This course will help engineers and scientists to draw valid insights and to produce empirical and physically- motivated models using collections of data from various sources. The course will also help managers to understand and review data analysis and modelling work performed for them. Although most examples involve engineering materials used in nuclear power plants, the approach is applicable to practical data applications in many other areas of engineering and science.
To reduce experimental costs and use the widest available range of data, engineers and scientists often collect data from several sources and analyse and develop models of the collected data. The data may come from experiments conducted over several years by various laboratories, periodic field inspections at several plants, and plant monitoring or surveillance programmes. Although the experiments, inspections, and monitoring programmes were individually well planned to produce the most useful information possible, the collection of data is usually not homogeneous, and it poses unique challenges for analysis and modelling. The collection of data usually does not fit the classic experimental designs, it often is incomplete in some variables and not well-balanced, and the underlying phenomena often require non-linear models with multiple variables for making valid estimates and predictions.
This course is focused on analysis and modelling of collections of technical data from various sources, with an emphasis on practical applications rather than statistical theory. The concepts will be presented via a combination of class discussion and lecture, examples and computer demonstrations, and presentation of data analysis and modelling case studies from professional practice.
Room 9, Manchester Meeting Place Sackville Street Manchester M13 9PL
Wednesday 29 June 2016 Registration 09:00 Programme start 09:30 Dinner 19:00 for 19:30
Thursday 30 June 2016 Programme start 09:00 Closes around 16:30
Cost per person for the two day course: £750
The course fee covers all printed course materials, refreshments during the morning and afternoon breaks, lunch on both days and the dinner on Wednesday evening.
Delegates are responsible for booking (and paying) their own accommodation. We recommend the Pendulum Hotel or Hotel IBIS (Manchester Centre Princess Street) which is a short walk to Manchester Meeting Place.
Car parking is available close by, at the multi storey on Charles Street. Cost £10 for 24 hours. Delegates parking overnight use levels L & M, otherwise park on any level.
The main train station (Piccadilly) is less than 10 minutes walk. Details on how to get to Manchester, can be found here, via the University website.
The Sat Nav postcode is M1 3BB.
Telephone: +44 161 306 4600 (note usual days of working are Monday ,Tuesday and Wednesday)
Visiting Professor Ernest Eason (founder of Modeling & Computing Services, LLC) will present, discuss and demonstrate his approach for analysis and modelling of collections of engineering data. The class discussions, lectures, demonstrations and examples will be based on decades of work in his professional practice analysing and modelling various ageing processes of nuclear materials in both water-moderated and graphite-moderated nuclear power reactors.
To ensure that all participants will have sufficient background, typical materials ageing issues in water-moderated and graphite- moderated reactors will be described on the first morning. The data analysis and modelling terms used in the course will also be defined at that time.
Course Content -- Day 1
- Water Reactor Ageing Issues Introduce important materials ageing issues in water-moderated reactors, describing the phenomena and the types of data that are available.
- Graphite Reactor Ageing Issues Introduce important materials ageing issues in graphite-moderated reactors, describing the phenomena and the types of data that are available.
- Data Analysis and Modelling Terminology Define the principal data analysis, modelling, and model evaluation terms and concepts that are discussed in the short course.
- Overview of Modelling and Data Analysis Procedures Present a brief conceptual framework for the short course by describing the main steps in conducting typical data analysis and modelling efforts.
- Issues Associated with Collections of Data Present analysis and modelling issues associated with collections of engineering data, and contrast such collections with the smaller data sets produced from individual planned experiments.
- Identifying Suitable Modelling Variables in Collections of Data Present various methods for identifying modelling variables that are both important and usable.
- Useful Fitting Functions Present the form and features of several fitting functions that are flexible and widely useful for modelling engineering phenomena.
Course Content -- Day 2
- Non-linear, Multi-variable Model Development Approaches Present practical model development approaches for the non-linear, multi-variable effects often found in engineering data.
- Evaluating Goodness of Fit and Goodness of Data Present several methods for evaluating how well a model fits the data. Discuss “goodness of fit”, “goodness of data” and how to know when a model cannot be improved substantially with the available data.
- Dealing with Outliers and Fitting Anomalies Present methods for identifying and dealing with isolated outlier points and local regions of poor fit that may be observed when modelling collections of engineering data.
- Experimental Issues Identified in Many Collections of Data Present observations about various experimental issues based on Professor Eason’s experience analysing many sets of data.
- Practical Case Studies Present one or more case studies (as time allows) of engineering modelling efforts that were conducted by Professor Eason, illustrating modelling issues and custom analyses with opportunity for questions and class discussion. The cases will be selected from projects such as:
- Reactor pressure vessel fatigue and corrosion fatigue crack growth rates
- Crack initiation versus time in steam generator tubing
- Stress corrosion crack growth rate of nickel alloys in PWR primary water
- Neutron embrittlement of reactor pressure vessel steel
- Dimensional change of Gilsocarbon graphite
- Young's modulus changes in Gilsocarbon graphite
- Bending strength changes in Gilsocarbon graphite
- Changes in coefficient of thermal expansion of Gilsocarbon graphite
THE COURSE IS PRESENTED BY
Visiting Professor Ernest D. Eason
Professor Eason is an expert on developing practical models of engineering laboratory and field data, mainly describing materials performance and materials ageing in service. His undergraduate degree was an unusual combination of mechanical engineering and operations research, including substantial coursework in statistics. Professor Eason began performing practical materials modelling at Sandia National Laboratories (a US Department of Energy National Laboratory) after earning his Ph.D. in Mechanical Engineering at University of California, Berkeley, where he was a US National Science Foundation Fellow. He later joined Failure Analysis Associates (now part of Exponent, Inc.), a consulting firm that analyses accidents and failures involving engineered products and materials. There he became a Principal and Manager of the mechanical engineering department, conducting several modelling projects related to nuclear and other power plants. Since 1987 he has specialised in developing empirical and physically-motivated models through his own consulting firm, Modeling and Computing Services, LLC.
Professor Eason's models can be found in many published papers, reports of the Electric Power Research Institute, parts of the ASME Pressure Vessel Code, and regulations of the US Nuclear Regulatory Commission. He has modelled a wide variety of materials ageing processes in steels, stainless steels, nickel alloys, and graphite, including fatigue, corrosion fatigue, stress corrosion cracking, irradiation-assisted stress corrosion cracking, ductile tearing, irradiation embrittlement of steels, dimensional change and changes to several graphite properties due to irradiation and radiolytic oxidation of graphite. He is a registered Professional Engineer and has been honoured for his work by the American Society of Mechanical Engineers and the American Nuclear Society.
Professor Barry Marsden
After working as a Research Fellow sponsored by Rolls Royce researching the structural integrity of jet-engine turbine flanges in the Department of Mechanical Engineering, University of Nottingham, 1980 to 1983, Professor Marsden joined the UKAEA (later to become AEA Technology). After initially working on the development of fast reactors, he became Head of Nuclear Graphite Technology in 1991 and led a key-team involved in safety assessments, life extension and decommissioning of graphite moderated reactors (including AGR, Magnox, HTR and RBMK Nuclear Power Plant). During 1996 Professor Marsden was a guest Research Fellow at the Japanese Atomic Energy Research Institute (JAERI), Japan, jointly funded by the Japanese Government and UK Office of Nuclear Regulation.
In September 2001, Professor Marsden founded the Nuclear Graphite Research Group (NGRG) in the School of MACE (Mechanical, Aerospace and Civil Engineering) at the University of Manchester. The NGRG is now an international centre of excellence in nuclear graphite technology. Since 2003 Professor Marsden has been an active member of the Graphite Technical Advisory Committee (an HSE (NSD) independent advisory committee), and he is a member of the editorial board of the Journal of Nuclear Materials. Professor Marsden took semi-retirement in September 2011 but is still working actively at the University of Manchester and as a nuclear engineering consultant.
PREPARATION FOR THE COURSE
Education or experience in a field of science or engineering and some awareness of common engineering materials will be helpful. Exposure to very basic probability and statistics concepts from experience or an introductory course will also be helpful, but expertise in that area will not be necessary. There will be no theorems or mathematical proofs in this course, as the emphasis will be on practical applications rather than theory.
Many excellent data analysis and modelling software packages and tools are available, some of which will be identified and described in the course. However, Microsoft Excel1 is usually available on computers used by technical people. Since the standard Excel tools are sufficient for demonstrating many key concepts of this course, and many technical people have some familiarity with them, Excel1 tools will be used in class demonstrations. Participants are welcome to bring their own computers with Excel installed and follow the class demonstrations on their own computers.