Authors:Tomislav Malvić, Josipa Velić, Janina Horváth and Marko Cvetković
Three examples of the use of neural networks in analyses of geologic data from hydrocarbon reservoirs are presented. All networks are trained with data originating from clastic reservoirs of Neogene age located in the Croatian part of the Pannonian Basin. Training always included similar reservoir variables, i.e. electric logs (resistivity, spontaneous potential) and lithology determined from cores or logs and described as sandstone or marl, with categorical values in intervals. Selected variables also include hydrocarbon saturation, also represented by a categorical variable, average reservoir porosity calculated from interpreted well logs, and seismic attributes. In all three neural models some of the mentioned inputs were used for analyzing data collected from three different oil fields in the Croatian part of the Pannonian Basin. It is shown that selection of geologically and physically linked variables play a key role in the process of network training, validating and processing. The aim of this study was to establish relationships between log-derived data, core data, and seismic attributes. Three case studies are described in this paper to illustrate the use of neural network prediction of sandstone-marl facies (Case Study # 1, Okoli Field), prediction of carbonate breccia porosity (Case Study # 2, Beničanci Field), and prediction of lithology and saturation (Case Study # 3, Kloštar Field). The results of these studies indicate that this method is capable of providing better understanding of some clastic Neogene reservoirs in the Croatian part of the Pannonian Basin.
Authors:Ana Brcković, Monika Kovačević, Marko Cvetković, Iva Kolenković Močilac, David Rukavina and Bruno Saftić
Lithofacies definition in the subsurface is an important factor in modeling, regardless of the scale being at reservoir or basin level. In areas with low exploration level, modeling of lithofacies distribution presents a complicated task as very few inputs are available. For this purpose, a case study in the Požega Valley was selected with only one existing well and several seismic sections within an area covering roughly 850 km2. For the task of expanding the input data set for lithofacies modeling, neural network analysis was performed that incorporated interpreted lithofacies (sandstone, siltite, marl, and breccia-conglomerate) in a single well and attribute data gathered from a seismic section. Three types of different neural networks were used for the analysis: multilayer perceptron, radial-basis function, and probabilistic neural network. As a result, three lithofacies models were built alongside a seismic section based upon predictions acquired from the neural networks. Three lithofacies were successfully predicted on the section while the breccia-conglomerate was either missing or underpredicted and mostly positioned in a geologically invalid interval. Results obtained by single networks differed from one another, which indicated that a result from a single network should not be treated as representative; thus, the facies distribution for modeling should be acquired from either an ensemble of neural networks or several neural networks. Analysis showed the initial potential of the usability of neural networks and seismic attribute analysis on vintage seismic sections with possible drawbacks of the applications being pointed out.