Blog
ORNL and INL find 4 GW at 2,600+ non-powered U.S. dams
Estimated reading time: 5 minutes · Last updated:
Oak Ridge National Laboratory and Idaho National Laboratory jointly identify 4 gigawatts of potential hydropower capacity at more than 2,600 non-powered dams across the contiguous United States. Using INL’s HydroGenerate tool together with USGS stream gage records and ORNL’s Dayflow reanalysis, the teams estimated those sites could produce about 15.2 terawatt-hours per year—roughly enough to serve 1.4 million homes. The assessment also reports an average candidate size of 1.5 megawatts and notes that only about 3 percent of the nation’s roughly 91,000 dams currently generate electricity, which is why a retrofit-focused approach changes the development calculus.
This refined data will help decision makers focus their efforts on the sites with the greatest possible capacity.
ORNL’s Carly Hansen
Key takeaways
- ORNL and INL identified 4 gigawatts of hydropower potential at more than 2,600 non-powered dams, capable of about 15.2 TWh per year.
- The average candidate facility in the assessment is about 1.5 megawatts and the resource would power roughly 1.4 million homes.
- Federally owned dams account for 86 percent of the identified capacity, concentrating much of the opportunity in the Upper Mississippi and Great Lakes regions.
Table of contents
What the ORNL–INL assessment found and why it matters
The joint ORNL and INL analysis selects more than 2,600 existing dams that have never been fitted with turbines or generators and estimates those sites could host 4 gigawatts of installed capacity in aggregate. That 4 gigawatts is not a theoretical maximum; the teams modelled practical plant sizes and daily output and report a projected annual yield of about 15.2 terawatt-hours. Converted to household demand, the authors state that output would be enough to supply roughly 1.4 million homes.
Those headline numbers matter because they come from retrofits rather than new dam construction: the dams already exist, so developers skip site construction and many early-stage costs. The assessment also reports an average candidate size of 1.5 megawatts, which highlights that most opportunities are small plants aggregated across many sites rather than a few large projects.
How the teams produced more realistic site estimates
The researchers combined INL’s HydroGenerate computational model with long-term hydrology to move past earlier, broader estimates. HydroGenerate calculates a design flow and factors such as hydraulic head and turbine efficiency curves to size turbines that could operate under real conditions rather than idealized assumptions.
ORNL’s Dayflow reanalysis provided consistent streamflow estimates across river networks where direct measurements are sparse, and the teams used daily USGS stream gage records where available. Crucially, the modelling built in operational constraints—schedules for flood control, navigation and other non-power uses—so the output reflects what is likely deliverable under existing water-management regimes. Developers can obtain HydroGenerate as an open-source package and run the same site-level feasibility checks themselves.
Where the resource sits and the role of ownership
Not all candidate sites are equal. The assessment finds federally owned dams account for 86 percent of the total identified capacity, a concentration that influences who makes decisions and how quickly projects can progress. The Upper Mississippi River and Great Lakes regions stand out in the dataset because they host a cluster of larger-capacity candidates.
Smaller, locally owned dams also appear in the inventory and may offer faster, lower-complexity retrofit paths for utilities or municipal owners. But the federal concentration means many of the highest-capacity opportunities will follow federal operational rules and multi-agency permitting pathways, which affects timelines and the set of stakeholders that must be engaged.
Tools, timelines and the barriers to deployment
The research team published the candidate list and supporting datasets through ORNL’s HydroSource site and built NPD Hydro, a platform with NPD Insights and an NPD Toolkit intended to help owners and planners rank sites and estimate payback and impacts. Those tools move an inventory into actionable, site-level analysis but do not shorten long regulatory or engineering lead times.
Permitting, design and construction for hydropower retrofits remain multi-year to multi-decade processes, and the modelling highlights a central constraint: water availability is controlled by competing uses such as flood control and navigation, which can sharply reduce how much energy a retrofit can actually produce. The teams also flag climate-driven streamflow changes as a future uncertainty that must be folded into long-term feasibility studies. The project is supported by DOE’s Hydropower and Hydrokinetic Office within the Office of Critical Minerals and Energy Innovation.
| Measure | Previous national estimates | ORNL–INL assessment |
|---|---|---|
| Total non-powered dam capacity (GW) | 12–30 GW | 4 GW (practical, constraint-aware) |
| Annual generation basis | Theoretical maxima | 15.2 TWh/year (operational modelling) |
| Data and tools | Limited datasets, ad hoc methods | HydroGenerate + Dayflow + USGS gages; open-source and site tools |
Cases for and against delivery
The case for
- Existing civil structures lower upfront capital and land costs because turbines and electrical works are added to dams already in place.
- Open-source HydroGenerate and the NPD Hydro platform give owners standardised, site-level feasibility tools that can shorten initial scoping and align technical assumptions across stakeholders.
The case against
- Operational constraints such as flood control and navigation can materially reduce available water for generation, limiting practical output at many sites.
- Most identified capacity sits on federally owned dams, which introduces multi-agency processes and long permitting timelines that can delay or block projects.
What to be careful about
- Regulatory and interagency clearance required at federally owned dams could add years or decades to project timelines.
- Hydrological constraints imposed by flood control and navigation can reduce the usable generation compared with design estimates.
- Future streamflow shifts from climate variability could lower actual output in some regions, changing project economics.
The bottom line
The ORNL–INL assessment reframes non-powered dams as a pragmatic source of new hydropower by grounding capacity estimates in operational data and constraints. Four gigawatts and 15.2 terawatt-hours per year are modest compared with some prior theoretical ranges, but they represent a deliverable resource that can be analysed at scale using HydroGenerate, Dayflow and the NPD Hydro tools. Realising that potential will hinge on federal agency processes, water-management rules and the incorporation of future streamflow changes into long-range planning.
What to watch
- Watch for the teams’ planned extension of the assessment to Alaska and Hawaii; no date has been set.
- Watch for HydroGenerate updates that incorporate future streamflow projections and new operational-constraint modules; no date has been set.
- Watch for updates to ORNL’s HydroSource and the NPD Hydro platform listing finalized candidate sites and detailed site data; no date has been set.
Frequently asked questions
What exactly did ORNL and INL measure?
They identified more than 2,600 non-powered dams and modelled realistic turbine sizes and daily output using HydroGenerate and ORNL’s Dayflow; the study reports an aggregate potential of 4 gigawatts and about 15.2 TWh per year.
How many homes could that output serve?
The teams estimate roughly 1.4 million homes could be supplied by the projected 15.2 TWh of annual generation from the 4 GW of candidate capacity.
Who owns the sites with the largest potential?
The assessment finds federally owned dams account for 86 percent of the identified capacity, concentrating much of the opportunity in regions such as the Upper Mississippi and the Great Lakes.
Related reading