The Foundation of a Good PdM Program

Issue 2 and Volume 3.





Click to Enlarge
Full scale model of the McGuire auxiliary feedwater supply system. Photo, Alden Research Laboratory.


Duke Energy’s predictive maintenance programs get measurable results.


By Drew Robb, Industry Writer

In May 2009, Duke Energy won the Nuclear Energy Institute’s Community Relations Process Award for its public outreach program “to educate the public and instill confidence in the safety, reliability and cost-effectiveness of nuclear energy.” For Duke, safety, reliability and cost-effectiveness are more than just a PR campaign. The company has a long-standing commitment to tracking down and fixing potential issues before they become a problem.

“We are always looking for ways to streamline our operations and find problems early,” said Ted Royal, senior engineer, predictive maintenance (PdM), at Catawba Nuclear Station in York County, S.C.

The 2,258 MW Catawba nuclear power plant is operated by Duke Energy and jointly owned by North Carolina Municipal Power Agency Number One, North Carolina Electric Membership Corp., Piedmont Municipal Power Agency and Duke Energy.

“In the predictive maintenance area, management support has been excellent,” said Royal. “Personnel resources are dedicated, we get the training we need and the latest equipment. I can’t say enough about how helpful that has been.”

Duke’s maintenance programs have been getting attention for a long time. In 1992 and 1994 respectively, the Oconee and McGuire nuclear sites received first place awards at the National Predictive Maintenance Conference and in 2006, received the Uptime Magazine Best Overall PdM Program Award at the Predictive Maintenance Technology Conference. The following year the Catawba plant took home that same award. In 2008, Catawba received industry recognition for strength in innovative monitoring and in 2009 was featured on the cover of Nuclear Professional magazine.

The PdM program has also gotten measurable results, with the Catawba site capacity factor going from the around 70 percent in the late 1980s to over 90 percent this past decade. In 2004, the Catawba Unit 2 set a company record of 531 days of continuous operation between refueling.

“I am convinced that a lot of the improvements that have occurred here, and in the industry as a whole, have been a direct result of being able to catch things early and fix them before they break,” Royal said.

No Component Left Behind

Royal has worked for Duke since 1981 and been in the PdM area for 18 years. Over the past two decades he has seen a lot of progress. He said that prior to 1985, the plant had a reactive and time-based maintenance approach. Today, condition-based monitoring, through the use of vibration, infrared and oil analysis technologies play key roles.

Click to Enlarge
Scaled hydraulic model of RWST withdrawal pipe. Photo, Alden Research Laboratory.

“We were more of a troubleshooting-type organization than we are today, where we catch problems in their earliest stage,” he said. “A lot of the advance has to do with industry initiatives, but a lot of it has also occurred through improvements in technology.”

In the early 1980s, PdM consisted of having one person who was a vibration expert lugging around large pieces of equipment on carts. The low-resolution infrared cameras available at the time, also quite large, required moving around a liquid nitrogen tank that was needed to supercool the detector.

“Technology changes have helped us tremendously, just as they have every other PdM program,” said Royal. “Now we can collect and analyze data with small handheld units that weigh 2 or 3 pounds and can download data into computer software programs where we can trend the data and analyze it in detail.”

Just as important have been the improvements in process, particularly INPO AP-913 Equipment Reliability Process Description. After the staff detects a problem, the process outlines whom they should inform, what the severity level is for the problem and whether the equipment is crucial for nuclear safety or plant reliability.

“The development of a good process is the foundation of any PdM program,” Royal said. “Between the technology and the process, we have been able to detect numerous instances of equipment degradation; not failures, because they never truly fail.”

Model Behavior

Improved sensors and analysis software provide one way to predict and prevent equipment problems. Another is to model plant systems to see what would happen during emergency or otherwise non-routine conditions.

For example, recently the U.S. Nuclear Regulatory Commission (NRC) advised that nuclear plant operators review their analysis of the tanks in their emergency core cooling systems (ECCS) and containment spray systems (CSS) to make certain the design was adequate to ensure that any vortices formed during drawdown of the fluid wouldn’t result in air getting sucked into the pump.

“We have a variety of different pump designs—some are single stage centrifugal, some are multi-stage pumps—and the latter are very intolerant to gas ingestion,” said Bryan Meyer, principal engineer in the Primary Systems Engineering Group at Duke Energy Corp.’s McGuire Nuclear Station.

Click to Enlarge
External view of scaled RWST hydraulic model, showing vortex viewing chamber and exit piping. Photo, Alden Research Laboratory.

As described in NRC Information Notice 2006-21 Operating Experience Regarding Entrainment of Air into Emergency Core Cooling and Containment Spray Systems, “Air entrained by more than a few percent by volume may degrade or cause failure of centrifugal type pumps. A degraded pump which successfully expels small amounts of entrained air (or other gases), may become air (or gas) bound to such a degree that it will not restart after being stopped due to the coalescing of air into the pump casing. Additionally, voids in the pumped fluid can cause excessive vibration and wear of important internal parts.”

While computational fluid dynamics (CFD) can be used to model certain types of flow problems, it cannot accurately predict the occurrence or severity of vortices. To ensure that its cooling systems could operate safely, Duke engaged Alden Research Laboratory (Alden) of Holden, Mass., to conduct physical modeling of its refueling water storage tanks (RWST) at all three of its nuclear plants (McGuire, Catawba and Oconee).

Since several different tanks with different geometries needed to be modeled, Alden used a circular tank with a diameter of approximately 40” at the bottom and a depth of approximately 5.5 feet, which would allow simulation of different depths. For the McGuire station (which consists of two 1,100 MW(e) Westinghouse pressurized water reactors (PWRs) with wet/ice containment) the model tank came to a scale of about 1:4. A 24” acrylic primary outlet pipe was installed at a 45 degree angle, with the elliptical entrance 12” above the tank bottom. Clear acrylic piping was used close to the tank to enable visual observations of air entrainment. A flow loop could operate closed, fully open or partially open. Partial return flow controls the rate of draw-down in the water tank, with the rest of the water going to the laboratory sump. Since water and acrylic have nearly identical refractive indexes, a rectangular acrylic viewing box was installed around the outlet pipe to compensate for the visual distortion of the curved pipe and to allow for good viewing and videotaping of the water bubbles.

A series of 10 tests were run at prototype flow rates from 1,600 gpm to 19,700 gpm. Five of the tests were conducted with a return flow rate representing a water level drop of about 1/2-inch per minute and five were done with the return pipe closed.

The tests showed that the tank could safely operate at much lower water levels than those required by ANSI’s Hydraulic Institute Standards (HIS). While the HIS specified a submergence of 2.85 feet for a flow of 1,600 gpm to 8.70 feet for 19,700 gpm, the hydraulic testing showed the tank was free of air entrainment to suction nozzle submergence depths as low as 0.045 feet at 1,600 gpm to 0.705 feet with no return flow at 19,700 gpm.

Similar results were achieved on the tests for the tanks for the two other Duke power plants. As a result, Duke was able to present verification to the regulators that there were no issues with its tank designs, which resulted in saving $50,000 per plant (not including down time) by not having to engineer and install vortex suppression devices on the tanks.

“Doing the physical model allowed us to recover some margin in our usable tank inventory and reduce our vortex allowance dramatically,” said Meyer.

Duke also had Alden evaluate the possibility of air entrainment during the operation of an auxiliary feed-water supply system. During an inspection, operators at McGuire discovered significant air pockets at high points in the auxiliary feed-water line, raising questions about possible air entrainment during operation. Once again, CFD modeling would not work.

“Numerical models can’t accurately capture all the physical processes that are going on, specifically the various bubble sizes, how the individual bubbles move in the water and how they coalesce into a larger bubble,” said Alden Vice President Dan Gessler, formerly a CFD research scientist at Colorado State University. “It is further complicated because the bubbles compress and expand as the pressure in the pipe changes. These are processes that are not well-captured in numerical models.”

In this case, rather than build a scale model, Alden built a full-sized model of McGuire’s 36” inlet pipe and the auxiliary feedwater piping to avoid bubbly flow scaling errors. Video cameras were set up at view ports to film any bubbles and sensors were placed along the pipe to measure changes in the water level as bubbles moved down stream. Over several weeks of testing, air was repeatedly introduced into the pipe and allowed to accumulate.

“We ran a variety of different flow scenarios for the service water; it may ramp up slowly or ramp up quickly,” said Phil Stacy, director of Alden’s Flow Measurement Division. “We conducted a lot of tests of the air volumes that accumulated in a steady state scenario and then tried to quantify how much air would enter the auxiliary feedwater branch during transient conditions.”

The data allowed Duke to evaluate the potential impact of bubbles on its pump and take any actions needed.

Building a Successful PdM Program

While no one can completely forecast the future, Royal says that his and other plants are getting much better at avoiding unpleasant ones.

“No matter what type of work you do, you should always strive for making improvements,” he said. “Never get complacent.”

While performing oil analysis right after an oil change, Catawba found that the particle count was unusually high. Catawba then tested the new oil in the drum, found that the brand new oil was actually dirty, and implemented a program where all oil is looped through a filtration process so it is pristine before going into the component.

Working with the flow accelerated corrosion (FAC) engineer, a program was implemented using thermography to determine which steam pipes need to be inspected for wall thinning. The program saves about $75,000 every refueling outage.

Catawba is also expanding the use of infrared with plans in place for 2010 to incorporate infrared scans as a part of plant engineering walk downs.

While the oil filtration and thermography programs were developed in house, Royal stresses working with others in the field to build on their experience and share your own. He has known his counterparts at Duke’s McGuire and Oconee plants, Dennis Roinick and Neil Watson, previous PdM Program winners, for 20 years and they speak several times per week.

Royal also does presentations for the Electric Power Research Institute (EPRI) and the Vibration Institute and regularly benchmarks plants overseas through Institute of Nuclear Power Operations (INPO), Areva and EPRI. For those looking at improving their own PdM program, he recommended using EPRI self-assessment guidelines and visiting other plants to see what programs they have in place.

The most import action, he said, is to ensure you have a good process in place.

John Wooden, a famous basketball coach once said, “Once you have the process in place, you know where your strengths and weaknesses are and you can focus your attention on areas where you need improvement.” Royal embraces this philosophy of “never mistake activity for achievement” and says, “The process defines your program. If you don’t have a process in place you will have a lot of activity but not a lot of achievement.”


More Nuclear Power International Issue Articles


Nuclear Power International Issue Archives


View Power Generation Articles on PennEnergy.com