The use of neural networks, seismic attributes, and the hydraulic flow unit concept for estimating permeability
- 出版日期：2002.
- 页数：191 p. :
- 第一责任说明：Herutama Trikoranto.
- 分类号：a355
- ISBN：0493765891(ebk.) :
MARC全文
02h0026345 20111125135339.0 cr un||||||||| 111124s2002 xx ||||f|||d||||||||eng | AAI3060914 0493765891(ebk.) : CNY371.35 NGL NGL NGL a355 Trikoranto, Herutama. The use of neural networks, seismic attributes, and the hydraulic flow unit concept for estimating permeability [electronic resource] : a case study / Herutama Trikoranto. 2002. 191 p. : digital, PDF file. Source: Dissertation Abstracts International, Volume: 63-08, Section: B, page: 3890.; ; Chair: Stephen A. Holditch. Thesis (Ph.D.)--Texas A&M University, 2002. Includes bibliographical references. The Hydraulic Flow Unit (HFU) concept is a breakthrough approach for correlating permeability with porosity, which in the past was correlated empirically. The concept is practical, and theoretically correct. However, there are some drawbacks to implement the concept, such as how to determine HFUs from core data, and how to infer the HFU s in other uncored wells more accurately.;In this dissertation, we introduce the application of Artificial Neural Network (ANN) to assign HFUs, which were determined previously using an unbiased least-square-based clustering approach from stressed core data, to uncored sections of the reference well and the surrounding uncored wells.;The implementation of ANN requires good quality openhole wireline log data. Therefore, the log data should be depth-shifted, environmentally corrected, normalized, and properly integrated with core data. A procedure for integrating log data with core data, called the rolling-ball averaging , was used. By using the rolling-ball averaging, the core data are averaged over a larger volume of rock so that the correlation of core data and log data is more compatible.;The ANN also requires that selected input (openhole wireline logs) should have a good tie, not necessarily functional, with target (core data). The stepwise discriminant analysis was introduced for selecting log data for HFU model generation. The selection is fully data-driven and helps the ANN yields better results.;Using ANN-derived HFU and log-derived effective porosity data, permeabilities in uncored sections of the neighboring wells were estimated by HFU basic equation. The results were comparable with permeability values derived by a nonparametric transformation of multivariate regression that involved HFU data as one of the independent variables. The results were acceptable after they were blind tested, and compared to the results of pressure build-up tests.;The advanced application of the HFU concept for estimating field-wide permeability distribution was demonstrated. 3-D seismic data were implemented to develop HFU and porosity maps. The ANN was used to relate HFU and porosity with the seismic amplitudes. The permeability was calculated using basic HFU equation. The results were acceptable after they were validated using average permeability derived from core data and neural net HFU profiles. Rocks ; Porosity Permeability ; Measurement. ; Measurement. Electronic books. aeBook. aCN bNGL http://proquest.calis.edu.cn/umi/detail_usmark.jsp?searchword=pub_number%3DAAI3060914&singlesearch=no&channelid=%CF%B8%C0%C0&record=1 NGL Bs1078 rCNY371.35 ; h1 bs1108