Matching water cuts of production wells is one of the most challenging tasks in the history matching process. When conducting history matching with a finite difference simulator, one of the challenges is how to define a suitable region for the permeability modifier to obtain a good match, and yet, retain major geological description. In the case of large models, the presence of several wells in a region where permability is adjusted makes simultaneous matching for these wells almost impossible. Commercial streamline simulators are best in reducing the mismatch in water cuts; however, it is cumbersome to include other uncertainty parameters such as water-oil contact, relative permeability, pressure-volume-temperature data, etc., during the optimization process. To combine the power of finite difference simulators and streamline technology, an in-house developed reservoir simulator was linked to a customized streamline tracing software package. This solution is capable of generating streamlines that represent a snapshot of the flow pattern within the reservoir, and providing well drainage region information and injector-producer relationships. This paper presents an automatic process in which the reservoir simulator is run via the control of a commercial optimizer, the simulation results are transferred to the streamline tracing package, which also tags the simulator cells to be modified by the optimizer without running a full streamline simulation. The optimizer then uses a global minimization method to fine-tune the permeability modifiers for reducing the water cut errors. This approach was verified by using two synthetic models in which the solution is known. Excellent results were obtained. Therefore, this approach makes it possible to employ the power and flexibility of finite difference simulators and optimization algorithms, with the crucial fluid flow information provided by streamline technology.
In phase behaviour calculations and compositional simulations with an equation of state, it is advantageous to represent the oil and gas systems by a small number of pseudo-components. This paper presents a scheme for determining these pseudocomponents based on the K-values of all constituents that are present in the reservoir oil and gas systems. The scheme is simple and applicable to both the light and heavy components. Sensitivity studies are carried out to investigate the effect of different pseudo-component representations on the shape of various phase diagrams, and on the compositional simulation results. Introduction Phase behaviour calculations and compositional simulations with an equation of state require the use of pseudocomponents to represent the oil and gas mixtures. Because of the large number of components which form the heavy fractions (e.g. C6 + fractions), it is necessary to group them into pseudo-components. Furthermore, to minimize simulation costs, it is also advantageous to lump the light fractions. This paper presents a procedure for characterizing the heavy fractions of oil and gas and a systematic scheme for lumping the components into pseudo-components based on the K-values at a specified operating condition which is typical of the process under study. Phase diagrams are then generated to investigate the effects of different lumping schemes on the shape of the phase boundaries and quality lines. The construction of these phase diagrams provides a rapid method for examining the sensitivities of the computed phase behaviour to different lumping schemes over a wide range of pressure, temperature and composition. Compositional simulation with different lumping schemes is also carried out to investigate the effect of different pseudo-component representations on the simulation results. Extended Analysis of Heavy Fractions Extended analyses of the heavy fractions (e.g. C6 + fraction) are required for accurate predictions of oil and gas phase behaviour from an equation of state (EOS)(1). True-boiling-point (TBP) analyses yield directly the boiling point, specific gravity and the molecular weight of each carbon-number (CN) group(2) from which molar distribution is found directly. Analyses from gas-chromatograph (GC) measurements, on the other hand, provides only the weight fraction of each CN group. The molar distribution is then obtained by using either the molecular weight of the normal alkane corresponding to each CN group, or the data of Katz and Firoozabadi(3, 4). The results obtained from ac measurements are less accurate than those obtained from TBP experiments. If only a partial extended analysis is available, it can be extended to higher CN group by using the method of Whitson(1), Pedersen et al.(5) or the one described in Appendix A. The critical properties and eccentricities of the CN groups are estimated from their specific gravities and boiling points by using Kesler-Lee correlations(6). If the specific gravity and boiling point of any CN group are not available, they can be estimated from the data of Katz and Firoozabadi(3, 4) or from the correlations reported by Whitson(7). Lumping into Pseudo-Components Because of the large number of components in the heavy fraction, it is necessary to lump them into pseudo-components before an EOS can be used efficiently.
The Underground Test Facility is designed to test the application of the steam-assisted gravity drainage process for the in situ recovery of bitumen, using horizontal wells drilled from tunnels below the pay zone.
SPE Members Abstract A horizontal well usually yields a high rate through its long perforation interval, resulting in a large frictional pressure drop that is believed to reduce the well productivity. Unlike conventional horizontal well methods which consider the horizontal section only, the proposed model also includes the vertical section of the well and the surface facility from the well head to the GOSP. The hydraulics model was validated stage-by-stage by various known solutions, and then applied to a high rate well. It was found that the effect of frictional pressure drop on oil rate was not as pronounced as suggested by analytical methods, or some numerical simulation studies. In summary, the pressure drop in the vertical section of the horizontal well plays an especially important role in deciding how much fluid can be produced. Therefore, it is recommended to be an integral part of horizontal well models. Introduction In models which consider the horizontal section only, frictional pressure drop was shown to reduce the productivity of horizontal wells. Dikken presented an analytical horizontal well model that combined the fluid flow in the horizontal section of a well and the reservoir flow. He concluded that the flow inside horizontal wells is either transition or turbulent in most practical situations. Furthermore, for single phase turbulent flows, he found that appreciable reduction in drawdown occurred at positions farther away from the start of the section. In a water coning example, he demonstrated that little additional production results from extending a 300 m long well with a diameter of 11.4 cm. Novy generalized Dikken's model so that it can be applied to the recovery of gas. In the process, he found that the friction factor correlation used by Dikken gives friction factors that are too high for rough tubes. Novy performed extensive sensitivity studies and he concluded that if the ratio of well-bore pressure drop to drawdown at the producing end exceeds 10%, friction is apt to reduce productivity by 10% or more. In a simulation study where the frictional pressure drop was included in the horizontal well model, Seines et al. found that by varying the effective roughness of the well, thereby changing the pressure drop, they obtained a 10% difference in cumulative production after 1 year. The well was producing at a fixed flowing bottomhole pressure (FBHP). This paper presents the work in which a complete hydraulics model for horizontal wells was developed and validated. The model was then applied to a high flowrate case. Calculated results were compared with analytical predictions. Investigation of the discrepancy in results provided some insights about the modeling of horizontal wells. P. 407
Production from reservoirs with large permeability contrasts or reservoirs consisting of several non-communicating layers can result in significant crossflow between the layers through multilayer wells. This entails the backflow of reservoir fluids from the wellbore to the reservoir. Standard line source/sink methods for handling wells in reservoir simulators are unable to cope with backflow because the wellbore is assumed to have the same saturations as the reservoir. To simulate backflow accurately, the well model must keep track of the fluid saturations within the wellbore. In this work, the discretized wellbore model developed by Collins et al. (1992) is extended and applied to vertical wells exhibiting backflow. The method described calculates fluid saturations along the wellbore implicitly. Comparisons of results between the standard line source/sink well model and the discretized wellbore model are made for several field-scale cases. It was found that the standard line source/sink model can result in unphysical values compared to the discretized wellbore model, which always yielded realistic production rates for the cases considered. Introduction Production from highly stratified reservoirs or from reservoirs with barriers between different production zones can result in significant backflow in producing wells. This corresponds to the flow of fluids from the wellbore back into the reservoir. This process happens quite frequently when the well is completed over multiple reservoir layers with poor vertical communication. The poor communication causes the potential gradient in the reservoir to be very different from the gradient in the wellbore. As the well acts as a channel for fluid flow between reservoir regions with different potentials, backflow occurs when the difference between the reservoir pressure and the wellbore pressure (pressure drawdown) becomes negative. Reservoir simulators where the well is represented as a line source/sink are not suitable for handling backflow. They give erroneous results, and when backflow is severe, they yield an unphysical solution. Holmes (1983) attempted to improve the source/sink representation of the well for backflow situations by including a global wellbore mass balance. In this approach, the fluid properties and saturations were assumed uniform throughout the wellbore. Modine and Coats (1990) introduced a superposition method in which fluid compositions resulting from an explicitly calculated head within the wellbore were used to determine individual layer inflow (layers producing into the wellbore) and outflow (layers receiving fluids from the wellbore). An efficient technique for modelling wellbore dynamics in reservoir simulation was developed by Collins et al. (1992). The wellbore flow equations were cast judiciously in a form similar to the reservoir flow equations. Thus, efficient techniques that were developed to solve the reservoir flow equations can readily be applied. Since Collins et al. considered horizontal production wells only, their method required extension to the case of vertical producers and injectors. This method will be referred to as the "discretized wellbore" technique. In this paper, the discretized wellbore technique is extended to vertical production/injection wells exhibiting backflow. Several field-scale cases are used to test the technique.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.