Carlson, Kristin, Murphy, Brendan, Owen, Andrew, Ermagun, Alireza, and Levinson, D. (2019) Safety in Numbers for Bicyclists at Urban Intersections Transportation Research Record. [doi]
Abstract: This study assesses the estimated crashes per bicyclist and per vehicle as a function of bicyclist and vehicle traffic and tests whether greater traffic reduces the per-vehicle crash rate, a phenomenon referred to as “safety in numbers” (SIN). We present a framework for comprehensive bicyclist risk assessment modeling, using estimated bicyclist flow per intersection, observed vehicle flow, and crash records. Testing a two-part model of crashes, we reveal that both the average of annual average daily traffic (AADT) over a 14-year period and the estimated daily bicyclist traffic (DBT) have a diminishing return to scale in crashes. This accentuates the positive role of SIN. Higher volumes of vehicles and cyclists lowers not only the probability of crashes, but the number of crashes as well. Measuring the elasticity of the variables, it is found that a 1% increase in the average of AADT across the time window increases the probability of crashes by 0.14% and the number of crashes by 0.80%. However, a 1% increase in the estimated DBT increases the probability of crashes by 0.09% and the number of crashes by 0.50%.
Sydney opened the long-awaited first Northwest section of its “Metro” line. Sydney has long had grade-separated, high-frequency train service (Sydney Trains) through its core, the “Metro” is different in that it is:
single-deck rather than double deck, with more doors, for faster boarding times
standing rather than sitting oriented (on a crowded train more standees than seated passengers, compared with Trains)
automated rather than manually driven
with platform-based as well as train-based doors, to improve safety.
In other words, while Sydney Trains is what Americans would think of as commuter rail, but on steroids, Sydney Metro is like late 20th century (early 21st century) trains built in much of the developed world, most similar to systems like BART, DC Metro, or MARTA.
To get to the Metro, we took Sydney Trains from Redfern to Epping. At Epping, one descends and descends to reach the Metro platform. The stations and controls from Epping to Chatswood were remodeled from the early 21st century trains line (when the corridor was expected to be a Trains rather than a Metro. We took the line west to Tallawong (a parking lot and near the train stabling facility), and alighted and boarded the eastbound train which we took to Rouse Hill, where we alighted for lunch, making a series of culinary choice errors at the Rouse Hill food court, though I am not clear one could do otherwise.
The good news is that demand was high (75,000 in five hours, the Sydney Morning Herald gushes), apparently exceeding expectations. People are curious about the line, want to see it succeed, want to be able to use public transport to reach the city. Even before the problems that I will soon describe emerged, it was Standing Room Only on the westbound run.
The trains had indicators showing where they were on the line. There was an emergency stop button located near the doors which look like a User Interface disaster waiting to happen (that is, there will be an enormous number of false positives as people will push the button accidentally or in the believe it is required to open the door, as in an elevator).
The braking sound of the train is very much like DC Metro, though deceleration did not induce the same kind of nausea that DC Metro does. There is nevertheless a significant uncomfortable jerk as the Metro train comes to a stop at many of the stops.
After thoroughly exploring the Rouse Hill Town Centre, we queued up to board the Metro back, to go to Chatswood, and then transfer to a Train back to Redfern.
The bad news is the service operator (MTR) was not quite ready to provide a reliable service. We may eventually discover whether someone(s) specifically screwed up, or whether failure is indeed an orphan. Apparently (I did not witness this, but people report) there were issues with platform and train doors aligning, and issues with doors closing properly and with trains overshooting the platform. This held up trains Chatswood and Macquarie Park, and thus eventually all the trains in the line, as shockwave of stoppage cascades backwards all the way to Tallawong.
It took 1 hour and 40 minutes from Rouse Hill to Chatswood. The first 40 minutes were queueing at Rouse Hill, so as not to overload the platform for the few trains making it through, it was no 90 second, or 4 minute, or 5 minute headway as variously promised by various people at various times. The remaining hour was on train from Rouse Hill to Chatswood. The scheduled time is 35 minutes station-to-station.
Epping Metro, as the train to Tallawong approachesNo up escalators at the spacious Rouse Hill Metro Station
This opening debacle will, as first impressions are important, likely create a perception that the service is unreliable. If this is coupled with a few well-publicised rush hour breakdowns, it will take years to fully regain a reputation for reliability, and people will clamor for restoration of more express bus services. Obviously some of this technology problem is teething issues, and will be eventually sorted out, but surely this should have been worked out in testing … unless it was rushed for some reason.
The queue management was professional if indicative of problems. The communications with customers about the problems was vague.
Now, to be fair, opening day often brings about unexpected outcomes.
The opening of the Green Line light rail between Minneapolis and St. Paul was marked by an automobile wrongly driving on, and getting stuck on, tracks; and the train hit multiple pedestrians in its first year.
The Opening of the Liverpool and Manchester railway killed a prominent Member of Parliament. So the delays on the Sydney Metro are perhaps small potatoes in the scheme of things. One just would have hoped for a better performance.
* I am not commenting on the strategic decisions about the location of the line, etc. here.
This article, by Somwrita Sarkar, Hao Wu, and David Levinson first appeared in The Conversation.
The Greater Sydney Commission has proposed a 40-year vision of a metropolitan region formed of three “cities”: the Eastern “Harbour” City, the Central “River” City, and the Western “Parkland” City. The plan aims to create 30-minute cities, where the community has access to jobs and services in three largely self-contained but connected regions. Thus, Sydney would be polycentric.
A polycentric city has multiple centres of employment, economic or social activity. Local labour markets and residential zones minimise long commutes, create a sense of place and neighbourhood, and strengthen economic agglomeration as companies, services and industries benefit from being close to one another.
However, it is still unclear whether Sydney is actually moving towards such a structure. In our recent work, we developed new ways of measuring polycentricity. We applied these to Journey to Work data from the 2016 Census to test how consistent the current centricity patterns of Greater Sydney are with the proposed plan.
How do you measure polycentricity?
Traditionally, employment densities are used as a measure of polycentricity. If the density of jobs in a location is higher than the average density for the entire region, then it is a centre.
However, this simple measure misses a key notion that makes cities what they are: network flows and spatial interactions. People “flow” from one place to another. Employment centres “attract” flows, and residential areas “produce” flows. Thus, a city is a collection of locations that interact dynamically, connected by daily commuting flows.
We proposed a set of new metrics to capture this idea of flows. We defined the net inflow of people to a location as the total number coming to this location to work minus the total number going from this location to work elsewhere. If the net inflows are positive, this place is a centre.
The chart below illustrates the idea. The base arc on the circle shows the number of people “flowing” out of a location to another location. The connecting arcs are coloured black if the net inflows into the focus regions (a), (b) or (c) are positive.
Testing polycentricity via net inflows: (a) Sydney City and Inner South (Sydney CBD), (b) Parramatta, (c) Eastern Suburbs.Source: The authors
Sydney CBD clearly emerges as a global centre for the whole region. Parramatta is a regional centre. Other locations such as the Eastern Suburbs are not centres at all.
The net inflow to a location can be divided by the total number of trips in the system, so inflow values are scaled from 0 to 1 using a standard statistical procedure. The higher the value, the higher the centre’s rank in the urban system. Here, a score of 1 means the centre is an absolute: all the trips in the system are a net inflow into the centre.
This gives us a trip-based centricity measure. And based on the area of the location, we can calculate a density-based centricity measure.
The maps below show trip-and-density-based measures – (a) and (b) respectively – for Greater Sydney at the Statistical Area Level 2 (representing a community that interacts together socially and economically).
Note the dominant role of the Sydney CBD. The other centres emerge as weak centres. Also, many of the second-order centres are very close to the CBD.
Visualising polycentricity in Sydney (red indicates highest values): (a) trip-based centricity, (b) density-based centricity, (c) transit-accessibility-based centricity, and (d) auto-accessibility-based centricity.Source: The Authors
The concept of accessibility
Counting the net inflow into a location may provide us with information about general centricity. However, it still does not tell us how easy or difficult it is for people to actually get to jobs. This brings us to the idea of accessibility.
Walter Hansen defined accessibility as “the spatial distribution of activities about a point, adjusted for the ability and the desire of people … to overcome spatial separation”. More practically speaking, a location is accessible if it can be reached within a set time (say 30 minutes) from another location.
We counted the net accessibility of a location by counting the number of jobs minus the number of workers (labour) that could be accessed from a particular location (SA2) in Sydney within 30 minutes. We counted travel time both by car and by public transport during a usual weekday peak hour (Wednesday 8am). Similar to the trip and density measures, accessibility centricities can also be scaled as values between 0 and 1. This allows us to compare across the four measures.
In the maps above, (c) and (d) show the transit and auto-based accessibility centricities based on accessibility for public transport and vehicles. Sydney CBD is highly accessible. The second-order centres show much weaker accessibility.
Takeaways for urban policy and the three-cities plan
The chart below shows the top-ranked centre, Sydney CBD (Level 1 centre), and the lower-ranked subcentres (Levels 2 and 3) emerging from our analysis.
Identified Level 1, 2, and 3 Centres for the Greater Sydney metropolitan region.Source: Authors
Accessibility planning should guide the design of a polycentric city
The design of polycentric Sydney should be guided by accessibility, the locations of jobs and homes, and subregional labour market organisation.
In short, the region should give priority to making jobs accessible by locating new jobs in emerging centres, instead of a mobility-focused system that takes people to jobs.
Reduce spatial mismatches between jobs and homes
Our results show that Sydney, paradoxically, remains strongly monocentric and strongly dispersed at the same time. The Sydney CBD accounts for 15% of jobs in the region, with the remaining 85% of jobs scattered around in weaker second-order centres and non-centres. Positive correlations exist between percentage of employed workers, trip-based centricity and the subcentre ranks.
But we see significant disparities between these ranks and accessibility centricities. This shows the spatial mismatches for commute lengths in the system.
A subcentre with high trip-centricity, employing a high percentage of workers, but relatively lower auto- and transit-based accessibility centricity, implies that even though a significant percentage of the population comes to this location to work, access to jobs at this centre within 30 minutes is low.
A policy response would be to increase the accessibility of jobs from this location, as it already serves as a centre. This situation is particularly clear in the cases of Parramatta-Rosehill and Macquarie Park-Marsfield. Penrith and Liverpool too have extremely weak accessibility centricity.
Parramatta CBD is emerging as a secondary centre in the Greater Sydney region, but with much weaker accessibility of jobs than Sydney CBD.haireena/Shutterstock
Polycentric cities should promote spatial justice
As cities grow in size, commute lengths increase if the labour market for the entire metropolitan region is integrated. Commute lengths will stabilise if a city has a clear polycentric or modular structure.
In the case of Sydney, spatial equity has always been a concern. However, inter-city comparisons show city size has a strong bearing on its equity and efficiency.
Our results show it’s increasingly important for larger cities to introduce a framework of subregional labour markets as part of the polycentricity agenda. Enabling shorter commutes for workers will improve spatial equity as well as efficiencies.
The dominant method for measuring values of travel time savings (VOT) and values of travel time reliability (VOR) is discrete choice modeling. Studies using revealed preference have tended to use travel times measured by devices such as loop detectors, and thus the perception error of travelers has been largely ignored. In this study, the influence of commuters’ perception error is investigated on data collected of commuters recruited from previous research. The subjects’ self-reported travel times from surveys, and the subjects’ travel times measured by GPS devices were collected. The results indicate that the subjects reliability ratio is greater than 1 in the models with self- reported travel times. In contrast, subjects reliability ratio is smaller than 1 in the models with travel times as measured by GPS devices.
Once upon a time (1888 to be precise), the United States and the world launched a huge building boom for urban streetcars. Companies like Twin City Rapid Transit laid miles of track in fast-growing cities, extending well past the built areas to serve greenfield sites for emerging suburbs waiting to be platted and built. They did this because the streetcar promoters benefited directly from the land sales. The availability of a new, fast transit system connecting to downtown made houses much more valuable. The fares from the new passengers covered the operating costs of the system.
What is the optimal size of a research paper? The answer, of course, is that it depends. Some research findings are complex and difficult to explain, and are highly intertwined. Others are much more straight-forward, using well-understood methods to observe something new. However, most papers in most journals are expected to be of a certain length. In transport journals, for instance, this length is typically 3500 – 8000 words. This leaves a lot of words to fill, and people often stuff them, or are asked to stuff them, with repetition of well-known and well-established theory, regurgitation of self-explanatory tables and figures, citation of tangentially related research, and other matters describing what was not done in the research. Without a tight word count restriction, the authors have no recourse but to include filler at the bequest of the almighty reviewers, or in anticipation of such bequest.
Transport Findings takes the opposite approach. With a 1000 word cap (plus a maximum of 3 figures and 3 tables), it demands authors get to the core of their results: what did they measure, how did they measure it.
It’s really surprising what you can say in a few words. The Gettysburg Address was 270 words, depending on version.
Some people think fewer words means less work. The opposite is often true. Omitting needless words requires editing, and good editing takes time. The amount of time spent typing is not anywhere close to the amount spent reviewing, revising, and redacting in a brief text. We do that not for ourselves, but for our readers, to save them time, to help them see the point clearly without having to wade through a morass of miscellany and nonsense.
There are other reasons for short papers in addition to the benefits for the reader. They are faster to review, and so can go from conception to publication in less time than it takes some journals to move an article from their inbox to the review queue. I’d hypothesize (without any actual data, but impressionistically) that review time increases with the square of article length. So a 4000 word article will take reviewers on average 16 times longer to be reviewed than 1000 word article, neglecting fixed costs of getting people to read their email. Several things factor into this, most obviously considering the interactions of words in a text (a 4000 word article has far more textual interactions than a 1000 word article), but also including dread at reading a long rather than a short document for precisely that reason. And words beget words, a long document citing everyone but me can easily be just a bit longer.
We are now able in the academic community to produce many different kinds of research outputs, ranging from raw data, to figures and charts, to regression analyses, to texts and papers. These can all be put online at data conservancies and given permanent identifiers. Peer review still has some cache as a quality filter, let’s not waste the scarce time of volunteer reviewers with noise.
Spatial weight matrix is unstable over time-of-day, while network weight matrix is robust.
Performance of network weight matrix in non-rush hour is better than rush hour.
The best look-back time window depends on the travel time between two study detectors.
The best look-back time window is shorter in uncongested than congested regimes.
Fig. 2. Traffic flow of 140 links for Tuesday January 6th, 2015.
Abstract
This study examines the spatiotemporal dependency between traffic links. We model the traffic flow of 140 traffic links in a sub-network of the Minneapolis-St. Paul highway system for both rush hour and non-rush hour time intervals, and validate the extracted network weight matrix. The results of the modeling indicate: (1) the spatial weight matrix is unstable over time-of-day, while the network weight matrix is robust in all cases and (2) the performance of the network weight matrix in non-rush hour traffic regimes is significantly better than rush hour traffic regimes. The results of the validation show the network weight matrix outperforms the traditional way of capturing spatial dependency between traffic links. Averaging over all traffic links and time, this superiority is about 13.2% in rush hour and 15.3% in non-rush hour, when only the first-order neighboring links are embedded in modeling. In addition, this study proposes a two-step algorithm to search and identify the best look-back time window for upstream links. We indicate the best look-back time window depends on the travel time between two study detectors, and it varies by time-of-day and traffic link.
Train riders have to get to stations somehow. This is often referred to as the “first mile” or “last mile” problem. There are many technical solutions to help travellers get from home to the station and back, ranging from cars to electronic scooters, but most people use a much older technology, their feet, to get from A to B. What is seldom considered is access to the train platform itself.
Stations are not points but places. They occupy a large area. A person walking at average speed takes about two minutes to walk from one end of a full-length eight-car train to the other.
Often platforms have a single access point on one side of the station, which makes it more difficult for people on the other side of the station to get to the platform. Passengers may need to almost circumnavigate the station to get to the platform. At an average walking speed, the extra distance they must backtrack adds up to six minutes per trip each way, our research has found.
Imagine being so unlucky to have an extra 12 minutes of travel time every day if you take the train. You might be tempted to drive instead.
Illustration of worst-case scenario, traveler lives west of the station with an East Platform and works East of a station with a West Platform, adding 6 minutes of travel each way, 12 minutes per day.
The table below shows the extra travel time in minutes depending on platform locations and access points for a traveller’s origin and destination. The average time for such a one-sided configuration of train stations is 3.25 minutes each way.
Work
East
West
Live
Platform
East
West
East
West
East
East
0
4
4
2
West
4
4
6
2
West
East
2
6
4
4
West
2
4
4
0
Table 1: Additional Travel Time Depending on Origin and Destination Residence and Workplace Location vis-a-vis Platform Location.
While this example is hypothetical, it is drawn from experience in Sydney, where 44 of 178 train stations have only a single side entrance.
So what impact will a second entrance have?
We examined those stations and access to their platforms: how many people lived within 5, 10 and 15 minutes of the station platform, considering actual entrance location, and how many jobs were within 5, 10 and 15 minutes of the platform. Using existing ridership data from Opal cards, we estimated a model that related the passenger entry and exit flows at each station to that station’s accessibility.
Accessibility at train stations across Sydney.Author provided
We sketched a second entrance at those 44 stations and measured accessibility again. It’s now higher, as having two entrances instead of one means more people can reach the platform in the same time. We then estimated the increase in ridership from the model due to the improved accessibility, assuming no change in population or employment.
Over all 44 stations, total morning peak period entries increased by 5%. But some stations benefit a lot, and others not at all, so prioritisation of investments matters.
It will be no surprise to locals that Erskineville station comes out on top with a nearly 35% increase. While many of the new apartment-dwelling residents west of the station make the extra hike every day, even more would catch the train if there were a convenient entrance.
Other top 10 stations include: Bankstown, Newtown, Villawood, Redfern, Burwood, Sydneham, Caringbah, Meadowbank and Penshurst. Planning is already under way to improve Redfern station.
While this result considers existing development, adding a second entrance can make new transit-oriented development that much more valuable. This is because it will likely increase activity on the previously less accessible side of the station, as the example of Erskineville shows below.
Author provided
Other considerations include accessibility for people who cannot use staircases, as many of the stations are older and will require lifts. The prospects of park-and-ride lots, the costs of construction, the presence of nearby stations, and site feasibility also play into final decisions.
Welcome to the May 2019 issue of The Transportist, especially to our new readers. We are testing a new newsletter platform, Substack. If you received this by email, you have been migrated, and no action is required on your part. As always you can follow along at the transportist.org or on Twitter.
Laws
Like Newton in Mechanics or Tobler in Geography, we ought to have laws in transport. I posit the following:
Law 1: Everyone complains about transport, independent of the quality of travel experienced.
David Levinson, Transportist 
You must be logged in to post a comment.