Data quality and reliability: a cornerstone for PHM processes

Jean-Baptiste Léger, Pierre-Jean Krauth, Guillaume Groussier, Maxime Monnin, Alain Mouchette, and Faycal Lawayeb
Submission Type: 
Full Paper
AttachmentSizeTimestamp
phmce_14_080.pdf1.27 MBJune 4, 2014 - 8:51am

Today, industrial measurement reliability is essential to answer the big challenges of European industries: improving product quality, creating high-added value products and improve process control. Indeed, industrial measurements are used to feed database and then analysed in order to improve process control. Therefore, the process control strongly depends on the reliability of measurements.
In industry, measurements devices are monitored thanks to: quality assurance and Predictive maintenance. However this is not satisfying since it relies on punctual verifications of measurements reliability and it covers only a fraction of instruments (less than 10%), and because controls are isolated. Such an approach does not guarantee full-time measurement reliability.
Besides, many sensors management software and process monitoring software are available on the market: one can find management of instruments’ calibration and maintenance actions, database of instruments’ measurements; process performance monitoring enabling the identification of process deviation.
In many cases such an approach does not allow identifying immediately a sensor drift. For example, during rolling, an out of gauge information does not allow discriminating a sensor drift from other causes such as rolling actuator failure or a problem linked to the metallurgical feature of the rolled product. These systems do not allow distinguishing individual sensor drift from the process, actuators and control systems behaviours. Moreover, industrial measurements are used to feed database and then analysed in order to improve process control. Therefore, the process control strategy strongly depends on the reliability of measurements affecting global performance.
In view of the above, maintenance team does not dispose of tools allowing anticipating sensors failures. This means that those failures are discovered during the analysis of incident on production lines or during the analysis of out-of-specification products. This is in fact a critical part within the deployment of PHM processes in order to ensure that further monitoring, diagnostic and prognostic methods are built and deployed on the basis of reliable and consistent data coming from the sensors.
To tackle such issue and further develop and deploy PHM system, Intelligent information technologies are required in order to enable sensor measurement validation taking into consideration steel process parameters correlation and operational conditions.
To this aim, an integrated software platform is currently develop under the umbrella of the PRIME project (ERA-Net / Manu-Net program) in order to enhance the reliability of on line and real-time industrial measurements. This innovative solution based on the CASIP/KASEM® platform integrates both an individual monitoring of sensors measurements and inter-measurements consistency monitoring.

In this paper, the main goals and objective of the approach will be presented along with an illustrative case study handled within this project. This case study is extracted from an ongoing application developed for the finishing line in ARCELORMITTAL plant at Florange in France. The proposed approached is described and results regarding measurement reliability assessment as well as sensor failure anticipation will be described.

Publication Year: 
2014
Publication Volume: 
5
Publication Control Number: 
080
Page Count: 
8
Submission Keywords: 
monitoring
diagnostics and prognostics
Data Quality in industry
Submission Topic Areas: 
Industrial applications
Submitted by: 
  
 
 
 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed