Facebook tracking pixel
To view this video please enable JavaScript, and consider upgrading to a web browser that supports HTML5 video

Research Case studies

Smart and Sustainable IT


Problem addressed: Nuclear Power Plant (NPP) Loss of Coolant (LOCA)

NPPs rely on coolants to retain a safe level of temperature. NPP ageing poses high risks of failure and LOCA would have serious implications. The Smart project aims to mitigate the risks of LOCAs through early failure prediction. Thus, safety of NPPS could be enhanced through building and deployment of accurate predictive models using artificial and convolutional neural networks to monitor thermal hydraulic parameters at the headers, where coolant leaves the nuclear reactor. 

Plus Icon Research that has taken place

Research that has taken place is the detection and prediction of LOCAs using dynamic neural networks, and deep neural networks. Other research conducted relates to: signal processing methods for verification and validation, and online modelling for the prediction of plume dispersion.

Firstly, neural networks in a simulation are trained to recognise a LOCA from data received from monitoring the operations of an NPP. The data are available as time-dependent measurements related to inlet header ‘break-size’. There are 35 analogue sequences, critical NPP operating measurements varying over time, and two digital values, representing the state of components like reactor-tripped (yes or no) and pump-pressure-high (yes or no).

Six different break-sizes are modelled ranging from ‘no-break’ to ‘double ended guillotine break’, with severity graduating between the two extremes. Around 541 signal measurements are taken for a transient duration of one minute. A neural network built has the following architecture: 37 neurons in the input layer due to the 37 variables available from the data; two hidden layers of a variable number of nodes and a final layer of three outputs for detecting the break-size and inlet header location where a failure occurred. The outputs are termed ‘softmax’ because they use a well-known mathematical formula to create discrete binary output, rather than a variable one.

Based on the simulation experiments, the neural network with 11 hidden nodes is found to be the optimal network with the highest accuracy of 98.5% on the test set. A network with 19 nodes has delivered the same accuracy of 98.5% but has eight more hidden nodes. Neural networks with fewer hidden nodes tend to have better generalised performance, which is why they are considered ‘optimal’. For an inlet header with large break-sizes, the optimal network achieved more than 95% accuracy, the highest being 99.2% for the largest break-size. Finally, the optimal network is robust when the test data is corrupted with ‘noise.’

Smart and Sustainable IT Education

Problem addressed:

According to Digital Skills Agenda Europe 2025, there is a need for a consolidated effort to strengthen human capital, employability and competitiveness. This is due to a predicted shortfall of 500,000 ICT professionals by 2020. Additionally, the European Commission views the ICT sector’s pivotal role as leading the way on climate and energy targets through a transition to an energy-efficient and low carbon economy.

Plus Icon Solution that has taken place

PERCCOM is an Erasmus Mundus Joint Master Degree (EMJMD) in Pervasive Computing and Communications for Sustainable Development (PERCCOM) provides world-class education on an amalgam of advanced Information and Communication Technologies (ICT) and environmental awareness. The aim is to equip ICT professionals with unique skills and competences so that they could build cleaner, greener, more resource and energy efficient cyber-physical systems.

The e-Infranet consortium develops an e-InfraNet Green Sustainability Policy which serves as a guiding principle to influence the future development, assessment and decision processes of Europe’s e-infrastructure. The consortium has also developed a free postgraduate course on Green Data Centres (still hosted in the Open University, Netherlands).

GLOBE[4] is an Erasmus Plus project aims to develop the green skills of apprentices, contributing not only to improve work-linked training process but to incorporate some transversal green skills (related to energy, waste management and water management) in the training pathway of apprentices to contribute to the concept of green economy.

Our Impact

Environmental and Economic Impact – Disaster Prevention
The deliverables of the SMART project are online monitoring and prediction tools for LOCA to enhance the safety of NPPs. The beneficiary is BARC, India. Predictive model for post-disaster plume dispersion will help reduce radioactive releases to the environment. The developed tools will maintain plant availability and reduce accident handling costs in nuclear power plants.

Measures of Change
Safety of nuclear reactors in BARC has addressed the following: design level safety; radiological protection during operations; management of radioactive waste; preparedness for nuclear emergency. The SMART project exploits the use of information (through monitoring), Artificial Intelligence and Signal Processing to enhance the safety of NPPs.

Transnational Education Provision
Projects that we are involved in have impact on current and future education and training of IT professionals. They are joint EU Smart and Sustainable IT postgraduate provision through the following projects: PERCCOM, ALIOT, e-Infranet, and GLOBE. Details of engagement is as follows:

    This project has been a great success because 94 students are trained in pervasive computing and communications for sustainable development. To date more than 50% of PERCCOM graduates have continued their study at doctorate level while the rest have found positions in the industries (e.g. Volvo, Ericsson, Garmin, Telecom Indonesia, etc…). The annual summer schools (from 2015-2018) are well attended by approximately participants including NGOs, private or public organisations (e.g. Greenspector, Orange France, Ellen Mac Arthur Foundation, Garmin, Fluidedge Innovation (Ireland), Lappeenranta City Council (Finland), IHE Delft (Netherlands), ENEA (Italy), Sky Captain Aerial Imaging (UK), etc… Current and alumni places of origin.
    14 modules on integrated IoT and smart systems will be developed for postgraduate students (Masters and Doctoral levels) and tutors in Ukrainian universities. Thus far, 3 training sessions have been conducted with more than 50 participants (BSc, MSc, PhD students, Professors, Lecturers, Heads of Departments including University Chancellors) per session. They are ALIOT 2017 Spring Training School, Synergy between ALIOT and TEMPUS Cabriolet Meeting 2017 (Coimbra, Portugal), and ALIOT Winter School 2018.

If Europe wants to lead, it needs to pool expert resources together and invest in digital skills.

Transnational Education provision

Next Steps

Future plans of the SMART project are: BARC links integrate SMART monitoring tools to their NPP safety system; continue UK-India Nuclear collaboration through EPSRC Phase 4 funding. Future plans is to incorporate control systems for NPP safety.

The future plan of PERCCOM is to secure further EMJMD funding via GENIAL.

Research Outputs

Plus Icon EPSRC Smart Project

Smart online monitoring tool for diagnosis and prediction of NPPs’ status during different stages of loss of coolant accident (LOCA) progression. The smart monitoring tool comprises a range of AI modules for detection, diagnosis, monitoring, and prediction: source term prediction based on Bayesian Belief Networks (BBN), Artificial Neural Networks (ANN), Deep Learning (Convolutional Neural Networks (CNN)), etc…

The tools act as early warning devices in order to facilitate emergency preparedness, prevent accidents from occurring, predict potential loss of coolant accidents, pinpoint their specific locations and predict possible radioactive release and dispersion patterns for different accident scenarios. They will also monitor accident progression at nuclear power plants and predict the onset and evolution of an accident, supporting operators in their decision-making process.

Plus Icon Research Publications

David Tian, Jiamei Deng, Gopika Vinod and T.V. Santhosh. (2018). A Constraint-based Random Search Algorithm for Optimizing Neural Network Architectures and Ensemble Construction in Detecting Loss of Coolant Accidents in Nuclear Power Plants, 11th International Conference on Developments in e-Systems Engineering, Cambridge, UK, 2-5 September 2018 (the 2nd Best Paper Award of the conference)

David Tian, Jiamei Deng, Gopika Vinod, T.V. Santhosh, Chris Gorse and Giuseppe Colantuono. (2017). Identification of Loss of Coolant Accidents of Nuclear Power Plants using Artificial Neural Networks, 4th International Conference on Nuclear Power Plant Life Management, Lyon, France 23–27 October 2017.

David Tian, Jiamei Deng, Gopika Vinod, T.V. Santhosh and Hissam Tawfik. (2018). Chapter 3: A Neural Networks Design Methodology for Detecting Loss of Coolant Accidents in Nuclear Power Plants, In Alani, M.M. et al (eds.). Applications of Big Data Analytics: Trends, Issues, and Challenges, Springer, 2018, DOI: DOI: 10.1007/978-3-319-76472-6_3.

David Tian, Jiamei Deng, Gopika Vinod and T.V. Santhosh. (2018). Selecting a Minimum Training Set for Neural Networks using Short-Time Fourier Transform in Detecting Loss of Coolant Accidents in Nuclear Power Plants,  The 2018 World  Congress in Computer Science, Computer Engineering, & Applied Computing, Las Vegas, USA, July 30 - Aug 02, 2018.

David Tian, Jiamei Deng, Gopika Vinod, T.V. Santhosh and Hissam Tawfik. (in press). A Constraint-based Genetic Algorithm for Optimizing Neural Network Architectures for Detection of Loss of Coolant Accidents of Nuclear Power Plants, Neurocomputing, Elsevier (accepted for publication)

David Tian, Jiamei Deng, Enrico Zio, Francesco Maio and Fucheng Liao . (2018). Failure Modes Detection of Nuclear Systems using Machine Learning, The Fifth International Conference on Dependable Systems and Their Applications (DSA2018), Dalian, China, September 22-23 2018.


Cutting edge research conducted by PERCCOM students relate to: (i) Software -  software eco-design, green software and storage; (ii) Network -  green network and network for greening; (iii) IoT – platforms, fog computing, and sustainability; (iv) Data centers – energy efficiency metrics,  green data centers, software defined data centers, and capacity management; (v) Smart Cities -  smart cities infrastructures, services, and systems.

Publications of PERCCOM

Details of all PERCCOM publications have been uploaded onto Researchgate.

Students’ Masters Theses could be found in LUT online repository.

Plus Icon ALIOT

The aim of this project is to develop MSc and PhD (multi-domain and integrated IoT programme) curricula for Ukrainian universities; facilitate intensive capacity building measures for tutors; establish two Ukrainian PhD incubators for multi-domain and integrated IoT infrastructures, platforms, services, and applications.


Training materials (for undergraduate and postgraduate courses)

IoT Cluster Network


  • Kharchenko, V., Kor, A. L., and Rucinski, A. (eds.). Dependable IoT for Human and Industry: Modeling, Architecting, Implementation (River Publishers Series in Information Science and Technology), ISBN-13: 978-8770220149, ISBN-10: 877022014X, Publisher: River Publishers (October 31, 2018).

Media Mentions

PERCOMM Master Programme

Worlds Most Sustainable Program

Back to Top Button
Back to Top Button