DOI: 10.5593/sgem2017H/15/S06.011


C. Popa, D. R. Jacota, C. Marinoiu
Thursday 23 November 2017 by Libadmin2017

References: 17th International Multidisciplinary Scientific GeoConference SGEM 2017, www.sgemviennagreen.org, SGEM2017 Vienna GREEN Conference Proceedings, ISBN 978-619-7408-26-3 / ISSN 1314-2704, 27 - 29 November, 2017, Vol. 17, Issue 15, 83-90 pp; DOI: 10.5593/sgem2017H/15/S06.011


Reservoir description is rapidly increasing with the help of advancing computing possibilities. Although machines become more efficient, it is necessary to have accurate input data and plenty of data points to achieve reliable results. However, in the case of both old Romanian hydrocarbon reservoirs and some of the new ones, not too many data points are available. In this situation, no matter the computing power of the machine or the computing algorithm, detailed reservoir description is limited. This paper extends some existing studies in the literature, where various methods of generating routine core analysis distribution maps were used, and compares the absolute permeability maps generated by universal kriging and artificial neural networks trusting the fact that high accuracy of the mentioned distribution can be obtained. The aim of this this paper is to point out which of the computational, statistical or numerical methods is more efficient residing on a small number of input data points. A first gain could consist in reducing computing time and obtaining the same quality results in a shorter amount of time. Other applications, once the reservoir image is created, can consist in multiform flow sets that take a considerable less time to run, from a large scale such as production forecasts to a pore scale like fluid phase redistribution inside the reservoir. From the latter, we would like to point out the tertiary migration of hydrocarbons, which forms the basis of production resumption potential and could have a high demand given the existence of many abandoned oil reservoirs from primary production that have a low recovery factor, without having secondary production undergoing.

Keywords: statistical methods, method kriging, artificial neural networks, absolute permeability distribution maps