Instrumentation & control | Tracking degradation

Proactive degradation management

9 May 2011



Extending the lifetime of existing nuclear power plant is an attractive proposition for utilities. However longer-term operations (LTO) will probably require new approaches to detecting, characterizing, monitoring, and managing degradation and predicting remaining life of critical systems and components. Prognostics could fill this role. By Leonard Bond and Pradeep Ramuhalli


Nuclear power is expected to play a vital role in satisfying the ever-increasing global demand for low carbon energy. Existing reactors globally (which have an average age of more than 20 years [1, 2]) are considered too valuable to decommission at the end of their current licensing lives, since the cost of building replacement generating capacity, whilst at the same time providing new plants that meet the growth in demand for electricity, would challenge available technical, regulatory and economic infrastructures.

However, the push towards longer-term operations brings significant challenges. It is essential that safe operation practices are maintained and capacity factors be kept high and potentially even improved. Still, ageing of components (particularly materials ageing due to the harsh environmental conditions) can limit the operating lifetime of critical systems. Processes are already in place in the USA to extend operation from 40 to 60 years, and the majority of plants are now expected to operate to 60 years. The challenge is now to look and to determine the feasibility of operation from 60 to 80 years, and potentially beyond. This longer-term operation (LTO) will most probably require that new approaches be deployed to address potential technical challenges related to detecting, characterizing, monitoring and managing degradation, and predicting remaining life (prognostics). These approaches will also need an enhanced ability to contain operation and maintenance costs, as well as enabling operators to avoid extended or unplanned outages [2].

There is a growing consensus that activities that enable LTO beyond 60 years, will be focused more on passive systems and structures. Active systems (such as pumps and valves) are, and can continue to be, well managed and routinely monitored, diagnosed, analyzed and upgraded as needed. In addition, there is an opportunity for adoption of online condition-based monitoring and maintenance practices [see also NEI April 2010, pp13-20] that can potentially be improved somewhat, through the use of remote monitoring and the availability of remaining useful life (RUL) prediction.

However, passive systems and structures such as pressure vessels and piping are not easily or economically upgraded when degradation is detected in an in-service inspection (ISI), and therefore, managing passive systems and structures degradation is likely to be the key to determining the economic viability of LTO [3]. The ability to estimate the RUL, using a prognostic methodology, for passive components with degradation such as corrosion will be important [4], since it provides a basis for determining whether continued safe operation (over some pre-defined interval) is possible, whether operating conditions need to be changed to mitigate further degradation growth [5] and/or other mitigation or repair actions need to be implemented.

The use of a prognostics-based approach contrasts with current practice in the nuclear industry, where periodic ISI, reactive responses and maintenance are the norm [8, 9]. With the current approach, one issue that is faced is that degradation in safety-critical components is usually only detected at the periodic inspection, and potentially then only after significant degradation has occurred. This has potential implications for LTO in terms of both safety and economics, extending an outage when a defect is found and resulting in increasing the cost of maintenance. An alternative to such periodic inspection is a proactive approach, using online monitoring, to identify, and hence enable the management and mitigation of degradation, especially if degradation can be detected early. This proactive approach in other fields is known by various names including prognostics-based health management (PHM) and integrated system health management (ISHM). Proactive approaches to manage component life and degradation require online tools to detect degradation early in the lifecycle (diagnostics), and estimate RUL of the degraded component or system (prognostics). There is potential for the nuclear industry to benefit greatly from advanced diagnostics and prognostics already deployed with active components in other high technology industries and to translate potential R&D outcomes into improved implementation strategies for NPP LTO [10].

The benefits of applying advanced diagnostics and prognostics technologies for LTO of nuclear power plants include:

  • Improved awareness, understanding, characterization, and prediction of coupled stressor and degradation effects, which if undetected, are associated with failures that threaten nuclear plant safety
  • Robustness and fault tolerance to component/subsystem/plant or system-level off-nominal performance, anomalous behaviour, faults, and damage
  • Improved/optimal performance of nuclear power plant systems at the plant-wide level
  • Potential to reduce cost associated with maintenance and unplanned down time
  • Safety assurance of advanced safety critical technologies for the current fleet of LWRs, new builds and also next generation plants

Diagnostics and prognostics

In both active and passive safety-critical NPP?components, an estimate of the current state of degradation is needed prior to determination of the RUL. Diagnostics or monitoring is aimed at signatures for degradation characterization; that is, determining the current state of the component. Within the context of proactive management of degradation (PMD), the diagnostics tool needs to be able to detect, and then monitor degradation early in its development cycle (i.e., at the degradation precursor stage) [2, 11, 12] so that appropriate mitigation strategies may be economically applied. The approach requires an integration of the science relating to stressors and component degradation, the sensors that measure stressors (which can be operational parameters) and signatures from which material and component conditions can be determined, all performed in the context of a risk-informed approach that identifies the most significant regions of a plant that are to be monitored.

The early detection of degradation is difficult, particularly in LTO where there is a risk of late-initiating and then fast-developing phenomena, largely due to the difficulty in defining and detecting appropriate precursors for every degradation mechanism. This is made additionally difficult in passive components due to the complexity of plants, and ongoing mitigation and response activities, including the move to replacements with new materials (such as HDPE piping) which are being used to replace some existing degraded components [see also NEI December 2010 pp30-1]. Further, diagnostics tools need to be capable of monitoring potentially large areas, with modest numbers of sensors, that operate reliably over the long term, and are capable of detecting relatively small changes in the level of degradation. The development of advanced measurement techniques (such as phased-array ultrasound for nondestructive measurement of material degradation) are being deployed in ISI, and these methods can also be used for online monitoring, as can acoustic emission and guided waves. Data from ISI can also potentially be integrated into accurate prognostics.

Assuming that appropriate NDE/diagnostics tools are available and locations identified that will enable a representative sampling of high risk areas in passive structures to be monitored, the challenge is then in assessing the degradation state from such measurements. This is an ill-posed inverse problem, with only limited or sparse data that is sensitive to measurement noise, and may not have a unique solution. However, many approaches have been proposed to address these issues, including physics-based methods. Given the current state of the component, a prognostics algorithm can then begin to assess the RUL. The accurate prediction of RUL is challenging, largely due to the dependence of the RUL on the current degradation state and expected future stressor levels—both of which to some extent are unknown—and the impact of past history of stressor levels which may only be partially known.

Several methods have been proposed for prognostics (in the nuclear and non-nuclear industry) [5]. Reliability data-based methods use historical time-to-failure data to model failure distributions (such as the commonly-used Weibull distribution) and predict the time-to-failure. Such methods, while commonly used, cannot take into account the operating condition of the specific component, and therefore are least accurate. Stressor-based methods, on the other hand, take specific operating conditions into account. These methods generally focus on trending the stressor characteristics (such as current, voltage, temperature, pressure, etc.) to build a relationship between the stressor value and the predicted failure time. This information is then used to predict the RUL.

Effects-based methods compute a degradation or damage index and correlate this quantity with the probability of failure. The RUL is typically estimated based on the time for the damage index to exceed some predefined threshold. Markov-chain models and general path models are some examples of an effects-based approach. These effects-based methods do require some level of knowledge of the probability of damage accumulation during each operational cycle. This may be obtained either through experiments performed a priori, or through the use of physics-based models. In particular, the use of a physics-based model of damage growth tends to provide the most accurate and reliable results, particularly if the model also uses information about the stressor values.

Current activities

In the nuclear power arena, the growing awareness of the need for prognostics integrated with diagnostics has led to several initiatives by industry, academia, and government. Globally, the IAEA has several activities focused on plant life management (PLIM) underway, including the development of guidance and safety standards documents for PLIM and ageing management, coordinated research programs (CRP), and international workshops. One of the CRPs is focused on the development of advanced diagnostics and prognostics techniques for NPP structures, systems and components (SSC) [13]. Separately, an International Forum for Reactor Aging Management (IFRAM) is being formed to facilitate the sharing of information among participants world-wide on advances in materials aging management (including diagnostics and prognostics) for NPPs. Individual activities related to diagnostics and prognostics are underway at the European JRC (and in member nations of the EU), South Korea, China, Japan, India, Canada and the USA, where the Electric Power Research Institute (EPRI), the US Department of Energy and the US Nuclear Regulatory Commission are all engaged. The US federally-funded activities, under DOE-NE, are being organized in the Light Water Reactor Sustainability Program (LWRS), with a programme office at the Idaho National Laboratory. Some people in the community addressing the technical challenges related to reactor life extensions, which involves universities and national laboratories, are also considering how to integrate advanced diagnostics into new build and next-generation NPPs.

To date, active components (such as pumps) have received more attention with regard to diagnostics and prognostics, with the result that the state of the art for the technology for these components is at a higher level of maturity [14, 15 & 16], with several pilot installations of online monitoring systems for diagnostics. Several researchers have demonstrated the application of prognostic tools to these components [11, 17, 12 & 18]. In contrast, the early detection of degradation in materials used in pressure vessels and piping is only now starting to receive attention. Work being performed at the Pacific Northwest National Laboratory (PNNL), Iowa State University and Northwestern University in the United States is starting to address some of the issues associated with early degradation diagnostics and prognostics in structural materials in NPPs [19, 20 & 21] and work at Oak Ridge National Laboratory and other sites is focused on underlying materials degradation issues.

While the use of prognostics in NPP clearly has potential, it also poses challenges. Prognostics for active components is well established in other industries, and can be implemented in NPPs. The major challenges for NPPs are in the development of prognostics for passive structures.

The issues include:

  • Models of degradation accumulation. Model-based approaches to prognostics typically are the most accurate, providing the best models of the rate of degradation growth. Developing and validating such models presents significant challenges, both experimentally and mathematically. In particular, the physics of failure (from damage initiation to failure of the component) is still poorly understood, especially for structural materials [22]. For instance, while the factors that impact the growth of a crack in materials are reasonably well understood, the dynamics of incipient crack growth are less well known. The impact of one or more stressors on the rate of growth of degradation is also needed. Numerical studies, backed by careful experiments, are being conducted at several institutions worldwide to obtain a better understanding of damage phenomena, especially in structural materials used in NPPs.
  • Diagnostics and damage state awareness. A related issue is the availability of diagnostic methods that are sensitive to early stages of degradation. At issue are both the sensitivity and specificity of the diagnostic method to the degradation mechanism of interest. Further, the issue of determining the current damage state (or level) from the diagnostic measurements is also challenging. It is likely that advances in diagnostics technology from other industries can be adapted to the unique needs of the nuclear power area. It is also likely that no single diagnostic method can provide adequate information about the damaged state of a material, component, or system. Instead, multiple orthogonal diagnostic tools will be necessary, as will novel data fusion methods, to uniquely determine the damaged state of the component.
  • Prognostics from precursors. To be useful, estimates of RUL are necessary from early stages of degradation (precursors). Challenges in this area include appropriate definitions of degradation precursors (that is, what is a degradation precursor), availability of measurement tools sensitive to precursors, and an understanding of degradation development from precursor states to component failure.
  • Uncertainty quantification. Given the various uncertainties associated with measuring the current state of components and those associated with stressors and degradation evolution, the RUL estimate is likely to be somewhat uncertain as well. Methods for quantifying the uncertainty associated with the RUL are available and constraining (bounding) estimates will need to be validated for NPP implementation.

The science base for advanced diagnostics and prognostics needed to support its use in NPPs for active components (pumps, valves etc) has been demonstrated in other industries – the challenge is in adaption for NPP deployment and the validation of the methods. Applications for passive structures are being researched and early laboratory work is demonstrating the potential for these methods. The transition from condition-based methods for active components and current ISI to online monitoring and prognostics for passive components will be helped if regulatory relief from some ISI inspections is granted when an online monitoring approach is deployed. The adoption of digital I&C systems provides a framework within which the added functionality provided with online monitoring can be deployed, and used to reduce operations and maintenance costs.


Author Info:

Leonard Bond ([email protected]), Pradeep Ramuhalli, Pacific Northwest National Laboratory, P.O. Box 999, Richland, WA 99352

Figure 4 Figure 4


Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.