(421g) Real-Time Monitoring, Fault Detection and Diagnosis for CNG Recharging Stations
- Conference: AIChE Annual Meeting
- Year: 2010
- Proceeding: 2010 Annual Meeting
- Group: Computing and Systems Technology Division
- Time: Wednesday, November 10, 2010 - 10:30am-10:50am
Korea is the second largest importer of LNG in the world [BP 2009]. LNG is gasified and distributed through the country-wide gas pipe network, which is more than 2,000 km long, and some part of it is compressed and charged into CNG buses at various CNG stations. As a cheaper and environmentally friendly alternative, Compressed Natural Gas (CNG) is being promoted as alternative fuel, especially for public transportation, and the shift to CNG is becoming popular. This research reports part of a continuing national project to improve the safety of the national energy infrastructure and its operations: Our reported research is mainly focused on facilities related to gas as energy carrier.
A CNG recharging station, usually located at the vicinity of the metropolitan city boundary, is the main focus of this research: It consists of compressors, dispensers and compressed gas storages. The natural gas from pipeline is compressed and charged into buses through unsteady-state operations. It uses the stored compressed gases as well as real-time compression of natural gas supplied through the pipeline. The compressor becomes operational discontinuously only when it recharges to storages or buses. A cycle of the operation of the station, including a complete start-up and shutdown of all the facilities, is similar to batch operations, but the number and amount of chages of CNG to buses are quite different from operation to operation. Thus, the information from the previous batch cannot be directly utilized for the monitoring and diagnosis of the next batch opration.
Even though a basic SCADA system is being operational for the CNG station used as testbed, the current state of monitoring does not consider the characteristics of unsteady operations. Operators simply ignore all alarms and only watch the sensor values directly to make the decision by themselves, because current monitoring is based on fixed threshold limits without matter of the state of operations and generates many nuisance alarms starting from the start-up and during ordinary operations. To improve the safety and operational uptime of the CNG station, an advanced monitoring and diagnosis system is proposed, in addition to the adoption of wireless sensors and sensor networks. Monitoring of the overall operations, including possible gas leaks, is performed by using sensor data collected over wired as well as wireless sensor networks. The proposed advanced monitoring and diagnosis system is based on a combination of PCA and moving boundary limits suggested by the dynamic model learned by the neural network.
In addition to preventing the unexpected leaks and failure of compressors through better monitoring, the other important concern for successful operation of CNG stations is achieving high energy efficiency. Based on our previous knowledge and experience on the optimization of compressor operations in LNG regasification plants, the optimization of the compressor operations to improve the energy-efficiency of the whole operation of the CNG stations is also proposed. Based on the diagnosis and prognosis result of the facilities and expected number of recharging of CNG for a particular day estimated from the previous history of operational records, start-up and shutdown time of compressor operations and the amount of CNG stored in the storage tanks are optimized in real-time. The figures below show the screenshot of current prototype system, being tested on-site and to be deployed in a few months.
Figure SEQ Figure \* ARABIC 1: Screenshot of the Prototype System
Figure 2: A Trend of Past Operational Data
Acknowledgements: This research has been financially supported by Korea Energy Management Corp.
1. BP Statistical Review of World Energy 2009, http://www.bp.com/productlanding.do?categoryId=6929&contentId=7044622