robmausser
Senior Member
Why are we discussing freight? The whole idea of HFR is that these will be dedicated passenger tracks without freight on them.
|
|
|
VIA Rail
- Thread starter ehlow
- Start date
Urban Sky
Senior Member
Okay, so ready for Part 2 of this "modelling HFR travel times on the existing Havelock alignment" exercise (Refer to post #7,265 for Part 1)?
3. Modelling
3.1 Speed limits
As I've already covered in a previous post, applicable speed limits are mostly determined by the tracks curvature (in combination with the superelevation applied to the curves) and is determined with the following formula:
Quoted from: my Master Thesis (p. 60)
The resulting speed limits can be found in the following table:
Source: own calculations, first presented in post #7,260
In terms of the curves, I reviewed my .kmz file (here the newest version) and found another 10 curves (of which 2 have a radius of less than 550 meters), for a total of now 280 curves (or one every 1.4 km):
Source: own calculations with track geometry data estimated by using the circle drawing function built into Google Earth Pro
As you can see in above table, a superelevation of 5 inches only allows HFR trains to keep their top speed (assumed at 110 mph) in curves of the highest category (i.e. a radius of 3000 meters), which means that more than 90% of all curves require the train to slow down and that means in the case of just under 80% of all curves to slow down below 76.4 mph (i.e. the average speed which is required to achieve a travel time of exactly 3:15 hours between Ottawa and Toronto). Conversely, with a superelevation of 10 inches, HFR trains could also keep their top speed in curves with a radius of 2400 or 1700 meters and would only have to slow down in just under 80% of all curves and below the 76.4 mph in only just over 40% of all curves.
3.2 Uniform acceleration
Without access to specialist software (one of the main limitations of my Master Thesis!), the only realistic way to calculate travel times is to assume uniform acceleration, i.e. by distinguishing between three different types of movement: acceleration (where the train speed increases by exactly the acceleration value), constant movement (where the speed is unchanged) and breaking (where the train speed decreases by exactly the deceleration value). The main advantage is that the acceleration and breaking behavior is reduced to only two variables, which can be applied in the following two textbook formulae:
Quoted from: my Master Thesis (p. 77)
Despite ignoring factors like gradients, this still get's quite complicated, but anyone really interested in this can read about it in Chapter 6 of you-know-what...
3.3 Fixed blocks vs. variable block
An important point in my Thesis was to compare the effects of fixed vs. variable blocks on achievable train frequency, as train capacity was an important consideration. If you recall what I just explained to Paul about PZB vs. LZB, variable blocks are basically a system with stationary signals (which in Germany would be protected with PZB magnets) and a system without signals, where moving authority and speed limits are communicated directly to the train (which in Germany would be the responsibility of LZB). For our purposes, however, we are more concerned about static speed limits (i.e. those permanently imposed to reflect track curvature than those temporarily imposed by signals) than about how long a train remains in a block, and therefore, it would be more accurate to talk about whether we assume speed limits which apply to individual curves or zones (which combine multiple curves, like Paul suggested).
Given that VIA's future fleet is based on the Siemens Charger locomotives, which run for Amtrak and (when it resumes service again) Brightline with mandated PTC, which requires a continuous and proactive enforcement of speed-limits (i.e. like LZB and unlike PZB), I believe it is reasonable to assume that the block lengths for speed limits would be fixed rather than variable, which means that I calculate speed limits with a granularity of 80 meters (i.e. 20 blocks per mile) rather than variable blocks with lengths of anything between 100 meters and multiple kilometers, which means that applicable speed limits may change a dozen time within a minute of a train's runtime, which would be far beyond what a human driver could process and safely reflect in his choice of acceleration and deceleration commands to the train's traction motors and breaks...
***
Having laid out the fundamentals of modelling train runtimes, I will show how to implement and solve the model to estimate the travel times for the existing Havelock alignment:
4. Model Solving
In the meanwhile, please let me know if I lost you somewhere...
3. Modelling
3.1 Speed limits
As I've already covered in a previous post, applicable speed limits are mostly determined by the tracks curvature (in combination with the superelevation applied to the curves) and is determined with the following formula:
Quoted from: my Master Thesis (p. 60)
The resulting speed limits can be found in the following table:
Source: own calculations, first presented in post #7,260
In terms of the curves, I reviewed my .kmz file (here the newest version) and found another 10 curves (of which 2 have a radius of less than 550 meters), for a total of now 280 curves (or one every 1.4 km):
Source: own calculations with track geometry data estimated by using the circle drawing function built into Google Earth Pro
As you can see in above table, a superelevation of 5 inches only allows HFR trains to keep their top speed (assumed at 110 mph) in curves of the highest category (i.e. a radius of 3000 meters), which means that more than 90% of all curves require the train to slow down and that means in the case of just under 80% of all curves to slow down below 76.4 mph (i.e. the average speed which is required to achieve a travel time of exactly 3:15 hours between Ottawa and Toronto). Conversely, with a superelevation of 10 inches, HFR trains could also keep their top speed in curves with a radius of 2400 or 1700 meters and would only have to slow down in just under 80% of all curves and below the 76.4 mph in only just over 40% of all curves.
3.2 Uniform acceleration
Without access to specialist software (one of the main limitations of my Master Thesis!), the only realistic way to calculate travel times is to assume uniform acceleration, i.e. by distinguishing between three different types of movement: acceleration (where the train speed increases by exactly the acceleration value), constant movement (where the speed is unchanged) and breaking (where the train speed decreases by exactly the deceleration value). The main advantage is that the acceleration and breaking behavior is reduced to only two variables, which can be applied in the following two textbook formulae:
Quoted from: my Master Thesis (p. 77)
Despite ignoring factors like gradients, this still get's quite complicated, but anyone really interested in this can read about it in Chapter 6 of you-know-what...
3.3 Fixed blocks vs. variable block
An important point in my Thesis was to compare the effects of fixed vs. variable blocks on achievable train frequency, as train capacity was an important consideration. If you recall what I just explained to Paul about PZB vs. LZB, variable blocks are basically a system with stationary signals (which in Germany would be protected with PZB magnets) and a system without signals, where moving authority and speed limits are communicated directly to the train (which in Germany would be the responsibility of LZB). For our purposes, however, we are more concerned about static speed limits (i.e. those permanently imposed to reflect track curvature than those temporarily imposed by signals) than about how long a train remains in a block, and therefore, it would be more accurate to talk about whether we assume speed limits which apply to individual curves or zones (which combine multiple curves, like Paul suggested).
Given that VIA's future fleet is based on the Siemens Charger locomotives, which run for Amtrak and (when it resumes service again) Brightline with mandated PTC, which requires a continuous and proactive enforcement of speed-limits (i.e. like LZB and unlike PZB), I believe it is reasonable to assume that the block lengths for speed limits would be fixed rather than variable, which means that I calculate speed limits with a granularity of 80 meters (i.e. 20 blocks per mile) rather than variable blocks with lengths of anything between 100 meters and multiple kilometers, which means that applicable speed limits may change a dozen time within a minute of a train's runtime, which would be far beyond what a human driver could process and safely reflect in his choice of acceleration and deceleration commands to the train's traction motors and breaks...
***
Having laid out the fundamentals of modelling train runtimes, I will show how to implement and solve the model to estimate the travel times for the existing Havelock alignment:
4. Model Solving
- Ignoring s-curves
- Respecting s-curves
In the meanwhile, please let me know if I lost you somewhere...
Last edited:
Allandale25
Senior Member
There is still freight service between Agincourt and Havelock. It is much less frequent than what VIA deals with on the CN line. I assume it can remain if the timing is adjusted. cc @crs1026
Northern Light
Superstar
Another report on the continuing saga of HFR and the Trudeau government's timelines for a decision.
/cloudfront-us-east-1.images.arcpublishing.com/tgam/7YOKBKU4VNIXJFQZTQH6NF46WI.jpg)
www.theglobeandmail.com
The suggestion is that a decision is hoped for in time for a spring budget.
So another 5-6 months yet.
The article also reports on concerns with blockades of rail and an apparent increase in tampering w/rail, including signal systems.
Ottawa on track for decision on high-frequency trains between Toronto and Quebec City in ‘coming months’
Transport Minister Marc Garneau says the $4- to $6-billion plan would allow for more frequent trains, shorter trip times and better on-time performance
www.theglobeandmail.com
The suggestion is that a decision is hoped for in time for a spring budget.
So another 5-6 months yet.
The article also reports on concerns with blockades of rail and an apparent increase in tampering w/rail, including signal systems.
Last edited:
crs1026
Superstar
The current freight activity consists of one train in each direction three times a week, hauling mostly minerals from mines near Havelock. Until recently there were some industrial customers and grain elevators in and around Peterboro, but that business has pretty much dried up. One wishes it would return. As @smallspy notes, 50 loads per train (westbound) and 50 empties per train (eastbound). Empties won’t bother superelevated track much, and 150 loads a week may add to wear and tear but not significantly.
- Paul
reaperexpress
Senior Member
Okay, so ready for Part 2 of this "modelling HFR travel times on the existing Havelock alignment" exercise (Refer to post #7,265 for Part 1)?
3. Modelling
3.1 Speed limits
As I've already covered in a previous post, applicable speed limits are mostly determined by the tracks curvature (in combination with the superelevation applied to the curves) and is determined with the following formula:
Quoted from: my Master Thesis (p. 60)
The resulting speed limits can be found in the following table:
Source: own calculations, first presented in post #7,260
In terms of the curves, I reviewed my .kmz file (here the newest version) and found another 10 curves (of which 2 have a radius of less than 550 meters), for a total of now 280 curves (or one every 1.4 km):
View attachment 272255
Source: own calculations with track geometry data estimated by using the circle drawing function built into Google Earth Pro
As you can see in above table, a superelevation of 5 inches only allows HFR trains to keep their top speed (assumed at 110 mph) in curves of the highest category (i.e. a radius of 3000 meters), which means that more than 90% of all curves require the train to slow down and that means in the case of just under 80% of all curves to slow down below 76.4 mph (i.e. the average speed which is required to achieve a travel time of exactly 3:15 hours between Ottawa and Toronto). Conversely, with a superelevation of 10 inches, HFR trains could also keep their top speed in curves with a radius of 2400 or 1700 meters and would only have to slow down in just under 80% of all curves and below the 76.4 mph in only just over 40% of all curves.
3.2 Uniform acceleration
Without access to specialist software (one of the main limitations of my Master Thesis!), the only realistic way to calculate travel times is to assume uniform acceleration, i.e. by distinguishing between three different types of movement: acceleration (where the train speed increases by exactly the acceleration value), constant movement (where the speed is unchanged) and breaking (where the train speed decreases by exactly the deceleration value). The main advantage is that the acceleration and breaking behavior is reduced to only two variables, which can be applied in the following two textbook formulae:
View attachment 272254
Quoted from: my Master Thesis (p. 77)
Despite ignoring factors like gradients, this still get's quite complicated, but anyone really interested in this can read about it in Chapter 6 of you-know-what...
3.3 Fixed blocks vs. variable block
An important point in my Thesis was to compare the effects of fixed vs. variable blocks on achievable train frequency, as train capacity was an important consideration. If you recall what I just explained to Paul about PZB vs. LZB, variable blocks are basically a system with stationary signals (which in Germany would be protected with PZB magnets) and a system without signals, where moving authority and speed limits are communicated directly to the train (which in Germany would be the responsibility of LZB). For our purposes, however, we are more concerned about static speed limits (i.e. those permanently imposed to reflect track curvature than those temporarily imposed by signals) than about how long a train remains in a block, and therefore, it would be more accurate to talk about whether we assume speed limits which apply to individual curves or zones (which combine multiple curves, like Paul suggested).
Given that VIA's future fleet is based on the Siemens Charger locomotives, which run for Amtrak and (when it resumes service again) Brightline with mandated PTC, which requires a continuous and proactive enforcement of speed-limits (i.e. like LZB and unlike PZB), I believe it is reasonable to assume that the block lengths for speed limits would be fixed rather than variable, which means that I calculate speed limits with a granularity of 80 meters (i.e. 20 blocks per mile) rather than variable blocks with lengths of anything between 100 meters and multiple kilometers, which means that applicable speed limits may change a dozen time within a minute of a train's runtime, which would be far beyond what a human driver could process and safely reflect in his choice of acceleration and deceleration commands to the train's traction motors and breaks...
***
Having laid out the fundamentals of modelling train runtimes, I will show how to implement and solve the model to estimate the travel times for the Havelock alignment:
4. Model Solving
- Ignoring s-curves
- Respecting s-curves
In the meanwhile, please let me know if I lost you somewhere...
Sounds good to me! What superelevation are you thinking of using? My vote is for 8" (5+3) west of Havelock, based on the Brightline precedent, and 9 or 10" (6 or 7 + 3) east of Havelock based on the limits set by Transport Canada.
I'm a bit surprised to hear that Amtrak's PTC system doesn't support variable blocks, but a granularity of 80 metres is still pretty decent - that's less than the length of an HFR train. It reminds me of a presentation I watched from someone in ProRail (the Dutch national railway management company) about their work to introduce moving-block signaling under ETCS Level 3 while still supporting some trains (freight trains) which lacked the safety systems required for purely communications-based control. They pointed out that in practice, a fixed-block system with extremely short blocks is functionally identical to a moving-block system, because even with moving blocks, the train only updates its movement authority periodically anyway (perhaps once per second). With very small fixed blocks it's easier to be backwards-compatible with older trains. You can mostly rely on tiny virtual blocks, which are then grouped into large physical blocks monitored by axle counters. Equipped trains can make full use of all the virtual blocks, while non-equipped trains can still operate on the line, albeit at a much longer headway due to their long blocks.
crs1026
Superstar
^(Digressing, but only slightly) I don’t have an opinion on which traffic control system is best for VIA, but - in practice - the control of authority over opposing trains in single track will matter more than the ability to run trains closer together at speed. If frequency reached even one train each way per half hour, there will be a significant number of meets at sidings. The signalling/control system will need to plan these meets to minimise delays. Do you hold the westbound train at a siding for an eastbound, or do you put the eastbound in the next siding to the west and advance the westbound? With slower moving freights, this is currently done (imprecisely) in the RTC’s head, with ”hot” trains taking priority, and others forced to wait - but for single track HFR, an automated solution is required.
Both fixed and moving block systems can work provided the spacing of the “approach” blocks at sidings are optimised - so trains slow down for meets at the last possible (safe) moment. It’s a game of “chicken” - with appropriate safety margins, of course.
This isn’t a concern for the modelling discussion, but the point is, for HFR, block spacing for approach to fixed meeting points likely matters more than block spacing for frequent movements in a single direction. (Except, perhaps, if GO service happened near the GTA). An HFR with RER level headways is a problem I’d love to have!
- Paul
Both fixed and moving block systems can work provided the spacing of the “approach” blocks at sidings are optimised - so trains slow down for meets at the last possible (safe) moment. It’s a game of “chicken” - with appropriate safety margins, of course.
This isn’t a concern for the modelling discussion, but the point is, for HFR, block spacing for approach to fixed meeting points likely matters more than block spacing for frequent movements in a single direction. (Except, perhaps, if GO service happened near the GTA). An HFR with RER level headways is a problem I’d love to have!
- Paul
Last edited:
reaperexpress
Senior Member
Yes of course. Given that it's going to be a single-tracked line, the only relevance that the block length has to HFR is the granularity at which speed limits can be applied while still being monitored by the train supervision system. I was sharing the reference to L3 ETCS in NL purely in that context, not the context of running trains close together.
Allandale25
Senior Member
Ottawa on track for decision on high-frequency trains between Toronto and Quebec City in ‘coming months’
Transport Minister Marc Garneau says the $4- to $6-billion plan would allow for more frequent trains, shorter trip times and better on-time performance
www.theglobeandmail.com
kEiThZ
Superstar
It lines up with the original JPO timeline but I hate how this is being dragged out. My fear is that a change of government will kill this.
micheal_can
Senior Member
I doubt we will see a federal election anytime soon. The NDP have more to gain by leaving them in power then going to an election.
kEiThZ
Superstar
This. I can't believe how fragile even even the HFR idea is. It really makes me sad for this country.
afransen
Senior Member
Yes. Canada really won't regret investing in HSR between Toronto - Ottawa - Montreal.
Urban Sky
Senior Member
Canada will regret if it continues trying to achieve HSR while skipping the creation of a frequent and semi-fast intercity rail network, which was the very foundation on which HSR was built on in every single HSR nation, save Saudi-Arabia and Uzbekistan...
nfitz
Superstar
Montreal-Toronto seems faster now, than the non-high-speed portion of the Seoul-Busan journey, when I took it 15 years ago, when it didn't yet get all the way to Busan yet.