Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Task Group Summary 7 How can we enhance robustness of engineered systems, and how can the methods of engineering analysis be extended to address issues of complexity and management in other fields? Challenge Summary The support of policy, industry, or private decisions involving complex, dynamic systems and uncertainty, is a challenge that presents common fea- tures across different fields. For example, lifecycle risk management in the automotive, space, and medical device industries involves complex physical systems, organizations, and uncertainties that vary with experience (test results, operational data, etc.). Similarly, the maintenance of the heat shield of the US space shuttle involves the physical characteristics of the tiles as well as human and organizational factors (including errors). The methods of engineering analysis can be extended beyond the realm of engineered systems to address issues of complexity and management in other fields. For instance, the design and operation of health care systems include both technological and human factors: how can information and incentives best be managed to enable affordable, quality healthcare, given the complex hierarchical domains involved, with levels ranging from clinical practices to the delivery of care and specific organizations? Another example is the management of the Internet, whose structure and interactions with dif- ferent markets evolve constantly, requiring an understanding of both the network and the complex behaviors of their users. One common thread is the engineering approach that can be adopted for the design and management of such complex systems, with an empha- sis on architecture (structure and functions) and a systematic, coherent treatment of both dynamics and uncertainties. One of the challenges is to build in and preserve robustness and adaptability, accounting for complex 59
60 COMPLEX SYSTEMS interactions among components; for example, to include interfaces and interactions of systems with the medium in which they operate and to anticipate future performance in situ (the human body for medical devices, soil/structure interactions, variations of external parameters of space for satellites, etc.). At another level, these interactions include those between the physical system and its operators (pilots, technicians, doctors), and between these operators and the managers who set the incentives and the information base for the people in charge of operations. The goal is to adapt the methods of engineering systems analysis to other types of complex sys- tems (human, climatic, etc.) in order to support policy decisions before full information has been gathered. Decisions pertaining to the management of design, tests, development and operations can be supported by a combination of systems analysis (stat- ic and dynamic), risk analysis and decision analysis. In addition, methods of economic analysis (including for instance, utility theory, principal-agent models and game theory) allow us to evaluate questions such as incentives as well as issues about budget optimization. Uncertainties are often at the core of the problem. In the context of risk management, one can rely on classical statistics when that information exists and the system is stable enough; but these data are not always available or relevant to all challengesâfor instance in the design stage of new devices. Bayesian probabilities are useful to support risk management decisions, in all phases of a device life (design, testing, approval, operation, and retire- ment). The challenge is to combine the powers of all existing methods to make the best possible use of incomplete information in the management of complex systems, both in industry and in government. Key Questions The key question is: how can the methods of engineering analysis of complex systems be extended to other types of systems (human, biological, physical), medical systems (e.g., anesthesia in operating rooms), threats of terrorist attacks, climatic phenomena, etc.? Problems can arise at the inter- face of engineered systems and the medium in which they operate, or the organizations that manage them. These interactions and the corresponding uncertainties have to be accounted for in a systematic way to support ratio- nal decision making. One focus can be the assessment and management of the risks of system failures and/or of reduced levels of performance based
TASK GROUP SUMMARY 7 61 on concepts of systems analysis, probability, stochastic processes, and eco- nomic analysis. The challenge to the working group is to come up with engineering strategies to address the fundamental problems of information and decision- making associated with the management of complex systems. Required Reading Carlson JN and Doyle J. Complexity and robustness. Proc Natl Acad Sci USA 2002;1:2538- 2545. Overview of the Vatican workshop of 1999. [Accessed online June 10, 2008: http://www. vatican.va/roman_curia/pontifical_academies/acdscien/documents/rc_pa_acdscien_ doc_20000530_survival_en.html.] PatÃ©-Cornell ME. The engineering risk assessment method and some applications. In: W. Edwards, R. Miles, and D. von Winterfeldt (eds.), Advances in decision analysis. New York: Cambridge University Press 2007. Suggested Reading Basole RC, Rouse W. Complexity of service value networks: conceptualization and empirical investigation. Systems Journal 2008;47(1):53-70. Davis JP, Eisenhardt KM, and Bingham CB. Complexity theory, market dynamism and the strategy of simple rules. Stanford University, Department of Management Science and Engineering, 2007. [Accessed online July 31, 2008:http://web.mit.edu/~jasond/www/ complexity.htm.] Murphy DM and PatÃ©-Cornell ME. The SAM framework: a systems analysis approach to modeling the effects of management on human behavior in risk analysis. Risk Analysis 1996;16(4):501-515. PatÃ©-Cornell ME and Fischbeck PS. Probabilistic risk analysis and risk-based priority scale for the tiles of the space shuttle. Reliability Engineering and System Safety 1993;40(3):221- 238. Rouse W. Complex engineered, organizational and natural systems: Issues underlying the complexity of systems and fundamental research needed to address these issues. Systems Engineering. 2007;10(3):260-271. TASK GROUP MEMBERS â¢ Fahmida Chowdhury, National Science Foundation â¢ Jeffrey Cooper, SAIC â¢ Tuan Duong, JPL/CIT â¢ Theirry Emonet, Yale University â¢ James Ferrell, Stanford University
62 COMPLEX SYSTEMS â¢ Panos Papadopoulos, University of California, Berkeley â¢ Michael Leo Parchman, South Texas Veterans Health Care Systems â¢ Vimla L. Patel, Arizona State University â¢ Steven Schiff, Penn State University â¢ Jessika Trancik, Santa Fe Institute â¢ Andreas Wagner, University of Zurich â¢ Joseph Wang, Virginia Polytechnic Institute and State University â¢ Cassandra Brooks, University of California, Santa Cruz TASK GROUP SUMMARY By Cassandra Brooks, Graduate Science Writing Student, University of California, Santa Cruz Human beings have long used engineering principles to solve complex problems, but these systems arenât infallible and increasing their robustness is a pressing concern. With this theme in mind, 11 scientists from different engineering and biological fields met at the 2008 National Academies Keck Futures Initiative Conference on Complex Systems to discuss their assigned question: How can we enhance robustness of engineered systems, and how can the meth- ods of engineering analysis be extended to address issues of complexity and management in other fields? Robustness refers to the ability of a system to preserve itself in response to perturbations. In other words, a robust system is one that can withstand variations with minimal damage or loss of function. Examples are buildings designed to maintain their integrity during an earthquake, power cords with built in surge protectors, and a mammalâs ability to maintain a constant internal temperature in different climes. Specific characteristics generate robustness in a system: redundancy, control systems, distributed robustness, error-correction and hardness. Redundancy is the duplication of critical components that will increase the reliability of a system. Control systems are devices that manage or regulate the system to keep it functioning properly. Distributed robustness means the robustness is spread throughout the system. Any system will fail at its weakest point. Error-correcting systems simply refer to the systemâs ability to detect and fix errors without perpetuating them. Lastly, hardness means over-designing something to make it stronger. For example, an aircraft
TASK GROUP SUMMARY 7 63 that has two engines (with one for back up) is redundant, whereas a bridge designed to remain standing in winds exceeding what engineers expect to see in nature is hardness. As the Task Group (7) discussed various specific engineering fields and broader aspects of biological systems, an underlying theme arose. Biological systems are inherently robust. Gene flow, genetic drift, natural selection, non-random mating, and mutation (the five mechanisms of evolution) result in the most robust of systems because with living organisms, health and proper function must be the norm. The group generated questions spanning biology and engineering and clustered them according to common themes. Why are most engineered systems rigid while biologic systems are soft? Which engineering principles for robustness are applicable to human/social systems? And which engineer- ing principles are not found in biologic systems and vice versa? The group focused on the latter part of the last question, âWhich biological systems are not used in engineering?â to address its ultimate ques- tion, âWhat complex biological behaviors or systems can be applied to solve engineering problems and make engineering systems more robust?â Uncertainty and human error are major problems compromising the robustness of engineering systems. An example would be the maintenance of the heat shield on United States space shuttles, which requires precise engineering as well as human and organizational factors. As we saw in 1993, human error and organizational problems at NASA led to the devastating Challenger explosion. The group began to ponder: Can we engineer a sys- tem to adapt and regenerate despite perturbations caused by human error or other uncertainties? Consider regeneration from a biological perspective. Regeneration, or the replacement of a defective limb, is a terrific example of redundancy in nature. Imagine if engineered systems could adapt to a problem by spontaneously fixing themselves. What if we could engineer a space ship to regenerate broken parts? What if we could somehow manufacture cells that would replace the damaged heat shield in the same way that our skin heals when cut? Once the group hit on this topic, it began free-associating. Could one apply regeneration to automobiles, robotic space probes (e.g. the Phoenix), and space shuttles? A biologist jumped in: how could we design a house that could repaint itself every other spring or replace the shingles on its roof after storms? Could we design roads and highways that would fill their own potholes?
64 COMPLEX SYSTEMS The Questions Seemed Endless Having begun the conversation wondering how to apply engineering concepts to biological systems, the group then asked whether understanding of biological systems could enhance the robustness of engineered systems. Engineers have long looked to biology for inspiration. The sleek and ef- ficient body plan of the bottlenose dolphin has been exhaustively studied by submarine designers. Detailed study of the albatross wing aided aircraft manufacturers, and our newest super-computers attempt to incorporate our limited understanding of neural networks to increase processing speed. Research proposals eluded the group but some felt that the discussions generated enough new ideas for a short perspective that could be published in a scientific journal. The perspective focuses on how engineered systems might learn from biological principles of regeneration to build more com- plex robust systems. Specifically, they are examining modular components at small scales. For example, using a limited number of building blocks (e.g. 21 amino acids) and using those blocks to build something new (e.g. heat shield on a spacecraft).