Version française
RSS feeds
Accueil
Home > Expertise > Research divisions > Geosciences > (Multi)scale access to reservoirs

Geosciences

(Multi)scale access to reservoirs

Digital modeling of petroleum reservoirs is an essential tool for optimal oilfield exploitation.

It requires the integration of available static (well log, seismic) data and dynamic (production) data.

These data are used to characterize the many parameters of reservoir models, and in particular to predict the spatial distribution of petrophysical properties.

However, the relationship between these parameters and the data is usually a complex one. Moreover, building a model that is consistent with the data thus requires the simulation of fluid flows for multiple models, which usually imposes a very heavy cost in terms of computing time.

To reduce this cost, it is possible to make use of intermediate digital models of lower resolution based on a coarser mesh size or a simplified description of the physical reality. Using the information generated by these models to facilitate the integration of data in the detailed model is the basis of multiscale modeling approaches.

Methods developed at IFPEN consist of producing the distribution of petrophysical properties(1) in a cascading sequence: the values calculated at a given mesh size provide the trend for generating a new distribution on a finer grid. More specifically, this method makes it possible to adjust the spatial resolution of the parameters to be calibrated to match the level of information contributed by the various dynamic data.

Another investigated approach is the construction of meta-models known as multi-fidelity models to approximate the results produced by the simulator at the most detailed level(2). These meta-models are defined by combining flow simulations performed with the detailed reservoir model and with lower resolution models. Integration of the less costly simulations reduces the demand on the simulator at the most detailed level to deliver appreciable time savings.

Given the demand for increasingly accurate and realistic reservoir models, multiscale methods appear to offer high-potential resources.

  

Scientific contact:  veronique.gervais@ifpen.fr

Published in Issue 22 of the information letter Science@ifpen - September 2015

Publications

  • (1) C. Gardet, M. Le Ravalec, E. Gloagen, Multiscale Parameterization of Petrophysical Properties for Efficient History-Matching, Mathematical Geosciences, 2014, 46(3), 315:336.
    >> DOI: 10.1007/s11004-013-9480-3
  • (2) V. Gervais, A. Thenon, M. Le Ravalec, Quantification of Uncertainties in Reservoir Models from Multi-fidelity Response Surfaces, Second EAGE Integrated Reservoir Modelling Conference, Dubaï, 2014
    >> DOI: 10.3997/2214-4609.20147475

links list

  • Print page