Browse Prior Art Database

Multi-Product Delivery System Optimization Procedure

IP.com Disclosure Number: IPCOM000115378D
Original Publication Date: 1995-Apr-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 4 page(s) / 160K

Publishing Venue

IBM

Related People

Gal, S: AUTHOR [+3]

Abstract

Disclosed is a novel optimization procedure for solving compound multi- product transportation/routing/scheduling optimization problems. It can be used for any distribution or delivery system operating a heterogeneous fleet of vehicles from one or several depots. Such distribution systems have to supply multi-product demand in a network of several production and destination points. Uniqueness of the tool lies in integrating the modelling and mutual optimization of a multi-product multi-depot delivery system including transportation, routing, scheduling, time-windows and queuing. The algorithm has been implemented in a package of programs written in Fortran, using the IBM* Optimization Subroutine Library (OSL). Large scale tests with real data proved it to be effective. Only the concept is described in this disclosure.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 36% of the total text.

Multi-Product Delivery System Optimization Procedure

      Disclosed is a novel optimization procedure for solving
compound multi- product transportation/routing/scheduling
optimization problems.  It can be used for any distribution or
delivery system operating a heterogeneous fleet of vehicles from one
or several depots.  Such distribution systems have to supply
multi-product demand in a network of several production and
destination points.  Uniqueness of the tool lies in integrating the
modelling and mutual optimization of a multi-product multi-depot
delivery system including transportation, routing, scheduling,
time-windows and queuing.  The algorithm has been implemented in a
package of programs written in Fortran, using the IBM* Optimization
Subroutine Library (OSL).  Large scale tests with real data proved it
to be effective.  Only the concept is described in this disclosure.

      Consider a network with nodes representing distribution and/or
production centers.  Each one of a large number of products
circulating in the system is produces at one or several nodes and is
demanded at some other nodes.  These demand nodes may be the
production nodes of some other products.  Distribution has to be
carried out by a heterogeneous fleet of vehicles with various
capacities which are to be routed and scheduled to satisfy demands.
At each node a vehicle must be unloaded and/or loaded with products
and requires special processing facilities at the nodes (servers).
Servers are available for service only during specified times.  There
is constraint on the number of vehicles which can be serviced at a
node at any one time; any additional vehicle waits in a queue.  The
primary concern is to supply all the customers with the products they
have ordered.

      Compound multi-product transportation/routing/scheduling/
optimization problems are difficult to solve efficiently.  Proposed
is a system or relaxations and approximations designed to reach a
near optimal solution of compound problems and to indicate scale of
deviation from the optimum.  The optimization tool consists of the
following components:

Route generation block: a preparatory stage which produces feasible
vehicle-route combinations using a preliminary screening.

Iterative routing block: based on a partial solution (vehicle-route
combinations included in the solution so far) that chooses and adds
to the final solution current vehicle-route combination with the best
estimated contribution to the overall objective function.

Workload re-allocation block: for the vehicle-route combinations
included in the partial solution so far, it re-allocates the
work-load among the combinations in an optimal way.

Product Flow block: after obtaining the optimal solution for the
workload re-allocation problem, which involves all the trucks; this
is a modified solution to take into account that the products can be
carried by an integer number of pallets.

Scheduling block: simulating...