Design of the Energy Storages for a (Battery) Electric Commuter Train

The use of battery-powered electric commuter trains instead of those which are propelled by internal combustion engines on only partially (or non) electrified railway lines can save considerable amounts of energy and can avoid exhaust emissions to a large extent with comparatively low investment costs. As path to these goals, the railway line, the operating and the trains are modeled for a particularly interesting, real-life example. Then, for the generated models, it is shown how the energy demands of model-compliant (battery) electric commuter trains can be calculated and how their energy storages can be designed. The models and the procedure can also be applied for a large number of comparable dimensioning cases when defining further railway lines, operating and commuter train models, corresponding parameterization as well as the occasionally necessary addition of slope and curvature resistances. The calculation of energy demands and the design of energy storages, however, show that there are a large number of different alternatives for technical specifications as well as decisions relevant to energy efficiency and business management, too. To determine the most suitable solutions, further measures such as the investigation of the effects of intelligent modifications to the underlying models, the development and implementation of computer-aided simulations as well as the construction and testing of real (battery) electric commuter trains make sense.


Introduction
In discussions about a future, more sustainable energy supply, measures to increase energy efficiency do not always receive adequate attention. Energy, which is not needed at all, is the most environmentally friendly energy and, with a predicted share of twothirds, almost the key to reach the German Federal Government's targets "Halving primary energy demand" and "Reducing electricity demand by one quarter" by 2050, which were defined in 2010. If the highly probable transition to e-mobility on the road by 2050 is excluded, which will have a favorable effect on primary energy demand as a result of improved conversion factors, and only moderate increases in demand are implied, the aforementioned electricity savings target could already be achieved alone by the measures listed in Table 1.  3 Schmetz, Struck From powertrain engineering, however, it is additionally known that the energy demand can often be significantly reduced by reducing friction and moving masses and, if possible, recuperating the energy required for unavoidable acceleration and lifting operations of the remaining (large) masses. In the transport sector, especially city buses, trams and commuter trains have a great potential for recuperation of kinetic (but also potential) energy due to their typical, strongly intermittent load profiles. This is the reason why some public transport companies already use (battery) electric city buses, which can (sometimes additionally) be supplied from occasionally existing direct current (DC) electricity grids. Likewise, some (battery) electric trams with the same property are already being tested [1]. In contrast, for railway applications only trials of diesel-electric hybrid commuter trains are known so far [2]. With regard to the currently fiercely focused emissions of internal combustion engines and the finite nature of fossil fuels, however, diesel-electric hybrid commuter trains do not represent a sustainable development goal anymore. In view of the increasing number of passengers and for protection of the environment, organizations such as the German Verkehrsverbund Rhein-Ruhr (VRR) are rather striving for the further expansion of the electrified railway network. Thus in its current plan for regional transport [3], the VRR calls repeatingly for the electrification of the 65 km long, non-electrified section of the left Rhine railway line (Nijmegen/NL-) Cleves-Geldern-Krefeld (-Dusseldorf/Cologne). This situation, the meanwhile usual tender for the operation of regional railway lines in rather smaller lots for periods of fifteen years, for which often own, specific trains are procured, as well as a number of other particularly typical features which simultaneously simplify the necessary calculations, are motivation according to [4] and [5] to investigate for this railway line once the use of (battery) electric commuter trains (BECTs) and carry out a design of the required battery storages.

Present Situation
Meanwhile, the left Rhine railway belongs with its northern section Cleves-Geldern-Krefeld (the former connection from Nijmegen/NL to Cleves was cut in 1991) and the section Krefeld-Dusseldorf to the diesel network Niederrhein/Ruhr/Münsterland (also called Niers/Rhine/Emscher-network) according to the left part of Figure 1, which contains yet some more strong frequented, but only partially electrified railway lines.   Figure 1). Nevertheless, and despite additional increases to triple traction during peak hours standing passengers (even in the first class) are not the 5 Schmetz, Struck exception but more likely in the southern part of the non electrified section of this railway line because of completely occupied seats. Therefore, several local authorities and political parties vehemently demand (in the meantime) the complete electrification (and the reconstruction of the second track in the section Cleves-Geldern). According to the German newspaper Rheinische Post of March 7, 2016, the network branch of the German railway company Deutsche Bahn AG estimates the cost for the electrification and reconstruction of the second track to about EUR 500 million [6]. Upon request in June 2016, about EUR 155 million from this money are needed for the (double-track) electrification. This figure, consisting of costs of EUR 2 million/km (double-track) and additional costs of EUR 25 million for two transformer stations including traction power lines, is in principle also confirmed by the spokesman of the German passenger association Pro Bahn [6]. It also roughly corresponds to the cost of EUR 12 billion proposed by the German company Siemens AG in promoting its eHighway concept in 2012 for the electrification of a highway backbone network with a length of 6000 km, but including the direct DC converter stations for feeding. In contrast, only about EUR 1.25 million/km (in the case of obviously existing transformer stations) are necessary for the fully projected completion of the electrification of the double track city line S 28 from Kaarst to Mettmann according to the standard Re 100, which is actually under construction [7]. Even with the use of this lower figure, the cost of electrification of the part Cleves-Geldern-Krefeld would still be in the order of EUR 100 million, a sum which, despite the rather dense traffic of light passenger trains, cannot be justified with regard to the actual cost of fuel and the return on investment, since the cancellation of freight and heavy passenger trains two decades ago. In addition, there would be the need to carry out a so-called planning approval procedure, in which, according to experience, many objections can arise, which elimination can lead to further, substantial cost increases.

Model of the Railway Line
On the railway line Cleves-Geldern-Krefeld-Dusseldorf, including the start and end stations, there are thirteen stations in the distances according to the table in the right part of Figure 1. All trains stop at all stations. Since there are no significant topographical obstacles on the left lower Rhine, despite the fact that the Rhine-Maas watershed has to be crossed twice, the railway line can be considered as straightforward and level, so that no slope and curvature resistances must be taken into account in the following 6 Schmetz, Struck calculations. Although the railway line is designed for a maximum speed of 140 km/h in the examined section, it is driven in normal operation only with a top speed of 120 km/h. For the sake of simplicity and estimation to the safe side, it is assumed for all travels between the stops that all trains accelerate as quickly as possible to the maximum speed of 120 km/h, which then is maintained as long as possible, before the trains brake down with a deceleration of -1.13 m/s². In reality, however, there are some inverter drives distributed to the four driven axles and one or more energy storages instead of the two internal combustion engines, the two gear transmissions and the two fuel tanks. This BECT, which consists of two single train units, is dimensioned to provide the same performance as the DCT. Beyond these settings, its further specification takes place step by step in the sections 3 and 4.

Calculation of Energy Demands
The calculation of energy demands is based on the DCT. For this purpose, the driving resistance to be overcome must be determined first. In principle, the driving resistance is composed of the components rolling resistance, slope resistance, acceleration 8 Schmetz, Struck resistance and aerodynamic drag [8], if the breakaway resistance is neglected, which only occurs for a very short moment at start. In case of railways, the resistance when passing through track curves must also be taken into account. But here it can be excluded as well as the slope resistance due to the above-defined railway line model. This results in the formula (1) for the driving resistance given in the upper left part of Figure 4. With the data of the DCT defined above and excluding the acceleration resistance according to formula (2), the speed-dependent driving resistance can be calculated, which is represented by the lower graph in the right part of this figure.

Figure 4: Driving Resistances as a Functiont of Speed
Because the formula (1) for the calculation of the driving resistance is rather complicated in praxis, in Germany there are additionally Strahl's formulas for the calculation of the driving resistance of freight trains and the also quite complicated Sauthoff formula for the calculation of the driving resistance of passenger trains (formula (3) in the lower left part of figure 4), valid for calculations at constant (final) speeds [8]. In the Sauthoff formula, the rolling resistance and the aerodynamic drag are combined by reference to the weight force to a speed-dependent driving resistance rL, 9 Schmetz, Struck which then only has to be multiplied by the mass of a train to obtain its overall driving resistance at a constant speed.
Since the Sauthoff formula (3) considers in addition to the components due to the difference of the formulas (1) and (2) also the increase of the rolling resistance coefficient due to the type of wheel bearings and the current speed, the effects of the length of a train, its induced aerodynamic undertow and a mean headwind on the aerodynamic drag, it delivers a stronger speed-dependent graph for the driving resistance according to the upper graph in Figure 4 than the difference of the formulas (1) and (2). Therefore, for the purpose of an estimation to the safe side, here the Sauthoff formula is used for the determination of the energy demands for overcoming the driving resistances. The results according to Sauthoff's formula, however, must be increased by the acceleration resistances due to formula (2), which have to be overcome after each stop for the accelaration from standstill to the maximum travel speed of 120 km/h. By analogy with the five-step transmissions of type Ecomat 5HP902 made by the German company Zahnradfabrik Friedrichshafen AG (ZF) which are actually installed in the LINT 41 trains and assuming a suitably matched combustion engine characteristic, it can be assumed for the DCT according to Figure 2, that the accelerations in the five steps are sectionally constant. Similar to normal operation, this leads to distances of 3000 m for each acceleration process after depature and after each stop as well as to the respective distances that can be traveled at the maximum speed of 120 km/h, and to the related energy demands according to Table 2. In contrast, the braking distances before each stop take only 500 m, and during each of the related braking phases, another 0.9 kWh of energy are needed to overcome the driving resistance. These energy demands, however, support the necessary withdrawal of kinetic energy during the braking phases. But they lead to a reduction in the recuperable energy, too.
In addition, the idling demands during standstill and braking phases must be taken into account for the DCT, during which electrical drive systems can largely be switched off or used for recuperation. Under the usual assumption that 10% of the indicated power is needed to overcome the internal friction in the four internal combustion engines of the DCT with an effective power of 315 kW, thus 35 kW each, these energy demands are at a total of 2.3 kWh per minute idle. Furthermore, it has to be taken into account, that for proper operation the internal combustion engines must be run in each case at least five minutes before departure from the starting station and after arrival at the destination 10 Schmetz, Struck station, and during the standstill and braking phases within the travel distances. Therefore, the related idle demands are listed in Table 2, too.  2 Energy demands during standstill and brake phases These energy demands for idling have to be increased by the requirements of the auxiliary drives yet. In case of the DCT, these requirements are taken into account with an average power demand of 50 kW for driving of generators for the on-board electrical system, compressors for the air brake systems, coolant pumps and fans for the waste heat of the internal combustion engines, and the lubrication of the gear transmissions. Another average power demand of 25 kW (corresponding to 50 kW only during the warmer half of the year) is added to drive the air conditioning compressors. Under the above operating conditions, this results in an additional energy demand of 125 kWh per train travel (corresponding to 1.25 kWh/min) for the auxiliary drives.
The same demand for the auxiliary drives is also assumed for the BECT. In the first instance, this may be surprising, since only for the DCT can be assumed that all the heat required for heating during the colder half of the year can be obtained from the cooling circuits of the internal combustion engines, while in the BECT even in case of use of 11 Schmetz, Struck liquid-cooled electric motors, inverters and battery storages including their management systems only the half of the heat is available, and therefore an additional electrically heating is necessary. But these assumptions are permissible with regard to the more efficient supply of energy to the on-board electrical system, the omission of the pumps for the lubrication of the gear transmissions and the same average power demand for driving the compressors of the air conditioning systems. Table 2 is the energy that could theoretically be recuperated when braking a (modified) DCT, for example by means of flywheel energy storages or, in case of the BECT, by restoring it to the batteries. It results from reducing the kinetic energy of 24.5 kWh supplied to the train models during each acceleration phase by the energy of 0.9 kWh required during each braking operation to overcome the driving resistance. Thus it is theoretically 23.6 kWh during each braking process or 283.2 kWh for a full train travel. But, since no recuperation is provided in the technical design of the DCT selected here, for the DCT it is not be examined further at this point.

Not listed in
With the energy demands for the single operating phases determined in this way, a total energy demand of 838.5 kWh, calculated as shown in

Electrotechnical Design of the Energy Storages
Determination of the capacity From Table 2 and Figure 5 can be calculated that the BECT requires for a travel on the considered railway line from the charging point in the starting station to reach the electrified part of the railway line model a total energy of 298.4 kWh. This amount of energy must be available in the batteries of the BECT at the begin of such a travel. In fact, the batteries must have a much higher nominal capacity as they lose capacity by aging, and should not be charged to their upper capacity limit or deeply discharged during normal operation. In addition, the energy demand of the battery management system must be taken into account, while here for simplification the rather low Schmetz, Struck discharge losses are added to the charging losses. These facts are considered by surcharges of 25% (corresponding to 74.6 kWh) for a 20% capacity loss during the service life of the batteries and a further 53.9% surcharge (equivalent to 201 kWh) for duty cycle operation with a limit of 90% capacity for charging and a limit of 25% capacity for discharging. It is assumed that, with a remaining capacity of 25% in normal operation, there is also a sufficient reserve for any emergencies. This means that the BECT must have a total nominal battery capacity of 574 kWh.

Configuration of the Energy Storages
Before continuing the electrotechnical design of the energy storages, the electric drive system of the BECT must be specified more precisely. For this purpose, the requirements for the largely identical designs of the DCT and the BECT and same performances have to be taken into account. In addition, the efficiencies listed in Figure   13 Schmetz, Struck 5 and some common, but here not further explained rules for the dimensioning of electric drives must be considered. Then, for example, eight 500 V three-phase asynchronous traction motors with 100 kW continuous power each are sufficient to drive the eight axles of a BECT. Furthermore, there are eight at least two-fold overloadable traction inverters required which are highly suitable for operation in the field weakening range, and which require a DC-link voltage of at least 710 V on the input side under full load and under these conditions. Overall, this results in eight single electric drives, which are connected in each single train unit on the input side expediently via a common DC-rail and a common, controlled rectifier with a high voltage part as shown in Figure 3, and regarding their control via a common data bus.
Any opportunities for simplification of this basic structure should not be dicussed at this point. Furthermore, it is assumed that the batteries are connected via their own DC/DCconverters to the common DC-rail, with which the charge and discharge of the batteries can be controlled in both the battery electric and the pure electric operation. Any potential measures to save these DC/DC-converters should not be examined here either. With regard to the number of batteries, and for reasons of a good mass distribution and low line losses, it is assumed that four batteries with a capacity of 143.5 kWh each are provided in the vicinity of the four pairs of axle drives of the BECT. These four batteries, for example, can be installed at the locations of the internal combustion engines and gear transmissions of today's DCTs.
Due to the relevant occupational saftey regulations like DGUV-I 8686, which apply in Germany, the four batteries should be designed as packs of fifteen stacks of 48 VDC each in series and with a sufficient number of stacks in parallel. Such battery stacks are usually constructed from single lithium-ion or lithium-iron phosphate battery cells. The characteristics and technical data of two suitable battery cells are shown in Figure 6.
Here lithium iron phosphate battery cells are chosen to maufacture the battery stacks because of their larger working temperature range, their suitability for higher currents, their nearly constant discharge voltage and their not shown, but commonly higher cycle stability, in spite of their lower capacity. Due to the increasingly large-scale production of high-performance energy storages for electric vehicles, there are now also other, fairly quickly available, sufficiently safe and cost-effective alternatives [9], which require adequate attention, too.
A favorable structure of the single battery stacks results, for example, from the arrangement of 240 battery cells of type YT26650FP according to figure 6 in 16S15P configuration (thus sixteen battery cells in series and fifteen in parallel each), which, for example, can be monitored with the aid of chips like Infineon-Aurix of type SAK-TC Schmetz, Struck 277 T. Then, a battery stack has an open circuit voltage of 51.2 VDC, a discharge current of 300 A at a voltage of 48 VDC and a capacity of 1.92 kWh. To obtain one single battery pack with a capacity of 143.5 kWh, fifteen of these battery stacks must be connected in series and each five in parallel, so that there are totals of 72,000 battery cells and 6000 management chips needed for the four required battery packs. In view of these figures, it becomes evident that in the real construction of the battery packs, additional effort is required to handle the heat released, above all during the charging and discharging processes, in the form of an efficient thermal management, which, however, can not be discussed here for reasons of scope. But a rough consideration of Further modifications (or measures which harm the energy efficiency) of the BECT are necessary if the energy recovery during braking is investigated. Because for a deceleration of a BECT with -1,13 m/s² (both the railway-technical and the electrotechnical admissibility of such a brake down in pure regenerative braking require separate checks) there is a braking power of almost 5900 kW required during the first second. After that, the braking power decreases by 200 kW with each additional second during each almost thirty-seconds long braking process. Due to the supportive driving resistance, the assumed reverse efficiency of the drive system of 90% and total charging losses of about 1 W per battery cell, but without taking into account the damping effect of the inductances of the electric motors and any filters, the real initial braking power can be assumed as 5000 kW, which decreases then by 170 kW per second. While the recovery of this braking power into the catenary on the electrified railway line section is not a particular problem, the reduced braking power during battery operation only falls after 26 seconds below the nominal net charging power of 540 kW, which is permitted 16 Schmetz, Struck for the four battery packs in normal operation (Figure 7). Even with the permission of quick-charging, which reduces the lifetime of the battery packs, the then permissible maximum net charging power of 1080 kW is not enough to absorb the recuperable braking energy. To solve this problem at least the measures listed in Figure 7 come into question.

Figure 7: Braking Performance and Measures to Absorb the Excess Energy
The wide range of possible measures according to Figure 7 shows that this is the actual problem of the design of the energy storages. Since the reason for the investigation is precisely the increase of energy efficiency (and the associated positive environmental effects), the first two measures do not have to be further investigated. Although the third measure is still feasible under the premises set here, it permanently uses almost all of the time reserves built into the operating model and thus lead to a particularly high susceptibility to delays (or to the need for procurement of another two additional BECTs). Likewise, only a slight acceptance of the longer travel times on behalf of the passengers as well as the resulting passenger migration can be expected. But even the last of the measures listed in Figure 7, the coverage of the high storage requirements by means of special short-term storages such as supercapacitors (hereinafter referred to as supercaps), implies considerable disadvantages. For storing the excess energy of 16 kWh per braking process in regular operation, for example, at least 232 pieces of suitable Eaton supercaps of type XLM-62R1137-R are required, which can be arranged in four groups parallel to the four battery packs, and, for the best use of their capacity, can be connected via separate DC/DC-converters to the common DC-rails of the single train units. As a result, apart from an increased space requirement, the mass of a BECT Schmetz, Struck increases by another 4 t and its price in the order of EUR 250,000.-. Therefore, all the measures listed in Figure 7 lead only to suboptimal solutions. Better solutions may be achieved if, in addition to minor adjustments to the operating and train models made so far, greater adjustments are allowed.
This becomes evident in the review of the highest power demands during the discharge cycles, which each occur at the end of the five single acceleration stages of the defined train model. These power demands can be calculated from the installed total power of the DCTs as 1120 kW, respectively 280 kW per battery pack, taking into account the efficiency assumptions shown in figure 2 and neglecting the needs of the auxiliary drives. The single battery packs, indeed, can provide a continuous discharge power of 1080 kW (as maximum even more than 2000 kW) when they are assembled from five parallel battery stacks. This means that the previously designed energy storages are also suitable for the operation of BECTs with a much higher installed drive power, which either let accomplish shorter driving times due to faster accelerations or a waiver of the installation of additional supercaps by shifting the acceleration and braking profiles without travel time extensions. In any case, it is to be expected that larger-sized electric drives will be required in order to slow down the BECTs in normal operation as far as possible in a regenerative manner.

Conclusions
In the previous sections, it was demonstrated for a real case, how the overall energy demand for a BECT can be calculated, its energy storages can be designed and its technical realization can be accomplished. It was also shown that in this manner great benefits in terms of energy demands can be achieved and exhaust emissions can be largely avoided. But it has also become clear that a large number of different alternatives exists for technical specifications as well as for decisions regarding energy efficiency and economic aspects. For implementation, therefore, not only further, considerable effort is required to determine the best possible configuration of train units, energy storages and electric drives, but fully developed BECTs will need investments for additional trains and charging infrastructures. An additional, periodic investment requirement will also be created by the not yet investigated, but limited service life of the energy storages. Because in case of the underlying operating model already 2500 full load cycles occur per year. Nevertheless, in order to achieve an economically necessary minimum service life of energy storages in the order of a half concession period, thus 7.5 years, a special emphasis has to be laid on the careful charging and Schmetz, Struck discharging of the energy storages, especially for part load cycles, in conjunction with a proper design and adequate battery management. Although a number of key figures and real components have already been identified in the previous sections and further missing information has been determined by internet research, an investment calculation was not carried out here because of the still insufficient database. According to an evaluation of the advertising of a well-known manufacturer at the Innotrans-fair in the German capital Berlin in September 2016 [10], this manufacturer expects, under the assumptions and figures given there, as well as by offsetting the also advertised, lower maintenance costs against any interest rates, and from a purely monetary point of view, at best an amortization within 28 years. In contrast, the tendency here is more likely to lead to a payback period in the order of ten years in the best case.