Novel approaches to benchmark capital project performance : an application to healthcare projects

Benchmarking is defined as a process of continuous improvement based on the comparison of an organization’s processes with those identified as best practice, thereby allowing for establishing improvement targets and promoting changes for better project outcomes. Despite its importance, incorporating it into an organization’s routine is a cumbersome and time-consuming endeavor as it entails considerable time and human effort. Moreover, it lacks a systematic approach to capturing the similarity of projects for generating credible performance comparisons. With the widespread implementation of Building Information Modeling (BIM) and technological advancements in the construction industry, new opportunities for improvement in benchmarking have emerged. In response, the overarching goal of this dissertation is to advance benchmarking practice by addressing major problems identified from current benchmarking processes in two different aspects. First, this research introduces a benchmarking framework that leverages BIM data for semi-automating a benchmarking data collection. To accomplish the goal, this research examines the feasibility and functional requirements of such an approach by investigating diverse BIM models created for real-world projects. As a consistent approach to obtaining reliable benchmarking data from BIM is essential, this research also develops a formalized representation schema that transforms information stored in BIM into benchmarking data focusing on neutral information models. Second, this research proposes a new approach that finds groups of similar projects by capturing project similarity. In this research, critical and flexible features are selected with the use of data analytics and data mining techniques. Based on the features, the method generates a set of rules that produces different groups of similar projects by performance metric, which enables reliable performance comparisons. The studies presented in this dissertation are carried out by focusing on a healthcare benchmarking program. This dissertation advances current benchmarking practices by streamlining the benchmarking process and allowing for more targeted metric comparisons. This dissertation contributes to integrating BIM with benchmarking practices by introducing a methodology to realize the BIM-based benchmarking and proposing a comprehensive and expandable representation schema to obtain reliable benchmarking data from BIM. It also contributes to establishing a systematic project grouping method that supports decision making for performance improvements.