MultiModal Network Solving for Multiple Modes

1477
7
03-03-2014 12:25 PM
New Contributor
Hello,

I am trying to create a multimodal network dataset using bus, trains, and streets in Chicago. I've broken the streets up between highways/expressways and all other streets (since pedestrians can't walk on highways).

I've created three connectivity groups: one for bus, train, and then streets. I have created two attributes: drive time and pedestrian time.

After I create the network dataset, I try to calculate a route to see if it works. I set the impedance to Pedestrian Time, but then it does not take into account drive time and vice versa.

I am thinking I will need to create an attribute in my network dataset for "all modes", which would incorporate pedestrian time, drive time, etc., however I am not sure how to duplicate the "streets" evaluators so I have an option to drive or to walk on those streets.

Tags (2)
7 Replies
New Contributor
I think I also am incorrectly calculating travel time for each edge. The Paris tutorial uses 3km/hr for walking, so the field evaluator reads: [meters] * 60/3000. Does this mean 60 minutes for 1 km (3000 meters)?

Using that logic, I calculate 3mi/hr walk rate to be: [miles] * 60/3, 50mi/hr train speed to be [miles] * 60/50. Is this logic correct? Or, am I calculating travel time for edges incorrectly?

Thank you!
Esri Regular Contributor
Hi Ada.  After you've created your network dataset, what's the goal of your analysis?  Do you want to calculate pedestrian travel time from one point to another?  Do you want to compare that to drive time?  The way you set up your network attributes should depend on what you're hoping to analyze.

I'm guessing you probably don't want to create an attribute that uses either walk time OR drive time on the streets.  You probably want to use separate attributes for these, like you have it set up now.  When someone is driving, they could conceivably park and then get on a train or bus, but they're not going to be able to continue driving when they get off the train or bus because they've left their car at the other end of their journey.  So, you're going to want one impedance attribute where the street evaluators are set up for walking and the bus and train ones are set up for bus/train time.  You'll want a second one where the street evaluators are set up for road speeds and the bus/train ones are never used (because this is a purely drive-time analysis). You can use a constant evaluator and set it to -1 to make train/bus edges never used.

Your logic for calculating travel speed in your evaluators looks correct, as long as the units of your cost attribute and the other attributes you're using in your functions are correct.  It's all just basic unit conversions.
New Contributor
Hello!

I am trying to solve for a pedestrian to travel from one point to another. I am using a streets file, a rail stations file, and a rail routes file.

I have two connector groups: one for streets (walking) and one for rail with the rail stations in both connector groups. I have an arribute for pedestrian time where I set field evaluators for walking speed on the streets and train speed on the train routes.

When I finish creating this network dataset, and solve for two random points, it does not take into account the rail mode, only walking.

I do not have a rail station entrances, like the Paris Tutorial. Is this my problem? My station locations only fall/intersect on the rail routes, not on the street network.

My geodatabase is too large to attach in a zip file to this thread.

Thank you for the help!
Esri Regular Contributor
If your rail stations are being used as the points that connect two separate connectivity groups, and those rail stations don't touch the streets at all, then yes, that is the problem.  The points have to touch both.  Furthermore, the points need to fall on an endpoint or vertex (depending on your connectivity settings) of the feature classes they touch.
New Contributor
This make much more sense, thank you!

If I am creating station entrances and transfers between the entrances and the stations in editor, will the entrances need to fall on endpoints or vertices? Or, would it be okay if I have station entrances (points) on edges of streets, and transfers that are lines connecting the points of stations to the points of station entrances?

I've built the entrances and transfers to connect the entrances, but when I create a multimodal network dataset, I still only get a route with walking.

Thank you!
Esri Regular Contributor
So a pedestrian route might go something like this: street -> station entrance -> station connector -> station -> Rail line -> station -> station connector -> station entrance -> street
(Tell me if I got that wrong).

Since your station entrances interact with streets, those station entrances need to fall on street vertices or endpoints, depending on the connectivity settings you've given it (End Point vs. Any Vertex).  If your street features don't have vertices at the locations of the entrances, you can use the Integrate tool to create vertices in the streets.  Without vertices or endpoints, the station entrances will never connect with the streets, and the pedestrians will never be able to get to the rail lines, so your routes will not show any rail usage.
New Contributor
Thank you! This was extremely helpful! Digitizing the station entrances at the verticies for streets is what I needed to do.