Stochastic control problems that arise in sequential decision making applications typically assume that information used for decision-making is obtained according to a predetermined sampling schedule. In many real applications however, there is a high sampling cost associated with collecting such data. It is therefore of equal importance to determine when information should be collected as it is to decide how this information should be utilized for optimal decision-making. This type of joint optimization has been a long-standing problem in the operations research literature, and very few results regarding the structure of the optimal sampling and control policy have been published. In this thesis, the joint optimization of sampling and control is studied in the context of maintenance optimization. New theoretical results characterizing the structure of the optimal policy are established, which have practical interpretation and give new insight into the value of condition-based maintenance programs in life-cycle asset management. Applications in other areas such as healthcare decision-making and statistical process control are discussed. Statistical parameter estimation results are also developed with illustrative real-world numerical examples.

This research employs a non-parametric, fractional, linear programming method, Data Envelopment Analysis to examine the Customer Service Representative resource allocation efficiency of a major Canadian bank’s model. Two DEA models are proposed, (1) to evaluate the Bank’s national branch network in the context of employment only, by minimizing Full Time Equivalent (FTE) while maximizing over-the-counter (OTC) transaction volume; and (2) to evaluate the efficacy of the Bank’s own model in meeting the desired customer satisfaction benchmarks by maximizing fraction of transactions completed under management’s target time. Non-controllable constant-returns-to-scale and variable-returns to-scale model results are presented and further broken down into branch size segments and geographical regions for analysis. A comparison is conducted between the DEA model results and the Bank’s performance ratios and benchmarks, validating the use of the proposed DEA models for resource allocation efficiency analysis in the banking industry.

This thesis investigates several uses of simulation methods to understand and control pandemic influenza in urban settings. An agent-based simulation, which models pandemic spread in a large metropolitan area, is used for two main purposes: to identify the shape of the distribution of pandemic outcomes, and to test for the presence of complex relationships between public health policy responses and underlying pandemic characteristics. The usefulness of pandemic simulation as a tool for assessing the cost-effectiveness of vaccination programs is critically evaluated through a rigorous comparison of three recent H1N1 vaccine cost-effectiveness studies. The potential for simulation methods to improve vaccine deployment is then demonstrated through a discrete-event simulation study of a mass immunization clinic.

Doctor of Philosophy / Department of Industrial & Manufacturing Systems Engineering / Chih-Hang John Wu / Quality is a result of product design and production control. Product design must maximize the ability to function across variations in production and environment. Production control must monitor and maintain the key design characteristics necessary for the intended function. Failure to do so results in premature part failure and increased costs. This has occurred in the production of modern cross ties. By designing quality into the product and production process, performance is maximized. This research presents a methodology for incorporating quality into the product design and production process. For product design, a relationship between product performance and design parameters is established by modeling techniques. These models provide a means to redesign the product to maximize performance and to understand the sensitivity of the design to fluctuation in production and environment. These models also establish the key design parameters that are critical for sustaining quality. For production, a method of monitoring the key design parameters is presented that provides an affordable means of automated inspection. Automated inspection removes operator error from the inspection process and allows for greater sampling rates to be achieved. The methodology presented allows for a potential of 100% inspection to be achieved with minimal impact to production costs. The research is applied to the analysis and quality control of pre-stressing steel reinforcement for concrete cross-ties. This application provides an opportunity to test and verify the research findings on a real world problem. Novel automated 3D spatial analysis algorithms are presented. This research furthers the state of the art of performing Geometrical Dimensioning and Tolerancing (GD&T). A cost effective method of non-contact surface profiling was developed with high resolution and high density surface profiles. The combined research findings present a methodology of achieving quality by design.

Doctor of Philosophy / Department of Industrial and Manufacturing Systems Engineering / Bradley Kramer / This research shows that industry and company related performance indicators enhance the business valuation process by providing a broader, more encompassing view of overall corporate health and a better understanding of improvement opportunity areas within a company. To incorporate performance indicators in the business valuation process, new methodologies are required to integrate the non-financial and soft data with the typical financial information used in business valuation. This requires a “re-think” of the standard business valuation process and the exploration and application of other methods and analytical techniques. The results of this research are the definition of a problem type and the development of a new business valuation process. The problem structure has as inputs industry specific performance metrics grouped into three primary areas Production Processes, Products/Services and Marketability and Management, a fuzzy logic model with fuzzy and approximate relationships between performance metrics and financial information and crisp financial information as output. The framework for a fuzzy logic model was developed and is used to approximate relationships and model a non-linear environment. The resulting crisp financial information is then input and integrated into the traditional business valuation process. The process was demonstrated with an example production company and with data from two regional airlines. A step-by-step example of the process was provided using the production company example to demonstrate how the results are generated and integrated with DCF business valuation. Heuristics to identify areas to improve company performance were described. Two regional airlines, individually and combined, were tested with actual data using the original fuzzy logic model structure and then the original fuzzy logic model structure was revised and new results generated. Tuning the model showed an improvement in the business valuation process performance. The benefits from this research include the definition of a new class of problems and a process to solve problems of this nature. The insights gained from this research can be applied in major disciplines such as accounting, business and finance, engineering and decision theory.

Doctor of Philosophy / Department of Industrial & Manufacturing Systems Engineering / Malgorzata J. Rys / This dissertation is the result of studying different methods of increasing guide sign visibility and legibility to drivers during nighttime, to increase safety on roadways. It also studies intersection lighting to indicate the lighting benefits on nighttime crash frequency reduction. From a survey conducted, practices related to overhead guide sign illumination and retroreflectivity in United States were summarized. A laboratory experiment was conducted to compare light distribution of five light sources: Metal Halide, Mercury Vapor, High Pressure Sodium, induction lighting, and Light Emitting Diode (LED). Cost analysis of the five light sources was performed. Combining results of the laboratory experiment and the cost analysis, induction lighting was recommended for states that want to continue external sign illumination. A retroreflectivity experiment was conducted to compare three types of retroreflective sheeting: Engineering Grade (type I), Diamond Grade (type XI), and High Intensity (type IV), to determine the sheeting that best increases visibility and legibility. Diamond Grade (type XI) was found to be the optimal sheeting that increases visibility and legibility to drivers during nighttime. A glare experiment was conducted to expand the retroreflectivity experiment results. Four sheeting-font combinations of High Intensity (type IV) and Diamond Grade (type XI) materials and Series E (Modified) and Clearview fonts were compared. Results revealed an optimal sheeting-font combination of Diamond Grade (type XI) sheeting and Clearview font which increases the visibility and legibility of guide signs to drivers under presence of oncoming glare source. The Highway Safety Information System (HSIS) database was used to study the effect of intersection lighting on the expected crash frequency. Illuminated intersections showed 3.61% and 6.54% decrease in the expected nighttime crash frequency as compared to dark intersections in Minnesota and California, respectively. In addition, partial lighting at intersections decreases the expected nighttime crash frequency by 4.72% compared to continuous lighting in Minnesota. The recommended sheeting-font combination for Departments of Transportation was Diamond Grade (type XI) and Clearview. This combination will increase signs’ visibility and legibility to drivers, and consequently increase safety on roadways. Adding partial lighting at intersections will reduce the expected nighttime crash frequency, and increase safety on roadways.

Doctor of Philosophy / Department of Chemical Engineering / Larry E. Erickson / Over the past decade, nanomaterials have been the subject of enormous interest. Their defining characteristic is a very small size in the range of 1-100 nm. Due to their nanometer size, nanomaterials are known to have unique mechanical, thermal, biological, optical and chemical properties, together with the potential for wide-ranging industrial applications. Here, we synthesized nanocrystalline metal oxides through the sol-gel process and used these materials as desulfurization adsorbents and photocatalysts. Deep desulfurization of fuels has received more and more attention worldwide, not only because of health and environmental consideration but also due to the need for producing ultra-low-sulfur fuels, which can only be achieved under severe operating conditions at high cost using hydrodesulfurization (HDS). Consequently, development of new and affordable deep desulfurization processes to satisfy the decreasing limit of sulfur content in fuels is a big challenge. Sol-gel derived Cu/Al[subscript]2O[subscript]3 and Zn/Al[subscript]2O[subscript]3 adsorbents have been demonstrated to be effective in the removal of thiophene from a model solution. Results showed that Cu[superscript]+ was the active site and thermal treatment under vacuum was critical for Zn/Al[subscript]2O[subscript]3 since a defective, less crystalline spinel led to stronger interaction between zinc ions and thiophene molecules in the adsorption process. The kinetic study suggested that most of the adsorption occurred in the first 30 min, and adsorption equilibrium was attained after 1.5 h. Both adsorbents showed good regenerative property. TiO2 is considered the most promising photocatalyst due to its high efficiency, chemical stability, non-toxicity, and low cost for degradation and complete mineralization of organic pollutants. However, the use of TiO[subscript]2 is impaired because it requires ultraviolet (UV) activation ([Lambda]<387 nm). The shift of optical response of TiO[subscript]2 from the UV to the visible light region would have a profound positive effect on the efficient use of solar energy in photocatalytic reactions. We shifted the optical response of TiO[subscript]2 and improved the photocatalytic efficiency through size modification and transition metal ion and nonmetal atom doping. Experimental results showed that C and V co-doped TiO[subscript]2 catalysts had much higher activity than commercial P25 TiO[subscript]2 towards the degradation of acetaldehyde under visible light irradiation. For the first time, we reported that activities were comparable in the dark and under visible light irradiation for co-doped TiO[subscript]2 with 2.0 wt% V. C and N co-doped TiO[subscript]2 exhibited higher activity for the degradation of methylene blue than pure TiO[subscript]2 under visible light and UV irradiation. Possible mechanisms were discussed based on the experimental results.

Caregivers injure their backs more than workers in any other industry. Efforts to reduce injuries have been on-going for decades with limited results. Mechanical lift devices have been incorporated into clinical practice over the past 30 years to reduce the risk of injury from patient lifting. Yet injury rates remain high. The use of mechanical lifts may be partly to blame. While these devices assist with lifting patients, they also introduce new activities that result in caregivers experiencing unsafe loading on the spine. We measured loads on the lower back during manoeuvres of the two most common lift types (overhead and floor) as well as during sling insertion. A new device called SlingSerterTM was evaluated for use in the clinical environment. We also investigated spine shrinkage as a measurement tool for estimating cumulative load. Caregivers worked alone and in pairs for both lift maneuvering and sling insertion activities. Overhead lift use resulted in much lower loads than floor lift use. We conclude caregivers can safely operate overhead lifts alone, while floor lift use remained unsafe even with two caregivers. Less-experienced caregivers had higher loads than more-experienced counterparts when using floor lifts. There was no corresponding effect of experience with overhead lift use and we found this to be a further benefit of overhead lifts over floor lifts. Most caregivers exceeded the safe limit for spine compression during sling insertion, though a single caregiver was at no higher risk of injury than two caregivers working together. Clinicians who tested SlingSerterTM agreed the device would be useful in clinical practice, particularly with bariatric patients and other special patient populations that are difficult to roll or turn. Finally, we investigated a novel method for estimating cumulative load based on spine shrinkage. There is growing recognition that excess cumulative load may be responsible for back injury. We found the variability in spine shrinkage was too large to estimate cumulative load directly. However, the technique may still be useful for determining the relative importance of the load from different activities to the cumulative total.

Many real-world systems experience deterioration with usage and age, which often leads to low product quality, high production cost, and low system availability. Most previous maintenance and reliability models in the literature do not incorporate condition monitoring information for decision making, which often results in poor failure prediction for partially observable deteriorating systems. For that reason, the development of fault prediction and control scheme using condition-based maintenance techniques has received considerable attention in recent years. This research presents a new framework for predicting failures of a partially observable deteriorating system using Bayesian control techniques. A time series model is fitted to a vector observation process representing partial information about the system state. Residuals are then calculated using the fitted model, which are indicative of system deterioration. The deterioration process is modeled as a 3-state continuous-time homogeneous Markov process. States 0 and 1 are not observable, representing healthy (good) and unhealthy (warning) system operational conditions, respectively. Only the failure state 2 is assumed to be observable. Preventive maintenance can be carried out at any sampling epoch, and corrective maintenance is carried out upon system failure. The form of the optimal control policy that maximizes the long-run expected average availability per unit time has been investigated. It has been proved that a control limit policy is optimal for decision making. The model parameters have been estimated using the Expectation Maximization (EM) algorithm. The optimal Bayesian fault prediction and control scheme, considering long-run average availability maximization along with a practical statistical constraint, has been proposed and compared with the age-based replacement policy. The optimal control limit and sampling interval are calculated in the semi-Markov decision process (SMDP) framework. Another Bayesian fault prediction and control scheme has been developed based on the average run length (ARL) criterion. Comparisons with traditional control charts are provided. Formulae for the mean residual life and the distribution function of system residual life have been derived in explicit forms as functions of a posterior probability statistic. The advantage of the Bayesian model over the well-known 2-parameter Weibull model in system residual life prediction is shown. The methodologies are illustrated using simulated data, real data obtained from the spectrometric analysis of oil samples collected from transmission units of heavy hauler trucks in the mining industry, and vibration data from a planetary gearbox machinery application.

Caregivers injure their backs more than workers in any other industry. Efforts to reduce injuries have been on-going for decades with limited results. Mechanical lift devices have been incorporated into clinical practice over the past 30 years to reduce the risk of injury from patient lifting. Yet injury rates remain high. The use of mechanical lifts may be partly to blame. While these devices assist with lifting patients, they also introduce new activities that result in caregivers experiencing unsafe loading on the spine. We measured loads on the lower back during manoeuvres of the two most common lift types (overhead and floor) as well as during sling insertion. A new device called SlingSerterTM was evaluated for use in the clinical environment. We also investigated spine shrinkage as a measurement tool for estimating cumulative load. Caregivers worked alone and in pairs for both lift maneuvering and sling insertion activities. Overhead lift use resulted in much lower loads than floor lift use. We conclude caregivers can safely operate overhead lifts alone, while floor lift use remained unsafe even with two caregivers. Less-experienced caregivers had higher loads than more-experienced counterparts when using floor lifts. There was no corresponding effect of experience with overhead lift use and we found this to be a further benefit of overhead lifts over floor lifts. Most caregivers exceeded the safe limit for spine compression during sling insertion, though a single caregiver was at no higher risk of injury than two caregivers working together. Clinicians who tested SlingSerterTM agreed the device would be useful in clinical practice, particularly with bariatric patients and other special patient populations that are difficult to roll or turn. Finally, we investigated a novel method for estimating cumulative load based on spine shrinkage. There is growing recognition that excess cumulative load may be responsible for back injury. We found the variability in spine shrinkage was too large to estimate cumulative load directly. However, the technique may still be useful for determining the relative importance of the load from different activities to the cumulative total.