Neural Network: What It Can Do for Petroleum Engineers
- Authors
- Shahab Mohaghegh (West Virginia U.)
- DOI
- https://doi.org/10.2118/29219-PA
- Document ID
- SPE-29219-PA
- Publisher
- Society of Petroleum Engineers
- Source
- Journal of Petroleum Technology
- Volume
- 47
- Issue
- 01
- Publication Date
- January 1995
- Document Type
- Journal Paper
- Pages
- 42 - 42
- Language
- English
- ISSN
- 0149-2136
- Copyright
- 1995. Society of Petroleum Engineers
- Disciplines
- 4.1.5 Processing Equipment, 5.6.1 Open hole/cased hole log analysis, 6.1.5 Human Resources, Competence and Training, 5.6.4 Drillstem/Well Testing, 5.1.2 Faults and Fracture Characterisation, 4.1.2 Separation and Treating
- Downloads
- 2 in the last 30 days
- 897 since 2007
- Show more detail
- View rights & permissions
Neural network, a nonalgorithmic, nondigital, intensely parallel anddistributive information processing system, is being used more and more everyday. The main interest in neural networks is rooted in the recognition that thehuman brain processes information in a different manner than conventionaldigital computers. Computers are extremely fast and precise at executingsequences of instructions that have been formulated for them. A humaninformation processing system is composed of neurons switching at speeds abouta million times slower than computer gates. Yet, humans are more efficient thancomputers at such computationally complex tasks as speech and other pattern-recognition problems.
Artificial neural systems, or neural networks, are physical cellular systemsthat can acquire, store, and use experiential knowledge. The knowledge is inthe form of stable states or mapping embedded in networks that can be recalledin response to the presentation of cues. In a typical neural data processingprocedure, the database is divided into two separate portions called trainingand test sets. The training set is used to develop the desired network. In thisprocess (depending on the paradigm that is being used), the desired output inthe training set is used to help the network learn by adjusting the weightsbetween its neurons or processing elements.
Once the network has learned the information in the training set and has"converged," the test set is applied to the network for verification.It is important to note that although the user has the desired output of thetest set, it has not been seen by the network. This ensures the integrity androbustness of the trained network.
A handful of articles on the use of neural networks in the petroleumindustry has appeared in SPE conferences, proceedings, and publications in thepast 2 years. These articles can be divided into two categories: those that useneural networks to analyze formation lithology from well logs and those thatuse neural networks to pick a reservoir model to be used in conventional welltest interpretation studies. These tasks are usually done by log analysts andreservoir engineers, and their automation using a fault-tolerant process mayprove valuable.
Neural networks can help engineers and researchers by addressing somefundamental petroleum engineering problems as well as specific ones thatconventional computing has been unable to solve. Petroleum engineers maybenefit from neural networks on occasions when engineering data for design andinterpretations are less than adequate. This is an especially common occurrencein the Appalachian basin, where some fields are quite old. Lack of adequateengineering data may also be encountered because of the high cost of coring,well testing, and so on.
Neural networks have shown great potential for generating accurate analysisand results from large amounts of historical data that otherwise would seem notto be useful or relevant in the analysis. An example of such a problem wasencountered by a gas company for a gas storage field in Ohio. In the absence ofappropriate data, which normally would make engineering design and evaluationof the fracturing jobs virtually impossible, a carefully designed neuralnetwork was able to predict the performance of fracturing jobs with greataccuracy. A linear plot of the actual fracturing job results (data never seenby the network during training) and network predictions resulted in acorrelation coefficient of 0.98, where 1.00 is a perfect match.
Neural networks have proved to be valuable pattern-recognition tools. Theyare capable of finding highly complex patterns within large amounts of data. Arelevant example is well log interpretation. It is generally accepted thatthere is more information embedded in well logs than meets the eye.Determination, prediction, or estimation of formation permeability withoutactual laboratory measurement of the cores or interruption in production forwell test data collection has been a fundamental problem for petroleumengineers. From geophysical well log data, it was possible to predict and/orestimate permeability of a highly heterogeneous formation in West Virginia, asshown in Fig 1.
Reference
1. Mohaghegh, S., Arefi, R., and Ameri, S.: "Design and Development ofan Artificial Neural Network for Prediction of Formation Permeability,"paper SPE 28237 presented at the 1994 SPE Petroleum Computer Conference, July31-Aug. 3, Dallas.
File Size | 82 KB | Number of Pages | 1 |