| |
 |
Corrosion rates are difficult to predict since they depend on process variables (concentrations, temperature, flow rates, etc.) that may not be known or, until now, predictable. An on-line Corrosion Monitoring System (CMS) is often needed to aid in the prevention of unscheduled shutdowns and lengthen the time between scheduled shutdowns. The CMS software system should seamlessly combine data from continuous corrosion sensors and process variable sensors to perform a non- intrusive , on-line inspection. This data should then be interpreted using an "expert" diagnostic and predictive corrosion tool. An ideal on-line CMS should automatically provide the process operator with information such as warning messages, the specific type of corrosion, predicted time frame for corrosion failure, and the specific reason for the corrosion problem.
The process of developing a CMS for specific application should include selecting candidate sensors for testing, defining a data collection strategy and required algorithms. Testing and implementation of the final CMS system should be preformed in-plant at chemical process facility at the conclusion of the project. Typical goals of such a project should be:
- Define Corrosion Monitoring System performance goals
- Evaluate and select corrosion sensors
- Design and fabricate test equipment including heat exchangers, reactor pots, distillation columns and piping for accelerated corrosion testing and data collection
- Assemble and test CMS hardware platform. Select, write and test software
- Develop diagnostic corrosion algorithms for general corrosion, pitting corrosion and stress corrosion cracking
- Develop prognostic corrosion algorithms to predict the time frame and likelihood of occurrence for a corrosion failure
- Conduct system performance tests
- Test and modify algorithms to reach performance goals
- Install CMS in a chemical plant for pre-demonstration testing
- Write test plan and perform demonstration test in a chemical plant of the diagnostic and prognostic decision algorithms
|
