Advanced Diagnostics with Statistical Process Monitoring

One of the strategies I recommend to improve process reliability is the use of product tiers.  Manufacturers of instruments have various tiers which should be applied with a specifically defined strategy.  I have presented these tiers as premium, core and conventional.  Premium tier products should be used in critical applications which are required to maintain availability, quality or safety of a process.  A feature that is only available on a premium tier product is Advanced Diagnostics.  Instruments with Advanced Diagnostics (AD) not only provide alerts for basic instrument and loop errors, they also provide alerts for equipment associated with the instrument.  A pressure transmitter with Advanced Diagnostics will monitor statistical deviations in the process variable.  This is known as Statistical Process Monitoring.

As discussed in previous posts, Statistical Process Monitoring can be used to detect cavitation pumps, plugged impulse lines and impeller loss on a process mixer.  It does this by measuring the mean and standard deviation of the process variable.  The example I most often use to explain SPM is that of a car driving down the highway.  While the driver will see the speed at 65 MPH (the process variable) the tires will detect every part of the pavement.  Should the car leave pavement and continue on a gravel road, the driver continues to see the speed, but the tires realize an entirely different condition.  This would be reported as an alert.  The challenge becomes how the alert gets reported.

When investing in a transmitter with Advanced Diagnostics, the benefit is completely realized with the use of an Asset Manager.  The Asset Manager, which is separate from the host control system, continually monitors the status of instruments associated with it.  This includes pressure transmitter with Statistical Process Monitoring.  Any error is then reported to the host control system.  This keeps the host available for other functions of process control.  Unfortunately, not all plants have made this investment.

A solution to this problem is to map the mean and standard deviation to the second and third-process variables, respectively.  This information is then feed to the host.  Host systems generally include a data historian that can be used to track this data.  A simple programming step can monitor these values and provide an alert for any deviation.  This does require the use of a HART card.  It also limits the availability of other information that can be mapped to the variables.  As mentioned, it also uses more host resources.  With this strategy, however, you get Statistical Process Monitoring on your existing host.

As with many strategies presented, there are always unique challenges and these may not always work as described.  However, there is usually a solution that will work with your process.  It is likely I have discussed it here (or will sometime soon).  As always, the goal is to develop and execute strategies that create a process plant that is reliable, available and profitable.

Tagged with:

Leave a Reply