Premium
A smart manufacturing methodology for real time chemical process diagnosis using causal link assessment
Author(s) -
Rathinasabapathy Rajan,
Elsass Michael J.,
Josephson John R.,
Davis James F.
Publication year - 2016
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.15403
Subject(s) - robustness (evolution) , granularity , computer science , discretization , risk analysis (engineering) , process (computing) , reusability , industrial engineering , reliability engineering , data mining , engineering , systems engineering , operating system , medicine , mathematical analysis , biochemistry , chemistry , mathematics , software , gene , programming language
Industrial chemical plant diagnosis is the task of analyzing process data to sufficiently pinpoint the causes of abnormal events as fast and as accurately as possible so corrective action can be taken in a timely manner. The need to identify failures explicitly and support human centered decision making becomes pronounced for enterprises. Qualitative diagnostic models offer robustness in capturing diagnostic behaviors when there is little or no data on fault conditions. This article develops, analyzes, and demonstrates a qualitative diagnostic methodology called Causal Link Assessment (CLA). CLA avoids the drawbacks of other methodologies while leveraging several new concepts that include dynamic pattern generation, single time step modeling with multitime step interpretation, and discretized, low granularity dynamic modeling. CLA is demonstrated for an existing ethylene production facility. Model building, robustness, reusability, unaccounted for faults and failures and alignment with emerging Smart Manufacturing infrastructure concepts are discussed. © 2016 American Institute of Chemical Engineers AIChE J , 62: 3420–3431, 2016