Is high train braking force better? Not really...

Post all other topics which do not belong to any other category.
Bacon117
Burner Inserter
Burner Inserter
Posts: 11
Joined: Mon Apr 01, 2019 10:31 pm
Contact:

Is high train braking force better? Not really...

Post by Bacon117 »

Summary:
Based on my investigation, a higher train braking force is not necessarily better. Based on the scenarios I tested, the general conclusion is this:
  1. High brake force benefits short train trips. The benefit decreases with longer train trips.
  2. At a given intersection (simple intersection), if a train has no conflict with any other train OR if a train is going to come to a complete stop regardless of the brake force, go with a high braking force. (This scenario doesn't always happen, keep in mind)
  3. At a given intersection, if a train will conflict with another train, it is BETTER for the train with the red signal to slow down so that it reaches the signal with some speed remaining, rather than stopping completely.
In a base where trains will interfere with each other at an intersection, it is better for a train to arrive at a stop signal with some speed when the signal opens, rather than being at a complete stop when the signal opens. A lighter braking force causes the train to arrive at the intersection a little later than a train with higher braking force, so that gives a chance for the intersection to clear before the train has to come to a complete stop. While this scenario may not happen all the time, the benefits of a higher brake force are not really THAT great to begin with.

Details of Test Setup:
One day, I was watching my trains move around my base (as one does), and I started thinking about how trains interact at an intersection. I know in real life, if you come to a stop light, it’s better to time it so that you reach the light with some speed, rather than speeding up and coming to a complete stop at the light. This should be true in factorion, so I wanted to investigate. I used a timer and various scenarios to see the time improvement vs brake force. I used the console to set the brake force between 0 and 2.5. I did look at other bonuses, but there isn’t any change <0. There is a little bit between 2.5 and 3, but not enough to spend the time to worry about it)

I am using a simple roundabout for my base. I know it’s not the best, but it hasn’t hurt me YET. I set up a comparison with various scenarios and used naknak’s really awesome train timer to measure the effect of changing the brake force bonus via the console. (viewtopic.php?t=53911)
I performed 4 scenarios and timing each of them and made a table based on the time. The four scenarios are:
  1. Measure the time of ONE 1-4 train station to station with a distance of about 600 tiles.
  2. Measure the time of TWO 1-4 trains station to station with a distance of about 600 tiles.
  3. Measure the time of TWO 1-4 trains across the roundabout only.
  4. Measure the time of ONE 1-4 train station to station with a distance of about 8300 tiles.
Here is a picture of my round about and my setup from the map view. When I switch a constant combinator on or off, the trains will swap stations (from West A to East B, for example) and cross the map.
Roundabout.png
Roundabout.png (3.41 MiB) Viewed 5730 times
Map View.png
Map View.png (11.77 KiB) Viewed 5730 times
Details of results
So, what are the results? Scenario 1, with one train travelling ~600 tiles had the most improvement with higher brake force. So with a brake force bonus set to 2.5 the time was faster than a bonus of 0 by about 14%. HOWEVER, from scenario 4, with one train traveling about 8300 tiles, the improvement was only 2% from a 0 bonus to a 2.5 bonus. This means that higher brake force really only benefits short train trips. Technically, the larger the ratio of brake time to total travel time, the more effective higher brake force is.

When comparing station to station time with TWO 1-4 trains travelling ~600 tiles and meeting at the roundabout at about the same time, the improvement in time isn’t that great. Scenario 2 shows this, with a 2.9% improvement between a bonus of 0 and 2.5. The reason for this small improvement is due to the overall improvement station to station of higher brakes, but the considerable reduction in time to go through the roundabout (Scenario 3). Scenario 3 measures the time change of one train through the roundabout. This train is the one that has to wait at the intersection. What this means is that it is FAR better for a train to approach an intersection slower so that when the track becomes free, the train has not come to a complete stop. With a brake force bonus of 2.5, the train got through the roundabout ~15% slower than with a brake force of 0.

Below is a chart of the percent difference in time vs. a change in brake force. A higher percentage means the time has improved (is faster) compared to the case of brake force bonus = 0.
PD Chart.png
PD Chart.png (21.64 KiB) Viewed 5730 times
Conclusion
So, the conclusion is that a higher brake bonus is better in some scenarios, but worse in others. It is worse for a train to come to a complete stop at an intersection rather than coming to the intersection with some speed. In my opinion, improving the situation when trains interact at intersections far outweighs the improvements that a higher brake force bonus does provide.

What do we do about it?
For now, don't research higher brake forces. Ideally, if a train is approaching a red signal, the train should slow to a speed that will allow the train to arrive at the braking point of that red signal just when the signal turns green, so that the train is travelling at maximum POSSIBLE speed when it crosses the signal rather than being at a complete stop. I don’t know if a mod can do this, because it could depend on many trains approaching a given intersection. But, that’s how to fully optimize trains when they interfere at an intersection.

It took me a few days to do this to see if brake force research is really better. So I hope it helps someone, I am certainly open to questions.

Thanks!

User avatar
TheRangerLOL
Long Handed Inserter
Long Handed Inserter
Posts: 74
Joined: Fri Sep 18, 2020 11:56 pm
Contact:

Re: Is high train braking force better? Not really...

Post by TheRangerLOL »

Why is this even a thing?
The one and only.

Qon
Smart Inserter
Smart Inserter
Posts: 2091
Joined: Thu Mar 17, 2016 6:27 am
Contact:

Re: Is high train braking force better? Not really...

Post by Qon »

Interesting post, thanks :)

I guess then that coal powered trains with single locomotive and many wagons (lower acceleration) would be more hurt by high breaking force then nuclear fueled trains with a high locomotive ratio.

User avatar
Khagan
Fast Inserter
Fast Inserter
Posts: 232
Joined: Mon Mar 25, 2019 9:40 pm
Contact:

Re: Is high train braking force better? Not really...

Post by Khagan »

Bacon117 wrote:
Sat Oct 10, 2020 2:55 am
I am using a simple roundabout for my base. I know it’s not the best
It would be much better if it were properly signalled: with the loop divided into quadrants by chain signals, your two sample trains wouldn't even have to stop at all.
This one is signalled for LHT, but you just have to flip the signals to get the RHT version:
Roundabout.png
Roundabout.png (1.14 MiB) Viewed 5661 times
As far as the actual thread topic is concerned, while it may be true in some specific circumstances that it would be better to brake more gently, don't forget that the distance ahead that a train has to reserve, and therefore the likelihood of it blocking other trains in the first place, is reduced by having a shorter braking distance.

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

Measuring the time 1 train at a time take for a certain trip is not enough to reach the conclusion. (imo)

If you take the analogy with car on the highway you could reason not only for 1 particular vehicule, but for a general flow, either of many vehicules, or at 1 specific portion.

1rst situation, the braking force is low: This means for car (like for train) that the distance between 2 vehicule is high when they have full speed. Like in a highway the security distance is high between 2 cars, and for trains it means that the amount of block in front of one train that are reserved is high. (Many yellow signals ahead).

2nd situation the braking force is high : This means that the distances between 2 vehicule is much lower. For car it would be with AI-controlled car, where there could be much smaller distances between all cars due to not relying on human reflexes for security braking but the faster reaction time of the machine that is aware of each car speed and can adapt the 10th car speed based on a slow-down of the 1rst one (the example of connected car transmitting to each other).

For trains it gets a bit weird, there could be fewer blocks reserved by each train in front of them ( less yellow signals) as Khagan says, but i'm not sure the game acts this way.
Khagan wrote:
Sat Oct 10, 2020 10:01 am
don't forget that the distance ahead that a train has to reserve, and therefore the likelihood of it blocking other trains in the first place, is reduced by having a shorter braking distance.
What does happen though is that a 2nd train trying to reserve a block already reserved by a 1rst train ( attempt at superposition of 2 yellow state ) will cause the 2nd train to slow down and re-accelerate once the 1rst train has cleared the junction. And it will occur from a larger distance to the junction in case of high speed.

This may add nuances to you statement :
Bacon117 wrote:
Sat Oct 10, 2020 2:55 am
A lighter braking force causes the train to arrive at the intersection a little later than a train with higher braking force, so that gives a chance for the intersection to clear before the train has to come to a complete stop.
Yes, but is this an effect of the braking force itself, or is this an effect of a lower speed from the 2nd train when crossing the junction ?

I would say that in your particular testing setup it is mostly synonymous because the lower braking force translate into a higher speed for the 2nd train when it cross junction. And the opposite with higher braking force. But it is not always the case.

Which could contredict this intuition in some cases :
Qon wrote:
Sat Oct 10, 2020 8:15 am
I guess then that coal powered trains with single locomotive and many wagons (lower acceleration) would be more hurt by high breaking force then nuclear fueled trains with a high locomotive ratio.
Here you can analyze differently, what you increase is momentum. A train with 1 loco and many wagons would also have a higher braking distance, hence a larger number of yellow signals in front of it, which means the moment the 2nd train start to slow down ( superposition of yellow state) is when it is much further away from the junction, it can or not give it enough time to accelerate again because of this distance despite its reduced acceleration. depending on the ratio between acceleration and braking capacity !.
Bacon117 wrote:
Sat Oct 10, 2020 2:55 am
So, the conclusion is that a higher brake bonus is better in some scenarios, but worse in others. It is worse for a train to come to a complete stop at an intersection rather than coming to the intersection with some speed. In my opinion, improving the situation when trains interact at intersections far outweighs the improvements that a higher brake force bonus does provide.
The balden part i think is true at all time. Which explains the fact that the underlined one is sometimes better, sometimes worse because the braking force doesn't always have the same effect on the speed at which a junction is crossed ON AVERAGE which is a much more important factor. The faster the junction is cleared the better, always.

For this topic i found some times ago lectures on traffic flow optimization for cars , they are quite more complex than i expected. And it changes a lot depending on wether or not all trains are the same LCCCC or if you have various trains size/acceleration. ( cars + trucks ). The distribution of cars and trucks impact the optimal speed limit for traffic.

Which could be analyzed as : the difference of braking distance require different speed for maximal flow.

The diagram that represent the flow on the vertical axis, and the speed on the horizontal one, would look like a hill, or an inverted U / inverted V. The smaller the speed the lower the flow at first. When increasing speed , the flow increases at first. It reaches a peak, and then the flow reduce, because the distance between 2 vehicules increases too much. ( for cars driven by humans the distance increase more than linearly with speed because of the increased distance the car travel before your reflexes makes you brake).

This has pratical and visible result : sometimes when you have many many cars on a highway they are much slower than maximal speed yet you have better flow overall if you count how many cars clear a portion of highway per hour. (than if each car was faster individually but more spaced out).

Apart from that junction analysis, i can see other reason to analyze the flow differently than taking 1 train trip time as reference.

The traffic jam effect, where one car braking too fast. It then reach a "local minimum speed", and cause the next one to reach an even lower "local minimum speed" and it cascade off to the point where one car end up in a complete stop. and all others behind that. This would play poorly in favor of the high braking force ONLY in case of lower acceleration than braking capacity !. ( which relate directly to Qon's intuition about momentum ).

Another point this time in FAVOR of the high braking force is the distance between the destination and the previous junction. With a higher braking force you preserve your speed until the very last moment, which could make ON AVERAGE , trains to clear a particular junction faster. Especially at low level of braking force and high momentum trains. This in the game translate as a train that start slowing down in the main network potentially triggering the traffic jam effect rather than having enough distance to decelerate without impacting the main network. In this case it [EDIT: can sometimes] increases the effiency of an already established network to increase the braking power. which is the case we encounter most of the time when we play.[EDIT: it depend on the network]


I don't have definite answer for "in general", because i think it highly depend on particular situation. hence why i think the test you did can't be conclusive.

It does help casting light on the complexty of the task though and provide good example of why it is not straighforward :D

A mod that could help/ be fun to test with is LTN. This way you could measure the time an arbitrary amount of request needs to be fulfilled. Given different networks, they would react differently to an increase of braking force. Averaging the differences of each travel times could give an answer for each network "globally", and then you could measure different junctions or portion throughput "individually" that would have mixed results. That would be closer to how it's done in real life ( as far as i can tell from the lesson i've read on traffic engineering).

Every junction has a different impact on the network; a main junction leading to multiples unload area in close proximity has more weight compared to a less populated junction in the middle of a long route to outpost for the perfomance of the network taken globally. The main junction may be impacted positively by an increase in braking force because of the "preserve speed" and "avoid traffic jam" effect, while the many lone away junction, negatively because of what is shown in your test. It is hard to give a precise answer/estimation. Hence the idea of doing a survey to car-travelers ( average of each travel time for each train) AND measure the flow at 1 junction ( amount of car/train that goes through in 1 hour).

sorry for long post but i wanted to respect the dedication that was shown for initial test :), please correct me if you think i'm wrong.
Last edited by mmmPI on Mon Oct 12, 2020 3:36 pm, edited 1 time in total.

astroshak
Filter Inserter
Filter Inserter
Posts: 597
Joined: Thu May 10, 2018 9:59 am
Contact:

Re: Is high train braking force better? Not really...

Post by astroshak »

Regardless of how fast a train was going, it wants to come to a complete stop at a Red signal.

For the sake of illustration, lets say a train takes 10 track tiles to slow down 10 KPH.

If it is traveling at 200 KPH, it will take 200 tiles to hit 0. Thus, it has to reserve that many signals ahead, and start breaking that far out.
On the other hand, if it takes 5 tiles to slow down 10 KPH, then it will only take 100 tiles to stop from 200 KPH. So it can travel an extra 100 track tiles at full speed, arriving there faster.

Lets say that the train signal clears when the train is 50 tiles away, so the train does not need to come to a complete stop?

10 tiles/10 KHP deceleration means it has travelled 150 tiles, and slowed to 50 KPH, when the signal clears and it can accelerate back to normal speed.
5 tiles/10 KPH means it has travelled 100 tiles at full speed, slowed to 100 KPH when the signal clears, and it can accelerate back to normal speed.

The more time a train spends braking, the less time it spends accelerating to, or at, full speed. The more times a train is required to brake, the more the research helps. If a train is not required to brake at all prior to approaching its destination, then improved brakes won’t matter much at all!

There should not be any circumstances where a lower braking force is better than a higher one. A lower braking force won’t allow a train to approach a red signal with a higher reserve speed (due to incomplete stopping) than a higher braking force; it is the other way around due to a train being able to maintain higher speed longer with a higher braking force than a lower braking force.

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

astroshak wrote:
Sat Oct 10, 2020 3:08 pm
For the sake of illustration, lets say a train takes 10 track tiles to slow down 10 KPH.
I don't know how the game handle it, but in real life it's not like this.

The deceleration is usually in proportion of the total speed at "high" speed. So you lose X % of speed per second while breaking, until a certain point.

Acceleration on the other hand is the opposite, the last 10 km/h to reach maximum speed takes a whole more time to get than the first 10 km/h to get moving.

This means that if a train breaks for 2 sec, it may need 1 second or 10 second to accelerate back to its previous speed. Depending on said speed.

I think it could gets worse if the braking is stronger. Since (re-)acceleration is capped by train composition and fuel.

This phenomenon can trigger the traffic-jam effect. Which then can impact negatively the whole network depending on existing traffic.

Maybe it can explain partly those circumstances.

astroshak
Filter Inserter
Filter Inserter
Posts: 597
Joined: Thu May 10, 2018 9:59 am
Contact:

Re: Is high train braking force better? Not really...

Post by astroshak »

Deceleration is negative acceleration. It is not percentage based of the top or current speed.

Acceleration is an increase in speed over time. Gravity on Earth accelerates objects towards the earth at 9.8 meter/sec/sec, aka 9.8 meters/(sec^2).

If a train in Factorio is traveling at 200 KPH, and one track tile is 2x2 meters square, then it is traveling 100,000 track tiles per hour, or 27.78 (rounded) track tiles per second (55.56 meters/sec).

If you put a negative acceleration, of -X meters/sec/sec, say -5 meters/sec/sec, it will come to a complete stop between 11 and 12 seconds later. 55.56 meters/sec at 0 seconds, 50.56 meters/sec at 1 second, 45.56 meters/sec at 2 seconds, and so on until 0.56 meters/sec at 11 seconds and been stopped at 12 seconds.

That’s assuming a constant deceleration. Saying its -x% up to a certain point is assuming a non-constant deceleration. We commonly use varying amounts of deceleration while driving, based on how hard we feel we need to hit the brake pedal. Your braking habits should not be assumed to be the case for Factorio. The trains are programmed to go as fast as they can for as long as they can, which means, maximum deceleration to come to a stop in the shortest amount of space and time. That means, a constant deceleration rate, rather than a variable deceleration rate.

Unless a dev wants to chime in on it. A brief look at the wiki found no information on train braking at all other than the 7 vanilla tech levels, their costs, bonus, and cumulative bonus.

Theikkru
Filter Inserter
Filter Inserter
Posts: 353
Joined: Wed Mar 27, 2019 2:18 pm
Contact:

Re: Is high train braking force better? Not really...

Post by Theikkru »

astroshak wrote:
Sat Oct 10, 2020 3:08 pm
[...]
For the sake of illustration, lets say a train takes 10 track tiles to slow down 10 KPH.

If it is traveling at 200 KPH, it will take 200 tiles to hit 0. Thus, it has to reserve that many signals ahead, and start breaking that far out.
On the other hand, if it takes 5 tiles to slow down 10 KPH, then it will only take 100 tiles to stop from 200 KPH. So it can travel an extra 100 track tiles at full speed, arriving there faster.

Lets say that the train signal clears when the train is 50 tiles away, so the train does not need to come to a complete stop?

10 tiles/10 KHP deceleration means it has travelled 150 tiles, and slowed to 50 KPH, when the signal clears and it can accelerate back to normal speed.
5 tiles/10 KPH means it has travelled 100 tiles at full speed, slowed to 100 KPH when the signal clears, and it can accelerate back to normal speed.
[...]
There should not be any circumstances where a lower braking force is better than a higher one. A lower braking force won’t allow a train to approach a red signal with a higher reserve speed (due to incomplete stopping) than a higher braking force; it is the other way around due to a train being able to maintain higher speed longer with a higher braking force than a lower braking force.
Your comparison is flawed because you're assuming that the signal clears when the two trains are at the same distance from the signal, but in the vast majority of cases the comparison should be between identical points in time (because the signal is being blocked by something time-dependent, such as another train passing through the junction, in Factorio, or a timer on a traffic light, IRL). This means that the train with more braking force will have traveled further towards the signal, and likely braked more than the slower one. Additionally, your simulation uses a distance-based deceleration, which is strange. A linear deceleration would be time-based, not distance-based.
If we rerun the hypothetical using linear deceleration (5tiles/sec² and 10tiles/sec²) and assume that the signal clears when the slower-braking train reaches 50tiles/sec, the math is as follows:
  • Set initial speeds to 200tiles/sec
  • Slowtrain will start decelerating 200÷5=40 seconds before the junction, and will reach 50tiles/sec (200-50)÷5=30sec after deceleration starts.
  • Set simulation start at 30 seconds before signal clears, when slowtrain starts decelerating.
  • Slowtrain will start decelerating (200+0)÷2×40=4000tiles before the signal, and will be (50+0)÷2×(40-30)=250tiles from the signal when the signal clears.
  • This means both trains are starting 4000tiles before the junction.
  • Fasttrain will start decelerating 200÷10=20sec before the junction, and will need (200+0)÷2×20=2000tiles to do so, so it will continue traveling for (4000-2000)÷200=10sec at full speed after the simulation starts.
  • Fasttrain will then decelerate for the next 20 seconds, covering the remaining 2000tiles, and reach a full stop just as the signal clears.
Result: Slowtrain will be 250tiles away from the signal at 50tiles/sec, while fasttrain will be at the signal at 0tiles/sec. Unless the trains have a far higher acceleration than deceleration, slowtrain will pass fasttrain in about 5 seconds.

Qon
Smart Inserter
Smart Inserter
Posts: 2091
Joined: Thu Mar 17, 2016 6:27 am
Contact:

Re: Is high train braking force better? Not really...

Post by Qon »

mmmPI wrote:
Sat Oct 10, 2020 2:04 pm
Which could contredict this intuition in some cases :
Qon wrote:
Sat Oct 10, 2020 8:15 am
I guess then that coal powered trains with single locomotive and many wagons (lower acceleration) would be more hurt by high breaking force then nuclear fueled trains with a high locomotive ratio.
Here you can analyze differently, what you increase is momentum. A train with 1 loco and many wagons would also have a higher braking distance, hence a larger number of yellow signals in front of it, which means the moment the 2nd train start to slow down ( superposition of yellow state) is when it is much further away from the junction, it can or not give it enough time to accelerate again because of this distance despite its reduced acceleration. depending on the ratio between acceleration and braking capacity !.
Apart from that junction analysis, i can see other reason to analyze the flow differently than taking 1 train trip time as reference.

The traffic jam effect, where one car braking too fast. It then reach a "local minimum speed", and cause the next one to reach an even lower "local minimum speed" and it cascade off to the point where one car end up in a complete stop. and all others behind that. This would play poorly in favor of the high braking force ONLY in case of lower acceleration than braking capacity !. ( which relate directly to Qon's intuition about momentum ).
Assuming wagons have breaks and contribute as much to decelarating force as locomotives:
Locomotives are heavier so would impact breaking negativly (compared to equal length but higher % wagon train). This is the opposite of what you say though. So wagons don't have breaks?

Continuing with the assumption anyways for now:
But while all wagons and locomotives contribute to weight and breaking, only locomotives accelerate the train. So breaking distance should increase with more locomotives (but same length), but the effect wouldn't be nearly as dramatic as the increase in acceleration.

Another point:
But higher accel fuel doesn't increase momentum (comparing equal speed trains) so the breaking distance should be equal while the acceleration changes.

And isn't acceleration usually lower than deceleration?

astroshak
Filter Inserter
Filter Inserter
Posts: 597
Joined: Thu May 10, 2018 9:59 am
Contact:

Re: Is high train braking force better? Not really...

Post by astroshak »

Momentum likely plays a part, in the vein of a larger or heavier train taking more time and distance to stop. Remove this from consideration by considering equal trains, such as only looking at the effects of braking force tech on LLLCCCCCCCC trains, or whatever train configuration you want to use.

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

I had anticipated those questionning and was doing the study when you posted. I hope it helps correct mistakes we all made :).

I did the Acceleration test on a looping circuit with 2 station A and B, train schedule to A, then B, without waiting condition. ( assuming that the game doesn't make any difference wether it's a curve rail or not).

I did the deceleration test on a straight line by placing a red signal and asking for a train to go past it making sure it had enough room to get full speed before it start braking.

I just went frame by frame and reported the measured speed in a spread sheet.

Here are my naming convention for the following charts :

3NL => 3 Nuclear Loco
3WL => 3 Wooden Loco
1NL => 1 Nuclear Loco
1NLCCCCC => 1 Nuclear Loco 4 Cargo-wagons
speedontimeaccel.jpg
speedontimeaccel.jpg (79.83 KiB) Viewed 5297 times
The acceleration is closer to linear if you use 3 nuclear loco.
Only 1 shows a little curvature when trying to reach the max speed.
The worse the fuel, the slower acceleration, and also the less linear.
The more locos, the closer to linear.
Adding wagons behind a loco reduce acceleration.
speedvarontimeaccel.jpg
speedvarontimeaccel.jpg (90.26 KiB) Viewed 5297 times
We can clearly see that acceleration is not linear.


speedontimedecel.jpg
speedontimedecel.jpg (93.02 KiB) Viewed 5297 times
3 nuclear loco, or just 1 nuclear loco, have similar max speed and braking behavior. (the orange and yellow line are almost identical).
3 wooden loco has less max speed but the braking behavior is similar to the 3 nuclear loco or the single nuclear loco.
The trains with many wagon shows "higher momentum" as it requires more time to come to complete stop that the other trains starting with same initial speed.

All behavior are close to linear. ( at least closer than acceleration).
speedvarontimedecel.jpg
speedvarontimedecel.jpg (119.46 KiB) Viewed 5171 times
deceleration isn't really linear either; only close to it.

It is best shown for the long train, it loses between 0.8 and 0.9 km/h/tick. Due to rounding i guess it alternate each tick between the two value. The frequency of the oscillation changes overtime to favor losing LESS km/h/tick, which become more apparent at low speed but is present during the whole test.
The braking behavior shown for 3NL 3WL or 1 NL is the same , the wooden one starting with less speed.
The long train has a different behavior, as its deceleration is lower.
astroshak wrote:
Mon Oct 12, 2020 1:09 pm
Momentum likely plays a part, in the vein of a larger or heavier train taking more time and distance to stop. Remove this from consideration by considering equal trains, such as only looking at the effects of braking force tech on LLLCCCCCCCC trains, or whatever train configuration you want to use.
It seems that a trains that has more wagons will take more time to stop than a train going at the same speed without wagons.
Qon wrote:
Mon Oct 12, 2020 8:46 am
Assuming wagons have breaks and contribute as much to decelarating force as locomotives:
Locomotives are heavier so would impact breaking negativly (compared to equal length but higher % wagon train). This is the opposite of what you say though. So wagons don't have breaks?
Wagons don't seem to have brake;
I don't know if the test i have performed can help conclude on the first part of your statement but it can very well be the case that i made a mistake while talking about "momentum" as it would be commonly understood using the weight of trains, please point it out if so, my nose is too close to the numbers atm to see them properly.

Qon wrote:
Mon Oct 12, 2020 8:46 am
Continuing with the assumption anyways for now:
But while all wagons and locomotives contribute to weight and breaking, only locomotives accelerate the train. So breaking distance should increase with more locomotives (but same length), but the effect wouldn't be nearly as dramatic as the increase in acceleration.
It seems that no matter how many loco, the train brakes at the same speed, each of them probably counter-acting its own weight by its own braking contribution.
Wagons however don't seem to contribute to reducing the braking distance, but the opposite.

Qon wrote:
Mon Oct 12, 2020 8:46 am
Another point:
But higher accel fuel doesn't increase momentum (comparing equal speed trains) so the breaking distance should be equal while the acceleration changes.

And isn't acceleration usually lower than deceleration?
Higher accel fuel also increases max speed for a train, which makes the braking distance change. (how to test your proposition ?).

Acceleration is usually lower than deceleration: not always, the 3nuclear loco trains reaches max speed in 123 ticks while it needs 274 to fully stop when going max speed.
th first second, acceleration is between 2.6 and 2.7 km/h/tick, and deceleration is only between -1.1 and -1.2 km/h/tick.
mmmPI wrote:
Sat Oct 10, 2020 4:02 pm
The deceleration is usually in proportion of the total speed at "high" speed. So you lose X % of speed per second while breaking, until a certain point.
No. Deceleration is higher when speed is higher, but it's very subtle unless speed is low in the game.

IRL it doesn't only depends on (hypothetical) driving habits but also air drag, and friction, and many other factors i guess, which implies a relation between the actual speed of a vehicule and those mentionned factors such as air resistance having a stronger impact at high speed than low speed.

Air resist seems to be a constant in game, based on the shape of the 1rst rolling stock.

I couldn't find any equation from the wiki to define the deceleration. The only formula i could get is the one that describe the movement of an accelerating/full speed train:
train_speed = max(0, abs(train_speed) - train_friction_force ÷ train_weight)
train_speed = train_speed + (10 × number_of_locomotives_in_moving_direction × fuel_acceleration_bonus ÷ train_weight)
train_speed = train_speed × (1 - air_resistance_of_front_rolling_stock ÷ (train_weight ÷ 1000))

Where train_friction_force is the summed up friction of each wagon and locomotive and train_weight is the summed up weight of each wagon and locomotive, see their individual pages for the weight values. The friction and air resistance of wagons and locomotives can be found in their prototypes. The calculated train_speed is capped to max_speed = 1.2 * fuel_top_speed_multiplier
maybe (hopefully) someones understand them better than me :)
Last edited by mmmPI on Mon Oct 12, 2020 8:39 pm, edited 4 times in total.

User avatar
Nosferatu
Fast Inserter
Fast Inserter
Posts: 228
Joined: Fri Jan 20, 2017 4:48 pm
Contact:

Re: Is high train braking force better? Not really...

Post by Nosferatu »

I suggest you test this using the mod and setup they use in this thread: viewtopic.php?f=194&t=46855

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

Nosferatu wrote:
Mon Oct 12, 2020 3:48 pm
I suggest you test this using the mod and setup they use in this thread: viewtopic.php?f=194&t=46855
I thought of it but i'd rather test something else first if i can find the time:

The deceleration shown on previous chart is what i would call "natural". The train brake when its braking point reach an obstacle.

BUT : If the obstacle appears in between the train and its braking distance, the trains suddenly is able to decelerate a lot faster, to the point where you can even have an instantaneous stop. This happens when you have a bunch of regular signals that you turn red using circuit just in front of a full speed train.

It may or may not happen depending on the junction that a train blocking the path of another train force the 2nd into a faster deceleration due to the signaling.

i want to investigate more on very simple things first so as to identify relevant parameter to test on a larger scale such as junction.

Also the location of the junction inside a wider network impact the flow in the junction itself because of the speed, inter-train distance, frequency of path crossing, lengh of train, lengh of train related to their direction, proximity of destination, proximity of other junction, and so on ...

so while a junction A could proove better in a test than junction B it doesn't mean that in a particular set of condition the opposite never occurs.

For example a junction could have better overall throughput, AND be balanced so each sides backs up evenly, allowing that overall optimal throughput.
But, be used in a network where train distribution cause 1 side to receive more faster and smaller trains, and another side less longer and slower train. Which would then cause another junction to jam or throw away the optimal flow. But another junction would have distributed the load differently due to , in this particular setup, with this particular speed and train composition, giving a priority to another side of the junction, causing no jam, or causing jam in a non-busy area.

This is very hard to test and attach the result to one particular junction, instead i prefer looking for some sort of list of those interaction that could impact the results. Hopefully giving clues to understand the complex behavior of every junction; braking force is one of those variable with an effect that cascade of to many other variables, so is acceleration, train composition, network topology, junction design amongst others :)

SoShootMe
Filter Inserter
Filter Inserter
Posts: 472
Joined: Mon Aug 03, 2020 4:16 pm
Contact:

Re: Is high train braking force better? Not really...

Post by SoShootMe »

mmmPI wrote:
Mon Oct 12, 2020 4:43 pm
BUT : If the obstacle appears in between the train and its braking distance, the trains suddenly is able to decelerate a lot faster, to the point where you can even have an instantaneous stop. This happens when you have a bunch of regular signals that you turn red using circuit just in front of a full speed train.

It may or may not happen depending on the junction that a train blocking the path of another train force the 2nd into a faster deceleration due to the signaling.
Huh, I hadn't realised you could get trains to decelerate a lot faster except "instant stop" (eg mine a rail on a train's only path)... But I don't think this should ever happen in normal circumstances, because when a train's stopping point reaches a signal it will either reserve the block (or blocks) after it or, if it can't because the block (or one of the blocks) is already reserved/occupied, it will start braking.

Yoyobuae
Filter Inserter
Filter Inserter
Posts: 499
Joined: Fri Nov 04, 2016 11:04 pm
Contact:

Re: Is high train braking force better? Not really...

Post by Yoyobuae »

SoShootMe wrote:
Mon Oct 12, 2020 5:11 pm
Huh, I hadn't realised you could get trains to decelerate a lot faster except "instant stop" (eg mine a rail on a train's only path)... But I don't think this should ever happen in normal circumstances, because when a train's stopping point reaches a signal it will either reserve the block (or blocks) after it or, if it can't because the block (or one of the blocks) is already reserved/occupied, it will start braking.
That situation can only happen (without player intervention) when circuit network is used. Here's what happens:

1. Train approaches rail signal
2. Train reserves the block after the rail signal.
3. Rail signal is closed via circuit network. The signal remains yellow because it was already reserved
4. Train moves closer to the rail signal.
4. Repath event is triggered for train. The train now abandons it's previous path and all previously reserved signals.
5. Train chooses a new path and starts reserving signals again
6. Train cannot reserve signal that's closed by the circuit network (train lost the reservation it had previously) and thus it must stop to avoid running a red signal
7. The train is now closer to the red rail signal than it's normal braking distance and thus it must brake a lot faster than normal.

There's also other possible setups by using enable/disable feature of train stations. Turning a train station off causes triggers a repath event on the train, if the new path has a rail signal that cannot be reserved (ie. it is red) then the train is forced to stop at that rail signal, regardless of braking distance.

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

Yoyobuae wrote:
Mon Oct 12, 2020 7:03 pm
That situation can only happen (without player intervention) when circuit network is used.
this map features one event such as the described, thanks for explaining.
speedbump.jpg
speedbump.jpg (141.5 KiB) Viewed 5197 times
Brutal Speed Bump.zip
(3.45 MiB) Downloaded 90 times

SoShootMe
Filter Inserter
Filter Inserter
Posts: 472
Joined: Mon Aug 03, 2020 4:16 pm
Contact:

Re: Is high train braking force better? Not really...

Post by SoShootMe »

Yoyobuae wrote:
Mon Oct 12, 2020 7:03 pm
SoShootMe wrote:
Mon Oct 12, 2020 5:11 pm
Huh, I hadn't realised you could get trains to decelerate a lot faster except "instant stop" (eg mine a rail on a train's only path)... But I don't think this should ever happen in normal circumstances, because when a train's stopping point reaches a signal it will either reserve the block (or blocks) after it or, if it can't because the block (or one of the blocks) is already reserved/occupied, it will start braking.
That situation can only happen (without player intervention) when circuit network is used.
I wasn't very clear but my meaning was to acknowledge the circuit network-closed signal case mentioned and for "normal circumstances" to implicitly exclude that, and more specifically to include the context of whether "depending on the junction that a train blocking the path of another train force the 2nd into a faster deceleration due to the signaling".

Is there a particular reason for the physics-busting braking? Other than "no rails ahead of train" for which there seems no practical alternative, why not eg brake normally to a stop following current path (perhaps including ploughing through an obstacle), then try to repath?

More importantly, can you abuse it to make a junction with better throughput? :D

Qon
Smart Inserter
Smart Inserter
Posts: 2091
Joined: Thu Mar 17, 2016 6:27 am
Contact:

Re: Is high train braking force better? Not really...

Post by Qon »

mmmPI wrote:
Mon Oct 12, 2020 3:34 pm

It seems that a trains that has more wagons will take more time to stop than a train going at the same speed without wagons.
Qon wrote:
Mon Oct 12, 2020 8:46 am
Assuming wagons have breaks and contribute as much to decelarating force as locomotives:
Locomotives are heavier so would impact breaking negativly (compared to equal length but higher % wagon train). This is the opposite of what you say though. So wagons don't have breaks?
Wagons don't seem to have brake;
I don't know if the test i have performed can help conclude on the first part of your statement but it can very well be the case that i made a mistake while talking about "momentum" as it would be commonly understood using the weight of trains, please point it out if so, my nose is too close to the numbers atm to see them properly.
Your deceleration graph has 2 curves with almost the same purple color.......
There's 1 yellow and purple that is showing almost identical data. 1LCCCCCC... and 1L could both have the same decelaration reading your graph, disproving your statement. Or you could be correct, I can't tell your colors apart.
mmmPI wrote:
Mon Oct 12, 2020 3:34 pm
Qon wrote:
Mon Oct 12, 2020 8:46 am
Another point:
But higher accel fuel doesn't increase momentum (comparing equal speed trains) so the breaking distance should be equal while the acceleration changes.

And isn't acceleration usually lower than deceleration?
Higher accel fuel also increases max speed for a train, which makes the braking distance change. (how to test your proposition ?).
Break from top speed. Ignore data until NL reaches top speed of WL, then compare deceleration graph.

mmmPI
Smart Inserter
Smart Inserter
Posts: 2675
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

Re: Is high train braking force better? Not really...

Post by mmmPI »

Qon wrote:
Mon Oct 12, 2020 8:27 pm
Your deceleration graph has 2 curves with almost the same purple color.......
There's 1 yellow and purple that is showing almost identical data. 1LCCCCCC... and 1L could both have the same decelaration reading your graph, disproving your statement. Or you could be correct, I can't tell your colors apart.
fair enough i made the line for the longer train green instead of purple, it is the one that takes the most time to come to a stop, 356 ticks.

1 nuclear loco or 3 nuclear loco have extremly similar results, it was a boring test. ( red and yellow ) 274 ticks, only the rounding from the game made it appears 2 different thing.

3 wooden fuelled loco have similar decelaration as previous example, but with lower base speed it only need 240 tick to stop. (blue) we can't see things but i can't move the line i can provide the data tomorrow if i'm guided on how to

Post Reply

Return to “General discussion”