BRIDGING RELIABILITY ENGINEERING AND SYSTEMS ENGINEERING

The increasing application of sensors, actuators, and complex algorithms for delivering artificial intelligence and connectivity in products and product-systems will drive an unprecedented growth in design complexity and software content, making it increasingly more difficult to ensure dependability in an economical manner. Much learning about the dependability of such new and innovative products is likely to happen as they are conceived and designed. Consequently, accelerated verification and validation iterations supported by easy and rapid storage and retrieval of failure knowledge must be enabled. No single software solutions provider effectively covers all three critical areas required for developing and delivering dependable smart connected products, namely, reliability engineering, systems engineering, and failure knowledge management. This paper mainly presents a potential map of the commonly used reliability engineering tools overlaid on the systems engineering technical processes. The paper recommends including a formal knowledge storage and retrieval system in the closed-loop between systems engineering and reliability engineering so that the details observed in past failures are not missed in future design iterations. INTRODUCTION Connectivity and artificial intelligence are major features of many upcoming products that are increasingly likely to be systems-of-systems. A large set of complex algorithms is needed for estimating accurately the instantaneous states of these systems and their operational environments and for exercising robust control over the systems to deliver the benefits desired by the end users. Intricate algorithms are used for sensor data fusion, remote diagnostics, remote repair, autonomous control, timely hand-off to humans, and many other functions. The increasing application of sensors, actuators, and the associated algorithms in products and product-systems will drive an unprecedented growth in design complexity and software content, making it increasingly more difficult to ensure dependability in an economical manner. Even without connectivity and artificial intelligence, the launch delays and recalls associated with today’s less sophisticated electronically controlled mechanical systems, due to performance issues, can very often be traced to design complexity and software-content. In case of connected, automated or autonomous systems, the problem can be expected to worsen. A US FDA study [1] has determined that software-related recalls measured as a percentage of all recalls in the medical devices industry have gone up from 14% in 2005 to 25% in 2011. This percentage has been trending upwards since 1983, with software-related recalls as a percentage of overall recalls averaging 6% between 1983 and 1991, 8% between 1992 and 1998, 11% between 1999 and 2004, and 19% between 2005 and 2011. The relationship between the number of recall events in a period of time and the number of units impacted by the recalls can be vastly different between industries [2]. For example, 150 recall events in the medical device industry may amount to 300,000 affected units but in the automotive industry, 30 recall events could impact 2 million vehicles. A financial advisory blog [3] mentions that there has been a substantial increase in software-related recalls in the automotive industry since 2012. The authors cite 32 software-related recalls that affected 3.6 million light vehicles between 2005 and 2012. However, they mention 6.4 million additional vehicles affected by 63 additional software-related recalls between 2013 and 2015. The blog also mentions going from 0.3% of recalls being softwarerelated in 2005 to 4.3% of recalls being software-related within the first 6 months of 2015, and this trend the authors state, is showing no signs of reversing. The authors of the blog also report a similar trend seen in NHTSA’s complaint data. Over the period covering 2005 to 2009, 55 softwareProceedings of the 2016 Ground Vehicle Systems Engineering and Technology Symposium (GVSETS) Bridging Reliability Engineering & Systems Engineering Page 2 of 6 related complaints were logged with NHTSA, whereas, over the period 2010 to 2014, 197 complaints contained the same reference to software related-issues, highlighting the increased role of software in automotive safety. Notable software-related issues in recent times, from the aerospace industry have been in connection with the Boeing 787 and the F-35 Joint Strike Fighter. A software bug in the Boeing 787 was found to be capable of shutting down the plane’s electric generators every 248 days because a software counter, internal to the generator control units (GCUs) could overflow after 248 days of continuous power. This could cause the GCU to go into failsafe mode, resulting in a loss of all electrical power regardless of the flight phase. The F-35 Joint Strike Fighter is expected to be further behind in its combat-readiness due to issues with its RADAR software and vulnerability to cyber-attacks, and these require the system to be rebooted every four hours of flight time while the desired reboot interval of the F-35 is eight to ten hours of flight time. The failure modes of software-intensive, control systems driven products are difficult to guess a priori due to the complexity of their functions and information flow, and consequently, crucial failure modes can easily be missed. Much learning about the dependability of such new and innovative products is likely to happen as they are conceived and designed. Consequently, accelerated verification and validation iterations supported by easy and rapid storage and retrieval of failure knowledge must be enabled. Today, enterprise level reliability engineering tools and systems engineering are not well-connected, preventing many lessons learned in reliability engineering from helping drive robust designs via systems engineering. This situation is further aggravated when we consider the silos of expertise and data among mechanical, electrical, and software disciplines. The product development tools used in these disciplines’ silos are very different and most often not connected with each other, rendering the ability to carry out systems engineering very difficult. A major challenge arises due to improper channeling of prior experience and knowledge about reliability into design, leading to repeated dependability issues of complex products. No single software solutions provider effectively covers all three critical areas required for developing and delivering dependable smart connected products, namely, reliability engineering, systems engineering, and knowledge management. Further, given the complexity of the development of the large number of specialty tools required to do this, it is perhaps unrealistic to expect that a completely integrated suite of solutions will be available from a single software solutions provider. The most practical solution for developing dependable, connected, and intelligent products could come through the application of specialty software tools that conform to applicable interoperability standards so that the enterprise level system integrators can seamlessly connect the tools needed for reliability engineering, systems engineering, and knowledge management. This paper mainly presents a potential map of the commonly used reliability engineering tools overlaid on the systems engineering technical processes. The paper adheres to the technical process of systems engineering described by INCOSE [4] and the commonly known tools and processes used in design-for-six-sigma and reliability engineering [5, 6]. The paper recommends including a formal knowledge storage and retrieval system in the closed-loop between systems engineering and reliability engineering so that the details observed in past failures are not missed in future design iterations. RELIABILITY ENGINEERING MEETS SYSTEMS ENGINEERING To enable fast learning cycles that will help identify potential failure modes of complex systems, a seamless enterprise level connection between the systems engineering technical processes, the reliability engineering tools, and a knowledge management system is needed, so that efficient storage and retrieval of failure modes information is possible. To accomplish the development of the enterprise level connection mentioned above, the activities related to the thirteen technical processes of systems engineering [4] have been considered in this work as a higher-level product lifecycle structure. Those thirteen systems engineering technical processes are: • Stakeholders’ Requirements Identification • System Requirements Definition • System Architectural Design • System Elements Definition • System Analysis • System Elements Realization • System Elements Integration • System Design Verification • Verified System Transition • System Performance Validation • System Operation • System Maintenance • System Disposal In the following section the diverse tools used in designfor-six-sigma and reliability engineering that should support the systems engineering technical processes are briefly described. Beyond the next section, each activity associated with the thirteen technical processes of systems engineering are associated with a set of design-for-six-sigma and reliability engineering tools, wherever those tools are likely to have a beneficial impact. The purpose is to present a connection Proceedings of the 2016 Ground Vehicle Systems Engineering and Technology Symposium (GVSETS) Bridging Reliability Engineering & Systems Engineering Page 3 of 6 between the tools used in reliability engineering and designfor-six-sigma, and the main technical process activities of systems engineering that need to be accomplished by enterprise level software systems integration so as to adequately deal with the complexity of designing and delivering smart connected products. Reliability Engineering Tools A wide variety of tools ar