22-07-2019
BLOCK SECTIONS AROUND STA- TIONS UNDER ETCS
THE EFFECT OF THE BLOCK LAYOUT ON THE HEAD- WAYS AT STATIONS
Willem Wagter - s1822497
Contents
1 Summary 2
2 Introduction 3
3 Literature 3
3.1 ETCS . . . . 3
3.2 Block lengths in NS’54 . . . . 5
3.3 Optimal block lengths . . . . 6
3.4 Summary literature . . . . 8
4 Method 9 4.1 Model unhampered succession . . . . 10
4.2 Model hampered succession . . . . 10
4.3 Analysing bottlenecks . . . . 11
5 Case studies 12 5.1 Almere . . . . 12
5.2 Schiphol . . . . 15
5.2.1 Direction Riekerpolder . . . . 15
5.2.2 Direction Hoofddorp . . . . 20
6 Hampered traffic 21
7 Conclusion and discussion 26
8 Recommendations 27
References 28
1 Summary
Railway networks use block signalling to prevent trains from entering occupied track, and an Automatic Train Protection (ATP) system enforces the drivers to follow the signals aspects. This work by fixed blocks and light signals. ETCS, a new digital European ATP and signalling system is developed to replace the legacy systems in countries that are often all different. ETCS has also as advantage that it can reduce driving times and increase capacity on the same infrastructure because of train dependant braking and accelerating.
Simulation studies have been performed that show that it actually is beneficial to the capacity, but it has not been researched how the infrastructure, and then mainly the signals need to be layed out to achieve this effect.
In this research is looked specifically into block layout around station. The main question is how different block layouts influence the headway around stations. This is important since current methods for determining block layouts are not suitable to use with ETCS and also reduce the capacity effect.
Different block layouts are tested for the Dutch stations Schiphol and Almere Cen- trum and simulated with RailSys to see the achieved headway in both hampered and unhampered conditions. The strategies used for the block design are twofold. A model developed to determine the signal locations is made and tested on the effect on the head- way. For hampered succession this model can actually decrease the headway with around 25 seconds, depending on the distances. These distances are an input in the model. On unhampered succession is this model not very effective, since the minimum headway is often limited by the setting time of switches of dwell time at platforms. It does reduces the occupation time of the blocks, so when more hinder is accepted, for example in delayed situations, the headway can be shorter.
The second method to reduce the headway at the platform is by manipulating the last critical block in multiple ways. For example, placing an extra signal at the begin of the platform, or on the platform. The effect on the headway unhampered is around 10 to 15 seconds while for the hampered situation the effect is of the same magnitude.
The model that was developed gave some promising results on the hampered headway;
the reduction of the headway was larger than with the other measures. The reason being is
that the model’s block layout facilitates the following train to accelerate fast and without
unnecessary slowdowns.
2 Introduction
Currently the Dutch railway system, and many others use some implementation of block signalling and a train protection system (ATP). This combination enables rail traffic to be safe by preventing trains to enter tracks that are already occupied, and enforces the driver to follow the signals. However, these systems mostly date from decades ago, which does not fit the growing demand for capacity today. Trains need to run at high speed and high frequencies, which puts a high demand on the rail infrastructure.
This high demand on the rail infrastructure faces the limits of the legacy national train protection and signalling systems. Those are usually relatively older systems, and do not allow for the short time headway necessary to increase the capacity of the existing tracks.
The development and implementation of ETCS Level 2 or 3, two subversions within ERTMS, in order to replace the legacy systems is a major option to further increase the capacity. Research already showed that the implementation of ERTMS on the Dutch rail network is likely to result in a significant increase in available capacity (de Pundert, van Touw, Bartholomeus, Verhagen, & Duijker, 2010).
However, it is still unclear how ECTS should be implemented exactly, e.g. locations of equipment, configuration of different systems, block layout. Those factors are important since they affect the potential capacity. A common practise when converting a track to ETCS is to replace all signals with ETCS equipment and not changing the block layout. This results in a sub-optimal capacity since the trackside equipment it replaces was positioned based on guidelines and principles of the legacy systems and does not work efficiently with ECTS. This increases the need to design for example the new block layout carefully to achieve the desired increase in capacity.
A problem that arises here is that since the current methods for determining block layouts are based on outdated signalling systems, these methods cannot be used when working on ERTMS tracks. Also, the existing methods and guidelines are not focussed on reducing headway, but just for a safe and functioning signalling system. The block layout is critical for the capacity, and because of the high intensities, the layout around stop locations are now more important then the block layout on the connecting tracks.
This leads to the need for new insights in block layouts near stations suitable for the use with ECTS that focus more on headway reduction.
To look at the effects of different block layouts on headways around stations, in this research different situations are simulated and compared. One of the situations is a simple model to predict the block layout. This is compared to the current situation, and a few other methods to shorten the headway. The simulations are based on the current infrastructure of a few Dutch stations.
3 Literature
3.1 ETCS
ERTMS (European Rail Traffic Management System) is a system of standards to har-
monise railway traffic in Europe. Part of this is ETCS (European Train Control System)
which is the digital European train protection system that is mend to replace the legacy
national train protection systems in the long run. The reason for the development was
once to promote interoperability between European countries. However, ETCS has other
benefits that makes it very important for the development of rail transport.
ETCS works with Movement Authorities (MA) transmitted via radio which gives the driver permission to drive to the next stopping point. Such a stopping point is always marked with a Stop Marker Board (SMB), Figure 1. The onboard computer calculates the braking distance at every moment, and knows via the MA where the stopping location is. When approaching this point, the computer calculates a braking curve, which consists of a maximum speed at each point advancing the stop location, and enforces this speed by activating an emergency braking when the speed limit is exceeded. This way trains can always brake as late as safely possible, and are not limited by trackside signals. Speed reductions are handled the same way. This concept within ERTMS is the braking model.
It is a fully described standard (European Railway Agency, 2016; ERTMS User Group, 2016) which uses the embedded parameters to capture train and track conditions in the model, for example, adhesion conditions, the train’s position, speed and acceleration, track gradients and speeds and distance limits, to calculate the braking behaviour of the trains.
Figure 1: Stop Marker Board
Within ETCS, three sub systems exist, which mainly differ in the way information is exchanged between different nodes in the system. The above described principles hold for both ETCS level 2 and level 3. In Level 1 the MA is not transmitted via radio, but via the ballises in the track and signals are still necessary. The braking model does apply to L1 as well. From here on, Level 1 is not discussed, since due to the technical nature Level 1 cannot provide the same capacity increase and cost reduction as Level 2 and 3 can.
The part where Level 2 and 3 differ is between stations where the trains are hindered by leading trains, and not by the end of the MA. In level 2, the track is divided in blocks which are each marked with an SMB, and the occupation of each block is measured with axle counters. If a block is occupied, the MA of an approaching train ends at the start of that block. Level 3 does not use the fixed physical blocks, but uses moving blocks. The safe distance to approach a train is again based on the braking distance at each moment, but now the location where the train must be able to stop before is the back of the train in front. This location is measured by odometers and GPS and transmitted to trains nearby.
This way trains can run very close to each other, while still able to stop in time. This
works for passenger trains, but a problem arises with freight trains, the train integrity
problem. If a car breaks off from the back of a train, the location of it is unknown since
the location of the back of the train is determined relative to the front. Therefore, the
divided part is not protected against collision with a successing train, since that train
only knows the rear location of the train if it had not divided. Level 3 with moving blocks is therefore an unattainable goal in the foreseeable future, not only because of the train integrity problem, but also because the traffic management systems work with blocks and to switch to operation with moving blocks would require a complete new system.
This lead to investigating the possibilities of a Hybrid Level 3 (HL3) version (Van Gom- pel, 2017). This overcomes the train integrity problem of level 3 and some operational difficulties by still using fixed blocks where freight train are detected and therefore can drive safely, but also dividing those physical blocks into smaller digital blocks that can be used for passenger trains of which the location of the rear end is known so those can still drive closer together.
The effects of the implementation of ETCS Level 2 or 3 are not directly quantifiable.
This is because the effects are very dependant on the Train Protection System that ETCS replaces. de Pundert et al. (2010) performed a simulation study with the specific goal to gain insight in the effects of implementation of ETCS Level 2 in the Netherlands. The only way to investigate this is by simulation, because the current tracks with ETCS Level 2 in the Netherlands are not representative any more because of recent major developments in the ETCS technology, and foreign tracks already were not representative in the first place due to national differences. Also in foreign rail networks the traffic is much more homogeneous than in the Netherlands. The study showed a decrease in driving times and track occupation which can be used to improve the robustness of the time schedule, for scheduling more trains per hour, or to keep from expensive infrastructural adaptations.
The effect of ETCS Level 3 has the same problem as mentioned above. Level 3 is not a proven technique, and has too many technical difficulties for it to be implemented right away (Furness, Van Houten, Arenas, & Bartholomeus, 2017). However, this can be overcome by the adaption of HL3, which is a realistic option. Since this is a very recent development, little is known about the effect. A recent study did show that HL3 offers an increased capacity compared to the legacy signalling an train protection system, while being cheaper to implement than a compared situation with Level 2 (Jansen, 2019).
It also mentioned that the effect of the use of HL3 is largely dependant on the layout near stations and platforms since those are becoming the bottlenecks when increasing the capacity on the connecting track by using HL3. Therefore it is very important to look into the effects here to optimise for the desired effect on the capacity.
3.2 Block lengths in NS’54
ATB-EG with NS’54, the Dutch legacy train protection system and signalling system, works with fixed blocks of which the entrance is guarded by signals (Figure 2). These blocks have a length of approximately 1.5 kilometres on older designed open tracks, the distance the longest and heaviest train needs for braking to a complete standstill. The distances between the signals, the block length, is determined with relatively aged guide- lines, of which is the most important the Design regulations for light signals (ProRail, 2019). This document describes the minimal and maximal distance between two signals.
The distance needed for a train to break from different speeds are defined in the Design
regulations braking distances (ProRail, 2014). This is what is used today to establish the
block layout.
Figure 2: Dutch block signalling (De Vries, 2018a)
Since every train that occupies a block needs to be covered by a red signal, the fixed block length limits how short the distance between trains can be which limits the capacity.
To deal with this, they implemented a scenario in which shorter blocks were realised with the legacy signalling system (Figure 3). This made some shorter headways possible, but can only be done with limited extra signals. Since the aspects of the signals, the speed reduction they show, can only be a few fixed speeds, this does not provide the needed flexibility to make the block even shorter. Besides, since ATB only checks if the brakes are applied at a speed reducing signal and not if the realised deceleration is enough, this also introduces safety risks.
4 4 4 4 44 4 4 4 4 44 4 4 4 4
4 4 4 4 4 44 4 4 4 4 4 4 44 4
Figure 3: Dutch block signalling with shortened blocks (De Vries, 2018b)
3.3 Optimal block lengths
Since L3 moving blocks is not feasible, the tracks still need to be divided in physical and virtual blocks. This means that choices still need to be made concerning the block lengths and layout, and since the use of the guidelines described in subsection 3.2 is obsolete, a different strategy is needed. As described above, the length of the blocks is proportional to the the distance between two trains and their speed, see also the difference between Figure 2 and Figure 3. It is desirable to drive closer together, because more trains can run in the same time over the same route (Goverde, Corman, & D’Ariano, 2013). So the ideal block layout provides the shortest possible headway.
So, the headway is important for the capacity, and due to the fact that is is mainly
determined by infrastructural choices, those choices need to be made well. The most
researched aspect of this headway is the headway on the track between stations. Abril et
al. (2008) showed that at a fixed speed the headway decreases when the block lengths are
shorter, see also Figure 4. This holds for trains following at the same speed. This study
looked, among other things, specially into the effects of block length when using ETCS
L2. The headway is then based on the time needed to pass a block, the braking time,
operating time which is a margin defined by the infrastructure manager, and the release
time which is the time that passes before the deparure of the first train from the block
has been detected and transmitted to the second train. This is illustrated in Figure 5.
80 100 120 140 160 180 200
500 1000 1500 2000 2500 3000 3500 4000 4500 50 00 5500 6000 Line Section Length (meters)
Seconds
200 Km/h 250 Km/h 300 Km/h 350 Km/h 400 Km/h 450 Km/h 500 Km/h
Figure 4: Relation between block length and headway (Abril et al., 2008)
Figure 5: Headway definition (Abril et al., 2008)
This principle can be used to calculate the optimal block length given a time schedule.
This gives the headway necessary to perform the planned schedule. Since the aspects of headway are all speed dependant, the maximum line speed is also needed as an input.
This all combined results in a signal spacing that is suitable to achieve the schedule.
Important here is that the block length calculated is specific for a type of train, because it is dependant on the braking characteristics of the train. To determine this for mixed train type traffic, the worst braking train is normative, and defines the block spacing.
This is is because if the worst braking train is not picked as normative, due to the longer braking time for this train, it is not able to attain the planned driving times.
Sangphong, Siridhara, and Ratanavaraha (2017) looked into the effects of block lengths
and headway on lines with different train speeds. They devised a model to determine the
minimum headway on a line with fixed block sections for two trains with different speeds
driving in the same direction. The headway in this situation is defined by the time after
which the second train departs behind the first one, while not being hindered by the front
train over the length of the line. When the first train is the fastest, the critical block is
the first one. This block defines the headway in that situation. When the second train is
faster, the last block on the line in the critical one. The situation with two train at the
same speed was found to have the highest capacity with fixed and equal block lengths.
They also found the same relation as Abril et al. (2008), that a shorter block length increases the capacity. These results are not directly interchangeable since Sangphong et al. did not account for the braking behaviour of the involved train, but they presumed a minimum length that would suffice. Abril et al. on the other hand this account for the braking characteristics. While not specifally stated in the study, the results and method of Sangphong et al. (2017) could be used to optimise the block lengths. When a the critical block as been identified, the length of it could be changed in order to improve the capacity.
Another study about the effects on and of block lengths is done by Landex and Kaas (2005). They looked into the relation between line speed on a specific track in a subway system with discrete blocks and the headway. They found that due to the discrete nature of the information supply to the train’s Automatic Train Control (ATC) system the capacity of a line could be improved by lowering the line speeds. This way the headway can be reduced and the trains can follow closer. They developed a new method to optimise the line speeds to the existing block sections. This is necessary since the rolling stock gets replaces and the infrastructure remains the same. This asks for a new optimum speed.
They adapted the speed profiles to fit the existing infrastructure the best as possible, where the best means the shortest headway. They did compare the discrete block sections with a continuous ATC where the train can receive information about the changed aspect of the signal at any time, and not only at a balise. This is similar to the functioning of ECTS level 2 and 3, so the same definitions as in Figure 5 hold here. The difference between the described ATC and ECTS is that ATC does not include a strictly prescribed way to calculated the braking distances which therefore was an important part of the study to calculate this. A conclusion they drew was that with the upgrade from discrete ATC to continuous ATC, of which the last is more or less comparable tot ETCS L2&3, slight fluctuations in speed do not have a large effect on the headway any more. Besides, they also found that even with the continuous ATC, optimising it to the signalling system and block layout still can improve the capacity. This results are not directly transferable to the problem in this research, but it shows that the it is important that the speeds should fit the block layout in order to achieve maximal performance in terms of capacity.
3.4 Summary literature
To conclude the previous sections about the theoretical background and the existing research, quite much is known about headways and capacity related to the layout of the used Train Protection System which is most often based on blocks. The definition of headway is also described, as well as the relation to capacity. What those studies have in common, is that they all assume that the capacity of a railway line depends on the capacity of the track connecting the stations. However, in highly congested rail traffic these approaches does not work any more since the location of bottlenecks shift from the connecting tracks as covered in the literature, to junctions and stations in the network (Jansen, 2019). What also is an important problem is that the capacity is highly limited by heterogeneous traffic. This means that the capacity drops when different trains, e.g.
speeds, acceleration and braking behaviour, lengths etc. share the same infrastructure.
Also the studies about the effects of ETCS did not consider block sections, since given
the state of development of the system that is too specific. Jansen (2019) even specifically
mentioned that now the global effects of ETCS both L2 and L3 are known, more specific
attention must be given to the actual implementation which involves block layout. Current
methods to determine block layouts are unsuitable either because they do not account for the nature and the advantages of ETCS, or because they do not account for stops and stations. So, in order to be able to determine the block layout for ETCS equipped tracks around stations to fit the growing demand, a new method is necessary. Therefore, in this research a new model that determines signal locations will be used to test if it has any effects on the headway. Also a few other strategies for signal placing to decrease the headway will be tested.
4 Method
In order to see the effect of block layout on the headway, different block layouts at two Dutch stations are simulated. The performance indicator here is the headway. Depending on the situation this is either the difference in time between two arrivals, or the difference between the departure and arrival. This is simulated by planning two trains right behind each other in the simulation software used and see what the shortest time between the two can be, considering the occupation of the blocks. As one variant to determine the block layouts a model is made and used that determines the signal positions for a given situation based on an algorithm. This is described in the paragraphs below.
The software used for the simulations is RailSys. This package is able to simulate ETCS trains accurately and can do both planning and simulation in the same package and in the same interface. The train types used in the simulation are SLT6, a 101 meter six car commuter train, and VIRM6, a 162 meter six car intercity double-decker train.
Those are commonly used train types on the Dutch rail network, but not exclusively. This are also the only train types with known ETCS characteristics.
The different situations are based on the conditions around two Dutch stations, Schiphol and Almere Centrum. Those stations are chosen because in the current timetable very short following trains already occur here, so it is interesting to see how changes in the infrastructure could affect this timetable.
Two different situations are simulated for each of the two stations and block layout variants. The first is an unhampered succession where the following train drives without influence of the first one, as the situation is in a normal timetable. The second represents a hampered situation that occurs when a train is that much longer on a platform that the train behind it comes to a complete stop af the first signal behind the train on the platform, see also Figure 6. The block layout should facilitate that the headway here, the time between the departure of the first train and the arrival of the second train, is as short as possible. This would result in a more robust timetable since delays are transferred less to the trains behind. For these two situations, a different model is created. All the different strategies that are used to determine the block layout variants are described in the next paragraphs.
Platform
Figure 6: Platform with stopped trains
To compare the effect of the different signal layouts, the main performance indicator is the headway. This is extracted from the simulation software. An other indication on the performance is the time-distance diagram. These are also the result of the simulation.
In those diagrams are visible on the horizontal axis the distance, and on the vertical axis the time with a scale in minutes and show the reservation and occupation of the blocks for the two trains.
4.1 Model unhampered succession
The model developed for this research consists of two separate parts, because for the unhampered and hampered situation the exact implementation is a little different. The algorithm behind it however, is the same.
The model for the unhampered situation works a follows. As an input the for braking and speed etc. for the train in a certain situation, the time-distance diagrams of those trains from RailSys are used. The reason to use RailSys for this is that it is harder to simulate this in a simple model, and the simulation of RailSys is already good. Then an extra line is created, namely the indication distance at each speed for the second train.
The reason is that the indication distance normative is for hinder. For unhampered traffic the block sections need to be cleared and released at a minimum of the indication distance at the travelled speed. So, in the model this condition is used to locate the signals. First, a first signal is predefined. This is the split point, the signal before the switch where the routes of the two followings trains divert. From this point, the model works back placing the signals. This is done as follows.
First it calculates the time where the back of the first train releases the last section, which was an input. The location of the next signal is then defined by the location of the indication curve of the second train at the preciously calculated time plus the setting time. This locates the next block in such a way, that that block is cleared for at least the setting time before the following train approaches that block at indication distance.
This process is then repeated with the new output block x
b(k + 1) of the last calculation serving as input for the next, see Equation 2. This way each block section is relatively positioned to the next one based on the actual situation.
x
1(t) = position rear train 1 x
2(t) = position f ront train 2
x
i(v) = indication distance as f untion of speed x
i,2(t) = x
2(t) + x
1(v)
x
b= block boundaries t
set= setting time
(1)
t
release=> where x
1(t) = x
b(k)
x
b(k + 1) = x
i,2(t
release+ t
set) (2)
4.2 Model hampered succession
The model for the hampered situation has the same principle, but works a little differently.
The initial situation consists here of two stopped trains, one on the platform, and the
second at the last signal before the platform, as displayed in Figure 6. The model assumes
the two trains start driving as closely after each other as possible, which in an ideal
situation would be with only the setting time in between. This can either be realised with a software solution within the ETCS system, or, and this is today already possible, by placing the first block boundary at the tail of the train on the platform. This way this section is occupied when dwelling, but it is released when the departing train has moved a few meters.
So, the model starts with the first block section at the tail of the first train. Also, the existing fixed block boundaries and the stop location are used as an input. When these conditions are defined, the paths of the trains are calculated. The first train accelerates with an acceleration of 0.7 m/s
2until the train has cleared the platform. The second train accelerates with the same acceleration, but is both limited by the maximum speed of the track and the braking distance corresponding to it’s speed. The braking distance here is the Permitted distance from the ETCS braking model, since this is the distance at which the train needs to brake in order to safely stop before the target. If at this point the braking distance curve does not intersect the curve of the back of the first train, the model is ready to determine the signal position. If not, the maximum speed of the second train is limited until it does not intersect any more.
The principal behind the signal locating algorithm is the same as before. The model first calculates the time the first defined block section intersects the braking distance curve, so the moment the second train reaches Permitted distance of the first section.
Then, the next signal is located at the point where the back of the first train is at the firstly calculated moment plus the setting time, see Equation 4. This way the back of the first train has cleared the created block section at Permitted distance for train two. This is then repeated with the output of the first calculation, x
b(k + 1), serving as the input for the second, until the first existing block boundary is reached. The output is then a string with locations for the signals.
x
1(t) = position rear train 1 x
2(t) = position f ront train 2
x
i(v) = indication distance as f untion of speed x
i,2(t) = x
2(t) + x
1(v)
x
b= block boundaries t
set= setting time
(3)
t
release=> where x
i,2(t) = x
b(k)
x
b(k + 1) = x
1(t
release+ t
set) (4)
4.3 Analysing bottlenecks
The third strategy to design the block layout is by analysing the time distance diagram
from the simulation. From these diagrams the critical block section becomes visible,
and manipulations in this critical block influences the headway. As the literature review
already showed, shorter block provide a shorter headway. So measures that are tried
are placing an extra signal between the platform and the current last signal and placing
signals along the platform. The first measure moves the point where hamper would occur
more to the front, so the following train can enter the extra section a little earlier. The
second measure is about clearing the platform as quickly as possible. This also moves the
point where hamper would occur forwards.
5 Case studies
The scenarios as described in section 4 are applied on the stations of Schiphol and Almere Centrum. For testing these scenarios there are a few assumptions and boundary condi- tions. In the hampered situation none of the block sections directly on the platform can be released before the first train starts the departure. This is necessary to prevent two trains on the same platform which is undesirable from operational standpoint. Also, the newly designed block layouts do not comply to the design regulations by definition. First of all because the regulations are very extensive, and designing the cases fully compliant to these take too much time and knowledge for a capacity study. Second, the goal of the research is to see the effect of certain measures, not to design a directly usable situation.
In the simulation of the cases a few assumptions are followed. Those are listed below:
• dwell time for regional trains is 60 seconds
• block reservation is at indication distance for unhampered situation (ProRail, 2018)
• only one train at the platform at each possible moment
5.1 Almere
The current infrastructure with the locations of the signals indicated with the kilometre locations is visible in Figure 7. Only the signals in the studied driving direction are visible to keep it clear. Also a part of the station is left out on the right side for the same reason.
The blue squares with the numbers 4 and 6 represent the stopping location for trains of 100 and 160 meters respectively.
In the studied situation, a regional train arrives at platform 3 (Figure 7) and behind it an intercity train that will stop at platform 4. The regional train here has to release the section with the last switch before the second train can follow.
14.680 14.040
13.475 12.660
12.260
15.323
4 6
15.240 15.220 1
2 3 4
Figure 7: Current situation Almere Centrum
The headway for this situation is 57 seconds. This is completely dependant on the release of the switch to platform 4. When applying the model described in section 4, the headway does not change. This is because, as described above, the last block is critical when a faster train follows a slower one (Sangphong et al., 2017). The block sections before this last section do not influence the headway in an unhampered situation, which can be seen in Figure 8.
The model described in section 4 is applied to this situation. The result is list of
locations for the block boundaries. This is visualised in Figure 9. Based on the model,
this would result in a headway that is a few seconds shorter than in the original situation.
However, when simulated this results not in a shorter headway, but a headway one second longer than before. This is not a very significant difference, and would fall within margin of error. The block occupation diagram is visible in Figure 10.
The effect of the model is therefore nothing, even a slight change for the worse. What does stand out is that even though the model strategy does not reduce the headway in the unhampered situation, it is likely that in a delayed situation the new block layout can provide a shorter headway. The reason is that the new blocks are released earlier, in a way that in a hampered situation braking can be postponed. This is also visible when Figure 8 is compared with Figure 10. In the last figure the gap between the two occupations and reservations is a little larger than in original situation.
1027 212
214 210 A 212
206 1026
1027
208 202 A 1028
1028 1025 A 1024
210
212 220
218 A 214
2202 2204
2256 2250 2218
2216 2214 2212
2220 2222 2224 2226
2254 2252 216
1029
214 204
202 1025
1031 1
2
1 2 3 4
23
Almm Alm_TT
4 4
6 6
101
201
Alm 202
Figure 8: Time distance diagram Almere Centrum current situation
0 100 200 300 400 500 600 Time (s)
0 1000 2000 3000 4000 5000 6000 7000
Dist an ce (m )
Position train 1 Position rear train 1 Position train 2
Indication distance train 2 New block boundaries
Figure 9: Model result Almere Centrum
1027 1024
1022 214
210 A212
206 1026
1027
208 202 A1028
1028 1025
A1024 210
218 A214
2202
2256 2250 2218
2216 2214 2212
2220 2222 2224 2226
2254 2252 216
1029
214 204
202 1025
1023 1021
1021 1 1
2 2
1 2 3 4
23
Ampo
Almm Alm_TT
4 4
6 201 6
101
Alm 202
Figure 10: Time distance diagram Almere Centrum blocks according to model
5.2 Schiphol
At Schiphol intercity and regional trains are mostly separated at the tracks. Therefore, only similar trains follow each other. See Figure 11 for the overview of the current infrastructure. This research addresses regional trains arriving at the two inside tracks, 3 and 4. Only two following regional trains are studied, this is a relevant situation at the moment, and for intercity trains it is almost the same, except for a different train types.
Also, intercity trains cannot follow as close as regional train in the first place, because they have longer dwell times that increases the headway. The train successions in both directions are discussed separately since they are different due to the asymmetrical layout of the station. The headway at Schiphol is the time between the departure of the first train at the platform and the arrival of the second train.
14.060 15.170
14.220 14.648
14.045 13.731
16.300 17.765
12.150 10.690
4
66 4
14.160 Shl
← Hoofddorp Riekerpolder aansl. →
1 2 3 4 5 6
Figure 11: Current situation Schiphol
5.2.1 Direction Riekerpolder
Trains coming from the direction of Hoofddorp approach Schiphol on the left of the two most right tracks and dwell on platform 3. They also leave straight, so no switches are passed when passing Schiphol. The front of the train is at the right end of the platform when dwelling. Combined with the length of the platform, around 400 meters, results in a very long block section at the platform.
In the current situation the headway is 01:53 min. The last block section is critical
again, and is mostly defined by the dwell time. See also Figure 12. What stands out in
this situation is that the last block is especially critical because its large length. Since
the whole block has to be reserved at indication distance of the approaching train, this
increases the headway. Therefore, solutions must be found in manipulating the last block,
and not the block sections when approaching the station.
88 88 1760
1758
1070 1066 1064 1072 1074
1062 1078
1058
1048 1034
1050 1032
1036 1030
1038 1028
1076 1060
1080 1056
1092 1090
1068 1094
1096 1054
1098 1052
11001778 1102
1776 11041774 1106
1772 1044 1738
1046 1736
1040 1742
1042 1740
1765 1764 1751 1088
1757 1082 1771
1769 1755 1084
1767 1762
1753 1086
1735
1722 1721
1733 1724 1719
1731
1726 1717
1728 1715
1729 1122
1120
1132 1110
1116
1124 1118
1130 1112
1114 1128
1126
1708
1707 1020 1022
1306 1304
1308
1302 1024
1714 1701 1026
123546
4
210 209
12 R4
R1
RA
21
3 RB2
Shl
Asra
10 10
6 6
8 8
12 12
14 14
302 303
Hfd
301
Figure 12: Time distance diagram Schiphol current situation
A possible solution is to divide the long platform in more shorter blocks that can be released earlier. This however leads to a problem. As noted in the assumptions at page 12, two trains cannot be on the same platform at the same time. When the platform is divided in shorter sections, normally the sections that are not occupied by the train on the platform are released. This means that the following train can already drive to the platform right behind the first train. This is what must be prevented, since this is confusing for the passengers in the train, and the waiting passengers on the platform. Therefore, simply dividing the platform in smaller sections does not work. A few alternatives are possible.
The length of the block at the platform does not only cover the platform itself, but as visible in Figure 11, also the switches before and after the platform. A way to handle this is to place a signal at the begin of the platform. This is showed in Figure 13. The headway gain with this measure is 11 seconds.
A possible solution to the problem described above is to place a signal just at the rear
end of the first train in a way that as soon it starts moving, the block section form there
back along the platform is released. This way the headway can be shortened while still
fulfilling the condition that two trains cannot be at the same platform together.
14.060 15.170
14.220 14.648
14.045 13.731
16.300 17.765
12.150 10.690
4
66 4
14.160 Shl
← Hoofddorp Riekerpolder aansl. →
1 2 3 4 5 6
14.466
Figure 13: Schiphol with signals at begin of the platform
14.060 15.170
14.220 14.648
14.045 13.731
16.300 17.765
12.150 10.690
4
66 4
14.160 Shl
← Hoofddorp Riekerpolder aansl. →
1 2 3 4 5 6
14.466
Figure 14: Schiphol with signal at rear end of the train
This layout (Figure 14) facilitates the long section behind the dwelling train to be
released 8 seconds earlier, and the total headway decreases with 11 seconds, so the same
improvement as with the signal at the begin of the platform. The effect is also visible in
Figure 15.
8 8
8 8
1070
1066 1064 1072
1074
1062 1078
1058
1048 1034 1050
1032 1036
1030 1038
1028 1076
1080
1056
1092 1090 1068 1094 1096
1054 1098
1052
1044
1738 1046
1736
1040
1742 1042
1740 1764
1751 1088 1757
1082 1755
1084 1762
1753 1086
1735 1733 1731 1729 1
2 3 5 4 6
21
10 10
8 8
12 12
303
Shl 301
Figure 15: Time distance diagram Schiphol with signal at rear end of the train
For this direction no further improvements can be made, for the reasons described
earlier. To show this, the last layout is simulated again, but with an extra signal between
the one at the rear of the train and the existing signal at the end of the platform, as
shown in Figure 16. Below in Figure 17 is visible that although the extra block section
results in that section to be released a little earlier, this effect cannot be used since for it
to be unhampered that section needs to be reserved at indication distance.
14.060 15.170
14.220 14.648
14.045 13.731
16.300 17.765
12.150 10.690
4
66 4
14.160 Shl
← Hoofddorp Riekerpolder aansl. →
1 2 3 4 5 6
14.466
Figure 16: Schiphol with signal at rear end of the train and extra halfway
88 88
1760 1758
1070 1066 1064 1072 1074
1062 1078
1058
1048 1034 1050
1032 1036
1030 1038
1028 1076
1080 1056
1092 1090 1068 1094 1096
1054 1098
1052
1044
1738 1046
1736
1040
1742 1042
1740 1765
1764
1751 1088 1757
1082 1771
1769 1755
1084 1767
1762
1753 1086
1735 1733 1731 1729 1
2 3 5 4 6
21
10 10
8 8
301
Shl 303
Figure 17: Time distance diagram Schiphol with signal at rear end of the train and extra
halfway
Table 1: Simulation results Schiphol direction Riekerpolder
Curren t blo ck la y out Additional signal at b egin platform Additional signal at tail of the stopp ed train Additional signal at b egin p latform and at tail of the train Additional signal at b egin platform and halfw a y to the end Headway 00:01:56 00:01:45 00:01:45 00:01:33 00:01:33 Improvement 00:00:11 00:00:11 00:00:23 00:00:23
5.2.2 Direction Hoofddorp
For the direction Hoofddorp the same problem exists as for the opposite direction, namely the large platform length and the large distance between the last signal before the platform and the platform itself. This is also recognisable in Figure 11. The headway in the original situation is 2:22 min, with again the last block that is critical (Figure 18). For the same reason as for the other direction, manipulating the block sections before the last one has no effect in the unhampered situation.
When applying the same strategies as for the opposite direction, the first step would be to also create an extra section between the last signal and the platform (Figure 13).
This reduces the headway by 10 seconds. When trying to reduce this even more, the
problem of two trains on the same platform does not occur in this direction. The reason
for this is that the stopping location is at the right end on the platform. So when the
first section on the platform is released, the first train is already moving.
88
88
1760 1758 1070
1066
1064 1072
1074 1062
1078 1058 1048 1034
1050 1032
1036 1030
1038
1028 1076
1060
1080 1056
1092 1090
1068 1094
1096 1054
1098 1052
1100 1778
1102 1776
1104 1774
1106 1772 1044
1738
1046 1736
1040 1742
1042 1740
1765 1764 1751
1088
1757 1082
1771 1769 1755
1084 1753 1762 1767
1086
1735 1722
1721
1733 1724
1719
1731 1726
1717 1728
1715 1729
1708 1707
1306 1304
1308 1302
1714 1701
1 2 3
5 4
6
21
20 20
16 16
18 18
22 22
304 302 303
Shl 304
302 303
Figure 18: Time distance diagram Schiphol in direction Hoofddorp, current situation
The next possibility is to place a signal at the front of the stopped train. This can facilitate the second train to have the entire route to the stopping location released when the first train releases only the stopping location. This measure results in a 13 second reduction of the headway in comparison with the original situation. When the above two measures are combined the headway is reduced with 16 seconds in comparison with the initial situation.
6 Hampered traffic
Besides the unhampered train succession which represents a regular timetable, the effects of block layout are also investigated for maximally hampered traffic. The infrastructure variants are mostly the same as for the unhampered succession, but the application of the model is slightly different. Because it is about the same block layouts as above, the same diagrams will be used as reference.
For reducing the headway in a hindered situation, two main parameters can be changed
to influence the headway. The distance of the waiting location to the platform, and the
moment the first section is released after the departure of the first train. The first variable
is changed by placing an extra signal at the begin of the platform, so the distance to be
travelled is shorter Figure 13. For Almere this results in a headway reduction of 17
seconds, and for Schiphol in 11 and 21 seconds reduction for the direction Riekerpolder
Table 2: Simulation results Schiphol direction Hoofddorp
Curren t blo ck la y out Additional signal a t b egin platform Additional signal at fron t of the stopp ed train Additional signal a t b egin platform and at fron t of the train Headway 00:02:22 00:02:12 00:02:09 00:02:06 Improvement 00:00:10 00:00:13 00:00:16
and Hoofddorp, respectively. The results are also visible in Table 3, Table 4 and Table 5.
Table 3: Simulation results Almere hampered
Curren t blo ck la y out Additional signal at b egin platform Headway 00:01:34 00:01:17
Improvement 00:00:17
A custom model was also developed to determine the block layout on the platform to reduce the hampered headway, the principle is the same as for the unhampered succession.
This is also described in section 4. To test the effect of this model more simulations are performed. However, the effects are not expected to be significant with short trains.
Therefore, these simulations are conducted with train of 202 metres, two six-car SLT trains combined. The downside of this is that the results cannot be directly compared with the previous simulation results.
The core concept behind the model is that the second train must be able to accelerate to
the maximum allowed speed or to the point where the deceleration needs to start. This
results in a very short driving time. The next important aspect is the moment the second
train can start moving. This needs to be as early as possible, which is the exact moment
Table 4: Simulation results Schiphol direction Riekerpolder hampered
Curren t blo ck la y out Additional signal a t b egin platform Additional signal at tail of the stopp ed train Additional signal at b egin p latform and at tail of the train Headway 00:01:34 00:01:23 00:01:10
Improvement 00:00:11 00:00:24
the second trains starts moving. If the distance between the waiting location and the stopping location on the platform is long enough, the second train can start moving at the same moment as the first one. In the simulation this is achieved by defining the first block boundary at the rear end of the stopped train, so it is released a the moment the first few meters are driven. In the future this can also be done by a position report over the ETCS system which could release the first block.
The new layouts for Almere and both directions of Schiphol can be found in Figure 19 and 20. An example of the output of the model can be found in Figure 21. The headway reduction for Almere is 25 seconds, and for Schiphol 19 second for direction Riekerpolder.
14.680 15.323
15.279 3
4
12
15.291 15.224
15.084
Figure 19: Block layout Almere according to model
Table 5: Simulation results Schiphol direction Hoofddorp hampered
Curren t blo ck la y out Additional signal a t b egin platform Additional signal at tail of the stopp ed train Additional signal at b egin p latform and at tail of the train Headway 00:01:42 00:01:21 00:01:09
Improvement 00:00:21 00:00:33
14.060
14.270 14.045
← Hoofddorp Shl Riekerpolder aansl. →
3 4
14.648
13.731 14.466
12
14.255 14.117 14.095 14.203 14.257 14.285
Figure 20: Block layout Schiphol according to model
0 200 400 600 800 Time (s)
0 100 200 300 400 500 600 700 800
Dist an ce (m )
Position front train 2 Position rear train 2 Position front train 1 Braking distance train 1 Existing block boundaries New block boundaries
Figure 21: Model output for Schiphol
Figure 22 shows the resulting blocking diagram of the hampered succession at Schiphol
in direction Riekerpolder. A few things stand out here. First, in the speed-distance graph
in the middle is visible that the second train can accelerate and brake the same way it
would when no train was before it. This results in short driving time. Furthermore, the
setting time starts very shortly after the departure of the first train. This facilitates that
the second train can start moving very shortly after the first train. That this provides the
highest use of the available infrastructure is also visible in the time between the release
of the block by the first train and the occupation of the second. Almost no time is in
between, which is good since that time would remain unused, while now it is used to
reduce the headway.
1074
1062 1078
1058
1050
1032 1036
1030 1038
1028 1076
1060 1068
1094 1096
1054 1098
1052 1082
1084
1086 3
5 4 21
140 km/h 140 km/h
105 km/h 105 km/h
70 km/h 70 km/h
35 km/h 35 km/h
0 km/h 0 km/h
130
60
0km/h
ETCS L2/3 sect. avail. ETCS L2/2 sect. avail. ETCS L2/3 sect. avail.
ETCS L2/2 sect. avail. ETCS L2/no sect.avail. ETCS L2/2 sect. avail.
ETCS L2
38 38
305 307
Shl
305
307
