Random Generation of Complex Data Structures for the Simulation of Construction Operations

Construction production systems are complex in nature and possess a high level of uniqueness. Modelling and simulation of such systems is challenging due to the randomness, complexity, and interdependency associated with many factors such as the type of product created, the steps of the production process, and the medium or the environment hosting the production process. These factors represent or control the working behaviour of a construction system and need to be realistically represented in a model in order to achieve accurate replication of real system behaviours. However, modeling and simulation of these factors require either a rich real life data set, which is seldom available for construction operations, or random generation of complex data structures with highly correlated attributes. This paper presents an investigation of mathematical techniques that can be used to generate random complex data structures while preserving the correlations between the embedded attributes. Generation of weather and pipelines data sets are selected in this study. We propose a non-parametric approach in the weather generation; its performance is measured against a parametric approach. For the generation of pipeline data sets, we propose a generation methodology based on a Markov chain model for a pipeline structure. It represents part of an ongoing research. A detailed description of the methodology and the progress in this part of the study are discussed.