This paper presents the development of a novel computational framework with which expert analysis knowledge can be captured, stored and dynamically retrieved when required to determine the performance of a building design with respect to user-defined requirements. By subdividing the engineer's and designer's knowledge into discrete steps, and storing these steps in a database along with a description of the context in which the knowledge is to be applied, a searchable knowledge base is created. Given a context consisting of a BIM model and one or more building requirements for which metrics are to be provided, the applicable knowledge can be iteratively retrieved from the knowledge base. Through computational reasoning an analysis is acquired as a chain of logically connected analysis steps. Foreseen benefits of use of the framework include safekeeping and disclosure of AEC expert's knowledge and automation of analyses without loss of analysis transparency.