Leveraging rail-based mobile energy storage to increase grid reliability in the face of climate uncertainty

The paper addresses how to maintain and enhance reliability and resilience of electric grids amid increasing extreme weather events, growing penetration of intermittent renewables, and evolving demand patterns during decarbonization. Traditional approaches relying on peaker plants and planning based on historical conditions are increasingly inadequate as risks become more variable, severe, and uncertain. Interregional transmission can diversify risk but faces political, logistical, and cost hurdles, and its value is concentrated in a small fraction of hours. The authors propose rail-based mobile energy storage (RMES)—containerized batteries transported via the extensive US freight rail system—as a flexible, shareable capacity resource to meet infrequent, high-impact events, relieve congestion, and provide redundancy without committing to fixed, duplicative stationary capacity or new transmission lines. The research questions are whether RMES is feasible operationally and predictably, and whether it is cost-effective compared to stationary storage and transmission for low-frequency, high-impact events.

Prior studies have examined mobile energy storage (MES) concepts including vehicle-to-grid and road-transported batteries, finding benefits such as reduced curtailment, operational flexibility, congestion relief, and peak shaving. However, road-based MES is constrained by weight limits and has focused on daily operational arbitrage with assumptions of perfect coordination. Rail-based mobility offers orders-of-magnitude higher carrying capacity (a single train ~1 GWh; roughly 1,000 trucks). Existing literature has not assessed real-world freight constraints, national-scale feasibility, or the potential to avoid redundant stationary assets for low-frequency reliability needs. The paper situates RMES within this gap, building on empirical transmission valuation and freight data to evaluate feasibility and value for infrequent, high-impact events.

Feasibility and value are assessed using empirical freight and power system data and comparative cost modeling: - Freight feasibility: 2019 Waybill sample data (origin–destination, volumes) are used to estimate average daily trains from each state to each ISO and travel times. Trains travel ~65–80 km/h for 23–24 h/day, implying ~1,480–1,930 km/day. Adding 1–2 days of scheduling time yields estimated RMES shipment times of 1–6 days between most ISOs, with many inter-ISO routes having daily freight flows. - Non-coincidence and value concentration: Using 2010–2021 day-ahead (DAH) and real-time (RTH) LMPs at representative pricing nodes across CAISO, ERCOT, ISO-NE, MISO, NYISO, PJM, and SPP, the authors compute inter- and intra-ISO synthetic transmission line values via hourly price differentials. Price correlations among most regions are 0.3–0.7, weaker across distant/AC-separate regions, implying non-coincident stress. They find 17–25% of total transmission value concentrated in the top 1% of hours. To reflect RMES unidirectional, non-instantaneous transfers, they compare bidirectional arbitrage value to unidirectional value in the day/week/month surrounding high-value events; unidirectional captures ~85% (day), ~70% (week), and ~40% (month) of bidirectional value nationally. - Predictability: For events within 1 day, DAH prices proxy tight supply; annual DAH and RTH price spikes align >90% of the time. For 2–7 days, gross load forecast data (2010–2020) are used; in the top 10% load hours, mean forecast error is under 5% in all ISOs, with CAISO having larger variance (upper bound ~20% at 7 days). This suggests sufficient lead time predictability for RMES scheduling. - Cost comparison framework: Three strategies are compared for serving low-frequency events between two regions: (1) duplicative stationary battery capacity in each region; (2) new interregional transmission; (3) RMES. Fixed costs include battery capital, siting/developer/interconnection (SDI), and fixed O&M; for stationary storage, costs are doubled to reflect duplicative assets. RMES fixed costs include one battery set and two interconnections; a four-hour duration and a moderate 2025 cost trajectory are assumed. Transmission fixed costs use a 345 kV single-circuit AC line with capital cost ~US$8.27 per (kW·km), inclusive of land, plus two substations. Variable costs: battery round-trip efficiency losses valued at US$30/MWh off-peak (assumed RTE 85%); transmission line losses at 0.01% per km; RMES incurs both battery losses and freight delivery costs. Freight cost uses ~US$0.03 per t·km; an illustrative per-container module of ~9 MWh at ~89 t is cited. Freight deliveries per year scale with annual event frequency (AEF), number of regions sharing the asset (n=2), distance between regions, and storage duration (4 h). Scenario analysis varies distance and AEF to map relative cost-effectiveness. A New York State case study applies the same LMP-based valuation intra-ISO and examines intrastate freight times (1–11 hours, plus scheduling).

- Freight and timing feasibility: Most ISOs already exchange daily freight; RMES can be moved between major power markets in 1–6 days, fitting within predictability windows for high-impact events. - Non-coincidence and value concentration: Inter-ISO LMP correlations are modest (0.3–0.7), especially across distant/AC-separated regions, indicating non-coincident stress. Nationally, 17–25% of modeled transmission value concentrates in the top 1% of hours, and unidirectional arbitrage can capture ~85% (day), ~70% (week), and ~40% (month) of bidirectional value around high-value events. - Predictability: DAH price spikes align with RTH spikes >90% of the time; gross load forecast errors in the top 10% load hours are on average <5% for 1–7 days ahead (larger variance in CAISO), enabling RMES scheduling lead times. - Cost-effectiveness: For low-frequency, high-impact events, RMES can be more economical than duplicative stationary batteries at short distances. Example: at ~400 km total travel, RMES is preferable when called on <2% annually per region. For extremely rare events (0.1% AEF), RMES outperforms stationary batteries at all distances. Versus new transmission, RMES is more economical for low-frequency events especially as distance increases; at low AEF, RMES becomes more economic than transmission beyond ~1,500 km because transmission capex scales faster with distance than RMES shipping costs. - Quantified savings: Relative to new transmission lines and stationary battery capacity, RMES deployment for such events could save upwards of ~US$300 per kW-year and ~US$85 per kW-year, respectively. - New York case study: Intrastate freight movement takes 1–11 hours (plus 1–2 days scheduling), with 1–3 daily trains between regions. In NYISO, 14% (DAH) to 26% (RTH) of potential transmission value concentrates in the top 1% of hours; unidirectional arbitrage retains ~85–95% of bidirectional value (day to month windows). RMES could mitigate downstate supply risk (e.g., wind drought) and transmission constraints, potentially substituting part of planned multi-billion-dollar transmission investments and expanding access to upstate/Canadian hydropower and out-of-state renewables within days by rail.

The findings show RMES can reduce the need for duplicative, underutilized stationary capacity and costly, inflexible transmission expansions to address rare but consequential grid stress events. By sharing mobile storage across regions with largely non-coincident peaks and stressors, RMES diversifies access to supply, provides capacity and transmission redundancy (supporting N−k contingencies), and enhances resilience against extreme weather and outages. Its cost advantage is strongest for infrequent events, at shorter distances relative to stationary capacity, and at longer distances relative to transmission. RMES can also create new import paths during emergencies, complementing existing lines and alleviating congestion. Practical deployment hinges on addressing interconnection logistics and regulatory classification (energy-only, capacity, or transmission resource), but technically RMES can participate. Cross-sector opportunities (e.g., freight rail decarbonization, maritime containerized batteries) could expand RMES availability and improve asset utilization, with supportive policies (e.g., California LCFS) improving economics.

The study proposes and evaluates rail-based mobile energy storage as a flexible, shareable reliability resource for a decarbonizing grid facing increasing climate uncertainty. Empirical freight data and power market analysis indicate RMES can be scheduled and moved within required lead times to address low-frequency, high-impact events, with substantial value concentrated in a small number of hours and largely capturable via unidirectional, non-instantaneous movement. Comparative cost modeling shows RMES can be more economical than duplicative stationary batteries for rare events (especially at shorter distances) and more economical than new transmission for rare events at longer distances, with potential sector-wide savings on the order of tens to hundreds of US dollars per kW-year. A New York case study highlights RMES’s potential to mitigate intrastate congestion and peak risks. Future research should integrate RMES into detailed system planning and dispatch models, evaluate operational benefits under high-renewables scenarios, refine event predictability using net load and weather forecasts, analyze future coincidence of stressors under climate change, and develop optimal siting/interconnection strategies. Policy work is needed to adapt market rules and interconnection processes to enable RMES participation at scale.

The analysis is high-level and does not fully model RMES in power system planning/operations or quantify dynamic operational benefits. Predictability assessments rely on historical gross load forecasts rather than net load and weather-based forecasts that better capture renewable variability; CAISO forecast variance suggests added uncertainty. The study does not project how event coincidence may change in a decarbonized, climate-impacted future. Cost analyses use assumed parameters (e.g., 4-hour duration, 2025 cost trajectories, average off-peak price) and simplified freight logistics; interconnection upgrade costs and siting constraints could be substantial and location-specific. Regulatory feasibility and revenue sufficiency under current market rules remain unresolved.