Many oil-field operations involve the injection of fluids into the formation around a well. In many cases, the fluids contain colloidal particles, either initially present or introduced during the operation through dirt or naturally-occurring particles. Therefore, all injection schemes suffer from injectivity decline. This injectivity decline is caused by clogging of the formation by particles, which form an external filter cake on the surface of the formation and block the pores inside the formation. This article reports on a set of crossflow experiments in which a suspension of particles is flowed over the surface of a sandstone core, with a portion of the suspension entering the core itself. An external filtercake is formed on the surface of the core, and internal deposition occurs inside the core. The experiments are conducted in a CT scanner, allowing measurement of the internal deposition in real time. Thus, although the time-dependent build-up of the external filtercake thickness cannot be measured directly, it can be deduced by mass balance, allowing for the crossflow and permeant effluents. This enables correlation of the external and internal mass deposition with the measured increase in injection pressure. We use the cumulative mass curves of the hematite deposited in the cake, core and in the effluent to model the dependence of the pressure drop on cake and core deposition, and compare the predicted pressures with the actual measured pressures. These results are very relevant to operational issues associated with injection or leak-off, including hydraulic fracture propagation, drilling-mud impairment and produced water matrix re-injection. The results are also relevant to water injection under fracturing conditions, since impairment of the wall of the fracture is one of the mechanisms governing fracture growth, and the implications of the results for this process are discussed.

The subject field of this study is a geologically complex Lower Cretaceous carbonate system of stacked reservoirs and dense seals, which have been affected by extreme diagenetic modification, including various types of faults & fractures. The purpose of this paper is to summarize several years of comprehensive study of identification, characterization and modeling of different fault/fracture systems within the reservoir stack, integrating different data types, conceptual models and geomodelling approaches for individual, or groups of, reservoirs. Timely identification and appropriate characterization of the fracture-fault systems is extremely important for correct well placement and completion strategy. The field has been producing for more than 30 years under a water-flood recovery strategy. Numerous amounts of different types of data including seismic, cores, image logs, tracers, interference tests, production/injection data etc. has been collected and studied carefully throughout those years. The integration of different data sets combined with regional structural study suggest existence of four different types of fault and fracture systems, with predictable and consistent orientations, that affect the reservoirs: Seismically mapped faults, fracture corridors, diffuse natural fractures, and thermally generated/artificially enhanced fractures. Their impacts on field performance vary, being dependent upon scale (length/throw) and stratigraphic positioning. For example, faults with large throw (30-60ft) can create inter-reservoir communication due to fault juxtaposition whereas diffuse fractures that are generally located at the top and base of the reservoirs (due to mechanical contrast between dense and reservoir rock strength) act as intra-reservoir permeability enhancement. Fracture corridors that can be either restricted to, or cross, reservoirs are most difficult to detect due to small offset and (when oriented NNE) possibly impose greatest impact in terms of flow heterogeneity. Thermally enhanced fracture effects are observed where newly drilled wells enter the water-flood affected area. These fractures can be seen on the image logs, especially where they fracture most brittle dense intervals limestones and their impact in reservoir is noted most from production water cuts. Modeling different types of fractures for simulation studies and field development planning are always very challenging due to limited availability of the critically appropriate data. As a result, an integration of different data types was key to identify and model different faults and fracture systems. Numerous faults were mapped from seismic but only those mapped with high confidence, supported by drilling reports and LWD log evidence, and which show impact on production performance, were included in the model – all faults are considered for other purposes (e.g. well planning). Diffuse fractures were modeled combining traditional methods (fracture density from cores) with ant-track attribute from seismic to determine directionality and distribution away from cored wells; and span the scale range to through-going fracture corridors. A new method has been introduced to assign effective permeability to fractures from seismic attributes by scaling up and down to the well test permeability. The displacement functions for fractures are estimated as effective pseudo-functions, representing the diffuse fracture and matrix media and conditioned to the performance of the well observing such fracture effects. Fracture corridors, whenever observed, were modelled deterministically as effective properties. Given the uncertainties associated with data and sub-surface knowledge, coupled with the classification of all reservoirs in the stacked sequence being fracture-assisted matrix zones, the objective of each and every fracture model approach was to create a representative, simple and fit-for-purpose model that can be modified and updated easily during history-match iterations and that serves business needs.

Lower Cretaceous-aged carbonate sediments in a supergiant Middle Eastern oil field are characterized by extensive diagenetic overprints (e.g., dolomitized burrows, dissolution fabrics and fractures) which occur over areas of several kilometers. Due to the permeability contrast with respect to surrounding fine-grained matrix, the diagenetic features are believed to play an important role in reservoir fluid-flow, particularly as a major contributor to early water breakthrough observed in the field. Uncertainties associated with three-dimensional subsurface reservoir models can be mitigated by incorporating the results of detailed reservoir characterization studies. How these study findings are successfully and meaningfully implemented into the reservoir model can be a challenge and requires an integrated effort by reservoir geologists, modelers, and engineers. This paper discusses a comprehensive reservoir characterization and modeling study conducted to capture the impact of diagenetic features on reservoir flow properties. A novel method was developed to map the spatial and stratigraphic distribution of these features from cores. Dolomitized burrows generally appear in core as randomly oriented features on a scale of cm to 10's of cm. They are characterized by grainier fill (packstone or grainstone), often dolomitized, within a background of muddier sediment (wackestone to packstone). Dissolution vugs are associated with algal rock type and can vary from few cms to 10's of cm. Fractures are generally layer specific and occur at the reservoir-dense boundaries. The origin of diagenetic processes and prediction of their occurrence is very difficult. However, the difference in texture and associated pore characteristics lead to heterogeneous porosity and permeability regimes that can have significant impact on sweep efficiency and recovery in oil fields that are subjected to waterflood. A simplified, fit-for-purpose, rapidly updateable static model has been developed to ensure accurate stratigraphic and lateral distribution of diagenetic features based on cores, logs and dynamic data. Whole core data revealed potential guidance for assigning permeability in diagenetic features. A consistent SCAL framework has been developed to capture the relative effects of these diagenetic features on flow. After incorporation in the model, simulation results clearly shows water movement through these features and rapid water cut. This is in agreement to the field observation that has experienced earlier than expected water breakthrough and steady increases in water cut over time.

In the current practice, ICD/ICV design parameters (e.g., number of compartments, compartment size, number of nozzles, and nozzle sizes) are optimized by a manual trial-and-error approach that requires tens to hundreds of iterations. To make the design process efficient and effective, an automated optimizer is desired. In addition, as more and more ICD/ICV wells are completed, reservoir simulation faces a challenge on how to efficiently run full field models with multiple ICD/ICV wells. This paper presents a new automated ICD/ICV design optimizer and an efficient way to run full field reservoir simulation with hundreds of ICD/ICV wells. The new optimizer uses oil recovery efficiency as its objective function. The optimizer works on injectors and producers separately. For injectors, the optimizer adjusts the packer locations, number of nozzles, and nozzle sizes to make the injection velocity along the wellbore as uniform as possible to ensure a uniform injection front. For producers, a five step optimization process is applied. Step 1 is to generate injected fluid flow travel times in 3D from injectors to producers and all major flow "highways" are identified. Step 2, the optimizer uses fluid travel times in a producer to automatically estimate number of compartments needed and adjust the compartment boundaries (packers) to match the "highways" identified, estimate number of nozzles needed and initial nozzle sizes to maximize oil production rate. No reservoir simulation is required in steps 1 and 2. Step 3 is to run a full field reservoir simulation with all design wells to tune and achieve the final nozzle sizes. Step 4 is to QC and analyze the results of all ICD/ICV wells and select all successful candidates for the final step, i.e., step 5 reconciliation of the designs with all other drilling/completion constraints. The optimizer is fully supported by the efficient well management logic which accurately and efficiently links ICDs/ICVs with reservoir simulation. Using the well management logic removes the needs of coupling between well simulation tools (e.g., NETool) and reservoir simulation software, and then makes full field simulations efficient. The new optimizer and well management logic have been applied and demonstrated significant values in a giant oil field in UAE. Compared to the traditional one-well-at-a-time well design, the new optimizer optimizes multiple ICD/ICV design wells at a time and results in better and faster designs with speedups in a range of several factors to an order of magnitude. The optimization is global and within the context of full field model. Running 370 ICD/ICV wells with the well management logic for a multi-million-cell reservoir simulation model only slows down the full field simulation around 10%.

As part of the ongoing development of a large offshore oil field, an asset owner places a strong emphasis on continuous improvement of the established framework for integrated post-drill well analysis. The geology of the candidate field is complex and the occurrence and distribution of the extreme permeability features that dictate early water production is highly uncertain. While much effort is devoted to mitigating their adverse impact through proper integration of surveillance data for accurate well planning, post-drill outcomes can still diverge significantly from pre-drill expectations. Several wells have been drilled in the production build-up campaign, including ground-breaking pilots and many more are following in very quick succession as part of the life cycle strategy for the field. Due to high drilling frequency, the challenges of assimilating learnings through conventional post-drill analysis for optimization of future drill wells can be enormous. To apply key lessons from these wells in building quick baseline knowledge for reservoir model update and drill plan optimization, the modeling and development team have developed an improved workflow for integrated post-drill analysis. The workflow leverages the full benefit of collaboration between multi-disciplinary teams to integrate 3D seismic data, multiple well information (including geologic reports, well logs and petrophysical results) and surveillance data from new drill wells to benchmark pre-drill expectations. An important aspect of the approach is the quick incorporation of drilling results into static and dynamic models via a cycled, closed-loop workflow for quick assessment of model fidelity through an evergreen update process. A multifunctional post-drill analysis facilitates critical consideration of well results to capture significant learnings that influence future drill well and data acquisition optimization, reservoir model history match and prediction enhancements, and identification of drilling hazards and geological features that affect reservoir performance. This paper describes the methodology used to plan and implement post-drill well analysis within a fast paced and high drill frequency environment. Key elements of the methodology are described through the use of a case study example, and include: Standardized subsurface workflow, comparison of post-drill well results with pre-drill well expectations, identification and documentation of significant observations and lessons learned improvement of history match & predictive capability of reservoir models and integration with other drill-well delivery processes.