Benchmarking operation of commercial buildings through text-mining maintenance work-orders

点击翻译按钮获取中文摘要

Indoor climate control in commercial buildings accounts for 13% of the total energy use and 11% of the CO2emissions in Canada; and, about 30% of the energy used in commercial buildings is wasted due to poorlymaintained, degraded, and improperly controlled equipment and components. Therefore, data-driven analyticaltools for building operation and maintenance have the potential to reduce our environmental impact and toprovide comfortable, healthy, and productive indoor environments.The objective of this research project is to develop algorithms that will benchmark the maintenanceperformance of building systems and components through text-mining within work-order managementsystems. Work-orders are the traditional form of information keeping in buildings. The work-ordermanagement systems contain operators' descriptions of maintenance routines and failure patterns in HVACsystems and their components, albeit as large amorphous documents. Consequently, they are seldom used toextract information about common HVAC faults and their occurrence frequencies. New algorithms will bedeveloped to extract useful information from text-mining work-order management systems. The algorithmswill identify top system and component-level failure modes, develop component level failure rate models, andintroduce failure modes and effects analysis tools for building systems and their components.The proposed research project will make significant intellectual, environmental, economic, and HQPcontributions to Canada. New datasets and methods will be created. Adoption of these methods by the industrypartner, Canada's largest property manager Bentall Kennedy, will contribute to our knowledge-based economy.Wider usage of the algorithms developed in this research project will reduce the environmental and economicimpact of commercial buildings. The HQP will work on data from real buildings, learn their systems andcomponents, and their shortcomings; and conduct interdisciplinary research on building performance anddata-science