Brownfields are often characterized by a varying degree of maturity, both within the field and within individual reservoir units. This variation makes infill drilling more prospective in areas with fewer well penetrations and completions and less production. However, these areas are inherently more uncertain, with geological, petrophysical, and structural parameters particularly affected. A novel workflow solves the complex problem of uncertainty assessment and risk management in a brownfield redevelopment. Traditionally, a single deterministic reservoir model is built, matched, and used for predictions and infill planning. The availability of sophisticated simulation workflow tools enable the team now to explore the practical aspects of performing sophisticated reservoir description, static model construction, history matching, and forecast uncertainty analysis. Incorporated into multiple equiprobable reservoir descriptions, uncertainties are carried from the static model construction throughout the entire dynamic modeling process. History matching is conducted for all realizations, and the match quality is assessed by means of statistical analysis. The workflow facilitates generating hydrocarbon thickness maps by using the average column thickness of many simulation models instead of a dedicated single one. Target selection also accounts for possible sweep and sand risks by means of maps showing the standard deviation of the column thickness. The new framework is applied to a conceptual redevelopment of a brownfield. It increases the understanding of fluid flow processes in the reservoirs and is a vital component of the decision and risk analysis for the concept selection stage.


History matching is traditionally conducted as a deterministic process with a single realization considered representative. It is well known that matching a particular model does not give a unique solution and, more importantly, that a good history-matched model could give a bad forecast, as shown by Tavassoli et al. (2004). Many researchers have dealt with the problem of history matching and uncertainty assessment in reservoirs. The focus has traditionally been in assisted or automated history-matching techniques. Bissel et al. (1992) demonstrated the usage of gradient methods for this purpose. Portella and Prais (1999) combined a history-matching technique with geostatistical modeling to provide equiprobable reservoir images, taking production data into account. The method includes an automatic history-matching procedure based on simulated annealing. Datta-Gupta et al. (1999) presented a streamline-based method to integrate production data into reservoir models using inversion techniques. Schulze-Riegert et al. (2001) applied evolution strategies to the problem of history matching based on genetic algorithms on parallel processors. Feraille (2003) presented an integrated methodology for constraining 3D stochastic reservoir models to well data and production history using a gradual deformation method. Sahni and Horne (2004) discussed an automated solution to generate multiple history-matched reservoir models with the inclusion of both geological uncertainty and varying levels of confidence in the production data by means of wavelet methods.

In this paper the problem is addressed of transferring uncertainty in the geological model through the flow simulation model up to the reservoir production forecasts. The uncertainty in the geological model is characterized by the differences among 15 equally probable reservoir descriptions that honor available data. While matching them all individually is prohibitive, a workflow is proposed that focuses on achieving an acceptable match for one representative model and transferring the match on the remaining realizations.

This content is only available via PDF.
You can access this article if you purchase or spend a download.