Sustainable approaches have been extensively proposed in product, process and system levels. However, a lack of applicable solutions for these methods is identified in the existing research. This research considers uncertainties affecting sustainable systems and comprehensively discusses the need for the optimal design in product and system levels under uncertainty. Based on the economic, social and environmental requirements of a sustainable product, and uncertainties in engineering systems, two innovative methods are proposed. The methods, including agent-based modeling (ABM) and Big Data, quantify effects of users’ preference changes as a significant uncertainty source in a product design process. The effect of quantified uncertainties on the product sustainability is then evaluated, and solutions to reduce the effects are developed. Through a novel control engineering method, uncertainties are modeled in the design process of a product. Using two mathematical models, the cost and environmental impacts in the design process are minimized under users’ preference changes. The models search for an optimal number of iterations in the design process to achieve a sustainable solution. The methods have been extended to model and optimize the sustainable system design under uncertainties. Design of Eco-Industrial Parks (EIPs) is a practical and scientific solution to achieve sustainable industries. To improve the feasibility of flow exchanges between industries in an EIP under several uncertainties, this research provides a perspective analysis for establishing flow exchanges between industries. The sources of uncertainties in the EIPs are then comprehensively studied, and research gaps are highlighted. Finally, models to optimize flow exchanges between industries are presented and the validity of models is evaluated using real data. A major is including all sustainability pillars in the proposed approach. The research addresses users’ preferences to highlight the role of individuals in the society. Moreover, the economic and environmental objective functions have been considered for optimal decision making in the design process. This research underlines the role of uncertainty studies in the sustainable system design. Multiple classifications, perspective analysis, and optimization objectives are presented to help decision makers with the optimal design of sustainable systems under uncertainties. / February 2017

Thesis: S.M., Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2016. / S.M. !c Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science 2016 / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 83-88). / Cities today are strained by the exponential growth in population where they are homes to the majority of world's population. Understanding the complexities underlying the emerging behaviors of human travel patterns on the city level is essential toward making informed decision-making pertaining to urban transportation infrastructures This thesis includes several attempts towards modeling and understanding human mobility at the scales of individuals and the scale of aggregate population movement. The second chapter includes the development of a browser delivering visual insights of the aggregate behavior of populations in cities. The third chapter provides a computational framework for clustering regions in cities based on their attraction behavior and in doing so aids a predictive model in predicting inflows to newly developed regions. The fourth chapter investigates the patterns of individuals' movement at the city scale towards developing a predictive model for a persons' next visited location. The predictive accuracy is then increased by adding movement information of the population. The motivation behind the work of this thesis is derived from the demand of tools that provides fine-grained analysis of the complexity of human travel within cites. The approach takes advantage of the existing built infrastructures to sense the mobility of people eliminating the financial and temporal burdens of traditional methods. The outcomes of this work will assist both planners and the public in understanding the complexities of human mobility within their cities. / by Fahad Alhasoun. / S.M. / S.M. !c Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Thesis: S.M., Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2017. / S.M. !c Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science 2017 / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 91-94). / For an insurance company, assessing risk exposure for Property Damage (PD), and Business Interruption (BI) for large commercial clients is difficult because of the heterogeneity of that exposure, within a single client (account), and between different divisions, and regions, where the client is active. Traditional risk assessment models attempt to scale up the single location approach used in personal lines: A large amount of data is collected to profile a sample of the locations and based on this information the risk is then inferred and somewhat subjectively assessed for the whole account. The assumption is that the risk characteristics at the largest locations are representative of all locations, and moreover, that risk is proportional to the size of the location. This approach is both ineffective and inefficient. Thus our first goal is to build a better risk assessment model through machine learning based on clients' data from internal sources. Further, we define a new problem, to predict whether a specific contract would be profitable or unprofitable for the insurance company. This problem turns out to be an imbalance classification, which attracts the second half of our research efforts in this thesis. In Chapter 2, we first review related literature on state-of-the-art risk assessment models in the field of insurance. Later in the chapter we move to the imbalance classification problems and review some popular and effective solutions researchers have proposed. In Chapter 3, we describe the data structure, provide some preliminary analysis over certain attributes and discuss the preprocessing techniques used for feature construction. In Chapter 4, we propose a new model with the objective to develop a new risk index which represents clients' potential future risk level. We then compare the performance of our new index with the original risk index used by the insurance company and computational results show that our new index successfully captures clients' financial loss pattern, while the original risk score used by the insurance company fails to do so. In Chapter 5, we propose a multi-layer algorithm to predict whether a specific contract would be profitable or unprofitable for the insurance company. Simulation shows that we can accurately label more than 83 percent of the contracts on record and that our proposed algorithm outperforms traditional classifiers such as Support Vector Machines and Random Forests. Later in the chapter, we define a new imbalance classification problem and propose a hybrid method to improve the recall percentage and prediction accuracy of Support Vector Machines. The method incorporates unsupervised learning techniques into the classical Support Vector Machines algorithm and achieves satisfying results. In Chapter 6, we conclude the thesis and provide future research guidance. This thesis builds models and trains algorithms based on real world business data from a global leading insurance and reinsurance company. / by Ziyu Wang. / S.M. / S.M. !c Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

A method for allocating tolerances to exactly constrained assemblies is developed. The procedure is established as an optimization subject to constraints. The objective is to minimize the manufacturing cost of the assembly while respecting an acceptable level of performance. This method is particularly interesting for exactly constrained components that should be mass-produced. This thesis presents the different concepts used to develop the method. It describes exact constraint theory, manufacturing variations, optimization concepts, and the related mathematical tools. Then it explains how to relate these different topics in order to perform a tolerance allocation. The developed method is applied on two relevant exactly constrained examples: multi-fiber connectors, and kinematic coupling. Every time a mathematical model of the system and its corresponding manufacturing variations is established. Then an optimization procedure uses this model to minimize the manufacturing cost of the system while respecting its functional requirements. The results of the tolerance allocation are verified with Monte Carlo simulation.

Thesis (M. Eng.(Mechanical Engineering))--University of Pretoria, 2002. / Summaries in Afrikaans and English. Includes bibliographical references.

Thesis (M.Eng.)(Mechanical)--University of Pretoria, 2005. / title from opening screen (viewed March 20, 2006). Includes bibliographical references. Includes bibliographical references.

Recently parallel platforms, also known as Stewart platforms, have been the subject of much active research because of their distinct advantages over serially linked manipulators. Parallel platforms may have a great impact, especially in the field of machine tools. Parallel platforms are however, not yet commonly used as machine tools. The main reason for this is the lack of a general and rationally based design system that is also easily implementable. The availability of such a system will allow for the set-up of a platform so that, not only will the task be executable, but it will also be performed in an optimum manner according to a criterion specified by the user. This study proposes an easy to use methodology that may by applied to the optimum design of planar parallel platforms for machining applications. The design methodology presented here is based on mathematical optimization. This approach is simple and intuitive, and all the most important design criteria can be implemented, in some mathematical form, in the application of the proposed optimization methodology. Six possible design variables are defined, all related to the physical dimensions and placement of the platform for a prescribed task. Different platform designs are studied by using different combinations of design variables, where some design variables are fixed at specific values while the remaining design variables are allowed to vary. Two types of design constraints are considered, namely geometrical constraints that specify physical bounds on the platform size and placement, and secondly, limits on the maximum and minimum allowable actuator leg lengths. The platform design is optimized according to a prescribed criterion. Two design criteria, also called cost functions, are considered in this study. The first is the minimization of the actuator forces as the manipulator executes a prescribed task. The actuator forces are calculated by means of a dynamical analysis software package, DADS. The other design criterion is the maximization of the so-called quality index over the prescribed tool path. The success of mathematical design optimization depends largely on the optimization algorithm that is used to solve the minimization problem. It is shown in this study that the minimization of the actuator forces is a difficult problem to solve by mathematical optimization. Important reasons for this are that some of the cost functions considered here have discontinuities in their gradients with respect to the design variables, and that they are also highly non-convex and non¬quadratic. Another difficulty is the presence of numerical noise superimposed on the cost functions. For this reasons the robust and reliable LFOPC optimization algorithm, of Snyman was used. This method, although relatively slow to converge, was highly successful in solving the various design optimization problems. Because of the slow convergence of the method and the computational cost of evaluating the actuator force cost function and gradients when many design variables are involved, it was decided to attempt to speed up convergence by the use of an approximation method. The approximation algorithm. Dynamic-Q also proposed by Snyman, was selected and its application illustrated by solving a design problem with many design variables. This algorithm quickly converged to an acceptable optimum design. The main conclusion of this study is that the design methodology proposed here can successfully be applied to the design of planar Stewart platforms and may easily be extended to also apply to spatial Stewart platforms. This methodology solves difficult design problems that are inherent to the design of parallel manipulators. Its future implementation in a more comprehensive design and operating system for Stewart platforms to be used for machining tasks is therefore imperative. / Dissertation (M Eng (Mechanical Engineering))--University of Pretoria, 2007. / Mechanical and Aeronautical Engineering / unrestricted

Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2017. / Thesis: S.M., Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 57-60). / Fluid flow and particle transport through porous media are determined by the geometry of the host medium itself. Despite the fundamental importance of the velocity distribution in controlling early-time and late-time transport properties (e.g., early breakthrough and superdiffusive spreading), direct relations linking velocity distribution with the statistics of pore structure in 3D porous media have not been established yet. High velocities are controlled by the formation of channels, while low velocities are dominated by stagnation zones. Recent studies have proposed phenomenological models for the distribution of high velocities including stretched exponential and power-exponential distributions but without an underlying mechanistic or statistical physics theory. Here, we investigate the relationship between the structure of the host medium and the resulting fluid flow in random dense spherical packs. We simulate flow at low Reynolds numbers by solving the Stokes equations with the finite volume method and imposing a no-slip boundary condition at the boundary of each sphere. High fidelity numerical simulations of Stokes flow are facilitated with the assist of open source Computational Fluid Dynamics (CFD) tools such as OpenFOAM. We show that the distribution of low velocities in 3D porous media is described by a Gamma distribution, which is robust to variations in the geometry of the porous media. We develop a simple model that predicts the parameters of the gamma distribution in terms of the porosity of the host medium. Despite its simplicity, the analytical predictions from the model agree well with high-resolution simulations in terms of velocity distribution. / by Mohammad S Kh F Sh AlAdwani. / S.M.

Indiana University-Purdue University Indianapolis (IUPUI) / Plastic injection molding is a versatile process and a major part of the present plastic manufacturing industry. Traditional die design is limited to straight (drilled) cooling channels, which dont impart optimal thermal (or thermos-mechanical) per- formance. Moreover, reducing the cycle time in plastic injection molding has become significantly important to the industry nowadays. One approach that has been pro- posed is to use conformal cooling channels. With the advent of additive manufacturing technology, injection molding tools with conformal cooling channels are now possible. However, optimum conformal channels based on thermo-mechanical performance are not found. This study proposes a design methodology to generate optimized design configurations of such channels in plastic injection molds. Numerical models have been developed here to represent the thermo-mechanical behavior of the molds and predict the stress and cooling time. The model is then validated experimentally and used in conjunction with DOE (Design of Experiments) to study the effect of differ- ent design parameters of the channels on the die performance. Design of experiments (DOEs) is used to study the effect of critical design parameters of conformal channels as well as their cross section geometries. These DOEs are conducted to identify op- timal designs of conformal cooling channels which can be incorporated into injection molds that are used to manufacture cylindrical and conical shapes of plastic parts. Though these are simplified forms, the study provides useful insight into the poten- tial deign parameters for all kind of injection molds.Based on the DOEs, designs for best thermo-mechanical performance are identified (referred to as ”optimum”). The optimization study is basically a trade-off and the solution is based on a specific sample size. This approach is highly result-oriented and provides guidelines for selecting optimum design solutions given the plastic part thickness.

As progress pushes the boundaries of both the performance of new hardware components and the computational capacity of modern computers, the requirements on the performance of robotic systems are becoming more and more demanding. The objective of this thesis is to demonstrate that concurrent design (Co-Design) is the approach to follow to design hardware and control for such high-performance robots. In particular, this work proposes a co-design framework and an algorithm to tackle two main issues: i) how to use Co-Design to benchmark different robotic systems, and ii) how to effectively warm-start the trajectory optimization (TO) problem underlying the co-design problem aiming at global optimality. The first contribution of this thesis is a co-design framework for the energy efficiency analysis of a redundant actuation architecture combining Quasi-Direct Drive (QDD) motors and Series Elastic Actuators (SEAs). The energy consumption of the redundant actuation system is compared to that of Geared Motors (GMs) and SEAs alone. This comparison is made considering two robotic systems performing different tasks. The results show that, using the redundant actuation, one can save up to 99% of energy with respect to SEA for sinusoidal movements. This efficiency is achieved by exploiting the coupled dynamics of the two actuators, resulting in a latching-like control strategy. The analysis also shows that these large energy savings are not straightforwardly extendable to non-sinusoidal movements, but smaller savings (e.g., 7%) are nonetheless possible. The results highlight that the combination of complex hardware morphologies and advanced numerical Co-Design can lead to peak hardware performance that would be unattainable by human intuition alone. Moreover, it is also shown how to leverage Stochastic Programming (SP) to extend a similar co-design framework to design robots that are robust to disturbances by combining TO, morphology and feedback control optimization. The second contribution is a first step towards addressing the non-convexity of complex co-design optimization problems. To this aim, an algorithm for the optimal control of dynamical systems is designed that combines TO and Reinforcement Learning (RL) in a single framework. This algorithm tackles the two main limitations of TO and RL when applied to continuous-space non-linear systems to minimize a non-convex cost function: TO can get stuck in poor local minima when the search is not initialized close to a “good” minimum, whereas the RL training process may be excessively long and strongly dependent on the exploration strategy. Thus, the proposed algorithm learns a “good” control policy via TO-guided RL policy search. Using this policy to compute an initial guess for TO, makes the trajectory optimization process less prone to converge to poor local optima. The method is validated on several reaching problems featuring non-convex obstacle avoidance with different dynamical systems. The results show the great capabilities of the algorithm in escaping local minima, while being more computationally efficient than the state-of-the-art RL algorithms Deep Deterministic Policy Gradient and Proximal Policy Optimization. The current algorithm deals only with the control side of a co-design problem, but future work will extend it to include also hardware optimization. All things considered, this work advanced the state of the art on Co-Design, providing a framework and an algorithm to design both hardware and control for high-performance robots and aiming to the global optimality.