Physics-of-Failure Approach for Fan PHM in Electronics Applications

Transcription

1 Physics-of-Failure Approach for Fan PHM in Electronics Applications Hyunseok Oh I, Michael H. Azarian', Michael Pecht'<, Clifford H. White 3, Richard C. Sohaney', and Edward Rherrr' I Center for Advanced Life Cycle Engineering, University of Maryland, College Park, MD 20742, USA 2 Prognostics and Health Management Center, City University of Hong Kong, Hong Kong, P.R. China 3 Dell, Inc, One Dell Way, Round Rock, TX 78682, USA pecht@calce.umd.edu Abstract - Fans have been a widely accepted solution for thermal management of electronic equipment. A fan is a critical component that affects the performance of expensive systems such as telecom equipment, power supplies, and server computers. The failure of a fan can lead to intermittent or catastrophic failures of target systems. There is a growing interest in prognostics and health management (PHM) of fans in the electronics industry. A physics-of-failure (PoF) based PHM approach enables the assessment and prediction of reliability under actual loading conditions. The PoF approach for fan PHM is based on a failure modes, mechanisms and effects analysis (FMMEA). FMMEA of brushless direct current (BLDC) fans was conducted while taking into consideration the expected application load conditions. The causalities between the failure causes and failure modes ofbldc fans were demonstrated by focusing on each failure mechanism. The prioritization of potential failure mechanisms was performed with the estimation of risk priority numbers. The life expectancies of fan ball bearings based on two potential failure mechanisms, including fatigue and lubricant deterioration, were calculated and compared to each other. The advantages and disadvantages of the PoF-based approach to PHM of fans are discussed. A potential application of the PoF based PHM approach to improve fan reliability assessment is addressed. Challenges for the PoF approach for fan PHM in the electronics industry are also addressed, including predicting overstress failures and reestimating reliability after design upgrades. 1 INTRODUCTION For electronics companies that use fans as a subcomponent oftheir electronic products, the reliability of fans from different suppliers must be understood. Reliability is the ability of a product to function properly within specified performance limits, for a specified period of time, under the life cycle application conditions. Fan reliability is commonly represented by life expectancy. The estimation of life expectancy commonly relies on the life calculation of a bottleneck component with the highest risk; e.g., ball bearings in a fan. As discussed in several references [1][2] dealing with fan reliability, the life estimation of subcomponents in a fan is not a simple task. A fan is an electromechanical system composed of multiple subcomponents. Each subcomponent can experience one or more failure modes and mechanisms during the life cycle. In order to estimate fan life, potential failure sites, modes, causes, and mechanisms must be understood. Interest is growing in fan PHM on the part of both suppliers and OEMs in order to assess the reliability and/or quality of fans quickly and reduce testing time during qualification and accelerated life testing. Physics-of- failure (PoF) based PHM permits the assessment and prediction of the reliability of a /10/$ IEEE MU3123 product under its actual application conditions [3]. The PoF approach uses knowledge of a product's failure mechanisms and life cycle loadings. The first step for the PoF based PHM is to analyze the hardware architecture of a product of interest. Physical connections and the functional relationship between components in a product can be identified. The second step is to estimate the application loads during the life cycle of a product. The product may experience various loading conditions, including thermal, chemical, electrical, and mechanical loads, and even radiation from the environment. The third step is to conduct a failure modes, mechanisms, and effects analysis (FMMEA) in order to identify and prioritize potential failure mechanisms. The FMMEA should be accompanied by a stress analysis, which offers knowledge of the stress levels at a specific location for a given application load. The last step is to assess the reliability of a product by calculating the time to failure for a dominant failure mechanism at failure sites based on PoF models. The purpose of this article is to illustrate the PoF approach for fan PHM. A brushless direct current (BLDC) fan was investigated to understand the hardware architecture. During the fan's life cycle, the expected environmental loads were determined from published articles. Potential failure mechanisms for the BLDC fan were identified by following FMMEA procedures. In order to minimize confusion about FMMEA terms, definitions are provided. This procedure demonstrates how to identify and prioritize potential failure mechanisms of fans under expected loading conditions. As a final step for the estimation of fan life expectancy, two calculations based on two potential failure mechanisms are compared with each other. 2 HARDWARE ANALYSIS A fan is an air moving device that utilizes a rotating blade or impeller driven by an electric motor with electronic or mechanical commands [4]. According to the definition of a fan, a rotating blade and electric motor are core components that help a fan achieve its desired function; i.e., air movement. In general, the kinds of components that a fan comprises may have variations depending on the requirements of suppliers and customers. For example, brushed motors instead of brushless motors can be used in a fan in order to reduce costs despite potential side effects such as generation of metal particles and electrical sparks due to degradation of metal brushes. However, the functions of the core components in a fan do not change regardless ofthe specific design. A BLDC fan for a consumer electronics applications was selected for hardware analysis. Figure 1 presents the two core elements of a fan; i.e., the electric motor and blades. In Figure 2, the electric motor is disassembled into two parts: a stator in the fan housing and rotor. The blades are directly mounted on the rotor in an electric motor. A bar-shaped permanent magnet in the rotor is flexible enough to fit into the housing of the rotor and interacts with the electromagnetic force generated by 2010 Prognostics & System Health Management Conference

2 ,.... ~ \ ~\,(~ «;,J ', 4"~ I Electric motor Blade Figure 1 Brushless Direct Current (BLDC) Fan for Consumer Electronics Applications Figure 2 BLDC Fan: Stator in the Fan Housing (left) and Rotor with Blades (right) Stator --.. Back view) Figure 3 BLDC Drive and Stator the stator so that the rotor rotates. The rotor is supported by two ball bearings. Figure 3 highlights a BLDC drive and stator mounted on the fan housing. The BLDC drive consists of transistors, resistors, capacitors, and ICs which control the current flow in the stator windings for the purpose of commutation. The current flow through wound wires allows the stator to generate electromagnetic forces for the rotation of the rotor. 3 LOADS In order to assess the reliability of a fan, environmental and operational loads must be understood. The application loads during a fan's life cycle can occur in manufacturing, assembly, storage, handling, transportation, and operation. The IPC-9591 standard [5] specifies two distinct loading conditions to be satisfied during the fan life cycle: nonoperating and operating. The loads during the non-operating portion of the life cycle include temperature, humidity, shock, and vibration. The environmental temperature should be limited to between -40 and +70 C. The environmental humidity variation should be within the range of 5 to 95%. The shock requirements are for a fan to withstand simulated conditions I with a single half-sign wave of 100 G for 2 msec or 50 G for 10 msec. And the vibration requirements also specify that a fan be able to withstand random and simulated vibration conditions for 30 minutes. The levels of random vibration are 0.03G 2/Hz from 10 to 100 Hz and 0.0IG 2/Hz from 100 to 700 Hz. The simulated vibration conditions are a sine wave with zero to peak amplitude of 0.5 G. The loads during fan operation include temperature, humidity, and dust. Details should be provided by each fan supplier. Fan specifications from two suppliers showed that the temperature should range from -10 to +70 C during fan operation. Other loads allowable during fan operation were not available in the specifications. 4 FAILURE MODES, MECHANISMS, AND EFFECTS ANALYSIS Failure modes, mechanisms, and effects analysis (FMMEA) is a systematic methodology which supports design-forreliability and root cause analysis of product failures. A failure mode is the effect by which a failure is observed. A failure mechanism is the physical, chemical, thermodynamic or other process or combination of processes that results in a failure. FMMEA helps to identify potential failure mechanisms and models for expected failure modes and prioritize them. The failure mechanism and model with the highest priority is a candidate for the PoF-based PHM approach to assess and predict a product's reliability. 4.1 Potential Failure Mechanisms The major components of a BLDC fan are chosen from the analysis of the hardware architecture in section 2: bearings, blades, a BLDC drive, and a stator. An overview of the potential failure modes, causes, and mechanisms of BLDC fans is listed in Table 1. Table 1 Potential Failure Modes, Causes, and Mechanisms ofbldc Fans Failure sites Failure modes Failure causes Failure mechanisms Seizure Thermal overload Lubricant deterioration Ball bearing Spalling Cyclic loading Fatigue Small furrows Moisture Corrosion Brinell mark Mechanical overload Yielding Blade Surface fouling Dust Adhesion Crack Cyclic loading Fatigue Open circuit High temperature Interdiffusion Short circuit Voltage bias, moisture Dendritic growth BLDC Open circuit on Thermal or De-adhesion drive wire bond mechanical shock BLDC motor Open circuit Moisture Corrosion Crack on solder Thermal cycling, Fatigue interconnect cyclic load Stator Cracked and Thermal overload, Thermal aging of peeling winding mechanical insulating materials wire film overload Ball Bearings: Lubricant Deterioration, Fatigue, Corrosion, and Yielding Ball bearings can have various failure modes and mechanisms depending on the application load. Deterioration of grease/oil in permanently lubricated ball bearings is caused by thermal overloads. Lubrication deterioration due to thermal overloads can be in the form of either physical or chemical degradation. The physical degradation is due to lubricant

3 evaporation. The chemical degradation is caused either by antioxidant consumption or lubricant oxidation. Lubricant deterioration leads to poor lubrication between bearing elements, resulting in seizure, the state of stopping relative motion between bearing components as a result of interfacial friction. When cyclic loads are applied to the bearing a very large number of times, the bearing will experience material fatigue. A dominant characteristic of fatigue failure is that the maximum value of the repeated stress is below the ultimate strength ofthe material and frequently below the yield strength. Cracks are initiated below the surfaces of bearing elements by cyclic loading and propagate to the surfaces, resulting in spalling. A spall is a defect where metallic particles flaked off from the surface. Bearing fatigue failure is usually determined by comparing the area of a spall and!or pit to the failure criterion value depending on the perspective of the bearing manufacturers. Ball bearings in fans are not usually sealed, but they are typically shielded from the environment. This provides an opportunity for ball bearings to be exposed to environmental contaminants such as moisture. Bearing metals under a humid environment are susceptible to corrosion, which is the destructive and unintentional attack of a metal. Oxidation products from the corrosion of bearing materials serve as abrasive media, generating small furrows on the surface of bearing components. Ball bearings are more vulnerable to mechanical overloads during non-operating rather than operating conditions. The viscosity of lubricant between bearing components becomes high during bearing operation. However, during non-operating conditions, the viscosity ofthe lubricant is not as high as during bearing operation. Localized damage to the surfaces of bearing components can be caused by mechanical overloads that exceed the yield point of the bearing materials. This can cause a permanent indentation, called a Brinell mark Fan Blades: Adhesion and Fatigue Fan blades are prone to contamination by dust from the environment. Contaminant particles can accumulate on the surface of fan blades due to adhesion. Several theories can be used to describe the adhesion phenomenon between fan blade surfaces and dust particles: physical adsorption, chemical bonding, electrostatic, and mechanical interlocking. The physical adsorption theory describes how two materials are held together by van der Waals forces, which are due to attraction between permanent and induced dipoles. The chemical bonding theory explains adhesion as the formation of covalent, ionic, or hydrogen bonds across the interface. The electrostatic theory explains the attraction force between double layers with electrically positive and negative charges. Lastly, the mechanical interlocking theory postulates that when a substrate has an irregular surface, dust particles may enter the irregularities. The fatigue mechanism is not described here again, since it was already explained in section Although spalls in the bearings and cracks on the blades in a fan are different failure modes, the underlying failure mechanism can be the same; i.e., fatigue BLDC Drive: Interdiffusion, Dendritic Growth, Deadhesion, Corrosion, and Fatigue A BLDC drive is composed of multiple subcomponents, including transistors, resistors, capacitors, and ICs on a printed circuit board. A number of potential failure mechanisms can be found in microelectronics packages within BLDC drives. Before explaining potentail failure mechanisms in a BLDC drive, it would be valuable to mention diffusion, since several failure mechanisms, including interdiffusion, dendritic growth, and corrosion, are driven by diffusion phenomena. Although diffusion is not usually a primary failure mechanism in mechanical, civil, and aerospace engineering applications, diffusion is a critical failure mechanism in electronic packages [6]. Diffusion is a time-dependent process involving the migration of an atomic, molecular, or ionic species within another material. In order to activate the migration of atoms or molecules from lattice site to lattice site, sufficient energy for disconnecting bonds must be provided to the atoms. The failures of electrical components by diffusion are manifested in the form ofan open or short circuit. Electronic packages in transistors and ICs can develop an open circuit by interdiffusion mechanism. When two different metals form an intimate contact, atoms in the metals diffuse across the interface. The difference of the diffusion rates between the two metals results in the depletion of atoms in one of the metals. Interdiffusion is a time-dependent mechanism. High temperature accelerates the interdiffusion process. An example of interdiffusion failure in electronic packages is the purple plague of aluminum wire bonds on gold bond pads [7]. Another potential failure mechanism in a BLDC drive is dendritic growth on a printed circuit board. Dendritic growth is an electrochemical process observed in a voltage biased condition. Metal ions migrate from the anode to the cathode regions developing dendrites. When the dendrite forms a complete bridge between the two electrodes, a short circuit occurs due to the increase of the leakage current across the bridge. The components in a BLDC drive such as transistors and ICs may fail when they are exposed to thermal or mechanical shock from the environment. This is an overstress failure. The failure is commonly observed in the form of an open circuit at the interface of a wire bond and the bond pad in the electronic packages of the components. The failure mechanism is deadhesion that is an opposite process to adhesion. De-adhesion of two materials is caused by mechanical work that exceeds the bonding strength at the interface. Details on two additional failure mechanisms, corrosion and fatigue, were already covered in section Potential failure causes and modes in a BLDC drive are discussed here. Moisture diffusion through plastic encapsulation in ICs and transistors may result in corrosion of the aluminum bond wires in electronic packages, which leads to an open circuit of the bond wires. Under thermal cycling or cyclic mechanical loads, the failure ofsolder joint interconnects is a common example of fatigue failures Stator in BLDC Motor: Thermal Aging ofinsulating Materials The principal function of a stator is sometimes lost due to a short in the winding wires. The failure is typically due to degradation of polymer films protecting and insulating the winding wires. Exposure to high temperature conditions can cause the polymer film to lose ductility and become brittle. When a mechanical load large enough to break the brittle

4 polymer film is applied to the wire, the exposed wires contact each other and high current flows through the short. The electrical short of the wires results in a reduction of the electromagnetic forces generated by the stator and leads to functional loss ofthe BLDC motor. 4.2 Prioritization ofpotential Failure Mechanisms The identification ofpotential failure mechanisms with high priority is achieved by past experience, stress analysis, accelerated tests, and engineering judgment. Past experience for fan failures can be acquired from the literature. Tian [2] reported fan failure data from qualification tests as well as analysis from field returns at the Hewlett-Packard Company. Bearings were the dominant failure site in mechanical failures followed by fan blades, which exhibited cracking. Dust adhesion should be considered a failure with a high occurrence, since consumer electronics products usually are not maintained in the field. The occurrence of electrical failure was lower than mechanical failure in the survey. The severity of the failure of mechanical components, including ball bearings and blades, is evaluated as performance degradation, while the severity of the failure of an electric motor is evaluated as a loss of function. The prioritization of potential failure mechanisms according to risk is summarized in Table 2. Lubrication deterioration, yielding of bearing materials, and dust adhesion were categorized as potential failure mechanisms with high risk. Table 2 Prioritization ofpotential Failure Mechanisms of BLDCFans Risk priority Failure sites Failure number Occurrence Severity mechanisms (occurrence Risk x severity) Lubricant deterioration High Ball bearing Fatigue Low Corrosion Low Yielding High Blade Adhesion High Crack Medium Interdiffusion Medium Dendritic BLDC growth Low drive De-adhesion Low BLDC Corrosion Medium motor Fatigue Medium Thermal Stator aging of insulating Medium materials Occurrence criteria: 5 (frequent), 4 (reasonably probable), 3 (occasional), 2 (remote), 1 (extremely unlikely) Severity criteria: 5 (total product failure), 4 (loss of function), 3 (performance degradation), 2 (operable at reduced performance), 1 (minor nuisance) Knowledge of life cycle application conditions and materials in the fan is used to estimate stresses at a specific location. The stress values calculated from environmental loading conditions are the input to PoF models that represent a relevant failure mechanism. A PoF model based on a wearout failure mechanism commonly offers time-to-failure under given loading conditions, while a failure model based on an overstress failure mechanism provides information on whether the failure occurs under a given stress level or not. In order to demonstrate a stress analysis ofball bearing failures in a BLDC fan, time-to-failure ofball bearings in a BLDC fan is calculated and compared for two failure mechanisms: lubricant deterioration and fatigue Bearing Life Expectancy for Lubricant Deterioration Failure The grease life equation represents the life for lubricant deterioration due to thermal overloads. The grease life equation has different constants depending on the type of lubricant. A grease life equation for general purpose grease used in fan bearings is as follows [8]. 10gL so = _ n _ - ( ṉ -Jr <: <: where l-ss is time to 50% failure (hours); nand N max are fan speed (rpm) and limiting speed (rpm) with grease lubrication, respectively; and T is operating temperature (OC). For example, two single-row radial ball bearings with an outer diameter of 8 mm, an inner diameter of 3mm, and a width of 4 mm were used in a BLDC fan for a consumer electronics application. The rated fan speed and limiting speed for the bearings were 4,800 rpm and 60,000 rpm, respectively. The operating temperature was 70 C. By substituting all these values into the grease life equation, the calculated grease life, Lso, was found to be approximately 5 years Bearing Life Expectancy for Fatigue Failure The term "rating life" of a group of nominally identical bearings is defined as the number of revolutions or hours at a constant speed that 90% of a group of bearings will achieve or exceed before the failure criterion is satisfied. Rating life is a term approved by the American Bearing Manufacturers Association (ABMA) to define bearing life. The rating life is a synonym for L IO life (time to 10% failure). The basic rating life is "a rating life for bearings manufactured with commonly used high quality material, good manufacturing quality, and operating under conventional operating conditions." In other words, the basic rating life represents the life for fatigue failure. The ISO 281 standard provides the basic rating life equation as follows [9]. where L IO is time to 10% failure (million revolutions); C, and P, are the dynamic load rating (N) and equivalent radial loading rating (N), respectively. For example, the dynamic load rating of the same bearing analyzed in section is 550 N from the bearing catalog. The stress applied to the bearings can be calculated at a rated operating condition of the BLDC fan. The radial and axial loads on the bearing are 0.70 N (half of the rotor blade weight) and 0.35 N (halfofthe pressure load applied to the rotor blade). The equivalent radial load from the radial and axial loads is 1.38 N. The calculated basic rating life is approximately 63.3x10 12 revolutions. When the fan rotates at 4,800 rpm, the basic rating life is 25,000 years. The estimated rating life is gigantic, which is reasonable, since the stress applied to the bearing is very small compared to the dynamic load rating of the bearing. In addition, the estimated rating life is only applied for fatigue failure. Other factors that affect bearing degradation were not considered in the estimation of bearing life. It was /10/$26.00 e 2010 IEEE MU Prognostics & System Health Management Conference

5 assumed that bearings are properly mounted, aligned, lubricated, loaded, sealed, and operated. 5 DISCUSSION OF THE POF APPROACH FOR FAN PHM In the FMMEA conducted for fan PHM, potential failure modes and mechanisms were identified. The potential failure mechanisms were prioritized by evaluating the occurrence and severity from the published literature. As a result, three failure mechanisms were identified as having the highest priority: lubricant deterioration in bearings, yielding ofbearing materials, and dust adhesion on the surface offan blades. Although the fatigue failure model under cyclic loading conditions is commonly used in bearing life calculation, the risk priority was estimated to be low compared to other failure mechanisms. The reason is that the applied load on the bearings is so small that the time required to generate damage on the surfaces of bearing components is extremely large. In comparing bearing life expectancies, the estimated time to 50% failure was 5 years for the lubricant deterioration mechanism, while the estimated time to 10% failure was 25,000 years for the fatigue mechanism. The time to 50% failure for fatigue failure would be longer than 25,000 years. Therefore, fan reliability assessment that is focused on the fatigue mechanism may not be as effective as an assessment that is focused on the lubricant deterioration mechanism. Dust adhesion on the surface of blades was identified as another potential failure mechanism with high priority. For consumer electronics applications, maintenance is not a common practice in the field. The user may regard the symptoms from dust adhesion as a failure. Therefore, the life estimation for dust adhesion can be an indicator which represents the susceptibility of a fan to dust contamination. A challenge for the life estimation for dust adhesion is that there is not a unifying PoF model for various kinds of dust particles. Damage accumulation models associated with dust adhesion need to be developed and validated for fans based on specific designs and applications in order to enable PHM for this mechanism. The yielding ofbearing materials was identified as another potential failure mechanism with high priority. Drops or shocks may occur during the non-operating life cycle of fans, including assembly, handling, and transportation. This may leave permanent indentations on the surface of bearing components in the form ofbrinell marks. A stress analysis can be conducted in order to estimate the allowable shock level. The risk associated with yielding of the bearing material could be estimated based on this stress analysis combined with a knowledge of the distribution of drop and shock events during the life cycle of a typical fan, obtained by continuous monitoring of units in the field. The estimated probability of failure over time could be updated by monitoring the occurrence of such events for individual fans during their life cycle. The PoF approach based on FMMEA has the potential to improve fan qualification and reliability assessment. In the traditional approach [5] for fan reliability assessment, loads applied during life testing are limited. Temperature and power cycling are the loads considered during the accelerated tests in the traditional approach, although other loads, including dust and mechanical overloads, were identified as equally critical in the FMMEA. The FMMEA is useful for selecting sensors for monitoring the load profile of fans. The PHM methodology helps to identify the real load profiles during the life cycle by in-situ monitoring. The identified load profile can be used as the basis for an accelerated test. The accelerated life test in the traditional approach takes more than 6 months despite the use of accelerated load conditions. This can be a large burden for both suppliers and OEMs. The PoF based PHM approach permits both the assessment and prediction of fan reliability. Failure precursors of a fan can be selected from the FMMEA or published studies [10]. A failure precursor is an event or a series of events that can be used to directly or indirectly indicate impending failure. By implementing sensors on a fan, the selected failure precursors can be monitored in-situ and the change of failure precursors can be detected during the accelerated life test. This approach is based on detecting impending failures of a fan, while the traditional approach waits until a fan completely fails. The PoF based PHM approach may reduce test time and save test cost significantly. In addition, time to failure can be predicted in real time during the accelerated life test by trending the precursor parameters. Conclusions The physics-of-failure approach for fan PHM, including hardware analysis, life cycle load analysis, and FMMEA, is presented. Common application loads during the non-operating life cycle of a fan are temperature, humidity, and shock, while the loads during the operating life cycle are temperature, humidity, and dust. Potential failure mechanisms with high priority are identified: lubricant deterioration in bearings, yielding of bearing materials, and dust adhesion on the surface offan blades. Life expectancies based on two failure mechanisms with high and low priority are compared. Fan bearing life expectancy for lubricant deterioration failure was calculated as 5 years, while fan bearing life expectancy for fatigue failure was calculated to be 25,000 years, which may have a low possibility of occurrence in the field. This demonstrates the importance of selecting a relevant failure mechanism for estimating the remaining useful life of a product. Future work for fan PHM will involve identifying the actual life cycle profile and health state of a fan based on sensor data. By in-situ monitoring failure precursors and dynamically predicting remaining useful life of a fan during an accelerated life test, the test time and cost required for fan reliability assessment can be significantly reduced. This work is expected to improve the qualification and reliability assessment process for fans. References 1. S. Kim, C. Vallarino, and A. Claassen, "Review of Fan Life Procedures," International Journal of Reliability, Quality, and Safety Engineering, Vol. 3 (1996), No.1, pp X. Tian, "Cooling Fan Reliability: Failure Criteria, Accelerated Life Testing, Modeling and Qualification," Proceeding of the Reliability and Maintainability Symposium, pp , Newport Beach, CA, USA, June 14-16, M. Pecht, Prognostics and Health Management of Electronics, Wiley-Interscience (2008), pp , New York, NY.

6 4. ISO 10302: Acoustics-Method for the Measurement of Airborne Noise Emitted by Small Air Moving Devices, 1st Edition, International Organization for Standardization (1996). 5. IPC-9591: Performance Parameters (Mechanical, Electrical, Environmental and Quality/Reliability) for Air Moving Devices, Association Connecting Electronics Industries (2006). 6. A. Dasgupta and M. Pecht, "Material Failure Mechanisms and Damage Models," IEEE Transactions on Reliability, Vol. 40 (1991), Issue 5, pp F. P. McCluskey, Y. D. Kweon, H. 1. Lee, 1. W. Kim, and H. S. Jeon, "Method for Assessing Remaining Life in Electronic Assemblies," Microelectronics Reliability, Vol. 40 (2000), Issue 2, pp NSK Rolling Bearing Catalog, Ell02e (2005), NSK Ltd, p. AI ISO 281: Rolling Bearings-Dynamic Load Ratings and Rating Life, 2nd Edition, International Organization for Standardization (2007). 10. H. Oh, T. Shibutani, and M. Pecht, "Precursor Monitoring Approach for Reliability Assessment of Cooling Fans," Journal ofintelligent Manufacturing, to be published. Biography Hyunseok Oh received the B.S. degree in mechanical engineering from Korea University, Seoul, Korea, and the M.S. degree in mechanical engineering from KAIST (Korea Advanced Institute of Science and Technology), Daejeon, Korea. He is currently pursuing a Ph.D. degree in mechanical engineering at the University of Maryland, College Park. His research interests include prognostics and health management ofelectronics. Michael H. Azarian received the B.S. degree in chemical engineering from Princeton University, Princeton, NJ, and the M.S. degree in metallurgical engineering and materials science and the Ph.D. degree in materials science and engineering from Carnegie Mellon University, Pittsburgh, PA. He spent 13 years in the disk drive, advanced materials, and fiber optics industries, most recently as Manager of Quality and Reliability with Bookham Technology. He is currently a Research Scientist with the Center for Advanced Life Cycle Engineering, University of Maryland, College Park. His research on reliability of electronic products has led to publications on electrochemical migration, capacitor reliability, electronic packaging, and tribology. He holds 5 U.S. patents for inventions in data storage and contamination control. Dr. Azarian was Technical Editor of the IEEE Standard 1624 on organizational reliability capability, for assessing suppliers of electronic products. He is on the editorial advisory board of Soldering and Surface Mount Technology. contributions to reliability research, 3M Research Award for electronics packaging, and the IMAPS William D. Ashman Memorial Achievement Award for his contributions in electronics reliability analysis. He served as chief editor ofthe IEEE Transactions on Reliability for eight years and on the advisory board of IEEE Spectrum. He is chief editor for Microelectronics Reliability and an associate editor for the IEEE Transactions on Components and Packaging Technology. He is the founder of CALCE (Center for Advanced Life Cycle Engineering) at the University ofmaryland, which is funded by over 150 of the world's leading electronics companies at more than US$6M1year. He is also a Chair Professor in Mechanical Engineering and a Professor in Applied Mathematics at the University of Maryland. He has written more than twenty books on electronic products development, use and supply chain management and over 400 technical articles. He consults for 22 major international electronics companies, providing expertise in strategic planning, design, test, prognostics, IP and risk assessment ofelectronic products and systems. Clifford H. White earned the B.S. degree in electrical engineering from Univ of Missouri at Rolla and a M.S. degree in electical engineering from Univ of Texas at Austin. His reliability engineering work experience started while working as a co-op student with Bell Telephone in 1963 and has continued with numerous electronics firms: Collins Radio, Texas Instruments, Apple Computer, Digital Switch, E Systems, CompuAdd, Micron Computer and Dell. His work assignments have ranged from individual contributor to director level. Richard C. Sohaney received the B.S. degree in mechanical engineering from Lehigh University, Bethlehem, PA. and M.S. degree in mechanical engineering from Purdue University, West Lafayette, IN. He has nearly 30 years experience in the sound and vibration field, including the aerospace, automotive, instrumentation, and information technology industries. Currently, he is in an acoustical engineering role at Dell, Inc. in the Product Development, Advanced Engineering group. Edward Rhem received the B.S. degree in electrical engineering from Prairie View A&M University, Prairie View Texas, the M.S. degree in business administration from St. Edwards University, Austin Texas, and engineering graduate studies in micro-computer engineering from Stanford University, Stanford California. He has over 27 years of development and test engineering experience gained from work at Bell Labs, IBM Corporation, and Dell Corporation. Michael Pecht is currently a visiting Professor in Electronic Engineering at City University in Hong Kong. He has an MS in Electrical Engineering and an MS and PhD in Engineering Mechanics from the University of Wisconsin at Madison. He is a Professional Engineer, an IEEE Fellow, an ASME Fellow and an IMAPS Fellow. He was awarded the highest reliability honor, the IEEE Reliability Society's Lifetime Achievement Award in He has previously received the European Micro and Nano-Reliability Award for outstanding