This session is geared toward new or potential users of ArcGIS Network Analyst—an extension designed to model road networks and solve transportation-related problems. Network Analyst is often used to route vehicles or fleets of vehicles, generate service areas, and perform site-selection analysis. The presenters will cover the basics of the extension’s modeling and analytic capabilities, including network creation, use of analysis tools in ArcMap, and integration with the geoprocessing framework.
00:01 So my name's Patrick Stevens. I'm a product engineer on the Network Analyst team.
00:04 And I'm Robert Garrity. I'm also a product engineer on the same team.
00:07 And we, as product engineers, we work closely with the developers designing the software, testing the software...
00:12 ...doing SDK work, and things like that.
00:17 Oh, and I also wanted to mention one other thing: there's another type of network that ArcGIS covers...
00:21 ...and that is geometric networks.
00:24 And those, instead of street networks or transportation networks, like we work on with Network Analyst...
00:28 ...those are utility and natural resource networks.
00:30 Directed flow, like electricity or rivers.
00:33 And if it's that type of network that you're interested in, public utilities and...and such...
00:37 ...then there's a presentation going on right now in room 3 about geometric networks.
00:41 And there's another one later on, I bel-...I believe.
00:42 It doesn't maintain geometry internally while it's solving, and it doesn't produce it when it's done.
00:44 When we get to the end, I'll show you the time for another one if you're interested in both kind of networks.
00:48 So sorry if it's a little bit confusing.
00:52 So Network Analyst is an extension of ArcGIS that allows you to perform analysis on a transportation network...
00:59 ...as I said; like streets, for example.
01:01 So if routing's a part of your job, then Network Analyst will be valuable to you...
01:04 ...help your organization save money, help you look like a hero.
01:08 So you see here the types of analysis we support.
01:11 These types are performed by what we call solvers. That's the term I'll use throughout this presentation.
01:16 And they... Solvers perform these...these algorithms on the street networks, of analyzing the networks...
01:22 ...including finding the best route through a set of locations, considering time windows and sequencing...
01:28 ...finding the nearest location from another location or sets of locations...
01:33 ...determining the area that's accessible around a facility location within a given cost or time service area...
01:41 ...finding a table of costs from a set of origins to a set of destinations...
01:48 ...optimizing a set of routes with a fleet of vehicles with the vehicle routing problem solver.
01:53 And new at 10.0, we're finding the best possible place to put a facility...
01:57 ...be it a warehouse, a station, or a bus stop, based on allocating demand efficiently.
02:03 So in this presentation, we're going to start out with going over how and where you'd work with Network Analyst in ArcGIS.
02:10 And then we'll talk a little bit about modeling street networks using our data model, which is called the network dataset.
02:15 We'll talk about each of these six solvers that I've just showed you and do a demonstration of each of them.
02:20 And then we'll go over a little bit of where you can further your Network Analyst education here at the Users Conference.
02:26 So the first thing you'll need to do with Network Analyst is enable the license.
02:30 So you need a network analysis license, and you'll enable it with the Extensions dialog...
02:35 ...that you'll see here that Bob will show you in a little bit.
02:38 Now Network Analyst is a complete GIS.
02:40 And what that means is we do visualization, we do analysis, we do data management, and we do dissemination.
02:47 And the data management portion of that is via the network datasets.
02:51 That's our core geodatabase model to represent undirected network or street networks.
02:56 And we're going to cover the power of this data model in a little bit through some slides and demos.
03:01 So you'll manage this data, these network datasets, within ArcMap.
03:05 That's where you do editing and viewing of the data.
03:07 And in ArcCatalog is where you'll do the creation wizard or building of the network.
03:12 And new at 10, you can dissolve and...and version networks.
03:14 And we'll go over that a little bit.
03:16 And now you can use the Arc...the Catalog window within ArcMap at 10.0.
03:23 So then you'll set up your analysis problems, and you'll do that within ArcMap or via geoprocessing.
03:27 And with that you'll work with the six solvers.
03:30 And this is the nitty-gritty of network analysis.
03:32 Your data's set up as network datasets, and you're creating scenarios in...in solving network problems or transportation problems.
03:41 And at 10.0, you can work with the 3D capabilities of Network Analyst using three-dimensional network datasets.
03:46 You'll do that in ArcScene and ArcGlobe.
03:48 And there's no out-of-the-box controls for Network Analyst in those two apps.
03:52 But all the geoprocessing tools are available, so you can do anything with Network Analyst that you want in three dimensions.
03:59 So the dissemination part of Network Analyst is serving out network analysis services.
04:04 And you'll serve those analysis maps through ArcGIS Server...
04:07 ...by publishing either as geoprocessing services or as a network analysis service.
04:12 By you... When you publish your service, you'll click the appropriate setting; it'll publish in the method that you choose.
04:17 And if that's what you want to do, there's a seminar on automating workflows using geoprocessing that'll...
04:21 ...help teach you how to use ArcGIS Server with Network Analyst.
04:25 So after you publish your services, they can be accessed via the REST or the SOAP endpoints...
04:35 Or you can use ArcMap and ArcGIS Explorer to connect to these connections as well.
04:41 Also I wanted to mention, if you are brand new to Network Analyst, please go through the tutorial.
04:45 You can install the tutorial with the tutorial data.
04:48 And it doesn't take very long, and it gives you a great overview of how to use the product and what it's capable of.
04:54 So in ArcMap, you'll want to use this Network Analyst toolbar pretty extensively.
04:59 With it, you'll create new analysis layers, you can add inputs to your analysis...
05:03 ...you'll solve the problems and generate directions and work with the network dataset.
05:07 You can also activate what's called the Network Analyst window or, as we call it, the NA window.
05:11 And that'll help you manage the inputs to each of these analyses, as...as well as manage your results.
05:18 There's also a complete set of geoprocessing tools, as I mentioned, in...in ArcMap as well as in ArcScene and ArcGlobe.
05:24 And anything you can do with Network Analyst through the toolbar, pretty much you can do through geoprocessing as well.
05:31 Now, Network Analyst is represented in ArcGIS the same way that other datasets are, and that's through layers.
05:36 You see here a table of contents with two layers shown.
05:39 The layer on the top is a network layer, and that references your network dataset on disk.
05:44 And the layer on the bottom is the network analysis layer...
05:46 ...and that holds the definition of the analysis that you're concerned with, that you're working with.
05:51 It's a composite layer that's made of sublayers like routes and stops and barriers as you see here for a route layer.
05:57 Now there's layers for each of the six types of analysis that I showed you.
06:00 But you'll work with all of them in the same way.
06:05 So Bob will show you some of the modeling capabilities of the...the network dataset.
06:10 Okay. Thanks, Patrick.
06:13 Let's see; try to get my microphone working so you can hear me.
06:16 Okay. The first thing that I like to do when I open up a fresh installation of ArcGIS is add the Network Analyst toolbar.
06:24 Do that by clicking Customize, Toolbars, and Network Analyst.
06:28 And then I'd add the Network Analyst window.
06:31 And to do that, I would click this button, but it's disabled.
06:33 And that tells me that I haven't enabled the Network Analyst extension yet.
06:37 So I'll go to Customize, Extensions, that'll open up the Extensions dialog box, and I just check on the extension here.
06:44 And it's...and you can see that the...the Show Network Analyst Window button is available now.
06:51 So I click there, and it's opened up the Network Analyst dialog box, which is here.
06:56 This will help me manage the inputs and outputs of my network analysis layers.
07:01 And I'll switch back to my table of contents because I'm going to add a network dataset to my map.
07:07 So to do that, I use a new Catalog window, and I'll look for my network dataset layer, or network dataset in this case.
07:15 And it's... I can recognize it by the grid pattern with the route that's drawn on top of it...
07:20 ...and I'll just drag that into the map and drop it.
07:23 And it asks if I want to add the source features that come with it.
07:27 I'm just going to say no for this analysis.
07:29 And here we have a map of...of San Francisco.
07:34 Just going to zoom in here.
07:35 And notice that it added the network layer.
07:38 I call it the network dataset layer just to make a...
07:41 ...make it more distinct between a network dataset layer and a network analysis layer.
07:45 And now that I have a network dataset layer in the map, I can create a network analysis layer.
07:51 And to do that, I just click the drop-down menu in the Network Analyst toolbar and choose the type of analysis I want to perform.
07:59 I'll choose Service Area in this case.
08:01 And now if we look at the Network Analyst window, the service area analysis layer is shown here.
08:08 Just going to create a facility by using the Create Network Location tool on the toolbar, and solve.
08:16 And this will give me a five-minute service area around that facility.
08:21 That's a quick overview of how you use Network An-...the UI components in Network Analyst.
08:32 So when you think about how you get from one location in town to another location in town, what does it take?
08:39 You have to get in your car, you drive along the streets, you obey the speed limits, you stop at traffic lights...
08:44 ...you drive on the correct side of the road, you try to take what you think is the quickest route to your destination.
08:49 And when you get there, you make sure that you park on the right side of the road for the place that you're visiting.
08:55 And it's our job to make a computer model that'll accurately reflect these conditions along a road network.
09:00 There's the conditions like the traffic laws I mentioned, as well as transitory conditions like...
09:05 ...congested traffic, weather, road closures, those kind of things.
09:09 And we'll do this through our network dataset model through static road conditions...
09:13 ...things that are inherent to the street network like speed limits...
09:16 ...as well as settings on the solvers that we offer for the more dynamic road conditions...
09:21 ...or things that are only appropriate to the analysis you're doing, like the placement of the stops.
09:26 Now, this is the... One of the many things that differentiates Network Analyst from our competitors...
09:30 ...it's that we provide a full set of tools that'll...that'll help you model the world.
09:33 The better the model, the more accurate the routes'll be.
09:36 Bob will go over the network dataset model with you now.
09:39 Okay. Let's see. To start using Network Analyst, you need data...
09:43 ...and Network Analyst supports source data in the form of shapefiles; file, personal, and SDE geodatabases.
09:51 It also supports StreetMap data, which is compressed and read-only.
09:55 Where can you get this kind of data?
09:56 We have a few options.
09:58 You can use the free Data & Maps DVD that comes with ArcGIS.
10:02 That contains nationwide street map data.
10:06 You can also convert U.S. Census TIGER data into a network.
10:12 And...and a new option that's available just a...is one that Esri made available on Friday...
10:20 ...was an ArcGIS Editor for OpenStreetMap.
10:23 So you can also use OpenStreetMap data; just download that onto your computer...
10:27 ...and then convert that into a network dataset.
10:30 The ArcGIS Editor for OpenStreetMap was designed so that you can contribute for the crowd sourcing project.
10:38 So you can add streets, add information.
10:40 And then lots of people around the world, they're doing the same thing...
10:43 ...and then you create your own network dataset from that information.
10:48 You can also use your own feature classes that represent roads or transportation networks.
10:53 Vendors like Tele Atlas and NAVTEQ also supply network data.
10:57 You'll need to pay for it, but it is really high-quality data.
11:02 And the data you'll use represents networks.
11:04 An important concept to understand is the difference between Euclidean and network paths.
11:09 So how do you travel from one place to another?
11:11 Say I want to get from point A to point B in this slide...
11:14 ...and if I take a straight-line path, I would be taking the Euclidean path.
11:19 But that would require swimming, and I probably wouldn't want to jump into that lake.
11:22 So instead, what I would do is travel along the roads that go around the lake.
11:27 And we call this path the network path.
11:31 Finding network paths is a fundamental ability of the Network Analyst extension.
11:35 And since people and goods tend to travel on network paths, its modeling and analysis tools are valuable.
11:44 So to correctly find these network paths, you also need to be able to accurately model your network dataset.
11:51 And the Network Analyst team has put a lot of effort into providing tools that will allow you to do this.
11:57 And one of the characteristics you need to be able to model is connectivity.
12:01 This is about how streets connect to one another.
12:04 Think about the lines, just normal lines in a simple feature class, a line feature class.
12:10 One line doesn't know that another line crosses it or even that another line exists.
12:14 But the network dataset, it keeps track of this information and knows what lines are connected...
12:19 ...so that it can quickly determine which lines or which paths along the network are possible.
12:24 It also allows you to set rule that...set rules that allow you to specify which intersecting lines truly connect.
12:32 This way you can model multimodal networks, overpasses and complex interchanges like the one shown here.
12:40 And each line, or edge as we call it, is...has attributes.
12:45 And there are four types.
12:46 You can have a cost, restriction, and...let's see...descriptor, and hierarchy attributes.
12:52 The most important of these is the cost attribute because all solvers or analyses you perform minimize the cost.
12:59 And whenever you create a network dataset, you need to provide at least a cost attribute.
13:04 And a moment ago, I was talking about finding a path around a lake...
13:07 ...and what I showed you was actually the shortest path, the graphic on the top.
13:12 It's the route that minimized the distance traveled.
13:15 And distance in this case was the cost.
13:17 And as we see here, it was 25 miles long, so we say the total cost was 25 miles.
13:22 To minimize distance, each edge or street needs to have an associated cost attribute, and that cost is in miles in this case.
13:32 And using distance as a cost attribute can be good when you're finding a route for a person who's walking.
13:38 However, if your route is for a person who's driving, you'd probably want to minimize the driving time...
13:43 ...which is a function of distance and speed or speed limits. In this case, each edge needs an associated driving time.
13:52 As these demon-...as these graphics demonstrate, the paths could be different...
13:55 ...depending on whether you're minimizing distance or driving.
14:00 And if you see, on the top one, person who would walk would walk through the city...
14:04 ...because it doesn't really matter if you're walking on streets or country roads.
14:08 But down below, you would take the path through the country so that way you don't have to stop at stoplights and stop signs.
14:18 A single network dataset often has multiple cost attributes.
14:21 This way you can find the shortest path from...for one analysis and then find the quickest path for another analysis.
14:28 You can even include other kinds of cost attributes in your network dataset as well.
14:34 So we've been looking at costs on edges. And...but network attributes always span edges, junctions, and turns.
14:42 So that means that the junctions at the ends of edges and the turns that model transitions from one edge to another can also have costs.
14:52 So I'm going to show you an example of a turn delay.
14:54 So the orange line here represents a simple route from point A to point B.
14:59 And the first road segment takes five seconds to traverse, and the second road segment takes another five seconds...
15:04 ...so it's a total travel time of 10 seconds.
15:07 But there's a left-hand turn light and some traffic there, so on average, it takes 15 seconds just to make that turn.
15:14 So what we can do to model this is add a turn feature, and...
15:18 ...represented by the yellow arrow, and then assign a cost of 15 seconds to that turn.
15:24 And then that changes the total cost of the...the route to 25 seconds.
15:31 But digitizing all the turns in a network would take a long time, so that's why we offer the global turn delay evaluator.
15:37 Global turns add a cost to every two-edge turn sequence in the network, unless a turn feature is already there.
15:44 In that case, the turn feature would override any global turn delay.
15:48 So this means that all you need to do is digitize the most important turn delays using turn features...
15:53 ...and then you can use the global turn delay evaluator to generically model the rest.
15:58 And all you need to do to create global turns is to fill out the dialog box shown here.
16:03 And one nice feature about this is that you can specify turns based on turn type and road class.
16:08 So if you're taking a left-hand turn, you can have that cost more than a right-hand turn.
16:13 Or if you're making a left-hand turn from a local road onto a highway...
16:18 ...that would take longer than a left-hand turn from a local road onto another local road...
16:22 ...because when you're trying to get onto a highway from a local road, you typically have to wait for a gap in traffic...
16:28 ...and it takes longer.
16:30 So here's a new feature that we added in the release of ArcGIS 10, historical traffic.
16:35 And its purpose is to capture how travel times change throughout the day and throughout the week.
16:41 So if you have a traffic-enabled network dataset, what you do is you provide a day and a time of day...
16:47 ...and then Network Analyst can determine the best route based on that time.
16:51 So here, the best route at 8:00 a.m. is along...from the city to the suburb is along the divided highway.
16:57 But at 6:00 p.m., the quickest route changes to the winding road on the left of the divided highway...
17:02 ...because there's more traffic going out to the suburb.
17:05 And it's important to note that Network Analyst won't just tell you...or won't just determine if there's traffic on a road and avoid it.
17:12 It'll determine what's better.
17:13 Is it better to wait in traffic?
17:15 Is it quicker that way?
17:16 Or is it quicker to find an alternate route?
17:21 So far, we've looked at the usefulness of cost attributes, but we also have restriction attributes.
17:26 And these allow you to model things like one-way streets, blocked intersections...
17:29 ...and turns that are prohibited by law.
17:32 You can turn these restrictions on and off when you solve an analysis.
17:35 So if you're finding the best walking path, you would turn off all these restrictions shown here.
17:39 But if you're going to find the best driving route, then you would turn these restrictions on so you don't break the law.
17:46 And another kind of attribute is a descriptor attribute, and it just describes a particular characteristic of the network.
17:53 It's not actually used by the solver, but what it's...
17:56 ...it's often used by is another restriction attribute to model a more complex restriction.
18:01 So let's say we have a descriptor attribute that stores the minimum clearances of bridges or overpasses.
18:07 So by itself, it wouldn't do anything, but if a...so a vehicle would still be routed underneath overpasses that are too low for them.
18:15 But what you can do is have a corresponding restriction attribute that references this descriptor attribute...
18:20 ...and also a vehicle characteristic that you enter at solve time, such as the vehicle height, then...
18:27 ...the route would avoid low overpasses.
18:31 So what I've shown you so far is mostly static characteristics of the road network, and they're built into the network dataset.
18:39 So the height of the overpass and the connectivity of the streets won't really change that often.
18:44 But what about temporary changes to the network that you can change when you're performing the analysis?
18:51 One of those changes would be the U-turn policy.
18:53 And this is about allowing or prohibiting U-turns at intersections.
18:59 So sometimes the quickest way to get from one place to another involves a U-turn.
19:02 It...it takes a long time to make a left-hand turn...
19:05 ...it might be quicker to go straight through the intersection to the next intersection, make a U-turn, come back...
19:10 ...and then turn right and go onto the same street you had originally intended to go on.
19:15 But maybe you don't want to make these kinds of questionable maneuvers...
19:17 ...or the vehicle you're driving, it really makes these kinds of maneuvers difficult.
19:22 So what you can do is prohibit U-turns at intersections.
19:27 We also have restriction barriers, which allow you to model parts of the network that are currently inaccessible.
19:33 So when river covers a road and makes it inaccessible, what you can do is simply create a barrier at analysis time...
19:43 ...and then when your analysis will restrict access...access to those roads.
19:49 But once the water recedes, all you need to do is delete the barrier, and then you're...
19:53 ...you can re-solve your analysis and have access to those roads again.
19:56 And the same is true for when a boulder falls on the road or a sinkhole forms.
20:01 You just add the restriction barrier, and once the boulder is removed or the sinkhole is filled, which might take a while...
20:08 ...but you could just delete it and use the roads again.
20:14 The scale cost barriers are a lot like restriction barriers, but instead of restricting travel completely...
20:19 ...what we do is scale the cost of traveling on the roads.
20:23 So, for instance, you can create a scale cost barrier to represent a snowstorm...
20:27 ...and the barrier will only increase the travel times of the roads it covers.
20:31 Once the storm is past, you can just delete the barrier and the travel times go back to their normal times.
20:37 And the same idea is true for roads that are under construction.
20:41 Curb approach specifies which side of the road you want your vehicle to be on when you arrive at a stop.
20:47 So if you look at the slide on the left side, we see the school bus has arrived at the school.
20:52 And since the school is on the left side of the bus...
20:54 ...the students have to cross the road, and this exposes them to a little bit more danger than the students on the right side of the slide...
21:02 ...who are dropped off on the same side as the school, so they don't have to cross the road.
21:06 And when you load or unload a truck, you also want to keep this in mind so you don't have to carry your cargo across the street.
21:13 So I'm going to give you a quick demo of some of these capabilities.
21:17 And the purpose of the demo is...
21:19 I'll show you using a two-stop route.
21:23 And the purpose of the demo isn't to show you that Network Analyst solves routes or how to use the user interface controls.
21:31 Rather, it's to show you or highlight the modeling capabilities of the network dataset...
21:35 ...and show you that how you model the network dataset affects your analysis results.
21:41 So I have two stops; I'm going to walk from the first stop down to the second stop.
21:47 And I'm going to open up the analysis properties, and I'm going to use a distance-based cost attribute to solve the analysis.
21:56 And this gives me the shortest path.
21:59 But there's a lot of Dumpsters and alley cats along this route...
22:05 ...so I've determined that this is not really within my walking distance anymore.
22:09 I would rather use my car instead.
22:12 So what I do is create a drive time or use my drive-time cost attribute, and then solve again.
22:20 And I get a slightly longer route.
22:22 And instead of going on these roads that we went on when we were walking, it takes these roads.
22:28 The reason is, is that the speed limits along here are faster than the speed limits along here.
22:34 Now, let's see. I'm not including the rules of the road in this analysis yet.
22:41 So I have one-way streets that I need to consider.
22:44 I'm driving against a one-way street here, which is not going to work.
22:47 I have restricted turns, so all these red arrows represent illegal turn maneuvers.
22:54 And then I also have a turn delay here.
22:57 There's a left-hand turn light that takes, on average, 25 seconds to drive through.
23:01 And there's a dedicated left-hand-turn lane here too.
23:05 And I have a height restriction.
23:07 There's a pedestrian bridge that crosses this road, and the minimum clearance of it is 12 feet, 6 inches.
23:13 And I also want to show you traffic.
23:16 So to show you traffic, just to visualize it, I'm going to use time in the ArcGIS and I just enable time on the map.
23:24 And I'm going to take my trip at 3:30 p.m. on Wednesday, so I've already entered that information here.
23:30 And what we see on the map is the roads are color coded.
23:35 And here we can see what the color codes mean.
23:38 So red is stop-and-go, and it goes on up to green, which is free flow or unimpeded traffic.
23:45 But I also have to set this information in my network analysis.
23:48 So I'm going to check on all these restrictions...
23:51 ...and, let's see, my vehicle height is 6 feet tall, and I'm also going to use a start time of 3:30 p.m...
23:58 ...and I'm going to travel on Wednesday.
24:01 And I'll solve, and I get a different route.
24:04 And this time, it makes a left-hand turn here...
24:08 ...and then goes underneath the pedestrian bridge and then arrives at the second stop.
24:13 Now, if I was driving a truck instead of a car, then what I would do is just tell the analysis that my vehicle is 13 feet tall.
24:23 Then solve again and get a slightly different route, or actually, pretty dramatically different route.
24:29 Now, I'm going to go in here.
24:31 I have turn delays enabled currently, but let's see.
24:37 Actually, I need to turn on my historical traffic; I didn't assign that before.
24:46 Okay. And I'll re-solve.
24:48 Okay, this is the route that I was actually expecting.
24:50 And this has three left-hand turns.
24:53 And when I'm driving a semitruck, I don't really want to make left-hand turns unless I have a protected intersection.
24:57 What I mean by that is that I have a left-turn light and that'll stop traffic from going for me...
25:04 ...and that way, I can make a left-hand turn pretty easily and safely.
25:08 So I can use the global turn delay evaluator.
25:10 So I go to my network dataset and...
25:12 ...open up the Global Turn Delay Evaluator dialog box...
25:19 ...and here I assign different turn delays based on the turn type and road class.
25:27 I'm just going to scale up all my left-hand turn lane...left-hand turns to take 10 minutes...
25:33 ...so that'll essentially make left-hand turns...or prohibit them in the analysis.
25:38 And then I'll re-solve.
25:40 And here I get a different route without any left-hand turns.
25:43 Instead, I make a right-hand turn, a U-turn, another U-turn, and then two more right turns.
25:48 But if I'm not making left-hand turns, I probably don't want to make U-turns either...
25:51 ...so I can shut those off in my analysis and say they're not allowed.
25:58 I'm also going to take a look at the curb approach.
26:00 And, see, I'm approaching the second stop, so it's on the left side of my vehicle...
26:04 ...and since I'm driving a truck and I'm going to unload it, I don't want it to...
26:07 ...I don't want to have to carry the cargo across the street.
26:09 So I'll change the curb approach of that stop to right side of vehicle...
26:15 ...re-solve, and now I'm arriving at the...the correct side of the road.
26:20 Let me zoom out.
26:23 And notice I'm also not taking any left or U-turns. Okay.
26:29 And lastly what I want to show you are the barriers.
26:33 So we have some construction on a road here, and one of the...
26:37 This construction has slowed down traffic, so it's about 10 times the normal speed.
26:40 One of the construction workers has broken a water main, and he's flooded out these roads here, and it blocks my road.
26:47 I'm going to load the construction as a scale cost line barrier, and then I'll load the flood as a restriction barrier.
27:04 And then I re-solve, and I'll generate directions while I'm doing that too.
27:09 Minimize those for the time being.
27:10 And now I'll just go over this route really quickly with you.
27:14 Leave the first stop, make a left-hand turn.
27:16 Even though I've used global turn delay evaluator to essentially prohibit left turns...
27:20 ...I have a turn feature there, and that has a cost attribute on it.
27:24 So it's not a prohibited left turn since...
27:28 ...if you remember, I told you that turn features with cost attributes override the global turn delay evaluator.
27:34 And that's what's happening here.
27:35 So this allows you to model dedicated left-hand-turn lanes.
27:40 And then you can drive down here, and it doesn't make a right-hand turn here because you've got a turn restriction.
27:45 It goes past this road; there's a one-way street going in the other direction.
27:49 It'll take this one-way street and make right-hand turns for the rest of the trip.
27:53 It enters into the scale cost barrier, and for this portion of its trip, it goes 10 times...
27:59 ...it takes 10 times longer than normal, and then arrives at the second stop.
28:03 And this is actually the quickest route from point...stop 1 to stop 2 given all the criteria I've given it.
28:09 And then, I can take a look at the directions; this gives me turn-by-turn instructions.
28:15 And I can also take a look at inset maps and highlight the turn with the highlight arrow.
28:23 So that's just a quick overview of the network dataset modeling capabilities.
28:30 And the things I want you to remember from that is I network dataset modeled the roads and the cost attributes...
28:34 ...modeled the best pedestrian and driving paths.
28:37 And we're able to model both static and dynamic modeling capabilities, and they changed the results of the analysis.
28:47 So I'll turn it back over to Patrick to talk about the different types of analysis we offer.
28:53 Thank you, Bob.
28:54 And if you ever need to deliver something during a flood in Chinatown in San Francisco...
28:58 ...now you guys'll know the correct way to go through all the one-way streets and all of that.
29:04 So we showed you the six types of analysis at the beginning of the presentation, and we call those our solvers.
29:10 And what Bob demonstrated was network dataset capabilities, but he used the route solver to do the demonstrations.
29:17 So this route solver helps you find the best way to get from one location to another location...
29:22 ...or to visit several locations along the way.
29:25 And we call these locations stops in the...in the route analysis.
29:28 You see in the image a four-stop route; it starts in the bottom right and...and...
29:33 ...and is optimized going from that stop 1 all the way to stop 4.
29:36 Now those stops could have been added to the route by clicking on the map interactively...
29:41 ...and it would snap the location to the network.
29:43 Or you could've entered an address in the geocoding...
29:47 ...with the geocoding capabilities, that would've geocoded that address to a point along the network.
29:52 Or you could've loaded the points from an existing feature class or feature layer.
29:56 Now with the route solver, there's a few options available to you, including time windows.
30:00 And these are a property of the stop.
30:02 And the route will help optimize when you want to visit that stop according to the time window that you set for...
30:07 ...for the...to...when best to visit.
30:11 Now the best...the best route can be determined in the order of locations specified by the user; in this case, 1, 2, 3, 4.
30:17 It'll find the shortest path from 1 to 2, from 2 to 3, from 3 to 4.
30:21 Or you can set the route solver to find the best sequence for you and do what's called the traveling salesperson problem...
30:26 ...where it will optimize these stop...it'll optimize the stop order for you.
30:30 And that's if you don't care when you visit the stops; you just want to get to all four of them...
30:34 ...you can have the route solver do that.
30:36 You can also create multiple routes.
30:38 You can specify the start time of the route, which you have to do if you're setting time windows.
30:43 And you can generate text directions for the route after it's been solved, as you saw Bob do.
30:47 And these directions can be exported as...
30:50 You can print them out or export them as CSV files or XML files to send to drivers and to distribute electronically.
30:56 As with all the solvers, you can add point, polyline, and polygon barriers if you want.
31:02 These barriers can be restrictive or additive for points.
31:05 So you can either block traversal or add some cost to the traversal...
31:09 ...like it takes an extra minute to go through this area, or this tollbooth costs an extra few dollars.
31:14 With polylines and polygon barriers, you can either restrict the area underneath the barrier...
31:18 ...or you can scale the cost within that barrier to make sure it's a little bit slower.
31:22 You might have seen in the plenary when he scaled Cleveland...
31:25 ...using a barrier that represented inclement weather in the form of snow.
31:30 So the route solver can be used in anything you need point-to-point routes for...
31:34 ...for example, organizing your days' worth of orders for one vehicle or inspector.
31:39 You can also access a free Network Analyst routing service via the Find Route tool within ArcMap and ArcGIS Explorer as well.
31:46 And that'll give you some simple point-to-point routing capabilities without needing the Network Analyst license.
31:52 So say you want to find the nearest hospital to an EMS incident or route the three closest police cars to a bank robbery...
31:59 ...or find the nearest ATM to your house, you'll do that using the closest facility solver.
32:04 Now this solver can be applied to emergency vehicle dispatching, to finding the nearest store to a customer...
32:09 ...or any use where you want to pick out a set of locations from the starting or ending point.
32:14 Now the scenarios are set up with what we call facilities and incidents with this solver.
32:19 Those are the names of the subclasses in the Closest Facility layer.
32:21 If you're setting up an EMS dispatching scenario for someone who called 911 from their house...
32:26 ...the closest facility solver would find the nearest facility or ambulance from the incident or accident site.
32:33 Now you can set a cutoff, for example, if you only want to find the police cars within 15 minutes of this bank robbery...
32:38 ...and you want to ignore the other vehicles outside of 15 minutes.
32:42 Or you can set a limit to the number of facilities to search for; only give me the five closest gas stations from my house, for example.
32:50 Now streets can have traffic modeled differently for each direction of travel, as you saw with one-way roads...
32:54 ...for example, or turn restrictions.
32:57 So you can generate your routes considering travel from the facility to the incident...
33:01 ...like routing fire engines from a fire station to a fire incident...
33:05 ...or from the incident to the facility, like taking a burn victim from the scene of the fire to the nearest hospital.
33:10 And note that you can also perform multiple closest facility analyses simultaneously if you have...
33:15 ...multiple incidents and multiple facilities.
33:18 So I'll give you a quick demonstration of...of dispatching using the closest facility solver.
33:25 Now this is a simulation of fire and police dispatching using the solver that I just talked about.
33:30 It's a Web application written against a Flex API accessing the closest facility solver via the REST endpoint.
33:38 So it's an ArcGIS Server application.
33:41 Let's see if I can...I was getting a little slow response earlier today; and it's all running local on the machine.
33:47 So I can click on the map, and...and it'll geocode an incident.
33:52 Let me refresh this.
33:57 Set up my solver again, get my police cars moving.
34:05 Slow network connection, looks like?
34:07 Yeah. There they go.
34:09 Okay. So you clip on...click on the map, and it'll snap to the network and find the nearest three police cars.
34:14 I've chosen a police incident.
34:16 I’ll click on the fire department incident and click, and it'll find the nearest three engines.
34:21 Or I can geocode an incident by entering an address and choosing a general incident type...
34:25 ...which will just get the nearest three vehicles.
34:30 So granted, this app is a simple version of dispatching...
34:33 ...but the important thing to note is how the nearest vehicles are located and dispatched...
34:38 ...and that...that there's a count of three of them.
34:40 And that the network analysis can happen across ArcGIS Server with GPS tracking of the vehicles without a problem.
34:46 Currently the route, closest facility, and service area solvers are available as REST endpoints...
34:53 ...and there's SOAP endpoints for all six of the solvers.
34:56 And you can also do geoprocessing and publish any Network Analyst workflow you want to.
35:01 And this is fun to play with too.
35:07 And you're welcome to see this demo later on if you want to...want to work with it a little bit.
35:11 [Audience question] Could you set up separate network analysts for your police vehicles, your fire trucks?
35:17 [Audience question] 'Cause you could set things like U-turn restrictions.
35:20 [Audience question] I mean, a police car can make a U-turn at a median where a fire truck can't [unintelligible]
35:25 The question is, Would you set up separate Network Analyst scenarios for police versus fire?
35:30 You'd probably use the same network dataset...
35:33 ...but within the analyses themselves, you can turn on or off restriction that apply only to those type of vehicles.
35:38 So you could set the...the restrictions you would set for fire engines against the restrictions you would set for police cars...
35:43 ...and you could solve them separately.
35:44 In this case, it was just they're all considered the same thing.
35:48 They were all facilities in the closest facility problem and it's dispatching them to the nearest incident.
35:56 So from that demonstration, you see they used the closest facility solver to route groups of vehicles to a location...
36:02 ...or groups of facilities, is the term we use...
36:04 ...and that the solvers can reroute moving vehicles using GPS tracking even though this was a simulation.
36:10 And this was all done through Web service through ArcGIS Server.
36:16 So if you want to know how many people live within 10 minutes of a proposed movie theater location...
36:22 ...or how many addresses don't have police coverage within 10 minutes...
36:25 ...you'll do that with what we call the service area solver.
36:28 Now this solver finds a region around a location or facility...
36:31 ...that can be accessed within a specific cost or cutoff or...or break as we call it.
36:36 Now with this solver, we use the term facility again like with the closest facility solver...
36:40 ...but that means it's the location you're solving against, the...the area on the network that you're solving outward from.
36:47 You can specify direction of travel again, like we could with the closest facility solver.
36:51 So are you considering how customers will arrive at your store...
36:54 ...or are you considering how far out from your store you can...you can service an area?
36:59 Now there's also many polygon and polyline generation options as you see here...
37:03 ...with polygons on the top and polylines on the bottom.
37:06 The polylines show the streets that were covered within that break.
37:09 The polygons are a generalized polygon around those covered streets.
37:13 And sometimes they're called drive-time polygons.
37:16 Now for scenarios with multiple facilities, the polygons or polylines can either overlap...
37:22 ...as you see on the lines on the bottom right or you can have them stop when they hit each other...
37:26 ...as with the service areas you see along the Eastern Seaboard of the United States there.
37:31 It just depends on the type of problem you're trying to solve what's most appropriate.
37:34 With service area, you can't generate directions because it's not point-to-point routing.
37:39 You're driving out in every...in every direction, so it wouldn't make sense to print directions in that way.
37:44 So Business Analyst was another application that they demonstrated at the plenary...
37:48 ...and that uses service areas quite often to help businesses manage and analyze their data.
37:54 This could be used to solve to find fire response zones or customer service areas, for example.
38:02 Go through a demonstration of it.
38:03 Now this is a map that I got from the Rancho Cucamonga, California, fire department...
38:08 ...that they used to show that there was deficient fire coverage in the center of the city.
38:12 They were tasked with the showing, finding any problems in the city with the fire engines reaching it.
38:18 They have seven fire stations, six operational and one in the northwest corner that was being built...
38:23 ...and they wanted to adhere to reaching every incident within four minutes, excluding the time leaving the station.
38:30 So this map was created and presented to the Rancho Cucamonga City Council...
38:33 ...as evidence that there's deficient fire coverage in this end of the city.
38:36 And the council approved starting the process for building a new fire station because of it.
38:40 So why did they choose to use service area for this kind of problem?
38:44 The reason is, is that they wanted street coverage and they wanted visualization.
38:48 They weren't trying to find the distance between individual points or sets of points...
38:52 ...they weren't locating facilities or trying to figure out where to put the new fire station...
38:56 ...and they weren't trying to route their fleet of engines.
38:59 What they wanted was to clearly see the areas that weren't sufficiently covered.
39:02 And that lends itself very well to the service area solver.
39:10 So I got the addresses of the Rancho Cucamonga fire stations from their public Web site...
39:14 ...and I created an empty service area layer.
39:18 Here in the NA window you can see it.
39:20 And I'll load these fire stations in as facilities by choosing to load, picking my fire stations, and loading them.
39:29 So now my seven fire stations are...are loaded into my service area problem.
39:33 And I click Solve...
39:35 ...and it'll go through and figure out which streets are covered from these fire stations within four minutes...
39:39 ...and generalize the polygon around that street coverage.
39:42 And you can see that there is definitely a coverage hole in the center of the map here.
39:46 It doesn't look exactly like their map because we have underlying street data.
39:50 They might have different turn restrictions or one-way streets or even different speeds along the streets.
39:56 So with the data I used with the free Data & Maps data, this was the hole that's generated in the coverage within four minutes.
40:03 So the question now, because clearly, there's a problem, is how do you fix the problem?
40:07 And we'll get to that when we talk about our next solver.
40:12 So the takeaways from that demo are that you should first think about your problem type...
40:17 ...and then decide on the appropriate type of analysis you want to do, the appropriate solver that you'll choose.
40:22 In our case, it was coverage and visualization, and that's perfect for service area.
40:27 Oh, yeah. And every time you see one of these intro slides in any of the presentations...
40:32 ...there's service areas on that little strip up there.
40:34 So you can thank the Network Analyst every time you go to any presentation for the rest of the UC.
40:39 Except there's laser beams coming out of one of the service areas, and I'm not sure why that did that.
40:46 Okay, so the next solver is the location-allocation solver.
40:48 And this one's new at 10.0; we just came out with this one.
40:51 The old cliché for real estate is there's three important things to consider, and that's location, location, and location.
40:58 And in that case, this solver is your best choice.
41:02 So it'll help you find where the optimum location is for a facility.
41:06 The name of the solver breaks down what it does.
41:07 It helps you locate facilities by allocating demand to those facilities.
41:12 Now, we call them facilities like with the closest facility and the service area solver...
41:16 ...but they're locations that you're asking the solver to pick from.
41:19 They can be warehouses, bus stops, police stations, hardware stores...
41:25 ...emergency vehicle predeployment areas, or whatever type of place you want to pick the best location for.
41:30 Evacuation sites is another example I've seen.
41:33 The other main input for the solver other than facilities is the demand points.
41:37 Now these are the people or the things that require the goods or services that you're providing from your facility.
41:43 This could be ZIP Code or census tract centroids.
41:47 It could be business customers.
41:48 It could be street intersections or whatever you want your facility to service.
41:52 Now the demand points can be weighted for importance.
41:55 Let's say the population in that ZIP code or census tract...
41:59 ...or the expected consumption by consumers at that demand point.
42:04 So remember also that this is all done along the network.
42:06 These aren't straight-line distances; these are network paths.
42:09 In the plenary, when they showed the location-allocation solver in Cleveland with the allocating for health facilities...
42:17 ...it was based on travel along the road network, not just straight-line, crow-flies distances.
42:22 This is a complex problem, so there's quite a few analysis types we offer, and they're listed here.
42:27 There's minimize impedance and what that'll do is it'll solve to reduce the travel costs...
42:32 ...from your customers to your location or from you to your customers.
42:35 Now, this is good for locating warehouses, for example, to minimize the cost of transporting goods to your outlets.
42:41 There's maximize coverage, which will solve to reach as many of the demand points as possible.
42:46 An example of this usage might be locating fire stations...
42:48 ...which might come in handy in our Rancho Cucamonga demo, or locating police stations to reach the most citizens.
42:54 You just want to get out as much demand as possible.
42:57 You can minimize the facilities...
42:59 ...and what that'll do is it'll keep the...try to keep the demand you capture the same
43:03 ...while reducing the number of facilities you use to do this.
43:05 And it'll help you allocate your resources more efficiently, more cheaply.
43:11 You can maximize attendance.
43:13 And that will maximize coverage but take into account that demand might be reduced...
43:18 ...the farther away somebody is from your facility.
43:22 That'll help you adjust for demand you lost to your...
43:24 ...your customer [sic], in case you don't have a lot of information about your competitors.
43:28 The idea is that the further away somebody is in this type of problem, the less likely they are going to go to your facility.
43:33 That could apply to grocery stores, especially.
43:35 No one's going to travel a long way to go to your grocery store if there's a closer one.
43:39 And you can maximize...maximize market share, and that'll get your facilities the most demand...
43:44 ...in the presence of competitor facilities.
43:46 And that works well with large discount stores or something where you have a lot of information about your competitors.
43:52 You can also target the market share.
43:54 And that's like maximized market share, except you'll, instead of asking it...
43:58 ...giving it a number of facilities you want to locate.
44:00 You tell it the percentage market share you want, 12%, 97%, whatever problem you're trying to solve...
44:05 ...and it will...the solver will determine the number of facilities you need in order to capture that level of market share.
44:12 So we'll go back to Rancho Cucamonga here and try to solve their problem of fire coverage.
44:21 So we're clearly aware that there's a problem in the center of Rancho Cucamonga; I'll zoom in to it.
44:29 Turn on my location-allocation layer.
44:33 So we're clearly aware there's a problem there, and the next step is to plug that hole in the coverage problem.
44:38 Now how do we best solve this?
44:39 We could just add locations one at a time by clicking and adding new facilities to the service area problem.
44:45 And after a lot of effort, that might give us a pretty good guess, but it wouldn't give us the optimal location.
44:51 For that, we should use the location-allocation solver; it's its job.
44:54 So since I'm a smart GIS professional, I'm going to read the documentation and do the tutorials.
45:00 And I can determine that the type of analysis that I want to do... Whoops. I'm on my service area problem.
45:08 The type of analysis I want to do for location-allocation is maximize coverage, and that's the one I mentioned before...
45:14 ...where you're trying to just get to as much demand as possible.
45:16 And that makes sense for a fire engine because you're not preferring one set of customers over the other...
45:20 ...and you're not less likely to serve people farther away.
45:24 So choose Maximize Coverage.
45:25 We know there's seven existing fire stations, and we want to add one more fire station to their...to their problem...
45:31 ...so we set it up for eight facilities to choose.
45:34 And we also set our impedance cutoff at four minutes just like our service area problem.
45:38 Now I've loaded our seven fire stations in as required facilities.
45:41 You can tell they're required by the little star...
45:44 ...and that's a setting on the...on the solver, or on the properties of the location.
45:48 And the next step is choosing which sites we want as our potential sites.
45:52 In our case, there's plenty of public parks in this area...
45:55 ...and that's a little bit easier for a city to take over than eminent domaining somebody's house...
45:59 ...so we'll suggest these public parks as potential fire stations.
46:03 So that's our first input.
46:04 We've got our facilities.
46:05 And the second set of inputs you need for location-allocation is demand points.
46:10 In our case, we just want to reach out as far as possible from each of these fire stations.
46:14 So I've set up points for every junction in the city.
46:17 And it's easy data to get, and I can load those in as demand points to my location-allocation problem.
46:22 And in our case, there's 9,129 demand points.
46:26 And you can see all these potential sites are loaded in...
46:29 ...and they don't have a star in them, meaning that they're candidate locations.
46:33 So I can click Solve, and the location-allocation solver will quickly go through...
46:37 ...out of these 19 facilities and nearly 10,000 demand points...
46:40 ...and tell us which of these facilities is...best serves to allocate to these demand points.
46:46 Yes, sir.
46:47 [Audience question] Your demand points are entered as such as all the intersections? Is that what it was?
46:51 Yes. I picked every junction in the city.
46:53 Because it's easy data to get out of a network dataset...
46:55 ...especially because there's your edge source, which is your streets, and there's a junction source, which is your intersections.
47:01 But you can use anything for your demand points.
47:03 You might want to use buildings or population information.
47:07 And I'm...
47:08 [Audience question] It was a dataset that was very convenient.
47:10 Yeah. And it also...I bet Rancho Cucamonga probably would've used incidents, used some historical incident data...
47:16 ...but I don't have their data, so I went with easily available data.
47:20 [Audience question] Excuse me. One question.
47:21 Sure. [Audience question] Is there anything...I noticed there's some gaps there...
47:24 [Audience question] ...[unintelligible] bottom middle screen.
47:26 Down here? [Audience question] Yeah. Is there any way to force those two points to meet?
47:31 You mean with these...
47:32 [Audience question] With your junctions.
47:34 [Audience question] You see, you've got a gap between those two junctions...
47:37 [Audience question] ...because the street between the junction points is not covered.
47:41 I'm not quite sure which area you're talking about...
47:42 Oh, he's using the junctions instead of the streets.
47:45 So he could use street centroids instead.
47:47 That might be why.
47:51 Oh, yeah, this is just as you see, each intersection.
47:53 You can see especially in this slide that it's just the intersections that were chosen as demand points.
47:58 Or it's street centroids with junctions.
47:59 [Audience question] Do you use the parcel, [unintelligible] parcel?
48:01 Sure. Yep.
48:02 That would've worked. Let's say you couldn't get centroids for all the parcels and try to reach every parcel as well.
48:06 And that would probably be more accurate too, 'cause you'd get a guess on exactly where they would be dispatched to.
48:11 And perhaps you could weight them by the...the population at those centers.
48:16 So you can see here that the location-allocation chose...
48:21 ...the southwest corner of Rancho Cucamonga Central Park, which makes sense; it's centralized there.
48:25 And it put a little star in it meaning it was chosen as our one extra facility.
48:29 So just to show, I will copy that facility...
48:34 ...into our service area problem, and re-solve our problem, and we'll see if it...if it plugs that hole in the coverage.
48:42 So it'll take a second again to solve outward from each of these facilities, eight now...
48:47 ...and to generalize these polygons around the facilities.
48:49 And you can see that it covers this area in the middle.
48:51 There's a little issue here because it doesn't quite cover across this...this park.
48:55 So that could be fixed by streets or however they were going to do it.
49:02 So for locating the fire stations...yes, sir?
49:05 [Inaudible audience question]
49:13 Your question was, What is the output of a location-allocation problem?
49:16 [Inaudible audience question]
49:25 Right. So the question is what kind of output would we have and is distance included. Yeah.
49:31 There is several parts, or several components to the output.
49:35 So one of the pieces of the component are the chosen facilities.
49:38 Another piece is the...how much demand is allocated to the facilities, or which demand in this case is allocated to the facility.
49:46 And it will also give you the cost of traveling from the facility to the demand point as well.
49:54 You can get that information.
49:55 So each of these lines contain the network cost to travel from each of those demand points to the facility.
50:01 As we said, it's not as the crow flies, the Euclidean cost.
50:04 It's the cost of traveling along that network.
50:07 The lines just demonstrate that there's an association between that chosen facility and the...
50:10 ...and the point you used as a demand point.
50:16 [Inaudible audience question]
50:18 [Inaudible audience question]
50:25 Yes. Yes.
50:27 This was simplified just to reach as far as you can within four minutes.
50:30 Or as many demand points as possible...
50:32 Yeah. Within four minutes.
50:33 And Rancho Cucamonga is an interesting problem, because there's a big mall on this side of the city.
50:38 So they would really want to weight demand more if they...if they truly were doing this analysis to...
50:42 Right. It sounds like the question is do we solve capacitated location-allocation.
50:46 And, no, we don't have capacitated allocation right now, location-allocation right now.
50:51 [Audience question] You don't. Do you want to define capacitated location-allocation for everybody?
50:54 Capacitated location-allocation is where you assign a particular number of demand points to a facility...
50:59 ...and here, we're just making sure that we have a coverage of four minutes to reach all our demand points.
51:05 And we don't really care how many demand points are covered within that four minutes, just as long as they're covered.
51:12 [Audience question] Okay. But you could use capacity to solve your other options using the location-allocation method, right?
51:22 You can weight the demand points themselves, yes.
51:24 [Audience question] Yeah. Yeah. So it was about the school application or, you know, you would set [unintelligible]...
51:29 [Audience question] ...how many children could go to school for public health centers or so? You use that?
51:36 Not out of the box, per se.
51:38 But if you attend the...what's the one Deelesh is giving?
51:41 Is it Automating Workflows with Geoprocessing?
51:43 Yeah. He has some geoprocessing tools that allow you to do...answer that type of question.
51:47 So what it'll do is it'll get the results of an OD cost matrix and export that into...
51:54 ...can use the results in a linear programming application and then have that run.
51:59 It uses Python, PuLP linear programming.
52:02 And he does a school allocation problem.
52:04 And that we have a script that'll do school allocation using the...one of our other solvers that we haven't gotten to yet...
52:09 ...the OD cost matrix and a linear programming package together to do it.
52:12 [Audience question] Okay.
52:16 So the takeaways from this are that, again, you want to think about the problem type you're solving...
52:21 ...and then pick the solver most appropriate for it.
52:23 In our case, we were locating a facility, so it made sense to use location-allocation.
52:28 And we also used the correct analysis type out of there for maximized coverage, just to find the...
52:33 ...the coverage we could reach from our fire station.
52:38 So our next solver is the vehicle routing problem solver.
52:41 Now this will help you determine how best to assign a group of customers to a fleet of vehicles...
52:47 ...as well as sequencing and scheduling the visits to those locations.
52:51 This is often used for distribution, for inspectors, for assessors, for technicians, for paratransit.
52:58 Now, there's many input classes for this solver, but there's three that you need to know to get started, the first one being orders.
53:04 Now, these are the places you want your vehicles to visit.
53:06 They could be orders, as they're called...
53:08 ...they can be inspection sites, they could be people you want to pick up and drop off.
53:12 Now the depots is the second one.
53:14 That's the starting and ending points of your routes.
53:17 That could be warehouses where you pick up the cargo.
53:19 It could be bus depots, et cetera.
53:21 And the third is routes.
53:23 Now, these are the separate routes you want created.
53:25 It's generally used to represent vehicles.
53:27 If you have five trucks, you'll use...set up five routes within your vehicle routing problem solver.
53:32 But it could be...also be days.
53:33 For example, if you have one truck and you just want to set up your orders throughout a week...
53:37 ...you can set it up to do Monday, Tuesday, Wednesday, Thursday, Friday, and it'll allocate the orders optimally for you.
53:44 So VRP will help you generate high customer service, because it'll honor time windows like with the routing problem solver...
53:49 ...and it'll keep your overall operating and investment costs for the fleet to a minimum.
53:54 Remember, shorter routes, optimized routes equal less gas, less tri-...
53:58 ...time driving, so less driver overtime, more customers reached for cheaper.
54:04 ArcLogistics is a product that was mentioned at the plenary as well.
54:07 That's an Esri product that's a stand-alone application built upon the vehicle routing problem solver.
54:12 Now it's a fleet routing solution that's meant to be easy for dispatchers instead of GIS professionals.
54:18 So, and it's also one of the pioneers in cloud computing and Software plus Services.
54:22 And it can save up to 30 percent in fleet-related costs just by optimizing these routes.
54:27 There's many possibilities as you have fleets of vehicles and can think of routes.
54:31 So Bob will give you a demonstration of appliance delivery using the vehicle routing problem solver.
54:35 Okay. For this demonstration, the study area is Indianapolis...
54:39 ...and I have an appliance store that's represented by that square in the middle of the map.
54:44 And each day, customers come in and they purchase appliances.
54:48 So I need to deliver the appliances and install them in their house.
54:52 So these are the deliveries I need to make today, represented by the circles...
54:58 ...and I have a fleet of four vehicles to make the deliveries.
55:02 So what I'm going to do is create a vehicle routing problem, and first I'm going to load my depot, which is the store.
55:10 That's where my vehicles start and end their route each day.
55:13 We can also have vehicles start and end from two different locations if we wanted.
55:22 So I'll load my appliance store as the depot, and I'm going to turn off some of these other layers.
55:31 I'll turn off the customers of the appliance store for now.
55:34 Next I want to load the information about my trucks, and I load those into routes.
55:38 And just think of routes as representing trucks or drivers, because the drivers drive the trucks along the routes.
55:47 And here's my vehicle information that I'm loading.
55:50 I have four vehicles.
55:51 We can take a look at that information.
55:52 There's lots of properties we can model.
55:56 So here's the depot that the vehicle starts from, that it ends at.
56:00 And one of the properties I want to point out is the capacities.
56:04 This is how much the vehicle can carry and I've chosen to have it represent cubic feet.
56:10 So 1,250 cubic feet is its capacity.
56:13 I could also use weight, you know, volume and weight together.
56:18 I could have multiple combinations.
56:20 I could use number of people if I'm transporting people or number of units.
56:26 So there's a lot of flexibility there.
56:28 And my drivers should only be on duty for nine hours, so I've set the max total travel time, or max total time to 540.
56:37 So that's nine hours.
56:40 And now I'm going to load my customers as orders, because the customers make orders.
56:49 And we'll take a look.
56:50 Most of these customers can make themselves available when I say that I'm going to be there at their house...
56:55 ...but some of them are only available at a certain time during the day.
56:59 So if we take a look at customer 1226, we see that they're only available between 1:00 p.m. and 3:00 p.m.
57:08 So I want the VRP solver to tell me, give me some routes that I...so that I can reach this customer within that time range.
57:16 Also, notice that they have the delivery quantities here.
57:18 This is the size of the appliances that are going to be delivered.
57:22 So this is paired up with the capacities of the routes...
57:25 ...and this is how the VRP solver will make sure that my trucks don't become overloaded.
57:34 Okay. One other thing I want to include here are the breaks for my drivers.
57:38 They're allowed to take a 30-minute lunch break.
57:42 So I've got information about the driver breaks here.
57:47 And notice that I have one break for each truck.
57:51 Open up the properties of the break, and I've given a time window of 11:30 a.m. to 12:30 p.m. ...
57:57 ...and what that means is the break can start sometime between that time range.
58:03 And the service time is 30 minutes so that means that they're allowed 30 minutes for their lunch.
58:10 And I just need to go in here and set a cost attribute.
58:14 And now I'll go ahead and solve the analysis.
58:18 And what it's doing, it's assigning orders to the different vehicles and then it's sequencing those orders in an optimal way...
58:26 ...and it's also giving consideration for those time windows that I had on some of my orders.
58:31 And these are the routes.
58:34 We look at the Network Analyst window, we see that the orders are now categorized by the truck that they're assigned to.
58:42 And this is the same customer I showed you a moment ago where he had the time window of...
58:47 ...1:00 p.m. to 3:00 p.m. when he was available.
58:49 And we'll take a look to see what time it arrives, and we see that it arrives within the time window at 1:39 p.m.
58:58 Now I like to visualize my results a little better...
59:01 ...so what I'm going to use is the new Search window to find the Apply Symbology From Layer tool.
59:11 And what this will do is take some symbology that I've defined for another vehicle routing problem analysis...
59:17 ...assuming that I assign this each day, I...I create new routes for my...
59:23 ...for my trucks, so that what I can do is just assign that symbology to my vehicle routing problem.
59:30 So that symbology is stored on disk here.
59:33 And I'll run the tool.
59:35 Here it's a little easier to cleave out the different routes.
59:39 And we can see they're color coded.
59:42 And what I can do is generate directions just like I did for the routes...
59:46 ...and to print these off or distribute them to my drivers electronically.
59:51 Okay. So the main things that I wanted you to take away from that is that if you need to route several vehicles to visit many locations...
59:58 ...either for inspections, visiting clients, pickups, deliveries, mixed pickup and delivery...
1:00:04 ...the vehicle routing problem will help you create optimal routes for...for them.
1:00:09 And it will help you reduce labor costs, fuel and maintenance costs, as well as carbon emissions.
1:00:16 And time windows will help you maintain a high level of customer service.
1:00:22 And lastly, remember that we have ArcLogistics available, both...
1:00:27 ...there's also an online version that you can use of that and give it a try.
1:00:32 And it's just another application that's based off this vehicle routing problem, and it's geared towards non-GIS professionals.
1:00:40 Give me a switch?
1:00:44 [Audience question] If you had school buses, and you were trying to route the school buses, elementary, this and that...
1:00:49 [Audience question]...and you have [unintelligible] at beginning of the year, would you use this tool to do that?
1:00:53 Yes. So the question is, Can you use VRP to solve school bus routing problems...
1:00:58 ...and I would say with a cautious yes, I guess.
1:01:02 You can do the main parts.
1:01:04 I used to be a school bas-...school bus dispatcher, and you can...
1:01:09 ...you can generate the routes, and you can get good routes quality.
1:01:11 [Inaudible audience question]
1:01:14 Right. But if you want more information, like a lot of these school bus routing programs will have information about students.
1:01:22 So like you have a substitute driver, they'll have a picture and information about the student and...
1:01:26 ...verify they're picking up the right child.
1:01:30 So it wouldn't do that type of stuff out...out of the box, of course.
1:01:33 But we also have a business partner, RouteSmart, and I think they have a...a...a booth downstairs, that they provide...
1:01:41 It is RouteSmart, right? I think...
1:01:43 Do school bus routing?
1:01:44 Yeah. They do the snowplow and garbage truck.
1:01:46 Do they do school bus as well?
1:01:48 Yeah, so they might be able to provide you some more like a specialized product for school bus routing.
1:01:54 Okay. You're welcome.
1:01:57 Our sixth and final solver is the origin-destination cost matrix solver.
1:02:01 And who here has seen the time and distance tables that you see in a route atlas or a road atlas?
1:02:07 So that's basically what the OD solver generates, and we call this the OD solver, the origin-destination cost matrix solver.
1:02:14 It sounds intimidating, but it's not.
1:02:16 Instead, it makes simply a table or a matrix of the cost to travel from one location in a set of locations...
1:02:23 ...to all the other locations in another set of locations.
1:02:27 So you could use this to solve logistical problems like finding all the distances from all the distribution facilities...
1:02:32 ...to each warehouse facility.
1:02:34 You could use it to export out the results to a linear programming package to do things like school-student allocation...
1:02:40 ...which we have a script that does.
1:02:43 And internally, we use it to optimize the stops in...in the VRP solver, the vehicle routing problem solver...
1:02:48 ...as well as the traveling sales problem...salesman problem done by the route solver.
1:02:52 Many of our clients use OD for their own TSP and VRP problems as well.
1:02:57 I've also seen this used to help manage firefighter response times, with concern for the fire stations and evacuation sites.
1:03:03 It also helps to validate historical incident response times.
1:03:07 And notice the lines you see connecting the origins and the destinations.
1:03:11 Like with the location-allocation solver, these lines don't demonstrate that the path found was a straight line.
1:03:17 It's still a network cost path.
1:03:18 The line just shows that there's an association between one stop and the other one.
1:03:27 So here we have a pretty map of the United States as a basemap...
1:03:31 ...and underneath it I have the free Data & Maps data that comes with the Data & Maps DVD.
1:03:37 This is the read-only SDC data as we call it, or StreetMap data.
1:03:41 So I've put in all 50 state capitals into an OD problem and solved it.
1:03:46 And what's generated is a set of 2,500 lines showing associations between 50 state capitals.
1:03:53 Oh, and there's no Honolulu, and Washington, D.C., has been added.
1:03:56 That's how it comes up to 50.
1:03:59 So with this OD solver, you do the solve.
1:04:02 And the real power of OD isn't in these lines; it isn't in symbology that you see there.
1:04:06 It's in this table that's generated...
1:04:08 ...this table of costs from each origin in the 50 origins to every destination in the 50 destinations.
1:04:16 And this can be used for data mining purposes or however you'd use big groups of data like this.
1:04:22 And I also want you to notice that...that the OD solver and the closest facility solver are very similar; they solve similar analysis.
1:04:29 M by n problems we call it, with two sets of inputs.
1:04:32 The difference is OD doesn't generate geometry; it just gives you lines...
1:04:36 ...where the closest facility solver will give you route geometries.
1:04:39 So this solver is much faster because of that.
1:04:46 So if you want to solve big problems, you don't care about the route geometries, then the OD solver is the one to go for.
1:04:51 It also has time cutoffs and number cutoffs if you want.
1:04:55 For example, here's a problem of the 50 origins to 50 destinations where you only find the nearest four destinations from every origin.
1:05:05 Which, by the way, includes itself, because each of these capitals are in both sets of inputs.
1:05:15 Go back here.
1:05:20 So the important thing to think of with the OD solver...yes, sir.
1:05:23 [Audience question] I want to know long it took to generate that [unintelligible]?
1:05:28 It's pretty quick.
1:05:30 I didn't do it because just to save time as we're running here at the end. You can see...
1:05:36 ...[Inaudible audience question]
1:05:37 Oh, yeah. Less than...less than 20 seconds, I would think.
1:05:41 Less than 30 seconds maybe.
1:05:42 So it ran through all the destinations and did a backward search...
1:05:45 ...and now it's going through all the origins and finding the distance to each destination.
1:05:49 So you could see it's tootling along there.
1:05:51 I did it for...
1:05:52 [Audience question] It's using the underlying map for all this.
1:05:55 Yes. It's using its...everything's done along the roads.
1:05:58 This is an entire country worth of streets that it's solving from every origin to every...
1:06:03 ...every destination in...in just a few seconds really.
1:06:08 There it goes. We'll let it finish up.
1:06:11 Three, two, one...yay! There it goes. Okay.
1:06:16 So the important thing, though, is with the...
1:06:17 ...with the closest facility solver, that would've been very difficult because of the route geometries.
1:06:21 And it would have to maintain internally 2,500 tremendously long routes.
1:06:27 So you...for this type of problem, you'd want to use the origin-destination cost matrix solver.
1:06:32 So as a summary of the presentation, I want to just go over the benefits of Network Analyst.
1:06:37 First and foremost is the accuracy of the transportation model, which is the network dataset model that you saw demonstrated.
1:06:43 And if you have a good accurate model, you should be able to solve your transportation problems fairly quickly...
1:06:49 ...and, in doing so, optimizing your routes and saving the company money.
1:06:52 And you can also save the company money by putting facilities in the right place.
1:06:56 I've seen examples of people locating facilities along...
1:06:59 ...where they think they're near a freeway because they have a straight-line distance to the freeway...
1:07:03 ...when really, they're far away from the onramp.
1:07:05 Because what you need to worry about is your transportation network as your distances, and not just straight-line distances.
1:07:12 Also, we showed you Network Analyst on the desktop as well as in ArcGIS Server.
1:07:16 And a little bit of geoprocessing.
1:07:18 But you can do everything we did through geoprocessing and everything you do with Network Analyst through geoprocessing.
1:07:22 There's also out-of-the-box controls if you want to write your own ArcGIS Engine application for Network Analyst.
1:07:28 And there's a few common questions we get, and I should've put school bus routing on this...
1:07:33 ...'cause that's one of the common questions we get as well.
1:07:35 There's something called high-density routing or arc routing, where you're not routing to points; you're routing to entire streets...
1:07:42 ...like with garbage collection or snowplows.
1:07:44 And in that case, Network Analyst doesn't support it out of the box because it's a different type of algorithm.
1:07:50 You don't want your snowplow going back across streets it's already covered, for example.
1:07:53 So you can...you can work it out, but we don't handle it out of the box.
1:07:57 And RouteSmart is one of our business partners that does do that, built upon our software.
1:08:02 We're also asked often if you can get alternative shortest paths.
1:08:06 I found one path; I want to know the next three shortest paths.
1:08:08 And there's definitely ways to do that.
1:08:10 There isn't a solver out of the box, but there's scripts available online that do it.
1:08:14 If you're interested in utility or natural resource networks, as I mentioned in the beginning, what you want is geometric networks.
1:08:20 And that's covered elsewhere in ArcGIS.
1:08:22 And they just ended a presentation, but there's another one, I believe, tomorrow at 3:15.
1:08:26 I have a slide that'll show that.
1:08:28 Transit schedules is another common question that we get.
1:08:31 We do work with multimodal networks...
1:08:32 ...where you can have bus lines connected to streets connected to bike paths.
1:08:37 But we don't work with is transit schedules.
1:08:39 In our case, you would get off your car, get on the bus and go...
1:08:42 ...where a transit schedule would want you to wait for half an hour for the next bus to arrive or something like that.
1:08:49 So now, for the rest of the Users Conference, if you're interested in learning more about Network Analyst...
1:08:54 ...I've handed out little small flyers that have some schedules of when all these tech workshops are and when our demo theaters are.
1:09:01 We highly recommend you come down to the Spatial Analyst Island if you have any questions for us.
1:09:06 There's developers there and more product engineers, and we'd love to talk with our clients and get feedback...
1:09:11 ...our users I guess as we say.
1:09:14 And so the next step you'd want to do is Performing Network Analysis, and that'll be in room 9.
1:09:18 And there Jay and Deelesh will help you choose your solver and tweak the settings to tweak the algorithms that the solvers use.
1:09:27 If you're working with your own data, I highly recommend the Creating Network Datasets session.
1:09:31 Alan will help you walk through the creation and build process.
1:09:34 I believe he does it with TIGER data.
1:09:37 By then, you'll start automating your efforts because you'll be an expert, so you want to do that using geoprocessing in Server.
1:09:42 And Deelesh will walk you through automating your efforts in that way.
1:09:47 There's quite a few demo theaters as well.
1:09:48 Don't worry about getting all this down; they're all written on the little half sheet.
1:09:52 And they're all at the Spatial Analyst Demo Theater, which is right next to our Spatial Analyst Island.
1:09:59 There's a VRP-specific demonstration if that's what you're interested in, and that's...that's at 10:00 a.m., so next.
1:10:05 There's also a real-time routing... He uses VRP as well on Server, if you're interested in that.
1:10:10 And the purple box is on the Server Island.
1:10:13 There's also a location-allocation-specific demo tomorrow afternoon and a 3D-specific demo as well tomorrow...
1:10:19 ...if that's what you're interested in.
1:10:21 Also, I forgot to mention at the beginning, you have two surveys.
1:10:24 Sorry if it's a little confusing.
1:10:25 There's the survey for this presentation, which we really appreciate you filling out and giving us feedback.
1:10:30 And those are the ones you drop in the little box outside.
1:10:32 And there's the survey about Network Analyst.
1:10:34 And I appreciate you filling that out as well.
1:10:36 And you can just leave that on your chairs and I'll collect those at the end, because it's a separate process.
1:10:43 There's some ArcLogistics sessions if that's what you're interested in attending, and they're in the Mezzanine Level.
1:10:48 And those are listed in your book as well.
© Esri 2013 http://www.esri.com