Multi-skill queueing models for call centers : approximations and performance optimization
(2010)
Files
ThèseJChrVandenSchrieck.pdf
Open Access - Adobe PDF
- 2.37 MB
Details
- Authors
- Supervisors
- Chevalier, Philippe
- Abstract
- (en) Our main objective in this thesis is to support the management of call centers by developing tools and methods to decide on the number of operators required in a call center. We use existing queueing theory models to analyze call centers with several types of calls. In this context the goal is to find the most efficient configurations, i.e. configurations that achieve the best performance at a minimum cost by combining different sorts of operators. Some can only answer calls of one type, others are skilled to treat calls of different types, but at a greater cost. We propose a Branch and Bound method to find the best combinations of operators while keeping the integrality constraints on the number of operators. The latter constraints matter in small-size call centers. Looking for efficient configurations requires assessing the performance in the models. We propose new methods to estimate the waiting probability, the average waiting time or the service level. These methods exploit the similarities observed in single-skill models between systems with no queue and the same systems with a queue of infinite size. Using Hayward's approximation, that permits to approximate the probability for a call to not be handled in a multi-skill system with no queues, we show how to estimate the performance of the same multi-skill system when blocked calls are put on hold. In the process we also present the peakedness functional. It is a measure of variability of the stochastic flows. We explain how to extract the peakedness from real-life data and to use it in our models. Different applications are proposed.
- Affiliations
Citations
Van den Schrieck, J.-C. (2010). Multi-skill queueing models for call centers : approximations and performance optimization. https://hdl.handle.net/2078.5/130977