Discrete Markov Approach for Building Component Condition, Reliability, and Service-Life Prediction Modeling

Michael N. Grussing, Liang Y. Liu, Donald R. Uzarski, Khaled El-Rayes, Nora El-Gohary

Research output: Contribution to journalArticlepeer-review


Condition indexes have been developed to measure building component condition degradation due to age, use, and deterioration in support of asset management tasks related to work identification, planning, and prioritization. With the development of these indexes, a vast amount of condition index data have been collected for a wide range of components in buildings of varying type, use, and geographic location. The U.S. Department of Defense has implemented a standardized condition-assessment approach applied to thousands of Department-owned buildings, resulting in a vast condition index dataset to support more in-depth study of building component condition and reliability. This paper explores the existing data and develops a rigorous definition of the relationship between component condition, failure, and reliability. Presented is an approach using Markov transition probabilities to analyze the existing component condition datasets. This approach provides an improved method for predicting future condition index values based on past inspection results that are not based solely on inputs traditionally prone to error, such as component age and expected service life. It also results in a reliability metric that relates to a component's probability of failure, providing a much needed measure to manage risk.

Original languageEnglish (US)
Article number04016015
JournalJournal of Performance of Constructed Facilities
Issue number5
StatePublished - Oct 1 2016


  • Asset management
  • Condition index
  • Deterioration modeling
  • Markov process

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Building and Construction
  • Safety, Risk, Reliability and Quality


Dive into the research topics of 'Discrete Markov Approach for Building Component Condition, Reliability, and Service-Life Prediction Modeling'. Together they form a unique fingerprint.

Cite this