An assessment of run of river hydropower potential in Great Britain

Hydropower is the most used renewable energy type globally, with 1330 GW installed as of 2020 – approximatively 16% of the total renewable electricity worldwide in 2019 (IHA, 2022a). In the UK, hydroelectricity produces 1.65 GW of energy (BHA, 2022a), which accounts for approximately 2% of the national capacity (BEIS, 2022). However, hydropower is an ideal complement to variable renewables like wind and solar due to its flexibility and energy storage services. Hydropower can also provide ancillary services, such as dispatchability, energy storage, frequency management, grid inertia and black start ability of other renewable types, providing additional energy security to the UK's power grid (IHA, 2022b). For example, pumped hydro storage like the Coire Glas scheme (SSER, 2023) could smooth the transition from oil, gas and coal to intermittent sources of energy like wind and solar.

The UK has committed to decarbonising its electricity system by 2035 (BEIS, 2021). The Department for Energy Security and Net Zero accepts that hydropower would be critical to delivering greater energy security and independence, economic growth and the UK's net zero ambitions (DESNZ, 2023). However, the Climate Change Committee recognises that the current pace of delivery and deployment of renewable generation and infrastructure may not be enough to meet the 2035 target (CCC, 2023). The committee recommends easing planning and regulatory regimes so that energy infrastructure can be built at the necessary speed. Large hydropower or pumped hydro storage schemes, like Coire Glas, take a long time (5–6 years) to build, mainly because of the complicated and extensive civil works (BBC, 2023). Additionally, large hydropower schemes usually require flooding of a large area in order to accumulate massive volumes of water behind a dam (Lenzen, 2010), which can cause negative environmental impacts (Agostinho et al., 2002; Bobat, 2015; Chaudhari and Pokhrel, 2022). Furthermore, most of the large hydropower developments in the UK have already been constructed (Sample et al., 2015).

Another type of hydropower that has a lesser environmental impact and a shorter construction time than large hydropower or pumped storage hydro is small hydropower. The environmental impacts of small hydro are less than those of large hydropower sites. Common impacts include water depletion, water quality degradation, loss of connectivity, habitat degradation and changes to biota communities (Kuriqi et al., 2021). The greatest impact is the discontinuation of flow – if the minimal flow is not assured or there are cascading small hydropower plants, downstream of the scheme can be affected, so it is necessary to assure a minimal flow downstream (Pang et al., 2015), also known as environmental flow. Kuriqi et al. (2019) noted the pronounced reliance of small hydropower generation on prescribed environmental flow releases.

Small hydropower is usually considered to be run of river (RoR) schemes, which divert water from a river to a turbine and then release it back into the river downstream. RoR systems work in the same way as large hydropower schemes, except they lack the ability to store water and are therefore dependent on the natural seasonality of flow (Mosier et al., 2016). RoR systems can be classified by installed capacity as pico (<5 kW), micro (5–100 kW), mini (between 100 kW and 1 MW) and small (1–5 MW) (BHA, 2022b). The benefits of RoR systems include smaller initial investments, shorter planning and construction times, less inundated area (if any) compared with large hydropower and the use of local labour and materials (Rojanamon et al., 2009). Considering its advantages over large hydropower and the Climate Change Committee's recommendation for fast renewables development, RoR hydropower could be developed in the UK to meet the net zero target.

However, selecting the location and assessing the feasibility of a RoR scheme is a complex task. Historically, the lack of direct observations of heads and flows meant that obtaining a realistic assessment of hydropower potential was challenging, especially at the national scale (Pasha et al., 2014). However, recent developments in the availability of geospatial and remote sensing data (e.g. Copernicus, 2016) could enable high-resolution hydropower resource assessments that have previously not been possible. Such assessments offer the possibility of increasingly complex hydropower analyses, incorporating assessments of financial feasibility, environmental impact, sustainability and social impacts.

Hydropower potential at a global scale has been assessed using different methods and different types of potential (gross, technical, economical, exploitable etc.). Hoes et al. (2017) and Zhou et al. (2015) used a systematic approach to determine the gross hydropower potential across the world. Hoes et al. (2017) calculated the potential per capita in the UK at less than 3300 kWh/year. Zhou et al. (2015) determined the global hydropower potential using stream flows from 1970–2000 and assessed the theoretical, technical, economic and exploitable potential. Although Zhou et al. (2015) presented a complete methodology, some of their assumptions and limitations include consideration of the use of only modern equipment for technical and economic assessments (which may not be realistic worldwide) and the inclusion of smaller hydropower sites (which may not be feasible), leading to a higher hydropower potential compared with other studies. Zhou et al. (2015) estimated that the UK has an exploitable potential of 9 TWh/year (∼138 kWh/year per person). While these global-scale studies cover a large surface, their results are less relevant at country or sub-regional level because their inputs are usually at a coarse resolution and do not capture local stream flows and terrain elevations.

On a smaller scale, other researchers (e.g. Garegnani et al., 2018; Pasha et al., 2014; Sammartano et al., 2019) have used geographical information systems (GISs), merit-based matrices and other tools to assess hydropower potential in several regions of the world. Merit-based approaches (Pasha et al., 2014) consider flow, head and flooded area with different weightings to determine the best locations for RoR hydropower. Sammartano et al. (2019), Garegnani et al. (2018) and Zaidi and Khan (2018) developed a GIS-based procedure to determine potential locations for RoR schemes in various locations (the Taw at Umberleigh river basin in south-west England, the Gesso and Vermenagna valleys in the Alps and Kunhar River in Pakistan, respectively) showing the versatility of GIS-based approaches. However, the selection algorithm employed by Zaidi and Khan (2018) was not rooted in an optimisation framework, which may limit the accuracy of the estimated maximum hydropower potential. Although detailed methodologies were presented in these studies, they were tailored for small regions and catchments and did not address the context of country-wide assessments.

There have been few studies in the last 35 years that have estimated the RoR hydropower capacity of the UK for the remaining financially and technically feasible water resource. The earliest – and still, to date, the only known UK-wide study – was that undertaken by Salford Civil Engineering Limited (SCE, 1989). This study estimated the RoR hydropower potential in the UK to be 322.3 MW in the 25 kW to 5 MW range. A general approach of regression equations, evapotranspiration maps and annual average rainfall estimates was used to calculate the mean annual flow for England and Wales. For Scotland and Northern Ireland, a desk study approach was used. While providing an indicative RoR estimate for the UK, this approach meant the results have not been validated.

The Department of Energy & Climate Change (DECC, 2010) updated the SCE (1989) study by considering current (for 2010) financial and technical constraints for England and Wales. The study identifying 1692 sites with a RoR potential of between 146 280 and 248 400 kW, which is a notable increase from the 36.3 MW reported in 1989 (SCE, 1989). In a separate study, Nick Forrest Associates (2008) used a GIS approach to interrogate flow and elevation data at points along a river network in Scotland using a digital elevation model (DEM). The approached analysed a range of factors that affected energy production and found the most sensitive factors to be the discount rate and electricity sales revenue. Nick Forrest Associates (2008) identified 36 252 technically feasible sites with a RoR hydropower potential of 2.6 GW. From these, 1019 schemes (657 MW) were also deemed to be financially feasible considering an 8% discount rate. However, key details such as the development of flow duration curves (which, from each gauging station were assumed to have been scaled based on catchment area and to provide estimates of flow at points on the river network) limits the usability of this study. More recently, Duncan (2012) identified financially viable schemes across Scotland considering the feed-in-tariff and different discount rates. However, the scope and usability of this study are now limited as it only considered schemes that were financially viable without calculating the hydrological, technical and realisable potential; furthermore a feed-in-tariff that is no longer used (Ofgem, 2019) was considered.

The limitations and ages of previous studies, the renewed interest in renewables development in the UK and the lack of a UK-wide study using the latest available data indicate the need for an updated assessment of the RoR hydropower potential in the UK. Integrating the various approaches from previous studies into a single, consistent methodology can be used to support the identification of a sensible hydropower potential. In this study, a methodological framework was created to determine the RoR hydropower potentials for pico, micro, mini and small schemes across Great Britain (GB). Northern Ireland was omitted from the study due to a lack of river flow data. The main objectives of this study were to

• ▪

  validate simulated flow data against observed river flows

• ▪

  generate points along the river network containing information about flow, elevation and so on

• ▪

  calculate hydropower potential across various RoR sizes, classified by hydrological, technical, financial viability and realisable potentials.

For the first time, the maximum hydrological and realisable potentials (Figure 1) for RoR hydropower were assessed in addition to the financial and technical potentials that previous studies have evaluated for GB, providing a holistic view of the hydropower landscape. This inclusive approach ensured that the assessment was not solely reliant on economic factors but also considered environmental and social implications, thus presenting a more balanced perspective on the feasibility and sustainability of RoR hydropower projects. Moreover, this research addresses a crucial gap in the existing literature by providing a comprehensive and up-to-date picture of the size and spatial distribution of RoR hydropower potential in GB. The significance of this study lies not only in its role in addressing the current gaps in knowledge but also in its practical implications for supporting the sustainable development of small hydropower to meet the UK's net zero targets through practical and feasible policy recommendations.

For this original research, a methodological framework containing the workflow steps was created to determine the RoR hydropower potential (Figure 2), offering a comprehensive evaluation that extends beyond conventional technical and financial considerations. The key steps in the methodology are as follows.

• (a)

  Validation of simulated flows

  • (i)

    Simulated flows for gauged catchments in GB were obtained from the UK Centre for Ecology and Hydrology (UKCEH) Marius (Managing the risks, impacts and uncertainties of drought and water scarcity)) project, using the grid to grid (G2G) hydrological model.

  • (ii)

    The simulated flows were validated against observed flows at 1498 gauged locations across GB, using the National River Flow Archive dataset (NRFA, 2023).

  • (iii)

    The coefficient of determination (R²) was calculated for various flow percentiles and flow statistics to assess the accuracy of the simulated flows.

• (b)

  Generation of analysis points

  • (i)

    Input data, including a DEM, a vectorised river network of GB and monthly flow data, were processed in ArcGIS.

  • (ii)

    A Python script was used to generate points along the river network at 25 m intervals, capturing information such as river segment ID, percentile flow, coordinates and elevation.

• (c)

  Hydropower potential calculations

  • (i)

    The maximum available potential RoR hydropower resource in GB was analysed, categorising it into pico, micro, mini and small RoR hydro.

  • (ii)

    Hydrological potential was calculated by considering the available power at each river segment, factoring in flow, head and turbine efficiency.

  • (iii)

    Technical potential was refined by considering factors like penstock diameter, turbine type and other technical limitations.

  • (iv)

    Financial potential was assessed by calculating the costs and net present value (NPV) of each technically viable RoR scheme.

  • (v)

    Realisable potential – schemes located in environmentally protected areas identified from Natura 2000 maps (EEA, 2022) were excluded from the realisable potential, considering environmental constraints.

Simulated flows for gauged catchments in GB (excluding the Isle of Wight, Anglesey, Kintyre Group, Inner Hebrides, Outer Hebrides, Orkney and Shetland due to lack of river flow data) from 1891 to 2015 were obtained from the UKCEH Marius project (Bell et al., 2018), in which the G2G hydrological model was used. The G2G is a hydrological model that operates on a national scale for GB. It uses a 1 × 1 km grid that aligns with the national grid, with a timestep of 15 min. The model is parameterised using digital datasets, such as soil types and land cover. It takes into account the impact of urban and suburban land cover on runoff and downstream flows. Studies (e.g. Bell et al., 2009; Formetta et al., 2018) have shown that G2G performs well for various catchments across Britain, particularly those with natural flow regimes, as it does not include artificial influences like abstractions and discharges on river flows. These were validated using the NFRA-observed flow dataset consisting of 1498 gauged locations spread uniformly (expect for the Scottish Highlands) across GB (NRFA, 2023). The coefficient of determination (R²) was used to determine how well the simulated flows predicted gauged flows. Bell et al. (2018) recommended that a comparison between simulated and observed flows should be made using statistics over lengthy periods of time, rather than time series. Therefore, R² was calculated for the percentile flows (Q₅, Q₁₀, Q₅₀, Q₇₀ and Q₉₅), minimum, mean and maximum flows at all gauges.

The input data (Figure 3) used were a DEM of the UK at 25 m resolution (Copernicus, 2016), a vectorised river network for GB (OS, 2022), monthly flow data from the Marius dataset (Bell et al., 2018) for GB ranging from 1891 to 2015. Monthly flow data were statistically processed into percentile flows from Q₁ to Q₉₅ in increments of 5% (Q₁ having an exceeding probability of once in 100 years etc.). Once these data was loaded in ArcGIS, a Python script generated points with unique IDs along the river segments spaced at 25 m (the same resolution as the DEM). Each point was assigned a unique ID along with the river segment ID (from OS (2022)), percentile flow (Q₁–Q₉₅), coordinates (easting and northing) and elevation. As the area considered for analysis was large (209 331 km²), it was necessary to divide it into smaller catchments to be able to compute the results. The hydrometric areas from NRFA (2014) were thus used to divide GB into catchments (Figure 3(d)). The names and areas of each hydrometric area are provided in Table S1 of the online supplementary material.

The maximum available potential RoR hydropower resource in GB was analysed. The total potential was split between pico, micro, mini and small RoR hydro, and between the hydrological, technical, financial and realisable potential (Figure 1).

Figure 4 shows the method used to calculate hydrological potential. The available power was calculated considering the first upstream point on a river segment to be the intake (step 1) and passed through all the downstream points (step 2), which were considered power houses, to calculate available power. After the algorithm went through all the possible power house locations for the first point, it considered the second downstream point as the intake and proceeded to calculate available power considering the next downstream points as power houses (step 3). This was done until the most downstream point of that river segment was reached. After this step was performed on all the river segments and all possible available power combinations were calculated, the algorithm chose the intake and power house locations on each river segment that generated the largest amount of power (step 4). Only one RoR scheme was considered for each river segment for computational ease and to ensure that the schemes would be independent (not cascading schemes). Another assumption made was that penstocks followed the river segment; therefore, they are not a straight line from intake to power house as shown in the sketch in Figure 4.

The power was calculated as:

where P is the available power in kW, g is acceleration due to gravity (= 9.81 m/s²), ρ is the density of water (= 1000 kg/m³), η is the turbine efficiency (considered to be 0.85 – an average value (Kabalci et al., 2021)), Q₄₀ is the flow with an exceedance probability of 40% (in m³/s), Q₉₅ is the flow with an exceedance probability of 95% and is considered to be the minimum flow that needs to be available downstream of the intake (in m³/s) and H is the available head between the intake and the power house (in m).

After the hydrological potential locations were chosen, the next phase was to refine the search considering technical limitations.

For this analysis only glass-fibre-reinforced plastic (GRP) pipes were considered since these are the most commonly used for small hydropower projects (Duncan, 2012). GRP pipes come in a variety of diameters, from 100 mm to 4 m, and pressure classes, from PN6 (resistant to pressures of up to 6 bar (1 bar = 100 kPa)) to PN32 (resistant to pressures of up to 32 bar). The choice of optimal penstock diameter was achieved by trialling different penstock diameters for a given design flow (Q₄₀–Q₉₅) to determine the optimal diameter.

GRP penstocks only come in diameters of 100 mm to 4 m. For RoR schemes that would require a diameter greater than 4 m, the flow was decreased at that location to try to find a suitable diameter. If the pipe diameter required was still greater than 4 m even with the smallest flow, that location was not considered since there are no suitable penstocks. In the case of an initial GRP pipe diameter smaller than 100 mm, the diameter was chosen to be 100 mm. If the total head was greater than 320 m (corresponding to approximately 32 bars (3200 kPa)), that RoR scheme location was not considered in this analysis because the penstock pressure would be too large. A possible solution to high heads is to reinforce the last part of the penstock (where the pressure would be greater than 32 bars) with a steel interior. However, this option was not considered for this analysis.

Five turbine types were considered: Pelton, Francis, Kaplan, Archimedean screw and Cross-flow. Considering the flow and the head available at the RoR locations, a turbine type was assigned for each location. If a suitable turbine could not be found for a particular location, the flow was modified to match the conditions of one the five turbine types; otherwise, that location was not considered in the analysis. The design criteria for each turbine are presented in Table 1. Detailed equations on the choice of penstock diameter and turbine efficiency are presented in Appendix 1.

The costs and NPV of each technically viable RoR scheme were calculated. In total, 14 costs were added to evaluate the total initial cost of the scheme (turbine, generator and control costs, penstock costs, civil structures costs, substation and transformer costs, transmission costs, access road costs, engineering costs, development costs, feasibility study costs and miscellaneous costs). The costing methodology was based on the work of Duncan (2012) and the RETScreen software developed for North American hydro projects (NRC, 2022). This approach was used as empirical cost functions are not available for GB.

RETScreen (NRC, 2022) uses pricing based on the Canadian dollar so a conversion factor for pound sterling was used based on the exchange rate at the time of the study (1 C$ = £0.58). Since the exchange rate for the current year was used, price inflation was not considered. Additionally, a map of the national power grid was not available, so the distance for the transmission lines was considered to be 1 km for all schemes. The cost of a Cross-flow turbine was taken to be the cost of a Pelton turbine. The cost for an Archimedean screw turbine was considered to be 22% less than that of a Pelton turbine (YoosefDoost and Lubitz, 2020). Cumulative NPV was calculated considering the lifespan of the RoR scheme to be 25 years (or 35 years) and the yearly discount rate was set to 4% to determine if a scheme was financially viable. Verán-Leigh and Vázquez-Rowe (2019) argue that RoR schemes could have a lifespan of up to 50 years. However, electromechanical equipment can have a shorter lifespan (40 years for Pelton turbines, 25 years for Francis turbines, 30 years for generators and 30 years for transformers (Flury and Frischknecht, 2012)). These shorter lifespans prompted the selection of a conservative lifespan for the RoR schemes. The yearly operational and maintenance cost was set to be 3% of the initial total cost as typical values range from 1% to 4% (Irena, 2021).

After a preliminary financial analysis, all the pico and micro RoR schemes had a negative NPV after a lifespan of 25 years. This prompted the design of a colour system

• ▪

  green RoR schemes have a positive NPV and are financially viable considering a lifespan of up to 25 years

• ▪

  amber RoR schemes have a positive NPV considering a longer time span of up to 35 years.

Additionally, for pico and micro schemes, since they are more likely to serve small communities or farms and local industries, the transmission line cost was considered to be zero, meaning they are not connected to the national grid. A detailed methodology of the financial potential is presented in Appendix 2.

The yearly income from selling electricity was calculated considering a capacity factor of 40% and a wholesale energy price of £45/MWh (accurate for 2020 according to BEP (2021)) as:

where P_d is the installed power (in MW), 365 is the number of days in a year, 24 is the hours in a day and 0.4 is the capacity factor (40%).

After computing the financially viable potential, the realisable potential was calculated. This was determined by excluding schemes in environmentally protected locations taken from Natura 2000 maps (EEA, 2022). Although the UK has exited the EU, which manages these sites, the UK government decided to continue to protect these areas (Rapleys, 2021), so they were considered suitable for exclusion of the construction of RoR schemes due to environmental concerns.

The flow dataset (Bell et al., 2018) was validated against gauged river flows taken from NRFA (2014). Table 2 shows the R² values for percentile flows for all gauges. The simulated flows performed better for higher and mean flows (Q₅₀, Q₁₀, Q₁) than for small and minimum flows. However, the simulated dataset was good to use in the analysis, with most R² values above 0.5 (Shen et al., 2018). Looking at the spatial distribution (Figure 5), the R² values were mostly over 0.7, with a few locations with R² < 0.3, showing that the simulated flows were predicted accurately by the dataset used in this study (Bell et al., 2018).

The total GB-wide (excluding islands) hydrological potential was assessed to be ∼20 GW, with a total technical potential of ∼11 GW. The total financial potential, however, was much smaller, estimated to be 390–420 MW, with a total realisable potential of 290–320 MW (Figure 6). The large drop from technical potential to financial potential is likely due to the assumptions and limitations of the financial criteria used in this analysis (i.e. a lifespan of 25 years (for green schemes) or 35 years (for amber schemes), lack of information on the national power grid (which prompted the selection of a fixed transmission line length) and not considering the benefits of selling renewable energy to the grid). Additionally, a fixed wholesale electricity price of £45/MWh was considered and it was assumed that the RoR scheme would produce electricity 40% of the time over a year (i.e. capacity factor of 0.4). These numbers may differ as the RoR scheme may sell electricity at higher prices during peak times and the capacity factor may change from year to year depending on river flow availability. However, increasing the wholesale electricity price from £45/MWh to £70/MWh (according to the latest figures from BEP (2021)) did not lead to a substantial increase in financial potential, suggesting that price does not play an important role in determining the financial feasibility of RoR schemes.

While sizable hydrological and technical potential was evident for pico and micro schemes (1.78 GW of hydrological potential and 1.19 GW of technical potential for schemes), the financial and realisable potentials from these types of schemes was zero (Figure 6). This is due to their smaller installed capacity (<100 kW), which, using the financial criteria from this study, does not produce and sell enough electricity to cover both the initial costs and the annual operation and maintenance costs. Two thirds of the hydrological potential for pico and micro schemes is technically viable. This suggests most of the pico and micro hydrological potential could be achieved with today's technology. On the other hand, only 42% of the mini and small hydrological potential is technically viable.

The spatial distribution of RoR hydrological potential across the four hydropower types (pico, micro, mini and small) was found to be uniform for pico schemes, while micro, mini and small schemes are mostly located in the west of GB (Figure 7). The largest clusters of RoR schemes are in the Taff group, Mid-Glamorgan group and Glaslyn group catchments in Wales and at the intersection between the Ouse, Mersey and Irwell and Ribble catchments in England (see Table S1 of the online supplementary material). Micro RoR schemes were found to be similarly spread throughout the country, with clusters in Wales and a cluster in England, in the Tyne, Wear, Tees group, Wyre and Lune, Kent group, Esk group, Derwent group and Eden catchments. Mini and small RoR schemes, however, are mostly located in the western side of the study area. Clusters were found in the Scottish Highlands for both mini and small RoR schemes in the Loch Linnhe group, Lochy, Beauly and Ness catchments. The Tay catchment in Scotland has the largest total hydrological potential at 1250 MW (6% of the total hydrological potential). This is perhaps expected due to the catchment size (5093.17 km² – the largest in Scotland), the mountainous terrain and high annual precipitation (Met Office, 2014).

Most of the technical potential was found to be in the western part of the study area, with clusters showing the largest technically viable potential to be in the same locations as those with the largest hydrological potential (Figure 7). Technically viable pico schemes are situated in south-eastern England (Norfolk Rivers group, East Suffolk Rivers, Thames and Sussex Rivers group catchments). Other clusters are located in the west of the study area in the Glaslyn group and Douglas group catchments. Technically viable micro RoR schemes are situated in the same locations as the hydrologically viable micro schemes. The mini and small RoR technically viable potential is in the west and north of the country. The largest total technically viable potential is in the Ouse catchment in Yorkshire (the second largest catchment in England), with a total potential of 476 MW (5% of the total technical potential).

Clusters of mini RoR schemes were identified in the west and south of the study area (Figure 7), with the largest ones in the Dify group, Taff group, Mid-Glamorgan group and Loughor group catchments in Wales. Small RoR schemes are situated in the west and south-west of the country, with the largest cluster in the Taff group catchment in Wales. The largest financially viable potential was found in the Taff group catchment, with a total potential of 21 MW (4.9% of the total financially viable potential).

Most of the small RoR plants were found to have a positive NPV before 25 years of operation (green schemes), with a few having a positive NPV after 25 years but before 35 years of operation (amber schemes). Conversely, half of the viable mini RoR schemes are green schemes, while the other half are amber schemes. This is because small RoR schemes reach a positive NPV earlier in their lifetime (11 years for small RoR and 22 years for mini RoR schemes on average). This leads to a higher NPV at the end of their lifetime compared with mini RoR plants. The median initial cost of a financially viable mini RoR scheme is £0.92 million, while the median initial cost for a financially viable small RoR is £1.80 million. However, the median NPV of a mini RoR scheme is £0.12 million, whereas the median NPV of a small RoR scheme is £1.33 million. These results suggest that, compared with mini RoR schemes, small RoR schemes are more financially viable and have the highest NPV.

In the case of mini RoR schemes, the most profitable schemes have a combination of flows of 0.75 m³/s with a head of 100 m and flows of 0.25–1 m³/s with heads of 175–300 m. For small RoR plants, locations with flows of 2–4.5 m³/s and heads of 75–200 m are the most profitable considering the financial criteria used in this analysis (Figure 8). The cut-off installed power above which RoR schemes become financially viable is approximately 312 kW for mini RoR and 1001 kW for small RoR (Figure 8).

Regarding the initial costs (see Figure S5 in the online supplementary material), for mini RoR, the largest cost is the turbine (∼18% of the total initial cost on average), followed by the civil structures cost (∼17% of the total initial cost on average). The largest initial cost for small RoR plants is the civil structures cost (∼25% of the total initial cost on average), followed by the turbine cost (∼19% of the total initial cost on average). For both mini and small RoR plants, the penstock cost is the most unpredictable, with a standard deviation of 4.67% for small RoR and 9.23% for mini RoR.

The spread of realisable potential across GB was similar to the spread of financially viable potential (Figure 7). The largest clusters were found to be in the same regions as the financial potential clusters. The largest realisable potential is located in the Taff group catchment, with a total potential of 20 MW (6.5% of the total realisable potential).

The methodology used in this study aligns with recent international studies on hydropower potential, including significant works by Garegnani et al. (2018), Zaidi and Khan (2018) and Sammartano et al. (2019). For example, in this study, potential hydropower values were calculated using the Q₄₀ flows and geospatial data techniques like GIS and remote sensing, similar to the methods used by Zaidi and Khan (2018). The characteristics of the hydrological, technical, financial and realisable potentials are similar to those reported by Garegnani et al. (2018) and Sammartano et al. (2019). This contextualises the present research within the broader global landscape of hydropower assessment and strengthens the overall body of knowledge in the RoR hydropower assessment domain.

Examining global studies (Hoes et al., 2017; Zhou et al., 2015), their assessment of the UK's gross hydropower potential yielded figures of less than 3300 kWh/year per person, with an exploitable potential estimated at approximately 138 kWh/year per person. Considering the population of GB (excluding islands) at approximately 60 million individuals, the hydrological potential findings from the present study are in accordance with prior research at approximately 3000 kWh/year per person. However, the determined realisable potential values are marginally lower than the exploitable potential (comparable with the realisable potential in the present study) reported by Zhou et al. (2015), standing at approximatively 47 kWh/year per person. This discrepancy is likely attributed to methodological variations, primarily in the selection of the design flow: Q₄₀ was adopted as the design flow in the present analysis to ensure a sufficient flow for at least 40% of the year, whereas Zhou et al. (2015) opted for a higher design flow of Q₃₀, consequently increasing their hydropower potential results. Another reason for the discrepancy in results may be the inclusion of smaller hydropower sites, which may not have been feasible in the study of Zhou et al. (2015).

Previous studies in GB have largely based their criteria for selecting and assessing suitable locations for RoR hydropower on technical and financial viability (e.g. DECC, 2010; Duncan, 2012; Nick Forrest Associates, 2008; SCE, 1989). However, in the present work, a framework that assesses hydrological and realisable RoR potentials alongside financial and technical potentials was developed and demonstrated. The method presented provides, for the first time, a clearer and up-to-date picture of the size and locations of RoR hydropower potential across GB. Furthermore, this study has raised important questions about the implications of the financial viability criteria. In this study, only the monetary value of selling electricity to the grid was considered, ignoring the other benefits of producing renewable energy – avoided emissions, green jobs created, improved health benefits and so on. Not considering the financial value of these benefits has most likely led to an under estimation of the financial potential.

The results are broadly consistent with the findings of SCE (1989), which determined the RoR hydropower financial potential to be 322.3 MW for the UK, although the 1989 study did not consider any financial constraints for Scotland and Northern Ireland. For Scotland, the estimated financial potential is 120 MW, less than the estimates of Duncan (2012) (204–898 MW, depending on the discount rate) and Nick Forrest Associates (2008) (657 MW). However, Duncan (2012) considered a feed-in-tariff that is no longer used and prioritised financial viability rather than harnessing the most available hydropower. Nick Forrest Associates (2008) determined the potential for a different range of RoR sizes than those considered in the present analysis. For England and Wales, the estimated financial potential is 277–310 MW, similar to the estimate produced by the Department of Energy & Climate Change (DECC, 2010).

When comparing the total financial and realisable potential results to the already installed RoR capacity in GB (Kennedy et al., 2024), the results are similar (Figure 9). However, the spatial distribution of RoR schemes in this study is different from the distribution of already installed schemes. For example, most installed RoR schemes are in the Scottish Highlands, but potential RoR locations in Wales and England (Figure 7) were identified in this study.

Based on the results of this study, the following practical policy recommendations are proposed.

• (a)

  Financial incentives for small RoR Schemes. Recognising the financial viability of small RoR schemes, policymakers could consider implementing targeted financial incentives to encourage their development. These incentives could include subsidies, tax credits or grants for initial construction costs, especially for schemes with optimal flow conditions (e.g. flows of 2–4.5 m³/s and heads of 75–200 m), where financial viability is higher.

• (b)

  Streamlining approval processes. Given the smaller initial investments and shorter planning and construction times associated with RoR schemes compared with large hydropower schemes, regulatory bodies could streamline approval processes to expedite project development. Expediting approvals can contribute to the faster deployment of RoR projects, aligning with the urgency of meeting renewable energy targets.

• (c)

  Local community engagement. Given the advantages of RoR schemes, such as smaller inundated areas and the use of local labour and materials, policymakers could actively promote community engagement in hydropower projects. Establishing community-based partnerships can enhance acceptance, address local concerns and ensure sustainable development practices in line with environmental and social considerations.

• (d)

  Strategic focus on technically and financially viable regions. Policymakers could strategically focus on regions with high technical and financial potential for RoR schemes, particularly in the western part of the study area (GB). Infrastructure and grid development initiatives can be targeted in these regions to harness the available hydropower potential efficiently.

• (e)

  Research and development for pico and micro schemes. While financial and realisable potential for pico and micro schemes may currently be limited, policymakers should encourage research and development efforts to enhance the feasibility of these smaller-scale projects. Funding for pilot projects and studies focused on overcoming the challenges associated with pico and micro schemes could pave the way for future expansion.

By incorporating these policy recommendations, policymakers can create an enabling environment for the sustainable development of RoR hydropower projects in GB, contributing significantly to the nation's renewable energy goals and environmental objectives.

While robust, the method used in this research may be improved by considering the cost of connecting to the national grid based on the spatial distribution of available substations, which was not available for this analysis. This limitation affected the financial and realisable potential results, which could be made more accurate by considering the actual cost of transmission lines (see Appendix 1). However, this only affects the viability of mini and small schemes because pico and micro schemes were considered too small to be connected to the grid in this analysis. In addition, while the primary function of this study was to determine the maximum hydropower potential across GB, the methods and tools used could also be adapted for a quick determination of suitable locations for turbines.

A comprehensive framework to determine the potential for RoR hydropower in GB (excluding islands) was created. Using this framework, the total hydrological potential was estimated to be 20 GW, with a technical potential of 11 GW, a financially viable potential of 320–420 MW and a realisable potential of 290–320 MW. Most of the realisable schemes were found to be either mini or small (100 kW–5 MW) situated in the west and north-west parts of GB, with the largest realisable potential in the Taff group catchment in Wales (20 MW).

This study is one of the first attempts to thoroughly examine the RoR hydropower potential across GB and the first to do so in the context of the drive to net zero. The analysis of RoR potential undertaken here has extended knowledge of the size and location of hydrologically, technically, financial and realisable potential in GB and about the impact of the financial criteria chosen. However, this study was limited by (a) the absence of a map of the national grid to allow for accurate transmission line pricing, (b) a lack of updated data on penstock and turbine prices, (c) a lack of flow and river network data for the Isle of Wight, Anglesey, Kintyre group, Inner Hebrides, Outer Hebrides, Orkney and Shetland and (d) the use of a monthly flow time series rather than a daily or even hourly flow time series. Despite these limitations, the results of this work certainly add to the understanding of RoR potential in GB and offer reliable estimates of the amount of hydropower that could be produced and favourable locations for RoR schemes. Furthermore, the findings presented offer valuable insights that extend beyond regional boundaries. Industries involved in hydropower development can benefit from the novel methodological framework proposed here, which could be applied to different regions. This adaptability makes the research relevant and transferable, offering a valuable resource for regions to evaluate their own hydropower potential and make informed decisions regarding renewable energy development.

A natural progression of this work would be to analyse the effects of climate change on RoR hydropower potential, considering the proposed lifespan of the more than 25 years of the schemes identified in this study. Climate change poses significant challenges to hydropower systems globally, impacting both their performance and sustainability (Marshall and Chen, 2018). Changes in precipitation patterns, increased frequency of extreme weather events and altered hydrological cycles are affecting river flow dynamics, ultimately influencing hydropower generation (Seneviratne et al., 2023). Based on future projections, in GB, a reduction in river flows during spring and summer is expected, with a marginal increase in winter flows and varied projections for autumn flows (e.g. Christierson et al., 2012; Kay, 2021; Prudhomme et al., 2012). These envisaged alterations in the hydrological regime across GB are expected to alter hydropower production, particularly affecting RoR schemes without energy storage capabilities for accommodating seasonal variations. For example, in Scotland, the mean annual flow is projected to vary between ±5% in the future (2050s) (UKCEH, 2023), suggesting that hydropower potential in that area could vary from the 120 MW determined in this study to 114–126 MW. A comprehensive study that incorporates up-to-date climate projections, for example the dataset from Hannaford et al. (2022), and considers both annual and seasonal changes in hydropower production to provide an accurate assessment of the potential impacts of climate change on RoR hydropower generation may be useful.

The authors extend heartfelt thanks to their families and friends for their unwavering encouragement and understanding during the completion of this research.

An initial estimate of penstock diameter was calculated using the Manning formula for a fully closed circular cross-section, considering a head loss of 4% of the gross head between the intake and the power house:

where d is the penstock diameter (in m), ε is the roughness coefficient for GRP pipe (= 0.029), Q_d is the design flow (= Q₄₀–Q₉₅ in m³/s), H is the gross head between the intake and the power house (in m) and L is the penstock length, calculated as:

where X_I, Y_I and Z_I are the coordinates of the intake and X_PH, Y_PH and Z_PH are the coordinates of the power house. This first estimate was used as a starting point by setting minimum and maximum diameter limits of ±10% to determine the nominal diameter from a penstock database for available diameters. The average water velocity V in the penstock was calculated as:

The Reynolds number Re was calculated:

where ν is the kinematic viscosity of water (= 1.31 × 10⁻⁶ m²/s). Using Re, the Darcy friction factor (f) was calculated by solving the Colebrook–White equation:

Using the Darcy–Weisbach equation, the head loss was calculated as:

The net head for each RoR scheme was calculated using the head loss from Equation 8:

The pressure rating of the penstock was chosen based on the calculated surge head that may be experienced by the penstock. The surge head (h_s) was calculated by multiplying the penstock pressure wave celerity rating (α, in m/s) and the maximum flow velocity V:

The total head was then calculated as:

The total head was transformed into units of bar and then the pressure class for each diameter was chosen from a catalogue.

After a turbine was chosen, the next step was to calculate its peak efficiency to determine the installed power. For the Francis and Kaplan turbines, which are reaction turbines, the peak efficiency was calculated by determining the runner size and the specific speed. The runner size was calculated using Equation 12 based on the design flow Q_d (in m³/s):

where k is an empirical constant with a value of 0.46 for Q_d < 1.8 m³/s or 0.41 for Q_d ≥ 1.8 m³/s. The specific speed (n_q) was then calculated using the following equation based on the net head:

where k_q is a constant based on the turbine type (k_q = 800 for a Kaplan turbine and k_q = 600 for a Francis turbine). Before calculating the peak efficiency, an adjustment to the peak efficiency (⁠$e^nq$⁠) must be calculated using the specific speed n_q:

A similar adjustment needs to be made based on the runner size:

Turbine peak efficiency was then calculated as:

where R_m is a turbine design coefficient that considers the characteristics of different manufacturers and was considered 0.45 in this project. The peak efficiency flow (Q_p) is less than the design flow (Q_d) and was calculated as:

The number of jets is a key configuration for the impulse turbine. In this analysis, the Pelton turbine was considered for this calculation. For simplicity, all Pelton turbines analysed have only one jet. The rotational speed (n) was calculated as:

where H_net is the net head (in m), Q_d is the design flow (in m³/s) and j is the number of jets. The runner diameter was calculated considering the rotational speed:

The turbine peak efficiency was calculated as:

The peak efficiency flow (Q_p) is less than the design flow (Q_d) and was calculated as:

Cross-flow turbines are impulse turbines, which means that the rotor is spinning air and is not fully flooded like in a reaction (e.g. Kaplan) turbine (Sinagra et al., 2014). However, the operation is different from other impulse turbines (e.g. Pelton) and the efficiency does not vary as much with the flow rate. Nevertheless, the peak efficiency is less than in a Pelton turbine and the literature around the calculation of the peak efficiency is minimal. Considering this, the peak efficiency was taken from experimental data as shown in Figure S2 in the online supplementary material. The peak efficiency for this type of turbine was taken as 81.6% and the flow at which peak efficiency occurs was 70% of the design flow (Q_d).

An Archimedean screw hydropower scheme allows for the transformation of the potential energy of a fluid into mechanical energy and is convenient for low-head hydraulic sites. As it is a less common technology, there are few references dealing design and performance optimisation (Dellinger et al., 2016). Similar to the cross-flow turbine, experimental data were used to determine the peak efficiency and flow. The peak efficiency was taken to be 82.8% and the peak flow was determined to be 78% of the design flow (Figure S3).

After the peak efficiencies and flows of all turbine types were calculated, the power was recalculated using these new values to determine the technically viable power.

The generator cost is common for all turbine types and was calculated as:

where n_T is the number of turbines (in this case 1), P_d is the design power in MW and H_g is the gross head. The turbine costs depend on the turbine type. For reaction turbines, the following formulas were used.

For a Kaplan turbine:

where d_run is the runner diameter in m.

For a Francis turbine:

For impulse turbines, RETScreen provides two equations based on the installed power to net head ratio. However, in this analysis, only one of the equations was considered for the Pelton turbine since the small Pelton turbine cost made it much higher than the other turbines analysed. Furthermore, a 0.5 coefficient was used to estimate the cost of a Pelton turbine and governor:

Formulas for the installed MW/unit are provided in Table 3.

The cost of a cross-flow turbine was taken as the cost of a Pelton turbine. The cost for the Archimedean screw turbine was considered to be 22% less than the cost of a Pelton turbine (YoosefDoost and Lubitz, 2020). Installation costs for the turbine, governor and generator were obtained using:

For the hydropower schemes analysed in this study, only GRP pipes were considered for the technical potential and the financial potential. However, the prices for penstock pipes were difficult to find so an estimation based on the work of Duncan (2012) was made for both rigidity classes considered in this project (SN 5000 and SN 10000). The prices are presented in Figure S4 of the online supplementary material. The installation cost for a penstock was based on its weight:

where W is the penstock weight in kg.

The civil structures cost includes the weir, intake works, power house and tailrace and was calculated as:

where l_b is the distance to borrow pits, assumed here to be 0.5 km and l_d is the length of the crest of the dam, assumed to be 10 m.

A substation and transformer are required to increase the voltage at the generator terminals from 1000 V to the transmission voltage. Their cost was assessed using the following formula:

where n_gen is the number of generators, chosen as 1 in this case and V_t is the transmission voltage, assumed here to be 33 kW. The installation costs for the substation and transformer were calculated as a percentage of the cost of the substation and transformer.

Transmission cost is fixed at C₉ = £150 091.5 considering the transmission line length to be 1 km and assumed that connections will be made to primary substations with a 33 kV busbar or to existing 33 kV lines. Connections to lines at voltages in excess of 33 kV typically require a new substation, costing in excess of £1 000 000. For these reasons, only connection at 33 kV was considered. The cost for 33 kV overhead lines based on published utility cost data suggest a range of £1550–8715 for a 70 m span line, giving a range of £22/m to £124.5/m (SHEPD, 2011). The midpoint of this range is £73.25/m which was taken as the unit cost of the transmission line. When connecting to an existing 33 kV line, it is assumed that a simple tee connection can be made. This is priced at £50 000 based on the cost of two additional sets of utility switchgear, one at the tee and a second at the hydro scheme step-up transformer. When connecting at a primary substation, it is assumed that an additional connection can be made to an existing 33 kV busbar at a cost of £150 000 based upon example 7A in SHEPD (2011). In both cases, an additional factor of 1.25 was applied to lone costs to represent non-ideal line routing. This gives the cost functions for substation connections, which was considered in this study to be a conservative approach.

Access road costs were calculated as:

where T is a cost reduction factor for unpaved roads (assumed here as 0.25), A represents the access difficulty set between 1 to 6 based on the terrain (assumed here to be 2), l_a is the distance to the nearest road, not considering existing water bodies or elevations.

Engineering costs were calculated using the following formula:

where E is an engineering cost factor which is 0.67 if a dam already exists at the location and 1 if a dam does not exist; here E = 1 was assumed.

Development costs were assumed to be the costs associated with obtaining the necessary permits to allow work and they depend on the other previous costs:

Feasibility study costs were calculated based on the other costs:

Miscellaneous costs represent unforeseen costs that may occur during the construction of the project and were calculated based on an interest rate (IR) assumed to be 10%.

Total cost is the sum of all costs:

The cost per kW was thus calculated by diving the total cost by the installed power:

All input data used in this study can be found at https://doi.org/10.15129/95db013c-d082-4aa9-a8ce-43b60568087d.

A

access difficulty set

C_ex

exchange rate (Canadian dollars to pound sterling)

C_inf

conversion factor taking inflation into account

E

engineering cost factor

$e^drun$

adjustment to peak efficiency based on runner size

$e^nq$

adjustment to peak efficiency based on specific speed

e_p

peak efficiency of turbine

d

penstock diameter

d_run

net head

f

Darcy friction factor

g

acceleration due to gravity

H

available head between intake and power house

H_g

gross head

H_net

net head

H_total

total head

h_s

surge head

hf

head loss

j

number of jets

k_q

constant based on turbine type

L

penstock length

l_a

distance to nearest road

l_b

distance to borrow pits

l_d

length of crest of dam

n

rotational speed of turbine

n_gen

number of generators

n_q

specific speed of turbine

n_T

number of turbines

P

available power

P_d

installed power

Q_d

design flow

Q_p

peak efficiency flow

Q_x

flow with exceedance probability of x%

Re

Reynolds number

R_m

turbine design coefficient

T

cost reduction factor for unpaved roads

V

average water velocity in penstock

V_t

transmission voltage

W

weight of penstock

X_I, Y_I, Z_I

coordinates of intake

X_PH, Y_PH, Z_PH

coordinates of power house

α

penstock pressure wave celerity rating

ε

roughness coefficient

η

turbine efficiency

ν

kinematic viscosity of water

ρ

water density