Development of offsite construction skill profile prediction models using mixed-effect regression analysis

Abstract

Abstract Offsite construction (OSC) transfers onsite construction activities to factory-based processes utilising technological advancements, resulting in new and emerging skills while causing some existing skills to be changed and others to be redundant. However, there are no established methods to systematically quantify these OSC skill requirements. This paper presents OSC skill prediction models while highlighting the process of model development for future research. The aim of these models is to predict skills using a comparable measure, manhours/m2. A skill classification with six skill categories was used to analyse OSC skills. Numerical model development methods were reviewed, and mixed-effect regression modelling was selected for model development. The skills data needed for regression modelling was collected using eight case studies. Predominantly panelised and modular OSC projects were used to collect skills data. The skill prediction models were validated using further case study data and an expert forum. Comparatively, modules OSC type requires higher skill quantities than panels, for all the six skill categories analysed. Onsite and offsite skill requirements vary for different OSC types. Additionally, complex, non-linear relationships were recognised between OSC types and the utilisation of their skills. This research presents unique OSC skill prediction models that can provide early-stage advice to policymakers, project planners and manufacturers on OSC skill requirements. It also provides a novel methodology to develop predictive models for specific industry scenarios that have non-linear and complex relationships.