Conference Agenda

EV adoption, a key transportation decarbonization strategy, is projected to drive global battery demand to nearly 4 TWh yr-1 by 2030. Growth in battery demand presents sustainability challenges driven by high demand for critical minerals, energy-intensive and costly production, and manufacturing and end-of-life infrastructure requirements. s. Batteries are responsible for approximately 40-60% of GHG emissions during EV production and approximately 27-30% of the purchase cost of EVs. A broad array of stakeholders and decision-makers are responsible for, or impacted by, battery life cycle management decisions that influence sustainability outcomes. We present a framework spanning the full battery life cycle and identify consequences and sustainability trade-offs for stakeholders. We identify 17 key drivers classified into 5 phases (battery design, materials production, manufacturing, use, and end of life) which determine the life cycle sustainability performance of batteries using a range of sustainability metrics. Across the framework we explore the influence of battery chemistry, identify sustainable principles in battery design, investigate material- and process-level life cycle impacts, trace the use-phase impacts of battery mass, efficiency, and degradation, evaluate the trade-offs between end-of-life pathways, categorize current battery management policy layers, and critique several current battery policies. The framework is designed to help decision-makers assess trade-offs and avoid unintended consequences across life cycle stages, stakeholders, and impact categories. Its development was informed by a critical review of the battery sustainability literature and stakeholder interviews across different sectors and fields of expertise. We highlight information transparency, infrastructure investment, systems-based policy and solutions, and multi-stakeholder collaboration as key enablers for sustainable battery life cycle management. Insights from the framework provide guidance for current and future battery management, by considering the outlook for management decisions throughout all phases. To remain competitive, the U.S. automotive industry must manage costs as well as environmental and social impacts – this framework provides guidance across all three pillars of sustainability (economic, environmental, and social). The framework encourages a holistic assessment of the battery life cycle as an integral part of management decisions and reinforces the need to consider the diverse range of sustainable development goals.

Hydrogen is a resource that is presently used as an industrial feedstock and is projected to play a pivotal role in future green energy storage systems. Of the multiple ways to generate hydrogen to meet future demand, water electrolysis is a strong contender owing in large part to its potential for reduced environmental burden compared to the incumbent fossil fuel-based production technologies. In order to understand electrolytic hydrogen’s full environmental impact (quantified by carbon emissions) and economic impacts one important variable to consider is the scale of the production plant. Centralized plants, or those supplying to more than one demand source, are generally bigger and therefore benefit from efficiencies of scale. They are simpler to manage and can deliver to many sites, supplying both large and small demand sources. The downside is, of course, the necessity to transport the hydrogen from production to demand site and lessened flexibility of having one big supplier providing hydrogen to multiple sites with potentially diverse needs. However, this defining variable has been given limited attention in literature. There are a small number of studies that focus on the economic scaling of these plants (1–3) but none that focus directly on the technology scale-up on environmental impacts. However, there are a significant number of studies that focus on individual electrolysis plants at specific scales both environmentally and economically, but this is limiting since these studies are highly context dependent without the ability to easily adapt findings to new scales, background systems, etc.

The aim of this research is to fill that gap by creating a python-based model integrating cradle-to-gate life cycle assessment (LCA) and technoeconomic assessment (TEA) that determines, given a certain context in terms of locations, scales, and requirements of the hydrogen demand sites, whether to deploy a centralized or decentralized hydrogen production system to minimize environmental and economic burdens. The model accounts for the infrastructural requirements of centralized hydrogen production by modelling pipeline network production and operation through linear optimization. To test this model, we conducted a case study using steel sites in Ontario, Canada as our demand sources, though the model is adaptable to other situations as well.

Preliminary results suggest that, at the scales modelled, that the economic benefits of centralized plants are substantial, lowering normalized production costs by close to 20%, but the additional distribution costs reduces that benefit. Environmentally, because the vast majority of emissions are from electricity demand for electrolysis (which is fixed regardless of scale), there are negligible benefits from larger scales, and considering distribution emissions there is effectively no scenario in which a centralized plant would have a lower greenhouse gas emission.

Ultimately, the model can be applied to situations other than the steel sector in Ontario, and in those other situations, it can help inform policymakers and planners regarding the optimal hydrogen production methodology, let it be centralized or decentralized based on the specific deployment context.

Bibliography

1. Reksten AH, Thomassen MS, Møller-Holst S, Sundseth K. Projecting the future cost of PEM and alkaline water electrolysers; a CAPEX model including electrolyser plant size and technology development. Int J Hydrog Energy. 2022 Nov 9;47(90):38106–13.

2. Paudel A, Paneru B, Mainali DP, Karki S, Pochareddy Y, Shakya SR, et al. Hydrogen production from surplus hydropower: Techno-economic assessment with alkaline electrolysis in Nepal’s perspective. Int J Hydrog Energy. 2024 July 12;74:89–100.

3. Schmidt O, Gambhir A, Staffell I, Hawkes A, Nelson J, Few S. Future cost and performance of water electrolysis: An expert elicitation study. Int J Hydrog Energy. 2017 Dec 28;42(52):30470–92.

Brackish water reverse osmosis (BWRO) systems are limited to ~70% water recovery due to mineral scaling from concentrated calcium and magnesium ions, requiring costly disposal of nearly one-third of the feed concentrate. Zero liquid discharge (ZLD) systems offer a pathway to further increase recovery and enable mineral valorization, yet the thermal concentration step in ZLD can face challenges from mineral precipitation and associated fouling. While individual treatment components such as chemical softening, secondary membrane processes, and thermal concentration exist, their integrated system-level performance, economics, and environmental impacts for BWRO concentrate management have not been comprehensively evaluated. This study develops and assesses an integrated treatment system combining chemical softening, recarbonation, and secondary reverse osmosis prior to mechanical vapor compression (MVC) and crystallization to enable high-recovery, sustainable ZLD for inland desalination. The research integrates process modeling, techno-economic analysis (TEA), and life cycle assessment (LCA) to evaluate the system-level performance. First, process simulation in Aspen Plus was coupled with an aqueous phase modeling software to quantify mass and energy balances of the complete treatment train. The techno-economic analysis calculated capital and operational expenses, and a discounted cash flow rate of return analysis was used to determine the levelized cost of water, net present value, and payback period. A cradle-to-gate LCA was conducted using open-source and commercial databases to evaluate ten environmental impact categories with a focus on global warming potential (GWP). In addition, credits for recovered minerals and systematic sensitivity analysis were included to evaluate the sensitivity of key economic and environmental outputs. Preliminary results indicated that the ZLD train achieved a LCOW of $2.43 per m3 and 15.4 kg CO2e m-3 of water recovered. Process-level analysis revealed that MVC represents the dominant cost contributor, accounting for 51% of total operational expenses due to its high energy intensity. This research provides critical quantitative evidence for the techno-economic and environmental viability of integrated ZLD systems, offering a pathway for water-scarce inland communities to achieve sustainable desalination while converting waste brine into recoverable resources and reducing disposal-related environmental burdens.

PFAS groundwater monitoring is costly, and the data we end up with are rarely uniform in space. In New Hampshire, sampling has focused on known or suspected source areas in the south and along the seacoast, while large rural regions remain sparsely tested. This targeted pattern is valuable for site investigations, but it also creates a clustered dataset dominated by non-detect and sub-MCL results—conditions that can mask exceedance risk and reduce confidence when predictions are extended to under-sampled areas

Here we frame that limitation as a sampling-design problem: how can monitoring be thinned in already dense hotspots, expanded into sparse regions, and still preserve the statewide PFAS signal needed for decisions? Using NHDES groundwater measurements (2016–2023) for four regulated compounds (PFOA, PFOS, PFHxS, PFNA), we train compound-specific random-forest classifiers to predict regulation-based classes (Low ≤ MDL; Medium MDL–MCL; High > MCL) on a 1-km statewide grid. Because overall accuracy is inflated in PFAS datasets, model performance is summarized using macro-F1 and class-wise recall.

The central contribution is a grid-weighted resampling strategy (A3) that deliberately declusters dense southern sampling and redistributes training support across space using cell-level information content (variability and area), with class balancing applied within the grid so minority-class samples are not concentrated in a few investigation zones. Compared with the baseline (A1) and global class balancing (A2), A3 reduces the training set from roughly ~17k samples to ~5.0–5.5k (about a two-thirds reduction), while retaining practical skill. Across compounds, macro-F1 values span 0.71–0.78 overall; under A3 they remain 0.71 (PFOA), 0.71 (PFOS), 0.72 (PFHxS), and 0.73 (PFNA), staying within ~0.04–0.06 of the best-performing approach for PFOS, PFHxS, and PFNA (and identical for PFOA). Grid-wise evaluation further indicates that A3 maintains—and in several cases improves—spatial generalization relative to the unmodified training set, whereas global balancing can reduce spatial robustness.

When projected statewide, predicted risk forms coherent belts in southern and southeastern New Hampshire: PFOA shows the broadest footprint, PFOS is similar but more compact, and PFHxS/PFNA hotspots are more localized and largely nested within the PFOA/PFOS belt. A pooled “maximum-class” surface highlights corridors along the Merrimack–Nashua region and parts of the seacoast where at least one regulated PFAS is likely to approach or exceed its MCL, supporting targeted follow-up sampling, private-well outreach, and source control. Overlaying risk with social vulnerability further suggests disproportionate burden in high-SVI communities (e.g., ~38.7% of residents in the highest SVI quartile falling in ≥MCL cells, and ~47% of exposed private wells in ≥MCL cells). Overall, the results point to a practical monitoring pathway: fewer tests, fewer blind spots, and clearer prioritization where modeled risk and vulnerability intersect.