Tandon, Saurabh (The University of Texas at Austin)
"Sweet spots" in the unconventional reservoirs such as organic-rich mudrocks are zones with high productivity. However, identifying such regions in unconventional reservoirs depends non-only on their petrophysical and but also on their geomechanical properties. Supervised learning methods can help in integrating numerical simulation and legacy field data in sweet-spot identification workflows and enhance their analysis in complex reservoirs. The objectives of this paper are to: (i) demonstrate the use of supervised learning in parameter selection and evaluation for fracture design and (ii) provide non-linear models for sweet-spot analysis in complex reservoirs.
We used fracture simulator that combines with fracture deformation with fluid-flow in discrete fracture networks. We started by selecting different geomechanical rock properties related to its fracability. We then used quasi-random design approach to obtain wide variation in aforementioned properties and performed 200 fracture simulations using the hydraulic fracturing simulator. We used the short-term Stimulated Reservoir Volumes (SRV) obtained at the end of numerical simulations, to quantify the performance of hydraulic fracturing operations. We used supervised learning techniques like support vector machines, decision trees, and random forests to perform parameter ranking and create non-linear regression models that can correlate the SRV to formation geomechanical properties.
The inputs for the analysis are: initial aperture, toughness, dilation angle, closure stress, and friction coefficient of initial fractures, stress anisotropy, shear modulus and a ratio of the reservoir rock. We analyzed the results using β-linear and multinomial regression, support vector machines, decision trees, and random forests. The linear models and non-linear models can explain up to 89.1% of output variance. The classification accuracy of support vector machines was at most 35% higher than other algorithms like random forests. Parameter rating using non-linear models showed that stress anisotropy and dilation angle demonstrated the highest effect on SRVs. Shear modulus and fracture toughness show minimal effect on the SRV but these parameters might still be useful they could be correlated to other formation parameters.
The outcomes of this paper demonstrated that parameters pertaining to unpropped fracture conductivity play a significant role in determining the success of hydraulic fracturing treatments. We have also compared the performances of supervised machine learning algorithms in assessing the impact of rock properties on fracturing treatments. Such supervised machine learning algorithms can help integrate field legacy data and numerical simulation outputs to develop proxy models that improve sweet-spot analysis and production estimates in unconventional reservoirs.
Borehole measurements are often subject to uncertainty resulting from the effects of mud-filtrate invasion. Accurate interpretation of these measurements relies on properly understanding and incorporating mud-filtrate invasion effects in the calculation of petrophysical properties. Although attempts to experimentally investigate mud-filtrate invasion and mudcake deposition have been numerous, the majority of published laboratory data are from experiments performed using linear rather than radial geometry, homogeneous rock properties, and water-based (WBM) rather than oil- or synthetic oil-based drilling mud (OBM or SOBM).
We introduce a new experimental method to accurately reproduce conditions in the borehole and near-wellbore region during, and shortly after the drilling process, when the majority of wellbore measurements are acquired. Rather than using a linear-flow apparatus, the experiments are performed using cylindrical rock cores with a hole drilled axially through the center. Radial mud-filtrate invasion takes place while injecting pressurized drilling mud into the hole at the center of the core while the outside of the core is maintained at a lower pressure. During the experiments, the core sample is rapidly and repeatedly scanned using high-resolution X-ray microcomputed tomography (micro-CT), allowing for visualization and quantification of the time-space distribution of mud filtrate and mudcake thickness. Because of the size of the core sample, the developed experimental method allows for accurate evaluation of the influence of various rock properties, such as the presence of spatial heterogeneity and fluid properties, including WBM versus OBM, on the processes of mud-filtrate invasion and mudcake deposition. Results indicate that our experimental procedure reliably captures the interplay between the spatial distributions of fluid properties and rock heterogeneities during the process of mud-filtrate invasion.
The goal of this work is to evaluate the applicability of a novel set of surfactants to enhance recovery from a viscous oil, high temperature, high permeability, clastic reservoir. A large number of novel short-hydrophobe based surfactants/cosolvents were designed and synthesized. As these surfactants do not require expensive aliphatic alcohols for their synthesis, they are likely to be less costly than conventional anionic surfactants. Here only phenol hydrophobe based non-ionic surfactants with varying number of propylene oxide (PO) and ethylene oxide (EO) groups are discussed. These surfactant molecules were investigated for their aqueous stability limits, interfacial tensions (IFT) with a viscous crude oil and oil recovery from sandpack or sandstone cores. Surfactant phase behavior experiments with viscous crude oil showed low IFT (not ultralow) for single surfactant systems. Only one surfactant (Phenol-7PO-15EO) formulation was chosen for coreflood in sandpack and sandstone cores. Water flood recovered about 50% original oil in place (OOIP) and reduced the oil saturation to about 48% in the high permeability sandpacks. The tertiary surfactant polymer flood with Phenol-7PO-15EO increased the cumulative recovery to 99% for sandpacks. The oil recovery was insensitive to injection brine salinity in the range studied. As the permeability decreased, the tertiary oil recovery decreased if the permeability is lower than 7 Darcy. Surfactant-polymer (SP) formulations with this surfactant can be recommended for high permeability sandstone reservoirs with viscous oils, but not for sub-Darcy sandstones.
Ahmadian, Mohsen (Advanced Energy Consortium, Bureau of Economic Geology, The University of Texas at Austin) | LaBrecque, Douglas (Multi-Phase Technologies, LLC) | Liu, Qing Huo (Duke University) | Kleinhammes, Alfred (The University of North Carolina) | Doyle, Patrick (The University of North Carolina) | Fang, Yuan (Duke University) | Jeffrey G, Paine (The University of Texas at Austin) | Lucie, Costard (The University of Texas at Austin)
Characterizing hydraulically induced fractures—height, length, orientation, and shape—is key to understanding reservoir performance. Our previous work has focused on the comparison of the state-of-the-art geophysical techniques currently used in hydraulic fracture imaging (microseismicity, tracer, tiltmeter, and distributed acoustic and temperature sensors) to perform a comprehensive set of electromagnetically active proppant (EAP)–assisted tomography methods (
Zhao, Bochao (Shell International Exploration and Production) | Ratnakar, Ram (Shell International Exploration and Production) | Dindoruk, Birol (Shell International Exploration and Production) | Mohanty, Kishore (The University of Texas at Austin)
Accurate estimation of relative permeability is vital for decision making in upstream applications from project appraisal to field development and evaluation of various field development options. As relative permeability is a function of both rock and fluid properties, it is harder to generalize it over a wide combination of rocks and fluids. In addition to this complexity, it is hard to gather coherent sets of data to develop a correlation covering the domain of interest for most projects. As a result, fast and reliable relative permeability prediction method is missing in literature. In this study, we identify Euler number (
In order to achieve our objective, first, we developed a machine learning model based on random forest algorithm (
The investigation based on machine learning of pore network simulation results in combination with the available data suggests that phase saturation and Euler numbers are the two dominant parameters affecting the relative permeability. In particular, it shows that Variation in relative permeability with different rock-fluid parameters (that along with intial fluid distribution can cause the variation in Euler number) is significant even when saturation is fixed. In other words, the relative permeability is multivalued function of saturation, as hysteresis models also indicate. This suggests neither of saturation or Euler number alone is sufficient for relative permeability prediction. At a fixed saturation (zero-dimensional volumetric abundance) and Euler number coordinates, the relative permeability is very consistent and vary insignificantly across different cases, suggesting thesetwo parameters as first-order predictors. Euler number characterizes the fluid connectivity/distribution, while saturation represents the net volumetric fluid quantity. We believe that Euler number has been the missing first-order predictor in traditional saturation-based predictive relative permeability models. Most importantly, we identify and present the quantitative relationship between relative permeability and Euler characteristic, and present a reliable correlation to determine the relative permeability based on Euler number and saturation.
Variation in relative permeability with different rock-fluid parameters (that along with intial fluid distribution can cause the variation in Euler number) is significant even when saturation is fixed. In other words, the relative permeability is multivalued function of saturation, as hysteresis models also indicate. This suggests neither of saturation or Euler number alone is sufficient for relative permeability prediction.
At a fixed saturation (zero-dimensional volumetric abundance) and Euler number coordinates, the relative permeability is very consistent and vary insignificantly across different cases, suggesting thesetwo parameters as first-order predictors. Euler number characterizes the fluid connectivity/distribution, while saturation represents the net volumetric fluid quantity. We believe that Euler number has been the missing first-order predictor in traditional saturation-based predictive relative permeability models.
Most importantly, we identify and present the quantitative relationship between relative permeability and Euler characteristic, and present a reliable correlation to determine the relative permeability based on Euler number and saturation.
To the best of our knowledge, this is the first successful attempt at directly investigating the quantitative relationship between Euler number and relative permeability based on machine learning of experimental SCAL data in combination with pore network simulation results. This work provides the necessary framework and lends itself for further research and development using additional data as time goes on along with more advanced numerical simulation and data analysis models.
Development of reliable models for hydrocarbon-in-place and water saturation estimation requires knowledge about wettability of mudrocks and the parameters (including rock properties and reservoir condition) affecting it. A significant volume fraction of organic-rich mudrocks is composed of kerogen. Therefore, wettability of kerogen affects the overall wettability of organic-rich mudrocks. The chemical composition and structure of kerogen varies with kerogen type and thermal maturity, which affects the surface properties of kerogen such as wettability. In a recent publication, we demonstrated using experimental techniques that kerogen could be water-wet at low thermal maturities and oil-wet at higher thermal maturities. However, the impacts of kerogen type and reservoir temperature/pressure conditions on kerogen and mudrock wettability is yet to be quantified. Therefore, the objectives of this paper include (i) quantifying the impacts of kerogen molecular structure and composition on water adsorption capacities, (ii) quantifying the impacts of reservoir pressure and temperature on water adsorption capacity of kerogen using molecular dynamics (MD) simulations.
In order to achieve the aforementioned objectives, we use a combination of molecular dynamics simulations and experimental work. The inputs to the molecular dynamics simulations include realistic models of kerogen, which are condensed to porous kerogen structures. Water molecules are filled in kerogen pore structure and MD simulation is performed. The outputs of the simulations include radial distribution function (RDF), and adsorption isotherms of water on kerogen for different kerogen types, thermal maturities, and temperature conditions. The adsorption processes are modelled for pressure and temperature conditions ranging from 0 to 35 MPa and 320 to 370 K, respectively. The outcomes of molecular dynamics simulations demonstrated that the water adsorption capacities of kerogen vary significantly with kerogen type, thermal maturity, and temperature and pressure conditions. The RDF results showed that the water adsorption capacity decreased from type I to type III kerogen. The water adsorption capacity of kerogen was found to increase by 128% with 38% increase in oxygen content. The increase in the adsorption capacity was attributed to the strong attraction between oxygen containing functional groups in kerogen and water. The adsorption isotherms of water and kerogen samples showed that the water adsorption capacity decreased by 0.19 mmol/g as the temperature increased from 320 K to 370 K. The average water adsorption capacity of kerogen was found to increase by 20% with increase in pressure by 34 MPa. The results obtained from molecular dynamics simulations were found to be in good agreement with experimental results. The results of this paper can be used to predict the adsorption capacities of any kerogen with the availability of geochemical information. This important property of kerogen is required for estimating kerogen wettability and can enhance understanding of fluid-flow mechanisms in organic-rich mudrocks.
The low-frequency dielectric response of sedimentary rocks is dominated by rock fabric, volumetric concentrations of fluids and minerals, and interfacial properties. The rock physics models for interpretation of multi-frequency complex permittivity measurements generally rely on simplified geometries for which analytical solutions are obtainable. Consequently, interpretation of permittivity measurements can be challenging in reservoirs with complex pore structure, mineralogy, and mixed-wet conditions. The objectives of this paper include the development of a rigorous numerical simulation framework to enhance the interpretation of multi-frequency, complex dielectric permittivity measurements and also to quantify the influence of polarization of the electric double layer, lithology, fluid properties, and pore-network geometry on dielectric permittivity measurements. We develop a simulator to calculate permittivity dispersion of sedimentary rocks by applying a combination of finite-difference and finite-volume methods to solve the nonlinear Poisson and Nernst-Planck equations in the time domain. We perform a sensitivity analysis of dielectric permittivity to the dominant mineral (e.g., quartz, calcite), pore geometry, and fluid properties (e.g., salt concentration). The main contribution of this paper consists of introducing a simulator that provides the complete and accurate description of electric field, ionic distribution, and effective dielectric permittivity in porous media for enhanced petrophysical interpretation of electromagnetic measurements. Results suggest that incorporating the introduced simulation into a workflow for broadband interpretation of dielectric measurements can improve petrophysical evaluation in formations with complex lithology, rock fabric, and in mixed-wet rocks. This unique approach provides a more rigorous characterization of the dielectric permittivity of rocks than previously documented analytical and numerical models.
Hydrocarbon production from Shale formations has become an increasingly significant part of the global energy supply since 2010. With the advent of horizontal drilling and multiple-stage hydraulic fracturing, the Utica Shale, which underlies the Marcellus Shale as a natural source rock, is one of the most promising and productive shale plays in the US. However, very few academic papers discuss its geo-stress, pore pressure, permeability, and corresponding DFIT applications, which are essential for the development of the Utica Shale. The objective of this study is to use Diagnostic Fracture Injection Tests (DFITs) data from the field to analyze minimum in-situ stress, closure pressure, reservoir pore pressure, key reservoir properties and fracture geometry in the Utica Shale by different DFIT interpolation methods. The analysis results are compared and discussed in detail to investigate the features of each DFIT interpolation method. In addition, DFIT numerical simulation based on Variable Compliance Model is performed to predict induced fracture geometry and effective formation permeability in the Utica Shale.
DFIT is a commonly applied technique to analyze stress regimes and reservoir properties, while its interpolation can be challenging and difficult for different formations. DFIT interpretation for Shale formations is even more complex. In this study, first overviewing the geology of the Utica Shale and continuing to the summary of DFIT analysis and its governing equations, one can gain a better understanding of the methods and processes used to analyze our DFIT data targeting the Utica Shale. Tangent Line method, Compliance method, and Variable Compliance method are reviewed, and the corresponding assumptions for each method are examined, compared and discussed. Our DFIT data, which is acquired from a horizontal well targeting the Utica Shale, is interpreted by all methods to analyze minimum in-situ stress, closure pressure, initial reservoir pore pressure, key reservoir properties and fracture geometry. The DFIT results are then discussed and compared in detail to investigate the features of each method with its diagnostic signatures. Following that, the induced fracture geometry and the effective formation permeability are predicted by numerical simulation and sensitivity analysis, which also evaluate the impacts of wellbore storage, formation properties and fluid properties on simulated pressure and pressure derivative profiles.
The results from DFIT analysis are very encouraging. The Tangent Line method oversimplified leak off dependence and fracture stiffness, while the obtained minimum in-situ stress, closure pressure, pore pressure, fracture geometry and effective permeability are consistent with the diagnostic plots and our petrophysics studies. The Compliance method is able to identify mechanical closure, but it overestimates the minimum principal stress. The Variable Compliance method can capture the variance in fracture stiffness and pressure dependent leak off during progressive fracture closure, and its estimated closure pressure is an average of the results from the Tangent Line and the Compliance methods. The formation permeability of the Utica Shale is estimated by performing a history match of the pressure and pressure derivative profiles. The physics behind the DFIT simulation and sensitivity analysis is analyzed and discussed in detail. Our study can significantly improve the understanding of pressure/stress regimes, fracture geometry, and reservoir properties in the Utica Shale, as well as features of different DFIT interpolation methods. The knowledge and results demonstrated in this article will indefinitely assist operators in their optimization of multistage fracturing and horizontal drilling design in order to develop the Utica Shale more cost-effectively.
Lara Orozco, Ricardo A. (The University of Texas at Austin) | Abeykoon, Gayan A. (The University of Texas at Austin) | Wang, Mingyuan (The University of Texas at Austin) | Argüelles Vivas, Francisco J. (The University of Texas at Austin) | Okuno, Ryosuke (The University of Texas at Austin) | Lake, Larry W. (The University of Texas at Austin) | Ayirala, Subhash C. (Saudi Aramco) | AlSofi, Abdulkareem M. (Saudi Aramco)
Reservoir wettability plays an important role in waterflooding especially in fractured carbonate reservoirs since oil recovery from the rock matrix is inefficient because of their mixed wettability. This paper presents the first investigation of amino acids as wettability modifiers that increase waterflooding oil recovery in carbonate reservoirs.
All experiments used a heavy-oil sample taken from a carbonate reservoir. Two amino acids were tested, glycine and β-alanine. Contact angle experiments with oil-aged calcite were performed at room temperature with deionized water, and then at 368 K with three saline solutions: 243,571-mg/L salinity formation brine (FB), 68,975-mg/L salinity injection brine 1 (IB1), and 6,898-mg/L salinity injection brine 2 (IB2). IB2 was made by dilution of IB1.
The contact angle experiment with 5-wt% glycine solution in FB (FB-Gly5) resulted in an average contact angle of 50°, in comparison to 130° with FB, at 368 K. Some of the oil droplets were completely detached from the calcite surface within a few days. In contrast, the β-alanine solutions were not effective in wettability alteration of oil-aged calcite with the brines tested at 368 K.
Glycine was further studied in spontaneous and forced imbibition experiments with oil-aged Indiana limestone cores at 368 K using IB2 and three solutions of 5 wt% glycine in FB, IB1, and IB2 (FB-Gly5, IB1-Gly5, and IB2-Gly5). The oil recovery factors from the imbibition experiments gave the Amott index to water as follows: 0.65 for FB-Gly5, 0.59 for IB1-Gly5, 0.61 for IB2-Gly5, and 0.33 for IB2. This indicates a clear, positive impact of glycine on wettability alteration of the Indiana limestone cores tested.
Two possible mechanisms were explained for glycine to enhance the spontaneous imbibition in oil-wet carbonate rocks. One mechanism is that the glycine solution weakens the interaction between polar oil components and positively-charged rock surfaces when the solution pH is between glycine's isoelectric point (pI) and the surface's point of zero charge (pzc). The other mechanism is that the addition of glycine tends to decrease the solution pH slightly, which in turn changes the carbonate wettability in brines to a less oil-wet state.
The amino acids tested in this research are non-toxic and commercially available at relatively low cost. The results suggest a new method of enhancing waterflooding, for which the novel mechanism of wettability alteration involves the interplay between amino acid pI, solution's pH, and rock's pzc.
Ghosh, Pinaki (The University of Texas at Austin) | Zepeda, Angel (The University of Texas at Austin) | Bernal, Gildardo (The University of Texas at Austin) | Mohanty, Kishore (The University of Texas at Austin)
Waterflood in low permeability carbonate reservoirs (<50 mD) leaves behind a substantial amount of oil due to capillary trapping and poor sweep. Addition of polymer to the injected water increases the viscosity of the aqueous phase and decreases the mobility ratio, thus, improving the sweep efficiency and oil production from the tight formations. Performance of current synthetic EOR polymers is limited by salinity, temperature and injectivity issues in low permeability formations. Mechanical shear degradation can be applied to high molecular weight synthetic polymers to improve the injectivitiy; but makes the process less economical due to significant viscosity loss and consequent increase in polymer dosage. Recently, a different class of polymer has been developed called "hydrophobically modified associative polymers (AP)". The primary goal of this work is to investigate the performance of associative polymers in low permeability carbonate reservoirs. We compare the performance of associative polymers with that of conventional HPAM polymers in low permeability formations. A low molecular weight associative polymer was investigated as part of this study. A detailed study of polymer rheology and the effect of salinity at the reservoir temperature (60 °C) was performed. Additional experiments were performed in bulk and porous media to investigate the synergy of associative polymers with hydrophilic surfactant blends at different brine salinities. Single phase polymer flow experiments were performed in outcrop Edwards Yellow and Indiana limestone cores of low permeability to determine the optimum polymer concentration to achieve the desired in-situ resistance factor (or apparent viscosity). Similar experiments were performed with HPAM polymer for a comparative study. Results showed successful transport of this associative polymer in low permeability formations after a small degree of shear degradation. The resistance factors for the associative polymer were higher than those for HPAM. Shear degraded polymers showed significant improvement in polymer transport in lower permeability cores with reduction in RRF.