Ds Kumar Mechanical Objective Pdf Free Download

[Billi Mayhue]

Liquidphase processing is often desired for selective product formation, such as during catalytic conversion of biomass, and for purification of waste-water streams. Since the liquid phase environment influences the adsorption properties and reactivity of solid catalysts and porous materials, significant performance changes have been observed as a function of solvent properties for various industrial catalytic and separation processes. There is a need to better understand these solvation effects on processes at solid-liquid interfaces and to develop novel multiscale models that permit one to explore solvation effects on surface catalyzed reactions and during adsorption in porous media such as zeolites. It is these tasks that this thesis addresses.

Specifically, the objective of our initial study is to identify the active site and understand the various solvation effects for the hydrodeoxygenation of organic acids derived from biomass over Pd catalysts to produce green diesel. This investigation finds that independent of solvent environment the Pd(100) surface is 3-7 orders of magnitude more active than the most dominant Pd(111). The presence of liquid water inhibits the most relevant decarbonylation mechanism and although decarboxylation is accelerated by the presence of water, over Pd catalysts the decarbonylation mechanism is the only relevant deoxygenation pathway.

We then developed an explicit solvation method based on a hybrid quantum mechanical and molecular mechanical (QM/MM) description of the total reaction system understand site-specific solvation effects on the adsorption behavior of molecules in porous materials such as zeolites. The developed method overcomes the conceptual limitations of both implicit solvation methods and classical force field methods. To test and validate our hybrid QM/MM-FEP (Free Energy Perturbation) technique, we studied methanol and ethanol adsorption at the acid sites in zeolite H-MFI in both the gas phase and in liquid water. Our explicit solvation calculations predicted an exergonic solvation free energy of adsorption for both methanol and ethanol, which is consistent with experimental data.

Finally, we applied our novel QM/MM-FEP method to study solvation effects on the adsorption characteristics and removal of sub-ppm level contaminants of emerging concern (CECs) such as clofibric acid, aspirin, and bisphenol A from waste-water streams with the help of nano-porous adsorbents. We found that in H-MFI zeolite all studied CECs exhibit an exergonic solvation effect during adsorption. In other words, the presence of liquid water increases the selectivity of the adsorbent to remove the CECs from water.

Legged animals navigate complex environments with incredible stability, agility and economy despite having significant neuromechanical constraints like large delays and highly compliant actuators. They do so partly by tuning the mechanics of their actuators (i.e. muscle-tendon units) to act in a context-dependent manner. This raises several questions, three of which are discussed in this thesis. (A) to what extent can you purely rely on the mechanics of your actuators? In particular, can muscle-tendon units reject perturbations like uneven terrain without changing neural control? (B) how does stability, agility and economy vary with changing muscle-tendon properties individually and how do they tradeoff? and (C) if morphology affects movement performance in animals, can we augment human function across multiple objective functions (namely stability agility and economy) simultaneously by augmenting the morphology of muscle-tendon units with passive wearable robots. To answer these questions in a causal, controllable and generative manner, we developed a framework where a single muscle-tendon unit is interacting with a mass in gravity through a lever arm in closed loop to generate cyclic movement with variable terrain (both in simulation and in-vitro closed-loop experiments), variable morphology (in simulation) and variable nervous system control (in simulation). Through our work, we show that (A) muscle-tendon units can rapidly stabilize a hopping body when faced with a sudden change in ground height despite zero change in neural control, (B) series elastic tendons variably influence stability, agility and economy of movement such that animals need to trade off stability, agility and economy when tuning their muscle-tendon properties and (C) passive elastic exoskeletons are able to simultaneously augment stability, agility and economy despite being 'spring-like' and unable to do net work themselves by shifting the mechanics of underlying muscle-tendon units. Through our research, : (1) we gain fundamental neuromechanical understanding of how animals enable stable, agile and economic movement by tuning their actuators and (2) we generate a template for the design of a new generation of bioinspired robotic actuators to enable legged and wearable robots to navigate the world in all its richness and complexity.

Prof. Suresh is the director of the Engineering Representations and Simulation Lab (ERSL) . The ERSL research group focuses on large-scale topology optimization, design for additive manufacturing, and high performance finite element analysis (FEA).

The novelty in the topology optimization approach is the concept of topological level-set that combines topological sensitivity and level-set in a simple and robust manner. This has resulted in innovative methods for handling manufacturing constraints, tracing Pareto curves in multi-objective optimization, designing multi-materials and compliant mechanisms.

In parallel, the research group has made several ground breaking advances in high-performance finite element analysis. The dual-representation strategy demonstrated how classic beam and shell theories can be used as efficient preconditioners for 3D FEA. The novel concept of tangled FEA extends classic FEA to tangled meshes containing inverted elements, bypassing the unsolved problem of mesh untangling. The group also proposed the idea of limited-memory deflated FEA to exploit modern multi-core CPUs and many-core graphic programmable units (GPUs).

For example, a Matlab-based design optimization toolbox Design Optimization Software accompanies the text Design Optimization using Matlab and SolidWorks, authored by Prof. Krishnan Suresh. This module also includes SolidLab, a Matlab-based interface to SolidWorks. Also available is a Matlab-based Medial Axis Generator (Matlab) for 2D objects.

A Matlab-based design reliabiity software toobox Design Reliability Software has also been recently added. In classic NURBS, the weights are equal along all physical coordinates. By allowing the weights to change independently in each physical coordinate, a new curve is generated which is called Generalized NURBS (GNURBS). GNURBS Lab is a MATLAB toolbox devised to generate and manipulate Generalized NURBS curves.

Large-scale FEA problems with millions of degrees of freedom are becoming commonplace in solid mechanics. The bottleneck in such problems is memory access. The objective of this project is to exploit assembly-free deflation techniques to accelerate FEA.

The primary computational bottle-neck in implicit structural dynamics is the repeated inversion of the underlying stiffness matrix. A fast inversion technique is proposed by combining the well-known Newmark-beta method, with assembly-free deflated conjugate gradient (AF-DCG) for large-scale problems.

Classic FEA breaks down if one or more elements gets inverted, i.e., if the mesh gets tangled. But, mesh tangling is unavoidable during mesh generation, mesh morhping and large-scale deformation. The objective of this research is to extend classic FEA to handle tangled meshes.

In quad meshes, nodes connected to exactly 4 quad elements are called regular; otherwise they are referred to as irregular or singular. Singular nodes are detrimental to FEA accuracy. A new singularity removal method has been developed to dramatically reduce the number of node singularities.

Topology optimization problems subject to several constraints are both theoretically and computationally challenging. The topological level-set method is combined here with augmented Lagrangian methods to solve such multi-constrained topology optimization problems.

As additive manufacturing expands into multi-material, there is a demand for efficient multi-material topology optimization. The classic approach is to impose constraints on the volume-fraction of each of the material constituents. This can artificially restrict the design space. Instead, the total mass and compliance are treated as conflicting objectives, and the corresponding Pareto curve is traced; no additional constraint is imposed on the material composition.

Hinges can lead to high stress concentration in compliant mechanisms. The topological sensitivity concept is exploited here to design hinge-free compliant mechanisms. These mechanisms exhibit high mechanical advantage and low stresses.

The wide-spread use of topology optimization has been deterred due to high computational cost and significant software/hardware investment. Our group has developed a cloud based implementation of topology optimization, hosted at www.cloudtopopt.com.

The objective in microstructural optimization is to find the distribution of one or more 'materials' that would result in a desired microscopic behaviour (ex: negative Poisson ratio). Our group has developed a highly efficient topological sensitivity based method for designing such microstructures and tracing their corresponding Hashin-Shtrikman curves.

The objective here is to exploit the inherent advantages of isogeometric analysis for multi-material topology optimization. Due to the unified parametrization of geometry, analysis and design space, the sensitivities are computed analytically.

In AM simulation, repeated meshing and insertion of new elements during material deposition can pose significant implementation challenges. Our group is developing an assembly-free framework for AM simulation that offers several advantages: (1) The workspace is meshed only once at the start of the simulation, (2) addition and deletion of elements is easy since the stiffness matrix is never assembled, and (3) the underlying linear systems of equations can be solved efficiently through assembly-free deflation methods.

3a8082e126