How Fast Do SEQ Trains Need To Be To Compete With The Car?

[#Metro]

How Fast Do SEQ Trains Need To Be To Compete With The Car?

We know from TransPerth and information publicly available that the TransPerth Mandurah line achieves patronage levels similar to that of the Sydney Metro Northwest Line (at the time of writing this terminates at Chatswood, Sydney). It manages to do so in an operating environment that has densities and population far, far below that for Sydney. And car usage that is the highest in Australia. 

Part of the success of the TransPerth Mandurah (and Joondalup) lines are the value proposition - high speed and high frequency combined. In the following exercise, a model is constructed to answer the question "How fast must trains go to equalise with a comparable car trip?"   

The TransLink definition of a journey being composed of multiple trips is used.

We assume that destinations are within 10 minutes access time of the train station (that could be a 10 minute walk, or a 10 minute car or bus trip). And that access time must be undertaken at both the start and end of train trips to form one complete journey.

Working Principles
- The main principle here is that the speed of the train should be adjusted such that the difference between a car journey and train door-to-door journey should be zero.
- To do this, we first need to work out the car travel time, and subtract any PT access or wait time components of the GTCE. The remaining time is available for train in-vehicle time.
- Because car average speeds vary during peak and non-peak times, a series of average speeds must be used. E.g. 30 km/hr for peak hour, higher speeds for non-peak times.

- Trips are modelled as spontaneous, that is the average user will experience a wait time of 1/2 x Frequency. (See note on "The 9-to-5 Commuter's Objection" for a discussion about this at https://humantransit.org/2009/04/why-transferring-is-good-for-you-and-good-for-your-city.html)

GTCE = General Travel Cost Equation, the sum of access time, in-vehicle time, and (average) waiting times. No penalties, weightings or adjustments were applied. No parking costs/PT ticket fares were considered.

Method
1. Set the distance to be travelled
2. Calculate the average time a car would take to travel to the same distance
3. Calculate the appropriate Public Transport constant (e.g. 10 min walk + 1/2 frequency average wait)
4. Subtract the PT constant from the car travel trip time to give the amount of time available (remainder) for the train to complete the given distance (the train "in vehicle" time)
5. Calculate the speed required to meet the distance in this given time

[#Metro]

Results



Notes: Values are average train speeds in km/hr * Any calculation errors- please advise. Thanks!

[#Metro]

Conclusions
So, what can we say about the results?

Firstly, it is easy to see why certain members of RBOT do not like freeway construction. The average speed required for a train to match the overall journey time provided by a car in free-flow freeway traffic at 15-minute frequencies requires dedicated HSR and is not possible on the narrow-gauge QR network. Model suggests that the average speeds required are >> 160 km/hr.

So, even if Gold Coast trains ran at 160 km/hr for their entire journey and came every 15 minutes, getting to Southern Gold Coast destinations would take about 30 minutes in-vehicle time. But once station access (2x 10 min) and wait time (1/2 x F = 7.5 minutes) are taken into account, its 57.5 minutes for a journey.

Secondly, let's look at urbanised areas where road speeds are generally 50 km/hr.
If we have even just a little bit of congestion - say 50 km/hr drops to 40 km/hr the picture starts to change quickly. At 15 minute frequencies, the average train speed required to match the car journey is less than 100 km/hr.

If we could get average train speeds up to around 60 km/hr - 80 km/hr combined with 15-minute frequency, trains would be very competitive vs car getting to places like Beenleigh, Caboolture and Ipswich.

[Gazza]

Im confused, so are you saying a 50km drive at 100kmh, the train must travel at 180kmh?

[#Metro]

Im confused, so are you saying a 50km drive at 100kmh, the train must travel at 180kmh?

Example
Alice and Bob both live in homes 10 minutes from Train Station A.
The distance between Train Station A (their local stop) and Train Station B (CBD) is 50 km.
But their ultimate destination is within 10 minutes walk of Train Station B in the CBD.

Choices:
1. Alice and Bob can either drive to the station (and get the train) or
2. Drive to the CBD directly.

Alice and Bob leave their homes at the same time.

Driving
Alice jumps in the car and drives 50 km to the city, on the freeway (off-peak). Alice gets to the CBD in 30 minutes (0.5 hour) maintaining an average speed of ~ 100 km/hr. Alice parks at the final destination, which is within 10 minutes walk of Train Station B (CBD).

Train
Bob jumps in the car and drives 10 minutes to Train Station A, and waits 1/2 F (7.5 minutes) for the train. The train travels at an average speed of X km/hr and reaches Train Station B in the CBD. Bob then walks 10 minutes to the final destination.

Alice and Bob both arrive at the final destination at the same time (thus meeting the condition that Car minus Train door to door journey time = 0)

How fast did Bob's train travel on average?

Bob took 10 minutes to drive to Train Station A and 10 minutes to walk from Train Station B in the CBD. He waited 7.5 minutes at the platform. The PT constant is thus:
- 10 min drive to Train Station A
- 7.5 min wait at Train Station A
- 10 minutes walk out of Train Station B
= 27.5 minutes

If Alice's Trip door-to-door was 30 minutes in total, then the train must have made the journey in 30 min minus 27.5 minutes.

This gives just 2.5 minutes to travel the full 50 km between Train Station A and Train Station B.

So the train travelling at average speed X must have been travelling at 50 km/2.5 minutes
which is 50 km/0.041666... hour, which gives an average speed of the train at 1200 km/hr.

Conclusion

As we know that no train on the QR suburban network can travel at speeds greater than 160 km/hr, and even with HSR (200-350 km/hr), we can conclude that at a 15 min train frequency there can be no train service that can match the car door-to-door journey time if the car can maintain a given average speed of 100 km/hr.

[Jonno]

This is nonsense thinking! Cars have destroyed our cities, finances and health! The concept that bigger and wider roads improved access and mobility is a blinkered "car-centric" perspective! The facts are it's done the opposite.

Further prioritising investment in driving has only made this worse and seen more people stuck in ever worse traffic!

This is why we need to de-engineer our cities and return our roads back to place people can enjoy...or at worst co-exist!

Public transport doesn't need to "compete" with driving! Driving needs to stop dominating and destroying our cities!

[#Metro]

This is nonsense thinking! Cars have destroyed our cities, finances and health! The concept the
At bigger and wider roads improved access and mobility is blinkered "car-centric" perspective! The facts are it's done the opposite.

Further prioritising investment in driving has only made this worse and seen more people stuck in ever worse traffic!

This is why we need to de-engineer our cities and return our roads back to place people can enjoy...or at worst co-exist!

Public transport doesn't need to "compete" with driving! Driving needs to stop dominating and destroying our cities!

Hi Jonno,

I'm sure it has. I just did this little exercise to see what would happen. I can see very clearly in the numbers why anyone would be opposed to freeway construction. It has a devastating patronage pull effect away from PT when you compare door-door (note, not station to station) overall journey times. Provided that the freeway doesn't max out capacity.


The output of the model is specific and quantitative. You can take this ideal case and modify some of the inputs to see what effect this has.  But I think these inputs are reasonable.

These results suggest measures to:
- Increase average train speed, particularly for those longer regional journeys
- Increase train frequency to at least 15 minutes all day (no hope of genuine mode share shift with just 30 minutes)
- Slow the car trip down
- Charge for road usage and/or decongestion charging
- Restrict and charge for parking in the CBD (and elsewhere if need be)

Public transport doesn't need to "compete" with driving! Driving needs to stop dominating and destroying our cities!

Well, if it wants to get mode shift it does. And this exercise just considers PT matching the car. If you make driving slower by say dropping the freeway speed limit to 50 km/hr (or making it a normal urban road) then the total area enclosed within any isochrone is going to shrink quite dramatically.

Let's look at what happens to a 30-minute isochrone ("everything within 30 minutes"):
(Happy to be corrected if there are any arithmetic errors - please advise)

At an average speed of 100 km/hr a 30 minute isochrone is 50 km from the CBD centre.
At an average speed of 50 km/hr a 30 minute isochrone is just 25 km from the CBD centre.
Which implies the accessible area shrunk by 75%.

Assuming an idealised city formed a circle, a 30 minute isochrone would enclose:
Area = pi x r-squared (with R being 50 or 25 km)
Try it yourself: https://www.omnicalculator.com/math/area-of-a-circle

- A 50 km radius isochrone encloses 7853 km²
- A 25 km radius isochrone encloses 1963 km²
- The area lost thus 5890 km² which leaves just 1963 km² accessible within 30 minutes.

The loss in freedom or accessibility can be measured or quantified as a drop in the accessible area:

Percent change = [(Old - New)/(Old)] x 100

Thus, [(7853 km² - 1963 km²)/7853 km²] x 100

= 75% drop in accessibility. 

[HappyTrainGuy]

Sorry but what dribble this is. This is where you start to come into issues when you compare direct areas and transit methods between different cities. The population might be similar but the actual method of people commuting is completely different. Take Carseldine/Fitzbiggon/Aspley. I'm using this as an example about the northern busway going to the PCH. The bulk of the population does not work in the city. Or how is having a 160kph train line to the Gold Coast going to fix the highway congestion at Helensvale when the bulk of the congestion is created by locals who are not going to Brisbane? North Lakes is another. While you might get lines and populations similar between cities the commuting patterns between both cities can be vastly different. The 330 buz is quite popular but outside of its main travel times its just an air parcel like the 340 buz that makes you try to justify it being maintained as a buz - which in its own has its own problems with the buz standard not being a one size fits all routes.

Once again those going to the city are get priority over locals who still don't have a decent public transport service. It was only 11-12 years ago now that the 335 outbound stopped at the Chermside interchange during peak hour for the first time. That meant for years those that lived to the north had no bus access to the a substantially large trip generator. Same for morning peak. If you lived along the 335 route which was a simple bus trip to Chermside you could not access the Chermside interchange because it went straight onto Webster road to the city as the P339.

[Gazza]

An average speed of 100kmh is largely impossible for most journeys in Australia.

If you're driving from one Roadhouse to another, both located on a highway with a 100 limit,you could achieve this very closely.

But most destinations urban areas are not immediately adjacent to a road with that speed limit.

[#Metro]

An average speed of 100kmh is largely impossible for most journeys in Australia.

If you're driving from one Roadhouse to another, both located on a highway with a 100 limit, you could achieve this very closely.

But most destinations urban areas are not immediately adjacent to a road with that speed limit.

- An average speed incorporates the total overall journey, no matter what combination of speeds are actually taken. There are lower average speed bands that we can refer to, we might want to use a speed band of 90 km/hr for example.

- If Alice maintains 50 km/hr average speed over 50 km length (even if a portion of her trip was on a 100 km/hr motorway), Bob's train still needs to reach an average journey speed of 92 km/hr over its trip for PT to match the Car door-to-door journey time (assuming train frequency = 15 min).

- The other thing is costs, which IMHO is in favour of PT. Here I was only interested in journey time comparisons. Even though car might be faster for certain trips someone might decide its not worth paying $20 for parking at the destination as the alternative trip you might take might be $5.

Sorry but what dribble this is.

HTG, I've set out my working, principles, inputs and method. You don't have to agree.
You can even do this yourself in Excel and you'll probably get the same figures.

Or how is having a 160kph train line to the Gold Coast going to fix the highway congestion at Helensvale when the bulk of the congestion is created by locals who are not going to Brisbane?

I don't think this table was constructed for that particular purpose? The question I wanted to explore was defined in the initial posts.

In any case, for the Gold Coast:

Bus Speed = Road Speed, OR
Bus Speed < Road Speed

Because buses can't drive faster than the road speed limit and must also make stops.

So, we wouldn't expect to see Bus Speed > Road Speed (as there is no bus priority). Therefore, we can be confident overall bus PT trips would be the same journey time as the car at absolute best, and more generally worse/longer than comparable car trips once we adjust for waiting and walking to the bus stop.

If roads become congested all-day, then we would use a lower speed in the car average speed band in the table. We would also consider separating the bus from the road.

I think the table and approach can be quite useful.

Let's go back to Alice who now can only maintain a 50 km/hr average car journey speed as the motorway and roads leading to it has now congested. At 15-minute frequency, the average train speed required to maintain PT/Car door-to-door parity is 92.31 km/hr.

Trains on the QR network might be able to go 130 km/hr, so rather than build an additional lane on the Pacific Motorway M1 to help Alice (who might live at Ormeau for argument sake), we could come in and say that money could be better spent on improving average train speed from 61 km/hr to 92 km/hr.

This is technically achievable on the QR network (green shaded cells), and therefore it would boil down to whether the overall benefit/cost for the enabling infrastructure was:

Motorway > Railway
Motorway < Railway

Not only that, because we have a model of this we can be very specific. Rather than just say "increase train speed" we can also now also say by how much - 31 km/hr. And based on this we can also see that we probably don't need it to be HSR either. It probably could be done with existing trains and a new alignment

[#Metro]

I've made an additional table, so both together are:





If members find any arithmetic errors, please let me know and I'll revise. Thanks!

Comments

This is a really interesting result. 

- We know PT usage is highest in the inner city, but here car has absolute advantage and the cells are thus greyed out. In the time it takes someone to walk in and out of a train station, a car would probably be at the destination. Interesting. 

- Either, inner city congestion is really bad. Or maybe something else is at play here like parking pricing and supply. Or maybe people's preferences have a threshold level? Perhaps people "expect" to spend at least 30 minutes and for them to take a car, they have to save a minimum parcel of time (say 15 minutes) before taking the car.

[Gazza]

Here's a real example, Beeneligh Station to Roma Street Station.

Both stations are Very close to motorways so the vast majority of the journey is on a motorway.

37.4km Drive
30 min travel time.

Thus an average speed of ~75kmh

Another example would be Richlands to Bowen Hills, both very close to motorway exits.

23km drive
21 min travel time

Thus an average speed of ~66 km/h


My take
-I don't think it is necessary for public transport to match the journey time inclusive of the wait time.
The reason being is in terms of promotion of the service, the headline travel time is what will grab people and get them to try it out.

Things like the access time, the wait time, the time to get a park, all too variable on the individual level IMO.

And we can see that even with a good frequency, adding 7.5 min wait + 10 min access is automatically adding an almost impossible hurdle.

Frequency is freedom I agree, but again as long as the headline frequency is "Good" then that's enough to get them to try it.

The access time is one I want to debate.
If the PT network is sufficiently developed, then when comparing drive versus train, you'll go near a station at some point anyway.

Eg this model assumes that 10 mins of access time is 10 mins going out of your way to get to the station but the driver doesn't suffer that. Why?

For the PT user, that 10 mins of access time could well be "on the way" anyway, so in effect its shortening the distance travelled on the train itself.

[#Metro]

My take
-I don't think it is necessary for public transport to match the journey time inclusive of the wait time.
The reason being is in terms of promotion of the service, the headline travel time is what will grab people and get them to try it out.

Things like the access time, the wait time, the time to get a park, all too variable on the individual level IMO.

- It is important to calculate the parity speed. This give us a stable reference point for comparison. I suspect that it also implies if the journey times were equal between modes, the PT vs Car mode shares would also be equal as Alice and Bob would be indifferent to their mode of travel (working hypothesis). 

Eg this model assumes that 10 mins of access time is 10 mins going out of your way to get to the station but the driver doesn't suffer that. Why?

Because it is already captured in their average car speed. It's door-to-door.
Even if that average car speed is brought down dramatically (e.g. from 100 km/hr to 50 km/hr), the difference in journey times is so wide that the train still needs to run very fast to achieve parity.

- Access time and exit time needs to be included. The fact that variance exists does not extinguish the fact that there is also a mean value. Otherwise one could argue that the time spent driving to and from motorway on- and off- ramps should be excluded. Or that most people live and work immediately above a train station (they don't).

- Example: Access time to the Airport is what got the HSR service barely across the line in the latest HSR business case. HSR proponents argued that access was easier if the HSR called at main inner-city stations whereas people had to take a taxi or do a trip to get out to the Airport, even if the train was slower than a plane.

- You do need to adjust for frequency because waiting time is a real experience, and it does impact patronage. We know that from BCC's BUZ introduction that patronage doubled when the waiting time was halved. The bus speeds themselves did not change.

[HappyTrainGuy]

Frequency plays a part but so too does travel demand and time of day. It's also based on a single trip journey or those directly near a station which is still flawed ie having to drive south to catch the train vs driving straight onto the motorway (and let's be realistic 10 mins is not a good example due to station spacing on some longer lines). Such as Springfield to Beenleigh or Caboolture to Redcliffe, Caboolture to Airport, Gold Coast to Cleveland. Cost is another issue that determines pt use. For example I've highlighted before how people at Nundah, Nudgee, Banyo and Virginia and even as far out as Boondall, Wavell Heights, Zillmere and Geebung prefer to drive to Northgate to catch a peak hour train than to use their local train station (it's sometimes longer than the train going to your local station depending where you live but more than $500 a year cheaper to do so - Northgate has frequency but we saw similar problems with Strathpine and Bald Hills with the fare zone there where people were preferring to drive a longer distance to get a cheaper fare which put congestion in areas where there previously wasn't congestion some of which has now gone with the zone change. We are also still seeing problems with people driving to bald hills from North Lakes and Murrumba Downs due to the fare zone + highway). BCC/TMR are starting to col-de-sack streets and add additional traffic lights on Sandgate road as traffic has continued to rise from people accessing the station.

Also remember long distance travellers also prefer timetables so frequency becomes less of a problem. And with a car you aim to catch a particular service.

Buz is a poor example as we still see incredible amounts of duplication and waste in that network that cause black spots elsewhere on the network which still makes locals preferring to drive more.

Car travel is always going to be the fastest mode of travel as it's personalised travel. PT is not personalised travel and there are always costs (using pt and the cost to provide it), span of frequency but most importantly where you want to go.