VALIDNET: A Network Validation Macro

(contributed by Denis Lebeuf, INRO Consultants)

Network validation is always a tedious and time consuming task. Not too long ago, the main tools available to accomplish this task were a pile of listings and luck. The advent of interactive graphics drastically reduced the amount of data to be scanned (an image is worth a thousand words, it is said) and thus also reduced the importance of the luck factor. Nevertheless, a more systematic approach is needed, if one wants to rely on the results from analysis carried out with these network data. We present here an EMME/2 macro which is a step toward such a systematic network validation method.

The VALIDNET macro uses mainly the network calculator (module 2.41) to perform a set of up to nine tests that detect all the occurrences of some potentially pathological conditions. We say "potentially pathological" in the sense that all the cases detected by the macro are not errors per se, but surely likely candidates. The macro assumes that the base network is syntactically correct (otherwise EMME/2 would not have accepted it) and draws the attention to some potentially questionable topological aspect:

Here are the nine tests that can be performed by VALIDNET:

• Dead ends and unused nodes on auto network. Nodes with no links, only incoming or only outgoing links or nodes connected to the network by a single two-way link are marked.
• Discrepancy between link length and Euclidian distance.
• Asymmetric link length on two-way links.
• Asymmetric link type on two-way links.
• Asymmetric number of lanes on two-way links.
• Asymmetric volume-delay functions on two-way links.
• Links with a defined volume-delay function but with no lanes or no car mode.
• Links allowing car mode but without lanes or defined volume-delay function.
• Links with zero volume after the assignment of a full demand matrix. These could indicate links with ill-defined characteristics, network connectivity problem (isolated regions) or links forming redundant loops.

The tests to be performed by VALIDNET can be specified by using the corresponding keywords as parameters to the macro. If no keywords are specified, all tests will be performed.

Each node or link that does not pass one of the tests is marked with a test specific value in user data UL3. They can then be exhaustively listed using a punch or report format, or can be viewed and quickly located on network plots by using the test specific marking as a color index value. For each test, the total number of instances found is also written in a summary file and kept in a scalar matrix.

The network validation process cannot be fully automated and we emphasize again that the VALIDNET macro uses a most likely candidate method. All data items identified by the macro must be examined on a case by case basis by someone familiar with the network to determine if it is a wanted feature or a coding error. The VALIDNET macro is far from doing exhaustive testing, but constitutes a very fast and efficient tool to do a first level validation.