Many industrial statisticians employ the techniques of Response Surface Methodology (RSM) to study and optimize products and processes. A second-order Taylor series approximation is commonly utilized to model the data; however, parametric models are not always adequate. In these situations, any degree of model misspecification may result in serious bias of the estimated response. Nonparametric methods have been suggested as an alternative as they can capture structure in the data that a misspecified parametric model cannot. Yet nonparametric fits may be highly variable especially in small sample settings which are common in RSM. Therefore, semiparametric regression techniques are proposed for use in the RSM setting. These methods will be applied to an elementary RSM problem as well as the robust parameter design problem. / Ph. D.

The expanding role of electronic nose instrumentation, as a quality-monitoring tool for food-packaging materials, is examined and reviewed. The food industry is interested in determining the applicability of using an electronic nose for odor analysis of retained printing solvent levels in packaging. Three electronic nose systems were optimized for this application and their performance assessed. These include the FOX 3000, the Cyranose 320, and the QMB6. Response surface methodology was used to generate 2nd order models of sensor response as a function of system and experimental parameters for the three electronic nose systems. Forty-seven of 50 sensor models generated were found to be significant at an a-level of 0.05. Optimum settings, that allowed adequate signals to be obtained for the full range of examined retained solvents levels, were selected for the remaining work using these models. Performance analyses of these systems, which use three leading sensor technologies, showed that the conducting polymer sensor technology demonstrated the most discriminatory power. All three technologies proved able to discriminate among different levels of retained solvents. Each complete electronic nose system was also able to discriminate between assorted packaging having either conforming or non-conforming levels of retained solvents. Each system correctly identified 100% of unknown samples. Sensor technology had a greater effect on performance than the number of sensors used. Based on discriminatory power and practical features, the FOX 3000 and the Cyranose 320 were superior. The results indicate that electronic nose instrumentation can be used as a complimentary discriminatory tool in quality control. / Master of Science

This thesis focuses on practical and quantitative methods for measuring effectiveness in naval ship design. An Overall Measure of Effectiveness (OMOE) model or function is an essential prerequisite for optimization and design trade-off. This effectiveness can be limited to individual ship missions or extend to missions within a task group or larger context. A method is presented that uses the Analytic Hierarchy Process combined with Multi-Attribute Value Theory to build an Overall Measure of Effectiveness and Overall Measure of Risk function to properly rank and approximately measure the relative mission effectiveness and risk of design alternatives, using trained expert opinion to replace complex analysis tools. A validation of this method is achieved through experimentation comparing ships ranked by the method with direct ranking of the ships through war gaming scenarios. The second part of this thesis presents a mathematical ship synthesis model to be used in early concept development stages of the ship design process. Tools to simplify and introduce greater accuracy are described and developed. Response Surface Models and Design of Experiments simplify and speed up the process. Finite element codes such as MAESTRO improve the accuracy of the ship synthesis models which in turn lower costs later in the design process. A case study of an Advanced Logistics Delivery Ship (ALDV) is performed to asses the use of RSM and DOE methods to minimize computation time when using high-fidelity codes early in the naval ship design process. / Master of Science

Yes / The deodorization of the refined palm oil process is simulated here using ASPEN HYSYS. In the absence of a library molecular distillation (MD) process in ASPEN HYSYS, first, a single flash vessel is considered to represent a falling film MD process which is simulated for a binary system taken from the literature and the model predictions are compared with the published work based on ASPEN PLUS and DISMOL. Second, the developed MD process is extended to simulate the deodorization process. Parameter estimation technique is used to estimate the Antoine’s parameters based on literature data to calculate the pure component vapor pressure. The model predictions are then validated against the patented results of refining edible oil rich in natural carotenes and vitamin E and simulation results were found to be in good agreement, within a 2% error of the patented results. Third, Response Surface Methodology (RSM) is employed to develop non-linear second-order polynomial equations based model for the deodorization process and the effects of various operating parameters on the performance of the process are studied. Finally, an optimization framework is developed to maximize the concentration of beta-carotene, tocopherol and free fatty acid while optimizing the feed flow rate, temperature and pressure subject to process constrains. The optimum results of feed flow rate, temperature, and pressure were determined as 1291 kg/h, 147 C and 0.0007 kPa respectively, and the concentration responses of beta- carotene, tocopherol and free fatty acid were found to be 0.000575, 0.000937 and 0.999840 respectively. / Prince of Songkla University, Songkhla, Thailand for providing financial support (Grant code: PSU2554-022)

Este trabalho apresenta uma ferramenta flexível e eficiente para o campo de projeto e desenho preliminar de cascos de embarcações. O estado da arte de projeto de cascos de embarcações consiste em um processo iterativo onde, primeiramente, definem-se parâmetros geométricos, de estabilidade, e hidrodinâmicos. Em seguida, o casco é modelado através de curvas paramétricas. submetido a uma análise de desempenho, e, após isso, é feita a tomada de decisão para otimizar os parâmetros iniciais. O ciclo se repete até que se atinja o resultado desejado. O que será apresentado cumpre com a parte de modelagem paramétrica, e é capaz de gerar de forma automática e otimizada curvas que compõem o projeto de uma embarcação de recreio. A ferramenta recebe parâmetros de alto nível, como razões adimensionais, coeficientes de forma e restrições, e desenha tridimensionalmente um casco intrinsecamente suave, de forma que todos os parâmetros especificados sejam atendidos. Uma metodologia de projeto de cascos de veleiros é utilizada em conjunto com curvas B-Splines, submetidas a um algoritmo de otimização, em cuja função objetivo se empregam critérios de suavidade (\"fairness criteria\"). O problema de otimização de curvas B-Splines é não linear, pela própria formulação da função objetivo e suas restrições. Para tal, utiliza-se um método de Programação Quadrática Seqüencial. Para a otimização dos parâmetros principais do casco, utiliza-se uma metodologia de baixo custo, baseada na técnica de Superfície de Resposta, que otimiza uma função de mérito, no caso, a resistência residual, variando-se dois parâmetros do casco. A interface com softwares CAD é desenvolvida com o objetivo de se realizar outras análises, como de curvatura da superfície, além do projeto do restante da embarcação. Neste também pode ser gerado o bloco sólido para integração com softwares de CAM para a usinagem e fabricação do mesmo. / This article presents a flexible and efficient tool to the field of ship project and preliminary design by using an approach of parametric modeling of hull forms. The state of the art of design of hulls consists in an iterative process where, first, geometric, stability and hydrodynamic parameters are defined. After that, the hull is modeled through parametric curves, simulated in a performance analysis software, and, after that, it is taken the decision to optimize the initial parameters. The cycle repeats until it reaches the desired result. This work deals with the part of parametric modeling, and is capable to generate, in an automatic way, optimized curves that compose the design of the hull of a yacht. The tool receives high level parameters, such as non-dimensional ratios, coefficients and restrictions, and draws a three-dimensional hull, intrinsically fair, fulfilling all the specified parameters. A methodology for the design of sail-boats is presented together with B-Splines curves, submitted to an optimization algorithm, using fairness criteria as the objective function. The optimization problem of B-Splines curves is not linear, because of its mathematical formulation, the objective function and constraints adopted. Therefore, a method of Sequential Quadratic Programming is used. For the optimization of the main parameters of the hull, a cost-effective methodology, based on the Surface Response Method, is proposed, optimizing a merit criterion, such as, the residuary resistance, changing two parameters of the hull. The interface with CAD software is implemented in order to allow other analyses, such as surface curvature, and the design of the other parts of the boat. The drawing also can be transformed into a solid model, and integrated with a CAM software to be ready for itsmanufacturing.

ABSTRACT PARAMETER OPTIMIZATION OF CHEMICALLY ACTIVATED MORTARS CONTAINING HIGH VOLUMES OF POZZOLAN BY STATISTICAL DESIGN AND ANALYSIS OF EXPERIMENTS Aldemir, BaSak M.S., Department of Industrial Engineering Supervisor: Prof. Dr. &Ouml / mer Saat&ccedil / ioglu Co-Supervisor: Assoc. Prof. Dr. Lutfullah Turanli January 2006, 167 pages This thesis illustrates parameter optimization of early and late compressive strengths of chemically activated mortars containing high volumes of pozzolan by statistical design and analysis of experiments. Four dominant parameters in chemical activation of natural pozzolans are chosen for the research, which are natural pozzolan replacement, amount of pozzolan passing 45 &amp / #956 / m sieve, activator dosage and activator type. Response surface methodology has been employed in statistical design and analysis of experiments. Based on various second-order response surface designs / experimental data has been collected, best regression models have been chosen and optimized. In addition to the optimization of early and late strength responses separately, simultaneous optimization of compressive strength with several other responses such as cost, and standard deviation estimate has also been performed. Research highlight is the uniqueness of the statistical optimization approach to chemical activation of natural pozzolans.

The focus of this dissertation is on improving decision-maker trade-offs and the development of a new constrained methodology for multiple response surface optimization. There are three key components of the research: development of the necessary conditions and assumptions associated with constrained multiple response surface optimization methodologies; development of a new constrained multiple response surface methodology; and demonstration of the new method. The necessary conditions for and assumptions associated with constrained multiple response surface optimization methods were identified and found to be less restrictive than requirements previously described in the literature. The conditions and assumptions required for a constrained method to find the most preferred non-dominated solution are to generate non-dominated solutions and to generate solutions consistent with decision-maker preferences among the response objectives. Additionally, if a Lagrangian constrained method is used, the preservation of convexity is required in order to be able to generate all non-dominated solutions. The conditions required for constrained methods are significantly fewer than those required for combined methods. Most of the existing constrained methodologies do not incorporate any provision for a decision-maker to explicitly determine the relative importance of the multiple objectives. Research into the larger area of multi-criteria decision-making identified the interactive surrogate worth trade-off algorithm as a potential methodology that would provide that capability in multiple response surface optimization problems. The ISWT algorithm uses an ε-constraint formulation to guarantee a non-dominated solution, and then interacts with the decision-maker after each iteration to determine the preference of the decision-maker in trading-off the value of the primary response for an increase in value of a secondary response. The current research modified the ISWT algorithm to develop a new constrained multiple response surface methodology that explicitly accounts for decision-maker preferences. The new Modified ISWT (MISWT) method maintains the essence of the original method while taking advantage of the specific properties of multiple response surface problems to simplify the application of the method. The MISWT is an accessible computer-based implementation of the ISWT. Five test problems from the multiple response surface optimization literature were used to demonstrate the new methodology. It was shown that this methodology can handle a variety of types and numbers of responses and independent variables. Furthermore, it was demonstrated that the methodology can be successful using a priori information from the decision-maker about bounds or targets or can use the extreme values obtained from the region of operability. In all cases, the methodology explicitly considered decision-maker preferences and provided non-dominated solutions. The contribution of this method is the removal of implicit assumptions and includes the decision-maker in explicit trade-offs among multiple objectives or responses.

No description available.

Doctor of Philosophy / Department of Chemistry / Stefan H. Bossmann / Pancreatic cancer is the fourth leading cause of cancer-related deaths in the USA. Asymptomatic early cancer stages and late diagnosis leads to very low survival rates of pancreatic cancers, compared to other cancers. Treatment options for advanced pancreatic cancer are limited to chemotherapy and/or radiation therapy, as surgical removal of the cancerous tissue becomes impossible at later stages. Therefore, there's a critical need for innovative and improved chemotherapeutic treatment of (late) pancreatic cancers. It is mandatory for successful treatment strategies to overcome the drug resistance associated with pancreatic cancers. Nanotechnology based drug formulations have been providing promising alternatives in cancer treatment due to their selective targeting and accumulation in tumor vasculature, which can be used for efficient delivery of chemotherapeutic agents to tumors and metastases. The research of my thesis is following the principle approach to high therapeutic efficacy that has been first described by Dr. Helmut Ringsdorf in 1975. However, I have extended the use of the Ringsdorf model from polymeric to nanoparticle-based drug carriers by exploring an iron / iron oxide nanoparticle based drug delivery system. A series of drug delivery systems have been synthesized by varying the total numbers and the ratio of the tumor homing peptide sequence CGKRK and the chemotherapeutic drug doxorubicin at the surfaces of Fe/Fe₃O₄-nanoparticles. The cytotoxicity of these nanoformulations was tested against murine pancreatic cancer cell lines (Pan02) to assess their therapeutic capabilities for effective treatments of pancreatic cancers. Healthy mouse fibroblast cells (STO) were also tested for comparison, because an effective chemotherapeutic drug has to be selective towards cancer cells. Optimal Experimental Design methodology was applied to identify the nanoformulation with the highest therapeutic activity. A statistical analysis method known as response surface methodology was carried out to evaluate the in-vitro cytotoxicity data, and to determine whether the chosen experimental parameters truly express the optimized conditions of the nanoparticle based drug delivery system. The overall goal was to optimize the therapeutic efficacy in nanoparticle-based pancreatic cancer treatment. Based on the statistical data, the most effective iron/iron oxide nanoparticle-based drug delivery system has been identified. Its Fe/Fe₃O₄ core has a diameter of 20 nm. The surface of this nanoparticle is loaded with the homing sequence CGKRK (139-142 peptide molecules per nanoparticle surface) and the chemotherapeutic agent doxorubicin (156-159 molecules per surface), This nanoplatform is a promising candidate for the nanoparticle-based chemotherapy of pancreatic cancer.

The Physical Vapour Deposition (PVD) magnetron sputtering process is one of the widely used techniques for depositing thin film coatings on substrates for various applications such as integrated circuit fabrication, decorative coatings, and hard coatings for tooling. In the area of coatings on cutting tools, tool life can be improved drastically with the application of hard coatings. Application of coatings on cutting tools for various machining techniques, such as continuous and interrupted cutting, requires different coating characteristics, these being highly dependent on the process parameters under which they were formed. To efficiently optimise and customise the deposited coating characteristics, PVD process modelling using RSM methodology was proposed. The aim of this research is to develop a PVD magnetron sputtering process model which can predict the relationship between the process input parameters and resultant coating characteristics and performance. Response Surface Methodology (RSM) was used, this being one of the most practical and cost effective techniques to develop a process model. Even though RSM has been used for the optimisation of the sputtering process, published RSM modelling work on the application of hard coating process on cutting tool is lacking. This research investigated the deposition of TiAlN coatings onto tungsten carbide cutting tool inserts using PVD magnetron sputtering process. The input parameters evaluated were substrate temperature, substrate bias voltage, and sputtering power; the out put responses being coating hardness, coating roughness, and flank wear (coating performance). In addition to that, coating microstructures were investigated to explain the behaviour of the developed model. Coating microstructural phenomena assessed were; crystallite grain size, XRD peak intensity ratio I111/I200 and atomic number percentage ratio of Al/Ti. Design Expert 7.0.3 software was used for the RSM analysis. Three process models (hardness, roughness, performance) were successfully developed and validated. The modelling validation runs were within the 90% prediction interval of the developed models and their residual errors compared to the predicted values were less than 10%. The models were also qualitatively validated by justifying the behaviour of the output responses (hardness, roughness, and flank wear) and microstructures (Al/Ti ratio, crystallographic peak ratio I111/1200, and grain size) with respect to the variation of the input variables based on the published work by researchers and practitioners in this field. The significant parameters that influenced the coating hardness, roughness, and performance (flank wear) were also identified. Coating hardness was influenced by the substrate bias voltage, sputtering power, and substrate temperature; coating roughness was influenced by sputtering power and substrate bias; and coating performance was influenced by substrate bias. The analysis also discovered that there was a significant interaction between the substrate temperature and the sputtering power which significantly influenced coating hardness, roughness, and performance; this interaction phenomenon has not been reported in previously published literature. The correlation study between coating characteristics, microstructures and the coating performance (flank wear) suggested that the coating performance correlated most significantly to the coating hardness with Pearson coefficient of determination value (R2) of 0.7311. The study also suggested some correlation between coating performance with atomic percentage ratio of Al/Ti and grain size with R2 value of 0.4762 and 0.4109 respectively.