also see Phyllosilicate Framework Fractures section on page 241), and aperture. Estimates of fracture aperture from a core are always a maximum value because the core expands during coring because of unloading (removal of the overburden) and associated stress release. Evidently, a key limitation of core is its small scale and essentially 1D nature relative to grid cells and reservoirs, and modelling of the 3D geometry and properties of fractures away from the well bore requires other data constraints.
Care should be taken to distinguish natural fractures from coring-induced fractures. The latter fractures form in response to stress changes that are induced by the drilling and coring processes (e.g., stress relief during unloading of a core). Although there are few definitive criteria for distinguishing coring-induced fractures, fractures parallel to the borehole axis and/or parallel to the maximum present-day horizontal stress direction most likely would be coring-induced (e.g., Kulander et al., 1990). An uncertainty remains on rubbled core intervals. These intervals are by default assumed to represent intervals where core damage was suffered during coring, retrieval, and core handling. Core damage was difficult to explain for some rubbled intervals, however, without invoking natural weaknesses in the rock (e.g., poorly consolidated rock, open natural fractures). Without other information (e.g., image logs of borehole wall, mud losses) to allow matching of the rubbled intervals to possible open fractures in the rock, the ambiguity remains non-resolvable.
In summary, knowledge of the geometry, mechanism, and orientation of the fractures from core will help to reconstruct the regional tectonic history, thereby allowing proper definition of conceptual and quantitative subsurface models.
For successful field development, a solid understanding is required of the 3D distribution of fracture flow properties at the well and interwell scales. Typical questions posed by a development team could relate to: (1) relationships between fracturing and reservoir layering, (2) 3D geometry (directions, spacing, length, etc.) of fluid-conductive fractures, and (3) fracture-prediction drivers from geomechanical tools and/or seismic attributes to predict fractures in 3D away from the wellbore. For this, a good conceptual understanding of the geometry (i.e., static aspects) of the fracture–fault system, also incorporating uncertainty, is an essential first step. However, a combination of the following factors typically precludes the definition of a single, deterministic fracture–fault model as input into simulation:
The sub-seismic nature of fluid-conductive fractures and faults
Wells are often lacking sufficient “hard” static and dynamic fracture data to pinpoint and fully characterise inflow zones. For example, during underbalanced well operations inflow is obtained only from the fractures and permeable matrix that have a pressure higher than the pressure in the well bore. A complication is that the well-bore pressure varies between toe and heel of horizontal well and that fractures with depleted pressure may actually cause losses, even in underbalanced well operations.
Often mediocre quality and resolution of seismic imgaging due to reservoir depth (typically ca. 3 km) and complex overburden that includes an overlying Zechstein salt sequence.
High structural complexity with multiple tectonic phases and variability across structural domains.
Limited deterministic understanding at the mesoscale and macroscales of fractures and fault diagenesis, if at all studied from available core. This results in moderate to large uncertainty on connectivity patterns at typical well-spacing scale (hundreds of metres to kilometres).
Last but not least, little or no deterministic constraints as to which faults, or parts thereof, have been reactivated, especially during the Alpine inversion stages. This com plicates finding clues to preferential localised development or reactivation of late-stage potentially open fractures.
Thus, proper uncertainty definition is essential, leading to multiple conceptual sound and data-driven 3D models to be defined. From these, plausible fracture models can be further assessed conceptually and quantitatively. One such study was done by De Keijzer et al. (2007). Even though that study did not concern Upper Rotliegend reservoirs, it does provide a good example of the full suite of data constraints used (including outcrops, core, fracture diagenesis, and isotope analyses) to derive data-driven conceptual fracture models with reduced uncertainty, which “quickly” led to an improved reservoir history matching.
As mentioned, every area in the Dutch Rotliegend has its specific tectonic, burial, and diagenetic history. Specific aspects to consider that could help in predicting open fractures include: present-day stress regime (open fractures expected parallel to the maximum principal stress), late-stage fault reactivation, inversion specifics, and height above free water level. However, present-day stress should be used with caution, because the pre-existing fault fabric can create local deviations in stress orientation and magnitude.
Figures 11 to 13 show examples of different conceptual models for the Rotliegend of the Greater Sole Pit area (located in the UK Southern North Sea ∼ 80 km north of Bacton on the Norfolk coast of England). There, most “fractures” are believed to be (reactivated) shear fractures and/or cataclasites associated with faults and their damage zones. This interpretation is based on various data sources (e.g., core, seismic, and drilling inflow data) as well as structural concepts (e.g., strong clustering of fractures along-hole and lack of relationship to layering). That said, there still is uncertainty in terms of mechanical stratigraphy and fracture–fault spacing, which is captured in the different conceptual models.
Although there is a reasonable understanding of the impact of fractures on production behaviour in the Rotliegend, it remains difficult to accurately quantify the impact on overall reservoir performance.
Rotliegend Fracture Types and Their Petrophysical Properties
Investigation of thousands of metres of Upper Rotliegend cores stored by NAM in Assen, Netherlands, and studies of outcrops show that in general the Rotliegend interval does not contain a high density of fractures. Typically, hardly any natural fractures can be observed in the core slabs of the many Dutch wells that were drilled. Exceptions exist for those few wells that have been drilled into a fault damage zone, but such wells are rare.
Cores from many wells have been analysed in detail with focus on fracture type and style. It is possible to categorise the Rotliegend fractures into six types, roughly defined by their clay content and depth of burial at time of faulting:
phyllosilicate framework fractures
Cataclastic fractures are the most abundant fracture type in the Dutch Upper Rotliegend reservoirs. Several cores show multiple phases of fracturing, in which the older fracture is usually a strongly consolidated cataclastic fracture, whereas the more recent fracture is often partially cemented (Appendices B6g and B6j). The latter are likely developed by reactivation, often related to the Alpine inversion phase or a late and more local extensional deformation.
Disaggregation zones are local zones of weakness and deformation in poorly consolidated sediments. They develop mostly in sediments after shallow burial. Their exact style depends on the amount of clay content in the rock. The more clay-rich material present in the sediments, the more the disaggregation zone will show a ductile, plastic deformation style. With only minor presence of clay-rich material, like most of the Upper Rotliegend where gas-field development has taken place, the sediments behave in a more “brittle” manner. No or very limited grain breakage takes place in the development of these more “brittle” disaggregation zones. Particulate flow is the deformation style. The grains are rearranged and rotated into a preferred orientation, such that the grains better align with the applied stress field on the rock. Some frictional grain-boundary sliding occurs.
The Upper Rotliegend interval in the Southern Permian Basin did not experience significant early deformation, and therefore these disaggregation zones are not often encountered in Rotliegend cores. The few disaggre-gation zones observed in the cores are very subtle features. The main reason for the fact that they are often difficult to distinguish in cores is that no grain breakage occurs, and therefore the grains in this zone have maintained their original grain size and also have the same sorting as their host rock. Photomicrographs of disaggregation zones confirm this (Fig. 14).
Disaggregation zones do not form barriers to fluid flow in the Rotliegend sandstone rocks. More clay-rich disaggre-gation zones may form a zone with reduced fluid flow due to the rearranging of clay minerals under seemingly plastic deformation behaviour and in fact forming some kind of clay smear.
Cataclastic fractures develop in very silica-rich host rocks, with less than 15% phyllosilicate material (Fisher and Knipe, 2001). In general, the Upper Slochteren and Lower Slochteren formations of the Rotliegend have a very high net-to-gross. In the most prolific hydrocarbon areas in the Netherlands, the net-to-gross is commonly above 90%. This explains the fact that cataclasis is the most dominant fracture type observed in Rotliegend cores, in addition to the fact that faulting occurred when the Rotliegend was more deeply buried.
Cataclastic fractures are typically lighter in colour than the host rock, its quality commonly very continuous and seemingly integral with the rock (see Appendices B6d, f, m, x, y). The host rock around the fracture sometimes has a lighter colour, which is referred to as a fracture halo. This bleaching of the host rock is interpreted to be related to flow of hydrocarbons through the rock (Gaupp and Okkerman, this volume; and Appendices B6a, b, c, d, f, t, v).
The main deformation process in cataclasis is grain fracturing (Fig. 15), frictional sliding, and subsequent quartz cementation. The process of grain crushing and frictional sliding results in significant reduction of porosity and permeability. Along the fracture, the smaller pieces of grains are rearranging and rotating to optimise compaction.
Various authors (Oelkers et al., 1996; Fisher and Knipe, 2001) show that the amount of grain-size reduction between the various cataclastic fractures in the Rotliegend can vary significantly, ranging from only minor grain fracturing down to a state in which all grains are strongly crushed into very small remnants (Fig. 16). The degree of grain-size reduction depends on the stress applied to the rock at the time of deformation, the original grain size, the variability in grain size of the host rock, the lithified stage of the rock at time of deformation, etc. The two most important factors defining the grain-size reduction are: (1) the effective stress at time of deformation, and (2) the total amount of strain accommodated.
The degree of grain-size reduction also appears to be defined by the maximum burial depth of the rock, which is not necessarily the depth at the time of deformation. A trend has been observed between permeability and maximum burial depth. The deeper the rocks were buried, the stronger the compaction and lithification, and the more likely the deformation and fracturing occurred at higher pressures and temperatures.
Local quartz cementation is often observed in addition to this breakage of grains. The source of quartz cement in cataclastic fractures is local, derived from grain crushing and subsequent quartz dissolution. The cement is subsequently filling the small pores proximal to their origin, commonly within or very closely along the fracture zone, building up a strong and impermeable fracture (Fig. 16, Appendices B6f, h, i. The many clean surfaces created by grain breakage are ideal locations for quartz crystal growth. The grain-to-grain quartz dissolution can be further enhanced by the presence of minor amounts of phyllosilicate material at the grain surfaces (e.g., Bjørkum, 1996). The irregularities caused by the presence of these micro-minerals create a preferred rough surface at which quartz prefers to precipitate.
A very important factor defining recrystallisation of quartz is the temperature of the rock at time of deformation. Quartz cementation takes place above approximately 90° C (Giles et al., 1992; Giles, 1997; Fisher et al., 2003). Examination of large databases shows that quartz cementation only occurs locally at temperatures from 80° to 90° C (Gluyas and Coleman, 1992).
When evaluating cataclastic fault or fracture seal properties it is crucial to take the temperature at time of faulting into account. Since the temperature of the rock at time of deformation is in most cases directly related to the depth of burial, it is relevant to understand the basin development. When cataclastic fractures or faults develop below 90° C, their surfaces may become polluted and less prone to later silica crystallisation. A direct correlation can be made between the degree of quartz cementation and cross-flow permeability of fractures. The more intense the quartz cementation, the lower the permeability for flow across the fracture.
The above describes quartz cementation at the time of deformation. Additionally, post-deformation quartz cementation can take place. This is steered by post-deformation stresses, temperature history following deformation, and conditions of the rock, such that diffusive mass transfer can occur.
It is interpreted that the Rotliegend cataclastic fractures are mainly related to an early deformation phase, during intermediate burial, likely developed during the Triassic and Jurassic extension phases (see section on Regional Tectonic History on page 230).
Cataclastic faults and fractures can form significantly strong barriers to fluid flow, holding back pressure differences of more than 200 bar. Depending on the processes and local conditions of the rock at time of deformation, the permeability can range from several mD down to ∼ 0.0005 mD.
Whereas cataclastic fractures derive their quartz cement from a very local source, the cemented fractures derive their cement from an external source. The cements in cemented fractures are precipitated from saturated fluids migrating through open faults and fracture networks. Cemented fractures are commonly observed in Rotliegend cores. Figure 17 shows a photomicrograph of a cemented fracture (Appendix B6b, c, g, t).
The appearance of cemented fractures is often bright because the colour of the precipitated cement often is lighter than the host rock. Anhydrite is the most common cement type. Other cement types are ankerite, barite, dolomite, quartz, and siderite. Whereas anhydrite, ankerite, and dolomite are likely sourced from fluids from the overlying Zechstein evapor-ite sequences, barite and siderite likely originate from fluids from the underlying Carboniferous intervals.
Cemented fractures are likely to form pronounced barriers at the production timescale, based on measured petro-physical properties. However, cementation typically varies along fracture surfaces. Locally, fracture cement is pervasive and seems to form a flow barrier and seal, but along the rest of the fracture a lot of irregularities, thinning, and absence of cement is commonly observed (Appendices B6g, j, n, r). This large variation of cementation quality is an important characteristic of this type of fracture. Thus, cemented fractures can form locally significant boundaries, but not effective hydrocarbon seals.
Phyllosilicate Framework Fractures
Phyllosiliciate framework fractures form in impure sandstones, with a mixture of sand and clay-rich material (15– 40% phyllosilicate-rich material). They do not develop in alternating layers of sandstones and shales.
Their appearance in the Rotliegend is often of darker colour, related to the darker clay-rich material in the fracture compared to the sandstone host rock (Appendix B6n, u). Since most of the Upper Rotliegend has very high net-to-gross, these phyllosilicate fault rocks have not been observed in great numbers. Commonly these fractures are encountered only where the Upper Rotliegend contains relatively high concentrations of detrital phyllosilicate material. Considering the depositional setting of the Dutch Upper Rotliegend, these fractures are expected to become more dominantly present towards the north, into the deeper parts of the Southern Permian Basin.
The fine-grained, platy, intergranular phyllosilicates and their micro-porosity network define the fracture-rock flow properties. The various processes being applied to the rock define the ultimate quality of the fracture. Similar kinds of processes take place as described above. The first process is rearrangement and sorting of material, since the rock consists of a mixture of sand-rich and clay-rich material, and with different grain sizes. This is a first means of reducing the pore space. Subsequently, with deeper burial the detrital grains and clay minerals are fragmented, similarly to that observed in the development of a cataclastic fracture. Thirdly, the same kind of local quartz cementation can take place as with cataclasis, assisting in further reduction of permeability of the fractured rock. Considering these processes, the sealing properties of these fractures can be continuous. Unlike cemented fractures, the phyllosilicate framework fractures can be strong baffles to barriers to fluid flow and may form a hydrocarbon seal that can hold several tens of metres of gas.
Clay-smear fractures normally develop in sections with alternating sandstone and shale layers. Due to the overall high net-to-gross of the Upper Rotliegend, clay-smear fractures are not typically encountered, similar to fractures of phyllosilicate framework type. Occurrence of clay-smear fractures does depend on the lateral facies development in the Upper Rotliegend. This fracture type is likely to be encountered more frequently in the more clay-rich facies towards the north in the basinal area.
The development process of these fractures is that the clay material is taken up from the shale layers and smeared within the fracture (Yielding et al., 1997; Yieding, 2002; Van der Zee and Urai, 2005). Examples in cores clearly show the dragging of the clay-rich material into the fracture. This process is known as shear-zone clay smear (Lindsay et al., 1993). There are two other types of clay smear: abrasion clay smear, by which a thin veneer of clay material is developed as a result of a sandstone sliding along a shale layer, and injection clay smear, by which clay-rich material is more ductile and takes up dilatant sections along the fault and in a way injected into this space.
Permeability of clay-smear fractures is usually very low, less than 0.5 µD. Thus continuous to semicontinuous clay-smear fractures and faults can form strong baffles to barriers to fluid flow and may form significant hydrocarbon seals if they are coherent and continuous.
Generally the Rotliegend has high net-to-gross ratios and good porosity and permeability, and gas-well production rates are generally economically attractive. However, where the Rotliegend has developed as a tight or very tight reservoir, for example due to severe illite diagenesis (Gaupp and Okkerman, this volume) open or partly cemented natural fractures or hydraulic fractures are required to achieve economically attractive production rates. The cores taken in the Dutch Upper Rotliegend intervals do not contain a high density of open fractures, and many lack fractures altogether.
The (partly) open fractures observed in the Upper Rotliegend cores are often associated with earlier-formed fractures (i.e., reactivated and dilated). The character and properties of the cemented parts of partly open fractures are comparable to the description provided above on cemented fractures. Care should be taken with completely open fractures in core, taking into account the diagnostics of coring-induced fractures (Kulander et al., 1990).
Several Upper Rotliegend core sections investigated display two fracture sets related to different times of deformation. It is not always easy to identify which fracture belongs to which deformation phase. In general, crosscut-ting and offset of other fractures would date relative fracture timing, but often these characteristics cannot be distinguished with certainty in the Rotliegend cores. Their character, type of faulting, and present-day depth can further assist in defining the approximate timing of fractures.
Fracture-aided production is rare in the Dutch Upper Rotliegend reservoirs, and conclusive evidence of fracture contribution to flow is difficult to obtain from results of well tests. Most of the fractures observed in core are cemented and form baffles to lateral flow. Cores and/or image logs are generally not acquired in deviated and horizontal production wells. Many of the earlier exploration and appraisal wells that did core and log the reservoir were drilled as vertical wells that have a sub-optimal orientation to allow a representative sampling of vertical to subvertical fractures that may be present. The detailed information to study the presence and distribution of subsurface fractures is therefore sparse.
A review of all Dutch Rotliegend cores acquired by NAM showed fractures only in a limited number of wells and a low fracture density and limited fracture lengths. The majority of the individual fractures that were recorded in the core review are cemented and have limited lengths (centimetres to metres) and are thought to impact mainly as subvertical baffles to flow in the reservoir at the laminae (millimetres), layer (centimetres to decimetres), bed (metres), and subunit (tens of metres) scale.
Outcrop examples in the Scottish Permo-Triassic Hopeman sandstones (Edwards et al., 1993) that were outlined earlier in this text and the database of fracture types from core-based studies guide the assignment of property ranges of fractures for NAM’s reservoir evaluation and simulation.
Examples of both open and closed fractures are found in the cores, often in the same well. The cemented fractures that were observed are expected to cause increased flow tortuosity, baffling, and compartmentalisation at micro (millimetres to centimetres) and meso (decimetres to metres) scale, depending on the continuity of the cement in fractures. Open or partly open fractures, where present, diminish lateral baffling, significantly decrease flow tortuosity, and increase the vertical and lateral communication in the Upper Rotliegend reservoirs. Due to the pronounced heterogeneity of the Upper Rotliegend reservoirs, Kv/Kh values are typically low to very low. The presence of open natural fractures greatly alleviates the negative impact on vertical communication caused by the pronounced lamination and layering of the Upper Rotliegend reservoir rock.
A good example of diminished communication at the microscale due to cementation is observed in Appendix B6t where the rock matrix is cemented on both sides of the fracture. At the macroscale the impact of fractures on well productivity can be pronounced. In the same vertical exploration well L13-15 subvertical open fractures were also observed from the core (Appendix B6r). The well-test results indicated a contribution from natural open fractures of approximately one-third of the total test flow rate of 250 thousand cubic metres per day.
In the Greater Sole Pit area in the UK Southern North Sea the development of tight Upper Rotliegend was aided by the intersection of swarms of open natural fractures in horizontal wells that were drilled in underbalanced mode. A few wells managed very high sustained production rates of greater than 2 million cubic metres per day. It is noted that the very high production rates were obtained only from a few discrete fracture swarms. The location and conductivity of the fractures has been difficult to predict at the well scale, even after extensive and detailed structural studies involving a wealth of expertise and subsurface data, including fracture and fracture-scenario modelling.
The large remaining uncertainty on presence, location, and conductivity of fractures has resulted in a drive by many operators to develop the tighter Rotliegend reservoirs with horizontal wells featuring multiple hydraulic fracs. Additionally there are efforts to boost the productivity and recovery from vertical and deviated wells in existing fields by remedial hydraulic fraccing. The failure to further reduce uncertainty and thereby move from stochastic models of fractures to deterministic subsurface models has led to a significant increase in the cost of well development.
Hydraulic fracturing operations are complicated further by the potential connection of hydraulic fractures with open natural fractures. Connection to open fractures has led to early screen-outs during pumping operations, due to the rapid and large fluid loss into the open fracture networks. The presence of natural fractures also influences the height and length growth of the hydraulic fractures. Fracture growth suffers from dissipation of energy into natural fractures, the variation of conductivity in the natural fractures, and the generation of multiple frac planes as the hydraulic fracs start following the path of lowest resistance in fracture networks.
Linking the mechanical properties of the rock (which can in part be measured in laboratory testing, if core sample material permits it) and the actual presence of natural fractures is very challenging. The natural fractures reflect the specific loading history at individual areas and the mechanical behaviour and its heterogeneity in the host rock at that location during failure (Laubach et al., 2009). Area-specific answers would logically need to be generated from well data and structural study. Large remaining uncertainty is a typical outcome of fracture study work, though—however well executed in view of the paucity of accurate input.
The economic impact of a better understanding of fracture presence and better-quantified fracture properties has increased for the remaining undeveloped Dutch Rotliegend opportunities. The remaining undeveloped Dutch portfolios contain a large portion of opportunities with small recoverable volumes and/or tight reservoir, and the economic margins for developing these remaining opportunities are smaller.
Fractures have been observed in the small number of depleted Dutch Rotliegend reservoirs that have been put into use, or are planned to be used in the future, for underground gas storage. The injection and production response in these reservoirs is controlled by the most permeable features (i.e., high-permeability zones, open natural fractures, induced fractures) and the least permeable features (i.e., cemented natural fractures, cemented zones, and shale layers). The core examples from the Norg underground gas storage, located in the northeastern Netherlands about 30 km west of the Groningen gas field, show cemented fractures from multiple phases of fracturing (Appendix B6w, x, y). Reservoir response during injection and production in the gas storages may be affected by the fractures, but requires study to quantify.
Because the injected gas in these underground gas storages typically contains a small amount of H2S, understanding the flow path of the gas and the scavenging of H2S in the reservoir by iron-bearing minerals in the Rotliegend rocks has added economic importance. If fractures do have a strong control on flow path, with injection and production, the depletion of natural scavenging capacity in those areas where inflow and outflow was largest may have occurred sooner than could be predicted using the scavenging values based on the analysis done on rock samples. The need for installing additional scavenging capacity at the surface would be better quantifiable with further study on impact of the fractures.
Detailed core analysis has shown that the common fracture types in the Rotliegend are cataclastic, cemented, phyllosilicate framework, clay smear, and open fractures. The first two types are the most common fractures encountered in the Rotliegend sandstones.
A catalogue with all fracture types encountered in cores from the Dutch Upper Rotliegend reservoir has been constructed. It remains difficult, however, to link the presence and properties of fractures quantitatively to observed reservoir response. There is thus still a large remaining and perhaps irreducible uncertainty regarding the presence and properties of fractures.
Fractures do not appear to play a key role in the production response of the Dutch Rotliegend reservoirs. Fractures are quite rarely encountered in core, with only a limited number of documented and published examples from other well data. Understanding the origin, development, and properties of fractures is of increased relevance where the Rotliegend sandstone reservoirs are tight. A significant number of remaining development opportunities in the Dutch onshore and offshore are contained in such reservoirs.
Fracture studies and evaluations are considered beneficial for optimising well trajectories and overall field development, even with the remaining fracture uncertainties.
The study of natural fractures also assists in improving our insight into the properties of faults that define the boundaries of individual prospects or compartments within fields.
Part of the data presented is derived from well logs and cores and internal reservoir study reports of the Nederlands Aardolie Maatschappij (NAM) and Shell U.K. Exploration and Production Ltd. Both companies are thanked for the permission to publish material contained in these reports. Support by numerous staff in NAM B.V. (e.g., Reina Van Dijk, Bettie Oosting, Jan Penninga, Jan Tillema) is greatly appreciated. We are grateful for constructive reviews by Franco Corona and Jan-Diederik Van Wees.
Figures & Tables
The Permian Rotliegend of the Netherlands
More than 50 years ago, the discovery of the giant Groningen Gas Field in the subsurface of the Netherlands by NAM B.V. marked a turning point inthe Dutch and European energy market initiating the replacement of coal by gas. Despite the fact that the Rotliegend dryland deposits in the Southern Permian Basin are one of Europe's most important georesources, no sedimentological overview is available to date for the subsurface of the Netherlands. This SEPM Special Publication presents for the first time such a summary of the present-day knowledge, including a comprehensive core atlas from on- and offshore wells.