Menu
Search
Filters
Close
Menu

96383 CLASSIFICATION AND PREDICTION OF CORROSION PHENOMENA VIA CLUSTER ANALYSIS

Picture for 96383 CLASSIFICATION AND PREDICTION OF CORROSION
Available for download
Product Number: 51300-96383-SG
ISBN: 96383 1996 CP
Author: C.P. Surrock, W.F. Bogaerts
$0.00
$20.00
$20.00
The data that describe corrosion phenomena in terms of individual cases and feature values specific a mapping onto an abstract multi dimensional feature space. Regions of high density, or clusters, significant areas of interest, and possibly, similar corrosion phenomena. This paper describes a method for calculating the similarity between cases based on a generalized measure of Euclidean distance. Evidence for a correlation between the similarity and observed corrosion phenomena is presented for a real-world database on chloride stress corrosion cracking (SCC) of Type 304 stainless steel in water. In contrast to conventional data analysis techniques, the method described can tolerate incomplete data. The performance of the method under different combinations of features was evaluated by calculating the error rate, which is the quotient of the erroneous predictions over the total number of cases examined. The error rate determined by considering incomplete data on pH and oxygen content in addition to complete data on temperature and chloride content was half that determined by considering these complete data alone. Additional data on evaporative service were found to correlate poorly with SCC behavior for the case examined. These results illustrate the importance of feature selection in empirical modelling. Keywords: aqueous corrosion, corrosion diita, empirical modeling, Euclidean distance, nearest neighbor, stainless steel stress corrosion cracking
The data that describe corrosion phenomena in terms of individual cases and feature values specific a mapping onto an abstract multi dimensional feature space. Regions of high density, or clusters, significant areas of interest, and possibly, similar corrosion phenomena. This paper describes a method for calculating the similarity between cases based on a generalized measure of Euclidean distance. Evidence for a correlation between the similarity and observed corrosion phenomena is presented for a real-world database on chloride stress corrosion cracking (SCC) of Type 304 stainless steel in water. In contrast to conventional data analysis techniques, the method described can tolerate incomplete data. The performance of the method under different combinations of features was evaluated by calculating the error rate, which is the quotient of the erroneous predictions over the total number of cases examined. The error rate determined by considering incomplete data on pH and oxygen content in addition to complete data on temperature and chloride content was half that determined by considering these complete data alone. Additional data on evaporative service were found to correlate poorly with SCC behavior for the case examined. These results illustrate the importance of feature selection in empirical modelling. Keywords: aqueous corrosion, corrosion diita, empirical modeling, Euclidean distance, nearest neighbor, stainless steel stress corrosion cracking
PRICE BREAKS - The more you buy, the more you save
Quantity
1+
5+
Price
$20.00
$20.00
Product tags
Also Purchased
Picture for 01013 A REVIEW OF CP CURRENT DEMAND AND
Available for download

01013 A REVIEW OF CP CURRENT DEMAND AND ANODE PERFORMANCE FOR DEEP WATER

Product Number: 51300-01013-SG
ISBN: 01013 2001 CP
Author: K.P. Fischer, B. Espelid, B.Schei
$20.00
Picture for 96382 DEVELOPING A KNOWLEDGE FRAMEWORK FOR THE
Available for download

96382 DEVELOPING A KNOWLEDGE FRAMEWORK FOR THE ORGANIZATION OF AIRCRAFT INSPECTION INFORMATION

Product Number: 51300-96382-SG
ISBN: 96382 1996 CP
Author: P.R. Roberge, L. Grenier
$20.00
Picture for 96389 ELASTOMER-INDUCED CREVICE CORROSION AND
Available for download

96389 ELASTOMER-INDUCED CREVICE CORROSION AND STRESS CORROSION CRACKING OF STAINLESS STEEL HEAT EXCHANGER PLATES IN SOUR AMINE SERVICE

Product Number: 51300-96389-SG
ISBN: 96389 1996 CP
Author: T.A. Moffat, M.G. Hay, J.J. Baron
$20.00
Categories
Industry
Recently viewed
Popular tags
View all

Association for Materials Protection and Performance

© AMPP All Rights Reserved.
Membership Terms of Content Use
Contact Customer Support