Enhancing Electrical Supply by Pumped

Storage in Tidal Lagoons

David J.C. MacKay
Cavendish Laboratory, University of Cambridge
[email protected]

March 13, 2007 – Draft 1.8 – first published 5/3/07

Summary
The principle that the net energy delivered by a tidal pool can be
increased by pumping extra water into the pool at high tide or by
pumping extra water out of the pool at low tide is well known in
the industry. On paper, pumping can potentially enhance the net
power delivered by a factor of about four. However, pumping seems
generally to be viewed as a minor optional extra, delivering only
a modest power enhancement. Two possible reasons why pumping
is not emphasized in tidal designs are that increasing the vertical
water range introduces additional costs (for example, higher walls),
and that alternating between pumping and generating worsens the
intermittency-of-supply problem from which simple tide pools suf-
fer.
The intermittency-of-supply problem also causes problems for
wind. How can we switch to wind power if the wind might stop
blowing for two days at a time? Chemical or kinetic-energy storage
systems are an economical way to smooth out the fluctuations of
wind power on a time-scale of minutes, but what about hours and
days?
Perhaps a shift of perspective on tidal lagoons is helpful. I sketch
designs for a large pumped-storage system located at sea-level with
a dual purpose: first, it can turn power that is poorly matched
to demand into high-value demand-following power; and second, it
can simultaneously serve as a tidal power station. Large designs
with a capacity of several gigawatts are the most economical.

1
 16
14 Figure 1. Cambridge mean wind
12 speed in metres per second,
10 daily (heavy line), and
8
6 half-hourly (light line) during
4 2006. The lower figure shows
2 detail from the upper. Thanks
0 to Digital Technology Group,
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
14
Computer laboratory,
12 Cambridge This weather station
10 is on the roof of the Gates
8 building, roughly 10 m high.
6 Wind speeds at a height of 50 m
4
2 are usually about 25% bigger.
0
Mar 14 Apr 1

30
25
Figure 2. Cairngorm mean wind
speed in metres per second,
20
daily (heavy line), and
15
half-hourly (light line), during
10 six months of 2006. Thanks to
5 Heriot–Watt University Physics
0 Department.
Jan Feb Mar Apr May Jun

Storage and wind

Offshore wind farms deliver, on average, about 3 W per m2 of sea-
floor area (or 3 MW/km2 , if you prefer).
Imagine that Britain had ‘30 GW’ of wind farms – fifteen times
as much as today. I put quotes round ‘30 GW’ to emphasize that
the nominal capacity of wind farms is much bigger than the av-
erage power delivered. The standard ‘capacity factor’ in the UK
wind industry seems to be 1/3, so 30 GW of wind farms would be
expected to deliver, on average, 10 GW.
Winds fluctuate (figures 1, 2). So this average of 10 GW would
be delivered burstily: 30 GW one hour, and 0 GW the next, on one
day; and perhaps 0 GW all day on the following day. How can such
bursty power be made useful to society?
The default approach is to build back-up stations using some
other sort of power – most likely fossil fuel – which sit idle when
the wind blows, and are switched on when it does not, or when
demand peaks. Another approach would be to manage demand –
using smart electric-car chargers, for example, which use electricity
when it is cheap; or running the Aluminium plant and the water-
purification factory only when the wind blows. A third approach
is storage. The storage required to deliver 10 GW for 24 hours is
 Figure 3. Dinorwig, in the
Snowdonia National Park. The
left map is a 10 km by 10 km
area. In the right map the blue
grid is made of 1 km squares.
Dinorwig is the home of a
9 GWh storage system, using
Marchlyn Mawr (615E, 620N)
and Llyn Peris (590E, 598N) as
its upper and lower reservoirs.
Images produced from
Ordnance Survey’s Get-a-map
service www.ordnancesurvey.
co.uk/getamap. Images
reproduced with permission of
Ordnance Survey. © Crown
240 GWh – twenty-six times as big as the 9 GWh of Dinorwig. Copyright 2006
The Scottish island of Fair Isle (population 70, area 5.6 km2 )
has pioneered several of these technologies. To solve the demand-
management problem, Fairisle has for over 25 years had two elec-
tricity networks that distribute power from two wind turbines and,
if necessary, a diesel electric generator. Standard electricity service
is provided on one network, and electric heating is delivered by a
second set of cables. The electric heating is mainly served by ex-
cess electricity from the turbines that would otherwise have had
to be dumped. Remote frequency-sensitive programmable relays
control individual water heaters and storage heaters in the individ-
ual buildings of the community. In fact there’s up to six frequency
channels per household, so the system behaves like seven networks.
Fair Isle also successfully trialled a kinetic energy storage system (a
flywheel) to store energy during oscillations of wind strength (with
a period of 12 to 20 seconds).

Designs for multi-purpose storage/tidal

systems
Key ideas for an energy-enhancing pumped-storage system:

1. It is said that connecting large numbers of wind turbines to

the national electricity grid could lead to instabilities. We
thus propose decoupling wind turbines from the grid, plugging
them directly into pumped storage systems instead. The wind-
to-pump connection could be a flexible grid with much wider
tolerances than the national network.

2. The pumped storage system is located in a region with large

 tides. Water is pumped to and from the sea in such a way
that (a) the power delivered can respond to the grid’s de-
mand, eliminating problems of intermittency; and (b) we get
more power out than we put in. (Yes, I mean that the en-
ergy delivered when generating exceeds the energy received –
in contrast to Dinorwig, which has a round-trip efficiency of
about 75%.)

3. When the demand for pumped storage is low (during a few

calm days, say), the facility can also function as a stand-alone
tidal power station. By using multiple lagoons, it’s possible to
turn the intrinsically intermittent tidal power into always-on,
demand-following capacity.

4. The facility could also buy electricity from the national grid
for pumped storage, just like Dinorwig.

In sum, it’s a storage system that is more than 100% efficient.

It’s a storage system that can also produce its own power when it’s
not needed for storage. Or, it’s a tidal facility that still provides a
valuable function even when the tides are small.

Rough models
Let’s assume a tidal range of 2h = 4 m throughout. I’ll also assume
that hydroelectric generators have an efficiency of 90% and that
pumps have an efficiency of 85%. (These figures are based on the
pumped storage system at Dinorwig, whose round-trip efficiency is
about 75%. I am not sure what the best figures are for low-head
tidal turbines. In their paper based on La Rance, Shaw and Watson
[2003a] assume pumping efficiencies up to 66%, with best efficiency
at large head, and generating efficiency 80%.)
Let’s start by finding some benchmarks for energy production.

Production on ebb and flow (no pumping, no

demand-following)
The power of an artificial tide pool. To estimate the power
of an artificial tide pool, imagine that it’s filled rapidly at high
tide, and emptied rapidly at low tide. Power is generated in both
directions. The change in potential energy of the water, each six
hours, is mgh, where h is the change in height of the centre of mass
of the water, which is half the range (figure 4). The mass per unit
 Sea Tidepool Figure 4. An artificial tide pool.
The pool was filled at high tide,
and now it’s low tide. We let
the water out through the
range electricity generator to turn the
water’s potential energy into
electricity.
h

land-area covered by tide-pool is ρ × (2h), where ρ is the density of

water (1000 kg/m3 ). So the power per unit area delivered by a tide
pool is
2ρhgh
6 hours
Plugging in h = 2 m, we find

Power per unit area of tide-pool = 3.6 W/m2.

Allowing for an efficiency of 90% for conversion of this power to

electricity, we get

Power per unit area of tide-pool = 3.3 W(e)/m2 .

(Or 3.3 MW/km2 .)

Tidal pools with pumping

The pumping trick artificially increases the amplitude of the tides
in the tidal pool so as the amplify the power obtained. The energy
cost of pumping in extra water at high tide is repaid with interest
when the same water is let out at low tide; similarly, extra water
can be pumped out at low tide, then let back in at high tide. Let’s
work out the theoretical limit for this technology.
I’ll assume that generation has an efficiency of ǫg = 0.9 and that
pumping has an efficiency of ǫp = 0.85.
Let the tidal range be 2h. I’ll assume that the prices of buying
and selling electricity are the same at high tide and low tide, so
that the optimal height boost b to which the pool is pumped above
high water is given by (marginal cost of more pumping = marginal
return of water):
b/ǫp = ǫg (b + 2h)
Defining the round-trip efficiency ǫ = ǫg ǫp , we have
ǫ
b = 2h
1−ǫ
For example, with a tidal range of 2h = 4 m, and a round-trip
efficiency of ǫ = 76%, the optimal boost is b = 13 m.
Let’s assume the complementary trick is used at low tide. (This
requires that the basin have a vertical range of 30 m!) The delivered
power per unit area is then
!,
1 1 1
ρgǫg (b + 2h)2 − ρg b2 T,
2 2 ǫp

where T is the time from high tide to low tide. We can express this
as the power without pumping, scaled up by a boost factor
1
 
,
1−ǫ
which is a factor of about 4.
Tidal amplitude Optimal boost height Power Power
h b with pumping without pumping
(m) (m) (W/m2 ) (W/m2 )
0.5 3.3 0.9 0.2
1.0 6.5 3.5 0.8
2.0 13 14 3.3
3.0 20 31 7.4
4.0 26 56 13

Unfortunately, this pumping trick will rarely be exploited to the

full because of the economics of basin construction: full exploitation
of pumping requires the total height of the pool to be roughly 4
times the tidal range, and increases the delivered power by a factor
of 4. But the material in a sea-wall of height H scales as H 2 , so
presumably the cost of constructing a wall four times as high will be
more than four times as great. Extra cash would probably be better
spent on enlarging a tidal pool horizontally rather than vertically.
The pumping trick can nevertheless be used for free whenever
the natural tides are smaller than the maximum tidal range. The
next table gives the power delivered if the boost height is set to h,
that is, the range in the pool is just double the external range.
 Tidal amplitude Boost height Power Power
h b with pumping without pumping
(m) (m) (W/m2 ) (W/m2 )
0.5 0.5 0.4 0.2
1.0 1.0 1.6 0.8
2.0 2.0 6.3 3.3
3.0 3.0 14 7.4
4.0 4.0 25 13
A doubling of vertical range is plausible at neap tides, since neap
tides are typically about half as high as spring tides. Pumping the
pool at neaps so that the full springs range is used thus allows
neap tides to deliver roughly twice as much power as they would
offer without pumping. So a system with pumping would show
two-weekly variations in power of just a factor of 2 instead of 4.
These benchmarks – 3.3 W/m2 without pumping and 6.3 W/m2
with pumping – assume that power is delivered and demanded at
exactly the optimal times, and that there is no limit to the flow
rate of water in the system. Such a system is highly intermittent
and spikey. We now examine a more reasonable, smooth, but still
intermittent, benchmark.

An intermittent solution that alternates between

steady pumping and steady generating
Figure 5 shows a pumping and generating schedule where the sys-
tem is always active; it spends exactly half the time pumping (at
constant power) and half generating (at constant power). The sys-
tem alternately sucks 7 W/m2 from the electricity grid (for three
hours) and delivers 20 W/m2 (for three hours). The net energy
contribution is thus 6.5 W/m2 . The range required is about 10 m –
slightly more than double the tidal range of 4 m.

Multiple-lagoon solutions
Using multiple pools – for example, a high pool and a low pool –
doesn’t increase the deliverable power, but does increase the flexi-
bility of when power can be delivered, thus enhancing the value of
a facility. A two-pool facility is ‘always on’, and would be able to
provide the same sort of valuable service as the Dinorwig station.
 20 Figure 5. A bursty tidal power
Power (W/m^2)

15
10 option using one lagoon at
5
0 sea-level. The tidal range is
-5
-10 2h = 4 m. The system
8
Sea
alternately sucks 7 W/m2 from
6
Lagoon
the electricity grid (for three
4 hours) and delivers 20 W/m2
2 (for three hours). The net
Height (m)

0
energy contribution is thus
-2
6.5 W/m2 .
-4

-6

-8
0 6 12 18 24 30 36 42 48

Sea

High Low
Pump Pump
at high tide at low tide

Generate on demand Generate on demand

(lower conditions) (higher conditions)

Figure 6. Design assumed: one, two, or three lagoons are located at

sea-level. While one lagoon is being pumped full or pumped
empty, the other lagoon may be delivering steady, demand-
following power to the grid. Pumping may be powered by
bursty sources such as wind, by spare power from the grid
(say, in the future, from nuclear power stations), or by the
facility itself, using one lagoon’s power to pump the other
lagoon to a greater height.
 10 Figure 7. A two-lagoon system
5
0
with no power input from the
-5 grid or wind. The tidal range is
Power (W/m^2)

30 2h = 4 m. Self-pumping takes
place with a power of 32 W/m2 .
20
10
0
12
Hi
After an initial set-up period of
10
8
Sea
Lo a couple of periods, the system
6 delivers a steady 4.5 W/m2 .
4
Top graph: solid line – power
Height (m)

2
0 delivered to grid.
-2
-4
Second graph: self-pumping
-6 power.
-8
-10
-12
0 6 12 18 24 30 36 42 48

A two-pool facility can do its own pumping.

Thus a tidal station can turn the intermittent tidal power into
demand-following power. As an extreme simple case, let’s assume
that demand is absolutely steady. Not realistic, but a challenging
target for a renewable source to deliver!
Figure 7 shows a possible schedule for a two-lagoon system. In
contrast to the single-lagoon schedule, where pumping periods and
generating periods alternate, each lasting 3 hours (with switches
from pumping to generating at high tide and low tide), here gener-
ating happens all the time and pumping lasts for three hours around
each high tide and three hours around each low tide. One lagoon’s
water level is always above sea-level; the other’s is always below.
In this figure, the pumping into or out of one lagoon is entirely
funded by the energy in the other lagoon. No energy is required
from the grid. After an initial set-up period of a couple of periods,
the system delivers a steady 4.5 W/m2 . The range is about 25 m
(about six times the tidal range).

The same facility can simultaneously be used for pumped

storage.
For simplicity and clarity, I again assume that the demand is steady.
I also assume in the computations that the power being stored is
steady, but the system would work equally well if the incoming
power fluctuated around its average value on a timescale of minutes
or one or two hours.
Figure 8 shows the result of using the same schedule, doing
self-pumping for three hours around each high and low tide, plus
pumping 5.5 W/m2 of ‘bursty’ wind power into the appropriate la-
 10 Figure 8. A two-lagoon system
5
0
receiving 5.5 W/m2 of bursty
-5 wind power and delivering
Power (W/m^2)

30
20
8.5 W/m2 of steady power. The
10
0 tidal range is 2h = 4 m.
12
Hi
Self-pumping takes place with a
10
8
Sea
Lo power of 10 W/m2 .
6 Top graph: solid line – power
4
delivered to grid; dashed line –
Height (m)

2
0 average power received from
-2
-4
intermittent source, e.g. wind.
-6 Second graph: self-pumping
-8 power.
-10
-12
0 6 12 18 24 30 36 42 48

goon all the time. The system is generating a steady 8.5 W/m2 .
The range is roughly 20 m (five times the tidal range).

The facility could also be used for pumped storage alone.

Perhaps pumps and generators are a valuable resource and none
are available for the self-pumping trick. Figure 9 shows results for
two incoming power conditions. In (a), 5.5 W/m2 of ‘bursty’ wind
power is turned into 7.5 W/m2 of steady power; the range between
the high pool’s maximum and the low pool’s minimum is 16 m.
In (b) 18 W/m2 of ‘bursty’ wind power is turned into 19 W/m2 of
steady power; the range required is about 26 m.

Other designs, future work

The next design I would like to explore uses three lagoons. The
two-lagoon solution (self-pumping) doesn’t deliver as much power
per unit area, and required larger vertical amplitudes, than the
one-lagoon solution with externally-funded pumping. I expect that
there are various three- or four-lagoon solutions in which one or
two of the lagoons follow trajectories like that of the one-lagoon
solution, with most or all of the required pumping funded internally.
An obvious piece of further work is to explore the economics
of realistic daily supply and demand inputs. It’s possible that the
economically optimal pumping and generating strategy might some-
times be to exploit just one high tide and one low tide for pumping
each night, and generate at appropriately selected times in the day.
 10 Figure 9. In these simulations,
5
0
no self-pumping takes place.
-5 (a) A two-lagoon system
Power (W/m^2)

30
20
receiving 5.5 W/m2 of bursty
10
0 wind power and delivering
12
Hi 7.5 W/m2 of steady power.
10 Sea
8 Lo (b) A two-lagoon system
6 receiving 18 W/m2 of bursty
4
wind power and delivering
Height (m)

2
0 19 W/m2 of steady power.
-2
-4 The tidal range is 2h = 4 m.
-6 Top graph: solid line – power
-8
-10
delivered to grid; dashed line –
-12
0 6 12 18 24 30 36 42 48
average power received from
(a) intermittent source, e.g. wind.
20
15
10
5
0
-5
-10
-15
Power (W/m^2)

-20
30
20
10
0
12
Hi
10 Sea
8 Lo

6
4
Height (m)

2
0
-2
-4
-6
-8
-10
-12
0 6 12 18 24 30 36 42 48
(b)

60 60

50 50

40 40

30 30

20 20

10 10

0 0
8 15 22 29 8 15 22 29
January 2006 June 2006

Figure 10. Electricity demand

in Great Britain (in GW)
during three winter weeks and
three summer weeks of 2006.
 120 Figure 11. Electricity prices in
100
80 Great Britain (in £ per MWh)
12 January 2006 60 on three days in 2006 and 2007.
40
20
0
0 6 12 18 24
120
100
80
13 June 2006 60
40
20
0
0 6 12 18 24
120
100
80
24 February 2007 60
40
20
0
0 6 12 18 24
Time in hours

To reduce costs associated with high walls we could look for lop-
sided schedules where lagoons are pumped down to lower extremes
and not pumped up so high.
Criticisms: I’ve ignored the true dependence of generating
and pumping efficiency on head. I’ve assumed the sea is an in-
exhaustible source or sink of water at the current sea-level. Once
the system reaches a sufficiently large size, its sucking and blowing
will have a significant effect on local sea-level.
I’ve not taken account of the cost of turbines, assuming that
we can install whatever pumping and generating capacity these
schedules call for.

Discussion
Some of these ranges are enormous. Where could such a system be
put? What would it cost, and what would it be worth?
One simple observation is that the value delivered scales as the
area of the lagoons, but the dominant part of the cost – the walls
– scales as the circumference. Very large systems are thus favoured
by simple economics.
Let’s pick a benchmark size. How about 10 km × 10 km?

ˆ A plain old intermittent tide-pool of this size (3.3 MW/km2 )

would deliver 330 MW on average (assuming, as usual, a 4 m
tidal range).
Figure 12. UK territorial waters
with depth less than 25 m
(yellow) and depth between
25 m and 50 m (magenta). Data
from DTI Atlas of Renewable
Marine Resources. © Crown
copyright.
 ˆ With pumping to 5 m above and below mean sea-level, a single
lagoon would deliver a net power of 650 MW.

ˆ The two-lagoon solution that does its own pumping would

deliver a steady 450 MW.

ˆ It could also be used as a pumped storage system for in-

termittent or unwanted electricity. For example, it could
turn 550 MW (average) of bursty wind power into 750 MW
of steady power. (A round-trip efficiency of 135%!) Or it
could turn 1.8 GW of bursty wind power into 1.9 GW of steady
power. A round-trip efficiency of 105% compares favourably
with Dinorwig’s 75%.

ˆ For comparison of pumped storage capacity with Dinorwig

– a profitable power station with a capacity to store 9 GWh
– these tide pools would have a capacity of about 20 GWh
(assuming a height change of 13 m). So this facility would
be worth two Dinorwigs. Indeed, it would be worth more,
since it would be better than 100% efficient, in contrast to
Dinorwig’s 75%.

We need at least half of this water to have a depth of about

13 m below mean sea-level. There’s lots of shallow water around
Britain. We need the tidal range to be as large as possible, too. An
ideal location might be an area of shallow sea surrounding a small
island, where the pumping facilities could be built. Alternatively
the high pool could be built on land. Offshore lagoons have many
advantages, as advocated by Tidal Electric limited, and Friends of
the Earth Friends of the Earth Cymru [2004]. I think the best two
locations in the British Isles are, on the East Coast, The Wash
(where the mean spring tidal range exceeds 7 m) and, on the West
Coast, anywhere in the Irish Sea from the Mersey to the mouth of
Morecambe Bay (where there have been proposals to build a 12-
mile bridge with built-in tidal and wind power). The mean spring
tidal range here is 7–8 m. Morecambe Bay already has a gas field,
so there is a precedent for energy exploitation. From wikipedia:
‘A lease has been granted for the development of two wind turbine
sites in Morecambe Bay, one at Walney Island and the other at
Cleveleys. Together these will have around 50 turbines.’ The Wash
would be big enough to fit one 10 km by 10 km tidal facility, but
perhaps not more. The Irish sea is bigger. Both locations have a
tidal range bigger than I assumed, so the potential power is bigger
– perhaps about twice as big, on average.
 Figure 13. Two locations with
plentiful shallow water and big
tides: The Wash, and the Irish
Sea (marked by two circles
each). UK territorial waters
with depth less than 25 m
(yellow) and depth between
25 m and 50 m (magenta). Data
from DTI Atlas of Renewable
Marine Resources. © Crown
copyright.

What about the cost? Two circular lagoons enclosing 100 km2
would require 50 km of walls. The low pool’s walls would be in
water of depth about 13 m. And (for the most ambitious schedules
described here) we need the high pool to have a wall of height 13 m
above sea-level. Let’s look at some costs from Tidal Electric limited.
Their plan for a small tidal lagoon in Swansea Bay (where the
mean tidal range varies between 4.1 m and 8.5 m) involved 9 km of
walls and would cost either £82 million (according to Tidal Electric,
AEA Technology, and W.S. Atkins Engineering) or £234 million
(according to critics of Tidal Electric’s scheme). The cost of the
wall was estimated to be £49 million or £114 million respectively.
Taking the larger of these two figures, the cost per km of wall
is £13 million. This wall was of height 16 m from sea bed to crest.
The walls I was imagining above would be slightly higher or perhaps
twice as high (if the high pool is built in water of the same depth as
the low pool). The wall, using this technology, would thus cost at
least £0.65 billion. Perhaps costs could be reduced by alternative
wall construction methods. And I think the wall heights could
be trimmed quite a lot without spoiling the results sketched here.
Doubling the wall’s cost to allow for all the other stuff, I’ll propose
£1.3 billion as the cost for a 10 km × 10 km two-lagoon system.
Dinorwig cost £0.4 billion in 1980 money, so £1.3 billion for a
facility superior to two Dinorwigs sounds a reasonable deal to me.
Another way of expressing the value of the facility is to take what
people currently spend on wind turbines – for example, £500 million
on the ‘650 MW’ Lewis wind farm, plus £375 million on the Lewis–
Mainland electricity connection: an expenditure of about £0.9 billion
on roughly 220 MW (average) of intermittent power. Scaling this
up, 550 MW of bursty wind power seems to be valued at £2.25 billion.
The pumped storage solution presented in figure 9(a), requiring
walls of height 9 m above mean sealevel, would turn this 550 MW of
bursty wind power into 750 MW of steady demand-following power.
It seems plausible to me that this service would be worth the esti-
mated cost of £1.3 billion.
If the cost needs to be reduced, we simply make the system
bigger. For example, we multiply the area by four (to 20 km ×
20 km) and double the length of all the walls. The estimated cost
roughly doubles (to £2.6 billion, say), but the storage quadruples
to 40 GWh (more than four Dinorwigs). As a source of tidal power,
this quadrupled station could deliver a steady 1.8 GW all day and
all night, and could serve peak demand.

Cost comparison with vanadium flow batteries

For comparison, VRB power systems have provided a 12 MWh en-
ergy storage system for the Sorne Hill windfarm in Ireland (cur-
rently ‘32 MW’, increasing to ‘39 MW’). I think VRB stands for
vanadium redox battery. This storage system is a big ‘flow bat-
tery’, a vanadium-based redox regenerative fuel cell, with a couple
of tanks full of vanadium in different chemical states. This storage
system can smooth the output of its windfarm on a time-scale of
minutes, but the longest time for which it could deliver one third
of the ‘capacity’ (during a lull in the wind) is one hour. The same
company installed a 1.1 MWh system on Tasmania. It can deliver
200 kW for four hours, 300 kW for 5 minutes and 400 kW for 10
seconds.
A 1.5 MWh vanadium system costing $480 000 occupies 70 m2
with a mass of 107 tonnes. Its efficiency is 70–75%, round-trip.
Scaling this up, and translating into British, a 10 GWh system
using vanadium would cost £1.64 billion; a 20 GWh system would
cost £3.3 billion. The tidal-pumped-storage system thus looks com-
petitive with the storage technology currently used for large wind-
farms. [Scaling up the Vanadium technology to 10 or 20 GWh might
have a noticeable effect on the world Vanadium market, but it
is probably feasible. Current worldwide production of Vanadium
is 40 000 tonnes per year. A 10 GWh system, assuming 1-molar
Vanadium solution, would contain 36 000 tonnes of Vanadium – one
year’s worth of current production. Vanadium is currently produced
as a by-product of other processes, and the total world Vanadium
resource is estimated to be 63 million tonnes.]
Compressed-air storage
Compressed air storage is said to be significantly less expensive
than other large-scale storage options [Denholm et al., 2005]. “En-
ergy is stored by compressing air in an airtight underground storage
cavern. To extract stored energy, compressed air is drawn from the
storage vessel, heated, and then expanded through a high-pressure
turbine, which captures some of the energy in the compressed air.
The air is then mixed with fuel and combusted, with the exhaust
expanded through a low-pressure gas turbine. The turbines are
connected to an electrical generator. Turbine exhaust heat and gas
burners are used to preheat cavern air entering the turbines. CAES
can be considered a hybrid generation/storage system because it
requires combustion in the gas turbine. The storage benefit of pre-
compressed air is the elimination of the turbine input compressor
stage, which uses approximately 60% of the mechanical energy pro-
duced by a standard combustion turbine. By utilization of precom-
pressed air, CAES effectively ”stores” the mechanical energy that
would be required to turn the input compressor and uses nearly all
of the turbine mechanical energy to drive the electric generator.”
1 kWh of electricity generated by the CAES turbine requires
4649 kJ of fuel (1.3 kWh) plus 0.735 kWh of compressor electricity.
This is said to be five times more efficient than the most efficient
plain fossil combustion technology.
Further details including a life-cycle analysis are in the paper
[Denholm et al., 2005].
I haven’t found a figure for the cost of such a storage system.

Additional opportunities
A pair of lagoons in the sea with 13 m-high walls and electrical
plumbing installed would be a good place to locate wind turbines.
The turbines would be offshore, which is good, but erection and
maintenance of turbines on the walls would be much easier and
cheaper than for regular offshore turbines. 100 m diameter turbines
(with ‘capacity’ 3.5 MW) could be placed every 500 m – 100 turbines
in total, with a ‘capacity’ of 350 MW. A good combination: wind,
pumped storage, and tidal energy, all enhancing each other.
Perhaps to kill four birds with one stone, we could sequester
carbon too: the walls could be built out of artificial limestone, or
coal!
Acknowledgements
I thank Stephen Salter, Denis Mollison, Adrian Wrigley, Tim Jervis,
Marcus Frean, and Trevor Whittaker for helpful discussions.

References
J. A. Baines, V. G. Newman, I. W. Hanna, T. H. Douglas, W. J.
Carlyle, I. L. Jones, D. M. Eaton, and G. Zeronian. Dinorwig
pumped storage scheme. Institution of Civil Engineers Proceed-
ings pt. 1, 74:635–718, November 1983.
J. A. Baines, V. G. Newman, I. W. Hanna, T. H. Douglas, W. J.
Carlyle, I. L. Jones, D. M. Eaton, and G. Zeronian. Dinorwig
pumped storage scheme. Institution of Civil Engineers Proceed-
ings pt. 1, 80:493–536, April 1986.
C. Baker, J. Walbancke, and P. Leach. Tidal lagoon power gener-
ation scheme in swansea bay, 2006. URL http://www.dti.gov.
uk/files/file30617.pdf. A report on behalf of the Department
of Trade and Industry and the Welsh Development Agency.
B. D. T. . R. S. C. A two-basin tidal power scheme for the severn
estuary. In Conference on new approaches to tidal power, 1982.
P. Denholm, G. L. Kulcinski, and T. Holloway. Emissions and
energy efficiency assessment of baseload wind energy systems.
Environ Sci Technol, 39(6):1903–1911, March 2005. ISSN 0013-
936X. URL http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?
cmd=Retrieve&db=pubmed&dopt=Abstract&list uids=15819254.

Department of Trade and Industry. DTI atlas of UK marine renew-

able energy resources, December 2004.
Friends of the Earth Cymru. A severn barrage or tidal la-
goons?, January 2004. URL http://www.foe.co.uk/resource/
briefings/severn barrage lagoons.pdf.

T. J. Hammons. Tidal power. Proceedings of the IEEE, 8(3):419–

433, March 1993.
Z. Kowalik. Tide distribution and tapping into tidal energy.
Oceanologia, 46(3):291–331, 2004.
S. H. Salter. Possible under-estimation of the UK tidal resource,
2005. URL http://www.dti.gov.uk/files/file31313.pdf. Sub-
mission for DTI Energy Review Question 2.
T. L. Shaw and M. J. Watson. The effects of pumping on
the energy potential of a tidal power barrage. Engi-
neering Sustainability, 156(2):111–117, June 2003a. ISSN
1478-4637. doi: 10.1680/ensu.156.2.111.37018. URL
http://www.atypon-link.com/TELF/doi/abs/10.1680/ensu.
156.2.111.37018?cookieSet=1&journalCode=ensu.

T. L. Shaw and M. J. Watson. Flexible power generation from a

severn barrage, April 2006. URL http://www.dti.gov.uk/files/
file31332.pdf.

T. L. Shaw and M. J. Watson. Flexible power generation from the

tides. Engineering Sustainability, 156(2):119–123, 2003b. ISSN
1478-4629.

S. J. Taylor. The severn barrage – definition study for a new ap-

praisal of the project, 2002. URL http://www.dti.gov.uk/files/
file15363.pdf. ETSU REPORT NO. T/09/00212/00/REP.
 Power translation chart
kWh/d each GW / UK TWh/y / UK Mtoe/y / UK
UK (2004)
310 2700 230
120 300 2600
290 220
2500
280 210
110 2400
270 200
260 2300
100 250 2200 190
240 2100 180
230 2000 170
90
220 1900
210 160
1800
80 200 150
1700
190 140
180 1600
70 130
170 1500
160 1400 120
60 150 1300 110
140 1200
130 100
50 1100
120 90 UK Electricity
1000
110 80 fuel input (2004)
40 100 900
90 800 70
80 700 60
30
70 600 50
60 500
20 50 40
400 UK Electricity (2004)
40 30
300
30 20
10 200
20
10 100 UK Nuclear (2004) 10
0 0 0 0
kWh/d each GW / UK TWh/y / UK Mtoe/y / UK

1 kWh/d the same as 1/24 kW ‘UK’ = 60 million people

GW often used for ‘capacity’ (peak output)
TWh/y often used for average output USA: 300 kWh/d each
1 Mtoe ‘one million tonnes of oil equivalent’ Europe: 120 kWh/d each