An infill well project on a turbidite gas reservoir has been intensively studied. The formation is a highly stratified sand shale sequence with a typical bed thickness of the order of 10cm. Over a period of 20 years, the reservoir has produced 2.0BCM of gas, or 46% of the original in place. However, all of the seven producers died out due to either excessive water production or sanding. A detailed inter-well correlation identified more than three hundreds pay sands, commonly covering more than half of the reservoir. The vertical variation of sand permeabilities, the lateral connectivity of sand bodies, and the lateral extent of interbedding shale barriers are modeled with geostatistical techniques. A high resolution simulation model is constructed with vertical gridding that corresponds to each sand. The model justifies that the early water production is a consequence of water encroachment through a few high permeability sands, leaving a large amount of gas unswept. Substantial reserves are located around the center of the crestal platform where no well exists, instead of the structure top. The production forecast of the infill wells results in 14% incremental recovery. Three important findings result from this study. First, the magnitude of gravity effect must be accounted for in well placement. Second, the behavior of a highly stratified reservoir is very sensitive to the permeability across interbedding shales, Finally, the basic strategy used for reservoir characterization has a direct impact on the flow simulation result.


Aga-Oki, the first Japanese offshore field developed in the Japan Sea, has been producing oil and gas for more than 20 years (Fig. 1). More than ten producing horizons are embedded in a highly stratified turbidite build up, which is believed to have formed along with the growth of the anticline structure. The field is currently at the final stage of decline and its economic limit is anticipated within a few years. However, a material balance survey indicates the existence of significant remaining reserves in some of the horizons.

N6 sand (Fig. 2) is among such reservoirs and has produced 2.0BCM of gas to date. A P/Z analysis calculates 3.5BCM of original gas in place and the existence of weak pressure support from the aquifer. It also indicates 0.6BCM of potential recovery if production is continued until a typical abandonment pressure is reached. However, all of the seven producers completed this formation has died out due to either excessive water production or sanding. As notable decrease in GWR began in very early stage of gas production, either water encroachment along high permeability sands or gas entrapment due to sand discontinuities have been suspected. In fact, the apparent water-out of the uppermost well (A4) disagrees with both the remaining reserves and the magnitude of water influx calculated from material balance by orders of magnitude when a uniform horizontal GWC is assumed over all sand bodies. Since 0.6BCM of remaining reserves could justify drilling of additional wells under the regional well-head price, it was decided to bring out a detailed reservoir simulation study to evaluate the feasibility of the infill well project.

Reservoir Modeling Strategy

Minimization of computational time has been a major concern ever since numerical reservoir simulation was introduced to the petroleum industry. In these days, that is still more pronounced in view of reservoir characterization. Advance in geostatistics-based reservoir modeling techniques makes it possible to quantify uncertainties in any outcome of a project. Many equiprobable stochastic reservoir images are generated while honoring information from a variety of resources recorded in different format and resolution. The variability of production forecast is obtained by simulating the fluid flow on them. However, it must be recognized that the aspect of geological structure to be focused is strongly subject to the recovery processes of interest. A sound modeling strategy is necessary to obtain an adequate understanding on the reservoir with a minimum computational efforts.

In this particular project, it is clear from the production history that the mechanism by which a large amount of gas has been left behind must be understood. Significant computational efforts have to be allocated to resolve the impact of vertical heterogeneities. Although upscaling and pseudoization algorithms that could reduce computational cost by several orders of magnitude are available, a fine scale simulation model is preferred to derive a definitive solution on the distribution of unswept gas. P. 543^

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