Design of Transport Infrastructure with Respect to the Specific Features of Case Study

abstract

en

Currently, the increasing emphasis is placed on the long-term sustainability of the supply and processing chain in all sectors. In addition to the economic aspect, the overall impact on the environment are often addressed. To implement appropriate changes and meet the set goals, support tools based on mathematical programming are widely used. These tools can provide appropriate insight into the issues addressed and draw attention to possible unfavorable conditions or overall inefficiency of operation. The main problems are location and allocation tasks, which are usually solved using flow in the network. For accurate results and their possible applicability, it is necessary to use a detailed infrastructure, which is modeled using graph construction. However, these tasks are fundamentally limited by extreme computational complexity if a real problem is addressed. In the case of thousands nodes with millions of edges, it is almost impossible to perform a calculation, where the non-linearity or integer variables are included. The computation can often be performed only on less detailed infrastructure, which is usually given by historical development or legislation. On the other hand, this structure does not reflect the specific conditions of the problem and the obtained results could be distorted and their usability can be considerably limited. It is therefore appropriate to deal with the design of own infrastructure. The aim is to create a changing detail of the transport infrastructure based on key elements in the system. The presented study introduces a proposal of a methodology for the design and reduction of transport infrastructure regarding an optimisation of network flow tasks. The principle of the procedure consists in keeping the finer detail of infrastructure in the vicinity of the monitored node and it is possible to have coarser detail with increasing distance. For this purpose, the transformation of the coordinates of the nodes is used, followed by the application of cluster analysis. Gradually, the k-means and hierarchical clustering algorithms are introduced, including their evaluation by silhouette. For the design of the transport network, the distances between municipalities and population are used. On the other hand, it is desirable to include other key factors such as production, capacity and taking into account multiple reference nodes, which will be the subject of consequent research.

Currently, the increasing emphasis is placed on the long-term sustainability of the supply and processing chain in all sectors. In addition to the economic aspect, the overall impact on the environment are often addressed. To implement appropriate changes and meet the set goals, support tools based on mathematical programming are widely used. These tools can provide appropriate insight into the issues addressed and draw attention to possible unfavorable conditions or overall inefficiency of operation. The main problems are location and allocation tasks, which are usually solved using flow in the network. For accurate results and their possible applicability, it is necessary to use a detailed infrastructure, which is modeled using graph construction. However, these tasks are fundamentally limited by extreme computational complexity if a real problem is addressed. In the case of thousands nodes with millions of edges, it is almost impossible to perform a calculation, where the non-linearity or integer variables are included. The computation can often be performed only on less detailed infrastructure, which is usually given by historical development or legislation. On the other hand, this structure does not reflect the specific conditions of the problem and the obtained results could be distorted and their usability can be considerably limited. It is therefore appropriate to deal with the design of own infrastructure. The aim is to create a changing detail of the transport infrastructure based on key elements in the system. The presented study introduces a proposal of a methodology for the design and reduction of transport infrastructure regarding an optimisation of network flow tasks. The principle of the procedure consists in keeping the finer detail of infrastructure in the vicinity of the monitored node and it is possible to have coarser detail with increasing distance. For this purpose, the transformation of the coordinates of the nodes is used, followed by the application of cluster analysis. Gradually, the k-means and hierarchical clustering algorithms are introduced, including their evaluation by silhouette. For the design of the transport network, the distances between municipalities and population are used. On the other hand, it is desirable to include other key factors such as production, capacity and taking into account multiple reference nodes, which will be the subject of consequent research.

Model-Size Reduction Techniques, Cluster Analysis, Infrastructure Transformation

30.08.2021

2653-8911

8

1

90–90

1