CCUG controls session includes deep dive into pattern recognition software

The Combined Cycle User Group (CCUG) conference session, Challenges of Modern Control Systems, kicked off with a presentation from a major DCS vendor on world class alarm management strategies. However, he might be the last person in the world of US power plant operation to think his major points were anything new. His recommendation to conduct an alarm audit yearly bears repeating however.

Two presentations from the same owner/operator focused on remote M&D and what they call “Smart Gen.” The company has over a hundred gas turbines in its fleet, including many combined cycles. Three and a half people are responsible for collecting, presenting, and maintaining real-time information on the combined cycle units, the first presenter said. To the extent possible, they translate results from their “target performance curves” into megawatts and dollar impact.

Advanced pattern recognition (APR) is being used to develop optimization routines. An example cited were correlations among auxiliary power consumed by the circulating water pumps, output, and ambient conditions. Perhaps the most important point he made, however, was that they are still focused on baseload operations. “Setting targets [using APR] for anything but baseload is difficult,” he concluded.

Polling Break: One third of the audience is currently using some form of pattern recognition for M&D and reliability programs.

The second presenter dove a little deeper. Five team members dedicated to the “reliability function” use APR to identify inflection points, the point where an “actual” measurement is deviating from the “expected” value. This allows them to flag potential O&M issues before plant operators see/hear an alarm. Then they can take inspection or maintenance actions before larger problems occur. He noted (as have many others at industry conferences) that equipment specialists are getting “thin,” and the goal is to extract the knowledge of those in their company and embed it in models.

One perennial gap with advanced M&D is direct measurement capability. Plants are traditionally outfitted with sensors critical for plant control but not necessarily real-time M&D. To plug this gap, the company has embarked on a long-term wireless sensor application program. Some advanced sensors identified include electromagnetic signature analysis, foreign particle detectors, radio frequency temperature sensors, fiber optic temperature sensor, acoustic leak detectors, and visual cameras stationed in key areas around the plant.

Once example of an important “catch”: they found a flange at significantly elevated temperature, which helped them pinpoint a critical leak in a gas turbine enclosure. Discovering these kinds of issues before they grow into safety or reliability impacts is the goal of these programs.

The connection between APR and advanced M&D to reliability-centered maintenance should be pretty obvious. Thus, this result from the polling was a bit disconcerting.

Polling break: Half of users in the audience have not set up equipment reliability priorities, i.e., given reliability ratings to all equipment, 35% have, and 15 % are in the process of doing so.

One user stressed that single point of failure analysis can turn up some rude surprises, such as the importance of a well-functioning purge system between gas and liquid fuel. Another user urged the audience to bring advanced M&D (on-line and off-line) to bear on transformers, the greatest single point of failure in the plant. “Transformers are not lasting the expected 30 years,” he said. His company now monitors eight transformer parameters every twenty minutes. Another audience member noted his company recently replaced all coated-type bushings on their transformers.

Replacement transformers have 12-18 month lead times. Insurance companies have flagged them as a leading cause of failure, plant downtime, and claimant payouts. Many owner/operator companies have acquired spares.

Posted in CCUG |

Comments are closed.