2015 MATC Fall Lecture Series - John Sangster

MATC Author

11/13/2017 Added

11 Plays

Description

New Perspectives on Delay and Level of Service at Intersections and Interchanges

Searchable Transcript

Search:

[00:00:00.585]So, it's on two papers that I submitted
[00:00:03.552]to TRB this summer, for my summers worth of work.
[00:00:07.970]Initially, when Dr. Rolette had asked me
[00:00:09.936]to do a presentation, I thought, oh great,
[00:00:12.417]I can put together a presentation
[00:00:17.319]on the lessons I've learned in the past six months
[00:00:19.715]as I've gone from grad school to academia.
[00:00:23.247]Okay, well, here are all the things
[00:00:24.965]that I didn't realize were important, taking the position
[00:00:27.385]here are all the things that matter,
[00:00:29.224]and ways to be helpful, and I spent about
[00:00:31.987]2 1/2 hours earlier this week trying
[00:00:34.372]to pull stuff together for that and realize
[00:00:36.303]there is so much information to compile that
[00:00:39.663]it wasn't going to happen for the this seminar.
[00:00:41.533]So hopefully, some other time, stop by and talk to me
[00:00:44.385]if you're interested or maybe some other seminar.
[00:00:47.846]Okay, so today I'm gonna talk about
[00:00:49.723]the work that I already had done,
[00:00:52.156]and could easily put together in a presentation.
[00:00:55.789]The title here, New Perspectives on Delay
[00:00:58.178]and Level of Service at Intersections and Interchanges
[00:01:00.545]is the title of one of the papers.
[00:01:02.660]But the content comes from actually two papers
[00:01:04.789]the first, Validating Analysis Methods for Alternative
[00:01:08.909]Intersections using Field Data.
[00:01:11.680]And then this New Perspectives thing kinda grew naturally
[00:01:14.714]out of that as a way to leverage the time
[00:01:17.504]and effort that went into the first one.
[00:01:19.715]This is a situation where I went to my committee
[00:01:23.145]and told them my dissertation topic
[00:01:25.518]and in my proposal explained here what I'm gonna
[00:01:30.417]do for the next six to nine months.
[00:01:32.985]And in passing one of my committee members said,
[00:01:35.678]you've done all this simulation work
[00:01:37.326]and you made lots of advances in our understanding
[00:01:40.963]of the operation alternative intersections and interchanges,
[00:01:44.214]but you have incorporated any field data.
[00:01:47.548]So, I think it would be good if you'd validate
[00:01:50.885]that we can stimulate these designs with field data.
[00:01:53.971]Why didn't you do that?
[00:01:55.795]And my initial thought was,
[00:01:57.881]because that's a different dissertation.
[00:02:01.137]But in order to complete your work and leave,
[00:02:04.890]it's a good idea if you address your committee's comments.
[00:02:08.391]I was fortunate or unfortunate depending
[00:02:12.081]on how you view it, to have some data to work with.
[00:02:16.863]So as I approach this topic of validating simulation
[00:02:20.396]for these intersection deigns,
[00:02:22.971]there's a couple different ways
[00:02:24.286]you can look at that problem.
[00:02:27.303]Possibly fortunate for me, the committee member did not
[00:02:29.278]define what they meant by validating the simulation.
[00:02:35.182]You can either look at the inputs or the outputs.
[00:02:37.882]So for the individual cars within the simulation,
[00:02:41.074]you've got their behavior in terms
[00:02:43.071]of car-following, lane changing, gap acceptance.
[00:02:46.992]So, how close behind another vehicle
[00:02:48.632]do they follow at any given speed?
[00:02:50.468]When do they choose to change lanes?
[00:02:52.380]How random is it changing lanes?
[00:02:54.338]If they're gonna make a left turn up ahead,
[00:02:57.377]how far in advance of that left turn
[00:02:58.844]do they choose to move toward that lane?
[00:03:01.475]It gives you so proportionate loading on the lanes.
[00:03:04.900]The gap acceptance, how large of a gap
[00:03:06.820]if they're making U-turn movement,
[00:03:08.633]which is critical for these designs.
[00:03:10.969]I'm just a making a left turn,
[00:03:12.769]you're waiting for a gap in those cars,
[00:03:15.457]how large does that need to be before you pull out?
[00:03:18.877]These are the inputs that go into the simulation.
[00:03:22.549]But then the output that you're looking at
[00:03:24.797]are the queue lengths.
[00:03:26.071]So is this intersection providing auxiliary lanes
[00:03:30.254]that are long enough for our turning queues?
[00:03:34.419]Ultimately the travel time is the measure
[00:03:37.964]that you're looking at overall.
[00:03:41.187]The average delay, the service measure out of that.
[00:03:45.274]Gets a little complicated with these designs,
[00:03:48.056]I'm gonna get into that some more today.
[00:03:52.482]So this data set that I had
[00:03:56.687]came from a summer internship that I did
[00:03:59.732]with Federal Highway.
[00:04:01.462]I was up in Washington, DC for the summer,
[00:04:04.435]commuting from Blackburn a couple of hours per week.
[00:04:08.982]They were collecting data to update Highway Capacity Manual.
[00:04:12.761]So the next version, the next iteration,
[00:04:14.630]will include a number of different alternative designs
[00:04:18.967]and subsequently those will then get into
[00:04:20.365]all the standard software applications.
[00:04:22.741]These are designs that are very rare right now,
[00:04:26.098]but about become much more common.
[00:04:28.468]You're standard consulting firm will have access
[00:04:31.991]to all of these now and will be on the table pretty soon.
[00:04:37.578]Federal Highway collected data from
[00:04:39.782]four different types of designs.
[00:04:42.080]All told, the Diverging Diamond Interchange
[00:04:44.990]at four different locations you set
[00:04:47.565]up 12 different cameras for each tower,
[00:04:50.503]so every approach, every direction,
[00:04:52.866]we've got concurrent video of everything going on.
[00:04:56.386]Ended up with 256 hours worth of video data
[00:04:59.302]to document those four locations.
[00:05:02.648]There was an awful lot of data at my fingertips.
[00:05:07.253]I had six months to do data reduction by myself
[00:05:12.081]and come up with some check that my simulation is valid,
[00:05:18.717]compared the other observations in the field.
[00:05:22.204]I narrowed it down to looking at one peak hour example
[00:05:27.335]for a restricted crossing u-turn, a displaced left turn
[00:05:29.926]and a Diverging Diamond Interchange,
[00:05:32.243]just to get some variety in there.
[00:05:35.053]That's not much a validation of
[00:05:36.958]all alternatives in motion but a case study,
[00:05:39.327]because that's what you can do with that in time.
[00:05:45.822]Here's an example how the videos could be set up.
[00:05:49.722]We've got the overall interchange location here.
[00:05:52.656]Here's the center of it, you can see the camera's
[00:05:54.769]mounted on a pole.
[00:05:57.608]In this case they were trying to be relatively efficient
[00:06:01.391]since they were flying us all around the country.
[00:06:04.339]To these different locations to get this data,
[00:06:06.823]we couldn't have and ITF ban or something like that
[00:06:11.500]with us so we had a duffle bag of GoPro cameras.
[00:06:15.082]With extra batteries and extra memory sticks.
[00:06:18.799]And then when we landed at each location
[00:06:20.921]would go to the local home depot
[00:06:22.582]and we'd buy these painter poles, 24 foot tall poles.
[00:06:27.464]And we had a mount to attach our cameras
[00:06:30.017]so we had two looking at an angle,
[00:06:31.886]one looking at the center that we'd attach this pole
[00:06:34.661]and we'd duct tape it to a street light
[00:06:36.734]or utility pole in the area to get the angle
[00:06:39.916]that we needed for our videos.
[00:06:42.666]In the midst so we could collect that video data,
[00:06:45.818]all over the country.
[00:06:47.010]Hopping on a plane one week, and then driving
[00:06:48.870]to the next location, hopping on a plane,
[00:06:50.577]going back someplace else the next week.
[00:06:57.614]From that video data we can then extrapolate
[00:07:00.829]what the signal timings are, what the demand flow rates are,
[00:07:06.489]and we can use the aerial images to generate the geometry,
[00:07:10.387]or using the designs and simulation.
[00:07:13.394]This is an example of a Diverging Diamond.
[00:07:16.182]Some call it a Double Crossover Diamond
[00:07:18.894]depending on who you ask and what month it is.
[00:07:21.192]In this case gonna, we've got major arterial
[00:07:24.711]from east-west, you've got a highway on the north-south,
[00:07:28.660]So, you've got a great separation here,
[00:07:31.705]I'm not sure on the highway cars,
[00:07:32.838]but there'd be significantly more running up
[00:07:35.446]and down the highway than on the arterial.
[00:07:37.509]The main arterial is the east-west passage here.
[00:07:41.197]And as we approach the interchange we're crossing over
[00:07:45.423]our traffic on the opposing side of the road
[00:07:49.993]at this point.
[00:07:51.229]So we let the right turners go free,
[00:07:53.688]we crossed the left turners and the through traffic over
[00:07:56.475]we then let the left turners go free on the other side
[00:07:58.998]and we crossed the through traffic back to the regular
[00:08:01.810]side of the road and they get across the other side.
[00:08:05.895]The main advantage of this design is
[00:08:07.728]it let's your left turners go free
[00:08:09.799]and it doesn't hold them at that signal,
[00:08:11.442]it doesn't build up queues across the bridge.
[00:08:13.860]This is a fairly large bridge, this was a new construction
[00:08:17.267]that Utah put in place.
[00:08:19.741]A lot of places that this has been utilized in the past
[00:08:23.794]has been at existing locations where demand flows
[00:08:27.296]are getting too high and the queues
[00:08:29.374]of left turns across the bridge
[00:08:31.113]are requiring another lane beyond
[00:08:33.113]what the bridge can sustain.
[00:08:35.830]So by converting this design at an exiting location,
[00:08:39.499]you can get more throughput without replacing
[00:08:42.667]a bridge, if the bridge structure itself
[00:08:44.524]has some life left in it.
[00:08:47.958]This is a partial displaced left turn.
[00:08:50.838]This is in Loveland, Colorado.
[00:08:54.958]In this case it's the minor road that
[00:08:56.455]they've done modifications to,
[00:08:58.212]and upstream of the signal on both ends,
[00:09:01.362]they take the left turn vehicles from the main intersection,
[00:09:04.794]and they cross them over to the opposing side
[00:09:07.635]of the through traffic from the opposite direction.
[00:09:11.321]The advantage of this is that your crews
[00:09:15.261]can go concurrently in both directions,
[00:09:17.178]and your left turns can go with them.
[00:09:21.831]So you end up with only one phase necessary
[00:09:24.176]for the north-south direction,
[00:09:26.215]so the other two phases left turns and throughs.
[00:09:29.253]This works well when you're going with left turning traffic
[00:09:32.945]per lane is approximately equal to
[00:09:35.106]the volume of through traffic.
[00:09:37.101]So the amount of time from the signal
[00:09:39.152]that they're demanding is about equal
[00:09:41.397]and then you can overlap those times.
[00:09:43.850]In in this case, the the main arterial
[00:09:45.604]on the east-west, the left turning traffic
[00:09:48.420]was much lower than the through traffic
[00:09:50.285]on that arterial so wouldn't have made sense
[00:09:52.801]to displace those left.
[00:09:58.023]Final example is the restricted crossing
[00:09:59.742]u-turn I looked at.
[00:10:01.183]This is a case where you're prohibiting through
[00:10:04.480]and left turn movements from the side street.
[00:10:07.474]You're forcing all that side street traffic
[00:10:09.296]to take a right and then use a u-turn bay
[00:10:11.394]that's upstream of the signal to come back around
[00:10:14.964]and complete the movement.
[00:10:18.004]So we're allowing on the main arterial here
[00:10:19.951]left's, throughs and rights.
[00:10:22.344]And then from the minor arterial,
[00:10:23.800]everyone has to take a right.
[00:10:25.533]And then if you have intended to go through
[00:10:26.930]or left use the u-turn bay up there to come back around.
[00:10:32.405]This is an excellent solution when you
[00:10:34.632]have a major arterial that's a critical roadway.
[00:10:37.911]You have to maintain the throughput on
[00:10:39.405]for your morning commute, but you want
[00:10:42.025]to provide access to those minor roads
[00:10:43.983]on the side, you want people to be able
[00:10:46.346]to get to that shopping center without necessarily
[00:10:50.068]a large concern for the travel time
[00:10:52.441]of vehicles coming from those side streets.
[00:10:58.807]So, this is what the video data looks like.
[00:11:03.003]Eight terabytes.
[00:11:06.176]About 10 days worth of video data.
[00:11:09.846]At this location we would have had eight cameras
[00:11:12.869]getting all the different angles.
[00:11:15.671]You can see the lights here,
[00:11:18.629]you can see the vehicles coming through.
[00:11:21.736]In order to line up all the videos
[00:11:25.533]in terms of chronological time, we would watch
[00:11:28.665]the cars coming through and we'd stop on the car
[00:11:33.166]we've seen in two different videos at once
[00:11:35.750]and then we'd figure out the offset
[00:11:37.477]between the start time of the two different videos.
[00:11:39.729]It was kind of done with a shotgun approach
[00:11:42.986]where there wasn't a methodology in place
[00:11:45.210]to determine the time, the actual time
[00:11:49.436]that each video started at, so we had to assume
[00:11:53.166]in time for one video and calculate everything else
[00:11:55.679]relative to that to line up all the other video feeds.
[00:12:00.477]If we'd had more time and had thought through the process,
[00:12:04.803]I would have liked to have a large digital clock
[00:12:08.023]that we held up in front of the camera
[00:12:10.063]when we started each videos so we had something
[00:12:12.004]that at least tied it back to.
[00:12:15.564]So, strictly from a perspective of getting the work done
[00:12:21.244]I had lots and lots of these videos
[00:12:23.646]and I had very little time to reduce the data
[00:12:26.856]and the process I needed to go through
[00:12:28.925]was to identify when the cars were coming
[00:12:32.496]into the network and when they were leaving the network
[00:12:35.125]in each given direction to develop travel paths.
[00:12:38.216]So I had to track each individual car,
[00:12:42.505]the time it entered, and the time it got to the exit.
[00:12:47.896]Fortunately, when I was in bin,
[00:12:49.762]the resolution on my screen is better
[00:12:51.773]than you're seeing here.
[00:12:54.410]After looking at this video,
[00:13:04.488]a section of the video up here with this light
[00:13:09.930]is this part here.
[00:13:12.653]So, at higher resolution you can see the vehicles
[00:13:16.160]queuing up on this side and you can see
[00:13:18.753]when they enter the network back here
[00:13:20.542]and you can see as the vehicles go out this way,
[00:13:22.361]this point is kind of where they leave the network
[00:13:26.593]and are beyond the limits of the impact of the intersection.
[00:13:31.076]So if you look at the aerial view
[00:13:32.985]that corresponds approximately to this spot.
[00:13:37.852]So for each eastbound lap based on through eastbound light,
[00:13:42.033]I had to define a location that I could see
[00:13:43.652]on the videos that was outside of the limits
[00:13:46.944]of the intersection where I could start measuring
[00:13:49.269]the entry and exit points. (mumbling)
[00:13:59.461]If you look at this Diverging Diamond intersection,
[00:14:02.415]example the purple lines all around the diagram
[00:14:07.032]show you where the video locations were
[00:14:10.161]that I was recording.
[00:14:11.641]So, for northbound left's, I would pick up the vehicle here.
[00:14:16.132]I would then track it through this signal
[00:14:19.740]and write down the time that it exited there.
[00:14:23.586]There's enough back and forth
[00:14:26.163]of vehicles exchanging positions
[00:14:29.034]that in order to track the vehicle that enters here
[00:14:33.820]to here, I ended up having to record the time across
[00:14:36.184]this stop bar and the time across that stop bar,
[00:14:40.412]and then track it through, I'd call it
[00:14:43.273]like a red sedan or a blue pickup or something like that.
[00:14:48.089]Because resolution wasn't good enough
[00:14:49.558]to see license plates anything like that.
[00:14:53.679]Gave me a little bit of difficulty though,
[00:14:56.969]because my ultimate goal is to compare the travel time
[00:15:01.257]of each approach, the northbound left,
[00:15:04.889]the northbound right, the southbound left,
[00:15:07.374]eastbound through.
[00:15:08.862]Compare that travel time on video
[00:15:10.586]against my simulation travel times.
[00:15:13.709]So I've got all these start and stop ones
[00:15:15.566]that aren't a uniform distance and I want
[00:15:18.525]to make them uniform distance so that I can
[00:15:20.395]have an approach that's
[00:15:24.696]consistent going into my research.
[00:15:31.437]The measure that I'm trying to validate
[00:15:35.232]is the control delay.
[00:15:39.943]Within the Highway Capacity Manual
[00:15:42.068]this is the ultimate sort of measure that we're looking at.
[00:15:45.391]Everything is based on this.
[00:15:47.581]If you're in consulting and you need
[00:15:49.361]to do intersection improvements
[00:15:50.773]because there's a new development
[00:15:52.498]they're going to say you need to have a control delay
[00:15:56.358]that's less than 40 seconds, once you're finished
[00:15:59.608]with your improvement, something like that.
[00:16:01.254]So, everything gets tied back to this control delay measure.
[00:16:05.724]It's defined in the Highway Capacity Manual
[00:16:08.016]as the delay brought about are the presence
[00:16:10.364]of a traffic control device.
[00:16:13.105]So whether it's stop sign, a signal, a roundabout,
[00:16:18.342]some sort of device that the controlling traffic
[00:16:20.671]it's causing delay.
[00:16:22.888]If you're looking at a simulation perspective of delay,
[00:16:25.948]you look at the desired free flow speed of the vehicle
[00:16:29.286]at a given time and you compare that with
[00:16:31.322]the actual speeds are traveling at that time
[00:16:33.905]and then you add up all the incremental, you know,
[00:16:36.260]I'm traveling this much slower than
[00:16:38.008]I want to for this long.
[00:16:40.076]Add up that extra time that the vehicle
[00:16:42.051]takes passing through the location
[00:16:44.781]and you gather the control delay for the vehicles.
[00:16:48.582]It includes the slowing down of the approach.
[00:16:52.481]The location includes the stopped time there,
[00:16:54.710]includes the speeding up as you exit,
[00:16:56.970]it's from before you're impacted by the intersection
[00:16:59.694]to after you're beyond its limits.
[00:17:02.552]Unfortunately, it explicitly does not include
[00:17:07.810]the additional travel time caused by geometrics.
[00:17:13.902]This poses a serious issue when we're now considering
[00:17:17.917]alternative intersections where our vehicles
[00:17:21.991]have a much longer travel path that they're going through
[00:17:26.475]to get to the same location.
[00:17:29.390]So, if you're going a comparative analysis
[00:17:31.891]between a conventional option of this location
[00:17:34.943]where we've got a signal in the middle
[00:17:36.249]and every would be the left and like whatever they want
[00:17:39.188]which is this condition, if you're using control delay
[00:17:43.173]as your service measure, there's no way
[00:17:45.436]to build in this additional travel time
[00:17:47.598]coming back around here.
[00:17:50.212]You'd have your northbound left
[00:17:51.278]at your conventional going this way
[00:17:53.432]and your northbound left u-turn going this way
[00:17:58.055]and the measure you'd be comparing them against
[00:18:02.290]is how much they were slowed down compared
[00:18:05.881]to what they wanted to go.
[00:18:08.267]It doesn't build in that extra travel time.
[00:18:11.166]So I needed to fix that for my research.
[00:18:26.451]I developed method using constant radius
[00:18:30.587]around the circle that basically says,
[00:18:34.026]here's how we help with the control delay
[00:18:36.923]and I'm going to go down the right side
[00:18:38.634]and I'm going to come up with a measure
[00:18:40.615]for geometric delay that I can apply consistently
[00:18:44.023]to every intersection design, whether it's a roundabout
[00:18:46.787]or an interchange or restricted crossing u-turn
[00:18:51.429]this methodology can be applied equally
[00:18:54.853]to give us a measure of junction delay
[00:18:57.844]that's equal to the control delay
[00:18:59.414]plus the geometric component of that delay
[00:19:02.307]or where you leave out.
[00:19:04.839]What it comes down to is looking at the path
[00:19:09.367]those people traveled and then comparing that against
[00:19:13.065]a theoretical node to node distance.
[00:19:17.738]So if you're going from your entry point
[00:19:20.121]to the centroid of the intersection and then straight out
[00:19:22.704]to your exit point, so you're doing, basically,
[00:19:25.740]the diameter of the circle for your distance.
[00:19:28.768]Every vehicle in terms of their desired path
[00:19:32.073]then has a distance equal to the diameter of the circle.
[00:19:36.688]If you have your speed limits and location
[00:19:39.113]or your vehicle speed almost approaches
[00:19:41.929]you can then directly calculate a base travel time
[00:19:45.489]that's uniform for every design.
[00:19:48.402]Whether it's an interchange at that location
[00:19:51.102]or whether it's an intersection or anything in between
[00:19:54.645]if you go from your entry point to the centroid
[00:19:57.294]and straight up to your exit point,
[00:20:00.039]every vehicle can have the same base travel time
[00:20:03.113]to compare things against.
[00:20:06.213]A problem in comparative analysis right now
[00:20:09.920]is that there's no base condition defined.
[00:20:13.387]You have your existing condition which is
[00:20:15.100]your signalized intersection, whatever it is,
[00:20:17.144]and you're comparing overall control delay,
[00:20:20.821]one against the other.
[00:20:22.873]But if you have additional travel time
[00:20:24.461]you can't then tie it back.
[00:20:26.625]So this method comes up with a way to
[00:20:28.869]give that base travel time and then build in
[00:20:32.372]the geometric travel on top of that.
[00:20:37.151]The grayed out, it's kind of hard to see
[00:20:38.678]on this screen here, but the grayed out portion
[00:20:42.436]is just the additional steps that were added,
[00:20:45.896]based on the videos having entry and exit points
[00:20:49.666]that didn't match up with where this radius touched in.
[00:20:53.242]So, if the exit point over here is back here
[00:20:57.190]or maybe here is where I'm measuring,
[00:21:00.658]the grayed-out boxes here give you a method
[00:21:04.634]for adjusting those appropriately so that
[00:21:07.454]you can have the same base.
[00:21:15.394]When we look at how the data works out
[00:21:18.302]for an example like this restricted crossing u-turn.
[00:21:22.282]Using the actual field data collection, demand volumes,
[00:21:27.123]the signal timings from the field,
[00:21:29.455]how its operating out there,
[00:21:32.234]we see that the geometric delay component,
[00:21:34.684]for example, the northbound left and through
[00:21:37.591]there's an extra 18 seconds here and 16 seconds here
[00:21:41.927]that's added to their travel time by making that movement.
[00:21:46.304]But when you look at just the radius that's added
[00:21:48.800]on the right turn it gets a bonus of 1.2 seconds
[00:21:53.773]that it's cut down on the travel time
[00:21:56.049]by just channelizing that right, it'd be offsetting
[00:21:59.333]the lanes a couple from the centroid
[00:22:01.453]and it's just traveling, just that much less distance
[00:22:04.583]to make that northbound right.
[00:22:06.778]This then gets applied through to the eastbound left
[00:22:12.090]and right, and westbound left and right.
[00:22:14.332]East and west through, kind of wash.
[00:22:20.179]And you see in the end where you've got 16
[00:22:23.115]or 17 extra seconds for making these side road left
[00:22:28.734]and throughs the overall impact in terms of geometric delay
[00:22:33.255]on your intersection is less than a second overall
[00:22:37.059]because you've got these minor bonuses
[00:22:38.721]to much larger flow rates and you've got
[00:22:42.119]these major impacts, major negative impacts,
[00:22:44.621]to much lower volume travel base.
[00:22:49.097]So we're disadvantaging the minor street movements
[00:22:51.331]to give an advantage to major Street.
[00:22:53.607]And overall in terms of geometric delay alone
[00:22:57.212]it's nearly a wash.
[00:23:09.931]Before my validation study for my committee member
[00:23:14.321]I looked at the observed field data,
[00:23:16.635]I looked at simulation using VISSIM,
[00:23:20.310]simulation using integration and then for fun
[00:23:23.998]I threw in the Highway Capacity Manual method
[00:23:28.398]with some adjustments to account
[00:23:29.658]for the extra travel time measures and designs.
[00:23:35.056]I was able to show that for the restricted crossing u-turn
[00:23:38.239]all the simulation is valid with the displaced left turn.
[00:23:42.294]I had some issues with progression on
[00:23:44.668]the integration simulation so the DDI were all valid.
[00:23:49.527]Now, one thing that stood out to me is that the DDI
[00:23:52.889]had valid results using the Highway Capacity Manual method
[00:23:55.767]from the 2000 version.
[00:23:59.999]It was not able to predict how things would go
[00:24:02.794]with the accurate intersections.
[00:24:08.685]But it's a condition dependent situation.
[00:24:18.121]So when I finished with my validation study,
[00:24:21.200]I'd put this time in and had this data,
[00:24:23.762]and I started thinking about OK, was that control delay now?
[00:24:27.891]Did that junction delay that I'm proposing
[00:24:30.646]in the existing control delay how does this work out
[00:24:34.081]to comparative analysis?
[00:24:37.012]If we've got an existing intersection that wants to
[00:24:40.953]be grade separated to maintain a bubble of service
[00:24:44.699]and then we've got all these alternative intersections
[00:24:46.718]that are potential solutions to extend the life
[00:24:49.298]of that location without going to an interchange solution,
[00:24:54.343]What are the levels of service results that
[00:24:58.804]get tied back to and I notice that we
[00:25:01.824]could stop-controlled intersection
[00:25:04.936]and the signalized intersection are very similar
[00:25:07.610]in terms of how controllable it maps back to a grade.
[00:25:12.306]This is the, the grades are basically
[00:25:14.391]what you go to the client with.
[00:25:17.271]You go to your municipality, you go to the town meeting
[00:25:20.100]and you tell them it's a level service D right now.
[00:25:23.171]We're going to make improvements it's going to
[00:25:25.198]be level service C when we finish.
[00:25:27.220]Well the amount of control delay
[00:25:30.365]at the different types of control
[00:25:34.880]result in different grades.
[00:25:36.735]So if you have 30 seconds of delay at your location
[00:25:40.855]well if that's stop-controlled intersection that's a D
[00:25:44.595]but if it's a signalized intersection it's a C
[00:25:47.942]and if it's an interchange it's a B
[00:25:51.571]for the same amount of time that you're delayed.
[00:25:56.127]The methodology that was used to determine this
[00:25:58.930]as far as I know was a group of engineers
[00:26:01.615]sitting around in a room saying, you know,
[00:26:04.822]when people come up to an interchange
[00:26:06.723]they're kind of expecting it's going
[00:26:08.144]to take longer to get through.
[00:26:09.977]So we need to modify this interchange
[00:26:14.522]to be higher amounts of delay compared to an intersection.
[00:26:21.137]And when you signalize the intersection itself
[00:26:23.781]if you've got an equal probability of people
[00:26:27.396]arriving, some people can arrive on red,
[00:26:29.426]you're going to have this base value of delay
[00:26:32.281]that you didn't have at a stop-controlled intersection
[00:26:35.223]had people just wait for the light to change.
[00:26:37.495]So let's let's increase that relative to the initial
[00:26:40.838]because people are going to expect to wait
[00:26:43.113]a little bit longer at the stop at
[00:26:45.787]the signalized intersection than they did at the stop sign.
[00:26:48.144]And they're going to be okay with waiting that much longer.
[00:26:50.679]Their feelings are gonna be they're gonna feel like
[00:26:54.040]this is level of service C, even though
[00:26:56.524]they're waiting a little longer.
[00:26:58.837]There's a little bit of research on this
[00:27:01.717]but it's not extensive and it's, for the most part,
[00:27:06.120]a committee sitting in a room saying
[00:27:08.453]we think the signalized interchange should be 1.5 times
[00:27:10.958]the signalized intersection.
[00:27:14.103]And they voted on it and that's how
[00:27:15.880]this regime compares to this regime.
[00:27:19.226]How they came up with those numbers.
[00:27:23.166]When roundabouts came along, they said,
[00:27:26.340]okay roundabout that's an unsignalized intersection,
[00:27:29.967]so we're gonna give that the same values
[00:27:32.939]as over here, let me get a stop sign or roundabout
[00:27:36.702]there's no signal there, people are expecting to wait
[00:27:38.888]about the same amount.
[00:27:40.354]Hence our method is to multiply by 1.5 for an interchange,
[00:27:44.507]compared to a stop sign let's just take this roundabout
[00:27:46.955]and when we have an interchange with roundabouts
[00:27:50.014]at the ramps we're gonna multiply that by 1.5 times
[00:27:54.316]what the roundabout was.
[00:27:55.874]So now we have this new column for level of service
[00:27:59.321]they're in control delay against, roundabout interchange.
[00:28:05.791]The research that needed to be done to validate
[00:28:10.142]how much delay equals what level of service
[00:28:13.967]is hard to define and hard to fund in terms of
[00:28:19.289]the benefit to the funding agency.
[00:28:24.648]So if you're going to the Department of Roads
[00:28:27.163]in Nebraska for instance, and you say
[00:28:29.745]this isn't legitimate I'd like to fund research
[00:28:32.295]where I looked at how people feel about waiting
[00:28:35.521]at an interchange versus waiting an intersection.
[00:28:37.998]What's the level service be with the level of service B?
[00:28:40.976]The Department of Roads is going to look at you
[00:28:43.127]like you have three eyes or two noses and say,
[00:28:46.083]we have a budget and we have lots of problems
[00:28:49.451]that are costing us money that we're trying to fix.
[00:28:52.163]we can't afford to pay you to look into this.
[00:28:56.226]So we're stuck with this situation (audio dropping, garbled)
[00:29:05.035]A small benefit to this is that I can sit in my field
[00:29:09.173]and I can say, well, I don't think this is right.
[00:29:11.160]I an quote my own thing without validating it with research.
[00:29:17.183]So, I did.
[00:29:19.303]Coming up with the control delay transition
[00:29:21.410]to junction delay, as we look at these
[00:29:23.304]new intersection designs.
[00:29:26.539]They're not interchanges, they're not grade separated,
[00:29:29.774]but they're also not single intersections.
[00:29:31.747]There's got multiple signalized points that vehicles
[00:29:35.059]are traveling through, it's kind of one at-grade interchange
[00:29:39.252]that these new designs fit into, and they don't fit
[00:29:42.424]in any of the existing categories.
[00:29:44.587]I don't want to add another column on that
[00:29:47.382]you know, long term bubble service
[00:29:49.029]and say alternative intersections,
[00:29:51.517]they're kind of halfway between an intersection
[00:29:53.645]and an interchange so we're gonna multiply by 1.25
[00:29:57.433]and let's vote on it.
[00:29:59.674]It's not a long term solution,
[00:30:04.371]especially when the problems we're facing
[00:30:06.957]are deciding what geometries to build in the future.
[00:30:12.017]If our intersection and demand volumes are increasing
[00:30:14.734]we need to make improvements but now it's a signal,
[00:30:17.840]We might keep it as a larger signal,
[00:30:19.672]we might go to an interchange,
[00:30:21.484]we might go to an alternative intersection solution.
[00:30:25.899]The level of service shouldn't be based on
[00:30:29.117]what geometry will bring in there.
[00:30:30.825]It should be based on instead lane closing,
[00:30:33.408]the total demand at that location.
[00:30:37.978]So if you're at a low volume situation,
[00:30:41.651]the total demand is going to determine
[00:30:43.914]how people perceive that intersection.
[00:30:46.917]If you've got only a few cars going through
[00:30:48.804]in your peak hour then people aren't going to circulate
[00:30:52.455]they don't care if there's a roundabout there or a signal
[00:30:54.634]or a stop sign, they're not gonna expect to wait.
[00:30:58.368]If you're at the larger signalized intersection
[00:31:01.945]volume threshold where the one-way roundabout
[00:31:04.556]doesn't work anymore, the stop sign doesn't work anymore,
[00:31:06.967]you're definitely warranted to have a signal there
[00:31:10.860]with its weight remains and things like that.
[00:31:12.911]But you're not yet to the point where you need
[00:31:15.356]a grade separated interchange to solve your traffic.
[00:31:18.358]There's an expectation there compared to
[00:31:20.609]the initial intersections that you're going to wait longer.
[00:31:23.948]There's cars around you have to wait for the cars
[00:31:25.923]in front of you to move either way is gonna
[00:31:28.056]be higher and you're not going to get mad
[00:31:30.425]at your city traffic engineer for not fixing
[00:31:33.845]the problem because your expectation is that
[00:31:36.287]you're going to have some delay there.
[00:31:40.090]If you get a high enough volume
[00:31:41.483]that you're now dealing with an interchange situation
[00:31:44.126]where you're taking ramps off to one side
[00:31:47.471]and coming back through and you've got multiple signals
[00:31:49.799]that you're going through, there's an expectation there
[00:31:52.486]that you're going to take some time
[00:31:54.155]to get through that geometry.
[00:31:58.949]I'd like to look at that straight across
[00:32:03.089]if we're looking at these different designs
[00:32:05.484]and not just act okay when we go to this
[00:32:08.108]the interchange we're allowed to have more delay.
[00:32:11.351]But OK, what's the delay value of
[00:32:14.367]all these different options whether we make
[00:32:16.967]a giant, accurate intersection
[00:32:19.218]or an interchange, but maybe we need fewer lanes
[00:32:22.212]but logical geometry or these alternatives in between.
[00:32:25.907]Let's compare apples to apples in terms
[00:32:28.244]of our level of service and say this one
[00:32:30.957]has 30 second delay and this one has 30 seconds delay,
[00:32:34.137]and this one has 30 seconds delay so those are all C.
[00:32:37.923]They are not B, C, and D depending on what we build,
[00:32:40.420]they're all the same.
[00:32:45.906]I was a little nervous presenting this
[00:32:47.143]to my committee because it's something
[00:32:50.572]where I just threw it on paper and said
[00:32:53.139]this is what I think and it's not, you know,
[00:32:57.114]a survey of drivers as they're going
[00:32:59.609]through these intersections and experiencing
[00:33:01.271]this delay and how they felt about it.
[00:33:03.099]It's not tied back to anything that I can document.
[00:33:07.953]It's just, here's an opinion and I presented my research
[00:33:12.652]and my committee all patted me on the back
[00:33:14.645]and said with a great job I'd done and that I
[00:33:16.838]was finally an academic.
[00:33:19.111]And I thought, what?
[00:33:21.732]So apparently, sometimes, you just need to go with it
[00:33:31.501]and that's what I have for today.

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Tags:
ntc
matc
nebraska transportation center
mid-america transportation center
John Sangster
civil engineering
transportation engineering
unl
university of nebraska-lincoln

Comments

0 Comments

2015 MATC Fall Lecture Series - John Sangster

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments