Abstract
Construction schedules are uncertain in nature; therefore, predicting construction duration is a difficult task. Extensive research has proposed mathematical models to predict construction duration based on regression analysis, Monte Carlo method (MCM), and so on. Yet regression analysis cannot capture duration uncertainties. Studies normally use Monte Carlo methods to simulate hundreds to thousands of activities in a project schedule. This can be complicated, time-consuming, and unrealistic because the statistical properties of all the activities cannot be readily determined in practice. Typical construction sequences in condominium building construction were first identified, and then the statistical distributions of controlling activities on the sequences were surveyed. Two-stage questionnaire surveys and goodness-of-fit statistical tests were conducted to achieve the mentioned objectives. Subsequently, a model for predicting the duration of building construction was proposed and applied to a high-rise building project. The results showed that the proposed model reasonably predicted the construction duration for this apartment building. The model fills the gap in knowledge of construction time forecast by introducing the concept of controlling activities to simplify the evaluation of the schedule uncertainty in multistory building construction. This research is beneficial for practitioners to estimate an overall construction schedule of building projects, especially in preconstruction phases.