Files in this item



application/pdfGRUSSING-DISSERTATION-2015.pdf (7MB)
(no description provided)PDF


Title:Risk-based facility management approach for building components using a discrete Markov process - predicting condition, reliability, and remaining service life
Author(s):Grussing, Michael N
Director of Research:Liu, Liang Y
Doctoral Committee Chair(s):Liu, Liang Y
Doctoral Committee Member(s):El-Rayes, Khaled; El-Gohary, Nora; Uzarski, Donald R
Department / Program:Civil & Environmental Engineering
Discipline:Civil Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Condition Prediction
Markov Chains
Remaining Service Life
Building Components
Abstract:The U.S. building stock is large, diverse, and of critical importance to the economic and social well-being of the country. A proactive facility asset management approach is required to ensure these buildings support the purposes, functions, and missions for which they were built and continue to be used. For federal organizations, particularly those with large building portfolios, the goal is to deliver an acceptable level of performance while minimizing life cycle cost and risk. This research presents a framework that explicitly measures the risk and uncertainty associated with building conditions, and uses this framework to support better decisions for facility investment and resource allocation. The end result of this research provides a model for optimizing the selection and application of building work activities ranging from inspection to repair to replacement and recapitalization. Realizing the importance of physical condition in the determination of a building’s performance, a major objective of this research was to improve the statistical accuracy of building component condition prediction models by using a probabilistic approach. To do this, a discrete Markov chain model was proposed and developed. The result of this work is a robust process for developing Markov transition probabilities to model the condition degradation process using existing condition assessment data that has been acquired and continues to be collected for large portfolios of facilities. It solves the problems with data quality issues, effects from major repair interventions, and variable inspection observation times. It also provides a direct means of measuring uncertainty, reliability, and risk of component failure. Finally, it supports an unbiased process of determining expected service life for components by using the Markov chain model to compute the average number of time cycles to reach the failure state. This probabilistic Markov chain prediction model provides a foundation towards a risk-based framework for facility management decision making. A Value of Inspection Information (VOII) model was developed by combining the probability distribution from the Markov prediction model with the decision tree logic from a value of information approach to calculate the benefit of inspection at a point in time using the last inspection results and the cost of component repair, replacement, and potential failure. In addition, the Markov prediction model was also applied to the work activity selection process, where the objective is the selection of the best activity to perform against a building component such that life cycle costs are minimized yet performance constraints are still satisfied. Traditionally these constraints have been condition based, but the proposed model also allows for risk based reliability performance measures as well. Including risk more explicitly in the decision framework has the potential to change the selection optimization process. The overall framework provides a logical approach that utilizes historic data to develop a more realistic model for building component condition and reliability. The approach analyzes component re-inspection information from large building assessment datasets (multiple inspections over time for a single component), to determine how past observed conditions correlate with future observed conditions to predict future reliability and service life. This model provides a stronger correlation to future condition and reliability estimates compared to an age-based deterministic model, and helps to counteract the situations where the recorded age of a component is not representative or expected design life is unknown. This allows a facility manager to proactively manage facility requirements using real-time risk-based metrics aligned with a data-driven probabilistic process.
Issue Date:2015-11-23
Rights Information:Copyright 2015 Michael N Grussing
Date Available in IDEALS:2016-03-02
Date Deposited:2015-12

This item appears in the following Collection(s)

Item Statistics