Braided fluvial reservoirs are important reservoirs around the world and being high net: gross, are often considered to be massive and relatively easy to characterise. This study shows that this is not always the case. Two wells in the offshore part of the fluvial Triassic Sherwood Sandstone reservoir in the Wytch Farm Field, Southern UK, have very similar permeability distributions with arithmetic average permeabilities of 400 and 625md, respectively. The effective permeabilities exhibit a similar and typical degree of heterogeneity in this fluvial reservoir in terms of core permeability. However, their well test effective permeabilities are radically different (44 and 1024md).
A full well test analysis is performed for each well, deriving all possible models that can be interpreted from the pressure derivative. The well testing software used (pansystem TM) allows filtration and resampling of the pressure data and the interpretation of the pressure derivative curve and identification of early, middle and late time flow regions. The models interpreted from the well test analysis are integrated with flow unit models derived from the core permeability, core sedimentology logs and knowledge of the reservoir geology. The flow unit interpretation can be directly related to the well test results. The spatial distribution of the permeabilities is different in the wells and this explains why the probability distribution functions alone fail to discriminate between the wells.
This work addresses the challenge of estimating effective permeabilities in high net:gross braided stream reservoirs. Permeability in this particular field varies by seven orders of magnitude and there is high grain size derived variability at the core plug scale. Interpretation of the well test in such environments can be greatly helped by an improved petrophysical description (probe permeameter), careful screening and averaging of the core data and identifying the flow units (which may differ from conventional net pay). In addition, we observe that the occurrence of negative skin in such environments can be attributed to natural simulation due to sedimentological features. We point out that the running of a production log (PLT) during the testing will greatly reduce the uncertainty in flowing interval and the test-derived parameters.
This work shows that the integrated interpretation of well test pressure data in a geologically-coherent fashion can enable the flow characteristics of the reservoir to be determined at the appraisal stage. This knowledge can then be incorporated in the model of the reservoir for improved performance prediction.