Miscible gas injection in the Kashagan platform started in 2017 and has been ongoing for three years. Gas injection in the platform carries both EOR and a facilities de-bottlenecking component. Initial flood assessment studies done using classical FD simulation have not provided the full picture of the reservoir dynamics to understand the performance of miscible gas injection. Areal reservoir depletion, pressure support and gas breakthrough events were challenging to quantify and characterize in terms of standalone full field FD simulation. Reservoir management strategy for operating the gas injection area was to maximize production from nearby producers and inject the full compressor potential. However, due to the fact that wells have different potentials driven by reservoir heterogeneity, areal distribution of cumulative reservoir fluid withdrawals and injection ended up being significantly different, which lead to flood pattern imbalances. The operator has implemented a modeling workflow that combines post-processing of history matched numerical simulation model with streamline tracing and integration of time-lapse well allocation factors (WAFs) to quantify and analyze flood performance. This paper presents how to use streamline modeling to estimate well-pair dynamic control volumes and a numerical integration workflow of the dynamic WAFs to evaluate pattern performance to guide flood balancing strategy. Streamline-based modeling workflow provides additional value by 3D visualization of the dynamic flood patterns and quantification of the individual pattern metrics. Numeric integration of injector-producer allocation factors (WAFs) and control volumes (CVs) allowed the construction of well pair conformance plots. Ranking of the patterns and I-P pairs filtered the outlier patterns with over injected and produced volumes and helped to focus on specific areas in need for pattern balancing. A list of producers with the highest pore volumes of gas injected were identified as at-risk wells for gas breakthrough and GOR elevation, which was confirmed by well test results. The first three wells with a rise in GOR and breakthrough sequence perfectly matched with the prediction of pattern performance. Those were identified as at-risk producers based on streamline modeling outputs. Verification of the analysis by field surveillance data gave confidence in reliability of the streamline-based flood evaluation approach. The outcomes of this study helped to understand miscible gas front movement and depletion dynamics in the gas injection area. This case study demonstrates how complementing finite-difference modeling with streamline analysis is necessary for achieving a complete assessment of the miscible gas flood performance.

Modeling and simulation of non-Darcy or turbulent flow are well documented in the literature and available in commercial reservoir simulators (E300, Intersect) only for gas wells rather than oil wells. There is a need to model non-Darcy or turbulent flow in reservoir simulation for oil wells in the carbonate reservoirs with highly connected and densely distributed fractures and karst. This paper proposes a new non-Darcy or turbulent flow modeling and simulation method for oil wells. Unlike the industry's existing methods for non-Darcy or turbulent flow that focus on the non-Darcy coefficient only, this paper presents a new method that models the ratio between non-Darcy and Darcy flows such that a unified model for a field or a region can be created, which significantly simplifies the non-Darcy or turbulent flow modeling process for multiple wells, especially for future wells. The ratio-based method is simple and comprehensive. It can be easily calibrated with MRT (multiple-rate test) data and implemented into in-house or commercial reservoir simulators using a simulator supported scripting language, e.g., Python etc. Kashagan is the world's largest oil reservoir discovered in the last 30 years that contains highly connected and densely distributed fractures and karst in its rim. The oil production rate for a well in the rim can be higher than several tens KSTB/D if it is not constrained by the facility. The current MRT data in all tested wells clearly show non-Darcy flow phenomenon and confirm that modeling non-Darcy flow is necessary to the field. Kashagan had experienced difficulties to match BHP (bottom hole pressure) and large errors in the blind test due to the OPEC's production curtailment and high-rate tests. Build-up pressure curves were miss-matched and HM (history match) of the crossflows (10 KSTB/D with less than 10 psi) in the bottomhole of a PLT (production logging tool) well during shut-in was challenging. Since modeling non-Darcy flow for oil wells in the commercial simulators, e.g., E300 and Intersect, is unavailable, the simulation team in NCOC has created a new method for the needs of non-Darcy modeling and simulation. The applications of the new method have resulted in the excellent results and solved the issues of history matching BHP, high/low-rate tests, build-up pressure trends, and bottomhole crossflows.

Dual porosity numerical models are widely adopted in the oil business to model the performance of complex systems characterised by two different porous media. However, for numerical models of a certain complexity due to the large number of active grid-blocks, the dual porosity approach is often computationally unaffordable especially when a compositional formulation must be used. This paper describes the methodology that was developed to mimic the dynamic performance of a complex triple porosity system by means of a single porosity model. The three porous systems were the matrix blocks, the fracture network and the dissolution karst bodies. The methodology was derived for a complex massive carbonate field not yet producing which is currently envisaged to be developed via miscible gas injection. The matrix and karst bodies were statically modelled independently from the fracture network system. The characterisation of the fracture network has been driven using a DFN approach by integrating seismic, continuity cube interpretations and well data, such as FMI and mud losses. Due to the lack of dynamic data, the fractures' petrophysical properties were calculated from correlations. The matrix, karst bodies and DFN derived fractures were then up-scaled to a dual porosity model. The dual porosity model dynamic performance was considered as the reference to be matched by an equivalent up-scaled single porosity model. An innovative procedure to up-scale matrix, fracture and karst properties into the equivalent single porosity model was tested for both a natural depletion and a miscible gas injection scenario. The methodology was firstly evaluated in representative sector models and then extended to the full field model. This methodology resulted to be very efficient being able to reduce the simulation time and model complexity drastically while capturing all the dynamic key performance indicators of the more complex and computationally expensive dual porosity model. Introduction In the recent years the detailed acquisition of field data and the use of specialist software packages enable geoscientists to characterize the distribution of fractures at different scales and to up-scale the information to field simulation models. Among the different techniques, the use of Discrete Fracture Network (DFN from now on) models is the most advanced one since it allows to model the complexities, heterogeneities and properties of the natural fractures at different scales of investigation. The definition of a DFN or several DFN models requires the joint effort of several disciplines from seismic, to structural geology, from core analysis to petrophyisics. DFN models represent the starting point for an upscaling process with the final aim of performing 3D numerical simulations necessary to understand the dynamic performance of the network. The optimum workflow would require to up-scale the fracture properties in terms of porosity, permeability and spacing to a dual porosity model. In the last 5 years this has become a quite common methodology which finds several applications worldwide. The field under investigation is envisaged to be developed via miscible gas injection; in order to properly model the complex thermodynamic of such process, a compositional model is required. In addition, the size of such field and the lateral and vertical reservoir heterogeneity require the use of a very large number grid-blocks in the dynamic model. As a matter of fact, the environment of this field can be regarded as a triple porosity system which consist of matrix, fractures and dissolution karst bodies. The DFN approach was adopted to characterize the fracture system. However, the use of the dual porosity option in a compositional model of such complexity is not computationally feasible and is unpractical. An innovative methodology was investigated for the simulation of the dynamic performance of such complex triple porosity system by means of a single porosity model. This methodology was tested for both a natural depletion and a miscible gas injection scenario. The methodology consists of defining the petrophysical properties of an "equivalent single porosity model" with performances identical of the more complex and computationally expensive dual porosity one (Figure 1).

No abstract

TX 75083-3836, U.S.A., fax 01-972-952-9435.