Avoiding overload

Today's process instruments for the measurement of flow, level, temperature, pressure and analysis already have a high degree of on-board diagnostics. However, it is important to consider the usefulness of this data.

A condition monitoring (CM) system that reports every item of diagnostics data available is guilty of committing 'information overload', thus reducing its effectiveness to almost zero. An effective CM system will, therefore, be engineered for a specific purpose, rather than bought as an 'off-the-shelf' product.

The information from a modern, networked, process measuring instrument can be divided into two distinct areas - fixed errors and variable data.

Fixed errors are generally displayed in the form of a diagnostic code, which may relate to a predetermined condition defined by the instrument manufacturer.

Variable data can be represented by the process measurements and may take the form of process pH and process temperature, for example, the primary, secondary (and, if available, tertiary) measurements made by the instrument.

In addition, other internal variables can be used, such as the signal-to-noise ratio within an ultrasonic level device that indicates the quality of the reflected signal. Rather than waiting until the signal is dysfunctional, the value can be read and compared with a preset minimum value within the CM software.

When engineering a CM project, consideration must first be given to which instruments should be monitored. It makes no sense to monitor every measuring point on the plant as not all measurements impact on safety or quality. Indeed, some estimates show that only 5% of the installed base is important enough to condition monitor. Therefore, defining process-critical instruments is essential.

The next stage is to define which diagnostic information is required, as only when all points have been considered can the useful and application-specific diagnostics be monitored. The specific application must be considered and advice sought from both maintenance and operation staff to determine the normal and potentially abnormal conditions that might arise. Measuring instrument suppliers may also provide advice on the operation and diagnostic capabilities.

When determining which error codes to include within a process measuring instrument, manufacturers must decide on the text to associate with each error. This default text is usually instrument-based and rarely application-based. For example, when applying a pH monitoring system, the meter will measure the impedance of the pH reference probe. If the impedance increases above a preset limit, the instrument will report a 'high reference impedance' error.

The reality, however, is that the high impedance is caused by 'contaminants in the process causing a blockage in the reference probe'. This message would be far more meaningful, and could also include maintenance advice such as 'please replace the reference probe within 24 hours to maintain an accurate measurement'.

Once meaningful, application-orientated error messages have been established, these messages must go to the correct onsite personnel. Some messages indicate issues within the production process to enable operations staff to make more informed decisions. Others have no impact on production and clearly signal maintenance to respond in an appropriate timely manner.

Mode of delivery should also be considered. The maintenance department may require messages sent to a mobile phone via SMS. Maintenance managers will require notification by e-mail but also need the information to be sent to their Computerised Maintenance Management System (CMMS) using Open DataBase Connectivity (ODBC) data transfer.

With modern data transfer technology, existing infrastructures and networks can be utilised to both gather diagnostic data and notify users of abnormal condition. Web-based CM modules deliver the notification using Simple Mail Transfer Protocol (SMTP) systems and make historical condition data available over web-based networks using a standard internet browser.

Ease of integration into existing CMMS packages is important. Unlike a process control or SCADA package showing process values and alarms against a plant-based backdrop, CM status should be displayed in an easy-to-understand format. Each asset can be associated with an area of the plant and the condition indicated by a simple 'traffic light' system on the display using a classification determined during the engineering phase that is, again, application- or plant-dependent.

Providing all diagnostic data from all assets will give a false indication of the true condition of the devices and may well conceal the useful data required to accurately determine when maintenance is required, or when a process is performing badly. Correctly engineering a CM project is paramount. Information must be sought from experts in all areas - on-site engineers, operators and instrument manufacturers.

Advanced diagnostics are now available in some assets. By using logical expressions ('AND', 'OR', 'IF', etc) or complex algorithms within the instruments, even more reliable performance related diagnostics will be made available. However, users must ensure that the data remains relevant to the application.

The real strength of any CM system lies in the availability of long-term data for interpretation. A single failure, although unfortunate, may be accepted, but, with the benefit of historical diagnostic information, trends and patterns can be identified. The failure may not always be the fault of the instrument but due to changes in the process conditions.

By far the greatest expenditure of any maintenance department is on unplanned or corrective maintenance. Waiting for an asset to fail is a false economy, as irreparable damage can occur and consequent process downtime can be highly expensive.

Identifying when to maintain requires analysing historical data and making the assumption that the data is accurate. Maintaining because 'that is what we have always done' can be a costly exercise.

As advanced on-board diagnostics develop and assets become ever more intelligent, maintenance staff will become dependent on the high quality information they produce to determine maintenance schedules. Ultimately, this will reduce the necessity for corrective or planned maintenance, and move CM into the area of predictive maintenance.



Andy Smith is business development engineer at Endress+Hauser Ltd



It is difficult to see how asset management systems can be implemented, without a concerted effort to embrace fieldbus technology, believes Alan Southward, a project engineer with many years' experience in the pharmaceuticals industry.

Investment in fieldbus control systems often requires a change of culture at project and business level, as project managers are reluctant to take on the cost of asset management. And, said Southward, "all too often the risk to the projects cost/time weighs too heavily against change."

Project engineers should do more to push for fieldbus technology, especially during the early capital investment stages, insists the project engineer. "It is not easy to change the mindset of companies, if their engineers and project managers are not prepared to take the risk."

According to Southward, engineers should be more prepared to take on the challenge of showing the benefits of fieldbus technology and highlight the long term savings on maintaining the plant with the diagnostics available.