Executing Design for Reliability Within the Product Life Cycle

Ali Jamnia, Khaled Atua

• Publisher: CRC Press
• Year: 2019
• ISBN:
  9781351165709
• Edition: 1
• Topic: Tecnología e ingeniería
• Subtopic: Ingeniería civil

At an early stage of the development, the design teams should ask questions such as, "How reliable will my product be?" "How reliable should my product be?" And, "How frequently does the product need to be repaired / maintained?" To answer these questions, the design team needs to develop an understanding of how and why their products fails; then, make only those changes to improve reliability while remaining within cost budget.

The body of available literature may be separated into three distinct categories: "theory" of reliability and its associated calculations; reliability analysis of test or field data – provided the data is well behaved; and, finally, establishing and managing organizational reliability activities. The problem remains that when design engineers face the question of design for reliability, they are often at a loss. What is missing in the reliability literature is a set of practical steps without the need to turn to heavy statistics.

Executing Design for Reliability Within the Product Life Cycle provides a basic approach to conducting reliability-related streamlined engineering activities, balancing analysis with a high-level view of reliability within product design and development. This approach empowers design engineers with a practical understanding of reliability and its role in the design process, and helps design team members assigned to reliability roles and responsibilities to understand how to deploy and utilize reliability tools. The authors draw on their experience to show how these tools and processes are integrated within the design and development cycle to assure reliability, and also to verify and demonstrate this reliability to colleagues and customers.

1

Reliability, Risk, and Their Significance

Introduction

The classical textbook definition of reliability as defined in many statistics and reliability engineering books is as follows: the probability that a product will perform its intended function for a specified time under specified conditions. However, given this definition, the interpretation of reliability varies based on the interest of the stakeholder. For instance, if you ask the end user or the buyer what the reliability characteristics of a purchased product should be, the answer may be one or all of the following:

• It never fails.
• No surprises, no unscheduled downtime.
• Get me up and running quickly after failures occur.
• Highest productivity and throughputs.
• Available whenever needed.
• Low cost of ownership.

For a manufacturer or a business, reliability is usually understood in terms of the following elements:

• Low cost of product surveillance, failure investigation, and field corrective action
• Low cost of service, including warranty, inventory for repair, and spare parts
• Low cost of liability, including regulatory or noncompliance, and low legal cost
• Maximized profit, including expedited product launch with low risk, increased customer loyalty, and elimination or rescue of lost sales

On the one hand, it is the responsibility of the reliability engineering team to examine the reliability of the product and provide feedback to the design team. In this effort, both the end user’s and the business’ concerns are addressed. On the other hand, it is the responsibility of the design team to define reliability requirements for the product. These requirements should fulfill both user and business concerns, i.e., end user expectation of failure-free life and longevity, and business needs in terms of low cost.

A History of Reliability and Background on Design for Reliability (DfR)

We find it interesting to consider the etymology of the word reliability as we delve into developing an understanding of how to develop and execute a plan to make products more reliable. The word reliable is a combination of rely and able, meaning the capacity on which one may depend on. Interestingly, reliable is based on the 1560s Scottish word reliabill, which was not commonly used prior to the 1800s.¹ The word reliability was (is) synonymous with dependability and consistency, meanings that were utilized in the fields of psychology and statistics when references were made to psychological tests and measurements (McLinn 2010). The use of the word reliability is quite widespread and is not limited to technical fields. As Saleh and Marais (2006) pointed out, there are over 3000 books in the Library of Congress on this subject or with this title, and a Google search on “reliability” returned over 12 million entries.

Reliability engineering owes its beginnings to Walter Shewhart (in the 1920s and 1930s) who founded statistical process control by applying statistical methods to identify and solve manufacturing problems. It is interesting to note that his initial paper was not well received by the engineering community because “laws of chance have no proper place among scientific production methods” (Freeman as quoted by Saleh and Marais, 2006). Another impetus for the field of reliability engineering was the American system of manufacturing. This system was based on the concept of mass production using interchangeable components—a process that was popularized by Henry Ford.

In the 1940s, electronic equipment played a significant role in World War II. Use of electronic devices such as radars, radios, and other electronic sensors became possible because of the vacuum tube. However, these tubes frequently failed and needed to be replaced repeatedly, almost as much as five times more than any other component (Coppola 1984).

Eventually, in 1952, the US Department of Defense created the Agree Group on Reliability of Electronic Equipment (AGREE) in collaboration with the American electronics industry. The goal of AGREE was, first, to identify and recommend improvements that would lead to more reliable products; second, to create a mechanism to influence increased reliability in both government and civilian programs; and, last, to provide training material on reliability (Coppola 1984).

In the same year, ARINC (Aeronautical Radio Inc.) published a report titled “Terms of Interest in the Study of Reliability.” Recall that in the 1920s and 1930s, the engineering community believed that the concept of chance would not play a role in scientific production methods. This report turned that opinion and mindset on its head by recognizing the probabilistic nature of reliability, and hence, a departure from a definition based on a deterministic dependability and consistency. During the same period, ARINC was tasked by the Navy to study the vacuum tube field failures to identify root causes. This information was then relayed back to manufacturers of these tubes for design improvements and ultimately better reliability (Saleh and Marias 2006).

The realization of the stochastic nature of reliability was not fully based on a study of electronic equipment. In part, the works of Weibull as well as Epstein and Sobel on fatigue and cumulative damage, which would ultimately lead to defining a general distribution model as well as life testing (in 1951 and 1953, respectively), were crucial to this realization (Saleh and Marias 2006).

In a way, one may say that the birth of reliability engineering was celebrated when formal societies of reliability were created by the IEEE and the US Department of Defense in the 1950s. IEEE Transactions on Reliability, founded during this period, continues to present the latest theories and discoveries in this field. The premise of reliability engineering, established in the 1950s, was based on three objectives (Saleh and Marias 2006):

1. 1. Collect field failure data followed by root cause identification to improve reliability
2. 2. Understand and develop specific and measurable reliability requirements
3. 3. Develop means of predicting product reliability prior to building and testing units

Two reports were issued by the end of the 1950s . The first was the aforementioned AGREE report and the second was “Reliability Stress Analysis for Electronic Equipment,” also known as TR-1100, from the Radio Corporation of America. TR-1100 was arguably the first document that provided a model for predicting the reliability of a product purely based on failure rates of its components. Thus, this document satisfied the need for developing means for predicting reliability. It was the predecessor for the military handbook MIL-HDBK-217, which was published in 1961 (Saleh and Marias 2006). The AGREE report provided evidence that developing and demonstrating specific and measurable reliability requirements is quite possible (Saleh and Marias 2006).

It is important that we recognize that one of the most important assumptions of reliability, namely, constant failure rate (or the exponential distribution), used to this date, was established in the 1950s. It was this model that enabled reliability predictions based on known component failure rates. This approach and the known component failure rates of military devices were collected in MIL-HDBK-217. Today, most engineers consider any calculated values based on this military handbook irrelevant and erroneous. Even though this may be a correct assessment, they do not realize that the handbook was primarily developed to predict reliability of military equipment. Later, Bellcore and Telecordia updated the same data for the components and products used within their own specific industries.

The 1960s brought a new degree of sophistication to the field of reliability engineering by witnessing higher degrees of specialization. One area of pursuit was to examine new statistical distribution functions to model field failures and their behaviors. At the same time, new reliability models departed from a straight part-count by developing deeper understandings of redundancies in a system as well as using prior knowledge of product behavior to predict future field failures. Alongside these theoretical pursuits, a great deal of attention was given to testing and test methodologies. As a result, different time-to-failure distributions were identified. In this era, focus began to shift from...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Dedication
• Contents
• List of Figures
• Preface
• Authors
• 1. Reliability, Risk, and Their Significance
• 2. Design for Reliability Process
• 3. The Design Process and the V-Model
• 4. Reliability Requirements, Modeling, and Allocation
• 5. Reliability Planning
• 6. Reliability Statistics
• 7. Predictive and Analytical Tools in Design
• 8. Component and Subsystem Reliability Testing
• 9 System Reliability Testing
• 10. Reliability Outputs
• 11. Sustaining Product Reliability
• 12. A Primer on Product Risk Management
• 13. Relating Product Reliability to Risk
• Appendix
• References
• Index