Hydrocarbon zone identification using support vector machine learning method

One of the basic steps of oil exploration is to define the hydrocarbon zone. Different methods have been used so far for defining such zones. For a specific dataset, finding the most appropriate method leads to more accurate estimates and predictions of analysis besides improving the speed of calculations. Support Vector Machine (SVM), which is one of the methods for analyzing the data, uses kernel functions. It finds a better relationship between data factors and hydrocarbon zone leading to better estimates and classifications.    In this article, hydrocarbon zone detection has been done using seismic and well data.    The purpose of facies analysis is to obtain important petrophysical parameters of the reservoir and to identify heterogeneous boundaries below the ground. The results of the interpretation of petrophysical parameters are the input of the three-dimensional reservoir modeling process and through these parameters, the reservoir parameters are distributed in three-dimensional space. This model is widely used in various sections such as exploration and drilling of new wells, overdraft from a reservoir, determination of suitable areas for overdraft, reduction of drilling risk and risk, determination of reservoir lithology and identification of key well and its extension to other wells in the region. The most important petrophysical parameters are shale volume, porosity, permeability, reservoir fluid saturation and reservoir lithology.    The study of seismic facies has been started since the 90's and due to its importance and application in reservoir description, it has always been considered by many researchers.    To perform the analysis above, first, the hydrocarbon zones were spotted across the Asmari Formation using well logs and well geology reports. Next, the SVM method was used to detect each hydrocarbon zone using well logs. There was an acceptable agreement between the results of SVM method and well geology reports. Second, hydrocarbon zones detection was done using seismic data by SVM. At this stage, seismic attributes were extracted from the seismic trace in the well location. Then, covariance matrix and cross plots of seismic attributes used to identify the most effective attributes to hydrocarbon zones detection. In order to validate the results, the seismic attributes of another trace near the well location were used for hydrocarbon zone detection. SVM results matched hydrocarbon zones with low error.