| dc.description.abstract | This dissertation investigates methodologies for reservoir simulation and model tuning of wells producing water with hydrocarbon in two particular and extreme reservoir environments: cases of extremely small and very large rock pore size. Both cases are of great practical interest and represent two distinct categories of producing fields; shale gas in tight reservoirs stimulated by slick water fracturing and extra heavy oil with strong aquifer drive in a highly porous reservoir.
In both the investigated cases, water production is a crucial factor. In the shale gas case, optimizing the schedule of back-flowing water has received limited priority and is often managed by intuition. In the heavy oil case, where the water production problem is damaging and costly and often dominates the production management of the field, the industry has not yet established good engineering practices to model, simulate and quantify the phenomenon.
In terms of characteristic pore sizes and the corresponding scale of the fluid flow pathway, the two cases are at the low and the high boundaries of the range encountered in producing hydrocarbon reservoirs. The large pore size range case investigated in this work approaches 1mm (10+3 μm) and the small case is close to seven orders of magnitude smaller, that is 1Å (10-4 μm). The oil industry classifies these two cases as non-conventional, implying difficulties and very complex models to predict fluid transport and storage in these reservoirs.
Conventional reservoir engineering methods utilized in the practical calculations, modeling and simulations of transport and storage of fluids in the reservoir employ the continuum fluid mechanics approach. Accordingly, macroscopic statements of the conservation laws, with certain adaptations to pore matrix flow, can capture the very complex microscopic scale phenomena involved in multiphase flow and the storage of multi-species solutions in porous rock. However, while the results satisfy applied petroleum engineering needs for most of the pore size range, the validity of these methods diminishes towards the extreme boundaries of the pore-size range prevailing in the non-conventional cases, shale gas and heavy oil reservoir flow.
Recognizing the advantages of using the well-established body of knowledge, practices and tools employed in reservoir engineering also for these extreme and non-conventional cases, this dissertation has embarked on an investigation to explore and conclude on the adaptations needed to extend the validity of the conventional tools for predicting water production in these two extreme cases.
The methodology employed in the research was an automated numerical experiment using a reservoir simulator and actual field data extracted from well reports in the two reservoir categories of interest, a North American shale gas formation and the Andean Heavy Oil Belt in South America. A central consideration guiding the numerical experiments was a need to minimize and simplify the parameters in the models and in the numerical tuning procedures. This is in light of the emerging practices to employ a model-based Integrated Asset Modeling (IAM) tool, linking the reservoir, wells and surface facilities models as an integrated tool for decisionmaking and optimization of field design and operations. The need to integrate the reservoir simulation output for the two cases investigated in an IAM system was, in fact, the trigger to launch the research reported here.
In both cases, the model and the simulations address a single well with a horizontal wellbore producing hydrocarbon and water. The tight reservoir case (shale gas) produces reservoir dry gas and water or aqueous solution originally induced into the reservoir by a multi-traverse hydraulic fracturing procedure using “slick” water. In the case of the very porous reservoir, the well produces heavy and highly viscous oil with rapidly increasing quantities of low salinity water originating from an active aquifer. The configuration of the wells in the study complies with the current practices of completing wells in shale gas and heavy oil fields.
A conventional commercial black-oil reservoir simulator was used in this work. 2D finite difference (FD) numerical models with a low number of grid cells (<6000) were built for a single horizontal well. Certain cases were modelled also in 3D, primarily for comparison with the 2D models used.
In the case of shale gas, the goal of the modeling and model tuning was to capture both the schedule of frac-water injection and the water flowback. The proposed approach utilizes two main modeling concepts: stress-dependent permeability and capillary pressure. The stressdependent permeability model has been used to represent the overall effect of the permeability increase occurring during the fracturing treatment in the ‘crushed zone’ in the rock matrix around the hydraulic fracture. It has been found that in the simulation it provided a sufficient injectivity increase to ensure the magnitude of the injected water rates and volumes achieved during actual fracturing operations. It also facilitates the modeling of fracture propagation in a predefined direction. This allows a more realistic pressure and saturation distribution in the shale matrix to be obtained. Inclusion of capillary pressure in the model helps to control the amount of water retained by the rock after the injection, and improves the model predictions of the water flowback rates and BHP. With capillary included, the impact of the water flow and recovery on the shortand long-term gas production performance was studied.
In the case of the heavy oil in highly porous rock, the modeling outcome is oriented towards studying well production performance in fields with a rapidly increasing water fraction driven by the active water aquifer. A set of key tuning parameters for a single horizontal well model has been proposed. Modeling of the oil–water mobility contrast as a function of water saturation was the main approach proposed in this study. It addresses apparent changes of oil and water mobility due to possible effects related to physical phenomena like viscous coupling and dispersed flow.
The dissertation describes the tuning and prediction experiments, and provides the relevant input and output data and information. It then reports the observations, the conclusions obtained and the suggested work processes derived.
Parts of the outcome of the dissertation have been published in two conference papers (SPE), authored (as main author) by the defender of this dissertation. The application of certain recommended methodologies and relevant observations has already been used successfully in commercial Integrated Asset Modeling tools and industry studies. These are cited, but are out of the scope of this work. | nb_NO |
