1. ¹Pontificia Universidad Católica de Chile, Santiago, Chile
2. ²University of Liverpool, Liverpool, Great Britain
3. ³COPPE/Federal University of Rio de Janeiro, Rio de Janeiro, RJ, Brazil

Correspondence: V Marianov, Department of Electrical Engineering, Pontificia Universidad Católica de Chile, Casilla 306 Correo 22 Santiago, Chile. E-mail: marianov@ing.puc.cl

Received September 2007; Accepted January 2008; Published online 7 May 2008.

Abstract

Most models for location of immobile congested facilities assume exponentially distributed service time at the facilities. Although the resulting formulations are tractable, they do not adequately represent service time distributions with small variances, as often occur in practice. In a recent paper, the authors utilized an order r Erlang distribution for the service time, applied to the simple case of single-server facilities. We generalize this approach to multiple-server facilities, which need a different mathematical treatment. The constraint on service availability is cast as a linear constraint on the proportion of time the servers are busy, and its right-hand side parameter is provided for different situations. Extensive analysis is offered on the influence of the parameters of the service time and the capacity of the facilities on the performance of the system. Numerical results are given for a data set relating to the municipality of Rio de Janeiro.