11 /2015 November 1 € 22 1 C 11 180 www.eurailpress.de/sd

Rail Signalling and Telecommunication

• INSTANDHALTUNG • SICHERHEIT • ETCS Deutliche Effizienzsteigerung GFR 2000: Eine Lösung zur The implementation of durch den Einsatz von Gefahrenraumfreimeldung an ETCS Level 1 in the mobilen Anwendungen Bahnübergängen Slovenian sector of Corridor D • Satellite-based train control systems

Cost efficiency analysis of the satellite based train control system 31nSat in Germany

Benedikt Seheier / Anja Bussmann / Florian Brinkmann / Uwe Wendland

DB Netz AG is aiming at cost effective several international project partners the highest comparability with the oth­ alternatives to conventional train con­ are involved, including DB Netz AG and er systems that have no sophisticated trol systems on low density lines. For the German Aerospace Center (DLR). fall-back technology), this purpose, the cost efficiency of the The project's work plan comprises sev­ • ERTMS-Regional, (a cost-reduced 31nSat system - a satellite-based plat­ eral work packages regarding nation­ variant of ETCS Level 3 for region­ form for train protection - is analysed al scenarios of a number of countries. al lines without conventional lineside by comparing the net present value of Besides the economic assessment, the signals and without track-side train equipment cost for 31nSat with the al­ German scenario includes a compari­ detection) and ternative systems ERTMS-Regional, son of the system's functionalities with • Germany's common conventional sig­ ETCS Level 2 without lineside signals the operational procedures and require­ nalling system (the so-called Ks-Sys­ and conventional signalling. From the ments of the German rail system. From tem). results it can be told that 31nSat offers this, further requirements for the 31n­ The current track equipment does not a cost saving potential, mainly due to Sat system are derived. The analysis of affect this analysis since all scenarios the fact that physical Eurobalises can the cost efficiency, the so-called Busi­ are assumed to have equal dismantling be replaced by virtual balises. ness Case German Scenario, is pre­ costs. A re-use of any part of the pre­ sented in this article. lt is carried out ceding track equipment, as weil as any by the DLR in close cooperation with questions of migration are also excluded 1 lntroduction and motivation DB Netz AG. The aim is to deliver a ba­ from the scenario definition. sis for a decision about whether or not In contrast to the application of 31nSat Germany's national infrastructure man­ DB Netz AG should continue their par­ in other countries, in the German scenar­ ager DB Netz AG operates about ticipation in the pursuit of this satellite­ io the telecommunication components 33 300 kilometres of the German rail net­ based technology. That is why the anal­ of 31nSat are not considered. This is be­ work, 12 000 kilometres of which are ysis is strictly bound to the framework, cause it is still unclear if voice radio will characterised as low traffic density lines. conditions and boundaries of the Ger­ be realised in 31nSat. This functionality Most of these lines belong to the region­ man railway network. Since the cost ef­ is however mandatory in the German rai l al network. A crucial aspect of cost effi­ ficiency analysis is worked out in par­ network for passenger transport. On the ciency is the train control and manage­ allel to many other activities at a rela­ other hand, there is a dense voice radio ment system. Currently, most of these tively early stage of the project and be­ GSM-R network available in Germany on lines feature a system with relatively high fore the completion of a prototype, not most of the considered lines. Yet, an up­ capital and operational expenditures in all information about the 31nSat system grade is necessary in order to meet the relation to their number of trains. In order architecture and functions are available European EIREN E requirements to pro­ to improve the cost efficiency of these in detail, so a number of assumptions vide ETCS data communication between lines, DB Netz AG scouted for innova­ have to be made. trackside infrastructure (Radio Block tive alternative systems, amongst them Centre (RBC)) and trains. 31nSat and ERTMS-Regional [1]. The aim The line characteristics are derived of the following analysis is to assess the 2 Method and approach from generic line standards according cost efficiency of one of the most prom­ to DB Netz AG regulations (3]. DB Netz ising ones, which is the 31nSat - Train ln­ AG's network of regional lines consists tegrated Safety Satellite System [2]. 31n­ 2.1 Scenarios of three standards, called R1 20, R80 and Sat offers satellite-based train position­ GSO, the characteristics of which are ing and enables the replacement of fixed In order to evaluate the cost efficiency of summarised in Tabel 1. Eurobalises by virtual balises. In doing 31nSat , the life cycle costs of this sys­ The most significant differences be­ so, it is supposed to meet the Safety ln­ tem are compared to alternative technol­ tween them are the number of tracks, tegrity Level 4 (SIL4) safety requirements ogies. Thus, the cost saving potential of the number of stations and the block and to be compatible with the ERTMS the 31nSat system due to the reduction lengths. standard. Furthermore, the system pro­ of fixed Eurobalises will become appar­ Since R120 and R80 are described by vides functionalities for satellite-based ent. an upper and lower limit, the arithmetic telecommunication between train and For the low density lines in focus, there mean of every attribute constitutes an infrastructure. are three alternative signalling systems additional "mean scenario" for R1 20 and In the 31nSat project, this system is to tobe considered: R80. Altogether, this leads to a number be developed, tested and validated in a • ETCS Level 2 without lineside colour­ of seven generic lines. The total length real set-up. Led by Ansaldo STS and co­ light signals (since a variant without of the network considered for the ap­ funded by the European Space Agency, conventional lineside signals provides plication of 31nSat in Germany adds up

36 SIGNAL + DRAHT (107) 11/201 5 Satellite-based train control systems • to around 10 000 kilometres of regional lines (as shown in Table 1) plus an ad­ R120 R80 G50 ditional 2 000 line kilometres of the G50 upper lower upper lower standard within the core network. The limit limit limit limit generic lines have a standardised length line length (generic) [km] 100 100 100 100 50 of 100 and 50 kilometres respectively. By track length [km] 180 120 100 100 50 means of the parameter "line length (re­ number of tracks [-] 1-2 1-2 1 gional network)" the cost analysis can however be extended to any network in number of stations [-] 4 4 9 5 3 focus. mean distance [km] 20 20 10 17 17 The number of trains to be equipped is between stations assumed to be seven for R120, five for block length [km] 5 10 no blocks --> station distance R80 and three for G50. Due to the fact line speed [km/h] 81-120 81-120 51-100 51-100 < 50 that they are presumed to vary in prac­ number of trains per tice, these figures are subject to a sensi­ day and direction [-] 50 25 30 18 5 tivity analysis. regional trains per day [-) 40 20 25 13 All in all the cost efficiency of the 31n­ and direction 0 Sat system will be analysed on the ba­ freight trains per day sis of a cost comparison between four [-] 10 5 5 5 and direction 5 equipment alternatives for seven refer­ line length (regional ence lines to allow for conclusions about [km] network) 4971 4639 287 the systems' suitability for different line characteristics. Table 1: Characteristics of generic lines {source: authors] 2.2 Assumptions for alternative systems

ETCS Level 3 ETCS Level 2 In this analysis, costs are not contrasted to any benefits. This is due to the fact system elements 31nSat ERTMS-R ETCS L2oS Ks-System that the benefits of the compared sys­ lnfrastructure tems are considered equally high, as no significant differences in track capacity, electronic IXL with OFC X X X X revenues, safety or external effects were lineside colour-light signals X identified. Table 2 shows a comparison lineside axle counters X X of the elements that determine the sys­ intermittent ATP (PZB) X tems' costs. The most significant feature of 31n­ ETCS RBC X X X Sat in terms of cost reduction is the fact ETCS Eurobalises (fixed) X X that no fixed Eurobalises are needed. GSM-R upgrade for In return 31nSat requires a special train ETCS L2/L3 X X X equipment in addition to the ETCS on­ 31nSat EGNOS-service x board equipment. This includes GNSS­ 31nSat TAL-Server x Receivers and antennas, a location de­ termination system (LOS) as weil as a Train track database in the form of a digi­ PZB on-board equipment X tal map that provides the position of ETCS on-board equipment X X X each Eurobalise and its contained in­ 31nSat on-board equipment x formation. The allocation of the digital map as an element of the train equip­ Table 2: System elements of equipment alternatives [source: authors], ment is based on an assumption, since OFC = Optical Fiber Gable; /XL = /nterlocking; ATP = Automatie Train Protection; no detailed information about its archi­ PZB = "Punktförmige Zugbeeinflussung"; TALS = Tracking Area LOS Server tecture is avai lable at present. lt is also possible that it may become a central­ ised element of the infrastructure. Also, in terms of the infrastructure, additions reasonable to assume the same for the tenance centre, and includes the cost have to be made in the form of the so­ rail sector. However, the possibility of a of all project phases from conception called TAL-server that provides the train charge cannot be absolutely excluded. to implementation. Each field element with high-precision positioning data All other cost components are the has a specific factor of SEU that it re­ from the European Geostationary Nav­ same as for ERTMS-Regional and ETCS sembles. By allocating a cost factor to igation Overlay Service (EG NOS) and Level 2, including the interlocking. For one SEU, the cost of the whole inter­ reference stations which are subsumed each system, the cost of the interlock­ locking can be calculated by summing under the 31nSat EGNOS-service cost ing is derived from the sum of its signal­ up the SEUs of its field elements. In the element. There is no detailed informa­ ling equivalent units (SEU). A SEU rep­ case of ERTMS-Regional for example, tion available at the moment whether resents outdoor and indoor installations where the axle counters have to be sub­ this service will be charged. In the avi­ of a field element as weil as its share tracted, the mere cost of the hardware, ation sector, it is a free service, so it is of the operations control and main- i. e. the physical field element and cen-

SIGNAL + DRAHT (107) 11/2015 37 -

• Satellite-based train control systems

tral axle counting equipment itself are tal expenditure. This value resembles straightforward allocation of expen­ discounted, since the logic behind the roughly the ratio of operational to capi­ ditures between these two actors, the clear track detection and its processing tal expenditures of other field elements. costs for trackside and train equipment within the interlocking is required for To cover this assumption, a ten times are disclosed separately. External cost s ETCS Level 3 applications, too. higher value is tested in the sensitivity like emissions or effects from modal The actual number of fixed Eurobalis­ analysis. shift are not taken into account, since es has an important influence on the The benefit of ERTMS-Regional and neither a usable line capacity increase cost reduction of 31nSat and results 31nSat in comparison to ETCS Level 2 nor a traffic increase is expected from a from the following assumptions: Eve­ lies in cutting out lineside axle counters. change of infrastructure equipment on ry home signal as weil as every start­ Despite the fact that a substitutional regional lines. er signal requires on average three fixed technical solution for train integrity con­ The initial capital expenditures are Eurobalises. Level crossings with au­ trol in both cases is not defined yet, no distributed in a 08-specific ratio over tomatic train protection hold on aver­ further cost for this functionality is as­ the five years of the start-up phase. age three Eurobalises for each track sumed here. Finally, the conventional Ouring the following operation phase, and direction. Since in the regional net­ Ks-System does not contain any ETCS­ the capital expenditures as weil as the work the density of level crossings is component but lineside signals and in­ operational expenditures of each sys­ high enough, there is no need to take termittent ATP equipment (the so called tem are determined for every year of the the re-positioning Eurobalises every 1.8 PZ8) at trackside and on board. life cycle. The reinvestment costs are kilometres into account. For the same lt is assumed that there is no differ­ considered in the respective years ac­ reason, the Eurobalises for block sig­ ence in operation costs between the al­ cording to the individual life span of the nals are not considered. The number of ternatives. Equally, in case of failure, op­ components. In order to calculate the Eurobalises was calculated individually erational rules provide the fall-back solu­ net present value of the life cycle cost, a for each scenario. At an average, there tion for every system. 08-specific int erest rate is applied. The are 1.9 Eurobalises per kilometre of sin­ capital expenditures include the hard­ gle track. 8esides that, the operation­ ware and all related project planning al costs are decisive for the profitabil­ 3 Calculation of life cycle cost and implementation costs. Operational ity of 31nSat, i.e. the costs for mainte­ expenditures consist mainly of mainte­ nance and replacement of fixed Euro­ lt is assumed that each system will nance costs and service charges. En­ balises. There are only few experienc­ be operated for 40 years after a span ergy costs are not considered due to es from the operation of Eurobalises in of five years installation time. Disman­ their minor share of the overall costs, the German network, so their opera­ tling costs are not considered. The cost and personnel costs are excluded be­ tional expenditures have to be estimat­ comparison of the alternatives wi ll be cause they are assumed to be equally ed. From the DLR's and DB Netz AG 's based on the net present value. This high in each t echnology. To account for point of view it is reasonable to assume view includes the expenses for the rail­ rising prices, the costs for capital and operational Eurobalise costs in the way undertaking as weil as for the in­ operational expenditures are subject to magnitude of about 1 % of the capi- frastructure manager. To allow for a a nominal increase per year. Finally, the most influential paramet ers as weil as parameters with high uncer­ Figure 1: Accumu­ t ainties regarding their value are under­ line standard RSO (5 trains, 100 km) lated costs of the going a sensitivity analysis, in order to alternatives find out how the results of the cost com­ • train equipment (source: authors) parison vary depending on those single • ETCS RBC cost factors.

• ETCS lineside GSM-R Upgrade • infrastructure without ETCS GSM-R and RBC 4 Results and outlook

The scenarios on all R80 and R120 lines show basically the same ranking

80% of the four compared technologies. The c.> R80 mean scenario gives a representa­ z tive impression of the overall outcome ~ (figure 1) . .5 "' 60% The life cycle costs only differ slight­ ~u ly between the four technologies. The ~ highest cost share is allocated to the in­ ....aJ rCI 40% frastructure, with the interlocking as the :; most influential cost component . As ex­ E :::1 pected, 31nSat has the lowest costs in u u infrastructure but highest in train equip­ ~ 20% ment. The reason why the conventional signalling system has the lowest over­ all costs seems to be the investment for the GSM-R-upgrade for ETCS. Since 3inSat ERTMS-R ETCSL2oS Ks-System these costs, as weil as those for ETCS train equipment, appear relatively high

38 SIGNAL + DRAHT (107) 11/2015 T Satellite-based train control systems • in relation to the sum of life cycle costs, further observation is required: ETCS target scenario As part of a sensitivity analysis, a sce­ nario is set up in which lower ETCS train line standard RSO (5 trains, 100 km) equipment costs are reached and the costs for lineside GSM-R upgrade for • t rain equipment the ETCS variants are negligible. This assumption is reasonable, since there • ETCS RBC will be future research in order to devel­ op alternative communication technol­ • ETCS lineside GSM-R Upgrade ogies that will exceed GSM-R in terms • infrastructure w ithout ETCS GSM -R and RBC of cost-efficiency. Lower ETCS costs might be achieved through the Euro­ pean research project openETCS. This project aims at a cost efficient and re­ liable implementation of ETCS train G.I equipment. Especially the development ::::1 process of the on-board ETCS software ~1 80% should be optimised by using open ...c standards (4)(5). The results are shown .,,G.I in figure 2. In this case, 31nSat and ERT­ „G.I Cl. MS-Regional become financially much 1 ...G.I more attractive and the conventional z 60% signalling system is the least favoura­ ...... \III ble one...... Furthermore there are a num ber of c III other uncertainties regarding decisive t; input data. The economic efficiency of 0 1.1 40% 31nSat mainly depends on the quanti­ -"'C ty and cost of Eurobalises, the number QJ of trains to be equipped and the cost +" for special 31nSat infrastructure and "S"' E train equipment. Whereas 31nSat spe­ :;:, 20% 1.1 cific infrastructure cost is presumably 1.1 negligible, the cost for train equipment "' like the track database will be a signif­ - icant cost driver and difficult to esti­ mate at this stage. With these param­ eters, a pessimistic and optimistic sce­ 3inSat ERTMS-R ETCSL2oS Ks-Syst em nario from the 31nSat point of view is set up. For the figures in question, oppo­ Figure 2: Accumulated costs in the ETCS target scenario (source: authors) sitional assumptions are made. Pessi-

SYSTEMLÖSUNGEN FÜR DEN VERl

SIGNAL + DRAHT (107) 11/2015 39 • Satellite-based train control systems

5 Conclusion 31nSat pessimistic scenario 31nSat optimistic scenario line standard RSO (7 trains, 100 km) line standard RSO (3 trains, 100 km) As a conclusion, it can be stated that • 31nSat train equip. (100 k( per lrain CapEx) • 31nSat train equip. (200 k ( per train CapEx) 31nSat offers considerable economic • TALS (100 k( per RBC CapEx) • TAL5 (100 kC per RBCCa pEx) • EGNOS service (2 kE: per lrain OpEx) • EG NOS service (0 t) potential and further studies should be • Eurobalise.s (1 per trac:k.km, 23 € OpEx) • Eurobalises (3 per track-km, 216( OpEx) conducted as soon as the cost-factors

100% database has improved . For this reason DB Netz will participate in further devel­ opments of the "virtual balise" concept ii:: 80% z using GNSS technology. Therefore, as soon as new cost information about the .: 60% § 31nSat system components will be avail­ :;;- able, the cost-efficiency analysis will be ~ 40% updated. The satellite based train con­ :; E trol system 31nSat will be developed fu r­ ther in the framework of the Horizon2020 I 20% project ERSAT-EAV; likewise, the cost figures for Eurobalises will be refined. 0% Meanwhile, there will be further investi­ 31nSat ETCS with ftx balises 31nSat ETCS with fiK balises gations about finding a more cost effi­ Figure 3: 3/nSat optimistic and pessimistic scenario (accumulated costs) (source: authors) cient alternative for the train radio and data communication in Germany, since GSM-R will be obsolete in the next dec­ ade. mistic assumptions are: a high number to be low. Additionally, a fee for the EG­ of trains to be equipped accompanied NOS service is assumed. The optimistic by high 31nSat equipment costs and a scenario is constituted by opposed as­ LITERATURE small number of Eurobalises to be re­ sumptions. [1] Coenraad, W.: ETCS Level 3: From High placed by virtual bali ses. The Eurobalis­ Figure 3 contrasts the pessimistic and Speed Vision to Rural Implementation, Sl­ es' operational expenditure is assumed optimistic view on the sum of cost el­ GNAL+DRAHT, 01-02/201 2, p. 47 - 49 ements that differ between 31nSat and [2] Rispoli , F. , et al. : Satellite localisation ETCS with fixed balises (ERTMS-R and and IP-based public telecoms for ERTMS, SIGNAL+DRAHT, 11 /201 4, p. 52 - 58 The authors ETCS L2oS). The effect of the different scenarios is substantial: In the pessi­ [3] Brückmann, G.: Deutsche Bahn Richtlinie 413 Bahnbetrieb - Infrastruktur gestalten, Benedikt Seheier mistic scenario the 31nSat system is not economically efficient whereas in the 2002 Researcher [4] http://openetcs.org/, 27 .08.2015, 15:30 opposite case it is distinctly efficient, German Aerospace Center (DLR), [5] https://itea3.org /project/openetcs.html, Institute of Transportation Systems compared to a conventional ETCS al­ 27 .08.2015, 15:30 Address: Lilienthalplatz 7, ternative. D-38108 Braunschweig E-Mail: [email protected]

Anja Bussmann Researcher German Aerospace Center (DLR), • ZUSAMMENFASSUNG Institute of Transportation Systems Address: Lilienthalplatz 7, Analyse der Wirtschaftlichkeit eines Systems zur satellitengestützten Zugortung D-38108 Braunschweig E-Mail: [email protected] Die DB Netz AG ist an kosteneffizienten, sicheren und zuverlässigen Alternativen für konventionelle Systeme der Leit- und Sicherungstechnik fü r regionale Strecken in Florian Brinkmann Deutschland interessiert. Das Projekt 31nSat hat als Ziel, einen Demonstrator zu ent­ Researcher wickeln, welches Satellitennavigation verwendet und SIL-4-Anforderungen und Kom­ German Aerospace Center (DLR), patibilität zu dem ERTMS-Standard erfüllen soll. Im Rahmen dieses Projekts haben Institute of Transportation Systems die DB Netz und das Deutsche Zentrum für Luft- und Raumfahrt (OLR) eine Wirt­ Address: Lilienthalplatz 7, schaftlichkeitsanalyse im Vergleich zu den alternativen Systemen ERTMS-Regional, D-38108 Braunschweig ETCS-Level 2 ohne konventionelle Lichtsignale und dem konventionellen Signalsys­ E-Mail: [email protected] tem (Ks-System) erstellt. Für die Analyse wurden generische Streckenstandards der DB verwendet, um die Ergebnisse auf die entsprechenden Teile des OB-Strecken­ Dipl. -Ing. (FH) Uwe Wendland netzes übertragen zu können. Für die jeweiligen Referenzstrecken und die Ausrüs­ ETCS Requirements Management tungsvarianten wurden mittels der Nettobarwertmethode die Kosten berechnet und DB Netz AG, l.NPS 321 vergleichend dargestellt. Ergebnis der Kostenrechnung und der anschließenden Sen­ Address: Mainzer Landstraße 201, sitivitätsanalysen ist, dass 31nSat ein hohes Kosteneinsparpotential bietet, welches D-60326 Frankfurt am Main maßgeblich aufgrund des Prinzips, physische Eurobalisen durch virtuelle zu ersetzen, E-Mail: zurückzuführen ist. Da aufgrund des Entwicklungsstadiums von 31nSat noch viele An­ uwe. [email protected] nahmen getroffen werden mussten, wird die Wirtschaftlichkeitsanalyse fortwährend aktualisiert.

40 SIGNAL + DRAHT (107) 11/2015