How Deregulated Markets Influence Hydro Revenue and Downstream Flow

Dams that impound water for hydroelectric facilities provide low-cost renewable electricity and are more flexible than thermal generation but can disrupt the “natural” flow regime of rivers. The authors explore the tradeoffs between hydropower revenue and ecosystem quality downstream from the dam.

By Jordan D. Kern

Efforts to reduce pollution and lower greenhouse gas emissions will benefit from increasing renewable energy production. Hydropower provides nearly two-thirds of total renewable electricity generation in the U.S., and these facilities can respond to short-term changes in electricity supply and demand (including fluctuations caused by intermittent supply of wind and solar energy) more rapidly and less expensively than thermal generators.1 However, concerns about the environmental impacts of hydroelectric facilities have motivated efforts to restore rivers.2,3 Consequently, it is critical to understand the tradeoffs between hydropower generation and ecosystem quality, as well as how this knowledge can be used to develop more sustainable natural resource management strategies.

Hydro facilities can have negative impacts on riparian vegetation,4 downstream geomorphology,5 and macroinvertebrate and fish communities.6 Often these impacts, which can include decreased biological diversity and abundance, occur as a result of disruptions to the natural flow regime of rivers.7,8

In this study, we explore the tradeoff between hydro revenues and downstream ecosystem quality and focus in particular on how participation in a deregulated electricity market impacts a utility’s revenue stream and downstream flow regimes. We compare six operating scenarios alongside unregulated flows (dam removal) using operations at a three-dam cascade in the Roanoke River basin. Flow regime statistics that reflect environmentally critical components of river flow are used to assess relative differences in downstream flows. These are then compared with the relative differences in revenues resulting under each.

Deregulated markets

The electricity market considered is the Dominion Zone of PJM Interconnection in the mid-Atlantic region of the U.S. In PJM, electricity is bought and sold in the “day-ahead” and “real-time” markets. In the day-ahead market, the variable cost of the last sell offer required to satisfy day-ahead forecasted demand generally sets the market-clearing price. The real-time market is used to schedule “load following” electricity to compensate for demand forecast error and unexpected changes, with prices determined in a manner similar to the day-ahead market.9 PJM also coordinates market-based pricing of three types of ancillary services. Frequency regulation, which corrects for short-term changes in electricity use that may cause the power system to operate above or below the 60 Hz standard, is the only one considered in this study. In the regulation service market, sellers agree to “ramp” generation by an incremental amount specified in their sell offers. They receive a market price for this service, but electricity produced from regulation service is sold in the real-time market.

Hourly price fluctuation is an important characteristic of deregulated markets. In general, there is a relatively low level of price volatility in the day-ahead market. But price volatility is inevitable (particularly in the real-time market) due to short-term changes in supply and demand.


The relationship between hydro revenues and downstream flow is explored under six “regulated” operational scenarios:

– Full-market participation (including real-time and regulation), with and without flow reregulation;
– Day-ahead electricity only; and
– Run-of-river, with and without flood control and flow reregulation.

For each scenario, the rules regarding how to allocate water over a multi-day operating horizon are dictated by the markets in which the facilities participate. “Day-ahead only” participation means hourly generation is scheduled according to hourly, day-ahead forecasted demand in PJM. Each hour of the operating horizon is ranked in terms of its forecasted demand. Given the cumulative discharge over the operating horizon, this volume of water is allocated one hour at a time in order of decreasing forecasted demand.

In “full-market participation,” dam operators schedule generation to take advantage of peak forecasted demand in the day-ahead electricity market and high real-time electricity prices. In addition, each hydroelectric plant makes a static hourly regulation sell offer of +/-10 MWh.

Two “run-of-river” strategies (with and without flood control) were designed to serve as a lower bound on dam operations in terms of environmental impacts. Without flood control, downstream flows are expected to more closely mimic unregulated flows by enforcing daily adherence to a “flat” reservoir guide curve (outflows equal inflows). An additional run-of-river scenario was developed, in which the ability to force outflows equal to inflows is limited by a requirement that reservoir volume be maintained at a level consistent with flood control objectives. These scenarios involve no participation in the real-time or regulation service markets.

When there are multiple dams upstream from sensitive ecosystems, there is interest in exploring how regulating flow at only the furthest downstream dam might impact flow regime and hydropower revenues. To test this, upstream reservoir levels are allowed to fluctuate but the downstream dam maintains a daily outflow proportional to inflows at the furthest upstream dam, and hourly flows are kept relatively constant (no peaking). Thus, flow re-regulation is added to the run-of-river scenario without flood control to form a presumed “lowest-impact” scenario; it is likewise added to full market participation to mitigate that strategy’s negative impacts on downstream flow regime.

Indicators of hydrologic alteration

Indicators of hydrologic alteration (IHA) are used to describe the flow regime resulting from each scenario. These are 32 flow statistics, subdivided into five categories – magnitude, duration, timing, frequency and rate of change – that quantify the degree of hydrologic alteration resulting from a disturbance in a river basin, such as changes in land use or a dam.10 Connecting changes in flow regime to specific biological and ecological endpoints is an ongoing challenge;6 this work focuses on evaluating changes in these IHAs relative to unregulated conditions and using these differences as a surrogate for environmental impacts. We use a process called principal components analysis (PCA) to yield a subset of the (statistically speaking) most important IHAs. PCA was performed using 82 years of simulated unregulated flow data (1929 to 2010) to select seven IHAs that collectively explain 84.5% of the statistical variation in unregulated flows.

Case study: Roanoke River basin

The Roanoke River basin flows southeast from Virginia to North Carolina, and the lower basin includes three hydro facilities:

206-MW John H. Kerr, owned by the U.S. Army Corps of Engineers, completed in 1953, with a turbine flow capacity of 35,000 cubic feet per second (cfs);
224-MW Gaston about 30 miles downstream (turbine capacity 44,000 cfs), built in 1963, owned by Dominion; and
– 104-MW Roanoke Rapids about 8 miles downstream (turbine capacity 20,000 cfs), built in 1955, owned by Dominion.

Due to constraints on reservoir level fluctuations at Gaston and Roanoke Rapids dams, the timing and magnitude of releases at Kerr Dam largely dictate the schedule of releases at the two downstream dams. There is little free-flowing river between Kerr Dam and the Gaston Reservoir and essentially no free-flowing river between Gaston Dam and Roanoke Rapids Reservoir. Thus operations at all three dams are modeled synchronously, with attention to environmental impacts focused on the long stretch of free-flowing river downstream of Roanoke Rapids.

This area includes one of the largest and least-fragmented river swamp forest ecosystems in the eastern U.S.11 This area and its floodplain have been identified as critical resources for the conservation of bottomland hardwoods and other riparian and in-stream biota.12 Numerous studies have explored the range of impacts river flow regulation (i.e. dams) has on the ecosystems of the Lower Roanoke River basin,12,13 but these studies pre-date the advent of deregulated electricity markets and their potential impacts on flows.

Current operations in the basin

The water available for release from Kerr Dam during any given week is established by the Corps and based on current storage, recent (and predicted) inflows and maintenance of flood storage capacity. Until 2005, the timing of hourly releases at Kerr Dam was also determined by the Corps, which coordinated generation with periods of high electricity demand for federal power customers. However, since May 2005 the scheduling of hourly releases at Kerr Dam has been largely determined by Dominion, which relays release requests to Kerr Dam.14 This change has been concurrent with Dominion’s active participation in PJM Interconnection.

To characterize Dominion’s current operations, generation release schedules were simulated under various combinations of operating horizon and level of market participation. Observational data (PJM market prices and day-ahead demand forecasts) were limited to the five years in which these hydro plants have been part of PJM. Comparison of simulated results with historical observations (see Figure 1 on page 34) strongly suggests that over the period 2006 to 2010 these hydro plants largely participated in the day-ahead market only, with the best model fit occurring with use of a four-day operating horizon. These two parameters are therefore used to characterize the power generation behavior under current operations. For the remainder of this article, Dominion’s current operations are considered synonymous with “day-ahead only.”

Day ahead vs observed hourly dam releases

Results and discussion

Figure 2 on page 38 shows differences between the simulated hourly operations under the day-ahead (left) and full-market (right) scenarios. Hourly dam operations under the full-market scenario reflect frequent participation in the real-time energy and regulation markets, which alters the hourly dam release schedule and translates to differences in revenues and downstream flow regime. In general, however, both scenarios reflect similar patterns of hourly operation due to season-specific variation in electricity demand.

modeled hourly market participation for roanoke rapids dam

The full-market participation scenario results in the highest revenues, followed by full market with reregulation, day-ahead only and then run-of-river scenarios. The cumulative difference in revenue between the three scenarios is primarily a function of the run-of-river scenarios’ guide curve storage constraints and one-day operating horizons. These constraints, respectively, result in significantly more spilling, more frequent generation during periods of relatively low electricity demand (and price), and access to fewer markets.

The difference in cumulative revenue between the day-ahead only and full-market scenarios primarily reflects the differences in market participation. Over five years, the day-ahead only scenario generates $346.1 million, while the full-market scenario generates $378.8 million selling energy in the day-ahead ($306.1 million), real-time ($66 million), and regulation service ($6.7 million) markets.

Results show that over the five years:

1) Strict daily adherence to a seasonal Kerr Dam guide curve results in $54 million and $86 million less in hydropower revenues than the day-ahead only and full-market scenarios, respectively; and

2) Removing flood control capacity at Kerr Dam increases these losses by $5 million. The reduction in revenues as a result of flow reregulation at Roanoke Rapids Dam is roughly $8 million under run-of-river conditions and $18 million under full market participation.

Figure 3 plots annual hydropower revenues against deviation from the unregulated flow regime for four operational scenarios. Deviation is defined as the root mean squared difference between regulated and unregulated flows at Roanoke Rapids Dam. Movement away from the origin along the y-axis signifies increased revenue and along the x-axis signifies increasing divergence from unregulated flow behavior.

hydropower revenue vs deviation from unregulated flows

An overall positive relationship between hydropower revenues and deviation from unregulated flows implies a tradeoff between downstream environmental quality and hydropower revenue. This tradeoff is present in results for one-day maximum flow data (all years), January mean flow (2006 to 2007, 2009 to 2010) and September mean flow (2009).

Results for most IHAs are somewhat mixed but overall positive, such as data for fall rate (2006 to 2009) and September mean flow (2006 and 2008), where day-ahead only leads to greater deviation than full-market participation but produces less revenue. Other IHA metrics, such as August and September mean flows (2006), yield a negative relationship, where greater revenue coincides with more “natural” flows. March mean flow presents an example of a negative relationship between revenue and deviation from unregulated flows under flood control conditions; then, as flood control capacity is removed, flows revert to a more natural state.

Annual results for many of the IHAs considered show a high level of inter-annual variability. Only two IHAs (March mean flow and one-day maximum flow date) demonstrate a consistent trend between revenue and deviation from unregulated flows over five years.

In general, IHA values for the day-ahead only and full-market participation scenarios are relatively similar, even as the full-market scenario generates more revenue in each of the five years; larger differences are observed when the full-market and day-ahead scenarios are compared with the run-of-river or full market reregulated scenarios.

Quantifying a particular scenario’s aggregate impact on flow regime is complicated by the use of IHAs with various units, so calculations of the percentage deviation from unregulated flows are used to gauge the impact of each regulated scenario. These calculated values correspond to 210 data points (six scenarios x five years x seven IHAs), which were found to generally follow a normal distribution. Grubbs’ test was used to identify and remove two statistical outliers (2008 fall rate data for the day-ahead and full-market scenarios). Figure 4 on page 44 shows results for full-market participation, day-ahead only, and run-of-river reregulated (without flood control). Each plot shows standardized revenue on the y-axis and percentage deviation from unregulated flows on the x-axis. The unregulated scenario is represented by the origin.

standardized hydropower revenue vs deviation from unregulated flows

The full-market participation and day-ahead only scenarios show a high degree of similarity in flow regime, as evidenced by the proximity of their respective IHA markers along the x-axis. Thus, changes in operations designed to take advantage of market deregulation appear to have a relatively small impact on flow regime. On the other hand, the difference in revenue between these two scenarios ranges from 4% to 11% annually (summing to $32.7 million over five years).


The general trend between hydropower revenue and deviation from unregulated flows over the five years considered appears positive (i.e. higher revenues correspond to greater deviation from the natural flow regime), albeit somewhat variable across the IHAs used and dependent on the year. This finding primarily reflects the similar nature of flows resulting from the full-market participation and day-ahead only scenarios, as well as the ability of the run-of-river scenarios to somewhat reduce deviation from unregulated flows. When comparing the day-ahead and full-market scenarios, the scale of the differences in flow regime is relatively small, while the added revenue-generating potential of the latter appears significant. Implementing a run-of-river policy would likely result in flow regimes that mimic unregulated flows somewhat better than the other scenarios in most years but yield significantly smaller generating revenues.

Interpreting the results of this study depends on the ability of the IHAs to measure hydrologic disturbance in a biologically and ecologically significant way. Use of IHAs in previous studies has established them as an accepted way to quantify hydrologic alteration after a disturbance. However, use of these metrics stipulates a maximum temporal resolution of one day, so IHAs do not fully capture the potential for impacts on downstream flow regime. Deregulated markets can significantly change hydropower operations on an hourly or sub-hourly basis, so one would expect the effects of these sorts of markets on flow regime to be more evident when analyzed on an hourly time step. This highlights the need for research into the effects of flow variations at smaller time resolutions, which would facilitate further investigation of the effects of market dynamics and strategies that aim to reduce deviation from natural flows.

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.


The author thanks funding sources: North Carolina Department of Environment and Natural Resources, Progress Energy, and Hydropower Research Foundation.


1Electricity from Renewable Resources: Status, Prospects, and Impediments, The National Academies Press, Washington, D.C., 2010.

2Babbit, Bruce, “What Goes Up, May Come Down,” Bioscience,Volume 52, No. 8, 2002, pages 656- 658.

3Hart, P. and L. Poff, “How Dams Vary and Why it Matters for the Emerging Science of Dam Removal,” Bioscience, Volume 52, No. 8, 2002, page 659.

4Townsend, P., “Relationships between Vegetation Patterns and Hydroperiod on the Roanoke River Floodplain, North Carolina,” Plant Ecology, Volume 156, No. 1. 2001, pages 43-58.

5Shields, F.D., et al., “The Stream Channel Incision Syndrome and Water Quality,” Ecological Engineering, Volume 36, No. 1, January 2010, pages 78-90.

6Poff N.L., and J.K.H. Zimmerman, “Ecological Responses to Altered Flow Regimes: A Literature Review to Inform Environmental Flows Science and Management,” Freshwater Biology, Volume 55, 2010, pages 194-20.

7Matthews, R., and B. Richter, “Application of the Indicators of Hydrologic Alteration Software in Environmental Flow Setting,” Journal of the American Water Resources Association, Volume 43, No. 6, 2007, pages 1400-1413.

8Vogel, R. et al., “Relations among Storage, Yield and Instream Flow,” Water Resources Research, Volume 43, 2007.

9Lambert, J., Creating Competitive Power Markets: The PJM Model, PennWell Corporation, Tulsa, Okla., 2001.

10Richter, B. et al., “A Method for Assessing Hydrologic Alteration within Ecosystems,” Conservation Biology, Volume 10, No. 4. 1996, pages 1163-1174.

11Lynch, J.M., “Roanoke River Preserve Design Project,” North Carolina Natural Heritage Program and The Nature Conservancy, Durham, N.C., 1981.

12Pearsall, S. et al., “Adaptive Management of Flows in the Lower Roanoke River, North Carolina, USA,” Environmental Management, Volume 35, No. 4., 2005, pages 353-367.

13Graham, B., and J. Cannon, “Effects of Load Following at Roanoke Rapids Power Station on Lower Roanoke River Main Stem and Floodplain Quality, September 18-29, 2000,” Dominion, Glen Allen, Va., 2000.

14Whisnant, R., et al., “Operating Policies and Administrative Discretion at the John H. Kerr Projects,” 2009.


Kern, Jordan D., et al., “Influence of Deregulated Electricity Markets on Hydropower Generation and Downstream Flow Regime,” Journal of Water Resources Planning and Management, Volume 138, No. 4, pages 342-355.

Jordan Kern is a PhD student in the Department of Environmental Sciences and Engineering at the University of North Carolina at Chapel Hill.

Previous articleKeokuk Energy Center: Harnessing the Power of the Mississippi
Next articleTransitions

No posts to display