Adopting “big loop” reservoir modelling workflow to handle subsurface uncertainty of the Gobe Field in PNG Highlands

Abstract

One of the fields in the greater Gobe area is the Gobe Main field. It is in the north-western area of Petroleum Development Licence 4, partially in both the Gulf and Southern Highlands provinces of Papua New Guinea. The reservoir is formed by the Gobe anticline, elongated, west-north-western trending surface expression fold in which Tertiary Darai Formation is exposed. The surface of the anticline is covered with weathered limestone producing rugged terrain covered with dense rainforest. The field is producing from two sandstone horizons, the Upper and Lower Iagifu Formations separated by a shale zone. Though originally developed in 1998 as an oil field under solution gas reinjection and gas cap expansion drive mechanisms, currently the Gobe Main field is predominantly in gas cap blow down stage feeding gas to the PNG LNG Project. The Gobe Main field has an abundance of reservoir performance data obtained from more than a dozen producing wells from about 20 years of field production history. At the same time, field subsurface description is carrying a significant structural uncertainty attributed mainly to the following factors: • Difficulty positioning seismic data acquisition equipment due to rugged terrain and dense vegetation; • Seismic data interpretation uncertainty due to signal degradation within the Darai Formation; • Structural discordance between surface geology and reservoir. In the recent work, a reservoir modelling study was initiated to optimize the late life exploitation of the Gobe field and better assess its remaining potential. The most important parameter that affected the study outcome was found to be the abovementioned structural uncertainty. There is a range of industry accepted solutions for managing subsurface uncertainties. Two end members of the range are: • A single realisation-based, deterministic two-step reservoir modelling approach where geological uncertainties are lumped together into one parameter-grid. This method has well-known limitations and may lead to a history match with lack of geological consistency and consequently a biased forecast. • A multi-realisation-based automated workflow, often referred as "big loop" that allows simultaneous direct analysis of static and dynamic uncertainties and to achieve a history match with geological consistency. Some consider the latter to be a more robust approach to simulation studies leading to improvement in reservoir management decision quality. But the method essentially requires parameterising subsurface uncertainties to enable an automatic solution for geological and reservoir engineering uncertainty scoping, followed by assisted history match and forecasting assessing the full range of probabilistic outcomes for field exploitation purposes. The nature of structural uncertainty in the Gobe field resulted in difficulty in describing it as a fully parameterised variable, at least in the given time frame, hence restricting applicability of that method. To complete the study and develop a numerical model that could become a robust tool for reservoir management purposes, a hybrid method was developed. This method was based on testing alternative deterministic structural realisations in the dynamic space through history matching them on reservoir performance data and improving those structural model realisations in the uncertainty between well control points based on history matching feedback. Such a iterative model refinement process was repeated till the best subsurface static model description could be obtained and history matched while preserving geological consistency in presence of significant structural uncertainty.

AAPG Datapages/Search and Discovery Article #90371 © 2020 AAPG Asia Pacific Region, The 1st AAPG/EAGE PNG Geosciences Conference, PNG’s Oil and Gas Industry: Maturing Through Exploration and Production, Port Moresby, Papua New Guinea, February 25-27, 2020