Pradeep George

Mercury and freon: temperature emulation and management for server systems

Power densities have been increasing rapidly at all levels of server systems. To counter the high temperatures resulting from these densities, systems researchers have recently started work on softwarebased thermal management. Unfortunately, research in this new area has been hindered by the limitations imposed by simulators and real measurements. In this paper, we introduce Mercury, a software suite that avoids these limitations by accurately emulating temperatures based on simple layout, hardware, and componentutilization data. Most importantly, Mercury runs the entire software stack natively, enables repeatable experiments, and allows the study of thermal emergencies without harming hardware reliability. We validate Mercury using real measurements and a widely used commercial simulator. We use Mercury to develop Freon, a system that manages thermal emergencies in a server cluster without unnecessary performance degradation. Mercury will soon become available from http://www.darklab.rutgers.edu.

A Compromise Experimental Design Method for Parametric Polynomial Response Surface Approximations

Abstract This study presents a compromise approach to augmentation of experimental designs, necessitated by the expense of performing each experiment (computational or physical), that yields higher quality parametric polynomial response surface approximations than traditional augmentation. Based on the D-optimality criterion as a measure of experimental design quality, the method simultaneously considers several polynomial models during the experimental design, resulting in good quality designs for all models under consideration, as opposed to good quality designs only for lower-order models, as in the case of traditional augmentation. Several numerical examples and an engineering example are presented to illustrate the efficacy of the approach.

A Compromise Method for the Design of Parametric Polynomial Surrogate Models

This study presents a compromise approach to augmentation of response surface (RS) designs to achieve the desired level of accuracy. RS are frequently used as surrogate models in multidisciplinary design optimization of complex mechanical systems. Augmentation is necessitated by the high computational expense typically associated with each function evaluation. As a result previous results from lower fidelity models are incorporated into the higher fidelity RS designs. The compromise approach yields higher quality parametric polynomial response surface approximations than traditional augmentation. Based on the D-optimality criterion as a measure of RS design quality, the method simultaneously considers several polynomial models during the RS design, resulting in good quality designs for all models under consideration, as opposed to good quality designs only for lower order models as in the case of traditional augmentation. Several numerical and an engineering example are presented to illustrate the efficacy of the approach.Copyright

Reliability-based optimisation of chemical vapour deposition process

In this paper, the process of Chemical Vapour Deposition (CVD) in a vertical impinging reactor is simulated and optimised using the reliability-based performance measure approach for the deposition of a thin film of silicon from silane. The key focus is on the rate of deposition and on the quality of the thin film produced. Proper control of the governing transport processes results in large area film thickness and composition uniformity. The effect of important design parameters and operating conditions are studied using numerical simulations. Response surfaces are generated for deposition rate and uniformity of the deposited film using compromise response surface method for the range of design variables considered. The resulting response surfaces are used to optimise the CVD system by considering the uncertainty in the design variables.

Optimization of Chemical Vapor Deposition Process

Chemical Vapor Deposition (CVD) process is simulated and optimized for the deposition of a thin film of silicon from silane. The key focus is on the rate of deposition and on the quality of the thin film produced. The intended application dictates the level of quality need for the film. Proper control of the governing transport processes results in large area film thickness and composition uniformity. A vertical impinging CVD reactor is considered. The goal is to optimize the CVD system. The effect of important design parameters and operating conditions are studied using numerical simulations. Then Compromise Response Surface Method (CRSM) is used to model the process over a range of susceptor temperature and inlet velocity of the reaction gases. The resulting response surface is used to optimize the CVD system.Copyright

2D/3D hybrid structural model of vocal folds

The spatial dimensionality of the vocal fold vibration is a common challenge in creating parsimonious models of vocal fold vibration. The ideal model is one that is accurate, with the lowest possible computational expense. Inclusion of full 3D flow and structural vibration typically requires massive amounts of computation, whereas reduction of either the flow or the structure to two dimensions eliminates certain aspects of physical reality, thus making the resulting models less accurate. Previous 2D models of the vocal fold structure have utilized a plane strain formulation, which is shown to be an erroneous modeling approach since it ignores influential stress components. We herein present a 2D/3D hybrid vocal fold model that preserves three-dimensional effects of length and longitudinal shear stresses, while taking advantage of a two-dimensional computational domain. The resulting model exhibits static and dynamic responses comparable to a 3D model, and retains the computational advantage of a two-dimensional model.

A two-dimensional/three-dimensional hybrid structural model of the vocal folds

The spatial dimensionality of the vocal fold vibration is a common challenge in creating parsimonious models of vocal fold vibration. The ideal model is one that provides acceptable accurate with the lowest possible computational expense. Inclusion of full three-dimensional (3D) flow and structural vibration typically requires massive amounts of computation, whereas reduction of either the flow or the structure to two dimensions eliminates certain aspects of physical reality, thus making the resulting models less accurate. Previous two-dimensional (2D) models of the vocal fold structure have utilized a plane strain formulation, which is shown to be an erroneous modeling approach since it ignores influential stress components. We herein present a 2D/3D hybrid vocal fold model that preserves three-dimensional effects of length and longitudinal shear stresses, while taking advantage of a two-dimensional computational domain. The resulting model exhibits static and dynamic responses comparable to a 3D model a...