Proefschrift
Ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,
op gezag van Rector Magnificus prof.dr.ir. J. T. Fokkema voorzitter van het College voor Promoties,
in het openbaar te verdedigen op dinsdag 2 september 2008
door
Jozijntje Krijntje Johanna TAAL – VAN KOPPEN
Doctorandus in de geologie
Prof. dr. S.M. Luthi
Samenstelling promotiecommissie:
Rector Magnificus voorzitter
Prof. dr. S.M. Luthi Technische Universiteit Delft, promotor Prof. dr. S.B. Kroonenberg Technische Universiteit Delft
Prof. dr. J. Bruining Technische Universiteit Delft Prof. dr. S.A.P.L Cloetingh Vrije Universiteit Amsterdam Dr. G. Bertotti Vrije Universiteit Amsterdam
Dr. A. Etchecopar Schlumberger
Dr. Y. Nassir Shell International Exploration and Production
Dr. G. Bertotti heeft als begeleider in belangrijke mate aan de totstandkoming van het proefschrift bijgedragen.
This research was financed by Shell International Exploration and Production
Karoo Kind
Blink skitter die liggies in die strale van die genadelose son. Om te probeer verlei die laaste bietje wat hy kon, - Op n arm waar hy reeds die water weggedamp het, - Maar die sout sal nooit deel wees van hom.
Karoo-kind, met jou liefde soos die van n kameeldoring, - Wys met lower en sagte geel die waarde van jou gees. Geanker in stollende gesteentes, deur tye wat net sy maker ken.
Terwyl n doring soos n assegaai uit pen.
Eg soos die brak van die grond waarvan niemand iets wil weet nie, Gee jy jou lewe aan ander sonder veel belangstelling.
Jy beteken nie veel,
maar hou vas aan jou deel, wat jou gaan dra tot n oorwinning.
Was originally signed:
Met liefde aan die Karoo Kourie Coetzee
Veel mensen hebben bijgedragen aan dit proefschrift en zonder een ieder van hen had ik dit werk niet kunnen voltooien.
Professor Luthi en Dr. Bertotti, Stefan and Giovanni, mijn dagelijkse begeleiders, jullie zijn beide, elk op een heel andere manier, van cruciaal belang geweest in de afgelopen zes jaar. Stefan, als mijn promotor was je een strenge begeleider die me uitdaagde tot het uiterste te gaan. Je deur stond altijd voor me open en er was altijd ruimte en tijd voor discussie. Giovanni, officieel mag je geen toegevoegd promotor zijn, voor mij ben je dat wel, want jouw begeleiding was net zo belangrijk als die van Stefan. Jouw creativiteit en nooit aflatende enthousiasme zijn een inspiratiebron voor me geweest. Ik heb genoten van de veldwerken die ik samen met jullie ondernomen heb en de heerlijke wijnen die we dronken na een dag hard werken in het veld.
This research would not have been possible without the financial support of Shell Exploration and Production. I want to thank Mr. Mercadier, Mr. Rawnsley, Mr. Schulte and Mrs. Yassir for this opportunity.
I would like to thank the people from the University of Stellenbosch and in particular Mr. deVille Wickens for their help with the preparation of the field trips and our stay in the core shed. Also thanked are the farmers in the Karoo area who were kind enough to let us work on their land. Baie veel dank.
The NOMAD consortium is thanked for allowing us to work with the core and borehole data.
Mr. Arnaud Etchecopar is kindly thanked for supervising me during my stay at Schlumberger Etudes et Productions Paris.
We were allowed to use various software packages with university licenses; the people from Sclumberger (Eclipse and Petrel), Golder Associates (FracMan) and Pangaea Scientific (SpheriStat) are thanked for their technical support.
Een gedeelte van dit werk is gebaseerd op het werk van MSc en BSc studenten. Tijs Beek, Martijn van Galen, Simone de Jong en Vincent Verlinden, bedankt voor jullie bijdrage.
Tijdens de vier jaar dat ik aan de TU Delft werkte ben ik in het gezelschap geweest van geweldige collega’s bij wie je niet alleen aan kon kloppen voor een ‘bakkie’, maar ook met de sores die een ieder van ons meemaakte of mee had gemaakt gedurende de jaren van het promotie onderzoek. Joep, Israël, Bob, Luc, Klaas, Remco, Marit, Wiebke, Tine, Irina, Albert, Rory, alle anderen en niet in de laatste plaats de secretaresses en de mensen van de IT afdeling, dank je wel.
For already two years I am happy to have returned to Total E&P NL, a company that has given me the opportunity to do an internship during my studies at the Vrije Universiteit and was kind enough to advise me when I decided to start my PhD. The company, the work and especially my colleagues make me bike to work every day with pleasure. Special thanks go to Laurance and Arjan who taught me so much in the first year. Sophie
Ik prijs mijzelf gelukkig met mijn familie, want de basis van wie ik ben en wat ik geworden ben, heb ik aan jullie te danken. Ma, ik ben blij dat je nog van het leven mag genieten en pa, ik denk dat we toch wel veel op elkaar lijken.
Vanaf de eerste dag hebben Aad en Ada mij als dochter opgenomen in hun gezin, dank je wel.
Vrienden komen en gaan en een ieder ben ik dankbaar omdat ook zij hebben bijgedragen aan wie ik ben. Judith en Arina, onze vriendschap zal altijd ‘zijn’; Brigitte en Elisabeth, de beste paranimfen die ik me kan wensen; Marleen en Ruud, voor al die mooie zomeravonden rond de vuurkorf; Ladies van de eetclub, het was altijd ‘hmmmm, erg lekker’; Marleen Pennings, er wacht een flesje port op je; Jeroen, mijn favoriete IT-er en gids in de wereld van de moderne muziek; Matthijs, de leukste Nederlandse Nigeriaan; en natuurlijk Sandra en Gediene, ’20 baantjes et un pour la route’.
Er zijn twee mensen die ik in het bijzonder wil bedanken, zonder hen was dit proefschrift er niet geweest.
Jan-Simon van der Linde, mijn leraar aardrijkskunde op de middelbare school, je was niet mijn inspiratiebron om geologie te gaan studeren, want jouw interesse lag meer op het vlak van de sociale geografie. Maar zonder jouw hulp en steun, die ik zo hard nodig had in die moeilijke tijd, had ik nooit mijn middelbare school diploma gehaald en had ik niet aan dit promotie onderzoek kunnen beginnen.
Jacco Taal, mijn man, mijn maatje, mijn lief, in de afgelopen tien jaar ben je niet van mijn zijde geweken en vanaf het moment dat ik besloot om aan dit promotie onderzoek te beginnen heb je me onvoorwaardelijk gesteund en die steun heb ik heel hard nodig gehad. Je hebt me geleerd om me in andere mensen te verplaatsten en in te leven, dingen van een andere kant te bekijken.
‘If you love something set it free, if it loves you back it will return’, het is een rode draad in ons leven. Elkaar vrij laten om onze eigen weg te zoeken en er tegelijkertijd voor elkaar zijn, daar ben ik dankbaar voor. Ik hou van je.
SAMENVATTING SUMMARY
CHAPTER 1 INTRODUCTION ... 1
1.1 Problem Definition and Objectives of the Thesis... 1
1.2 Approach... 2
1.3 Theoretical Background of Rock Fractures ... 4
1.3.1 Fracture scale, shape and spacing ... 8
1.4 Structure of the Thesis ... 9
CHAPTER 2 FIELD AREA DESCRIPTION AND DATA ACQUISITION METHODS ... 11
2.1 Field Area... 11
2.1.1 Geography of the Study Area ... 11
2.1.2 Regional Geological Setting ... 14
2.1.3 Geology of the Field Area... 16
2.2 Data Acquisition Methods ... 19
2.2.1 Surface Data... 19
2.2.2 Subsurface Data ... 22
CHAPTER 3 FAULTS AND FOLDS ... 25
3.1 Introduction... 25
3.2 General Architecture of Deformation Zones ... 26
3.2.1 Outcrops... 26
3.2.2 Boreholes ... 28
3.3 Secondary Tectonic Features ... 30
3.3.1 Secondary Faults... 30
3.3.2 Folds... 32
3.3.3 Breccias... 33
3.4 The Nature of Cement in Fractures and Breccias ... 34
3.4.1 Introduction... 34
3.4.2 Fracture Mapping... 34
3.4.3 Microscopic Analysis of the Cement... 35
3.4.4 Interpretation... 37
3.5 Tectonic Framework ... 38
CHAPTER 4 FRACTURES AND FRACTURE SWARMS... 41
4.1 Introduction... 41
4.2 Statistical Methods for Processing of Fracture Orientations ... 42
4.3 Fracture Orientations: Results from Statistical Analysis... 45
4.3.1 Outcrop Data... 45
4.3.2 Borehole Data ... 52
4.3.3 Summary of Fracture Patterns: Surface versus Subsurface ... 59
4.3.4 Timing and Dynamics of Fracturing... 61
4.4 Fracture Density and its Relation to Bed-Thickness... 63
4.4.1 Introduction... 63
4.4.2 Outcrop Data: Processing of Fracture Density Data... 63
4.4.3 Fracture Densities from Outcrops Measurements... 64
4.4.4 Subsurface Data: Processing of Fracture Density Data ... 66
4.4.5 Fracture Densities from Borehole Measurements... 68
4.4.6 Interpretation of the Fracture Densities in Outcrops and Boreholes... 71
4.5 Fracture Swarms ... 72
4.5.1 Orientation of the Swarms ... 73
4.5.2 Fracture Densities and Recurrences of Fracture Swarms ... 74
4.5.3 Timing and Dynamics of Fracture and Fracture Swarms ... 77
CHAPTER 5 IMPLICATIONS FOR FLUID FLOW BEHAVIOUR... 79
5.1 Introduction... 79
5.1.1 Fracture Network Modelling... 79
5.1.2 Fluid Flow Modelling ... 81
5.2 Fault-Related Fracture Study ... 82
5.2.1 Model Setup and Flow Parameters ... 83
5.2.2 Fluid Flow Results ... 84
5.3 Sensitivity Study ... 88
5.3.1 Model Setup and Flow Parameters ... 89
5.3.2 Base-case Model ... 90
5.3.3 Permeability ... 94
5.3.4 Fracture Density... 96
5.3.5 Fracture orientation... 98
5.3.6 Fracture Length... 98
5.4 Summary and Conclusions ... 100
CHAPTER 6 DISCUSSION AND CONCLUSION... 101
REFERENCES ... 105
APPENDICES ... 115
Reservoirs die worden gekenmerkt door scheuren en breuken in het gesteente (“fractured reservoirs”) zijn in de ondergrond over het algemeen moeilijk te karakteriseren omdat de resolutie van seismische gegevens te laag is om de scheuren te detecteren terwijl boorput gegevens zeer gedetailleerd maar erg gelokaliseerd zijn. Veldgegevens van dagzomende gesteentelagen die als analoog voor “fractured reservoirs” worden beschouwd kunnen van groot belang zijn om de kennis van breuken en scheursystemen te vergroten, met name om inzicht te krijgen in de driedimensionale geometrie, de onderlinge connectiviteit en individuele eigenschappen. Of en hoe we observaties van oppervlaktegegevens kunnen gebruiken om reservoirs in de ondergrond te beschrijven is een belangrijke vraag die zelden wordt gesteld.
Om inzicht te krijgen in dit probleem, wordt gebruik gemaakt van unieke data set van een veldanaloog in het Tanqua Karoo bekken in Zuid-Afrika. De data bestaat uit luchtfoto’s, veldgegevens en boorput gegevens van een aantal ondiepe onderzoeksputten die door de veldanaloog zijn geboord. Veldgegevens zijn verkregen door gebruik te maken van een nieuw ontwikkelde, digitale techniek die ons in staat stelt alle gegevens direct in de ruimte te positioneren en op te slaan in een digitale database. Boorput gegevens bestaan uit gesteente kernen en zogenaamde “borehole images” verkregen met verscheidene technieken.
De doelstelling van deze studie is tweeledig: het kenmerken en vergelijken van breuken en breukpatronen aan het oppervlak en in de ondergrond en het verklaren van de verschillen in geomechanisch gedrag van het gesteente. Daarnaast is het belangrijk inzicht te krijgen welke eigenschappen van breuken en scheursystemen en op welke manier deze ondergrondse vloeistof stromen beïnvloeden.
De veldanaloog bevindt zich in het zuidwesten van het Zuid-Afrikaanse Karoo bekken. Dit is een “retro-arc” voorlandbekken van Permische ouderdom dat is gevormd als gevolg van Z-N gerichte compressie die verantwoordelijk is voor de vorming van de “Cape Fold Belt”. Het gesteente wordt gevormd door vier opeengestapelde “basin floor fans” van de Skoorsteenberg Formatie die geleidelijk naar het noorden en oosten uitdunnen. Boorput gegevens van deze formatie zijn verkregen gedurende een stratigrafische studie (het NOMAD project). Zeven ondiepe onderzoekputten werden geboord waarvan complete gesteente kernen genomen zijn. Omdat sommige putten zich dicht bij de verticale, gesteentewanden bevinden kunnen we een directe vergelijking van de breuken in de boorgaten met die in de dagzomende gesteentes maken. Georiënteerde beeldopnames van de wand van het boorgat zijn verkregen met verschillende technieken waaronder elektrische, optische en akoestische beelden. In de kernen alsook op de beelden van het boorgat worden intensieve scheursystemen waargenomen.
Een Landsat beeld en een aantal recente (1998) luchtfoto's die het volledige gebied bedekken, tonen lineamenten die in twee dominante richtingen georiënteerd zijn, O-W en NW-ZO, en die de morfologie en het drainagesysteem lijken te controleren. Een uitgebreide veldstudie bevestigt de aanwezigheid van deze twee dominante lineamenten en toont aan dat de O-W gerichte lineamenten kunnen worden geassocieerd met de plooiassen en de strekking van de breuken. De NW-ZO georiënteerde lineamenten worden geassocieerd met zones van hoge scheurdichtheid, zogenaamde scheurzones.
plooien en scheurzones zijn in detail bestudeerd alsook gebieden waar alleen kleinschalige scheursystemen de gesteentes doorkruisen, waardoor ook inzicht in het regionale scheursysteem verkregen wordt. Om het effect van laagdikte op de scheursystemen te bestuderen werden drie laagdiktes geselecteerd: massieve zandsteen, dungebankte zandsteen en schalies. Veldgegevens zijn verkregen met een nieuwe digitale opname techniek die ontwikkeld is om een groot aantal breuken in een bepaald studiegebied te kwantificeren en op te slaan in een “geo-referenced” gegevensbestand. Naast scheursystemen die direct met breuken of plooien in verband kunnen worden gebracht, bestaat er aan het aardoppervlak een achtergrond patroon dat over het hele gebied word waargenomen. Dit patroon bestaat uit een dominante richting en twee ondergeschikte richtingen die overeen lijken te komen met waarnemingen op de luchtfoto’s. De interpretatie van de boorgatbeelden laat dezelfde dominante richting zien, echter maar een van de ondergeschikte richtingen. Ondanks dat er verschillen bestaan, lijken de scheursystemen op alle schalen vergelijkbaar te zijn.
Een interessante observatie werd gemaakt betreffende scheurdichtheid. Aan het oppervlak, laten dungebankte zandsteen een grote spreiding van scheurdichtheid zien, maar over het algemeen is de dichtheid min of meer constant. Deze tendens wordt doorgaans niet waargenomen in veldanalogen. De boorgatbeelden laten zien dat in de ondergrond scheuren voornamelijk in de dikkere gesteentelagen voorkomen. De statistische analyse toont, verrassend genoeg, een omgekeerd verband tussen laagdikte en breukdichtheid, d.w.z. wanneer de lagen dikker worden, wordt scheurdichtheid ook groter. Mogelijke verklaringen voor deze observaties zijn: De dungebankte lagen zijn meer ontvankelijk voor veranderingen in het spanningsveld, bijvoorbeeld door erosie, en aan de oppervlakte gerelateerde processen zoals thermische verwering. Daarnaast worden dungebankte lagen geassocieerd met schalielagen die ervoor zorgen dat de zandsteenlagen niet bros maar ductiel reageren wanneer ze onder spanning komen. In tegenstelling tot dungebankte lagen, zullen dikgebankte lagen de spanningen in zich opnemen en daarom op een brosse manier reageren.
Om te bestuderen welke eigenschappen van scheursystemen de prestaties van een (olie) reservoir beïnvloeden, werd een sensitiviteitsstudie uitgevoerd waarbij een eenvoudig reservoirmodel werd bevolkt met typische reservoirparameters voor zowel de gesteentematrix als de scheuren. Onderzochte parameters omvatten permeabiliteit van de scheuren, intensiteit en scheurlengte. Wanneer de permeabiliteit van de scheuren vermindert, word geïnjecteerd water gedwongen om door de matrix te stromen waar het de olie naar de scheuren verplaatst. Dit heeft een hogere olie productie en een uitgestelde waterdoorbraak tot gevolg. Als de scheurintensiteit zeer laag is, zal de waterdoorbraak vroeger komen en is de olieproductie is lager, mits er een aaneengesloten cluster van scheuren is. Bij een zeer hoge scheurintensiteit wordt de zogenaamde “percolation threshold” bereikt, de invloed van het scheursysteem vermindert en zal uiteindelijk verdwijnen. Wanneer bij gelijke scheurintensiteit de lengte van de scheuren minder wordt, produceert het reservoir langer en met een constant snelheid, terwijl verlenging van de scheuren geen duidelijk effect op reservoirprestaties aantoont.
Fractured reservoirs are notoriously difficult to characterize because the resolution of seismic data is too low to detect fractures whereas borehole data is detailed but sparse. Therefore, outcrops that are considered analogues of fractured reservoirs can be of great support in gaining knowledge of the three-dimensional geometry of fractures and fracture networks, their connectivity and individual properties. Whether surface observations can be extended to the subsurface is, however, a rarely addressed question.
Using a unique data set, we try to gain insight into this problem. The data set consists of multi-scale fracture data obtained from an outcrop area through which a number of shallow research wells were drilled. It includes aerial photography that covers the entire area, digital outcrop measurements in a number of geological settings, borehole images obtained with several different imaging techniques and continuous cores. Our goals are twofold: to characterize and compare fractures and fracture patterns at the surface and the subsurface, and to explain the differences in geomechanical behaviour of the rocks. In addition, we want to gain insight into how fracture (network) properties affect fluid flow behaviour using a simulated reservoir situation with water injection.
The study area is located in the south western part of the South African Karoo basin. This Permian retro-arc foreland basin developed in response to S-N shortening that caused the evolution of the Cape Fold Belt. In the study area, the prominent outcrops of the Tanqua-Karoo basin are formed by four stacked basin floor fans of the Skoorsteenberg Formation that gradually thin towards the N and E. During a stratigraphic study of this formation (the NOMAD project) seven research wells were drilled through the outcrops and all are fully cored. Oriented borehole images were obtained using electrical, optical and acoustic techniques. Some of these wells are located near the cliffs, allowing a direct comparison of the fractures in the boreholes with those of the outcrops. Extensive fracturing can be observed in the cores and on the borehole image logs.
A Landsat image and a set of recent (1998) aerial photographs covering the entire study area in the Tanqua basin show two dominant lineament directions, striking E-W and NW-SE, that both seem to control the morphology and the drainage system. An extensive field study confirmed the presence of these two dominant lineaments. It shows that the E-W trending lineaments are associated with folds and low-angle thrusts; while the NW-SE trending lineaments are associated with zones of high fracture density i.e. fracture swarms. Locations for detailed fracture studies were chosen near boreholes in order to compare field data with borehole measurements. Several deformation zones and fracture swarms were studied in detail as well as areas affected solely by fracturing. The latter allows us to gain insights into the regional fracture pattern.
To study the effect of bed thickness and lithology three rock types were distinguished: massive sandstone, thin-bedding sandstones and shales. Outcrop data was gathered using new digital acquisition and processing techniques specially developed to quantify a large number of fractures in a given study area. They are stored in a georeferenced database. Fracture orientations in the outcrops show a background pattern with one dominant and two minor trends that coincide with those observed on aerial photography. Superimposed on this “far-field” pattern are fractures caused by local structural effects such as folds and faults. Interpretation of the borehole images revealed the same dominant trend but only
coinciding at all scales.
A different observation, however, was made concerning fracture densities in the outcrops. Although the thin-bedded sandstones demonstrate a large scatter the overall fracture density at the surface is more or less constant, a trend which is generally not observed in other outcrops studies. In contrast, the borehole images revealed that most fracturing occurs in thicker beds. Statistical analysis shows surprisingly an inverse relationship between bed thickness and fracture density, i.e. thick beds have a higher fracture density than thin beds.
Possible explanations for these observations are: Thin beds can be more susceptible to stress unloading and surface effects such as thermal weathering and can therefore be more fractured at the surface compared to the subsurface. Thin beds are also associated with shale layers that could have caused them to react in a ductile manner to stresses. In contrast to thin beds, thick beds tend to concentrate stresses and are more prone to react in a brittle manner.
After characterizing and comparing fractures in both data sets, we studied how and which fracture properties influence the productivity of a fractured hydrocarbon reservoir. A sensitivity study was performed with a simple reservoir model that was populated with typical reservoir parameters for both the matrix as well as the fractures. Fracture parameters investigated include permeability, intensity and length. We found that decreasing fracture permeability causes the injected water to flow into the matrix, which increases the recovery and causes a delay in water breakthrough. When the fracture intensity is very low, the water breakthrough occurs earlier and recovery is lower as long as there is a percolating fracture cluster. When the fracture intensity becomes very high and reaches the percolation threshold, its influence decreases and eventually disappears. With decreasing fracture length (equal fracture intensity), the reservoir produces longer at a constant rate, whereas increasing the fracture length does not show a clear effect on reservoir performance.
1.1 Problem Definition and Objectives of the Thesis
The Earth and its interior are constantly in motion, causing earthquakes and volcanic activity, and shaping the Earth’s surface. Rocks are therefore always under stress. When applied stresses exceed the strength of a rock, the rock fails either in a brittle manner by breaking or in a ductile manner by internal reorganization of the rock (Figure 1-1).
The term fracture is a generic term for ‘surfaces along which rocks or minerals have broken and are surfaces across which the material has lost cohesion’ (Twiss and Moores, 1992). Fractures are therefore the result of brittle deformation of a rock. Due to loss of cohesion of the rock along the fracture surface, fractures can cause stability problems in underground mining and civil engineering constructions. When fractures create spaces (voids) within the rock, they can act as conduits for fluid movements and can therefore have a positive effect on groundwater management or hydrocarbon production. On the other hand, when fractures are filled by mineral precipitation they can cause considerable compartmentalization of a reservoir, and therefore have a negative effect on fluid flow.
Figure 1-1 a) Brittle deformation of a sandstone succession accommodated by normal faulting. b) Ductile deformation in a limestone succession accommodated by folding (Skinner and Porter, 1995).
While fractures can be detected and resolved in wells in the subsurface either through coring or borehole logging techniques like imaging, their three-dimensional distribution over an area of interest such as an oilfield cannot be deterministically assessed. In addition, seismic resolution is commonly too low to perceive the fractures, although recent advances in seismic processing techniques provide new tools for sub-seismic fracture imaging between wells (Gray et al., 2002; Parney and LaPointe, 2003; Qian et al., 2006).
Outcrop analogues are widely used to study fractures and fracture networks, and from there, to predict reservoir behaviour in the subsurface. As to what extent outcrop
observations are representative for the subsurface is a crucial yet unanswered question related to such studies.
The principal objective of this Thesis is to examine how surface data of fractures and fracture networks occurring in various geological settings compare and can be combined with borehole data to build realistic fracture networks for the subsurface. An additional and more reservoir-engineering oriented goal is to study which specific aspects of such models are of major influence on fluid flow behaviour through fracture networks.
1.2 Approach
The study area chosen lies in the setting of the Tanqua-Karoo basin where a number of Permian basin-floor fans form excellent outcrops through which several shallow research wells were drilled during a previous, stratigraphic research project (the NOMAD project, Hodgson et al., 2006; Luthi et al., 2006). The layers are only mildly deformed but nevertheless exhibit a variety of conspicuous tectonic features such as low-angle thrusting and folding. Fractures are found throughout the outcrop area in all tectonic sub-environments and are well exposed. Therefore, all structural features can be studied and the observations can be placed in a regional tectonic context to improve the understanding of the geological history of the area. This also provides insight into the mechanical behaviour of the sedimentary rocks.
In this project, the structural features and the fractures in the area are studied over a wide range of scales using Landsat images, aerial photography, on-site outcrop inspection, and analysis of well data such as cores and borehole images (Figure 1-2). This multi-scale approach to fracture network studies can be related to the various tectonic settings with the following questions in mind:
• What is the width of the affected area around basin-scale features such as faults and how can we predict the affected area around structural features that can be observed with seismic imaging techniques?
• How can outcrop data be used to model fracture networks in the subsurface?
• How can borehole data be used to predict fracture networks in the inter-well area?
One of the immediate problems that can be addressed with such a data set is to what extent an outcrop analogue is representative of the subsurface fracture network. In order to do this in a scientifically correct manner, both surface as well as subsurface data has to be acquired with methods that eliminate any bias as much as possible. At the surface, fracture data is gathered using new GIS-based acquisition techniques that allow building of a fully digital database directly in the field with all data contained in a geo-referenced manner. The borehole data are all in digital form and are systematically analyzed using state-of-the-art software. This resulted in a database that contains a large number of quantitative and symbolic fracture properties such as orientation, height, length, types of fracture fill etc. From these original data, average properties such as fracture densities and orientations can be extracted, and the fracture network can be
subdivided into fracture sets through clustering. The fracture densities in the subsurface are obtained using a novel approach similar to what is normally used in outcrops, i.e. the fracture density is calculated perpendicular to the fracture planes; the results are therefore considered to be directly comparable with surface fracture densities.
The outcrop and borehole data are combined to produce several fracture network models as a function of the structural settings. These are subsequently probed for their influence on fluid flow behaviour using simulations, whereby the fracture parameters are changed such that their sensitivity on fluid flow behaviour can be evaluated.
Figure 1-2 Data used for the multi-scale approach used in this study; a) aerial photography, b) outcrop data, c) borehole imaging, d) cores and e) thin sections.
1.3 Theoretical Background of Rock Fractures
Prior to failure, rocks are assumed to behave elastically when subjected to stress i.e. the rock will return to its original size and shape as soon as the applied stresses are removed (Price, 1966). When stresses build up further, they eventually exceed the
strength of the rock (i.e. yield strength) and the rock will then deform permanently due to ductile deformation (i.e. no loss of cohesion of the rock) or brittle deformation (i.e. the rock loses cohesion; Figure 1-1). How the rock fails depends on the properties of the rock, e.g. lithology, isotropy and porosity, although pressure, temperature and strain rate (i.e. the amount of strain per unit time) play a dominant role in the behaviour of a rock under stress. At low confining pressure for example, a rock will behave brittle whereas increasing confining pressure will move the rock from the brittle domain into the ductile domain (Price, 1966). Pore fluid pressure on the other hand works against the confining pressure and reduces the strength of a rock (Price and Cosgrove, 1990). When the pore fluid pressure approaches the confining pressure, the rock will behave brittle at higher confining pressures (Price, 1966). Brittle deformation generally occurs at moderate temperatures (240° – 600° C, depending on the lithology) and is therefore limited to the brittle, upper crust (Mandl, 2005). Increasing temperatures allow ductile processes to prevail. This is also the case under very low strain rate conditions, e.g. a rock can slowly change its shape by ductile processes under its own weight whereas the same rock will behave brittle when subjected to a high increase in strain rate.
Many laboratory experiments have shown that in the brittle domain a rock fails along consistent angles relative to the three principal stresses (Price, 1966; Goodman, 1989). Two main types of fractures are distinguished (Figure 1-3): Shear fractures have displacements parallel to the fracture plane whereas the displacement of opening mode fractures is perpendicular to the fracture plane. Shear fractures commonly develop in conjugate sets with the fractures forming an acute angle to the maximum compressive stress direction (σ1) and an obtuse angle to the minimum compressive stress direction
(σ3). When the displacement along a shear fracture is significant, it is called a fault
(Dennis, 1967; Price and Cosgrove, 1990). Anderson (1951) described three different types of faulting; normal faulting, thrust faulting and strike-slip faulting that develop depending on the direction of the three principal stresses. Obviously, the difference between a shear fracture and a fault is difficult to determine.
Figure 1-3 Different types of fracturing developed during experiments on rocks. a) Tension fractures; b) Longitudinal splitting; c) Extension fractures and d) Conjugate shear fractures (Twiss and Moore, 1992).
Extension and tension fractures are opening mode fractures, generally referred to as joints, forming in the plane of the maximum (σ1) and medium (σ2) compressive
stresses and show displacements perpendicular to the fracture plane (Figure 1-3, Twiss & Moore, 1992). The difference between these two types of fractures is that the minimum stress direction is compressive (positive) in the case of extension fractures while the minimum stress direction is tensile (negative) in the case of tensile fractures (Nelson, 2001). Fractures due to longitudinal splitting form when the minimum compressive is close to zero. They are aligned with the maximum compressive stress (Twiss and Moore, 1992).
There has always been a lot of debate in the literature about the terminology of fractures, see Pollard and Aydin (1988) for an historical overview of fractures and a discussion on terminology. Here, the term fracture is preferably used for all planes along which the rock lost its cohesion because the term joint for example directly implies opening mode fractures and is therefore already an interpretation of the origin of the fracture.
The relationship between fractures and the three principal stresses has been established in laboratory experiments under well defined conditions and can help to analyze the fracture patterns observed in (sub) surface rocks. Natural fractures represent the local state of stress at the time of fracturing and may develop at any time in the history of a rock. They can develop as soon as the rock is deposited and is still in unconsolidated condition or due to large-scale tectonic processes that causes consistent fracture patterns over large areas (e.g. at basin-scale, here these type of fractures are referred to as regional fractures). They can also be related to small(er) scale processes such faulting or folding. And they may be associated to surface-related processes such as thermal weathering.
As mentioned regional fractures may develop over extremely large areas in undisturbed, horizontal strata cross-cutting local structures (Figure 1-4; Mandl, 2005). They have a simple geometry and consistent orientations although variability in orientation as much as 30-45° has been reported (Lorenz and Finley, 1991; Laubach, 1991).
Figure 1-4 Regional fractures in Jurassic Navajo sandstone, Lake Powell, southern Utah (Nelson, 2001).
Regional fractures have relatively large spacing without evidence of offset across the fracture plane (Nelson, 2001) and are generally referred to as a systematic set of fractures. At the surface, these fractures are commonly accompanied by a shorter, discontinuous set of fractures that are more or less orthogonal to the systematic set and classified as a non-systematic set. The origin of regional fractures is at present still unclear. Several authors have proposed that regional fracture originate from stress release due to unloading of strata during uplift and subsequent erosion (Currie and Nwachukwu, 1974; Engelder, 1987). An increasing amount of studies from subsurface data such as oriented cores however, describe only one, well developed set of fractures at depth suggesting regional fractures can be present and formed at considerable depth in unfolded, horizontal strata (Lorenz and Finley, 1991 and references therein). This suggests that cross fracturing due to stress release during uplift and erosion can be a younger effect. Lorenz et al. (1991) proposed that regional fractures can form at depth due to high pore pressure in the rocks in combination with tectonic compression induced by, for example, a nearby active fold- or thrust-belt.
As already stated, faults are shear fractures along which significant displacement parallel to the fracture plane took place. Faults can be pre-dated by zones of intense fracturing that prepares the rock for the eventual offset (Nelson, 2001; Peacock, 2001; Mandl, 2005). When, for some reason, deformation ceased at an early stage of growth and large-scale slip did not occur, a so-called fracture swarm is the only evidence of deformation that is left. Fracture swarms are long (typically several kilometres) and continuous, but very narrow (several meters) clusters of fractures with little shear displacement. The terminology for fracture swarms in the literature is variable: Their names range from joint zones (Hodgson, 1961; Engelder, 1987) to zoned joints (Dyer, 1983), joint swarms (Laubach, 1991), joint zones or fracture swarms (Cruikshank and Aydin, 1994) and fracture corridors (De Joussineau, 2004).
Faults can be accompanied by small shear fractures of which one fracture set is parallel while the other is conjugate to the fault (Figure 1-5).
Figure 1-5 Fracture patterns typically observed around a normal fault (a) and reverse fault (b; Petit et al., 2000).
Another possible fracture set is extensional and bisects the acute angle between the shear fractures. They are so-called pinnate fractures that form en echelon arrays oriented similar with respect to the fault they are associated with. The acute angle between the shear fracture and the fault can be used to indicate sense of shear because it points in the direction of the relative movement along the fault (Price, 1966; Twiss and Moore, 1992).
Fractures that post-date faulting are influenced by the local stress field around the fault. Several authors have described examples and made experiments showing opening mode fractures rotate in the vicinity of pre-existing faults. Pollard and Segall (1987) for example show fractures that curve to be parallel to a fault whereas Rawnsley et al. (1992) and Petit et al. (2000) show more complex fracture paths in the vicinity of conjugate strike-slip faults.
Fold-related fractures are of great variety because the orientation and magnitude of the principal stresses changes with position in the fold and the development of the fold trough time. In general, a set of opening mode fractures forms perpendicular to the fold axis and the bedding of the sedimentary layers. Another set may form parallel to the fold axis also perpendicular to the bedding and will therefore vary in dip depending on the position compared to the crest of the fold (Figure 1-6; Price, 1966; Price and Cosgrove, 1990).
Figure 1-6 Fractures associated with folding (http://earthsci.org).
Shear fractures that develop in association with folding show more complex relationship with the fold geometry. Shear fractures with high angles compared to the bedding are parallel to the fold axis and associated with the convex side of the fold, they can have normal displacement. In the convex side of the fold, shear fractures with low angle compared to bedding tend to develop, they can exhibit reverse displacements.
Other types of fractures described in the literature are sheet fractures, columnar fractures and desiccation fractures. Sheet joints are sub-parallel to the topography and develop as extension fractures due to uplift during which the maximum stress remains horizontal and the minimum stress vertical or as a result of residual stresses in a rock due to cooling of for example a plutonic igneous body. Columnar fractures probably result from thermal stresses by unequal cooling in igneous extrusions and shallow intrusions.
Columnar fractures commonly are organised in a hexagonal pattern, why they form in such a pattern is not fully understood. A similar pattern of fractures is observed in desiccation fractures that form during dewatering and associated contraction in mud layers at the surface.
1.3.1 Fracture scale, shape and spacing
Fracture dimensions can vary from the grain size of the rock (micro- cracks) up to several hundreds of meters and even kilometres (master fractures). Shape and scale are controlled by the type of rocks (e.g. sedimentary, metamorphic or igneous), the layering of the rocks (e.g. massive, layered or foliated), the thickness of the unit through which they cut, their orientation compared to the layering and the process of fracturing (e.g. cooling). In uniform rock types such as granites, fractures tend to be circular or elliptical of shape with the long axis in the horizontal direction. In sedimentary rocks where layers of different lithology and mechanical properties alternate, fractures are generally constrained to the upper and lower boundary of a sedimentary bed or unit (Price, 1966; Pollard and Aydin, 1988).
Fractures tend to occur in series of sub parallel fractures at systematic distance from each other (i.e. spacing). The spacing between fractures (or fracture density) has long been studied and it has been recognized by many authors that fracture spacing mainly depends on lithology and bed thickness (Hobbs, 1967; Priest and Hudson, 1976; Price, 1966; Ladeira and Price, 1981; Narr and Lerche, 1984; Huang and Angelier, 1989; Narr and Suppe, 1991; Wu and Pollard, 1995). Fracture spacing for example is much higher in limestone than in greywacke (Twiss and Moore, 1992) and stronger rock show more closely spaced fractures than weaker rock (Huang and Angelier, 1989). Ladeira and Price (1981) noted that the fracture density not only depends on the lithology of the fractured bed, but also on the thickness of the overlying and underlying weak layers. They observed that fractures are more closely spaced when the weak layers are very thin.
Several relationships between bed thickness and fracture spacing or fracture density have been proposed, they range from negative exponential (Priest and Hudson, 1976), to linear (Ladeira and Price, 1981) and log normal (Narr and Suppe, 1991), but widely accepted is the fact that fracture densities decrease when bed-thicknesses increase. Different explanations for this inverse relationship have been proposed, they are mainly based on the assumption of stress release or a drop in fluid pressure around the freshly formed fracture. The accompanying so-called stress shadow around the fracture prevents new fractures from growing within a certain distance to it. This distance depends on the thickness of the layer, but this seems not to be valid for layers thicker than ~2m as is demonstrated by Price and Cosgrove (1990) and Cruikshank and Aydin (1994). It must also be noted that all these observations and relationships have been made for opening mode fractures although Huang and Angelier (1989) have verified the same relationship for shear fractures. The theory can however, not explain the formation of closely spaced fractures observed in fracture swarms. Among other theories, Olson (1989) suggested that the propagation velocity of fractures may cause fracture localization.
The relationship between fracture density and bed thickness is mainly derived from outcrop data and is widely used to estimate fracture densities in the subsurface (Narr and Lerche, 1984; Acquilera, 1988; Narr, 1991; Narr, 1996). Peacock (2006) however,
recognized the difficulties to estimate fracture densities from one well to the other when fractures occur in clusters.
As shown in this small literature review fractures are not simple, planar features but can in fact be very complex and many studies and literature that deals with fractures exist but were not all mentioned here. This thesis tries does not aim to resolve all the complexities involving fracturing but tries to answer some of the issues of fracturing that is observed in the Tanqua-Karoo basin.
1.4 Structure of the Thesis
The Thesis is organized as follows:Chapter 2 gives an overview of the regional geology of SW South Africa, and the Tanqua-Karoo basin in particular. It describes the outcrop and borehole data and the new techniques that were developed to gather outcrop data. Furthermore, it explains why and at what measurement locations the data were collected.
Chapter 3 concentrates on the compressional tectonic features observed in the field and on the borehole images. It describes their general features such as orientation, basin-scale recurrence, sense of movement etc. Secondary features such as fault-related folding, brecciation and cementation are also described, and comparisons are made between the surface and subsurface observations. Finally, a mechanical model is proposed that places the observations in a regional tectonic framework.
Chapter 4 is dedicated to the fracture patterns observed and measured in the field and on the borehole images. It describes the statistical procedures that were performed in order to classify the fractures into sets based on their orientation. Fracture characteristics such as length, height, cementation etc. are summarized. The processing procedures developed to extract fracture parameters from the geo-referenced database are outlined, as well as a new approach for calculating fracture densities from boreholes. The results of both techniques are placed in a tectonic/mechanical framework. As an additional aspect of the observed fracture networks, fracture swarms are separately discussed at the end of this chapter, and their properties such as orientation, basin-scale recurrence pattern, and internal fracture density are described.
Chapter 5 presents the results of two fluid flow simulation studies. One study involves the flow through a fault-related fracture network, while the other is a sensitivity study that tests different fracture properties and their influence on flow behaviour. It also examines the effect of the fracture swarms described in chapter 4, and it shortly discusses discrete fracture network modelling and some more general aspects of fluid flow simulations through fractured media.
Chapter 6 discusses the results of this Thesis and the general conclusions that follow from it. It furthermore contains some recommendation for further studies.
DATA ACQUISITION METHODS
2.1 Field Area
2.1.1 Geography of the Study Area
The area of interest in this study is the Tanqua-Karoo basin which is located ~300 kilometres northeast of Cape Town, South Africa (Figure 2-1). The south-western part of South Africa is geographically characterised by two mountain ranges that form the branches of the Cape Fold Belt. They form the western and southern borders of the large Karoo basin that extends over much of the country (Figure 2-1a). Sedimentary deposits in the southwest Karoo basin comprise the upper Carboniferous glacial Dwyka Group, the shales and sandstones of the Permian Ecca Group and the Triassic fluvial Beaufort Group (Figure 2-2). The two branches of the Cape Fold Belt meet near the town of Ceres, where they form a NE trending basinal high. Located to the SE of this high is the Laingsburg subbasin and to the NW lies the study area, the Tanqua subbasin (Figure 2-1b).
Figure 2-1 a) Map of South Africa and adjoining countries illustrating the extent of the Karoo basin. b) Detailed map of southwest South Africa with the two branches of the Cape Fold Belt and the location of the Tanqua and Laingsburg subbasins (modified from Anderson and Worden, 2004).
The outcrops of the Tanqua-Karoo basin cover an area of about 650 km2
(Figure 2-3). The landscape is characterised by a broad flood plain in the West with steep cliffs in the East overlooking it. The highest point, the Skoorsteenberg, from which the main formation in the area derives its name, rises almost 1200 meters above sea level (for a detailed description of the Skoorsteenberg formation see section 2.1.3). Climatic conditions in the area are arid to semi-arid resulting in good outcrop exposure.
Figure 2-2 Stratigraphical overview of the Karoo supergroup showing the main formations present in the Tanqua-Karoo basin (modified from Bouma & Wickens, 1991).
Figure 2-3 Landsat image of the study area showing the alluvial floodplain in the West, the outcrops of the Skoorsteenberg Formation in the central part. The mountains in the East are formed by younger slope and shelf-edge sediments (Note that the coordinates are UTM).
2.1.2 Regional Geological Setting
During late Carboniferous, basin development in the southwestern part of Gondwana started in response to subduction of the palaeo-Pacific plate underneath the southern edge of Gondwana (Figure 2-4; Johnson, 1991, Ransome and de Wit, 1992; Cole, 1992; Veevers, et al., 1994). It initiated the evolution of a magmatic arc and a retro-arc foreland basin (pre or early Karoo basin) as a consequence of flexural tectonics due to a combination of orogenic and dynamic loading (Figure 2-5; Johnson, 1991; Veevers, et al., 1994; Johnson, 1997; Catuneanu et al., 2002; Catuneanu, et al., 2005). Deflection of the lithospheric plate resulted in three flexural provinces; the foredeep, the forebulge and the back-bulge (Figure 2-5; Catuneanu, 2004b). The foreland basin was covered with large ice sheets that, at the end of the Carboniferous, started to break up and consequently initiated glacial sedimentation in the basin (Visser, 1987). Glacial conditions prevailed into the early Permian and resulted in an up to 800 meter thick succession of glaciogenic sediments that form the Dwyka Group (Figure 2-2).
Figure 2-4 Foreland basin development in southern Gondwana during late Palaeozoic (modified from Ransome and de Wit, 1992).
At the end of the early Permian, ongoing subduction led to large scale shortening which caused folding and thereby initiating development of a mountain range (Cole, 1992; Visser, 1992). Further disintegration of the ice sheets was accompanied by a marine transgression and glacial sedimentation was replaced by widespread deposition of suspended muds (the lower Ecca Group; Wickens, 1992). SW-ward thinning of the muds indicates the presence of an incipient mountain range (Cole, 1992) that, at the time, was still beneath sea level. The mud-rich succession is followed by black carbonaceous shales that indicate deposition under anoxic conditions and reflect a period of tectonic quiescence (Wickens, 1992).
A second major contractional phase during late Permian led to the evolution of a fold-thrust belt, the Cape Fold Belt. In the southwestern part of the Karoo basin the foredeep deepened and fine-grained sheet turbidites and deep-water shales were deposited (the Collingham formation). Associated volcanic ash layers provide evidence for active volcanism, probably in the magmatic arc located to the South
(Cole, 1992; Visser, 1992; Johnson et al., 2001). The Cape Fold Belt evolved in two branches along pre-existing structural basement trends (Hälbich, 1983), forming the western and southern borders of a wide flexural basin (the Karoo basin) that evolved due to lithospheric loading. North-directed shortening in the southern branch is accommodated by E-W trending structures which include bedding-parallel thrusts and North-verging folds. Deformation in western branch is milder (de Beer, 1992) and is expressed by N to NW trending faults and open folds. Slickenslides and lineation indicate the presence of a strike-slip component with a dextral movement. N-directed shortening in the southern branch is believed to have occurred simultaneously with NE-directed shortening in the western branch (de Beer, 1990). The two branches meet in a NE-trending structural domain which is characterized by a complicated interference pattern of faults and folds. A tectonic model of rotating lithospheric microplates is proposed by Ransome and de Wit (1992). It includes clockwise rotation of the crust located to the West and counter-clockwise rotation of the crust located immediately to the East of the NE-trending domain of structural interference.
Figure 2-5 Schematic representation of a retro-arc foreland basin (Catuneanu, 2004b).
The NE-trending structural domain is believed to have caused the initiation of two, more or less equivalent subbasins in the foredeep, the Tanqua basin to the NW and the Laingsburg basin to the SE (Figure 2-1b); de Beer, 1990; 1992; Cole, 1992; Ransome and de Wit, 1992; Visser, 1992; Wickens, 1992; 1994). In both subbasins a stack of fine-grained siliciclastic turbidites were deposited; these are respectively the Skoorsteenberg and the Laingsburg formations. The sediments of the Laingsburg basin are intensely deformed with indications of syn-depositional tectonics. The sediments of the Tanqua basin, however, are only mildly deformed by post-depositional tectonics.
During late Permian times, further thrust-loading caused local, rapid deepening of the foreland basin (Cole, 1992). In both basins, a succession of submarine slope, shelf-edge and shoreface sediments was deposited (upper Ecca group), which can reach thicknesses of up to 1000-1500 meters (Pysklywec and Mitrovica, 1999; Hodgson et al., 2006). These sediments are overlain by the fluvial dominated deposits of the Beaufort group. The entire sedimentary succession reflects a northward progradation of the system (Catuneanu et al., 2005) and a gradual shallowing of the basin, as sedimentation rates exceeded subsidence rates. A
subsequent epeirogenic uplift of South Africa during early Triassic time resulted in an abrupt transition from a depositional to an erosional environment across large areas of the Karoo basin (Cole, 1992; Visser, 1995; Johnson et al., 1997). Uplift is related to coalescence of Pangaea and by middle Triassic times the entire area was sufficiently uplifted to start erosion of the Beaufort group, resulting in a hiatus in the depositional history of the region (Pysklywec and Mitrovica, 1999). During middle Jurassic time, Gondwana started to break-up due to the rise of a very extensive mantle plume, the Karoo mantle plume, which intruded the lithosphere of central Gondwana and caused widespread volcanism in the eastern part of South Africa.
2.1.3 Geology of the Field Area
In study area, the outcrops mainly comprise the sediments of the Skoorsteenberg formation. They consist of a series of four stacked basin-floor fans overlain by an intraslope fan (Bouma and Wickens, 1991; Wickens, 1994; Wickens and Bouma, 2000; Johnson et al., 2001; Hodgson et al., 2006). The fans are organized in a north- and east-ward progradational sequence (Figure 2-6), in which the first three fans reflect basin-ward stepping, fan 4 illustrates basin wide deposition and the intraslope fan reflects back-stepping of the system (Hodgson et al., 2006). These authors suggest that the fans are deposited during low-stand sea level as a fifth-order sequence from a source area interpreted to be located south of the Cape Fold Belt. Based on the mismatch between the petrography of the fans and the nearby formations of the Cape Fold Belt (Scott et al., 2000; van Lente, 2004; Anderson & Worden, 2004), and the frequency of occurrence of the fans which is much higher than the tectonic cycles of the Cape Fold Belt (Goldhammer et al., 2000; Hodgson et al., 2006), their deposition is interpreted to be driven by glacio-eustatic sea-level cycles (Hodgson et al., 2006).
Figure 2-6 Development of the basin floor fans of the Skoorsteenberg Formation (Hodgson et al., 2006).
The individual fans are 20 – 60 meters thick and are separated by thick basinal shale units. The entire formation reaches a thickness of more than 400 meters. Hodgson et al (2006) recognized that the internal distribution of each individual fan displays three development stages, i.e. progradation, aggradation and retrogradation. The fans mainly consist of fine- to very fine-grained sandstones, virtually without any porosity or permeability as a result of cementation and compaction during burial (Johnson, 1991; Cole, 1992; Wickens & Bouma, 2000). They form steep cliffs whereas the interfan shales are more erosive and form slopes (Figure 2-7). As much as 14 different types of lithofacies were distinguished (Johnson et.al., 2001). Here the sediments are divided into three major lithofacies: 1) massive sandstones, structureless and frequently amalgamated; 2) thin-bedded sandstones, parallel or ripple laminated; 3) shales. Palaeocurrent analysis from both outcrop (Wickens, 1994; Hodgson, et al., 2006) as well as borehole data (Luthi et al., 2006) show a redirection of the turbidity current, from an initial eastward in the South to a northward direction in the North, suggesting a topographical obstacle in the East, possibly related to the uplift of the Cape Fold Belt (Hodgson, 2006).
Structural deformation in the area is mild. The stratigraphy dips gently towards the East at an average dip of 4° and is only occasionally disturbed by minor faults, folds and fracture swarms (Figure 2-8). A detailed description of these features is given in chapter 3 and 4. Besides these local areas of tectonic deformation, the rocks show extensive fracturing.
Figure 2-7 The submarine fans form steep cliffs while the more erosive shales form scree slopes (Photo by S.M. Luthi).
Figure 2-8 The area’s major tectonic features include minor thrust faults (a; at Kleine Gemsbok fontein), minor folds (b; at Dadelboom) and fracture swarms (c; at Blesmerrie); locations are shown in figure 2-10.
2.2 Data Acquisition Methods
Fracture data was collected at three scales; the basin-scale with the aid of aerial photographs, the field-scale with outcrop data and the well scale from shallow research wells. Outcrop data was collected using new acquisition techniques which are described in detail in the following section (2.2.1). Borehole data was obtained with various imaging techniques. They are described in section 2.2.2.
2.2.1 Surface Data
The surface data set consists of a set of aerial photographs and comprehensive outcrop measurements. The aerial photographs used in this study have a scale of 1:80.000 and cover the entire field area. Their interpretation not only provided insights into the large-scale fracture pattern but also served as a guide for the field studies, particularly for choosing study sites.
Several tens of outcrop locations were selected for detailed studies and measurements of fracture parameters. These outcrop studies generated a data set containing over 1500 surface measurements. Outcrop locations were selected using several criteria. Where possible, fractures were measured on outcrops located in the vicinity of the boreholes (Figure 2-9). Furthermore, outcrops were chosen over the entire area so that the regional fracture pattern could be studied (Figure 2-10). Fractures were measured in areas located far from deformation zones to distinguish the regional fracture pattern from the fault- or fold- related fractures (Figure 2-11). In addition, several fault and fold zones as well as fracture swarms were selected for detailed study (Figure 2-8). This provides not only insights into the character of the deformation zone itself, e.g. the width of the deformation zone, the amount of brecciation etc, but also its effect on the regional fracture pattern.
In order to study fracturing in sedimentary layers or a sedimentary succession outcrops in which a range of different bed thicknesses is present and outcrops that contain a variety of lithologies were selected. Although more types of lithofacies were described by other authors (Wickens, 1992; 1994; Hodgson, 2006) here, only three main types are distinguished i.e. shales, thin-bedded sandstones and thick-bedded to massive sandstones (Figure 2-12).
Figure 2-9 Photograph panel of outcrops East of borehole NB-2 (with wireline logging operations in progress). FMI image and GR log are overlain onto outcrops of fan 4 (figure courtesy of NOMAD Consortium).
Figure 2-10 Landsat image showing the outcrop locations where fracture measurements were made (black dots) and the locations of seven boreholes (white dots). Also shown are various farming areas.
Figure 2-11 Fracture pattern observed in areas affected by regional deformation (“far-field” fracture pattern). Note the back-pack for scale.
Figure 2-12 Illustration of the three main lithofacies used; shales (a), thin-bedded sandstones (b) and thick-bedded sandstones (c). Note author for scale (upper circle). Photo by S.M. Luthi, author for scale.
In practice, fractures in the shales turned out to be difficult to measure due to erosive character of the shales.
Fracture data is usually obtained using a simple scan-line method, meaning that the fractures are measured along a measuring tape which is placed roughly perpendicularly to the fractures. This method necessitates subjective definition of layers, or groups of layers, that geomechanically behave in a similar fashion (Gillespie et al., 1993). The new acquisition and processing techniques that were developed for this study (Bertotti et al., 2007a & b) not only allows building of a fully digital database, but it does not necessitate a priori definition of mechanical layers. The technique is based on digital outcrop photographs that are georeferenced using a specially developed software package based on GIS principles (Geographical
Information Systems) and installed on a portable tablet PC. The position of the outcrop is determined with a conventional GPS and a reference baseline is marked on the outcrops before it is photographed (Figure 2-13).
Figure 2-13 Positioning of a marked baseline at a georeferenced outcrop site.
The UTM coordinates of a representative point and the length and direction of the baseline are then used to georeference the outcrop photograph. All objects that are now drawn on the screen will be automatically positioned with respect to the reference point. The technique requires subhorizontal strata in a near-vertical outcrop typically a few meters in width and height, i.e. large enough to expose different layers but small enough so that all fractures can be measured within a reasonable amount of time. The method requires two operators; one that operates the software and is guided by the second operator at the outcrop, who marks the beginning and the end of a fracture, measures its orientation and makes observations such as the type of cementation. In order to compare the mechanical stratigraphy obtained with this technique with the sedimentary stratigraphy of the outcrop, a detailed stratigraphic column is drawn of each outcrop.
The technique allows positioning of fractures in space, defining fracture orientation, length and fill, storing all in one large data base. Because the data is georeferenced, information such as bed thickness can be conveniently extracted at a later stage, therefore saving field time. A detailed description of these processing techniques is given in chapter 4.
Whenever field conditions were not suitable for this new technique, the plain scan-line method was used.
2.2.2 Subsurface Data
In the recently completed NOMAD project, a comprehensive data set was collected that included the drilling of seven boreholes at strategic locations with full coring as well as wireline logging (Hodgson et al., 2006; Luthi et al., 2006).
Three large diameter boreholes (~15 cm) were drilled along a N-S line in the West relatively close to the outcrops (Figure 2-10) in which a full logging suite comparable to industry standards was acquired (i.e. electrical image logs, petrophysical logs and geochemical logs, Figure 2-14).
Figure 2-14 A full suite of wireline logs obtained from the large diameter wells. From left to right: (a) Gamma Ray, (b) Fullbore Micro Image, (c) Elemental Capture Spectroscopy, (d) Spectral Gamma Ray, (e) Density Neutron Log and (f) Array Sonic Log. Courtesy of NOMAD Consortium.
These wells are marked with NB in Figure 2-10. Four small diameter wells were drilled along a second line running on the plateau further to the East (~10 cm) in which initially only slim-hole gamma ray logs were run, at a later stage supplemented with slim-hole imaging tools. These wells are marked by NS in Figure 2-10. All wells were fully cored with a total recovery of 1247 meters, corresponding to 97% of the total depth drilled. The lithology of the cores was systematically described at a resolution of 0.5 cm.
The electrical imaging tool (FMI1) works according to the physical principles of the dipmeter (Luthi, 2001) and contains a large number of electrodes such that continuous resistivity images of the borehole wall are obtained. The method is limited to conductive borehole muds, and, although fresh water was used in the boreholes there was sufficient conductivity for the currents to pass into the formation.
This condition of a low-conductivity mud and a high-resistivity formation, in fact, resulted in a high degree of focusing of the current, providing very sharp images that in general respond very well to bedding and fracturing (Figure 2-15a). The tool has a resolution of 1 cm and fractures as thin as 10 µm can be detected (Luthi and Souhaité, 1990).
Figure 2-15 Examples of borehole images obtained in the seven wells. a) Electrical borehole image (FMI) showing a partly cemented fracture; b) optical borehole image (OBI) showing two cemented fractures; c) acoustic borehole image (ABI) showing several open fractures; d) core photograph showing a partly cemented fracture (Courtesy of NOMAD Consortium).
The optical imaging tool (OBI2) generates a continuous image of the borehole wall using a down-hole CCD camera. Its use is limited to clear borehole fluids and under favourable circumstances information on bedding and fracturing can be obtained. The image quality strongly depends on the colour contrasts of the features to be imaged. In our case, the contrast between the white, cemented fractures and the
1 Mark of Schlumberger
grey colour of the formation was sufficiently large to provide a good characterisation of cemented fractures (Figure 2-15b). However, the contrast between the rock matrix and thin open fractures was too small to be detected.
The ultrasonic imaging tool (ABI2) generates an image of the borehole wall by transmitting ultrasound pulses and recording the amplitude and travel time of the signals reflected from the borehole wall. The reflected amplitudes respond to the acoustic properties of the rocks surrounding the borehole, whereas the travel time depends on the borehole shape. The technique is very useful for detecting open fractures because of the scattering and wave conversion associated with them (Figure 2-15c).
Extensive fracturing was observed in the cores and on the borehole image logs, although to various degrees, due to the different nature of the techniques. The electrical borehole images show all types of fractures and are therefore the main source of subsurface fracture data. All fractures that were observed on these images were systematically described and properties such as orientation and height were measured. The small diameter wells primarily provided information on fracture orientation due to limitations of the acoustic and optical imaging techniques. The cores were used to verify the character of the fractures that were interpreted on the borehole images (Figure 2-15d). For example, they showed that many fractures that were interpreted on the FMI images as open fractures (conductive response) appeared to be partially cemented. Because the cores were not oriented their use for structural analysis was found to be limited.
Analysis of all borehole data not only provides accurate geometrical information on the fractures, but also a thorough characterization of the fracture properties. It allows conducting statistical analyses per borehole, lithology, sedimentary unit (fan or interfan) or structural setting.
3.1 Introduction
Although the geology of the Tanqua-Karoo basin has been studied since a long time, these studies concentrated on the sedimentary geology rather than on the structural geology (for an overview, see chapter 2). The sedimentary rocks of the Tanqua-Karoo basin are mildly deformed by contractional tectonics that is associated with to the evolution of the Cape Fold Belt. Shortening is accommodated by thrusting and folding, some of which is clearly expressed in the field and in the boreholes (chapter 2.2). A set of aerial photographs covering the entire outcrop area was interpreted prior to the field work. They show clear lineaments with two distinct orientations, E-W and NW-SE (Figure 3-1).
Figure 3-1 Landsat image with lineaments as interpreted from aerial photographs. The two major river valleys are aligned with the E-W oriented lineaments. Note the rose-diagram in the upper left corner representing the orientation of the lineaments.
Field studies showed that the E-W oriented lineaments coincide with the strike direction of the faults and the orientation of fold axes whereas the NW-SE oriented lineaments are fracture swarms. The latter will be described in chapter 4.
Outcrops, borehole images and cores allow studying of the deformation zones at a high resolution. At a larger scale, aerial photographs outline the basin-scale geometries of the zones. A number of deformation zones were studied in detail in order to analyse the architecture of fault- and fold-zones and their surrounding fracture patterns. The results are described and discussed in the following sections in which the focus initially lies on the general architecture such as orientation, sense of displacement, width of the affected area, amount of shortening, basin scale recurrence etc. Besides the general aspects, secondary features such as brecciation, folding and cementation, are described. These are indicative of the geomechanical behaviour, i.e. the response of various lithologies to the applied stresses. Fault related fractures will be dealt with in chapter 4.
Finally, the contractional features of the area will be placed in the regional geological context and a conceptual geomechanical model will be proposed to explain the field observations.
3.2 General Architecture of Deformation Zones
As mentioned above, contractional deformation in the area is mild and shortening is accommodated by structures such as those shown in figure 2-8a & b. Besides these deformation types, more complex structures are present in which combinations of faulting and folding resulted in ramp-flat structures, fault-related folding and kink-folding (Figure 3-2). In the following sections, the general architecture of the deformation zones studied in the outcrops (section 3.2.1) and in the boreholes (section 3.2.2) is described.
3.2.1 Outcrops
Faults generally dip at low angles (between 25° and 45°) towards the SSE (Figure 3-3a) although the aerial photographs (Figure 3-1) illustrate a slightly radial pattern of lineaments that seem to be irregularly spaced. Striations on the fault planes indicate dip-slip movements with N-vergent directions. Some of the fault planes show two sets of striations, one dip-slip and one sub-horizontal, indicating strike-slip movements during a later, second pulse of deformation. Unfortunately, the sense of shear could not be determined. Along strike, the faults can reach lengths of several kilometres as can be observed on the aerial photographs (Figure 3-1). The width of the affected areas (across strike) is limited; horizontal displacement ranges from 50 to 100 meters. Over the entire field area, which is ~50 kilometres long in the shortening direction, the horizontal displacement accommodates a total amount of shortening that is estimated to be in the order of 5%. Vertical displacements of the faults are small, in the order of 10 to 20 meters. Nevertheless, the faults commonly affect multilayer stratigraphic successions comprising various lithologies, providing good opportunities to study the mechanical response of various lithologies to contractional faulting. Fault zones are usually accompanied by secondary features such as fracturing, brecciation, small-scale (<5m) folding and secondary faulting. These features are described in section 3.3.
Figure 3-2 Photograph of a ramp-flat structure with fault-related folding (a) and a kink-fold (b), both structures are located in the Loskop area (for locations see figure 2-10).
The zones that are affected by folding have similar dimensions as the faulted zones. Fold amplitudes are generally in the order of several meters whereas the wavelength can reach several tens of meters. Some of the folds can be linked to faulting, such as the folded hanging wall of the ramp-flat structure illustrated in figure 3-2a. In other areas, only minor or no indications for faulting were found. Fold axes are sub-horizontal (dipping less than 5°) and trend N80E (Figure 3-3b), parallel to the strike of the thrust faults. Striations on the folded bedding planes indicate flexural slip movements.
