Chapter 3
Novel Hydroelectric Storage Concepts
Frank Escombe EscoVale Consultancy Services, Reigate, Surrey United Kingdom
Abstract
This chapter examines the use of pumped hydro techniques in configurations other than well-established pumped hydroelectric storage (PHES). Most are of similar scale, aiming to bring high-power, high-energy storage to regions where PHES is not feasible. The largest geological piston-in-cylinder systems can also address emerging markets that require energy levels beyond the capability of PHES. Future networks, relying more heavily on variable renewables, may require 100 h or more of gigawatt-class storage capacity to harvest large energy surpluses and to ride through prolonged periods of low power generation. Other concepts covered include deep-level reservoir systems and offshore seabed and lagoon storage.
Keywords
pumped hydro
gravitational energy storage
hydraulic accumulator
energy membrane
tidal lagoon
energy island
energy storage market
1. Introduction
1.1. Scope and Purpose
Following on from coverage of conventional pumped hydroelectric storage (PHES), this chapter examines other concepts that share the same principle: using hydroelectric equipment (usually reversible pump-turbines and motor-generators) to convert electrical energy to and from gravitational potential energy. It explores their ability to complement PHES by:
• Extending use into regions where the terrain is unsuitable for conventional PHES or where public opposition/approval difficulties (areas of outstanding natural beauty or villages or agricultural land or conflicting priorities for water resources) make PHES projects impractical.
• Opening up new applications—in particular, to meet future requirements in electricity networks with high concentrations of wind and solar resources that need much longer duration/higher energy solutions than are usually feasible with conventional PHES.
1.2. Constraints
Developers of these novel concepts hope to benefit from the status, technical expertise, and market dominance of PHES, which accounts for >95% of global grid-scale installed capacity. They also have to live within its limitations. The main problem with gravitational storage is gravity, which is incredibly weak compared with the other fundamental forces. This is a good thing for the universe but rather inconvenient for energy storage developers.
As an exercise in the absurd, a record-breaking weightlifter can support an astonishing mass of over 400 kg. If an electricity company could persuade him to store energy by carrying this from sea level to the top of Everest, it could afford to pay him almost $1 for his trouble. A simple time-shifting energy storage service is not worth much to the average electricity supplier—<$100 (MW h)–1. Any form of gravitational storage has to accept some uncompromising facts of physical life.
Kilowatt-hours make a sensible starting point for electrical energy storage calculations. The SI unit of energy is the joule (J), equal to the energy expended when applying one watt of power for one second (1 W s). Therefore, 1 kW h (1000 W for 3600 s) is equal to 3.6 × 106 J (3.6 MJ).
A joule is also equal to the energy expended in moving a force of one newton through one meter (1 N m). A mass of one kilogram exerts a downward force of “g” newtons, where g is the acceleration due to gravity (about 9.81 m s–2):
• In a perfect world, raising/lowering 1 kg by 1 m will store/generate 9.81 J.
• Storing/generating 1 kWh (3.6 MJ) therefore requires 367 tonne-meters (367 t m).
• The world is not perfect. Even with the best turbines, hydroelectric generation needs about 400 t m (tonnes of water × meters of vertical travel) per kilowatt-hour. We might require 450 t m with less efficient turbines.
• For our purposes: each kilowatt-hour from storage requires (400–450) t m (tonnes × meters vertical travel).
It is sometimes more convenient to express hydroelectric energy in terms of its volume/pressure relationship:
• In a perfect world, E = V × P, where E = energy in joules; V is the volume of water passing through the turbine in m3; and P is the pressure in pascals. One pascal (symbol Pa) = 1 N m–2. One megapascal (MPa) = 106 Pa = 1 MPa = 10 bar pressure.
• Storing/generating 1 kW h (3.6 MJ) requires 3.6 m3 MPa (cubic meters × megapascals). Real world hydroelectric generation has to allow for turbine efficiency.
• For our purposes: each kilowatt-hour from storage requires (4.0–4.5) m3 MPa or (40–45) m3 bar.
For comparison, we could retrieve 1 kW h from a battery weighing less than 10 kg. Or we could electrolyze less than 1 kg of water and use the resulting hydrogen to generate 1 kW h in a fuel cell. Or, if only we knew how to use really strong fundamental forces (E = mc2), we could convert our kilowatt-hour to and from 40 × 10–9 g (40 nanograms) of matter—less than 1 mm of fine hair!
PHES is the elephant in the storage room, but it is there for good reasons. Novel hydroelectric storage technologies will also have to offer real advantages if they are to make real progress. This may be in terms of capital and operating costs; cycle life and longevity; or other performance characteristics, including power and energy capability (where PHES has often been seen as the only practical answer).
1.3. How Did We Get Here?
Pumped hydro concepts can trace their origins back to 18th century water towers or to 19th century hydraulic accumulators.
Given enough time, a small steam-driven pump could elevate a considerable quantity of water into an 18th century water tower. The water was then used for intensive operations, by releasing it rapidly through high-power hydraulic machinery. The 20th century saw the adoption of PHES as the dominant method of storing electricity. Conventional PHES uses mountains rather than water towers, but the principle is the same.
Nineteenth century hydraulic accumulators [1] provided attractive alternatives to costly and occasionally fallible water towers. Early systems were vertical cylinders at ground level, sealed at the top by massively weighted pistons. Operating pressures were an order of magnitude higher than available from water towers at that time, with a corresponding reduction in the volume of water required for a given task. The 21st century may see history repeat itself, with the introduction of accumulator-type systems to store electricity in high-pressure, low-level water, without needing mountains and, perhaps, on a scale that eclipses PHES.
1.4. Novel Hydroelectric Storage Categories
The technologies covered in this chapter are grouped in the following sections:
• Piston-in-cylinder storage
• Energy membrane storage
• Novel land-based and seabed pumped hydro configurations
• Offshore lagoon and energy island storage.
The first two categories use pumped hydro techniques to elevate a solid working mass (similar in principle to hydraulic accumulators). The remainder use water (or seawater) as the working mass in configurations that differ in some respects from conventional PHES.
1.5. Future Applications and Markets
Most of the concepts covered here are intended for high-power, high-energy applications: multiple hours of power delivery at ratings from tens of megawatts to multigigawatts. Looking ahead, we can envisage four nominal overlapping segments, discussed further in Section 2.4
• Day-ahead storage (say, (3–20) h capacity). This accounts for almost all of today’s high-energy business and will remain extremely important. Such systems can provide many reserve and regulation services, help meet the next day’s peak, and tailor the erratic production profile of variable renewables. There is little need for anything more than day-ahead storage at present.
• Week-ahead storage (say, (15–40) h capacity). Such systems will be used by future networks that rely heavily on variable renewables, subject to multi-day extremes in terms of high or low production.
• Strategic storage (say, (>50–250) h capacity). In addition to providing fairly long-term storage and quasigenerator services, the objective is to harvest energy that would otherwise be lost from the electricity supply system.
• Seasonal electricity storage (say, >200 h capacity). A hypothetical seasonal storage system may use summer sunshine to power winter loads. In practice, there usually comes a point where the marginal cost of adding energy capacity cannot be justified by the additional energy throughput.
2. Piston-in-cylinder electrical energy storage
2.1. Background and Operating Principle
The idea of storing energy by using a hydraulic piston to raise a solid working mass has been suggested many times, usually by people who do not realize the miniscule energy released when a few tonnes are lowered by a few meters. Others have proposed constructing large commercial and public buildings on hydraulic jacks for energy storage and earthquake resilience. Buildings could be elevated by (5–10) m using cheap overnight electricity and lowered slowly to provide daytime power. Unfortunately, power levels are unexciting. One would need an Empire State Building (400 000 t) to get close to 1 MW over a working day.
Clearly, we are going to need something much larger if we want to be in the same ballpark as PHES (in fact, a ballpark may be about the right size!). For context, 20 GW h is a useful amount of energy for a network—it might keep a city of a million going for a day in Europe or for half a day in the United States. Using the 400 t m (kW h)–1 metric, 20 GW h requires some 8 × 109 t m. PHES could store 20 GW h by pumping 20 × 106 t of water 400 m uphill. In the absence of a suitable mountain, we could get the same result by pumping 400 × 106 t of water up a 20 m hill. That would be absurd because of the size and cost of the two reservoirs, each the size of a small country. We would not need reservoirs if we were to raise a 400 × 106 t concrete disk by 20 m, but the disk would be far too expensive (even if we could work out how to lift it).
However, since we live on a rocky planet, we might be able to make a comparable disk cheaply by excavating material from around a disk shape. This is the basis of several storage ideas, including EscoVale’s GBES (ground-breaking energy storage), which incorporates the components of a gravitational storage system in a single construction. The concept is straightforward and construction looks simple—ground-breaking in the literal sense! In practice, the method and sequence would differ greatly from this highly simplified outline (Fig. 3.1) but, in essence:
• Excavate and reinforce a deep trench to form the periphery of the piston and the cylinder wall.
• Excavate horizontally to form and reinforce the base of the piston.
• Install a seal between the trench and the working area beneath the base.
• Store energy by pumping water into the working area to raise the piston in the cylinder.
• Recover energy by releasing the water through hydroelectric generators.
Figure 3.1 Piston-in-cylinder energy storage.
A GBES system could be constructed as above, but the concept is versatile with many possible configurations of reservoir, piston, and power plant, some of which are outlined in Fig. 3.2:
• The walls need not be vertical above the travel zone, simplifying construction tolerances, reinforcement, and local geology issues.
• The piston can be excavated within the upper reservoir, reducing the footprint and cost.
• It could be submerged to be aesthetically pleasing (the reservoir level remains constant, avoiding the unsightly cyclic drainage issues of PHES).
• A reservoir would not be needed if constructed within a lake or offshore.
• A large piston might accommodate some (perhaps all) of the power plant—short, cheap pressure tunnels and more piston control options.
• GBES can be integrated with other energy infrastructure. It could add energy storage capability to a conventional hydroelectric station without requiring a second reservoir. It could increase substantially the capacity of an existing PHES plant. GBES could be located within tidal lagoons or tidal barrage schemes and would fit within turbine separation distances in offshore wind farms.
• Alternatively, GBES could accommodate a large public water supply reservoir, using excavated material to create a retaining barrier, for example, thousands of hectares inside a 10 m to 20 m barrier.
• A similar arrangement could store surplus seasonal water or be used for flood relief, temporarily using part of the power plant to pump external water into the impoundment area. Energy supply and water management rank among the century’s greatest challenges. A technology that tackles two high-value issues could be useful in sharing costs and gaining support (water provision, flood control and coastal protection have tangible and popular local benefits, unlike energy storage).
Figure 3.2 Ground-breaking energy storage schematic.
At least two other programs share the same principle as GBES, starting from different origins within the past ten years or so.
EscoVale [2] commenced work on GBES in 2007, initially as an idea for gigawatt-class seabed storage, using piston diameters of around 500 m within large offshore wind turbine arrays. Onshore applications of interest include PHES lookalikes in flat terrain and future high-energy markets that are beyond the reach of other electrical energy storage systems—even the great majority of PHES plants.
Gravity Power (GP) [3] is believed to be the longest established and furthest advanced firm, initially based on relatively small–diameter pistons fabricated from high-density concrete, moving a considerable vertical distance in very deep shafts. GP’s focus is now on somewhat larger units, from about 30 m to 100 m diameter, giving the option of geological or fabricated pistons. The market entry target is in peaking applications, typically providing 4 h of power delivery from about 40 MW/160 MW h for the smallest unit and up to 1.6 GW/6.4 GW h with the 100 m module.
Heindl Energy (HE) [4] based its work on geological pistons from the outset, with proposed applications on the same general scale as large PHES plants—typically, gigawatt power levels and at least tens of gigawatt-hours energy. HE was ahead of EscoVale in making its plans public, including extremely large designs intended for 1012 W h (terawatt-hour) seasonal storage.
GP and HE have advanced plans for demonstration projects. EscoVale is completing an analysis of market opportunities for systems with GBES characteristics to provide a rational framework for development choices [5].
2.2. Piston versus PHES?
The job of the piston (and of the high-level PHES reservoir) is to create working pressure at the power plant. This puts us on familiar ground. If PHES and piston storage operate at the same pressure and pump the same amount of water at the same rate for the same time, they will store a similar amount of energy and require hydroelectric equipment of similar rating. The two systems have much in common, but they are not competitors. In any event, PHES would normally win—it works brilliantly, is immortal (near enough), and is backed by many millions of hours in service. It will take decades for piston storage to gain comparable experience. In the meantime, it must seek opportunities that can exploit some of its five major points of difference:
• Location
• Operating pressure
• Energy rating
• Discharge power rating
• Charging power rating.
2.2.1. Location, Location, Location
The most important attribute of piston storage is that it brings pumped hydro down from the mountains, overcoming fundamental limitations of PHES:
• Most networks do not have mountains, making PHES a nonstarter.
• Where there are mountains, there are not many people and so PHES has to serve distant applications.
• Public opinion opposes knocking mountains about to create PHES reservoirs.
2.2.2. Pressure by Design
PHES operating pressure is dictated by the terrain (typically, a few megapascals or low tens of bar). In general, designers prefer higher pressures, enabling physically smaller hydroelectric equipment, reservoirs, and pressure tunnel diameters. Pressure is a design variable for piston systems, dependent on piston height. For example, a piston height of 500 m usually provides about 7.5 MPa (75 bar) pressure at the power plant [6]. Given this freedom, the piston storage industry is likely to opt for pressures toward the top of the range for which there is PHES experience—there are few examples over 8 MPa (80 bar). In principle, piston storage might go well above 10 MPa (100 bar) and use equipment other than the industry norm of reversible Francis pump turbines. In any event, designers will probably adopt a small number of preferred operating pressures, selecting from standard hydroelectric pump motor/turbine generator packages, in much the same way that airlines and aircraft manufacturers select engines.
2.2.3. No Energy Limitations
The energy rating of a PHES system is constrained by the reservoirs. As an approximation, E ≈ 2.5VH × 10–9, where E is the stored energy (in units of gigawatt-hours); V is the usable volume of the smaller of the two reservoirs (in units of cubic meters) or water mass in tons; and H is the difference in elevation between reservoirs (expressed in meters).
PHES systems generally serve day-ahead markets where the largest systems require tens of gigawatt-hours. This is difficult, even for PHES. Information on energy ratings is sparse, but there are probably only ∼30 PHES systems >10 GW h worldwide. This compares with ∼150 medium energy (1–10) GW h plants and 200+ smaller systems of <1 GW h. Most are entirely adequate for their applications, but there is no doubt that sites suitable for PHES of >10 GW h are already hard to come by (see chapter: Pumped Hydroelectric Storage).
This limits the ability of PHES to serve future requirements for gigawatt-scale week-ahead and larger strategic storage markets (see Sections 1.5 and 2.4), with system ratings of tens or hundreds of gigawatt-hours. A 500 GW h PHES would need to transfer water between 500 Mt reservoirs (assuming an elevation difference of 400 m). There are few locations where such systems are feasible, requiring two large natural bodies of water (and even with 100 km2 lakes, there may be objections to 5 m changes in water levels).
Naturally, 500 GW h (100+ hours) piston storage would also be a huge engineering task, but of a different nature. Prospective sites are almost as common for 100 h as for 10 h systems, and are as likely (in fact, more likely) to be found in flattish terrain or fairly close to load centers, rather than in mountains. The 100 h systems need not require much more space than 10 h schemes and marginal costs are moderate.
We can illustrate this with a 3 GW GBES design (matching the current highest power PHES plant). A piston diameter of 800 m (0.5 km2 area) and operating pressure of 8 MPa (80 bar) delivers a convenient 1 GW h for every 1 m of travel [6].
A 10 h, 30 GW h day-ahead unit (e.g., Fig. 3.2) requires 30 m vertical travel.
Energy capacity could be doubled at negligible cost by extending the vertical cylinder wall section to allow 60 m travel. There is a limit to the height that the piston can be raised (one reason being that the operating pressure at the power plant increases slowly as the piston rises above the surface). It should be possible to get well into the week-ahead sector—in our example a 40 h, 120 GW h, 120 m travel system would result in a pressure increase of about 10%.
Ultrahigh-energy strategic storage requires a different approach, with a radical increase in the piston diameter, travel distance or operating pressure—the latter seems unlikely. A 600 GW h, 200 h system, with the same travel distance and pressure would require a massive (but not inherently more difficult) piston of ∼1800 m diameter. The marginal cost of the additional storage capacity would be low (perhaps 10% of the dollars per kilowatt-hour cost of the 40 h design and <5% of the 10 h unit).
Alternatively, the cylinder depth of the smaller diameter system could be increased from ∼600 m to, say, 1200 m by excavating a deeper trench. It would be necessary to take ∼600 m off the top of the piston to maintain the design operating pressure. This gives an extra 600 m of travel (600 GW h, 200 h) before the piston reaches the surface. The main issue is the huge increase in excavated material, mostly from the top of the piston—over 500 Mt for our example. The cost of extraction is surprisingly low (a few dollars per kilowatt-hour of energy storage capacity) and so disposal costs are critical. We live in a world that will have to accommodate, feed, and improve the conditions of a few billion additional people while coping with rising sea levels and increasingly extreme weather conditions. This will push up demand for material associated with coastal defense, flood protection, water management, and land reclamation projects. If blocks of rock have enough value to cover extraction and local transport, the marginal cost of deeper cylinders would be even lower than that of larger diameters.
With either approach, stored energy is limited only by system requirements and by economics. The 10th hour of storage has a higher cycle count and more earning opportunities than the 100th, and the 1000th might only be used once a year. Multiterawatt-hour storage systems are technically feasible. At the other end of the scale, GP’s modular system starts at little more than 100 MW h.
Another point to note is that piston storage projects are physically small, at least compared with PHES. A deep cylinder system can store >1 MW h m–2 in terms of cylinder area and >100 kW h m–2 of site area. This is one or two orders of magnitude better than PHES—indeed, the energy storage capacity of the entire global PHES portfolio could be matched by just two of the above 600 GW h piston storage systems.
2.2.4. No Power Limitations
Geological piston storage has no practical upper power limit—the bigger, the better because of strong economic benefits of scale. If required, power ratings far above the present storage maximum of ∼3 GW are feasible. Gigawatt-class systems will remain the norm for large-scale storage, although some locations could accommodate >10 GW. The lower threshold will depend on other design features, especially the energy rating, but may be ∼100 MW. GP’s fabricated piston design is a special case, enabling use of relatively small diameters at powers down to tens of megawatts in applications that are not accessible to HE and GBES.
2.2.5. Asymmetric Charging
Asymmetric charging is an interesting design quirk for slow-speed piston designs such as GBES or HE. It should be one of the most valuable features for future networks with a very high proportion of variable renewables.
Slow speed gives many hours of power delivery from a limited travel distance and more options in terms of seal subsystems. It also simplifies transition from discharge to charge and makes it much easier to handle extreme faults, for example, loss of grid connection when the piston is traveling at its highest descent speed and generating maximum power. Basically, if the turbines are unable to extract energy and send it to the grid, we need to deal with the kinetic energy in the system and the large amount of potential energy that would be released if the piston continued to descend an appreciable distance before it is brought to a halt. That is relatively easy if the highest descent speed is a brisk snail’s pace [6].
Piston speed is not an issue during the charging phase, with energy taken from the grid and converted via pumps into potential energy as the piston rises. Gravity assists (rather than counteracts) a switch from charge to discharge mode and aids piston deceleration if grid power is lost. Also, some of the more complex components of seal systems may not be needed when the piston is ascending. Consequently, the storage system can incorporate extra pumping capacity (at low cost), increasing the input power rating during the charge cycle. The charge rating can be several times the power delivery rating, if this is beneficial.
Asymmetric charging has little value at present. Large-scale storage systems usually have plenty of time to gather sufficient overnight energy to provide services during the following day. The situation will be quite different in networks that rely on variable renewables for a substantial part of their energy (see Section 2.4).
2.2.6. Other Performance Characteristics
The above piston storage features (power, energy, footprint, pressure, location) are quite different from PHES. In other respects, performance and longevity should be similar. Roundtrip efficiency may be marginally higher [6] and should exceed 80%.
The construction phase would be as disruptive as it is for PHES and more evident—PHES schemes are usually in very remote locations. Completed piston storage projects can be unobtrusive and may be aesthetically pleasing or locally beneficial, for example, submerged piston designs and storage integrated with public water supply or flood prevention.
Storage has a much better safety record than electricity production, but all energy projects have safety issues. PHES is the main culprit as far as storage is concerned, with its virtual monopoly of grid-scale systems. Risks are largely confined to construction and supply industry personnel. Piston systems carry an extra element of risk because they will be somewhat closer to the general public. However, in the unlikely event that a terawatt-hour of gravitational energy escapes, it will be comforting if this happens hundreds of meters below the general public’s feet, rather than above its head.
This is not just a question of up and down. As in a traffic accident, the speed at which it happens is critical. A megaton nuclear explosion can release 1 TWh almost instantaneously, destroying a city. The same energy, released in a Richter Scale 7 earthquake, can cause widespread damage in less than a minute. Catastrophic failure of a terawatt-hour hydroelectric dam might drain a 109 t (thousand million ton) reservoir in tens of minutes, with serious consequences along downstream watercourses. In comparison, fragmentation of a 1-TW h GBES piston would be a slow-motion car crash as rock and water jostle to swap places in a confined space—not a disaster, but hugely expensive and best kept a sensible distance from population centers.
2.2.7. What Could Possibly Go Wrong?
Gigawatt-class piston storage systems will face formidable engineering issues. Those raised most frequently relate to construction difficulties; the seal subsystem; structural integrity of the piston; preventing leakage from the high-pressure chamber beneath the piston; safety considerations (see Section 2.2.6, for example); “parking” the piston for maintenance; and preventing piston tilt (GBES only [6]). These are serious challenges, but they are not obvious showstoppers:
• Construction: In essence, piston storage is much like any other very large construction project and could be undertaken by adapting equipment and techniques that are well understood in the civil engineering, electromechanical, hydroelectric, and extractive industries. It may encourage development of advanced technologies, for example, for rock cutting, but it does not depend on them.
• Seal: The seal subsystem is absolutely critical to the success of piston storage. All the developers believe that they have plausible solutions for evaluation during the crucial technical development stages.
• Piston integrity: There is never a cubic kilometer of seamless basalt around when you want one. It may seem that there is little chance of a real world piston surviving the huge forces exerted by the high-pressure water needed to lift it. In fact, the force on the base of the piston is virtually identical to that which was provided by the underlying rock for millions of years before the piston was separated. This, of course, was just enough to support it. If the exerted force is fractionally more, the piston starts to move up: if fractionally less, it starts to move down.
• Leakage: For the same reason, containment of the water beneath the piston is relatively easy, because it is surrounded by material of very similar lithostatic pressure (except for the seal, of course).
The nonengineering challenges are just as important. Given a choice, potential backers prefer charismatic technologies that can be demonstrated convincingly on a laboratory bench. They should promise short-haul development, with opportunities to gain experience (and early revenue) from niche markets. Modular systems are preferred, with economies of production scale rather than physical scale. Gigawatt-class piston storage ticks all the wrong boxes, especially for most venture capital investors. It will need to make a particularly convincing case, both technically and economically.
2.3. Piston Storage Economic Performance
2.3.1. Capital Cost
Cost comparisons for new technologies are not very meaningful—early estimates are notoriously unreliable and the comparison point is a moving target. For piston systems, PHES is the logical benchmark and current gold standard for high-power storage. PHES specific costs vary widely from project to project, but $1500 kW–1 to $2000 kW–1 (in terms of power) or about $200 (kW h)–1 to $300 (kW h)–1 (in terms of energy) are reasonable averages. Such figures exclude interest costs during construction, land acquisition, approvals, hookup costs, and dedicated transmission links. Multibillion dollar annual markets suggest that these cost levels are broadly acceptable for day-ahead storage.
The range spans ∼$1000 to >$5000 kW–1. Below-average costs are more prevalent in developing economies, which account for much of the global activity. High-end costs are sometimes attributed to PHES by those advocating alternative we-can-do-better-than-that technologies. We are not aware of any significant PHES activity toward the top of the cost range.
Capital cost estimates put forward by piston storage companies suggest that parity (or better) with PHES costs is plausible in key markets from hundreds of megawatt to multigigawatt (Table 3.1). One cannot yet put much confidence in such figures, and coverage in terms of project scope is uncertain. However, there is nothing to suggest that dramatic cost reduction will be needed to match PHES.
Table 3.1
Piston Storage Capital Cost Estimates
Power/ MW |
Energy/ (GW h) |
Time/h | Power capital cost/ ($ kW–1) |
Energy capital cost/ $ (kW h)–1 |
Comments (see text) |
| 40 | 0.16 | 4 | 3200 | 800 | GP—compare with battery storage |
| 250 | 0.8 | 4 | 1700 | 425 | GP |
| 1000 | 6.4 | 4 | 870 | 220 | GP |
| 20 | 0.5 | 24 | 2400 | 100 | HE—construction costs only |
| 330 | 8 | 24 | 500 | 21 | HE—construction costs only |
| 750 | 125 | 168 | 800 | 4.7 | HE—construction costs only |
| 2750 | 2000 | 720 | 800 | 1.1 | HE—construction costs only |
| 2000 | 20 | 10 | 1300 | 130 | GBES |
| 2000 | 80 | 40 | 1400 | 35 | GBES |
| 2000 | 400 | 200 | 1800 | 9 | GBES |
GP has probably made the most thorough of these cost appraisals, including assessments undertaken by German engineering companies. The GP figures in Table 3.1 are taken from a German 2013 source [7], converted to US dollars at the exchange rate prevailing at that time. These are for the 4 h modular system, targeting peak power markets. This relatively short duration leads to fairly high dollar per kilowatt-hour figures, especially for the 40 MW unit. An important point is that $3200 kW–1 and $800 (kW h)–1 are competitive when compared to high-energy battery systems—a more realistic comparison point at this power and energy level. In any event, the 40 MW design is best regarded as a demonstrator since GP’s interests are mainly at higher powers, where projected costs compare well with PHES. This is noteworthy as GP’s figures are for relatively small diameter pistons—larger diameters should offer further economies of scale.
HE’s figures [8] are intended to underline the steep fall in specific costs as size increases, rather than to represent the all-in costs of practical storage systems. They relate to the core mechanical construction costs of the piston/cylinder (excavation, reinforcement, seal system, etc.), where the HE model assigns costs to the various construction tasks [9]. They exclude major items such as hydroelectric equipment, the hydraulic circuit, upper reservoir, and powerhouse. The full system cost will be appreciably higher in dollar per kilowatt terms (although below the PHES average). System costs in dollar per kilowatt-hour terms will remain very low for the larger HE systems—say $2 (kW h)–1 for the 2000 GW h design. This is an extreme example, operating at ultrahigh pressures with variation between 15 and 20 MPa (150 and 200 bar) during the cycle.
GBES features should keep costs low, but preparation of authoritative estimates is not a priority at this stage. The figures in Table 3.1 are tentative (back of envelope plus ∼50% contingency) and intended to illustrate the progression in moving to higher levels of stored energy, as discussed in Section 2.2.3. Moving from day-ahead to week-ahead applications is virtually cost free. Longer duration strategic storage incurs considerable additional cost, but the dollar per kilowatt-hour metric continues to fall sharply.
Capital cost estimates require caution, but piston storage economics seem likely to be acceptable in proven day-ahead markets. Indeed, piston storage might outcompete PHES, if necessary. More importantly, piston storage is a potential frontrunner in areas where PHES cannot be used, and where competition is perceived as being much weaker. There are provisos, of course: piston storage has yet to demonstrate its technical and economic prowess; other emerging technologies may set tough new benchmarks in day-ahead markets; and smaller-scale storage is becoming much more prominent.
Piston storage is a promising candidate, not a shoo-in, but it has a couple of extra shots in its locker, if there is value in going beyond the energy needed for day-ahead markets:
1. Even if we are mistaken and piston storage holds no cost advantage over other contenders at 10 h capacity, the marginal cost of adding energy storage is inherently very low and should be far less than for other electricity-in/out storage technologies.
2. Asymmetric charging enables networks with highly variable production to capture large surpluses (far in excess of the power delivery rating of the storage unit).
Capital costs are important, but the overall cost and value of the storage service are crucial. As well as the return required to recover the investment, additional factors include operation and maintenance costs, input energy costs (including roundtrip losses), annual utilization, and the value of the output energy.
2.3.2. Finance Cost
In its simplest form, the finance cost (Cf, in terms of dollars per unit of energy) depends on the specific capital cost (C in terms of dollars per unit of power), utilization (U which is the equivalent annual power delivery time at the rated output), and the required rate of return (r expressed as a percentage) according to:
As an example: assuming C = $2000 (kW)–1; r = 6% a–1 (where a is annum) and U = 3000 h a–1 then 
In practice, adjustments may result in a lower figure to take account of the tax position or specific incentives for this type of investment, or the fact that storage is said to increase the value of other assets in the network’s portfolio.
2.3.3. Operation and Maintenance (O&M) Costs
O&M costs for PHES are probably the lowest of all storage technologies, typically around $5 (MW h)–1 for high-power installations.
Ignoring special issues for the moment, piston storage conditions bring benefits that might yield a 40% reduction to $3 (MW h)–1 (∼3000 h utilization, use of variable speed technology as standard, higher average rating).
In practice, this advantage is likely to be offset by seal subsystem maintenance and other deep-level work and so $5 (MW h)–1 is a reasonable target. In the first instance, it would be prudent to budget for a higher figure— say, $10 (MW h)–1.
2.3.4. Energy Costs
For a storage system with 80% roundtrip efficiency, the input energy is 1.25 times the delivered energy. Input energy costs can be calculated in several ways:
• One approach is to operate the storage system as a network asset in which participants can park surplus energy and share revenue from subsequent energy sales, reserve services, etc. The input energy could be treated as having zero value, since it is being parked rather than sold at that stage. Other essential participants, for example, transmission and distribution operators, government agencies, and those financing the system, would also be included as beneficiaries. An agreement would account for energy and other costs as part of the equitable distribution of revenue.
• Alternatively, the storage system can be regarded as an independent entity, with input costs based on the market price of electricity at the time it is used to charge the system. Storage can take advantage of low off-peak power (there are times when it is near zero or negative in some networks). This is attractive initially, but is not particularly stable. Introducing storage increases demand and narrows the window during which low-cost surplus power is available. The sale of power from storage adds to supply and reduces the price of on-peak electricity. At some point, it becomes uneconomic to invest in further storage capacity (other than for low-energy reserve and regulation services). This will fall short of the optimum capacity in terms of overall benefit to the network (independent storage requires a production surplus to drive down the market price of input electricity and a subsequent supply deficit to drive up the price of peak power). A totally competitive storage market is unlikely to be in the best interests of the sector, or of the wider electricity supply industry. However, if this happens, piston storage could be the biggest bully on the street: it can afford an input energy price that would put some competing storage technologies out of business; asymmetric charging enables it to grab more than its fair share of input energy during shorter intervals of low prices; higher capacity systems (>20 h, say) will be the only buyers left when prolonged periods of high renewables production exceed the charging capacity of day-ahead storage.
• A third option is to base input energy costs on an estimate of the levelized cost of electricity (LCOE) delivered to the storage site. This gives stability, but introduces anomalies regarding the “correct” LCOE for input energy. For example, onshore wind power may be a principal energy input in a renewables-rich network. In some parts of the world, this would be low enough for profitable high-energy storage—perhaps a good deal less than $50 (MW h)–1. At the other extreme, ∼$150 (MW h)–1 might be regarded as appropriate and storage is left with the weaker argument that it loses less money than the alternatives (dumping surplus energy or using it for other purposes).
2.3.5. Overall Storage Costs
At its simplest the LCOE delivered from the piston storage service is the sum of the costs of finance, O&M, and input energy.
This might put the output LCOE at around $100 (MW h)–1, comprising:
| Finance cost | $40 (MW h)–1 |
| O&M | $5–$10 (MW h)–1 |
| Input power price | $40 (MW h)–1 |
| Energy losses | $10 (MW h)–1 |
A storage service costing ∼$60 (MW h)–1 (on top of the input energy price) would probably be seen as viable by most networks, especially as it does not take account of other revenue opportunities, such as provision of reserve and regulation services.
Other analyses are more ambitious. An estimate on GP’s website [3] gives the output LCOE as $76 (MW h)–1 for their 4 h, gigawatt-class design, based on $40-(MW h)–1 off-peak charging power and a capital cost of ∼$1000 (kW h)–1. The storage service cost of $36 (MW h)–1 includes ∼$10-(MW h)–1 energy losses. The balance of ∼$26 (MW h)–1 for finance and O&M costs probably requires very low cost finance since funding has to be recovered through relatively few annual hours of power delivery with a 4 h system.
HE goes further still, estimating that it would take just four years to recoup the investment in a much higher energy gigawatt-class system [10].
2.4. Markets and Competition for Piston Storage
2.4.1. Size Matters, or Does It?
With few exceptions, piston storage power ratings will be >100 MW and energy ratings will equate to at least four hours at rated delivery power. This puts piston storage at the top of the power/energy spectrum—in fact, it goes well beyond the present storage envelope. There is not much advantage to be gained from output ratings above the present 3 GW achieved by PHES, but ultrahigh-energy/long-duration storage should be very important in future (and difficult for most other storage technologies to achieve).
Competition will also come from much further afield. Storage services are delivered by wire and users requiring high-power, high-energy storage can get it from a large number of low-power, high-energy systems or, for that matter, from an even larger number of low-power, low-energy units. Low-power technologies (<100 MW in a piston storage context) account for less than 5% of grid-connected installed storage capacity at present, but well over 95% of R&D expenditure and press coverage, heralding attractive new challengers.
Equally importantly, delivery by wire means that piston storage can serve low-power or low-energy applications for which a large dedicated unit would be totally inappropriate. The user of storage services does not need to know what is on the other end of the wire.
This assumes that the wires are still there by the time that piston storage becomes commercially available. The popular “utility death spiral” scenario postulates that we are already locked into a total transformation in electricity supply, whereby large central networks will be completely swept away by local energy provision, based on distributed resources. Electricity will be provided by renewables, other generators and cogenerators, integrated with storage in locally connected networks and microgrids. If this were the future, there would be little point in gigawatt-scale concepts, including piston storage. In practice, it is far more likely that we are heading for a future in which distributed resources form a significant component of a fully interconnected system or, failing that, one that maintains the status quo. In a straight fight between the alternatives, it is the small-scale systems that would usually be swept away if they were deprived of the support provided by the central network. The argument should revolve around whether the optimum distributed resource share should be 25% or 50% or 75%—not whether it should be 0% or 100%. Piston storage can live equally comfortably with any of these, other than 100%.
2.4.2. Short-Duration Markets (<4 h Storage Capacity)
There are unlikely to be significant sales of short-duration piston storage systems, but applications in this area represent an important source of revenue.
Piston storage is excluded from one of the most active areas. Modular distributed storage is extremely responsive and can do a far better job than central units in handling local power quality issues and in smoothing the output of highly dispersed wind, solar, and other erratic power sources. Systems can be constructed to any power level (from the smallest behind-the-meter units to ∼100 MW thus far).
Piston storage will seldom be associated with a specific resource, unless the site is chosen for its proximity to gigawatt-scale wind farms or solar parks. The main short-duration market opportunity is in broader renewables support, frequency management, and other reserve services, extending into the lower end of time-shifting and peak supply. Large piston storage can ramp up or down at >100 MW min–1 (100 MW per minute) and react more quickly than spinning reserves or peaking plant (although not as quickly as batteries or flywheels). High-power storage is also valuable in emergencies, responding to line faults or failure of a large power plant, and in black-starting following network shutdown.
Naturally, these applications require that the storage unit is partially charged and not fully committed to its primary longer duration function at the time it is needed. However, part of the capacity of a large storage system could be ring-fenced for reserve and regulation services (permanently, or on a schedule agreed with the network operator).
2.4.3. Day-Ahead Markets (∼3–20 h)
Day-ahead markets have been the mainstay of PHES and so account for the great majority of today’s storage portfolio. It will remain extremely important in future. Storage capacity averages (8–10) h at present, but is seldom fully cycled. The average will decrease with the advent of new technologies optimized for somewhat lower stored energy. Annual utilization currently averages about 1000 h (equivalent hours at rated output). There is seldom any problem in gathering sufficient off-peak and surplus power to meet the following day’s requirements.
This is a densely populated sector, both in terms of its share of the grid-connected storage market and in development effort. Around 20 storage technologies are vying to serve day-ahead applications and to build a platform from which they can tap into valuable shorter duration revenue streams. Day-ahead targets range from residential storage, through a host of distributed resource applications (extending well beyond 100 MW) and into traditional central network storage systems.
Piston storage should be an effective contender, but there are many high-power alternatives, including PHES, conventional and advanced compressed air energy storage (CAES), several electricity-out storage technologies based on thermal techniques and numerous battery electrochemistries.
Competition will also come from “power-to-X” systems, where surplus electrical energy is used to produce another energy vector such as heat, methane, hydrogen, and electrofuels, which can be used in other energy markets or stored cheaply for subsequent use. The focus is on energy harvesting rather than power delivery, since the roundtrip efficiency of power-to-X-to-power cycles is generally too low to compete in the day-ahead electricity business, if another storage solution is available. However, there is a good case for power-to-X:
• for low cost utilization of prolonged energy surpluses that would otherwise be discarded
• in the many networks that will choose not to use electricity-in/out storage among their techniques for network management
• as a means of using renewables to decarbonize natural gas networks and other fossil fuel usage (especially after decarbonizing electricity production, which is more effective in carbon terms, has run its course).
2.4.4. Week-Ahead Markets (∼15–60 h)
There is little demand for systems with tens of hours of storage, although it is a useful feature for PHES or CAES if large lakes or caverns are available. Longer durations will become important in future networks that rely heavily on variable renewables, with prolonged periods when power production is generally in surplus and others when it is in deficit. As well as providing day-ahead services, higher energy units can also act as quasigenerators. When charging prospects are good, an equivalent capacity of fossil plant can be taken offline, in the knowledge that adequate notice can be given when it needs to be brought back into service. Similarly, storage units can be brought to a high state of charge in the days before a forecast period of prolonged low renewable production.
A market for these types of high-energy storage may develop fairly quickly, even though it may take several decades before problems become widespread. As with major road projects, large storage systems should be built to cope with traffic growth over a long working life. That is easier to justify if the cost of adding spare capacity is low or if the design can accommodate a low-cost upgrade, both of which apply to piston storage.
Most other storage technologies have relatively high dollar per kilowatt-hour capital costs. The exception is power-to-X, which is expected to be the main competitor in networks that use high-energy storage as one of their management tools. Power-to-X will be cheaper, but piston storage should represent a better investment under realistic input and output energy values.
Within the piston storage sector itself, moving from day-ahead to week-ahead energy capacity will often be very inexpensive (see Section 2.3.1). In this case, higher energy (tens of hours) systems may become the norm, even if most of the revenue comes from day-ahead and short duration applications.
2.4.5. Long-Duration Markets (>50 h Storage Capacity)
Compared with week-ahead systems, the main additional objective is to harvest energy that would otherwise be lost from the electricity supply system. This is not much of an issue at present, but will become prevalent in networks where variable renewables are the principal energy source.
There is no doubt that strategic storage would be extremely useful and there is unlikely to be much competition from other electricity-in/out technologies at this level. As noted in Section 2.2.3, a 600 GW h strategic store equates to about half the world’s electrical energy storage capacity. It is inconceivable that half the present global PHES population would be installed in California or a large European country or a Chinese province, where 600 GW h (or more) strategic storage might be appropriate. This could be an attractive application for piston storage. The key questions are whether the cycle count and cycle value of the additional energy capacity is sufficient to justify its cost, and whether this is a better proposition than alternatives such as long-distance interconnectors or power-to-X systems.
Low marginal cost helps. So does asymmetric charging, which can radically improve utilization—ultramarathon runners would be at a huge disadvantage if their “recharging rates” at feeding stations were restricted to the rate at which they burn energy while running (say 40 kJ min–1), especially if the energy gels are taken off the table every time the wind drops!
Opportunities for strategic (and possibly seasonal) storage will probably be confined to large supply networks. However, piston storage should greatly expand the envelope within which it makes sense to retain electrical energy within the network. It opens prospects for electrical energy storage at much higher energy levels than are envisaged with today’s technologies, and in networks that have made little or no use of storage in the past.
3. Energy membrane–underground pumped hydro storage
3.1. The Energy Membrane Concept
Invented in Denmark by JolTech [11], energy membrane-underground pumped hydro storage (EM-UPHS) stores energy by pumping water into a cavity bounded by two layers of membrane, sealed at the edges. The proposed commercial design is a 30 MW, 200 MW h unit with a working area of 0.2 km2. The cavity is buried beneath a 25 m layer of soil. The cavity inflates when water (or seawater) is pumped in at about 5 bar, raising the soil until it is approximately 10 m higher than the surrounding ground level. Energy is recovered when the water is released in turbine mode. EM-UPHS is designed for use in areas where the terrain is unsuitable for PHES, but it can take advantage of any difference in elevation between the water source and the cavity installation site. For example, a difference of just 5 m would add about 10% to the operating pressure and energy rating.
The concept has been refined with the help of a pilot project operated over a period of several years. This is a cooperative venture involving EnergiNet.dk, JolTech ApS, GODevelopment ApS, RisøDTU, DTU Byg, SDU/MCI, GEO, Syd Energi, Danfoss A/S, Arkil A/S, Lean Energy Cluster, Sloth-Møller A/S, and Sønderborg Kommune.
The pilot plant (Fig. 3.3) is of appreciable physical size (2500 m2) but on a small scale compared with the commercial design (∼10% linear, 1% area, and 0.01% energy). Table 3.2 compares key design parameters. Early results led to a design revision in which the edges of the membrane are “clamped” within a profiled trench constructed around the perimeter of the membrane area. Together with a geotextile between the upper membrane and the soil in the edge zone, this prevents excessive strain and movement in this vulnerable area.
Figure 3.3 Energy membrane–underground pumped hydro storage schematic. Schematic representation of the underground energy storage concept.
(a) Isometric view of the L × L = 50 m × 50 m field test plant; (b) Section view showing details of the design. (A) inflatable cavity, (B) connecting pipe, (C) pump system, (D) water reservoir, (E) strainer, (F) lower membrane, (G) top soil, (H) level meter.
(a) Isometric view of the L × L = 50 m × 50 m field test plant; (b) Section view showing details of the design. (A) inflatable cavity, (B) connecting pipe, (C) pump system, (D) water reservoir, (E) strainer, (F) lower membrane, (G) top soil, (H) level meter.
Table 3.2
EM-UPHS Design Parameters
| Design parameters | (5 × 5) m Lab test | (50 × 50) m Field test | (500 × 500) m Plant |
| Pump/Turbine power/kW | Not measured | 11/5.5 | 30 000 |
| Underground storage area/m2 | 25 | 2 500 | 250 000 |
| Soil layer lift distance/m | 0.1 | 0.6 | 10 |
| Soil layer thickness/m | 0.5 | 3 | 25 |
| Soil layer weight/t | 17 | 15 000 | 10 000 000 |
| Cavity volume/m3 | 1.5 | 1 500 | 1 500 000 |
| Water pressure/(100 kPa) or/bar | 0.07 | 0.6 | 5 |
| Stored energy/kWh | 0.003 | 25 | 2 900 000 |
| Percentage energy loss in soil layer/% | 5 | 0.4–1.2 | <0.1 |
| System efficiency/% | Not measured | 30 | >80 |
EM-UPHS topics for investigation include:
• Energy losses due to soil deformation
• Soil migration
• Longevity of the membrane and stability of the water cavity
• Verification of efficiency and economics at commercial scale.
3.2. Energy Losses due to Soil Deformation
Pumping water into the cavity stores energy by raising the soil mass. Energy losses result from relative movement within the soil layer, especially in the zone around the perimeter of the membrane where a ramp develops as the cavity is inflated. One of the main tasks of the trials was to measure these losses during charge/discharge cycles. They were found to average <1% of the input energy [12] and to correspond well with a finite element model incorporating soil properties. These losses are insignificant compared with the overall roundtrip losses of a commercial system (∼20%). Indeed, the model predicts that scaling benefits should reduce deformation losses to ∼0.1% [12] if the model remains valid at higher energy densities (about 1 kW m–2 in the commercial design, compared with 10 W m-2 in the trials).
3.3. Soil Migration
The surface of the soil layer takes up a similar profile to the cavity, rising and falling during the charge/discharge cycle. When the cavity is inflated, some of the soil tends to migrate horizontally toward the (lower) surrounding ground level. Key factors include soil characteristics, rainfall (rainwater that reaches the membrane must drain laterally), and, most importantly, the cycle history (the number and depth of energy cycles and the gradient profile at different states of charge). It will take longer with clay than with sand but, basically, a block of soil will tend to level out, especially if it is raised and lowered once a day.
The average of the pilot trials appears to be about 1 mm of horizontal movement per meter of vertical travel (combined up and down). The paper reporting the results [12] shows a slowdown in soil movement toward the end of the 200 day trial, suggesting the approach of steady-state conditions. There were few charge/discharge cycles after 50 days and so it is unlikely that a plot of horizontal displacement against number of cycles or aggregate vertical travel would show such a pronounced slowdown. Another caveat is that most of the trials were gentle. The paper shows only two charge/discharge cycles exceeding 50% of the rated volume change/energy capacity.
Soil migration continues after steady-state conditions are reached, slowly reducing the working mass and, therefore, the megawatt-hour energy rating for a given cavity volume. This could be dealt with by undertaking occasional remedial earth movement.
Another issue is that thinning (or thickening) of the soil layer changes the cavity profile and is progressive if not corrected.
3.4. Membrane and Water Cavity
The developers are confident that further trials will confirm membrane longevity and the long-term stability of the water cavity. The surface area of the upper membrane (which is anchored along its edges) increases when the cavity is inflated. The intended shape of the water cavity is roughly similar to the space between two flattened domes. The separation between the upper and lower membranes increases gradually with distance from the clamped edge [12]. Compared with a near cuboid shape, this requires much less membrane expansion during charging and spreads the strain over a larger area. If expansion can be accommodated by membrane stretch, there is no need for excess folded material when deflated (with greater potential for abrasive movement and the possibility of soil accumulation in progressively deeper folds as the cycle count increases). Restricting the depth of discharge (to about 80% in the trial unit) helps membrane management by retaining some water within the cavity, preventing frequent contact between the membrane layers.
The required membrane stretch is small, provided that the cavity shape can be controlled and maintained. The target shape depends on applying the correct pressure profile over a large area of the cavity. This, in turn, requires controlled variation in soil layer depth. Maintaining the correct depth pattern over such a large area may be difficult, since the soil layer is subject to frequent cycling, gradient changes, and substantial movement.
3.5. Efficiency and Economic Performance
Capital costs for the 30 MW, 200 MW h commercial design are estimated [12] at €1111 kW–1 and €208 (kW h)–1—roughly $1200 kW–1 and $230 (kW h)–1 at May 2015 exchange rates. Estimates of the cost of the storage function vary from about $60 (MW h)–1 [12] to $100 (MW h)–1 [13]. Encouragingly, the lower estimate is from the more recent source, with details in terms of cost of finance (7%), energy losses, and utilization.
Energy losses are based on roundtrip efficiency of 80%—a figure generally associated with systems operating under much higher power and pressure conditions, but one which the developers expect to achieve with the 30 MW commercial unit probably using a reversible Francis pump turbine custom-designed for this purpose. Utilization equates to some 1800 h of annual power delivery at rated output.
These are difficult targets and the cost of the storage function will probably edge up with inclusion of O&M costs. However, if further work verifies these general levels of cost and performance, and also confirms the durability of the concept at the commercial scale, EM-UPHS should have a strong competitive position. The majority of the funding is in place [12] for the next planned step: a 100 kW, 400 kW h unit at the trial site, which should clarify these issues.
4. Novel land-based and seabed pumped hydro configurations
4.1. Background
Chapter: Pumped Hydroelectric Storage focused mainly on conventional PHES systems, where water is pumped up mountains to a high-level lake or reservoir, which provides operating pressure at the low-level pump/turbine plant. The chapter also touched on alternative arrangements, operating on the same principle. The purpose is usually to provide a pumped hydro function in regions that do not have the mountainous terrain required for conventional PHES. Options include systems with:
• Two surface reservoirs
• One surface and one subterranean reservoir
• Seabed systems, where the sea acts as the high-pressure pressure reservoir and water is pumped from a low-pressure chamber anchored to the seabed.
4.2. Surface Reservoir Systems
These are variants of the standard PHES arrangement moving water between two accessible reservoirs. Possible configurations include:
• Use of worked-out open-pit mines as the lower reservoir. Some mines are of impressive depth [14], capable of accommodating large PHES-type reservoirs and operating at typical PHES pressures. The missing components are an equally large surface reservoir and, perhaps, readily accessible water (many mines are in arid areas, where provision of the initial water stock may be a problem, as may the modest ongoing requirement to make up for evaporation and other losses). However, major sites could provide storage systems of more than 10 GW h capacity.
• Use of smaller mining and quarry sites in projects of, say, (10–1000) MW h storage capacity [15,16]. QBC—the Quarry Battery Company—is a leading advocate of this approach, with plans to construct a 600 MW h facility in Wales [16], making use of two former quarry sites with an average elevation difference of about 250 m. This type of project tends to use fairly deep reservoirs of comparatively small area, resulting in substantial pressure variation during the cycle, due to changing water levels.
4.3. Subterranean Reservoir Systems
As discussed in chapter: Pumped Hydroelectric Storage, there is a long history of proposals (with little success to date) for pumped hydro projects in which the low-pressure reservoir and power plant are located underground, fed from a surface reservoir or other water source. The difference in elevation is typically comparable with a PHES plant but there is no particular restriction. Deep mines in South Africa, for example, have been developed over depths of thousands of meters, with workings at different levels. If storage systems were to be developed in deep mines, they would probably operate between worked-out levels, rather than at very high pressure. The main limitations are usually the capacity of underground chambers and the stability of mine workings when subjected to rapid water flows through the shafts and tunnels connecting chambers.
Most proposals envisage adapting old mine works [17], but underground reservoirs could be constructed specifically for the project, as with Riverbank Power’s gigawatt-scale Aquabank concept [18]. Riverbank’s website seems to have been inaccessible for some time, but a project of similar magnitude has been proposed by Sogecom and O-PAC with their FLES (flat land electricity storage) system [19]. The original plan to utilize a disused coal mine was taken further to propose excavating the lower reservoir within underlying solid rock to develop a 1.4 GW, 8 GW h project.
On a much smaller scale, the University of Colorado, Boulder (United States) [20] has explored underground pumped hydro storage systems operating from aquifers at power levels below 1 MW (down to tens of kilowatts) for use in agricultural and other rural applications.
4.4. Seabed Hydroelectric Storage
More than 99.9% of global hydropower is derived from freshwater lakes and rivers (which account for less than 0.1% of global water). The onshore/offshore balance is not going to change dramatically, but there is considerable interest in power from tidal, wave, and ocean current sources, including gigawatt-scale schemes. This encourages the development of hydroelectric equipment designed for saline conditions, which should also benefit energy storage applications.
Most seabed storage ideas use CAES techniques (see chapter: Underwater Compressed Air Energy Storage). Underwater pumped hydro concepts store energy by pumping water out of concrete pressure vessels anchored to the seabed at depths of 400 m to more than 1000 m. Energy is recovered when water reenters the vessel in turbine mode [21]. The operating pressure at 400 m is about 4 MPa (40 bar)—sufficient to generate a convenient 1 kW h for each cubic meter of water that flows back into the pressure vessel (or 2 kW h at 800 m depth). There are at least two programs investigating this.
The MIT proposal illustrated earlier (see Ref [22] and also Chapter: Pumped Hydroelectric Storage Section 4.2 and Fig. 3.3) envisages use of a large array of concrete spheres of (25–30 m) diameter, each storing around 6 MW h at 400 m depth (an operating volume of 6000 m3). A depth of 750 m depth is considered optimal at current estimated installed costs [21], aiming for an energy storage cost of about $60 (MW h)–1. With a wall thickness of up to 3 m, the design could be used at greater depths. The mass of the concrete is sufficient to offset buoyancy, keeping the structure on the seabed without elaborate anchoring. The tank weight is presumably about 10 000 t, raising some transport issues.
Work in Norway by Subhydro [23] and SINTEF [24] emphasizes advanced concrete technology. The objective is to take weight and cost out of the pressure vessels, which would be buoyant when empty and could be towed to the operating site. These pressure vessels are cylindrical tanks with hemispherical ends, where the design volume depends on the cylinder length. Each hydroelectric unit serves a group of interconnected tanks. A ventilation shaft to the surface provides air at atmospheric pressure above the water in each tank [24] and may have other functions.
A ventilation shaft is a substantial structure, over 400 m in length and capable of withstanding the surrounding water pressure. The MIT approach does not require a ventilation shaft, with a near vacuum above the water, as the sealed tank is pumped out. This also provides a slightly higher operating pressure (one extra bar), but presumably requires a purging mechanism to prevent outgassing products building up as “new” seawater is depressurized during each cycle.
Seabed pumped hydro is intended to be a good match with deep-water floating wind turbine arrays, where it is envisaged that several hundred (perhaps 1000) tanks could be deployed to provide output smoothing and day-ahead storage services. It could also serve other applications in countries where appropriate depths are found reasonably close to shore.
5. Offshore lagoon and island storage systems
5.1. Background
Seawater is an attractive resource for energy storage—PHES proposals can fail in areas with good geology because of interference with, or scarcity of, freshwater resources. Disadvantages include the corrosive nature of seawater and the somewhat lower efficiency and high specific cost of low-head hydroelectric plant. Greater use of ocean power resources will encourage development of saline low-head hydroelectric equipment. This will also help with the types of storage covered in this section, the first of which is closely related to tidal lagoon power generation.
5.2. Shallow-Water Lagoon Energy Storage
A simple tidal power generation lagoon is constructed by building a seabed-to-surface enclosure around a substantial area of water at a location with a large tidal range. Water is impounded within the lagoon until low tide, when it is released through low-head hydraulic turbines to generate power. Water is then excluded until high tide, when it flows back into the lagoon, generating further power.
It often makes sense to incorporate a pump storage function in tidal lagoon and barrage systems. It takes very little energy to pump additional water into the lagoon at high tide, because the water is raised only a small distance. The same water can yield much more energy when released through a greater distance at low tide. For example, if it takes 10 MW h of input energy to raise the level of a lagoon by 1 m at high tide, then one might, in principle, recover 100 MW h of additional energy when the water is discharged at low tide (the extra water is raised by an average 0.5 m but falls by more than 5 m in a good tidal area). This trick can be repeated at low tide, by pumping some of the remaining water out of the lagoon, reducing its level by 1 m.
In essence, this adds a PHES function at very low cost. It does not require a larger lagoon or bigger turbines (these are sized to handle the most extreme tidal conditions that the project is expected to encounter, and pumping just makes use of some of the surplus water storage capacity that is nearly always available). It has the intriguing quirk of an intuitively impossible ≫100% round-trip efficiency due to the input of “free” external energy (if you ignore the fact that tides slow the planet’s rotation!). This makes a nice change from the usual scenario where 60% might be a struggle for a small turbine when most of the water is transferred at less than half the turbine design pressure.
Technically, this qualifies as energy storage, but there is very little control over the timing of charge and discharge cycles. It is best regarded as a means of augmenting tidal generation, rather than as part of the energy storage business. However, there is at least one proposal—by David MacKay of Cambridge University (United Kingdom)—where energy storage takes center stage [25]. This is based on more aggressive pumping regimes between multiple lagoons, at least one of which is built to a height well above the high-tide level. At its simplest, this would firm up intermittent tidal energy and act as an on-demand resource without any input energy from the grid. This flexibility is obtained by self-pumping (using tidal energy to pump between lagoons). Unsurprisingly, annual energy delivery is less than that in intermittent mode (about 70% in the example in [25]).
A more interesting example stores off-peak or surplus grid power (not confined to short periods just after high and low tides), perhaps with additional input from wind turbines built as part of the project (possibly driving pumps rather than generators). This can be regarded as a storage system that time-shifts the grid power used for charging and returns it at a time when it has higher value to the network, much as any other storage system, except that it returns more energy than it takes from the grid. There is some sleight of hand here, in that the extra energy is derived from the tidal power plant (and any integrated wind turbines). However, this is really a matter of accounting. The lagoon system has to recover high capital costs, and the way in which revenue is earned is not very important unless we could get a better result by optimizing the lagoon for power production, with storage services provided by another resource.
Near-shore lagoon storage is characterized by low-energy density and large physical size (maybe 100 km2 for a system rated at hundreds of megawatts and multiple gigawatt-hours). Operating pressures are low and vary widely during the cycle, generally requiring separate pumps and turbines, rather than reversible machines. Potential sites are located close to shore in shallow water (of the order of 10 m, say), in areas with a wide tidal range.
5.3. Deeper Water Energy Island Storage
A similar technique can be employed in deeper water (of the order of 50 m, say). Suitable locations are fairly close to shore but tidal enhancement is not a factor—or, at least, not of much importance. Much of the work on this approach has been undertaken by KEMA (now within DNV GL) and by Gottlieb Paludan Architects [26,27].
A barrier is constructed to impound a substantial area, ranging from a few square kilometers up to perhaps 100 km2 to form an island with an internal lagoon. Energy is stored by pumping water out of the lagoon to lower its level by, say, 45 m. Energy is recovered when water is allowed to flow back in through turbines. In principle, the system could be fully discharged, allowing the level to rise to that outside the island. However, little energy would be recovered during the final stages, at much reduced powers and low efficiency. Most designs are intended to operate within a narrower range (perhaps between (45 and 35) m head in our example).
Using the 400 t m (kW h)–1 metric, an average head of 40 m and a 10 m difference in the charge and discharge levels would store 1 GW h for each square kilometer of lagoon area. Proposals reviewed for this report have energy densities in a range of about (0.3–0.9) GW h km–1. Difficulties with working at higher energy densities in deeper waters (100 m, say) include: the rapid increase in construction costs with barrier height; the scarcity of suitable sites with good seabed geology and reasonably constant depth over a large area fairly close to land; and more demanding sealing issues at higher pressures.
If developed, energy storage lagoons will be huge projects, usually incorporating other infrastructure (which could sometimes be the main motivation for the scheme). Port, airport, marina, and recreational facilities might be integrated with the barrier, and part of the lagoon could be infilled to service these facilities and for residential, commercial, and industrial use. The surrounding barrier is a massive structure and extends well above sea level to prevent overtopping under extreme conditions. More infrastructure can be built on the barrier, including proposals for wind turbines driving mechanical pumps to help charge the system.
6. Conclusions
Conventional PHES has dominated the grid-connected storage market and, given the opportunity, it remains fully capable of defending its corner (high-power, day-ahead storage). It does not have that opportunity in areas where the terrain is unsuitable, or where appropriate sites have already been developed, or where projects have little chance of overcoming public opposition or approval difficulties. Conventional PHES cannot participate in the rapidly growing distributed storage sector and its role will also be quite limited within any future markets for week-ahead and strategic storage.
The storage sector looks set to become much more important—it is no longer acceptable to take the view that if you cannot have PHES you cannot have storage (or, at least, not of much significance). This book underlines the diversity of technologies that will help build markets around the PHES core and, sometimes, in competition with it.
The novel hydroelectric concepts discussed in this chapter are not yet market proven, but they share many of the characteristics of PHES. If they can get close to their economic and performance targets, they should help take PHES-type solutions into territories that are currently inaccessible. Some of them can also serve emerging markets for longer duration storage, which will be beyond the reach of the great majority of electricity in/out storage technologies, including conventional PHES.
Acknowledgment
The author wishes to thank Jan Olsen of JolTech ApS for helpful advice.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.