Deterioration prediction of community buildings in Australia

Local government infrastructure assets in Australia represent a vast investment built up over many generations. The $15 billion value of community buildings ranks them as the second largest class of infrastructure assets managed by local councils in Australia. Buildings are complex assets, where the large number of elements and deterioration regimes complicate the process. Reliable deterioration prediction of building infrastructure enables local government asset managers to plan and manage the projected expenditures throughout the life cycle of elements of the infrastructure. The research study presented here is funded by the ARC Linkage project scheme with six local government councils as partners. The thesis explores a best practice asset management framework and the gaps identified in the current project partners’ practice in deterioration prediction models and decision making process. The subsequent research focussed on deterioration forecasting and cost optimisation for community buildings using discrete condition data collected by local councils.

The major challenge in deterioration forecasting with condition data is the high variability of data. Several deterministic and reliability-based methods were considered during the literature research and the Markov Process was selected as the base model. A building hierarchy including almost 400 elements was selected to represent each of the community buildings.

Current practice in building asset management was captured through a number of research visits to the partner councils. Currently available condition data were compiled, analysed and mapped to five condition states. Wherever possible, the data were linked to the hierarchical structure of building elements.

The data were then used to derive transition matrices to define the Markov Chain for building the deterioration process. Derivation of the matrices was found challenging with a number of standard methods indicating poor convergence. A new method for calibration of transition matrices entitled “Direct Absolute Value Difference” has been developed and shown to improve the accuracy of the prediction by almost 25% when validated using an independent set of data.

In order to utilise the developed models in sustainable financial planning, a cost optimisation method was developed and the application of the method is demonstrated using one type of building elements: ceilings of community buildings. In addition, a probabilistic risk-cost projection methodology is introduced and developed, which considers component groups of the building hierarchy.

Finally, an algorithm for a user-friendly software tool has been developed to integrate the system hierarchy, condition data registry, deterioration forecasting and risk-cost optimisation.