Making Your Process Operator-Proof

Dec. 5, 2006
Senior Editor Dirk Willard discusses Shigeo Shingo's take on mistake-proofing, or Poke Yoke, if you speak Japanese. The practice suggests the following devices: eliminate — redesign, facilitate — guide, mitigate — lessen the damage caused by the error, or flag — identify the error.

“It happens all the time,” Mike complained. Caustic cleaning solution had been emptied by mistake into a product tank. He was right: different operator, same mistake, three times, for two-months running. This time, the problem had a twist: the handle for a valve was missing; another operator borrowed it to save himself a few steps.

Mike had opened the wrong valve by mistake — an all too common mistake. “Small wonder,” I told myself as I filled out the quality investigation: no pipe labels, poor design — no mistake-proofing. Running the plant with iron discipline alone wasn’t working. If you subscribe to Demming’s management philosophy, it was a failure of leadership. According to Demming, 95% of the blame for business failure rests with management, not the workers.

So, how can you make your process operator-proof? Shigeo Shingo, the originator of mistake-proofing, or Poke Yoke, if you speak Japanese, suggested the following devices: eliminate — redesign, facilitate — guide, mitigate — lessen the damage caused by the error, or flag — identify the error. Although, all are applicable in some circumstances, only elimination and mitigation are generally appropriate for the chemical industry given the safety and environmental impact of operator error.

Here are a few examples I’ve seen for eliminating potential operator error:

  1. Reducing the number of valves controlled by operators by pairing valves;
  2. Automating the entire process instead of using one automatic valve and one hand valve;
  3. Constructing with unique fittings or a one-way orientation to avoid piping errors, i.e., swing panels;
  4. Installing spring-check valves that close the line during a break, e.g., when unloading chlorine tankcars;
  5. Spot-welds on valves to prevent removal of the handle or opening the valve too far;
  6. Computer-generated labels for sample bottles;
  7. Welding of tools in-place or avoiding tools altogether so that tools aren’t lost in the process;
  8. Programming fixed limits, e.g., a stroke limit on a control valve;
  9. Using the same units in batch records, particularly those involving measurement totals;
  10. Orienting graphic (and equipment) layouts so as to avoid confusion between equipment and processes (layouts from operator’s natural perspective); and
  11. Configuring instrument fault alarms so that poor readings can’t be ignored.

Often, the solution is simple: why do you think manhole covers are round? If they were square they could be dropped into the hole. No investment in a manhole alignment technology was necessary.

Vendors can sometimes help in avoiding future operator errors. One well-known supplier of centrifuges designs its plates so they can only be assembled a certain way and each plate has a number identifying its place in a stack. I remember being amused when a new operator complained that he couldn’t get the cover on the centrifuge after cleaning; he had 160 plates in the wrong order so the cover couldn’t be bolted. At another company, I saw chemicals separated by class: acid, base, oxidizer, fuel, etc., to avoid accidental fires or reactions; the vendor supplying the chemicals used different shaped totes.

Sometimes, just labeling equipment and pipe can avoid problems. Far too often such details are ignored at the close of a project — this happens almost as much as the failure to produce as-built drawings. While avoiding a problem is better, sometimes the best we can hope for is mitigation.

One of the best devices I’ve seem for avoiding the effects of high pressure was a lift ball. A nickel-coated lead ball was loaded in a whistle box. With sufficient pressure, the ball would be lifted eventually crashing into the stops at the end of the box. Despite the presence of a rupture disc, which needed to be replaced if the ball lifted, this was a failure-proof device that avoided the risk of a relief valve failure. Another mitigating device was a chain at the end of a lift plate. If the tank reached sufficient pressure, due to a chain reaction, the plate would lift. The chain prevented the plate from flying off and injuring someone. In a fire, there may not be time for operators to act; an automatic valve can be returned to failed position by connecting the actuator by plastic hose.

Without a doubt, the best way to mistake-proof your process may be a dedicated, motivated, trained — and content workforce. In Mark Tweeddale’s book, “Managing Risk and Reliability of Process Plants,” p. 371,” he described a study of a commuter who saved money by calling his wife and hanging up. She knew when to wait by the phone and that her husband needed to be picked up at the train station. Because it was important to him, in more than 500 calls, he never dialed the wrong number.