Design phase constructability assessment model

Abstract

Constructability is an important feature of a design where it deals with the ability to build. Constructability problems that are encountered during construction are normally associated with design deficiencies. The problems are more common in the traditional contracting system where the design is separated from the construction. Recent study found that in the Malaysian construction industry constructability has been neglected for quite some times. Many designers have failed to give proper consideration to design constructability during the design process. In other studies it was established that various principles of constructability needed to be considered in order to improve design constructability. The focus of this study is to develop model that can be used to assess design constructability based on the different principles of constructability. Improvement can be made to design that fails to meet the minimum level of design constructability. The study was conducted in three phases that are literature review, questionnaire survey and model development. Artificial Neural Network (ANN) and regression techniques were used in the model development. An extensive review of the literature resulted in the identification of the design-related problems in construction, design phase constructability principles and constructability improvement methods. Through questionnaire survey, six out of eighteen design phase constructability principles identified from literature were found to be very important by the engineers and architects in the Malaysian construction industry. The outcomes of the literature review and questionnaire surveys form a basis for the formulation of a beam-design constructability assessment framework, which is based on the relationship between the degree of application of constructability principles and design constructability. The beam-design constructability assessment framework has enabled design constructability data to be collected. By applying Artificial Neural Network (ANN) and regression methods models of beam-design constructability assessment were developed. The best performance model was found to be the multiplayer back-propagation neural network model consisted of twelve input nodes, five hidden nodes and one output node. Test results indicate that the Artificial Neural Network (ANN) method can produce a sufficiently good prediction even with a limited data-collection effort, and thus provide an efficient tool for design constructability assessment. In this study it was concluded that the model based on the relationship between the level of application of constructability principles and design constructability can be used to assess constructability of project at the design phase.