Why are the results of an assignment different in a subnetwork?
Assignment results of subnetworks generally cannot be compared to the assignment results of their original networks, because paths get cut off and shortened, impedances sum up differently and volumes are shifted. Using subnetworks is only useful when analysing local effects which are independent of the surrounding network. Using subnetworks is not useful to implement changes in the subnetwork and port them back to the entire network. Using subnetworks also is not useful to save memory requirements in the assignment. In that case, use other strategies instead:
1) Reduce the relevant demand for filtered zones and the setting "only pairs of active zones" and "considered relations".
2) PuT: Reduce the relevant PuT lines by a filter for PuT lines and use the option "Regard only active vehicle journey sections" which is offered in multiple dialogs.
3) Stop non-used protocols.
4) PuT: Use a connection file for repeated assignments.
5) Critically check the assignment parameters.
6) PuT: Be careful when opening links for transport systems of type PuT Walk, use this only for single links when really needed.