ROBUSTNESS REMAINS the number-one deliverable in the process automation system. Connectivity is great, openness is wonderful, and cool graphical configuration tools rock, but operators will care less if the control system, causes process upsets or spurious shutdowns. Most of us who have been around for a while have seen how a single unplanned process interruption can negate months or years of groovy advanced-control benefits, and Ill bet nearly everyone has seen a colleague or two who confused his or her hobbyhorse with the real objective of manufacturingproducing positive cash flow through reliable and predictable production.
If you have spent any significant portion of your career as a process analyzer specialist, as I have, youve been acquainted with the term data validation for many years. Analyzers, in contrast to other process instrumentation, were once notoriously unreliable (ahem some still are), but the extraordinary value of having the data in near-real-time was obvious to even the most backward operations supervisor. It didnt take many instances of driving the process into the weeds for operations to stop calling maintenance when the analyzer went offline.
To keep our analyzer control loops functioning, we began employing levels of data validation. Our gas chromatograph (GC) vendors began to can maximum change and minimum change (flat-line detection) algorithms in the GC itself. Advanced controls vendors also made analogous computations on their end, and sometimes ran models and/or inferential schemes to compare and validate the process analyzer results. On the instrument end, we could get pretty sophisticated too. We could look at temperatures, flows, sensor voltages and other diagnostics, and write code to flag whether an analytical result could be believed or should be used for computing a new process set point or valve position.
Analyzers were the original smart instruments and benefited greatly from having integrated microprocessors. Not long after the introduction of smart transmitters, we began finding ways to use simple diagnostics: a crude status bit, and in many instances a case temperature, which those of us in cold climates used as a rudimentary plugged impulse line diagnostic. These simple features alone helped prevent many an abnormal situation or shutdown. More than 10 years ago, an operating supervisor at a neighboring refinery said he was able to stay running when others in our area froze up solid, because he could see which instruments were freezing up. The only time our refineries made money was when the other guy was unexpectedly down, so diagnostics were already improving process availability and profit.
Fieldbus devices have now been around for a while. Even those old devices had extensive diagnostics, but many were cryptic or relatively meaningless. Have you ever gotten theMaintenance_needed_now message? No kidding. You usually see it upon examining an already-broken instrument. This sort of diagnostic is not likely to lead to vastly improved process availability. Curiously, I have yet to see any device that overtly supports flat-line detection or a too big step change diagnostic, such as we knew from our analyzer days. Even today, I cant set a low temperature limit in a leading vendors DP cell to generate an alarm or diagnostic. Ironically, I can do this in the same companys valve positioner (if you reside below the Arctic Circle, you probably dont care if your positioner is cold). Do you suppose the valve boys just did this to mess with their transmitter brethren across the prairie?
Many users and most suppliers see process availability as the big payout benefit of Foundation Fieldbus (FF) diagnosticsa good portion of it realized through data validation.
I suspect many of us arent getting the most from fieldbus diagnostics, and we cant lay this deficiency at the feet of our suppliers.
If the engineers who configure your DCS are still behaving like their system stops at the marshalling cabinet in the rack room, you may need to bring them into the fold. DCS configuration specialists, even those who reside in a vendors shopreally dont understand or appreciate what their field instrumentsand Foundation Fieldbushave to offer. Data validation can help many plants achieve noticeable improvements.
|About the Author|