|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Cai, Hanpeng (Center for information geoscience & School of Resources and Environment, UESTC) | Qin, Qing (Center for information geoscience & School of Resources and Environment, UESTC) | Li, Huiqiang (Center for information geoscience & School of Resources and Environment, UESTC) | Wang, Yaojun (Center for information geoscience & School of Resources and Environment, UESTC) | Zhang, Yuejing (Research Institute of Petroleum Exploration and Development, Sinopec Shengli oilfield Company) | Wang, Qianjun (Research Institute of Petroleum Exploration and Development, Sinopec Shengli oilfield Company) | Wang, Jinduo (Research Institute of Petroleum Exploration and Development, Sinopec Shengli oilfield Company)
ABSTRACT The existing AVO inversion method based on Bayesian theory only considers the prior distribution of noise in seismic data, but does not take into account the prior distribution of reflection coefficient. In view of this problem, the paper studies the statistical distribution characteristics of seismic reflection coefficient, and proposes the concept and idea of AVO inversion that considering the statistical distribution of reflection coefficient. Moreover, we construct the AVO inversion objective function containing the statistical distribution that constraints of reflection coefficient under the framework of AVO inversion based on Bayesian theory. Considering the difference in the prior distribution of stratigraphic reflection coefficient in different regions, we choose the generalized extremum distribution(GEV) with parameter adaptive adjustment features to describe the statistical distribution of reflection coefficient. Aiming at the high nonlinearity of GEV, the objective function in this paper is solved by an improved particle swarm optimization (PSO) algorithm. The experimental analysis results not only verify the effectiveness and reliability of the proposed method, but also confirm the importance and criticality of constraints of the reflection coefficient distribution. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 9:20 AM Presentation Time: 10:35 AM Location: Poster Station 7 Presentation Type: Poster
Wang, Xiaokai (Chinese Academy of Sciences and the University of Texas–Austin) | Yang, Changchun (Chinese Academy of Sciences) | Li, Xueliang (Chinese Academy of Sciences) | Chen, Wenchao (Xi'an Jiaotong University) | Zhao, Haixia (Xi'an Jiaotong University)
The producing oil-field need the second/third time seismic data acquisition to evaluate the remaining production of crude oil. In the repeated data acquisition, the non-random well-pump noise affects the quality of seismic data. In order to know the characteristics of well-pump noise, we design a small single-point receiver array, and use this array to receive sufficient time duration of the typical wellpump noise. After analyzing the obtained well-pump noise, we find it has some characteristic differences with reflections, such as limited bandwidth, locating in lowfrequency area, low velocity, dispersion and bad spatial correlation. Taking into account these differences, we propose an iterative method for well-pump noise attenuation in the time-frequency domain. By applying the proposed method to synthetic signal and real field data, we demonstrate the superior performance of our method.
Presentation Date: Wednesday, September 27, 2017
Start Time: 11:25 AM
Presentation Type: ORAL
Beam forming is widely used in seismic exploration, such as denoise and beam based migration. A high-resolution beam is essential for those processing techniques. But beam data extracted by traditional methods suffers from noise leakage and low resolution. In this work, a CLEAN inversion scheme is described. This method suppresses the leaking noise and enhances the beam resolution through the exploitation of the second order statistics. To improve the performance of CLEAN method, a post-processing block is proposed to work with regular CLEAN inversion. The clustering and locating block further optimizes the beams generated by CLEAN inversion. Consequently, the proposed method generates high-resolution beam. We use synthetic data and real data to prove its effectiveness and robustness.
Beam forming is widely used in Radar technology to detect the angle of arrival (Capon, 1969). It has been introduced in the seismic data processing to suppress noise (Ozbek, 2000). The high-resolution beams can result in efficient migration procedure and yield a high S/N ratio imaging result. Traditionally, the beam is generated by local tau-p transform. But, it suffers from low resolution and the leaking noise (Stoffa, 1981). The semblance can be used to reduce the leaking noise, but the threshold is difficult to choose. In order to enhance quality of beam data, the inversion-based beam forming method can be used.
CLEAN technique is effective in radar target imaging (Högbom, 1974; Bose, 2011). It aims at reducing the spurious sidelobes and hence enhances the resolution in beam forming. CLEAN assumes that the maximum of the dirty beam corresponds to a source point. The effect of this source is extracted in an iterative way that minimizes the residual energy and eliminates the spurious peaks simultaneously.
In this paper, we adopt CLEAN inversion to suppress the leaking noise and enhance the beam resolution. In the CLEAN inversion, second order statistics of the spatially local data is utilized to automatically determine the beam components and suppress the leaking noise in an iterative way. In order to further enhance the beam resolution, a post-processing block is also proposed to work with CLEAN inversion, including clustering and locating. The block further optimizes beam components generated by the regular CLEAN inversion, which significantly enhances the beam resolution.
Growing numbers of National Oil Companies (NOCs) have experienced great changes since the beginning of the 21 st century which have transformed them into International Energy Companies. The challenges of internationalization have led NOCs of China to pay more attention to building technology innovation systems, which promoted indigenous innovation capacity and technological competitiveness in the global market. This paper is based on the study and analysis of Chinese NOCs' technology innovation systems, and emphasizes the building of these systems in terms of how they have facilitated the companies' transition. A detailed review based on case studies in China indicates that organization, facilities and mechanisms are the key elements and emphasis in the building of Chinese NOCs' technology innovation system. NOCs of China restructured their technology innovation organization structures, optimizing the allocation of innovation resources from R&D to commercialization in both upstream and downstream operations. Facilities are the foundation of technology innovation activity, and in this respect NOCs of China invested in building or upgrading labs/pilots to support their technology innovation activities and promote the development of science and technology in the national petroleum industry. Mechanisms such as stable investment, technical talent and commercial deployment policies incentivized the application of new technology and accelerated open innovation. This paper describes the practices of three NOCs of China in constructing technology innovation systems and the significant achievements they have made in enhancing their indigenous innovative capacity and strengthening international cooperation, contributing to Chinese NOCs' transition from technology followers to technology leaders.
Joint inversion of PP and PS reflection data has been hindered by the difficult task of registration or correlation of PP and PS events. It can perhaps be achieved by registering the events during inversion but the resulting algorithm is generally computationally intensive. In this paper, we propose a stochastic inversion of PP and PS data which have been registered to the same PP time scale using a new interval velocity analysis technique. The prestack PP and PS wave joint stochastic inversion is achieved by using the PP and PS wave angle gathers using a very fast simulated annealing (VFSA) algorithm. The objective function attempts to match both PP and PS data; the starting models are drawn from fractional Gaussian distribution constructed from interpolated well logs. The proposed method has been applied to synthetic and real data; the inverted results from synthetic data inversion compare very well with model data, and inverted results for real data inversion are consistent with seismic data and log data. These also show that the proposed method has a higher accuracy for estimating rock physics parameters while it circumvents the horizon registration problem in the data interpretation. We also estimate uncertainty in our estimated results from multiple VFSA derived models.
Abstract Data sharing is an integral component of strategies to support critical decisions during the life of field value chain. Petroleum organizations face major challenges in the integration of legacy database systems to support an effective methodology for maximizing asset value. Case histories are presented that show effective integration of data, information, and knowledge in the asset value chain. Introduction In 1997, a "Delphi Survey" was conducted to determine what the shape of exploration and production (E&P) data management would be in the next three years. In that survey, a geophysicist at what was then Conoco, Inc. in Lafayette, Louisiana, in the United States, was asked how he imagined data management would be in the year 2000. His response was that he wanted to "draw a rectangle on a map" and have access to everything that pertained to his project. Like many predictions, the vision has remained consistent, but the strategies for accomplishing it have matured. Information management solution providers now recognize a spectrum of approaches to accessing, integrating, and sharing all the data pertaining to an asset. Methods - Consolidation Strategies At the time of the survey, the most common strategy could be identified as "consolidation", in which all the data relevant to an asset was collected, validated, formatted, and loaded into a single relational database management system (RDMS) with interfaces to query and report on the data in a map view for integrated asset management. The sophistication of the mapping interfaces gradually improved, but the strategy still depended on the use of a single data store (Figure 1). This strategy emphasizes the value for the organization has a whole, and is the most precise, but carries a high cost in both time and resources for successful implementation. If information is available from a single centralized location it makes logical groupings of data readily available for a wide range of applications, and can simplify later tasks of data management, such as entitlements to distribute data types to multiple data owners. Having a single master location also eliminates the difficulties of data duplication. The disadvantage with this approach is the cost resulting from the effort required for the initial implementation; the complexity of transforming the data, and the time involved in carrying out the necessary data quality checks. In addition, some categories of information cannot be fit into a single repository in an effective way and data may be held in a form that does not readily meet the needs of applications. This method also ignores the value of essential related or contextual information. An example of a consolidation project comes from the Kuwait Oil Company (KOC). In line with many other Arab oil producing states, Kuwait began negotiations in the early 1970s to restore control over its own natural oil resources. By mutual agreements with the Company's two original partners, the State's shareholding in KOC was increased progressively until full control was achieved. On March 5th, 1975, the State of Kuwait signed an agreement with two oil companies (British Petroleum and Gulf) giving Kuwait complete control of its oil resources.
Wang, L. (Geo Visual Systems Australia Pty Ltd) | Tyson, S. (Geo Visual Systems Australia Pty Ltd) | Song, X. (RIPED of PetroChina) | Cao, H. (RIPED of PetroChina) | Wong, P.M. (Univ. of New South Wales)
Abstract This paper demonstrates a case study of a hybrid methodology based on the combination of radial basis function neural network and sequential Gaussian simulation. The methodology is demonstrated with an application to modelling the porosity distribution in an oil reservoir of the Lower Tertiary in the north of Dongying depression, Shengli Oilfield, East China. The methodology first uses radial basis function neural networks to estimate the porosity trends (fluvial directions) from high-dimensional data system with both well and seismic data. Gaussian simulation helps to do the local uncertainty analysis for the reservoir model. The final results from the hybrid methodology assure our confidence on the reservoir model both horizontally and vertically. They are realistic and honour the geological rules of the oilfield. The technique is fast and straightforward, and provides an effective computational framework for conditional simulation. Introduction Spatial descriptions of reservoir properties such as porosity and permeability are key components for performance evaluation and field development planning. When wells are sparse and limited, the statistics of the well data become unrepresentative. It is critical to integrate well data with all the available soft data to construct reliable reservoir models. In this paper, we will make use of the extensive seismic attributes and seismically interpreted results. Although some purely geostatistical techniques are capable of providing some of these functionalities, cross-correlation modelling is often difficult and tedious in practice. Also the information such as two-point statistics and linear relationships extracted from conventional reservoir data (well and seismic) may not be sufficient for describling the complexities of the reservoir model. Wang et al., Wong et al. and Caers and Journel show that the neural network approach is a promising tool to deal with non-linear and multidimensional data system, hence to achieve the reliable and realistic approximation showing also the basic geological rules for reservoir model. On the other hand, Gaussian simulation provides an opportunity to do local uncertainty analysis. This paper will employ a hybrid methodology, called neural network residual simulation (NNRS), to construct a porosity model for an oil reservoir of the Lower Tertiary in the north of Dongying depression, Shengli Oilfield, East China. The objective of this paper is to provide a case study of NNRS methodology. The reservoir in the case area is characterised by a fluvial channel with well data, seismic attributes and seismically interpreted results. The next section will first present a description of NNRS methodology, followed by the demonstrations of the case study. Methodology The integrated technique used in this paper is developed based on a combined use of neural networks and geostatistics. The technique assumes that any spatial prediction is composed of a predictable (trend) component and an error (noise or residual) component. Firstly neural networks (inexact estimators) are used to model the former component and residual kriging and simulation (exact estimators) to model the latter component. Hence the name "neural network residual kriging" (NNRK) and "neural network residual simulation" (NNRS) are used. The final estimate is simply the sum of the two components, and hence the estimator honours all the conditioning data. The methodology can be briefly described as four steps:use of radial basis function neural networks to model the regional trends by integrating all the suitable seismic attributes and well data in the low resolution level; the residuals at the wells are calculated in the high resolution level; Gaussian simulation is used to generate multiple realisations for the residuals; and adding the results of 2) and 3) together as the final predictions.
Byun, Joongmoo (SensorWise Inc.) | Hatch, Rena M. (TomoSeis division of Core Laboratories) | Yu, Gang (TomoSeis division of Core Laboratories) | Zhang, Yong (Sinopec) | Zhou, Jian Yu (ShengLi Oilfield) | Xu, Yi Wei (JiangHan Oilfield)