· · · Institutional Repository

The SLOPE3 project investigates the depositional architecture of a Permian age, deep water / complex channelized slope system which is of interest to the oil industry as a potential analogue for real oil / gas reservoirs. This study forms a part of the SLOPE3 project and is focused on an integration of cores and wireline logs from six research boreholes, each forming a 1-D datasets showing a hierarchy of genetic depositional elements: -channels, complexes, complex sets and equivalent hierarchy channel sands, channel margin, levee and overbank deposits. These elements were previously characterized in outcrop by the University of Liverpool in an earlier phase of the SLOPE project. The University of Liverpool presented their data as a high resolution outcrop map / correlation panel some five kilometre long showing a unit truncated by a 120 m deep slope valley (see Figure frontpage). For SLOPE3 a set of six boreholes were drilled along a line sub-parallel and about 300 metres behind that outcrop section. The borehole positions had been picked on the basis of the outcrop correlation panel. All six boreholes were fully cored and five boreholes were logged with GR / Sonic / FMS. The ultimate objective of the project is to provide a set of analogue-based input data for 3D reservoir modelling. The wireline data of GR, Sonic, and FMS borehole images formed the basis for this MSc thesis and was processed and presented as graphics/logs by making use of GeoFrame software. With this software manually dip-picking of planar sedimentary structures was done. With the focus on constraining the 3D architecture of the channelized slope system, bed boundaries as well as the following other dip classes were picked: erosion surfaces, climbing ripples, low-angle cross-bedding, syn-sedimentary faults, and slump folds. By making use of the FMS images and the GR / Sonic log results the following lithofacies associations / genetic sedimentary units were identified in the five logged boreholes: massive sandstone, channel margin, external levee, internal levee, all in a background of shale. All measured sedimentary dip picks were corrected to palaeocurrent / palaeoslope indicators by subtraction of the measured and averaged structural dips. Sedimentary dips were plotted on stereonets for statistical analysis and interpretation in terms of palaeocurrent flow direction / palaeoslope orientation. This combination of a detailed outcrop data set with a suite of boreholes located on the basis of the outcrop data is a near ideal test set for the quality of geological predictions of a 3D reservoir architecture. The different type of architectural reservoir elements each with their specific lithofacies associations, can be characterized by FMS, Sonic, and GR logs when run in boreholes. However in this thesis it is shown that by doing a prediction (‘prognosis’) of the distribution of the lithofacies associations and their thicknessencountered by the boreholes based on an extrapolation from the nearby outcrop, it is rather difficult to predict the ‘actual’ geology away from well control. Moreover the value of palaeocurrent and palaeoslope indicators as identified from the FMS images for the prediction the distribution (orientation and the position) of the different genetic bodies, proved to be surprisingly minor. This is a reflection of the large spread / variation in measured palaeocurrent directions (local variation), and the additional scatter of those indicators away from the boreholes (areal variation). This latter aspect is characteristic for a meandering submarine slope channel complexes. The CD Ridge complex channelized slope system is hence interpreted to be a ‘jig-saw’ to even a ‘labyrinth’ reservoir type. The shale-prone external levee deposits are laterally extensive, thinly bedded sequences have a more layer-cake architecture. Therefore it is suggested to use deterministic modelling for these layer-cake parts, where the ‘jigsaw’ parts require a probabilistic modelling approach. The analysis of borehole images and logs resulted in: A complete input dataset including: • Zonations of the genetic bodies for every well. • A whole bunch of stereonets for 3D analysis. Both could not be achieved without BHI. Value of the use of borehole images is the 3D aspect and to distinguish between different thin-bedded lithofacies associations.

The Permian Skoorsteenberg Formation in the Tanqua-Karoo basin in South Africa provides excellent exposure of submarine basin-floor fans. Because of this, the European Union sponsored various outcrop studies, drilling as well as the data acquisition of the research boreholes in the NOMAD project. It was a unique opportunity to study a submarine fan system by combining widely exposed outcrop and research borehole data. Because both data sets are available, this MSc. project aims to acquire bed thickness data from core and borehole image (Fullbore Formation Micro Imager) analysis and to extrapolate these data to 3D in a variety of methods based on the results of first objective. Analysis has been done on the turbidites of Fan 3 in wells NB-4, NB-3, NB-2, and NS-2. The cumulative distributions of turbidite bed thicknesses in the studied wells were found to follow a power law. Therefore, the cumulative bed thicknesses plot can be used for the following purposes: (i) to derive certain parameters for the bed geometries and distributions; (ii) to calculate turbidite volume connected to the well; (iii) to suggest accommodation space degree of confinement; and (iv) to derive qualitative information on the extent of erosion and bed amalgamation (thus, may suggest depositional setting). The change of slope (the change of exponent) in the cumulative bed thicknesses can be interpreted due to confinement or alternatively, due to the variation of the flow rheology. The clustering of data in the cumulative bed thicknesses plot may suggest the flow rheology affecting the turbidite deposition in particular time or location, which may represent a shift in the lobe depocenter. The turbidites volume connected to the wellbore can be calculated using three different methods, these are: using the mathematical model developed by Malinverno (1997) (volume is calculated from the bed thickness distribution); using the facies model developed utilizing Petrel 2010.1 software; and using the discrete convolution method (volume is calculated by relating the experimental data and the well data). The application of each method has to be done with care, taking into consideration the data availability and the limitation of each method. Moreover, information of the lobe internal geometries is needed in the volume calculation.