Developing An Intelligent Logistics and Distribution System

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 1

DEVELOPING AN INTELLIGENT LOGISTICS AND DISTRIBUTION SYSTEM

FOR A LARGE NUMBER OF RETAIL OUTLE

The optimal route is expected to minimize cost and time. Generating retail outlet visit routes in
Indonesia is very complex. Generally, the route to visit retail outlets in Indonesia has a visit pattern that
repeats every two weeks. TSP concept is applied in the formation of a route, which has a depot with
several destination nodes to be visited on a trip, where the destination node can only be visited once.

In the establishment of an optimal route, a strategy is needed to decide which route will be selected and
used. VRP is an operational research issue, where the goal is to design the route at the minimum cost for
a fleet of vehicles with the same capacity to serve customers who are geographically dispersed. VRP
often applies to goods delivery activities or route visits. The primary purpose of VRP is to get an optimal
route to minimize mileage, minimize cost, or balance routes. In contrast to TSP, VRPs need to generate
at least one or more routes based on vehicle capacity or certain constraints. Each route must start and
finish at the same depot point, and all node points must be visited. Route Generation Using Minimum
Spanning Tree Approach MST is a tree from a graph that has the minimum total weights.

A route can be constructed from the MST solution by replacing edges from nodes with degrees more
than one with other edges. Completion of MST Preorder Walk from Kruskal's Algorithm The solution
formed from Figure 1 is a tree and it is not necessarily a route solution because some nodes may need to
be passed more than once to get to the depot. After performing the preorder walk, the resulted route is
D-A-B-CE-D, with total distance = 44.

In the route formation case, clustering data based on point position can make the route generation
efficient. The data then are used as a basis for generating optimal route schedule visits using the
clustering and heuristic methods. Scoring process based on consumer id on the customer base The next
step after clustering is the route forming. From those routes, the distance of each route and the total
distance are calculated, which aims to compare with the existing route at the current customer base
condition.

Figure 4a shows a scatter diagram for the current condition, while Figure 4b is a scatter diagram for the
routes formed by the proposed algorithm.

A) routes configuration with an outlier; b) routes configuration with an outlier; c) routes configuration
without an outlier; d) routes configuration without an outlier Figure 4.

Scatter Diagram of Routes From Figure 4a and Figure 4b, there is one outlier point that can be visible to
the human eye with a location at the point. Total Distance Results Based on the Euclidean distance
formula, the total distance from the route formed at the current condition is 297.81, while the total
distance from the new route is 410.14. The standard deviation value of the total daily route distance
under current conditions is 5.42, whereas the standard deviation value of the total daily distance of the
newly formed route is 0.23.

The establishment of a more balanced route is vital to consider. The total distance from the route can
also be affected by the center. The study proves that the combination of the K-means and the MST-Walk
balanced methods can generate more balanced routes with faster computing time for large amounts of
data. From ten runs of the program, the minimum total Euclidean distance generated from the
proposed route is 410.14. The total route distance at the current condition is 297.81, shorter than the
proposed total route distance. The current standard deviation for the daily route is quite high, i.e., 5.42,
while the newly proposed standard deviation of the daily route is very much lower at 0.23.

You might also like