Mês: agosto 2026

The triple porosity/ permeability system in carbonate reservoirs, plays an essential role in controlling fluid storage and flow dynamics. Matrix porosity is typically small and homogeneously distributed, while non-matrix porosity is related to post-depositional process and often results in excess permeability that significantly impacts reservoir performance. In the pre-salt carbonate reservoirs of the Barra Velha Formation (BVE) in the Santos Basin, non-matrix porosity is a key driver of enhanced production. Consequently, numerous studies have been conducted to understand the occurrence and relationships of these features within the reservoirs, revealing a correlation between non-matrix porosity, silicified intervals, and fractured zones.
To improve the knowledge on the distribution and magnitude of the vuggy areas away from the boreholes, this work applies a machine learning approach into 3D seismic interpretation using multi-attribute analysis, by establishing a mathematical relationship between the seismic data and vuggy porosity through the application of a Multiattribute Analysis in a dataset composed of a PSDM seismic volume and wells with conventional and special logs, including vuggy porosity, and Dry Weight Silicon (DWSI) logs. Once a transform is found and validated, through the quantification of the final prediction error, we produce a seismic volume whose Z value is the predicted property at the target interval, which is then correlated with production rates and silica content.
Illumination differences, seismic resolution, and presence of coherent noise in the pre-salt section of the seismic volume are the main factors that challenge the application of this quantitative interpretation workflow. To address this, we have established low, middle, and high confidence zones in the seismic data and incorporated them into result evaluation. Finally, this study provides a more accurate understanding of the spatial distribution and magnitude of non-matrix porosity, enhancing reservoir characterization and offering valuable insights into production optimization and geological modeling in the BVE.

The objective of this work is to present a new practical methodology to manage petroleum fields considering three stages (life-cycle, short-term, and real-time) that can run alongside different model fidelities and characteristics. The model-based field management process follows the general methodology proposed by Schiozer et al. (2019) with four activities: (1) fit-for-purpose models construction, (2) data assimilation for uncertainty reduction, (3) life-cycle production optimization and (4) short-term optimization for real-time implementation. The selection of the production strategy for field management comprehends the last two activities. Life-cycle optimization is the first stage of the process and generates control setpoints for short-term analysis. Short-term optimization is then used to improve the quality of the solutions considering the control parameters of the next cycle (considering a closed-loop procedure). Real-time solution is then implemented considering operational disturbances from real operations. The methodology was applied to a benchmark case (UNISIM-IV-2026) which is a case based on a typical carbonate field from the Brazilian Pre-salt, with light oil and submitted to Water-Alternate-Gas injection with CO2 (WAG-CO2). The results show that the methodology is applicable to real and complex fields. As the three stages can run simultaneously, one can (1) use different model fidelities to improve the quality of the solutions and (2) use model-based solutions for real-time implementation. Life-cycle optimization using complex simulation models and long-term objectives can run in the background to generate control setpoints for short-term analysis in which lower fidelity models and simplified solutions can be used for the control and field revitalization parameters of each closed-loop cycle. Real-time solutions can be implemented considering operational problems and disturbances. This work presents a novel procedure to integrate three stages for production optimization that can run in parallel, allowing the integration of life-cycle and real-time solutions. The methodology (1) allows the use of complex reservoir simulation models from the life-cycle production strategy optimization, (2) focuses short-term control parameters that improve the quality of the short-term solution, and (3) guides real-time implementation, so it can be the basis to a digital field management.

The chemical injection (CI) lines in oil-producing wells have been plagued by issues related to clogging, hindering the injection of demulsifiers into the oil reservoir, which directly impacts oil productivity. This problem might require stoppages in the oil production for cleaning and/or pigging the line, and even replace downhole equipment (chemical injection valves). Given this scenario, the present study aims to investigate the compatibility of fluids present in the chemical injection line. If these chemicals are not compatible and result in solid formation, it could lead to pipe blockages. To assess this, monoethylene glycol (MEG), commonly used as a flushing fluid and hydrate inhibitor, and an ethoxylated polymeric surfactant, used as a demulsifier, were mixed to observe any physical change in the solution. The tests were conducted at room temperature (≈30 °C) and 60 °C, for up to 72 h, with visual monitoring during this period. In addition, rheological tests were carried out on pure fluids and their mixtures to evaluate if there were viscoelastic changes. These studies made it possible to detect a whitish gel at the bottom of the test tube formed through contact between the MEG and the demulsifier (phase separation) at both temperatures. This allows us to conclude that physical changes occurred in the mixture (MEG + demulsifier), forming a higher viscosity gel. Importantly, preventing this gel formation could possibly prevent clogging in the CI lines, as the gel could adhere to solid contaminants and contribute to blockages.

Optimization problems often involve expensive computational calculations. For example, in petroleum well place-ment problems, optimization techniques are applied to reservoir simulation models to identify the optimal number and positions of exploration wells. Each evaluation of a well placement plan requires extensive calculations of differential equations to forecast fluid behavior. To enhance the quality of the optimization techniques without increasing the computational burden, re-searchers have explored combining optimization methods with Surrogate Models, approximation techniques used in regression problems to replace expensive objective functions with more computationally efficient evaluations. However, since most of the research on surrogate modeling has been performed for single-objective problems, surrogate modeling for multi-objective problems remains a challenging task. For this reason, this paper provides a comparative analysis of three supervised learning techniques, Random Forest (RF), Multi-Layer Perceptron (MLP), and k-Nearest Neighbors (k-NN) when applied as multi-objective surrogate models to approximate the real objective function val-ues of five datasets of a real-world expensive discrete petroleum well placement problem. We also investigate if an ensemble of these three methods surpasses their individual performances. The results demonstrate that the ensemble prediction outperforms the individual methods in 4 out of 5 datasets in terms of Root Mean Square Error, while k-NN provides the best Mean Absolute Error in 4 out of 5 datasets. Nonetheless, the Friedman and Bonferroni-Dunn tests indicate that there are only statistically significant differences between MLP and the remaining methods. Therefore, the choice between k-NN, RF, or the ensemble is indifferent.

Along the production lines, crude oil and water are produced together, being subjected to high shear and turbulence. Thus, the flow conditions can provide the necessary energy to disperse one phase into another, producing emulsions along the flow lines. Among the main types of emulsion, the water-in-oil (W/O) type is the most common and the one that most impacts the production. The main characteristic of emulsions is the increased viscosity in comparison to the isolated fluids (water and oil), which results in lower flow rates in the production system. The internal coating of pipes and other surfaces exposed to the flow has been pointed out as a strategy to minimize this problem, as it can change the physicochemical properties of the surface, being able to reduce surface-fluid interaction. Thus, this work aims to analyze whether a coating on the inner wall of the pipes can reduce the drag of the fluid with the pipe wall and, consequently, the pressure drop in emulsion flows. For this purpose, two samples of the same steel surface were evaluated, one commercial coated and the other uncoated. The wettability of the surfaces by fluids was determined by measuring the contact angle (θ) using a tensiometer based on the Sessile Drop methodology. The fluids used in the study were deionized water, two mineral oils, brine (sodium chloride solutions), crude oil, production water, and w/o emulsions. The temperature varied between 25 and 45°C and the test time was fixed at 10 minutes. Based on the contact angle tests, mineral oils presented a higher affinity for the uncoated surface compared to the coated surface. Regarding the water/oil emulsion, the results indicated a lower affinity for the coated surface, suggesting a reduction of the fluid/ pipe wall drag and an improvement of the flow under dynamic conditions. The effect of coating pipes will be done in a flow loop with water-oil emulsions and the results will be evaluated in terms of pressure drop.

Emulsions are common during oil production, and this is the case of the Peregrino field, e.g. . In several fields, oil and reservoir fluids can be produced through Electrical Submersible Pumps (ESP) and the formation of emulsion affects the ESP system, its run life, and the production itself. The viscosity increase due to emulsion formation severely hampers the ESP performance, requiring an increase in energy consumption on the platform to keep the same production rate. This work presents an initial validation of the model proposed by Bulgarelli (2021) to predict the water-in-oil emulsion effective viscosity within the ESP and its impact on lift capacity. The model can aid the design of the ESP system and can be used as a flow rate estimation tool during ESP production. The initial validation was performed using data from water-in-oil emulsion production of the two wells (A and B) in the Peregrino field, which use two different ESP models. The validation was divided into two parts: the first corresponds to Technology Readiness Level (TRL) 3 and the second, TRL 4, according to ISO 20815:2018. Firstly, well tests data were used to calibrate and test the model for initial validation of the TRL 3. Operation data of these two wells were then used for the initial validation of the TRL 4 in the second part. In the first step, measured and predicted ESP head were compared. Mean Absolute Percentage Errors (MAPE) of 3.6% and 0.8% were obtained for Wells A and B, respectively. In the second part, both the measured and predicted ESP dimensionless head and the measured and predicted differential pressure were compared. From these comparisons, promising results were observed for both wells’ operations. However, more oil data field is necessary to complete the TRL 3 and TRL 4 validation.

Electric Submersible pumps (ESPs) are critical equipment for the artificial lift and transportation of viscous fluids. ESPs also work in severe conditions, including abrasion, gas, and emulsion, requiring continuous monitoring and control. Model Predictive Control (MPC) is a process control approach that can reduce energy consumption, keeping the operation safe within operational limits. This paper presents a non-linear ESP model for MPC based on the Hammerstein-Weiner modeling strategy. The proposed model presents a simplified solution for pumping dynamics considering the changes in fluid viscosity, enabling safe and continuous operation. The experimental arrangement consists of an 8-stage ESP operating with oil. The experiments were in steady-state and transient-state. The experimental procedure consists of varying pump rotation and choke valve aperture. The pump rotation varied from 1000 to 3500 rpm in steps between 0 and 300 rpm. The choke valve opening varied from 19 to 100% of the opening in steps between 0 and 18%. The fluid’s viscosity in the experiments was monitored, varying between 70 to 250 cP. This paper proposes a Non-linear model that considers the viscosity effects of the fluid in the system dynamics. The model formulation involves a non-linear model in which the dynamic state-space model (DSS) inputs are the outputs of a non-linear steady-state model (SSM). The SSM considered the pump and the downstream elements, including the choke valve. The proposed model was fitted and validated using experimental data. A simulation comparing the model and the experimental setup behavior shows errors below 5% between prediction and plant behavior in at least 80% of the dynamical simulation.