3D Reservoir Research Articles

Three‐dimensional static reservoir modelling of Kareem sandstone reservoir, Tawilla Oil Field, at the southern region of Gulf of Suez, Egypt

The Gulf of Suez, which contains Egypt's oldest oil fields, is one of North Africa's most well‐known oil regions. There are more than 80 conventional oil fields in Egypt's Gulf of Suez, some of which have reservoirs that stretch back to the Precambrian and Quaternary. In close proximity to the southern entrance of the Gulf of Suez is the Tawilla West oilfield. The oil field Tawilla West is believed to consist of rotating fault blocks that descend in a south‐west direction. The main producing reservoirs are the Miocene section reservoirs, the Belayim and Kareem sandstones. The current research is focusing on the structural elements affecting this giant field to update the field structural model using the newly processed 3D seismic survey, the acquired data from newly drilled wells and the associated different logging techniques. The seismic information quality varied from poor to fair. The quality of the interpreted stratigraphic horizons and geological faults was mainly controlled by the seismic information quality. The research used seismic attribute analyses to improve interpretation and incorporate additional features, enabling better hydrocarbon potential identification and characterization of the reservoirs. Several geological structure contour maps and cross‐sections were generated to help in delineating and understanding the reservoir's extension. Based on the detailed correlation study, we were able to detect the faults that affected the structure of the Tawilla West field in detail, define their throw amounts and directions, and identify the missed sections across the studied area. This study introduces an updated model scenario to show the differences and their effect on the field development plan and recommendations. By examining subsurface geologic structural characteristics and evaluating petrophysical data, a 3D static reservoir model was created to resolve structural settings and hydrocarbon trapping, providing detailed information on the field and identifying new opportunities for future development. The research discovered that the updated detailed 3D structural model may support the Kareem Reservoir development plans and encourage drilling, workover and dynamic operations to assign development possibilities in the correct area. According to the established model, there are at least three options in the study's attic areas that might boost oil output and oil reserves for the field while avoiding further failures.

Integration of seismic and well log data for evaluating the upper cretaceous reservoirs of Horus oil field, Alamein Basin, Northern Western Desert, Egypt

Horus field stands out as one of the main oil-rich areas in Alamein Basin. Formation and evolution of Horus field were influenced by tectonic processes occurring from Jurassic to Cretaceous periods. The main objectives of this study are to characterize structural elements, identify hydrocarbon potential, and pinpoint promising zones within Early Cretaceous rocks for hydrocarbon accumulations in Horus reservoirs. By utilizing high-quality 2D seismic reflection sections enhanced with structural smoothing attribute, which help eliminate random noise while preserving structural details and horizon geometry, building a 3D structural model becomes facile. Analysis of well logging data provides a comprehensive understanding of Cretaceous reservoirs, simplified by 3D static reservoir modeling to depict their varying ranges and anisotropies. 3D structural model shows that the trap of Horus field is linked to a structural uplift in an ENE-WSW trend, bordered to north and south by sediment-filled lows. This anticlinal ridge within the model is segmented into blocks by a series of NW-SE normal faults. 3D static reservoir models reveal high-quality reservoirs in Upper Bahariya and Abu Roash (G) Dolomite, characterized by substantial thicknesses of net pay zones. 3D static modeling of Cretaceous reservoirs guides the oilfield development and optimizes production strategies in Horus area.

Investigating forced convection in porous media-filled cavities for designing supplementary passive cooling systems

This study aims to propose innovative passive cooling methods within the reservoir section of a closed-loop liquid cooling system. The techniques employed in this study include using a cold wall, a porous medium, varying inlet positions, and nozzles of different sizes and shapes. The effects of parameters associated with these techniques on outlet temperature reduction, pressure drop, and average Nusselt number (Nu¯) have been analyzed in detail. The reservoir section has been numerically modeled as a two-dimensional (2D) and a three-dimensional (3D) cavity. The numerical model implements mass, momentum, energy, and conservation equations for the Darcy model in the porous region. A fluid flow with Re=50 with a laminar generalized porous media modeling approach employing Brinkman’s and Forchheimer’s terms has been used. The optimized inlet position with respect to maximum cooling (55.60 K) and moderate pressure drop (2.30kPa) for the 2D cavity has been identified in this work for a given Darcy number and porosity. Results show that decreasing porosity improves convective heat transfer efficiency and increases pressure drop. Similarly, varying Darcy’s number affects heat transfer, with lower Da values resulting in reduced outlet temperature reduction and lower Nu¯, with higher pressure drop. A similar analysis was conducted for the 3D reservoir. The use of a converging nozzle can lead to an increase in cooling by 61.61% and 3.44% for the 2D and 3D reservoirs, respectively.

A latent space method with maximum entropy deep reinforcement learning for data assimilation

Data assimilation aims to calibrate the uncertain state or parameter vectors of a system by matching simulation results with observations, which is crucial for uncertainty quantification and optimization. Although traditional methods of data assimilation have shown promising results in practical applications, they need well-designed iteration rules (i.e., gradient, covariance, search strategies). Deep reinforcement learning (DRL) can solve data assimilation problems by trial and error, which does not require convoluted iteration rules. However, previous DRL methods cannot tackle high-dimensional data assimilation problems efficiently. In this work, we propose a latent space method with maximum entropy DRL for data assimilation to extend DRL to complicated systems with lots of parameters. Solutions can be found in a reduced space, through the interaction between the agent represented by artificial neural networks and the environment of numerical simulation. The proposed method contains two key points. First, to make the method applicable to high-dimensional problems, we construct a latent space method by integrating a dimensionality reduction method with the state, action, and reward settings for the agent. Second, a maximum entropy DRL algorithm, Soft Actor–Critic (SAC), was employed to efficiently explore the parameter space. The proposed method overcomes the limitation that previous DRL algorithms are only suitable for small-scale cases, and enables DRL to deal with medium-scale or even large-scale data assimilation problems. The performance of the proposed method was validated by comparison with a deep reinforcement learning algorithm and traditional data assimilation algorithms on 2D synthetic and 3D reservoir cases.

(Invited) Solution-Processable Materials and Structuring Processes for Electrochemically Gated Thin Film Devices

Printed and flexible electronics enable interesting novel applications in the fields of sensors [1], bioelectronics [2], and security applications [3]. Most of them gain their functionality from a layered structure composed of different materials that interact with each other. For semiconducting layers, organic and inorganic materials can be used; in this regard, the respective functional materials need to be formulated into inks that are then printed on various substrates. The structuring process of printing is a complex process in which many chemical and physical process parameters need to be controlled at the same time, such as drop volume, the waveform of the electrical signal controlling droplet formation, or the drying time. Also, the surface of the substrate plays a decisive role in printing accuracy and layer morphology. After printing, post-treatments are often necessary to obtain the desired features, since the properties of bulk materials often differ from printed nanoparticulate or precursor-derived films. For instance, temperature sintering of printed layers can improve conductivities as well as UV-curing which also allows sintering at lower temperatures.As functional materials for thin-film transistors and other devices, in particular inorganic materials such as metal oxide precursor and nanoparticulate inks are used. These can show advantages of high carrier mobilities, transparency, stability in air, non-toxicity, and can be prepared mostly by using water or alcohol-based solvents. We employ a large portfolio of various printing techniques, including inkjet, gravure, screen, and even direct laser printing. Nevertheless, for all of these techniques, printing thin pinhole-free dielectrics is a challenge and crucial for the whole device functionality, since pinholes can lead to short circuits or other defects of the device. Especially for transistors, this is a fundamental problem, since very thin dielectric layers are needed between the top gate and the semiconducting channel.To circumvent this problem, we have developed in the past years a composite solid polymer electrolyte (CSPE) consisting of two solutions, namely LiClO4+PC and PVA+DMSO. This approach allows employing a different geometry of the transistors and a thicker electrolyte layer while keeping the functionality. The electrolyte ink is inkjet-printable and dries at room temperature. The advantage of such electrolytes is their ability to cover well even rough film surfaces and create a large Helmholtz double layer capacitance, which causes low threshold voltages and hence allows for low-voltage operation of the printed devices. We have further expanded the electrolyte to chemically cross-linked ion gels that even show better device reliability concerning the humidity of the environment at similar electrical performance. Our recent works on device and component levels include ultra-low-power latches operating down to 0.6 V, pn- and Schottky-diodes, resistors, and logic gate standard cells such as inverter, NAND, and NOR-gates, as well as circuits such as ring oscillators, rectifiers, and security primitives [1].This presentation starts by introducing various semiconducting inks based on either nanoparticles or precursor solutions and discusses thin film properties, microstructure, and finally the step-by-step device fabrication processes. We employ several structuring processes along the process chain, including laser ablation and lift-off, digital printing techniques such as inkjet, superinkjet, and direct laser printing. In addition, doping concepts to tune the threshold voltage and composite solid polymer electrolytes that are compliant with silver electrodes will be presented. The talk will further discuss a plethora of realized devices, including transistors, memristors, and even fully integrated circuits. In more detail, the fabrication and electrical performance of inkjet and laser-printed ZnO digital memristors as well as WOx analog memristors which show “learning” behavior will be discussed [4]. The conductance of analog-type devices can be gradually manipulated to emulate different behaviors of the biological synapse, such as short-time plasticity, long-term potentiation, and long-term depression.In addition, progress towards transistor architectures using nanoparticle-passed InO as the semiconductor and Pt or ITO as electrode materials is shown. Along these lines, transistors with InO-semiconducting nanoparticles printed into a 3D laser-written polymer reservoir with a 10 µm height are presented and their electrical characteristics discussed. Our devices show high ion currents in the range of ~mA for typical widths of a few hundreds of micrometers. These 3D hybrid devices have a huge potential for biohybrid systems, in which the reservoir structure can be used to accommodate biological material or tissue, or even single cells.[1] Marques et. al., Advanced Materials 31, (2019), https://doi.org/10.1002/adma.201806483.[2] Saghafi et. al., Adv. Funct. Mat., Early View, (2023), https://doi.org/10.1002/adfm.202308613[3] Zimmermann et. al., IEEE Transactions on VLSI Systems 27 (2019), https://doi.org/10.1109/TVLSI.2019.2924081[4] Hu et. al., Adv. Funct. Mat., EarlyView (2023), https://doi.org/10.1002/adfm.202302290

Coalbed methane reservoir properties assessment and 3D static modeling for sweet-spot prediction in Dahebian block, Liupanshui Coal field, Guizhou Province, southwestern China

Guizhou Province has many multi-layer, thin-thick coal seams; however, complex geology, incomplete reservoir characterization, and sweet-spot selection technology prevent large-scale coalbed methane (CBM) development. This study evaluates the CBM reservoir properties within the Dahebian block using logging data, coal sample analysis, and well-testing data and develops a 3D static reservoir properties model to analyze their spatial and vertical propagation. A sweet spot evaluation model was established using a multi-level fuzzy method based on 9 parameters extracted from a 3D static reservoir properties model. The coal measure has 22 coal seams, and seams >2 m thick have 2 or 3 thin non-coal layers intercalated. Coal seams 1#, 7#, and 11# are thin to thick, deeply buried, widely distributed, and have high gas content and saturation. Undeformed and cataclastic coal predominates the coal seam 1# and 7#, whereas coal seam 11# is dominated by cataclastic and granulated coal. The southern and central parts of coal seam 7# and 11# have less tectonically deformed coal (TDC). Coal seams 1# and 7# have low permeability relative to seam 11# and are localized, while coal seams 11# have high permeability, are extensively distributed, and contain substantial gas concentrations. Comparative analysis of evaluation scores and CBM production statistics shows that high scores indicate sweet spots for CBM development. Sweet-spot potential was classified as high, medium, and low. Scattered sweet spots are found in single layers, while combined development (1# + 7#+11#) reveals a wider high-potential area in the south-central region. This area, featuring deep, thick coal seams, high permeability, gas saturation, reservoir pressure, and low TDC proportion, indicates significant development potential. This study validates CBM development statistics, identifies future development areas, and guides the development of geologically complex Guizhou CBM.

SURFACTANT–POLYMER FLOODING: GEOCHEMICAL MONITORING OF THE PROPERTIES OF FORMATION FLUIDS AT THE IMPACT AREA

Chemical flooding methods improve oil extraction after standard waterflooding processes. Chemical EOR methods modify different properties of fluids and/or rock to mobilize the remaining oil. It is expected that the residual oil will be different in properties than conventional. A new technology is presented to study a surfactant–polymer flooding, based on a periodic sampling and measuring properties of oil and water during 15 months. The study area is a terrigenous reservoir with 3 chemical injection wells and 16 production wells. High-precision mass spectrometry methods were carried out on each sample, which made it possible to evaluate the effectiveness of ongoing work on chemical flooding. At each sampling date, 2D reservoir model was built to track these changes in the study area. It is shown that the composition of the water has changed - highly permeable channels were closed and formation waters were involved in the production. The properties of oil have also changed - depending on the site of impact, there are areas with more degraded oil in production after the injection of chemicals. Geochemical monitoring shows the redistribution of filtration flow directions in the reservoir volume.

IMMISCIBLE VISCOUS FINGERING MODELING OF TERTIARY POLYMER FLOODING BASED ON REAL CASE OF HEAVY OIL RESERVOIR MODEL

In immiscible displacements, lower viscosity injected fluids with higher mobility than crude oil can create viscous fingers, affecting displacement efficiency. The Buckley–Leverett approach for relative permeabilities (kr) may not represent accurately 2D features like increased water saturation in viscous fingering. Based on the physics issue, this work applies Sorbie's 4-Steps methodology to a 3D simulation of an offshore heavy oil reservoir focusing on waterflooding and tertiary polymer flooding, assessing their impact on oil production forecasts. It also explores the application of this methodology to coarse grid simulation models, employing pseudo kr functions by data assimilation. During tertiary polymer injection, two processes were identified in oil displacement: viscous crossflow mechanism and oil bank mobilization by a second finger. This combination resulted in earlier and increased oil production. For both strategies, refining the grid increased simulation runtime from minutes to days compared to coarse grids, making it impractical for intensive processes. From data assimilation, the best solution with matched field indicators reduced runtime from days to minutes. This study expanded the 4-Steps methodology for 3D reservoir simulation, proposing kr as uncertainties. Data assimilation enhances the methodology, generating pseudo kr for coarser grid simulations, reducing computational costs, and capturing small-scale phenomena.

3D Marine Controlled-Source Electromagnetic Numerical Simulation Considering Terrain and Static Effect

The marine controlled-source electromagnetic (CSEM) method is an important geophysical method for the exploration of marine hydrocarbon resources. In marine CSEM forward modeling the uplift terrain, such as submarine hills and seamounts, the static effects caused by polymetallic nodules and hydrothermal sulfide are ignored which can lead to the deviation of marine CSEM data. To improve the accuracy of data processing and interpretation, this study realizes efficient 3D numerical simulation considering submarine uplift terrain and static effect based on the finite difference (FD) algorithm in a fictitious wave domain. First, based on the correspondence principle between the fictitious wave domain and a real diffusive domain, we derive the FD electromagnetic field iteration equations of the fictitious wave domain, and apply the fictitious emission source and the boundary condition of complex frequency shifted-perfectly matched layer (CFS-PML). We use the inverse transformation method to convert the electromagnetic response to the diffusive domain. Then, we carry out simulations on typical 1D and 3D reservoir models to verify the correctness and effectiveness of the algorithm. Furthermore, we design an uplift terrain model and a static effect model and study the influence of parameters such as top width, bottom width, height and volume of uplift terrain on the CSEM field response characteristics through the forward modeling of multiple models and discuss the influence of parameters such as electrical conductivity, length, width, thickness, depth and volume of a shallow anomaly on the marine CSEM response. Finally, we analyze the characteristics and rules of electromagnetic field propagation of uplift terrain and static effect, which provides theoretical guidance for the design of marine CSEM exploration systems.

Integrated 3D Reservoir Modeling and Tectonic Evaluation of the Upper Cretaceous Bahariya Formation in the Burg El Arab Area, Egypt: Implications for Hydrocarbon Exploration and Production Strategies

Integrated 3D Reservoir Modeling and Tectonic Evaluation of the Upper Cretaceous Bahariya Formation in the Burg El Arab Area, Egypt: Implications for Hydrocarbon Exploration and Production Strategies

A maximum entropy deep reinforcement learning method for sequential well placement optimization using multi-discrete action spaces

Well placement optimization is a crucial method to solve the planar conflicts in reservoir development, mainly to determine the optimal well locations and drilling sequence to maximize the economic benefits of reservoir development. However, the current well placement optimization methods face the problems of high-dimensional discretization of optimization variables and lack of effective incentives for policy exploration, which make it challenging to improve the global optimization ability (the ability to jump out of locally optimal solutions in time and continuously search for better solutions in the whole optimization process) and real-time adjustability of the well placement optimization methods under limited numerical simulation times. In this paper, we propose a new sequential well placement optimization method, based on the Discrete Soft Actor-Critic Algorithm (DSAC), which incorporates the maximum entropy mechanism to formulate well placement and drilling sequencing schemes more efficiently and maximize the net present value (NPV) over the entire life cycle of reservoir development. Specifically, the method models the well placement optimization problem as a Markov Decision Process (MDP) and achieves sequential well placement optimization by training a Deep Reinforcement Learning (DRL) agent that maps reservoir states to a stochastic policy of well placement variables as well as evaluates the value function of the current policy. The DRL agent can determine the optimal infill well location in real-time based on the reservoir state at different times during the development process, thus obtaining the optimal drilling sequence. The proposed method in this paper has two innovations. First, by reconstructing the large-scale discrete action space of well placement optimization variables into multi-discrete action spaces, and with the maximum entropy mechanism, policy exploration is encouraged to improve the global optimization capability. Second, the trained policy can swiftly adapt the subsequent well placement scheme for a specific state of the target reservoir without the requirement to initiate training from scratch, which can realize the offline application of the trained policy and has better real-time adjustability. To verify the effectiveness of the proposed method, it is tested in 2D and 3D reservoir models. The results show that DSAC not only outperforms the gradient-based optimization method, classical evolutionary algorithms, and existing reinforcement learning proximal policy optimization (PPO) method in terms of global optimization ability but also shows better real-time adjustability of the trained policy when applied offline.

An efficient transformer-based surrogate model with end-to-end training strategies for automatic history matching

In recent years, surrogate models have gained popularity as a tool to tackle the challenges presented by time-consuming numerical simulations in automatic history matching (AHM). Although there are many different surrogate models designed to alleviate this issue, it is still challenging for most of them to handle the high dimensionality and strong non-linearity in reservoir models and dynamic production data. Inspired by the rapid development of Transformer, we propose a novel hybrid hierarchical Vision Transformer (HHVT) approach for history matching, which utilizes a unified architecture to predict production data for specific physical fields with an end-to-end strategy. For predicting the production data of wells, the Transformer supports parallelism computation among multiple timesteps, which shows more superiority than traditional recurrent neural networks. Specifically, our approach constructs a novel encoder-decoder Transformer architecture to learn the implicit features of high-level spatial parameters to match the features of time-series production data. With this architecture, HHVT achieves fast training and inference, which is suitable for large-scale datasets and high-dimensional features. The proposed HHVT model is integrated with a multimodal optimization algorithm to find history-matching solutions. We first validated the effect of hyperparameters of HHVT on a simple 2D reservoir. Moreover, the proposed method was verified on the complex 3D Brugge model. The results demonstrate that the training speed of the Transformer-based model is approximately twice as fast as the surrogates based on convolutional and recurrent neural networks. The proposed HHVT also shows better prediction accuracy in two cases, compared with other surrogate models, which enhances the applicability of surrogate-based history-matching methods in large-scale complex reservoir scenarios.

3D Seismic Data Interpretation and Static Reservoir Modeling of Mamuniyat Formation, I and R Oil Fields, Murzuq Basin, Libya

3D Seismic Data Interpretation and Static Reservoir Modeling of Mamuniyat Formation, I and R Oil Fields, Murzuq Basin, Libya

Graph theory based estimation of probable CO2 plume spreading in siliciclastic reservoirs with lithological heterogeneity

Estimating plume spreading in geological CO2 storage reservoirs is critical for several reasons including the assessment of pore space utilization efficiency, preferential CO2 migration pathways and trapping. However, plume spreading critically depends on lithological heterogeneity of the reservoir and CO2 injection rate. It might require numerous high fidelity full physics numerical simulations to constrain the uncertainty in plume spreading for a given reservoir. This might not always be practical due to computational limitations. Hence, reduced physics approaches, such as invasion-percolation method and machine learning, could be useful to answer certain questions on plume spreading in the subsurface. This study presents a new reduced physics approach based on graph theory for estimating probable CO2 plume migration under very low and very high injection rates. The two end-member scenarios are assessed by performing random walk in the 3D reservoir space to constrain 20,000 possible paths of CO2 flow away from the injection well. The resistance to CO2 flow associated with each path is computed for viscous, capillary and gravity forces. The resistances are then transformed into the likelihood of CO2 migration along the path. The algorithm was applied to 45 reservoir models with varying degrees of lithological heterogeneity and the results were compared to those from full physics and invasion percolation simulations. The graph theory results showed a close match with the results from full physics approach for both flow regimes and with results from invasion-percolation approach for capillary-gravity dominated flow regime. The algorithm was further applied to answer key questions on reservoir screening such as pore space utilization potential. The graph theory approach was also integrated with machine learning to predict CO2 saturation. Testing suggested that the graph theory approach can be as much as 50 and 20 times faster than the full physics numerical simulations and invasion-percolation simulations, respectively.

Few-shot learning and modeling of 3D reservoir properties for predicting oil reservoir production

Few-shot learning and modeling of 3D reservoir properties for predicting oil reservoir production

3D Reservoir modelling for Sidi AbdelRahman oil field, Alamein basin, Western Desert, Egypt

3D Reservoir modelling for Sidi AbdelRahman oil field, Alamein basin, Western Desert, Egypt

3D static modeling and sequence stratigraphy using well logs and seismic data: An example of Abu Roash G member in Bahga oilfield

This research uses both three-dimensional (3D) modeling and geologic well control to piece back together the architectural parts of the Late Cretaceous formations. The goal is to figure out the sizes, directions, locations, and controls of the layers of the fluvial sandstone reservoirs. Sequence stratigraphy is essential for 3D reservoir modeling and petroleum geology understanding in the Bahga oilfield. The purpose of this work is to create a static model that shows the layers and facies distribution in the reservoir interval. We will use data from nine well logs and 22 seismic lines calibrated by the Abu Roash G Member reservoir core intervals to accomplish this. The petrophysical study discovered three parts in the Abu Roash G Member reservoir rock: channel fill that is affected by tides, channel fill that is dominated by tides (intertidal sands), and channel top with lenticular bedded sandstone. The model's findings point to the existence of an NNW-oriented sand body, which could be a prime location to produce hydrocarbons. The original oil in place (OOIP) is about 3,438,279 Stock Tank Barrels (STB), and the oil reserve reaches up to 1,031,484 (STB). Sequence stratigraphic analysis using seismic and well log information (SB) reveals that the Upper Cretaceous AR/G reservoir of the Bahga field is characterized by third- and fourth-order stratigraphic sequences, which are constrained by three Maximum Flooding Surfaces (MFS) and two Sequence Boundaries. The integration of the derived geological model and sequence stratigraphic results can lower future extraction risk by identifying the locations and trends of the geologic facies with the necessary petrophysical properties for the hydrocarbon accumulations.

Reservoir Static Modelling towards Safe CO2 Storage in Depleted Oil Reservoirs of ‘CRK’ Field, Niger Delta, Nigeria

Geological carbon storage (GCS) is gradually gaining acceptance as the technology of highest repute for the mitigation against greenhouse gas effect. This involves the injection and long-term or permanent storage of carbon dioxide (CO2) in subsurface geological formations. The development of a robust 3D reservoir model then becomes necessary to understand the facies changes and the petrophysical properties distribution of candidate CO2 storage reservoirs. This study attempts the use of static modelling technique to investigate the depleted oil and gas reservoirs of ‘CRK’ field in the Niger Delta sedimentary Basin as a potential CO2 storage site. Using an integrated approach of 3D seismic and a suit of well logs from the study area, two reservoir sands (D10C0 and D6200) were mapped. 3D static modelling of discrete and continuous property distribution of the reservoirs revealed the reservoirs are composed majorly of clean sands with water saturation ranging from 8 to 75%. The average porosity and permeability values are between 20 to 29% and 250 to 890mD respectively. A theoretical storage capacity of 24.85Mt is estimated for the two reservoirs combined, while the median effective storage capacity is 6.22Mt. Comparison of the results obtained in the study with recommended standard values show the reservoirs are suitable for CO2 storage. The property models constructed can serve as primary input for dynamic simulation of the oilfield in future studies.

An effective multiphysics toolkit for lithium prospecting: From geophysics to the static reservoir model in the Pozuelos salt flat, Argentina

Energy transition drives the energy sector to renewable energy and electrification, being the critical minerals key players in the industrial development map. They comprise rare-earth elements and 35 other elements, including lithium, which holds 60% of its world reserves in the so-called lithium triangle located in Argentina, Bolivia, and Chile. The low electrical resistivities, variations in salt concentrations, low acoustic impedances, and dynamics of the hydrogeologic system make brine monitoring a complex geophysical exploratory problem. So, the objective is to find a suitable combination of geophysical techniques that fit the lithium exploration objectives, which are the characterization of the salt flat in-depth, fluid detection, basement delineation, definition of the main structures and main faults, and detection of semifresh water aquifers that contribute to its recharge and that are key to the reservoir water balance of an endorheic basin, which has the resource in solution. For this purpose, the evaluation of several prospecting methods in different salt flats was executed, concluding that full tensor magnetotellurics (MT), electrical resistivity tomography, and gravity comprise a toolkit that fits the objectives set. The methodology was validated at the Pozuelos salt flat. The results show that the fresh water-brines contact and the recharge system were well-defined and imaged by ERT. The full tensor MT detected two ultraconductive units: a shallower multilayer system saturated with brines, has a 400 m thickness, whereas the deeper unit has a 500 m thickness. Both MT and gravity characterized the basement, and gravity successfully delineated the main structures. The geophysical interpretation is in concordance with shallow and deep exploration wells. Finally, the integration of geophysical and well data allowed the construction of a 3D static reservoir model that finds the deepest basement area at approximately 900 m depth and discriminates eight lithofacies.

Optimizing geothermal reservoir modeling: A unified bayesian PSO and BiGRU approach for precise history matching under uncertainty

This research focuses on optimizing geothermal reservoir modeling tackling issues related to non-uniqueness, subsurface uncertainties, and computational intensity. The proposed workflow integrates Bayesian Particle Swarm Optimization (PSO) and a Bidirectional Gated Recurrent Unit (BiGRU) network to enhance the efficiency of history matching in the context of geothermal resource development under uncertainties. The methodology involves four key steps: firstly, identifying crucial uncertainty parameters such as injection rates and reservoir temperature; secondly, constructing the BiGRU network to capture the nonlinear relationship between these parameters and time series outputs, with Bayesian PSO automating hyper-parameter tuning; thirdly, using Bayesian PSO for the inversion of uncertainty parameters, utilizing the BiGRU as a forward model to reduce computational costs; and finally, conducting a thorough analysis and refinement of errors in predicted responses between the high fidelity model and Bayesian PSO, ensuring the accuracy of the history-matching process. Validation using a 3D fractured geothermal reservoir scenario demonstrates the Bayesian PSO and BiGRU method's efficacy in inverting uncertainty parameters with narrow uncertainties. The Obtained results are compare the performance of predicting the recovery factor (RF) using different surrogate methods, including Gaussian Processes (GP), Radial Basis Function (RBF), and Random Forest Regression (RFR). The proposed Bayesian PSO and GRU-based history-matching approach proves highly accurate and robust for efficiently addressing significant uncertainties in geothermal extraction processes.