Managing Service Capacity Under Uncertainty
Open Access
- Author:
- Robbins, Thomas R.
- Graduate Program:
- Business Administration
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 11, 2007
- Committee Members:
- Terry Paul Harrison, Committee Chair/Co-Chair
Susan Xu, Committee Member
Douglas J Thomas, Committee Member
Tom Michael Cavalier, Committee Member - Keywords:
- stochastic optimization
manpower planning
call center - Abstract:
- This dissertation addresses the issue of capacity management in a professional services context; specifically a call center based support operation with contractually committed Service Level Agreements (SLAs). The focus of this research is on capacity planning in the face of uncertainty. I investigate the impact of uncertainty on the capacity management decision and develop models that explicitly incorporate uncertainty in the planning process. A short term scheduling model develops detailed staffing plans given variable and uncertain demand patterns. A medium term hiring model seeks the optimal hiring level for the start up of a new project with learning curve effects. A cross training model seeks to determine the best number of agents to cross train on multiple projects. The analysis employs stochastic programming, discrete event simulation, and a simulation based optimization heuristic. This dissertation is very much an applied OR analysis. The research focuses not on developing new theory or methodology, but on applying existing methods to a real problem. In the process I create several new and unique models that contribute to the literature. The research is motivated by work I performed with an IT Support outsourcing company. That company was kind enough to give me access to a great deal of data upon which to base my analysis. I find that incorporating uncertainty into the planning process yields solutions with better outcomes and also provides better insight into key management tradeoffs. The short term scheduling model shows that hedging against arrival rate uncertainty lowers the total cost of operation by improving the probability of SLA attainment. It also shows that increasing the flexibility of the staffing model, by scheduling even a few part time resources, can significantly lower costs. I also find that increasing the probability of achieving the service level goal becomes increasingly expensive. The medium term hiring model shows that learning curve issues during start-up have a significant impact on total costs. The cross training model shows that adding even a moderate amount of flexibility into the workforce can significantly lower costs through the dynamic reallocation of capacity.