appendix 1: photogrammetry techniques, fieldwork...
TRANSCRIPT
1
Appendix 1: Photogrammetry Techniques, Fieldwork Planning and Outcrop Models
Contents GIS analysis on planning fieldtrip ..................................................................................................................................... 1
Terrestrial digital photogrammetry (TDP) ................................................................................................................... 4
Fieldwork planning ................................................................................................................................................................. 6
Ground resolution .............................................................................................................................................................. 6
Accuracy and precision .................................................................................................................................................... 7
Mapping area and physical/topographic constraints ......................................................................................... 8
Registration approaches .................................................................................................................................................. 9
Differential global positioning system (DGPS) ..................................................................................................... 10
Geodetic principles .......................................................................................................................................................... 11
Differential positioning .................................................................................................................................................. 12
Image processing ................................................................................................................................................................... 13
Features extraction ............................................................................................................................................................... 18
Complex surfaces .............................................................................................................................................................. 19
Joints structures ................................................................................................................................................................ 20
Outcrops overview ................................................................................................................................................................ 22
Carnarvon Highway road cut ....................................................................................................................................... 22
Isla Gorge 1 .......................................................................................................................................................................... 22
Isla Gorge 2 .......................................................................................................................................................................... 22
Cabbagetree Creek ........................................................................................................................................................... 22
Other methods considered: drone and laser scanner ............................................................................................ 22
1
GIS analysis on planning fieldtripDue to the large extent of Precipice Sandstone formation, a preliminary GIS analysis was carried out to develop a targeted field plan for photogrammetric survey in the eastern part of the outcrop line. Several data, available for free at http://qldspatial.information.qld.gov.au/catalogue/, have been used. In particular they are:
� 1 sec Digital Elevation Model (DEM)� Baseline roads and tracks Queensland shapefile � Surface geology 1974 - Surat Basin shapefile � Exploration and production permits shapefile � Queensland Boreholes Web Map Service � Queensland Topographic Web Map Service
The study area has been chosen on the basis of: known exploration permits boundaries, known position of CTSCo lease area and regional geology (distribution of Precipice Sandstone outcrops and regional geological structures). An area placed between Taroom Trough (Mimosa Syncline) and Burunga – Leichhardt Fault System was considered to be the most suitable region to conduct our analysis. It isn`t significantly deformed and geological and geotechnical features can be extrapolated to CTSCo's West Wandoan 1 well site.
In order to find out the steepest areas and locate possible outcrops, slope analysis and contour line generation (with 5m spacing from each other) were performed (Figure 1).
���������� ����������������� �� ����������������� ���
2
Since DEM (digital elevation model) resolution was too low (approximately 30m per pixel), it proved to be not detailed enough for our terrain analysis and, so, unsuitable for recognition of scarps and cliffs. In fact, these topographic features are generally 30/40 meters high and tend to be smoothed by automatic interpolation (Figure 2).
���������� ������������������� ������������ � ��� �� ���� ����!"�������������������������#���$���#��!�
Because no orthophotos were accessible in the public domain (except for a fee), research for adequate areas was conducted using available Web Map Services (WMS). At this stage, several survey locations have been selected especially taking into account ease of access.
Also a literature review helped us to close off different sites on the basis of formation thickness. Martin`s thesis on “Sedimentology of the Precipice Sandstone, Surat Basin, Queensland” (Martin, 1976) provided an excellent documentation of outcrops. A few images have been captured from this paper and georeferenced to improve our knowledge on the outcrop’s line (Figure 3 and Figure 4).
3
�������%�� ������������������� ���������� &������������ � ���
�� ���� ����!"�������������������������#���$���#��!���������������
'��(��
����������� �������������������)*+,-.�����/�������������������� ���� ���
��������/ �����0���� 1�����
4
Initially, three different sites that expose the change in palaeocurrent direction on the eastern margin were selected for reconnaissance work. Later, three short fieldtrips were performed (March, April, June 2015) to explore the outcropping belt and extend the facies mapping eastwards (Figure 5).
�������'����2����������� ����3���������� ���� �������(�������
Terrestrial digital photogrammetry (TDP)One of the main aims of this project is the application of 3D ground based photogrammetric models to capture digital images that can be archived and analysed for sedimentary and structural features that will impact on reservoir properties. The term Photogrammetry, meaning measuring on photographs, has been defined as the science and technique of interpreting and evaluating the form, dimension and position of objects by analysing and measuring images of them. The result of photogrammetric procedures is a precise three-dimensional geometric reconstruction of the object that can be orthogonally projected onto a plane (normally, horizontally or vertically) at a certain scale or visualized in a perspective static or dynamic representation in a computer monitor for further evaluation.
In photogrammetry, images are georeferenced through the determination of the six exterior orientation parameters that describe the original spatial relation existing between the photo and the object coordinate systems at the moment the image was captured. This set of parameters (Figure 6) is called the exterior orientation of the image and consists of three object
5
coordinates of the projection center (X0, Y0, Z0) and three rotation angles around the object coordinates axes defining the spatial orientation of the photographic axis in the object space.
�������,��� (������ ���3�������(����� �� ����� �1����� ����!�)������# � ���4����'-��
A process called resection is used to calculate the camera exterior orientation based on relative or absolute coordinates of a number of object points (the more there are the more the redundancy). A crucial assumption in this process is the principle of collinearity, which states that an object point, the perspective center and an image point on the focal plane of the camera are aligned in a straight line. In order to respect this principle, the interior geometry of the camera during exposure must be known. Indeed, if correction for distortion and other internal parameters is not applied, collinearity will not be respected. The three main elements of interior orientation are the focal length and the x0 and y0 coordinates of the principle point (offset). Other parameters include lens radial (K1, K2, K3… Kn) and tangential (P1, P2) distortion. The interior orientation is determined during the process called “calibration”. To address the aim of the project, the research team used a photographic system accurately calibrated by CSIRO in their laboratories and formed by a Nikon camera body and two fixed lenses. The calibration file (Figure 7) has been provided in order to process correctly the images acquired and obtain in this way the most reliable 3D models.
�������+���� 2��� ������ (�5������������ ���������� ����
6
Photogrammetric software uses an automated matching process to find corresponding image points on pairs of photographs. In order to optimize the matching process, it is important to respect a number of recommendations when taking photographs (Figure 8).
�������6���� 2��� �������/ �7���� ((������ ����
Fieldwork planningBefore undertaking a field campaign using terrestrial remote-sensing techniques, some essential planning components should be considered. These include:
� specifying the resolution (ground point spacing) necessary for the purpose of a project,� specifying the required accuracy and precision,� defining the area to be mapped, taking into account physical/topographic constraints.
Careful planning allows a better understanding of the effect of these components on the subsequent geotechnical discontinuity measurements from 3D models in terms of potential orientation bias, truncation level and observation scale effect.
Ground resolutionIn a photogrammetric project, the focal length of a lens, the pixel size and the distance to an object determine the instantaneous angular field of view (IFOV), i.e. the area covered by one pixel on the ground. The ground point spacing (ground resolution) is slightly different, being
7
dependent on the step size. Step size quantifies the number of pixels used, both horizontally and vertically, to generate one spatial point (typical step sizes range between 4 and 8 pixels). In other words, the ground pixel size (IFOV) should be multiplied by the step size to obtain the ground point spacing.
It should be emphasized that the ground point spacing used in digital photogrammetry surveysis an average value. Because of the object topography, areas closer to the scanner have a closer ground point spacing than the average value whereas distant objects have a wider ground point spacing.
As a general rule, it is suggested that a slightly higher ground resolution than required is used,so that there is enough redundancy to guarantee adequate interpolation of a 3D model surface.
SirovisionTM was the available and recommended system during the early stages of the project. It is a geology/geotechnical mapping and analysis tool that can generate accurate, scaled 3D images, georeferenced using a wide range of surveying equipment. SirovisionTM
provides also a very useful “fieldwork planning tool” that enables the user to estimate the results based on several different criteria (Figure 9).
�������*���� 2��� �������/ �7����������� ��) !&���� 2��� ��2�,������-��
Accuracy and precisionPhotogrammetric stereomodel accuracy and precision depend on calibration, automated stereomatching, ground resolution and network geometry. The expected planimetric accuracy is typically 0.3 pixels ([0.05-0.5]) on the ground assuming an adequate calibration and orientation. The depth accuracy depends on the geometric relationship between the camera positions and the object being photographed according to the following equation:
8
This equation suggests that a large distance/baseline ratio provides a better depth accuracy. However, a large distance/baseline ratio results in difficulties for the image matching process (Figure 10 and Figure 11) causing errors that may show up as “spikes”. Consequently, a ratio situated between 5/1 and 8/1 is recommended.
����������� (������������ ��3������/������(����� ���� ��������3�� �1�����������3 � ����3���)����5�������4����-��
��������8&�� ���������9��3 /�����3���������� ����������������1��������3����(����� ���� ���):���34����*-��
Mapping area and physical/topographic constraintsDepending on the size of the area to be mapped, several sets of photographs will need to be taken. When there are major changes in perspective, such as around a corner, the area should be split into smaller windows. When there is a potential for occlusion and/or orientation bias, separated pairs of photographs should be taken from different angles. Occlusion occurs when parts of a rock face cannot be sampled because it is obscured by protruding features. It may create holes (shadow zones) with missing spatial points in a 3D model (Figure 12). Orientation bias occurs when the camera line-of-sight is sub-parallel to a discontinuity, resulting in a linear trace if viewed from the camera position. Furthermore, obstructions (like tall trees or shrubbery) and occlusions in front of the outcrop spoil the construction of a correct and reliable
9
model. For this reason, future work may require laser scanning technology, in combination with photogrammetry.
���������0��������� �� �� ������ ������2�������� �������� �������)����5�����������������4����*-�
Registration approachesTDP 3D models can be registered in a variety of coordinate systems, including the Universal Transverse Mercator (UTM) Geographic Coordinate System and relative (local) systems oriented with respect to North. The process of projecting 3D models into one of these systems is called “registration” and it can quickly become the most time-consuming part of a terrestrial remote sensing field survey. Depending on the accuracy/precision required for a specific project several approaches can be adopted.
�������%�������� ��2���������������� ������ ��3����;3���<������������������(����� ���� ��������3��������� ��� ��� ������;3�����������(� ������������(������(������ ���� ���)����5�����������������4����*-��
10
These approaches (Figure 13A-D) can be used to register 3D models varying in terms of setup (time), and equipment required (cost). The simplest one (approach A) uses compass clinometer readings allowing registration in a relative reference system oriented with respect to North. The setup is quick, easy and inexpensive. On the other hand, approaches B, C and D use a total station and/or DGPS, which require a longer and more expensive survey. When access to a rock cut is limited, approaches A and B are very convenient.
In order to obtain the required detail for the project, method B was used to observe the outcrops at a distance from the exposure, due to topographic constraints and inaccessibility.Basics on DGPS theory, technique used to record the camera stations coordinates, andchallenges faced during the field operations will be shortly described in the next chapters.
Differential global positioning system (DGPS)By tracking the microwave radio signal that GPS satellites are transmitting continuously, a GPS (Global Positioning System) can calculate the latitude, longitude and height of a receiver. The principle is illustrated in Figure 14. On this figure in two dimensions, three distances from three satellite positions are required to determine a unique receiver position, the equal-distance trace to a fixed point (satellite) being a circle. In three dimensions, to find a unique receiver position, four satellites and four distances are required, the equal-distance trace to a fixed point being a sphere. There is an additional unknown as the receiver clock is usuallydifferent from the GPS clock, consequently, an additional satellite is required, i.e. five in total. However, by using only four satellites, two possible solutions can be obtained, one being in space which can thus be neglected. Consequently, four satellites are required to satisfy four unknowns, the X, Y and Z coordinates and the receiver clock delay. Additional satellites allow redundancy.
����������������� �� ���3�����<��������2���� ���� �������4�7� /�����3���������������� ���� ���)�4��4��%-������3��������������)=4�=�4�=%-�)� (������4����*-��
11
Geodetic principlesThe fact that the topographic surface of the Earth is highly irregular makes it difficult for geodetic calculations (Figure 15). To overcome this problem, geodesists adopt a smooth mathematical surface called an “ellipsoid”, to approximate the Earth surface. In Australia, a common ellipsoid (or horizontal datum) is the Geocentric Datum of Australia (GDA94).The World Geodetic System from 1984 (WGS 84) is another datum usable worldwide. The vertical datum, i.e. the surface of zero height, is often the geoid (equipotential surface, along which the gravity potential is constant). On a global basis, it represents the mean sea level. Across Australia, mean sea level and its onshore realisation, the Australian Height Datum (AHD), correspond to within approximately ±0.5m of the geoid.
The height above or below the geoid is called “orthometric height”. GPS obtained heights are referred to the ellipsoid and are called the “ellipsoidal heights”. AUSGeoid09 is Australia's geoid model for converting ellipsoidal heights to AHD heights and is accurate to 0.03m across most of Australia. Many GPS receivers and software packages have built-in models for automatic conversion between orthometric height and ellipsoidal height.
�������'�>��3 (������3���3��)?-4������� �����3���3��)3-� ���3��� ����#������� ����������� ��)"-)� (������4����*-��
GPS data are measured in a three-dimensional geodetic coordinate system (latitude, longitude, height). Map projection is the transformation of the geodetic coordinates into rectangular grid coordinates (northing, easting, height) or Cartesian coordinates (x, y, z). The projection should minimize the distortion due to transforming an ellipsoidal shape to a flat surface. The Universal Transverse Mercator (UTM) is the more common map projection. The earth ellipsoid is divided into 60 zones, which are projected separately, using a secant cylinder. It is not suitable for the polar areas, where the projection results in significant distortion.
The photogrammetric software solution used for this project needs coordinates to be imported as grid values, i.e. metric values, (Figure 16). Accordingly, GPS 3D geodetic data (latitude, longitude and ellipsoidal height) have been transformed in grid coordinates (MGA 94 zone 55 and 56).
12
�������,� ���������� �2���� ����������3����#��$������ ��� ������
Differential positioningPositioning can be both absolute and relative (differential). Using absolute positioning, i.e. using a single receiver (like commercial handheld GPS, integrated or external camera GPS), the accuracy is in the order of meters or tens of meters, due to the large distance between the satellites and the antenna, the small magnitude of the time increments and other systematic errors (Figure 17). The accuracy also depends on whether the receiver is a single L1 frequency or a dual L1 and L2 frequency GPS. Indeed, the data coming down to the ground are transmitted by the satellites as a complex coded signal modulated by the L1 and L2
frequencies, which, once they`re combined together, permit the correction of many errors and delays enhancing in this way the accuracy.
Using relative (or differential) positioning, i.e. two or more receivers, an accuracy of a few meters to millimetres can be obtained. Usually, one receiver, called the base, is positioned at a known point or left on a point for a long time, so that its position is recorded with accuracy. Another receiver, called the rover, is placed at the position, which is to be surveyed. By simultaneously tracking the same satellites, the base and rover are subject to the same errors and bias. The known position of the base is used to calculate corrections to the GPS derived position and these corrections are subsequently applied to the rover. The shorter the distance between base and rover receivers, the more similar the errors.
Various field measurements methods exist for differential surveys, but, according to technical requirements and budget available, just one has been chosen to perform surveys needed. This is called RTK (Real Time Kinematic) and it merges the information of code and carrier phase observables received at both base and rover receivers and instantaneously compute the precise position on the spot. The RTK method calculates new positions from the old ones, through continuous tracking of the satellites in real time. Consequently, although post-processing is not required, the system and direct line-of-sight between rovers and the base must be initialized.
The survey carried out allowed us to get a series of geographic points with an average 3D quality of about 3 cm.
13
�������+�@;A�(��3 ���� (�����)� (������4����*-��
Survey time at individual outcrops usually spans over a few hours to a day, which precludes from surveying a base station for a longer time. Consequently, the position accuracy of the base station can be limited. To overcome this problem, the base receiver is located at an unknown point and logged for as long as possible period. The shift of the measured values with respect to the true position will be approximately constant for all the receivers: base and rover receivers may not thus be accurately positioned, although correctly oriented. In many cases, such a setting is adequate for geological applications.
Also, it is extremely important try to avoid and remove every possible obstruction (foliage, rock wall closeness), identify easy accesses to set DGPS stations, check for satellite (number of satellites, cutoff angle, GDOP, visibility) and weather forecast (geomagnetic activity) status, take care of baselines layout. All these precautions have significantly reduced the number of sites potentially surveyable for this study.
Image processing Photogrammetry has been captured in three main sites, one along the Leichhardt Highway in Isla Gorge National Park, the second one near Cabbagetree Creek and the last one along Carnarvon Highway.
To completely cover the first outcrop length, 88 pictures have been taken from 31 positions for a total of 44 stereo couples (Figure 18). The average distance from the outcrop is 14 meters. To correctly georeference the model using only camera positions, three other stations have been arranged closer to the outcrop. In this way everything is more spatially constrained. Accuracy on the outcrop surface (ground pixel size) is 1.4 mm.
To survey the second outcrop it has been necessary to take 44 pictures from 15 different positions for a total of 22 stereo couples (Figure 19). The average distance from the outcrop is 20 meters. The geographical constraint represented by the presence of a large creek just in front of the outcrop forced us capturing the images from different distances to the rock face.
14
Proceeding in this way allowed us to obtain an adequate 3D spatial configuration for georeferencing the model. Accuracy on the outcrop surface (ground pixel size) is 1.8 mm.
3D models have been created for every stereo couple thanks to the powerful SirovisionTM
algorithm (Figure 21). If necessary, spikes generated automatically because of perspective errors (outliers) or deformations at image borders have been cleaned (Figure 22). All single models have been then merged together by manually digitizing anchor points between them to form a large unique mosaic.
Since Carnarvon Highway road cut was surveyed using a different approach and processed by means of an alternative software called Agisoft Photoscan. Although it belongs to the family of programs known as “Structure from Motion” (SfM), Photoscan is not a strictly photogrammetric package. SfM relies on algorithms that detect and describe local features for each image and then match those 2D points throughout the multiple images. Using this set of correspondences as input, SfM computes the locations of those interest points in a local coordinate frame (also called model space) and produces a sparse 3D point cloud that represents the geometry/structure of the scene. The camera position and internal camera parameters are also retrieved.
15
�������6�$���33�����?��3/���� ������%��( ����)0����� ����"��� ����#��7-��>���� �������3�������� =�(������+�(��
16
�������*� ����������� ���7�%��( �����;3�������3� ���3�� ���� �����B,�(���
17
Figu
re 2
0: C
arna
rvon
Hig
hway
road
cut o
utcr
op (p
ersp
ectiv
e vi
ew).
The
outc
rop
leng
th is
app
roxi
mat
ely
180
m.�
18
�������������� �� �����( ����2��/����-���=������(��3��-�/������(���-�� ����� ����!�������� =�(������'(��<������
�������������7���������������������������
Features extraction The extraction of both geological and geotechnical features provided a real challenge for the research team (Figure 23). Indeed, SirovisionTM is only able to follow planar properties such as joints, faults and beddings planes for instance. On the contrary ripples, planar stratification (PPS), cross-planar stratification (CPS) are “unfortunately” wavy surfaces, so it`s not possible to interpolate them through complex surfaces for their entire length but only to discretise them in different parts. This workflow has no sense and no scientific validity in order to describe the spatial trend and predict their influence underground.
19
��������%�>2��2��/� ���3���� � ������(�������� ���3�� ����������� ���7�( ������������� 2��� ���������B�,(�� ������*(�3��3�
Complex surfaces The software was not able to extract curvilinear surfaces, making difficult to use to pull out inclined beddings. Hence, non-linear geometries were described as 3D polylines in which every node was accurately snapped to the model surface (Figure 24).
Figure 24: Sedimentary features described spatially by 3D polylines. Outcrop is 70m long.
20
Joints structures Due to the 3 dimensions attributes of the point cloud, photogrammetry allowed to create and extract joint sets out of the models by means of best fitting planes detection.The joint sets have been extrapolated from Cabbagetree Creek model because it represents the most clear and thick outcrop surveyed with photogrammetry, in this case the joint sets result immediate to recognition (Figure 25). The extrapolation and elaboration have been performed with SiroJointTM, plugin of SirovisionTM. In the Cabbagetree Creek model 3 families of joint sets have been recognized. In the summary table these families of joint sets are categorized with colours (Figure 27).The green family is most frequent compared to the other two families (Figure 28). The direction of the three families is consistent with the general structural geology of the area (Figure 26).
��������'�������� ���������3 ���3 /�����3�������� ��1 ������������ ���5�������3�� ���� �� �� ����������� ���7��$����3����,(�����*(�3��3��
21
��������,���1 ������������������(������� (���4������ � ������(����
��������+���((����������� ����(������ ��1 ����������
��������6������ ����/��3�%���(������ ��1 ������������ ���������� ����������� ���7��
22
Outcrops overview Carnarvon Highway road cut Carnarvon HWY road cut outcrop (Figure 20) has been surveyed from both sides of the road, providing full coverage and spatial continuity. Pictures were taken from 117 positions with a 4 meters long baseline (distance between two following camera positions). The acquisition was performed from every camera station according to a fan schema in which every photo has both 50% of horizontal and vertical overlap. The 3D model was georeferenced and scaled by means of 16 markers whose positionswere recorded using a Differential Global Positioning System (DGPS) in RTK (Real Time Kinematic) mode. The model has been generated using Agisoft Photoscan, a SfM (structure from motion) software which exploits the most recent computer vision algorithms in order to reconstruct faithful 3D duplicates.
Isla Gorge 1CTSCo, made available one 3D model for the major road cut outcrop located in proximity of Isla Gorge National Park. The model was not considered suitable for this project due to the poor resolution, low photographic quality and unknown coordinate system used. Every attempt to generate a new model was unsuccessful because of the lack of the original frames. Furthermore, it wasn't possible to acquire new pictures since lush vegetation now covers the outcrop surface.
Isla Gorge 2A second and minor model was built in order to provide a record coming from the same area. Pictures were taken according to a stereo couple configuration, to satisfy SirovisionTM requirements. In particular, just two images with 50% horizontal overlap were acquired from every camera positions. The model, shown in Figure 18, has proved to be suitable for facies analysis and sedimentary mapping, although any geotechnical element was found.
Cabbagetree Creek in Nathan GorgeCabbagetree Creek model has been built according to the geometric configuration required by SirovisionTM. 22 stereo couples have been acquired from 23 camera stations and surveyed with a DGPS in RTK mode. The model itself is 55 meters wide and 11 meters tall at its highest point and clearly exhibits two families of joints trending according the regional structural setting.
Other methods considered: drone and laser scanner
Other methods considered to enhance the high-resolution outcrop imagery include laser scanner and drone photography and photogrammetry.
� The laser scanner or Lidar technique considers the controlled steering of laser beams followed by a distance measurement at every pointing direction. The power of the Laser scanner technique allows a large degree of freedom during the management of resulted point clouds, removing for example vegetation coverage.
23
� A drone was initially planned to be used for mapping large and vertical cliffs.Aerial photogrammetry represents the only way to survey, for example,Carnarvon Gorge National Park because of the presence of high cliffs. Unfortunately the drone was been used due to National Park regulations that limit their use in proximity to the public area (Isla Gorge camping ground).
Appendix 1A: Approach to Facies Analysis and Modelling Contents Facies analysis and stratigraphy adopted .................................................................................. 1
Nested approach (FAKTS database) ............................................................................................ 2
Facies analysis and stratigraphy adoptedTo help in the construction of reservoir flow units in this project, the facies model approach has been used. Although facies models, which provide an overall summary of a particular sedimentary environment (Walker, 1984), are a dated concept, they still form the backbone of reservoir modelling and become more effective with continuous innovation (Howell et al., 2008; Martinius et al, 2014).
Deposits were subdivided into sedimentary units or allounits, namely stratigraphic units bounded by both unconformities and correlative-conformity surfaces (North American Commission on Stratigraphic Nomenclature, 1983). Sedimentary units represent different sedimentary environments (i.e. facies associations). Facies association analysis was adopted since it is a powerful tool to define the depositional history and basin-scale geometry of sedimentary successions (Bianchi et al., 2014; Ghinassi, et al., 2009; Martini et al., 2011, 2013). Allounits are preferred to depositional sequences (Vail et al., 1977), UBSU (Unconformity Bounded Stratigraphic Units, Salvador, 1987) or synthems (Chang, 1975; International Subcommission on Stratigraphic Classification, 1987; Salvador, 1994) because of their wide applicability in heterogeneous, confined basins characterized by erosional, non–depositional and correlative-conformity surfaces, either in marginal or depocentral areas.
The integration of elements as facies units, architectural elements, sub-environments and depositional environments provides the workflow that can be codified into a quantitative database for application to reservoir modelling (for example the Fluvial Architecture Knowledge Transfer System (FAKTS): a database of fluvial-reservoir analogues described in Colombera et al, 2012). The use of these types of database increases information transfer between the real and virtual, and provides a range of scenarios for application within a reservoir, that can be verified through model validation. An overview of FAKTS follows, as it is an option for this style of project, and its different elements can also be used in stratigraphic forward modelling.
�������������� ���������� ���������� ������������������ ��������������������� �!�
In this report, the authors propose a schema (Table 1: facies schemaTable 1, Table 2, Table 3) that groups all the facies recognised in outcrops into a table that shows the relative dimensions as well as spatial and genetic associations that allow them to be grouped or upscaled into adepositional environment (Bianchi et al., 2014; Ghinassi, et al., 2009; Martini et al., 2011, 2013). As an example, a lower and upper allounit subdivision was proposed for the Precipice (and follows general convention), but the upper allounit is characterised by three different sub-environments, all grouped into the coastal depositional environment.
Nested approach (FAKTS database)With the enhancing of the sedimentological knowledge the relation of facies unit to architectural element is no longer one-to-one, but rather different facies units describe one architectural element, in turn different architectural elements define one depositional element. This database comprises data from literature or case study of modern and ancient fluvial successions described in a qualitative and quantitative way, for a total of 111 case studies (Colombera et al, 2012, 2013). The power of this approach is the inclusion of genetic units recognised in stratigraphic and geomorphological realms (������� ). Following the sedimentological approach and the concepts of Miall (1996), the classification has a hierarchy, which starts with the classification of bounding surfaces even if the ranking could increase the difficulties. It follows the classification of depositional elements that in large scale are classified as channel complexes and floodplain (������� �). Architectural element classification corresponds to sub-environments representing the architecture variability. Finally, the facies unit classification is the broadest list of feature and concerns all of the set containing same lithofacies and paleocurrents. The database provides also statistical parameters, as percentage of units and range of dimensions (�������). The geometrical parameters in the database are given along strike downstream. Moreover the catalogue provides filters based on fluvial style, climate and hydraulic regime, in this way the research for the most suitable case is cleverly driven.
Our aim is to use the same nested hierarchical approach in the subdivision of sedimentary features for the Precipice Sandstone. For our purpose we limit the genetic units to be recognised in the stratigraphic record and not only in the fluvial realm. We want to test the extension of this approach to transitional coastal environments since in the literature there is
no evidence of a quantitative facies database for this type of setting. Unfortunately, in the facies schema authors cannot guarantee in dimension parameters presented in the FAKTS database in detail, due to the reduced number of outcrops and the difficulty of access to many vertical cliff faces. For this reason, photogrammetry was used in this analysis.
���������"#��$������%��������������������������������������������� & ���'�������&��������������� � ��������������������� ('�)���������� ������ ��*�����$�����������$������������������ +������������������������������'�
Tabl
e 1:
faci
es sc
hem
a
Cod
eN
ame
Des
crip
tion
Dep
osit
inte
rpre
tatio
nD
imen
sion
s
1
�
,� ����
��� �+
�$$������
������ �
Mas
sive
cla
st-s
uppo
rted
smal
l pe
bble
to g
ranu
les
size
cla
sts
(up
to C
obbl
es) w
ith m
ediu
m s
andy
m
atrix
floo
ring
an e
rosi
ve s
urfa
ce.
Coa
l cla
sts,
logs
and
pla
nt d
ebris
.
Cha
nnel
lag
W~
500
cmT
~ 50
(max
80)
cm
2a
�
���
�����-��
�����
+������
��� +��
����
� ����
Ver
y co
arse
to m
ediu
m s
and
orga
nise
d in
pla
nar-
cros
sst
ratif
icat
ion.
Dip
~ 2
3/25
º. W
ell-
deve
lop
norm
al g
radi
ng. P
lant
de
bris
and
som
e sm
all p
ebbl
es
conc
entra
ted
on th
e to
eof
the
cros
sse
ts.
Fron
tal
accr
etio
n
(tran
sver
se
bar)
W ~
900
cm
T ~
80 (m
ax
150)
cm
2b
�
���
�����-��
���*+������
��� +��
����
� ����
Ver
y co
arse
to m
ediu
m s
and
orga
nise
d in
pla
nar -
cros
sst
ratif
icat
ion.
Dip
~ 1
0/20
º tra
nsve
rse
toth
elo
calf
low
. Wel
l-de
velo
p fin
ing
upw
ard
som
e sm
all
pebb
les
conc
entra
ted
on th
e to
eof
the
cros
sse
ts.
Late
ral
accr
etio
n (s
ide
bar)
W ~
500
0 cm
T ~
180
cm
3
�
���
�����-��
�$���
��$�
�������
���������� ���
�
Med
ium
to v
ery
fine
sand
with
pl
ane
para
llel s
tratif
icat
ion
or v
ery
low
ang
le p
lana
r cro
ss-
stra
tific
atio
n.Se
dim
enta
ryun
itsw
ithsh
arp
base
and
coa
rsen
ing-
upw
ard
grai
nsi
zetre
nd.
Fron
tal
accr
etio
n (lo
ngitu
dina
l ba
r)
W ~
200
cm
T ~
10 c
m (m
ax
20)
4
�
���
�����-��
����
������� +
������
� ����
Ver
y co
arse
to m
ediu
m s
and
orga
nise
d in
trou
gh c
ross
3D
dun
es o
n ac
tive
chan
nel
fill
W ~
130
cm
(m
ax 2
20)
stra
tific
atio
n. P
ocke
ts o
f sm
all
pebb
le. F
inin
g-up
war
d se
quen
ce.
T ~
100
(max
20
0)5
�
.������������
���
����� ����
Het
erol
ithic
bed
ded
sand
and
silt
or
gani
sed
in ri
pple
s an
d pl
ane
para
llel l
amin
atio
n. G
ently
dip
ping
of
10º
tran
sver
seto
the
loca
lflo
w.
Fini
ng-u
pwar
d tre
nd.
Late
ral
accr
etio
n (p
oint
bar
)
W ~
100
0 cm
T ~
300
cm
6
�
)��
+�������
��-��
���� +��
����
� ���
Ver
y co
arse
to m
ediu
m s
and
orga
nise
d in
pla
nar-
cros
sst
ratif
icat
ion.
Wel
l-dev
elop
ed
norm
al g
radi
ng.
Pro
xim
al
crev
asse
spl
ayW
~ 3
00 c
mT
~ 30
cm
7
�
)��
+������
��-��
�$���
���
������
� ����
Med
ium
to v
ery
fine
sand
org
anis
ed
in p
lane
par
alle
l stra
tific
atio
n.S
and-
shee
ts
depo
sits
W ~
200
cm
T ~
5-10
cm
8a
�
/��
����� ��
����
����
� ����
Mas
sive
mud
s an
d si
lt w
ith
disr
upte
d pl
ane
para
llel l
amin
atio
n,
root
s ca
sts,
sid
erite
nodu
les.
Floo
dpla
in w
ith
inci
pien
t pa
leos
oils
W ~
800
0 cm
T
~ 20
0 cm
8b)��
0���������
��������
���
��B
lack
mas
sive
woo
d de
bris
with
br
ight
and
dul
l ban
ds
Mire
dep
osits
W?
T ~
5-10
cm
9
�
)��
+������
��-��
���$
$���
������
����
Rip
pled
fine
san
ds w
ith c
oars
enin
g-to
fini
ng-u
pwar
d gr
ain
size
trend
.D
ista
l cr
evas
se s
play
W ~
100
0 cm
T
~ 10
cm
10
�
,� ����
����
��� ��
Mas
sive
cla
st-s
uppo
rted
gran
ules
flo
orin
g ho
rizon
tal e
rosi
ve s
urfa
ce.
Tran
sgre
ssiv
e la
gW
~ 1
5000
cm
T ~
10 c
m
11
�
/����
�$�
������+�
����������*���+
����
�� ���
�
Pla
ne p
aral
lel-s
tratif
ied
med
ium
to
fine ,
wel
l-sor
ted
sand
. Ver
y lo
w d
ipan
gle
(dip
3-4
º).
Fore
shor
e de
posi
tsW
~ 1
5000
cm
T ~
500
cm
12
�
��� +
����������*���+
����
�� �����
Trou
gh a
nd p
aral
lel c
ross
-stra
tifie
dfin
e, w
ell-s
orte
d sa
nd. 2
D d
unes
. S
pora
dic
pebb
ly p
ocke
ts a
nd
float
ing
mud
cla
sts.
Upp
er
shor
efac
e de
posi
ts
W ~
150
00 c
mT
~ 50
0 cm
14
�
���
����
�+���*���$$�� �
com
bine
d-flo
w ri
pple
sin
fine
-gr
aine
d to
med
ium
san
d w
ith
sym
met
rical
bid
irect
iona
l acc
retio
n in
tabu
lar b
eds.
Ver
tical
bur
row
s (�0������
!.
Low
er
shor
efac
e de
posi
ts
W ~
800
00 c
mT
~ 15
0 cm
15
�
.��
���0&�
���� *
���&�
��� +�
����������
���
'�
Low
ang
le h
umm
ocky
cro
ss-
stra
tific
atio
n an
d sw
aley
cro
ss-
stra
tific
atio
n in
med
ium
san
d (5
0x10
cm
),w
ithin
terc
alat
ions
of
mas
sive
mud
dyla
yers
.
Offs
hore
tra
nsiti
on
depo
sits
W ~
50
cmT
~ 10
cm
16
�
1�����������
������
�����
�P
lane
par
alle
l lam
inat
ed s
ilt a
nd
mud
in ta
bula
r bed
s w
ith v
ertic
al
and
horiz
onta
l bur
row
s (�0������
�an
d����-����
).
Offs
hore
de
posi
tsW
~ 1
0000
cm
T ~
50 c
m
17
�
)��
+������
��-��
��� ���� ��
��
Mas
sive
med
ium
to c
oars
e sa
nds
with
inve
rse
grad
ing.
31º
dip
. P
rese
nce
of b
acks
et b
eds
at th
e to
e.
Deb
ris-fl
ow
depo
sits
on
fore
set o
f G
ilber
t del
ta
(800
x8 m
eter
s)
W ~
100
0 cm
T ~
800
cm
18a
�
���
�����-��
����������
� ����
�������
�
Ero
sion
al s
cour
ed s
ilt a
nd m
ud w
ithtig
htla
min
atio
ns d
rapi
ng th
e ba
sal
eros
iona
lsur
face
.
Aba
ndon
ed
chan
nel i
n m
outh
bar
W ~
200
(max
50
0) c
mT
~ 75
cm
18b
�
���
�����-��
��� ���� ��
��E
rosi
onal
sco
ured
mas
sive
med
ium
sa
nd, w
ell-s
orte
d an
d sp
orad
ic
gran
ules
.
Dis
tribu
tary
ch
anne
lW
~?
(cor
e oc
curre
nce)
T ~2
000
m
19
�
)��
+������
��-��
���� +
���������
� ����
Low
ang
le lo
bes
with
coa
rsen
ing
upw
ard
trend
. Rip
pled
fine
to v
ery
fine
sand
inte
rfing
ered
with
silt
and
or
gani
sed
in lo
bate
d ge
omet
ries
(silt
y at
the
toe)
. Fre
quen
t ver
tical
bu
rrow
s �0������
'�C
oars
enin
g-up
war
dlo
bate
units
Pro
xim
al
mou
th b
ar
W ~
100
0 (m
ax
5000
) cm
T
~ 15
0 (m
ax
250)
cm
20
�
2�������
����
�M
ud d
efor
med
by
soft
defo
rmat
ion
and
sedi
men
tary
load
.D
ista
l mou
th
bar
W ~
200
cm
T
~ 10
cm
21�
��� �����
�*��&���� ����
���
�
Flas
er to
wav
y be
ddin
g of
ver
y fin
e sa
nd in
lam
inat
ed o
rgan
ic-ri
ch m
ud.
Tida
l fla
t de
posi
tsW
~ 1
000
cm
T ~
50 c
m
Table 2: Facies association table with associated sedimentary environments.
Facies code
Facies association code Facies Facies association
Depositional Environment
1 FA1 Channel lag Channel belt
Fluvial environment (FAbr 1/2a/2b/3/4/6/7 FAme 1/4/5/9/8a/8b )
2a Transverse bar
2b Side bar 3 Longitudinal
bar 4 3D bedform
in channel fill 5 Lateral
accretion 6 FA2 2D bedform
in small tributaries
Floodbasin
7 Sand sheet in small tributaries
8a Floodplain and paleosols
8b Peat deposits 9 Crevasse
splay 10 FA3 Transgressive
lag Shoreline Coasts wave-
dominated 11 Foreshore
deposits 12 Upper
shoreface deposits
13 Middle shoreface deposits
14 Lower shoreface deposits
15 Offshore transition deposits
16 Offshore deposits
1 FA4 Channel lag Gilbert Delta
Topset 4 3D bedforms 17 Debris-flow Foreset
20 Soft-deformed mud
18a FA5 Abandoned channel
Shoal water delta
Coast river-dominated
18b Distributary channel
19 Proximal mouth bar
20 Soft-deformed mud
8b Mire 1 Channel lag 8a Paleosols 8b FA6 Mire Tidal
deposits Tidal plain
5 Lateral accretion
8a Paleosols 21 Flaser and
wavy bedding
Tidal flat
Table 3: table showing the occurrence of facies in each outcrop and core considered in this project.
FFacies code OOutcrops CCore Cn FH Fl IG1 IG2 Cr Cn Hgw WW1 GW4 11 X X X X X
22a X X X X
22b X
33 X X X
44 X X X X X X X X
55 X X X
66 X X X X X
77 X X X X X
88a X X X X
88b X X
99 X X
110 X X
111 X X
112 X X X
114 X X
115 X X
116 X X X
117 X X X
118a X X X X X X
118b
119 X X X
220 X X X X X X
221 X X X
Appendix 2: Static Geomodels of the Outcrops
Contents The workflow ............................................................................................................ 2
Resulting Models...................................................................................................... 4
Modelling the outcrop Cabbagetree Creek ..................................................................... 4
Geometries ...................................................................................................................... 4
Sedimentary Fabric Modelling ......................................................................................... 5
Grain Size distribution.................................................................................................... 14
Modelling the outcrop Isla Gorge#2 .............................................................................. 18
Geometries .................................................................................................................... 18
Sedimentary Fabric Modelling ....................................................................................... 18
Grain Size Distribution ................................................................................................... 23
References ....................................................................................................................... 27
The workflow The Precipice Sandstone outcrop modelling has been conducted following the workflow represented below (Figure 1). After a sedimentological characterisation, the outcrop has been imaged by photogrammetry techniques in order to allow the extrapolation of surfaces detected through the facies analysis. This leads to populatinga static model with geometries, fabrics and grain size (Table 1).
�������'�3��0���*�� �����������$��4����
Different properties have been modelled to achieve simulated features. Modelling the outcrop uses just observation data; no interpretation is taken into account at this stage.
��������������������������*����$��$����� ��������������������� '�
Properties Features1. Geometry Units2. Fabric Internal Bedding3. Grain size Sedimentary Facies
The jump to up-scaling the outcrop into the subsurface required an understanding of the effective traceability of outcrop facies and the recognition of these facies in core
and wireline (Table 2). This required the interpretation of observational data, and the extrapolation of geobody surfaces based on their dip, since they were commonly only partially exposed (e.g. lobes, channel belt, etc.). Based on core, the Precipice Sandstone in the Glenhaven area is made of vertical and lateral stacking of geobodiesthat change character. This required a composite interpretation based on observations from all outcrops which variably exposed different parts of the Precipice section. To understand the geometry of the stacked channel belts, the area of dense open file drilling within the APLNG area along strike to the Glenhaven area was used to test the continuity of the geobodies and geobodies groups also called storeys.
����������������������� ��$����*���������������4��� '�
This workflow provides a conceptual framework within which the spatial dimensions of different scale sedimentary features can be measured in outcrop, as well as the nature of their bounding surfaces and internal variability in grain size and bedding that influence reservoir flow units. Ideally, outcrops should have two cuts running parallel to each other (such as the unpublished in-house CTSCo model of the Carnarvon Highway) to achieve the best measurements, but the remoteness and character of the eastern flank outcrops did not have this feature because they are shown only in one cut instead of two cuts. The workflow for the outcrop modelling encompassed three methodologies:
1. Outcrop analogue mapping. This study used outcrop mapping, through an allostratigraphic-sedimentological approach based on facies-analysis principles (more information in the appendix: Facies Analysis). Facies associations and bounding surfaces were characterized via bed by bed logging and outcrop architectural panels, with the aim of identifying associations of spatially and genetically related facies. Two approaches were used to define the bounding surfaces of architectural elements: hand drawn line tracing on scaled 2D photomosaics and georeferenced 3D photogrammetry using Datamine Sirovision and later Maptek I-SiteTM. Within these bounding surfaces, variation in grain size, fabric and sedimentary structures were recorded and used to develop a detailed schema.
2. Photogrammetry. The application of 3D ground based photogrammetric models captures digital images that can be archived and analysed for sedimentary and structural features that can be related to reservoir properties. The result of photogrammetric procedures provides a precise three-dimensional geometric reconstruction of an object at a certain scale, which can then be visualized in perspective either as a static or dynamic (e.g. animation) representation. The assemblage of photogrammetric stereo-couples and the development of photogrammetric models have been completed in Datamine Sirovision, although
Outcrop modelling Regional modellingUnits GeobodiesInternal bedding Internal beddingSedimentary facies Depositional elements
alternative platforms are also being evaluated. For more information regarding the acquisition technique and processing please visit Appendix 1.
� Static geological modelling. Static geological models of the outcrops were created using the bounding surfaces mapped in outcrop to create subunits that were populated by field data such as grain-size and fabric (internal bedding). The approach undertaken for nested geometry and fabric modelling includes building geometries for different units using outcrop measured boundary lines; building a fabric model by multiple-point statistical modelling (MPS); and building grain size distribution by a hierarchical Gaussian random function simulation with fabric distribution as a constraint.
The resulting models are detailed below.
Resulting Models
Modelling the outcrop Cabbagetree Creek
Geometries Models were developed for each of the outcrops. The model size is constrained by the outcrop, rather than being extrapolated to the interpreted size of the lobe geobodies.Deposits cropping out in the Cabbagetree Creek area are interpreted as compensational stacked lobes belonging to a shoal-water delta, based on its low gradient and the presence of bioturbation, wave ripples and climbing ripples. In our modelling, we tried to remain consistent for geometry (geobodies with bounding surfaces), internal sedimentary structures that control bedding (fabric) and grain size distributions (texture).
The boundary lines of units were interpreted based on the surveyed outcrop at Cabbagetree Creek (Figure 2A). This outcrop description is an update of the model already presented by Bianchi et al. (2015b), considering the real dip calculated by photogrammetry. The bounding lines were exported from Datamine Sirovision™ and imported into Petrel2014TM as shown in Figure 2B. The bounding lines are georeferenced and the structural dip was also measured by compass in the field. In order to control the generation of horizons within the units model, the bounding lines were projected 200 m northward away from the outcrop (Figure 2C) with the structural dip angle and azimuth (constraints which are listed in Table 3). All the measured points for the outcrop section are projected using the following equations:
� = �� + tan(�) × 200 (1)� = �� + 200 (2)� = �� + tan(�) × 200 (3)Where X, Y and Z are projected coordinates in m, X0, Y0 and Z0 are coordinates of ���������� ������������ ������������������������ ����������������������������(strike compared to the true north) respectively. Note that the projections are used to
find the locations of those boundary lines at 200 m northward away from the outcrop. We assume the widths of them stay constant when projecting because of the shortage of data.
������('��������������$������������-�������������� ��������������������0'�)���������������$�� ����������� �������*���������$�$��$����� '�
Unit Boundaries ������������ ��������������A 2.4 181.9B 4.0 184.4C 3.6 206.3D 3.1 153.0E 18.7 158.3F 5.8 311.0Bottom 5.7 197.0
Surfaces lines were first generated using the measured points at the outcrop; the surface lines were then projected away from the outcrop according to the measured dip and dip direction. Based on these surfaces, geometries were constructed with x- and y- range of 54 m and 45 m respectively. Results show that units D and F pinch out toward the West and North (Figure 3).
The grid size in the x- and y-direction is 0.1 m and in the vertical is varied. The top eroded surface and units A through D, are layered in proportion according to unit thickness. Units E, F and Bottom are layered with a constant thickness of 0.1 m reflecting their lateral continuity. Figure 4 shows the cell vertical size distribution for each unit. The average cell thickness is 0.09 m with standard deviation of 0.02 m. The number of layers for the Top, A to F and Bottom are 17, 17, 17, 22, 15, 53, 18 and 22, respectively. Finally, the total cell number is 539×452×181=44,096,668.
Sedimentary Fabric Modelling Different interpolation options are available when assigning properties to the grid models created from the bounding surfaces. Sequential indicator simulation (SIS), object modelling (ObjM) and multiple-point statistics (MPS) are used widely for models with categorical variables, such as sedimentary facies, lithology, and flow unit.
SIS is based on spatial variation in a given data property as described by the variogram and is known as two-point geostatistics (Journel, 1983; Journel and Isaaks, 1984; Journel and Alabert, 1988; Deutsch, 2006). The modelled properties have an equiprobable chance to be assigned to a grid, but are guided by the variogram.
ObjM, also known as a marked point process (Holden et al., 1998), is object-based instead of two-point geostatistics. ObjM forces the grid assignment to follow a preordained pattern interpreted from knowledge of the object or unit such as a channel
belt, mouth bar, shoreline, etc. The facies fraction, orientation, amplitude, wavelength, width and thickness for a given object, for example a channel belt, are important inputs for ObjM (Holden et al., 1998; Petrel, 2014).
MPS, a pixel-based sequential simulation algorithm, is capable of producing models with more geological complexity (Strebelle, 2000; Strebelle and Journel, 2001) than SIS and it is also easier for conditioning to hard data than ObjM. In outcrop modelling, MPS was used to guide the assignment of grain size into patterns that reflected the fabrics, e.g. parallel, planar cross and trough cross etc (see Figure 5).
In this study, three relative grain sizes, e.g. coarse, medium and fine, are used to simulate the internal bedding fabric. Then the interactive modelling (hand drawing) method was used to generate three fabric types, e.g. trough cross, parallel and planar cross, which are used as training images in MPS modelling. Those fabric types are a combination of the above-mentioned three relative grain sizes (Figure 5). In these models, grid size is 0.1 m and cell number is 100-10-100 in the x-y-z direction. The grid size and grid number was selected to reflect the average internal bedding size observed in outcrop, and the promote definition in the lateral variation of cross beds and larger scale bounding surfaces. The grid size in the fabric modelling is also considered in selecting the grid size for the training image of internal bedding.
An ellipsoid search mask with radius of 10-3-10 in x-y-z direction and a multi-grid (Gómez-Hernández, 1991; Tran, 1994; Petrel, 2014) with number 3 is used to create training image patterns. The ellipsoid search mask with radius of 10-3-10 in x-y-z- direction will cover 80 cells (=2(3-1)×10×2) in x and z direction and 24 cells (=2(3-
1)×1×2) in y direction (Okabe and Blunt, 2005; Petrel, 2014) which meets the requirement of covering 80% of cells of the training image when creating patterns. Figure 6a is a training image with a channel areal ratio of 26%. This area is gridded into 400 cells. According to the events’ template with four neighbour conditioning data (Figure 6b), the data event number is 24=16; then we can account for the probability of each event globally as shown in Table 4. For example, there are 168 grids are cross connected with four neighbouring grids at event #1 (EV1); 5 of those 168 grids are channel. Figure 7 presents the steps to conduct a MPS with one realisation for a simple case without soft data. This is also the process to take the training images from Figure 5 and populate them throughout each of the depositional elements into Figure 8 which is the generated fabric model.
Based on the bedding inspected from the outcrop, Table 5 lists all parameters used in modelling for the different units. Figure 9 shows that the modelled percentages of relative grain sizes, coarse sandstone, medium sandstone and fine sandstone are 12.8%, 76.3%, and 10.9%, respectively. Note that these values are calculated from the modelled fabric, not from outcrop.
��������������&���������$���!��� ����$��������������&�� ��� ��� ���$� ������������ �� ����������$� ��������1!���� �$��4�������������&����� ������$� ������������ ���!'�)���������������'���������������&����� �����������0������������������������� �$��4������� �������*������������������������ ������������&����� 5���$��-������������$��������������������������0�������$'�
A
B
C
10m
10m
53m
53m
Measured outcrop
Projected boundary lines based on dip angle and azimuth
������� (�2 � ������������ ���� � �� ��(2 �� �����������$��� �� �������� �����1 ������� �$�� �������� ������ �����������& � ������ '�������� ����� ��*�������������������� ������������������$��$�������������������������������0�������$'������ �-��*��������������� �� � ��*������������6�����$��$�����������������*����������������� ����������������7'�)���������������$� ���������������� �������������&��������������� ����������������������$��������&���$������� �$��������$��4������'����������5 ����� ����������&�������������������������$��&���'����
Top
ABC
D
E
F
Units
�������6��������������� �-���� ������������������������ ��*������������('��� ���� ������ ����� �����������������������������&�������0�� �*���������������'�
36 90
90
0.12
0.12 0.09 0.09
0.09
4
8
12
16
20
24
0.040 0.08
4
8
12
16
20
24
0 0.03 0.06
4
8
12
0 0.03 0.06
4
8
12
16
0.030 0.06 0.12
4
8
12
16
20
24
28
32
0.040 0.08 0.12
60
30
0.040 0.08 0.12 0.16
60
30
0
0
0 0 0
0 0.04 0.08 0.120 0.04 0.08
8
16
24
32
40
48
00
0
Perc
enta
ge, %
Cell height, m
Top A B
C D E
F Bottom
�������7���� �������� ���������&��������$������ ������!�$����������1!�$��������� �������!����������� � ���������������&����������������������'�8������ ��* ������� ���������������������������� �-������������� �����������������������������������$������ '���� ��$������ �*����� ����������� ����������������$&���� 9��������������'�8������:����������� �-�;�*� � �����������������$������������������������ ���������������������������$'�
�������<������������������*�������0������������������!���������������� �<����������� �*�����������������������������!'������������ =�������>�����0=�����$������?���������'��� <!'��
A B
C
10m
10m
1m
10m
1m1m
1m
Coarse grain
Medium grain
Fine grain
Relative grain size
1 1 1 1 9 4 5 1 1 1 1 3 9 4 5 1 1 1 1 11 1 1 3 9 14 5 1 1 1 1 9 12 8 5 1 1 1 1 11 1 1 9 12 8 5 1 1 1 9 9 8 8 1 1 1 1 1 41 1 3 9 8 8 1 1 1 9 9 8 8 1 1 1 1 1 9 11 1 9 7 8 1 1 1 9 9 8 8 1 1 1 1 1 3 4 81 1 9 8 5 1 1 9 9 14 8 1 1 1 1 1 1 9 7 51 9 7 8 1 1 9 9 13 14 11 1 1 1 1 1 3 9 8 53 9 8 5 1 9 9 8 13 12 11 11 1 1 1 1 9 2 8 19 2 8 1 9 9 8 8 3 12 16 5 5 1 1 9 4 8 1 11 8 1 9 9 8 8 1 3 12 8 8 1 1 3 9 8 5 1 12 1 9 9 8 8 1 1 9 2 8 1 1 1 9 2 8 1 1 41 9 9 8 8 1 1 9 4 8 1 1 1 9 1 8 1 1 9 49 9 8 8 1 1 9 9 8 5 1 1 9 4 14 1 1 3 9 83 8 8 1 1 9 3 8 8 1 1 3 9 13 5 5 1 9 7 82 2 1 1 3 4 8 2 1 1 4 9 2 8 2 1 9 9 8 51 1 1 1 9 7 5 1 1 9 9 5 8 1 1 9 3 8 8 11 1 1 3 9 8 5 1 9 3 8 8 1 1 9 4 8 2 1 11 1 4 9 7 8 1 9 1 8 2 1 1 9 9 8 5 1 1 11 9 9 10 8 5 9 4 8 1 1 1 9 3 8 8 1 1 1 13 3 8 8 2 3 3 8 5 1 1 3 1 8 2 1 1 1 1 1
(b)
(a)
uu1
u2u3
u4
������6������ ��������������������$���������&������������������$���������&��?���������'�� <!'�
Data event
Grid numbers Event probability, %
Channel probability, %Channel Floodplain
EV1 5 163 42 3.0EV2 5 9 3.5 35.7EV3 7 16 5.75 30.4EV4 10 4 3.5 71.4EV5 3 19 13.6 9.1EV6 0 0 0 0.0EV7 6 0 1.5 100.0EV8 30 37 16.75 44.8EV9 26 43 17.25 37.7EV10 1 0 0.25 100.0EV11 1 2 0.75 33.3EV12 5 0 1.25 100.0EV13 2 1 0.75 66.7EV14 3 1 1 75.0EV15 0 0 0 0.0EV16 1 0 0.25 100.0
�
�������@����*��������� ��*��������$���� � ������� ��$���,/���?���������'�� <!'�
������7�/��������&$������������$���������������������������� �-�������������������� ������������������������������0'�
Units Sedimentary Structure or Fabric type
Percentage of Relative Grain Size, %Coarse Grain
Medium Grain
Fine Grain
Top Parallel 10 70 20A Parallel 10 80 10B Planar cross 10 80 10C Parallel 10 45 45D Parallel 5 5 90E Trough cross 10 80 10F Trough cross 80 15 5Bottom Parallel 10 70 20
�
Scan training image to build cumulative prior distribution function (cpdf) for data events
Define random path; uniform random probability; to visit all unsampled grids
Is this grid match any conditional event?
Yes
Draw one of the facies from variable’s uniform
random probability for the visiting grid
Draw data event based on randomly generated uniform random probability
No
Conflict with conditional data?
No
Yes
Assign facies for neighbouring grids
Seed number
�������A�B
�������������� �-���� ����������*��������
�����������������$'���� �����������&� ��
* �������������������� �-���� �����
����������
����$����������������
������'�
Facie
sRe
lativ
e gr
ain
size
Coar
se g
rain
Med
ium
gra
in
Fine
gra
in
�������C�B�������������� �-��$��$������ �������������� ��*������������A'�
Grain Size distribution The grain size and its sorting are related to reservoir properties of sandstone, especially porosity and permeability (Chapuis and Aubertin, 2003). In this study, the grain size distributions for different units are listed in Table 6. These values are estimated based on the outcrop observed grain size.
Based on Table 6 listed parameters, hierarchical Gaussian random function simulation is used to generate the distribution of grain size by using fabric as constraints. An omnidirectional variogram with horizontal range of 23 m which is half size of the model and vertical range of 0.2 m was used (for minor sedimentary fabrics, there are only several cells have the same fabric in vertical direction as shown in Figure 5); standard deviations of grain size were assumed as 0.07 mm which is nearly equal to the minimum average grain size of the fabric (0.05 mm) and the rough estimation of standard deviation by the experience equation of (max-min)/4 because of the shortage of statistical data.
Figure 10 shows one realisation of real grain size distribution and Figure 11 shows the histogram of real grain size in the model. Results show that the real grain size distributions for all units except unit C have two peaks.
Coarse grain
Medium grain
Fine grain
Estimated relative grain size proportions, %
������<�D����� �-�� ������������������������� �������������'�
Units Relative grain size
Real grain size, mmmin max mean
Top Coarse grain 0.5 1 0.6Medium grain 0.25 0.5 0.27Fine grain 0.125 0.25 0.2
A Coarse grain 0.5 1 0.52Medium grain 0.25 0.5 0.3Fine grain 0.125 0.25 0.23
B Coarse grain 0.5 1 0.55Medium grain 0.25 0.5 0.3Fine grain 0.125 0.25 0.23
C Coarse grain 0.125 0.25 0.13Medium grain 0.062 0.125 0.065Fine grain 0.0078 0.062 0.05
D Coarse grain 0.5 1 0.55Medium grain 0.25 0.5 0.27Fine grain 0.125 0.25 0.13
E Coarse grain 0.5 1 0.55Medium grain 0.25 0.5 0.3Fine grain 0.125 0.25 0.23
F Coarse grain 0.5 1 0.9Medium grain 0.25 0.5 0.45Fine grain 0.125 0.25 0.22
Bottom Coarse grain 1 0.5 0.6Medium grain 0.5 0.25 0.26Fine grain 0.25 0.125 0.2
������� �D��
���������
��������� �-���� �������������(2��&��������������D�� ����������
���������� ���������������������
����
���������0�������$'�)
�����������
���������*��������
���������������� �� ���
�� ����� ��� ����������
���7��������
������������� �-���� ����������� ����������� ��� �������������<'
Grai
n siz
e, m
mRe
al g
rain
size
, mm
��������.� ������ ��������������� �-�������������������� �*����������������$'�
48
12162024
0
Top
C
F
A B
D E
Bottom
2832364044
0
4
8
12
16
20
24
0.2 0.4 0.6 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
0.2 0.4 0.6 0.80.2 0.4 0.6 0.80.04 0.12 0.20
0.15 0.45 0.75 0.2 0.4 0.6 0.8
0
4
8
12
16
20
24
0
4
8
12
16
0
4
8
12
16
20
0
4
8
12
16
20
24
0
4
8
12
16
0
4
8
12
16
20
24
28
Perc
enta
ge, %
Grain size, mm
Modelling the outcrop Isla Gorge#2
Geometries
Deposits present in the Isla Gorge 2 area are interpreted to be fluvial, presenting as a multichannel non-sinuous belt. The style is consistent with a braided ephemeral stream (Bianchi et al., 2015, Milestone 1 report).
The Isla Gorge #2 section is about 5 m in height and 62 m in length. The process for geometry modelling for Isla Gorge #2 is similar to that for Cabbagetree Creek. Figure 12 shows the surveyed outcrop for Isla Gorge #2, the interpolated boundary lines of depositional units and parallel extended boundary lines.
Based on these boundary lines, horizons on top of units #Top, A, B, C and D were constructed first; then sub-horizons are generated on tops for units # B2 to B7, and D2 to D10. The grid size in x- and y-direction is 0.5 m and in the vertical is varied. The grid size is selected by considering the computing time, the modelling area and the heterogeneity/ geometry of the fabric. Figure 13 shows the generated units with ranges of 62 m and 55 m in x- and y- direction.
Figure 14 shows the cell vertical size distribution for each unit. The average cell height is about 0.08 m with standard deviation of 0.04 m. Finally, the total cell number is 129×117×132=1,992,276.
Sedimentary Fabric Modelling For Isla Gorge #2, the types of fabric are same as Cabbagetree Creek. Three fabrics, coarse grain, medium grain and fine grain, are used as basic fabric elements. Training images including those three fabrics are shown in Figure 5. Based on these image patterns, MPS is used to generate the fabric models. According to the bedding inspected from the outcrop, Table 7 lists all the parameters used in modelling for the different units. Figure 15 presents the vertical fabricproportions in the model. Results show that the lower section has more fine grains than the upper section. Figure 16 shows the modelled fabric distribution. In this model, the percentages of coarse grain, medium grain and fine grain are 14.5%, 69.0% and 16.4% respectively based on the geological model.
������� ��� ����&������ ���$�� ��� E ���D �����F��� �!��� ����$��������� �����&�� ��� ��� ��� $� ����������� � �� 1!���� ��#��������������&����� ������$� ������������ ���!'�)���������������'��������������$���� ������������&����� �����$��4��������� ��������� ����� ��*�&����������������$�����*����-������$������������ ��*������������$�����������������'���� ��$��4���������� ��������&�� ������������������ ���������������'�
C
B4.7m
A62m
OutcropMeasured outcrop
Parallel Projected boundary lines azimuth
�������(�2� ����������������� ����(2�������������$������������������1�����'�������� ����� ��*�������������������� ������������������$��$���������������E ���D�����������$'������ �-��*����������� ��*������������6�����$��$�����������������*����������������� ����������������@'�
�
�������6��������������� �-���� ������������������������ ��*������������('�
UnitsPe
rcen
tage
, %
Cell height, m
�
������@��������������������������� ���������������������������������� ��������������E ���D�����F�'�
Units Fabrics Percentage, %Coarse grain
Medium grain
Fine grain
Top Trough cross stratification 10 30 60A Trough cross stratification 10 80 10B Trough cross stratification 10 80 10C Trough cross stratification 10 80 10D1-D8 Planar parallel stratification 5 15 80D9-D10 Planar parallel stratification 50 30 20
�������7�B�������������� �-��$��$������ �������������'�
Coarse grain
Medium grain
Fine grain
Estimated relative grain size proportions, %
�������<������� �����
����������
������������� �-��*������������������� ����
�����������
��������������&�� �������
�E ���D�����������$'��
Rela
tive
grai
n siz
e
Coar
se g
rain
Med
ium
gra
in
Fine
gra
in
Grain Size Distribution
The grain size distributions for different units of Isla Gorge #2 are listed in Table 8.These values are measured based on the outcrop observed grain size and the Wentworth Scale of siliciclastic grain-size (Wentworth, 1922). Based on the parameters listed in Table 8, a hierarchical Gaussian random function simulation is used to generate the distribution of grain size by using measurement of outcrop fabric as a constraint. An omnidirectional variogram with a horizontal range of 23 m and avertical range of 0.2 m was used; standard deviations of grain size were assumed as 0.07 mm because of the shortage of data. Figure 17 shows one generated realisation of grain size distribution and Figure 18 shows the histogram of grain size in the model.
������A�D����� �-�� ������������������������� �������������'�)�������������� �-�� &���� ��������0������������������� �-��������������������������3���*�����C��!'�
Units Relative grain size Real grain size, mmmin max mean
Top Coarse grain (ms) 0.25 0.5 0.3Medium grain (fs) 0.125 0.25 0.13Fine grain (vfs) 0.0625 0.125 0.12
A Medium grain (vcs) 1 2 1.5Fine grain (cs) 0.5 1 0.8Coarse grain (sPb) 2 64 3
B Coarse grain (vcs) 1 2 1.8Medium grain (cs) 0.5 1 0.8Fine grain (ms) 0.25 0.5 0.4
C Medium grain (vcs) 1 2 1.5Fine grain (cs) 0.5 1 0.8Coarse grain (sPb) 2 64 3
D1-D8 Coarse grain (vfs) 0.0625 0.125 0.1Medium grain (silt) 0.0156 0.031 0.02Fine grain (mud) 0 0.0039 0.003
D9-D10 Coarse grain (vfs) 0.0625 0.125 0.12Medium grain (silt) 0.0156 0.031 0.029Fine grain (mud) 0 0.0039 0.003
Compared to Cabbagetree Creek results of simpler design due to the characteristic downstream dipping facies, Isla Gorge 2 has more architectural complexity, exhibitingthe intense “cut and fill” behavior of a braided plain. Nonetheless, mouth bar deposits of Cabbagetree Creek belong to a fluvial-dominated deltaic shallow marine system, which means that mostly of the deposits can be similar to the pure fluvial facies. The difference in grain size distribution between Cabbagetree Creek and Isla Gorge 2 is because the grid size and input grain size are different though they are built with the
same training images. The grid size Cabbagetree Creek is 0.1 m and Isla Gorge is 0.5 m which makes the beddings in Isla Gorge coarser than Cabbagetree Creek. The input coarser grain size for units A, B and C of Isla Gorge 2 leads a coarser grain size distribution for those units.
�������@�D��
���������
��������� �-���� �������������(2��&��������������D�� ����������
���������� ���������������������������������� ���
����&�� ������
��E ���D�����������$'
Grai
n siz
e, m
mRe
al g
rain
size
, mm
26
�������A�.� ������ ��������������� �-�������������������� �*����������������$'�
��
Perc
enta
ge, %
Grain size, mm
Top A B1
C D1 D9
14
12
10
8
6
4
2
00.15 0.30 0.450
8
16
24
32
40
0
48
8
16
24
32
40
00.8 1.6 2.4 3.2 0.3 0.6 0.9 1.2 1.5 1.8
8
0
16
24
32
40
0.8 1.6 2.4 3.2 0.02 0.04 0.06 0.08 0.10 0.12
4
0
20
16
12
8
24
28
5
0
10
15
20
25
30
0.02 0.04 0.06 0.08 0.10 0.12
27
References Ashraf, M., 2014. Geological storage of CO2: Heterogeneity impact on the behavior of pressure. International Journal of Greenhouse Gas Control 28, 356-368.
Bachu, S., 2008. CO2 storage in geological media: Role, means, status and barriers to deployment. Prog Energ Combust 34, 254-273.
Cavanagh, A.J. and Haszeldine, R.S., 2014. The Sleipner storage site: Capillary flow modeling of a layered CO2 plume requires fractured shale barriers within the Utsira Formation. International Journal of Greenhouse Gas Control 21, 101-112.
Deutsch, C.V. and Journel, A.G., 1998. GSLIB Geostatistical Software Library and User's Guide. Second Edition.
Georgi, D.T., Bergren, P.A. and Devier, C.A., 1997. Plug gamma ray: Key to formation evaluation. Poster presentation at the 1997 SCA International Symposium, Calgary, 8-10 September. SCA-9732.
Martin, K., 1976. Sedimentology Of The Precipice Sandstone, Surat Basin, Queensland. Unpublished PhD Thesis, The University of Queensland. 224 pp plus appendices.
Martinsen, O.J., Ryseth, A., Helland-Hansen, W., Flesche, H., Torkildsen, G. and Idil, S., 1999. Stratigraphic base level and fluvial architecture, Ericson Sandstone (Campanian), Rock Springs Uplift, W. Wyoming, USA. Sedimentology, 46, 235–260.
OGIA model, 2014. Office of Groundwater Impact Assessment, Queensland, 2014.
Petrel, 2014. Schlumberger, user manual.
Chapuis, R.P., Aubertin, M., 2003. Predicting the coefficient of permeability of soils using the Kozeny-Carman equation. EPM-RT-2003-03, 1-30. http://www.polymtl.ca/biblio/epmrt/rapports/rt2003-03.pdf.
Deutsch, C.V., 2006. A sequential indicator simulation program for categorical variables with point and block data: BlockSIS. Computers & Geosciences 32, 1669-1681.
Gómez-Hernández, J., 1991. A stochastic approach to the simulation of block fields conductivity upon data measured at a smaller scale. Ph.D. thesis, Stanford University, Stanford.
Journel, A.G., 1983. Nonparametric estimation of spatial distribution. Mathematical Geology 15 (3), 445-468.
Journel, A.G. and Alabert, F.G., 1988. Focusing on spatial connectivity of extreme valued attributes: stochastic indicator models of reservoir heterogeneities, SPE Paper 18324.
28
Journel, A.G. and Isaaks, E.H., 1984. Conditional indicator simulation: application to a Saskatchewan uranium deposit. Mathematical Geology 16 (7), 685-718.
Holden, L., Hauge, R., Skare, O. and Skorstad, A., 1998. Modeling of fluvial reservoirs with object models. Mathematical Geology 30, 473-496.
Okabe, H. and Blunt, M.J., 2005. Pore space reconstruction using multiple-point statistics. Journal of Petroleum Science and Engineering, 46, 121-137.
Strebelle, S.B., 2000. Conditional simulation of complex geological structures using multiple-point statistics. Mathematical Geology 34, 1–22.
Strebelle, S.B. and Journel, A.G., 2001. Reservoir modeling using multiple-point statistics. SPE paper 71324 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, 1–11.
Tran, T., 1994. Improving variogram reproduction on dense simulation grids. Computers & Geosciences 20, 1161–1168.
Wentworth, C.K., 1922. A scale of grade and class terms for clastic sediments. The Journal of Geology, 30(5): 377-392.
Zhou, F., Shields, D., Tyson, S. and Esterle J., 2016. Nested approaches to modelling swamp and fluvial channel distribution in the Upper Juandah Member of the Walloon Coal Measures, Surat Basin. APPEA Journal, in press.
0
Appendix 3: Subsurface Modelling Of Lithologies Within The Precipice Sandstone, A Case Study From Densely Drilled Well Data
Contents
Background .............................................................................................................. 1
Data ........................................................................................................................... 1
Comparison of wireline and core............................................................................ 2
Modelling .................................................................................................................. 5
Structural framework ...................................................................................................... 5
Correlation approach and relation to outcrop observations........................................ 6
Petrophysical upscaling ................................................................................................. 7
Variogram Analysis ......................................................................................................... 9
Modelled Distribution of GR ......................................................................................... 11
Volume rendering of the different lithologies...................................................... 17
Comparison of subsurface model with outcrop models .................................... 23
Discussion and future work .................................................................................. 24
References.............................................................................................................. 25
1
SUBSURFACE MODELLING OF LITHOLOGIES WITHIN THE PRECIPICE SANDSTONE, A CASE STUDY FROM DENSELY DRILLED WELL DATA
Background
The ultimate goal in subsurface static reservoir modelling is an accurate realisation of the heterogeneity likely to be encountered that will influence flow unit behaviour during the injection of CO2. The spatial heterogeneity of porous and permeable versus impervious lithologies, along with pressure and stress, will affect the flow pathways for CO2 (Dance, 2013; Issautier et al., 2013; Cavanagh and Haszeldine, 2014). These reservoir properties are related to the internal architecture of a reservoir, in particular the lateral continuity of bounding surfaces, sedimentary bedding, grain texture and fabric, often referred to as lithofacies. In the outcrop studies, the channel and lobe stacking patterns were complex and the actual size and extent of a complete geobody was difficult to determine because their exposure is limited compared to their actual size. In the absence of 3D seismic survey, a densely drilled area can be used to correlate bounding surfaces and model the distribution of lithologies that comprise the different geobodies.
This report presents a case study for modelling the distribution of lithologies within the Lower Precipice Sandstone (considered the main reservoir) using Gamma Ray (GR) wireline logs as a proxy. This wireline log was used as it was available in all wells and able to be normalized for the interpretation of lithologies by simple cut off values. Correlation of the wireline cut off values to lithofacies interpretations were validated from previous work on core in the Precipice Sandstone and overlying units (Bianchi et al, 2015b and c). Although simple in its approach, this subsurface model was used to test the insights gained from outcrop facies studies within the Precipice Sandstone (Bianchi et al, 2015b and c), to develop the geostatistics that influence the assignment of rock and reservoir properties into a grid model, and to determine the gaps needed to improve conceptualization and subsurface modelling of sedimentary geobodies for the Glenhaven site.
Data
A wireline dataset from the densely drilled Durham Ranch and Spring Gully field area (Asia Pacific Liquified Natural Gas or APLNG project) was used to develop a preliminary workflow for lithofacies modelling in the Precipice Sandstone. This area is along depositional strike from the less densely drilled EQP7, the exploration license area where injection for a CO2 storage project in Precipice Sandstone is proposed (Figure 1). In the absence of a large scale 3D seismic volume, the densely drilled APLNG area, with an approximate well spacing of 1 km, can provide insight to the
2
distribution and connectivity of lithologies, in particular the fine grained siltstone and mudstone (high gamma) units that potentially form baffles.
������� � �� ������ ��� ����� ����� ����� ��� ���� ������ ������ ���� ���� �� ������� ���� ������ ����� ��� ��� �� ������� ���� ������������� ���������������� �!���� "���������� ���������������� "������������������ �#� ������������$#��������������������� ������������ �#� ����������������������!��������������#��������
Comparison of wireline and core
Figure 2 shows the comparison of GR and Density logs (DEN) with core observed lithology and measured porosity and permeability for the West Wandoan 1 well. Porosity (67 points), permeability (51 points), and overburden permeability (27 points) for well West Wandoan 1 were collected from the milestone report by Golab et al. (2015). If the comparisons between these data sets are good, then the wireline logs can be used to estimate the lithologies, but more importantly their reservoir character that is important to dynamic modelling. The more detailed sedimentary facies are also shown to remind the reader of the variability in the architectural elements or geobodies (i.e. large scale bars and sand sheets within the main belt of a braided river, point bars within a meandering river, lobate mouth bars and so on) interpreted within the Precipice Sandstone. In future subsurface models, these geobodies can be modelled as objects or controlled through the application of multi point statistical methods, but for this model, the intent was to see how well probabilistic modelling within different subunits could, if at all, define the position and size of the trunk channel and the distribution of the finer grained units within them.
3
����
���%
�&��
#����
����
������
����
���'
�����
�(��
��#��
#����
����
����
����
����
�����
���
������
����
���
����
�#�
���
����
����
����
��(�
�����
����
������
�����
���
����)
����)
����
���
�
4
From core observation, there are mainly three basic lithologies in the Lower Precipice Sandstone that are easily discriminated: quartzose coarse sand, quartzose fine sand and fine lithic sandstone with mudstone interlaminations (dirty sand+mud). Results show that it is easy to discriminate the different lithotypes using GR for the Lower Precipice Sandstone. However, the thin clean fine sand at the top of the Lower Precipice Sandstone, which is interpreted to have some tidal reworking from outcrop observations, is too thin to identify in wireline. Similarly, the quartzose conglomeratelayers are also not recognisable in wireline. Hence the preliminary model for the LowerPrecipice Sandstone was kept simple and modelled using gamma cut off values.
Figure 3 shows a good relationship of porosity with GR and DEN. If GR less than 40 API, then the porosity of most samples except for one is greater than 16.5%; whereas the porosity of most samples is less than 11% for GR greater than 80 API. If porosity is less than 11%, the permeability is less than 0.1 mD as shown in Figure 4. These correlations suggest that simple wireline logs can be used as a first pass estimation of reservoir character, and that discrimination of different lithologies and their geometry (which can also reflect geobodies) within a reservoir unit can be used to model spatial heterogeneity of permeability.
�������*�+��������+,-�����������#���������������������������./����������#���������!����������)����)��������
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
��������'�0'�1�*2�*3
040
8012
016
020
0
040
80120
160200
./
'�1�4
�53
Symbol legend
Porosity, % vs. GR (68456) GR vs Porosity
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
��������'�0'�1�*2�*3
1.9
22.
12.
22.
32.
42.
52.
62.
72.
8
1.92
2.12.2
2.32.4
2.52.6
2.72.8
+,-
'�1�2
�*3
Symbol legend
Porosity, % vs. DEN (68456) DEN vs Porosity
Porosity(%)=211.85-81.93·DENCorrelation coefficient: -0.905628Covariance: -0.576573
(a)
(b)
Porosity(%)=28.05-0.168·GRCorrelation coefficient: -0.84Covariance: -281.02
5
�������6������������������#�����!��������������)����)��������
Modelling
Structural framework
Six well correlation sections (example section in Figure 5 and full suite in Appendix; Locations are shown in Figure 1) were used to quality control (QC) stratigraphic well tops provided through the Queensland Petroleum Exploration Database (QPED July 2015) and the Office of Groundwater Impact Assessment (OGIA unpublished 2014), and to correlate three allostratigraphic subdivisions within the Lower Precipice Sandstone: the-Meandering unit, the Blocky unit and the Basal Blocky unit. Note that the Basal unit is not always present in every well. These well tops were used in building an unfaulted static geological model (Figure 6). Well spacing varied and the surfaces were modelled using a 100 m by 100 m x- and y- grid. Normally, maximum cell dimensions are dictated by the minimum sizes of the features to be resolved. In this study, the selection of cells size of 100 m in x- and y- directions is based on well spacing and computing time because of the 250 layers in the vertical direction. After gridding, the convergent gridder method (Petrel, 2014) was used firstly to build the surfaces on the tops of the Lower Precipice Sandstone and the Bowen Basin Unconformity (Buncon) constrained by the picks from the well logs of 89 wells as shown in Figure 1b. Then the allostratigraphic surfaces were modelled proportionally into 120 (meandering), 100 (blocky) and 30 (basal blocky) vertical layers, within which yields an average cell height of 0.3 m for each of the units.
y = 3E-05e0.7782x
R² = 0.9287
0.001
0.01
0.1
1
10
100
1000
10000
0 5 10 15 20 25 30
Perm
eabi
lity,
mD
Porosity, %
6
Correlation approach and relation to outcrop observations
The three subunits were recognized – the Meandering, Blocky and Basal Blocky - onthe basis of different GR trends bounded by higher GR layers (Figure 5). The high GR layers correspond to fine-grained layers that can be widely recognized in the area, associated with overbank/floodplain deposits. The subdivisions, that define a multi-story channel system, record the stratigraphic evolution in fluvial style, which reflects the variation in accommodation space (Martinsen et al., 1999). The lateral continuity of the fine grained units in outcrop was only observable in Carnarvon Gorge, where they could be traced for a distance of at least 2 kilometres.
�������78�,���#�������������� ��������������������������!��(��������������������������� �#� �����������������������������������!����� ���������
The Basal Blocky subunit is characterized by a blocky low GR pattern which extensively floors most of the model area with a maximum thickness of 13 m; locally it appears to be eroded by the overlying Blocky subunit. The peculiar feature of this subunit is its widespread spatial extent; these deposits could be interpreted as unconfined sand sheets, representing the first big supply of sediment across the basin,perhaps coming from erosion of the strata exposed by the late Triassic to early Jurassic uplift in the Bowen Basin.
Between the Basal Blocky and the Blocky subunits, a widespread fine grained zone divides the two, except where it is eroded by the overlying Blocky subunit. The Blocky subunit is characterized by an average of 35 m to 40 m of blocky low GR signature, divided into smaller intervals up to 10 m thick. Locally, if the GR log has high enough resolution it is possible to appreciate small scale (a few metres), stacked, upwards
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
433.3
548.8
437.
544
7.5
457.
546
7.5
477.
548
7.5
497.
550
7.5
517.
552
7.5
537.
5
Sand
ston
eSa
ndst
one
Sand
ston
eto
ne
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
402.5
518
410.
642
0.6
430.
644
0.6
450.
646
0.6
470.
648
0.6
490.
650
0.6
510.
6
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
403.9
519.4
411.
842
1.8
431.
844
1.8
451.
846
1.8
471.
848
1.8
491.
850
1.8
511.
8
Mud
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECLOW
LP1
PRECUP
BUNCON
LP2
406.4
521.9
414.
842
4.8
434.
844
4.8
454.
846
4.8
474.
848
4.8
494.
850
4.8
514.
8
Sand
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
391.9
507.4
398.
240
8.2
418.
242
8.2
438.
244
8.2
458.
246
8.2
478.
248
8.2
498.
2
Sand
ston
eSa
ndst
one
Sand
ston
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
394.3
509.8
404.
541
4.5
424.
543
4.5
444.
545
4.5
464.
547
4.5
484.
549
4.5
504.
5
Sand
ston
eSa
ndst
one
PRECLOW
LP1
BUNCON
LP2
403.5
519
407.
641
7.6
427.
643
7.6
447.
645
7.6
467.
647
7.6
487.
649
7.6
507.
6
Mud
ston
eSa
ndst
one
Sand
ston
e
PRECLOW
LP1
BUNCON
LP2
398.7
514.2
403.
941
3.9
423.
943
3.9
443.
945
3.9
463.
947
3.9
483.
949
3.9
503.
9
PRECLOW
LP1
BUNCON
LP2
1118 m 1780 m 1276 m 1018 m 1233 m 1991 m 2508 m59739 [SSTVD]MD
-9 213gAPI
GR
Color fill
Color fill
Litho
59805 [SSTVD]MD
1.1 g/cm3 2.9
DEN
-14 263gAPI
GR
Color fill
Color fill
Litho
59617 [SSTVD]MD
1.1 g/cm3 3.0
DEN
-18 263gAPI
GR
Color fill
Color fill
Litho
59764 [SSTVD]MD
1.1 g/cm3 3.1
DEN
-10 260gAPI
GR
Color fill
Color fill
Litho
60875 [SSTVD]MD
1.0 g/cm3 2.8
DEN
-24 264gAPI
GR
Color fill
Color fill
Litho
60937 [SSTVD]MD
0.9 g/cm3 2.8
DEN
-21 263gAPI
GR
Color fill
Color fill
Litho
60948 [SSTVD]MD
0.9 g/cm3 3.0
DEN
-17 263gAPI
GR
Color fill
Color fill
Litho
59583 [SSTVD]MD
1.1 g/cm3 2.8
DEN
-2 183gAPI
GR
Color fill
Color fill
Litho
Basal
Blocky
Meandering?
7
coarsening packages (sort of a zig-zag of the GR). This could be the GR signature of single-story braided bars.
The Meandering subunit has a heterogeneous pattern and grain-size mostly resembling fining-upward packages capped by mudstones, and in places coal (where observed in Woleebee Creek GW4 and West Wandoan 1 cores; locations of these two wells are shown in Figure 1a). Coals were not observed in the wireline density logs. The GR signature is highly serrated, recording the heterolithic character of the fine-grained lithologies. Some mudstone zones in this subunit may not be as extensive as those in the lower subunits because they can represent the mud plug formed during the channel abandonment phase; this is a frequent process in meandering environments. From the Basal Blocky to the Meandering subunit it is clear that the accommodation space is increasing from highly amalgamated sandstone dominated channels upwards to more isolated channels in the floodplain. Thus the Basal Blockyand Blocky subunits record a low accommodation space period transitioning to higher accommodation in the Meandering subunit (Martinsen et al., 1999), with a rise in base level.
������� ��) ���� ������� ���������!�� ���� ���� ������ ���� ��#� ��� ������ ��� �#� ��� ��������� ����� ������ �����9� �������������� ����������������������:��(���� ����������������
Petrophysical upscaling
Before petrophysical modelling, the GR logs from the 83 wells were upscaled by using the Arithmetic averaging method because GR is continuous and non-directional. Figure 7 shows the upscaled GR and the comparison of histograms of GRs before
8
and after the upscaling. Results show that the average GR is about 28 API with standard deviation of 19 API and the histogram of upscaled GR is nearly the same as that from logs. This suggests that the upscaling cell height is small enough to capture thin intervals with high GR, which are also observed in the core descriptions.
Figure 8 show the histogram of upscaled GR for different units. Results show that the Blocky unit has a lower average GR of 20.5 API with standard deviation of 13.7 API while the Basal Blocky unit has higher average GR of 26.4 API with standard deviation of 17.7 API. The slightly higher GR could reflect the inclusion of lithic clasts in the conglomeratic sandstone that forms this unit.
�������;�9#� �����./��� �%:��(���� ����������������������������������./������������������!�������#� ������������#� ����� ������!��
�������<�=�������������#� �����./�����������>���������'���� "��������������� "���
GR, API
Mean=28 APIStd.=19 API
(a)(b)
High GR mudstone
100
2030
4050
200 40 60 80 100 120 140 160 180 200
Meandering unit(Mean=24.4; Std.=17.7)
Blocky unit(Mean=20.5; Std.=13.7)
Basal unit(Mean=26.4; Std.=17.7)
Prob
abili
ty, %
GR, API
9
Variogram Analysis
Kriging required the analysis of the horizontal variogram based on the upscaled GR. Considering the average well spacing which is about 1 km, and the size of modelling area is about 17 km in x- and y-direction, a search distance in x- and y- direction of 10km and lag numbers of 10 were used to generate the horizontal variogram which were used to identify the orientation of the variogram range. Figure 9 shows the generated horizontal variogram map which indicates a major orientation of about 135� within the study area. This figure also shows that the major and minor ranges are 3.9 km and 2.5 km respectively at a normalized variance of 0.75. This orientation is consistent with regional isochore maps and palaeoflow directions that define the main channel system within the Precipice Sandstone (Bianchi et al, 2015b; Martin, 1976).
Based on the horizontal variogram, a directional variogram was generated along major and minor orientations, respectively. Table 1 lists the parameters used in calculating the experimental variogram. The lag distance is 800 m and the number of lags is 20along both major and minor orientations. The band width is 800 m; tolerance angle is ������������ ���������������In vertical direction, the number of lags is 20 and lag distance is 0.4 m. Figure 10 shows the vertical experimental semivariance and its regressions for each of the Meandering, Blocky and Basal Blocky units. Results show that the type variogram is exponential for GR in the vertical direction. With same vertical distance, e.g. 2.6 m, the semivariance is least for the Blocky unit while it is largest for the Basal Blocky unit. Note that the steep increase of semivariance in Figure 10c may be because the thickness of the Basal Blocky subunit is commonly less than 7 m in most wells. However, the horizontal experimental semivariance points are very scattered and it is difficult to get a good variogram at the current well spacing for this study area, which suggests it is even more difficult in areas of sparse drilling without us of an analogue model. Geostatistics derived from this study might assist in modelling the different units within the Glenhaven area, where drilling is less dense.
10
�������?�=���@������(�����������#����������#� �����./�����������A������������������������*7B�������������
C�!���������������������#����������(��������� �� ��������
Direction Azimuth from north,
�
Number lags
Lagdistance,
m
Band width, m
Tolerance �������
Lagtolerance,
%
Vertical // 20 0.4 // 45 50
Major 135 20 800 800 45 50
minor 45 20 800 800 45 50
0.75
0.75
0.75
0.75
0.750.75
0.75
0.75
0.875
0.875
0.875
0.875
0.875
0.875
-8000 -6000 -4000 -2000 0 2000 4000 6000 8000
-8000 -6000 -4000 -2000 0 2000 4000 6000 8000
-800
0-6
000
-400
0-2
000
020
0040
0060
0080
00
-8000-6000
-4000-2000
02000
40006000
8000
0 1000 2000 3000 4000 5000m
1:129119 0.650.680.700.730.750.780.800.830.850.880.900.930.950.981.001.031.051.081.10
Variance
3.9 km
2.5 km
Distance, m
Dist
ance
, m
N
11
�������:�����(����� ������������� ���������(���� ������� ����'���� �������������./�(�������������������������� "��������
Modelled Distribution of GR
Simple Kriging (SK; Deutsch and Journel, 1998) was used to interpolate the distribution of GR values (without ��#� ���� interpretation of lithologies) with different variogram ranges the in x-y-z direction. Table 2 lists three scenarios; the models are shown in Figure 11. The choice of a vertical variogram range compared well to outcrop observations for sandstone, where cosets of bedforms varied between 3 to 5 m, although individual beds were much thinner. The fine grained units observed in outcrop, which would form the lower permeability zones, varied in thickness from about 1 to 3 m, which is below the range applied in our modelling.
The variograms in the x and y directions were investigated as the sedimentary geobodies, such as channels or mouthbars, were larger than the outcrop. Even at the Carnarvon Gorge, which is an extensive outcrop, fine grained units that bound the subdivisions could only be tracked for a few kilometers before disappearing behind vegetation. Therefore, different arbitrary variogram ranges were trialled in the subsurface modelling to determine which captured the spatial variation. With a simple interpretation that low gamma signatures represent sandstone and higher gamma signatures the finer grained units, a central sinuous pattern bordered by fine grained
(a) Meandering unit (b) Blocky unit
(c) Basal unit
(a) sill=0.54; vertical range=3.1 m;
(b) sill=0.40; vertical range=5.0 m;
(c) sill=0.85; vertical range=6.0 m;
12
lithologies emerges, regardless of variogram range. As expected, the areas of fine grained lithologies merge with a greater variogram range, but define the margins of a main channel.
Figure 12 to Figure 14 present the comparison for histograms of GR from modelled cells, upscaled cells and well logs from scenarios S1 to S3 respectively. Results show that the distribution of GR in those models is not the same as those from well logs.
� For the Meandering and Basal Blocky units in scenarios S1 and S2, the percentage of GR values for range from 20 to 30 API in models is higher than those from upscaled and well log values. It is lower in models for other ranges.
� For the Blocky unit, the percentage of GR values for range from 10 to 20 API in models is higher than those from upscaled and well log values; while it is lower in models for other ranges.
� For the Meandering and Basal Blocky units in scenario S3 that was modelled using a long variogram range of 10 km in x- and y- direction, the percentage of GR values for ranges from 20 to 30 API and 30 to 40 API in models is higher than those from upscaled and well log values; while it is lower in models for other ranges.
� For the Blocky unit, the percentage of GR values for ranges from 10 to 20 API and 20 to 30 API in models is higher than those from upscaled and well log values; while it is lower in models for other ranges. This yields a lower percentage of cells with high GR values in the models compared with that from logs (i.e. the models do not always capture the distribution of the thinner fine grained units).
Nevertheless, the model captured units less than 1 m because the average cell height in these models is about 0.3 m. Results also show that with an increasing variogram range, the modelled GR distribution is closer to the upscaled and logging GR.
13
C�!���%�D������������������������������������� ��������
Scenarios Units Exponential variogram range in x-y-z direction
Nugget/Sill
S1 Meandering 3 km× 3 km×3.1 m 0/0.54
Blocky 3 km× 3 km×5.0 m 0/0.40
Basal Blocky
3 km× 3 km×6.0 m 0/0.85
S2 Meandering 5 km× 5 km×3.1 m 0/0.54
Blocky 5 km× 5 km×5.0 m 0/0.40
Basal Blocky
5 km× 5 km×6.0 m 0/0.85
S3 Meandering 10 km× 10 km×3.1 m 0/0.54
Blocky 10 km× 10 km×5.0 m 0/0.40
Basal Blocky
10 km× 10 km×6.0 m 0/0.85
14
(a) S1
(b) S2
15
��������*+�+�����!���������./��������������������������(��������������������������C�!���%��������������������:��(���� ����������������
�������%�&��#������������������������./��������������� ����'��#� ����� ������������������������� ���������
(c) S3
(a) Meandering unit (b) Blocky unit
(c) Basal unit
0 40 80 120 160 200 2400
10
20
30
40
50
60
10
20
30
40
50
60
00 40 80 120 160
Perc
enta
ge, %
Perc
enta
ge, %
Perc
enta
ge, %
GR, API0
10
20
30
40
50
60
70
0 40 80 120 160 200GR, API
GR, API
16
�������*�&��#������������������������./��������������� ����'��#� ����� ������������������������� ��������%�
�������6�&��#������������������������./��������������� ����'��#� ����� ������������������������� ��������*�
(a) Meandering unit (b) Blocky unit
(c) Basal unit
0
8
16
24
32
40
48
0 40 80 120 160 200 240
Perc
enta
ge, %
GR, API
0 40 80 120 160
0 40 80 120 160 2000
10
20
30
40
50
60
Perc
enta
ge, %
GR, API
8
16
24
32
40
48
0
Perc
enta
ge, %
GR, API
(a) Meandering unit (b) Blocky unit
(c) Basal unit
8
16
24
32
40
00 40 80 120 160 200 240
0 40 80 120 160
0 40 80 120 160 200
Perc
enta
ge, %
8
16
24
32
40
0
Perc
enta
ge, %
48
56
GR, API GR, API
GR, API
Perc
enta
ge, %
8
16
24
32
40
0
17
Volume rendering of the different lithologies
The volumetric percentages of the different lithologies and their spatial distributions allow the visualisation of potential reservoir and baffle units within the different allounits of the Precipice Sandstone. In some software packages these are also called “geobodies”, but should not be confused with the sedimentary elements described in the main report, although in this instance they define the main trunk of the channel system. In this exercise, the lithologies were kept simple: Type 1 was defined by cells with GR<=40 API (coarse sand), Type 2 (dirty fine sand) was defined by cells with GR>40 and <=80 API, and Type 3 (dirty fine sand + mud) with GR>80 API (Figure 15).These were assigned to the grid models using different variogram ranges.
Table 3 lists the volumetric percentage for different lithologies within the different modelling scenarios using different variograms. Results shows that with horizontal variogram range increasing from 3 to 10 km (from S1 to S3), the percentage of Type2 increases from 4.6% to 10.6%; from 1.1% to 2.6% and from 3.1% to 8.2% for the Meandering, the Blocky and the Basal Blocky units, respectively; the percentage of Type 3 increases from 0.2% to 0.9%; from 0.2% to 0.6% and from 0.2% to 0.8% for the Meandering, Blocky and Basal Blocky units, respectively.
C�!���*������������(�������� �#�� ���������������������������������#���
ScenariosLithology Type Volumetric percentage, %
Meandering Blocky Basal BlockyS1 Type 1 95.2 98.8 96.8
Type 2 4.6 1.1 3.1Type 3 0.2 0.2 0.2
S2 Type 1 92.5 98 94.5Type 2 7 1.7 5.1Type 3 0.5 0.3 0.4
S3 Type 1 88.6 96.8 90.9 Type 2 10.6 2.6 8.2Type 3 0.9 0.6 0.8
There is commonly a question of “how thick does a fine grained impermeable unit have to be before it forms a baffle to flow”. Although that question cannot be answered by this study, Figure 16 shows that the thickness of the fine grained Type 3 occurrences are less than 2.5 m and more than 90% are less than 1 m across the wells. The thickness ratio of Type 3 from borehole data is about 3.1%, which is higher than the percentage of high gamma classes from models showing in Figure 11 which is about 0.2%, 0.4% and 0.8% from S1 to S3 respectively over the study volume. This suggests
18
that stochastic modelling of GR alone would underestimate the distribution of the lithotypes.
�������7����#�������./������������������������������������������� �#� ����������������� �����!���(��������������)����)��������
���������C�� "������������������C�#��*������./E<:4�5������������
In order to map the distribution of the ply number of the fine grained units (Type 3), a code was used to count the quantity of Type 3 within the 83 wells. Figure 17 shows the number distribution of Type 3, which was generated by using the borehole data and an isochore interpolation method. Results show that the number varies from 0 to
GR, A
PI
(1=mud; 2=dirty fine sand+mud; 3=dirty fine sand; 4=clean fine sand; 5=clean coarse sand; 6=clean conglomerate.)
10points
50points
10points
121points
556points10points
Litholoies in Lower Precipice Sandstone
Max
75th percentilesMedium
25th percentiles
Min
Lithologies of the Lower Precipice Sandstone
GR, A
PI
Thickness, m0 1 2
Perc
enta
ge, %
0
10
20
30
40
50
60
19
11 with an average of 3; the west, middle and south areas have more fine grained units. These finer grained units potentially mark the boundaries of the high porosity and permeability reservoir.
�������;�5�� �������#����������������!���������!���������C�#��*���� ������./E<:�4�5�
Figure 18 shows the distribution of filtered Type 3 for different variogram scenarios S1 to S3, or increasing the chance for the fine grain units to amalgamate. Results in each scenario show a “baffle free” low sinuosity belt with a maximum width of about 2.7 km trending southeast. From scenarios S1 to S3, the extent of the fine grained layers witha potential to baffle flow increases with the increase of horizontal variogram range (as expected).
Figure 19 shows the distribution of reservoir and potential baffle volumes from scenario S1, S2 and S3 along a well section. Results show well Spring Gully 37 has 11 layers of Type 3 with length from 0.2 km to 1.3 km. These high GR layers may represent low permeability intervals which would assist in delaying the arriving time of a CO2 plume to the top of Lower Precipice Sandstone. Such a pathway would increase the contact volume of reservoir fluid with the injected CO2 and hence increase the CO2
storage. This could be modelled in a dynamic flow simulation.
0
7
5
10
5
9
106
0
1
73
0
0
1
3
3
20
3
7 010
25
7
22
0 5
3
0
3
5
1
2
6
21
66 2
143
010
0
30
1
2
1 23
3
6
1
4
0
2
0
0
0
3
7
1
110
2
116
6
0
25
6
4
3
6
4
1
149°02'E 149°04'E 149°06'E 149°08'E 149°10'E 149°12'E
149°02'E 149°04'E 149°06'E 149°08'E 149°10'E 149°12'E
26°0
4'S
26°0
2'S
26°0
0'S
25°5
8'S
25°5
6'S
25°5
4'S
26°04'S26°02'S
26°00'S25°58'S
25°56'S25°54'S
0 1 2km
012345678910111213
N
Number of Geobody 3
+ Well location and layer numbers
20
(a) S1
2.7 km
(b) S2
2.5 km
21
�������<����������./�������!����������� ��������F������*�:��(���� ����������������
(c) S3
2.3 km
22
�������?��� �����������������������!�������������� ��������'��!��� ��������%'������ ��� ��������*'��������������������� �������������C�#��*G�������������������� ����C����� ���������������� �������������������������
DURHAM RANCH 23
SPRING GULLY 37
SPRING GULLY 43
DURHAM RANCH 64
SPRING GULLY 39SPRING GULLY 38
DURHAM RANCH 40DURHAM RANCH 50
DURHAM RANCH 51
SPRING GULLY 49SPRING GULLY 47
DURHAM RANCH 15
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
-160
-80
080
160
-160-80
080
160
0 0.5 1 1.5 2 2.5km
Body with GR<=40Body with GR(>40,<=80)Body with GR(>80)
(c) S3
DURHAM RANCH 23
SPRING GULLY 37
SPRING GULLY 43
DURHAM RANCH 64
SPRING GULLY 39SPRING GULLY 38
DURHAM RANCH 40DURHAM RANCH 50
DURHAM RANCH 51
SPRING GULLY 49SPRING GULLY 47
DURHAM RANCH 15
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
-160
-80
080
160
-160-80
080
160
0 0.5 1 1.5 2 2.5km
Body with GR<=40Body with GR(>40,<=80)Body with GR(>80)
(b) S2
DURHAM RANCH 23
SPRING GULLY 37
SPRING GULLY 43
DURHAM RANCH 64
SPRING GULLY 39SPRING GULLY 38
DURHAM RANCH 40DURHAM RANCH 50
DURHAM RANCH 51
SPRING GULLY 49SPRING GULLY 47
DURHAM RANCH 15
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
-160
-80
080
160
-160-80
080
160
0 0.5 1 1.5 2 2.5km
Body with GR<=40Body with GR(>40,<=80)Body with GR(>80)
(a) S2
Basal unit
Meandering unit
Blocky unit
23
Comparison of subsurface model with outcrop models
In this study, the observed lithology in core from West Wandoan 1 was used to relate GR with lithology as shown in Figure 2. The vertical channel filling pattern and lateral continuity from outcrop was also compared. However, for further comparison of this preliminary model with outcrop observations, the grain size has to be related with GR. Georgi et al. (1997) found that the core-plug-derived grain size and plug GR can be used to infer GR cutoffs for differentiating fine-grained sands from silty-shaley sands as shown in Figure 20. However, to quantitatively calibrate the GR with grain size is still a challenge. Another issue is that the grid sizes in outcrop models are about 0.5 m in horizontal and 0.1 m in vertical while the local model is 100 m×100 m× 0.3 m as shown in Figure 21. Upscaling is needed before this comparison can be made.
In the future work, we will develop the relationship between the GR wireline log and the sedimentary facies using West Wandoan 1 and Woleebee Creek GW4 logs, potentially with a core from the APLNG area, and use these to develop a facies distribution model. The geometry of the geobodies in which the facies occur, will be constrained by parameters obtained in outcrop. The static geological grid model is this way populated by the facies, using either (or both) object modelling or multiple-point simulation. Based on facies models, we will generate distributions for grain size or petrophysical properties which can be used for dynamic reservoir modelling. A pictorial workflow is shown in Figure 21.
�������%:�/����������#�!�������#����./���(����� ���!���������4�5�(��������������������������������@������ ����� ����#���������������.������������??;��
24
�������%���@�� ��#������������ ���������������� ��#�����������.�������@��������!������������ ��#����5����.�����% ��!��./�������!������!������������ �� ���� ����������������������������������������-���������5����.����������������������H�������������!'�!������������!����#���������(��
Discussion and future work
This preliminary model was set up to develop a workflow for further, more sophisticated subsurface facies modelling for the Precipice Sandstone, and to test the influence of the variograms on the interpretation of lateral connectivity of the lithologies, in particular the finer grained baffle units within the Lower Precipice Sandstone allostratigraphic unit. The dense drilling in the APLNG area was selected for modelling, as the jump from outcrop scale to an area with sparse drilling, such as in the Glenhaven area doesn’t allow a good test of architectural element or geobody modelling. Observations of the lower braided fluvial facies are best exposed at the Carnarvon Gorge, and here discrete, albeit thin, finer grained units could be tracked for about 2 km. This suggests that one can correlate, or at least model, the finer grained units and facies in the braided facies for variogram correlation lengths in the range of 3-10 km. In the detailed model, regardless of variogram used, a “baffle free” belt trending southeast was present and, coupled with a southerly marching, cross
(a)
(b) (c)
25
bedded bar forms observed in core and outcrop in the lower unit, could act to control flow pathways in the reservoir. Based on the geometry of the baffle free zone, we’d interpret the trunk channel belt of the Precipice to average 2.5km in width, although smaller scale macroforms and channels would be modelled within this.
Regarding the vertical distribution of baffles in Basal Blocky and Blocky subunits they are thinner but more laterally extensive, whereas in the Meandering subunit they appear locally developed and thicker, suggesting the latter originated as abandoned channel fill in meandering systems in response to the base level rise. It is recommended that available core from this APLNG area be logged to test the interpretation of facies in comparison to those from the Glenhaven area.
Future work, should apply a more sophisticated approach to facies modelling, utilizing the geobody dimensions obtained in the outcrop studies, constrained by further subdivision within the wireline logs according to the three allostratigraphic units- the lower braid plain, meandering facies transitioning into more deltaic facies as the sealing unit of the Evergreen is approached. A test of the geometry of the different sedimentary facies and their geobodies might be to object model their distribution as interpreted for a given outcrop, then “cookie cut” that outcrop and compare to the measured data. This is outside the scope of this project. Included in the appendix are the preliminary correlations that can be used for further modelling and/or extended to other areas. Knowledge gained from this study, e.g. the lateral extent of subunit bounding surfaces, the statistics for assignment of lithofacies properties to the grid, the relationship between the wireline, core and porosity and permeability, and thedistribution and extent of the potential baffles and the baffle-free reservoir within the Precipice Sandstone; can be applied to the more sparsely drilled EPQ7, to improve confidence in the subsurface static geological model.
References
Ashraf, M., 2014. Geological storage of CO2: Heterogeneity impact on the behavior of pressure. International Journal of Greenhouse Gas Control 28, 356-368.
Bachu, S., 2008. CO2 storage in geological media: Role, means, status and barriers to deployment. Prog Energ Combust 34, 254-273.
Begg, S., Carter, R., and Dranfield, P., 1989. Assigning effective values to simulator gridblock parameters for heterogeneous reservoirs. SPE reservoir engineering, 4(04), 455-463.
Bianchi, V., Pistellato, D., Boccardo, S., Zhou, F. and Esterle, J. (2015a) Outcrop mapping and photogrammetry of the Precipice Sandstone ANLEC Milestone 2 Report: Photogrammetry and preliminary mapping results.
26
Bianchi, V., Pistellato, D., Boccardo, S., Zhou, F. and Esterle, J. (2015b) Outcrop mapping and photogrammetry of the Precipice Sandstone ANLEC Milestone 3 Report: Detailed facies schema and associated trial models.
Bianchi, V., Pistellato, D., Boccardo, S., Zhou, F. and Esterle, J. (2015c) Outcrop mapping and photogrammetry of the Precipice Sandstone ANLEC Milestone 4 Report: Geobody dimensions database and associated models.
Bowman, M., McClure, N., and Wilkinson, D., 1993. Wytch Farm oilfield: deterministic reservoir description of the Triassic Sherwood Sandstone. Paper presented at the Geological Society, London, Petroleum Geology Conference series.
Cavanagh, A.J., Haszeldine, R.S., 2014. The Sleipner storage site: Capillary flow modeling of a layered CO2 plume requires fractured shale barriers within the Utsira Formation. International Journal of Greenhouse Gas Control 21, 101-112.
Dance, T., 2013. Assessment and geological characterisation of the CO2CRC Otway project CO2
storage demonstration site: from prefeasibility to injection. Marine and Petroleum Geology, 46, 251-269.
Davies, D.K., Williams, B.P.J., and Vessell, R.K., 1993, Dimensions and quality of reservoirs originating in low and high sinuosity channel systems, Lower Cretaceous Travis Peak Formation, east Texas, U.S.A., in C. P. North and D. J. Prosser, eds., 1993, Characterization of fluvial and eolian reservoirs: Geological Society Special Publication 73, 95-121.
Deutsch, C.V. and Journel, A.G., 1998. GSLIB Geostatistical Software Library and User's Guide. Second Edition.
Geehan, G. W., Lawton, T.F., Sakurai, S. Klob, H., Clifton, T.R., Inman, K.F., and Nitzberg, K.E., 1986. Geologic prediction of shale continuity, Prudhoe Bay field, in L. W. Lake and H. B. Carroll, eds., Reservoir characterization: Orlando, Florida, Academic Press, 63-82.
Georgi, D.T., Bergren, P.A., and Devier, C.A. 1997. Plug gamma ray: Key to formation evaluation. Poster presentation at the 1997 SCA International Symposium, Calgary, 8-10 September. SCA-9732.
Golab, A., Arena, A., Khor, J., Goodwin, C., Young, B., Carnerup, A., Hussain, F., 2015. Milestone 1.4. Final report of RCA and SCAL data on plugs from West Wandoan-1 Well. Technical report for Project 7-0311-0128, Sub-project 1: Derive a full suite of special core analysis (capillary pressure, supercritical CO2: brine relative permeability) data sets on plugs from Surat Basin reservoir and seal rock samples.
Issautier, B., Fillacier, S., Gallo, Y.L., Audigane, P., Chiaberge, C., and Viseur, S., 2013. Modelling of CO2 injection in fluvial sedimentary heterogeneous reservoirs to assess the impact of geological heterogeneities on CO2 storage capacity and performance. Energy Procedia, 37, 5181-5190.
Ma, J., Couples, G.D., and Gardiner, A.R., 2008. Numerical modeling of the fluid flow impact of thin baffle laminae in cross bedding. Water Resources Research, 44, 1-6.
27
Martin, K. (1976). Sedimentology Of The Precipice Sandstone, Surat Basin, Queensland. Unpublished PhD Thesis, The University of Queensland. 224 pp plus appendices.
Martinsen, O.J., Ryseth, A., Helland-Hansen, W., Flesche, H., Torkildsen, G. and Idil, S., (1999). Stratigraphic base level and fluvial architecture, Ericson Sandstone (Campanian), Rock Springs Uplift, W. Wyoming, USA. Sedimentology, 46, 235–260.OGIA model, 2014. Office of Groundwater Impact Assessment, Queensland, 2014.
Petrel, 2014. Schlumberger, user manual.
Shepherd, M., 2016. Braided fluvial reservoirs. AAPG Memoirs. http://wiki.aapg.org/Braided_fluvial_reservoirs.
28
APPENDIX
�������48�)���� ������������� �����F���C�� "�����>+'�./2+�������'����������'������#� �����./����'��������������������
�������4%8�)���� ������������� �����F��%�
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
Sand
ston
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
415.7
554.2
421.
143
1.1
441.
145
1.1
461.
147
1.1
481.
149
1.1
501.
151
1.1
521.
153
1.1
541.
1
Sand
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
395.8
534.3
407.
641
7.6
427.
643
7.6
447.
645
7.6
467.
647
7.6
487.
649
7.6
507.
651
7.6
527.
6
Sand
ston
eSa
ndst
one
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
404.9
543.4
415.
442
5.4
435.
444
5.4
455.
446
5.4
475.
448
5.4
495.
450
5.4
515.
452
5.4
535.
4
Silts
tone
Silts
tone
Sand
ston
eSa
ndst
one
Sand
ston
eSi
ltsto
nee
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
419.5
558
428.
343
8.3
448.
345
8.3
468.
347
8.3
488.
349
8.3
508.
351
8.3
528.
353
8.3
548.
3
Sand
ston
eSa
ndst
one
Sand
ston
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
431.6
570.1
439.
244
9.2
459.
246
9.2
479.
248
9.2
499.
250
9.2
519.
252
9.2
539.
254
9.2
559.
2
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
414.1
552.6
422
432
442
452
462
472
482
492
502
512
522
532
542
Sand
ston
eSa
ndst
one
Sand
ston
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
404.4
542.9
411.
942
1.9
431.
944
1.9
451.
946
1.9
471.
948
1.9
491.
950
1.9
511.
952
1.9
531.
9
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
413.2
551.7
422.
343
2.3
442.
345
2.3
462.
347
2.3
482.
349
2.3
502.
351
2.3
522.
353
2.3
542.
3
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
400.3
538.7
406.
541
6.5
426.
543
6.5
446.
545
6.5
466.
547
6.5
486.
549
6.5
506.
551
6.5
526.
5
Sand
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
398.1
536.6
408.
841
8.8
428.
843
8.8
448.
845
8.8
468.
847
8.8
488.
849
8.8
508.
851
8.8
528.
8
Sand
ston
eSa
ndst
one
ston
e
PRECLOW
LP1
BUNCON
LP2
438.7
577.2
451.
546
1.5
471.
548
1.5
491.
550
1.5
511.
552
1.5
531.
554
1.5
551.
556
1.5
571.
5
PRECLOW
LP1
BUNCON
LP2
1096 m 1063 m 1022 m 896 m 1016 m 1281 m 839 m 1851 m 2522 m 1405 m58593 [SSTVD]MD
0.7 g/cm3 3.7
DEN
-12 218gAPI
GR
Color fill
Color fill
Litho
59300 [SSTVD]MD
1.0 g/cm3 2.9
DEN
-16 261gAPI
GR
Color fill
Color fill
Litho
59007 [SSTVD]MD
1.1 g/cm3 3.1
DEN
-6 211gAPI
GR
Color fill
Color fill
Litho
1 ohm.m 100
RT
59028 [SSTVD]MD
1.1 g/cm3 2.9
DEN
-6 255gAPI
GR
Color fill
Color fill
Litho
1 ohm.m 100
RT
59000 [SSTVD]MD
0.8 g/cm3 3.0
DEN
-12 212gAPI
GR
Color fill
Color fill
Litho
59005 [SSTVD]MD
1.0 g/cm3 2.8
DEN
-9 184gAPI
GR
Color fill
Color fill
Litho
58583 [SSTVD]MD
1.2 g/cm3 3.3
DEN
-24 264gAPI
GR
Color fill
Color fill
Litho
58585 [SSTVD]MD
1.2 g/cm3 3.3
DEN
-24 263gAPI
GR
Color fill
Color fill
Litho
59058 [SSTVD]MD
1.0 g/cm3 3.0
DEN
-7 259gAPI
GR
Color fill
Color fill
Litho
1 ohm.m 100
RT
59546 [SSTVD]MD
0.6 g/cm3 2.9
DEN
-12 178gAPI
GR
Color fill
Color fill
Litho
61102 [SSTVD]MD
0.3 g/cm3 3.3
DEN
-15 256gAPI
GR
Color fill
Color fill
Litho
Basal
Blocky
Meandering?
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
Sand
ston
eSa
ndst
one
Silts
tone
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
379.2
532.4
385.
439
5.4
405.
441
5.4
425.
443
5.4
445.
445
5.4
465.
447
5.4
485.
449
5.4
505.
451
5.4
525.
4
Sand
ston
eSa
ndst
one
Sand
ston
e
PRECLOW
LP1BUNCON
LP2
411.2
569.2
422.
143
2.1
442.
145
2.1
462.
147
2.1
482.
149
2.1
502.
151
2.1
522.
153
2.1
542.
155
2.1
562.
1
Sand
ston
e
PRECLOW
LP1
BUNCON
LP2
407.6
565.4
413.
542
3.5
433.
544
3.5
453.
546
3.5
473.
548
3.5
493.
550
3.5
513.
552
3.5
533.
554
3.5
553.
5
Sand
ston
eM
udst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
406.5
575.4
413.
442
3.4
433.
444
3.4
453.
446
3.4
473.
448
3.4
493.
450
3.4
513.
452
3.4
533.
454
3.4
553.
456
3.4
Silts
tone
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
406.5
584.9
416.
742
6.7
436.
744
6.7
456.
746
6.7
476.
748
6.7
496.
750
6.7
516.
752
6.7
536.
754
6.7
556.
756
6.7
576.
7
Sand
ston
eSa
ndst
one
Sand
ston
eSi
ltsto
nePRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
410.1
590.2
423.
443
3.4
443.
445
3.4
463.
447
3.4
483.
449
3.4
503.
451
3.4
523.
453
3.4
543.
455
3.4
563.
457
3.4
583.
4
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
393
554
407.
241
7.2
427.
243
7.2
447.
245
7.2
467.
247
7.2
487.
249
7.2
507.
251
7.2
527.
253
7.2
547.
2
Sand
ston
eSa
ndst
one
Sand
ston
eM
udst
one
PRECLOW
LP1
BUNCON
LP2
405.7
588.6
419.
742
9.7
439.
744
9.7
459.
746
9.7
479.
748
9.7
499.
750
9.7
519.
752
9.7
539.
754
9.7
559.
756
9.7
579.
7
Sand
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
Silts
tone
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
397.4
578.6
404.
341
4.3
424.
343
4.3
444.
345
4.3
464.
347
4.3
484.
349
4.3
504.
351
4.3
524.
353
4.3
544.
355
4.3
564.
3
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
1355 m 996 m 1889 m 1579 m 2623 m 1057 m 1186 m 2985 m58590 [SSTVD]MD
0.7 g/cm3 3.4
DEN
-7 193gAPI
GR
Color fill
Color fill
Litho
58591 [SSTVD]MD
1.1 g/cm3 2.9
DEN
-22 261gAPI
GR
Color fill
Color fill
Litho
59033 [SSTVD]MD
0.7 g/cm3 3.2
DEN
-13 206gAPI
GR
Color fill
Color fill
Litho
59738 [SSTVD]MD
0.9 g/cm3 3.0
DEN
-19 263gAPI
GR
Color fill
Color fill
Litho
59721 [SSTVD]MD
1.2 g/cm3 2.9
DEN
-16 263gAPI
GR
Color fill
Color fill
Litho
59804 [SSTVD]MD
1.2 g/cm3 3.0
DEN
-15 262gAPI
GR
Color fill
Color fill
Litho
60841 [SSTVD]MD
0.6 g/cm3 3.3
DEN
-12 215gAPI
GR
Color fill
Color fill
Litho
60824 [SSTVD]MD
1.1 g/cm3 2.9
DEN
-22 264gAPI
GR
Color fill
Color fill
Litho
61247 [SSTVD]MD
0.9 g/cm3 3.0
DEN
-22 263gAPI
GR
Color fill
Color fill
Litho
Basal
Blocky
Meandering?
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
433.3
548.8
437.
544
7.5
457.
546
7.5
477.
548
7.5
497.
550
7.5
517.
552
7.5
537.
5
Sand
ston
eSa
ndst
one
Sand
ston
eto
ne
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
402.5
518
410.
642
0.6
430.
644
0.6
450.
646
0.6
470.
648
0.6
490.
650
0.6
510.
6
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
403.9
519.4
411.
842
1.8
431.
844
1.8
451.
846
1.8
471.
848
1.8
491.
850
1.8
511.
8
Mud
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECLOW
LP1
PRECUP
BUNCON
LP2
406.4
521.9
414.
842
4.8
434.
844
4.8
454.
846
4.8
474.
848
4.8
494.
850
4.8
514.
8
Sand
ston
eSa
ndst
one
Sand
ston
eSa
ndst
one
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
391.9
507.4
398.
240
8.2
418.
242
8.2
438.
244
8.2
458.
246
8.2
478.
248
8.2
498.
2
Sand
ston
eSa
ndst
one
Sand
ston
e
PRECMID
PRECLOW
LP1
PRECUP
BUNCON
LP2
394.3
509.8
404.
541
4.5
424.
543
4.5
444.
545
4.5
464.
547
4.5
484.
549
4.5
504.
5
Sand
ston
eSa
ndst
one
PRECLOW
LP1
BUNCON
LP2
403.5
519
407.
641
7.6
427.
643
7.6
447.
645
7.6
467.
647
7.6
487.
649
7.6
507.
6
Mud
ston
eSa
ndst
one
Sand
ston
e
PRECLOW
LP1
BUNCON
LP2
398.7
514.2
403.
941
3.9
423.
943
3.9
443.
945
3.9
463.
947
3.9
483.
949
3.9
503.
9
PRECLOW
LP1
BUNCON
LP2
1118 m 1780 m 1276 m 1018 m 1233 m 1991 m 2508 m59739 [SSTVD]MD
-9 213gAPI
GR
Color fill
Color fill
Litho
59805 [SSTVD]MD
1.1 g/cm3 2.9
DEN
-14 263gAPI
GR
Color fill
Color fill
Litho
59617 [SSTVD]MD
1.1 g/cm3 3.0
DEN
-18 263gAPI
GR
Color fill
Color fill
Litho
59764 [SSTVD]MD
1.1 g/cm3 3.1
DEN
-10 260gAPI
GR
Color fill
Color fill
Litho
60875 [SSTVD]MD
1.0 g/cm3 2.8
DEN
-24 264gAPI
GR
Color fill
Color fill
Litho
60937 [SSTVD]MD
0.9 g/cm3 2.8
DEN
-21 263gAPI
GR
Color fill
Color fill
Litho
60948 [SSTVD]MD
0.9 g/cm3 3.0
DEN
-17 263gAPI
GR
Color fill
Color fill
Litho
59583 [SSTVD]MD
1.1 g/cm3 2.8
DEN
-2 183gAPI
GR
Color fill
Color fill
Litho
Basal
Blocky
Meandering?