By Jim Montague, Executive Editor
How do you get process control knowledge out of someone's head? Ask them and then pay attention. Simple, right? Well, it's easy to say, but it can be devilishly hard to do.
First, you have to ask the right questions about operations, best practices and exceptional events. Second, you must document that collected know-how and coordinate it with the input from all the other veterans you're interviewing. Third, and most challenging, you have to convert this documented know-how into instructions that rookies can use for training and to make better decisions.
"Sometimes those people with the most knowledge are the least likely to write it down, so we've had some success pairing them with technical writers and then collecting their knowledge into a coherent form that others can use," says Eric Cosman, engineering solutions IT consultant at the Dow Chemical Co. (www.dow.com) in Midland, Mich. "We even have some videos steeaming on our corporate intranet."
Likewise, Finland-based Neste Oil Naantali (www.nesteoil.com) and Metso Automation (www.metso.com) recently cooperated to implement a refinery-wide metsoDNA CR automation system, which forms a complete process monitoring and control entity that covers phases that used to be controlled as separate units (Figure 1). Besides unifying the refinery's controls, Neste Oil's engineers report that metsoDNA gives them knowledge management tools, such as "diary" and "process helps," which store and allow sharing of experience, enhance continuous learning and develop operational practices. They add that metsoDNA's knowledge management also builds an "organization memory" to further boost performance and learning of Neste Oil's production staff.
"It's now is possible to use operator resources and increase the value of the oil fractions through the whole production chain without skill or technology limitations, since all processing units are now included in the same DCS version," says Kari Koivisto, who directed Neste Oil's automation upgrade project. "The operators share a similar process view through a uniform metsoDNA user interface, which enables uniform operation and feedback. Of course, the old veterans know the process right down to the last detail, but the new system is also useful to them because it gives them an overview of the whole process and helps when they're training new operators."
To help other users learn how knowledge management differs from what they do now, Eddie Habibi, founder and CEO of PAS Inc. (www.pas.com), explains that its heart is: 1) aggregating know-how into large collections after it's captured; 2) contextualizing that knowledge after it's retained, and 3) simplying it for easier sharing. "Knowledge is of no value until it's turned into an output," says Habibi. "This is also how our Integrity software brings in know-how; normalizes it, so everything looks alike; gives context to the data, so all the pieces make sense; and simplifies it, so the results are easy to find and use."
Habibi adds that PAS recently surveyed about 40 plant personnel who work with automation systems and found 25 to 40% of their time is spent searching for information instead of solving problems with that information. "Integrity can shorten those searches, so they only take up 1% to 2% of automation professionals' time," says Habibi. "For example, Integrity can indicate when an operator is preparing to start a pump and then suggest reviewing incident reports for it; or remind him about a procedure that needs to be done with this type of pump; and finally capture more knowledge at this point of transaction."
Extracting and Converting
While it's not easy to turn in-the-brain expertise into digital documentation that others can use, there are more than a few projects seeking to do it. "The key isn't capturing knowledge, but putting it to use. The know-how in books, papers and even DCS databases is no good if you can't apply it to the situation at hand. Fortunately, we're finding ways to put this knowledge in play automatically without so much effort," says George Buckbee, PE, Expertune's (www.expertune.com) vice president of product development and marketing. "This all begins with the ability to recognize patterns. For example, a sticky valve will generate patterns, such as a square wave on process variables or a saw-tooth wave on control outputs, but users need to be able to recognize and look for them. An expert like F. Greg Shinskey would recognize these patterns as a sticky valve, while a novice might just see it as something cycling. Capturing enough knowledge can help software like our PlantTriage perform these diagnostics, rather than just presenting data, and lead to direct action to correct problems. One of our PlantTriage users says it tells him what not to work on. Now, it's still difficult to capture all expert knowledge, but we can take key parts, make them accessible and apply them to actual situations." In fact, Expertune has codified knowledge from experts like Shinskey and used it to add active-model capture (AMC) and large-scale interaction analysis capabilities to its software.