Dam Renovation to Prolong Reservoir Life and Mitigate Dam Impacts

Abstract

Dams are essential to society, yet have tremendous environmental impacts, for which there is an increasing interest in mitigation. At the same time, sedimentation threatens the sustainability of reservoir storage and reservoir functions. We use the term dam renovation to encompass a wide range of measures, including dam rehabilitation, a term commonly used for structural retrofits, typically of the dam structure or spillway, fishway retrofits for migratory fish passage, reservoir reoperation, which involves modifying dam operations to improve flow regimes for ecological purposes, and sustainable sediment management, which includes measures to pass sediment through or around dams, as well as other mechanical measures to restore sediment connectivity. Compared to dam renovation, an inordinate amount of literature has been published on the topic of dam removal. While in some cases dam removal is a practical way to improve river condition and to resolve the safety problems of aging dams, the reality is that most dams in existence today will remain for the foreseeable future, provided they do not fill with sediment, or their structures deteriorate to the point of failure. Thus, it is imperative that we understand the options available to renovate dams with poor environmental performance or whose sustainability is threatened.

1. Introduction

Dams are essential components of modern infrastructure, impounding water for municipal supply, irrigation, hydroelectric generation, navigational flow releases, and flood control. There are over 2.8 million medium and large dams globally, which have altered flow regimes and longitudinal connectivity on most of the world’s rivers [1]. By altering flow regimes, blocking migration by fish and other organisms, trapping sediment, nutrients, and other materials in flux, dams have multiple impacts on river systems up- and downstream. Moreover, trapped sediments fill reservoirs, displacing storage space for water.

There are many reasons why dams may require renovation. These can include deterioration of the dam itself (e.g., the commonly occurring problems with concrete swelling and weakening due to alkali-aggregate reaction and similar processes), risk of failure from earthquake or landslides into the reservoir, risk of overtopping due to changed hydrology, accumulation of sediment threatening the long-term sustainability of the dam and its services, and evolution of the expectations of the dam, such as increased environmental standards and public expectations of environmental quality and recreational opportunities along rivers.

Societal expectations of the benefits from rivers have evolved since the dam-building frenzy of the mid-20th century. In the US, dams built in the 1930s through the 1960s, were generally not required to account for potential environmental impacts. Exceptions were rivers with important runs of anadromous salmon, but, even in these cases, the imperative for hydropower or water supply nearly always trumped environmental considerations, as illustrated by the closure of the Friant Dam on the San Joaquin River, California, which caused the extinction of a distinct run of chinook salmon. Since the 1970s, in the US, requirements under the National Environmental Policy Act (NEPA), the Endangered Species Act (ESA), and the Federal Power Act have included analysis of environmental impacts, modifications of dam operation to minimize impacts, and actions to mitigate impacts. Federal projects are required to undergo consultation with relevant federal agencies, so, for example, the US Bureau of Reclamation, or the US Army Corps of Engineers, must undergo an ESA ‘Section 7’ consultation with the US Fish and Wildlife Service or National Marine Fisheries Service for any species present that are listed under the ESA. Non-federal projects must operate under a license from the Federal Energy Regulatory Commission (FERC). These licenses must be renewed upon expiration, usually after 40–50 years. As discussed below, this has resulted in many studies of fish and other environmental resources affected by dams and has involved extensive consultation with local communities, and with other stakeholders, such as the regulatory agencies tasked with the conservation of fish and wildlife.

In Europe, the adoption of the EU Water Framework Directive in 2000 required member states to assess the ecological conditions of water bodies, including their geomorphological conditions, and to develop measures to improve them where needed. In many countries, the assessments highlighted dams as key disrupters of longitudinal connectivity, and recommended measures, including dam removal or modification, to permit the passage of fish, sediment, and large wood. The EU project, AMBER, developed a detailed atlas showing such barriers across Europe and case studies of how longitudinal barriers were resolved.

With climate change, even dams that currently meet the expectations set upon their construction decades ago may be inadequate to meet safety standards in the future, as the magnitude and intensity of storms and floods (for a given return interval) increase. Virtually every river basin in the world is expected to experience some changes in flow regime, increasing the threat of flooding in many populated areas, and altering the hydrologic basis on which hydroelectric power systems were designed. For example, the majority of hydroelectric power in California USA is generated by plants at high elevation. These projects have very little storage, but rather depend on storage in the snowpack. However, with climate change, less precipitation will fall as snow and more as rain, meaning less inflow to these high-elevation hydroelectric plants during the snowmelt season of later spring-early summer, with impacts on potential hydroelectric power production [2]. Similarly, climate change is expected to produce shifts in vegetation cover that may lead to increased erosion rates [3], which can be expected to increase the rates of sedimentation in reservoirs. Even larger changes are documented from land-use changes in river basins draining to dams, with road construction, mining, and land-clearing for agricultural expansion being among human actions with the greatest impact on flow and sediment loads reaching reservoirs.

These environmental changes are being imposed on an aging population of dams. As summarized by Pittock and Hartmann [4], the majority of the world’s dams were built in the latter half of the 20th century, mostly by the 1980s. When the World Commission on Dams published its report in 2000, the world’s large dams were, on average, 35 years old [5] (WCD 2000). Like any infrastructure, dams require maintenance, and, if they are not maintained adequately, there is increased risk to downstream populations from dam failure. The American Society of Civil Engineers [6] reported over 4000 ‘deficient’ dams in 2008, of which over 1800 had ‘high hazard potential’, and the list of deficient dams was growing faster than dams were being repaired.

An increasing number of dams are being removed when they are unsafe, no longer needed, their costs (notably hydroelectric relicensing costs) exceed their economic return, or when there is strong social demand for removal to restore fish migration, or to achieve other valued outcomes [7]. Compared to dam renovation, an inordinate amount of literature has been published on the topic of dam removal. While, in some cases, dam removal is a practical way to improve river condition and to resolve safety problems of aging dams, the reality is that most dams in existence today will remain for the foreseeable future. The exceptions are those affected by structural defects, deterioration, or by sedimentation at rates rapid enough to affect reservoir/dam function, dams that no longer serve a compelling engineering/economic function, and dams that block important migratory routes of fish. Most dams removed to date have involved two of these three criteria (e.g., safety, low economic return, or ecological impact). Given that the numbers of new dams being built far exceed the numbers of old dams being removed, and, given that most existing dams are here to stay for some time, it follows that our first priority should be to find ways to improve the performance of dams, to lessen their environmental impacts, and to sustain their benefits over time. Because most reservoirs will persist, it becomes critical that we understand how to better manage them, how to modify their operation with existing controls and facilities, or what kinds of retrofits are needed. Such changes in the operation and structural features of a dam constitute ‘dam renovation’

In this paper, we use the term dam renovation to encompass a wide range of measures, from dam rehabilitation, a term commonly used for structural retrofits, typically of the dam structure or spillway, fishway retrofits to dams to allow migratory fish to pass through or around the dam, to reservoir reoperation, which involves modifying dam operations to improve flow regimes for ecological purposes, and sustainable sediment management, which includes measures to pass sediment through or around dams, as well as other mechanical measures to restore sediment connectivity. These topics are developed after a summary of dam impacts, and then the approaches above are illustrated by case studies. Most cases described in the literature as dam renovation were for projects that could be considered rehabilitation.

2. Effects of Dams

Before considering how dams should be renovated or re-operated for environmental and social benefits, we consider the effects of dams on river systems more broadly, to better understand the effects of dams for which mitigation or restoration may be needed. We briefly summarize these effects and their ecological consequences here, with citations to more detailed treatments published elsewhere.

2.1. Dam-Induced Changes in Connectivity and Flow

Rivers are characterized by connectivity in three dimensions: longitudinal (i.e., upstream-downstream), lateral (i.e., channel to floodplain and side channel), and vertical (surface-groundwater) [8]. Dams interrupt the longitudinal continuity of river systems directly, and, through changes in flow regime and induced changes in channel form, can also reduce lateral connectivity. These changes can reduce vertical connectivity as well. Dams trap sediment, nutrients, and large wood, depriving downstream reaches of these elements, as discussed below.

One of the best-understood ecological impacts of dams is blocking of the passage of migratory fish. While dams block fish migration in rivers globally, research relevant to this impact and its mitigation with fishways has been concentrated on coldwater rivers and streams with salmon and trout [9]. The classic example is the anadromous salmon who seek to swim upstream from the sea (where they spend their adult life) to their natal freshwater streams to spawn, but are blocked by dams [10]. However, many fish species in other river systems also depend on migration, often long-distance, to reach spawning grounds, and when they cannot reach their spawning areas, the fish runs can be extirpated. Likewise, populations of catadromous fish, notably eels, which reproduce in the ocean, but spend their adult lives in freshwater until they redescend seaward to spawn, have been devastated by dams that interfere with their longitudinal movements [9].

The passage requirements and capabilities of non-salmonid species, especially in tropical rivers, are essentially unknown, due to a lack of research and lack of experience in the implementation and monitoring of fish passage measures in these river systems. Dams can block the movement of fish both upstream and downstream in different ways. The image of the adult salmon swimming upstream and running into a concrete wall is well-established in the public imagination, but, equally important are the problems of downstream migrants, who may pass through hydroelectric turbines, whose blades may kill them, as documented for seaward migrating adult European eels (Anguilla anguilla) by Pedersen et al. [11], a source of mortality that has contributed to the loss of nearly three quarters of this population in recent decades. Moreover, while upstream migrants tend to be adults of stronger swimming species, downstream migrants include eggs and larvae that passively drift downstream [9].

Dam-induced changes in river flow regimes vary widely, depending on the natural pre-dam hydrology of the river, and the purpose, size, geometry, and operation of the dam and reservoir. In some cases, water is diverted directly from the reservoir, so the net flow in the river downstream is reduced, while, in other cases, the dam releases all the stored water (less water lost to evapotranspiration and seepage into underlying strata) but in a changed temporal pattern. Dam-induced changes to the temporal pattern of flows occur over multiple time scales, from inter-annual, to seasonal, to daily, to hourly. Many larger dams are designed to provide year-to-year carryover storage, altering flow patterns on a multi-year scale, while hydroelectric dams are often designed to provide peaking power and to thereby take advantage of higher electricity rates at periods of high demand. Unless there is a re-regulating reservoir downstream, this hydropeaking can cause unnatural rises and falls in water levels, with deleterious effects on aquatic biota and very apparent signatures in the river hydrograph [12].

The changes in flow regime induce changes in river ecology in multiple ways. River flow and sediment load create physical habitats, such as bars, riffles, pools, and floodplains. Changes in flow regime can result in changes to these features. Because aquatic species have evolved to adapt to natural seasonal patterns of flow, artificially modifying the seasonal flow regime (termed ‘de-synchronization’) can disrupt various life stages of organisms, such as reproduction, juvenile growth and migration, and adult migration.

Reservoirs also alter temperature regimes, especially large reservoirs with deep water, which can stratify into layers with different temperatures. When dams draw water from surface layers in summer months, the result can be unnaturally high temperatures, but drawing from deeper layers may yield unseasonably cold water. These cold waters maintain a run of endangered winter-run Chinook salmon (Oncorhynchus tshawytscha) below Keswick Dam on the Sacramento River (outside of their original habitat range), and a run of rainbow trout (O mykiss) below Glen Canyon Dam on the Colorado River, where the trout are completely exotic and survive only because of the artificially cold temperatures.

Many species depend on longitudinal and lateral connectivity (for reproduction, migration, feeding, and/or growth); disrupting this connectivity interferes with these functions. As native species are adapted to natural flow variability and temperature regimes, this gives them an evolutionary advantage over potential invaders, but if the flow regime is artificialized, it can allow invasive species to flourish at the expense of native fauna [13]. Many livelihoods are linked to the riverine ecosystem, including fishing, floodplain agriculture dependent on natural flooding cycles that recharge water tables and renew soil fertility through deposition of silt, and food sources from riparian forests. As these natural riverine functions are disrupted, it can affect the local communities that depend on them. As discussed below under Reservoir Reoperation, to achieve environmental improvements, the flow release pattern from dams can be modified to better reflect natural patterns that support native species and local livelihoods.

2.2. Sediment Trapping by Reservoirs and Dam-Induced Decrease in Downstream Sediment Load

In a natural river basin, sediment is transported through river basins from rapidly eroding mountainous headwaters down to river deltas. There are many ways that human actions can alter erosion, sediment transport, and deposition throughout the river basin, such as increased erosion rates from road construction and clearing of steep mountain slopes, mining of sand and gravel from river channels, and, most importantly, construction of dams that block transport of sediment downstream (Figure 1). Dams typically store water by design, and store sediment as an unintended consequence, although some dams have been built as debris basins or sediment-control (sabo) dams [14].

Dams are the principal disruption to the natural continuity of sediment transport through a river basin, because dams trap 100% of the river’s coarse bedload (the gravel and sand that moves along the channel bed by rolling, sliding, and bouncing), and a percentage of the finer suspended load (sand, silt, and mud held aloft in the water column by turbulence) that depends on the ratio of the reservoir storage capacity to the volume of inflowing water on an annual basis. Thus, dam-induced changes in flow regime are typically accompanied by reductions in the river’s sediment load as reservoirs trap sediment, creating conditions of sediment starvation directly below the dam [16]. Storing water and sediment results in changes in flow and sediment load downstream of dams, almost always producing downstream changes in alluvial channel form (e.g., incision, narrowing) and bed material composition (e.g., clogging, armoring) [17]. Reservoir sediment trapping can have severe consequences for the downstream river, which is deprived of its natural sediment load, and for coastal landforms, such as beaches and deltas, which depend on river-supplied sediment (especially sand). As sediment supply is cut off to coastal deltas, deltas subside more rapidly and are eroded by wave energy, as illustrated on the Mississippi and Mekong Deltas [18,19].

As reservoirs trap and begin to accumulate sediment, the trapped sediment can begin to interfere with the functioning of the reservoir, long before water storage capacity itself is seriously compromised [20,21]. For example, small sediment accumulations can interfere with the functioning of hydropower inlets, and, where the river enters the reservoir, the buildup of delta deposits can cause backwater flooding upstream. In quiescent landscapes with relatively low sediment yields, rates of sediment build-up within large reservoirs may be sufficiently gradual that managers are not confronted with serious problems for decades or even centuries. However, in landscapes with high erosion rates and high sediment yields, the rate of sediment accumulation in reservoirs can be rapid, and reservoirs can fill within decades, forcing managers to confront this problem sooner. In Taiwan, a geologically active landscape prone to intense typhoonal rains, the naturally high sediment yield was increased further by land use in steep mountain areas, resulting in very high sediment loads for many rivers, and rapid rates of reservoir sedimentation. These in turn have motivated the belated adoption of sustainable sediment management approaches [22].

2.3. Vegetation Encroachment

Reservoirs with large storage (relative to flow in the river) reduce high flows, decreasing the dynamism of the river channel downstream. The reduced flows downstream of dams may be insufficient to transport sediment delivered by tributaries, resulting in channel aggradation and potentially increasing flooding risk. Gravel beds, formerly mobilized every year or two, may go for years without being moved, allowing riparian vegetation to establish in the active channel, and fine sediments to accumulate within the gravel (so-called clogging). Without frequent mobilization of the bed, the active channel typically narrows by encroachment of woody riparian vegetation and deposition of sediment [23], but not universally. If peak flows are only slightly reduced, but bedload is fully starved by the reservoir, the channel can degrade and become progressively armored and clogged. If both are significantly reduced, the channel may not degrade but instead undergo a significant narrowing due to riparian vegetation encroachment. The complexity of alluvial channel forms depends upon the availability of coarse sediment (sand and gravel) that compose bars and riffles. When the supply of coarse sediment is reduced because of trapping by upstream dams, the alluvial channel form typically simplifies, as bars and riffles are eroded away without being replaced by deposition of such sediments from upstream [17].

2.4. Large Wood

Dams also trap large wood, and, historically, such wood has mostly been collected and ‘disposed’ of (e.g., by burning), because it was viewed as creating management problems, such as debris jams at downstream bridges and blocking of the passage of migratory fishes [24]. In many river systems, large wood is critically important in creating channel complexity [25], and, where wood is denied to the river system below dams, downstream channel complexity can decline as a consequence.

3. Dam Renovation Strategies

Dam renovation strategies and remediation measures can be undertaken to achieve one or both of two broad goals: sustaining reservoir storage capacity over time or sustaining fluvial and ecological processes up- and downstream (Figure 2). Here we describe (1) dam rehabilitation (structural retrofits), undertaken to sustain reservoir function over time), (2) fish passage retrofits (e.g., fish ladders) and (3) reservoir reoperation (e.g., modifying dam rule curves), both undertaken to sustain or restore fluvial and ecological processes up- and downstream, and (4) sustainable sediment management to restore sediment connectivity, which can both sustain reservoir storage capacity and restore/sustain fluvial and ecological processes up- and downstream. The latter three strategies include opportunities for environmental improvement listed in

Table 1. Structural opportunities during renovation of dams to improve environmental performance and reduce impacts. (Adapted from [4]).

Opportunity	Description
Wildlife (‘fish’) passage devices	These devices provide passage for wildlife around barriers, such as dams (e.g., fish ladders, rock ramps and by-pass passages, and fish locks). The design is critical for the effectiveness [26,27].
Fish-friendly or aerating turbines	These reduce mortality rates for fish passing through the turbines and improve water quality [28].
Thermal pollution mitigation devices	Expensive multi-level off-take towers can draw water from different levels in the dams, releasing water close to ambient/natural temperature of improved quality. Cheap devices (e.g., propellers, adjustable pipes or curtains) are less widely used and have high operating costs [29].
Water-release structures	Many dams, particularly hydropower dams, were built with small outlet valves as most river flows were to be diverted. Decisions to increase environmental flow can be difficult to achieve because the dam cannot release enough water [29]. Retrofitting larger outlets that increase turbine and spillway water-release capacities is possible (e.g., rebuilding of the Jindabyne Dam in Australia).
Sediment flushing	Bypass tunnels and bottom outlets can restore sediment continuity through/around dams by moving sediment around or through the reservoir, delivering it to the downstream river channel. There are dual benefits: to the life expectancy of the reservoir (less sedimentation) and to river health downstream (mitigating sediment starvation in downstream reaches) reducing channel erosion, maintaining sediment supply to support coastal landforms, such as deltas, and maintaining nutrient loads associated with fine sediments [30].
Reregulating dams	Hydropower dams often generate peaking power, which requires the sudden release of unnaturally high flows to generate power at times of high demand, often at short notice. These high flows, and the abruptness of flow increases and decreases, have an impact on downstream geomorphology and ecology, and pose a hazard to humans. Reregulating dams downstream can store these ‘peaking’ flows, evening out flow releases downstream [29].
Coordinated dam operations	Coordinated operations of dam cascades may improve services (e.g., power generation), enabling lower dams to reregulate flow and minimise environmental impacts downstream [29,31]. Similarly, flushing of fine sediments can be coordinated from dam to dam in a cascade, allowing sediment to pass downstream through a series of dams without deposition [30].
Removal	Increasingly, unsafe or redundant dams are removed for safety, to restore fish passage, to restore sediment and nutrient transport downstream, and restore continuity for recreational kayaking and canoeing [32,33]

.

3.1. Dam Rehabilitation

The term rehabilitation has been used for structural retrofits, typically of the dam structure or spillway, or by constructing new bypass tunnels. The government of India’s Manual for Rehabilitation of Large Dams [34] states the motivation for dam rehabilitation as follows: “Rehabilitation of large dams is required to counter various deficiencies which develop with time and also to correct inadequacies on account of revisions in various standards/guidelines. Deficiencies that are caused primarily by the ageing of a dam include degradation caused due to weathering, wear and tear of equipment due to normal use or misuse, loss of serviceability with prolonged operation, damage from natural events including floods, earthquake or landslides, damage from vandalism and war, etc.” Specific measures include seepage control, correcting problems in embankment dams, such as cracking, piping, excessive seepage, and disturbed riprap, fixing or improving spillways, increasing spillway capacity to account for new, higher estimates of the design discharge, improved dam safety implementation and implementation of emergency and disaster management plans.

A number of projects termed ‘dam renovation’ in the literature fall under the category of rehabilitation. A case study on a tributary to the Yangtze River, described by Zheng et al. [35] involved analyses to determine the dam’s vulnerability to mass movement failure, and measures such as reshaping the dam’s downstream slope to improve its stability, installation of anti-seepage grout curtains, and improvements to the spillway. Similarly, Roux et al. [36] describe the ‘renovation’ of an underground gallery of the Génissiat Dam on the Rhone River, France. The gallery is used regularly to flush the reservoir, and, to counteract abrasion by sand particles, the tunnel walls were strengthened, as described further below in a case study of sustainable sediment management. An unusual use of the term renovation was on a historic hydroelectric powerplant on the Shoshone River, Wyoming. The powerplant was to be returned to its original (1922) condition, but made operable so it can generate power needed locally [37].

Some deformation of concrete is inevitable over time, though the rate will be much higher in some climates and with some water chemistry profiles. Many of the case studies of dams requiring rehabilitation described in the literature are caused by two chemical reactions that commonly result in slow expansion (swelling) of concrete: alkali-aggregate reactions and internal sulfate attacks. While these are distinct chemical reactions, they have a similar effect on the structural behavior of dams, described as “delayed start of a slow rate expansive process that influences the dam equilibrium in time” [38]. The swelling process is normally so slow that the first evidence is apparent only after 20–30 years, often longer. As summarized by Amberg [38], “The concrete expansion is a very complex phenomena, which is influenced by many parameters: moisture, temperature, confinement, creep, alkali content, oxygen (for ISA), aggregate dimensions and shape, pore characteristics and discontinuities in dams, e.g., lift or contraction joints”. It is difficult to replicate under laboratory conditions, and theoretical knowledge needs to be improved. However, once recognized, it is critically important that the concrete swelling be addressed. As summarized by Amberg [38], remedial works to mitigate expansion or slow the reaction can include installing an impervious membrane on the upstream dam face to prevent infiltration of water, grouting cracks with cement grout or epoxy resins to enhance the structural continuity, reducing leakage or increasing sliding safety on the lift joints, improving drainage and/or grouting curtains to reduce uplift pressure and seepage, reducing temperature, and slot-cutting to relieve compression.

3.2. Fishway Retrofits to Provide Passage for Migratory Fish

Fishways (or fish passes) are structures to provide safe and timely movement past an obstacle. One of the key tools in the dam renovation ‘toolbox’ (Table 1), fishways encompass engineered structures, such as the well-known traditional fish ladders consisting of a series of pools stepping gradually upwards, as well as new technologies, including more ‘natural’ bypass channels and fish ramps that climb in elevation parallel to the dammed river channel, and structures specific to certain species and migratory patterns, such as eel ladders [9]. Despite an increase in the construction of fish ladders, their performance is poor in many settings, attributable at least in part to a lack of knowledge about passage needs for species other than salmonids and outside of coldwater environments [9]. Fishway technologies now include unconventional approaches, like fish lifts, such as on the St. Stephen Dam on the Santee River, South Carolina, and the Milford Dam on the Penobscot River, Maine, USA [39,40]. Still more unconventional is the “salmon cannon” used on the Yakima River in Central Washington State, USA, which shoots adult salmon through a flexible tube filled with an air-water mix over a dam and into the reservoir above, as documented in a CGTN Global Business report accessible as a video on YouTube (https://www.youtube.com/watch?v=eGzdOpCisnQ&t=6s (accessed on 16 April 2022)). Not all migrating fish use a fish ladder, so, at best, ladders can be only partially successful. It is generally easier to attract upstream adult migrants to enter a fishway than to direct downstream migrants into fishways, especially larvae and juveniles, who are not strong swimmers and mostly drift with the current. When they enter a reservoir, they find no distinct downstream current, become disoriented, and are easy prey. Moreover, fish pooling around below the entrance to fishways make easy targets for predators, and, once in a fishway, such as a traditional fish ladder, fish are vulnerable to poaching if the site is not monitored.

Perhaps the most problematic aspect of fishways is the lack of knowledge of the passage needs of fish outside of the well-studied cold-water salmonids. This is a critical oversight given the boom in construction of hydroelectric dams in tropical rivers. Even if dam operators seek to incorporate fishways into their dams, they lack the information needed to design effective structures [9].

3.3. Reservoir Reoperation

Reservoir re-operation encompasses opportunities and strategies to modify dam operations to restore natural flow regimes and associated ecosystem health and services. As argued by Richter and Thomas [41], a number of case studies have illustrated the potential recovery of ecological function through “carefully targeted, partial restoration of natural flow regimes”. Using an “environmental flow vocabulary” of five flow types—extreme low flows, low flows, high-flow pulses, small floods, and large floods—changes to these flow categories can be assessed. Typically, each of the river components is affected differently based on the way the dam is operated. In this context, Richter and Thomas hold that “it is critically important to develop testable hypotheses linking river flows to ecological or social conditions, which are subsequently translated into well-defined goal statements. This provides the basis for assessing the success of the re-operation project. However, it is no small challenge to identify appropriate indicators of success” [41]. Strategies for re-operation include modification of flood control facilities to permit seasonal inundation of floodplains for environmental benefit, which, in some cases, will require enlarging outlet works to pass a higher percentage of current flows. Reoperation can provide an important path to achieving improved environmental performance of dams (Table 1).

3.4. Sustainable Sediment Management

Many reported examples of dam renovation have been largely driven by sediment issues, such that the motivation for intervention is either vanishing water storage capacity in the reservoir or sediment starvation downstream. The rehabilitation measures are focused largely on allowing sediment to move through/around dams or otherwise restoring reservoir capacity, and/or compensating downstream reaches for the sediment supply lost to trapping by dams. We first review the artificial supply of sediment to reaches below the dam and the deliberate release of high flows from the dam to mobilize and redistribute the sediments. Next, we review options for using the river’s energy to transport the sediment through/around the dam.

3.5. Morphogenic Flows and Sediment Augmentation

To mitigate dam-induced impacts, controlled high-flow releases designed to mimic the action of natural floods are increasingly required in licenses for dams and as part of programs to restore river function. These deliberate, high-flow releases thus constitute one component of environmental flow requirements for the maintenance of aquatic and riparian habitat, and reflect an evolution of environmental flow requirements from simple minimum flows, to include periodic high flows to mimic flood effects on channels or on ecological processes [42,43]. Various names have been applied to these high-flow releases (e.g., flushing flows, channel maintenance flows), but Loire et al. [44] propose using morphogenic flows as better reflecting the explicit geomorphic objectives for the flow releases. The magnitude of morphogenic flows can be set by calculating the flow providing the tractive force needed to mobilize the bed sediments, or by specifying a flow of a given return interval (such as the two-year flow) assumed to mobilize the bed [42]. Release hydrographs for wet, normal, and dry years may be specified, prescribing the magnitude, duration, and pattern of morphogenic releases. Such morphogenic flow releases are usually for ecological purposes, but they can also be implemented for other objectives, such as risk management (e.g., maintaining channel capacity to prevent overbank flooding or bank erosion, preventing reservoir siltation). To restore habitat lost to sediment starvation and the lack of wood downstream of dams, restoring flows alone may not achieve the objectives. Sometimes it is better to introduce coarse sediment and wood, as has been done in many rivers mechanically, or by promoting bank erosion. Coarse sediment added to downstream channels is commonly not taken from the reservoir itself due to consideration of land ownership, technical challenges, or the objections of nearby residents to the truck traffic that would be generated.

There is a consensus in the scientific literature about the necessity to release morphogenic flows (e.g., [45,46]). The last two decades have seen morphogenic flows prescribed to restore river channels by mimicking the effects of natural floods [42,45,47]. While some studies have emphasized the interaction of water flows with sediment [48], such as sediment supplied from downstream tributaries (e.g., [49]), most published work has focused on flow releases needed to accomplish geomorphic goals. Even if a post-dam flow regime were to mimic precisely the pre-dam flow regime, the river system would be severely altered by the loss of its sediment load [42,48]. Thus, increasingly, partial restoration of sediment load is prescribed along with morphogenic flows [17,30,50]. Coordinating morphogenic flows with sediment augmentations has seldom been reported, but is likely to become more common (e.g., [51]).

3.6. Designing Dams to Pass Sediment

As a more sustainable alternative to costly and energy-intensive mechanical gravel augmentation, it is often possible to use gravity to deliver sediment to the channel downstream of one or more dams by passing sediment through or around the dams, for which a range of techniques can be used (Figure 3) [21,30,52,53,54]. The summary below is drawn primarily from [30].

For smaller dams, the most sustainable approach (where feasible) is to pass the sediment load around or through the dam. River water can be diverted to an off-channel reservoir during lower flows (in between floods) when the water is relatively sediment free, while allowing sediment-laden floodwaters to continue down the main river channel. To avoid deposition of sediment in mainstem reservoirs, a sediment bypass can divert part of the incoming sediment-laden waters into a tunnel that is routed around the reservoir, so the sediments never enter the reservoir at all, but rejoin the river below the dam (Figure 3). Sediment can also be sluiced, by maintaining sufficient velocities through the reservoir to let high flows pass through without slowing enough to permit sediment to deposit. Alternatively, the reservoir can be drawn down, to scour and re-suspend sediment in the reservoir and transport it downstream (i.e., flushing). This involves the complete emptying of the reservoir through low-level gates. In some cases, when flows with high suspended sediment concentrations enter a reservoir, they can remain intact as denser turbidity currents flowing along the bed of the reservoir at the bottom of the water column, without mixing with the overlying, less dense waters. Density current venting involves opening dam bottom release outlets to allow such density currents to exit the reservoir via the outlets, carrying most of their sediment with them. One advantage of this approach is that the reservoir need not be drawn down, which makes the approach suitable for reservoirs with year-to-year carry-over storage. Sluicing, flushing, and density current venting all pass only those sediments that are in suspension, which tend to be the finer fractions of the sediment load, but can include significant amounts of sand.

Sluicing and flushing work best on reservoirs that are narrow, have steep channel gradients, and have a storage capacity that is small relative to the average annual discharge of the river. Otherwise, backwater zones might form in wider reservoirs where the hydrodynamic forces are insufficient to mobilize sediment. As shown by Sumi [56] (2008), flushing has been effective on reservoirs that impound less than 4% of the mean annual inflow. Larger reservoirs, with year-to-year carry-over storage, are poor candidates for such sediment pass-through approaches.

It is generally most efficient to take sediment management and environmental flow regimes into account at the outset of the design and to plan the operation of dams so that dams are properly equipped to successfully sluice (or flush) sediments (e.g., with sufficiently large low-level outlets), and the operations are planned to account for some periods of reduced power generation (or other functions) to allow sediment to be passed. In dam renovation, retrofits to allow sediment passing through existing dams may be possible, but often raise dam safety concerns. Bypasses can be safely built around existing dams without threatening the integrity of the dam. Despite the demonstrated effectiveness of these measures, incorporation of sustainable sediment management is, unfortunately, the exception rather than the rule in dam planning and design globally. The infrequency with which these measures are employed can be attributed to the additional costs to construct dams with the needed outlets or bypass tunnels, potential lost power revenue from operating dams to pass sediment, and the fact the standard economic analyses conducted for dam planning use discounting rates that render future costs from sedimentation irrelevant to current decision making [21]. Sediment bypasses, for example, have been employed at individual sites worldwide, with technologically advanced Japan and Switzerland having five tunnels each [57]. In its Guidelines for Mainstem Dams, the Mekong River Commission calls for use of sustainable sediment management technologies to allow suspended sediment to pass through dams to downstream reaches [58]. Plans for the dam built on the mainstem of the Lower Mekong at Xayaburi included sediment management measures, although an expert review panel questioned the likely effectiveness of these measures [59].

4. Case Studies

4.1. Introduction to the Case Studies

Dam renovation (i.e., repair of deficiencies, re-operation to improve environmental performance, etc.) is possible in many cases, and it is often possible to retrofit poorly-designed dams to improve their environmental performance (e.g., through installation of fish ladders, ‘fish-friendly’ turbines that have a lower mortality rate, or aerating turbines), but these retrofits are usually more expensive and less effective than if the dam had been initially designed with these functions taken into account. Potential measures to improve environmental performance of dams in dam renovation are presented in Table 1.

We focus on examples of dams that have been renovated to improve their performance economically, energetically, socially, and environmentally. The case studies begin with examples of the rehabilitation of dams for which repairs, or upgrades, were needed for safety reasons, such as to address seismic risk or to repair deteriorating concrete, then case studies of fish passage are described, then examples of reservoir reoperation to improve environmental performance are considered, and, finally, a case study of sustainable sediment management is presented.

4.2. Rehabilitation of BF Sisk Dam (San Luis Reservoir), California USA

The BF Sisk Dam (formerly San Luis Dam) is a 115m-high earthfill dam along the west side of the Central Valley of California in Merced County. Completed in 1967, it impounds runoff from small local drainages (Pacheco and Portuguese Creeks), but its principal purpose is to provide off-channel storage for the large federal and state water projects, the Central Valley Project and California State Water Project. Both projects divert from the southern portion of the inland delta of the Sacramento-San Joaquin River system and convey waters southward through canals along the western margins of the Central Valley. The massive capacity of the San Luis Reservoir (2.5B m³) was built to provide storage in the water supply system from winter to summer and to mitigate interannual fluctuations in water availability. It is the largest off-stream reservoir in the US.

The dam is near the San Andreas Fault so potential seismic risk was recognized when the dam was constructed. However, with the identification of the Ortigalita Fault, which runs through the reservoir, and the recent re-evaluation upward of its potential to generate earthquakes, and reanalysis of foundation data indicating some sections to be weaker than originally assumed, a corrective action study was initiated in 2007 [60].

In 2008, a series of scoping meetings was held with stakeholders and members of the public, and field studies initiated of the dam’s geotechnical features and potential sources of construction materials (‘borrow sources’). From the geotechnical perspective, there was recognition that the Ortigalita Fault was longer than originally believed, and thus capable of generating larger earthquakes than had originally been assumed. Studies also recognized the potential for slopes around the reservoir to generate large landslides into the reservoir, which, in turn, could generate waves large enough to overtop the existing dam, The selected ‘preferred alternative’ was to raise the height of the dam to increase the available freeboard, which, in this case, could be done with relatively minor impacts because the lands surrounding the reservoir subject to inundation were uninhabited and mostly publicly owned. In addition, downstream berms and shear keys were to be added to stabilize the embankment, along with the installation of a new “zoning filter along the downstream side of the crest to protect against cracking” [61].

4.3. Rehabilitation of Hydropower Dams Southwestern Connecticut, USA

Deterioration of concrete in 21 hydroelectric dams operated by the Bridgeport Hydraulic Company was addressed by a range of methods, as detailed in Bernard [62]. The dams, built in the early 20th century, exhibited both freeze-thaw and seepage damage, with concrete spalling evident in many sites. Bernard [62] describes the process of identifying the issues specific to the Easton and Means Brook dams, and the process of selecting fixes appropriate to each dam, a process he terms “renovation”.

4.4. Rehabilitation of Pian Telessio and Isola Dams, Italian and Swiss Alps

The Pian Telessio Dam, a concrete arch dam in the Italian Alps was affected by swelling concrete, experiencing a total expansion of almost 250 μm/m, illustrative of the problems of unfavorable loading conditions for arch dams at low reservoir levels, “when compressive stress in vertical direction at the upstream dam heel reached values up to 15 MPa” [38]. By 2007, the dam had reached 60 mm of permanent displacement in the upstream direction [63]. To rehabilitate the dam required “systematic slot-cutting within the upper half of the dam”, i.e., removal of a small slot of concrete to provide space for the concrete to expand into [38]. Similarly, Isola Dam in the Grison Alps of Switzerland experienced concrete swelling, with upstream displacement of the dam of about 60 mm. The solution chosen to remediate the problem and relieve stress (termed ‘renovation’) was, likewise, to remove a thin slot in the arch dam (when the reservoir is drawn down in the winter) using a diamond-wire cutting saw [64,65].

4.5. River Murray Fishways, Australia

On the River Murray, fish migration was blocked by multiple dams, contributing to a 20% decline in fish biomass. Relicensing for these dams triggered an assessment of dam safety, and this, in turn, triggered dam renovation, and in the process, other improvements required by current law that were not required when the dams were first built [4]. The result was an AUD 45 million program (2001–2010) to add fishways to 12 weirs and 5 barrages, reconnecting fish passages over 2225 km, called the ‘Sea to Hume Dam’ program [66].

Prior to installation of the fishways, extensive consultations were carried out with Aboriginal groups, and with heritage and environmental agencies, to avoid impacts to important cultural sites and to other species and ecosystem functions. Project consultation extended to other stakeholders, river management agencies, and the public, to gather feedback on project elements and to increase acceptance of the proposed works by the public and stakeholders. Interestingly, the trigger for the program “was not biodiversity conservation but a safety audit of dams, revealing some failures to meet occupational health and safety standards. These safety renovations triggered state laws, adopted after the dams were built, requiring fish passage on new instream infrastructure” [4].

4.6. Penobscot River, USA

On the Penobscot River in Maine, northeastern USA, populations of migratory Atlantic salmon (Salmo salar) and American shad (Alosa sapidissima) had fallen precipitously, due in large measure to the presence of eight dams on the lower mainstem Penobscot, which blocked access to headwater spawning areas for the fish. When the power company, PPL Corporation, applied to the Federal Energy Regulatory Commission to renew the license for hydropower production from the cascade of dams, it triggered a relicensing process that involved negotiations with environmental NGOs, indigenous tribes, and relevant government agencies. The parties reached agreement in 2004 to restore fish passage by removing the two lowermost mainstem dams and retrofitting the remaining dams to allow fish passage. Through capacity and operational changes, these renovated dams will increase their electricity generation such that total hydropower production from the basin will remain the same or increase slightly [31]. The project will expand the range of habitat accessible to migratory fish from 60 to 615 km and is thus expected to increase fish populations over time [67].

In both the Murray and Penobscot River cases, the relicensing process triggered a range of requirements, and highlighted the potential of dam reoperation and renovation to increase both economic and environmental benefits of the dams. The Penobscot River project “illustrates that a basin-scale approach can potentially yield more comprehensive solutions for sustainable hydropower than can be achieved at the project scale…such large-scale planning processes can improve the sustainability of both regulatory licensing of existing dams as well as the planning of future dams in regions undergoing the expansion of water-management infrastructure.” [31].

4.7. Pelton Fish Ladder, Deschutes River, USA

The construction of the Pelton Dam on the Deschutes River in central Oregon was controversial because of its anticipated impacts on the river’s abundant wild salmon runs. From the moment the project was announced in 1949, fishing interests actively opposed construction of the dam. The dam was to produce peaking hydropower, which would cause rapid unnatural fluctuations in river stage downstream, and the dam would cut salmon off from excellent spawning and rearing habitat upstream. The dam was ultimately built in 1956–1958, but the proponent, Portland Gas and Electric, was required to build a small reregulating dam below Pelton to even out the flows released to downstream reaches. The utility also built a 5-km fish ladder, the world’s longest, to allow fish to bypass both the reregulating dam and the main Pelton Dam to reach the river and tributaries upstream [68]. Adult salmon successfully ascended the fish ladder, but juveniles were unable to descend, because once they reached impounded waters of the hydropower reservoir and had no distinct downstream current to follow, they would become lost and easily eaten by larger predatory fish. Moreover, in 1964, the utility constructed another dam 11 km upstream, which reduced the area of habitat accessible to upstream migrants who negotiated the Pelton fish ladder. In 1968, the utility ceased using the Pelton fish ladder as a fish ladder, acknowledging the fish passage project as a failure [68]. The Pelton experience illustrates the challenges of using artificial measures to restore longitudinal connectivity in rivers with anadromous fish runs, and the common problem of finding a solution for upstream adult migration but failing to provide for downstream migration.

4.8. Savannah River Reservoir Reoperation, USA

The US Army Corps of Engineers operates three large dams on the Savannah River (which forms the border between Georgia and South Carolina). The lowermost dam is the Thurmond Dam, which is operated primarily for flood control. As a result of a collaboration between the Corps, the Nature Conservancy, and the Natural Heritage Institute initiated in 2002, and after engaging in a comprehensive stakeholder outreach exercise, the Thurmond Dam was successfully re-operated. The dam now releases high-flow pulses to “benefit fish spawning and access to low-lying floodplain areas, flush oxbow lakes, and disperse the seeds of floodplain trees” [41].

However, the goal of restoring frequent, small floods has faced some obstacles. The Thurmond Dam has done its job of flood control very well, perhaps too well, in reducing the magnitude and frequency of floods large and small. Pre-dam frequent floods had ranged from 2500 to 6000 m³ s⁻¹. While the total annual flow has been reduced by only 15% by diversions, flows have been so effectively evened out over the year that even minor pulse flows exceeding 450 m³ s⁻¹ have occurred only rarely post-dam. Since construction of the dam and shrinking of the area expected to be inundated in floods (and thus under any land-use restrictions), there are now houses built on the floodplain that would be inundated at about 1500 m³ s⁻¹, and a public facility in the city of Augusta (located along the river) that would be flooded at 1000 m³ s⁻¹ [41]. One proposal was to construct a bypass for floodwaters around the city, a strategy that has been implemented successfully in urban areas elsewhere [69]. In the meantime, the deliberate high-flow releases have been limited to 850 m³ s⁻¹. The way the Corps achieves these small high-flow releases is worthy of explanation. Prior to releasing a high flow, the Corps allows reservoir levels to climb about 0.5 m above the normal target level for the season (i.e., above the ‘rule curve’) to store water to support the release. While this involves encroaching on the flood pool, the encroachment is minor, and, if a large storm is predicted to affect the basin at the time, the reservoir can be dropped back to its normal level and the high-flow pulse can be postponed. As Richter and Thomas [41] note, “This case study illustrates the fact that certain aspects of environmental flow restoration can be accomplished with little to no infringement on other existing water uses or dam purposes”.

One of the specific goals of the Savannah River dam reoperation was to provide adequate flows for shortnose sturgeon (Acipenser brevirostrum) to migrate upstream to spawn. The sturgeon’s movements are monitored to provide feedback on the adequacy of flow releases for fish migration. One result of this monitoring was evidence that the fish were responding, not only to levels of flow and velocity, but also to water temperature, so, in response, subsequent releases were timed to coincide with suitable water temperatures [41].

4.9. Roanoke River Reservoir Reoperation, USA

The Roanoke River (Virginia and North Carolina, USA) provides a good example of reservoir reoperation involving a change in release patterns from hydroelectric dams to prevent artificially prolonged inundation of the river’s floodplain during seasons when tree seedlings would normally sprout in moist, but well drained, soils of the recently flooded floodplain. By artificially maintaining high flows and inundating the floodplain during this period, young trees were drowned and unable to establish. The operator of two dams on the river (a private hydropower company) agreed to reduce the frequency of these artificial pulses and to allow sufficient time between them to allow the floodplain soils to drain. Fortunately, the utility could make up for lost power from this change through other hydropower operations in its system. In addition, negotiations expanded to involve an upstream dam operated for flood control by the US Army Corps to explore options to release small floods of around 1000 m³ s⁻¹ to restore system dynamics [41]. A 2017 settlement agreement established a new (more ‘environmentally friendly’) flow regime, which was informed by results of an adaptive management program over the preceding decade [67].

4.10. Normandy Dam, Duck River, USA

The Normandy Dam provides a good illustration of a relatively small dam reoperation/retrofit that yielded substantial ecological benefits. The Normandy Dam was completed in 1976 on the Duck River, Tennessee. It is the largest Tennessee River Valley (TVA) dam that does not generate hydroelectric power. Rather, it was created for water supply and flood control (with a capacity of 156 M m³) [70]. Two endangered freshwater mussels inhabit the Duck River, one of which is the Cumberland monkeyface mussel (Theliderma intermedia), formerly widespread throughout the Upper Tennessee River system, but since 1970 found only in the Clinch, Powell, Tellico, and Duck Rivers [71]. Despite being within the historical range of the species, surveys of the Duck River downstream of the dam did not encounter the mussels, which was hypothesized to result from low dissolved oxygen levels in waters released from the dam. In response, beginning in the 1990s, the TVA retrofitted the dam to oxygenate the waters released from the dam, resulting in a population “in the tens of thousands, extending for 50 river kilometers” by the 2000s [72].

4.11. California System Reoperation Program, USA

The California Department of Water Resources, as directed by the state legislature in 2008, launched a study to identify options for the reoperation of the state’s existing flood protection and water supply systems. The system reoperation program and study was intended “…to identify opportunities for enhanced systems efficiencies through coordinated operations between the SWP [State Water Project] and CVP [Central Valley Project, a federal project] and between water supply and flood management infrastructure. California can do much more with its existing water infrastructure by taking advantage of the physical interconnections (and enhancing them) while also operating the system in a coordinated manner to optimize the benefits” [30]. Much of the effort has been devoted to evaluating the potential for releasing high flows that can inundate floodplain areas and make possible recharge of the alluvial aquifer, termed “managed aquifer recharge”.

4.12. Sustainable Sediment Management at Verbois and Génissiat Dams, Rhône River, Switzerland and France

The Verbois Dam is located on the Rhône River about 10 km downstream of the city center of Geneva. The mainstem Rhône drains Lake Geneva, so its waters are generally clear and sediment-free, but only 1.5 km downstream of the outlet from the lake, the Rhône is joined by the Avre, a river whose headwaters are located on the slopes of Mont Blanc, and which carries a large load of sediments derived from the alpine catchment. The Verbois Dam, a 34-m high gravity dam impounding 12 million m³, traps much of this sediment, and aggradation of the reservoir and riverbed upstream create a risk of backwater flooding to the urban Geneva. To avoid this problem, sediment is flushed from the Verbois Dam regularly. However, sediment flushing must be coordinated with downstream dams, notably the Génissiat Dam, located about 25 km downstream in France. The management of the Verbois and Génissiat dams illustrates important aspects of dam renovation relating to both operational changes and structural retrofits.

The first consideration is managing sedimentation to avoid loss of reservoir capacity or induce backwater flooding, so sediment flushing is coordinated [30]. Of the 19 dams on the French Rhône River from the Swiss border to the Mediterranean Sea operated by the Compagnie Nationale du Rhône (CNR), all but one are run-of-the-river dams that operate by short-circuiting the “old river” with a straight canal, leaving abandoned meander bends with greatly reduced flows, some of which have been the loci of ecological restoration efforts [73]. The exception is the Génissiat Dam, completed in 1948, a 104-m high concrete gravity dam across the mainstem Rhône impounding 56 M m³. Using water storage in Lake Geneva, sediment is flushed from Verbois downstream to Génissiat by flushing, such that the opening of gates is coordinated from dam to dam as a pulse of sediment moves downstream.

The flushing from the Génissiat Dam occurs every three years, and requires mobilizing a staff of about 400 for approximately 10 days to implement the flushing and to monitor sediment concentrations and other metrics, at a cost of about EUR 1.4 million, based on the 2003 flushing [74]. The approach is termed “environmentally friendly flushing,” and is designed to limit the potential impacts of flushing on downstream aquatic life, water supply intakes, and restored side-channel habitats. This approach is of particular interest because this flushing is conducted under extremely strict restrictions on turbidity and suspended sediment concentrations, not to exceed 5 g/L, on average, over the entire operation, and not to exceed 15 g/L over any 15-min period [74]. The dam is equipped with outlets at three levels: a bottom gate, an outlet halfway up the dam, and a surface spillway. Concentrations are controlled by mixing waters with high sediment concentrations from the bottom of the water column with enough “cleaner” water from higher in the water column (normally via the mid-level outlet) to stay within the required concentrations.

In the four decades during which flushing was conducted every three years, an estimated 23 million tons of sediment would have been expected to deposit in the Génissiat reservoir by 2003 absent flushing, but only 4.5 million tons have actually deposited. To remove an equivalent volume by dredging would have been far more costly than the flushing operation [30,75].

To support the operation of the flushing, the Génissiat Dam required rehabilitation, i.e., a structural retrofit of its bottom outlet. The outlet is a tunnel originally used to bypass the river to allow for construction of the dam, but is now used to flush sediments from the reservoir. However, passing the sediment laden water through the tunnel resulted in abrasion, so that a renovation project was undertaken to increase the resistance of the tunnel to abrasion. The renovation specifically involved “extending the thick metallic armoring of the valve and setting up of a new high wear-resistant layer of ultra-high performance fiber reinforced concrete (UHPFRC) on all other surfaces subjected to high hydraulic stresses.” [36]. Abrasion is an issue in all sediment bypass tunnels and has been dealt with through use of high-resistance coatings on the tunnel surfaces. In one Swiss dam, the bottom of the tunnel is paved with granite slabs to serve as an abrasion-resistant surface (Prof Robert Boes, ETH, personal communication, 2021).

Although there are 19 dams on the French Rhône, flushing is coordinated only downstream through Génissiat. On the Lower Rhône, storage is lacking, and, while it would theoretically be possible to coordinate flushing with high tributary inflows, disruptions to navigation must be arranged a year in advance, so flushing is not attempted, and, instead, sediment is removed mechanically from these downstream impoundments [76].

4.13. Sediment Management at Sanmenxia Reservoir, Yellow River, China

The Yellow River is China’s second largest, with a drainage area of 750,000 km² and a length of 5460 km from its headwaters in the Tibetan Plateau to its mouth in the Bohai Sea. The river takes its name from the color imparted by its notoriously high sediment loads, driven mostly by high erosion rates in the 640,000-km² Loess Plateau in the central part of the river basin, attributable to both the inherent erodibility of the loess soils and to its land use history. Concerted efforts over five decades to reduce erosion rates and to retain sediment in the plateau through soil conservation measures and check dams have resulted in a remarkable decrease in sediment yield from the Loess Plateau and the sediment loads in the Yellow River [77]. However, by virtue of the inherent erodibility of the landscape, sediment loads remain high compared to many other rivers globally.

The Sanmenxia Dam was constructed 250 km upstream of Zhengzhou on the Yellow River in 1957–1960. Its planning and design did not account for the river’s high sediment loads. The Sanmenxia controls 89% of the basin’s runoff and 98% of its sediment load. During the first 18 months of the dam’s operation, the reservoir trapped 93% of incoming sediment load and consequently lost 17% of its water storage capacity (below elevation 335 m) of 9.75 × 10⁹ m³ [78]. In response to sedimentation in Sanmenxia Reservoir, sustainable sediment management measures were undertaken, and in addition, the outlet structures were rehabilitated to resist abrasion from sediment passing through.

To reduce trapping of sediment, the dam operation was changed to detain water only during the flood season, but the dam’s ability to release water at other times was limited due to the small outlets. As a result, by October 1964, a volume of 4 × 10⁹ m³ of sediment had accumulated in the reservoir, representing a 41% loss of water storage capacity (below elevation 330 m). The riverbed upstream of the reservoir was aggrading and backwater flooding was threatening the city of Xi’an, located on the Weihe River, a major tributary joining the Yellow River about 100 km upstream of the Sanmenxia Dam. To increase the discharge capacity, the outlet structures were improved in two stages. In the first stage (1966–1968), two tunnels were excavated along the left bank and four penstocks were repurposed to sluice sediment. In the second stage (beginning in 1970), eight bottom outlets were repurposed to sluice sediment. To allow hydropower generation at the lower river level during flood stage, penstock intakes were lowered, and new generating units were installed. This resulted in an overall increase in outlet capacity from 3180 m³ s⁻¹ to 9300 m³ s⁻¹ [78]. However, the bottom sluices sustained damage from abrasion and cavitation. The repairs (1984–1988) involved lining the sluices with abrasion-resistant material, which constricted their area and reduced the discharge capacity by about 470 m³ s⁻¹. To compensate for this, two new bottom sluices were installed in 1990 [78]. The dam now has 27 outlets (not counting penstocks) with a capacity of 9700 m³ s⁻¹ (Figure 4). Since 1973, it has been operated in a controlled release mode, wherein the dam stores clearwater during non-flood months (November-June) but passes sediment-charged water during flood months (July-October). Since adoption of these sediment passing measures, the accumulation of sediment in the reservoir has basically stopped.

Through adoption of the controlled release mode of operation, the Sanmenxia Reservoir has maintained a flood control capacity of around 6 × 10⁹ m³, and has thereby reduced the peaks of floods downstream on the Yellow River [78].

5. Conclusions

5.1. Overview of Case Studies

The case studies illustrate the diverse and often complementary objectives and measures used in dam renovation. The rehabilitation project at the Fisk Dam was motivated by a reappraisal of risk to the dam from earthquakes and landslides into the reservoir, while the Connecticut and Alpine cases both involved deterioration of structures from concrete expansion and other factors, such as seepage. All rehabilitation projects were motivated by safety and maintaining the functionality of the dams into the future. The Murray and Penobscot River case studies both involved installation of multiple fishways in programs to restore fish passage on the river basin scale, both of which proved successful based on the numbers of fish that have migrated through the dams since. By contrast, the Pelton Dam fishway case study documents a failed attempt to mitigate for the disconnection of the spawning and rearing habitats of the upper basin by the Pelton Dam. Despite its unusual length (5 km), adult upstream migrants followed the Pelton fish ladder to reach the reservoir upstream of Pelton Dam, but downstream, migrating juveniles were unable to pass through the reservoir, and the passage project was ultimately acknowledged as a failure.

Both the Savannah and Roanoke River case studies involved reservoir reoperation to partly naturalize the annual hydrograph, specifically, to release high flows sufficient to partly reactivate the floodplain (Savannah) and to cease maintaining unnaturally high water levels that were drowning the seedlings of riparian trees (Roanoke). The Duck River case study, by contrast, involved modifying the temperature of waters released from the dam by changing the reservoir level from which waters were released, thereby making conditions suitable for freshwater mussels, whose populations rebounded in the reach below the dam. The California dam reoperation case study reports on a programmatic initiative to reoperate dams to inundate floodplains and increase recharge into floodplain soils.

The Verbois and Génissiat dam case study involves, first, a sustainable sediment management program to preserve reservoir capacity (and to avoid flooding of urban Geneva), coordinated between the upstream and downstream dams. It also involves rehabilitation of the outlet tunnel of the Génissiat Dam, which experienced cavitation during periodic flushing of the reservoir. Passing sediment through the dams both maintains reservoir capacity and supports fluvial processes in the reaches downstream of the dams, thereby contributing to both broad goals of dam renovation.

The case studies vary widely in setting, scale, problem confronted, renovation strategy employed, and cost. Costs were not reported for most dam renovations described in the literature, but data available for some provide an indication of the costs. For example, the rehabilitation of the Chimoni Dam in Kerala to address seepage problems cost INR 94 million (about USD 1.2 M) [34]. For the US Bureau of Reclamation BF Sisk Dam in California, the ‘crest raise’ alternative cost was USD 830 M, while the crest raise with shear key cost USD 1134 M [79]. Dealing with sediment accumulated in reservoirs can also be costly. Mechanical removal of sediment typically costs a minimum of USD 2.50 per m³ to remove, as reported in 1998 [54], with costs increasing with distance to the disposal site, the depth of dredging, and the complexity of disposal. Sluicing and flushing can prevent deposition, or remove at least some of the deposited sediment, using gravity in lieu of fossil fuel energy, but these methods entail some loss of hydroelectric generation capacity or other water storage functions. The economic tradeoffs associated with investment in sediment management technologies, especially over the longer term, have been inadequately recognized to date.

5.2. Increasing Importance of Dam Renovation

Reservoir storage capacity is best viewed as a non-renewable resource. If a reservoir is allowed to fill with sediment, its services are not easily replaced by building a new dam because the most favorable reservoir sites have already been developed, and thus any replacement dam will inevitably be less efficient. With ongoing changes in climate, and increased flow variability in rivers, the storage provided by reservoirs will become more important, not less. Thus, preserving dams, their reservoir capacity, and their functions is increasingly important to society. As summarized by Randle et al. [20], “Sustaining the critical functions of reservoirs for future generations requires adoption of a new paradigm of sustainable use, as opposed to the traditional design life concept which ignores consequences beyond a project’s formal planning horizon... The design life approach has critically hindered management of our aging infrastructure, much of which is past or near the end of its design life…”. The design life paradigm is based on assuming a ‘design life’ of typically 50–100 years, without accounting for what happens after that. Even if the costs of dealing with a completely sedimented reservoir were to be factored in, the use of a discount rate tends to trivialize the burden placed on future generations who will inherit deteriorated and increasingly risky infrastructure [20,21].

Dam renovation is undertaken either to sustain reservoir capacity/functions over time, or to sustain/restore fluvial/ecological processes, or both (in the case of sustainable sediment management that restores sediment continuity). Measures include repairs and retrofits to address safety risks, long-term maintenance needs, increased hydroelectric generating efficiency, as well as modifications to structures or operation to meet environmental and social goals. The scale of dam renovation ranges from repair of joints in concrete, up to modifying flow releases from large dams.

The inevitable deterioration of concrete exposed to high stresses, the accumulation of sediment in the reservoir, and other ‘wear and tear’, means that some level of maintenance and retrofitting will be needed to maintain basic dam functions. With changing climate and river basin conditions, the existing infrastructure will be stressed, making significant retrofits more likely at many dams. These safety-related modifications may open the door to broader re-operational requirements.

Separate from renovation actions aimed at keeping a dam functioning, there are many measures that can be taken to improve the environmental performance of dams (Table 1). These include reoperation of reservoirs to yield more favorable conditions downstream for species of concern. We will likely see dam renovation implemented more widely in the future, but each dam has a unique set of opportunities and constraints, for which an informed assessment is required.

While many of the geomorphic and ecological impacts of dams could be reversed by removing dams, it is likely that dam removal will remain the exception rather than the rule. As their functions are increasingly needed, most dams are here to stay, at least over the time scale of decades to centuries, provided they do not fill with sediment, or their structures deteriorate to the point of failure. Thus, repairing structural weaknesses, increasing safety factors, retrofitting dams with fishways, reoperating to improve flow and temperature conditions downstream, and restoring sediment continuity through, or around, reservoirs, can sustain reservoir life expectancy and functions, as well as benefit fluvial and ecological processes in the river reaches downstream of the dam.