Produção Científica



Material Didático
06/02/2023

MANUAL DE MODELAGEM DE DADOS DE GPR NO DOMÍNIO DO TEMPO (GPRMAX) APLICANDO EFEITOS DA DISPERSÃO ATRAVÉS DE UMA ABORDAGEM LINEAR PARA SIMULAR UM MODELO DE Q CONSTANTE USANDO VÁRIOS POLOS DEBYE.

Artigo em Revista
10/10/2022

Viscoelastic and viscoacoustic modelling using the Lie product formula
We present efficient and accurate modelling of seismic wavefields in anelastic media.We use a first-order viscoelastic wave equation based on the generalized Robertsson’sformulation. In our work, viscoacoustic and viscoelastic wave equations use the stan-dard linear solid mechanism. To numerically solve the first-order wave equation, weemployed a scheme derived from the Lie product formula, where the time evolutionoperator of the analytic solution is written as a product of exponential matrices, andeach exponential matrix term is approximated by the Taylor series expansion. Theaccuracy of the proposed scheme is evaluated by comparison with the analytical so-lution for a homogeneous medium. We also present simulations of some geologicalmodels with different structural complexities, whose results confirm the accuracy ofthe proposed scheme and illustrate the attenuation effect on the seismic energy duringits propagation in the medium. Our results demonstrate that the numerical schemeemployed can be used to extrapolate wavefields stably for even larger time steps thanthose usually used by traditional finite-difference schemes.

Artigo em Revista
10/10/2022

Multiparameter vector-acoustic least-squares reverse time migration
The use of dual-sensor acquisitions enabled different studies using the so-calledvector-acoustic equations, which admit particle velocity (displacement or accel-eration) information instead of solely the pressure wavefield. With cost far fromelastic formulations but comparable with the usually used second-order acousticequation, previous works involving the use of vector-acoustic equations along withmulticomponent data have been applied to conventional reverse time migration andfull-waveform inversion, always emphasizing the benefits of using wavefields con-taining directivity information, which make the receiver ghosts interact constructivelywith the backpropagated reflected wavefield. Thus, generated results are superior tothose of conventional single-component data imaging techniques, particularly withspatial subsampling of marine seismic data. To assess whether the effects of applyingthe vector-acoustic equations persist in a linearized inversion, we developed a multi-parameter vector-acoustic least-squares reverse time migration, inverting reflectivitiesassociated with velocity and density. To demonstrate the method’s performance, weapply it to two-dimensional numerical examples and compare the results with thoseobtained by the conventional acoustic least-squares reverse time migration. Theresults obtained by the vector-acoustic least-squares reverse time migration methodare accurate for all inverted parameters and also deliver better convergence whencompared with the conventional least-squares reverse time migration.

Dissertação de Mestrado
10/10/2022

Avaliação do modelamento da propagação de ondas sísmicas pelo método lowrank
O modelamento da propagação da onda Ă© uma parte fundamental para o imageamento sĂ­smico, pelo qual se requer mĂ©todos precisos para representar o campo de onda. O mĂ©todo lowrank Ă© aplicado Ă  matriz propagadora determinada pela extrapolação da solução da equação da onda. Nesse trabalho, Ă© avaliado o desempenho absoluto e relativo do mĂ©todo lowrank em meios homogĂȘneos nos quais tem-se expressĂ”es teĂłricas, assim como em meios heterogĂȘneos. Realizamos vĂĄrios testes em modelos de meios homogĂȘneos com diferentes discretizaçÔes espacias e temporais, e a resposta do modelamento foi comparada com a solução analĂ­tica, mostrando boa acurĂĄcia cinematicamente, enquanto o desempenho na dinĂąmica foi afetado pelo intervalo de amostragem temporal, embora para cada modelo foi possĂ­vel determinar uma discretização temporal no qual conseguiu-se um erro dinĂąmico menor que 1% para modelar a onda em meios homogĂȘneos. Para determinar o desempenho relativo foi considerado o mĂ©todo de diferenças finitas tradicional de segunda ordem em tempo e espaço. Os coeficientes de reflexĂŁo aproximados pelo mĂ©todo lowrank considerando um modelo com um refletor plano foram comparados com os coeficientes de reflexĂŁo de ondas planas para os quais tem-se expressĂ”es teĂłricas. Neste caso, observa-se que com o mĂ©todo lowrank obtĂ©m-se os coeficientes com erro menor que 1% quando comparada ao coeficiente teĂłrico, mostrando melhor resultado do que obtido por diferenças finitas. O tempo computacional dos mĂ©todos foi avaliado, sendo que lowrank presenta um crescimento linear do tempo com o tamanho do modelo, isto sugere que para modelos de grande porte, lowrank economize o tempo de modelamento em comparação com diferenças finitas.

Artigo em Revista
18/07/2022

Characterizing seismic facies in a carbonate reservoir, using machine learning offshore Brazil.
The authors propose an approach that uses machine learning to characterize carbonate facies in a wildcat (Gato do Mato) prospect in the Santos basin, offshore Brazil. We analyzed different seismic attributes and selected those that best responded to the seismic patterns identified in the study area as input for an unsupervised classification.
The classification method used is the self-growing neural network (SGNN) technique that consists of the following steps:

1. Seismic pattern identification in seismic amplitude. The main patterns identified are build-up, debris, carbonate platform and bottom lake facies.
2. Generation and analysis of seismic attributes to characterize seismic patterns. We chose Eigen coherence, dip-steered enhancement, relief and relative acoustic impedance to help in seismic characterization.
3. We performed principal component analysis (PCA) of the attributes: amplitude filtered from dip-steered enhancement (seismic-driven structural filtering), Eigen coherence and relief.
4. Unsupervised seismic classification from the PCA of seismic attributes (item 3 above).

Using this approach, we associated the classified seismic facies with the patterns identified in the amplitude data. The seismic facies allowed us to differentiate the carbonate platforms from the build-up facies. However, the classification encountered difficulties in identifying the patterns associated with lake bottom facies and the chaotic seismic pattern of debris facies.

Artigo em Revista
18/07/2022

Probabilistic Estimation of Seismically Thin-Layer Thicknesses with Application to Evaporite Formations
The identification of potassium (K) and magnesium (Mg) salts prior to the well drilling is a key factor to avoid washouts, closing pipes, fluid loss damage, and borehole collapse. The Bayesian classification combines the outcomes from statistical rock physics and seismic inversion, providing the spatial occurrence of the most-probable salt types. It serves as a facies identifier of Mg–K-rich salts (bittern salts) before drilling. Nevertheless, the most-probable classification is limited to the seismic resolution which may underestimate seismically thin-layer thicknesses. Along with the most-probable facies, the Bayesian classification renders the facies probability volume. We demonstrate that the facies probability and facies-specific total thickness highly correlate to each other even under the threshold of seismic resolution. Thus, we employ the bittern-salts probability volume to predict thin-bed bittern-salts thickness in undrilled locations. To capture the variability of the seismic estimation, we resort to Monte Carlo-assisted simulations of wells that emulate the layering patterns of a site-specific deposition environment. These simulations are crucial to assist the estimation of the joint probability density function between the facies volume and the total thickness. Therefore, given the facies probability, the joint probability density function enables us to derive the conditional expectation and percentiles of thin-bed thicknesses. Furthermore, this paper proposes a method to quantify the negative influence of seismic noise in the estimation of thin-bed thicknesses. The blind well confirms the consistency of this technique to unfold the uncertainty in the seismic predictability of thin layers. We argue that this procedure is extendable to other facies.

Artigo em Revista
18/07/2022

Mineralogy based classification of carbonate rocks using elastic parameters: A case study from Buzios Field
Brazilian presalt carbonate reservoirs are highly heterogeneous. This feature is mostly justified by the nature of the original depositional system and subsequently diagenetic processes. Consequently, reserve estimates and production forecasting are under large uncertainties. In this geologic context, it is of great relevance to develop techniques that helps to obtain a detailed description on the spatial distribution of these different rocks. In doing so, it contributes to the understanding of presalt carbonate sedimentary deposits, providing inputs for more predictive reservoir models. Traditionally, these carbonates are grouped into three classes, from which only one exhibits reservoir properties. Using a dataset from Buzios Field, this work proposes a characterization of presalt carbonate reservoir rocks by grouping them in terms of their mineral composition. Taking advantage of rock physics concepts, we aim to potentialize the use of elastic parameters for multiple rock type discrimination. We explored several attempts for rock classification by using a Bayesian approach. Among all the tested propositions, a two-step workflow for five lithotypes classification, emerges as the most appropriate for the Buzios Field. In this scheme, three lithotypes represent good-quality reservoirs and the other two are low-porosity and Mg-clay-rich carbonates. The average root-mean-square error of the most likely a posteriori rock proportions is around 8.4%, only approximately 1% higher than the conventional three lithotypes configuration. To support that, we compared different methodologies for Bayesian classification at well-log scale through acoustic impedance and compressional to shear velocity ratio. Potential applicability of the proposed methodology at field scale is reinforced by similar results achieved using well-logs filtered to the seismic bandwidth.

Artigo em Revista
18/07/2022

The influence of subseismic-scale fracture interconnectivity on fluid flow in fracture corridors of the BrejÔes carbonate karst system, Brazil
The present study used a multitool approach to characterize fractures of several orders of magnitude in large fracture corridors, caves, and canyons to investigate their impact on fluid flow in carbonate units. The study area is the BrejĂ”es carbonate karst system that is located in the Neoproterozoic Salitre Formation in the IrecĂȘ Basin, SĂŁo Francisco Craton, Brazil. The approach included satellite imagery, used for interpreting the regional structural context, Unmanned Aerial Vehicle (UAV) and ground-based Light Detection And Ranging (LiDAR) imagery, used for detailed structural interpretation. Regional interpretation revealed that fracture corridors, caves and canyons occur along a N–S-oriented anticline hinge. An advanced stage of karstification caused fracture enlargement and intrabed dissolution, and the formation of caves and canyons. A river captured by the highly fractured zone along the anticline hinge played an important role as an erosive agent. Detailed characterization of fracture corridors comprised structural analysis, topological studies, persistence estimations, power-law fitting of fracture trace length distributions, and identification of network backbones. Our results indicate that fracture corridors comprise four subvertical fracture sets: N–S and E-W and a conjugate pair, NNE-SSW and NW-SE. Fractures observed in the caves show the same dominant directions. Fracture directions are consistent with a common origin associated with the anticline folding. Fracture traces range from 1.0 m to 300 m, comprising both subseismic (<50 m) and seismic scale fractures (>50 m). Networks have dominance of node terminations Y and X (notably Y), CB values higher than 1.8, high P20 and P21 persistence values, and highly interconnected backbones. Fracture network connectivity is associated with power-law exponents greater than 2.5 for the fracture trace distributions, indicating large influence of subseismic-scale fractures on fluid flow. As the final result of folding and karstification, large volumes of secondary macroporosity were created, particularly in the zone of maximum fracture intensity around the hinge zone of the anticline. This scenario can be used to understand better oil reservoirs formed in similar structural controls in near-surface conditions.

Artigo em Revista
18/07/2022

A constrained version for the stereology inverse problem: Honoring power law and persistences of the fracture traces exposed on arbitrary surfaces
We present a stereological study in a cave setting that is part of a karstic carbonate system located in the SĂŁo Francisco Craton, Brazil. Using a Lagrangian approach, a constrained version of the nonlinear inverse problem of stereology is solved. Besides the classical demand of fitting the histogram of fracture traces measured on arbitrary exposed surface, it is imposed that the solution honors also measures of surface intensity (p21) and power law exponent obtained from fracture traces on the same exposed surfaces. Estimates of volumetric intensities (p32) of conjugate fracture pairs might be also imposed to be close values. The resulting cost functional is minimized using the Particle Swarm Optimization (PSO) method. The implemented version of PSO furnishes the best solution and a set of suboptimal quasisolutions, from which the solution uncertainty is evaluated. A key aspect of the implemented approach is that all terms composing the cost functional are normalized, obtaining as a result, robustness for the weighting parameters to changes in the input data. The resulting stochastically simulated discrete fracture networks honor all statistical observations but, in general, do not reproduce the positions of the observed fracture traces. The methodology is applied to synthetic and field data examples. All obtained solutions are stable and geologically reliable.

Artigo em Revista
18/07/2022

Well log analysis for lithology and fluid contacts in Rovuma Basin – Mozambique: Application of cluster and discriminant analyses
This study applies the cluster and discriminant analyses in geophysical well log data from the Rovuma sedimentary Basin - Mozambique. The main objective was to determine the lithological profile and fluid contacts in reservoirs. Well log data from five wells drilled on the same basin were used. For the discrimination, a reference well was chosen for training, and the obtained functions from it were then applied to the remaining wells. The classification process comprehended three main phases, namely, the separation of shale/non-shale layers along the entire logged section, separation of water/hydrocarbon within reservoirs and the separation of oil/gas within hydrocarbon bearing zones. The two methods, cluster analysis and discriminant analysis, were applied in parallel and the results are compared in each classification phase. The quality of reservoirs was also assessed by applying cutoffs in relation to shale content and effective porosity, delineating net reservoirs. In general, both methods converged to the same lithological model and fluidtypes in reservoirs. Gas has been indicated as the most predominant hydrocarbon in the basin.

<<  <   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65  66  67   >  >>