Multi-criteria decision making and machine learning method research in petroleum natural gas explorations

Oil and natural gas exploration is a long process in which many criteria should be evaluated together in terms of time, cost and workforce. Hydrocarbon activities start with purchasing a license in the area to be examined, then continue with geological, geophysical and drilling studies. If a hydrocarbon discovery has been made in the area, reservoir determination studies are carried out for this area. In oil and natural gas exploration, such as seismic, gravity and magnetic techniques are used to determine the hydrocarbon location. In this study, for the estimation of hydrocarbon potential areas in an area of 27000 km2 in the Ergene Basin, Multiple Criteria Decision Analysis (MCDA) and machine learning were utilized, which are frequently used in studies such as the most appropriate site selection, energy policy determination etc. Analytical Hierarchy Process and Fuzzy Logic Analytical Hierarchy Process are preferred, which is widely used in MCDA. The Random Forest Algorithm, which is one of the machine learning methods, has also been used to predict the locations of oil and natural gas containing areas. It was estimated whether the randomly generated spots for the study area contain hydrocarbons. A forecast map was produced using forecast points. Accuracy analyzes were performed by forecast maps obtained from all three methods matched with drilling data. The result obtained by averaging the prediction points obtained from the Random Forest Algorithm is 78%, FAHP is 73% and AHP is 69%. It has been demonstrated that all three methods can be used for hydrocarbon potential estimation.