About geometric networks

In this topic

The geometric network

Just as with network datasets, there are two views of a network: geographical and logical. The geometric network and its associated logical network represent these in a geodatabase. The geometric network is the set of features that participates in a linear system. The geometric network corresponds to the geographical view of the network with features that have positions and shapes. The ends of lines, such as stream reaches and electric lines, join in space at confluences and poles.
The logical network is a set of junction and edge elements, without any geometry, that forms a connected set. The logical network is concerned only with the connectivity of the system. This abstraction of network features to a graph is what enables tracing analysis and sophisticated editing behavior, such as rubber banding of adjacent edges when moving a junction feature and constant feature validation when editing network features.
The following geographic view is concerned with features and their geographic locations. The logical view considers only the connectivity of elements:

Features in a geometric network

A geometric network is a collection of features that comprises a connected system of edges and junctions. An edge has two junctions, and a junction can be connected to any number of edges. Edge features can cross in two-dimensional space without intersecting. An example is a bridge over a road—this is nonplanarity.
The features that represent edges and junctions are network features. Only network features can participate in a geometric network. There are three types of network features:
  • Junction features—These features represent the location where multiple edges come together.
  • Simple edge features—These features represent the line between two adjacent junctions.
  • Complex edge features—These features represent a set of connected lines with two or more junctions. Complex edge features are used to aggregate secondary parts of a network into a compound object corresponding to a domain object, such as a line section of an electrical network.
The following simple geometric network has one complex edge, two simple edges, and six junctions:
As shown in the following catalog view, this geometric network has two complex edge feature classes and three junction feature classes:
A network feature class is a homogeneous collection of one of these types of network features: junction, simple edge, or complex edge. More than one network feature class can represent edges or junctions in a geometric network. A network feature class is associated with exactly one geometric network. Network features in a geometric network have all the same characteristics as other features:
Network features are edited within a framework that preserves connectivity and automatically updates network elements in the logical network. Four connectivity rules guide how features are connected to each other while editing—the edge-junction rule, the edge-edge rule, edge-junction cardinality rule, and default junction type rule. Geometric network  features cannot participate in a topology. A network feature class participates in exactly one geometric network.
A geometric network contains the features that participate in a network. These features—junctions, simple edges, and complex edges—are stored in network feature classes. See the following illustration:

Elements in the logical network

Like a geometric network, a logical network is a collection of connected edges and junctions. The key difference is that a logical network does not have coordinate values. Since edges and junctions in a logical network contain no geometry, they are not features but elements. The main purpose of the logical network is to store the connectivity information of a network along with certain attributes.
A junction feature is associated with one junction element in the logical network. A simple edge feature is associated with one edge element in the logical network. A complex edge feature is associated with a set of junction elements and edge elements in the logical network.
Five edge elements are discovered from this simple network of one complex edge and two simple edges. See the following illustration:
A logical network contains the connectivity of the network. The following connectivity table lists all the adjacent junctions to a given junction, along with the edge that connects them:

Network flow

In a geometric network, sources and sinks are used in determining flow direction. Any junction feature class can take on the ancillary role of a source or a sink. A source is a junction from which a commodity flows. A sink is a junction where a commodity terminates.
Features in a geometric network can have an enabled or disabled state. That is, you can specify no flow through any specified features. This is useful for modeling open switches or closed valves.
In a geometric network, edges can have an indeterminate flow state. Edges that cannot be reached in the network are said to have an uninitialized flow. An edge has a flow attribute that can have the following values:


Edges and junctions have any number of weights as an attribute. Weights store the cost of traversing across an edge or through a junction. Weights can be lengths, line capacity, travel time, speed limit, and tolls.

Network analysis

A program that works with a geometric network to perform network analysis is a solver. A solver takes the following items as input:
Solvers include upstream trace, downstream trace, isolation trace, and path trace. Weights can represent attributes such as length, diameter, impedance, time to traverse, number of lanes, road classification, miles per hour, and toll.

See Also:

Inside a geometric network
Network flow
Connectivity rules
How to create network features
Network analysis