Thesis (Ph. D.)--University of Washington, 2008. / Vita. Includes bibliographical references (leaves 126-133).

Master of Science / Department of Industrial & Manufacturing Systems Engineering / Malgorzata J. Rys / In this thesis, various safety aspects of center line rumble strips were evaluated. Based on the literature review center line rumble strips (CLRS) are considered to be effective safety countermeasure for reducing crossover accidents on two-lane, two-way roadways. CLRS are indentations milled into the center line of undivided two-lane, two-way roadways to warn driver of drifting into upcoming traffic. Researchers at Kansas State University (KSU) have conducted studies on CLRS and retroreflective pavement markings (RRPM) installed over them. Based on the literature review and the survey conducted on motorcycle riders it can be concluded that majority of riders believe in the effectiveness of CLRS and they recommend the KDOT to implement CLRS in more locations. From the survey conducted on residents of US 40 it can be concluded that RRPM help them in providing visual guidance. They also noticed that there is considerable deterioration of RRPM over CLRS on US 40. From the studies conducted on US 24, US 50 and US 40 it can be concluded that wet retroreflectivity of pavement markings installed over CLRS is considerably lower than dry retroreflectivity. In locations without CLRS wet retroreflectivity of RRPM is higher than dry retroreflectivity. Also, the analysis performed on retroreflectivity measurements from US 24, US 50 and US 40 show that retroreflectivity follows a linear reduction in performance over time. In addition, a new methodology was developed for evaluating RRPM over CLRS. Various tests and analysis were performed and the new method seems effective.

Master of Science / Department of Industrial & Manufacturing Systems Engineering / Malgorzata J. Rys / Centerline rumble strips (CLRS) are raised or indented patterns installed in the center of undivided rural two-lane highways. Their main function is to alert drivers who are encroaching or leaving the intended travel lane, by producing vibration and noise when crossed by vehicles’ tires. CLRS have been demonstrated to be an effective way in reducing head on and opposite direction sideswipe on two-lane highways (cross-over accidents). However, there are some disadvantages in their utilization, such as the exterior noise created by the strips, which may disturb residents in the highway vicinity. The objective of this study was to verify if the amount of noise created by CLRS is enough to impact negatively on residences and businesses, and to discover if the mean level of noise created by CLRS is statistically different than the noise generated by vehicles driving over smooth pavement. Two types of vehicles were driven over two different patterns of milled-in CLRS (rectangular and football-shaped) and over smooth asphalt pavement, at two different speeds. Researchers collected the noise levels at three distances 50, 100, and 150 feet, measured orthogonally from the center line, in 8 different open space locations. Results indicate that vehicle type, vehicle speed, pavement type, location and distances affect the levels of noise. In addition, both football and rectangular CLRS produced significantly higher levels of noise as compared to the smooth asphalt pavement. A 15 passenger van produced higher levels of noise in comparison with a sedan. Moreover, lower the vehicle speed, noise levels were lower. At every 50 feet of distance, the noise levels dropped significantly. CLRS do increase levels of noise relative to smooth pavement at distances up to 150 feet.

Master of Science / Department of Industrial & Manufacturing Systems Engineering / Malgorzata J. Rys / The community of Manhattan, Kansas was recently chosen as the future site of the National Bio- and Agro-Defense Facility (NBAF). At this site, research of agricultural and animal diseases and pathogens will take place. Due to the fact that the site will be in close proximity to a university, as well as many residents, a risk assessment must be completed to determine whether or not the current road infrastructure would be sufficient for evacuating the city in the event of an emergency. It should be noted that while NBAF is a large concern for this report, risk management is important in other scenarios as well, such as natural disasters or chemical spills, and this information can be applied to such events. This paper discusses the creation and analysis of a discrete-event simulation using ARENA software. The simulation described several scenarios. They were a base case scenario with only campus traffic evacuating; a scenario in which campus and outside traffic evacuate; a case with increased outside traffic; a case in which a vehicle breaks down; a case which includes guardians of children attending campus childcare are re-routed to pick up their children before evacuating; a case which accounts for reduced traveling speeds due to cell phone usage; and a case which closes a direction outside of Manhattan due to wind direction. Such simulations are an ideal performance measure of traffic flow under certain conditions due to the fact that physical resources are not needed to make a realistic comparison between each of them. Each of the situations described above were compared based on percentage of traffic leaving Manhattan and arriving at a defined safe zone each hour. Based on the findings, those involved with disaster management planning can determine if the percentages of vehicles leaving the system per hour are acceptable. They should be evaluated against potential spread rates of diseases to ensure that all residents may evacuate without the danger of becoming infected. For applications outside of NBAF, the results give insight into the degree of change in evacuation percentage that changes within the system may cause, and change any routing accordingly.

Flexible manipulators provide significant advantages over the commonly-used rigid robots due to their lightweight properties, but an accurate control of these manipulators is more difficult to attain, and it is especially demanding in task executions involving changing payloads. This thesis addresses the problem of payload adaptive control of flexible manipulators. The nonlinear model describing the manipulator dynamics is completely derived and is then used for an accurate computer simulation of the flexible manipulator motions. Payload identification is implemented by using a novel neural network approach to identify distinct payload classes from tip deflection patterns which result from different payloads. The identification procedure is then used to select a controller which best meets the control objectives specifying hub speed and maximum tip deflection. Two distinct controller synthesis procedures, one using a pole-placement design and one employing a variable structure technique, are developed. The merits of payload adaptive control are shown by several simulation experiments.

This thesis utilizes artificial intelligence techniques and problem specific knowledge to assist in the design of an manufacturing test cell for electronic products. The electronic printed circuit board (PCB) is subjected to one or more functional evaluation(s) during the manufacturing process. The purpose of these evaluations is to assure product quality. This thesis is focused on, with historical knowledge, the configuration of this testing environment and associated fault isolation processes. By using such knowledge, an improvement in the testing efficiency will be realized which will allow the overall product cost to be minimized.

Industry and society are now aware of risk more than ever before. Organisations whose activities pose risk to individuals and society are accountable to manage and reduce risk to an acceptable level. In this regard, utilisation of Safety Instrumented Systems (SISs) as an independent protection layer is a practical method of achieving the required risk reduction. The role of a SIS is to maintain the safety of equipment under control by providing a safety-related function. The International Standards, IEC 61508/61511, provide a set of guidelines to promote consistency for implementation of SISs used for risk reduction. In accordance with IEC 61508, the performance of a SIS shall be established by computing the associated Probability of Failure on Demand (PFD) as a reliability measure using a suitable technique. The principal purpose of this research is to provide the basis for reliability assessment of redundant SISs affected by process demand as well as component failures. This dissertation introduces four new reliability models for redundant SISs subject to process demand for the first time using Markov analysis technique. The proposed reliability models 1 and 3 incorporates process demands in conjunction with Common Cause Failure (CCF) and evaluates their impacts on the reliability quantification of 1oo2 redundant configuration using different repair philosophies. In model 3, the proposed Markov model was also compared with the IEC 61508 approach for redundant SISs and a reliability improvement between 9% - 15% were observed. The model 2 on the other hand integrates the Dangerous Detected (DD) failure rates in the unavailability assessment of redundant SISs subject to process demand assuming that CCF does not occur. An additional reliability model was developed in this research for a 1oo3 redundant configuration subject to process demand excluding CCF and its construction was verified using partial verification method. Furthermore, a generic framework for reliability assessment of 1oon redundant SISs is provided in this thesis in conjunction with some guidelines for future researchers as how to conduct reliability assessment of SISs subject to process demands. The accuracy of the proposed Markov models is verified for industrial application case studies. It is demonstrated that the proposed approach provides a sufficiently robust result for all demand rates, demand durations, common cause failures, dangerous detected and undetected failure and associated repair rates for SISs. The effectiveness of the proposed models offers a robust opportunity to conduct reliability assessment of redundant SISs subject to process demands.

The purpose of this dissertation is to develop a predictive maintenance plan to diagnose contact degradation on load tap changers, LTCs, and voltage regulators using pattern recognition techniques The following sensor technologies have been applied to the LTC and voltage regulator. Strain gages were attached to the drive arm of the devices to detect changes in the strain associated with each tap change. Acoustic transducers and accelerometers were used to measure changes in the sound wave and vibration associated with each tap change. Both the load tap changer and the voltage regulator were run with new contacts and repeatable signals or signatures were obtained for each sensor type Burned contacts and failed finger contacts, two common modes of failure for these machines, were obtained from Entergy. The burned contacts were installed on the voltage regulator and on each phase of the LTC and the tests were repeated. These sets of data showed that there exist distinct signatures for each equipment type experiencing this failure mode. The magnitudes of the good and bad contact signals were different enough that classification could be performed A power spectral density was also computed for the two cases, operational contacts and burned contacts. These frequency plots show significant differences in the two modes which provide useful information for classification as well Two classification methods were derived for the two-case study under consideration, to determine whether the unit being tested is in need of maintenance. These are template matching and a linear classifier used to partition the feature space into two non-overlapping areas. Graphical results show the validity of each method The future of the project involves field testing the predictive maintenance program on energized substation equipment for a significant period of time. The predictive maintenance system's performance will be measured by its ability to identify equipment in need of maintenance or repair. The completion of this stage of the project will depend upon future funding and upon the cooperation of Entergy Services Inc / acase@tulane.edu

The effects of cure time, cure temperature, and air circulation on the release of 4-phenylcyclohexene (4-PCH) in tufted carpet backed with carboxylated styrene-butadiene rubber (SBR) latex were examined in this research project. 4-PCH is a suspect cause of certain "building-related" illnesses such as headache, upper respiratory and eye irritation, and fatigue. A standardized method for preparing carpet backed with SBR latex and analyzing its 4-PCH content was developed. The cure parameters investigated included three cure temperatures (225, 250 and 275°F), six cure times (0, 5, 10, 20, 40 and 60 minutes), and three air circulation rates (zero, three air exchanges per minute, and ten air exchanges per minute with 1500 linear feet per minute airflow). Increased airflow (1500 linear feet per minute) over the carpet surface produced the greatest impact on 4-PCH release from carpet. Increasing the cure temperature and cure time also significantly increased the release of 4-PCH.

Systematic design and analysis of fuzzy dynamic systems has been a problem which attracted much attention from researchers in recent years. In this dissertation, we propose a methodology for analysis and design of fuzzy dynamic systems. First, we introduce a new way to treat fuzzy sets: fuzzy sets as points in fuzzy state space. We investigate the relationship between membership functions and their corresponding fuzzy set points in fuzzy state space. We then examine the formulation and stability issues of fuzzy dynamic systems based on the geometric structure of fuzzy state space, resulting in the generalization and extension of classical stability definitions to fuzzy dynamic systems. We also introduce cellular structure to fuzzy state space, allowing a discrete cell-to-cell mapping method to be developed to approximate a fuzzy dynamic system model. This method leads to an efficient global behavior analysis algorithm based on a simple cell-to-cell mapping search. Finally, we outline the cellmapping-based search algorithm for fuzzy optimal control design and demonstrate its validity and advantages by applying it to time-optimal trajectory generation for coordinated manipulator systems with uncertain parameters.