U.S. Air Force operations require staffing several operation centers with trained and certified personnel. Scheduling the staffing of these operation centers is currently a time consuming manual process. The schedules must be prepared for the operation centers, training events, training resources, and trainers necessary to maintain current certification. The goal of this project is to develop an optimization scheduling model that will enable the automation of the scheduling tasks, currently performed manually, for a USAF group-level organization.
The project team developed an optimization model that provides for
the automatic scheduling of fifteen operation centers. The
optimization model also schedules required training, training
resources and instructors. The model is capable of handling daily
changes and producing a re-optimized solution while still adhering
to all objective and constraints.
The emphasis of the project was to develop, test and analyze an
optimization algorithm to improve efficiency and performance over
the existing manual scheduling process. The scope of the project was
limited, and no user interface or software code to interface with
the current scheduling tool, TimePiece, was developed.
Design and development of the optimization algorithm, the model, was
conducted through an iterative approach. The model was developed
using Linear Programming and Integer Programming techniques. Team
AFOCS used AIMMS v3.11 as the primary software tool.
The optimization model produces monthly schedules for personnel in
four phases. The first phase focuses on scheduling the shifts for
the operation centers, as fulfilling shift demand is the first
priority. After shifts are scheduled, the model’s second phase
focuses on scheduling training events for each person to meet the
monthly training requirements. The training events require personnel
being trained, instructors to perform the training, and the
necessary training resources, with each component having limited
availability and potential scheduling conflicts. The third phase of
the model entails scheduling the backup crew for each day. The
fourth phase performs the assignment of students, instructors, and
evaluators to the simulator training and evaluation events. After
creation of the monthly original schedule for personnel, the last
portion of the model development was to create the capability to
reschedule mid-month due to an absence of one or more personnel.
The optimization algorithm was tested iteratively throughout the
model development. Final testing was accomplished with simulated
data provided by the project sponsors. The automated model provides
a solution after about three hours of runtime. The model performs in
much less time than the current scheduling process, which is
approximately 1.5 weeks. Not only does the model reduce the
scheduling process time down significantly, it also finds a solution
within 2% of the relaxed linear programming solution’s optimality.