Intelligent Condition Monitoring for Civil Nuclear Structures

Lead Research Organisation: University of Bristol
Department Name: Aerospace Engineering

Abstract

Building on its successful feasibility study, where a potential 25% time-saving for machine fault diagnosis was
demonstrated, the consortium aims to achieve Intelligent Condition Monitoring (ICM) for the civil nuclear industry. This
project will enable Beran and EDF Energy to extend health monitoring to a wider range of machinery, structures and new
civil nuclear plants.
The technology adopted is derived from a physical understanding of non-linearities arising from degradation of
components, utilising expertise in prognostic algorithms from the University of Bristol. The analysis tools will utilise methods
already used extensively in the aerospace and medical sector, which will be adapted for use within nuclear power plant
systems. Industry benefits include increased accuracy, reliability and safety, greater availability of plant through predictive
maintenance, improved operator intelligence, resulting in enhanced plant performance and reduced operating costs.

Planned Impact

This project will develop new technologies to support UK business in the civil nuclear industry and strengthen the supply
chain. The previous TSB-supported feasibility study demonstrated an EDF-endorsed potential 25% time-saving for
vibration condition monitoring. The technology developed during this project will be embedded within new and existing off- line and on-line monitoring instrumentation for structural health monitoring.
Economic Impacts -
The project will provide additional skilled employment and job security within the rural Torridge community (Tier 3 status),
consisting of 1.5 new engineers at Beran Instruments Ltd during development, and an additional 12 new employees 5
years following project completion (Appendix A of TSB Application). Both EDF and Beran will benefit from increased levels
of R&D within their organisations, with increased SME R&D spend estimated at 10%. Beran will be ideally placed to apply
these new techniques and equipment within a short timeframe, through its existing business with EDF Energy, where its
On-Line Vibration Monitoring Systems are currently being utilised to assess and maintain the health of plant assets,
including Turbines, Main Boiler Feed Pumps, Gas Circulators and Reactor Fuel Channel Components. The new ICM
product will form a spearhead for Beran's export development, contributing ongoing economic growth as part of the UK's
Manufacturing Export Strategy.
Economic benefits to the nuclear industry include:
1) Reduction in the number of unplanned plant outages;
2) Maintenance cost savings, as defective components are replaced before catastrophic damage occurs;
3) CM benefits to the smaller BOP market, where operators previously could not make a business case due to the
additional administration / cost burden associated with continuous monitoring.
Societal Impacts -
ICM as part of the Nuclear Industry's overall strategy will positively enhance the public perception in terms of EDF Energy's
operational safety. Indirectly, the new ICM product will help reduce the risk of blackouts, along with the resulting impact to
consumers in the wider economy.
The results of the project will also be proposed as part of the Royal Society Summer Exhibition, to demonstrate the
importance of the energy sector and its effect on society.
Environmental Impacts -
ICM's advanced anomaly detection will reduce the risk of catastrophic plant failure and pollution, where the consequences
of mechanical break-up are serious and costly to repair. The solutions prognostic capabilities will allow informed decisions
to be made about maintenance cycles and plant life extensions, helping optimise fuel usage and minimise waste.
Diversification of the technology within the Fossil Fuel and Renewables industries will help reduce CO2 emissions.

Publications


10 25 50
 
Description Excessive false alarm firing was established that is due to changing operating conditions. In order to deal with the challenge a regime recognition system was developed, which is generic and learns from the data and it is capable to identify the any number of unique operating regimes. The increase in the vibration level indicates developing fault but it can also can be caused by changing operating condition. A regime recognition system was developed which consists of two main parts - clustering based on k-means where the number and the signatures of the unique regimes are identified. The separate clusters, which represent the separate regimes are adopted as classes. Further a Bayesian classifier is trained using the identified unique regimes. Then the vibration signal is separated in different regions according the operational regimes and for each operational regime a threshold which separates the faulty from no faulty is calculated. Usually when calculating the threshold an assumption is made that the distribution is Gaussian, which in practice is a very rare case. In order to provide an adequate threshold Extreme Value Theory is applied as assumptions about the distribution are not required. Instead of establishing the distribution of the vibration signals we establish the distribution of the maxima as the faults manifest themselves in the vibration signal maxima. It was proven that the possibilities for the distribution of maxima are only three - Weibull, Frechet and Gumbel.An algorithm for the parameter estimation based on simulated annealing was developed. Having estimated the parameters then the calculation of the suitable threshold is trivial.
Exploitation Route The work has direct relevant to various industrial sectors interested in whole life sustainment of physical systems. The output has generated new tools which are currently being deployed by the industrial partner.
Sectors Aerospace, Defence and Marine,Energy,Transport
 
Description The methodology was applied in the power generation industry and the rate of false alarms was decreased by 75%. The developed system is learning and therefore generic and it can be applied in any industry.
First Year Of Impact 2016
Sector Aerospace, Defence and Marine,Energy
Impact Types Economic
 
Description EDF Energy 
Organisation EDF Energy
Country United Kingdom of Great Britain & Northern Ireland (UK) 
Sector Private 
PI Contribution understanding of monitoring techniques required to carry out in-situ health monitoring of large continuously operating machinery in the Nuclear power generation industry.
Collaborator Contribution Access to machines and data.
Impact https://hal.inria.fr/hal-01022050
Start Year 2010