It was the night after Christmas; I had the beeper. Except for the whirl of a distant blower the winter scene was something out of Norman Rockwell. But I was in a part of the plant I seldom visited because the production supervisor was having trouble with pH probes. I took a moment to take in the whole picture while he explained the problem. We had one technician each shift whose sole purpose was to maintain the pH probes in six batch tanks that fed a dryer. Yet, operators took measurements every hour to check the probe readings. “So, what was the purpose of the tank probes anyway?” I asked myself.
This was an old familiar problem. An engineer thought the pH probes would eliminate the need for operator measurements but production management quickly became leery of the idea when the probes fouled and caused mistakes. Another brilliant design idea rusted in place by the relentless progress of corrosion.
Where did the design engineer go wrong? Accounting for all the factors that affect reliability can be extremely tough. Take the simple dP transmitter — Eric Wickberg lists several criteria for instrument robustness, including resistance to corrosion, over-pressure and physical shock during handling (www.ChemicalProcessing.com/articles/2002/91.html). In detail engineering, these items can be systematically evaluated. Unfortunately, control decisions often are fixed in stone during basic design. So, detail design at best can only attempt to build control logic to protect the process from a reliability problem that will occur. Now, it helps that instruments often are specified based on reliable past service but that experience frequently must be extrapolated.
Once, I was asked to design a level alarm for a chlorine tankcar. It was new territory for me. So, I contacted several companies that claimed to have solved the problem, including soft-drink makers who used carbon dioxide. My first call was to our chlorine supplier but the soft-drink people got back to me first, suggesting that a target flow meter, my first idea, would work. I was under a time constraint so I went with the idea. Well, to be blunt, it didn’t work. I later suggested using the approach recommended by our chlorine supplier: a dP transmitter and a static loop. But, by then, I’d lost the support of production for my project. Generally, if an instrument fails to perform even once, or gives results that are hard to interpret, production will learn to ignore it.
The worst problems with reliability are when a plant is built for someone else, often by somebody else — especially when it’s overseas. A client in the Far East wanted to automate the make-up-water flows to its HCl absorber. This required a laboratory measurement. I knew we were in trouble the minute the vendor explained that the sampler and other equipment necessary for the proper functioning of the instrument weren’t integrated into the analyzer. Reliable is often synonymous with simplicity. Sure enough, the plant couldn’t even properly install the unit. How do you avoid these horrors?
Obviously, you need to nip them off early — in basic design. To keep yourself out of trouble, ask the vendor five simple questions: 1) How complex is the measurement, in steps? 2) Does the instrument require special training to maintain or use? 3) Does the vendor offer a sophisticated package of self-diagnostics? 4) Does the vendor even offer training to repair or calibrate the instrument? 5) Are case histories available with applications similar to yours?
Look for details in any case histories — and consider how they differ from your installation (www.chemicalprocessing.com/articles/2007/206.html). Ideally, you’d like to hear about instruments that have been operating for more than a year without trouble. What’s said about ease of installation (as it often plays a key role in reliability)? If the case history doesn’t contain a glowing recommendation from the production manager, ask more questions.
And, add one more item to your checklist: a reliability review. Just like a HAZOP, consider how the instrument could fail and how much will it cost, in time, downtime, lost production and good will with production.
One question you may be asking is: “How did I solve my pH problem?” We relocated the pH probes to the pump discharges and held a contest. I dangled a $50,000+/year contract in front of five instrument companies and we tested each analyzer and probe against a scorecard I created. Usually, all that’s needed to solve a problem is a little imagination — and the support of a management ready to cultivate new ideas.
Dirk Willard, contributing editor