Service Life Prediction Beyond the 'Factor Method'
Johann, Mc Duling, Pr Eng, Emile Horak Pr Eng, Chris E. Cloete · 2008
The ability to quantify changes in condition over time is important to ensure sustainable development in the built environment. The current ‘state of the art’ Factor Method [ISO 15686-1:2000] for service life prediction calculates an estimated service life, but not changes in condition. The application of the stochastic Markov Chain is restricted by limited availability of historic performance data on degradation of building materials required to populate transition probability matrices. This paper, based on a PhD thesis, looks at the application of neuro-fuzzy artificial intelligence to translate expert knowledge into probability values to supplement historic performance data for the development of Markovian transitional probability matrices, towards prediction of service life, condition changes over time, and effects of maintenance levels on service life of buildings. Expert knowledge is used to express durability and degradation factors in “IF-THEN” rules, which are translated into crisp probability values with neuro-fuzzy artificial intelligence to populate the Markovian transitional probability matrices. A case study is presented to proof that the limited availability of historic performance data on degradation of building materials can be supplemented with expert knowledge, translated into probability values through the application of Fuzzy Logic Artificial Intelligence, to develop transition probability matrices for the Markov Chain towards calculating the estimated service life of a building or component, quantifying changes in condition over time and determining the effects of maintenance levels on service life.