Performance prediction for designing shared services

Abstract

Since early 1980s, shared services have been utilized by public and private organizations with the purpose of reducing the administrative cost. Currently, shared services evolve into an efficient and flexible tool in optimizing resources and capitals, raising service qualities, promoting strategic growth and generating greater profits for both public and private organizations. One important aspect of design and implementation of shared services is to ensure the quality services delivered by a shared service center. This thesis presents a new family of approximate Mean Value Analysis algorithm for solving multi-class product-form queuing network models. The proposed algorithms are capable of quickly and accurately predicting the average completion time of different types of tasks to be delivered by a shared service center. The computational and numerical properties of the proposed algorithms are analyzed. This thesis demonstrates the usefulness and effectiveness of the proposed algorithms for facilitating the design and implementation of shared services.