ENBIS: European Network for Business and Industrial Statistics

My abstracts

The following abstracts have been accepted for this event:

• Predictive Maintenance and Model Based Control in Zero Defect Manufacturing
  
  Authors: Jan Post (Philips, University of Groningen)
  Primary area of focus / application: Process
  Keywords: Predictive maintenance, Industry 4.0, Data analytics, Model based control
  

  Submitted at 2-Mar-2017 16:29 by Jan Post
  Accepted (view paper)
  

  30-May-2017 08:45 Keynote - Predictive Maintenance and Model Based Control in Zero Defect Manufacturing
  Big data and knowledge based design become more and more popular in Industry as a part of Industry 4.0. On the one hand, in the development area, knowledge based design of process and product become mature and more easy to use in CAE (Computer Aided Engineering), on the other hand sensors, data analytics and predictive maintenance are introduced in – new - production platforms.
  Complex numerical models, like non-linear FEM models, of process chains combined with optimization techniques like DACE (Design and Analyses of Computer Experiments) doing hundreds of complex simulations lead to the analyses of enormous amounts of data and data processing.
  High volume production using automated production platforms, including real time sensing, data storage and data analytics also lead to big data issues and data processing.
  Data based on numerical models has its specific characteristics and is a priori (before the production platform is running) and the data based on real-time measurements has its characteristics but is a-posteriori. The question is how to combine these two domains of information and make it useful for future process control.
  Philips is working in both domains in the area of high precision parts made from stainless steel metal. This presentation will give an overview about what Public Private partnership are running at Philips on these subjects, how they contribute to the domains of process development, predictive maintenance and process control and will envision how these worlds can be combined to realize model based process control, including predictive maintenance.
• The Role of Statistics in Data-Driven Predictive Maintenance Systems
  
  Authors: Chris Gray (Uptime Engineering)
  Primary area of focus / application: Process
  Secondary area of focus / application: Reliability
  Keywords: Process, Reliability, Predictive maintenance, Live data
  

  Submitted at 3-Mar-2017 08:44 by Chris Gray
  Accepted
  

  29-May-2017 08:45 Keynote - The Role of Statistics in Data-Driven Predictive Maintenance Systems
  Predictive maintenance is practiced in a variety of industries in order to maximise overall system reliability. Detection of incipient failure, or early warning of elevated failure probability provides a trigger for critical inspection or repair tasks. If performed in a timely manner, the system is able to continue operation and expensive unplanned failure and the associated downtime is avoided. Such practices have reached a high level of maturity in the aerospace industry, although there is increasing interest from automotive, conventional and alternative power generation industries.
  
  Effective failure detection and prediction relies on detailed knowledge of the loading history for a specific system, its current operational state and the expected duty cycle. Furthermore, accuracy can be increased if information about the behaviour of a complete fleet is available. Overall the task of performing relevant analysis for fleets consisting of perhaps hundreds of vehicles or thousands of wind turbines becomes highly time consuming. A large volume of information is must be considered, typically including time series performance data, control system alarms and events, service logs, system configuration, component failure rates and component exchange histories.
  
  Manual interpretation of such information at large scale is prohibited by cost and resource limitations, therefore it is necessary to automate as much as possible the many required process steps. Data must be gathered and stored, data validation and analysis must be performed, unusual system behaviour must be detected and results must be converted into advisory statements relating to maintenance processes. Across this process, statistical methods must be effectively applied to ensure that accurate results can be produced with high efficiency.
  
  In order to place this in context, a case study is presented based on predictive maintenance of a large fleet of offshore wind turbines. The use of operational data as well as historical records of component failure to provide a basis for an optimised service strategy is presented.
• Empirical Bayes Approach for Site-Specific Wind Rose Prediction from Short-Term Data
  
  Authors: Antonio Lepore (University of Naples), Biagio Palumbo (University of Naples), Antonio Pievatolo (CNR-IMATI)
  Primary area of focus / application: Modelling
  Secondary area of focus / application: Reliability
  Keywords: Empirical Bayes approach, Wind rose, Directional statistics, Markov chain Monte Carlo
  

  Submitted at 3-Mar-2017 10:52 by Antonio Lepore
  Accepted
  

  29-May-2017 10:05 Empirical Bayes Approach for Site-Specific Wind Rose Prediction from Short-Term Data
  Feasibility of wind-farm projects emphasises the need for timely and site-specific evaluations of wind potential (i.e., electrical power production from wind sources) that, unfortunately, are usually hampered by lengthy and costly long-term anemometric sampling. On the other hand, short-term anemometric data may be poor if collected when the wind is not blowing from the prevailing direction(s). A reliable site-specific wind rose plays a strategic role in the wind-farm layout design, which basically aims to minimize wake loss due to any turbulent effects among adjacent wind turbines (WTs). An incorrect wind-farm layout forces wind sector management operators to systematically shut down selected WTs or to operate them in load-reducing modes that drastically increase operations and maintenance (O&M) costs.
  In this work, by means of a real case study from Southern Italy, an empirical Bayes approach is introduced to cope with actual site-specific problems faced by renewable energy companies. The proposed approach is capable of integrating short-term samples (collected at the candidate site) with both historical information (from a neighbouring survey station) and expert opinion in order to enhance the prediction of the annual wind rose at the given candidate site.
• Using JMP for Censored and Degradation Data: An Overview
  
  Authors: Volker Kraft (SAS Institute/JMP Division)
  Primary area of focus / application: Reliability
  Keywords: JMP, Censored and degradation data, Bayesian methods, Software
  

  Submitted at 5-Mar-2017 10:22 by Volker Kraft
  Accepted
  

  29-May-2017 11:15 Using JMP for Censored and Degradation Data: An Overview
  Whether in manufacturing or service delivery, high product reliability and high availability are strongly linked to business success. Although JMP has capabilities for getting the most from the many diverse data that arise when attempting to improve reliability and availability, this presentation focuses mostly on the treatment of censored and degradation data in JMP. As usual, an important aspect centers around how the functionality is surfaced, and this will be exemplified by some live examples. These examples will also show the use of Bayesian methods to incorporate prior engineering knowledge, and simulation methods to get more reliable measures of uncertainty.
• Using JMP for Censored and Degradation Data: Hands-on Session
  
  Authors: Volker Kraft (SAS Institute/JMP Division)
  Primary area of focus / application: Reliability
  Keywords: JMP, Censored and degradation data, Bayesian methods, Software
  

  Submitted at 5-Mar-2017 10:33 by Volker Kraft
  Accepted
  

  29-May-2017 14:45 Using JMP for Censored and Degradation Data: Hands-on Session
  This hands-on session relates to the companion presentation of the same title, and leads you through some more detailed examples of how to approach the analysis of censored and degradation data in JMP. Examples will be drawn from a variety of areas, and will illustrate the use of Bayesian methods to incorporate prior engineering knowledge, and simulation methods to get more reliable measures of uncertainty. All workshop content will be shared with the attendees, who are welcome to follow along using their own laptops. A free 30-day license of JMP 13 for Windows or Mac can be downloaded at www.jmp.com/try and should be pre-installed before the workshop.
• Software-Based Reliability Management
  
  Authors: Esther Lichtenegger (Uptime Engineering)
  Primary area of focus / application: Process
  Secondary area of focus / application: Reliability
  Keywords: Reliability, Maintenance, Validation, Software
  

  Submitted at 6-Mar-2017 11:01 by Chris Gray
  Accepted
  

  29-May-2017 11:55 Software-Based Reliability Management
  Reliability as a key performance indicator of technically complex products such as vehicles, construction equipment or wind turbines needs to be focused throughout the entire development process. Starting with product definition and component structures, corresponding reliability targets need to defined. To track such targets and ensure as early as possible that the customer expectation regarding reliability is fulfilled, usage and environmental conditions have to be known to be able to consider relevant stress occurring in real application already in the product design phase. The management of prototypes and functional development ensures that at the point where reliability validation starts, the product has sufficient maturity. The validation as one of the most costly parts of the product development needs to be tailored precisely to the requirements by applying both accelerated and representative testing in a balanced way to minimize the overall warranty risks. Finally, as soon as the product is released, monitoring of its operation is required for predictive maintenance and early indication of field problems.
  Uptime SOLUTIONS is a web-based tool designed to manage these reliability requirements adequately. It collects the required data of all relevant product development phases and makes them available transparently for all users. The module PROVE provides process optimization for the functional development phase. LOCATE is designed to assess, evaluate and optimize the product development and HARVEST manages the operation control including preventive maintenance. All three modules allow interactive work and provide comprehensive reporting tools.
  The presentation will give an overview of the most important features of Uptime SOLUTIONS and highlights some practically important results.