Network Analyst Not Routing to Destinations

4567
7
Jump to solution
12-04-2015 11:06 AM
Highlighted
Occasional Contributor

Hello GeoNet Community,

I have a similar question to the post from this thread: Trouble carrying out closest facility analysis in ArcMap

I am running OD Cost MAtrix to get the total transportation cost between points based on the ESRI U.S. Roads and Canada Detailed Streets network.  This road network has one-way assignments as a restriction.  The OD model worked for a majority of my Destination points, there are just 2 Destination points that the program will not route to.  The points are located in the initial loading, and I have tried everything suggested in the post referenced above, including:

     1. Fixing the one-way FT and TF directions around my Destinations to make sure they are all going the correct direction.

     2. Making all layers the same coordinate system

     3. Running the model with the "Exclude restriction" box checked

     4. Rebuilding the network dataset after each change

I have run this model with 5,000+ Origins and set to find 15 destinations, and it finishes in a matter of minutes.  Running the above model with 256 Origins and 2 Destinations takes about an hour and a half to run, and it fails to find a solution each time.  The only warning is that "No Destinations found for XXX point in Origins".

I would appreciate any feedback you can give.  Thanks!!!

Reply
0 Kudos
1 Solution

Accepted Solutions
Highlighted
MVP Frequent Contributor

As Dan Patterson said, looks can be deceiving.  It just takes one tiny disconnect to break the network, but many of the disconnects are so small that they are almost impossible to see, even if zoomed in.  I've pulled my hair out on several network datasets trying to figure out why they wouldn't work, only to finally discover some tiny breaks.

Here's the quick and dirty way to find the disconnects:

Run a "New Service Area" on the two problem points using Network Analyst.  In the setup, check off the box to "Generate Lines".  Then set a long distance so it will cover a good bit of your network.  Then run it.  After the run, check the resulting Service Area and the lines outputted over the street linework.  If there are breaks, the Service Area will end abruptly and you can use the generated lines to find where the ending.  After that, edit the gap so the segments snap or use topology to fix it (and then rebuild the Network).

If you don't find breaks, then I would double-check the network-specific attributes.  Maybe one of them is set to NULL?

Chris Donohue, GISP

The reality of Network Analysis:

Network Analysis = 1 % analysis + 99% data cleanup.

View solution in original post

7 Replies
Highlighted
MVP Esteemed Contributor

breaks in the line so that they arent connected and hence unreachable?

Reply
0 Kudos
Highlighted
Occasional Contributor

I dont think so.  I checked and it all looks fine.  Below is a screenshot of the network edges with the Restriction symbology (one way and two-way) turned on. 

Reply
0 Kudos
Highlighted
MVP Esteemed Contributor

Looks can be deceiving.  When confronted with problem like that, I just try to go around a corner in both directions or straight ahead from the offending point.  Sometimes zooming helps,  unless there is no intersection where two lines meet (ergo, an overpass or underpass) and the network isn't configured to handle all realities in the field.

Reply
0 Kudos
Highlighted
MVP Frequent Contributor

As Dan Patterson said, looks can be deceiving.  It just takes one tiny disconnect to break the network, but many of the disconnects are so small that they are almost impossible to see, even if zoomed in.  I've pulled my hair out on several network datasets trying to figure out why they wouldn't work, only to finally discover some tiny breaks.

Here's the quick and dirty way to find the disconnects:

Run a "New Service Area" on the two problem points using Network Analyst.  In the setup, check off the box to "Generate Lines".  Then set a long distance so it will cover a good bit of your network.  Then run it.  After the run, check the resulting Service Area and the lines outputted over the street linework.  If there are breaks, the Service Area will end abruptly and you can use the generated lines to find where the ending.  After that, edit the gap so the segments snap or use topology to fix it (and then rebuild the Network).

If you don't find breaks, then I would double-check the network-specific attributes.  Maybe one of them is set to NULL?

Chris Donohue, GISP

The reality of Network Analysis:

Network Analysis = 1 % analysis + 99% data cleanup.

View solution in original post

Highlighted
MVP Frequent Contributor

Based on the image you provided, I'll throw out a guess as to what the issue is for the one you provided.  It appears the Destination that is highlighted is on a divided highway, with one-way traffic flow.  If the data is for a country with left-hand drive (like the United States), the segment it is on is going "South".  At least given the image shown, there is no way to reach that destination given the only segment visible is one-way away from the destination - i.e. the destination is "North" but the only segment near it goes "South".

Also, the traffic lane going the other way also appears to be reversed from the expected travel direction, at least the arrows on your data make it seem so that way on the last segment approaching the Destination.  It appears the directionality of that segment is reversed from all the other segments - it is going "South" while all the other ones connected to it are going "North".  So traversing the network to the point the other way (via the "Northbound" lane will not be workable either (and this will defeat a search or snap distance to some extent).

And if your data is for a place like England, my apologies for the fact that the Brits do everything backwards...

Chris Donohue, GISP

Highlighted
MVP Frequent Contributor

Visual of some potential issues specific to the image you posted:

pastedImage_0_modified.png

Chris Donohue, GISP

Highlighted
Occasional Contributor

A HUGE THANK YOU to both Chris Donohue, GISP and Dan Patterson for answering my question.  Chris, you were right in that both of my points were on the wrong side of the one way (love making these kinds of mistakes...), and additionally, some of the one-way directionalities in the road network attribute table were going the wrong direction (FT instead of TF) so the routing was broken, even when they were placed on the right side of the one way.  Your recommendation to run a service area and turn on the lines was spot on.  THANKS!!!!

Reply
0 Kudos