Development of a rule-based fault diagnostic advisory system for precut fractionation column
This research presents a Fault Diagnostic Advisory (FDA) System which can be used to detect and diagnose unexpected process deviation in the operation of fatty acid precut fractionation column. The developed algorithm is expected to enhance the safety of operation in oleochemical industry. Early det...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2005
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/4293/1/HengHanYannMFChE2005.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research presents a Fault Diagnostic Advisory (FDA) System which can be used to detect and diagnose unexpected process deviation in the operation of fatty acid precut fractionation column. The developed algorithm is expected to enhance the safety of operation in oleochemical industry. Early detection and diagnosis is useful to avoid abnormal condition that might lead to the loss of both human live and economic values. The advisory system algorithm used process history based method and presented by rule-based approach. It was developed using Borland C++ Builder 6.0 and had a user friendly interface. Plant model was simulated by using commercial simulator, HYSYS.PlantTM and verified with real plant data. Univariate Statistical Process Control technique (Individual and Moving Range (x-MR) chart) and Hazard and Operability (HAZOP) Study were used for the diagnostic task. The system detected and diagnosed process deviations using the saved data and the set limit. Fault occurred if the data value was out of limits. The interface of FDA System then displayed the results in the form of charts. Finally, the causes and consequences of fault were displayed. Although the scheme was developed based on data of fatty acid precut fractionation column, the algorithm of fault detection and diagnosis can be extended to other chemical process by changing the x-MR chart and HAZOP for each selected monitoring variables. |
|---|