火山岩貯留層のモデリングとフローシミュレーション：多点法地球統計学と確率摂動法の応用

Abstract

A domestic gas field was modeled and flow-simulated. Its reservoir body comprises stacked rhyolite lava domes erupted under submarine environment. Porous network developed inside and chilling by seawater formed hyaloclastite to deposit around it. Although hyaloclastite is also porous, its permeability has been dramatically reduced by clay minerals. Impermeable basaltic sheets and mud seams are also present. Each facies plays a specific role in the pressure system of the field. Stratigraphic correlation originally identified multiple reservoirs. Gas has been produced according to priority assigned to each, assuming that it is sealed from others. However, it is noticed by now that pressures of unexploited units have also declined with variety of rates. In addition, some unusual local pressure behaviors have been recorded. It was decided to re-model the whole pressure system to reasonably explain these observations. Combination of multi-point geostatistics and probability perturbation theories was employed. It successfully reproduced non-linear features of stacked lava domes, while allowing pressure data for controlling geo-body distributions in larger scale. A common difficulty of building a proper training image, further pronounced for a volcanic reservoir, was solved by iteratively adjusting its prototype until the number of perturbations for history matching was minimized. Ambiguous reservoir boundary, due to lava growth in random directions, was stochastically represented by populating a predetermined modeling space with both pay and non-pay pixels. Resulted realizations closely simulate pressure history to every detail and look realistic in facies distribution. They ascribe the uneven pressure decline between different units to tortuous flow channels connecting them. They also uncovered the cause of the unusual smaller-scale pressure performances. OGIP evaluated by 20 equally probable realizations stay within ±15% around the mean. Incremental recovery by adding a new well indicates wider scatter than that by installing a booster compressor, which quantifies geological risk associated to it.