Abstract
Late Ordovician (c. 445 Ma) glacial sandstones form important gas reservoirs in the Illizi Basin, SW Algeria. These reservoirs have a high degree of depositional and diagenetic complexity, such that understanding and predicting reservoir quality (RQ) presents a major challenge to their economic development. Porosity is typically 1–10%, but reaches up to 15% and permeability is typically <10−15 m2 (<1 mD), but locally reaches >10−13 m2 (>100 mD). The key questions addressed herein concern the development and distribution of this RQ variability, specifically why has good RQ been locally preserved?
Primary depositional fabric exerts a strong control on RQ. Muddy sandstones are either highly compacted or pervasively cemented by quartz and microporous illite, and have very poor RQ. Only fine- to medium-grained, moderately well sorted, clean sandstones can contain good RQ, but texturally and mineralogically similar sandstones span a wide range of porosity and permeability. This range is primarily driven by the degree of quartz cementation, with incomplete cementation resulting in the best RQ. Quartz overgrowths in incompletely cemented clean sandstones are patchy and non-luminescent in scanning electron microscopy with cathodoluminescence (SEM-CL), possibly indicating slow growth rates. There is tentative evidence to link incomplete quartz cementation with oil charging of the reservoir. An alternative or additional explanation of RQ preservation may be related to limited silica supply in the centres of the thickest, stacked, clean sandstones, where the better RQ tends to reside.
The results of this study imply that sustained high-energy depositional processes, coupled with an early oil charge, are prerequisites for retaining the best RQ. This has important implications for the exploration and development of Late Ordovician glacial sandstones in the Illizi Basin, and potentially similar plays elsewhere.
Lower Palaeozoic sandstones host important hydrocarbon resources across North Africa and the Middle East (e.g. Le Heron et al. 2009). Sandstones deposited during the Late Ordovician (Hirnantian) glaciation form one of the most important reservoirs. These reservoirs display considerable depositional complexity (e.g. Hirst et al. 2002) and have been subject to more than 400 myr of burial diagenesis. Consequently, there are many reservoir quality (RQ) controls that interact in complex ways and are often difficult to predict; the ability to unravel the various depositional and diagenetic RQ controls is of considerable scientific and economic significance. Previous RQ studies of Ordovician glacial sandstones from North Africa have highlighted the importance of depositional fabrics and clay content (Hirst et al. 2002), syn-sedimentary emplacement of clay coats by pressurized glacial meltwater (Tournier et al. 2010), and eo- and meso-diagenetic cements (Hirst et al. 2002; El-ghali et al. 2006a, b; Tournier et al. 2010).
This paper documents a case study from the In Amenas and Bourarhat Sud area of the Illizi Basin, Algeria (Fig. 1), where Late Ordovician gas-bearing sandstones typically have porosities of <10% and permeabilities of <10−15 m2 (<1 mD). However, sandstones recording porosities up to 15% and permeabilities of >10−13 m2 (>100 mD) exist and have dominated production to date. This study integrates detailed sedimentology, petrography and routine core analysis data with results from other analyses including X-ray diffraction (XRD), scanning electron microscopy (SEM), SEM with cathodoluminescence (SEM-CL) and fluid-inclusion analysis from a large subsurface dataset to address two key questions:
What are the depositional and diagenetic features that resulted in the typically poor RQ?
What depositional and diagenetic features have caused the local preservation of moderate–good RQ?
(a) Lower Palaeozoic stratigraphic column for the Illizi Basin, Algeria (compiled from Galeazzi et al. 2010; Hirst 2012). Strata belonging to the Hirnantian glacial episode are the focus of this study. (b) Palaeogeographical reconstruction of Gondwana at approximately 440 Ma showing the distribution of Hirnantian ice sheets (after Le Heron et al. 2009). (c) Palaeozoic basins and arches of Algeria and Libya, showing the location of the Illizi Basin and In Amenas–Bourarhat Sud study area (after Hirst 2012). T–R cycles, transgressive–regressive cycles
The studies cited above illustrate that subtly different RQ controls appear to have operated in different areas of North Africa. However, it is hoped that the outcome of this study may shed further light on RQ controls in other Lower Palaeozoic reservoirs, both across North Africa and worldwide.
The geological setting of the Late Ordovician glacial deposits of the Illizi Basin, Algeria is first summarized before the various methods employed are reviewed. A large dataset is then interrogated to better understand the likely controls on RQ. Finally, two hypotheses regarding the preservation of RQ in these sandstones are outlined, including hydrocarbon inhibition and silica budgets.
Geological setting
Regional context
The Palaeozoic of North Africa records mainly fluvial and shallow-marine sedimentation on a low-relief passive-margin ramp more than 1000 km wide on the northern edge of the Gondwana supercontinent (Fig. 1) (Scotese & McKerrow 1990; Scotese et al. 1999; Le Heron et al. 2004; Ghienne et al. 2007; Craig et al. 2010; Eschard et al. 2010; Galeazzi et al. 2010). Reactivation of basement lineaments in the Cambrian–Lower Ordovician led to the segregation of the passive-margin ramp into areas of lower and higher subsidence (arches and basins, respectively: Fig. 1). The Illizi Basin was flanked by the Amguid-el Biod High to the west, the Ahara Arch to the north and the Tihemboka Arch to the east (Fig. 1). By the Late Ordovician, when the sediments that are the subject of this study were deposited, subsidence rates equalized across the basins and arches marking a period of tectonic quiescence that extended into Silurian times (Boote et al. 1998; Eschard et al. 2010).
During the Lower Palaeozoic, Gondwana was located over the South Pole, and a short-lived (c. 1–2 myr) glacial episode is recorded at the end of the Ordovician (Hirnantian, c. 445 Ma) (Fig. 1) (Beuf et al. 1971; Hambrey & Harland 1981; Vaslet 1990; Moreau et al. 2005; Saltzman & Young 2005; Ghienne et al. 2007; Le Heron & Dowdeswell 2009). The Late Ordovician ice sheet covered a large part of North Africa and Arabia (McClure 1978; Vaslet 1990; Powell et al. 1994; Ghienne & Deynoux 1998; Senalp & Al Laboun 2000; Ghienne et al. 2007; Le Heron & Dowdeswell 2009), with glacial influence extending as far north as Spain at the glacial maximum (Guitierrez-Marco et al. 2010). Algeria and Libya were situated towards the margin of the North African ice sheet, with glacially transported sediment generally directed north to northwestwards towards the ocean (Beuf et al. 1971, pp. 77, 165–170).
Algeria and Libya are interpreted to have been extensively ice-covered during the ice maximum (Le Heron & Craig 2008). During this time and the subsequent ice retreat, a series of glacial valleys were carved into the pre-glacial semi-consolidated substrate (Hirst et al. 2002; Moreau et al. 2005; Le Heron & Craig 2008; Hirst 2012). The valleys were largely filled during phases of overall ice-sheet retreat by a complex assemblage of glaciogenic and glacially influenced sediments (e.g. Hirst et al. 2002; Hirst 2012; Lang et al. 2012).
Late Ordovician glacial deposits: facies scheme and depositional model
Eleven descriptive lithotypes have been identified from outcrop and core studies (Table 1) (Hirst 2012). They include:
fine- to coarse-grained, bedded and cross-bedded sandstones (Sb and CSx);
unstructured to faintly laminated sandstones with water-escape features (Su and SMu);
heterolithics displaying various portions of Bouma sequences (HS and HM);
laminated mudstones with rare outsized grains (MM and MMg);
very poorly sorted massive to highly deformed sandstones and sandy mudstones with floating outsized grains (SSg, SMg and MSg).
Summary of the lithotypes identified in the subsurface cores used in this study (after Hirst 2012)
These lithotypes have been consistently applied on a cm-scale to the more than 3 km of core in the In Amenas–Bourarhat Sud study area. The distinction of ‘clean’ v. ‘muddy’ sandstones (e.g. Su v. SMu) is made based on the colour of the core (muddy sandstones tend to be a darker grey) and integration with wireline log data (especially gamma ray, where Su is typically <30 g API and SMu >30 g API).
Five facies associations have been identified (Table 2) (Hirst et al. 2002; Hirst 2012). FA1 (fine- to coarse-grained, bedded and cross-bedded sandstones) typically occur towards the base of glacial valley fills (i.e. towards the base of Unit IV: Fig. 1a) and have limited lateral extent at outcrop and in the subsurface (<200 m perpendicular to palaeoflow). They are interpreted as subglacial channel-fills to ice-proximal proglacial outwash fans (Fig. 2) (Hirst et al. 2002; Dixon et al. 2008; Hirst 2012). FA2 (massive to low-angle cross-laminated, dewatered sandstones) are more common towards the top of glacial valley fills and in strata that overtops the valleys to form a regionally extensive blanket (i.e. towards the top of Unit IV: Fig. 1a). The sandstones occur in both laterally extensive (10 s km) amalgamated sheets and channels, up to tens of metres thick, and, more rarely, in sinuous ribbon channels, which are typically 50–400 m wide and 4–12 m thick, capping the glacial stratigraphy (Beuf et al. 1971). The presence of transcritical bedforms and dewatering fabrics distinguish these sandstones from FA1. FA2 is interpreted as the deposits of relatively ice-distal proglacial high-density turbidites (Fig. 2) deposited by glacier meltwater events, possibly related to periodic catastrophic draining of subglacial lakes (Hirst 2012). FA3 (heterolithic deposits with localized deformation) and FA4 (laminated mudstones and siltstones with rare granules) are interpreted as ice-distal deposits comprising the distal expression of meltwater turbidites and hemipelagic sedimentation with iceberg rain-out material (Fig. 2). Finally, FA5 (poorly sorted argillaceous sandstones with granules and deformed fabrics) are interpreted as the deposits of debris flows shed from the unstable ice-margin and subaqueous valley walls (Fig. 2).
Depositional model of pro-glacial outwash-fan-related deposits, Late Ordovician of the Illizi Basin, Algeria. Schematic depositional dip profile axial (a) and off-axis (b) to a pro-glacial outwash fan. (c) Block diagram illustrating a late deglaciation depositional model. Colour scheme represents facies associations, which are summarized in Table 2. Based on the models of Lønne et al. (2001), Powell & Cooper (2002), Russell & Arnott (2003) and Hornung et al. (2007).
Summary of the facies associations identified at outcrop and in the subsurface cores used in this study (compiled following the facies scheme of Hirst 2012)
The proglacial turbidite sandstones are the dominant reservoir facies in the subsurface: hence, understanding their distribution is highly valuable. Their extensive development in the uppermost glacial stratigraphy is thought to be related to stillstands during the final, relatively rapid retreat of the marine-terminating ice sheet. The glacial valley network and the end-glacial ribbon channels both occur as segments, tens of kilometres long, with abrupt beginnings and ends (e.g. Le Heron & Craig 2008; Girard et al. 2012). Mega-regional mapping of the glacial sediment isopach reveals local thickening orientated perpendicular to the ends of these valley trends (Le Heron & Craig 2008). These large-scale patterns are consistent with a phased ice retreat, with the development of grounding lines during periods of ice stillstand (Le Heron & Craig 2008). Grounding-line fans would be expected to form in front of these grounding lines (e.g. Lønne et al. 2001; Powell & Cooper 2002; Dowdeswell et al. 2015), providing a model to predict the presence of reservoirs in the subsurface.
No grounding-zone wedges have been explicitly identified in the outcrops of southern Algeria and Libya; however, candidate grounding-zone wedges have been interpreted from 3D seismic surveys to the north (Decalf et al. 2013). The mapped grounding lines in the subsurface are spaced at approximately 30 km, consistent with the spacing of valley and ribbon-channel segments observed elsewhere.
Methods
With more than 3 km of core available, covering a geographical area of around 5500 km2, there is a large dataset on which to draw. In order to unravel the controls on reservoir quality, the detailed sedimentological descriptions outlined above have been integrated with:
routine core analysis (RCA) from 3519 core plugs from 36 wells;
detailed qualitative and quantitative petrography on 235 thin sections from 25 wells;
X-ray diffraction (XRD);
scanning-electron microscopy (SEM);
scanning electron microscopy with cathodoluminescence (SEM-CL);
fluid-inclusion data.
Routine core analysis (RCA)
The RCA database is a combination of data acquired using a CMS-300 (Core Measurement System) machine since 2008 and historical data (from the 1970s to 2000s) collected using older machinery. The majority of the data used here (c. 70%) were collected using the CMS-300: an automated computer-directed, unsteady state pressure decay permeameter and porosimeter with a sleeve pressure of 5.5 MPa (800 psi). Owing to the different vintages of equipment used to collect the data presented here, and the lack of measurements at different effective stresses for all of these machines, no overburden correction has been applied and measurements should be treated as unstressed. Porosities and permeabilities, especially at the lower end of the scale, will be lower at reservoir conditions. However, unstressed data are sufficient for the purpose of this study: that is, identifying causes in the variability of RQ rather than highly quantitative petrophysical rock typing, for example.
The RCA database was first quality controlled to remove fractured plugs and low-permeability samples analysed with older machinery that was unable to resolve permeabilities below about 10−16 m2 (0.1 mD). This left 3519 plugs with useable data. The sedimentary lithotypes were interpolated onto the RCA data, together with other pertinent information such as true stratigraphic thickness below top reservoir seal.
Petrography
The petrographical dataset comprises textural analysis and point-count data (200 mineral point counts + porosity) from 25 wells, with 235 out of a total of 342 thin sections sampling the Late Ordovician glacial stratigraphy. The remainder of the thin sections sample the pre-glacial Cambrian–Ordovician stratigraphy and the Late Ordovician–Early Silurian post-glacial stratigraphy. Grain size was categorized on the Udden–Wentworth grain-size scale with Ø upper and lower divisions. Sorting was calculated as the standard deviation from the mean length (expressed as Ø), and is categorized according to the scheme of Folk & Ward (1957).
Porosity measurements derived from RCA are unable to distinguish macroporosity from microporosity: however, these different pore architectures have an important impact on permeability. An attempt has been made to quantify macroporosity v. microporosity in all of the petrographical samples. Macroporosity is derived from modal analysis and microporosity is calculated as: [measured routine core analysis porosity]–[point counted total macroporosity].
Petrographical data have been collected by several different petrographers. Therefore, caution must be exercised when integrating all of the data, as there is likely to be a degree of operator bias. Of the 235 thin sections, 212 have accompanying porosity–permeability data. A review of the petrography samples shows that they cover the range of porosity and permeability in the larger RCA database (Fig. 3f). This adds confidence to conclusions drawn from the petrography database regarding RQ controls. There are detailed proprietary petrographical reports for four of the wells and two proprietary overview reports covering 15 of the remaining wells. The detailed qualitative observations included in these reports have been instrumental in understanding the RQ controls.
(a) Porosity–permeability cross-plot for the cores used in this study, shaded by lithotype (see Table 1). The data fall into two broad trends: a higher gradient trend comprising lithotypes CSx, Sb and Su (b & c); and a lower gradient comprising lithotypes SMu, HS, SSg, SMg and MSg (d & e). (f) Petrographical samples with RCA data cover a representative range of lithotypes and porosity–permeability compared to the larger RCA dataset. The locations of four samples chosen to illustrate various controls on RQ in Figures 7 and 11 are shown.
X-ray diffraction (XRD)
Whole-rock and <2 µm clay fraction XRD data has been collected using standard practices (Appendix 1) from 89 of the petrographical samples, covering a range of lithotypes (NB SSg, SMg and MSg lithotypes are not represented), and has assisted in the identification of the major mineral phases present.
Scanning electron microscopy (SEM)
Backscattered and secondary electron SEM imaging of rough surfaces was used primarily to investigate the morphologies, compositions, mineralogy and paragenetic relationships of the clay minerals. SEM and SEM-CL imaging of polished thin sections was primarily used for detailed analysis of quartz cements and compaction, but was also employed for observation of the clays and their paragenetic relationships.
Fluid inclusions
Fluid inclusions were first identified and characterized using routine petrographical techniques in plane polarized and ultraviolet (UV) light to determine a primary or secondary origin, and to define an aqueous or hydrocarbon content. The fluid inclusions were almost exclusively aqueous, with rare hydrocarbon-filled examples, and of primary origin: trapped in quartz either at grain–cement boundaries (dust rims) or within the main bodies of the quartz overgrowths (typically along crystal-growth-zonation boundaries). Rare secondary aqueous and hydrocarbon fluid inclusions were observed: trapped along microfractures within grains of detrital quartz. Ultraviolet epifluorescence was used to identify and differentiate different types of hydrocarbon based on the colour of any fluorescence. Microthermometric analysis was carried out to determine temperatures and salinities of the pore-filling fluid phase at the time of entrapment (Appendix 1).
Results
Plots of porosity v. permeability shaded by lithotype show a strong relationship between lithotype and RQ (Fig. 3), highlighting the importance of depositional textures and sedimentary structures on RQ. Two broad trends are observed, both of which are well represented in the petrographical dataset:
A ‘high-gradient’ porosity-permeability trend defined by lithotypes CSx, Sb and Su (defined in Table 1). These are all clean to very clean sandstones (light grey in core, low gamma-ray log values).
A ‘low-gradient’ porosity-permeability trend defined by lithotypes SMu, SSg, SMg, HS and HM (defined in Table 1). These are muddier sandstones (darker grey in core, higher gamma-ray log values).
There is a broad range in porosity and permeability for each lithotype. Therefore, although lithotype controls which porosity–permeability trend a given rock will occur on, additional factors control the position on that trend.
The following subsections provide systematic descriptions of the detrital and diagenetic characteristics of the sediments, with the objective of better understanding the porosity–permeability cross-plot (i.e. RQ controls).
Mineralogy and sandstone classification
Petrographical point-count data and XRD analysis illustrate that, in general, these are highly chemically mature sediments (Fig. 4). The detrital mineralogy is typically dominated by monocrystalline quartz, with only minor lithic material, feldspar and detrital clays. The lithic assemblage is dominated by polycrystalline quartz with lesser amounts of typically highly altered grains of unidentifiable original composition. There has been some minor (<1% by volume) secondary porosity generation through feldspar dissolution: however, depositional abundances of plagioclase and K-feldspar are considered to have typically been very low. Given the low volumes of porosity generated, the causes of this dissolution have not been explored in more detail. Detrital clay is relatively scarce, and has been partially altered to authigenic illite and, more locally, chlorite. Illite and chlorite, and rarely kaolinite, also occur as discrete authigenic cements and grain-replacive components, which makes it difficult to assess original detrital clay content.
(a) Sandstone classification for Late Ordovician sandstones derived from point-count data. All of the lithotypes plot in the same area of the quartz–feldspar–lithics diagram and are strongly quartz-dominated with negligible feldspar. (b) Summary of whole-rock XRD data categorized by lithotype. ‘Clay’ includes illite + illite–smectite, kaolinite and chlorite. ‘Other’ typically includes minor calcite, siderite, dolomite and K-feldspar. These data reinforce the quartz-dominated nature of the deposits, but illustrate that clay contents are subtly higher in lithotypes SMu and HS compared to lithotypes CSx, Sb and Su.
Sandstones classify as either quartzose, sublithic or lithic sandstones (Fig. 4a). All of the lithotype groupings occupy broadly the same space on the sandstone classification diagram (Fig. 4a), suggesting similar derivation. This is consistent with erosion and mixing of the relatively thick (>200 m) underlying pre-glacial Cambro-Ordovician stratigraphy. In places, Late Ordovician syn-glacial sandstones directly overlay crystalline basement, proving that glacial erosion did reach basement locally (e.g. Hirst 2012). This may be a source for the more chemically stable lithics (e.g. polycrystalline quartz) observed in thin sections (local basement typically comprises meta-granite in the study area).
XRD data confirms that all of the sandstone lithotypes are dominated by quartz (from 49 to 100% bulk XRD, with a mean of 95%: Fig. 4b). Lithotypes CSx, Sb and especially Su have the lowest total clay content (mean <2% bulk XRD), followed by SMu (mean 4.4% bulk XRD), then HS (mean 13.5% bulk XRD: Fig. 4). All other identified minerals (most commonly K-feldspar, plagioclase, calcite, siderite, dolomite, ferroan dolomite, pyrite, barite and muscovite and minor anatase, analcime, sphalerite, halite, hornblende, and marcasite) are very rare (mean <1% bulk XRD) in CSx, Sb and Su lithotypes, becoming slightly more common, although still rare, in the muddier lithotypes (Fig. 4). These data corroborate the more qualitative metrics used to define the lithotypes (core colour and gamma-ray log values: Table 1).
The <2 µm clay fraction is dominated by illite (45–100% clay fraction XRD with a mean of 80%) and mixed layer illite + illite–smectite (0–23% clay fraction XRD, with a mean of 11%). Chlorite is typically minor by comparison (0–33% clay fraction XRD, with a mean of 6%). The dominance of illite is interpreted to reflect both the cold-climate weathering origin of the detrital clays and extensive recrystallization during prolonged diagenesis (e.g. Worden & Morad 2003).
Grain size and sorting
There is a broad relationship between grain size, sorting and lithotype (Fig. 5). Lithotypes SMu and HS tend to have the finest grain sizes and are the most well sorted. Lithotypes Su, Sb and CSx overlap the SMu and HS trend, but extend to coarser grain sizes and poorer sorting. Lithotypes SSg and SMg tend to have poorer sorting for a given grain size compared with the other lithotypes. This is consistent with the core-based observation of scattered outsized granules in SSg and SMg lithotypes that are absent in lithotypes Su, SMu and HS. This difference in grain size and sorting suggest that the debris flows (SSg, SMg lithotypes) were unlikely to have evolved into the turbidity currents (Su, SMu and HS lithotypes) inasmuch as high-energy turbidity currents should be able to entrain the granule-grade material if it were available (Hirst 2012).
Cross-plots of grain size v. sorting shaded by (a) lithotype (see Table 1); (b) porosity; and (c) permeability. The data illustrate that lithotypes SMu and HS (muddy turbidites) tend to be the finest-grained, most-sorted sandstones. Lithotypes SSg, SMg and MSg (debrites) are the least-sorted deposits. Lithotypes CSx, Sb and Su (tractional deposits and clean turbidites) have intermediate grain size and sorting. There is a high degree of scatter in the grain size and sorting v. porosity and permeability data, but, generally, the finest-grained, most-well-sorted sandstones have the lowest porosity and permeability. Very-fine- to medium-grained, moderately well-sorted sandstones retain the best porosity and permeability, but can also have very low porosity and permeability.
Permeability and, especially, porosity relationships with grain size and sorting are more difficult to discern, as might have been expected from the porosity–permeability cross-plot. However, it appears that the finest grained, most-sorted sandstones tend to have the worst permeabilities. The coarsest-grained, least-sorted sandstones have moderate permeabilities, and the highest permeabilities are associated with sandstones of intermediate grain size (medium-grained) and sorting (moderately well sorted to moderately sorted: Fig. 5). However, medium-grained, moderately well-sorted to moderately sorted sandstones can also have low porosities and permeabilities. This indicates that, in spite of diagenetic overprinting exerting a very strong control on RQ, an underlying primary depositional texture control is still locally evident.
Paragenetic sequence and burial history
The main diagenetic degraders of RQ in these rocks are strong compaction together with quartz and illite cementation (Fig. 6a). Other diagenetic minerals such as kaolinite, siderite and chlorite are present in lesser abundance. Poor RQ is to be expected considering the burial history of these reservoirs (Fig. 6b). There is considerable uncertainty surrounding the burial history of the Ordovician reservoirs since Carboniferous rocks currently crop out at the surface, leaving an approximately 300 Ma unconformity. However, the preferred burial history model takes into account the major tectonic events and is calibrated using fluid-inclusion data. Although burial history modelling suggests that the reservoirs have not been especially deeply buried (maximum burial depth of c. 2.5–3 km), they have been at moderate burial depths from the Carboniferous (c. 330–320 Ma) to present day. Modelled sustained effective stress of approximately 5500–6500 psi is consistent with strong compaction, and modelled sustained temperatures of >90°C is consistent with strong quartz cementation. Indeed, a more pertinent question might be why do these reservoirs have any porosity and permeability remaining at all? Before returning to this question, the factors impacting compaction and cementation are examined using the datasets available.
(a) Generalized paragenetic sequence based on thin-section observations. (b) Typical burial history plot for the study area. Modelled temperatures are denoted by the shaded fill, with the ‘quartz cementation window’ (i.e. above 80°C) marked by diagonal white hatching. Isolines of modelled effective stress are shown in dashed white lines. Note that the reservoir (the top is marked by the bold black line) has been at elevated temperatures (inside the ‘quartz cementation window’) and at moderately high effective stress for a long period of time (>300 myr).
Ductile content (detrital and authigenic clay): impact on compaction diagenesis
A variety of datasets suggests a strong relationship between ductile content and RQ. Muddier lithotypes (SMu and HS: as defined by core observations and gamma-ray logs, and corroborated by XRD and modal analysis data) have poorer RQ (Fig. 3). Cleaner lithotypes (Su, Sb and CSx) can have better RQ (Fig. 3).
Two principal habits of clay are observed in thin sections: pore-filling detrital and authigenic clay (Fig. 7); and grain-rimming detrital/recrystallized clay (Fig. 8). XRD data confirm that the dominant clay mineral is illite. Other clay minerals are rare, but can include kaolinite and chlorite. Pore-filling, finely crystalline, loosely packed illite cements with significant intercrystalline microporosity are relatively common in the muddier sandstones (Fig. 7b).
Thin-section photomicrographs of pore-filling clay habits. (a) The relatively high clay content (24%) in this sandstone has resulted in strong compaction, as shown by the tightly compressed matrix clay (i), together with numerous elongate, sutured and stylotized (ii) grain contacts. No macroporosity has been preserved and only minor (3.1%) microporosity is present. Consequently, porosity (5.4%) and permeability (9 × 10−18 m2 (0.009 mD)) is negligible (Fig. 3f). (b) The slightly lower clay content (14%) in this sandstone comprises finely crystalline, loosely packed, webby to fibrous authigenic illite (ii). The presence of significant intercrystalline microporosity (7.8%) leads to a green tinge in plane polarized light (PLL) due to partial resin impregnation. The illite cements are best viewed in polished and rough surface backscattered electron microscopy (BSEM) images shown to the top and bottom right of the PPL image, respectively. Although the porosity is higher (9%), the permeability (7 × 10−14 m2 (0.07 mD)) remains negligible (Fig. 3f).
Thin-section photomicrographs illustrating the impact of grain-coating clays on compaction and quartz cementation. (a) A clean, moderately sorted, coarse-grained sandstone that has experienced pronounced compaction, with highly elongated, often interpenetrating grain contacts (i). Quartz cement is relatively minor (8%), and is dominated by a single generation forming highly discontinuous, yet thick, pore-filling overgrowths and outgrowths (ii), post-dating pronounced grain-contact elongation. Earlier, thin, non-luminescent quartz overgrowths are only rarely observed. Although the sample only has 2.5% grain-coating clay, these coats are generally continuous and typically from 1 µm thick (iii) to 20 µm thick (iv). They appear to have inhibited earlier quartz overgrowths, which in turn allowed for strong compaction. Strong compaction means that porosity (2.8%) and permeability (7 × 10−14 m2 (0.07 mD)) is negligible. (b) A clean, well-sorted, fine-grained sandstone that has preserved a very high intergranular volume (32.5%). Compaction is weak, with only minor, patchy grain-contact elongation observed (i). Quartz cement is abundant (22%), with two generations identifiable. The earliest generation is volumetrically minor, forming non-luminescent, fairly continuous, thin (<15 µm thick) syntaxial overgrowths (ii), which coat the surface of most quartz grains. The later generation is volumetrically dominant, forming thick, pore-filling overgrowths, which pervasively occlude the primary pore network (iii). The sample has minor grain-rimming detrital clay (3%) that forms very thin (<1 µm thick), discontinuous coats around most detrital quartz grains (iv). These clay rims do not appear to have inhibited an early phase of quartz cement, which stabilized the grain framework, and inhibited compaction. Nonetheless, pervasive quartz cementation has resulted in negligible porosity (2.8%) and permeability (6 × 10−18 m2 (0.006 mD)).
The most ductile-rich samples (>15% total ductiles: lithotypes SMu, HS, SSg and SMg) have no macropores, moderate–negligible microporosity that decreases with increasing ductile content (Fig. 9), and low–negligible permeability (<2 × 10−16 m2 (<0.2 mD)). Thin-section observations illustrate that these samples are strongly compacted, facilitated by the high ductile content. Ductile clays have deformed between rigid quartz framework grains destroying all macroporosity and preventing the precipitation of microporous authigenic illite in significant quantities (Fig. 7a).
Cross-plots from petrographical samples illustrating the impact of detrital ductiles and authigenic clays on RQ: (a) detrital ductiles and authigenic clay v. macroporosity from modal analysis; (b) detrital ductiles and authigenic clay v. calculated microporosity; and (c) detrital ductiles and authigenic clay v. permeability from RCA. Only the cleanest sandstones have significant macroporosity and permeability remaining. However, equally clean sandstones can have negligible macroporosity and permeability. Intermediate ductile and clay content are associated with microporosity-dominated systems with negligible permeability. At higher ductile and clay content, not even microporosity is preserved.
Sandstones with 5–15% total ductiles (lithotype SMu) have limited–no macroporosity and are typically dominated by microporosity (Fig. 9), with microporosity increasing as total ductile content increases. Rare macropore-dominated samples can have good permeability (up to 10−13 m2 (100 mD)), but micropore-dominated samples have poor–negligible permeability (<10−15 m2 (<1 mD): Fig. 9). This suggests that the microporosity does not provide an effective flow path, even for gas. Stylolites are often observed at the core to thin-section scale and a strongly compacted fabric is observed in thin sections. Thin-section observations suggest that slightly lower initial detrital clay volumes, including less-well-developed grain-rimming clays (see below), relative to the most ductile-rich sandstones resulted in a slightly more open fabric post-compaction, allowing the precipitation of microporous illite cements (Fig. 7b).
The cleanest sandstones (<5% total ductiles: lithotypes Su, Sb and CSx) tend to be dominated by macropores, with lower microporosity (Fig. 9), and these are the only sandstones where good permeabilities are encountered (>10−13 m2 (>100 mD)). However, even in these sandstones, permeability can range from >10−13 to <10−17 m2 (from >100 to <0.01 mD: Fig. 9).
Quartz cements
Quartz cements are typically the volumetrically dominant authigenic phase in these sandstones and are responsible for occluding much of the post-compaction porosity.
Cross-plotting total quartz cement against other variables, such as permeability, can give a distorted view on the importance of quartz cements on RQ as the absolute abundance of quartz cement is limited by the amount of space available for precipitation. For example, the muddier sandstones may have lower volumes of quartz cement than the cleaner sandstones because there was limited space for quartz cements to precipitate following strong compaction and/or authigenic illite precipitation. Making the simplification that the majority of the quartz cements were the last authigenic phase to precipitate, and thus post-date the main phase of compaction, the degree of quartz cementation can be calculated from the point-count petrographical data: [(present-day macroporosity + total quartz cement)/total quartz cement]. There is a reasonably strong negative correlation between degree of quartz cementation and permeability; higher permeabilities are associated with less complete quartz cementation (Fig. 10). The factors controlling the degree of quartz cementation are clearly complex and challenging to consistently relate back to depositional fabrics (lithotypes, grain size, sorting and ductile content: Fig. 10). A deeper understanding of the variability in quartz cementation was achieved through the use of SEM-CL to reveal multiple quartz-cement generations and their habit, not consistently differentiable using optical microscopy.
(a) Degree of quartz cementation v. core permeability; shaded by (b) depositional lithotype; (c) grain size; and (d) total ductile content. Degree of quartz cementation is calculated as: [total quartz cement/(total macroporosity + total quartz cement)]. Permeability decreases as the degree of quartz cementation increases. The most quartz-cemented sandstones tend to be fine-grained and argillaceous, reflecting both a large surface area for quartz-cement nucleation, and limited pore space available post-compaction and illite cementation. The least quartz-cemented sandstones tend to be medium- to coarse-grained and less argillaceous, reflecting both a smaller surface area for quartz-cement nucleation, and more pore space available due to less compaction and less illite cementation. However, notable exceptions to these generalizations exist; medium- to coarse-grained, less argillaceous sandstones can also be strongly quartz-cemented with very poor permeability.
An earlier generation of quartz is overlain by kaolinite, illite, siderite, residual oil and, locally, barite. It forms volumetrically scarce–minor (<3%: based on point-counting of SEM-CL images from selected samples), thin, variably continuous, non-luminescent (dark) overgrowths, which post-date minor–moderate grain-contact elongation. These overgrowths tend to be less well developed in highly compacted samples, but are thicker and more continuous in less-compacted samples. Furthermore, the early overgrowths appear to form more readily in exceptionally clean sandstones with thin, discontinuous clay rims around detrital quartz grains (Fig. 8b). Conversely, thicker, more continuous clay rims appear to have inhibited the early phase of quartz cement (Fig. 8a).
A later generation of quartz is inhibited by, and also engulfs, kaolinite, illite, siderite, residual oil and, locally, barite (Fig. 6). In pervasively quartz-cemented sandstones, this later, dominant phase typically forms thick, well-developed, anhedral, pore-filling overgrowths that occlude the remnant primary pore networks. The overgrowths are typically moderately luminescent and can show growth zonation (Fig. 11a). These moderately luminescent, zoned overgrowths are absent in sandstones with the highest permeabilities, which instead display patchy, non-luminescent overgrowths that, at least in part, post-date significant compaction (Fig. 11b). This may suggest slower quartz-cement precipitation rates in these sandstones, leading to there being fewer of the defects in the crystal lattice that give rise to luminescence in quartz (Götze et al. 2001). Given the degree of compaction, the non-luminescent overgrowths are interpreted to represent an amalgamation of the earlier and later generations of quartz cements. In other words, cementation continued in the incompletely cemented sandstones, possibly synchronously with the moderately luminescent, zoned cements in the pervasively cemented sandstones, but at a much slower rate.
Plane polarized light (PPL) and SEM-CL photomicrographs illustrating typical quartz-cement habits in strongly cemented v. incompletely cemented sandstones. Pores are shaded white in the PPL and SEM-CL images. (a) A clean, well-sorted, fine-grained sandstone that has been pervasively quartz-cemented (24.5%), with two distinct generations. The earlier is thinly developed and non-luminescent (i). The later phase is more abundant, well-developed (thick and often continuous/interlocking), moderately luminescent and with zoned overgrowths (ii). The pervasive quartz cementation has resulted in negligible porosity (2.1%) and permeability (10−17 m2 (0.01 mD)) (Fig. 3f). (b) A clean, moderately well-sorted, medium-grained sandstone that has been moderately quartz-cemented (9%). The quartz cement comprises patchy, euhedral, non-luminescent (dark) overgrowths (iii) that, at least in part, post-date significant compaction/grain interpenetration (note the stylolitized grain contacts (iv). This sample has good porosity (9.7%) and permeability (1.8 × 10−13 m2 (183.3 mD): Fig. 3f) due to the absence of the moderately luminescent quartz seen in (a).
Summary: pore evolution pathways
From all of the observations and interpretations given above, four main pore evolution pathways can be identified based on precursor sediment and major diagenetic processes (Fig. 12).
Pore evolution diagram summarizing the main controls on RQ. Four end members can be recognized: very argillaceous, fine-grained, well-sorted sandstones are strongly compacted; argillaceous, fine-grained, well-sorted sandstones are strongly compacted with the remaining pore space filled by authigenic illite and quartz; clean, medium-grained, moderately well-sorted sandstones with pervasive quartz cements; and clean, medium-grained, moderately well-sorted sandstones with patchy, incomplete quartz cementation.
Very argillaceous, fine-grained, well-sorted sandstones (SMg & SMu)
Fine-grained sandstones both with (SMg) and without (SMu) outsized grains are prone to very strong compaction, especially where detrital ductile content increases above 15%. Ductile grains and clays are deformed and totally occlude the pore space as the sediment is buried. Macropores are not present and micropores are typically minor; RQ is negligible.
Argillaceous, fine-grained, well-sorted sandstone (SSg & SMu)
Typically, fine-grained, well-sorted sandstones (SMu) with 5–15% detrital ductile content are also strongly compacted. Well-developed clay rims around detrital quartz grains prevented an early phase of quartz cement rendering the sandstone more prone to later compaction. However, some post-compaction pore space was available for the precipitation of quartz cements and microporous illite. Macropores are rare and microporosity dominates; RQ is poor, being composed of poorly connected microporosity.
Clean, medium-grained, moderately well-sorted sandstone (Su, Sb & CSx) with pervasive quartz cementation
Very clean (<5% detrital ductiles), medium- to coarse-grained, moderately well-sorted to moderately sorted sandstones (Su, Sb & CSx) are compacted, but often not as strongly as muddier sandstones. Thin, discontinuous clay rims around detrital quartz grains allowed an early phase of quartz cement to precipitate, which appears to have held open the pore network during initial compaction. However, protracted, moderately deep burial (Fig. 6) resulted in pervasive quartz cementation. Both macropores and micropores are rare, and RQ is very poor in these highly cemented sandstones.
Clean, medium-grained, moderately well-sorted sandstones (Su, Sb & CSx) with incomplete quartz cementation
These sandstones appear to have the same fabric and early diagenetic history as the pervasively quartz-cemented very clean sandstones above. However, in this case, some macroporosity has been preserved due to incomplete quartz cementation. These sandstones have the best RQ (up to 15% porosity and 2 × 10−13 m2 (200 mD) permeability). SEM-CL observations suggest that the quartz cements that do exist were precipitated slowly.
The primary control on RQ is, therefore, depositional fabric: most importantly, the detrital ductile content. The muddy sandstones always exhibit poor RQ, due either to very strong compaction or to strong compaction followed by illite and quartz cementation. The clean sandstones have typically preserved more porosity through the compaction phase of diagenesis, but pervasive quartz cementation has often occluded this porosity. Yet, against expectations from burial history modelling (Fig. 6b), some clean sandstones have preserved good RQ due to incomplete quartz cementation. The degree of quartz cementation in these cleanest sandstones does not appear to be related to grain size or to total ductile content. SEM-CL observations suggest that the highest permeability sandstones have patchy, non-luminescent overgrowths. Understanding the preservation of good RQ, therefore, becomes a question of what is preventing pervasive quartz cementation by slowing down or stopping the precipitation of quartz cements.
Discussion: preservation of RQ in the cleanest sandstones
The rate of quartz cementation is a function of the rate of silica supply from various sources, the rate of diffusion from the source to the cementation site and the rate of precipitation at the cementation site (e.g. Walderhaug 1996). Three theories to explain the preservation of RQ in the cleanest sandstones are investigated here: surface area available for quartz cementation; hydrocarbon inhibition of quartz cementation; and reservoir-architecture control on quartz cementation.
Surface area available for quartz cementation
Surface area available for cementation is suggested to be an important control on quartz cementation (e.g. Walderhaug 1996). Finer-grained sandstones have a larger surface area for quartz cements to nucleate and so are expected to be more cemented than coarser-grained sandstones, all else being equal and if precipitation rate is the rate-controlling step (Walderhaug 1996). This is broadly corroborated by the data in this study: the finer-grained, more detrital-clay-rich sandstones are typically pervasively quartz-cemented and have poor RQ (Fig. 10b–d). Whilst the absolute volume of quartz cement in these sandstones may be less than in some of the cleaner sandstones, this reflects lower pre-quartz-cement porosity in the muddier sandstones (i.e. there was less space available for quartz cementation). As a function of space available, they are more strongly quartz-cemented. Conversely, in terms of space availability, the least quartz-cemented sandstones tend to be medium- to coarse-grained and have a low–very low ductile content (Fig. 10b–d). This can be explained by less compaction, which itself might be a function of discontinuous clay coats and the precipitation of a volumetrically minor early quartz cement that held the pore space open during compaction (Fig. 8b), and less illite cementation. However, there are almost as many exceptions to this rule as there are applications. Although some of the strongly quartz-cemented clean sandstones may be explained by their smaller grain size, there are also medium- to coarse-grained, clean sandstones that have been strongly quartz-cemented (Fig. 10b–d).
Surface area available for quartz cementation is also controlled by the presence and distribution of grain coats, such as clays or microcrystalline quartz (e.g. Heald & Larese 1974; Pittman et al. 1992; Aase et al. 1996; Walderhaug 1996; Bloch et al. 2002; Taylor et al. 2010). No microcrystalline quartz has been observed in these sandstones, consistent with a lack of volcanic glass, tuffs and sponge spicules or interbedded radiolarian-bearing mudstones, which tend to be precursors to microcrystalline quartz (e.g. Vagle et al. 1994; Taylor et al. 2010; Weibel et al. 2010). In this study, the samples with the thickest, most continuous clay coats are typically highly compacted with poor reservoir quality (Fig. 8). Thin, discontinuous clay rims in the cleanest sandstones appear to have allowed the precipitation of an early phase of quartz cement that resisted strong compaction. Presumably, these early overgrowths would provide ideal nucleation sites for the main phase of quartz cementation. In other words, discontinuous clay coats appear to be required to preserve some porosity pre-main-phase quartz cementation. However, these sandstones have a high risk of subsequent pervasive quartz cementation.
Surface area available for quartz cementation therefore appears to explain the preservation of good RQ in some, but not all, medium-grained, moderately sorted, clean sandstones.
Hydrocarbon inhibition of quartz cementation
Hydrocarbons may be able to inhibit (prevent or slow down) the transport of aqueous silica and the precipitation of quartz cement by lowering the water saturation in a reservoir and preventing aqueous silica reaching quartz grain surfaces (e.g. Worden et al. 1998). This is a contentious concept and has been the subject of debate since the 1990s (e.g. the summary in Worden & Morad 2000). In oil-wet sandstones, quartz cementation could theoretically be halted by preventing aqueous silica reaching the quartz-grain surfaces. However, sandstone reservoirs are often water-wet and a thin film of water remains around quartz grains, meaning that quartz cementation could continue, possibly at a reduced rate.
In this study of gas reservoirs, residual oil is observed in thin sections typically as oil-impregnated illite fibres and as very thin (<1 µm thick) discontinuous drapes on the surfaces of framework grains and, rarely, incompletely developed quartz cements (Fig. 13). Rare hydrocarbon-bearing primary and secondary fluid inclusions are also observed (Fig. 13c). All samples have been cleaned in toluene before RCA and thin-section preparation: therefore, further residual oil may well have been removed, leading to an underestimation of the residual oil. There appears to be a relationship between permeability preservation and the presence of residual oil in very clean sandstones (<5% total ductiles: Fig. 13). Seventy per cent of very clean sandstones with permeabilities >10−15 m2 (>1 mD) contain at least traces of residual oil, which has been identified in thin section. Conversely, residual oil has only been identified in 30% of very clean sandstones with <10−15 m2 (<1 mD) permeability (Fig. 13).
Observations of residual oil and possible implications on RQ. (a) Loosely packed, finely crystalline, wispy, pore-lining to pore-bridging illite fibres (i), engulfed by thick, pore-filling quartz overgrowths (ii). The illite is locally coated in residual oil and appears brownish to black in colour (iii). (b) Thin, discontinuous films of residual oil (iv) locally draping well-developed quartz overgrowths (v). (c) Hydrocarbon-filled fluid inclusion seen within authigenic quartz; note the strong blue fluorescence in ultraviolet light. (d) Cross-plot of permeability v. depth below the Silurian Hot Shale top seal (Fig. 1). Only the cleanest sandstones (<5% total ductiles) are plotted and they are shaded by the presence of residual oil as derived from thin-section observations. Data points are sized according to the distance to the nearest macroscopically defined likely silica source: shales, stylolites and argillaceous sandstones.
Fluid-inclusion data can be used to tentatively place the timing of hydrocarbon charge relative to quartz cementation (Fig. 14). Fluid inclusions are very rare and small in the early quartz cements and dominantly occur in the later phase of quartz cement. Hydrocarbon fluid inclusions have only been observed in about the top 20 m of the reservoir beneath the Silurian Hot Shale top seal. Aqueous primary fluid inclusions suggest that quartz-cement growth commenced at around 80°C, which is consistent with models of quartz kinetics (e.g. Walderhaug 1996), and, according to the preferred burial history, was attained by the formation during the Carboniferous (Fig. 14). The majority of the aqueous fluid inclusions span the temperature range 120–140°C, attained by the formation post-Carboniferous (c. 300 Ma). Homogenization temperatures of fluid and gas phases from primary and secondary aqueous inclusions cogenetic with hydrocarbon inclusions indicate oil emplacement above about 120–125°C throughout the sample set, up to approximately 155°C (Fig. 14). Only eight of the measured 146 hydrocarbon fluid inclusions imply emplacement below 120°C. The hydrocarbon fluid inclusions exhibit yellow-green–blue-green fluorescence and have spectrometrically determined API gravities of 26–43°. Minor populations of wet-gas inclusions were found and possible dry-gas inclusions were observed in the outer part of some quartz overgrowths. Raman spectrometry indicates that the wet-gas contains only hydrocarbons, and has a composition of approximately 67 mol% CH4, 19 mol% C2H6 and 14 mol% C3H8+.
Thermal history (derived from burial history: Fig. 6b) plotted alongside quartz-cement homogenization temperatures from fluid inclusions. Quartz cementation commenced at about 80°C, consistent with kinetic models (e.g. Walderhaug 1996), and, according to the preferred burial and thermal history model, occurred during the early Carboniferous. Significant hydrocarbon fluid inclusions are not noted until significantly higher homogenization temperatures (c. 120–140°C). This is consistent with direct downward migration of hydrocarbons from the Silurian Hot Shale source rock immediately above the reservoir. It also implies that, although hydrocarbons were present in the pore system at the time of quartz cementation, they did not arrive in significant quantities until after quartz cementation had commenced.
These data have several implications. First, it seems probable that these reservoirs were oil charged before a later gas charge displaced the oil and/or the oil was cracked to gas in the reservoir, leaving the observed gas charge. Second, the inferred modal hydrocarbon (light oil) emplacement temperatures are consistent with higher temperatures that the reservoir would have attained at the time the immediately overlying Silurian Hot Shale was oil mature. This supports a component of direct downward migration as the main charging mechanism, as opposed to long-distance lateral migration. Third, it suggests that quartz cement had already started to precipitate before significant volumes of light oil reached the reservoir (Fig. 14). Given the sustained moderately high thermal stress that the reservoir has been subjected to and the current gas charge, the likely source of the residual bitumen rarely observed impregnating illite fibres and coating quartz overgrowths (Fig. 13a, b) is through thermal cracking of the earlier oil charge or possibly gas-induced asphaltene flocculation, as opposed to lower temperature biodegradation (e.g. see Rogers et al. 1974; Blanc & Connan 1994). Clearly, this late-stage development of grain- and cement-coating residual oil relative to the onset of quartz cementation impacts the extent to which oil can have inhibited quartz cementation by preventing aqueous silica reaching grain and cement surfaces. Furthermore, the mere presence of, albeit very rare, hydrocarbon fluid inclusions implies that the initial oil charge failed to completely stop quartz cementation by lowering the water saturation. However, both factors could possibly have slowed the rate of precipitation of quartz cement in the later stages of burial history. This is consistent with the apparent relationship between the presence of residual oil and preservation of permeability (Fig. 13d), and the petrographical observation that the least-cemented clean quartz sandstones have non-luminescent, presumably slowly precipitated, quartz cements (Fig. 11). The present-day gas charge may be continuing to inhibit quartz cementation in the remaining macroporous sandstones that otherwise remain in the quartz-cementation temperature window (at c. 100°C).
Reservoir architecture and silica budgets
Aqueous silica in quartz-cemented sandstones may be sourced internally (relatively locally) or externally (basin-scale convection currents). The latter would appear to be less likely given that ‘the abundant evidence that quartz cementation typically takes place at advanced burial depths and high temperatures where flow of meteorically derived groundwaters does not take place’ (Walderhaug 1996, p. 732). Internal silica sources include:
dissolution of quartz grains at grain contacts and stylolites due to pressure solution (chemical compaction), which can be clay- or mica-catalysed (e.g. Walderhaug 1996);
clay reactions (e.g. smectite to chlorite or illite; kaolinite to chlorite or illite: Worden & Morad 2003);
feldspar reactions (e.g. K-feldspar to illite: Worden & Morad 2003);
biogenic silica (e.g. sponge spicules or radiolarians; Vagle et al. 1994; Weibel et al. 2010);
volcanic glass and tuffs (e.g. Taylor et al. 2010).
No volcanic glass, tuff, sponge spicules or radiolarians have been observed in any of the studied thin sections: hence, a volcanogenic or biogenic source of silica is deemed unlikely in these Late Ordovician glacial sandstones. Similarly, negligible feldspars (Fig. 4) or secondary porosity from dissolved feldspars have been observed, suggesting an insignificant contribution of silica from feldspar reactions. Conversely, the interpreted thermal history (Fig. 6b) and observation of illite as the dominant clay mineral (Fig. 7) suggests that diagenetic clay reactions have taken place, providing a potential silica source. Quartz-grain dissolution features are relatively common at the core- and thin-section-scale, including stylolites, microstylolites, grain-contact elongation and suturing, and grain interpenetration. These features are products of the sustained mechanical and thermal stresses that these sandstones have encountered, and are deemed a significant source of the quartz cements observed. Interpenetrative grain contacts and stylolites are observed to be more common in the muddier sandstones (SMu) than in the cleanest sandstones (Su). This illustrates the impact of detrital clays in facilitating strong mechanical compaction and catalysing chemical compaction.
It is difficult at best, and sometimes impossible, to definitively quantify the amount of silica produced at stylolites, but an attempt has been made to quantify the amount of silica generated by grain-contact dissolution using SEM-CL in a few selected samples following the methodology of Tournier et al. (2010). Comparison of quartz cements (measured from SEM-CL images) with this estimation of internal silica supply illustrates that most sandstones, even moderately argillaceous ones, are silica importers at the thin-section-scale (Fig. 15). In other words, an additional source of silica is typically required to explain the amount of quartz cement in these samples. This is a common observation in quartz-cemented sandstones, with nearby (within tens of cm) stylolites often supposed to supply the remaining silica in kinetic models of quartz cementation (Walderhaug 1996). Stylolites are present in the Late Ordovician rocks of this study, and clay transformations in interbedded mudstones may provide some additional silica. However, neither of these potential silica sources are uniformly distributed. The distance from each clean sandstone sample (<5% total ductiles) to the nearest mudstone, stylolite or argillaceous sandstone (i.e. a macroscopically identified potential silica source) was measured and the results are plotted in Figure 13d. These data have their limitations inasmuch as stylolites may have been missed in the core description, microstylolites and grain interpenetration are omitted, and shale, stylolite or argillaceous sandstone frequencies are not accounted for. For example, a clean sandstone sample may plot <1 m from a stylolite, but this could be the only stylolite for many metres. The amount of silica being exported in this case would clearly be less than in an interval of more frequent stylolites. Nonetheless, the data provide a useful, readily measureable proxy for the amount of silica available for cementation.
Quantification of silica import v. export using detailed SEM-CL analysis. (a) A very heavily compacted (pre-quartz-cement porosity = 10%) sandstone with sutured, strongly interpenetrating grain contacts (i). Quartz cement (ii) is relatively scarce and the earlier generation evident in (b) is not present. (b) A relatively lightly compacted (pre-quartz-cement porosity = 31%) sandstone, with predominantly tangential to elongate grain contacts (i) and relatively minor grain interpenetration (ii: overlapping areas highlighted on the higher magnification inset). Quartz cement is very abundant (17.5%) and two generations are easily differentiable due to differences in their luminescence (iii, older, non-luminescent; and iv, younger, moderately luminescent). (c) Cross-plot of total quartz cement (silica import: point-counted from SEM-CL images) v. overlapping grains (proxy for silica export: areas measured from SEM-CL images). Data points corresponding to the samples shown (a) and (b) are highlighted in bold. Total ductile content (point-counted volume %) is shown next to the data points.
In thin-bedded, heterolithic packages undergoing strong compaction, stylolites and grain dissolution in interbedded muddy sandstones may produce more silica than can be accommodated, thus leading to silica export (Fig. 16a). Any given clean sandstone bed within such a heterolithic package will be relatively close (within tens of cm or less) to a potential supply of aqueous silica and, hence, prone to a higher degree of cementation (Fig. 16a). This is supported by the generally lower permeability encountered in clean sandstones <2 m away from a macroscopically identified potential silica source (Fig. 13d) and by the porosity–permeability profile of the example well in Figure 16a.
Cartoon illustrating the possible control of bed thickness and reservoir heterogeneity on silica budgets and the amount of quartz cementation. (a) In thin-bedded clean sandstones, local silica supply is at least the same, if not greater than, the pre-quartz-cement porosity, and strong quartz cementation is likely. (b) In thick-bedded clean sandstones, the local silica supply may be less than the pre-quartz-cement porosity, especially in the centre of the sandstone package and this may result in less pervasive quartz cementation. Core plug porosity and permeability data corresponding to the selected intervals are shown to the right of the gamma-ray log profile. Core porosity is plotted on a linear scale, while core permeability is plotted on a logarithmic scale.
By comparison, in thick-bedded, stacked, clean–very clean sandstone successions, any given bed will be further away from a potential silica source (Fig. 16). Stylolites are rare (several metres apart) to absent in these clean sandstones and an early phase of quartz cement appears to have held the pore space open during compaction, leading to few interpenetrative grain contacts. Viable silica sources may be confined to underlying and overlying muddy sandstones and mudstones. A scenario could theoretically develop whereby sandstones closest to a silica source at the top and base of the stacked sandstone became preferentially cemented, leading to silica starvation in the centre of the clean sandstone package (Fig. 16). It is somewhat difficult to conclusively prove this hypothesis, as within stacked sandstones subtle variations in grain size, sorting and clay content complicate the picture. However, clean sandstones >2 m away from a macroscopically identified potential silica source tend to record good permeabilities (Fig. 13d), and the best RQ is typically encountered in thick (>50 m), stacked, clean sandstone successions (Fig. 16b).
Conclusions
A detailed study of >3 km of core from the In Amenas–Bourarhat Sud area, Illizi Basin, Algeria has helped to elucidate the main controls on reservoir quality (RQ). The primary reservoir target comprises Late Ordovician glaciogenic sandstones that have been subject to moderate mechanical and thermal stresses for a prolonged period of time (>400 myr).
Integration of sedimentological and petrographical techniques shows that a primary depositional fabric exerts a strong control on RQ:
Sandstones with a high detrital clay content (>15%) have been intensely compacted, and have negligible porosity and permeability.
Very-fine-grained, well-sorted sandstones with a moderate detrital clay content (5–15%) have been strongly compacted, but some post-compaction pore space remained. Macroporosity is typically pervasively occluded by quartz cements derived from readily available local silica sources, including clay-catalysed dissolution of quartz grains at grain contacts and along stylolites. The fine grain size provided a large surface area over which the quartz cements can precipitate. Variable amounts of microporosity are present in fibrous illite cements; however, low permeabilities show that the microporosity does not provide an effective flow pathway, even for gas.
Very-fine- to medium-grained, moderately well-sorted, very clean sandstones (<5% detrital clay) preserve the best RQ. They have been strongly compacted and variably quartz-cemented. The best RQ occurs in the least-cemented sandstones. Such low quartz-cement volumes are at odds with the burial (temperature) history. Close inspection of SEM-CL photomicrographs shows that the quartz overgrowths in the most permeable sandstones are non-luminescent – interpreted to be due to slow quartz-precipitation rates.
The best RQ sandstones are typically, but not exclusively, observed to occur in thick (>50 m) stacked sandstones at the top of the reservoir interval immediately below the regionally extensive lowermost Silurian Hot Shale seal and source rock. Although downward charge from the overlying Silurian Hot Shale occurred after quartz cementation had begun, there are observations to suggest that this hydrocarbon charge may have slowed the rate of quartz cementation and preserved RQ. Consideration of silica budgets suggests that the centres of thick, stacked sandstones may become silica-starved as they are more distant from silica sources, in contrast to thinner-bedded, more heterolithic intervals.
In reality, both of these processes are likely to be, at least in part, responsible for the observed RQ variations in this sedimentologically and diagenetically complex system. However, this study provides some simple concepts that can be tested in the search for better-quality reservoirs in the late Ordovician play in North Africa: that is, that the thicker, clean glaciogenic outwash sandstones close to the overlying Silurian Hot Shale should be targeted.
Acknowledgements
We would like to take this opportunity to thank those colleagues who have contributed to the data presented and to discussions on the contents and ideas herein, in particular Dorothy Payne, Norman Oxtoby, Richard Dixon, Barnaby Roome, Chereé Stover and Ed Jones. We also thank BP and Sonatrach for permission to publish this material. Finally, thanks to Richard Worden and two anonymous reviewers for their insightful comments on an earlier version of this manuscript, and to Pete Armitage and James Churchill for editorial comments.
Appendix 1
X-ray diffraction (XRD) analysis standard practices
A Hiltonbrooks generator with a Philips PW1050 goniometer was used to carry out all XRD measurements. Samples were first cleaned, where necessary, using distilled water and then toluene, and then gently crushed until fragments <5 mm in diameter were obtained. Whole-rock samples were disaggregated using a pestle and mortar: 2 g of material was then placed into a McCrone micronizing mill for 5 min. The slurry was dried, and the resulting powder tightly packed and reverse-mounted into an aluminium sample holder. The whole-rock powders were run from 5° to 60°2θ, with a step size of 0.05°, and were exposed to X-ray radiation from a copper anode at 40 kV and 30 mA.
Clay fraction (<2 µm) samples were totally disaggregated using a pestle and mortar: 5 g of the sample was mixed with 50 ml of distilled water and 5 ml of Calgon (sodium hexametaphosphate and sodium carbonate solution). The samples were then placed in an ultrasonic bath, followed by mixing table, ultrasonic bath and centrifuge (×2) before decanting 200 ml into a beaker. The total weight of clay extracted was obtained by removing 20 ml of the clay suspension, and evaporating this to dryness. A clay mount was made using the ‘Millipore filter’ (filtering the clay suspension until an even clay layer is achieved) before mounting into an aluminium XRD slide. Analysis of the clay-fraction mount was carried out in the following sequence: (1) after air-drying; (2) supersaturated in ethylene glycol overnight, using a temperature of 65°C; (3) heated for 2 h at 380°C; and (4) further heated at 550°C for 1 h. The clay-fraction samples are run from 3° to 32°2θ, with a step size of 0.05°, and were exposed to X-ray radiation from a copper anode at 40 kV and 30 mA.
Identification of minerals in both whole-rock and clay-fraction traces was made by comparison of their diffraction profiles with standard traces.
Fluid-inclusion analysis standard practices
For the determination of all phase changes during microthermometry, a ‘cycling’ protocol, following the general methodology for the study of fluid inclusions in diagenetic cements outlined by Goldstein & Reynolds (1994), was applied. The errors in the determination of phase changes were therefore controlled by the temperature increment used during the approach to the end point of the phase change being measured. Typical errors for final ice melting (Tice) are between ±0.2 and ±0.1°C, and, for homogenization temperatures (Th), between ±2 and ±5°C. No pressure corrections were applied to the homogenization temperatures.
Salinities were modelled in the system NaCl–H2O using ‘Calcicbrine’ (Naden 1996). Very limited first ice-melting temperature data indicate that this is a simplification of the reality and that other divalent cations are also present. However, owing to the generally very small size of the inclusions, it was not possible to derive data on the temperatures of other phase changes (i.e. hydrate melting, etc.) that could be used to model salinities in a more sophisticated manner.
- © 2018 The Author(s). Published by The Geological Society of London. All rights reserved
References
Deciphering multiple controls on reservoir quality and inhibition of quartz cement in a complex reservoir: Ordovician glacial sandstones, Illizi Basin, Algeria
