The personnel scheduling problem is common in practice and is a very important application field for operations research techniques. Researches in this area in the past have been fruitful but are largely limited to single-duty scheduling problems. In this research we propose a dual-duty personnel scheduling problem and develop a mathematical model as well as solution algorithm to figure out an optimal schedule. The problem is characterized in the sense that each personnel have to serve two types of duties. For each person, the schedule for type II duty is known and fixed, while only the number of service hours for type I duty is given. The purpose is to arrange a schedule for type I duty such that the total working days of all personnel are minimized. We model the problem as a network design problem and discuss solution methods. We also present a real world case study based on scheduling midterm exam supervisors for a school. Computational testing yields promising results.