Abstract
Freight railroads are exploring alternative energy technologies such as battery electric locomotives (BELs). A BEL produces tractive effort from electricity stored in onboard batteries that are recharged at terminal stations, through regenerative braking, or both. To aid railroads in identifying the types of routes and train services benefiting most from BEL implementation, this paper investigates the sensitivity of BEL energy benefits to various parameters across multiple rail corridors. A simplified train energy model was developed to quantify BEL performance on four rail corridors in the United States. For each route, diesel energy consumption of conventional diesel-electric locomotives was compared with diesel consists supplemented with one BEL across a range of train weights, railcar resistances, battery efficiency, and
Introduction
To reduce fuel consumption and carbon emissions, heavy haul freight railroads are exploring alternative energy technologies such as the battery electric locomotive (BEL). A BEL is a self-contained locomotive that produces tractive effort using electric power from onboard batteries. The batteries onboard the BEL are recharged from energy captured through regenerative braking, or from external electric power at terminal facilities. Although the entire mass of the trailing train contributes to the energy that can be recaptured through regenerative braking, because current prototype BELs do not have the ability to pass electricity “across the coupler” from other locomotives, only the powered axles of the BEL can be used to generate electricity for battery charging. Current prototype BELs do not have a diesel engine prime mover, but BELs are capable of operating in a multiunit consist (i.e., a group of locomotives assigned to a particular train) with other conventional diesel-electric locomotives and/or BELs, or independently on their own. Although onboard battery storage for freight trains is not a new concept, limitations in battery technology have largely confined previous BEL models to yard switching service ( 1 ). However, there is potential for more widespread use of BEL technology in mainline freight rail operations given recent improvements to battery technology, decreasing battery costs, and increasing environmental regulation. As an emerging alternative energy technology in the early stages of experimental deployment, it is unlikely that mainline freight trains will be solely powered by multiple BELs that completely replace diesel-electric locomotives until BELs have demonstrated sufficient range and reliability in revenue service. In the near term, a likely mainline BEL deployment scenario will be to supplement a conventional diesel-electric locomotive consist with one BEL. The tractive effort provided by the BEL can then be used to reduce the traction demands on the diesel-electric locomotives and correspondingly reduce diesel fuel consumption. In such an application, important information about and practical experience of BEL operating characteristics and efficiency could be observed while the conventional diesel-electric locomotives are still available to power the train if the BEL battery is completely discharged. Although this hybrid-consist BEL deployment strategy has the potential to reduce diesel fuel consumption, at the time of writing, BEL technology has not been employed for routine revenue mainline Class 1 freight rail service in the United States. Thus, the exact energy and emissions benefits of BELs under various mainline freight railway operating conditions and charging strategies represent a research frontier.
Despite the lack of BEL technology deployed within the North American freight railroad industry, several studies have investigated the feasibility of BELs to replace conventional diesel locomotives for specific trains on certain case study corridors ( 2 – 7 ). In California, BNSF Railway experimented with a 2.4 megawatt-hour (MWh) BEL added to a diesel-electric locomotive consist and experienced 12% diesel fuel savings on average over a 3-month test period ( 8 ). The BNSF test route profile contained both flat and steep grade sections, and the experiment used terminal charging. Lengthy charge times of up to 11 h were experienced, at times causing the BEL to depart at less than a full charge. Union Pacific Railroad recently agreed to purchase 20 BELs for testing in various yards in California and Nebraska ( 9 ), and CN Railway has also agreed to purchase a BEL for testing. Internationally, heavy haul operations in Australia and Brazil have agreed to purchase BELs, including eight-axle BELs with an onboard battery storage capacity of 14.4 MWh. Although these prototype demonstrations of BELs are aimed at proving the reliability of the technology in the heavy haul mainline freight service environment, the observed fuel and emissions benefits may not be representative of more widespread deployment. Quantifying more general diesel savings from BELs is critical to justifying investments in this technology as the rail industry continually seeks to become more efficient and increasingly implements sustainable practices to reduce carbon emissions.
The physics of heavy haul freight trains with their low power-to-weight ratio dictates that the vertical grade profile of a given rail corridor greatly influences the overall energy demand and potential for a BEL to save fuel by recharging through regenerative braking. Grade profiles need to have sufficiently steep downhill sections before regenerative braking occurs. If a descending grade is not steep enough, then the potential energy gained from moving downhill is unable to overcome the inherent train rolling resistance, and additional energy must still be drawn from the diesel, BEL, or both, to propel the train forward. However, downhill grades that produce beneficial regenerative energy in one direction may be problematic for trains moving in the opposite direction, as the traction demand of sustained uphill grades may quickly deplete a battery. Once a BEL is depleted, it cannot contribute tractive effort and imposes an additional train weight and resistance penalty that increases traction demand from the conventional diesel-electric locomotives in the consist. Sustained downhill grades may quickly charge a BEL battery to full capacity, with additional regenerative braking energy wasted as heat in a manner similar to dynamic braking with a conventional diesel-electric locomotive. Thus, there is hypothesized to be an interplay between train weight, rolling resistance, grade profile, and battery capacity that influences the potential energy and emissions benefits of supplementing a conventional diesel-electric locomotive consist with a BEL for a given train trip.
In addition to these interactions between train, route, and locomotive characteristics, the possible benefits of BELs are further complicated by the potential implementation of different terminal charging scenarios. If a BEL must be charged to full capacity before each trip, a BEL with a large battery storage capacity must spend considerable time recharging at each terminal, particularly if the terminal charger has a low charge rate. This additional terminal dwell time will decrease locomotive utilization, resulting in fewer mainline trips per year, and potentially less cumulative annual diesel fuel savings and emissions reductions than a BEL with a smaller battery or terminal with a higher charge rate. It is hypothesized that there will be a trade-off between a BEL with a larger battery capacity that experiences higher savings per trip but makes fewer trips per year, and a BEL with a smaller battery capacity that experiences lower savings per trip but makes more trips per year because it can fully charge faster. Further, a railroad may elect to implement a BEL in conjunction with conventional diesel-electric locomotives but without any terminal charging. By eliminating charging time, this approach will maximize the number of locomotive trips per year, but energy and emissions benefits are limited by the regenerative braking energy that can be captured on a given route.
Objective and Research Questions
To investigate these hypotheses and aid railroads in identifying the types of routes and train services benefiting most from BEL implementation, and thus maximize the potential diesel fuel savings and carbon emissions reductions of this technology, this paper investigates the sensitivity of BEL energy benefits to various train and locomotive parameters across multiple rail corridors and charging strategies. An energy model was developed to predict energy savings from the implementation of one BEL in a diesel-electric consist. Several routes with varying grade profiles representative of conditions found throughout the U.S. rail network were selected for detailed energy analysis. Furthermore, a sensitivity analysis was conducted on energy model input parameters to determine how altering the inputs affected energy savings per trip, and on an annual basis considering the impact of the charging strategy on locomotive utilization.
This study aimed to answer the following research questions:
How much does route topography influence BEL energy benefits?
How sensitive are mainline freight BEL energy benefits to train and locomotive characteristics?
Are BEL energy saving results from a simplified mathematical approach comparable with those from detailed train performance simulation?
What is the trade-off between the BEL storage capacity and charging strategy in relation to annual energy savings?
From an economic perspective, what is the optimal combination of battery size and charging strategy for each of the study corridors?
The remaining sections of this paper discuss previous research on BELs, summarize an exploratory BEL energy analysis, describe the methodology used to calculate BEL energy savings and perform the economic analysis, outline the experimental design, present the results of the analysis, and review the conclusions and limitations of this study along with possible future research directions.
Previous Research
Railroad researchers have long sought to save diesel fuel through various means of capturing, storing, and reusing the electrical energy generated by dynamic braking. In 1979, the Federal Railroad Administration (FRA) studied the performance of a diesel-electric switching locomotive enhanced with flywheel energy storage in a boxcar coupled to the locomotive ( 10 ). The concept was technically feasible but not economical given the small amounts of energy captured and returned by the flywheel in yard switching service. In the early 2000s, Painter ( 11 ) and Painter and Barkan ( 12 ) examined the potential fuel savings associated with recovering dynamic braking energy on mainline freight trains descending Cajon Pass from Barstow to San Bernardino, CA. Given the traffic levels at the time, an idealized system with 100% recovery efficiency could save over 1 million gallons of diesel fuel per year. Although the specifics were not examined, the authors acknowledged that the actual efficiency and storage capacity limitations of onboard storage with batteries, supercapacitors, or flywheels would substantially reduce potential fuel savings. Iden further notes that the ability of onboard energy storage systems to use regenerative energy is constrained by train marshaling restrictions that limit the number of dynamic braking axles at the front of the train ( 13 ). These restrictions effectively require some braking energy to be provided by the train air brakes and thus prevent a portion of the potential regenerative energy from being recovered.
The development of modern BELs began in the early 2000s with diesel/battery hybrid switcher locomotives that primarily used battery power but could use a small diesel engine to recharge if necessary ( 14 ). In 2007, Norfolk Southern developed a true BEL for switching service, but limitations of the initial battery technology required a new battery management system in 2009, and an upgrade to newer battery technology in 2014 ( 14 ). The California Air Resources Board examined the potential of the BEL concept in mainline freight service with the aid of battery tender cars, each with a storage capacity of 6.2 MWh ( 14 ). A subsequent study of strategies to eliminate mobile-source locomotive emissions on mainline routes in California indicated that using only batteries to power a typical mainline freight train between terminals required an impractical number of battery tenders from the perspective of train weight and length ( 15 ). A 2017 FRA study examined the application of battery technology for rail propulsion, with a focus on the ability of various battery chemistries to absorb regenerative energy and provide traction energy within mainline diesel-electric battery hybrid locomotives and BELs for switching applications ( 16 ). The report estimates that an onboard battery can provide 23% of mainline freight train traction energy across a particular study route, but also predicts a battery lifespan of 7 to 10 years under the corresponding duty cycle. In 2019, Thorne et al. surveyed the state of BEL technology, noting that although there was a lack of mainline freight rail applications, onboard battery power had gained a market in the passenger rail and transit sector ( 17 ). This finding is consistent with contemporary research that examines battery storage for light rail vehicles ( 18 ), quick charging of overhead-wire-catenary-free electric trains with onboard energy storage in Japan ( 19 ), and the development of algorithms to optimize passenger train speed profiles when operating on battery power through track segments without an overhead catenary traction power supply ( 20 ). The latter study notes that batteries can only power the passenger train for up to 20 min, and that the batteries must be charged from the overhead catenary since the energy captured from regeneration is insufficient to recharge. In contrast to the 0.5-MWh battery considered for the referenced passenger application, this study considered freight locomotive platforms with sufficient axle load and available carbody space to support significantly larger batteries (2.4 to 14.5 MWh) that can sustain BEL traction power for much longer periods of time with increased operation range.
Since the BNSF and Wabtec BEL pilot study was conducted in California, the energy, emissions, and economic benefits of mainline application of BEL technology have drawn more research attention. Popovich et al. describe the economic feasibility of implementing BELs for mainline freight service when environmental factors are considered ( 21 ). The study assumes widespread use of fast-charging infrastructure, and a fixed range for a BEL with a 14-MWh battery tender based on national averages of freight train energy consumption. The authors conclude that a switch from diesel-electric to BELs would save an estimated $94 billion over 20 years. A follow-on study by Moraski et al. examines the potential use of the U.S. rail system as a nationwide backup transmission grid over which containerized batteries, or rail-based mobile energy storage (RMES), are transferred to meet demand peaks and increase resilience ( 22 ). In estimating the transportation costs of RMES, the study makes similar assumptions about average train energy consumption. Whereas national averages are applicable to these high-level assessments of BEL economics, multiple studies have shown that freight rail energy efficiency varies greatly between routes and is also influenced by various train and locomotive characteristics ( 23 – 26 ). The work presented in this paper improves on this previous work by examining BELs on specific route topography and quantifying the sensitivity of BEL energy benefits to various train-, operating-, and charging scenarios.
In the first of two recent works most closely related to this study, Zenith et al. analyzed two nonelectrified railway lines, one in Norway and the other in the United States, for their potential to be electrified with overhead catenary, batteries, hydrogen, or hydrogen-battery hybrid powertrains ( 27 ). Single freight trains were simulated, and the long-term economics of the different technologies were compared. The results suggest the potential of batteries and fuel cells to replace diesel on rail lines with low traffic volumes. In the second paper, Cipek et al. considered a hybrid consist of conventional diesel-electric locomotives and BELs and developed a model to optimally control the hybrid BEL consist ( 28 ). The authors then used simulation to demonstrate fuel savings between 22% and 30% for a freight train operated with a hybrid BEL consist over a mountainous railway route in the Lika region of Croatia. Whereas these two papers focus on detailed economic analyses of BEL technology under a limited number of specific implementation scenarios, the work presented in this paper makes a novel contribution by taking a more general approach to conduct a structured experiment that examines the sensitivity of BEL energy savings to various route topographies and to a variety of implementation scenarios.
Exploratory Analysis
To illustrate the potential variability in BEL mainline freight performance, an exploratory analysis was conducted with the web-based Synthesis of COnsists as Rolling Energy (SCORE) micro-grids train energy toolset ( 29 ). The SCORE tool was developed by Pennsylvania State University with funding from the Department of Energy Advanced Research Projects Agency–Energy. The online SCORE interface is prepopulated with route topography for 18 different mainline corridors radiating from Chicago. The routes range in length from approximately 200 to 1,000 mi. SCORE allows the user to build a train consist from a set of available locomotive and freight car models, and then to conduct a detailed train performance simulation and energy consumption calculation for that train operating from Chicago to the remote terminal on each of the 18 routes. SCORE outputs the time and energy consumed for each train run, allowing for comparisons between routes, train consists, and available locomotive technologies.
Approach
To quantify the potential freight transportation productivity of BEL technology on each route, a train consist of two BELs with 7 MWh of battery storage each (14 MWh total) and unpowered trailing freight railcars was iteratively simulated on a given route. In this case, to highlight the limitations of current BEL storage capacity, no diesel-electric locomotives were included in the consist. With each iteration, the number of freight cars in the train consist was increased by one until the SCORE model returned an error, indicating that the BEL battery state of charge (SOC) had reached zero at some point along the route before the train reached its destination. With a depleted BEL battery charge, the train stopped mid-route because the BEL could no longer generate tractive effort, and that size train consist was considered infeasible. The number of railcars in the largest feasible train on each route was recorded. This number of railcars was then multiplied by the gross railcar load and length of the route to yield the ton-miles of freight transportation productivity provided by the two BELs. Since SCORE assumes that both BELs are fully charged before each trip, and the largest feasible train nearly depletes the battery at some point on the route, the total ton-miles were divided by 14 MWh to yield an estimated normalized BEL efficiency metric for each route in ton-miles per MWh of initial terminal electrical charge. Similar to the miles per gallon used as a primary energy efficiency metric for light-duty highway vehicles, heavy haul freight railroads often express their energy efficiency in ton-miles per gallon of diesel fuel when using diesel-electric locomotives. Ton-miles, the product of total weight of freight and the distance transported, provides a measure of transportation productivity, whereas the amount of energy consumed in diesel gallons (or electricity MWh) represents the energy input required to achieve that transportation productivity. Dividing the transportation productivity value by its input energy provides a measure of efficiency in the form of a normalized transportation productivity per unit of input energy. Thus, for a BEL, it is natural to use ton-miles per MWh as an analogous efficiency evaluation metric to ton-miles per gallon.
A potential challenge in interpreting the results of the SCORE experiment is that the feasible train size (number of railcars) is likely to be route-specific based on its length and topography. As the number of trailing railcars increases, with a fixed BEL consist, the balancing speed, acceleration, and braking of the train will decrease, and the running time of the train will increase. Longer trains that travel at these slower speeds will experience less resistance per railcar, and thus may exhibit higher BEL efficiency than a similar route that is operated with a shorter train that travels faster and encounters correspondingly more resistance per railcar. Unfortunately, the SCORE model does not have a provision to fix the running time of the train at a target value but instead attempts to run the train as close to the maximum authorized speed as possible over the length of the route. Thus, the results for ton-miles per MWh may be influenced by train size and speed, in addition to the specific route topography. However, these combined effects provide additional insights into the potential variation in expected BEL efficiency.
Results
Examining the results of this process across all 18 corridors (Table 1), the values of BEL ton-miles per MWh showed considerable variation from 53,000 to 92,000 ton-mi/MWh. In general, routes covering longer distances and featuring larger changes in elevation tend to yield fewer ton-miles per MWh. When climbing uphill, the BEL battery charge must be used to overcome both the inherent train resistance per unit distance traveled, and to provide the work required to increase the potential energy of the train as it rises in elevation over this distance. The larger the elevation change, the more battery energy required to transport each ton over a distance of 1 mi, and therefore fewer ton-miles can be accumulated for a fixed amount of charge (MWh). It was hypothesized that longer routes are more likely to have extended upgrades or long, level sections of track that are expected to deplete the BEL battery before it can be recharged via regenerative braking on an extended downgrade. As a means to test this hypothesis, a normalized index considering the relative change in elevation, highest elevation, and length of each route was calculated for each route based on the elevation profile displayed in the SCORE tool. Based on the concept of evaluating the efficiency of a railroad route elevation profile by total rise and fall in elevation, the metric of excess feet of rise and fall per mile of route (H) is calculated with Equation 1, where
Routes with low values of excess feet of rise and fall per mile are either relatively level or feature a sustained upgrade (or downgrade) from origin to destination with limited potential for a BEL to capture (or reuse) regenerative braking energy. Routes with higher values of excess feet of rise and fall per mile will feature both upgrades and downgrades that potentially offer BELs the opportunity to capture and reuse regenerative braking energy, increasing their ton-mile productivity per MWh of initial charge. When the ton-miles per MWh and excess feet of rise and fall per mile were plotted for each route (Figure 1), the resulting trend indicated that BELs exhibited better performance on routes with more rise and fall. However, there was still considerable variation that was not explained by this factor, suggesting that BEL performance was also sensitive to the other factors investigated later in this paper.
BEL efficiency in ton-miles per MWh of charge as a function of route topography.
Given the different route lengths, the BEL efficiency metrics in Table 1 corresponded to train sizes from a low of 3 to a high of 41 loaded railcars for two BELs. Even if the number of BELs was doubled to four, on most simulated routes, the resulting train lengths would still be far below typical mainline freight trains of over 100 railcars ( 30 ). These train sizes suggest that BELs have a limited capability to independently transport economical mainline freight train sizes over long distances. Instead, BELs are likely to be implemented by forming hybrid consists of both BELs and conventional diesel-electric locomotives.
Exploratory Route Characteristics and Corresponding BEL Efficiency
Note: BEL = battery electric locomotive; H is excess feet of rise and fall per mile as defined in Equation 1.
All routes begin in Chicago and start at an elevation of 600 ft.
Although the SCORE tool proved useful for this exploratory analysis, at the time this research was conducted, it had several limitations that prevented it from being used for the main experimental design presented in this paper. Most significantly, the web version of SCORE had no provision to upload different routes for analysis, and the tool only analyzed train runs originating in Chicago. The tool lacked the capability to examine a round trip that covered the same route in both directions. Finally, the SCORE tool was coded to assume that each BEL departed its initial terminal with a full charge; there was no provision for examining implementations without terminal charging at one or both ends of a given route. Because of these limitations, SCORE was not used for further analysis. Instead, the authors developed their own simplified train energy model so that the main experimental design presented in this paper could include different terminal charging strategies and a set of routes selected to provide greater contrast in topography than the set of SCORE routes that all originated in Chicago.
Methodology
The methodology section first describes the simplified train energy model formulated to investigate the study research questions. This is followed by a brief description of the detailed train performance simulation that is compared against the simplified train energy approach. The methodology concludes with an outline of the approaches used for the sensitivity analysis, annual energy saving analysis, and economic analysis.
Simplified Train Energy Model
To evaluate the potential diesel fuel savings of BELs across multiple corridors and various operational parameters, a simplified train energy model was developed based on first principles of locomotive tractive effort, train resistance, work, and energy. The main function of the energy model is to track estimates of both the potential energy from changing track elevation and the work needed to overcome train rolling and aerodynamic resistance. In converting train resistance and changes in potential energy into demands for diesel and battery charge for a train of a given mass, the model considers the efficiency of the BEL in supplying electrical power for traction and storing energy recovered through regenerative braking. The efficiency factor dictates that proportionately more diesel or electrical energy is required to overcome the energy demands of a given track segment, and proportionately less electrical energy is captured by the battery compared with the energy removed from the train during regenerative braking. In this manner, as the BEL traverses a route, the model uses traction and braking demands to track the estimated battery SOC.
To perform an energy analysis, the model requires as input the vertical gradient or elevation profile for a mainline rail corridor and the values for several locomotive and train parameters that compose the experimental design. Based on the track elevation profile, the rail corridor is divided into segments of constant (or relatively constant) vertical gradient between points of vertical inflection. For a given segment, n, the energy required to overcome inherent train resistance over the length of the segment and the change in elevation between the start and end of the segment is calculated using Equation 2,
where
E n is energy (at the wheels) required to traverse segment n (MWh);
W is total weight of the train under study, including locomotives (tons);
R is inherent rolling and aerodynamic resistance of the train (lb/ton);
X n is distance from the start of the route to the start of segment n (mi);
Z n is vertical track elevation at the start of segment n (ft); and
a and b are coefficients for unit conversions.
When En≥ 0, energy is consumed since the locomotives must supply traction force to overcome the combined train and grade resistance. When E n < 0, no energy is consumed as the locomotives must supply braking force, and the BEL (if present) will have an opportunity to recapture braking energy. To determine the baseline energy consumption of the train when traversing the entire corridor with conventional diesel-electric locomotives, all positive values of E n are summed across all n segments whereas negative values are ignored.
This analysis assumes a single BEL is added to the conventional diesel-electric locomotive consist. When E n < 0, the BEL can use regenerative braking to increase the SOC, S, of its onboard battery, as calculated using Equation 3 and appropriate unit conversions,
where
S n is BEL battery SOC at the start of segment n,
U is efficiency of the BEL battery and powertrain,
B is total battery capacity of the BEL.
In Equation 3, the “minimum” function is used to ensure that the energy recaptured in regenerative braking does not exceed the total dynamic braking energy required over the segment, exceed the maximum possible single BEL regenerative braking force and work limitation, or cause the battery SOC to exceed its maximum storage capacity. The constant maximum regenerative braking force of 135,520 lb provided by the BEL in the simplified model is based on published dynamic braking curves for an equivalent 4,400-hp six-axle diesel-electric locomotive ( 31 ).
When E n ≥ 0, the BEL provides traction force and the SOC, S, of its onboard battery decreases, as calculated using Equation 4 and appropriate unit conversions,
where T is maximum tractive effort force provided by the BEL and Smin is minimum BEL battery SOC (assumed to be zero in this study).
In Equation 4, the “maximum” function is used to ensure that the traction energy supplied by the traction battery does not exceed the total traction energy required over the segment, exceed the maximum possible single BEL traction force and work limitation, or cause the battery SOC to drop below its minimum acceptable level (or zero as assumed in this study).
To determine the total diesel energy saved when traversing the entire corridor with a combined BEL and conventional diesel-electric locomotive consist, the change (decrease) in battery SOC when the BEL is providing traction energy is calculated for segments where
where D n is diesel energy savings for segment n.
The total diesel energy savings for the entire route is calculated by summing
Several assumptions are made by this simplified train energy model. Each train is modeled using a point-mass assumption relative to the vertical grade profile. With this simplifying assumption, the conditions on each segment are assumed to instantaneously and uniformly act over the length of the train regardless of the length of the segment relative to the actual length of the train. Consist-level energy management is driven with a greedy approach; when the BEL SOC exceeds zero and the train has a demand for tractive effort, the battery is depleted without preparation or optimization of the stored energy. All braking is dynamic without the use of train air brakes. The energy required for idle locomotives is not included in the model and curve resistance is ignored. Since maximum authorized speed data were not available for all of the study routes, the model assumes a constant average train speed of 30 mph when in traction, and therefore the corresponding tractive effort is equal to 45,173 lb-force per locomotive (both diesel and BEL). Dynamic braking speeds were set to 20 mph, yielding a maximum 135,520 lbf braking per locomotive (both diesel and BEL). These simplifications were made to facilitate a subdivision-level screening analysis of potential BEL benefits across large portions of the U.S. freight rail network with a minimum of input data, with four of these corridors used in the experimental design presented in this paper. In practice, it is envisioned that this screening analysis would be used to quickly identify corridors where BEL implementation appears promising, with these corridors then being subject to a more detailed train performance simulation exercise to quantity more exact BEL benefits. One objective of this study was to determine the relative difference between the BEL energy savings predicted by this simplified train energy model and a more detailed train performance simulation.
Detailed Train Performance Simulation
To evaluate the effectiveness of the simplified train energy model, its results are compared against those obtained from the detailed train performance simulation conducted with the Advanced Locomotive Technology and Rail Infrastructure Optimization System (ALTRIOS). A fully integrated, open-source software, ALTRIOS simulates and optimizes the rollout of locomotive technologies for energy efficiency and decarbonization ( 32 ). ALTRIOS simulates real-world, multidecade rail network operation in a single unified model, including the energy-affecting aspects of rail operation such as freight demand-driven train scheduling constrained by technology-specific operating requirements, algorithmically generated target speed traces corresponding to sequences of train departures, and meet and pass events to ensure conflict-free operation. ALTRIOS is highly applicable to this study because it was validated against diesel-electric and BEL train performance data obtained from locomotive event recorders during the BNSF and Wabtec tests in California ( 32 ).
The single-train simulation modules were the main ALTRIOS components used in this study to provide accurate estimations of energy consumption. ALTRIOS simulates train performance and dynamics, including the effects of grade, aerodynamic drag, inertia, and rolling resistance, in modeling the acceleration and deceleration process of the train to reach the maximum authorized train speed. Locomotive powertrain technology modules that simulate diesel-electric locomotives and BELs are used by ALTRIOS to transform traction and braking force demands into realistic energy consumption from diesel fuel and electricity. For train consists with BELs and diesel-electric locomotives, the consist power distribution controls are set to minimize powertrain-specific energy consumption. Compared with the point-mass assumption made by the simplified train energy model, ALTRIOS uses a mass-strap train weight model assumption to enable better estimates of grade resistance and the corresponding power requirements. Thus, direct comparison of the energy results from the simplified train energy model and the detailed ALTRIOS simulation can illustrate the relative impact of including acceleration, braking, and a mass-strap model in the BEL savings analysis.
The ALTRIOS single-train simulation requires the number of diesel-electric locomotives and BELs plus several locomotive parameters as input. The BEL battery capacity is subject to change, providing the flexibility of differential energy storage for different optimization scenarios. For the BELs, the ALTRIOS simulation of a round trip was set to either inherit the arrival SOC or to fully charge at terminals, with the arrival SOC recorded at each end of the round trip for the purpose of estimating necessary battery charging. Independent of battery capacity, the effective horsepower, tractive effort, dynamic braking, and adhesion limit of the BEL is taken to be the same as a common 4,400-hp six-axle diesel-electric locomotive.
The properties of the freight train railcars can be adjusted to reflect 263,000-lb, 286,000-lb, and 315,000-lb gross rail weight railcars with customized empty-loaded railcar weights to explore another dimension of sensitivity. However, this study assumes that all railcars are loaded or empty 286,000-lb gross rail load railcars.
The ALTRIOS network data and location files that specify the vertical alignment were created from the same railroad track chart data used by the simplified train energy model approach.
ALTRIOS directly reports round-trip diesel fuel and BEL energy consumption in its standard single-train simulation output. These values were used to determine diesel fuel savings whereas the battery SOC on arrival at terminals was used to estimate the battery charging requirements.
Experimental Design and Procedure for Sensitivity Analysis of Individual Round Trips
To evaluate the potential diesel fuel savings of BELs across a range of mainline heavy haul freight rail applications, the experimental design included the following seven factors:
Route topography,
Locomotive consist,
BEL charging strategy,
Train mass,
Train rolling resistance,
Battery efficiency, and
Battery storage capacity.
Four routes derived from actual North American rail corridors were selected for study to represent the different types of grade profiles a BEL might encounter (Figure 2). Routes A and B both represented mountainous elevation profiles where significant energy savings might be expected owing to the substantial elevation changes of over 700 m and the associated potential for battery charging via regenerative braking energy. Route C was flatter and followed a river with a nearly level (0%) grade. Route D was selected to determine whether hilly terrain with minor undulations but an overall relatively flat grade profile could still recapture sufficient regenerative braking energy to obtain BEL benefits.
Elevation profiles of four mainline study routes: (a) Grade Profile A, (b) Grade Profile B, (c) Grade Profile C, and (d) Grade Profile D (elevation scale on vertical axis varies).
For all four routes, elevation data were interpolated from industry track charts containing grades and milepost markings. For routes without track charts, FRA grade crossing data were overlaid with a public digital terrain model to obtain the elevations at grade crossing locations. These fixed points of elevation with known mileposts were then used to interpolate the entire grade profile.
Two options were considered for the locomotive consist factor: a conventional diesel-electric locomotive consist, and a conventional diesel-electric locomotive consist supplemented with one BEL. Note that the BEL is added to the conventional consist and does not replace a diesel-electric locomotive, to ensure that the train will always have sufficient tractive effort even if the BEL reaches a SOC of zero.
For BEL consists, an additional experimental factor is the terminal charging strategy. Rail operators may elect to avoid the expense and locomotive utilization constraints of BEL terminal charging infrastructure. To investigate the implications of this approach, two scenarios were modeled:
“Without terminal charging” has no battery charging at the end terminals and relies entirely on regenerative braking to charge the BEL while it is operating in a train over the route, and
“Terminal charging” assumes that the BEL is fully charged at terminals at both ends of a route.
The final four experimental factors (Table 2) were designed to capture a variety of train properties and a range of possible BEL characteristics to form the basis of the sensitivity analysis:
Three different levels of train weight were considered to represent 3,500-ton empty bulk unit, 10,000-ton mixed manifest, and 15,000-ton loaded bulk unit freight trains.
Three levels of combined train rolling and aerodynamic resistance were modeled to represent different types of railcars and operating conditions.
Three levels of battery efficiency were included to capture the influence of losses associated with using stored battery energy and capturing regenerative braking energy. A wide range of battery capacity was studied to reflect the current large degree of uncertainty in the efficiency of BEL-scale batteries across a range of climate conditions and over the lifespan of the BEL battery.
Three levels of BEL battery capacity were considered to represent the prototype 2.4 MWh BEL, current 4.8 MWh production models, and the planned BEL models with a larger 9.6 MWh storage capacity.
Sensitivity Analysis Experimental Design Factors and Levels
A train corresponding to the baseline set of input parameters (10,000 tons, 80% battery efficiency, 4.8 MWh battery capacity, and 5lb/ton average train resistance) was analyzed on each of the routes, A to D. For each route, energy needs were calculated for both the conventional diesel-electric locomotive consist, the consist with the BEL and no terminal charging, and the consist with the BEL and terminal charging. For equivalent conditions, the diesel energy demands of a train with and without a BEL can be compared to determine the equivalent diesel fuel energy savings in MWh for a round trip over the corridor.
To understand how energy savings respond to changing operational conditions, a sensitivity analysis was conducted by varying a single factor in Table 2 from its baseline value to its low or high value, and analyzing this new scenario with all other factors at their baseline values. The resulting change in the BEL energy savings response to changes in train and operating factors was quantified using arc elasticity, as seen in Allen and Lerner’s study ( 33 ). Fullerton et al. similarly used this approach to quantify the sensitivity of freight train fuel efficiency to different factors ( 26 ). Referencing the baseline values, arc elasticity quantifies how BEL energy savings react to normalized unit changes of each experimental factor. A larger arc-elasticity value indicates that changes to that factor have a greater impact on BEL diesel energy savings compared with factors with lower arc-elasticity values. For a given experimental factor, arc elasticity is calculated by dividing the percentage change in diesel savings resulting from a change in the study parameter from its baseline value to its high (or low) value relative to its baseline value, with the percentage change in the study parameter from its baseline to high (or low) value relative to its baseline value, as shown in Equation 6 where subscript “0” indicates the baseline value and subscript “1” indicates the high (or low) value.
The full experimental design and sensitivity analysis was first conducted with the simplified train energy model. For the purpose of comparison, the experiment and sensitivity analysis was then partially repeated with ALTRIOS. Since ALTRIOS performs more complicated train resistance and powertrain efficiency calculations than the constant values used by the simplified train energy model, the sensitivity of these parameters was not investigated in the ALTRIOS portion of the study.
Experimental Design and Procedure for Comparison of Annual Diesel Fuel Energy Savings
To investigate the trade-off between BEL storage capacity and charging strategy in annual energy savings, a secondary experimental design was developed to consider combinations of the two factors:
BEL battery capacity, with factor levels of 2.4, 4.8, 9.6, and 14.4 MWh to investigate the range of prototype and proposed BEL models, and
Charging strategy, with factor levels of no terminal charging, and terminal charging with either a 0.5 MW, 1.0 MW, or 2.0 MW charger to reflect the range of actual and proposed BEL charger configurations.
All factorial combinations of these two factors were analyzed with ALTRIOS while the other model parameters were held at their baseline values of the 10,000-ton train weight.
To determine the annual diesel fuel energy savings, the annual number of BEL trips was estimated based on the time required for the train to make a mainline round-trip over the study corridor, minimum connection times at the end terminals, and additional battery charging time at the end terminals (for scenarios with terminal charging). Charging time (in hours) was calculated by dividing the amount of electrical energy (MWh) needed to fully charge the battery at the terminals (based on the arriving battery SOC and the battery capacity) by the terminal charger power rating (MW). This linear charge curve is appropriate for high-level analysis and allows the annual energy savings to be calculated using Equation 7,
where
Economic Analysis
To help determine the optimal combination of battery sizes and charging strategy (and terminal charging facility investments) on each study corridor, a high-level preliminary economic evaluation is conducted to investigate the sensitivity of the optimal combination to route parameters and thus aid stakeholders in making capital investment decisions. The experimental design follows that of the annual diesel fuel energy saving analysis.
The high-level economic evaluation metrics are the net present value (NPV) of diesel savings, electricity charging cost, charger investment, BEL conversion investment, and also the benefit–cost ratio of these same factors. The high-level NPV calculation is shown in Equation 8.
where
Monetary Value of Energy Benefit and Cost
The two ALTRIOS simulation outputs that need to be transformed into monetary values are the amount of diesel fuel savings and electricity consumed by battery charging in terminals, as previously calculated from the annual energy saving studies. Their corresponding values are monetized by rates of dollars per MWh. The railroad diesel price varies over time ( 34 ), but a general diesel fuel price is assumed to be $3/gal. According to U.S. DOT, Bureau of Transportation Statistics, the energy density of diesel fuel is 138,700 British thermal unit (Btu)/gal ( 35 ), equivalent to 0.0406 MWh/gal. Thus, the diesel price per MWh was calculated as $74 per MWh. The electricity price was estimated to be approximately $12 per MWh, per the Energy Information Administration’s 2022 data ( 36 ). The flat rate of energy cost was applied to diesel savings and electricity charged at terminals. Acknowledging uncertainties in future energy retail prices, the monetary values of these energy sources were subject to change.
Number of BELs Supported per Charger Pair
Although allocating the entire capital cost of a pair of terminal chargers to a single BEL is appropriate for a prototype demonstration of a single unit, it is not representative of a more widespread adoption with multiple BELs operating on a given route and using the same pair of chargers. Under these conditions, the capital cost of each pair of terminal chargers should be allocated over the fleet of BELs that they can support. To determine the appropriate share of the terminal charging cost to allocate to a single BEL within the fleet, the total number of BELs that can be supported by a pair of chargers must be determined.
The maximum number of BELs that can be supported by a pair of chargers is influenced by the charger utilization rate. Charging and refueling operations are not expected to be seamless during routine railroad operations, as some buffer time between charging sessions is needed to account for repositioning equipment and the variability of arrival times encountered in real-world yard operations ( 37 – 40 ). The research team assumed an 85% utilization rate of chargers in terminals as a basis for this allocation.
Conceptually, the maximum number of BELs that can be supported by a pair of chargers is the maximum number of BELs that can be consecutively charged at one terminal (considering charger utilization) before the first BEL to be charged completes a round trip and returns for recharging. Thus, the maximum number of concurrent BELs on the specific corridor with one sole pair of terminal charging ports (i.e., one charger at each end terminal) is calculated by dividing the total round-trip cycle time by the longest of the two end charging times, shown in Equation 9,
where NBEL is maximum number of concurrent consists on a corridor, and
Infrastructure Cost
The infrastructure cost involves the charger capital cost and the capital cost required to convert existing diesel-electric locomotives into BELs, as shown in Equation 10.
where
is allocated charger installment cost per BEL.
Since multiple Class 1 railroads are already engaged in programs to remanufacture older AC (alternating current)-traction locomotives, instead of acquiring newly manufactured BELs, a cost-efficient approach to creating a BEL fleet may be to instead rebuild these older AC-traction locomotives as BELs. Since it is assumed that the railroad would incur similar costs for remanufacturing labor and miscellaneous material when converting to a BEL as opposed to an upgraded diesel-electric locomotive, these costs as well as the salvage values of the retired diesel-electric prime movers and alternators have been omitted for simplification purposes. The battery cost is estimated to be $100,000 per incremental MWh of onboard storage capacity ( 41 ) and is thus calculated as a linear relationship to battery capacity, as shown in Equation 11.
where
Previous research by the National Renewable Energy Laboratory has estimated the cost of a terminal charger with a fixed 1 MW (i.e., 1 megawatt-hour per hour [MWh/h]) charging speed as $1 million per charging port. Acknowledging the incremental cost of an improved charging speed, a linear relationship between charging speed and charger cost can be approximated based on similar observations from battery electric truck charging stations ( 42 ). From the cost data listed by Bernard et al., the incremental cost for charger speed was approximated as $950,000 per MW, and thus the cost function for a BEL charging station is estimated by Equation 12,
where
The allocated charger cost per BEL is calculated by dividing the cost of a pair of charging ports with the maximum number of concurrent consists (calculated by Equation 9), as shown in Equation 13.
Other Economic Factors
Owing to rapid battery technological development and the anticipated short battery lifespan under the demand of mainline freight duty cycles, the system lifespan is estimated to be 7 years ( 16 ). Since battery and charger technology is likely to evolve over this 7-year lifespan, the analysis assumes that the charger also has a service life of 7 years with no salvage value. Since infrastructure costs are calculated up front whereas diesel energy savings and electricity costs are distributed across the lifespan of the system, an NPV calculation is needed to evaluate the investment and compare this across the different BEL deployment approaches. This study implemented the 7-year NPV calculation with a 5% discount rate consistent with FRA benefit–cost guidelines. A cumulative energy benefit value throughout lifespan calculation is shown in Equation 14,
where
To simplify the comparison cases, charger costs are calculated based on the maximum number of concurrent BELs on the specific corridor for the corresponding operation combinations. A smaller fleet of BELs will incur a larger allocated charger capital cost per BEL, and thus have less favorable economics. A larger fleet of BELs will require the installation (and associated cost) of multiple chargers at each terminal.
Results
The results section first presents the sensitivity analysis for individual train round trips based on the simplified train energy model calculation, followed by those for the ALTRIOS simulation. Then the results of the annual diesel saving analysis and economic analysis based on the ALTRIOS results are summarized.
Sensitivity Analysis for Individual Round Trips
The results of the baseline train modeled on Route B illustrate a typical SOC profile for mountainous terrain (Figure 3, top). When SOC is zero, the BEL stops producing tractive effort and the train must rely entirely on diesel fuel for its energy. When the SOC plateaus at its maximum, in this case the battery storage capacity of 4.8 MWh, the battery is full and additional regenerative braking energy is wasted as heat. To maximize BEL benefits, it is best to exhibit a continuously changing SOC without extended distances at a SOC of zero or at a SOC equal to the maximum storage level. A BEL SOC that reaches a sustained plateau suggests that a larger capacity battery may be employed to better capture energy from regenerative braking. To demonstrate this concept, the results of the same baseline train on Route B but fitted with a 9.6-MWh capacity BEL (Figure 3, bottom) indicated that the SOC just briefly hit capacity and never hit zero. The 9.6-MWh BEL was therefore ideal for this route; an even larger capacity BEL would be of no additional energy benefit but could provide a buffer of additional electrical charge to add resiliency to unexpected train work events during a given trip.
Battery state of charge on Route B for trips in increasing and decreasing milepost directions with 4.8 MWh battery capacity (top) and 9.6 MWh capacity (bottom) using the simplified train energy model (note difference in vertical scale corresponding to different battery size).
Round-trip energy savings outputs (Figure 4) for the baseline train configuration (4.8-MWh BEL, 10,000 tons, 5 lb/ton, and 80% efficiency) are consistent with the average percent diesel fuel savings reported for the BEL tests on BNSF in California over a mix of flat and steep gradient terrain. The baseline train configuration consumed 72.1, 71.0, 21.6, and 42.5 MWh of diesel energy, respectively, for the four study routes. Both the simplified approach and ALTRIOS provided similar results in relation to the differences among the corridors. Owing to the differences in the calculations and underlying assumptions, ALTRIOS provided slightly lower values for absolute diesel savings, whereas the percentage savings were rather consistent with the simplified approach. The numerical MWh saving result from ALTRIOS was lower than that from the simplified model because of differences in the train mass model, resistance calculations, and efficiency calculations that tended to reduce the amount of dynamic regenerative braking that could be captured by the BEL in the complex simulation. However, the percentage savings value tended to be the same or slightly higher in the ALTRIOS results because the same differences in the train mass model, resistance, and efficiency tended to decrease the baseline diesel-electric energy consumption. Where the mass-strap assumption in the detailed simulation balanced the uphill and downhill portions of the train at vertical curves in a manner that reduced demand for traction and braking energy, the point-mass assumption of the simplified model created more extreme energy demands at these locations, increasing baseline energy consumption and potential regenerative braking energy with the simplified approach. Both models indicated that on the two mountainous routes (A and B), terminal charging provided fewer incremental benefits per round trip than on the two flatter routes (C and D). Terminal charging is likely to be required in flatter regions because of the lack of regenerative braking opportunities. Although the mountainous routes (A and B) exhibited greater absolute energy savings from the numerous opportunities to capture and reuse regenerative energy, their normalized percentage savings were lower since those routes required greater amounts of energy overall. Thus, it is important to examine both absolute and relative savings when comparing BEL performance across routes.
(a) Absolute and (b) percent round-trip diesel energy savings from one 4.8 MWh BEL on each study route under different charging scenarios and baseline train configuration with simplified approach, and (c) absolute and (d) percent savings from ALTRIOS simulation.
In general, the sensitivity analysis of the simplified train energy model results (Figure 5) revealed that fuel savings from adding one BEL per train were more sensitive to changes in battery efficiency and train weight than changes in train rolling resistance and battery capacity. These findings were supported by the sensitivity of the ALTRIOS simulation results (Figure 6). Since ALTRIOS had a more detailed modeling approach that considered nonconstant efficiency and rolling resistance, only train weight and battery capacity were investigated by the ALTRIOS sensitivity analysis.
Arc-elasticity quantifying relative change in percent diesel energy savings to percent change in experimental factors across Routes A to D and terminal charging options with simplified train energy model: (a) Route A with charging, (b) Route A without charging, (c) Route B with charging, (d) Route B without charging, (e) Route C with charging, (f) Route C without charging, (g) Route D with charging, and (h) Route D without charging.
Arc-elasticity quantifying relative change in percent diesel energy savings to percent change in experimental factors across Routes A to D and terminal charging options with ALTRIOS simulation: (a) Route A with charging, (b) Route A without charging, (c) Route B with charging, (d) Route B without charging, (e) Route C with charging, (f) Route C without charging, (g) Route D with charging, and (h) Route D without charging.
The cases without terminal charging were more consistent between the two methods, whereas the simplified approach estimated higher sensitivity for train weight compared with the ALTRIOS results. Despite observing these general trends, the savings were still highly variable between routes of different grade profiles, thus highlighting the necessity of analyzing each potential BEL route separately in detail. The sensitivity of train weight and battery capacity showed similar patterns for ALTRIOS and the simplified model. The consistency between the simplified and ALTRIOS results suggests that instead of employing a detailed train performance simulation, a simplified train energy approach may be effective in screening for the best BEL routes from the many origin–destination pair possibilities within a larger rail network.
Examining each factor individually revealed that, consistent with expectations, increased battery efficiency resulted in increased energy savings. With increased train weight, the percent energy savings decreased. Heavier trains depleted the battery capacity faster on both level terrain and uphill grades, maximizing the use of BEL energy to offset diesel fuel. The increased train weight also allowed the BEL to recharge faster on downgrades because the heavier train required additional braking energy. This increased rate of charging and discharging the BEL, particularly on undulating terrain, increased the number of charge cycles along the route. Routes that were perfectly flat or traveled in one direction up or down a sustained grade yielded at most one charge or discharge cycle, limiting the BEL benefits from regenerative braking and increased train weight. However, although the mountainous routes increased the number of charge cycles and the productive use of the BEL, heavier trains require proportionately more energy overall, resulting in a decrease in percent diesel savings as the train weight increases. On flatter routes, increasing the train weight more quickly will deplete the BEL battery and increase the distance it must be transported while not contributing tractive effort, potentially increasing overall diesel energy consumption, and further diminishing percent savings as the train weight increases.
Increasing the battery capacity had the effect of increasing fuel savings on routes with undulation. Both mountainous routes (A and B) yielded increased energy savings but with diminishing returns. Each observed route had a maximum amount of regenerative energy that could be produced for a specific trip profile regardless of battery capacity, and therefore further increasing battery capacity was of no benefit (Figure 5 a–d and Figure 6 a–d). The lower capacity in the mountainous regions had a negative effect on energy savings because the battery remained at zero SOC for longer periods of time. Both the flatter routes (C and D) exhibited a slight increase in energy savings for cases with terminal charging. This was consistent with expectations because a larger initial charge enables the BELs to travel longer distances on battery power before diesel energy is required. The flatter routes showed no energy savings without the support of terminal charging regardless of BEL capacity. This was expected because there was little opportunity to use regenerative braking to charge the BEL and it spent extensive miles at a SOC of zero. In general, if a BEL is used in captive service (i.e., dedicated to operating back and forth between a pair of terminals) on one particular route, it is possible to select a battery capacity that is optimized for that unique grade profile and other operational conditions.
It was hypothesized that increasing train resistance would result in decreased energy savings because 1) less regenerative braking is required and thus few BEL charge cycles can be obtained, and 2) increased train resistance increases overall energy consumption and thus diminishes energy savings on a percentage basis. Similarly, decreased resistance would result in increased energy savings. With no terminal charging, a decreased resistance results in a greater increase in energy savings compared with terminal charging scenarios because more regenerative braking is required on downgrades and therefore the battery charges faster and to a higher SOC. Also, once charged in this manner, with less train resistance, the battery is depleted at a slower rate and less time is spent with a SOC of zero, increasing the diesel energy savings. Operation with terminal charging relies on quickly using the stored battery energy to facilitate additional recharge cycles that save additional diesel energy. A higher train resistance will more quickly dissipate the initial terminal battery charge and free up more battery capacity to absorb regenerative energy on downgrades. If the resistance is low, the initial terminal charge will last longer, and there may not be sufficient battery capacity to absorb regenerative energy on downgrades, effectively “wasting” this potential energy.
The ALTRIOS simulation sensitivity results indicated that the simplified approach had overestimated the benefit of a lighter train weight, partially because of the nature of the train mass model used in each approach. ALTRIOS uses mass-strap simulation, which will decrease the sensitivity of train weight since grade changes along the route become more gradual from a train force perspective, and less time is spent with the BEL at maximum tractive effort or regenerative braking. The simplified train energy model makes a point-mass assumption with the entire train weight either moving entirely uphill or downhill, resulting in more time spent at throttle and brake extremes, and no time with a portion of the train effectively balancing each out on sag or crest curves as in the detailed ALTRIOS simulation. Further, because of its detailed dynamic powertrain efficiency functions, the current SOC contributes to BEL efficiency, which could explain why the ALTRIOS results exhibited a lower sensitivity to battery size.
Comparison of Annual Diesel Energy Savings
The results of the annual diesel energy savings analysis from the ALTRIOS output (Figure 7) illustrate the expected trade-off between BEL battery capacity and charging capability, and further reinforce findings on the sensitivity of BEL performance to route topography and charging strategy.
Annual diesel energy savings and required battery charging resulting from combinations of battery electric locomotive (BEL) battery capacity and charging strategy (no terminal charger and 0.5, 1.0, and 2.0 MW terminal charge rates) across Routes A to D with all other factors held at their baseline values: (a) Route A with charging, (b) Route A without charging, (c) Route B with charging, (d) Route B without charging, (e) Route C with charging, (f) Route C without charging, (g) Route D with charging, and (h) Route D without charging.
For the experimental scenarios, annual diesel savings performance with terminal charging was always improved by faster charge rates. On mountainous Route B, where terminal charging had the smallest incremental impact on round-trip energy savings, a strategy with no terminal charging yielded the highest annual savings when compared with all the other routes. When charging scenarios were compared on Route B, the no-charging scenario also provided better annual diesel savings than all but the case with the 2-MW charger. On the other mountainous route, A, the strategy with no terminal charging outperformed scenarios with the 0.5-MW charger. These results suggest that when route topography is favorable, similar or higher annual energy savings may be achieved by deploying BELs to operate solely on regenerated energy, while avoiding the infrastructure installation cost and decreased locomotive utilization of terminal charging.
The flatter routes, C and D, exhibited the opposite trend. Because these routes offered little available regenerative energy, the time penalty of terminal charging was offset by the greatly increased energy savings this offered compared with the no terminal charging strategy. On the flattest route, C, the no terminal charging scenarios actually led to negative savings (i.e., higher overall diesel consumption) owing to the weight penalty of the BEL on the segments of the route where its charge had been depleted.
Another observation was that on all four study routes, the battery size yielding the highest annual diesel savings was most likely to be the largest battery capacity available, the 14.4-MWh battery. There was one exception, on the flattest route, C, where the 14.4-MWh battery was outperformed by the 9.6-MWh battery, indicating the round-trip energy saving improvement from the larger battery storage was offset by the excess terminal charging time. The penalty of operating fewer trains annually as a result of low BEL utilization stemming from long charge times placed the 14.4-MWh batteries at a disadvantage.
From the perspective of electricity consumed by terminal charging, a larger battery capacity will generally result in more annual charging consumption. For certain combinations of corridor topography, battery size, and charging rate, the electricity consumed by terminal charging could decrease slightly for the largest battery sizes since that scenario results in fewer annual BEL round trips. Given that electrical energy for charging has different economic and environmental costs compared with diesel energy, and that these costs can vary greatly by the regional electrical grid generation mix, the energy consumed by terminal charging could not be directly deducted from the annual diesel energy savings values to identify an optimal scenario.
Economic Analysis
The results of the economic analysis are presented as the NPV (Figure 8) and benefit/cost ratio (Figure 9) of each investment scenario. One of the most significant findings is that all corridors favored the 9.6-MWh battery when all charging options were considered. The relatively small incremental diesel savings owing to poor utilization and additional electricity costs for charging the 14.4-MWh BEL did not appear to justify its higher initial battery capital costs. Despite the higher capital cost of faster chargers, the increase in concurrent BEL fleet size supported by each charger combined with facilitating more annual round trips for each BEL caused the 2-MW per hour chargers to offer the best economical outcome across all scenarios that used terminal charging. Although these benefits were achieved by sharing the charger capital cost across a larger BEL fleet size, potentially forcing more widespread adoption for the railroad, the additional benefit compared with the slower charger suggests that developing rapid charging at power rates of 2 MW or higher is a critical research and development need for the long-term economic success of BELs. When only considering the currently available range of 0.5- to 1-MW power chargers, smaller BEL battery sizes of 4.8 MWh produced the largest returns on three of the study corridors.
Net present value resulting from combinations of battery electric locomotive (BEL) battery capacity and charging strategy (no terminal charger and 0.5, 1.0, and 2.0 MW terminal charge rates) across Routes A to D: (a) Route A, (b) Route B, (c) Route C, and (d) Route D.
Benefit/cost ratio resulting from combinations of battery electric locomotive (BEL) battery capacity and charging strategy (no terminal charger and 0.5, 1.0, and 2.0 MW terminal charge rates) across Routes A to D: (a) Route A, (b) Route B, (c) Route C, and (d) Route D.
The economic analysis has shown that for the flat and undulous corridors (Routes C and D), terminal charging scenarios yielded a significantly higher return than the scenario without terminal charging. The additional energy savings from terminal charging greatly surpassed the corresponding infrastructure cost and cost of electricity. The mountainous corridors (Routes A and B) suggested a different outcome. The 2-MW per hour charger was the best deployment strategy for Route A, whereas the no-charging scenario was not far behind. The no-charging scenario appeared to be the best strategy for Route B, suggesting that regenerative braking alone could generate enough energy savings to justify the BEL battery cost, and that terminal charging could not provide enough incremental savings to compensate for the cost of the additional terminal time and infrastructure associated with battery charging. The benefit/cost ratio results (Figure 9) showcased the advantages of not investing in charging facilities, as the smaller overall capital investment resulted in higher benefit/cost ratios for no charging and the smallest BEL battery size.
Although the NPV and benefit–cost analysis provide information on the overall economic efficiency of each deployment scenario, they do not offer insights into the relationship between the capital cost and energy savings of each alternative. To visualize how each deployment strategy leverages the capital cost of BEL batteries and charging infrastructure to obtain energy savings (monetized as the difference between diesel savings and electricity cost), Pareto-optimal plots were created for each corridor (Figure 10).
Pareto-optimal front of combinations of battery electric locomotive (BEL) battery capacity and charging strategy that maximize monetized energy savings while minimizing capital cost across Routes A to D: (a) Route A, (b) Route B, (c) Route C, and (d) Route D.
Each deployment strategy is plotted as a single data point, with the capital cost of batteries and charging infrastructure serving as the horizontal axis, whereas the net monetary value of the diesel saving by adding one BEL (after subtracting the cost of electricity for charging) serves as the vertical axis. The ideal solution would feature a low capital cost while achieving high annual energy savings, placing it toward the top-left corner of the graph. Since these are opposing objectives in practice, the Pareto-optimal front is indicated by a dashed line that shows the preferable scenarios that exhibited the best trade-offs between incrementally more energy savings and added capital cost over the “dominated” solutions that fall below or to the right of the front. These dominated solutions feature lower energy savings for the same capital cost as a Pareto-optimal solution, or more capital cost for the same energy savings, and are thus not preferred. Each study route had a different Pareto-optimal front, indicating that route topography influenced how incremental capital cost translated into additional monetized energy saving benefits.
Based on the results of the economic analysis, BEL deployment strategies can be recommended for each study corridor (Table 3). The number of concurrent BELs indicates the fleet size over which the charger cost is allocated and is thus needed for this best-case benefit to be achieved. Otherwise, the increased share of the charger cost per BEL may result in another strategy being more favorable. Since Route B favored a scenario without terminal charging, fleet size was not a consideration: the same NPV would be obtained for each BEL regardless of fleet size (as long as sufficient trains are operated on the corridor to fully utilize the entire BEL fleet).
Recommended BEL Deployment Strategy for Study Corridors
Note: BEL = battery electric locomotive; na = not applicable.
A significant limitation of the economic analysis is that it views the BEL investment from the private railroad perspective and thus does not consider any monetized public health and climate benefits from the reduced diesel consumption and associated locomotive emissions relative to those produced in generating the electricity used to charge the BEL. If a mechanism were available to capture the value of these public benefits and transfer this to the railroads, some portion of these monetized benefits could be included in the analysis and the BEL economics would be improved.
Conclusions
The research presented in this paper helps to further industry understanding of the capability of BEL technology, and the sensitivity of its ability to reduce diesel fuel consumption to various routes, train, locomotive, and operating factors. Because of the high energy demands of stand-alone BEL operations relative to battery capacity, the likely mainline freight implementation of BELs will be to supplement conventional diesel-electrics by forming hybrid consists. All three experiments demonstrated that the range of BEL performance (both stand-alone and in hybrid consists) was a function of the route topography and availability of terminal charging. Terminal charging had the greatest benefit on flatter routes with limited regeneration potential, whereas mountainous routes may achieve the best annual energy savings by implementing BELs without terminal charging, powered only by recaptured braking energy. The energy savings from adding one BEL per train were more sensitive to changes in battery efficiency and train weight than to changes in train rolling/aerodynamic resistance and battery capacity. However, savings were highly variable between the routes and should therefore be analyzed on a case-by-case basis with particular attention to the trade-off between BEL battery size and terminal charging rate. Because of the time required to recharge a BEL, a smaller battery that yields less savings per trip, but charges faster may result in larger annual energy savings. The economic analysis further quantified the trade-offs between different charging strategies and battery sizes to suggest the most favorable BEL deployment strategies. In all cases, the largest BEL battery size did not produce the most favorable economics, underscoring the overall conclusion that the largest practical BEL battery size is not necessarily always better, but should instead be sized to the particular needs of the corridor topography and planned train operations.
This conclusion raises a limitation of this analysis in that it only considers four discrete BEL sizes. Based on the Pareto-optimal front, intermediate BEL sizes may produce the most favorable BEL deployment strategies. Future work could repeat this analysis at a finer resolution to determine the true optimal BEL size for each study corridor.
A key limitation of the simplified train energy modeling approach is that it makes basic assumptions about train speed, uses greedy BEL energy management, and ignores curve resistance while using a point-mass train resistance assumption. However, the results produced by this approach were generally comparable with the more detailed train performance simulation conducted with the validated ALTRIOS model. Thus, despite these limitations, the simplified train energy modeling approach should be adequate for a high-level examination of the general effects of various parameters on BEL performance relative to diesel-electrics subject to the same assumptions, or to conduct an initial screening analysis to identify promising BEL deployment segments from within a larger freight rail network. For more detailed studies, ALTRIOS provides a more comprehensive energy consumption analysis with its advanced simulation and tractive effort requirement calculations. However, owing to a lack of curvature data on some segments of the case study corridors, this study could be further improved by including horizontal curves in the analysis, although this has been shown to have a much smaller effect relative to vertical gradient. For the simulation results based on vertical data, ALTRIOS confirmed the findings from the simplified approach with only minor differences in the sensitivity analysis of the train weight and battery capacity.
A final limitation and subject of further research is that the economic analysis makes various assumptions about some of the costs for emerging technologies such as batteries and terminal chargers. The analysis was also only conducted for one set of baseline economic parameters (discount rate of 5% and analysis period of 7 years) and a charger utilization rate of 85%. The sensitivity of the economic analysis to these parameters is the subject of future research, as these values are subject to change as a result of continued technological developments and evolving economic circumstances.
Footnotes
Acknowledgements
The authors thank Henry Stuckey, Undergraduate Research Assistant at the University of Texas at Austin, for conducting the SCORE tool analysis.
Author Contributions
The authors confirm contribution to the paper as follows: study conception and design: D. Shi, M. Copley, T. Dick; data collection: D. Shi, M. Copley, T. Dick; analysis and interpretation of results: D. Shi, M. Copley, T. Dick; draft manuscript preparation: D. Shi, M. Copley, T. Dick. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: C. Tyler Dick is a member of Transportation Research Record’s Editorial Board.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Department of Civil, Architectural and Environmental Engineering at the University of Texas at Austin. The lead author received additional support from the National Renewable Energy Laboratory in addition to his position at the University of Texas at Austin. The second author was supported by the Association of American Railroads and the CN Railway Engineering Research Fellowship at the University of Illinois Urbana-Champaign.
