Use Cases > PTV xCluster > How to Plan Cluster

How to Plan Cluster

The method planClusters optimises a set of orders considering a given number of clusters. An order corresponds to exactly one visit. The procedure considers the following conditions:

Set Order

The object order contains a set of parameters to specify values for the visits as follows:

Attribute Description
Quantity Indicates specific values such as turnover, service period or other potential values. Quantity should be used in order to receive balanced clusters according to the total sum of values per cluster.
Index for cluster Unique index for each cluster. Number of clusters is user-defined.
Index for order Unique key to identify each order.
Fixed order Previous planned clusters can be used in the optimisation process. Hereby, a fixed order means that the order is already assigned to a cluster. It isn`t permitted to move this order to another cluster.

Define Cluster Parameters

Cluster parameters are necessary to specify the purpose of clustering. Partially these attributes are optional. The following attributes are relevant to be considered according to the given use case:

Attribute Description
Number of clusters Define the number of clusters whereby the following condition is to be taken into account: Minimum one cluster and maximum 364 clusters.
Distance calculation

Distance calculation The PTV xCluster Server provides the necessary default values per configuration file. The following values can be modified:

  • Coordinate format such as MERCATOR, GEODECIMAL, GEOMINSEC, UTM etc.

  • Type such as road or geometric (airline distance)

  • Vehicle type such as car or truck and corresponding speed profile

  • Optimisation factor: relation between distance and time to be considered in the routing calculation.

  • Specific values such as a unique index, delete an existing distance matrix before or after usage. This means a previous calculated distance matrix can be used for further calculation processes.
Costs

Costs per minute or per distance can be set to weight the distance and driving periods
Procedure

Procedure The following steps are provided according to the given use case:

  • Construction step to build clusters as compact as possible. This step should be carried out to start a new planning.

  • Improvement step to relocate the visits to another cluster and to improve the distribution of the specific values and simultaneously to maintain sufficient compactness.

  • Post optimisation step is to be used to improve the condition “compactness” in case of varying quantities.

Return Output of the Planned Clusters

The output contains two result lists: The orders which were assigned to the given clusters (whereby an order is exactly addressed to one cluster) and the cluster with additional cluster information. The following table describes the results of a successful computed cluster plan.

Attribute Description
Orders

This class contains the list of unique order keys and the assigned clusters.
Clusters

Clusters This object contains a set of parameters to provide additional information such as the unique key, the target/actual comparison of the quantity per cluster and the number of visits per cluster. Considering the percentage of load, the relation between compactness and actual distribution is presented.

Using the PTV xMap Server as rendering engine, the clusters could be visualised considering best compactness and balanced distribution as exemplarily shown in the next figure:

Best compactness of the given locations

Balanced distribution is taken into account using “turnover” or “service period” as potential values

 

 

Copyright © 2024 PTV Logistics GmbH All rights reserved. | Imprint