Determination of lockbox collection points via simulated annealing
- PDF / 308,707 Bytes
- 8 Pages / 595 x 842 pts (A4) Page_size
- 91 Downloads / 157 Views
#1999 Operational Research Society Ltd. All rights reserved. 0160-5682/99 $12.00 http://www.stockton-press.co.uk/jor
Determination of lockbox collection points via simulated annealing PR McMullen and RA Strong University of Maine, USA This research presents a heuristic to solve the lockbox location problem via a search-based technique known as simulated annealing. In the past, more traditional mathematical programming techniques have been used to address this problem, but with limited success due to its combinatorial nature. Because simulated annealing is a search-based technique, an optimal solution is not guaranteed, but past research has demonstrated that search-based heuristics can provide reasonable solutions without the dif®culties associated with the more traditional formulations. In this paper, the simulated annealing methodology is used to solve a large lockbox location problem at several differing levels of cost. The results compare favourably to solutions obtained from a K-means clustering heuristic. Keywords: simulated annealing; heuristics; lockbox; optimisation
Introduction and simulated annealing Determination of cash collection points is an important component of accounts receivable management. These cash collection points are frequently referred to as lockboxes Ð a destination for customer remittances. Proper location selection for these lockboxes can help reduce travel time (frequently referred to as ¯oat). Reduction of this travel time in turn reduces the time required for the check to clear the bank, ultimately reducing the ®rm's opportunity costs. The more lockboxes a ®rm has the more it will be able to reduce ¯oat because the remittances do not, typically, have as far to travel. On the other hand, while this variable cost is decreased, the ®xed cost component is increased due to the increased number of lockboxes. It is also important that however many lockboxes are used they be positioned according to some type of policy that considers the population densities of the prospective lockbox locations. Because of this tradeoff between variable and ®xed costs as well as the travel time considerations previously mentioned, the lockbox problem is far from trivial. Levy1 was one of the ®rst to address this problem with a `greedy' heuristic that was intended to ®nd `good' but not necessarily optimal solutions to the lockbox problem. Nauss and Markland2,3 followed with approaches dedicated to ®nding optimal solutions to the lockbox location problem. Stone4 has also offered a heuristic to solve the lockbox location problem that does not guarantee optimality but does consider computational ef®ciency. Feilitz and White5,6 extended the work of both Stone and Nauss-Markland to obtain optimal solutions to the lockbox problem with Correspondence: Dr PR McMullen, University of Maine, Maine Business School, 5723 Donald P.Corbett Business Building, Orono, Maine 044695723 USA. E-mail: [email protected]
reasonable computational ef®ciency. Maier and Vender Weide7,8 also presented approaches to the lockbox locat
Data Loading...
