Mechanical Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/2263
Browse
2 results
Search Results
Item Direct Numerical Simulation of Non-Premixed Flame Extinction Phenomena(2010) Narayanan, Praveen; Trouve, Arnaud C; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Non-premixed flame extinction phenomena are relevant in a variety of com- busting environments, including but hardly limited to diesel engines, pool fires, and fire suppression scenarios. These disparate phenomena are controlled by various parameters that contain information on flame stretch, heat losses, composition of the fuel and oxidizer supply streams, etc. Direct Numerical Simulation (DNS) is used in the present study to provide fundamental insight on diffusion flame extinction under non-adiabatic combustion conditions. The list of DNS configurations include: (C1) counterflow laminar flames with soot formation and thermal radiation transport; (C2) coflow turbulent flames with soot formation and thermal radiation transport; (C3) counterflow laminar and turbulent flames interacting with a mist-like water spray. Configurations C1 and C2 use single-step chemistry while configuration C3 uses detailed chemistry (all cases correspond to ethylene-air combustion). Configuration C1 is also treated using large Activation Energy Asymptotics (AEA). The AEA analysis is based on a classical formulation that is extended to include thermal radiation transport with both emission and absorption effects; the analysis also includes soot dynamics. The AEA analysis provides a flame extinction criterion in the form of a critical Damköhler number criterion. The DNS results are used to test the validity of this flame extinction criterion. In configuration C1, the flame extinction occurs as a result of flame stretch or radiative cooling; a variation of configuration C1 is considered in which the oxidizer stream contains a variable amount of soot mass. In configuration C1, flame weakening occurs as a result of radiative cooling; this effect is magnified by artificially increasing the mean Planck soot absorption coefficient. In configuration C3, flame extinction occurs as a result of flame stretch and evaporative cooling. In all studied cases, the critical Damkohler number criterion successfully predicts transition to extinction; this result supports the unifying concept of a flame Damköhler number Da and the idea that different extinction phenomena may be described by a single critical value of Da.Item End-of-Life and Constant Rate Reliability Modeling for Semiconductor Packages Using Knowledge-Based Test Approaches(2009) Yang, Liyu; Bernstein, Joseph B; Reliability Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)End-of-life and constant rate reliability modeling for semiconductor packages are the focuses of this dissertation. Knowledge-based testing approaches are applied and the test-to-failure approach is approved to be a reliable approach. First of all, the end-of-life AF models for solder joint reliability are studied. The research results show using one universal AF model for all packages is flawed approach. An assessment matrix is generated to guide the application of AF models. The AF models chosen should be either assessed based on available data or validated through accelerated stress tests. A common model can be applied if the packages have similar structures and materials. The studies show that different AF models will be required for SnPb solder joints and SAC lead-free solder joints. Second, solder bumps under power cycling conditions are found to follow constant rate reliability models due to variations of the operating conditions. Case studies demonstrate that a constant rate reliability model is appropriate to describe non solder joint related semiconductor package failures as well. Third, the dissertation describes the rate models using Chi-square approach cannot correlate well with the expected failure mechanisms in field applications. The estimation of the upper bound using a Chi-square value from zero failure is flawed. The dissertation emphasizes that the failure data is required for the failure rate estimation. A simple but tighter approach is proposed and provides much tighter bounds in comparison of other approaches available. Last, the reliability of solder bumps in flip chip packages under power cycling conditions is studied. The bump materials and underfill materials will significantly influence the reliability of the solder bumps. A set of comparable bump materials and the underfill materials will dramatically improve the end-of-life solder bumps under power cycling loads, and bump materials are one of the most significant factors. Comparing to the field failure data obtained, the end-of-life model does not predict the failures in the field, which is more close to an approximately constant failure rate. In addition, the studies find an improper underfill material could change the failure location from solder bump cracking to ILD cracking or BGA solder joint failures.