Methodology for Prediction of the Locations of Giant Oil and Gas Reservoirs: Field Results

Abstract

Abstract In 1985, we developed methodology for oil and gas exploration by combining morphostructural zoning and pattern recognition. We applied this methodology in the Andes of South America. In this paper, we will present the results of testing:the suggestion that giant oil/gas fields are located in morphostructural nodes,the decision rule for recognizing knots promising giant oil/gas fields,the 1985 prediction of future discoveries in the Andes. Since 1986, only three new giant oil/gas fields (more then 1 billion bbl of oil-in-place equivalent each) were found in the Andes - Cano Limon, Cusiana-Cupiagua (Colombia) and Camisea (Peru). All three are located in predicted sites. There are eight more sites in that region recognized as highly perspective, these are sites of special interest for future drilling. Introduction Modern methods of prospecting for oil and gas fields are based mainly on models of the deep structure of the Earth's crust obtained by means of geological and geophysical methods. These models provide good results in mapping of probable oil and gas traps, but they do not allow a distinction between productive and non-productive reservoirs. The geochemical and geophysical methods for direct detection of oil and gas pools do not considerably increase the success ratio of exploratory drilling: the uncertainty in the interpretation of data obtained is still high. Therefore, selection of licenses for oil and gas exploration is inevitably associated with a high risk of obtaining negative results. This is particularly true when one is hopeful of finding large fields. In order to reduce exploration risk, reliable and inexpensive methods are needed that could be instrumental in assessing the areas prior to bidding. Efficient and proven methods for handling such tasks are as yet unknown. In the last decade, the authors have developed a methodology for prediction of the largest oil and gas accumulation. Mainly it is based on the methods developed in the early seventies for geophysical problems:morphostructural zoning andpattern recognition. Initial data contain topographic and hypsometric maps, satellite images, regional geologic, tectonic, and other special maps. In 1985, we applied this methods to the problem of oil and gas exploration in the Andes. We found that almost all giant oil and gas fields in that region coincide with morphostructural knots (intersections of lineaments). We applied computer pattern recognition for recognizing knots that contain (or can contain) giant oil fields. As a result, we found 11 knots that should contain giant oil fields but had not been developed since 1985. The location of future discoveries are predicted with a precision of 45 miles. The rate of success was estimated as 0.75 or more. The reliability of the prediction was checked in a set of special statistical and logical tests. The objective of the paper is to present the results of the most important test - comparison of the drilling results in the last 10 years with the prognosis. The structure of the paper will be as follows. First, the basics of used methods will be given (morphostructural zoning and pattern recognition). Next, the description of knots and the results of the recognition of knots in the Andes will be presented. Then, special efforts to increase the reliability of the prediction and the results of two field tests will be discussed. Finally, we will present the recursive recognition technique and will discuss the results and emphasize the limits of the methodology. P. 321^