Improving reliability in predicting the degradation of building assets

Abstract

Predictive modelling of building component degradation optimises project management and maintenance costs. Typically, the visual inspection-based generic condition ratings of building components are collected over time and analysed to determine age-related degradation trends and corresponding life cycle costs. Using two datasets, this research proposes two new approaches: (i) deficiency-based (as opposed to generic) condition ratings of building components from seven local councils in Sri Lanka were analysed to develop Markov models at the component level (engineering-based approach); (ii) nominal replacement costs and times for building components assigned by estimators from the City of Melbourne were used to arrive at degradation rates for component groups through a novel concept of cumulative lost value ratio (CLVR) (monetary-based approach). In the engineering-based approach, snapshot data were collected using both deficiency-based and generic deterioration-based condition scales, and the Markov Chain Monte Carlo technique was used to develop reliability-based models. The results showed that deficiency-based models were more accurate and reliable. The monetary-based approach explored the CLVR concept and the validity of using Markov models for component groups, where stochasticity is based on component mix rather than degradation process randomness. The study's theoretical contribution was to interpret "degradation" in terms of curable and incurable deterioration from a maintenance perspective, estimate component maintenance-free ages using data screening, establish new monetary indices such as LVR and CLVR, evaluate the impact of influencing factors using GRG NLO categorisation, and utilise a monetary-based degradation forecasting paradigm utilising nominal cost data as an alternative to using physical condition data. In practice, the deficiency-based approach will directly improve predictive maintenance reliability, lead to longer maintenance intervals, convert deficiency-based ratings to cost-based ratings using the LVR-based index, and bundle maintenance through categorisation via degradation patterns. The monetary-based approach will eliminate inconsistent physical condition assessments and enable more building assets to be modelled. Keywords: Building components; Deficiency; Degradation; Markov modelling; Influencing factors