Although valves are non-rotating mechanical assets, they can still sustain damaging vibrations and incur other problems. Monitoring their condition is important – particularly on the large valves used in nuclear steam turbine service. In this 2-part article, we explain what these valves do, why they are monitored, and how they are monitored.
There are typically more than 1,000 valves in a nuclear power plant1. As mechanical assets, they can wear out and due to the highly critical roles that many of these valves play, the consequences of failure are very high. The inability to open, close, or modulate a valve can be catastrophic. Those used as part of the steam loop for the turbine are particularly critical, but so are many others throughout the process, whether in BWR or PWR plants.
Statistics compiled over a 4-year period by the Electrical Power Research Institute (EPRI) and through the docket records of the Nuclear Regulatory Commission (NRC) in the late 1970s and early 1980s both quantified and increased the industry’s awareness of the impact of valve failures.
Understandably, this resulted in a renewed focus on the reliability of valves – initially in nuclear plants, but also spreading to conventional fossil plants as well as more broadly to the process industries – particularly those where valves were used in handling hazardous substances.
For the first decade after these findings were released, a preventive maintenance focus was employed; i.e., a reliance on calendar-based inspections and replacement of parts at pre-determined intervals. However, because this approach is intrusive and entails disassembly, inspection, and reassembly of the valves – often in areas where personnel are exposed to radiation – it is less than ideal. Figure 1 shows an in situ valve repair but a manual inspection as part of a preventive maintenance strategy is also done in situ and can be equally invasive.
Figure 1. In situ repair being performed on a large steam turbine valve. Photo courtesy of Reliable Turbine Services, LLC.
Regardless of these difficulties, the more significant problem is that preventive maintenance was not found to be sufficiently effective. It is a well-known fact, based on pioneering work2 commissioned by the US Department of Defense and based on data from the airline industries in the 1970s, that nearly 90% of failure modes are not age-related and thus cannot be predicted based on running hours or calendars. It is also a well-known fact – based on even earlier research dating back to WWII – that preventive maintenance can actually introduce problems where none previously existed3.
Thus, after a decade of unsatisfactory results using purely preventive maintenance approaches for valves, the industry began turning in the late 80s and early 90s to an exploration of condition-based (predictive) maintenance technologies. This is readily apparent by conducting a search of the literature and simply noting the proliferation of papers that began appearing during that time period.
Although there was a proliferation of papers during the aforementioned period, most were scientific publications and patents, and thus dealt with technology. As scientific publications – not business justifications – they assumed that the reader already understood why it was advisable to monitor valves and merely explained various methods. Methods are the “how” of monitoring, not the "why".
One particularly valuable article4 published in 1990 stands out. This is because it conveyed not just the “how” of a particular manufacturer’s solution (Westinghouse), but also an answer to the larger question pertaining to all manufacturers and all users: why. To answer this question, the article relied heavily on the previously cited EPRI and NRC data compiled between 1979 and 1982 (the authors presented this as Table 1, reproduced here in summary form for convenience). Although the “hows” may have changed and evolved over the intervening three decades, the fundamental “whys” have not.
Here is what the authors had to say in 1990 (adjusted for 2023 economic conditions):
Bear in mind that this reflects only the US nuclear market and that the US accounts for only about one-quarter of global nuclear power production. Bear in mind also that most fossil-fired plants use steam turbines and thus essentially the same critical valves as a nuclear plant uses for its steam loop. Additionally, fossil-fired plants still account for nearly two-thirds (63.3%) of global electricity while nuclear plants account for only 10.4%. The implications of all of this combined are that the costs of lost production in power generation attributable to valve-related issues are easily in excess of $5B annually. When the costs of parts and labor are added, the figure is even higher.
Thus, valve problems are problems worth addressing and a condition-based approach has generally been found to be the most cost-effective means of improving reliability for large, highly consequential valves. This is why suppliers of plants and machinery to the nuclear industry (such as Westinghouse and GE) began releasing valves with condition monitoring sensors and even special monitoring systems beginning in the 1990s.
“… the costs of lost production in power generation attributable to valve-related issues are easily in excess of $5B annually.”
At this point, we have addressed “why monitor”. However, there are really two “whys” associated with the topic of valve monitoring. We now turn to the second one.
Much research on the failure mechanisms of valves also occurred during the 80s and 90s. Valves are often in extremely harsh environments that include elevated temperatures and steam. Because valves move and involve parts that contact one another, such as stems and packing, they wear out. Other parts that can wear out are gear mechanisms, motors on motor-operated valves (MOVs), and diaphragms on air-operated valves. Steam quality and flow friction is also part of the equation. Because packing does not seal perfectly, steam leaks and valve stems and other parts develop mineral deposits and corrosion over time. Additionally, erosion of valve seats and other components occurs due to flow even without the inevitable contributions of corrosion. Parts may thus no longer seat properly and fully seal. And because many valves are not constantly moving and may indeed be in one position for very long periods of time (e.g., Main Steam Isolation Valves and Main Stop Valves), they can freeze in place due to deposits, corrosion, and other problems unless regularly stroked/tested. Even those valves that are moving regularly, such as governor or other control valves, may develop so-called stiction6,7,8 and thus behave erratically.
In addition to all of these, there is another phenomenon that causes valves to fail: vibration. Because vibration is involved, it explains why solutions from within our vibro-meter product line have been developed, given our domain expertise. We will discuss those solutions shortly, but let’s first examine how this vibration arises and why it is so damaging.
Steam valves can be subject not only to mechanical vibrations from piping and surrounding equipment (such as the turbine), but to vibrations arising in the valve itself from the acoustics of the steam flow9. These vibrations can separately be damaging to the valve, but they can also interact with one another in sympathetic fashion to exacerbate the problem and prematurely wear out the valve. At one time, designers simply tried to model and predict these frequencies and ensure that valves did not operate in such regions10. However, the demands on plant operational flexibility has made such an approach less desirable and online valve monitoring for such vibrations is now preferred. This is particularly true of valves used in steam turbine service and even more particularly, those in nuclear service.
Based on the foregoing discussions, it should come as no surprise that steam turbine manufacturers and valve OEMs have placed increasing emphasis on valve condition monitoring and now offer valves with such capabilities pre-installed. Consider, for example, these next gen steam valve offerings from GE for turbine main stop and control (governor) applications. Other examples could also be cited and indeed, the emergence of such valve monitoring systems was occurring as early as 30 years ago as previously mentioned4.
Although an OEM offering is certainly an option on a new plant, it generally does not address valves that are already installed. It also means that if valves from different manufacturers are installed, a plurality of monitoring systems will then exist – just for valves. A more desirable solution is to monitor valves in the same system that handles the steam turbines in nuclear and conventional thermal plants, and both gas and steam turbines in combined cycle plants. This points to the vibration monitoring system which is almost always accompanied by a companion condition monitoring system. Our VibroSight condition monitoring software and our VM600Mk2 hardware represents an ideal solution that is independent of whose valves you may be using. We also supply the necessary vibration sensors and interface seamlessly with other necessary sensors, such as those using a process variable 4-20mA output.
In Part 2, we present this solution in greater detail and examine the valves that are most often candidates for our valve condition monitoring solution.
1Au-Yang, M K. “Non-intrusive valve diagnosis” Power Engineering vol. 100, no. 1, Jan. 1996.
2 Nowlan, F.S., Heap, H.F. “Reliability-Centered Maintenance” US Department of Defense, Report AD-A066-579, Dec 1978
3 Waddington, C.H., “O.R. in World War II: operational research against the U-boat” (London: Elek Books, 1973)
4 Schreurs, J., Bednar, F., "On-line valve monitoring and diagnosis," IEEE Computer Applications in Power, Vol. 3, No. 1, pp. 25-29, Jan 1990.
5 1990 amount was $150 million per year at an electricity price of $0.10/kWh. Amount here reflects $0.16/kWh and a 227% adjustment for inflation.
6 Stiction stands for “static friction” and describes the tendency for the friction between the valve stem and the packing to prevent smooth movement. Consult footnotes 7 and 8 for additional reading on this topic.
7 Ruel, M. “Stiction: The Hidden Menace – How to Recognize This Most Difficult Cause of Loop Cycling” Control Magazine, Nov 2000. https://www.idc-online.com/technical_references/pdfs/instrumentation/STICTION.pdf (retrieved Feb 13, 2023)
8 Stewart, L. “What is Stiction?” Exida Explains, Feb 2013 https://www.exida.com/Blog/what_is_stiction (retrieved Feb 13, 2023)
9 Glaun, A. “Avoiding Flow-Induced Sympathetic Vibration in Control Valves” Power Magazine, pp. 80-82, Feb 2012.
https://www.powermag.com/proper-valve-selection-reduces-downtime-increases-process-efficiency/ (retrieved Feb 13, 2023)
10 Potdar, Y.K., Welch, D.E., Chan, C.P., Forte, G.F., Methods and Apparatuses for Monitoring Steam Turbine Valve Assemblies US 7596428 B2, United States Patent and Trademark Office, 29 Sept 2009.