The development of fluvial stochastic modelling in the Norwegian oil industry: A historical review, subsurface implementation and future directions

摘要

Fluvial sandstones are an important reservoir type for the petroleum industry. In the late 1970's and early 1980's, large hydrocarbon discoveries in the Norwegian North Sea in fluvial strata prompted the need for generating geologically meaningful, stochastic, object-based models of fluvial deposits. The aim of this focus was to allow the geologist to provide the reservoir engineers with a more realistic representation of permeability contrasts within channelised, fluvial deposits by being able to use appropriate measurements from outcrop analogues as direct input data into the modelling software. This initiative resulted in the development of a suite of geologically driven, stochastic modelling algorithms supported by an extensive fieldwork program aimed at collecting stratigraphic and quantitative data from ancient outcrop analogues to support enhanced reservoir characterisation and geological modelling. Today, these reservoirs are still important hydrocarbon producing fields with accurate reservoir description and 3D modelling capabilities playing a vital role in targeting remaining oil, especially now that many of the fields on the Norwegian continental shelf are past peak production and are in a decline phase. As both computing capabilities and quantitative outcrop analogue studies have increased the understanding of, and the ability to model fluvial reservoirs, so have stochastic modelling techniques continued to provide the most suitable and robust means of building geologically realistic 3D reservoir models that incorporate increased geological understanding and heterogeneity complexity, hi the recent past, a multitude of data, such as seismic and production data have been used to condition the stochastic algorithms. This review paper aims to outline the role of stochastic algorithms in building geologically-realistic, 3D fluvial reservoir models and highlight the success of these developments with case studies from both producing fields and ancient outcrop analogue studies. Finally, the paper will allude to possible improvements in stochastic fluvial modelling and future directions in the modelling of fluvial petroleum reservoirs. These include the use of physical or process-based models, high-resolution near wellbore models, and multi-point statistics, that allow for more realistic representations of heterogeneities of fluvial deposits at a variety of scales and by a variety of methods.