Mitigation of environmental effects of frequent flow ramping scenarios in a regulated river

Abstract

In the transition to a society based on renewable energy, flexibility is important in balancing the energy supply as more intermittent sources like wind and solar are included in the energy mix. The storage-based hydropower systems are a renewable energy source that provides the needed flexibility since a hydropower plant can be started and stopped in minutes, and the reservoirs provide stored energy that can be utilized when the demand arises. The main objective of this study is to describe a method of designing the necessary volume of water required to mitigate a fast ramping turbine, and present the effect this has on the downstream river reach. We used a 2D hydraulic model to find the areas affected by hydropeaking operation and, furthermore, to define areas with a faster ramping rate than 13 cm/h which is used as a limit in Norwegian guidelines. Based on this, we developed a ramping regime that would prevent fast dewatering of critical areas and provide this as a basis for mitigating the effects of fast dewatering in the downstream river (River Nidelva in Norway was used as a test case). Furthermore, the effect of increasing the frequency of start–stop cycles was studied, and the proposed mitigation was evaluated for the new operational regime.

1. Introduction

The transition into a society based on renewable energy sources has received considerable attention over the past years as a way of combatting global warming (Edenhofer et al., 2011). The renewable energy mix contains intermittent energy sources like the wind and sun, and methods of balancing the energy supply are important to secure a constant supply. Hydropower with water storage in reservoirs is a very flexible renewable energy source which can be started and stopped in minutes and is thereby ideally suited to balance the intermittent energy sources. The reservoirs provide stored energy that can be utilized when the demand arises, which further extends the flexibility and usefulness of the hydropower systems for load balancing purposes. However, the flexible operation of hydropower plants can cause nonnatural, sub-daily fluctuating flows into receiving water bodies, often referred to as "hydropeaking" (Harby et al., 2013). This can be a serious environmental issue, particularly when the hydropower outlet is in a river (Bruder et al., 2016) as opposed to power plant outlets in fjords, reservoirs, or lakes (Halleraker et al., 2022).

Hydropeaking in rivers can cause serious consequences for the physical environment and ecosystems downstream (e.g., Harby et al., 2013). The key factors in the assessment of peaking are the vertical ramping rate (the drop in water level per time unit), and the rate of dewatering (drying) of the river bed when the water level is reduced (the lateral ramping rate). The lateral ramping rate might be as important as the vertical ramping rate in the impact assessment study. Literature covers a wide range of ecological impacts from hydropeaking operations. Auer et al. (2017) reported on the flushing of biota as a turbine starts, and this can lead to changes in fish distribution. When the turbine stops, stranding or pool trapping of juvenile fish in river areas that had dried out was observed (Saltveit et al., 2001; Schmutz et al., 2015). Periods with a dry river bed can also have impacts on salmon eggs and alevins buried in the gravel (Casas-Mulet et al., 2014; Casas-Mulet et al., 2016).

Hydropeaking operations can have severe impacts on fish populations in downstream rivers. Hayes et al. (2021) have shown large reductions in biomass of graylings between reference sites and sites with peak flow regime. Schmutz et al. (2015) have found that typical hydropeaking indices like peak frequency and ramping rate have explained most of the indices of biotic integrity developed for rivers in Austria. Saltveit et al. (2020) have shown that year classes of brown trout older than the young of the year disappear from heavily peaked river reaches in Norway. Through a modeling study using an individual-based model, Hedger et al. (2018) have shown that stranding of smolts may have a large impact on the population through a direct impact on returning adult Atlantic salmon.

Impacts from peaking operations have to be handled to reduce the adverse effects on nature, and mitigation is necessary in maintaining sustainable energy production, particularly when a hydropower plant in the energy mix has an outlet in a river rather than in a fjord or reservoir (Halleraker et al., 2022). Flow rules or operational ramping restrictions, as proposed by Hayes et al. (2019), are examples of mitigation measures related to peaking operations which aim at adapting the peaking hydrograph within safe ecological limits for different life stages of target species. However, designing flow rules is a complex process since the dewatering of river sections is dependent on the distance from the outlet of the hydropower plant and the morphology of the river (Hauer et al., 2017). The relationship between the vertical and lateral ramping rates is dependent on the morphology and shows large spatial variations (Tekle, 2021), which is best estimated through hydraulic modeling.

The challenges related to mitigation are both related to technical turbine constraints and the extra use of water caused by gradual stops. For creating a gradual stop of the power plant, it is necessary to operate the hydropower turbines at the rate of discharge defined in the mitigation protocol. This might not be possible without violating the turbine's technical constraints. For example, overheating could occur at low turbine flow and turbine blades can be damaged by cavitation if operated outside the normal flow range. Juarez et al. (2019) proposed a mitigated turbine stop for the Storåne river to reduce stranding, but here, the turbine constraints made it only possible to run a slow shutdown until the turbine reached 40% of the full capacity before a full stop was required.

In addition, new operational regimes for turbines are possible and also substitution of turbines allows for a more adapted shutdown regime. An alternative to handling the restricted shutdowns directly through the turbine is to employ a dampening reservoir just downstream of the turbine outlet (Meier et al., 2016; Tonolla et al., 2017). The purpose of this is to fill the reservoir during the start of the turbine and thereby dampen the flushing effect and to gradually release water when the turbine stops to reduce the dewatering effect. An alternative solution using water storage for dampening the effect of hydropeaking has been proposed by Storli and Lundström (2019) who presented the Air Cushion Underground Cavern (ACUR), providing underground storage where compressed air is used to control the storage and release of water.

Based on the need for a new and more flexible hydropower turbine as described above, the HydroFlex project was initiated and funded by the European Union. The objective of the HydroFlex project is to develop a variable-speed turbine that can handle high ramping rates and multiple starts and stops per day (possibly up to 30) without compromising the lifetime and efficiency of the turbine and to develop equipment like generators and converters that are required for the new operation. The project also describes the possible environmental impacts of operating the new HydroFlex turbine and evaluates potential mitigation measures.

The objectives of this study can be summarized as follows: (1) Investigate the effects of hydropeaking in River Nidelva in Trondheim, Norway, on stranding areas for juvenile Atlantic salmon using a hydraulic model under the current operational regime. (2) Use the results from (1) to evaluate potential mitigation measures for Nidelva as a basis for an evaluation of the HydroFlex turbine. (3) Investigate the effect of a high-frequency start and stop operation according to the HydroFlex turbine design on conditions in the river reach downstream of the powerplant outlet and how mitigation measures can be applied to this situation.

2. Materials and methods

2.1 Study site

River Nidelva in Norway was used as a study site in the project (Figure 1). The river is regulated through the Nea-Nidelva scheme which consists of the regulation of Nea upstream of Selbusjøen and the regulation of Nidelva downstream of Selbusjøen. This project used the lower part of the river from Nedre Leirfoss to the outlet in the fjord in Trondheim city center. This reach of the river is regulated through the Nye Leirfoss power plant and the Bratsberg power plant. Bratsberg has two Francis turbines with a capacity of 50 m³s⁻¹ each, while Nye Leirfoss has two Francis turbines of 30 m³s⁻¹ and 60 m³s⁻¹, respectively. In addition, there is a minimum flow release of 10 m³s⁻¹ between Øvre and Nedre Leirfoss. Bratsberg is running the hydropeaking operation, while Nye Leirfoss contributes to a total minimum flow of 30 m³s⁻¹ downstream of the hydropower outlet.

2.2 Data collection

Water discharge data for the river were taken from the Rate gauge (123.20.0) some hundred meters downstream of the power plant outlet and from the Bratsberg power plant, both available from the database of the Norwegian Water Resources and Energy Directorate. The frequency of future ramping was built on model simulations of the future energy market developed for the HydroFlex project by the University of Aachen (Siemonsmeier et al., 2020). The bathymetry from the river was compiled from different sources; mainly, a measurement campaign was carried out by the project using a SonTek M9 HydroSurveyor towed by a kayak. Particular emphasis was made to collect detailed data from areas known to have dried out during minimal flow and thereby become a potential stranding area for fish. The collected data were interpolated into a Digital Elevation Model (DEM) using ArcMap 10.8 with a resolution of 0.5 × 0.5 m.

2.3 Hydraulic model and mitigation analysis

To properly capture the details of the dewatering areas, a two-dimensional hydraulic model was used (Vanzo et al., 2016; Juarez et al., 2019). The hydraulic simulations were carried out in HEC-RAS v.6.0 (Brunner 2021). Different grid sizes were tested, and finally, a model grid of 1 × 1 m was used, providing a good compromise between accuracy in estimating the stranding areas and computational speed. The model was calibrated against the observed water levels by adjusting Manning's n-value in river sections, and the model was further evaluated using aerial imagery from the Norwegian mapping authority. Atlantic salmon (Salmo salar) was used as a target species in this study. Data on hydropeaking severity for Atlantic salmon and the necessary ramping rate threshold to avoid fish stranding was based on a limit of 13 cm/h proposed by Bakken et al. (2021). The ramping rate was computed for several locations along the river using the standard hydrograph with particular focus on areas where total dewatering was observed in the field.

Based on the observed turbine operation, several simulations were carried out with input hydrographs that gradually extended the time it takes for the power plant to go from full to zero production. The results were used to correlate the dewatering rate with the stoppage time to find a shutdown scenario that reduced the dewatering rate below the acceptable threshold. Using the findings, an environmentally friendly shutdown hydrograph was made by extending the time of the shutdown and creating a recession with a milder slope. Extending the time of the stop requires more water to be released from the power plant, and the extra volume of water required was computed from the mitigated hydrograph (Kenawi 2021). In addition, to find a flow regime that prevents critical dewatering, we have also tried to conserve the use of water in the mitigation procedure to reduce the impact on power production.

2.4 Analysis tools

All plotting and analysis of the spatial extent of peaking were carried out in ArcMap 10.8 and ArcGIS Pro, and all other statistical analyses and plotting were done using the R programming language (R Core Team 2019).

3. Results

3.1 Model calibration

The comparison of the measured and simulated water levels was done for 42 points measured between Nedre Leirfoss and Tempe. The Manning numbers of the model were adjusted to minimize the differences between the observed and simulated water levels, leading to a mean error of −0.017 cm (SD = 0.08 cm). The Manning number ranged from 0.045 to 0.090. The correlation between the observed and simulated water levels gave an r² of 0.99. The difference between the simulated and the observed water-covered area is 3%.

3.2 Modeled dewatering

An observed hydrograph measured in the Bratsberg power plant at the upstream boundary is shown together with the standardized hydrograph derived from the measured data and used in the development of the mitigation measures. The environmental flow of 30 m³s⁻¹ was added to the turbine flow. The ramping simulations with multiple peaks per day show the translation of peaks downstream after each stop in Leirfossen. The duration between the falling and rising limbs of the hydrograph is shortened, and eventually, the rise of the next peak starts before the previous peak has reached its potential minimum, thereby reducing the dewatered areas in the reach.

3.3 Mitigation measures

The simulations identified the potential dewatered areas and provided an understanding of the dewatering rates in different locations along the river. New simulations were carried out by gradually increasing the stoppage time of the turbine. Changes in depth were extracted from the two most severe stranding locations in the river. Table 1 shows that for all regions, the dewatering rate is above the acceptable threshold of 13 cm/h (Bakken et al., 2021). To estimate the necessary level of mitigation, simulations were carried out for different shutdown times for the turbine (range 0–380 min), and the dewatering rates were computed for each case. For the longest shutdown time of 380 min, the rates were acceptable for all regions. Using the relationships between dewatering rates, duration of turbine stop, and the simulations with the hydraulic model, a time of 270 min was found to be acceptable. Based on this, the extra water required to obtain the gradual recession amounts to 0.61 mill m³. Using the proposed mitigation method, simulations were done with the original and mitigated hydrographs to evaluate the effects.

4. Discussion

As a part of the investigations on the feasibility of developing a hydropower turbine capable of multiple starts and stops in a day to provide increased flexibility in an energy market with an increased level of intermittent sources, the environmental impacts and potential mitigation were evaluated. The 2D hydraulic model HEC-RAS was used to simulate the effect of various peaking hydrographs, and an operational strategy that satisfied the 13 cm/h dewatering rate from the guidelines (Bakken et al., 2021) was created and used to evaluate the potential for mitigation of the adverse effects of peaking operation.

A detailed assessment of stranding areas and dewatering rates at a small spatial scale requires fine-scale hydraulic modeling, and it is therefore important to have a detailed digital bathymetric model of the terrain and a well-calibrated model. In this project, a special emphasis was made on measuring the potential stranding areas in detail to generate the DEM using sonar and GPS systems. The evaluation of the model against measurement shows good accuracy, and this was further controlled by comparison with orthophotos from a drone survey of the stranding site. When compared to other studies (Hauer et al., 2014; Sauterleute et al., 2016; Juarez et al., 2019), the dried-out areas in Nidelva are moderate, with around 9% of the total wetted area of the river at a discharge of 130 m³s⁻¹ dry at the environmental flow at 30 m³s⁻¹. This is explained by long reaches with steep bank slopes and embankments, and a relatively high environmental flow which also contributes to reduced stranding areas.

From the analysis of the shutdown procedure, a stop time of 270 min was found to reduce the dewatering rate to a level less than the limit for stranding for each turbine stop. The extended stop time is most likely not technically feasible for the turbines currently in operation in Bratsberg, and probably also not economically feasible for the power plant due to the extra water required. The economic cost of this has not been estimated in this project, but the amount of water required per peak for mitigation is significantly larger than the value reported by Juarez et al. (2019). It is worth noting that the extra water released both in this study is released through the turbine and will generate income.

It is not yet clear if the turbine setup from HydroFlex will have the potential to release water as proposed in the mitigation hydrograph due to possible operational restrictions on low flow releases. A potential mitigation strategy could be to configure the power plant such that the proposed extended stop could be handled by the turbines. Retrofitting a dampening reservoir could be a solution to the need for extra water, but this is not a possible solution at the outlet in Nedre Leirfoss due to the terrain and infrastructure. Given the amount of water required for mitigation, Saberi et al. (2021) show that the ACUR system could provide the necessary storage and thereby provide the dampening of the peaking operation. Structural solutions could also be an option as a mitigation measure, such as installing a weir at the side channel inlet or constructing a channel that leads water into the side channels at minimal flow.

In the scenario shown, the frequency and duration of the peaks are such that the river will reach a minimum flow state between each peak. If the number of sub-daily peaks is further increased toward the goal of 30 as specified in the HydroFlex project, we will have periods where the next peak starts before the previous peak has reached the minimum flow level. This will thereby reduce the peak amplitude and reduce the dewatering effect downstream. This is even more the case if we observe the mitigated hydrographs. Similarly, it is observed that frequent sub-daily ramping dampens the dewatering effect far downstream in the river as the river starts rising again before the previous stop has reached the minimum level.

5. Conclusion

Through the detailed hydraulic modeling of frequent hydropeaking, a potential mitigation strategy is devised. The strategy involves increasing the duration of the stop of the power plant to ensure that dewatering rates are below the critical level of 13 cm/h. This involves releasing an extra 0.61 million m³ at each shutdown and requires a turbine that allows the slow shutdown speed. This can be a challenge both for the water budget and economy of the hydropower plant and for the operation of the turbine. A potential solution to this could be to use a dampening reservoir built in the mountains based on the ACUR technology that could store the required water during the start of the turbine and then release this water into the river when the turbine stops. When the frequency of the start and stops increases, an overlap of waves is seen in the river downstream which reduces the dried-out area for each start–stop cycle.

Data availability statement

The data sets presented in this article are not readily available because some data are collected by third parties and made available to the project, and these cannot be released without the consent from the data collectors. The data collected by the project can be had from the corresponding author on request at knut.alfredsen@ntnu.no.

Author contributions

KA supervised the work and was the main author of the study; KA and AG designed the study; AG collected the data and prepared and calibrated the hydraulic model; MG prepared and ran the simulations for the HydroFlex turbine with multiple start and stops in 1 day; MK and SS carried out the simulation of the mitigation measures, designed the mitigation flow regime, and ran the simulations to evaluate its impact. All authors have contributed to the text and have read and edited the final manuscript.

References

Full reference list (over 50 sources) available in the original PDF.