Predictive Maintenance Through Monitoring of Control Loops

Today's open market requires companies to increase productivity to face global competition. Making reductions in production costs and operating at optimal conditions are the primary objectives. Effective instrumentation, process control and maintenance are essential elements to fulfill these objectives. Even though the development of equipment and process control systems progresses at an impressive rate, maintenance is too often performed in only a corrective way. Unexpected shutdowns and reduction of equipment availability generally lead to increased production costs and high production loss. Increased use of the information generated by control equipment and systematic documentation of interventions carried out on control loops are interesting avenues to follow to find solutions to implement the most advantageous maintenance procedures. In this article, the real-time monitoring of control loops is presented as a predictive maintenance tool.

Maintenance Objectives
The objective of maintenance is to replace or repair used or defective equipment to preserve its functionality. Generally, the different types of maintenance applied in the industry can be found in one of the following categories:

Corrective maintenance is carried out after a failure is detected and located in the process. Most of the time, it is performed off-line and requires the shutdown of one or more sections of the process. Preventive maintenance is generally applied systematically, i.e., planned and performed at regular time intervals. It is then possible to coordinate these activities with planned shutdowns. However, this type of maintenance does not offer any protection against faults occurring between interventions. Finally, predictive maintenance is based on real-time monitoring of systems to prevent failures before they occur. It thus requires an on-line analysis of signals and information from the process.

The main advantages of predictive maintenance are:

The fact that some unexpected shutdowns can be avoided justifies the efforts deployed to implement this type of maintenance. Taking for an example a pulp and paper plant, Dottori [1] estimates that eliminating five unplanned shutdowns of an average duration of two hours each, will save approximately $250,000/year in additional production costs. These costs, on different scales, can be transferred to different types of industries.

The on-line monitoring of the performance of control loops supplies an effective indicator as to the good working order of a dynamic system and its associated control equipment. Solutions and tools are necessary to process and analyse the information required for this task. These tools (software) can easily be connected to existing control systems such as distributed control systems (DCS) or programmable logic controllers (PLC).

This real-time monitoring can be integrated to predictive maintenance procedures. The fact that the performance of a control loop, even when correctly tuned initially, decreases over time is a straightforward indicator of a change in the properties of the controlled system:

When the occurrence of a failure is anticipated, actions can be taken on the control loop equipment or the process system to restore proper performance. Contrary to the usual tendency to retune the controller, this approach gives the possibility of resolving the problems at the source, instead of masking equipment malfunction by an unjustified readjustment of the controller parameters.

Many approaches or detection criteria can be used to analyze control loop signals in order to anticipate different types of failures. Typical techniques are as follows:

These methods become more powerful when combined with other sources of information. Using historical data that characterizes the typical signature of a fault or a network of intelligent sensors that complete the diagnosis, are some options that show very interesting possibilities.

Conclusion
In conclusion, the equipment and systems related to process control provide an important quantity of information. Due to the complexity and the time required to analyze the data manually, this information is underused most of the time. Integrating autonomous tools (e.g. software) for signal processing and data analysis gives rise to the possibility of fully exploiting the potential of this data. On-line monitoring of the performance of control loops as a tool in predictive maintenance programs is one of the solutions recommended to help companies reduce production costs while maintaining and improving the quality of its products.

References
1. Dottori F.A., "Process Control Benefits and the Bottom Line: How I see it !", Control Systems 2000, Quantifying the Benefits of Process Control, Victoria,BC,Canada, May 2000.

Eric Poulin is with Breton Banville and Associates. For more information visit www.bba.ca. ET