Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
United States
NMR Logging Data Processing Using Deep Learning
Luo, Gang (China University of Petroleum) | Luo, Sihui (China University of Petroleum) | Xiao, Lizhi (China University of Petroleum) | Blunt, Martin J. (Imperial College London) | Bijeljic, Branko (Imperial College London) | Liao, Guangzhi (China University of Petroleum) | Shao, Rongbo (China University of Petroleum)
ABSTRACT Nuclear magnetic resonance (NMR) is a powerful tool for formation evaluation in the oil industry to determine parameters such as pore structure, fluid saturation, and permeability of porous materials, which are critical to reservoir engineering. With the rapid development of deep learning, its application to further solve problems in NMR data processing has become a new research direction such as in exponential signal reconstruction, MRI reconstruction, and peak selection for protein NMR spectra. These successful applications show that deep learning has the potential to address the shortcomings of traditional methods for processing NMR logging data. This paper concentrates on two common practical problems in NMR logging process and proposes the denoising framework for echo signals and networks for the inversion of spectra based on deep learning. NMR forward simulation dataset is implemented to satisfy the training requirements for deep learning. The designed auto-encoder network and the attention multi-scale convolutional neural network (ATT-CNN) are trained by the simulated dataset. In addition, core data measured in the laboratory are used to verify the denoising and inversion effect of deep learning models. The results demonstrate that a priori knowledge added to the simulation dataset can adequately constrain the network model. Furthermore, the ATT-CNN network structure can be adapted to low SNR echo signals and more sparse and stable spectra can be obtained. Finally, we discuss the problems that need to be overcome and the possible development potential of using deep learning to process NMR logging data. INTRODUCTION Nuclear magnetic resonance (NMR) is a powerful tool in biomedical (David, 2019; Daniel, 2019), materials science (Michael et al., 2018), chemical analysis (Arthur, 2014), oil industry (Venkataramanan et al., 2014), and other scientific fields. It provides pore structure, fluid saturation, and permeability of porous materials and is an indispensable tool for evaluating the reservoir potential of the formation. In recent years, artificial intelligence techniques have attracted great interest in various research areas, and deep learning is a representative artificial intelligence technique utilizing neural networks. Deep learning algorithms extract features from higher dimensions by using complex structures or multiple nonlinear transformations and do not require any prior knowledge or formal assumptions. The employment of deep learning methods to deal with NMR data processing problems is a new research field. Examples include exponential signal reconstruction (Huang et al., 2021), MRI reconstruction (Qin et al., 2019), and peak picking of protein NMR spectra (Piotr et al., 2018). The existing work has demonstrated the potential of deep learning methods for dealing with NMR data. In this work, we will study its application to the processing of NMR logging data.
- North America > United States (0.93)
- Asia > Middle East (0.68)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Open hole/cased hole log analysis (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Neural networks (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Artificial intelligence (1.00)
Predictive Modeling of Relative Permeability using a Generalized Equation-of-State
Purswani, Prakash (John and Willie Leone Family Department of Energy and Mineral Engineering and The EMS Energy Institute, The Pennsylvania State University, USA) | Johns, Russell T. (John and Willie Leone Family Department of Energy and Mineral Engineering and The EMS Energy Institute, The Pennsylvania State University, USA) | Karpyn, Zuleima T. (John and Willie Leone Family Department of Energy and Mineral Engineering and The EMS Energy Institute, The Pennsylvania State University, USA) | Blunt, Martin J. (Department of Earth Science and Engineering, Imperial College, London, UK)
Abstract Reliable simulation of enhanced oil recovery processes depends on an accurate description of fluid transport in the subsurface. Current empirical transport models of rock-fluid interactions are fit to limited experimental data for specific rock types, fluids, and boundary conditions. In this paper, a general equation-of-state (EoS) approach is developed for relative permeability (kr) based on a set of geometric state parameters: normalized Euler characteristic (connectivity) and saturation. Literature data and pore-network modeling (PNM) simulations are used to examine the functional form of the EoS. Our results show that the new kr-EoS matches experimental data better than the conventional Corey form, especially for highly nonlinear relative permeabilities at low saturations. Using hundreds of PNM simulations, relative permeability scanning curves show a locus of residual saturation and connectivity which defines an important limit for the physical kr region. The change of this locus is also considered for two contact angles. PNM data further allows for the estimation of the relative permeability partial derivatives which are used as inputs in the EoS. Linear functions of these partials in the connectivity-saturation space renders a quadratic response of kr, which shows excellent predictions. Unlike current empirical models that are based on only one residual saturation, the state function approach allows for dynamic residual conditions critical for capturing hysteresis in relative permeability.
- Research Report > New Finding (1.00)
- Research Report > Experimental Study (1.00)
Summary We present a new semianalytic method to solve the nonlinear pressure-diffusion equation at early time, before reservoir boundaries are encountered, and under constant bottomhole pressure (BHP), applicable to the analysis of unconventional reservoirs. We assume that the flow rate is inversely proportional to the square root of time since the beginning of production. The method is an extension of the semianalytic solution proposed by Schmid et al. (2011) for spontaneous imbibition; we replace the solution for saturation with one for pressure, while extending the functional form of the governing diffusion equation. The solution can accommodate arbitrary pressure-dependent nonlinear rock and fluid properties as well as production caused by desorption. The mathematical formulation is presented for a general nonlinear case and tested by use of synthetic data. Field production from the Barnett Shale is then used to estimate effective matrix permeability. The model can be used to predict production if the rock and fluid properties are known, or can be used to constrain reservoir properties from production data. It is a complement to traditional pressure- or rate-transient analysis; if the response of a well for constant-pressure production can be determined, our method can be used to determine reservoir properties, without any approximations inherent in linearizing the flow equations.
Low Salinity Waterflooding: From Single Well Chemical Tracer Test Interpretation to Sector Model Forecast Scenarios
Spagnuolo, Marco (eni S.p.A.) | Callegaro, Chiara (eni S.p.A.) | Masserano, Franco (eni S.p.A.) | Nobili, Marianna (eni S.p.A.) | Sabatino, Riccardo (eni S.p.A.) | Blunt, Martin J. (Imperial College)
Abstract We study Enhanced Oil Recovery (EOR) through Low Salinity (LS) waterflooding in a brown oil field. LS waterflooding is an emerging EOR technique in which water with reduced salinity is injected into a reservoir to improve oil recovery, as compared with conventional waterflooding, in which High Salinity (HS) brine or seawater are commonly used. The efficiency of this technique can be quantified at the well-scale by a Single Well Chemical Tracer Test (SWCTT), which is an in-situ method for measuring the Remaining Oil Saturation (ROS) after flooding the near-wellbore region with a displacing agent. Two SWCTTs were executed on a sandstone North African field. The tests were realized in sequence with seawater and LS water to evaluate the EOR potential at the well-scale. Here, we propose the interpretation of these two SWCTTs. They were modeled through numerical simulations because of the presence of several non-idealities in the complex scenario considered. A recently-developed tracer simulator was employed to solve the reactive transport problem. This was used as a fast post-processing tool coupled with a conventional reservoir simulator. Model parameters were estimated within an inverse modeling framework, on the basis of an assisted history matching procedure that exploits the Metropolis Hastings Algorithm (MHA). Results were scaled up on a sector model of the field, and forecast scenarios that consider a field-scale implementation of this technique were defined. The well-scale displacement efficiency gain associated with LS water, as compared with seawater, was evaluated. It was quantified as a ROS reduction of 8 saturation unit (s.u.), with a P10–P90 range of 3–15 s.u. Reservoir-scale simulations suggest that the associated ultimate oil recovery of the EOR pilot may be increased by 2% with LS water, with a P10–P90 range of 0.7–4.3%. Overall, the LS EOR potential for a selected field was quantified through a robust and original workflow, based on SWCTT interpretation. This state-of-the-art procedure is now available for further applications. The simulated oil recovery improvement with LS water is promising, and leads the way to the implementation of an inter-well field trial.
- Asia (1.00)
- North America > United States > Texas > Terry County (0.34)
- North America > United States > Texas > Gaines County (0.34)
- Europe > United Kingdom > North Sea > Southern North Sea (0.34)
- Research Report > Experimental Study (0.68)
- Research Report > New Finding (0.46)
- North America > United States > Texas > East Texas Salt Basin > East Texas Field > Woodbine Formation (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6507/8 > Heidrun Field > Åre Formation (0.99)
- Europe > Norway > Norwegian Sea > Halten Terrace > Block 6507/8 > Heidrun Field > Tilje Formation (0.99)
- (18 more...)
Abstract This paper describes a simulation study of the low-salinity effect in sandstone reservoirs. The proposed mechanistic model allows differentiation of water composition effects and includes multi-ionic exchange and double layer expansion. The manifestation of these effects can be observed in coreflood experiments. We define a set of chemical reactions, to describe the contribution of van der Waals forces, ligand exchange, and cation bridging to mobilization of residual oil. The reaction set is simplified by incorporating wettability weighting coefficients that reflect the contribution of different adsorbed ions to the wettability of the rock. Changes in wettability are accounted for by interpolation of the relative permeability and capillary pressure curves between the low and high salinity sets. We also construct and test simplified phenomenological models, one relating the change of the relative permeability to the concentration of a dissolved salinity tracer and another one to the concentration of a single adsorbed tracer. The full mechanistic model, with multiple ion tracking, is in good qualitative agreement with experimental data reported in the literature. A very close agreement with the mechanistic model was obtained for a coreflood simulation using single tracer phenomenological models. The similarity of the results is explained by the fact that the most critical factor influencing the flow behavior was the function used to interpolate between the oil- and water-wet sets of saturation curves. Similar interpolation functions in different models lead to similar oil recovery predictions. This study has developed a detailed chemical reaction model that captures both multicomponent ion exchange and double layer expansion effects, and can be used to improve understanding of low-salinity recovery mechanisms by analyzing their relative contributions. The approach of matching a tracer model to a detailed mechanistic model promises a route to the development of simplified, less computationally demanding proxy models for full field simulation studies.
- Europe (1.00)
- North America > United States > Oklahoma (0.30)
- North America > United States > Wyoming (0.28)
- (3 more...)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Sandstone (0.54)
- Geology > Geological Subdiscipline > Geochemistry (0.46)
- North America > United States > Wyoming > Powder River Basin (0.99)
- North America > United States > Montana > Powder River Basin (0.99)
- North America > United States > Alaska > North Slope Basin > Duck Island Field > Endicott Field > Kekiktuk Formation (0.99)
- (6 more...)
Summary We quantify the influence of the initial nonwetting-phase saturation and porosity on the residual nonwetting-phase saturation using data in the literature and our own experimental results on sandpacks and consolidated sandstones. These experiments were conducted at ambient or elevated pressure and temperature (ETP) conditions. The principal application of this work is for carbon capture and storage (CCS) where capillary trapping is a rapid and effective way to render the injected CO2 immobile, guaranteeing safe storage. We introduce the concept of capillary-trapping capacity (Ctrap) which is the product of residual saturation and porosity that represents the fraction of the rock volume that can be occupied by a trapped nonwetting phase. We show that the measured trapping capacity reaches a maximum of approximately 11% for porosities of 22%, which suggests an optimal porosity for CO2 storage.
- North America > United States > Texas (1.00)
- Asia > Middle East (0.68)
- Europe > Netherlands (0.68)
- North America > United States > Alaska > North Slope Borough (0.28)
- Government > Regional Government (1.00)
- Energy > Oil & Gas > Upstream (1.00)
- North America > United States > Texas > Permian Basin > Yeso Formation (0.99)
- North America > United States > Texas > Permian Basin > Yates Formation (0.99)
- North America > United States > Texas > Permian Basin > Wolfcamp Formation (0.99)
- (21 more...)
Abstract We present a coreflood experiment that measures three properties that control the long-term fate of CO2 injected into the subsurface during carbon capture and storage (CCS): the residual saturation of CO2 after brine flooding as a function of initial saturation; the amount of CO2 that can dissolve in the brine; and the primary drainage capillary pressure. We employ a porous plate method to establish initial saturation, a stirred reactor ensures that the injected brine is pre-equilibrated with CO2 and we use isothermal de-pressurization to determine CO2 saturation. The fluid pressure was 9 MPa and the temperature varied between 33 and 70°C - the CO2 is in a super-critical (sc) phase at a temperature and pressure typical of likely storage aquifers with a density between 211 and 705 kgm. We find that significant quantities of the CO2 can be trapped, with residual saturations up to 35%; the variation of trapped saturation with initial saturation is accurately matched using the Spiteri et al. model (a quadratic function). We compare the results with experiments performed at similar conditions using decane as the non-wetting phase. More decane is trapped than CO2, suggesting that the CO2-brine systems are not completely water-wet. We show that temperature (density) variation has no effect on the saturation of scCO2 that is residually trapped. The measured dissolution constant lies between 0.84 and 0.97 moles CO2/kg brine. The primary drainage capillary pressure is consistent with a strongly water-wet system and the same - to within experimental error - as that measured on an analogue decane-brine system.
- North America > United States (1.00)
- Europe > Norway (0.69)
- North America > United States > West Virginia > Appalachian Basin (0.99)
- North America > United States > Virginia > Appalachian Basin (0.99)
- North America > United States > Tennessee > Appalachian Basin (0.99)
- (42 more...)
Abstract The aim of this paper is to investigate the effects of CO2 sequestration on groundwater abstraction above the storage formation. Large-scale (10Mt/yr) CO2 injection in underlying saline aquifers is considered using a fully compositional simulator to study the pressure distribution, CO2 leakage and inter-layer brine flow. Structural, residual and solubility CO2 trapping are taken into account while the model domain is considered to have no-flow boundaries to simulate CO2 injection under the context of either pressure not being able to be dissipate quickly (due to other CO2 injection processes, for instance) or the formation being self-contained. Changes in salinity are affected by groundwater abstraction and although not caused by injection itself, can be magnified by such processes. We also conclude that the time at which CO2 leakage, pressure perturbations and upward brine flow are at their peak in overlying aquifers (layers above the injection site) may be significantly after injection has ceased (50-150 years in some cases) and could potentially cause groundwater movements, land surface uplift or rock fracturing long after the injection phase has ended.
- North America > Canada (0.68)
- North America > United States > Texas (0.28)
- Europe > United Kingdom > England (0.28)
- Geology > Geological Subdiscipline (1.00)
- Geology > Rock Type > Sedimentary Rock (0.68)
- Energy > Oil & Gas > Upstream (1.00)
- Government > Regional Government > North America Government > United States Government (0.68)
- Reservoir Description and Dynamics > Storage Reservoir Engineering > CO2 capture and sequestration (1.00)
- Reservoir Description and Dynamics > Improved and Enhanced Recovery > Chemical flooding methods (1.00)
- Health, Safety, Environment & Sustainability > Sustainability/Social Responsibility > Sustainable development (1.00)
- (2 more...)
Abstract Displacement experiments using the porous plate method were conducted on water-wet sandstones to measure the capillary trapping of oil by waterflooding as a function of its saturation after primary drainage. Three sandstone samples ranging in porosity from 12.2% to 22.1% were considered. Experiments on two samples were conducted at an elevated temperature and back-pressure of 343K and 9MPa respectively; experiments on the third sample were conducted at ambient conditions (292 to 297K and 0.06 to 0.17MPa). Residual oil saturation increases monotonically, but with a decreasing gradient, as initial saturation increases. The dependence of residual saturation on initial saturation is accurately predicted by a two-phase pore-network simulator when a uniform distribution of intrinsic contact angles between 35° and 65° is assumed. The networks were extracted from X-ray microtomography images of small samples of the same rock as those used in the experiments. The laboratory measurements are also accurately described by trapping models proposed by Land (1968) and Spiteri et al. (2008). The residual saturations we measured were higher than in previous displacement experiments, suggesting, for example, that capillary trapping may be an effective way to store substantial quantities of carbon dioxide in aquifers.
- Europe (0.93)
- North America > United States > Texas (0.68)
- Asia > Middle East (0.68)
- North America > United States > Texas > Permian Basin > Yeso Formation (0.99)
- North America > United States > Texas > Permian Basin > Yates Formation (0.99)
- North America > United States > Texas > Permian Basin > Wolfcamp Formation (0.99)
- (21 more...)
- Reservoir Description and Dynamics > Reservoir Simulation (1.00)
- Reservoir Description and Dynamics > Reservoir Fluid Dynamics > Flow in porous media (1.00)
- Reservoir Description and Dynamics > Improved and Enhanced Recovery (1.00)
- Reservoir Description and Dynamics > Formation Evaluation & Management (1.00)
Measurement of Nonwetting-Phase Trapping in Sandpacks
Pentland, Christopher H. (Imperial College London) | Itsekiri, Endurance (Imperial College London) | Al Mansoori, Saleh K. (Imperial College London) | Iglauer, Stefan (Imperial College London) | Bijeljic, Branko (Imperial College London) | Blunt, Martin J. (Imperial College London)
Summary We measure the trapped nonwetting-phase saturation as a function of initial saturation in sandpacks. The application of the work is for carbon dioxide (CO2) storage in aquifers, where capillary trapping is a rapid and effective mechanism to render the injected fluid immobile: The CO2 is injected into the formation followed by chase-brine injection or natural groundwater flow that displaces and traps it. Current models to predict the amount of trapping are based on experiments in consolidated media; while CO2 is likely to be injected at depths greater than approximately 800 m to render it supercritical, it may be injected into formations that tend to have a higher porosity and permeability than deep oilfield rocks. We use analog fluids—water and refined oil—at ambient conditions. The initial conditions are established by injecting oil into vertical or horizontal sandpacks 0.6 m long at different flow rates and then allowing the oil to migrate under gravity. The packs are then flooded with water. The columns are sliced, and the residual saturation is measured with great accuracy and sensitivity by gas chromatography (GC). This method allows low saturations to be measured reliably. The trapped saturation initially rises linearly with initial saturation to a value of approximately 0.13, followed by a constant residual as the initial saturation increases further. This behavior is not predicted by the traditional Land (1968) model but is physically consistent with poorly consolidated media where most of the larger pores can be invaded easily at relatively low saturation and there is, overall, relatively little trapping. The best match to our experimental data is achieved with the Aissaoui (1983) and the Spiteri et al. (2008) trapping models.