Explore the vast range of titles available.

In this work, we have examined the predictive capability of a machine leaning model based on the XGBoost framework as regards the solubility of active pharmaceutical ingredient-like molecules in organic solvents over a wide range of temperatures. A total of 30 binary mixtures has been investigated. The dataset was divided in two sets, with one set for training, testing and validation including solubility data for four solute compounds (butyl paraben, fenofibrate, risperidone, fenoxycarb) consisting of a total of 224 data points, and the second set used for prediction consisting of the solubility data for butamben, with 50 data points in total. The calculated root mean square errors (RMSLE) for the calculated solubility (train, test, validation) were 0.05, 0.09, 0.13 and 0.15, respectively, while the average RMSLE for the predicted solubility of butamben was 0.41. A total of 10 descriptors were considered in this work, comprising parameters for solute (heat of fusion, melting temperature, heat capacity and Hansen solubility parameter), two descriptors representing the solvent (dielectric constant and boiling temperature) as well as temperature, and drug and solvent names. The temperature-dependence of solubility has been captured accurately by setting a constraint on the XGBoost algorithm. A comparison between the performance of the machine learning model proposed and evaluated in this work, and the semi-predictive Flory-Huggins and the temperature-dependent NRTL-SAC models on the other hand, shows that for all the studied compounds, the machine learning model can deliver significantly improved capability to model as well as predict solubility.

This work demonstrates an effective procedure to correlate and predict viscosity of petroleum fluids using friction theory (FT) viscosity model coupled with perturbed-chain statistical associating fluid theory (PC-SAFT). The FT is used for viscosity prediction in reservoir fluids. The PC-SAFT is applied for calculation of equilibrium composition and density of vapor and liquid phases. The FT has a few characteristic parameters for each component for viscosity prediction. These parameters are not available for petroleum fractions. In this study, such a problem is addressed by finding a model to predict FT characteristic parameters for different petroleum fractions as a function of molecular weight and critical pressure. 20 real reservoir fluid samples are used to develop the model. Afterward, for 5 real reservoir oil samples in the evaluation step, the viscosity modeling results are compared against experimental data, and the methods of Tan et. al., Lohrenz et al., and Pedersen et al. for showing the accuracy of proposed model. It is concluded that with suitable characteristic parameters for the FT viscosity model and PC-SAFT, improvement in liquid viscosity estimation can be achieved. The average absolute deviation percent (AAD%) is 10.22% for FT + PC-SAFT (this work), 13.71% for Lohrenz et al. and 23.48 for Pedersen et al. In addition, since the free-volume (FV) theory like FT belongs to semi-empirical viscosity models, a comparison with the FV model (published in the work of Khoshnamvand and Assareh in In J Thermophys 39:1, 2018) is performed. The results demonstrate that the FT viscosity model with presented characteristic parameters in this study gives a comparable accuracy in viscosity prediction for the studied real reservoir fluids. Compared to FV, the FT is less dependent on the EOS calculation which is an advantage of FT.

Acid gases, containing CO2 and H2S, are by‐products of gas sweetening. Geological sequestration of these gases in naturally fractured reservoirs (NFRs) is a practical method to reduce greenhouse gas emission. An industrially accepted approach to simulate fluid flow in NFRs is the dual‐porosity method; however, this method needs multiple parameters' specifications. The main goal of this study is to develop a dual‐porosity model with improved parameters that can be used for simulation of both hydrocarbon gas gravity drainage and acid gas injection in the gas‐invaded zone of NFRs. To do so, a single‐porosity model, as the reference model, is constructed for a single matrix block (SMB) with which the equivalent dual‐porosity model's (DP) parameters are determined and matched. Then, DP is improved by a dual‐porosity vertical discrete (VD) model to consider gravity drainage. This was later enhanced by non‐neighborhood connections (NNCs) to account for re‐infiltration in stacked matrices, yielding comparable results to the reference CPU‐intensive single‐porosity simulation. A thorough sensitivity analysis is performed on acid gas injection in VD model. The results show that the most effective parameter is porosity. The permeability and NNC transmissibility only change the rate of acid gas storage and more acid gas is trapped as H2S content increases. Also, the heterogeneous distribution of porosity only influences the rate of storage when the mean porosity is constant, while permeability heterogeneity does not affect acid gas storage. The recovery factor is considerably increased to nearly 100% when the acid gas replaces hydrocarbon gas in fractured surrounding. About 7000 kmole of acid gas is stored in SMB over 4.5 years. Similar results are obtained for stacked matrices, and trapped gas is about 22,000 kmole, after 9 years.

For some of the EOS models the dimension of equilibrium problem can be reduced. Stability and difficulties in implementation are among the problems of flash calculation. In this work, a new reduction technique is presented to prepare a reduced number of equilibrium equations. Afterwards, a number of appropriate solution variables are selected for the prepared equation system to solve the equations in an efficient numerical scheme. All the derivatives and solution procedures for the new reduced flash calculation framework were prepared based on Peng-Robinson equation of state. One reservoir oil sample and one gas condensate sample were selected from published literature to evaluate the proposed method for the calculations of reservoir fluids equilibrium. The equilibrium calculations with the proposed reduction technique were compared to full flash calculations. The reduced formulation implementation is simple and straightforward as it is derived from full flash fugacity equality criteria. The presented technique not only reduces the number of equations, and hence simplifies flash problem, but also presents a comparable convergence behavior and offers the same solution system for different reservoir fluid types. The results, demonstrates the proposed method performance and the accuracy for modeling with complex equilibrium calculations like compositional reservoir simulation when there are many components available in the mixture fluid description.

Predictions of salt solubility, brine vapor pressure and density are necessary to plan subsurface and surface operations. In this work, a comparative study was conducted to investigate prediction capabilities for salt solubility, equilibrium and volumetric properties of several minerals in “Formation” water by the Pitzer model (a commonly used industrial model) and by the ePC-SAFT equation of state (EOS). To achieve this, several validation and comparison steps were considered for reliable implementation of both models. Firstly, solubility, brine densities and vapor pressures for single-salt solutions were predicted. Afterward, the models’ applications were extended to several mixtures of two salts to predict solubility compared to experimental data from available literature. ePC-SAFT shows promising accuracy. The average relative deviation for prediction of liquid density (which cannot be estimated with the Pitzer model) for single-salt solutions of NaBr, KCl, NaCl and Na2SO4 are 1.13%, 0.70%, 1.02% and 0.36%, respectively. The average relative deviation from experimental data of vapor pressure for mixtures of NaCl and KBr is 0.75% and for mixtures of NaBr and KCl is 0.74% for systems with different concentrations. Moreover, a stochastic regression procedure is presented to evaluate the ePC-SAFT parameters. This technique is used for strontium in different salt solutions. The average relative deviation for mean ionic activity coefficients calculated by ePC-SAFT from experimental data using the ePC-SAFT parameters of strontium from the work of Held et al. [1] for single-salt solutions of strontium chloride, strontium bromide, strontium iodide, strontium nitrate and strontium perchlorate are 19.73%, 19.96%, 15.15%, 14.95% and 25.70%, respectively. These values using the ePC-SAFT parameters of strontium regressed in this work are 6.98%, 10.13%, 8.33%, 9.59% and 10.08%, respectively. Finally, the solubility of different salts was studied for mixing of formation water (FW) and injection water (SW).

There are several approaches to model Asphaltene deposition process in the wellbore. There are different assumptions to simplify the problem in the previous investigations for specific conditions, limiting the prediction range of the models. In this work, the effect of precipitated asphaltene particles size is included, to extend the available modeling approaches for deposition profile. To do so, two-dimensional partial differential equations based on asphaltene micro aggregates material balance including asphaltene aggregation, diffusion and deposition are numerically discretized and solved to find asphaltene deposition profile, in radial and vertical directions of vertical oil wells. The modeling results are verified with the results of the well-known ADEPT (asphaltene deposition tool in flow lines) model of Kurup et al. (2011). The size dependent diffusion coefficients of Escobedo and Mansoori (2010) are used to extend the base model. In addition, the population balance method (PBM) was included to improve the aggregation process description with size distribution of asphaltene particles. Based on the developed model a parametric study is performed to study the effect of asphaltene particles average size, flow rate, wellbore radius and fluid viscosity. The model evaluation shows the importance of asphaltene particle size in the deposition profile. In addition, the evaluation results show that as the average asphaltene particle size increases for a given distribution, the amount of deposition in the wellbore decreases.

The measurement of the minimum miscibility pressure (MMP) is one of the most important steps in the project design of miscible gas injection for which several experimental and modeling methods have been proposed. On the other hand, the standard procedure for compositional studies of miscible gas injection process is the regression of EOS to the conventional PVT tests. Moreover, this procedure does not necessarily result in an accurate calculation of the MMP. In this study, an effective procedure is presented using both conventional PVT and slim tube data in the regression to provide appropriate EOS parameters for field studies including miscible gas injection. In the first step, the EOS parameters were subjected to regression to the conventional PVT data. In addition, these parameters were then used as inputs for simultaneous regression to the conventional PVT and MMP data. MMP is modeled through the automated execution of a series of compositional simulation of slim tube. Moreover, the regression uses a stochastic optimization for minimizing an objective function (regression) have been coupled with two separate core calculations, (1) equilibrium calculations of the conventional tests and (2) compositional simulation of the slim tube. For evaluation, a number of real reservoir fluids from field data are used from reliable datasets in the literature. Finally, the promising results demonstrated that this procedure is capable to provide EOS parameters for accurate predictions in the miscible gas injection processes.

This paper introduces an effective production optimization and a water injection allocation method for oil reservoirs with water injection. In this method, a two-stage adaptive simulated annealing (ASA) is used. A coarse-grid model is made based on average horizon permeability at the beginning iterations of the optimization to search quickly. In the second stage, the fine-grid model is used to provide the accuracy of the final solution. A constrained optimization problem to maximize an objective function based on net present value is implemented. Allocation factors from the streamline simulation are used to help for the appropriate estimation of initial water injection rates. The proposed optimization scheme is used for a field sector simulation model. The results show that the optimized rates confirm the increment of total oil production. Optimized oil production and total water injection rates lead to an increase in the total oil production from 385.983 (initial guess) to 440.656 Msm3. This means a recovery factor increment by 14.16%, while the initial rates were much higher than the optimized rates. Moreover, the recovery factor of optimized production schedule with an optimized total injection rate is 2.20% higher than the initial production schedule with an optimized total water injection rate. The allocation of the water injection rates and the revision of allocation rates result in 446.383 and 450.164 Msm3. The revision of the water rates allocation provides a reduction of water cut during production.

COVER CAPTION: The cover image is based on the article Dual Porosity Simulation of Gravity Drainage Mechanism Induced by Geological Acid Gas Storage in Naturally Fractured Reservoirs by Mehdi Assareh et al., https://doi.org/10.1002/ese3.70094.

Accurate predictions of equilibrium, transport, and interfacial properties of H2S-brine systems play a significant role in improving acid gas sequestration efficiency in saline aquifers. In the current study, molecular dynamics (MD) simulations were used to simultaneously predict the interfacial tension (IFT), mutual solubility, viscosity, and density of the H2S-brine solutions to compensate for the lack of experimental data on these crucial properties. The effects of temperature, pressure, salt type and concentration on the properties influencing the storage process are investigated in the ranges of pressures up to 30 MPa, temperatures of 323.15–393.15 K, and salinities of 1–3 mol/kg, which represent typical conditions of acid gas geological storage. We employed mixed brine systems with the most common monovalent and divalent salts to study the impact of different ions on the desired properties with detailed microstructural insight. A comprehensive validation was performed to verify the accuracy of MD model by comparison with experimental data of the H2S-water and H2S-NaCl solutions. The average absolute deviations percent (AAD%) of 4.27, 5.20, 3.74, and 4.93% were obtained for reproducing the H2S solubility in water, H2S-rich phase water content, density, and IFT of the H2S-water system, respectively, close to the experimental uncertainties. The increasing ion concentration results in a decrease in the mutual solubilities and a linear increase in IFT values because of forming contact ion pairs, which is more remarkable in the CaCl2-containing solution. However, the salting-out effect is more pronounced for the solubility of H2S than water content. IFTs of the H2S-NaCl + CaCl2 (aq) are greater in comparison with those of H2S-NaCl + KCl (aq) solution, while this trend is reversed for mutual solubilities. The H2S dissolution reduces the density values of brine solutions compared to fresh brine, which has an unfavorable impact on the density-driven convective process and has the higher opposite effect on the viscosity values. The most effective parameters on the density and viscosity values are salinity and temperature, respectively, under our studied operating conditions.