BOOK REVIEWS
This section will review those books whose content and level reflect the general editorial policy of Technometrics. Publishers should send books for review to Ejaz Ahmed, Department of Mathematics and Sciences, Brock University, St. Catharines, ON L2S 3A1 ( [email protected]).
The opinions expressed in this section are those of the reviewers. These opinions do not represent positions of the reviewers’ organization and may not reflect those of the editors or the sponsoring societies. Listed prices reflect information provided by the publisher and may not be current.
The book purchase programs of the American Society for Quality can provide some of these books at reduced prices for members. For information, contact the American Society for Quality at 1 (800) 248-1946.
Achieving Product Reliability: A Key to Business
Success
Necip Doganaksoy, William Q. Meeker, and Gerald G. Hahn, Eds.
Roelof L.J. Coetzer562
Mathematical Statistics
Lawrence M. Leemis
Robert Lewis and Jason Wilson564
Linear and Non-Linear System Theory
T. Thyagarajan and D. Kalpana
Antony Ndolo565
Advanced Engineering Mathematics
K.A. Stroud and Dexter J. Booth
Stan Lipovetsky566
Subjective Well-Being and Social Media: Reconciling Big Data and Statistics
Stefano M. Iacus and Giuseppe Porro
Stan Lipovetsky570
In the text, Achieving Product Reliability, the authors managed to address the most important aspects to achieve reliability assurance throughout the product life cycle. The chapters are very well organized and follow a chronological order of discussing every step in the process of achieving reliability assurance. More specifically, the following aspects are discussed; designing for product reliability assurance, reliability development steps and assessment methodologies, reliability validation testing and demonstration, assuring reliability in manufacturing including statistical process monitoring, tracking reliability in the field and pro-active feedback to ensure continuous reliability improvement. The text is concluded with a discussion on statistical aspects that are important in product lifetime data analysis.
The text contains little to no technical detail on statistical and fundamental reliability 1 theory, and consequently is very easy to read. Therefore, it is very applicable and suitable as a post-graduate course in mechanical or reliability engineering to give them exposure to the role of statistical thinking and analysis in reliability assurance initiatives and projects. The text is also applicable for post-graduate studies in the statistical and operations research sciences to make them aware of the opportunities in reliability design, development, testing and data analysis in product development and manufacturing. Furthermore, the text will be very suited for an advance course in management sciences to expose potential top management candidates of the importance of pro-active reliability assurance and how to achieve it, as well as to the complexity and multidisciplinary effort between engineers and statisticians required to ensure product reliability from development, manufacturing to assurance in the field and creating continuous value add. The authors are successful in explaining reliability in simple terms, why is it important throughout the product life cycle, and use many real-life examples to explain the concepts, dangers of poor or unquantified reliability, and the correct data analysis required to achieve overall reliability assurance. I will now provide some more detail on each of the chapters.
Chapter 1 sets the seen for the rest of the text and the following chapters. Reliability is defined in terms of how the customer sees it. Specifically, reliability assurance is both failure avoidance and quality over time. The aspect of time is very important in any product development because reliability assurance must be quantified for a specific time period into the future, say no more than 1% failures over the next 10 years at 95% confidence, through two proper testing and statistical data analysis. The latter is critical, and the example of the challenger disaster in 1986 is discussed where the wrong decisions were made due to the lack of data and improper data analysis. Therefore, the greatest challenge is to collect the right data to quantify product reliability correctly, and the importance of statistical design of experiments and data analysis cannot be over emphasized.
In Chapter 1 the authors also provide an excellent motivation to move from reactive to proactive reliability assurance, which entails employing the methods and statistical thinking which are discussed in the remaining chapters of the text. Proactive reliability assurance requires the use of statistics in design, testing and validation, and to understand and predict failures before it occurs, which create value add for the business.
Chapter 2 is about system reliability evaluation of a conceptual design and addresses the important concept of reliability by design, such as, reliability assurance starts at the design phase. An example of a washing machine, with its various components subjected to reliability failure, is used to explain how the reliability of any system can be investigated and quantified. For any conceptual product design, the impact of the lack of reliability can be quantified, alternative designs can be compared and requirements for redundancies can be tested, as well as the necessity of more reliable but more expensive components.
Reliability block diagrams (RBD) are explained and how the probability of failure or reliability is calculated using some elementary probability theory. However, for more complex systems and products with many components or assemblies, more complex engineering and three statistical models or simulation models are required to assess and quantify the reliability. The concepts of standby redundancy and k-out-of-n redundancy are explained as ways to improve system reliability. The aim is to achieve as high as possible reliability assurance with minimal cost. Greater redundancy creates greater reliability but is more expensive.
Product reliability development is discussed in Chapter 3. The statistical design and analysis of experiments are very important in product design and development to ensure the collection of the right data for proper reliability analysis and for developing accurate predictive models. Use-rate acceleration, accelerated life testing (ALT), and accelerated degradation testing are discussed in detail, with examples, to illustrate how reliability over time can be quantified.
Failure modes, causes, and mechanisms must be understood for reliable product design. Engineering tests to discover failure models and statistical based tests to estimate product reliability are both important in proper product design. Engineers often conduct highly accelerated life testing during product development for discovering failure models or potential failures as soon as possible. Statistical design of experiments and robust design are used to assess the performance of new materials, compare alternative designs, evaluate design changes, test prototypes and optimization.
Companies cannot afford to collect data over a long period of time for reliability evaluations. Therefore, use-rate acceleration, accelerated life testing (ALT) and accelerated degradation testing are used to collect data on components and assemblies under various 4 user rate conditions, stress conditions and environmental factors. Data on the development of a washing machine motor is provided and analyzed to illustrate the analysis of failure data and the prediction of lifetime. The development of a new designed insulation for generator amateur bars is discussed as an example of ALT.
Degradation measurements can be used when life testing does not provide sufficient failure data. Furthermore, degradation data allow for developing predictive lifetime models. Resistance measurements on 40 batteries are used as example of degradation data and the prediction, or extrapolation, of failure.
Chapter 4 is about reliability validation, which is aimed at evaluating product reliability under actual field operating conditions. Validation can be done through in-house testing and/or by beta site testing. The latter is where a sample of items are tested in the field under normal operating conditions before full-scale production. Valuable data can be collected from field testing to perform, for example, reliability growth analysis i.e., the rate of failures over time. This provides data for continuous improvement in reliability, new designs and product development.
Product safety is discussed as a criterion for reliability assurance. The aim is to ensure no personal arm or injury is sustained with the use of a product. This may require the introduction of additional measures for safety hazards or design alternatives for a safer product. A number of examples are provided to illustrate the concepts on reliability validation.
Reliability assurance during manufacturing is addressed in Chapter 5. The goal is that the reliability targets are maintained in the manufacturing of the products. In a manufacturing process there are many inputs or variables that could potentially affect the quality of the product. Therefore, quality and reliability must be improved by reducing process variability. Process capability analysis must be performed to understand the capability of the process to manufacture the product on target specifications and with minimal variability. Statistical process monitoring (SPM) and control charts were developed as early as the 1920’s and has been applied with great success in manufacturing processes to monitor over time whether the characteristics of a process or product are within statistical control i.e., within expected variability. The authors discuss different control charts and their use in manufacturing processes. Algorithmic Statistical Process Control (ASPC), which is the hybrid application of SPM and engineering or advanced process control (APC), is an exciting new discipline to explore.
Stator bar manufacturing for electric power generators is used as an example of applying statistical process monitoring and data analysis for process improvement. Specifically, quantifying measurement error and identifying sources of variability, and the process followed to reduce process variability are discussed. This is related to the so-called plan-do-check-act cycle or Six Sigma process. The statistical thinking and data analysis process followed in the example are relevant for any manufacturing process.
Chapter 6 is concerned with field reliability tracking, which aims to identify early failures in operational units or products. Collecting field data is used to improve reliability continuously, implementing mitigation actions, prevent potential harmful impacts of failures, and inform management for improved decision making. However, an understanding of the operational environment of the product is required for correct data analysis and informed reliability decision making. The development of an informative failure reporting system is critical to ensure fast, just-in-time and pro-active corrective actions of reliability problems.
In Chapter 7 the authors provide a view of the future in terms of collecting reliability data and the analysis of big data in particular. With 4IR and the rapid development of the Internet of Things (IoT), greater volumes of data are collected on all process and products in industry. Data are collected from online sensors and instrumentation on all process variables, and stored on SCADA systems for process monitoring, advanced process control, and predictive modelling. There is great opportunity to use online data for pro-active reliability analysis and development.
A digital twin is a complex computer model or a surrogate model of a physical system that is updated with actual operational, environmental and maintenance data. Such a computer model can be used for operational and reliability improvement of the actual process or system. Online and in-time data can be used to model and prevent failures, reduce unplanned maintenance, and other potentially more harmful failures.
The authors conclude with Chapter 8 which discusses various statistical methods for product lifetime data analysis. They address all the important aspects for the correct statistical analysis of lifetime data, as well as the assumptions of the analysis. Different models are discussed for reliability and lifetime data, as well as different types of data that may be encountered in reliability studies. The chapter is a very good summary of the important statistical models and data analysis approaches for analyzing reliability data. The authors provide many references which the interested reader can acquire for a more detail understanding of the statistical theory and methods.
I thoroughly enjoyed reading Achieving Product Reliability, and recommend it to engineers, technical and management level, and statisticians, both lecturers and research fellows, who are interested in reliability theory and data analysis.
≪INLINE FIGURE≫
North-West University, South Africa