The study was carried out in North carolina to identify which procedure gave good bus schedules for realistically sized problems using a reasonable amount of computer time, so that multiple runs with parameter changes were economically possible. The main objective is to combine trips to minimise the size of the required fleet and travel costs. The three heuristics used were 'assmt', 'orl', and 'ilp' and were used on a group of 30 sample problems from actual bus data gathered from 113 routes and 15 schools with 10 distinct opening times. Results suggested that each of the heuristics could provide good schedules. However, although significantly faster, assmt appeared to produce some combinations needing manual adjustment. Although icp provided consistently good solutions, it required a large linear programming package, and could give non integer solutions. Feasible combinations were generated by orl, but it required the largest amount of computer time and often gave inferior solutions. It is suggested that for trips of 200 or more the potential savings in capital costs could not justify the use of ilp and orl, at the same time, the manual adjustments required by assmt could be excessive. (TRRL)
[1]
Henry L. W. Nuttle,et al.
Addendum to "a comparison of heuristics for a school bus scheduling problem"
,
1987
.
[2]
Clifford S. Orloff,et al.
Route Constrained Fleet Scheduling
,
1976
.
[3]
Shen Lin.
Computer solutions of the traveling salesman problem
,
1965
.
[4]
Arthur J. Swersey,et al.
Scheduling School Buses
,
1984
.
[5]
Paolo Toth,et al.
Algorithm 548: Solution of the Assignment Problem [H]
,
1980,
TOMS.
[6]
Amos Levin.
Scheduling and Fleet Routing Models for Transportation Systems
,
1971
.