Most profit oriented organizations are constantly striving to improve their revenues while keeping costs under control, in a continuous effort to meet customers' demand. After its proven success in the airline industry, the revenue management approach is implemented today in many industries and organizations that face the challenge of satisfying customers' uncertain demand with a relatively fixed amount of resources (Talluri and Van Ryzin 2004). Revenue management has the potential to complement existing scheduling and pricing policies, and help organizations reach important improvements in profitability through a better management of capacity and demand.

The work presented in this thesis investigates the use of revenue management techniques in the service sector, when demand for service arrives from several competing customer classes and the amount of resource required to provide service for each customer is stochastic. We look into efficiently allocating a limited resource (i.e., time) among requests for service when facing variable resource usage per request, by deciding on the amount of resource to be protected for each customer and surgery class. The capacity allocation policies we develop lead to maximizing the organization's expected revenue over the planning horizon, while making no assumption about the order of customers' arrival. After the development of the theory in Chapter 3, we show how the mathematical model works by implementing it in the healthcare industry, more specifically in the operating room area, towards protecting time for elective procedures and patient classes. By doing this, we develop advance patient scheduling and capacity allocation policies and apply them to scheduling situations faced by operating rooms to determine optimal time allocations for various types of surgical procedures.

The main contribution is the development of the methodology to handle random resource utilization in the context of revenue management, with focus in healthcare. We also develop a heuristics which could be used for larger size problems. We show how the optimal and heuristic-based solutions apply to real-life situations. Both the model and the heuristic find applications in healthcare where demand for service arrives randomly over time from various customer segments, and requires uncertain resource usage per request.