Adams, Vance H. (Ph.D. Mechanical Engineering)
Three-Dimensional Study of Combined Conduction, Radiation, and Natural Convection From an Array of Discrete Heat Sources in Horizontal Narrow-Aspect-Ratio Enclosure
Three-dimensional natural convection flow and heat transfer are studied for a three-by-three array of discrete protruding heat sources on a horizontal substrate in an air-filled, rectangular, narrow-aspect-ratio enclosure with length, width, and height ratio of 6:6:1. The geometry is such that it describes heat transfer within compact, portable electronic equipment. The study examines the complex thermal interactions between the heat sources, substrate, fluid, and enclosure walls as affected by the thermal conductance of the walls and substrate and includes both numerical modeling and experiment. The influence of radiation on the overall heat transfer is given particular attention. The three-dimensionality of the problem was evident in the overall flow characteristics and in the convective heat transfer edge-effects on the heat source surfaces. Excellent agreement between temperature predictions on the heat sources and substrate and experimental measurements was obtained for modified Rayleigh numbers in the range of 9.7x105 to 1.5x107. The combined numerical and experimental study spans a range of Rayleigh numbers from the onset of natural convection to the transition to time-dependent behavior. It was found that the values of Rayleigh numbers obtained for these key points were lower than those obtained in other studies of simpler geometries. Aspects of compact thermal modeling of electronic components are also addressed, including a model validaiton methodology and analysis of the thermal behavior of the compact models. It was found that thermal models could be validated in simple but realistic systems and that these validated thermal models could be used in studies of more complex system geometries with multiple heat source interactions. Results show that system level considerations, including thermal radiation, dominate the accurate prediction of electronic package device temperature.
Tang, Lan (Ph.D. Mechanical Engineer)
A Multi-Scale, Conjugate Thermal Analysis Methodology for Convectively Cooled Electronic Enclosures
A novel multi-scale, integrated approach is developed for efficient thermal modeling of electronic systems. This involves setting up of coarse grid global (system level), and fine grid local (board and component level) models, and appropriate communication between the two. Several possible techniques for this communication have been evaluated and a preferred strategy has been adopted. The integrated modeling approach has been applied to one natural convection and two mixed convection and forced convection air-cooled configurations, populated with circuit cards containing discrete heating elements.
In order to reduce the computational effort in system level simulations, Symmetrically Coupled Gauss-Seidel (SCGS) based multigrid method has been implemented. Representation of components with reduced (or compact) models during system level analysis has also been studied. Reduced models with effective homogenized properties not only reduces number of grid points required for the system level modeling, but also smoothes the spatial variations in the thermal properties within the computational domain.
Experimental validation of the approach has been carried out for the two mixed and forced convection air-cooled configurations. Computations of the component juction temperature using the integrated approach were in good agreement (10%) with the measured values.
Upadhyayula, Kumar (Ph.D. Mechanical Engineering)
An Incremental Damage Superposition Approach for Surface Mount Electronic Interconnect Durability Under Combined Temperature and Vibration Environments
Combined thermal and vibration induced stresses can be more benign than vibration induced stresses at room temperature in stimulating interconnect fatigue failures depending on the applied load histories. This observation is in apparent contradiction of popular Palmgren-Miner?s hypothesis because interactions between applied loads are ignored in Miner?s hypothesis. Therefore, unwarranted application of simplified hypothesis to all instances can lead to erroneous field life estimates. Therefore, obtaining a meaningful acceleration transform, to extrapolate test predictions to field predictions, hinges upon being able to capture interactive effects in Physics-of-Failure (PoF) failure models. An incremental damage superposition approach (IDSA) is presented to assess and quantify the interactive damage caused by simultaneous application of thermal and vibration loads. In IDSA the rate of damage accumulation due to vibration loads is quantified as a function of applied temperature. Further, the slowly changing thermo-mechanical mean stresses experienced due to thermal cycling is treated as a mean stress that biases the fatigue damage accumulated by rapidly changing vibration induced stresses. These two features allow IDSA to capture the complex synergy between temperature and vibration loads. The confidence in applying a macroscopic approach such as IDSA is enhanced by the micromechanical understanding of fatigue failures. The confidence in applying a macroscopic approach such as IDSA is enhanced by the micromechanical understanding of fatigue failures. The thesis hopes to provide a micromechanistic perspective of a commonly observed macro phenomena. The contribution made by Prof. Mura and Nakasone in proposing a dislocation mechanics based model to account for fatigue crack initiation is profoundly acknowledged. The thesis proposes to enhance their model by evaluating the role played by mean stress effects in the fatigue crack initiation process. The dislocation based micromechanical perspective lets us verify whether a macro-approach, such as the generalized Coffin-Manson model, is appropriate for including mean stress effects on fatigue damage or whether research should focus on a different functional dependence of fatigue damage on mean stress.
Zhang, Yibin (M.S. Mechanical Engineering)
Case Studies of Semiconductor Component Reliability
Semiconductor component reliability has exhibited dramatic improvements over the past two decades. The first case study of this thesis assessed the trend in the semiconductor component reliability based on some key studies which describe the time-to-failure of semiconductor components as a function of steady state temperature and ambient relative humidity, using activation energy based reliability models. The quantitative findings of this study show that time-to-failiure of semiconductor components has doubled every fifteen months on average since 1975. This trend suggests that it is no longer justifiable to assume that the average activation energy, often used in qualification specifications and in predictions is constant. In addition, this study shows that the acceleration factor used for test-to-field analysis is doubling every five years. As a result, future accelerated life tests willl have to be modified to save test time and money.
Improper storage is one factor that can precipitate failures in electronic products, especially in a hostile environment. In the second case study of this thesis, the effects of storage environments and intermittent operation of electronic equipment were investigated. In this case, failure were observed in th epropulsion logic modules on trains that were stored in Taiwan for nearly three years. Four ICs from these modules were analyzed to determine the possible causes of failures. Results suggested that the fialures of two plastic encapsulated packages were due to the occurrence of ionic ingress and corrosion, while no such failures were found in two hermetic ceramic packages.