Data Use – Classroom Presentations
CLEAN CURRENTS 2026
Time: 3:45 PM - 4:45 PM
Day: 9/24/2026
New: Add Sessions to Your Calendar
Interested in this session? Click "Add to My Calendar" and save specific sessions so you don’t miss out. Available for Mobile and Desktop
Calendar Note: Selecting “Add to Calendar” may download a calendar file (.ics) to your device. On mobile, tap the downloaded file to open it in your calendar app. On desktop, open the downloaded file to add the event to your calendar.
Details about each presentation and the speakers are below:
In this session:
Hydro Fleet Intelligence (HFI) – AI-Enabled Asset Condition Analytics
Data as Regulatory Infrastructure: Faster Reviews and Stronger Records for Modern Hydropower Relicensing
Advanced AI and Optimization for Flexible Hydropower Cascades in Modern Electricity Markets
Hydro Fleet Intelligence (HFI) – AI-Enabled Asset Condition Analytics (Oak Ridge National Laboratory)
Presented by Srijib Mukherjee, Oak Ridge National Laboratory
The Hydropower Fleet Intelligence (HFI) project addresses a critical gap in the hydropower sector: the limited availability and adoption of data-driven asset condition models and advanced analytics capable of delivering actionable insights for asset management. Current practices rely heavily on expert elicitation and fragmented data sources, which constrain the ability of hydropower owners to systematically evaluate asset reliability, operational risk, and operations and maintenance (O&M) costs.
DOE's HFI being performed at Oak Ridge National Laboratory (ORNL) advances a comprehensive analytics framework that integrates asset condition modeling, artificial intelligence (AI)-enabled multi-sensor data fusion, and scalable software tools to support evidence-based decision-making across the hydropower fleet. The project has developed modular analytics and software components that can be directly adopted by industry or extended with additional analytics and data management capabilities to address a broad range of operational and planning use cases.
Our presentation will reflect a key innovation of HFI: the development of AI foundation models designed to integrate numerical measurements with unstructured textual data routinely generated during hydropower operations, such as maintenance logs, inspection reports, and event records. These models will eventually support automated data quality assessment, alignment, and integration across disparate data types, enabling the synthesis of insights that are not attainable from individual data streams alone.
In conjunction with ongoing efforts within HFI and the broader hydropower community to characterize asset-specific failure modes, the analytics framework also facilitates the identification of data gaps and prioritization of additional data sources that can yield high operational value. HFI is extending its capabilities through the development of physics-informed machine learning models as part of an integrated digital twin effort, further enhancing predictive accuracy and interpretability. Collectively, the HFI project establishes a scalable, data-centric foundation for risk-informed, cost-effective, and resilient hydropower asset management.
Data as Regulatory Infrastructure: Faster Reviews and Stronger Records for Modern Hydropower Relicensing (Kleinschmidt Associates)
Presented by Joetta Zablotney, Kleinschmidt Associates
Hydropower relicensing has become one of the most data-intensive regulatory processes in the energy sector. Study scopes are expanding; interdisciplinary coordination is more complex than ever, and agencies increasingly expect transparent, reproducible analyses. In this environment, data management is no longer an administrative function, it is regulatory infrastructure.
Relicensing efforts generate vast and diverse datasets across hydrology, fisheries, water quality, habitat mapping, geomorphology, operational modeling, and stakeholder engagement. When data are fragmented, inconsistently structured, or poorly documented, project teams face delayed reviews, repeated information requests, diminished agency confidence, and vulnerability in the administrative record.
This session demonstrates how disciplined data architecture, standardized schemas, spatial consistency, transparent metadata, and automated quality control workflows strengthens regulatory defensibility while accelerating review timelines. Attendees will learn how separating preserved source datasets from reproducible analytical pipelines enhances scientific integrity and auditability, ensuring every conclusion ties back to verifiable evidence.
A licensee case study will illustrate how implementing a structured, cross-disciplinary data framework reduced rework, improved interagency coordination, and shortened review cycles. Participants will leave with practical strategies for building resilient, scalable data systems that support faster decisions and stronger evidentiary records throughout pre-application studies, license development, and post-license compliance.
Advanced AI and Optimization for Flexible Hydropower Cascades in Modern Electricity Markets (Argonne National Laboratory)
Presented by Quentin Ploussard, Argonne National Laboratory
Hydropower operations are increasingly asked to deliver flexibility, grid reliability, and climate resilience, while honoring complex physical, environmental, and regulatory constraints. This classroom presentation introduces a next-generation cascading hydropower optimization model developed under a DOE WPTO-funded project to bridge the gap between physical realism, digital innovation, and operational decision-making.
The model integrates detailed physical representations traditionally simplified in production cost models, including nonlinear storage-to-elevation relationships, explicit water mass balance equations, accurate water-to-power conversion functions, and coupling constraints between hydraulically connected reservoirs. A novel water travel-time distribution framework captures delayed flow propagation across the cascade, enabling realistic inter-reservoir coordination, which is critical for operational flexibility, environmental compliance, and downstream safety considerations.
To preserve fidelity without sacrificing performance, the model combines physics-based structure with data-driven regression components that learn operational relationships directly from empirical plant data. These embedded regression models illustrate a practical artificial intelligence (AI) application in hydropower operations: leveraging plant data to improve forecasting, production planning, and asset utilization. The regression models are integrated into advanced mixed integer linear programming (MILP) and mixed integer quadratic constraint programming (MIQCP) formulations compatible with modern open-source solvers, supporting transparency, digitalization, and long-term modernization of hydro decision-support tools.
A strong mathematical formulation enables efficient solve times without compromising accuracy. For example, an hourly scheduling problem of a three-dam cascade over a one-week horizon solves in approximately five minutes on a standard laptop. This computational performance makes the framework suitable for operational scheduling, water resource forecasting, renewable integration studies (including pairing with wind and solar), and scenario analysis under extreme weather conditions.
The model has been successfully tested on two real-world systems managed by the Bureau of Reclamation: the Aspinall Unit and the North Platte Project. Results demonstrate improved coordination across reservoirs, realistic flow transitions, and tractable solve times without sacrificing operational details.
Key takeaways for attendees:
How to embed high-fidelity hydropower physics into optimization models without exploding complexity
How AI-driven regression components can enhance forecasting and production planning
Practical formulation strategies that dramatically reduce solve time
Lessons learned from applying advanced analytics to real federal hydropower assets
Pathways for digital modernization of hydro operations and improved flexibility in evolving electricity markets
