PhD Oral Exam - Zahra Yousefli, Building Engineering
Development of A Multi-Agent System for Automated Maintenance Resource Allocation in Hospital Buildings
This event is free
School of Graduate Studies
When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.
Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.
Facility managers of hospitals face complex maintenance decisions as they address a multitude of maintenance requests in an environment of limited resources and segmented information. The complex, uncertain, and dynamic nature of the maintenance management environment is a source of concern to facility managers in hospitals due to unexpected failure of building components, the daily arrival of maintenance orders, and changes in related schedules.
Responding to a growing demand for maintenance, on one hand, and lack of proper maintenance management systems, on the other, has led to delays in repair and maintenance of the building components and systems in hospitals. Such delays could cause significant distress to patients and health-care personnel. In such circumstances, centralized systems become inadequate because of their top-down approach which lacks a feedback mechanism and ignores new information. Therefore, to address any change, centralized systems have to be reformulated making it impractical, short-sighted, and problematic to adopt them in hospitals. As such, the use of centralized systems can lead to financial losses and dissatisfaction of patients. It, therefore, becomes necessary to establish a distributed maintenance management system to support the decision-making process of facility managers.
To address the issues stated above, this thesis presents three newly developed automated models (1) a computer resource allocation model for integrating fragmented maintenance information; (2) two simulation models that represent the dynamic environment of maintenance resource allocation; and (3) a Simulation-Based Optimization (SBO) model for resource allocation that minimizes the down-time of building components being considered for maintenance.
Accordingly, this research initially focuses on the maintenance workflow and resource allocation issues pertinent to hospitals. A distributed system is developed to integrate segmented information at different levels of maintenance management with the aim of minimizing maintenance delays in hospital buildings. Multi-Agent Facility Management System (MAFMS) is conceptually designed as a distributed interactive system. This design employed Unified Modeling Language (UML) diagrams that illustrate the specific agents of the system and how these agents interact with each other.
Two simulation models are then developed to demonstrate the benefits of the developed method in reducing the response time to maintenance requests. The developed simulation models consist of two components: a Status Tracking System (STS) and a Resource Allocation System (RAS). A Discrete Event Simulation (DES) is used to simulate the maintenance process flow while a Multi-Agent System (MAS) is used to simulate the process of allocating resources for maintenance activities in hospital buildings. The STS simulation is a DES process, which registers, arranges, and distributes maintenance tasks (orders) to the appropriate resources. For the RAS component, a Multi-Agent Resource Allocation Simulation (MARAS) is developed to simulate different resource allocation scenarios accounting for interactions among various agents (decision-makers) in the maintenance process. A case example is presented to demonstrate the essential features of the developed method. Maintenance data of a hospital building is used to initiate the multi-agent simulation for workflow management process. The simulation results show the benefits of the developed method, to reduce the response time to maintenance requests. Sensitivity analysis method is used to validate the simulation model.
Finally, the third model, i.e. the SBO model, is developed using OptQuest. This model is proposed to optimize the use of limited resources and reduce the down-time of building services components. The SBO model is validated using the sensitivity analysis method.