2012 Journal RIAC Web - 2

The Reliability Information Analysis Center (RIAC) is a DoD Information Analysis Center sponsored by the Defense Technical Information
Center
i RAC
VOLUME 20, NO. 1 JANUARY 2012
JOURNAL
OF THE RELIABILIT Y INFORMATION ANALYSIS CENTER
0 2 F I F T Y Y E A R S O F P H Y S I C S O F FA I LU R E 0 8 P O W E R S Y S T E M P R O G N O S T I C S F O R T H E U. S . A R M Y O H - 5 8 D
H E L I CO P T E R
2 3 N E W M I L I TA R Y H A N D B O O K 1 8 9 C , R E L I A B I L I T Y G R O W T H
MANAGEMENT
24 RIAC ANNOUNCES MA JOR NEW PRODUC T/SERVICE RELEASES 2 8 D E V E LO P M E N T O F T H E W E B - A CC E S S I B L E R E P O S I TO R Y O F

P H Y S I C S - B A S E D M O D E L S ( WA R P )
F I F T Y Y E A R S O F P H Y S I C S O F FA I LU R E
Kaushik Chatterjee, Mohammad Modarres and Joseph B. Bernstein Center for Risk and Reliability Department of Mechanical Engineering University of Maryland College Park This year marks the fiftieth anniversary since Physics of Failure (PoF) was formally conceptualized in the first of a series of symposia in 1962 organized by the Rome Air Development Center (RADC)1 of the US Air Force. The driving force that established this approach to reliability was concerns in the 1940s and 1950s in US military establishments regarding the reliability of electronic systems. However, mechanistic treatment of failures had its roots in the late nineteenth century when, in 1870, A. Wohler summarized fatigue test results on railroad axles, and concluded that cyclic loads are more important for determining life than peak loads. Thereafter, much of the reliability work in the first half of the twentieth century was related to fatigue and fracture of materials (fatigue failure was the main concern during World War I). For example, Basquin (1910) proposed a log-log relationship between stress and life (the so-called S-N) curves using Wohlers fatigue test data. Griffith (1921) introduced his theory of fracture while exploring the strength of elastic brittle materials. Miner (1945) popularized the linear damage hypothesis suggested by Palmgren (1924) as a practical design tool in which the expended fatigue life of metals was empirically modeled. Epstein (1948) published the statistical foundation for assessment of the life of materials subject to fracture. At the start of World War II, it was discovered that over 50% of the airborne electronics equipment in storage was unable to meet the requirements of the Air Core and Navy (McLinn, 2011). In 1950 the US Department of Defense (DoD) initiated an ad hoc group on reliability of electronic equipment, which stated that to improve part reliability it was essential to develop better parts, establish quantitative reliability requirements for parts, and collect field failure data to determine the root cause of problems (Ebel, 1998). However, the formation of the Advisory Group on the Reliability of Electronic Equipment (AGREE) in August 1952 by the DoD is often considered the turning point in modern reliability engineering. The AGREE committee recommended that the high cost of ownership due to low reliability could be controlled by developing reliability test programs using stress tests such as high and low temperature, vibration, and other cyclic environments to understand causes of failures and ways to correct them. It also recommended the development of a reliability demonstration program in terms of the estimated mean life of equipment and the confidence associated with such estimate. The reliability techniques recommended by AGREE were accepted by the DoD, and later by NASA and many other organizations supplying high technology equipment. Thereafter, several conferences began in the 1950s to focus on various reliability topics. One conference that warrants special mention is the Holm Conference on Electrical Contacts, begun in 1955, which emphasized reliabil-
1 RADC became Rome Laboratory (RL), but is now known as Air Force Research Laboratory (AFRL) Information Directorate. * Also a Professor at the Bar Ilan University, Ramat Gan, Israel.
THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER
JANUARY 2012
ity physics. This conference established itself over the years as the primary source of reliability physics information on connectors. Reliability work related to fatigue and fracture of materials continued through the 1950s and early 1960s. For example, in 1957 G.R. Irwin proved that the fracture of materials was due to plastic deformation at the crack tip and generalized Griffiths theory (Irwin, 1957). Between 1955 and 1963, Waloddi Weibull produced several publications related to modeling of fatigue and creep mechanisms in which methods for evaluating fatigue and creep failure data were discussed (Weibull, 1959). In 1961, Weibull published a book on materials and fatigue testing while working as a consultant for the US Air Force Materials Laboratory (Weibull, 1961). Another important milestone was the introduction of methods for predicting the rate of growth of fatigue cracks building on Irwins work on stress intensity factor (Paris, 1961). Against the backdrop of the developments in mechanistic-based life models (particularly in the assessment of fatigue and fracture failures) and the AGREE recommendations, RADC introduced a PoF program in 1961 to address the growing complexity of military equipment and the consequent higher number of failures observed. In 1962, researchers from Bell Labs published a paper on High Stress Aging to Failure of Semiconductor Devices that justified using the kinetic theorys interpretation of the Arrhenius equation, a simple yet accurate formula for the temperature dependence of the reaction rate constant as a basis for assessment of temperatureinduced aging of semiconductor devices (Dodson and Howard, 1961). Later, RADC and the Armour Research Foundation of the Illinois Institute of Technology (now IIT Research Institute) organized the first PoF symposium in electronics in September 1962 in Chicago. This symposium laid the groundwork for future research and development activities related to PoF by RADC and several other organizations. Numerous original papers and ideas introducing and explaining the PoF concepts and methods were presented in these symposia. In one of the original PoF papers presented at the first PoF symposium, Vaccaro (1962) opined that PoF should seek to relate the fundamental physical and chemical behavior of materials (i.e., failure mechanisms) to reliability parameters. This approach is based on the principle that to eliminate the occurrence of failures, it is essential to eliminate their root causes, and to do that one must understand the physics of the underlying failure mechanisms involved. Davis (1962) described the need for identifying probable failure mechanisms by which components fail as a function of time, environmental and operating stresses. He also developed mathematical models to represent these mechanisms in order to meet reliability requirements of components. Various companies and universities conducting research on failure mechanisms were active participants. These included Raytheon, Syracuse University, and Motorola. Although PoF was key to improving the design and reliability of components, higher costs in terms of facilities and manpower were identified as the critical impediments to using PoF at that time (Ryerson, 1962). The various key elements of the PoF approach, such as identification of the failure mode, mechanism, and cause, were defined for the first time in this symposium (Zierdt, 1962; Earles and Eddins, 1962).
Due to the success of the first symposium in 1962, four PoF symposia were held in subsequent years (until 1966), with many more papers describing concepts related to PoF. For example, Tamburrino (1963) provided key points about the requirements of a reliability physics program: e.g., materials, measurement techniques, and failure mechanisms. The need for part vendors to be kept abreast of available knowledge in failure physics was identified. It was also stated that any changes in pre-established part processing or fabrication could potentially be a key factor in inducing new failure mechanisms, and should be closely coordinated with reliability engineers. Bretts, et al. (1963) provided accelerated test results for resistors, which they correlated with physical degradation models to estimate time to failure. PoF was identified as an essential step in planning accelerated tests as well as evaluating them. In the third PoF symposium, Ingram (1964) described performance characteristics and failure mechanisms of a device in probabilistic terms. He suggested, Environmental and stress conditions applicable to the device, and its performance and strength characteristics, are expressed in the form of multidimensional probability distributions. By joint evaluation of these probability distributions, a quantitative estimate of the reliability of the device can be obtained. Beau (1964) described methods for managing and assessing the role of the human elements in PoF. He described three classical causes of failure: reliability limitation inherent in the design; reliability degradation caused by the factory process; and reliability degradation caused by the user. According to him, the factory operator, in the form of poor workmanship or operator error, introduced the human element in reliability of devices. Workman (1964) described the failure analysis practices followed by Texas Instruments at that time, and the need for incorporating information gained from failure analysis in new reliability test design, process control, and device design. Shiomi (1965) introduced a generalized cumulative degradation model for estimation and prediction of component life under successive different stress levels. Partridge, et al. (1965) argued that vendors supplying semiconductor parts and the parts themselves should be screened based on engineering evaluations that rate the parts reliability. They further stated that qualification tests alone were insufficient to determine the ability of vendors to provide reliable parts, but production procurement data from screening and burn-in could provide sufficient vendor history. Church and Roberts (1965) presented different causes of failure of a component, such as failure due to accidental damage during manufacture, assembly, testing, storage, or failure in service due to operating conditions or failure of another component. Thomas (1966) used basic concepts of dimensional analysis to make a general examination of mathematical models, e.g., Eyrings equation. He opined that the concepts of signal, noise and dimensionless variables could be used to formulate mathematical models, physical laws, and probability distributions. Schenck (1966) presented two forms of progressive failure mechanisms of a commercial silicon diode, and studied them as a function of various stress and measurement variables. Several papers were also presented that proposed nondestructive inspection and screening procedures based on PoF, which later formed the basis for prognosis and health management continued on next page http://theRIAC.org
FIFTY YEARS OF PHYSICS OF FAILURE
continued from page 3 et al. (1990) advocated use of the PoF approach for reliability assessment as opposed to the part count technique. Dasgupta and Pecht (1991) published a series of tutorial papers to review important material failure mechanisms and damage models. Engel (1993) presented failure models for mechanical wear modes and mechanisms. Cushing, et al. (1993) of the U.S. Army Material Systems Analysis Activity (AMSAA) compared empirically-based reliability prediction approaches (e.g., MIL-HDBK-217) with the PoF approach. They identified several limitations of MIL-HDBK-217 that could be addressed using the PoF approach. Although studies related to PoF were published through the 1990s and 2000s, a trend towards probabilistic2 consideration of PoF also emerged from the early 1990s. For example, Hu, et al. (1991) presented a probabilistic approach for predicting thermal fatigue life of wire bonding in microelectronics. Mendel (1996) formally described probabilistic PoF (PPoF) as a technique in which the statistical lifetime model is derived considering the PoF, and presented a case for applying PPoF in design for reliability. Later, Modarres, et al. (1999) also emphasized that prediction of failure is inherently a probabilistic problem due to uncertainties associated with PoF models and their parameters, and with failure-inducing agents that can result from changes in environmental, operating, and use conditions. Several publications related to the PPoF then appeared from the early 2000s. For example, Haggag, et al. (2000) presented a PPoF approach to reliability assurance of high-performance chips that considered common defect activation energy distribution. Hall and Strutt (2003) presented PPoF models for component reliabilities by considering parameter and model uncertainties. Azarkhail and Modarres (2007) presented a Bayesian framework for physics-based reliability models. Matik and Sruk (2008) highlighted the need for PoF to be probabilistic in order to consider variations of variables involved in processes contributing to the occurrence of failures. Chatterjee and Modarres (2011) presented a PPoF approach for estimating tube rupture frequency in advanced nuclear plant steam generators that considered the PoF and various uncertainties associated with environmental conditions, geometrical and material properties, PoF models, and model parameters. Another important milestone in the discussion on PoF was the publication in 2008 of the Physics-of-Failure Based Handbook of Microelectronic Systems (Salemi et al., 2008) through the Reliability Information Analysis Center (RIAC). The handbook was the first of its kind to present an approach for microelectronic system reliability assessment and qualification based on PoF in a sum-of-rate approach to account for multiple mechanisms. Another important activity currently underway is the development of the Web Accessible Repository of Physics-Based Models (WARP) under the aegis of the RIAC3. The objective of WARP is to collect and analyze the characteristics of important PoF models for electronic, electromechanical and mechanical components in order to provide a centralized webbased repository accessible to researchers and engineers4.
2 Probabilistic consideration of mechanistic life-models had its roots in the 1940s even before the term PoF was introduced, when Epstein (1948) presented the statistical foundation for life assessment of materials failing by fracture. 3 http://theriac.org 4 Chatterjee, Modarres and Christou are directly participating in the development of WARP
concepts. For example, Gill and Workman (1966) presented a reliability screening procedure (consisting of destructive tests and nondestructive inspections) based on identifying failure mechanisms resulting from high-stress tests and failure analysis. Potter and Sawyer (1966) presented an optical scanning-based nondestructive technique to study various semiconductor device phenomena and identify causes of anomalous device behavior in order to improve device reliability. From 1967 on, IEEE sponsored the Reliability Physics Symposium (IRPS) to present a wide range of PoF related research. For example, Ryerson (1967) presented mathematical models for semiconductor diodes illustrating how failure mechanisms, part strengths, and application stresses interact and affect the failure rate of component parts. Keen et al. (1967) presented mechanisms of failure in ohmic and expanded contacts, from metal-semiconductor contacts and bonds to metallization in semiconductor devices. Payne (1967) presented a failure mechanism for barium titanate capacitors by studying the PoF. Frankel and Kinsolving (1970) discussed the need for reliability testing of components for hostile environments, by first simulating field conditions and then developing accelerated laboratory conditions. Hollingshead (1970) introduced a technique for optimizing the selection of parts for system application by reliability and quality levels through systematizing the compilation and processing of necessary data. The comparative influences of performance parameters such as repair cost, storage time, and cost of failure were discussed. Schwuttke (1970) showed that peripheral yield loss in silicon wafers can be minimized whenever temperature gradients arising during cooling of a row of wafers are eliminated. The IEEE IRPS continued to disseminate a plethora of knowledge on PoF through the 1970s and 1980s. Several failure mechanisms and mathematical models were reported for a wide range of electronic components such as capacitors, semiconductors, resistors, and interconnects. Metallization, metallurgical effects and bonding dominated the key presentations and papers published by IRPS. For example, Black (1974) presented a model for predicting electromigration time to failure. Macpherson, et al. (1975) introduced the concept of fast temperature cycling as a key agent of failure in transistor metallization. Crook (1979) presented a model for time dependent dielectric breakdown (TDDB) of semiconductors as a function of operational and environmental conditions and the devices physical parameters. Lloyd (1983) presented the initial analysis of electromigration effects in multilevel geometries. Hieber and Pape (1984) presented a creep-rupture equation that calculates time to rupture as a function of applied mechanical load and temperature. Chen, et al. (1985) presented a quantitative breakdown model for thin gate and tunneling oxides based on the physical understanding of oxide breakdown. Christou, et al. (1985) presented reliability investigation results for high electron mobility transistor devices. Conrad, et al. (1988) presented a methodology to monitor and predict early life reliability failure mechanisms. While there was a noticeable decline in PoF techniques and uses during this period, the IRPS continued PoF related publications until a resurgence of interest in PoF began in the 1990s that has continued until today. By the late 1980s and early 1990s, several publications on PoF-related research outside of the IEEE IRPS also appeared. For example, Pecht,
JANUARY 2012
As we celebrate fifty years of PoF, RADC deserves special recognition, including its chief and founder Joseph J. Naresky, under whose leadership PoF was first formally conceptualized in the symposium on Physics of Failure in Electronics organized in 1962, with considerable contributions by RADCs Joseph Vaccaro. It is remarkable that many of the original ideas introduced in these symposia continue to have a significant impact on todays understanding of failures in electronics and have offered enduring models for estimating life characteristics. As was observed for the first time in the 1962 symposium, the PoF approach encompasses multiple disciplines, such as reliability engineering, physics, metallurgy, mathematical statistics and probability. The symposia of the 1960s provided PoF approaches for non-destructive test methods and for improving and predicting component reliability with limited access to mass test data. While PoF analysis is complex and costly to apply, it provides the strongest characterization available of reliability of components, structures and systems. As an approach for reliable product development, PoF has gained wide acceptance today in the commercial sector (e.g., at Microsoft), as well as in several countries (e.g., Japan, Singapore, and Taiwan)a tribute to its strong foundation established fifty years ago.
monitor and predict early life reliability failure mechanisms, Proceedings of 26th Annual Reliability Physics Symposium, pp: 126-130, 1988 Crook, D.L., Method of determining reliability screens for time dependent dielectric breakdown, Proceedings of 17th Annual Reliability Physics Symposium, pp: 1-7, 1979 Cushing, M.J., Mortin, D.E., Stadterman, T.J., and Malhotra, A., Comparison of Electronics-Reliability Assessment Approaches, IEEE Transactions on Reliability, Vol. 42, Issue 4, December 1993 Dasgupta, A., and Pecht, M., Material failure mechanisms and damage models, IEEE Transactions on Reliability, Vol. 40, Issue 5, pp: 531-536, December 1991 Davis, H., Introduction, Proceedings of the First Annual Symposium on Physics of Failure in Electronics, September 26-27, pp: 1-3, 1962 Dodson, G.A., and Howard, B.T., High stress aging to failure of semiconductor devices, Proceedings of Seventh National Symposium on Reliability and Quality Control, Philadelphia, PA, January 1961 Earles, D.R., and Eddins, M.F., Reliability physics (the physics of failure), Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 179-193, September 26-27, 1962 Ebel, G.H., Reliability physics in electronics: a historical view, IEEE Transactions on Reliability, Vol. 47, Issue 3, September 1998 Engel, P.A., Failure models for mechanical wear modes and mechanisms, IEEE Transactions on Reliability, Vol. 42, Issue 2, June 1993 Epstein, B., Statistical Aspects of Fracture Problems, Journal of Applied Physics, Vol. 19, February 1948 Frankel, H., and Kinsolving, W., Reliability testing for hostile environments, Proceedings of Eighth Annual Reliability Physics Symposium, pp: 219, 1970 Fischer, A. H., et al., Experimental Data and Statistical Models for Bimodal EM Failures, in 38th Annual Proceedings of IRPS, pp: 359-363, 2000 Gill, W., and Workman, W., Reliability screening procedures for integrated circuits, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 101-140, 1966 Griffith, A. A. The phenomena of rupture and flow in solids, Philosophical Transactions of the Royal Society, Vol. 221, pp: 163-198, 1921 Haggag, McMahon, Hess, Cheng, Lee, and Lyding, A Probabilistic-Physicsof-Failure/Short-Time-Test Approach to Reliability Assurance for High-Performance Chips: Models for Deep-Submicron Transistors and Optical Interconnects, Proceedings of IEEE Integrated Reliability Workshop, pp: 179-182, October 23-26, 2000 Hall, P.L., and Strutt, J.E., Probabilistic physics-of-failure models for component reliabilities using Monte Carlo simulation and Weibull analysis: a parametric study, Reliability Engineering & System Safety, Vol. 80, Issue 3, pp: 233-242, June 2003 Hieber, H., and Pape, K., Lifetime of bonded contacts on thin film metallization, Proceedings of 22nd Annual Reliability Physics Symposium, pp: 128-133, 1984 Hollingshead, C.O., A system oriented components selection optimization technique, Proceedings of Eighth Annual Reliability Physics Symposium, pp: 220-225, 1970 Hu, J.M., Pecht, M., and Dasgupta, A., A Probabilistic Approach for Predicting Thermal Fatigue Life of Wire Bonding in Microelectronics, Journal
References
Azarkhail, M., and Modarres, M., A Novel Bayesian Framework for Uncertainty Management in Physics-Based Reliability Models, Proceedings of the ASME International Mechanical Engineering Congress and Exposition, November 11-15, 2007, Seattle, WA Basquin, O. H. The Exponential Law of Endurance Tests, Proceedings of American Society of Testing Materials, Vol. 10, pp: 625-630, 1910 Beau, J.F., Management of the human element in the physics of failure, Proceedings of Third Annual Symposium on the Physics of Failure in Electronics, pp: 264-279, 1964 Black, J.R. Physics of electromigration, Proceedings of 12 Annual Reliability Physics Symposium, pp: 142-149, 1974
th
Bretts, G., Kozol, J., and Lampert, H., Failure physics and accelerated testing, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 189-207, September, 1963 Chatterjee, K., and Modarres, M., A probabilistic physics-of-failure approach to prediction of steam generator tube rupture frequency, Proceedings of the International Topical Meeting on Probabilistic Safety Assessment and Analysis, March 13-17, 2011, Wilmington, NC Chen, I.C., Holland, S. and Hu, C., A quantitative physical model for timedependent breakdown in SiO2, Proceedings of 23rd Annual Reliability Physics Symposium, pp: 24-31, 1985 Christou, A., Tseng, W., Peckerar, M., Anderson,W.T., McCarthy, D.M., Buot, F.A., Campbell, A.B., and Knudson, A.R., Failure Mechanism Study of GaAs MODFET Devices and Integrated Circuits, Proceedings of 23rd Annual Reliability Physics Symposium, pp:54-59, 1985 Church, H.F., and Roberts, B.C., Failure mechanisms of electronic components, Proceedings of Fourth Annual Symposium on the Physics of Failure in Electronics, pp: 156-178, 1965 Conrad, T.R., Mielnik, R.J., and Musolino, L.S., A test methodology to
continued on next page http://theRIAC.org
FIFTY YEARS OF PHYSICS OF FAILURE
continued from page 5

Potter, C.N., and Sawyer, D.E., Optical scanning technique for semiconductor device screening and identification of surface and junction phenomena, Proceedings of Fifth Annual Symposium on the Physics of Failure in Electronics, pp: 37-50, 1966 Ryerson, C.M., Project control to provide for the physics of failure in electronics, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 68-72, September 26-27, 1962 Ryerson, C.M., Mathematical Modeling For Predicting Failure Rates Of Component Part, Proceedings of IEEE Sixth Annual Reliability Physics Symposium, pp: 10-15, 1967 Salemi, S., Yang, L., Dai, J., Qin, J., and Bernstein, J.B., Physics-of-Failure based handbook of microelectronic systems, Reliability Information Analysis Center, Utica, NY, 2008 Schenck, J.F., Progressive failure mechanisms of a commercial silicon diode, Proceedings of Fifth Annual Symposium on the Physics of Failure in Electronics, pp: 18-35, 1966 Schwuttke, G.H., Yield problems in LSI technology, Proceedings of Eighth Annual Reliability Physics Symposium, pp: 274-280, 1970 Shiomi, H., Cumulative degradation model and its application to component life estimation, Proceedings of Fourth Annual Symposium on the Physics of Failure in Electronics, pp: 74-94, 1965 Tamburrino, A.L., Analysis of requirements in reliability physics, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 189-207, September, 1963 Thomas, R.E., Some unifying concepts in reliability physics, mathematical models, and statistics, Proceedings of Fifth Annual Symposium on the Physics of Failure in Electronics, pp: 1-17, 1966 Vaccaro, J., Reliability and Physics of Failure Program at RADC, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 4-10, September 26-27, 1962 Weibull, W., Statistical evaluation of data from fatigue and creep rupture tests, Part I: fundamental concepts and general methods, Wright Air Development Center, Technical Report 59-400, Sweden, September 1959 Weibull, W., Fatigue testing and analysis of results, Pergamon Press, London, 1961 Workman, W., Failure analysis techniques, Proceedings of Third Annual Symposium on the Physics of Failure in Electronics, pp: 238-263, 1964 Zierdt, C.H., Diagnostic techniques in semiconductor device stress response analysis, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, pp: 91-97, September 26-27, 1962
of Electronic Packaging, Vol. 113, Issue 3, pp: 275-285, 1991 Ingram, G.E., Prediction of Device Reliability by Mechanisms-of-Failure Principles, Proceedings of Third Annual Symposium on the Physics of Failure in Electronics, pp: 200-209, 1964 Irwin G. R., Analysis of stresses and strains near the end of a crack traversing a plate, Journal of Applied Mechanics, Vol. 24, pp: 361364, 1957 Keen, R.S., Loewenstern, L.R., Schnable, G.L., Mechanisms of Contact Failure in Semiconductor Devices, Proceedings of IEEE Sixth Annual Reliability Physics Symposium, pp: 216-233, 1967 Lloyd, J.R., Electromigration-induced extrusions in multi-level technologies, Proceedings of 21st Annual Reliability Physics Symposium, pp: 208210, 1983 Macpherson, A.C., Weisenberger, W.H., Day, H.M., and Christou, A., Effects of Fast Temperature Cycling on Aluminum and Gold Metal Systems, Proceedings of 13th Annual Reliability Physics Symposium, pp: 113-120, 1975 Matik, Z., and Sruk, V., The physics-of-failure approach in reliability engineering, Proceedings of IEEE International Conference on Information Technology Interfaces, pp: 745-750, June 23-26, 2008 McLinn, J., A short history of reliability, The Journal of the Reliability Information Analysis Center, January 2011 Mendel, M., The case for probabilistic physics of failure, Chapter in Reliability and Maintenance of Complex Systems, Edited by Ozekici, S., Springer, 1996 Miner, M.A., Cumulative Damage in Fatigue, Journal of Applied Mechanics, Vol. 12, No. 3, pp: A- l59- l64, September, 1945 Modarres, M., Kaminskiy, M., and Krivtsov, V., Reliability engineering and risk analysis: A practical guide, Marcel Dekker, New York, 1999 Naresky, J.J., Foreword, Proceedings of First Annual Symposium on the Physics of Failure in Electronics, September 26-27, 1962 Palmgren, A., Durability of ball bearings, ZDVDI, Vol. 68, Issue 14, pp: 339, 1924 (in German) Paris, P.C., Gomez, M. P. and Anderson W.E., A rational analytic theory of fatigue, The Trend in Engineering, Vol. 13, pp: 9-14, 1961 Partridge, J., Hall, E.C., and Hanley, L.D., The application of failure analysis in procuring and screening of integrated circuits, Proceedings of Fourth Annual Symposium on the Physics of Failure in Electronics, pp: 96-139, 1965 Payne, D.A., Concerning the physics of failure of Barium Titanate capacitors, Proceedings of IEEE Sixth Annual Symposium on Reliability Physics, pp: 257-264, 1967 Pecht, M., Dasgupta, A., Barker, D., Leonard, C.T., The reliability physics approach to failure prediction modeling, Quality and Reliability Engineering International, Vol. 6, Issue 4, pp: 267-273, September/October 1990
THE APPEARANCE OF PAID ADVERTISING IN THE RIAC JOURNAL DOES NOT CONSTITUTE ENDORSEMENT BY THE DEPARTMENT OF DEFENSE OR THE RELIABILITY INFORMATION ANALYSIS CENTER OF THE PRODUCTS OR SERVICES ADVERTISED*
*
Paid advertising appears on pages 22 and 34.
JANUARY 2012
Introduction to Maintainability Engineering

On-Line Training Course
Course Description:
This course was developed by the RIAC as an introduction to maintainability engineering concepts and practices. It begins with a brief overview of the RIAC, and because maintenance demand is driven by system reliability, provides an overview of basic reliability engineering concepts before proceeding into the core maintainability topics. Section 3 begins the core content by covering basic maintainability concepts and definitions, the need for maintainable systems and the elements of a comprehensive maintainability program. Section 4 covers the basic mathematical foundations of maintainability by describing various probability distributions that are commonly used to model system maintainability. Section 5 describes establishing maintainability requirements at the system level as well as at lower levels of design. These include establishing quantitative requirements, such as mean time to repair (MTTR), as well as techniques such as quality function deployment (QDF) that are used to ensure user requirements are met. Section 6 delves into maintainability design techniques such as allocations, predictions, human engineering and standardization. Section 7 covers maintainability verification approaches, including inspection, analytical methods and testing. Section 8 describes trending methods to track and improve maintainability over time and Section 9 discusses ensuring maintainability is not degraded by production decisions. Section 10 provides a review and wrap-up of the course. The course also includes two quizzes to reinforce the concepts covered.
Get instant online access for only $
299
Features & Benefits:

X X X 3 Days of live recorded feed including quizzes and demos, totaling over 14 Hours Synchronized PowerPoint presentations with video for maximum learning content On-Line training lets you learn at your own pace, anywhere, anytime, and save lots of money on travel expenses
View the online demo at: http://theRIAC.org
http://theRIAC.org
P O W E R S Y S T E M P R O G N O S T I C S F O R T H E U. S . A R M Y O H - 58 D H E L I CO P T E R
Jeffrey Banks, Todd Batzel, Robert Keolian, Matt Poese, Terrance Lovell, Mitch Lebold, and Karl Reichard, Applied Research Laboratory at The Pennsylvania State University State College Kevin Cunningham, Bell Helicopter Textron
Abstract
The OH-58D Kiowa Warrior helicopter has been a workhorse for the U.S. Army for decades and is projected to continue to accrue flight hours for years to come as a highly capable platform applied against various mission profiles. The U.S. Army is interested in the implementation of condition based maintenance (CBM) for this platform to increase operational availability of the aircraft, reduce the required number of maintenance activities and increase the inspection interval period. The CBM methodology and these objectives are directly dependent upon the capability of the health and usage monitoring system and its ability to detect diagnose and provide an estimate of remaining useful life (RUL). The Applied Research Laboratory at The Pennsylvania State University (ARL Penn State) has developed prognostic technologies for helicopter electrical power systems that can be integrated into an existing on-board Health and Usage Monitoring System (HUMS). The goal of the program was to develop a capability to monitor, detect and provide an estimated remaining useful life RUL for the starter/generator, battery and power inverter before they lead to electrical power system failures. The focus of this effort was to develop technologies that provide actionable prognostic information for forecasting part replacement and to extend existing time and usage-based maintenance intervals.
1. Introduction
ARL Penn State in collaboration with Bell Helicopter developed advanced diagnostic and prognostic health management technologies for helicopter electrical power system components that can be integrated into existing on-board Health and Usage Monitoring Systems (HUMS). The program was supported through the Aviation Applied Technology Directorate (AATD) Operations Support and Sustainment Technology (OSST) program that was jointly funded by AATD and Bell Helicopter. The objective rotorcraft platforms for this program were the OH-58D Kiowa Warrior and the Bell 407, although this paper will focus on the OH-58D platform1. The goal of the program was to develop an embedded capability to monitor, detect, fault isolate and provide a remaining useful life estimate for starter/generator, battery and power inverter faults. The intent was to develop technologies that provide actionable advanced diagnostic and prognostic information to helicopter maintainers for forecasting part replacement and to extend existing maintenance intervals. The development of helicopter electrical power system health management technologies involved five phases. The first phase in the process for all of the technology areas consisted of conducting a failure modes, and effects analysis (FMEA) for the OH-58D Kiowa
1 978-1-4244-7351-9/11/$26.00 2011 IEEE. IEEEAC paper #1654, Version 2, Updated October 26, 2010 Article is reprinted with permission of the authors.
JANUARY 2012
Warrior (sufficient information was not available for a separate analysis of the Bell 407 helicopter). The second phase consisted of conducting operational helicopter field tests to collect data to enable a better understanding of the power system function and dynamics. This data was useful for test bed design, algorithm development and platform integration. The third phase in the process involved the design of technology development approaches that included the creation of component models, evaluation of the application of sensors, data processing requirements, building of test benches, laboratory data gathering, and algorithm development. The fourth phase of the effort consisted of the application of the developed technologies to laboratory test benches for technology testing, validation and refinement. The fifth phase involved the design and development of embedded hardware/software and a health management information interface.
and functions as a back-up emergency source of DC power during flight operations in the event of a DC generator failure. The objective of the battery effort was to develop an accurate State of Charge (SOC) prediction capability for the NiCd battery. The FMEA results for the battery did not result in comprehensive indications of the battery failure modes due to the lack of battery failure post-mortem information. Information from the battery manufacturer and laboratory test data indicated that the battery health management system should be designed to regularly measure accurate SOC during helicopter operation in order to predict when the battery maintenance procedure should be implemented.
Power Inverter FMEA Results

The power inverter provides emergency backup power to the 115 VAC Essential Bus in the event of a AC generator failure. The inverter receives DC power from the Battery-Generator Bus. The objective of the power inverter effort was to develop a diagnostic and potentially a predictive capability for the failure modes that have the highest probability for detection. The FMEA results for the inverter only provided a very limited set of failure data to assess the most dominant failure modes, but based on the interviews and anecdotal evidence from Tobyhanna Army Depot it was determined that the current source inductor (CSI), bulkhead transistors and the circuit card are the dominant failure modes. It was determined that extensive modification to the circuit card would be required to conduct diagnostics for this device, so based on engineering judgment this was not pursued. The primary focus of the effort was to develop a diagnostic and potentially predictive technology for the CSI and bulkhead transistors as a lumped component using external sensors that were applied to the input and output cable assemblies. Additionally, the approach developed for the CSI and bulkhead transistors was applied to the output resonant tank (ORT).
2. Power System FMEA Results

In order to provide a focus for the technology development effort, a FMEA was conducted for each power system component. The activities performed for this analysis included: Acquired maintenance data and conducted interviews with the OH-58D contract maintenance personnel at the Aviation Center Logistics Command at Fort Rucker. Conducted interviews with Kiowa Warrior Maintenance Test Flight Examiner (United Stated Army Safety and Standardization Directorate (USASSD)) tasked to inspect maintenance operations in all U.S. Army OH-58D units. Conducted interviews with Army employee who repairs Kiowa Warrior inverters at Tobyhanna Army Depot. Based on the FMEA results for each power system component, each health management technology development team focused their development on the most critical failure modes.
3. Operational Helicopter Field Test

An operational helicopter field test was conducted at Fort Rucker on an OH-58D Kiowa Warrior helicopter as shown in Figure 1 to assess the electrical power system signal and noise environment.
Starter-Generator FMEA Results

The starter-generator is an integrated dual purpose device that provides the ability to start the helicopter turbine engine with on-aircraft battery power or external auxiliary power. The generator provides the ability to continuously create DC electrical power on-aircraft during flight operations. This power is provided to the Battery-Generator Bus and the 28VDC Essential Bus that supports various avionics and electronic systems. The FMEA results for this device only provided a limited set of failure data to assess the most dominant failure modes. Based on the analysis and anecdotal evidence it was determined that brush wear was the most critical failure mode. Other failure modes such as rotor and stator short cicuit and open circuit conditions were found to occur significantly less often and were not considered critical failure modes. The ability to detect these failure modes was explored and described in this paper but the focus of the effort was to develop a prognostic technique for brush wear.
Battery FMEA Results

The vented aviation nickel-cadmium (NiCd) helicopter battery provides power to the starter-generator for turbine engine starting
Figure 1 Field Test on a OH-58D Kiowa Warrior Helicopter
continued on next page http://theRIAC.org
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER The helicopter power system was instrumented with 20 sensors including current, voltage and vibration transducers as well as a tachometer for measuring engine speed. Data was gathered at a sample rate of 50 kHz per channel and for durations of up to 80 seconds. The focus of the data gathering was to acquire data for two modes of operation. The first was to gather data when the power buses were energized and the avionic systems were turned on. During this mode the engine was not running and the battery was providing bus power. The second mode consisted of gathering data during the engine start period through a full engine RPM/flat pitch operating state. This test was conducted multiple times both from a battery start and a ground power start configuration. The data that was generated provided information about the power system levels and dynamics that were useful for the technology development. The data in Figure 2 shows the time sequence and amplitude range for the generator current during the engine start sequence.
maintenance technologies. In addition, this modeling allowed for the advance selection of sensors, initial algorithm design, and an estimate of data processing requirements. For the starter-generator, the modeling consisted of a magnetic finite element analysis of a similar (or surrogate) starter-generator since the construction details of the objective machine were not available. From machine dimensions, coil design and coil configuration, a time-stepping magnetic finite element analysis was performed. At each time step, the FEA yields the magnetic flux vector at every location in the machine as depicted in Figure 3.
Figure 3 Magnetic FEA Flux Vector Results at a Single Time-Step Figure 2 Generator Current Data during a Start Procedure The DC characteristics of this data were used to design the test bench for the generator to emulate the characteristics of the helicopter. The AC characteristics of this data were used to facilitate the development of the fault detection algorithms. An understanding of the high frequency dynamics provided information that was useful for the development of current signature analysis fault detection techniques. The field testing also helped determine the effectiveness of the sensors that potentially will be used for conducting health monitoring on the helicopter. The magnetic flux and the machine state variables at that time step allows calculation of all coil voltages. Once the induced (due to mechanical motion) coil voltages are determined, the rotor position dictates the electrical circuit configuration (which is brush position dependent). The resulting circuit configuration is then solved using standard circuit analysis techniques. In summary, the FEA calculation and subsequential circuit analysis performed at a progression of time steps yields armature voltage, armature current, field voltage, field current, and even quantities that would not be feasible to measure on a physical system such as the current in an individual coil. Using the FEA model, it is very straightforward and relatively fast to perform the time-stepping analysis for a starter/generator with various degrees of winding failures or brush/commutator defects. The FEA simulations were performed on the surrogate machine for normal (baseline) operation as well as a variety of winding and brush/commutator failure modes. Scenarios evaluated included: Field winding short circuit Armature winding faults Brush resistance increase (uneven wear) Commutator bar short circuit Commutator bar open (bad segment)
4. Health Management Detection and Prognostic Technology Development

The health management technology development for each power system component will be discussed separately by component.
Starter-Generator Health Management Technology

The process for starter-generator diagnostic and prognostic technologies consisted of developing and evaluating models of the machine. This initial modeling allowed for the identification of potential observables that can be used to implement predictive
JANUARY 2012
Each of these scenarios was evaluated at various operating speeds and load currents to determine observable quantities that are indicative of the failure mode. From these studies, several observables were identified including: transfer impedance, which is defined as the induced (internal) armature voltage per unit of field current (E0/ Ifld), field current at the pole-passing frequency, and the component of the field current at the commutation frequency. A summary of the three identified observables and how they are affected by various faults as determined from the FEA studies is included in Table 1. Table 1 Summary of Fault Effects on Observables as Determined from FEA Studies
current at the pole-passing frequency and that the field current at the commutator frequency showed potential for detecting brush sparking, which is a parameter that is useful for monitoring brush wear. The next step in the starter-generator technology development process was to conduct seeded fault tests on the helicopter startergenerator. Since the FMEA determined that brush wear was the most dominant and critical failure mode, the testing was focused on the development and validation of a brush wear prognostic technique. To quantify brush wear, the starter-generator was coupled to a spindle motor, as shown in Figure 4, and to a variable electrical load bank.
The model analysis indicated that the transfer impedance provides a detection capability for all six of the fault types. It was also determined that the field current at the pole-passing frequency is also a very useful parameter for identifying various armature and commutator related faults. The second step in the starter-generator diagnostic and prognostic technology development was to conduct laboratory seeded fault testing on a surrogate DC machine to confirm the simulation results. The faults seeded into the machine include shorted field windings, open/short circuit armature windings, open and shorted commutator segments, brush surface area reduction, and increased brush sparking. A summary of the three identified observables and how they are affected by various faults as determined from the seeded fault tests is included in Table 2.
Figure 4 Prime Mover (left) and Starter-Generator (right) The armature load could be varied from no load through a rated current of 200 A, with speeds ranging from 7,000 to 13,000 RPM. The brush wear during the test was determined from initial and end-of-test brush length measurements. Given the nearly constant brush pressure provided by the brush springs, the wear rate was found to be approximately linear with respect to load current and rotational speed. It should also be noted that the wear rate increases with sparking at the brush/commutator interface. An increase in sparking activity has been shown in previous literature [1] to increase wear rate by a wear rate factor. The combination of the wear rate and the wear rate factor leads to an algorithm to determine brush wear rate. The algorithm, shown in block diagram form in Figure 5 utilizes several sensor inputs to track the brush wear based on observed operating conditions and sparking levels.
Reliability Modeling
Table 2 Summary of Fault Effects on Observables to Reliability Prediction, Assessment and Estimation The RIAC Guide as Determined from Seeded Fault Tests
Figure 5 Brush Wear Prognostic Algorithm The brush wear tracking algorithm uses speed and armature current from the wear rate algorithm in combination with the observed sparking index to obtain the wear rate in inches per hour. The spark index is determined from the component of the field current at the commutation frequency. Using previously published visual spark continued on next page http://theRIAC.org
The results of the seeded fault testing indicated that all six of the fault types can be detected using the transfer impedance parameter, that armature and commutator shorts can be detected using field
11
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER index identifiers [1], the observed brush sparking level and corresponding field current at the commutation frequency was used to relate field current at the commutation frequency to a wear rate factor. That is, field current measurements and analysis are used to determine the wear rate factor. The wear rate is integrated to obtain the cumulative brush wear in inches. Given the initial conditions (new brush length in inches) the remaining brush percentage can be determined. As long as the proper initial conditions are established (i.e. when new brushes are installed) the algorithm will provide the maintainer estimates of remaining useful brush life as shown in Figure 6.
the terminal voltage that the manufacturer uses to define empty for a 20 cell sintered plate vented NiCd battery like the SP-170A (20 cells times 1.0 V/cell). In our work we have chosen 19 V as the terminal voltage using the nearly equivalent convention of (20-1) cells times 1.0 V/cell. The fully charged, ready-for-service condition that yields this nameplate capacity entails charging the battery with a voltage source capable of reaching well over 28 V to nearly 35 V. Most secondary (i.e. rechargeable) batteries do not maintain the nameplate capacity indefinitely over their lifespan. For purposes of our discussions, the term fully-charged capacity will refer to the number of amp-hours a user can obtain from a battery that has been fully charged (by some definition) before the battery terminal voltage falls below 19 V. Often, this charging is done at a fixed voltage and the charging profile used may or may not return the battery to the nameplate capacity. Many factors can contribute to a reduction in the amount of charge (measured in Ah) that a user can extract from a fully charged battery before the terminal voltage is less than 19 V. Factors that reduce the fully-charged capacity from the nameplate capacity include cycle count, depth-of-discharge of previous cycles, temperature during discharge, and charge voltage. These factors impact the underlying mechanisms that contribute to the reduction of nameplate capacity (electrolyte or electrode chemical or mechanical change, for example). For many battery chemistries, the reduction from nameplate capacity can be gradual and permanent. However, for the SP-107A NiCd battery the nameplate capacity can usually be fully restored by following a prescribed reconditioning maintenance action that charges the battery more fully than occurs autonomously on the helicopter, followed by a restoration of electrolyte levels that may have dropped from heavy charging. It was the purpose of this project to create a system which can determine and predict when this maintenance action should be performed for individual batteries. It will also be useful for discussion purposes to define a batterys state of charge, which is the instantaneous amount of charge that is left in a battery, expressed as a percentage of its nameplate capacity. A measurement of SOC can theoretically be made anytime during a cycle and can vary from 0% to greater than 100% in a healthy battery. A physics based model approach was selected for the battery diagnostic and prognostic technologies based on several factors including: the battery chemistry, battery operational usage profile and the charging scheme on the helicopter. The physics-based model of the battery outlined in Figure 7 is motivated by the Bode reaction scheme [2], [3].
Figure 6 Brush Wear Trend and Prognostic Indication This starter-generator display provides a condition indicator and historical trend data for brush wear based on a simulated operational usage profile as well as a RUL prediction in flight operation days. Data samples as shown by the blue data points are representative of snapshots taken at helicopter start-up, one snapshot during each flight hour and one snapshot at shutdown. The yellow and red horizontal lines represent the brush wear warning and alert limits, respectively. The red line extending from the blue data points represents the predicted data trend. The peach colored confidence bounds that track with the wear prediction provide a forecasting confidence range for the prediction. Where this forecasted trend crosses the alert level provides the RUL prediction in flight operation days. The viewing time history of the plot is user selectable between one minute, one hour, one day, one month, six months, one year and three years.
Battery Health Management Technology

Based on the FMEA results and an engineering analysis, the focus of the battery health management technology task was to develop accurate SOC capability that could be implemented on the helicopter and gather data during flight operations. The helicopter battery, MarathonNorco Aerospace model number SP-170A, has a nameplate capacity of 17 Ah. This can be verified for a new, healthy and fully charged (using the manufacturers specification) battery by discharging that battery at 17 amps and measuring 17 Ah of charge flow before the terminal voltage drops to 20 V, which is
JANUARY 2012
linearly with temperature, the dependence coming from literature values [5]. The model was implemented in LabVIEW, a graphical programming language that interfaces to the user through a front panel shown in Figure 8. The inputs to the model are displayed by large indicators at the top left portion of the interface and the upper graph shown in Figure 8. Outputs are shown by the large indicators at the top right and the lower graph. Small controls and indicators scattered around the panel control and indicate operation of the model. The panel is a screen shot during a run on a historical data set of 52 full discharges to 19.0 V and subsequent charges at a constant potential of 28.0 V. The model can accept historical data as shown in the figure, measured real time data, or synthetic data as inputs. The stepped green trace of the lower graph indicates the amount of charge the battery actually supplied at discharge. The white trace in the lower graph shows the models prediction of the SOC, which drops to near zero at discharge events then rises during the charging events. With the parameters in the model during this run, the model does a good job with its predictions, although it slightly overestimates the SOC at early discharge cycles and slightly underestimates SOC at later discharge cycles.
Figure 7 Block diagram of the Physics-Based Model of the Battery. (Inputs enter the model to the left of the dotted line. To the right, wide colored arrows represent the flow of charge.) The model is centered around three buckets that hold charge Q1, Q2 and Q3from which the state of the battery is primarily derived. Q1 nominally represents the amount of charge stored in the charged beta-NiOOH phase, Q2 nominally represents the charge stored in the gamma-NiOOH phase (or more accurately, the gamma-[2NiO2NiOOH](KOH2H2O) phase), and Q3 represents the amount of lost overcharge that goes into the electrolysis of the electrolyte. The current and the change in time stamp between two successive iterations is used to calculate, based on its sign, an input or output incremental charge dQ that either charges or discharges the battery. During charge, the Charge Acceptance Model passes a certain fraction of the input charge to Q1. That fraction is a function of Q1, Q2, current and battery temperature. The amount left over, dQ, is available for Q2 to accept. What remains after that, is used to increment Q3. A conversion factor is then used to convert Q3 into an equivalent volume of lost electrolyte. During discharge, the Discharge Model first determines what fraction of the measured output charge dQ is to come from Q1. What is leftover, dQ, is taken from Q2. The decrement of Q1 is in general larger than the difference between dQ and dQ because of discharging inefficiencies that are a function of current and temperature. Largely for simplicity, the model as a whole attributes all charge and discharge inefficiencies that are functions of current and temperature to Q1 because it is the dominant source of charge, and we have little experimental basis with which to split the inefficiencies between Q1 and Q2. The state-of-charge is calculated in an Available Charge Model from the sum of Q1 and only a fraction, at present taken to be 16%, of Q2, to represent the sequestered nature of the available charge in the alpha-gamma couple due to the poor mobility of potassium in the nickel electrode. (A better model would include a time dependence for the diffusion of potassium in the lattice.) Corrections are made for current and battery temperature [4], [5], with the option (usually taken) of calculating the SOC for a fixed output current such as 17 A as is done by the manufacturers rating method. Both Q1 and Q2 are assumed to decay slowly in a Charge Retention Model, exponentially in time, with a time constant that depends
Figure 8 User Interface to the Model, Showing Results Based on Historical Data In the upper graph, the green trace is the battery temperature, orange is the battery voltage and magenta is the battery current. In the lower graph the red trace is Q1, blue is Q2, white is the predicted state-of-charge, yellow is the amount of charge accumulated continued on next page http://theRIAC.org
13
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER in the input data over the previous charging event, and green is the amount of charge in the input data that left the battery during the previous discharge event. Internally, the model stores various partial derivatives that are used to calculate changes in Q1 and Q2. Figure 9 shows the change of Q1 with incoming charge dQ during charging, as a function of Q1 and Q2.
Figure 10 Charge Acceptance Model for Q1, Nominally Representing Available Charge Stored in the Beta-Beta Couple of the Nickel Positive Electrode, as a Function of Battery Temperature and Cell Voltage Whatever charge, dQ, not accepted by Q1 is available for Q2. The fraction that is accepted is shown in Figure 11. Figure 9 Charge Acceptance Model for Q1, Nominally Representing the Readily Available Charge Stored in the Beta-Beta Couple of the Nickel Positive Electrode, as a Function of Q1 and Q2
The shape of these functions are controlled through additional front panels of the model where the user can adjust the various heights, widths, boundaries, etc. that define the functions. In Figure 9, the height of the plateau determines the charge efficiency with which Q1 accepts charge. The boundary where the function goes to zero controls the maximum charge that can be held by Q1. This is affected by Q2. In effect Q2 is used as a signal for Q1 to accept more charge on the initial cycles after a reconditioning event. To some extent this is an ad-hoc assumption made to match the long term decay seen in the datathe memory of how long it has been since reconditioning is stored by the model, effectively, in Q2. The charge acceptance of Q1 is also modeled as being influenced by temperature and cell voltage, as seen in Figure 10. At low voltages, Q1 keeps most of the incoming charge for itself. At larger voltages it allows more incoming charge to pass to Q2. In other words, the model allows Q2 to only be charged at the relatively high voltages associated with reconditioning the battery, thus allowing a reset to the beginning of the long decay of charge acceptance and SOC as a function of cycle number. The charge acceptance of Q1 is also lowered by cold temperatures and especially by high temperatures. From literature values, the NiCd battery holds less charge at high temperatures [6], [7].
Figure 11 Charge Acceptance Model for Q2 as a Function of Q2 and Cell Voltage, Nominally Representing the Sluggish Charge Stored in the Alpha-Gamma Couple. (Whatever charge that is not accepted here is assumed to electrolyze water.) The function shown in Figure 11 is limited as a function of Q2 to express the finite charge holding ability of the electrodes. The fraction drops as a function of voltage to allow more charge to flow to Q3 and electrolyze the aqueous electrolyte. A conversion factor of 0.245 cc/Ah, given by the battery manufacturer [4], is used to estimate the volume of electrolyte lost from the accumulated Q3. This factor differs from the theoretical stoichiometric value of 0.336 cc/Ah for reasons we do not yet appreciate. Perhaps the difference represents charge loss due to mechanisms other than electrolysis.
JANUARY 2012
The Discharge Model is simpler than the Charge Acceptance Model as shown in Figure 12.
Figure 13 Discharge Model for Q1, as a Function of Temperature and Current Figure 12 Discharge Model for Q1, as a Function of Q1 and Q2 The graph shows that for most values of Q1 and Q2, most of the charge leaving the battery is assumed to come from Q1. This is how the model allows Q2 to decay slowly (motivated by the slow diffusion of potassium in the alpha-gamma phase) and account for the slow decay of charge acceptance and SOC. Only if Q1 is empty, i.e. near zero, and, nevertheless, charge is measured to be leaving the battery, is sizable charge taken from Q2. If Q2 is empty all the charge comes from Q1, although Q1 is never allowed to become negative (thus in effect resetting the model if somehow not enough charge is in Q1 and Q2 to account for the amount of charge measured to be leaving the battery). The Discharge Model takes more charge from Q1 than is measured leaving the battery at extreme currents and temperatures, as shown in Figure 13. This function is constant as a function of current at low currents and increases logarithmically with current at higher current as described in the literature [5], [4]. Its temperature dependence is a spline fit to literature data modified to account for our limited data on these particular batteries [5], [6]. The same current and temperature correction factors are used in the estimate of SOC in the Available Charge Model of Figure 7. Lastly, the decrement of Q2 per whatever charge dQ that Q1 did not take responsibility for is unity unless Q2 is running out of charge, as seen in Figure 14.
Figure 14 Discharge Model for Q2, as a Function of Q2 The zeroing of the derivative at small Q2 is another mechanism by which the model resets itself from long term drift. It takes advantage of the knowledge that Q1 and Q2 cannot both be negative and at the same time be supplying current out the battery terminals. The model so far makes little use of the battery voltage information. It may be possible in a more advanced model to use such information to help the model correct itself from long term drift in the absence of occasional full discharge. The focus of this program was to provide the Army with the ability to maintain aviation NiCd helicopter batteries (specifically the SP-170A) using an on-condition approach. Compared to the on-schedule approach which is currently utilized, this on-condition approach is expected to allow a longer time between maintenance continued on next page http://theRIAC.org
15
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER periods which will reduce maintenance costs and increase up-time of the helicopter. The physics based battery model provides the maintainer with the ability to track the available charge/SOC in amp-hours and predict the remaining useful life of the battery as shown in Figure 15.
the components required to build a circuit model of the inverter). Component values were identified through direct measurement, inferred from measured performance characteristics, and inferred from comparison of simulated performance (with candidate component values) to measured characteristics. A circuit simulation model was developed to compare the signal characteristics of the parameterized inverter model to theoretical models. The model was developed in Multisim and at the same time the circuit was parameterized in order to confirm our model calculations. Development of the circuit model required several key assumptions. The output resonant tank (ORT) was treated as an ideal voltage source. The model was then developed working back toward the input voltage source until the current source inductor (CSI) and MOSFETs, and the RMS to DC converters gave correct outputs. At this point the ideal AC source was replaced by a RLC resonant tank circuit. The Pulse-Width Modulation (PWM) of the MOSFETs was simulated as a circuit input. The datasheet for the PWM IC was used to build the PWM IC and placed into the model so that there was automatic voltage regulation [8].
Figure 15 Historical Trend Data and RUL Prediction for the Battery This battery display provides condition indicator historical trend data for electrolyte level and available charge based on a simulated operational usage profile as well as a RUL prediction in flight operation days. Data samples as shown by the blue data points are representative of single reports based on continuous measurement of battery voltage and current to accumulate charge/discharge current snapshots taken at helicopter start-up, one snapshot during each flight hour and one snapshot at shutdown. The yellow and red horizontal lines represent the electrolyte level and available charge warning and alert limits, respectively. The red line extending from the blue data points represents the predicted slope for the data trend. The peach colored confidence bounds that track with the prediction trend provide a forecasting range for the prediction. Where the forecasted trend for either the electrolyte level or available charge crosses the alert level provides the RUL prediction in flight operation days.
The circuit models that were developed for this program were used to identify faults in the inverter components. Measured parameters or performance were compared to the model components and the predicted performance to identify faults in inverter sections or components. Looking into the inverter from the output reveals two basic modes, the first when the MOSFETs are on and the second when the MOSFETs are off. Knowing that the inverter operates in two different modes and that they happen at regular intervals, the output voltage and current signals can be separated based on the inverter mode. Both topologies need to be separated further to account for initial conditions. This makes sense for the MOSFETs ON case where there are two well defined switches changing the polarity of the current from the CSI that may have a non-zero initial current. The reasoning for keeping track of which MOSFET OFF case the inverter is in is less apparent, because the topology is the same for both of these segments. If the circuit equations includes an integral there is most likely a constant that represents the initial condition of the integral. This initial condition alternates signs thus requiring the knowledge of the operational mode of the MOSFET. The structure of the detection algorithm is to determine the inverter topology mode of operation, generate the needed derivatives and integrals, perform coefficient estimation using Batch Least-Squared Estimation (BLSE), and scale the output for reference as shown in Figure 16.
Inverter Health Management Technology

The inverter health monitoring technology is based on a modelbased approach. Real-time measurements of the observed inverter operating parameters are compared to those predicted by a circuit model of the inverter. Deviations between observed and predicted behavior provide indicators of faults in the critical components identified in the FMEA. The first step in developing the model of the inverter involved identifying the topology used in the OH-58 inverter; different inverter topologies use different configurations of components and will therefore have different circuit models. After identifying the inverter topology, we identified the specifications for and operational characteristics of the actual inverter components (both the components of interest from a health monitoring standpoint and
Figure 16 Coefficient Estimation Block Diagram

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER JANUARY 2012
In order to develop an algorithm to segment the signals so they can be used in coefficient estimation routines, a well defined understanding of the operation of the inverter and the expected signals is necessary. It is possible to use either the output voltage or current, although the current tends to be dependent on the load, while the voltage tends to depend more on the supply voltage and not the load. A quick inspection of the output voltage waveform as shown in the top plot in Figure 17, does not show any signs of when switching occurs, but by looking at the first derivative of the output voltage as shown in the middle plot of Figure 17, the switching events are observable. These transitions are even more noticeable as a peak in the second derivative of the output voltage as shown in the bottom plot of Figure 17.
The process starts by identifying the turn-off transition because it produces a larger spike on the second derivative of the output voltage as shown in Figure 18. The larger relative size of this spike is due to the CSI having more current when the MOSFETs turn off than when they turn on. A single cycle of data is used to identify the largest positive and negative peak.
Figure 18 Initial Turn-Off Peak Detection This is where the frequency accuracy is needed and is noticeable. An FFT with sufficient zero padding for a nominal 400-Hz signal sampled at 50 kHz (about 125 samples per cycle), could result in a 3 to 4 sample drift over 50,000 samples and cause inaccurate identification of turn-on and turn-off times. Using the relationship that these peaks occur once every cycle at one cycle intervals, the expected peak location is projected over the entire one second of data. Because the expected peak location is only accurate to 1 sample, the interval around the expected peak is searched for the local positive or negative peak, and the peak location is updated. Since the turn-off peaks are larger than the turn-on peaks, the turnoff peaks and the area around them are zeroed out after they are identified so the largest remaining peaks are the turn-on peaks as shown in Figure 19.
Figure 17 KGS Inverter Output Voltage, First Derivative and Second Derivative To automate the process of finding the topology transitions, an algorithm was developed to identify these peaks in an accurate and orderly fashion. This is done in a three-step process: determine the operating frequency, filter the signal, find the turn-off and turn-on transitions. The detection algorithm is designed to work on one second of data but is capable of handling longer or shorter lengths of data. The frequency of the voltage is needed to calculate the samples per cycle used in both the turn-off and turn-on event detection algorithms. These algorithms estimate the location of peaks over the entire length of the sampled signal, and require accurate voltage measurements to accomplish this process. The accuracy needed can be accomplished by taking a Fast Fourier Transform (FFT) of the output voltage. Zero padding to the next larger power of two is advised for FFT speed, but is not necessary to increase accuracy. The second derivative of the output voltage is run through a high pass filter with a cut-off frequency set to 2.5 times the measured output frequency. This cut-off frequency was set to maximize the attenuation of the fundamental (400 Hz in this case) without degrading the peak amplitudes.
Figure 19 Initial Turn-Off Peak Detection continued on next page http://theRIAC.org
17
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER Once the MOSFET turn-off transitions have been zeroed out the peak identification algorithm is applied again to identify the MOSFET turn-on transitions. The final results of the topology transition location algorithm provides a map of the MOSFET states as shown in Figure 20 that are used for the fault detection procedure.
XSC1
Ext T rig + _ A + B _ + _
L_csi
R_csi
R_ll
L_l
XCP1
V_bus
L_t
Figure 21 BLSE Model and Component Table for MOSFET ON Topology A simplified model of the output as shown in Figure 22 was developed. It consists of an inductor capacitor resonant tank and the load. This model worked well at the load level being tested on the long term test bed, but showed load disturbances. This model was kept simple because it was observed that the inverter spent most of its time in the MOSFETs ON topology. XSC1
Ext T rig + _ A B _ + _
Figure 20 Inverter Operational Topology Segments In order to track changes in the inverter a method of Batch LeastSquared Estimation is used. The method accepts inputs of measured signals, their derivatives as well as integrals and then outputs the parameter estimates [9]. These parameters estimates are the coefficients found in front of each of the derivatives and integrals in the mathematical equation describing the MOSFET ON and MOSFET OFF models. When the model is generated it lumps values together such as the MOSFET drain to source-on resistance and the CSI resistance into the CSI resistance, consequently reducing the ability to detect changes in the MOSFET drain to source resistance. The information in Table 3 below provides an indication of the detectability of each failure mode using coefficient estimation.
XCP1
L_tank
C_tank
Z=A+jB
Z_load
Figure 22 BLSE Model for MOSFET OFF Topology Table 3 Component Failure Detection Probability using Parameter Estimation Seeded fault testing was conducted and one second worth of data from the inverter was recorded every ten minutes. Each of these one second samples is considered an index in time. Faults and load changes were implemented instantaneously, which resulted in step changes in the parameter estimation. Three seeded faults were induced to validate the detection capability of the parameter estimation algorithm. The faults induced in the inverter included those listed in Table 4. Table 4 Inverter Seeded Fault Indexes A higher order BLSE model shown in Figure 21 was used for the MOSFET ON topology to reduce loading changing effects. Models with only capacitors and inductors, and without leakage inductance and primary winding resistance were tested but showed noticeable change when loads were changed.
JANUARY 2012
Z=A+jB
R_c
C_t
Faults seeded in the ORT, output capacitance, should show up in both the MOSFETs ON and MOSFETs OFF parameter estimations, but faults in the CSI, CSI inductance and resistance, should only show up in the MOSFETs ON parameter estimation. At the same time load changes should not show up in the parameters estimations, but are noticeable in the MOSFETs OFF estimation as indicated in Figure 23 at index 1969. The results for the MOSFETs OFF case correctly identified the capacitance change that occurred at index 6298 in Figure 23 as an increase of capacitance, but also identified other capacitance and inductance changes that did not occur. The MOSFET OFF parameter estimation picked up cross talk between phases at index 1969 when a change in capacitance in phase A was also indicated in phase B as an increase in capacitance. The spike that occurs on phase B at 11034 to 11185 is the result of installing a damaged CSI on that phase; once again this should not show up on the MOSFET OFF parameter estimation, but is a result of a low component count model being used for parameter estimation. This in itself does not constitute an issue because the inverter would have to get replaced, but the parameter estimation also indicated a capacitance change at index 1969 that was a load change. The change in load effect is more noticeable when the load current is varied and the parameters are estimated as shown in Figure 23. The result of this data analysis does not indicate that a correction curve should be used, but instead that a higher component count mathematical model should be used. Parameter estimation for the MOSFET OFF case was done using a two component model, which allowed parameter estimation of actual circuit parameters. The MOSFET ON case used 10 components which resulted in the estimated parameters to be made up of combinations of the actual circuit parameters. The model equation used for the MOSFETs ON estimation uses a more complete model, and as a result does not show the loading effect that the MOSFETs OFF estimation suffered.
Though the intent of this program was to potentially develop an inverter prognostic capability, this was not achievable due to inability to achieve results from the accelerated life testing in time for this paper.
5. Health Management Hardware and Interface Implementation

The power system health management technology includes sensors, three component health nodes, one power system health hub and an interface display as shown in Figure 24.
Figure 24 Power System Health Management Hardware Block Diagram The nodes and health hub are comprised of a PC-104 computer network (small format computer running Windows XP) and a Windows-based graphical user interface (GUI) was developed to display the results from the development of the health management technology team in a single, cohesive environment. The majority of the data processing occurs at the node level, which provides data acquisition capability and condition indication fault algorithm processing capability. The power system health hub is representative of the HUMS system and provides the system level health assessment and display processing capability. The power system health hub is an Intel Core 2 Duo based computer system with an Intel Graphics Media Accelerator 4500MHD to support the 3D animation graphics in the user interface. The data can potentially be downloaded to a ground station from the power system health hub for use by maintenance personnel. The graphical user interface provides an informational data link between the three subsystem data collection/processing nodes (battery, starter/generator and inverter) and the power system health hub. The intent of the hardware/software architecture was to collect pertinent condition based maintenance condition indicators and functional parameters from each of these subsystem nodes and present it in an efficient and user friendly manner for the final continued on next page http://theRIAC.org
Figure 23 MOSFETs OFF Two Component Parameter Estimation
19
POWER SYSTEM PROGNOSTICS FOR THE U.S. ARMY OH-58D HELICOPTER technology demonstration as shown in high level block diagram of the software architecture in Figure 25.
health. This provides a visual aid to the user on the exact location of the failing component, which can aid in the troubleshooting and replacement of the components. The bottom portion of the display provides text based alerts from the health monitor.
6. Conclusions
The helicopter power system health management technology is designed to be embedded on the helicopter and integrated with the HUMS. The objective of the technology implementation is to provide diagnostic and prognostic condition indicators in order to lengthen the maintenance interval which will help to increase operational availability and reduce sustainment costs. Figure 25 Software Architecture Design of the Power System Health Monitor User Interface The software architecture layout consists of a main overview screen display that leads to component level overview screens and historical as well as trend plotting for each component. All of the displays are setup and initialized through a configuration screen. The graphical user interface is a Windows based application meant to provide the user with an intuitive interface that will provide diagnostic and prognostics predictions for the helicopter power system components. A three dimensional (3-D) helicopter model is displayed on the main screen of the helicopter power system health monitor application as shown in Figure 26. This effort focused on development of prognostic condition indicator algorithms for the detection of power system faults including the starter-generator, battery and inverter. Hardware, software and information interfaces were developed for the technology implementation to show the utility of the technology for power system health monitoring.
Acknowledgment
This work was supported jointly by the Aviation Applied Technology Directorate (AATD) and Bell Helicopter through a U.S. Army Operations Support and Sustainment Technology (OSST) program. The content of the information does not necessarily reflect the position or policy of the Government or Bell Helicopter, and no official endorsement should be inferred. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied of the Aviation Applied Technology Directorate or the United States Government.
References
[1] [2] [3] R.J. Hamilton, DC Motor Brush Life, 1998 IEEE Industry Applications Conference, vol. 3, pp. 2217-2224. A. H. Zimmerman, Nickel-Hydrogen Batteries Principle and Practice. El Segundo, CA: The Aerospace Press, 2009, pp. 87107. D. Berndt, Maintenance-Free Batteries: Based on aqueous electrolyte, 3rd ed. Baldock, Hertfordshire, England: Research Studies Press Ltd., 2003, pp. 208214. MarathonNorco Aerospace, Operating and Maintenance Manual for Nickel-Cadmium Aircraft Batteries. Waco, TX, 1997. J. M. Evjen and A. J. Catotti, Vented sintered-plate nickel-cadmium batteries, in Handbook of Batteries, 2nd ed., D. Linden, Ed. New York: McGraw-Hill, 1995. General Electric Company, Battery Products Section, Nickel-Cadmium Battery Application Engineering Handbook. Gainesville, FL,1971. Varta Batterie AG, Sealed Nickel Cadmium Batteries. Dusseldorf: VDIVerlag GmbH, 1982. http://focus.ti.com/lit/ds/symlink/sg2524.pdf strm, Karl J., and Wittenmark, Bjrn, Adaptive Control, 2nd edition, Dover Publications, Inc., 2008, pp 41-49.
Figure 26 The Main Screen of the Helicopter Monitor User Interface The purpose of the main screen is to display high level status and fault isolation information for the power system components. The left portion of the display provides status icons for each component. The right portion of the display provides a 3-D version of a OH-58D Kiowa Warrior model that provides a intuitive capability for providing fault indication and isolation information. The fault information is indicated using simple green, yellow, red type symbolic indicators for each component based on its current
[4] [5]
[6] [7] [8] [9]
JANUARY 2012
http://theRIAC.org
21
A SHORT HISTORY OF RELIABILITY
REPERTOIRE
RELIABILITY TRAINING FOR PROFESSIONALS ON THE GO
REPERTOIRE is the RIACs set of interactive reliability engineering training courses developed around the American Society for Quality (ASQ) body of knowledge for the Certified Reliability Engineers (CRE) exam. Whether you are preparing for the CRE exam, or just need some basic training or refreshing in reliability, youll appreciate the convenience of training at your own pace, on your own schedule. The combined set of five courses contains approximately thirty hours of narrated training, with around six hours of content in each course. The available courses cover: Reliability Management (REPER-01) Probability and Statistics for Reliability (REPER-02) Reliability in Design and Development (REPER-03) Reliability Modeling and Prediction (REPER-04) Reliability Testing (REPER-05) Each of the five courses is divided into independent modules that typically take about one hour each to complete. The web-access version of REPERTOIRE contains hundreds of quiz questions and interactive exercises (about 10-20 reinforcement questions per module), so students can assess their progress and review those areas where they may need improvement. The questions are
ORDER CODE: US PRICE NON-US WEB ACCESS REPER-DVD REPER-FULL REPER-01 REPER-02 REPER-03 REPER-04 REPER-05 $370 NA NA NA NA NA NA $405 NA NA NA NA NA NA NA $449 $99 $99 $99 $99 $99
automatically graded and stored by REPERTOIRE for future reference. (Note that the quizzes and interactive exercises are not included in the DVD version). Purchase of the entire fivecourse set on DVD (REPER-DVD) or via web access (REPERFULL) includes a copy of the Quanterion Solutions Inc. QuART PRO software set of automated reliability tools.
A REPERTOIRE demo is available at: http://theRIAC.org
THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER // JANUARY 2012

The Journal of the Reliability Information Analysis Center is published quarterly by the Reliability Information Analysis Center (RIAC). The RIAC is a DoD Information Analysis Center (IAC) sponsored by the Defense Technical Information Center (DTIC) and operated by a team led by Wyle Laboratories, and including Quanterion Solutions Incorporated, the Center for Risk and Reliability at the University of Maryland, the Penn State University Applied Research Lab (ARL) and the State University of New York Institute of Technology (SUNYIT ).
i RAC
View the online demo at: http://theRIAC.org Representative, RIAC Contracting Officers
Air Force Research Laboratory Joseph Hazeltine RIAC Director, Technical Area Task (TAT) Manager Preston MacDiarmid RIAC Technical Director Valerie Hayes RIAC Deputy Director TATs/SAs David Nicholls RIAC Operations Manager David Mahar Software & Database Manager Patricia Smalley RIAC Training Coordinator
Paul Engelhart
P 315.330.4477 paul.engelhard@rl.af.mil P 256.716.4390 // FAX 256.721.0144 joseph.hazeltine@wyle.com Toll Free 877.808.0097 P 315.732.0097 // FAX 315.732.3261 pmacdiarmid@quanterion.com P 301.863.4301 // FAX 301.863.4281 valerie.hayes@wyle.com Toll Free 877.363.RIAC (7422) P 315.351.4202 // FAX 315.351.4209 dnicholls@theRIAC.org Toll Free 877.808.0097 P 315.732.0097 // FAX 315.732.3261 dmahar@theRIAC.org Toll Free 877.363.RIAC (7422) P 315.351.4200 // FAX 315.351.4209 psmalley@theRIAC.org
The Reliability Information Analysis Center 100 Seymour Road Suite C101 Utica, NY 13502-1311 Toll Free: 877.363.RIAC (7422) P 315.351.4200 // FAX 315.351.4209 inquiry@theRIAC.org http://theRIAC.org RIAC Journal Editor, David Nicholls Toll Free: 877.363.RIAC (7422) P 315.351.4202 // FAX 315.351.4209 dnicholls@theRIAC.org
2012 No material from the Journal of the Reliability Information Analysis Center may be copied or reproduced for publication elsewhere without the express written permission of the Reliability Information Analysis Center.
JANUARY 2012
N E W M I L I TA R Y H A N D B O O K 189 C , R E L I A B I L I T Y G R O W T H MANAGEMENT
Lisa I. Carroll, AMSAA
Summary
The U.S. Army Materiel Systems Analysis Activity (AMSAA) published the new MIL-HDBK-189C, Reliability Growth Management, which is critical for implementing the new Office of the Secretary of Defense (OSD) and Army reliability policies.
Background
Reliability growth management procedures have been developed to improve the reliability of Department of Defense (DoD) weapon systems. Reliability growth techniques enable acquisition personnel to plan, evaluate and control the reliability of a system during its development stage. The reliability growth concepts and methodologies have evolved over the last few decades by actual applications to military systems. Through these applications, reliability growth management technology has been developed to the point where considerable payoffs in system reliability improvement and cost reduction can be achieved.
Conclusion
AMSAA published MIL-HDBK-189C, Reliability Growth Management, in June 2011 to reflect recent development of reliability growth concepts and methodologies based on applications to military systems. Comments from Reliability Subject Matter Experts within the Army, Navy and Air Force were incorporated. The updated handbook supports new OSD and Army reliability policies and was posted to the Acquisition Streamlining and Standardization Information System (ASSIST) database for use by all of DoD.
Whats New
Reliability growth encompasses 3 areas: planning (prior to test data), tracking (using test data), and projection (using test data and applying fix effectiveness factors). Thirty years of lessons learned has culminated in the recent development of several models in each of these areas. Collectively these reliability growth models are referred to as the AMSAA Visual Growth Suite and are available free of charge to US government personnel and their supporting contractors. One of the most significant models to note is the Planning Model Based on Projection Methodology (PM2). It develops a system-level reliability growth planning curve that incorporates the developmental test schedule and corrective action strategy. Benefits include risk reduction, construction of feasible reliability test programs, and bridging the gap between engineering efforts and program constraints with the overall reliability program.
Biography
Lisa Carroll is a member of the Reliability Analysis Team at the U.S. Army Materiel Systems Analysis Activity. She earned her bachelors degree in Mathematics at Albright College in Pennsylvania and her masters degree in Statistics at the University of Delaware. Lisa I. Carroll Operations Research Analyst, AMSAA ATTN: AMSRD-AMS-LR 392 Hopkins Road APG, MD 21005-5071 USA
http://theRIAC.org
23
RIAC Announces Major New Product/Service Releases

The RIAC Web-Accessible Repository of Physics-based Models (WARP) The RIAC Web-Accessible Repository of Physics-based Models (WARP)
The current trend in systems engineering and reliability assessment within the DoD and industry is to supplement traditional empirically-based reliability prediction models with techniques that apply more physics-based approaches to the identification and elimination of life-limiting failure modes and mechanisms through proactive Design for Reliability (DFR) and reliability growth techniques. Many Physics-of-Failure (PoF) models, covering both mechanical and electronic device failure modes and mechanisms, have been published in the open literature over the years or developed as part of membership-funded consortiums (e.g., CALCE, AVSI, VITA-51). Up until now, virtually no effort has been made to synthesize these widely scattered models into a single, readily-accessible repository that can be used to support reliability practitioners and researchers in model implementation and development. RIAC proudly announces the release of its free Web-Accessible Repository of Physics-based Models (WARP) website. WARP provides systems engineers, design engineers, reliability practitioners and the research community at large with a centralized resource for dozens of physics-based models, and opportunities for registered members to become part of a rapidly growing community of model submitters and evaluators. Come see what WARP is all about, at http://www.theriac.org/WARP/login/main_login.html
Achieving System Reliability Growth Through Robust Design and Test Test Achieving System Reliability Growth Through Robust Design and
Historically, the reliability growth process has been thought of, and treated as, a reactive approach to growing reliability based on failures discovered during testing or, most unfortunately, once a system/product has been delivered to a customer. As a result, many reliability growth models are predicated on starting the reliability growth process at test time zero, with some initial level of reliability (usually in the context of a time-based measure such as Mean Time Between Failure (MTBF)). Time zero represents the start of testing, and the initial reliability of the test item is based on its inherent design. The problem with this approach, still predominant today, is that it ignores opportunities to grow reliability during the design of a system or product, i.e., opportunities to go into reliability growth testing with a higher initial inherent reliability at time zero. In addition to the traditional approaches to reliability growth during test, this book explores the activities and opportunities that can be leveraged to promote and achieve reliability growth during the design phase of the overall system life cycle. The ability to do so as part of an integrated, proactive design environment has significant implications for developing and delivering reliable items quickly, on time and within budget. This book offers new definitions of how failures can be characterized, and how those new definitions can be used to develop metrics that will quantify how effective a Design for Reliability (DFR) process is in (1) identifying failure modes and (2) mitigating their root failure causes. Reliability growth can only occur in the presence of both elements. For more information, come visit the RIAC web page at http://theRIAC.org.
Achiev ing Sys tem Re liability Growth Throug h Robu st Design and Te st
RIAC is is opera a DoD Informatio State Unted by a team of n Analysis Ce nter iversity Ap Wy plied Re le Laboratories,sponsored by the search La Qu boratory anterion Solut Defense Technic ion and the al State Un s, the Universit Information Ce iversity of y nte New Yo of Maryland, the r. RIAC rk Institu Pe te of Tech nn nology.
JANUARY 2012
Techniques to Evaluate Long-Term Aging of Systems Techniques to Evaluate Long-Term Aging of Systems
The Reliability Information Analysis Center mission has continued to evolve over the years, with greater recognition that reliability, maintainability, and quality are critical military system and commercial product attributes. Coincident with this evolution has been the recognition that these attributes have much more encompassing impacts than their specific attained metrics. For example, where at one time the measure of reliability called mean-time-between-failure (MTBF) was of interest in itself, now management and system planners have expanded their interest because of how that MTBF impacts such issues as sustainability and total ownership costs. Part of the refocusing of reliability-related needs is the concern over long-term aging effects on systems in operation and in storage brought about by systems having to last longer. The objective of this book is to highlight the impact of longterm aging effects on parts, assemblies and equipment by investigating characteristics of aging as they impact specific material classes. More information is available at the RIAC website: http://theRIAC.org.
by a erated C is op Research er. RIA ed n Cent rsity Appli ormatio Unive nical Inf Penn State y. se Tech log the Techno e Defen Maryland, th of titute red by ty of sponso e Universi New York Ins Center of s, th alysis n Solution University tion An ate terio Informa ies, Quan y and the St a DoD or rator RIAC is Wyle Labo Laborat of team
i RAC
i C RAC
Probability Distributions Used in Reliability Engineering Probability Distributions Used in Reliability Engineering
Reliability Engineers are required to combine a practical understanding of material science and engineering with statistics. The reliability engineers understanding of statistics is focused on the practical application of a wide variety of accepted statistical methods. Most reliability texts provide only a basic introduction to probability distributions or only provide a detailed reference to few distributions. Detailed statistician texts provide theoretical detail which is outside the scope of likely reliability engineering tasks. As such, the objective of this book is to provide a single reference text of closed-form probability formulas and approximations used in reliability engineering. This book, developed by the University of Maryland Center for Risk and Reliability, published by the RIAC, and used as a supplemental reference book in courses offered by the University of Maryland, provides details on 22 probability distributions. Each distribution section provides a graphical visualization and formulas for distribution parameters, along with distribution formulas. Common statistics such as moments and percentile formulas are followed by likelihood functions and, in many cases, the derivation of maximum likelihood estimates. Bayesian non-informative and conjugate priors are provided, followed by a discussion on the distribution characteristics and applications in reliability engineering. Each section is concluded with online and hardcopy references which can provide further information, followed by the relationship to other distributions. More information is available at the RIAC website: http://theRIAC.org.
Proba Used i bility Distrib utions Reliab n ility En gineer ing
RIAC is is opera a DoD Inform State Unted by a team ation Analysis Ce of iversity Applied Wyle Laborat nter sponsored ories Researc h Laborat , Quanterionby the Defense So ory and Te the Sta lutions, the Unchnical Inform te Unive ati rsity of iversity of Maryl on Center. RIA New Yo rk Instituand, the Penn C te of Te chnolog y.
http://theRIAC.org
25
RELIABILITY INFORMATION ANALYSIS CENTER
i RAC
training
March 20-22, 2012 // RIAC Training in San Diego, CA

Choose From:
March 20 - 22, 2012 Reliability 101
Daniel Gonzalez, Quanterion Solutions, Inc.
$1,295.00 per attendee

Weibull Analysis
Wes Fulton, Fulton Findings Hosted at the San Diego Training & Conference Center
June 12-14, 2012 and June 19-21, 2012 // RIAC Training in Virginia Beach, VA
Choose From:
June 12 - 14, 2012
Steven Arndt, PhD, Nuclear Regulatory Commission
Software Reliability & Integrity

Accelerated Reliability Testing
Mohammad Modarres, PhD, University of Maryland Hosted at the Spring Hill Suites by Marriott

June 19 - 21, 2012
Daniel Gonzalez, Quanterion Solutions, Inc.
Reliability 101

A Systems Approach to Reliability in Design
Stewart Kurtz, PhD, Penn State University John McCool, PhD, Penn State University
$1,495. per attendee

00
For more information and to register visit http://theRIAC.org or call 877.363.7422

Discounts apply to multiple registrations from an organization. Please contact the RIAC for details.
JANUARY 2012
Accelerated Reliability Testing

Course Description Introduces accelerated testing approaches that can be effectively used to characterize and improve product reliability. Shows the students how to develop and optimize test plans for accelerated testing and presents graphical and analytical analysis methods. Discusses models for life testing at constant stress, and for step and time varying stresses. Includes hands-on exercises using software tools. Who Should Take the Course Designers, reliability specialists, and program managers can benefit from a knowledge of how accelerated testing techniques can be used to improve and characterize product reliability.
Patricia Smalley RIAC Training Coordinator Toll Free 877.363.RIAC (7422) P 315.351.4200 // F 315.351.4209 psmalley@theRIAC.org http://theRIAC.org/Training
A Systems Approach to Reliability in Design

Course Description The Systems Approach to Reliability in Design training course is a practical application of fundamental systems engineering to system and component design for reliability. This course covers the theories of systems' reliability and demonstrates the supporting mathematical theory. The essential tools of reliability analysis are presented and demonstrated. These applications are further solidified by practical problem solving and open discussion. The RIAC course, "Reliability 101" is an excellent preparatory course. Who Should Take the Course Designers, reliability specialists, and program managers can benefit from knowledge of how various systems' analysis tools, accelerated testing and life cycle prediction techniques can be used to characterize and improve product reliability.
Software Reliability & Integrity

Course Description The 3-day course offers software reliability models, their domains of application and their limitations as well as hands-on software reliability measurement, analysis, and design. A background in software engineering, software development, or reliability engineering is not required to attend the course. Course hand-outs include a detailed course manual. Who Should Take the Course This course is specifically designed for reliability engineers, systems engineers, quality assurance engineers, and software engineers and testers who are responsible for measuring, analyzing, designing, automating, implementing, or ensuring software reliability for commercial and government programs.
Reliability 101
Course Description The course covers reliability program concepts, design and analysis methods, and test and evaluation approaches. The course is ideal for persons with newly assigned responsibilities in the reliability and maintainability area, as well as managers who want to increase their awareness of the payoffs of different approaches and techniques. Who Should Take the Course Individuals who need to know the basics of reliability engineering as they apply to developing and fielding better products and systems. Design engineers, reliability specialists, and product/program managers will benefit from the course.
Weibull Analysis
Course Description The 3-day course gives a complete overview of Weibull analysis, starting with the basics of performing a Weibull analysis and interpretation of the results. The course progresses through more complex applications of Weibull as well as other distributions. Weibull software applications are utilized to give students hands-on examples of performing analyses and interpreting the results. Class problems round out the in-depth presentation. Who Should Take the Course Engineers responsible for reliability, safety, supportability, maintainability, materials, warranty, life cycle cost, design, structures, instrumentation, and logistics will find these Weibull techniques extremely useful.
http://theRIAC.org
27
D E V E LO P M E N T O F T H E W E B - A CC E S S I B L E R E P O S I TO R Y O F P H Y S I C S - B A S E D M O D E L S ( WA R P )
Alex MacDiarmid, Quanterion Solutions, Inc. This article discusses the development of the Web-Accessible Repository of Physics-Based Models (WARP), a Physics-of-Failure model database developed by Quanterion Solutions Incorporated and the University of Marylands Center for Risk and Reliability as part of the operation of the Reliability Information Analysis Center (RIAC). specific nature of PoF models, combined with a general uncertainty amongst analysts as to where one should search for them, can often inhibit the use of these models despite the benefits of their predictive capabilities. In order to alleviate some of the difficulties associated with the search for suitable models, the RIAC initiated an effort to develop the Web-Accessible Repository of Physics-Based Models (WARP). Essentially, the WARP database provides a single-source repository of Physics-of-Failure models for both electronic and non-electronic component types. Since the definition as to what constitutes a PoF model can be somewhat inconsistent, models submitted and accepted into the WARP database must meet the following criteria: A quantitative mathematical model Identifies a root cause failure mechanism Identifies the operational stresses that accelerate/decelerate the failure mechanism Includes parameters that are based on test, field and/or simulated data The interactive WARP webpage (http://www.theriac.org/WARP/ index.php) which hosts all related activities for the repository, is also accessible through a link on the RIAC homepage (http:// theriac.org/). The single-source collection of applicable failure models is of significant value to the engineering community, as it not only simplifies the search process, but also facilitates the use of an accurate modeling technique for various reliability purposes. The numerous features of the WARP repository, including the submission, approval and retrieval of individual models, are described in the following sections.
The Web-Accessible Repository of Physics-Based Models (WARP)

Of the various competing approaches to forming a reliability prediction for parts or assemblies, there has been a growing preference in the use of Physics-of-Failure (PoF) models. Physics-based models are revered because of the fact that they model the life-limiting failure mechanisms of a part or system, accounting for the various operating conditions (e.g., temperature, voltage, load cycling, etc.) that either accelerate or decelerate failures. This approach is widely regarded as one of the most accurate predictors for the behavior of a part or assembly in a given environment, overcoming some of the documented limitations of both empirical models and surrogate data. Despite their popularity, there are some disadvantages to the use of PoF models, which generally result from the narrow focus of the individual models. The amount of detail that is required to model the failure mechanisms of a particular component type, part interaction, or material process is such that these models are often limited in their applicability. Consequently, the search for an appropriate model often becomes a time-consuming process. Adding to the issue is the fact that developed models are often scattered throughout several different types of publications, including conference proceedings, technical reports and scientific journals. The
JANUARY 2012
Figure 1: The WARP Home Page for a WARP Contributor
WARP Features
WARP was designed as an interactive database, providing users with the ability to view and potentially submit PoF models, while also allowing for forum discussions regarding the strengths, weakness and overall applicability of submitted models. In order to accommodate these features, a great deal of functionality was built into the design of this repository and the supporting interface. The features available to a potential user are subject to the status for which they have registered and been approved. There are three general levels of participation in the WARP repository. Warp Users, which are not required to register, are given access to all of the features of the WARP website, but cannot submit models or participate in the WARP Forum. Essentially, WARP Users are limited to searching and viewing the available models. WARP Members, on the other hand, are required to register and, once approved, receive the same privileges as a WARP User and can also participate in the WARP Forum. Those interested in submitting models or validation data can request contributor status on their user registration form. Once approved, they will become a WARP Contributor, with full privileges to contribute models and/or data to the WARP repository. Figure 1 provides an illustration of the WARP Home Page and the various user features for a WARP Contributor. Individuals applying for contributor status should generally be interested in promoting the PoF methodology, with a desire to directly contribute to the development and growth of this field. In some instances, the WARP repository provides users with an opportunity to present their published model(s) to the engineering community. In other cases, individuals that have not developed a model may choose to submit one with which he or she has had success or some level of familiarity. Those who are only interested in viewing the models within WARP can simply disregard this option and choose the user status that best represents their interests. As part of this effort, the database was initially populated with a collection of models. However, it is important to note that, ultimately, WARP is intended to be a user-populated database, where the repository will rely on submissions from the user community for sustained growth. Another feature of the WARP database is the ability to perform, or provide input to, a gap analysis. Users can suggest a specific failure mechanism, component type, or mechanism-component relationship that the repository does not already include. This feature can be used for a number of purposes. When a WARP Contributor sees that there is a need for a specific type of model, he or she may know of an appropriate model, or perform a search of relevant sources to find a model fitting the criteria, which can be submitted to the WARP database. On the other hand, the gap analysis may identify a future opportunity for researchers in industry or academia that are involved with development of PoF models. This feature is helpful in identifying any gaps in the repository, while ensuring that the technologies and failure mechanisms included are as comprehensive as possible. continued on next page http://theRIAC.org
29
DEVELOPMENT OF THE WEB-ACCESSIBLE REPOSITORY OF PHYSICS-BASED MODELS (WARP)
WARP Functionality
The main functions of the WARP repository are to allow for the submission, approval and retrieval of PoF models. A structured process has been developed for each of the three, ranging in complexity depending on the details of the task. Retrieving a model is simple and straightforward. Model submission, on the other hand, requires additional input to sufficiently summarize the details of the model. Nonetheless, users will find that the WARP interface allows these functions to be performed with ease and requires minimal, if any, instruction. For convenience, however, the Users Guide that can be downloaded directly from the WARP website.
The main input fields used to gather the necessary data for a submitted model are as follows: Model Title Failure Mechanism Model Image Equation Model Parameters Model Component Types Model Characteristics Model Assumptions Model Limitations Model Constraints Technical Point of Contact Publication Status Bibliographic Citation Uncertainty Limits Model Data/Info Needed from Outside Sources
Model Submission
A significant portion of the WARP effort was devoted to developing a process that allows users to submit PoF models through the WARP interface. After several iterations, each altering the format and type of information requested, the process has been simplified while maintaining the detail necessary to explain the application of the model. For WARP Contributors, model submissions are initiated on the WARP Home Page, after which the user is directed to the Model Specification Page (see Figure 2), where he or she can populate the various fields that are used to summarize a candidate model. Much of the refinement to the model submission process involved modifying the different ways that a user can input the necessary information. Looking at the Figure, one will notice that there are several different ways to enter data into the repository. A title for the model, for instance, is entered through a text box while the failure mechanism is selected from a drop-down box. An image of the models mathematical equation is uploaded from a saved file. Inputs that require multiple fields are submitted through a pop-up window. Thus, there are a number of methods for data entry, each of which has been carefully considered to expedite and simplify the model submission process. The WARP database not only contains a collection of relevant PoF models, but it also presents a summarized description of the model in a viewable format. In order to do this, key pieces of information about the models development, application and validation are needed. Different fields are used to request this data from the user, instructing them as to where and when specific information should be provided.
While many of these fields are fairly straightforward, others (such as characteristics and limitations) can include a wide variety of user inputs. The Model Characteristics field is used to enter any information that may be of interest, including the models general purpose, how it is applied or how it was developed (e.g., theoretical basis, extension of previous modeling efforts). Model Limitations typically address the scenarios in which the model can be applied, which may be limited to a specific class of materials or loading conditions. Similarly, Model Constraints address the quantitative limits
Figure 2: Illustration of the Model Specification Page to Provide Relevant Information During Model Submission
JANUARY 2012
for which the model has been proven valid, and often include a range of temperatures, voltages or some other operating condition. It is critical that the summary provided for a model submission include only the information that is directly available from the source document from which the model is taken. In other words, WARP Contributors should not provide their own interpretations and descriptions of the model since they are not responsible for its development. Doing so can be troublesome because it puts words in the authors mouth, with which they may not agree. By simply summarizing the information explicitly stated in the models source document, WARP Contributors can avoid these types of issues. The bibliographic citation field reflects the importance of the models traceability, both for the approval process and for WARP Users interested in the development or application of the model. Accordingly, it is necessary to provide a citation for individuals and organizations that are responsible for the models development, so as to avoid copyright infringement. This data also allows WARP users to see what work has been done in a specific field, and who has performed it. More importantly, once a dialogue is started, there may be a possibility for collaboration to apply new ideas that could potentially improve existing models, or develop entirely new models. Before a model can be formally submitted, the WARP interface verifies that all of the necessary fields have been populated, preventing a user from omitting important details. Those interested in a more detailed explanation of the model submission process can download the WARP User Manual from the WARP Help Page.
Once the review process has been completed, the WARP interface generates an email that notifies the user of the results. The notification will include one of the three possible outcomes: 1. The model has been approved for display in the WARP repository 2. The model has been rejected from the WARP repository 3. The model is being considered for display in the WARP repository but requires revisions The approved and rejected outcomes require no further action, though a user can submit feedback if it is unclear as to why a particular model was rejected. When the user is notified that the model requires revisions, it generally means that a valid PoF model is presented; however, there are some issues with the model summary that require attention. The WARP interface allows reviewers to provide comments on the model submission, so that users can quickly identify any errors in the model summary. Once corrected, the user can resubmit the model in the same fashion as the original submission. It is important to note that the presence of a model in the WARP repository does not reflect an endorsement of that model, and further that the RIAC does not warrant the applicability or accuracy of these models. Instead, the approval/rejection process simply determines whether or not a users submission satisfies the aforementioned criteria to be classified as a physics-based model.
Model Retrieval
The model retrieval process represents the main goal for the WARP development effort. The WARP database provides engineers, scientists and other reliability practitioners with a single-source repository of physics-based failure mechanism models. The features of the WARP interface allow users to search for relevant models and view a summarized report of the models that meet their search criteria, allowing them to select the appropriate model for their analysis. The search and retrieve functions are arguably the most important features of the WARP repository. To simplify the process of identifying suitable models, the WARP interface provides multiple techniques to filter the models of the database. Users can select models by component type, where the interface presents a list of the various component technologies under which one can view the relevant models. Similarly, WARP also allows users to select models by failure mechanism, where models are presented in the same manner. Users also have the option to use both fields to further narrow the selection of models. In addition to the general filtering techniques, WARP also provides a Search Panel, as previously depicted in Figure 1, where users can search for particular terms or phrases. The Advanced Search goes a step further and allows users to search within specific fields of the model summary, such as the title, author, characteristics and so on. Filtering or searching through the database will produce a list of models that meet the specified criteria. By selecting a title of interest, the user is directed to the webpage for that specific model, where he or she will have access to the complete model summary, including continued on next page http://theRIAC.org
Model Approval
Once submitted, the model approval process is performed by appropriate RIAC Subject Matter Experts (SMEs). Initially, analysts will use the models component technology and failure mechanism(s) to assign an appropriate SME to perform a comprehensive review. Before a review can begin, a reviewer must have access to the supporting documentation (e.g., journal article, conference proceeding paper, etc.) from which the model was extracted. Users may upload the document (if not subjected to copyright restrictions), provide a URL link to its location in the Bibliographic Citation, or loan a hard copy of the article to the RIAC. Note that all loaned documents are used solely for the model review process and are returned to the WARP Contributor upon completion of the review. The review process compares the users summarized data on the Model Specification Page to the information presented in the source documentation. This ensures that all relevant details are included in the model summary, since a user who is familiar with the model may overlook an important piece of information. The review also verifies that the submitted equations and parameter definitions match those in the original paper, and that the provided text and numerical values do not contain any mistakes. Essentially, the review process ensures that a users submission is, in fact, a physics-based model, and that the provided information clearly and concisely summarizes the characteristics, limitations, application and development of the model as intended by the author(s) of the source document.
31
DEVELOPMENT OF THE WEB-ACCESSIBLE REPOSITORY OF PHYSICS-BASED MODELS (WARP)
Figure 3: A Portion of the Webpage Summarizing a Model in the WARP Repository
all of the equations and necessary details. An illustration of the top portion of a model webpage is presented in Figure 3. As a repository for a multitude of physics-based models, the WARP database is only intended to provide a summary of the included models, to allow for a quick review and evaluation of its suitability for a given purpose. Individuals interested in applying a particular model for a reliability analysis are encouraged to acquire the source document in order to gain a complete understanding of the context under which the model was developed. The WARP interface reveals where such documents can be found, while also providing an opportunity to contact the corresponding author (using appropriate measures of privacy) with additional questions. WARP Members are also given an opportunity to comment on individual models based on their experiences and general impressions.
WARP Summary
Physics-of-Failure models are recognized by both government and industry for their ability to account for the effects of operational conditions on the failures experienced by a part or system. However, the aforementioned obstacles, particularly regarding the availability or accessibility of these models, have prevented them from seeing widespread use in the engineering community. The Web-Accessible Repository of Physics-Based Models (WARP) has been designed to alleviate the difficulties in the search for relevant models. In providing a single-source repository for PoF models, WARP will allow engineers in industry and academia to identify technologies for which models do (and do not) exist, while facilitating a growth in the use of informed physics-based modeling.
JANUARY 2012
Introducing...
...the newest member of the Quanterion Automated Reliability Toolkit family...

QuART PRO $189 Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes No Yes Yes Yes No No Yes No Yes Yes No No No Yes No No Yes No Yes Yes No Yes Yes No No Yes Yes Yes No No Yes No Yes Yes QuART ER $399 Yes Yes + Yes Yes Yes + Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes + Yes Yes Yes Yes Yes + Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes + Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes + Yes Yes Yes + QuART
With over two dozen new or improved tools!

General Knowledge Reliability Advisor Reliability Program Cost Reliability Imp rovement Cost Warranty Cost Statistical Distributions Checklists Definitions Acronyms Concept Reliability Potential Reliability Allocation Reliability Tailoring Reliability Approach Assessment Design Analysis Derating Thermal Design Failure Modes FMECA Worksheets Material Durability Improvement Parts Count Analysis Software Reliability Prediction Redundancy Modeling Reliability Adjustment Factors Interference Stress Strength Analysis Availability Calculator Testing Benefit of Reliability Reliability Growth Testing - Duane Method Reliability Growth Testing - AMSAA-Crow Method Reliability Growth Testing - Crow Extended Method Reliability Demonstration Test Equal Risk RDT Reliability Demonstration Test Variable Risk RDT Reliability Test Cost Accelerated Reliability Test WeiBayes Substantiation Testing Test Results / Confidence Exponential Distribution Test Results / Confidence Binomial Distribution Production Design of Experiments 2 Factor DOE Design of Experiments 3 Factor DOE Design of Experiments 4 Factor DOE HASS - ESS HASA Sampling Plans Process Capability Field Weibull Analysis Optimal Replacement Interval Sparing Analysis (Graphical) Spares Analysis (Tabular)
$99 No No No No No No Yes Yes Yes No Yes No Yes No No No No Yes No Yes Yes No No No Yes No No Yes No Yes Yes No Yes No No No Yes No No No No Yes No Yes Yes
Enhancing Reliability
Quanterion Automated Reliability Toolkit
Download for only $399
QUANTERION
SOLUTIONS INCORPORATE D
(315)-732-0097/(877) 808-0097
quanterion.com
+ indicates improved features or functions
Aordable tools for improving reliability
Quanterion Solutions Incorporated the team member responsible for the day-to-day operations of the Reliability Information Analysis Center (RIAC) and is the prime contractor for the operations of the Data and Analysis Center for Software (DACS).
JANUARY
2012 Annual Reliability and Maintainability Symposium Reno, NV January 23, 2012 thru January 26 2012 Contact: RAMS // P 603.863.2832 // F 603.863.2832 // http://rams.org/
MAY
2012 Industrial and Systems Engineering Research Conference Orlando, FL May 19, 2012 thru May 23, 2012 Contact: Institute of Industrial Engineers // P 800.494.0460 or 770.449.0460 // http://www.iienet2.org/Annual2/Default.aspx 2012 Environment, Energy Security and Sustainability Symposium and Exhibition New Orleans, LA May 21, 2012 thru May 24, 2012 Contact: Kari King, CMP, Associate Director // P 703.247.2588 // kking@ndia.org, Systems & Software Technology Conference 2012 Salt Lake City, UT TBD Contact: Software Technology Support Center // P 801. 777.9828 // sstc@hill.af.mil
JANUARY
S 1 8
M 2 9
T 3
W 4
T 5
F 6
S 7
FEBRUARY
ICRSS 2012 : International Conference on Reliability and Structural Safety Penang, Malaysia February 22, 2012 thru February 24, 2012 Contact: http://www.waset.org/conferences/2012/penang/icrss/
10 11 12 13 14
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
FEBRUARY
MARCH
S 7 M 1 8 T 2 9 W 3 T 4 F 5 S 6
10 11 12 13
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
RIAC Open Training Program San Diego, CA March 13, 2012 thru March 15, 2012 Contact: Pat Smalley, Reliability Information Analysis Center // P 877.363.7422 or 315.351.4200 // F 315.351.4209 // psmalley@theRIAC.org Reliability-Centered Maintenance and Root Cause Analysis Conference Ft. Meyers, FL March 20, 2012 thru March 23, 2012 Contact: Reliability Performance Institute // P 239.333.2500 // crm@reliabilityweb.com 28th Annual National Logistics Conference and Exhibition Miami, FL March 26, 2012 thru March 29, 2012 Contact: Megan Edwards // P 703.247.2566 // medwards@ndia.org
JUNE
Condition Monitoring and Predictive Maintenance Forum Ft. Meyers, FL June 5, 2012 thru June 8, 2012 Contact: Reliability Performance Institute // P 239.333.2500 // crm@reliabilityweb.com RIAC Open Training Program Virginia Beach, VA June 5, 2012 thru June 7, 2012 and June 12, 2012 thru June 14, 2012 Contact: Pat Smalley, Reliability Information Analysis Center // P 877.363.7422 // F 315.351.4209 // psmalley@theRIAC.org Mega Rust 2012: U.S. Navy Corrosion Conference San Diego, CA June 12, 2012 thru June 14, 2012 Contact: Teresa Denchfield // P 703.836.6727 // tdenchfield@ navalengineers.org International Applied Reliability Symposium New Orleans, LA June 13, 2012 thru June 15, 2012 Contact: ReliaSoft // P 888.886.0410 or 520.886.0410 // info@ ARSymposium.org 11th International Probabilistic Safety Assessment and Management Conference & The Annual European Safety and Reliability Conference Helsinki, Finland June 25, 2012 thru June 29, 2012 Contact: CONGREX / Blue & White Conferences Oy // P: +358 9 5607 500 // F +358 9 5607 5020 // psam11@congrex.fi
MARCH
S 6
M 7
T 1 8
W 2 9
T 3
F 4
S 5
10 11 12
13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
APRIL
S 3
M 4
T 5
W 6
T 7
F 1 8
S 2 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
APRIL
2012 IEEE International Reliability Physics Symposium Anaheim, CA April 15,, 2012 thru April 19, 2012 Contact: Prasad Chaparala // P 408.886.9381 // prasadc@ altadevices.com 13th Annual Science & Engineering Technology Conference Charleston, SC April 17, 2012 thru April 19, 2012 Contact: Brant Murray // P 703.247.2572 // bmurray@ndia.org ESTECH 2012 Orlando, FL April 30, 2012 through May 3, 2012 Contact: HeatherWooden // P 847. 981.0100, ext. 20 // marketing@iest.org The Fifth International Conference on Communication Theory, Reliability, and Quality of Service Mont Blanc, France April 29, 2012 thru May 4, 2012 Contact: http://www.iaria.org/conferences2012/CTRQ12.html
MAY
S 1 8
M 2 9
T 3
W 4
T 5
F 6
S 7
10 11 12 13 14
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
JUNE
S 5
M 6
T 7
W 1 8
T 2 9
F 3
S 4
10 11
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
JULY
S 3
M 4
T 5
W 6
T 7
F 1 8
S 2 9
JULY
4th International Conference on applied Human Factors and Ergonomics San Francisco, CA July 21, 2012 thru July 25, 2012 Contact: Laura Abell, Conference Administrator // aeiadmin@ purdue.edu
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
JANUARY 2012
i RAC
JANUARY 2012
Please fax completed survey to 315.351.4209 ATTN: Journal Editor

R 20 TE AR . 2 QU NO ND 18 , E CO SE LU M VO 10
) AC r (RI by nte ed Ce onsor sis aly r sp r An nte Cente on Ce ati sis ation orm Analyorm y Inf on Inf ilit ati ical liab orm hn e Re Inf Tec Th DoDfense is a De the
i C RA
The Reliability Information Analysis Center (RIAC) is a DoD Information Analysis Center sponsored by the Defense Technical Information Center
i RAC
VOLUME 19, NO. 1 JANUARY 2011
Th is e Re thea DoD liabilit De Inf y fen orm Inf orm se Tec ation ati hn An on ica aly An l Inf sis aly orm Ce sis ati nter Cente on sp Ce on r (RI nte sor AC ed ) r by
RA iC
VO JU LU M LY 20 E 19 , 11 NO
RIAC JOURNAL SURVEY

Hard Copy
JO
OF TH E RE
LIA
BI
UR
LIT Y IN FO
NA
AT IO N AN AL YS
L
CE NT
ER
IS
.2
RM
Journal Format
Web Download
ss f ner hA to en Ow ealt of pm dH ost elo s an lC Dev osi tion OF THE RELIABILIT Y INFORMATION ANALYSIS CENTER Tota and gn Pro In0 2ec educe sp ign S) ty CR Des HM bili ive is to O S r the s (V elia truct lR alys F O foT - E em ura 0-DesThermal CyclingTtFailures - Part One of Two An ( V isR H BstN E Sy ct on 2 ility1 8 g H alys M b ki F An emen E U .S n I T A Stru ng N S SO 0 8ila rader History ) Trac Reliability L 0 2 Usi g Ava A Shortanag D E of P R O G . A R MN A LY OF th TH eg h L E R n RA Y V SIS E RE D2 8lt M A D -row C R G Usi M EH TR LIA ty F R E E E P BI A form Hea T H E R IC LIT ES 10 bili LE DEY Plat hicle O L FA T IN elia M A L H E S PA FO DE IGU RM eR A LT C 1 8 Ve CY AT RJ E CL IO cret H E TO N OI MO IN MA AN Dis O NT D G AL NA L A YS FA R E E L PA IS S 24 G
essm
to en
am irfr fA ship
es
journal
jo ur
I LU
na
CE NT
RE
LIA EM A D BI RAM EN E SLIT E T S SIG Y E TER PA YS N-A RT ST S F TE I IM TW O M D AT R S O AC IO OF N & TW SN O AG
ER
How satisfied are you with the content (article technical quality) in this issue of the Journal?
Very Satisfied Satisfied Neutral Dissatisfied Very Dissatisfied
How satisfied are you with the appearance (layout, readability) of this issue of the Journal?
How satisfied are you with the overall quality of this issue of the Journal (compared to similar magazines, newsletters, etc.)?
How did you become aware of this issue of the Journal?

Subscribe Colleague Library RIAC Website RIAC Email Conference/Trade Show
Would you recommend the RIAC Journal to a colleague?

Definitely Probably Not Sure Probably Not Definitely Not
Please suggest general changes / improvements to the RIAC Journal that would improve your level of satisfaction:
Please suggest further topics for the RIAC Journal that might help it to better meet your needs:
Overall satisfaction
Very Satisfied (5) Satisfied (4) Neutral (3) Dissatisfied (2) Very Dissatisfied (1)
CONTACT INFORMATION Name Organization Address Country Phone My Organization is:
(OPTIONAL)
Position/Title Office Symbol City Email Fax Army Navy Air Force Other DoD/Government Industry Academic Other State Zip
I need help with a reliability, maintainability, quality, supportability, or interoperability problem.
Please contact me
http://theRIAC.org
35
Reliability Information Analysis Center 100 Seymour Road | Suite C101 | Utica, NY 13502-1348
i RAC
PRSRT STD U.S. POSTAGE PAID UTICA, N.Y. PERMIT NO. 566
THE RIAC ONLINE

The RIACs Desk Reference is a Virtual Knowledge Base of reliability know-how on best practices, analyses and test approaches. Save time and money at the RIACs online store where you can browse, order, and immediately download electronic versions of most of the RIACs products.
Online Product Store RMQSI Library Technical Answers The RIAC Journal Upcoming Training Courses Whats New at RIAC

2012 Journal RIAC Web - 2

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

2012 Journal RIAC Web - 2

Caricato da

Copyright:

Formati disponibili

The Reliability Information Analysis Center (RIAC) is a DoD Information Analysis Center sponsored by the Defense Technical Information

VOLUME 20, NO. 1 JANUARY 2012

24 RIAC ANNOUNCES MA JOR NEW PRODUC T/SERVICE RELEASES 2 8 D E V E LO P M E N T O F T H E W E B - A CC E S S I B L E R E P O S I TO R Y O F

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

FIFTY YEARS OF PHYSICS OF FAILURE

continued on next page http://theRIAC.org

FIFTY YEARS OF PHYSICS OF FAILURE

continued from page 5

Paid advertising appears on pages 22 and 34.

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

Introduction to Maintainability Engineering

Get instant online access for only $

Features & Benefits:

View the online demo at: http://theRIAC.org

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

Power Inverter FMEA Results

2. Power System FMEA Results

3. Operational Helicopter Field Test

Starter-Generator FMEA Results

Battery FMEA Results

Figure 1 Field Test on a OH-58D Kiowa Warrior Helicopter

continued on next page http://theRIAC.org

continued from page 9

4. Health Management Detection and Prognostic Technology Development

Starter-Generator Health Management Technology

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

continued from page 11

Battery Health Management Technology

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

continued from page 13

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

continued from page 15

Inverter Health Management Technology

Figure 16 Coefficient Estimation Block Diagram

Figure 19 Initial Turn-Off Peak Detection continued on next page http://theRIAC.org

continued from page 17

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

5. Health Management Hardware and Interface Implementation

Figure 23 MOSFETs OFF Two Component Parameter Estimation

continued from page 19

[6] [7] [8] [9]

A SHORT HISTORY OF RELIABILITY

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER // JANUARY 2012

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

RIAC Announces Major New Product/Service Releases

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

Proba Used i bility Distrib utions Reliab n ility En gineer ing

RELIABILITY INFORMATION ANALYSIS CENTER

March 20-22, 2012 // RIAC Training in San Diego, CA

$1,295.00 per attendee

$1,495.00 per attendee

Software Reliability & Integrity

$1,295.00 per attendee

$1,295.00 per attendee

$1,295.00 per attendee

$1,495. per attendee

For more information and to register visit http://theRIAC.org or call 877.363.7422

THE JOURNAL OF THE RELIABILITY INFORMATION ANALYSIS CENTER

Accelerated Reliability Testing

A Systems Approach to Reliability in Design

Software Reliability & Integrity

The Web-Accessible Repository of Physics-Based Models (WARP)