David Croft, M.Eng. 2014
Supervisor: Dr. Rodrigo Mora
There is a huge amount of knowledge accumulated within the building industry that is poorly shared and often difficult to access. Local building codes and construction references often fail to identify parameters that are relevant for a particular building assembly, for example climatic conditions such as the average annual rainfall and the driving rain index. As a result, a building might be constructed the same in Vancouver as it would be in Los Angeles despite the two cities having very different climates. Furthermore, since building knowledge and industry best practices are often difficult to access or hard to find for the context of a specific problem, builders and engineers might end up implementing a solution that has known deficiencies and could have been easily remedied with a change of materials or modifications to the assembly.
This project expands upon research within the Building Information Modeling and Design Rationale fields which attempts to store the intent behind design decisions — in this case the design of building assemblies and being able to store and retrieve knowledge related to causes for failure, the function of materials/components and solutions that were implemented to address issues. The project introduces a flexible set of core entities and relationships that were used to develop a graph-based database to input, organize, and provide access to building construction and building science knowledge. Physical parameters such as roof and wall geometries, as well as properties related to heat, air and moisture flow can be associated with specific building assemblies or materials to catalogue the building and index it with other buildings in the knowledge base to evaluate benefits and risks. A key principle that was carried through this research was to maintain a simple and flexible ontology for representing building knowledge, so that additional parameters such as material costs or life expectancy can be included if and when they are relevant to a study.
The problem with many traditional databases commonly known as “relational databases” is that they require a fixed schema of tables and columns to be established before any data is entered. The graph database proposed for this project can expand and change as new case studies are added. Initially the database was developed to store examples from Depreciation Reports that were legislated by the Province of British Columbia for Strata Corporations in 2013, however the proposed model can also store examples from single family and commercial construction projects.
As described in greater detail in this paper, the user interface will be essential to maintain the integrity of the knowledge that is stored and allow users to extract it. A web-based user interface has been developed to input use cases for Depreciation Reports, and a mobile Android application is under development. Once further user interface development has been completed, the knowledge base will be tested by building science and construction professionals to determine its usefulness and to refine the ontology used to represent and organize knowledge within the system.