Scholarly record

MULTITHREADING ANALYSIS OF SEISMIC DATA ON THE HYBRID SUPERCOMPUTER

Yuri Ilyushin

Abstract

When preparing the field for industrial operation, it is necessary to build a conceptual model that reflects the internal structure of the objects under study. Conceptual modeling, in turn, is based on the methods of system analysis and interpretation of all available geological, geophysical and seismic information in the study area using various types of system analysis, which allows to predict changes in lithology and filtration-capacitive properties of rocks, as well as assess the reserves of the field. The conceptual geological model summarizes the sedimentation, petrophysical and seismic-facies models, combining them in the 3D geological model of the field. A characteristic feature of such model is its versatility both in the choice of the proper attributes and in the spectrum of the tasks for which it is used. Given the large amount of geophysical data, the use of high-performance systems and parallel algorithms is an integral part of their analysis. As an example, consider the Kislovodsk mineral water field, a 2D survey of which was carried out in 2018: the observation area was approximately 5 thousand square kilometers, and the total amount of data collected is more than 10 terabytes. In this article, we will use Nvidia CUDA technology to process seismic data. That will require significant processing of the existing algorithm. In this regard, the development of methods for organizing parallel computing, simplifying the construction of parallel programs for processing seismic data, is an urgent task. The initial data for processing were obtained by aerospace photography by means of remote sensing. The method described in this article was tested at the mineral water extraction company of the JSC ?Narzan-Hydro Resources? city of Kislovodsk. Software algorithms given in this article have copyright certificates issued by the Federal Institute of Industrial Property of the Russian Federation.

Publication Impact Profile

• Captures
• Mendeley - Readers: 3

PLUMX see details

DOI resolver: doi.org/10.5593/sgem2019/2.1/s08.126

Publication details

Back to publication list