Nathalie Voisin

Hydropower plants will operate under new conditions as water regimes shift, reservoir operating rules evolve, and the grid requires flexible balancing of wind and solar. Hydropower plant turbine systems are traditionally designed to operate at the highest efficiency within narrow ranges of flow and reservoir levels while deviations from design conditions can degrade HP efficiency and trigger failures, especially in aging infrastructure. Turbine replacements typically occur every 35 to 50 years, offering a rare opportunity to modernize turbine designs by adjusting turbine head and discharge capacity to fit new conditions, rather than replicating legacy designs. The central challenge is to determine when to replace instead of extending lifetime of existing fleet, which system signals indicate the need for redesign, and which configurations will deliver the most robust gains. This study introduces HyTUNE (Hydropower Turbine Upgrade and Next‐generation Planning), a dynamic decision‐support tool that integrates basin hydrology, plant hydraulics, and adaptive optimization to guide turbine replacement timing and configuration. HyTUNE shifts planning from reactive, end‐of‐life replacement toward proactive, information‐driven strategies. The framework learns from evolving system states, identifies threshold conditions where adjustments in head or capacity improve performance, and prescribes phased interventions that balance energy production, reliability, and cost under uncertainty. Application to the Hoover Hydropower Plant demonstrates that HyTUNE’s adaptive policies consistently outperformed conventional reactive strategies across diverse hydrologic futures. HyTUNE is of considerable practical relevance as hydropower systems worldwide confront the dual challenge of modernization and adaptation to future uncertainty and demand growth.

Reservoir drought, a form of hydrological drought, is characterized by abnormally low reservoir storage levels, reflecting combined impacts of natural water shortages and water management practices. Despite their profound impacts on water, energy, and agricultural sectors, systematic projections of how reservoir drought and resilience may evolve remain scarce. This study addresses this gap by quantifying shifts in reservoir drought signatures (onset, severity, duration, recovery and frequency) across five heavily regulated US basins under rising temperatures. By coupling atmosphere‐land‐river models with reservoir operations at ∼12‐km resolution, we simulated multi‐decadal reservoir storage dynamics and characterized drought signatures for individual reservoirs. We follow a no‐adaptation scenario, holding water demands and operational rules fixed, to isolate warming‐driven stress on reservoir systems under present‐day management conditions. Our findings reveal significant regional and functional disparities in reservoir drought resilience under the scenarios of rising temperature. Reservoirs in the Texas‐Gulf region are projected to become more resilient, while systems in the Upper Colorado and South Atlantic‐Gulf regions face increased risk due to prolonged drought durations (increasing by ∼6–10 months) and slower recovery (with recovery times increasing by up to three‐to fourfold). With respect to primary function, reservoirs used for irrigation and hydropower, particularly those with smaller storage capacity and lower degrees of regulation, are most susceptible to future drought stress. Overall, this study provides a novel continental‐scale benchmark for reservoir droughts to support evolving multi‐sectoral drought mitigation efforts.

In recent years, several reservoirs in the western U.S. have approached or exceeded critically low storage during drought conditions; such events could be particularly impactful to water and energy availability if they are widespread across the region or exacerbated by climate change. However, projected changes in low storage and near-minimum-power-pool (MPP) risk remain poorly quantified across large spatial domains. Here, we assess simulated low storage frequency (LSF) across 94 western U.S. reservoirs using a calibrated offline hedging release model driven by projected reservoir inflows under future climate scenarios. We evaluate near-MPP risk for 30 reservoirs for which reservoir-specific thresholds are known. Projected low storage responses are spatially and seasonally heterogeneous. Across the ensemble, annual LSF declines in many future simulations, and near-MPP exposure remains limited for most reservoirs. Thus, within the climate-driven inflow projections and hedging framework, future projections do not produce a widespread increase in critically LSF across western U.S. hydropower reservoirs. This broad pattern masks important regional and seasonal differences, with elevated late-season vulnerability emerging in California and parts of the Southwest. LSF is more strongly associated with precipitation than with temperature, with drier years producing elevated risk across nearly every reservoir and warming amplifying risk in California and the Southwest. These results indicate that estimating operational threshold risks from climate-driven inflow projections remain challenging, particularly for low storage conditions that matter most for hydropower vulnerability.

Extreme weather events, such as Heat Waves (HW) and Cold Snaps (CS), pose significant risks to the power grid. The United States (U.S.) Federal Energy Regulatory Commission Order No. 896 mandates regional coordination standards that account for extreme thermal events. However, the lack of a universal definition for extreme thermal events may lead to inconsistent compliance efforts among neighboring entities, undermining the reliability of the transmission system. This study directly addresses this challenge by systematically evaluating how varying technical choices in defining HW and CS fundamentally impact the characterization and ranking of extreme events for power grid reliability studies. We used 12 event definitions and multiple temperature spatial aggregation approaches to construct historical (1980–2024) regional extreme thermal event libraries across North American Electric Reliability Corporation (NERC) subregions in the conterminous U.S. We examined the sensitivity of event characteristics (e.g., duration, frequency, intensity, and spatial coverage) to different definitions. While some definitions produced similar libraries and top event rankings, definitions based on moving‐window‐averaged temperatures yielded markedly different characteristics. Spatial aggregation methods had minimal impact on heat wave or cold snap intensity, frequency and duration but significantly influenced spatial coverage. The top events identified across different aggregation methods were consistent, but their ranking order varied. These findings offer critical insights for characterizing and selecting extreme thermal events and for supporting local and cross‐regional coordination as required by reliability standards.

Large Language Models show promise for translating natural language into database queries, but deploying such systems in safety-critical domains requires high reliability. We present an application of GridCoPilot to thermal event analysis (heatwaves and coldwaves) that affect power grid reliability. Our approach uses a LangChain SQL Agent to translate natural language queries into auditable SQL statements, with deterministic visualization routines that parse the structured query results. We introduce structural framing as a design principle, we integrate a NERC-region-level event library with county-level meteorology and decompose the combined data into three relational tables (event metadata, county-level event details, and a county-to-NERC subregion mapping), using prompt-guided joins to direct the model toward correct multi-table queries. For two core analytical patterns (identifying worst events by region and by region-year), the system achieved 100% SQL accuracy across all 16 NERC subregions and both event types (64 queries total). These results validate the approach for target use cases, though performance on diverse natural language formulations requires further investigation. We discuss design trade-offs, failure modes including JSON output truncation, and pathways for extending this approach to other hazard domains.

Hydropower is the largest source of renewable electricity and a central component of the water–energy nexus and the net-zero transition. Aging infrastructure, combined with climate variability and evolving grid demands, is reducing the efficiency and operational flexibility of Hydropower Plants (HP) designed for past climate and grid conditions. Meeting future energy needs requires modernizing the existing fleet, for example through turbine replacement which occurs every few decades.In practice, turbines are often replaced with identical units that replicate legacy configurations optimized for past conditions, but replacements constitute an opportunity to redesign turbine head and discharge capacity to match evolving hydrology, reservoir operations, and grid needs. In drought-prone systems, for example, installing units optimized for lower head can sustain generation at reduced reservoir levels, as demonstrated by recent upgrades at Hoover Dam on the Colorado River. What is missing is a rigorous method to determine when and how turbines should be upgraded to ensure efficient, reliable, and sustainable outcomes.This study addresses this need for large-scale hydropower upgrades by using a newly developed framework, HyTUNE (Hydropower Turbine Upgrade and Next-generation Planning). HyTUNE is a dynamic decision-support tool that integrates basin hydrology, HP hydraulics, and adaptive optimization to inform turbine replacement timing and configuration. HyTUNE learns from evolving system states and identifies threshold conditions where adjustments to design head or discharge capacity improve hydropower performance.Application to the Hoover Hydropower Plant in the Colorado River Basin shows that HyTUNE’s adaptive policies consistently outperform benchmark replacement strategies across diverse hydrologic futures. The approach increases economic returns, measured as net present value, and enhances plant capability through higher firm power, peak-period generation, and operational efficiency, with fewer turbine replacements. Climate variability still shapes outcomes, with the largest benefits of HyTUNE compared to benchmark expected under wetter conditions, and the strongest improvements in firm power and operational efficiency under drier conditions. HyTUNE offers a practical framework for hydropower systems facing the combined challenges of modernization, climate uncertainty, and growing demand.

Hydropower is a reliable renewable energy source that plays a central role in the water–energy nexus and in integrating emerging energy loads and generation technologies. Hydropower plants (HPs) are designed to operate efficiently within defined ranges of reservoir releases and water levels. Deviations from these conditions, driven by changes in water availability, regulatory constraints, and evolving energy grid demands, reduce operational efficiency. As a result, less energy may be produced per unit of water in a future where hydrology and water operations are meaningfully different from what the HPs were designed for.However, current state of the art large-scale, long-term energy planning models assume constant, near-optimal turbine efficiency or even constant hydraulic head, ignoring variability in HP efficiency and losses. These simplifications lead to systematic overestimation in our future projections of hydropower generation and capacity. They can also lead to underestimation of future resource adequacy needs, and consequent underinvestment in complementary energy infrastructure, increasing risks to future grid reliability. Models and approaches that provide more accurate, temporally and regionally resolved assessments of hydropower potential are therefore needed to support informed planning decisions.Here we introduce HEADFIT (Hydraulic-Energy Analysis and Dynamic Fitting), a physics-informed framework for analyzing and calibrating hydropower system performance in long-term water–energy planning models. HEADFIT integrates plant hydraulics, including frictional and minor head losses, tailwater dynamics, and operational limits, with turbine efficiency curves for a high-fidelity estimation of how hydropower generation is expected to change in a changing climate. These relationships are approximated at the plant level using physics-informed relationships calibrated with observed hydrological and operational data. The calibrated plant-level models are then used to project hydropower generation under future hydrological scenarios. Lastly, we employ a Western United States power system model to propagate refined hydropower projections into more accurate grid performance assessments across time scales.Preliminary analysis for 15 major hydropower plants across the Colorado and Columbia River basins shows that constant-efficiency assumptions overestimate annual hydropower production by an average of five percent, with larger biases during periods of high releases combined with low reservoir levels. These discrepancies reduce the accuracy of capacity and flexibility estimates that support essential grid services. They can also misguide investment and design decisions, increasing risks to grid reliability as climate and demand variability intensify.

Power outage datasets report on the spatial and temporal characteristics of customers without electricity and provide insights into trends in reliability and compliance. As the availability and quality of such datasets vary, an understanding of fit-for-purpose is needed. We compare two U.S. outage datasets, EAGLE-I and IEEE, from 2018-2022 to evaluate their strengths and weaknesses for different scientific questions and identify complementarities. The datasets differ in collection methodologies and in temporal and spatial scales. EAGLE-I captures a snapshot of outages sourced from the web, whereas IEEE compiles data from utilities directly. We use outage statistics, correlations of daily Customers Interrupted (CI), and case studies of extreme events to illustrate their differences. Despite processing the datasets to a common spatial and temporal resolution, direct comparisons are challenging due to differences in their methodologies and data collection approaches. EAGLE-I covers a larger share of customers, contributing to systematically higher CI and the associated estimated Customer Minutes Interrupted (CMI). Pooled across NERC (NERC LTRA) regions, we find weak linear mapping in CI between the two datasets, indicating that, in general, their trends in reliability metrics would not be consistent. Consistency is higher for some NERC regions, specifically Florida, Northwest, Southwest Power Pool, and Texas. EAGLE-I is advantageous for fine-scale, event-focused analysis. IEEE is more suitable for long-term regional benchmarking and analyses that support mapping between power outages and evolving asset characteristics. We conclude that efforts to promote coordination and harmonization between outage datasets are needed to expand analytical capabilities about reliability events.

Hydropower facilities represent a key electricity generating resource in the U.S. Western Interconnection. These facilities rely upon forecasts of inflow when scheduling releases to generate electricity. However, hydropower operations represented in bulk power systems models do not reflect uncertainty in inflow forecasts. This study aims to evaluate how inflow forecast uncertainties impact hydropower generation and revenues at the scale of an entire power grid at a spatial scale relevant to power system modeling. The question is critical and timely as more flexibility is called upon to integrate other technologies without understanding the flexibility already exercised. New advances are needed to represent hydropower contributions under operational uncertainty at the interconnection scale. Our contribution includes the development of consistent and coincident medium‐range (0–7 days) inflow forecasts and a generic hydropower scheduler, Forecast‐Informed Scheduler for Hydropower (FIScH), that captures non‐powered water management objectives and constraints and allows for varying electricity prices. This scheduler was applied at 242 hydropower facilities representing 86% of the conventional nameplate capacity in the Western Interconnection. Hydropower revenues were examined for schedules developed using three sets of inflow forecasts with differing levels of accuracy over a 20‐year period from 2000 to 2019. In aggregate, we find that annual hydropower revenue decreases 0.08% when using more skillful forecasts, and 0.11% when using baseline persistence forecasts as compared to revenue using perfect forecasts. Regional and interannual results were more varied and ranged between −1 and 4%. The translation of improved forecast skill into higher revenues is non‐linear and varies regionally, with larger revenue changes on the west coast and smaller responses across the interior western U.S. Overall, we demonstrate that scheduling mostly alleviates the impact of inflow forecast errors on hydropower revenue. The study motivates the need for a more detailed evaluation into which specific hydrologic events impact hydropower scheduling and revenue at the system scale.