Call Center Trunks Optimization
Client: ORM Technologies
Team: Lexi Farrar, Ashley Hall, Neimy Sarmiento
Faculty Advisor: Dr. Barr Year: 2014
Documents: Presentation, Report, Video
ORM Technologies currently has a call center client looking to make their call center more efficient. Currently, ORM Technologies has a functioning system that uses web-based inputs that allowing the customer to input data and generate optimized resource requirements. The system runs an Erlang C profile for every agent level in the pre-model. From there, the optimization selects the minimum headcount required by 30 minute intervals by day, week, month and year. Some models are 12 months, but most are 5 years.
Our team had the task of using Erlang B probability distribution to dynamically estimate and model the most cost-effective number of trunks, or phone channels, needed to service a live call center by trunk type. The optimization model is written in GAMS, General Algebraic Modeling System, for integration into the client’s existing system. (Our team has no prior experience with GAMS.)
For testing, ORM provided a data set of call traffic, segmented by 30-minute time intervals from 7AM to 6:30PM, Monday through Friday for 60 months. Initially, we found the peak call volume in the given data and used that value as the input for an online Erlang B calculator.
The initial model was expanded to include a cost analysis for three separate trunk types, resulting in a minimum annual trunk cost of $7,140, with quarterly peak trunk volumes at March, June, September, and December resembling a cosine wave.