Modern Approach to Estimation of Uncertainty of Predictions With Dynamic Reservoir Simulation—A Case Study of a German Rotliegend Gas Field

Abstract

Abstract A deep German onshore Rotliegendes gasfield in a complex sedimentary and structural environment is subject to evaluation of the remaining potential for additional wells and possible recovery acceleration. A graben structure filled with aeolian sediments in an overall depth of about 4900m comprises the dry gas bearing reservoir. The field is developed by a limited number of expensive wells with a relatively wide well spacing. The wells face complicated drilling conditions and a necessary deviation to honour environmental issues which severely limit the number of well site locations. The quality of the 3D seismic is limited because of Zechstein salt intrusions in the overburden. Hence the structural and interpreted fault models are subject to significant uncertainties. Characterisation of aeolian sedimentary facies distribution and quality, which suffers from diagenetic effects, is achieved by a geostatistical approach. The model is mainly based on information from cores and logs. Seismic information could only be used to guide the distribution of the aeolian reservoir facies and a probably unproductive fanglomerate facies type. One main question is the level of compartmentalisation of the field due to faults and flow barriers related to reservoir facies distribution. During the dynamic modelling process possible scenarios are checked against the observed production data regarding communication and GIIP distribution. This gives a range of possibilities leading to a significant uncertainty regarding the prediction of future production and the effect of acceleration wells. There is room for only a few more wells. The decision where to place these is supported by the integrated modelling approach. This has been chosen to address a wide range of uncertainty of both static and dynamic effects. The static effects have been addressed by a combination of geostatistical static reservoir modelling with fully 3D static uncertainty estimation. The basis for the dynamic modelling part is a family of realisations defined by a process to identify "most diverse" representative scenarios out of bigger number of geological realisations. Computer aided history matching, experimental design and proxy solutions gave a very effective approach to handle the uncertainty estimation with the dynamic reservoir simulation. The fully integrated workflow allowed an optimized process to estimate the uncertainty of prediction within a limited time frame. The main subject of this paper is the process of dynamic reservoir simulation using these techniques and discussion of the results.