A comparative approach in the definition of minerals resource outline

E. Tawo, F. Boukadi, A. Al-Bemani, S. Al-Alawi

Research output: Contribution to journalReview article

Abstract

The paper compares the performance of Geostatistical, Artificial Neural Network, and Geoquest Reservoir Technology (GRT) Grid approaches in defining either an oil reservoir or a mineralized outline. The objective is to find the most suitable technique that would reduce exploratory drilling cost without sacrificing on accuracy. Each technique will be applied to the same complete data set, and subsequently to a reduced data set. Their performances will then be assessed and compared to each other and to known values at sampled locations. While a geostatistical technique could effectively be applied in the optimization of a drilling pattern and a sampling program, its use is limited at the early stage of exploration when data is scarce. However, its kriging variance provides a measure of the quality of estimation that gives it an edge over the other two techniques whose errors of estimation are not measured. On the other hand, ANN estimates are found to be very close to the actual values, this is largely due to its emulating capabilities. The absence of data at the boundaries affects on the distribution of the training patterns and in turn might lead to biased estimates. Although GRT Grid does not associate an error to its estimates, its overall performance is very similar to that of Geostatistics. By virtue of this similarity, an inferred estimation error from the geostatistical technique could be extended to GRT Grid estimates. A general drawback in all three approaches is possibly the lack of knowledge of the exact volume of data needed at an early stage of exploration that will ensure an effective estimate that reduces the drilling cost [1]. However, a re-estimation method gives to a large extent, an insight into a preferred approach amongst the three techniques analyzed in this paper. The ease of use, level of discretization, number of iterations for convergence, a good appreciation of the intrinsic function and their limitations together with an appreciation of the geology contribute to a choice of a technique.

Original languageEnglish
Pages (from-to)3-16
Number of pages14
JournalArabian Journal for Science and Engineering
Volume23
Issue number1A
Publication statusPublished - Jan 1998

Fingerprint

Minerals
Technology
Geology
Oil and Gas Fields
Costs and Cost Analysis
Spatial Analysis
Datasets

ASJC Scopus subject areas

  • General

Cite this

A comparative approach in the definition of minerals resource outline. / Tawo, E.; Boukadi, F.; Al-Bemani, A.; Al-Alawi, S.

In: Arabian Journal for Science and Engineering, Vol. 23, No. 1A, 01.1998, p. 3-16.

Research output: Contribution to journalReview article

@article{ca19b97195b84dbfa9db0a4e068b84b1,
title = "A comparative approach in the definition of minerals resource outline",
abstract = "The paper compares the performance of Geostatistical, Artificial Neural Network, and Geoquest Reservoir Technology (GRT) Grid approaches in defining either an oil reservoir or a mineralized outline. The objective is to find the most suitable technique that would reduce exploratory drilling cost without sacrificing on accuracy. Each technique will be applied to the same complete data set, and subsequently to a reduced data set. Their performances will then be assessed and compared to each other and to known values at sampled locations. While a geostatistical technique could effectively be applied in the optimization of a drilling pattern and a sampling program, its use is limited at the early stage of exploration when data is scarce. However, its kriging variance provides a measure of the quality of estimation that gives it an edge over the other two techniques whose errors of estimation are not measured. On the other hand, ANN estimates are found to be very close to the actual values, this is largely due to its emulating capabilities. The absence of data at the boundaries affects on the distribution of the training patterns and in turn might lead to biased estimates. Although GRT Grid does not associate an error to its estimates, its overall performance is very similar to that of Geostatistics. By virtue of this similarity, an inferred estimation error from the geostatistical technique could be extended to GRT Grid estimates. A general drawback in all three approaches is possibly the lack of knowledge of the exact volume of data needed at an early stage of exploration that will ensure an effective estimate that reduces the drilling cost [1]. However, a re-estimation method gives to a large extent, an insight into a preferred approach amongst the three techniques analyzed in this paper. The ease of use, level of discretization, number of iterations for convergence, a good appreciation of the intrinsic function and their limitations together with an appreciation of the geology contribute to a choice of a technique.",
author = "E. Tawo and F. Boukadi and A. Al-Bemani and S. Al-Alawi",
year = "1998",
month = "1",
language = "English",
volume = "23",
pages = "3--16",
journal = "Arabian Journal for Science and Engineering",
issn = "0377-9211",
publisher = "King Fahd University of Petroleum and Minerals",
number = "1A",

}

TY - JOUR

T1 - A comparative approach in the definition of minerals resource outline

AU - Tawo, E.

AU - Boukadi, F.

AU - Al-Bemani, A.

AU - Al-Alawi, S.

PY - 1998/1

Y1 - 1998/1

N2 - The paper compares the performance of Geostatistical, Artificial Neural Network, and Geoquest Reservoir Technology (GRT) Grid approaches in defining either an oil reservoir or a mineralized outline. The objective is to find the most suitable technique that would reduce exploratory drilling cost without sacrificing on accuracy. Each technique will be applied to the same complete data set, and subsequently to a reduced data set. Their performances will then be assessed and compared to each other and to known values at sampled locations. While a geostatistical technique could effectively be applied in the optimization of a drilling pattern and a sampling program, its use is limited at the early stage of exploration when data is scarce. However, its kriging variance provides a measure of the quality of estimation that gives it an edge over the other two techniques whose errors of estimation are not measured. On the other hand, ANN estimates are found to be very close to the actual values, this is largely due to its emulating capabilities. The absence of data at the boundaries affects on the distribution of the training patterns and in turn might lead to biased estimates. Although GRT Grid does not associate an error to its estimates, its overall performance is very similar to that of Geostatistics. By virtue of this similarity, an inferred estimation error from the geostatistical technique could be extended to GRT Grid estimates. A general drawback in all three approaches is possibly the lack of knowledge of the exact volume of data needed at an early stage of exploration that will ensure an effective estimate that reduces the drilling cost [1]. However, a re-estimation method gives to a large extent, an insight into a preferred approach amongst the three techniques analyzed in this paper. The ease of use, level of discretization, number of iterations for convergence, a good appreciation of the intrinsic function and their limitations together with an appreciation of the geology contribute to a choice of a technique.

AB - The paper compares the performance of Geostatistical, Artificial Neural Network, and Geoquest Reservoir Technology (GRT) Grid approaches in defining either an oil reservoir or a mineralized outline. The objective is to find the most suitable technique that would reduce exploratory drilling cost without sacrificing on accuracy. Each technique will be applied to the same complete data set, and subsequently to a reduced data set. Their performances will then be assessed and compared to each other and to known values at sampled locations. While a geostatistical technique could effectively be applied in the optimization of a drilling pattern and a sampling program, its use is limited at the early stage of exploration when data is scarce. However, its kriging variance provides a measure of the quality of estimation that gives it an edge over the other two techniques whose errors of estimation are not measured. On the other hand, ANN estimates are found to be very close to the actual values, this is largely due to its emulating capabilities. The absence of data at the boundaries affects on the distribution of the training patterns and in turn might lead to biased estimates. Although GRT Grid does not associate an error to its estimates, its overall performance is very similar to that of Geostatistics. By virtue of this similarity, an inferred estimation error from the geostatistical technique could be extended to GRT Grid estimates. A general drawback in all three approaches is possibly the lack of knowledge of the exact volume of data needed at an early stage of exploration that will ensure an effective estimate that reduces the drilling cost [1]. However, a re-estimation method gives to a large extent, an insight into a preferred approach amongst the three techniques analyzed in this paper. The ease of use, level of discretization, number of iterations for convergence, a good appreciation of the intrinsic function and their limitations together with an appreciation of the geology contribute to a choice of a technique.

UR - http://www.scopus.com/inward/record.url?scp=0032369131&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032369131&partnerID=8YFLogxK

M3 - Review article

VL - 23

SP - 3

EP - 16

JO - Arabian Journal for Science and Engineering

JF - Arabian Journal for Science and Engineering

SN - 0377-9211

IS - 1A

ER -