Flexible space is becoming increasingly common in many types of buildings because of its enhanced space efficiency and contribution to sustainable development. Space-use analysis (SUA), a performance analysis method conducted during project development, predicts space utilization based on user activity and space information. SUA automatically maps user activities onto available spaces and then calculates each space’s utilization, thereby helping architects to improve space utilization without harming building functionality. However, current SUA theory is not tailored to flexible space, requiring architects to predict and update the utilization of such space manually, a time-consuming and often inconsistent practice.

To address this issue, this research has extended existing SUA method to the flexible space arena. This research makes two theoretical contributions. First, this research has formalized an activity and space ontology for the SUA of flexible spaces. There are 10 classes and 32 properties in the activity ontology capturing the flexible and non-flexible space-use of user activities and five classes and 22 properties in the space ontology abstracting information on flexible and non-flexible space. These two ontologies enable architects to gather and store activity and flexible space information in a computer-interpretable format to conduct SUA seamlessly in contrast to existing practice, which requires architects to keep track of different types of flexible and non-flexible activity space-use and the various properties of flexible spaces manually. The two ontologies were found to be formal as 20 activities and 14 spaces identified in four cases were successfully represented, and were found to be comprehensive as they captured all seven space-use type differentiators. They were also found to be reproducible, being used to model two cases in a prototype SUA system, with consistent results achieved across architects.

In addition, this research has also developed a method for mapping user activities onto building spaces (i.e., flexible and non-flexible space) and generating activity-space pairs in SUA. Each activity goes through four phases: choosing spatial requirements, finding spaces, computing the number of usable units for flexible space, and mapping activities onto the available space. The proposed method formalizes knowledge on how to distinguish different types of flexible and non-flexible space-use, which affects SUA, and how to divide a flexible space into an appropriate configuration that supports a particular activity during mapping. Such knowledge is not available in current SUA theory, and architects are thus required to manually generate activity-space pairs for use in SUA on an ad hoc basis. The proposed method was validated by measuring the conformity of the activity-space mapping results it generated with those generated by five architects in four cases and by inviting those architects to rate its acceptability.

Drawing on the two aforementioned contributions, this research has defined the implementation process of automated SUA for flexible space as comprising three steps: “entering project data” (input), “mapping user activities onto spaces” (analysis), and “computing the space utilization” (output). The effectiveness of the proposed automated SUA was validated by developing a prototype system and inviting six architects to participate in a charrette test on two cases. The results of that test demonstrated automated SUA to improve both the speed and consistency of utilization prediction.

The extension of SUA to the prediction of flexible space utilization will allow architects to check the space efficiency of their designs immediately and consistently, thereby helping them to reduce under-utilized and over-utilized space to achieve economic and environmental sustainability. This research contributes to performance-based building theory by providing a means of predicting the utilization of both flexible and non-flexible space during project development.