Appointment scheduling in complex stochastic service systems

Appointment scheduling assigns service start times to customers and is central to efficient, reliable service delivery. It entails a fundamental trade-off: schedules that reduce customer waiting typically increase provider idle time, while tighter schedules improve utilization but risk delays, overtime, and deteriorating user experience. These tensions are amplified in modern service systems where uncertainty is pervasive. Service durations vary, demand fluctuates, and disruptions such as no-shows and last-minute cancellations are common. Traditional static scheduling approaches often perform poorly under such variability.
This thesis develops analytically grounded, computationally efficient methods for appointment scheduling in complex stochastic environments. Building on stochastic operations research and queueing theory, the proposed frameworks enable scalable optimization while capturing rich operational structure. Three non-standard settings motivate the work: (i) integrated routing and appointment scheduling for a mobile service provider visiting customers, where travel and service uncertainty interact; (ii) appointment scheduling with parallel servers when each customer requires multiple tasks to be processed simultaneously, creating synchronization and dependence across service resources; and (iii) the design of delivery time windows and dynamic update policies that balance reliability and customer convenience in last-mile logistics.
Across these settings, the thesis introduces queueing-based approximations and recursive evaluation techniques, validated through simulation, to optimize schedules and time windows with realistic problem sizes. The resulting methods provide practical tools for designing robust, adaptive scheduling systems applicable across healthcare, professional services, and logistics.