Premium
The use of multiple correspondence analysis and hierarchical clustering to identify incident typologies pertaining to the biofuel industry
Author(s) -
Riviére Carine,
Marlair Guy
Publication year - 2009
Publication title -
biofuels, bioproducts and biorefining
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.931
H-Index - 83
eISSN - 1932-1031
pISSN - 1932-104X
DOI - 10.1002/bbb.187
Subject(s) - diversification (marketing strategy) , typology , production (economics) , multiple correspondence analysis , biofuel , work (physics) , risk analysis (engineering) , cluster analysis , supply chain , business , computer science , operations management , transport engineering , operations research , engineering , geography , marketing , economics , waste management , artificial intelligence , mechanical engineering , archaeology , machine learning , macroeconomics
Biofuel production has been expanding for more than five years, leading to an increasing number of production sites worldwide and also to a tremendous diversification of processes and approaches to producing biofuel. Such a fast move in industry has sometimes proven in the past to potentially lead to underestimating safety management needs. The significant number of existing facilities producing so called first generation biofuel allows for a reasonable survey of safety issues from incidents. In 2006, INERIS initiated research work devoted to the analysis of safety‐related issues including the implementation of an incidents database. Its purpose is to collect known and reasonably well documented incidents (explosions, fires, spills, derailments, and road accidents) that relate to the life cycle of biofuel supply chains. This paper focuses on the analysis of this database, which contains 100 incidents that occurred from January 2000 to early 2009. From the database, an attempt has been made to identify the root factors of incidents potentially impacting biofuel supply chains, using statistical methods like multiple correspondence analysis and ascendant hierarchical clustering. This multivariate analysis exercise has led us to identify five main incident typologies, which in turn allows us to draw appropriate information on safety issues pertaining to first‐generation biofuel supply chains. Each typology is illustrated by actual cases of accidents. Copyright © 2009 Society of Chemical Industry and John Wiley & Sons, Ltd