A. James Clark School of Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/1654
The collections in this community comprise faculty research works, as well as graduate theses and dissertations.
Browse
8 results
Search Results
Item Single- and Multi-Objective Feasibility Robust Optimization under Interval Uncertainty with Surrogate Modeling(2022) Kania, Randall Joseph; Azarm, Shapour; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation presents new methods for solving single- and multi-objective optimization problems when there are uncertain parameter values. The uncertainty in these problems is considered to come from sources with no known or assumed probability distribution, bounded only by an interval. The goal is to obtain a single solution (for single-objective optimization problems) or multiple solutions (for multi-objective optimization problems) that are optimal and “feasibly robust”. A feasibly robust solution is one that remains feasible for all values of uncertain parameters within the uncertainty interval. Obtaining such a solution can become computationally costly and require many function calls. To reduce the computational cost, the presented methods use surrogate modeling to approximate the functions in the optimization problem.This dissertation aims at addressing several key research questions. The first Research Question (RQ1) is: How can the computational cost for solving single-objective robust optimization problems be reduced with surrogate modelling when compared to previous work? RQ2 is: How can the computational cost of solving bi-objective robust optimization problems be improved by using surrogates in concert with a Bayesian optimization technique when compared to previous work? And RQ3 is: How can surrogate modeling be leveraged to make multi-objective robust optimization computationally less expensive when compared to previous work? In addressing RQ1, a new single-objective robust optimization method has been developed with improvements over an existing method from the literature. This method uses a deterministic, local solver, paired with a surrogate modelling technique for finding worst-case scenario of parameter configurations. Using this single-objective robust optimization method, improved large-scale performance and robust feasibility were demonstrated. The second method presented solves bi-objective robust optimization problems under interval uncertainty by introducing a relaxation technique to facilitate combining iterative robust optimization and Bayesian optimization techniques. This method showed improved feasibility robustness and performance at larger problem sizes over existing methods. The third method presented in this dissertation extends the current literature by considering multiple (beyond two) competing objectives for surrogate robust optimization. Increasing the number of objectives adds more dimensions and complexity to the search for solutions and can greatly increase the computational costs. In the third method, the robust optimization strategy from the bi-objective second method was combined with a new Monte Carlo approximated method. The key contributions in this dissertation are 1) a new single-objective robust optimization method combining a local optimization solver and surrogate modelling for robustness, 2) a bi-objective robust optimization method that employs iterative Bayesian optimization technique in tandem with iterative robust optimization techniques, and 3) a new acquisition function for robust optimization in problems of more than two objectives.Item ESTIMATING THE RELIABILITY OF A NEW CONSUMER PRODUCT USING USER SURVEY DATA AND RELIABILITY TEST DATA(2022) Shafiei, Neda; Modarres, Mohammad; Herrmann, Jeffrey W.; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Because new products enter the market rapidly, estimating their reliability is challenging due to insufficient historical data. User survey data about similar devices (e.g., older versions of the new device) can be used as the prior information in a Bayesian analysis integrated with evidence in the form of product returns, reliability tests, and other reliability data sources to improve reliability estimation and test specification of the new product. User surveys are usually designed for purposes other than reliability estimation. Therefore, extracting reliability information from these surveys may be tricky or impossible. Even when possible, the extracted reliability information contains significant uncertainties. This dissertation introduces the critical elements of a reliability-informed user survey and offers methods for collecting them. A generic and flexible mathematical approach is then proposed. This approach uses the survey and reliability test data of similar products, for example, an older generation of the same product as prior knowledge. Then it combines them through a formal Bayesian analysis with the reliability test data to estimate the life distribution of the new product. The approach models continuous life distributions for products exposed to many damage-induced cycles. It proposes discrete life distribution models for products whose failures occur within several damaging cycles. The actual cycles for various applicable damaging stress profiles are converted into the equivalent (pseudo) cycles under a reference stress profile. When damage-induced cycles are estimated from user surveys, they may involve biases, as is the nature of most nontechnical users’ responses. This bias is minimized using an approach based on the Kullback-Leibler divergence method. The survey data and other evidence from similar products are then combined with the test data of the new product to estimate the parameters of the reliability model of the new product. The dissertation developed approaches to design reliability test specifications for a new product with unknown failure modes. The number of samples, stress levels, and the number of cycles for the accelerated life test are determined based on the manufacturer’s requirements, including the desired warranty time, the desired reliability with some confidence level at the warranty time, and the maximum number of samples. The actual use conditions (i.e., actual stress profiles and usage cycles) are grouped using clustering techniques. The centers of clusters are then used to design frequency-accelerated or stress-accelerated reliability tests. The application of the proposed reliability estimation approach and the test specification design approach is illustrated and used to validate the proposed algorithms using the simulated datasets for a hypothetical handheld electronic device with the failure mode of cracking caused by accidental drops. The proposed approaches can adequately estimate the reliability model and design test specifications for a wide range of consumer products. These approaches require reliability data about an existing product that is similar to the new product, however.Item CLASSIFICATION AND PROBABILISTIC MODEL DEVELOPMENT FOR CREEP FAILURES OF STRUCTURES: STUDY OF X-70 CARBON STEEL AND 7075-T6 ALUMINUM ALLOYS(2011) Nuhi Faridani, Mohammad; Modarres, Mohammad; Reliability Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Creep and creep-corrosion, which are the most important degradation mechanisms in structures such as piping used in the nuclear, chemical and petroleum industries, have been studied. Sixty two creep equations have been identified, and further classified into two simple groups of power law and exponential models. Then, a probabilistic model has been developed and compared with the mostly used and acceptable models from phenomenological and statistical points of view. This model is based on a power law approach for the primary creep part and a combination of power law and exponential approach for the secondary and tertiary part of the creep curve. This model captures the whole creep curve appropriately, with only two major parameters, represented by probability density functions. Moreover, the stress and temperature dependencies of the model have been calculated. Based on the Bayesian inference, the uncertainties of its parameters have been estimated by WinBUGS program. Linear temperature and stress dependency of exponent parameters are presented for the first time. The probabilistic model has been validated by experimental data taken from Al-7075-T6 and X-70 carbon steel samples. Experimental chambers for corrosion, creep-corrosion, corrosion-fatigue, stress-corrosion cracking (SCC) together with a high temperature (1200 0C) furnace for creep and creep-corrosion furnace have been designed, and fabricated. Practical applications of the empirical model used to estimate the activation energy of creep process, the remaining life of a super-heater tube, as well as the probability of exceedance of failures at 0.04% strain level for X-70 carbon steel.Item Advanced methodologies for reliability-based design optimization and structural health prognostics(2010) Wang, Pingfeng; Youn, Byeng Dong; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Failures of engineered systems can lead to significant economic and societal losses. To minimize the losses, reliability must be ensured throughout the system's lifecycle in the presence of manufacturing variability and uncertain operational conditions. Many reliability-based design optimization (RBDO) techniques have been developed to ensure high reliability of engineered system design under manufacturing variability. Schedule-based maintenance, although expensive, has been a popular method to maintain highly reliable engineered systems under uncertain operational conditions. However, so far there is no cost-effective and systematic approach to ensure high reliability of engineered systems throughout their lifecycles while accounting for both the manufacturing variability and uncertain operational conditions. Inspired by an intrinsic ability of systems in ecology, economics, and other fields that is able to proactively adjust their functioning to avoid potential system failures, this dissertation attempts to adaptively manage engineered system reliability during its lifecycle by advancing two essential and co-related research areas: system RBDO and prognostics and health management (PHM). System RBDO ensures high reliability of an engineered system in the early design stage, whereas capitalizing on PHM technology enables the system to proactively avoid failures in its operation stage. Extensive literature reviews in these areas have identified four key research issues: (1) how system failure modes and their interactions can be analyzed in a statistical sense; (2) how limited data for input manufacturing variability can be used for RBDO; (3) how sensor networks can be designed to effectively monitor system health degradation under highly uncertain operational conditions; and (4) how accurate and timely remaining useful lives of systems can be predicted under highly uncertain operational conditions. To properly address these key research issues, this dissertation lays out four research thrusts in the following chapters: Chapter 3 - Complementary Intersection Method for System Reliability Analysis, Chapter 4 - Bayesian Approach to RBDO, Chapter 5 - Sensing Function Design for Structural Health Prognostics, and Chapter 6 - A Generic Framework for Structural Health Prognostics. Multiple engineering case studies are presented to demonstrate the feasibility and effectiveness of the proposed RBDO and PHM techniques for ensuring and improving the reliability of engineered systems within their lifecycles.Item Data-Informed Calibration and Aggregation of Expert Judgment in a Bayesian Framework(2009) Shirazi, Calvin Homayoon; Mosleh, Ali; Reliability Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Historically, decision-makers have used expert opinion to supplement lack of data. Expert opinion, however, is applied with much caution. This is because judgment is subjective and contains estimation error with some degree of uncertainty. The purpose of this study is to quantify the uncertainty surrounding the unknown of interest, given an expert opinion, in order to reduce the error of the estimate. This task is carried out by data-informed calibration and aggregation of expert opinion in a Bayesian framework. Additionally, this study evaluates the impact of the number of experts on the accuracy of aggregated estimate. The objective is to determine the correlation between the number of experts and the accuracy of the combined estimate in order to recommend an expert panel size.Item STRUCTURING A PROBABILISTIC MODEL FOR RELIABILITY EVALUATION OF PIPING SUBJECT TO CORROSION-FATIGUE DEGRADATION(2009) Alseyabi, Mohamed Chookah; Modarres, Mohammad; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Pipelines are susceptible to degradation over the span of their service life. Corrosion is one of the most common degradation mechanisms, but other critical mechanisms such as fatigue and creep should not be overlooked. The rate of degradation is influenced by many factors, such as material, process conditions, geometry, and location. Based on these factors, a best estimate for the pipeline service life (reliability) can be calculated. This estimate serves as a guide for maintenance and replacement practices. After a long period of service, however, this estimate requires reevaluation due to the new evidence gathered from monitoring the conditions of the pipeline. Several deterministic models have been proposed to estimate the reliability of pipelines. Among these models is the ASME B31G code, which is the most widely accepted method for the assessment of corroded pipelines. However, these models are highly conservative and lack the ability to estimate the true life and health of the pipeline. In addition to the limitations embedded in these deterministic models are the problems with inspection techniques and tools that may be inadequate and susceptible to error and imprecision. What is needed, therefore, is a best-estimate assessment model that estimates the true life of these pipelines and integrates the uncertainties surrounding the estimate. Hence, this dissertation proposes a probabilistic model that is capable of addressing the limitations of these models (by accounting for model uncertainties), the inspection data (by characterizing limited and uncertain evidence), and subjective proactive maintenance (by involving the decision-making process under uncertainty). The objective of this research is to propose and validate a probabilistic model based on the underlying degradation phenomena and whose parameters are estimated from the observed field data and experimental investigations. Uncertainties about the structure of the model itself and the parameters of the model will also be characterized. The proposed model should be able to capture wider ranges of pipelines rather than only the network ones, so that the proposed model will better represent the reality and can account for material and size variability. The existing probabilistic models sufficiently address the corrosion and fatigue mechanisms individually but are inadequate to capture mechanisms that synergistically interact. Given that capturing all degradation mechanisms would be a challenging task, the new model will address two of the most important mechanisms: pitting corrosion and fatigue-crack growth. The field data is very limited, and the experiments required an extensive and expensive set-up before they could produce suitable results. Hence, relying primarily in the initial stage on the generic data available from the literature facilitated the construction of the empirical degradation model and provided an order-of-magnitude estimate of the parameters of the degradation model. The proposed model in it is simplest form has the capability to estimate the degradation outputs with the least parameter inputs available. The Bayesian approach was implemented to incorporate the experimental data to further improve the proposed model and estimate the two constants' values. The proposed empirical model can be used to estimate the aging life expended, which will enable inspection and replacement strategies to be developed.Item Methodology for Evaluating Reliability Growth Programs of Discrete Systems(2008-04-25) Hall, J. Brian; Mosleh, Ali; Ellner, Paul M.; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The term Reliability Growth (RG) refers to the elimination of design weaknesses inherent to intermediate prototypes of complex systems via failure mode discovery, analysis, and effective correction. A wealth of models have been developed over the years to plan, track, and project reliability improvements of developmental items whose test durations are continuous, as well as discrete. This research reveals capability gaps, and contributes new methods to the area of discrete RG projection. The purpose of this area of research is to quantify the reliability that could be achieved if failure modes observed during testing are corrected via a specified level of fix effectiveness. Fix effectiveness factors reduce initial probabilities (or rates) of occurrence of individual failure modes by a fractional amount, thereby increasing system reliability. The contributions of this research are as follows. New RG management metrics are prescribed for one-shot systems under two corrective action strategies. The first is when corrective actions are delayed until the end of the current test phase. The second is when they are applied to prototypes after associated failure modes are first discovered. These management metrics estimate: initial system reliability, projected reliability (i.e., reliability after failure mode mitigation), RG potential, the expected number of failure modes observed during test, the probability of discovering new failure modes, and the portion of system unreliability associated with repeat failure modes. These management metrics give practitioners the means to address model goodness-of-fit concerns, quantify programmatic risk, assess reliability maturity, and estimate the initial, projected, and upper achievable reliability of discrete systems throughout their development programs. Statistical procedures (i.e., classical and Bayesian) for point-estimation, confidence interval construction, and model goodness-of-fit testing are also developed. In particular, a new likelihood function and maximum likelihood procedure are derived to estimate model parameters. Limiting approximations of these parameters, as well as the management metrics, are also derived. The features of these new methods are illustrated by simple numerical example. Monte Carlo simulation is utilized to characterize model accuracy. This research is useful to program managers and practitioners working to assess the RG program and development effort of discrete systems.Item Integrated Methodology for Thermal-Hydraulics Uncertainty Analysis (IMTHUA)(2007-01-25) Pour-Gol-Mohamad, Mohammad; Modarres, Mohammad; Mosleh, Ali; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation describes a new integrated uncertainty analysis methodology for "best estimate" thermal hydraulics (TH) codes such as RELAP5. The main thrust of the methodology is to utilize all available types of data and information in an effective way to identify important sources of uncertainty and to assess the magnitude of their impact on the uncertainty of the TH code output measures. The proposed methodology is fully quantitative and uses the Bayesian approach for quantifying the uncertainties in the predictions of TH codes. The methodology also uses the data and information for a more informed and evidence-based ranking and selection of TH phenomena through a modified PIRT method. The modification considers importance of various TH phenomena as well as their uncertainty importance. In identifying and assessing uncertainties, the proposed methodology treats the TH code as a white box, thus explicitly treating internal sub-model uncertainties, and propagation of such model uncertainties through the code structure as well as various input parameters. A The TH code output is further corrected through a Bayesian updating with available experimental data from integrated test facilities. It utilizes the data directly or indirectly related to the code output to account implicitly for missed/screened out sources of uncertainties. The proposed methodology uses an efficient Monte Carlo sampling technique for the propagation of uncertainty using modified Wilks sampling criteria. The methodology is demonstrated on the LOFT facility for 200% cold leg LBLOCA transient scenario.