IN
DEGREE PROJECT THE BUILT ENVIRONMENT, SECOND CYCLE, 30 CREDITS
STOCKHOLM SWEDEN 2019 ,
Modeling fault probability in single railroad turnouts in Eastern
Region, Sweden, with the use of logistic regression models
A step from preventive to predictive preventive maintenance in railway maintenance planning FILIPP ZAROV
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF ARCHITECTURE AND THE BUILT ENVIRONMENT
1
Modeling fault probability in single railroad turnouts in Eastern Region,
Sweden, with the use of logistic regression models
A step from preventive to predictive preventive maintenance in railway maintenance planning
Filipp Zarov
Master thesis December 2019
School of Architecture and Built environment KTH Railway Group
KTH Royal Institute of Technology
Stockholm, Sweden
2
Abstract
Turnouts are an important part of railway infrastructure for two reasons: infrastructure and maintenance. For the infrastructure they provide the flexibility to allow the formulation and branching of railway network and for maintenance they consume a large part of maintenance budget and have a prominent place in maintenance planning policy and activities. This is because as a “mechanical object”, a turnout often experiences malfunctions. The problem becomes even more complicated, since a turnout is composed of many different parts and each of them fails for very different reasons (e.g. switch blades vs crossing part). This is reflected in the different needs for maintenance activities, as railways are forced to pour in excessive amounts of resources to carry out emergency repairs, or to carry out unnecessary scheduled maintenance works in turnouts, which do not need to be inspected or repaired.
Therefore, it is difficult to plan and organize maintenance activities in turnouts in an efficient manner.
This raises the question of whether malfunctions in turnouts can be predicted and used as information for the maintenance planning process in order to optimize it and develop it into a more reliable preventive maintenance planning.
The aim of this analysis is to attempt to model the probability of various malfunctions in turnouts as a function of their main geometric and operational characteristics by using logistic regression models and then input these results into the maintenance planning process in order to optimize it. First, it was important to objectify the railway track system and the turnout components, both in terms of parts and interrelationships. Furthermore, the process and basic elements of railway maintenance planning were defined, as well as arguments that motivate a turn towards preventive maintenance planning methodologies. This was done through a comprehensive literature study.
The basis of this research was case studies, which described the relationship between geometrical and operational characteristics of turnouts and their wear, as well as risk-based modelling methods in railway maintenance planning. To create the analysis model, data from turnouts in eastern region provided by the Swedish Transport Administration were used, both from the point of view of describing the underlying causes of turnout malfunctions and to formulate an object-oriented database suitable for using in logistic regression models. The goal was a logit model that calculated the malfunction probability of a turnout, which could be used directly into a maintenance planning framework, which ranked maintenance activities in turnouts.
The results obtained showed that although the model suffers from low correlation, different relationships between input variables and different functional errors were established. Furthermore, the potential of these analytical models and modeling structures was shown to be able to develop preventive, predictive railway maintenance plans, but further analysis of the data structure is required, especially regarding data quality. Finally, further possible research areas are presented.
Keywords: maintenance planning, railway turnouts, malfunction probability modeling, object- oriented databases, logistic regression models
Sammanfattning
Spårväxlar är viktiga delar av järnvägens infrastruktur av två orsaker: infrastruktur och underhåll.
För infrastrukturen ger de möjlighet till flexibla tillåter de formulering och grenning av järnvägsnät och
för underhållet konsumerar de en stor del av underhållsbudgeten och de har en framträdande plats i
underhållsplaneringspolitiken och aktiviteterna. Detta beror på att som ett ”maskinellt objekt”, har
spårväxeln ofta fel. Problemet blir ännu mer komplicerat, eftersom en spårväxel består av många olika
delar och var och en av dem bryts ner av mycket olika skäl (t.ex. tunganordning vs korsningsdel). Detta
återspeglas i olika behov av underhållsaktiviteter. Eftersom järnvägarna tvingas hålla alltför stora
mängder resurser för att utföra akuta reparationer eller för att utföra onödiga schemalagda
underhållsarbeten i spårväxlar, som inte behöver inspekteras eller repareras. Därför är det svårt att
planera och organisera underhållsaktiviteter för spårväxlarna på ett effektivt sätt. Detta ställer frågan
om funktionsfel i spårväxlar kan förutsägas och användas som information till
3 underhållsplaneringsprocessen för att optimera den och utveckla den till en pålitligare förebyggande underhållsplanering.
Syftet med denna analys är att försöka modellera sannolikheten för olika funktionsfel i spårväxlar som en funktion av deras huvudsakliga geometriska och operativa egenskaper med användning av logistiska regressionsmodeller och sedan mata dessa resultat in i underhållsplaneringsprocessen för att optimera den. För det första var det viktigt att objektifiera järnvägsspårsystemet och spårväxlarkomponenterna, både vad gäller delar och inbördes förhållanden. Dessutom definierades processen och grundelementen i järnvägsunderhållsplaneringen, samt att argument som motiverar förändring till förebyggande underhållsplaneringsmetoder. Detta gjordes genom en omfattande litteraturstudie.
Grunden i denna analys var fallstudier, som beskrev förhållandet mellan geometriska och operationella egenskaper hos spårväxlar och deras förslitning samt riskbaserade modelleringsmetoder i järnvägsunderhållsplanering. För att skapa analysmodellen användes data från spårväxlar i östra regionen som tillhandahölls av Trafikverket, både ur synpunkten att beskriva de underliggande orsakerna till spårväxlarsfel och för att formulera en objektorienterad databas lämplig för användning i logistiska regressionsmodeller. Målet var en logitmodell som beräknade sannolikheten för fel i en spårväxel, som kunde användas direkt i en underhållsplaneringsram, som rangordnar lämpiga underhållsaktiviteter i spårväxlar.
Erhållna resultat visade att även om modellen lider av låg korrelation, konstaterades olika samband mellan ingående variabler och olika funktionsfel. Vidare visades potentialen hos dessa analysmodeller och modelleringsstrukturer för att kunna utveckla förebyggande, förutsägbara järnvägsunderhållsplaner, men det krävs troligtvis ytterligare analys av datastrukturen, speciellt angående datakvaliteten. Slutligen presenteras ytterligare möjliga forskningsområden.
Nyckelord: underhållsplanering, spårväxlar, felsannolikhet modellering, objektorienterad databas, logistiska regressionsmodeller
Professional Acknowledgements
Throughout the writing process of my master thesis, I received a great deal of support and assistance.
I would like to thank 1) my supervisor Anders Lindahl, member of KTH railway group
1and transport planning division, as well as teacher in M.Sc. in Transport and Geoinformation technology for his invaluable support, teaching, guiding and assistance not only during the master thesis but also through the entire Master programme, 2) Arne Nissen
2, responsible for managing Trafikverket railway maintenance databases, who kindly provided not only access to data related to railway traffic and turnout malfunctions from Ofelia and BIS databases in eastern region, but also actively supported and explained the variables and structure of the databases, 3) Carlos Casanueva Perez, associate professor in rail vehicle technology, department of Aeronautical and Vehicle engineering, CMG at ECO2 vehicle design and director of M.Sc. programme in railway engineering for his technical and practical advices, as well as for providing me with bibliographic material related to turnouts, 4) Fredrik Andersson, for being the inspiration behind this master thesis topic and helping formulating the basis of it and finally 5) anyone who directly or indirectly was involved in the implementation of this topic.
Personal Acknowledgements
I would like to thank Marija Rubil, who wholeheartedly supported me through the whole process of realizing this master thesis, with her endless psychological and practical support, as well as helping me with the collection of critical reading material, which proved invaluable during the writing of this master thesis.
1 Education and project support
2 Trafikverket, Luleå, Spårtekniker, UHjsp
4
Table of Contents
1. Introduction ... 6
Background ... 6
Aims ... 7
Method ... 7
Limitations ... 7
2. Literature review ... 8
2.1. The railway wheel-track system – The dynamic interaction between rail infrastructure and rail vehicles – basic components of rail infrastructure ... 8
2.1.1. Track structure in a nutshell ... 8
2.1.2. Forces acting on a track ... 8
2.1.3. Other factors that amplify the forces ... 10
2.1.4. Track components ... 10
2.1.5 Conclusions – Track system, loading and parts ... 17
2.2. Turnouts ... 17
2.2.1. Designation of a turnout ... 18
2.2.2. Parts of a turnout ... 18
2.2.3 Other important parts and aspects of a turnout ... 23
2.2.4. Geometry and operational characteristics of a turnout ... 26
2.2.5 Conclusions: Turnouts, parts and interaction between operational and geometrical characteristics ... 31
2.3. Structural analysis of a rail track and turnout considerations ... 32
2.3.1. Theoretical foundation and main parameters of track structural analysis ... 32
2.3.2. From main parameters of structural analysis to stress calculations ... 38
2.3.3. Structural considerations in a turnout... 42
2.3.4. Conclusions: Structural considerations in railway infrastructure and turnouts ... 47
2.4. Railway maintenance planning in Sweden and turnouts ... 48
2.4.1. Railway market in Sweden, railway maintenance and Trafikverket ... 48
2.4.2. Maintenance from Trafikverket’s perspective ... 49
2.4.2.1. The maintenance plan 2019-2022 ... 49
2.4.3. Maintenance planning considerations ... 51
2.4.4. Classification of maintenance in relation to response- from corrective to predictive maintenance ... 54
2.4.5. Modelling maintenance planning ... 54
2.4.6. Conclusions: Maintenance planning, modeling and incorporation of asset degradation ... 57
3. Methodology ... 58
3.1. Fault risks in a turnout and data analysis: the case of east region in Sweden ... 58
5
3.1.1. Turnouts in Sweden and Trafikverket’s perspective regarding turnout failures ... 58
3.1.2. Creating an object- oriented synthetic database for analysis ... 60
3.1.3. Data analysis of turnout malfunctions in eastern region – General trends ... 60
3.1.4. Data analysis of turnout malfunctions in eastern region: relation between malfunctions and operational/geometric characteristics of turnouts ... 61
3.1.5. Analysis of Causes of turnout malfunctions as recorded in Ofelia DB ... 64
3.1.6. Conclusions: data analysis, database construction and results of analysis of turnout malfunctions in eastern region ... 65
3.2. Modeling the probability of turnout faults by using logit models – An application to predictive maintenance planning ... 66
3.2.1. Logistic regression – principles ... 66
3.2.2. Modeling malfunction probability as a function of turnouts characteristics- separate influence of variables and statistical validation ... 68
3.2.3. Modeling malfunction probability as a function of turnouts characteristics- Combined influence of variables and statistical validation ... 69
3.2.4. Application of the model – an example ... 69
3.2.5. Conclusions: modeling turnout malfunctions with logistic models ... 70
4. Results ... 71
5. Topics for further research ... 75
Bibliography ... 76
Appendix 1 ... 80
Appendix 2 ... 82
Appendix 3 ... 83
Appendix 4 ... 86
Appendix 5 ... 87
Appendix 6 ... 88
Appendix 7 ... 91
6
1. Introduction
Railway maintenance across Europe is one of the most important endeavors a national railway organization must undertake – with an increasing cost around EU. Sweden is not exception:
according to Trafikverket’s maintenance plan 2019-2022 (Trafikverket, 2019) , for 2018 alone, around 8 billion Kr were disposed towards railway maintenance and by 2022 another 30 billion Kr will be disposed, with an increased expenditure each year. This includes not only preventive/remedial maintenance, but also reinvestments. In this grand scheme, one of the main points of focus is track infrastructure, with turnouts having a prominent position. Are turnouts important in the maintenance scheme and in railway network? Experts think so. Lichtberger (2005) underlines their importance, as they are prerequisite for the development of a highly productive network. However, they require high investment and large-scale maintenance, which makes them, as a premium component, expensive (1 meter of switch has up to 4 times higher cost than one meter of track). Atop of that, due to their structure, turnouts are prone to faults, which can seriously hinder rail operations.
In the light of the rising maintenance requirements and costs, experts seek a way to optimize the planning of infrastructure maintenance, stressing the need for departing from corrective and preventive maintenance regimes (both are highly costly), as well as from ad-hoc planning principles and general KPIs, to predictive preventive maintenance (Zoeteman & Esveld, 2004).
This approach comes from the risk analysis realm, where infrastructure faults are treated as probabilities, rather than certainties. This approach is highly promising, as it can truly optimize maintenance planning, for it can almost guarantee that maintenance intervention will occur neither after, nor before the damage but just when it needs to occur, Experts approach the optimization problems by frequently using mathematical models.
In Sweden, rail infrastructure is old and exposed to high traffic volumes, which are expected to increase. That triggers the need for high maintenance and reinvestment costs and turnouts are no exception, given their special place in it. At the same time, maintenance optimization is a way of improving the financial result and effectiveness of maintenance. There are many models, which attempt it, mostly through the prism of track possession optimization and few from the perspective of risk analysis. So, what if the infrastructure is itself positioned at the center of modeling? That means modeling the probability of individual infrastructure to fail as a function of its attributes, can give rise to a bottom-up pattern of maintenance, which will be very effective, since it reflects the condition of the assets. A central argument of this thesis is that can be done with the use of logistic regression models. However, which properties of a turnout must be taken under consideration and which are the parts and mechanics of a turnout and rail infrastructure in general that must be considered? Can this approach be viable in terms of improving the maintenance? These thoughts comprise the core of this study.
Background
Railway engineering literature has established well both the structural analysis of a track, as well as the description of the parts of a turnout, mostly from the scope of national railway organizations. Maintenance optimization modeling has also several representatives (see Liden , 2015; Liden & Joborn, 2017 for examples). However, there is not a single compiled text regarding these practices and the practice of maintenance modeling in general, in combination with structural analysis of a trunout, as well as describing its parts. Therefore, references from Kerr (2003), Lichtberger (2005), Esveld (2001), as well as the AREMA practical guide for railway engineering (2003) are primarily used to establish a structural analysis of the turnout and its parts.
Regarding the implementation of logit model in studying turnout faults, data about turnouts
provided by Trafikverket for the east region in Sweden is used, which can be considered as the area
under consideration. An attempt to use a risk analysis model in maintenance planning is done by
Consilvio, et al. (2018), but no previous study of that kind exists for the area under consideration. As
for logistic models, there is a sufficient body of literature , as well as practical applications in
transport planning (see Ortuzar & Willumsen, 2001).
7
Aims
The aims of this thesis is to 1) justify the need for a turn towards predictive preventive maintenance planning for the entire rail infrastructure – not only for turnouts, 2) to establish why turnouts are important in maintenance planning and why is important to forecast their faults, 3) to establish a framework for structural analysis of a turnout and 4) underline its peculiarities as a part of the track, 5) to examine what are the factors causing faults in turnouts and finally 6) to develop a logistic model which can be used to calculate the probability of a turnout to malfunction. To what degree aims are achieved is discussed on “Results” and “Topics for further research” chapters.
Method
Apart from using elements of structural engineering and applied railway engineering, the basic method is consisted of:
I. Using turnouts failure data from 2018 for the east region in Sweden from two different maintenance databases: Ofelia and BIS.
II. Combining them, along with inputs from Trafikverket technical documentation, a new synthetic database is constructed, where the malfunction is expressed with a binary manner (0-1) as the dependent variable and technical, geometrical and traffic characteristics are composing the independent variables. This will produce a logistic model, which will be able to forecast the probability of a turnout to malfunction or not.
III. Statistical verification of the model.
Limitations
The thesis scope is restricted due to several factors:
Geography: Due to the difficulty and restrictions imposed by Trafikverket regrading handling and distribution of data, only data for the east region were provided.
Time: In order to conduct a thorough study of the topic this thesis is dealing with, more time is required, which obviously is not enough in the context of a master thesis.
Data quality: database building is a whole topic by its own. This process of building a complete database for the purpose of modelling requires cross-reference of many data sources and it is time consuming. As a result, database will contain only basic characteristics of a turnout.
Types of turnouts: For this thesis only simple turnouts (left or right-handed) are considered. There are many more types of them, which won’t be discussed extensively.
Interviews: For this master thesis, no interviews were conducted, as the input from Trafikverket
was deemed sufficient. Furthermore, this is a topic based mostly on existing bibliography, technical
documents and statistical data, rather than on data collected on site.
8
2. Literature review
2.1. The railway wheel-track system – The dynamic interaction between rail infrastructure and rail vehicles – basic components of rail infrastructure
Turnout is a premium component of track infrastructure, meaning it has a configuration and parts that typical track does not. However, in terms of basic components and forces applied on it, it resembles a combination of a straight track and tracks on turns, without superelevation. Therefore, it is useful to revisit some key aspects of track parts and dynamics that are in play, in order to comprehend the loading environment in a turnout. The focus here is describing a ballasted track structure.
2.1.1. Track structure in a nutshell According to Selig & Waters (1994), a rail track must serve as a stable guideway with appropriate vertical and horizontal alignment. Therefore, each component of the system must perform its specific functions satisfactorily in response to the
traffic loads and environmental factors imposed on the system. These points are made more clear by Lichtberger (2005), who stresses that the track consists not only of individual components which must be viewed separately, but it must be viewed as the "railway wheel-track" system as a whole (Figure 1).
Finally, Edwards & Ruppert (2018) adopt an even wider perspective of the functions of a track. More specifically, a track has primary as well as secondary functions, which are
3:
2.1.2. Forces acting on a track
The main force applied on a track is the vertical wheel load (static or dynamic). The configuration and the components of the track itself are designed and arranged in such a way that allow the distribution and the reduction of the very high force of pressure in the wheel-rail contact point. Progressively, each layer distributes and reduces the initial load and as a result, the track remains in place (Figure 2).
In reality, since a train moves on the track, the interaction between wheel and track is much more dynamic and multidimensional. According to Esveld (2001), in rail track litterature, the wheel-rail forces are applied vertically (Q is the denomination for the vertical force-z-direction), laterally to the track –to its sides ( Y for the lateral force - y-direction), and longitudinally –in parallel to the track ( T for the longitudinal force - x-direction). In addition, one must consider the internal longitudinal temperature forces, which may be present and are indicated with symbol N (Figure 3).
3 R = Rail, S = Sleeper, FS = Fastening System, B = Ballast
Figure 1: Schematic representation of the Wheel-track system. Source:
(Lichtberger, 2005, p. Chapter 1)
Primary Secondary
1. Support and distribute train loads [R, S, FS, B] 6. Transmission of signal circuit[R]
2. Guide the vehicle[R,S,FS] 7. Broken rail detection[R]
3. Provide adhesion at wheel‐rail interface[R] 8. Path of ground return for traction power[R]
4. Provide a smooth running surface[R, S, FS]
5. Facilitate drainage[B]
Table 1: Functions of a track. Source: (Edwards & Ruppert, 2018)
9 Esveld (2001) provides a more analytical insight
of the causes and nature of track loads. He states that the forces acting on the track, as a result of train loads, are considerable and sudden and are characterized by rapid fluctuations. The loads can be considered from three main angles:
vertical
horizontal, transverse to the track
horizontal, parallel to the track
In addition, loads can be divided according to their nature:
quasi-static loads as a result of the gross tare, the centrifugal force and the centering force in curves and switches, and cross winds
dynamic loads caused by:
o track irregularities (horizontal and vertical) and irregular track stiffness due to variable characteristics and settlement of ballast bed and formation
o discontinuities at welds, joints, switches etc.
o irregular rail running surface (corrugations)
o vehicle defects such as wheel flats, natural vibrations, hunting.
He adds that there are also the effects of temperature on CWR track, which can cause considerable longitudinal tensile and compressive forces, which in the latter case can result in instability (risk of buckling) of the track.
Examining the loads from the perspective of their
angle, Esveld (2001) states that the vertical rail force is the total vertical wheel load, which is made of the following components:
𝑄
𝑡𝑜𝑡= (𝑄
𝑠𝑡𝑎𝑡+ 𝑄
𝑐𝑒𝑛𝑡𝑟+ 𝑄
𝑤𝑖𝑛𝑑) + 𝑄
𝑑𝑦𝑛(1)
𝑞𝑢𝑎𝑠𝑖 − 𝑠𝑡𝑎𝑡𝑖𝑐 𝑓𝑜𝑟𝑐𝑒𝑠
In which: 𝑄
𝑠𝑡𝑎𝑡: Static wheel load = half the static axle load, measured on straight horizontal track, 𝑄
𝑐𝑒𝑛𝑡𝑟: Increase in wheel load on the outer rail in curves in connection with non-compensated centrifugal force, 𝑄
𝑤𝑖𝑛𝑑: Cross winds forces, 𝑄
𝑑𝑦𝑛= dynamic wheel load components resulting from:
Sprung mass : 0-20 Hz
Unsprung
4mass: 20-125 Hz
Corrugations, welds, wheel flats: 0-2000 Hz
The total horizontal lateral force exerted by the wheel on the outer rail is:
𝑌
𝑡𝑜𝑡= (𝑌
𝑓𝑙𝑎𝑛𝑔𝑒+ 𝑌
𝑐𝑒𝑛𝑡𝑟+ 𝑌
𝑤𝑖𝑛𝑑) + 𝑌
𝑑𝑦𝑛(2)
𝑞𝑢𝑎𝑠𝑖 − 𝑠𝑡𝑎𝑡𝑖𝑐 𝑓𝑜𝑟𝑐𝑒𝑠
4 Unsprung mass: all the mass that is below the suspension of the train– weight of the wheel on the axle and parts of the track frame. Everything above the springs is sprung mass
Figure 2: Pressure distribution of the wheel force Q in the individual system components of the track. Source:
(Lichtberger, 2005, p. Chapter 2)
Figure 3: Rail forces and displacements. Source: (Esveld, 2001, p. Chapter 1)
10 In which: 𝑌
𝑓𝑙𝑎𝑛𝑔𝑒: Lateral force in curve caused by flanging against the outer rail, 𝑌
𝑐𝑒𝑛𝑡𝑟: Lateral force due to non-compensated centrifugal force, 𝑌
𝑤𝑖𝑛𝑑: Cross wind forces, 𝑌
𝑑𝑦𝑛: Dynamic lateral force component; on straight track these are predominantly hunting phenomena.
Finally, the horizontal longitudinal forces occur in the track as a result of:
Temperature forces, especially in CWR track. These forces can be considered as a static load
accelerating and braking
shrinkage stresses caused by rail welding
Rail creep (or creepage)
2.1.3. Other factors that amplify the forces
Considering all these forces applied on the track, it is of utmost importance to set some requirements for the bearing strength and the quality of the track. According to Esveld (2001), these requirements depend largely on the loading parameters:
axle load: static vertical load per axle
tonnage borne: sum of the axle loads
running speed
Track horizontal and vertical geometry The static axle load level, to which the dynamic increment is added, in principle determines the required strength of the track.
The accumulated tonnage is a measure that determines the deterioration of the track quality and as such provides an indication of when maintenance and renewal are necessary.
The dynamic load component, which depends on speed and horizontal and vertical track
geometry, also plays an essential part here (Esveld, 2001). In general, Esveld implies that as variables, speed, axle load and track geometry are closely related. Especially speed and track geometry increases the static forces applied on the track.
2.1.4. Track components
Selig and Waters (1994), state that a track structure is consisted of two main parts:1) the superstructure and 2) the substructure (Figure 4, Figure 5). Superstructure is consisted of the following elements:
Rails
Fastening system
Sleepers (ties or corssties)
The substructure on the other hand is consisted of:
Ballast
Subballast
Subgrade
Finally they note that these parts are separated by the sleeper-ballast interface. The components of a track will be analyzed below.
Figure 4: Track Structure components (parallel section). Source: (Selig & Waters, 1994)
Figure 5: Track Structure components (cross-section). Source: (Selig
& Waters, 1994)
11 2.1.4.1. Rails
Rails are the longitudinal steel members that directly guide the train wheels evenly and continuously. They must have sufficient stiffness to serve as beams, which transfer the concentrated wheel loads to the spaced sleeper supports without excessive deflection between supports (Selig &
Waters, 1994). In general, modern railways use the Vignole rail (or T-rail) as the rail of choice, come in sections, which size is measured in 𝑙𝑏𝑠/𝑦𝑎𝑟𝑑 or in 𝑘𝑔/𝑚𝑒𝑡𝑒𝑟. Rail brands (Figure 7) include the following information (Lichtberger, 2005) along with rail size: company, year of rolling, profile, steel sort.
The length of rail sections may also differ.
Lichtberger (2005) states that it is possible to roll
rails of 120 m length and to deliver them to the site, but standard rails used today are still only 60 m. In USA, rail sections are less than 39 feet (approximately 12 m) or 80 feet (approximately 24 m). He also states that greater rail section lengths have many advantages, in terms of productivity, economics and mechanical properties after installation. Finally, he states
that the rails are always laid on the track with an inward inclination, but in switches they are usually laid without inclination.
A big distinction between rail sections is the way of installation. According to Selig and Waters (1994), steel rail sections may be connected either by bolted joints or by welding. Bolted rails are most commonly used on curves to provide stress relief from thermally induced length changes or on secondary lines. However, joints increase track
deterioration in total (as many rail discontinuities) and increase maintenance costs. The other way is to use continuous welded rail (CWR). This approach is preferred on lines with high speed, with high axle loads, or with high traffic density. Rails can have a different gauge between them.
The most important properties of a rail section are the ones given by the manufacturer. Every rail section has its own properties related to its internal geometry, resistance to angular acceleration (moment of inertia), tension stress limits etc. These properties are important for conducting a structural analysis of a track in relation to the rails and their capacity to accommodate loads without breaking (Figure 6, Table 2). Values related to the hardness of a rail and its tensile strength (how big loads can it accommodate before breaking) can be given by Brinell hardness and the tensile test respectively (see Lichtberger, 2005 for more details).
In general, it can be said that heavier rail sections have bigger durability and higher tensile strength, which is also related to the quality of steel and existence of internal rail defects.
Figure 7: Rail Branding in USA. Source: (Edwards
& Ruppert, 2018)
Figure 6: Schematic representation of the stress-strain diagram. Source: (Lichtberger, 2005, p. Chapter 3)
Table 2: Moments of inertia and section modulus of rails. Source: (Lichtberger, 2005, p. Chapter 2)
12 2.1.4.2. Fastening systems
According to Selig and Waters (1994), the connections between the sleeper and the rails are achieved by the fastening system, which have many variations. The purpose of the fastening system is to retain the rails against the sleepers and resist vertical, lateral, longitudinal, and overturning movements of the rail produced by forces exerted on the track and by changes in temperature.
The fastening system to be used depends on the type
of sleeper (crosstie) which will be used and Selig and Waters (1994) describe these differences. With wooden sleepers, steel plates (tie plates) to distribute the rail force over the wood surface are used, which provides suitable bearing pressure and restrain lateral movements of the rail through friction. The size of sleeper plate is an important factor as it defines the
stress applied on wooden sleepers. The other parts of the fastening system of this type are the spike fasteners (essentially big nails) and the rail anchors. Selig & Waters state that the spike fasteners restrain the sleeper plates horizontally. Longitudinal rail movement must be restrained by separate anchors clipped to the rails and placed against the sides of the sleepers in the cribs. They also note that driven spikes provide little rail uplift restraint. Finally, they stress that this fastening system is used with jointed rails (Figure 8, Figure 9, Figure 10).
Selig and Waters (1994) also provide an insight in the fastening system used in concrete sleepers, which have spring fasteners (elastic fastening system). They provide vertical and longitudinal restraint as well as lateral. The main components of that system are
the clips (which secure the rail in all directions) and the rail pad assembly (mainly for load distribution). Pads are required between the rail seat and the concrete sleeper surface to fulfill the following functions (Selig & Waters, 1994):
1. Provide resiliency for the rail/sleeper system 2. Provide damping of wheel induced vibrations 3. Prevent or reduce rail/sleeper contact attrition, and 4. Provide electrical insulation for the track signal circuits
This type of fastening system is suitable for modern track and especially for a CWR track. In addition, the elastic downward pressure is essential for the smooth control of the rail's upward movement and high creep resistance (Lichtberger, 2005) (Figure 11). Finally, elastic fastening systems handle forces better, thus reducing the need for maintenance (Figure 12).
Apart from the previous categorization, there is another way of categorizing fastening systems.
According to Esveld (2001), fastening systems can be categorized as follows:
Direct fastenings entail that the rail and, if necessary, the baseplate are fixed to the sleeper using the same fasteners. Direct fastenings also include the fastening of track on structures without ballast bed and sleepers (Figure 10).
Figure 8: Typical fastening system in wooden ties.
Source: (Edwards & Ruppert, 2018)
Figure 9: Cross-section of a fastening system in wooden ties. Source: (Edwards & Ruppert, 2018)
Figure 10: Standard permanent way, USA. Source:
(Lichtberger, 2005, p. Chapter 4)
Figure 11: Fastening system in a concrete sleeper.
Source: (Edwards & Ruppert, 2018)
13
Indirect fastenings entail that the rail is connected to an intermediate component, such as the baseplate, by other fasteners than those used to fix the intermediate component to the sleeper. The advantages of indirect fastenings are that the rail can be removed without having to undo the fastening to the sleeper and the intermediate component can be placed on the sleeper in advance (Figure 13).
For Esveld (2001), indirect fastening has many more advantages, such as:
The vertical load is distributed over a
larger area of the sleeper. This lengthens the service life of the sleepers.
The horizontal load is absorbed better because of friction and because it is distributed over all the fastenings anchored in the sleeper. Baseplates are excellent for sustaining large lateral forces if large cant deficiencies are provided
The overturning moment causes less force on the fastenings in the sleeper
Baseplates have a high bending stiffness and grooves in the ribs provide good fastening locations for the rail
Baseplates give extra weight to the sleeper.
The only drawback of indirect fastening is the relatively high price.
2.1.4.3. Sleepers (crossties)
According to Lichtberger (2005), sleepers, ties or crossties have several important functions:
1. Establish and maintain track gauge.
2. Distribute and transmit forces to the ballast bed, such as, vertical, horizontal and longitudinal forces
3. Hold the rails in height, to the sides and in the longitudinal direction
4. to secure the track in cases of rail breakage and derailment
5. to dampen rail vibration and
6. to reduce the influence of sound and impact waves on the environment
Figure 12: Comparison of rigid and elastic rail fastenings. Source:
(Lichtberger, 2005, p. Chapter 4)
Figure 13: Cross section of the K permanent way.
Source: (Lichtberger, 2005, p. Chapter 4)
Figure 15: Forces on a baseplate. Source: (Esveld, 2001)
1 2
Figure 14: Indirect fastening to concrete sleepers (1), Vossloh 300 fastening system (2). Source: (Lichtberger, 2005, p.
Chapter 4)
14 In bibliography it is stated that there are many types of sleepers, but here, wooden and concrete sleepers are going to be analyzed, which are the predominant ones (Selig & Waters, 1994). Wooden sleepers are produced from different types of wood. In Europe, wooden sleepers are made of oak or beech and they have dimensions of 16 × 26 × 260 𝑐𝑚, (height-width-length) while in USA, various species of wood are used with typical dimensions around 18 × 23 × 240 − 270 𝑐𝑚 (Lichtberger, 2005). Of course these dimensions vary from country to country, depending on the standards set by the national railways (see Table 3). Wooden sleepers are divided in the following categories of wood (Esveld, 2001):
Softwood sleepers (pine-wood)
hardwood sleepers (beech, oak, tropical varieties). This type is stronger and has a longer service life. Hardwood sleepers are used, for instance, in switches and crossings and where fastenings are applied without base plates.
Esveld states that wooden sleepers (Figure 16) have to go a series of procedures before laid on the track. The total service life in years of some types of timber sleepers is: pinewood 20 - 25, beech 30 – 40 and oak 40 -60 (Esveld, 2001). In spite of all these procedure applied above, it is fair to mention that wooden sleepers are more elastic and lighter than concrete, but are more susceptible to climate and environmental conditions.
The other category of sleepers are the ones made from pre-stressed concrete. According to Esveld (2001), the development and use of concrete sleepers became significant after the Second World War owing to the scarcity of wood, the introduction of CWR track, and the improvements in concrete technology and pre-stressing techniques. As Kerr (2003) elaborates, the strength of concrete in tension is 1/10 of its compressive strength. At the same time steel is strong in both conditions but expensive.
The combination of them, by placing steel wires to cover the tensile stresses, gave birth to reinforced concrete. However, that was not a viable alternative for concrete ties, as cracks under the rail seat and in the center part of the tie were formed, which led to penetration of the tie by water that corroded the metal reinforcement.
Kerr (2003) narrates that the tendency of conventional reinforced concrete to crack when subjected to vertical loads may be eliminated by subjecting the concrete of a beam, artificially, to a compression stress, a method which is called prestressing. For concrete ties, this eliminates the tensile stresses in the concrete, however should the vertical load be
increased further and the prestressing force 𝑁
𝑜remain the same, tensile stresses will occur in the concrete. When the beam cracks due to overloading, the crack will close when the load is removed, provided that no solid material penetrates the formed crack. The basic idea of prestressed concrete ties is demonstrated in Figure 17 and Figure 18.
Table 3: Typical sleeper Dimensions. Source: (Selig & Waters, 1994)
Figure 16: Wooden sleeper and baseplate. Source:
(Esveld, 2001)
Figure 17: Principle of prestressed concrete ties. Source: (Edwards
& Ruppert, 2018)
15 The advantages of concrete ties over wooden
ones are the following (Lichtberger, 2005):
Longer life cycle and service life,
Less expensive than hardwood sleepers
Lower maintenance of the fastenings
Higher resistance to lateral displacement due to higher weight.
Esveld adds also that:
heavy weight( 200-300 kg), useful in connection with stability of CWR track.
great freedom of design and construction
relatively simple to manufacture.
However, concrete ties have also disadvantages such as (Lichtberger, 2005):
susceptible to shock and impact:
difficult handling due to greater weight, and
maintenance of longitudinal level is somewhat more difficult because of the higher moment of inertia and the lower elasticity
Esveld (2001) adds:
Less elastic than wood. On poor formation, pumping may occurs
Susceptible to corrugations and poor quality welds
Risk of damage from impacts (derailment, loading/unloading, tamping tines)
Dynamic loads and ballast stresses can be as much as 25% higher.
Esveld also states that there are two basic types of concrete sleepers (Figure 19):
Twin-block sleeper. This type consists of two blocks of reinforced concrete connected by a coupling rod or pipe (synthetic pipe filled with reinforced concrete). Advantages of that design are well-defined bearing surfaces in the ballast bed and high lateral resistance in the ballast bed because of the double number of surface areas.
Monoblock sleeper. This is based on the shape of a beam and has roughly the same dimensions as a timber sleeper. it is considered to endure the higher and intensive loadings better than the twin-block sleeper. Advantages of that design are lower price, little susceptibility to cracking and that it can be pre- stressed.
Regardless of the type of sleeper, its key parameters are a good bearing and bending capability: in other words, a sleeper should be able to bear the stress caused by load on a rail and be able to bend without braking. In addition, a second consideration is their spacing, which affects the
magnitude of forces applied on them.
Figure 19: Reinforced twin-block sleeper (up) and prestressed monoblock sleeper (down). Source: (Esveld,2001)
Figure 18: The effect of concentric prestressing on the concrete beam stresses. Source: (Kerr, 2003)
16 2.1.4.4. Ballast and Subballast
Selig and Waters (1994) define the ballast as the select crushed granular material placed as the top layer of the substructure in which the sleepers are embedded. They state that angular, crushed, hard stones and rocks, uniformly graded, free of dust and dirt, and not prone to cementing action have been considered good ballast materials, but no universal agreement exists as to proper specifications for the ballast material index characteristics. In addition, economics have to be considered when choosing ballast. Thus, a wide variety of materials have been used for ballast such as crushed granite, basalt, limestone, slag and gravel.
Selig and Waters also enumerate the functions of ballast in detail, which are:
1. Resist vertical (including uplift), lateral and longitudinal forces applied to the sleepers to retain track in its required position
2. Provide some of the resiliency and energy absorption for the track
3. Provide large voids for storage of fouling material in the ballast, and movement of particles through the ballast
4. Facilitate maintenance surfacing and lining
operations (to adjust track geometry) by the ability to rearrange ballast particles with tamping.
5. Provide immediate drainage of water falling onto the track
6. Reduce pressures from the sleeper bearing area to acceptable stress levels for the underlying material.
In Figure 4 and Figure 5, the subdivisions of the ballast can be seen. Ballast gradation changes over time because of (Selig & Waters, 1994):
1. Mechanical particle degradation during construction and maintenance work, and under traffic loading 2. Chemical and mechanical weathering degradation
from environmental changes
3. Migration of fine particles from the surface and the underlying layers.
Thus the ballast becomes fouled and loses its open- graded characteristics so that the ability of ballast to
perform its important functions decreases and ultimately may be lost. Subballast on the other hand, is the layer between the ballast and the subgrade. It fulfills two functions, which are also on the ballast list. These are (Selig & Waters, 1994):
1. Reduce the traffic induced stress at the bottom of the ballast layer to a tolerable level for the top of subgrade
2. Extend the subgrade frost protection
More specifically, subballast layer fulfills also functions that ballast cannot and these are related to separating ballast and subgrade, migration of particles between the various layers, prevent subgrade attrition by ballast, shed water and permit drainage of extra water (Selig & Waters, 1994). According to Selig and Waters (1994), the most common and most suitable subballast materials are broadly - graded naturally occurring or processed sand-gravel mixtures, or broadly-graded crushed natural aggregates or slags. They must have durable particles and satisfy the filter/separation requirements for ballast and subgrade.
In general, as discussed before the main key characteristics of any ballast layer is to distribute weights and facilitate drainage. Therefore ballast must have a specific height, big enough to absorb and distribute weights. As for the drainage function, it can be measured with ballast fouling index (Figure
Figure 20: Healthy track ballast. Source:
https://d1p2xdir0176pq.cloudfront.net/wp- content/uploads/Graham-Ellis-Track-3.jpg
Figure 21: highly fouled ballast. Source: (Dersch &
Ruppert, 2018)
17 20, Figure 21). This index measures health ballast, but also determine the hydraulic conductivity of it- in other words how much water can pass through the system. Fouling index can be measured in situ and defined as:
𝐹
1= 𝑃
4+ 𝑃
200(3)
Where: 𝑃
4is the percentage of particles passing 4.75 mm sieve and 𝑃
200is the percentage of particles passing a 0.075 mm sieve. The higher the index, the more fouled the ballast is and therefore the more water it holds, leading to the overall deterioration of the track.
2.1.4.5. Subgrade
The final layer of track system is the subgrade. According to Selig and Waters (1994), the subgrade is the platform upon which the track structure is constructed. Its main function is to provide a stable foundation for the subballast and ballast layers. The influence of the traffic induced stresses extends downward as much as five meters below the bottom of the sleepers. This is considerably beyond the depth of the ballast and subballast. Hence the subgrade is a very important substructure component which has a significant influence on track performance and maintenance. For example subgrade is a major component of the superstructure support resiliency, and hence contributes substantially to the elastic deflection of the rail under wheel loading. In addition, the subgrade stiffness magnitude is believed to influence ballast, rail and sleeper deterioration. Finally, subgrade also is a source of rail differential settlement. They also mention that anything other than soils existing locally is generally uneconomical to use for the subgrade.
2.1.5 Conclusions – Track system, loading and parts
It is vital to understand that trains and track infrastructure are a system: whatever happens to one affects the other and vice-versa. In this sense, trains apply forces on the track, vertical ,lateral and longitudinal. The purpose of a track is to handle these forces. The application of these forces is not static but dynamic, due to movement of train , geometry and track/wheel irregularities. Speed contributes to the increase of these forces. A typical ballasted track is comprised of several components;
each of them is performing a key function. Different configuration of these components (e.g. size of rail section or type of sleeper) affect the performance of the track, creating a totally different loading environment each time. Also, each component has several key parameters which define its performance:
rails have a certain stress limit, while sleepers have a maximum compression stress , affected by the sleeper spacing. These are important considerations when someone tries to understand how a ballasted track works and how different components affect track performance.
These notions become even more important when someone considers a turnout: its geometry is different from a typical track, forces are applied differently and some of them are even more important, as they are crucial in turnout deterioration, like the lateral and longitudinal forces. Furthermore, the selection of components for a turnout can distinguish turnouts which fail and turnouts which are robust and perform their purpose.
2.2. Turnouts
Railway turnouts can be considered as one of the most important parts of track infrastructure in railway networks. However, the terminology surrounds
them is somewhat baffling, in the sense that the same device is referred as a “turnout”, a “switch”, a “set of points” or Switches and Crossings (S&C’s). So, for purposes of clarity, this device is going to be referred as a railway turnout (turnout for short or spårväxel in Swedish). That confusion comes from the fact that in reality, a railway turnout is comprised of several parts, with the main being the switch and the crossing.
According to Kerr (2003, p. 1), main line tracks are consisted of three major parts: the superstructure, the substructure and special structures. Turnouts can be
Figure 22: A typical right hand turnout. Source:
https://sc01.alicdn.com/kf/UT8yEAUXxNXXXagOFbXy /202615050/UT8yEAUXxNXXXagOFbXy.jpg
18 categorized as a superstructure element. It must be noted that here only simple left or right handed turnouts are considered here.
The devices in the railway superstructure that allow trains to change from one track to another are called switches
5. The devices that allow trains to cross tracks are called crossings. This changing or crossing of tracks is a necessity to use the railway tracks in the most optimal way and to allow trains to be directed in different directions (Zwanenburg, 2007). The combination of a switch and a crossing constitutes a turnout (Figure 22). Esveld (2001, p. 333) states that turnouts are used to divide a track into two, sometimes three tracks and he purpose of crossings is to allow two tracks to intersect at the same level. A typical example of a left hand turnout is presented below (Figure 23). According to Lichtberger (2005), switches are of special importance for railways, as they are the prerequisite for the development of networks, i.e., for the branching and joining of tracks. He also states that the productivity and line speed of a railway is essentially influenced by the number and type of its switches.
Finally, he notes that the structure of a switch is far more complicated and expensive than that of the track grid. According to Bianculli (2003, p. 114), switch origins are ancient, but in 1303 A.D., similar devices were used in forestry industry, in a similar manner as in railways for timber transport. Finally, Gridley Bryant, the builder of the Granite Railway in 1826, was credited with inventing the original track switch. Of course, today, a railway switch is a way more complicated railway device, which has been refined and enhanced over the course of 200 years, through the process of small incremental improvements, a process that was implemented in other parts of railway infrastructure.
2.2.1. Designation of a turnout
According to Lichtberger (2005), turnouts are designated according to the following criteria:
Design type: single switch, double switch, single diamond crossing, double diamond crossing, etc.
Rail shape/profile: S49. S54, UIC60 etc.
Radius: 190, 300, 500, 1200 m
Ratio of inclination: this is the angle formed at the end of the switch by the tangent of the curve axis to the axis of the Main track; this angle is expressed by the ratio of inclination of both axles to each other (1:9, 1:12, 1:18.5 etc.),
Direction of the branch track (to the left L, to the right R)
Type of blade (loose heel switch, flexing point or spring switch blade)
Type of sleepers:(wooden sleepers, steel sleepers or concrete sleepers) and
Type of crossing/frog (fixed vs movable crossing/frog).
2.2.2. Parts of a turnout
As part of track superstructure, a turnout has many similarities with a straight or curved track.
However, it has also some premium or special components, which cannot be found in any other part of the track superstructure. In addition, its geometry makes it a distinct part of track superstructure.
5 “Switches” is the American-English term. The same device is also more generally known as “turnout” or in the UK as “set of points”.
Figure 23: A single left-hand Switch. Source: Lichtberger (2005)
19 According to Esveld (2001), a turnout is consisted of three major
parts, presented based on their sequence along a turnout (Figure 26):
1. Switch blades 2. Closure rails 3. Crossing (or frog) Switch blades or switch
According to Lichtberger (2005), the blade device or switch blades consists of the blades or tongue blades and the stock rails to which the blades cling. The initial point of the blade is called the switch tip and the end is referred to as the heel. The switch blades are connected by switch rods and operated as a unit (AREMA, 2003) with the help of a point machine. Geometrically, the switch blades part is defined from the Point of Switch (PS) to the Heel of Switch (HS), where according to AREMA (2003) Point of switch (PS) is the location where the diverging or straight route is determined and Heel of switch (HS) is the location at which the switch point pivots about.
Switch blades are mostly characterized by the angle they form with the stock rail, the geometry of their profile and their connection with the closure rails. Regarding the switch angle, it is the angle formed between the stock rail and the switch blades. According to Ruppert (2017), the angle of the switch blade is connected with the crossing/frog angle and with the heal spread, which according to Hay (1982) is the distance between the gauge sides of stock rail and the switch blades at the heel, in order to avoid an abrupt deflection (Figure 25). In addition, Esveld (2001) notices that the smaller the angle of the switch, the longer the switch blade is.
Figure 26: Parts of a turnout. Source: (Ruppert, 2017)
Figure 25: relation between frog angle (green angle), switch angle (red angle)
and heel spread (blue line) in a turnout. Source: (Ruppert, 2017) Figure 24: Cross-sectional drawing of switch blade and stock rail (asymmetric section). Source: (Esveld, 2001)
20 Regarding the geometry of their profile, as Hay (1982) states, the point itself is ground to a knife-edge and has a snug fit against the stock rail. Esveld (2001), describing the cross-section of a switch blade, states that in modern designs is an asymmetrical section that is lower than the standard rail profile. The asymmetric section also affects the moment of inertia of the blades: because of their asymmetric base, the moment of inertia is higher compared to a switch blade made of standard rail. Of course, some railways still use switch blades made of standard rails.
Finally, an important aspect of the switch blades is the fact that they can have simple or complex geometry, in terms of how the turnout negotiates the curve. Turnouts with straight switch blades are referred to standard geometry, while turnouts with curved (clothoid) geometry are referred to tangent geometry.
Regarding the connection with closure rails, AREMA (2003) states that the heel of each switch rail is connected to its lead rail by means of special joint bars, or in some cases is continuous, and the switch as a unit pivots about these connections. These connections are placed on the heel block assembly. According to
AREMA (2003), the heel block assembly maintains the correct distance between the gauge side of the stock rail and the gauge side of the points. It adds strength and rigidity. The block will be different for each switch and rail section. AREMA discusses two types of heel blocks: The conventional bolted heel block assembly, (Figure 29) permits movement of the point rails at the heel block. In the floating heel block the point flexes over its length. The floating heel block merely acts as a bearing point between point and stock rail to limit movement. Special plates are used under the heel block assembly.
Lichtberger (2005) discusses the topic further. He states that blades can move, as a joint or a spring element is arranged at the heel of the blade. On that remark, he distinguishes the blades into four categories:
loose heels switches,
flexing blades with switch rail plates
spring switch blades
flexing blades without switch rail plates.
Figure 27: Standard rail section switch blades. Source:
http://www.ostpubs.com/wp-
content/uploads/2016/05/Beginning-End_00.jpg
Figure 28: Asymmetric rail section switch blades. Source:
https://thumbs.dreamstime.com/z/railroad-switch-rails- device-connection-railway-tracks-intended-transfer-train-one- way-to-another-turnouts-track-127254651.jpg
Figure 30: Cross-sectional drawing of T-rail switch blade (standard rail switch blade). Source: (Esveld, 2001)
Figure 29: Bolted heel block assembly.
Source: (AREMA, 2003)
21 In conclusion, modern switch blades have an asymmetric rail profile, which enhances the handling of forces and durability. In addition, they tend to be welded to the closure rails. Finally, switch rails vary in length from 11 to 39 ft. (3.3528- 11.8872 m) and even longer for high turnout numbers, depending on the weight of the rail and the curvature of the turnout (AREMA, 2003).
Closure rails and stock rails
According to AREMA (2003), closure rails are the rails connecting the switch blades with the crossing/frog of the turnout and stock rails are the outside rails in a turnout that the switch blades bear against. In cases of left or right handed turnouts, one of these rails serves as a contact for the closed blade and as a running rail for the open blade (Lichtberger, 2005). Two of these rails are straight, while the other two are curved (corresponding to the straight and diverging route). The curvature is expressed in radius in Europe or in degrees of curvature in USA (Hay, 1982)and depends of the number of frog/ratio or the inclination of the turnout. Stock and closure rails are made of standard rail steel.
Finally, it should be noted that closure/stock rail section is defined geometrically from the Heel of the Switch to the Toe of Frog/crossing. These two points, according to AREMA (2003) are defined as:
Heel of Switch (HS) is the location at which the switch point pivots about
Toe of frog (TF) is the joint location ahead of the frog/crossing point connected to the closure rails.
Also, regarding the connection of closure rails and frog/crossing Hay (1982) states that joint between closure rails and crossing/frog are one of the few in a turnout, but Esveld (2001) states that these are welded or if necessary glued.
Crossing/Frog
The last part of a turnout is called the crossing or frog. According to AREMA (2003), a frog is a device at the intersection of two running rails to permit the flange of a wheel moving along one rail to cross the other rail. A frog is comprised of the following parts (University of Wisconsin - Madison, 1899) based on Figure 31:
The wedge shape part 𝐴 is the frog Tongue
The extreme point of tongue 𝑎 is the point
The space 𝑏 between the ends 𝑐 and 𝑑 of the rails is the mouth
The rails comprising the mouth have a narrow point 𝑒 which is called throat
The curved ends 𝑓 and 𝑔 are the wings
AREMA (AREMA, 2003) states that toe of the frog (TF) is the joint location ahead of the frog point connected to the closure rails, while Heel of frog (HF) is the joint location behind the point of frog. Hay (1982) suggests that the distance from frog toe to the point is the toe distance, while the distance from the point to the heel of the frog is called heel distance. Ruppert (2017) suggests that toe is simply the part from the joint ahead of the point to the point, while heel is the distance from point to the joint behind the point. Finally, as Lichtberger (2005) suggests, that the rails 𝑐 − 𝑔 and 𝑑 − 𝑓 are called the wing rails, which are the continuation of the two closure rails of the switch which are bent laterally leaving a flange groove. It can be stated that up to the throat of a frog, wing rails support the wheel, while from the throat to their end they guide the wheel flange.
Special note about the frog point must be made. According to Ruppert (2017), in theory, the frog comes to a distinct point (THEO. PT), but in reality, it is rounded off to what is called a ½ inch point
Figure 31: A detailed view of a turnout crossing/frog. Source: (University of Wisconsin - Madison, 1899)
22 (1/2 PF) in USA railway practice, because otherwise the impact of the wheels would destroy it. This point is not referred to the distance from Theoretical PT but where the width of this point is ½ inch (approx. 1,27 cm). Hay (1982) adds that for greater stability, the base of the point is continued into the throat and ground back from the theoretical point. So, for geometrical calculations, two points are used:
the theoretical point and the ½ in. point of frog (actual point of frog).
Turnout frogs or crossings are characterized by the mechanics of the point of more specifically by the way a frog is negotiating the gap between frog throat and the point as well as its geometry or more specifically the size of the frog angle or the frog number. Ruppert (2017) categorizes frogs into two big groups, depending on if they eliminate the gap or not: Fixed point and Movable point frogs (Figure 32). According to Esveld (2001), depending on the traffic load, different types of crossings are used.
For normal to medium axle loads and speeds up to 200 km/h, rigid/fixed crossings are used. For higher axle loads and higher speeds, crossings with movable parts have to be used (for pictures, see Appendix 3, Figure 101):
Bolted Rigid frog: They are fabricated from machined rail pieces with specialized blocks and bolts to guide and support the assembly. Bolted-rigid frogs are generally limited to use in yard and industrial tracks, where traffic is fairly light
on both sides, self-guarded frogs are not available, or the usage of second- hand frogs is desired.
Rail-bound Manganese (RBM) Frog: This Frog is made from high-strength manganese steel and held in place between the frog wing rails. It is the most common type of frog, used in many moderate to high speed and tonnage lines. It is also typically used for sidings, passing tracks, spurs, leads, etc.
Solid Manganese –Steel frog: In comparison
with RBM, this entire frog is a solid casting. These type of frogs are not that common (in North America) but some are used in transit service as alternative to RBM frogs.
Self-Guarded frog: This is a type of casted steel frog with raised “guard” cast onto wing rails: it eliminates the need for a guard rail opposite the frog but is only suitable for low-speed operation.
In this type of frog, the frog flange is simply not strong enough to withstand high - speed operation.
However, it is cost-effective, because it eliminates both the initial cost and the maintenance expense of two guard rails. It is commonly used in yard and industrial tracks. Finally, it is possible to be used on main lines, but only in cases where the speed is less than 30 m/h.
All the frogs presented above are fixed-point frogs. This implies that they all have a gap in the running surface of the rail to allow a wheel flange to pass through, but this causes impact loads as the wheel tread jumps the gap. Therefore, loads are concentrated on the frog point and as a result, it suffers damage and wear. In that process, flangeway width and depth are also affected. These damages are typically addressed by welding and grinding maintenance activities to restore the point and the top surface. In order to address all these issues and mitigate the problem, eliminating the gap can be considered as a main strategy, which produced a completely new series of frogs: Frogs with movable point. Esveld (2001) provides a good review:
Swing Nose frog: In this type of frog, the entire point “swings” between the two guardrails to close the gap. It has a nose made out of a machined and heat-treated block. Smaller crossings of this type have an expansion joint in the heel to compensate for the difference in length after switching from one position to the other.
Crossing with movable wing rails or spring frog: This type is used for small turnouts and when the length of the turnout is restricted. The elimination of the gap is done by moving the wing rails instead of the point (AREMA, 2003).
Figure 32: Fixed vs movable point frogs. Source: (Ruppert, 2017)
