Thesis (M.S.)--University of Missouri--Rolla, 2007. / Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed February 5, 2008) Includes bibliographical references (p. 82).

Thesis (Ph.D)--Mechanical Engineering, Georgia Institute of Technology, 2009. / Committee Chair: Joshi, Yogendra; Committee Member: Gallivan, Martha; Committee Member: Graham, Samuel; Committee Member: Yeung, Pui-Kuen; Committee Member: Zhang, Zhuomin. Part of the SMARTech Electronic Thesis and Dissertation Collection.

The complex turbulent flow regimes encountered in many thermal-fluid engineering applications have proven resistant to the effective application of systematic design because of the computational expense of model evaluation and the inherent variability of turbulent systems. In this thesis the integration of the Proper Orthogonal Decomposition (POD) for reduced order modeling of turbulent convection with the application of robust design principles is proposed as a practical design approach. The POD has been used successfully to create low dimensional steady state flow models within a prescribed range of parameters. The underlying foundation of robust design is to determine superior solutions to design problems by minimizing the effects of variation on system performance, without eliminating their causes. The integration of these constructs utilizing the compromise Decision Support Problem (DSP) results in an efficient, effective robust design approach for complex turbulent convective systems. The efficacy of the approach is illustrated through application to the configuration of data center server cabinets. Data centers are computing infrastructures that house large quantities of data processing equipment. The data processing equipment is stored in 2 m high enclosures known as cabinets. The demand for increased computational performance has led to very high power density cabinet design, with a single cabinet dissipating up to 20 kW. The computer servers are cooled by turbulent convection and have unsteady heat generation and cooling air flows, yielding substantial inherent variability, yet require some of the most stringent operational requirements of any engineering system. Through variation of the power load distribution and flow parameters, such as the rate of cooling air supplied, thermally efficient configurations that are insensitive to variations in operating conditions are determined. This robust design approach is applied to three common data center server cabinet designs, in increasing levels of modeling detail and complexity. Results of the application of this approach to the example problems studied show that the resulting thermally efficient configurations are capable of dissipating up to a 50% greater heat load and 15% decrease in the temperature variability using the same cooling infrastructure. These results are validated rigorously, including comparison of detailed CFD simulations with experimentally gathered temperature data of a mock server cabinet. Finally, with the approach validated, augmentations to the approach are considered for multi-scale design, extending approaches domain of applicability.

Future generation of all-electric ships will be highly dependent on electric power, since every single system aboard such as the drive propulsion, the weapon system, the communication and navigation systems will be electrically powered. Power conversion modules (PCM) will be used to transform and distribute the power as desired in various zone within the ships. As power densities increase at both components and systems-levels, high-fidelity thermal models of those PCMs are indispensable to reach high performance and energy efficient designs. Efficient systems-level thermal management requires modeling and analysis of complex turbulent fluid flow and heat transfer processes across several decades of length scales. In this thesis, a methodology for thermal modeling of complex PCM cabinets used in naval applications is offered. High fidelity computational fluid dynamics and heat transfer (CFD/HT) models are created in order to analyze the heat dissipation from the chip to the multi-cabinet level and optimize turbulent convection cooling inside the cabinet enclosure. Conventional CFD/HT modeling techniques for such complex and multi-scale systems are severely limited as a design or optimization tool. The large size of such models and the complex physics involved result in extremely slow processing time. A multi-scale approach has been developed to predict accurately the overall airflow conditions at the cabinet level as well as the airflow around components which dictates the chip temperature in details. Various models of different length scales are linked together by matching the boundary conditions. The advantage is that it allows high fidelity models at each length scale and more detailed simulations are obtained than what could have been accomplished with a single model methodology. It was found that the power cabinets under the prescribed design parameters, experience operating point airflow rates that are much lower than the design requirements. The flow is unevenly distributed through the various bays. Approximately 90 % of the cold plenum inlet flow rate goes exclusively through Bay 1 and Bay 2. Re-circulation and reverse flow are observed in regions experiencing a lack of flow motion. As a result high temperature of the air flow and consequently high component temperatures are also experienced in the upper bays of the cabinet. A proper orthogonal decomposition (POD) methodology has been performed to develop reduced-order compact models of the PCM cabinets. The reduced-order modeling approach based on POD reduces the numerical models containing 35 x 109 DOF down to less than 20 DOF, while still retaining a great accuracy. The reduced-order models developed yields prediction of the full-field 3-D cabinet within 30 seconds as opposed to the CFD/HT simulations that take more than 3 hours using a high power computer cluster. The reduced-order modeling methodology developed could be a useful tool to quickly and accurately characterize the thermal behavior of any electronics system and provides a good basis for thermal design and optimization purposes.

Thermal characterization of electronic cabinets is becoming increasingly important, due to growing power dissipation and compact packaging. Usually, multiple length scales of interest and modes of heat transfer are simultaneously present. A steady reduced order thermal modeling framework for electronic cabinets was developed to provide an efficient method to model thermal transport across multiple length scales. This methodology takes advantage of compact modeling at the chip or component level and reduced order modeling at subsystem and cabinet levels. Compact models, which were incorporated into system level simulation, were created for components, and reduced order models (ROMs) were developed using proper orthogonal decomposition (POD) for subsystems and system. An efficient interfacial coupling scheme was developed using the concept of flow network modeling to couple the heat and mass flow rates and pressure at each interface, when interconnecting ROMs together to simulate the entire system. Thermal information was then subsequently extracted from the global modeling and applied to the component model for detailed simulation. A boundary profile-matching scheme for ROM of each subsystem was developed to broaden the applicability of the multi-scale thermal modeling methodology. The output profiles of the subsystem upstream can be transferred to the input profiles of the subsystems downstream by adding necessary flow straightening ducts during the snapshots generation process. A general method to create dynamic multi-layer compact models for components and modules was developed. These dynamic compact models were incorporated into enclosure level simulation. The dynamic reduced order model for the enclosure was developed using POD. The transient multi-scale thermal modeling approach was illustrated through an electronic enclosure with insulated gate bipolar transistor (IGBT) module. The multiscale thermal modeling methodology presented here was validated through experiments conducted on a simulated electronic cabinet and the test vehicle with hybrid cooling technique. The latter incorporated double-sided cooling with hybrid forced air convection, thermoelectric cooling, and micro-channel liquid cooling. The overall multi-scale modeling framework was able to reduced numerical models containing 107 DOF down to around 102, while still retaining an approximation accuracy of around 90% in prediction of chip junction temperature rises, compared to measurements.