3D Multi-Level Facies Constrained Geostatistical Inversion for Predicting Carbonate Thin Reservoirs

Carbonate reservoirs are highly heterogeneous and poor in interwell connectivity. Therefore, it is difficult to predict the thin oil layers and water layers inside the carbonate reservoir with thickness less than 10 ft by seismic data. Based on the petrophysical analysis with core and well logging data, the carbonate target layers can be divided into two first level lithofacies (reservoir and non-reservoir) and three second-level lithofacies (oil, water and non-reservoir) by fluids. In this study, the 3D lithofacies probabilistic cubes of the first level and second-level level lithofacies were constructed by using the simulation method of well-seismic cooperative waveform indication. Afterwards, constrained by these probability cubes, the prestack geostatistical inversion was carried out to predict the spatial distribution of thin oil layers and water layers inside the thin reservoir. The major steps include: (1) Conduct rock physics analysis and lithofacies classification on carbonate reservoirs; (2) Construct the models constrained by two-level lithofacies; (3) Predict thin reservoirs in carbonates by prestack geostatistical inversion under the constraint of two-level lithofacies probability cubes. The prediction results show that through the two-level lithofacies-controlled prestack geostatistical inversion, the vertical and horizontal resolution of thin oil layers and water layers in carbonate reservoirs has been improved significantly, and the accuracy of thin oil reservoir prediction and the analyzing results of interwell oil layer connectivity have been improved significantly. Compared with the actual drilling results, the prediction results by 3D multi-level lithofacies-controlled inversion are consistent with the drilling results, and the details of thin carbonate reservoirs can be predicted. It has been proved that this method is reasonable and feasible. With this method, the prediction accuracy on thin reservoirs can be improved greatly. Compared with the conventional geostatistical inversion results, the 3D multi-level lithofacies-controlled inversion can improve significantly the vertical and horizontal resolution of prediction results of thin reservoirs and thin oil layers, and improve the reliability of interwell prediction results. For the prediction of thin carbonate reservoirs with serious heterogeneity, the 3D multi-level lithofacies-controlled inversion is an effective prediction method.

A Multi-Process Reservoir Modelling Workflow as the Key for Unlocking Reservoir Prediction in Carbonates. Application to a Sector Model from the Albion R&D Project

In carbonate reservoirs, because of the diversity of geological processes involved in the reservoir construction, the extrapolation of properties directly from well data to reservoir model gridblocks may lead to poorly predictive reservoir properties and then production forecasts. This paper proposes a modelling workflow in which new tools from disruptive technologies are associated in order to produce reservoir models consistently with reservoir geological construction. The workflow combines the simulation of the depositional facies and their transformation after diagenesis overprint. Original depositional facies are carried out from SED-RES™, a stratigraphic forward modelling software that generates and transports carbonate sediments according to ecological conditions and wind-induced currents. Then GODIAG™, a lattice gas, reproduces the evolution of the properties of the sediment after it has been deposited. The diagenesis history can be multi-stage and can involve different kinds of physical and chemical reactions. This new workflow has been evaluated in the framework of the ALBION R&D Project dedicated to the study of the Barremian-Aptian rudist-rich carbonate platform from south France that is known as an analogue of the Kharaib and Shuaiba reservoirs (UAE). Thanks to its multi-scale and multi-site aspect, ALBION offers the opportunity to test new modelling tools. The efficiency of the new workflow has been successfully applied on a sector model from an ALBION site on which a rich geological and petrophysical dataset is available from outcrops and numerous wells,

Deep carbonate reservoir characterisation using multi-seismic attributes via machine learning with physical constraints

Seismic characterisation of deep carbonate reservoirs is of considerable interest for reservoir distribution prediction, reservoir quality evaluation and reservoir structure delineation. However, it is challenging to use the traditional methodology to predict a deep-buried carbonate reservoir because of the highly nonlinear mapping relationship between heterogeneous reservoir features and seismic responses. We propose a machine-learning-based method (random forest) with physical constraints to enhance deep carbonate reservoir prediction performance from multi-seismic attributes. We demonstrate the effectiveness of this method on a real data application in the deep carbonate reservoir of Tarim Basin, Western China. We first perform feature selection on multi-seismic attributes, then four kinds of physical constraint (continuity, boundary, spatial and category constraint) transferred from domain knowledge are imposed on the process of model building. Using the physical constraints, the F1 score of reservoir quality and reservoir type can be significantly improved and the combination of the effective physical constraints gives the best prediction of performance. We also apply the proposed strategy on 2D seismic data to predict the spatial distribution of reservoir quality and type. The seismic prediction results provide a reasonable description of the strong heterogeneity of the reservoir, offering insights into sweet spot detection and reservoir development.

Research and Application of Globally Optimized Sequence Stratigraphic Seismic Interpretation Technology: Taking the Lower Cretaceous Shahezi Formation of Xujiaweizi Fault Depression as an Example

In the study of sequence stratigraphy in continental rift basins, the use of seismic data to track different levels of sequence stratigraphic boundaries laterally is the key to the division of sequence stratigraphic units at all levels and the establishment of an isochronous sequence stratigraphic framework. Traditional seismic interpretation and the establishment of a 3D sequence stratigraphic structure model are a difficult research work. This paper introduces the concept of cost function minimization and performs global stratigraphic scanning on 3D seismic data to interpret horizons and faults in a large grid. Constrained by the results, human-computer interactive intelligent interpretation, by adding iterative interpretation of geological knowledge, established a global stratigraphic model with a relative geological age. The application in the Lower Cretaceous Shahezi Formation of Xujiaweizi fault depression shows that this technology has improved the accuracy and efficiency of sequence stratigraphic interpretation, and the application of this technology has achieved the interpretation of each event horizon under the current seismic data resolution conditions. In this way, a continuous sequence stratigraphic model is established. From this stratigraphic model, any high-frequency sequence-interpreted seismic horizon can be extracted, which provides a basis for the combination of lateral resolution and longitudinal resolution of subsequent reservoir prediction.