Minnesota planners begin to envision driverless future | Star Tribune

Eric Roper at the Strib writes: Minnesota planners begin to envision driverless future.

My quote:

“We’re sitting here fighting about a train — a billion-dollar train that it tells you when you have to be there and where to go,” said state Rep. Pat Garofalo, R-Farmington, who owns a Tesla. “Pretty soon here you’re going to have cars that can just pick you up wherever you are and take you to wherever you want to be.”

But others like David Levinson, the lead author of the U report, said large cities will still need high-capacity transit to serve busy urban areas.

“Cars, even driverless cars, can’t move as many people per hour past a point as trains can,” Levinson said.

Now my fuller response to Eric Roper (to be clear, Eric never said his questions were off the record, and he was talking to a blogger, so he should have been aware):1self073017gr_online_525px

Do you think we’re planning enough for the arrival of this technology? It seems like there’s enough unknowns that folks like the Met Council don’t have much to say about how it will affect land use. And I’ve gotten some vague answers from Minneapolis, which is looking into it.

What should cities and states like the Twin Cities and Minnesota be doing at this point, if anything, to prepare and plan for the shift?

As you gathered no one [at the Metropolitan Council] is planning for [driverless cars]. Now, it is hard to say what the effects will be (I have my ideas), but my concern is not that we are NOT planning for driverless cars, but that we ARE planning for nothing to change. I.e., all of the plans and forecasts assume today’s technologies remain unchanged 30 years into the future, which seems implausible. This is a good time for alternative scenario planning rather than forecasts.

As a consequence of extrapolative forecasts (both in computer models and in people’s mental models of how cities work), cities like the Twin Cities (and others) are planning for more highway capacity when all the expectations for driverless cars should be more efficient use of road space (closer spacing between vehicles both laterally (narrow lanes) and longitudinally (shorter headways or gaps between vehicles). Given that roads are very long term investments and hard to reverse (i.e. roads are historically unbuilt at a much slower pace than they are built), building road capacity for needs that may soon disappear seems unwarranted and a classic example of a white elephant.

Nevertheless, there are factors which may increase travel demand (deadheading (i.e. empty and relocating vehicles), faster speeds, less driving responsibility (train passengers e.g. are willing to travel more time than car drivers because they can do something else while in motion), less expensive vehicles (EVs should cost less than the internal combustion engine when built to scale) and less expensive energy (with renewables, the price of electricity will continue to fall)).
However, there are other offsetting factors which could dampen this (switching from car ownership to a more taxi-like model in urban areas that charges on a per-trip basis, moving from an individual shopping to a delivery model for goods). Also one has to question whether the 5 day a week daily commute will remain as common in a world 30 years from now (with more telecommuting and more flexible work environments).
And all of this can be controlled with policy, we underprice roads by failing to recover both their direct costs (like infrastructure – fuel and related taxes pay for less than half of infrastructure costs) as well as their externalities (like pollution (most of which disappears with EVs), noise (similarly), crashes (most of which disappears with AVs), and congestion (which could disappear with proper road pricing)).
So what we should be doing: Don’t build new roads, or widen existing roads, until we implement a time-of-day based road pricing system that discounts off-peak use and recovers the full cost of car and truck travel.
In response to the Garofalo comment, I wrote:
Driverless cars don’t resolve the fundamental space issues of large cities (we can argue whether Minneapolis qualifies). Cars, even driverless cars, can’t move as many people per hour past a point as trains can. So if you have large demands to move people between fixed places, as you see in places like New York, Chicago, San Francisco, Sydney, or just about any city in Europe or Asia, AVs cannot solve that problem.
And that’s even assuming there’s some added capacity, right? Or do you disagree with the Tom Fisher argument that we’ll have a lot more roadway in urban areas to repurpose for other things?
To be clear, we will have more roadspace to repurpose (see our book), (we already do). But there are limits. If there are cities with large demands, trains will always move more people in a given amount of space than cars. Are the demands large enough now (or in the future) to justify the costs?  The biggest gains in space will come in (a) skinny cars (which will allow doubling the number of lanes), and (b) more widespread adoption of bikes and e-bikes and increased walking, with some additional gains from eliminating on-street parking, narrowing lanes for full-size cars and trucks, and shorter headways between vehicles.

Is it fair to say then that they could be commonplace by 2030?


Common, but far from universal. The median age of a car on the road is about 11.5 years, so if that still holds, the median car on the road will have been built in 2019.

There are also degrees of autonomy, (see the SAE report on this  ). I think most discussion refers to Level 4 (or 5) as Autonomous. Today we are between level 2 and 3 in production cars (Tesla autopilot is better than level 2 on freeways effectively, but not really quite level 3, the Waymo (Google) Car is level 4 in mapped environments), and the trials of test vehicles at the Level 4 and 5 range still require human intervention (“disengagements”) from time to time.

The Transportist: July 2017

Welcome to the July 2017 issue of The Transportist. As always you can follow along at the blog or on Twitter.

I was in Brisbane for the WSTLUR 2017 conference at the beginning of the month, and the family moved down, and it’s winter break, so slow this month.

Last call (closes July 30)

Transportist Posts


Media Appearances


Transport News





(Almost no Uber news this month)





Land Use

Equity / Justice

[As a Minneapolitan relocated to Australia, I am the mirror of Justine Damond, so I have been following this. Really, the police shouldn’t be killing anyone, there are better solutions in almost every case. And, as the Guardian reports, police manage to kill hardly anyone at all in many developed countries. In Australia, police killed 94 people in 20 years (5 per year), US police kill that many in one month. After controlling for population (325M vs. 25M (13x), police homicides in the US are still hugely more (1093 per 325M in 2016, vs. (scaling Australia levels to the US) less than 65 per 325M in Australia). For the math inclined, I am 16 times more likely to be shot by a police officer in the US than Australia. This is not random. It is not bad luck. It is systemic and systematic and it is not getting noticeably better. I have no special transport angle here, though the police apparently shot from the inside of the car, because why? Unlike Philando Castile, last year’s wrongful police shooting in the Twin Cities, after a traffic stop, where the police were outside the car and shot inside.]



Internet of Things


  • Elon Musk says he has ‘verbal’ OK to build N.Y.-D.C. Hyperloop – NPR [Is all news reporting so credulous? Famous person tweets something absurd, and it is just repeated far and wide by NPR, among others. My response: Today we launch a new solution to America’s traffic, The Pricing Company. We received tweets of approval from all 50 states to toll roads.]




By Us

Call for Papers

CALL FOR PAPERS – Deadline October 31, 2017
55th International Making Cities Livable Conference on Healthy, 10-Minute Neighborhoods
May 14-18, 2018, The Shaw Center, Ottawa, Canada

Commuter car parking

The Parliament of New South Wales is asking about commuter car parking.

Terms of Reference

That the Committee inquire into and report on commuter car parking in NSW, including:

  1. The effectiveness of current state government policies and programs covering commuter car parking;
  2. Processes for selecting the location of commuter car parks;
  3. The potential for restricted access or user pays commuter car parks;
  4. Consideration of alternative modes of first mile/last mile travel, including point to point transport, active transport and on demand buses; and
  5. Any other related matters.

This is my response …


It is my pleasure to provide information to the New South Wales Parliament’s Committee on Transport and Infrastructure regarding Commuter Car Parking. I am Foundation Professor in Transport Engineering at the University of Sydney, with more than 25 years experience in the field in the United States. While I cannot comment on individual car parks or their location, as the appropriate designs are usually context-specific, I can provide some general background and ways of thinking about the question.

The problem of commuter car parking is more generally the problem of accessing public transport stations, sometimes referred to as the “last mile” or “first and last mile” problem. While having fast, direct, frequent, and reliable public transport service is important, being able to get to that service is also critical. The travel times involved in accessing transit stations at either end are often as long as the time spent moving aboard the transit vehicle.

There are a variety of means that can be used to access public transport service, including walk, bike (including both traditional privately owned bikes and electric bikes (e-bikes) and bikes from newer bike-sharing and e-bike sharing systems), taxi (including ride-hailing like Uber), other public transport (like local bus or multi-party ride sharing vehicle), as a car passenger (‘Kiss and Ride’), or as a car driver (park-and-ride). (One can imagine other modes as well (e.g. car-sharing (like GoGet or CarNextDoor, but those are usually less practical). The best choice varies by individual and location, and most public transport stations will have a mix of arrival modes.

Historically, dating from the age of trams and stream railways, public transport was accessed primarily on foot. For this reason tram lines were spaced closely together (say every half-mile (or 800m)) so that walking to stops was convenient. Older suburbs in cities like Sydney developed around this transport mode, and had the residential population density to support frequent public transport service by tram, train, and later bus. Walk had and continues to have numerous advantages over other access modes, as it is low cost and has no environmental externalities.

Location efficiency (land use) with Walk and Bike Access

In a transit-based city, public transport and land use have historically evolved together, and new transit lines should be complemented with appropriate land development (and vice versa). Everyone in such places everyone can walk to public transport.

From a cost per commuter perspective, walk and bike are the least expensive modes, both for the traveler and for society as a whole. The advantage of bikes over walking is their larger catchment area. Biking (at 20 km/h) allows coverage of about 16 times the area of walking (at 5 km/h). This implies significantly more customers in the same amount of time, and should be strongly encouraged. The disadvantage is that bikes are sightlier costlier than walk as a mode to support, as bikes will require secure parking and safe access paths (walking of course requires sidewalks, which generally already exist, unlike separated bike lanes in NSW), as well as a supportive rather than hostile public policy environment. Nevertheless, the cost of bike storage is significantly lower than the cost of park-and-ride for automobiles. The Dutch are the world’s experts at bicycle transport and bike-and-ride, and many lessons about how best to do this can be learned by studying practice in the Netherlands.

Another cost-effective way to increase the catchment area of public transport is to construct entrances at each end of the station. Long platforms take nearly 2 minutes to traverse, so travellers who live, say, south of a station with an entrance at the north end may need to walk the length of the platform before entering the station, and then, depending on the preferred car to optimise their exit, may need to walk back again (an extra four minutes), which could be reduced with a second platform entrance. (They may need to walk another 2 minutes depending on their final destination vis-a-vis the exit at their destination station.)  This could be repeated on the evening commute, resulting in up to 12 minutes of lost time per day because of inconvenient entrances and exits.

Stationless bikesharing is becoming hugely popular in China, and Reddy-Go has introduced the service to Sydney. The advantage of such a system is that bikes will be located near frequent origins and destinations, and tend to cluster at stations. By encouraging bike access or egress, they make transit more desirable as a mode for more people.  Storage areas for shared bikes need to be set aside, clearly designated, and enforced should this become popular in order to ensure these bikes do not interfere with pedestrian access.

Bicycle parking at Train Station in Houton, Netherlands
Bicycle parking at Train Station in Houton, Netherlands


Pick-up and Drop-off.

The earliest pick-up and drop-off at transit stations date from the earliest days of the motorcar and suburban railway stations, have evolved into what are referred to as “kiss-and-ride”, whereby the driver (typically a spouse, parent, or child) drops off their family member at a transit station, and then proceed onto their final destination (after exchanging affections). (The mirror trip is logistically more complex and includes pick-ups in the evening, before returning home).  This is more efficient than park-and-ride as it avoids the need for parking at the station, and the costs of an extra vehicle for the household. While the multi-car family has resulted in this type of trip becoming less popular, saving time for the traveler chauffeuring the passenger at the cost of higher parking and car ownership costs, this type of trip may see an upsurge.  The advent of app-summoned taxis and their equivalent (Uber, Lyft, and so on) can provide access to or egress from transit stations, complementing transit service. Lyft, the main US competitor to Uber, reports that transit stops are their most popular category of destination.  While this is an added cost, more expensive than walking, biking, or well-used buses, one can imagine with the emergence of autonomous vehicles the costs will drop and this will become more popular, especially in lower density suburban areas.


Newer suburbs developed in the age of the automobile, and while many have grown to include train and bus services, the car is a far more dominant mode in these areas in terms of market share, and transit access is more difficult on foot because of the greater spacing between stations and lines and lower density of residential development. In these areas park-and-ride lots (commuter car parks) have been constructed.

The advantage of commuter car parks lies in basic geometry. It takes about 28 square meters to store a parked car on a surface lot (including access lanes, etc.), or about 360 cars/hectare. For a fully occupied 1 hectare lot, if every one of those parked cars carried 1 person, that produces 360 public transport boardings from that station in the morning (and 360 boardings elsewhere in the evening, assuming symmetry). That hectare generates 720 daily public transport trips.

In contrast, let’s say we had zero commuter car park spaces, and those car users could not otherwise access the station because of distance and lack of other access modes. Instead we had transit-oriented development. Let’s further assume that adjacent land uses have a 50% public transport mode share for work trips and 0% for non-work trips. We would need 720 resident workers on that hectare to have a similar number of public transport trips generated. Since only half the population works, we are looking at 1440 total persons on that ha of land to generate as many trips as transit oriented development. The point is not that anyone should (or shouldn’t) build a structure with a 1 ha footprint housing 1440 people, just that park-and-ride generates a large number of riders that cannot be easily made up with low-density transit-oriented development.

Low, or even medium, density residential development around the station will not enable as many public transport users as the park-and-ride lot. Now that doesn’t mean it is cost-effective to build a park-and-ride lot, which depends on the value of land, on maintenance costs, whether park-and-ride spaces are given away for free or can be charged for, and levels of demand. It certainly doesn’t mean it is cost-effective to construct a parking structure, which cost on the order of $50,000 per space (amortised that is about $5000 per space per year, or $20 per space per work day)

Even after accounting for construction, surface parking lots are far from cost-free, maintenance costs are surprisingly high: in Minnesota, a 288-stall lot generated $AU 43,000 per year in maintenance costs which amounts to a subsidy of at least $AU 147 per parked car per year. (Divide by occupancy, the share of spaces used daily, for the actual subsidy, which is higher), or at least $AU 0.58 per day per car. While most of Australia can avoid the snow plowing costs of Minnesota, lighting and other maintenance issues remain.

If the charge for car parks were free, this adds to the cross-subsidy from people who walk to public transport to people who drive to public transport. To speed revenue collection, parking should be paid for with Opal cards. The rate should be set separately for each lot as costs, demands, and conditions vary.

As the market evolves over time, surface park-and-ride lots can be thought of as a land bank, which can be developed at higher intensities when conditions warrant. The simplest way to ensure land is developed to the highest and best use, be that park-and-ride surface lots, structured parking, or more intensive land development is to place it in the hands of organisations with the right incentives. This may require allowing the transit service provider to develop land adjacent to and above (and below) stations. Land value capture techniques (like the land value tax and joint development, among others) can be used to ensure that the transit system benefits from the land value uplift created by transit services.

Trains running alongside freeways and freeway express/bus rapid transit lanes are especially appropriate for park-and-ride, as the drivers converging on downtown can be persuaded to divert to transit upstream of the city and avoid downtown parking costs (and the resulting congestion between their diversion point and the city).

Impacts of Millenial Student Loan Debt on Transportation Choices

The Federal Highway Administration’s Transportation Policy Unit has a series of reports on Transportation Futures. I was involved in one of them as an advisor to the consultant, though my name is not on the report, so I am not responsible. The report  is now online:

Figure 18. Transportation and Loan Payment as Percentage of Income
Figure 18. Transportation and Loan Payment as Percentage of Income


Impacts of Millenial Student Loan Debt on Transportation Choices

Now the largest generation in America, the Millennials are not driving at the same rates of their predecessor generations, the Baby Boomers and Generation X. There have been plenty of studies about the millennial generation’s lack of interest in driving. Many conclude that Millennials are fascinated by technology or urban culture.

According to AAA’s findings of the 2013 ‘Your Driving Costs’ study, annual automobile spending for an average sedan owner are $9,122 (Based on 15,000 miles annual usage). For someone newly out of college with student loan debt, automobile ownership may feel out of reach. Millennial student loan debt is a widely discussed topic. Approximately 40 million Americans hold student loan debt. Currently more than 70 percent of U.S. students who graduate with a bachelor’s degree leave with debt, averaging $28,400. According to the White House Council of Economic advisors, 61% of adult millennials attended college, compared to 46% of their Baby Boomer parents. In 2014, the total outstanding student loan debt in the US surpassed $1 trillion.

This paper attempts to investigate the impact of student loan debt along with other variables on the millennial transportation choices.


The data was tortured looking for a relationship. If there is one, it is weak. For instance see the finding buried on p. 34

“In general, cutting back on transportation expenses may not be a central priority for those with student loans, as their job earnings enable such individuals to handle rising transportation costs. Indeed, the data shows a positive relationship between income and student loan value (Figure 15).

Taking into account all student loan holders, the relationship between student loans and transportation expenses appears unclear. While some analyses suggest a slightly positive relationship (i.e., the uptick in transportation expenses for loan-holders), many of the other trends can be explained by Figure 15. Student loan holders in our data are generally well-off, which would contribute to higher transportation expenses.

Across the board, people spent less on transportation as a percentage of household expenditures post- 2008, but Millennials showed a particularly large difference between loan-holders (Figure 16, right) and those without loans (Figure 16, left). While we might consider this drop to be connected to student loan commitments, a number of analyses seem to refute this idea.”

Closing the Cahill Expressway at Circular Quay

Map of Cahill Expressway, Red line indicates section under discussion in this post.
Map of Cahill Expressway, Red line indicates section under discussion in this post.

The Cahill Expressway in Sydney, the city’s first expressway, opened in 1958, connecting the Eastern suburbs to the Harbour Bridge. After the Harbour Tunnel opened in 1992, traffic was halved, the section’s Eason or being eliminated. Looking at a map, you can see the Harbour Tunnel and Harbour Bridge approaches join north of the Harbour, and basically form an upside-down V-shape, with the Circular Quay section forming a cross, the segment turning the upside-down V into an A.

Traffic counts for the Cahill Expressway at Circular Quay are given for 2012 as about 20000 average annual daily traffic in each direction. While certainly non-trivial, this is also not a lot for two lanes in each direction, equivalent to a four-lane arterial. And when the system is working, all of this traffic has alternative routes, as the route is topologically similar to the classic Braess Paradox.

The Braess Paradox observes that under certain circumstances an additional link increases total travel time, and is dysfunctional, because of the difference between the costs that travellers pay and the costs they impose by congesting others. While it is hard to prove such cases in the real world, there is no reason for this link to exist in the post-Tunnel configuration except as a backup when the Harbour Tunnel is closed or constricted to divert traffic to the Harbour Bridge.

Circular Quay Railway Station, Cahill Expressway on the top deck.
Circular Quay Railway Station, Cahill Expressway on the top deck.

If this section of the Cahill were to be removed, many of its access and egress links could be removed as well, creating additional space and sunlight in the constricted central business district. Southbound traffic would decide north of Sydney whether to diverge for the East or West and then take the Bridge or Tunnel, with no recourse except for city streets. Northbound traffic from the East would take the tunnel to cross the Harbour or exit onto city streets. The operators of the tunnel should be pleased.


Suppose the Circular Quay section were closed. The expressway lies on the upper deck of a double-deck elevated structure, with an elevated railway (the under-rated John Bradfield‘s City Circle, completed in 1956)  immediately below. So the whole structure cannot easily come down. Instead the expressway deck can be repurposed, much like New York’s High Line and other infrastructure reuse projects, as a pedestrian overlook (there is already a sidewalk) on the north side, with the south side hosting restaurants and open-air cafes with a gorgeous view of the Harbour.* I am sure urban designers could come up with some lovely watercolour renderings.

Views like this from the Cahill Expressway at Circular Quay (photo from the train deck below)

While all of this undoubtedly requires study and many, many consultant contracts, it is really easy to test the actual traffic effects (and would make a nice Master’s Thesis project). Close the ramps for a few weeks “for repairs”. This must happen from time-to-time anyway.  Perhaps there is a ‘natural experiment’ coming up, or recently passed, when this happened. Monitor traffic elsewhere in the system. Evaluate the consequences.

The hypothesis is that traffic conditions are no worse overall (system travel time is unchanged or lower), though selected links may in fact be worse off while others are improved. Given the reduction in merges and diverges, I suspect more links are improved than worsened.

If this hypothesis is borne out, there is less total travel (fewer vehicle kilometres traveled) in the city, travel is faster, and most travellers are better off.

In recent decades there has been a trend for cities to close obsolete freeway sections. San Francisco famously took down the Embarcadero Freeway for instance, opening up the waterfront. Seoul removed the Cheongyecheon freeway and restored a river. There have been others. While removal of this section of the Cahill is not likely to have the same effects, as the elevated railway will remain, it still could be beneficial.  Proposals to demolish the entire Cahill, which bisects major parks the Botanical Gardens and the Domain have also been discussed, though burying them under air rights park seems a far simpler and less controversial proposal, and less like to strand the Harbour Tunnel.

Aerial Photo (via Apple Maps) of Cahill Expressway at Circular Quay, facing south. Red marks indicate extent of section.
Aerial Photo (via Apple Maps) of Cahill Expressway at Circular Quay, facing south. Red marks indicate extent of section.

Update July 28: A reader writes:

I think you are seriously wrong about the Cahill Expressway and its utility.

It is effectively the artery that feeds and drains the eastern side of the CBD for we who live on the north side (and who I might add paid for it!) and without it the eastern side of the CBD would be near impossible if not extremely inconvenient to access. It cannot be accessed from the tunnel and otherwise requires traversing the city not fun normally and a nightmare right now.

And I think it is a lot prettier – if that can ever be used about 1950’s engineering – than the much loved EL in Chicago and other insertions into older cities to make them work.

And you can at the very least watch the NYE fireworks from it! Or pre 911 you could.

– apart from anything else it is part of JJC Bradfield legacy and that is by popular consensus untouchable!

My response

Fort Street Public School
Fort Street Public School.  A school surrounded by freeway ramps. They are very elegant and nearly perfectly circular ramps (and must be fun to drive).

I am referring only to the section on Circular Quay. How hard would it be to connect  Bridge Street only to the tunnel? I know everything takes too long and costs too much, but I bet with a concerted effort, if there weren’t any significant underground utilities, this would be under a month.  This configuration is only the way it is  for historic reasons (the Bridge was first), no one would configure it that way now.

Cahill AADT (at Circular Quay) 40k 2-ways (4 lanes 10k/lane)
Harbour Tunnel AADT 90k AADT (4 lanes 22k/lane)
Harbour Bridge AADT 160k AADT (8 lanes (1 bus only), 20k/lane)
A better argument for keeping it might be that the Harbour Tunnel is more congested than the Bridge. But surely they are in equilibrium because traffic has sorted itself out, and will do so with any other change, and road changes would be reflected in different effective catchment areas and changed patterns (longer distance trips might use the bridge to the Western Distributor to the Cross City Tunnel instead of the Harbour tunnel for instance. And with all of the development going west of the city (rather than East, where the Ocean lies), shouldn’t traffic from the east be steered away from the Bridge toward the Tunnel)
Now I guess Kirribilli is more difficult to access via the Tunnel than the Bridge, but isn’t that what the ferry is for?
Of course the irony of Bridge Street leading to the tunnel is also a worthwhile reason.
I will leave the aesthetics to the eye of the beholder, but the structure wouldn’t fully come down unless there was a solution for the trains.



* A single lane passage for emergency vehicle could be maintained if necessary.





Mo’ Rankings, Mo’ Bragging

I posted skeptically late last year On Academic Rankings. Some new rankings have come out, so it is time to brag or fret some more. The world renowned ARWU has come out with new rankings for Civil Engineering, and better still, for Transport.

I am pleased to report Sydney comes in 7th globally in Transport. None of this is my doing, I just got here, but nevertheless it is good to hear.

ARWU Transport Rankings

World Rank Institution* Country Total Score Score on
1 Delft University of Technology 274.9 100
2 Beijing Jiaotong University 267.5 79.9
3 University of California, Berkeley 265.5 86.3
4 University of Leeds 245.6 72.6
5 National University of Singapore 241.8 61.8
6 The Hong Kong University of Science and Technology 237.4 51.1
7 University of Sydney 236.7 78.7
8 The Hong Kong Polytechnic University 234.7 62.8
9 The University of Texas at Austin 230.9 73.8
10 Massachusetts Institute of Technology (MIT) 228.7 69.3
11 University of Maryland, College Park 228 74.9
12 Purdue University – West Lafayette 224.9 74.3
13 Georgia Institute of Technology 224.4 67.8
14 Tsinghua University 223.2 74

This is not how I would rank the universities, but feel free to look at the methodologies and construct your own.

The ARWU indicators and methodology are provided at: http://www.shanghairanking.com/Shanghairanking-Subject-Rankings/Methodology-for-ShanghaiRanking-Global-Ranking-of-Academic-Subjects-2017.html#3

Sydney comes in at 32 globally in Civil Engineering (a bigger arena than transport usually). Again I am not responsible, and this is not how I would rank them, and it sure is puzzling how this is how it came out, and sadly we are behind local rival UNSW, but that is being worked on …

ARWU Civil Engineering Rankings

World Rank Institution* Country Total Score Score on
1 Tongji University 313.1 100
2 Lehigh University 252.1 39.7
3 Swiss Federal Institute of Technology Zurich 250.9 63.9
4 The University of Texas at Austin 248.8 66.2
5 Polytechnic University of Madrid 226.9 59.4
6 Tsinghua University 226.1 85.8
7 National Technical University of Athens 217.2 54.7
8 University of California, Berkeley 213.7 72.6
8 University of Illinois at Urbana-Champaign 213.7 67.9
10 University of Canterbury 204.2 36.4
11 The University of New South Wales 202.4 59.2
12 Western University 201.9 42
13 University of California, San Diego 197.3 39.6
14 Nanyang Technological University 192.8 56.4
15 University of California, Davis 189.9 56.1
16 University at Buffalo, the State University of New York 185.3 40.3
17 The University of Tokyo 183.9 47.1
18 Texas A&M University 183.7 77.8
19 University of Toronto 181.7 50.9
20 Purdue University – West Lafayette 180.2 63.6
21 Dalian University of Technology 176.2 75.8
22 The Hong Kong Polytechnic University 174.4 76.6
23 Seoul National University 174.3 56
24 The University of Hong Kong 174.2 56.4
25 Swiss Federal Institute of Technology Lausanne 173.8 53.8
26 Delft University of Technology 173.3 75.5
26 University of Stuttgart 173.3 36
28 Virginia Polytechnic Institute and State University 172.8 59.2
29 McMaster University 172 36.4
30 Harbin Institute of Technology 171.7 69.4
31 Southeast University 170.6 73.8
32 University of Sydney 169.5 43.9
33 University of Michigan-Ann Arbor 168.3 48

An Introduction to the Network Weight Matrix

Recently published:

This study introduces the network weight matrix as a replacement for the spatial weight matrix to measure the spatial dependence between links of a network. This matrix stems from the concepts of betweenness centrality and vulnerability in network science. The elements of the matrix are a function not simply of proximity, but of network topology, network structure, and demand configuration. The network weight matrix has distinctive characteristics, which are capable of reflecting spatial dependence between traffic links: (1) elements are allowed to have negative and positive values capturing the competitive and complementary nature of links, (2) diagonal elements are not fixed to zero, which takes the self-dependence of a link upon itself into consideration, and (3) elements not only reflect the spatial dependence based on the network structure, but they acknowledge the demand configuration as well. We verify the network weight matrix by modeling traffic flows in a 3 × 3 grid test network with 9 nodes and 24 directed links connecting 72 origin-destination (OD) pairs. Models encompassing the network weight matrix outperform both models without spatial components and models with the spatial weight matrix. The network weight matrix represents a more accurate and defensible spatial dependency between traffic links, and offers the potential to augment traffic flow prediction.

Talk: Full Cost Analysis of Accessibility

The following research talk will be held on the University of Sydney campus on July 10. Let me know if you are interested in attending.

Comparing alternative paths
Comparing alternative paths

Title: Full Cost Analysis of Accessibility

Abstract: Accessibility measures the ease of reaching valuable destinations. For transport systems, accessibility combines travel costs and opportunities into a single metric, which represents both the transport network and land-use. Traditional accessibility metrics have been analyzed from the perspective of travel time – considering the time cost of travel. This fails to fully capture the full travel costs, especially the external costs of travel. In this presentation, a framework of extending accessibility analysis is proposed combining the significant cost components of travel, time, safety, emission and money, with accessibility analysis. By examining both private costs, which individual travelers consider when making travel decisions, and external costs, which society should consider when making investment decision, this approach better aligns accessibility with the goals of evaluation for transportation planning. A proof-of-concept analysis based on a toy network was conducted to prove the practicability of the framework. The current studies focus on the implementation of the framework on Minneapolis-St.Paul metropolitan region.

Bio: Mengying Cui is a Ph.D. Candidate in the Department of Civil, Environmental and Geo-Engineering at the University of Minnesota, and working in the Accessibility Observatory of the Center for Transportation Studies at the university. Mengying earned her bachelor of engineering degree (2011) in logistics engineering from Tongji University and a master of engineering degree (2014) in transportation planning and management from Dalian University of Technology. Her research interests in accessibility evaluations, transport economics, network reliability and GIS in transportation. She has been granted Matthew J. Huber Award for Excellence in Transportation Research and Education (2017).

When: Monday July 10, 2017, 14:00 – 15:00 pm AEST

Where: Room 438 | School of Civil Engineering Building J05  | The University of Sydney | NSW | 2006

Elon Musk’s Chicago Tunnel makes a dumb idea even dumber | Wired

I was interviewed by Aarian Marshall of Wired for Elon Musk’s Chicago Tunnel makes a dumb idea even dumber. My quote:

Be skeptical, too, of Chicago’s promises that any contractor—Boring Company included—will pay for this tunnel’s construction and maintenance all on its lonesome. Public transportation construction and operations are rarely profitable, says David Levinson, a transportation engineer at the University of Sydney. “There’s no way fares will cover the capital costs of a tunnel,” he says. “If you made everyone take this train, maybe.”