08.b Boek sewtha.pdf

6 — Solar 8

power (kWh/d)

7 6

total heat generated immersion heater

5 hot water used

4 3 2 1

solar power

controller

0 J

F M A M J

J

A S O N D

would be about 10 m2 of panels per person – and let’s assume these are 50%-efficient at turning the sunlight’s 110 W/m2 into hot water (figure 6.3). Multiplying 50% × 10 m2 × 110 W/m2

Figure 6.3. Solar power generated by a 3 m2 hot-water panel (green), and supplementary heat required (blue) to make hot water in the test house of Viridian Solar. (The photograph shows a house with the same model of panel on its roof.) The average solar power from 3 m2 was 3.8 kWh/d. The experiment simulated the hot-water consumption of an average European household – 100 litres of hot (60 ◦ C) water per day. The 1.5–2 kWh/d gap between the total heat generated (black line, top) and the hot water used (red line) is caused by heat-loss. The magenta line shows the electrical power required to run the solar system. The average power per unit area of these solar panels is 53 W/m2 .

we find solar heating could deliver 13 kWh per day per person. I colour this production box white in figure 6.4 to indicate that it describes production of low-grade energy – hot water is not as valuable as the highgrade electrical energy that wind turbines produce. Heat can’t be exported to the electricity grid. If you don’t need it, then it’s wasted. We should bear in mind that much of this captured heat would not be in the right place. In cities, where many people live, residential accommodation has less roof area per person than the national average. Furthermore, this power would be delivered non-uniformly through the year.

Jet flights: 30 kWh/d

Solar photovoltaic Photovoltaic (PV) panels convert sunlight into electricity. Typical solar panels have an efficiency of about 10%; expensive ones perform at 20%. (Fundamental physical laws limit the efficiency of photovoltaic systems to at best 60% with perfect concentrating mirrors or lenses, and 45% without concentration. A mass-produced device with efficiency greater than 30% would be quite remarkable.) The average power delivered by south-facing 20%-efficient photovoltaic panels in Britain would be 20% × 110 W/m2 = 22 W/m2 . Figure 6.5 shows data to back up this number. Let’s give every person 10 m2 of expensive (20%-efficient) solar panels and cover all south-facing roofs. These will deliver 5 kWh per day per person.

Car: 40 kWh/d

Solar heating: 13 kWh/d

Wind: 20 kWh/d

Figure 6.4. Solar thermal: a 10 m2 array of thermal panels can deliver (on average) about 13 kWh per day of thermal energy.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

The conclusion so far: covering your south-facing roof at home with photovoltaics may provide enough juice to cover quite a big chunk of your personal average electricity consumption; but roofs are not big enough to make a huge dent in our total energy consumption. To do more with PV, we need to step down to terra firma. The solar warriors in figure 6.6 show the way.

4500 4000 3500 3000 2500 2000 1500 1000 500 0

h/d

Apr

Mar

Jan

Feb

Dec

Oct

kW

Nov

Sep

Jul

Aug

Jun

Apr

12

May

Since the area of all south-facing roofs is 10 m2 per person, there certainly isn’t space on our roofs for these photovoltaic panels as well as the solar thermal panels of the last section. So we have to choose whether to have the photovoltaic contribution or the solar hot water contribution. But I’ll just plop both these on the production stack anyway. Incidentally, the present cost of installing such photovoltaic panels is about four times the cost of installing solar thermal panels, but they deliver only half as much energy, albeit high-grade energy (electricity). So I’d advise a family thinking of going solar to investigate the solar thermal option first. The smartest solution, at least in sunny countries, is to make combined systems that deliver both electricity and hot water from a single installation. This is the approach pioneered by Heliodynamics, who reduce the overall cost of their systems by surrounding small high-grade gallium arsenide photovoltaic units with arrays of slowly-moving flat mirrors; the mirrors focus the sunlight onto the photovoltaic units, which deliver both electricity and hot water; the hot water is generated by pumping water past the back of the photovoltaic units.

Sustainable Energy – without the hot air

Energy delivered (kWh)

40

Figure 6.5. Solar photovoltaics: data from a 25-m2 array in Cambridgeshire in 2006. The peak power delivered by this array is about 4 kW. The average, year-round, is 12 kWh per day. That’s 20 W per square metre of panel.

Figure 6.6. Two solar warriors enjoying their photovoltaic system, which powers their electric cars and home. The array of 120 panels (300 W each, 2.2 m2 each) has an area of 268 m2 , a peak output (allowing for losses in DC–to–AC conversion) of 30.5 kW, and an average output – in California, near Santa Cruz – of 5 kW (19 W/m2 ). Photo kindly provided by Kenneth Adelman. www.solarwarrior.com

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

41

6 — Solar

Fantasy time: solar farming If a breakthrough of solar technology occurs and the cost of photovoltaics came down enough that we could deploy panels all over the countryside, what is the maximum conceivable production? Well, if we covered 5% of the UK with 10%-efficient panels, we’d have

≃

10% × 100 W/m2 × 200 m2 per person 50 kWh/day/person.

I assumed only 10%-efficient panels, by the way, because I imagine that solar panels would be mass-produced on such a scale only if they were very cheap, and it’s the lower-efficiency panels that will get cheap first. The power density (the power per unit area) of such a solar farm would be 10% × 100 W/m2 = 10 W/m2 . This power density is twice that of the Bavaria Solarpark (figure 6.7). Could this flood of solar panels co-exist with the army of windmills we imagined in Chapter 4? Yes, no problem: windmills cast little shadow, and ground-level solar panels have negligible effect on the wind. How audacious is this plan? The solar power capacity required to deliver this 50 kWh per day per person in the UK is more than 100 times all the photovoltaics in the whole world. So should I include the PV farm in my sustainable production stack? I’m in two minds. At the start of this book I said I wanted to explore what the laws of physics say about the limits of sustainable energy, assuming money is no object. On those grounds, I should certainly go ahead, industrialize the countryside, and push the PV farm onto the stack. At the same time, I want to help people figure out what we should be doing between now and 2050. And today, electricity from solar farms would be four times as expensive as the market rate.

So I feel a bit irresponsible as I include this estimate in the sustainable production stack in figure 6.9 – paving 5% of the UK with solar panels seems beyond the bounds of plausibility in so many ways. If we seriously contemplated doing such a thing, it would quite probably be better to put the panels in a two-fold sunnier country and send some of the energy home by power lines. We’ll return to this idea in Chapter 25.

Figure 6.7. A solar photovoltaic farm: the 6.3 MW (peak) Solarpark in Mühlhausen, Bavaria. Its average power per unit land area is expected to be about 5 W/m2 . Photo by SunPower. Total UK land area: 4000 m2 per person

buildings: 48 m2 gardens: 114 m2 roads: 60 m2 water: 69 m2

arable land: 2800 m2

Mythconceptions Manufacturing a solar panel consumes more energy than it will ever deliver. False. The energy yield ratio (the ratio of energy delivered by a system over its lifetime, to the energy required to make it) of a roof-mounted, grid-connected solar system in Central Northern Europe is 4, for a system with a lifetime of 20 years (Richards and Watt, 2007); and more than 7 in

Figure 6.8. Land areas per person in Britain.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

42

Sustainable Energy – without the hot air

a sunnier spot such as Australia. (An energy yield ratio bigger than one means that a system is A Good Thing, energy-wise.) Wind turbines with a lifetime of 20 years have an energy yield ratio of 80. Aren’t photovoltaic panels going to get more and more efficient as technology improves? I am sure that photovoltaic panels will become ever cheaper; I’m also sure that solar panels will become ever less energy-intensive to manufacture, so their energy yield ratio will improve. But this chapter’s photovoltaic estimates weren’t constrained by the economic cost of the panels, nor by the energy cost of their manufacture. This chapter was concerned with the maximum conceivable power delivered. Photovoltaic panels with 20% efficiency are already close to the theoretical limit (see this chapter’s endnotes). I’ll be surprised if this chapter’s estimate for roof-based photovoltaics ever needs a significant upward revision.

Jet flights: 30 kWh/d

PV farm (200 m2/p): 50 kWh/d

PV, 10 m2/p: 5

Car: 40 kWh/d

Solar heating: 13 kWh/d

Wind: 20 kWh/d

Solar biomass All of a sudden, you know, we may be in the energy business by being able to grow grass on the ranch! And have it harvested and converted into energy. That’s what’s close to happening. George W. Bush, February 2006 All available bioenergy solutions involve first growing green stuff, and then doing something with the green stuff. How big could the energy collected by the green stuff possibly be? There are four main routes to get energy from solar-powered biological systems: 1. We can grow specially-chosen plants and burn them in a power station that produces electricity or heat or both. We’ll call this “coal substitution.” 2. We can grow specially-chosen plants (oil-seed rape, sugar cane, or corn, say), turn them into ethanol or biodiesel, and shove that into cars, trains, planes or other places where such chemicals are useful. Or we might cultivate genetically-engineered bacteria, cyanobacteria, or algae that directly produce hydrogen, ethanol, or butanol, or even electricity. We’ll call all such approaches “petroleum substitution.” 3. We can take by-products from other agricultural activities and burn them in a power station. The by-products might range from straw (a by-product of Weetabix) to chicken poo (a by-product of McNuggets). Burning by-products is coal substitution again, but using ordinary plants, not the best high-energy plants.

A power station that burns agricultural by-products won’t deliver as much power per unit area of farmland as an optimized biomass-growing facility, but it has the

Figure 6.9. Solar photovoltaics: a 10 m2 array of building-mounted south-facing panels with 20% efficiency can deliver about 5 kWh per day of electrical energy. If 5% of the country were coated with 10%-efficient solar panels (200 m2 of panels per person) they would deliver 50 kWh/day/person.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

43

6 — Solar advantage that it doesn’t monopolize the land. Burning methane gas from landfill sites is a similar way of getting energy, but it’s sustainable only as long as we have a sustainable source of junk to keep putting into the landfill sites. (Most of the landfill methane comes from wasted food; people in Britain throw away about 300 g of food per day per person.) Incinerating household waste is another slightly less roundabout way of getting power from solar biomass. 4. We can grow plants and feed them directly to energy-requiring humans or other animals. For all of these processes, the first staging post for the energy is in a chemical molecule such as a carbohydrate in a green plant. We can therefore estimate the power obtainable from any and all of these processes by estimating how much power could pass through that first staging post. All the subsequent steps involving tractors, animals, chemical facilities, landfill sites, or power stations can only lose energy. So the power at the first staging post is an upper bound on the power available from all plant-based power solutions. So, let’s simply estimate the power at the first staging post. (In Chapter D, we’ll go into more detail, estimating the maximum contribution of each process.) The average harvestable power of sunlight in Britain is 100 W/m2 .

The most efficient plants in Europe are about 2%-efficient at turning solar energy into carbohydrates, which would suggest that plants might deliver 2 W/m2 ; however, their efficiency drops at higher light levels, and the best performance of any energy crops in Europe is closer to 0.5 W/m2 . Let’s cover 75% of the country with quality green stuff. That’s 3000 m2 per person devoted to bio-energy. This is the same as the British land area wood (commercial forestry) rape rape to biodiesel maize sugar beet short rotation coppice calorific value energy crops calorific value miscanthus to electricity switchgrass corn to ethanol wheat to ethanol jatropha tropical plantations (eucalyptus) 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 power density (W/m2 )

1.5

Figure 6.10. Some Miscanthus grass enjoying the company of Dr Emily Heaton, who is 5’4” (163 cm) tall. In Britain, Miscanthus achieves a power per unit area of 0.75 W/m2 . Photo provided by the University of Illinois.

Figure 6.11. Power production, per unit area, achieved by various plants. For sources, see the end-notes. These power densities vary depending on irrigation and fertilization; ranges are indicated for some crops, for example wood has a range from 0.095–0.254 W/m2 . The bottom three power densities are for crops grown in tropical locations. The last power sugarcane (Brazil, Zambia) density (tropical plantations∗ ) assumes genetic modification, tropical plantations∗ fertilizer application, and irrigation. In the text, I use 0.5 W/m2 as a 1.8 summary figure for the best energy crops in NW Europe.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

44

Sustainable Energy – without the hot air

currently devoted to agriculture. So the maximum power available, ignoring all the additional costs of growing, harvesting, and processing the greenery, is 0.5 W/m2 × 3000 m2 per person

Biomass: food, biofuel, wood, waste incin’n, landfill gas: 24 kWh/d

= 36 kWh/d per person.

Wow. That’s not very much, considering the outrageously generous assumptions we just made, to try to get a big number. If you wanted to get biofuels for cars or planes from the greenery, all the other steps in the chain from farm to spark plug would inevitably be inefficient. I think it’d be optimistic to hope that the overall losses along the processing chain would be as small as 33%. Even burning dried wood in a good wood boiler loses 20% of the heat up the chimney. So surely the true potential power from biomass and biofuels cannot be any bigger than 24 kWh/d per person. And don’t forget, we want to use some of the greenery to make food for us and for our animal companions. Could genetic engineering produce plants that convert solar energy to chemicals more efficiently? It’s conceivable; but I haven’t found any scientific publication predicting that plants in Europe could achieve net power production beyond 1 W/m2 . I’ll pop 24 kWh/d per person onto the green stack, emphasizing that I think this number is an over-estimate – I think the true maximum power that we could get from biomass will be smaller because of the losses in farming and processing. I think one conclusion is clear: biofuels can’t add up – at least, not in countries like Britain, and not as a replacement for all transport fuels.

Even leaving aside biofuels’ main defects – that their production competes with food, and that the additional inputs required for farming and processing often cancel out most of the delivered energy (figure 6.14) – biofuels made from plants, in a European country like Britain, can deliver so little power, I think they are scarcely worth talking about.

PV farm (200 m2/p): 50 kWh/d

Jet flights: 30 kWh/d

PV, 10 m2/p: 5 Solar heating: 13 kWh/d

Car: 40 kWh/d

Wind: 20 kWh/d

Figure 6.12. Solar biomass, including all forms of biofuel, waste incineration, and food: 24 kWh/d per person.

0.45

Notes and further reading page no.

0.4 0.35 0.3 0.25

38 . . . compensate for the tilt between the sun and the land. The latitude of Cambridge is θ = 52◦ ; the intensity of midday sunlight is multiplied by cos θ ≃ 0.6. The precise factor depends on the time of year, and varies between cos(θ + 23◦ ) = 0.26 and cos(θ − 23◦ ) = 0.87. – In a typical UK location the sun shines during one third of daylight hours. The Highlands get 1100 h sunshine per year – a sunniness of 25%. The best spots in Scotland get 1400 h per year – 32%. Cambridge: 1500 ± 130 h per year – 34%. South coast of England (the sunniest part of the UK): 1700 h per year – 39%. [2rqloc] Cambridge data from [2szckw]. See also figure 6.16.

1960

1970

1980

1990

2000

Figure 6.13. Sunniness of Cambridge: the number of hours of sunshine per year, expressed as a fraction of the total number of daylight hours.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

45

6 — Solar

Figure 6.14. This figure illustrates the quantitative questions that must be asked of any proposed biofuel. What are the additional energy inputs required for farming and processing? What is the delivered energy? What is the net energy output? Often the additional inputs and losses wipe out most of the energy delivered by the plants.

Sunlight 100 W/m2

2 carbohydrate 0.5 W/m energy delivered by plants

Energy used or lost in farming and processing

delivered energy

net energy

additional inputs required for farming and processing

39 . . . that would be about 10 m2 of panels per person. I estimated the area of south-facing roof per person by taking the area of land covered by buildings per person (48 m2 in England – table I.6), multiplying by 1/4 to get the southfacing fraction, and bumping the area up by 40% to allow for roof tilt. This gives 16 m2 per person. Panels usually come in inconvenient rectangles so some fraction of roof will be left showing; hence 10 m2 of panels. – The average power delivered by photovoltaic panels. . . There’s a myth going around that states that solar panels produce almost as much power in cloudy conditions as in sunshine. This is simply not true. On a bright but cloudy day, solar photovoltaic panels and plants do continue to convert some energy, but much less: photovoltaic production falls roughly ten-fold when the sun goes behind clouds (because the intensity of the incoming sunlight falls ten-fold). As figure 6.15 shows, the power delivered by photovoltaic panels is almost exactly proportional to the intensity of the sunlight – at least, if the panels are at 25 ◦ C. To complicate things, the power delivered depends on temperature too – hotter panels have reduced power (typically 0.38% loss in power per ◦ C) – but if you check data from real panels, e.g. at www.solarwarrior.com, you can confirm the main point: output on a cloudy day is far less than on a sunny day.

This issue is obfuscated by some solar-panel promoters who discuss how the “efficiency” varies with sunlight. “The panels are more efficient in cloudy conditions,” they say; this

180 delivered power (W/sq m)

38 The average raw power of sunshine per square metre of south-facing roof in Britain is roughly 110 W/m2 , and of flat ground, roughly 100 W/m2 . Source: NASA “Surface meteorology and Solar Energy” [5hrxls]. Surprised that there’s so little difference between a tilted roof facing south and a horizontal roof? I was. The difference really is just 10% [6z9epq].

160 140 120 100 80 60 40 20 0 0

200

400

600

800

1000

irradiance (W/sq m)

Figure 6.15. Power produced by the Sanyo HIP-210NKHE1 module as a function of light intensity (at 25 ◦ C, assuming an output voltage of 40 V). Source: datasheet, www.sanyo-solar.eu.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

46

Sustainable Energy – without the hot air average sunshine (W/m2 ) 0 Anchorage, AK Edinburgh, UK Oslo, NO Dublin, IE Hamburg, DE London, UK Brussels, BE Munich, DE Paris, FR Bern, CH Toulouse, FR New York, NY Seattle, WA Boston, MA Chicago, IL Portland, OR Kansas City, KS Rome, IT Madrid, ES Atlanta, GA Lisbon, PT Algiers, DZ Salt Lake City, UT Denver, CO Athens, GR Tunis, TN Houston, TX Malaga, ES Freetown, SL San Francisco, CA Albuquerque, NM Yaounde, CM Liberia, LR Bangui, CF Limassol, CY Accra, GH Rabat, MA Miami, FL Las Vegas, NV Phoenix, AZ Los Angeles, CA Tarabulus, LY Dakar, SN Abuja, NG Nairobi, KE Cairo, EG Gambia, GM Conakry, GN Addis Abeba, ET Honolulu, HI Ouagadougou, BF Muqdisho, SO Bamako, ML Niamey, NE Al Khur-tum, SD Djibouti, DJ Nouakchott, MR

50 100 150 200 250 87 94 95 99 105 109 112 124 125 131 143 147 147 149 155 159 171 176 177 182 185 186 189 190 190 196 197 199 200 204 207 208 210 214 215 217 217 219 221 224 225 229 232 234 234 237 240 241 243 248 251 252 254 255 263 266 273

Figure 6.16. Average power of sunshine falling on a horizontal surface in selected locations in Europe, North America, and Africa.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

47

6 — Solar

ultraviolet

infrared 500

400

400

300

300

200

200

100

100

0

ultraviolet

0 0

0.5

1

1.5

2

2.5

3

photon energy (eV)

3.5

4

0

0.5

1

1.5

2

2.5

3

3.5

4

photon energy (eV)

may be true, but efficiency should not be confused with delivered power. 39 Typical solar panels have an efficiency of about 10%; expensive ones perform at 20%. See figure 6.18. Sources: Turkenburg (2000), Sunpower www. sunpowercorp.com, Sanyo www.sanyo-solar.eu, Suntech. – A device with efficiency greater than 30% would be quite remarkable. This is a quote from Hopfield and Gollub (1978), who were writing about panels without concentrating mirrors or lenses. The theoretical limit for a standard “single-junction” solar panel without concentrators, the Shockley–Queisser limit, says that at most 31% of the energy in sunlight can be converted to electricity (Shockley and Queisser, 1961). (The main reason for this limit is that a standard solar material has a property called its band-gap, which defines a particular energy of photon that that material converts most efficiently. Sunlight contains photons with many energies; photons with energy below the band-gap are not used at all; photons with energy greater than the band-gap may be captured, but all their energy in excess of the band-gap is lost. Concentrators (lenses or mirrors) can both reduce the cost (per watt) of photovoltaic systems, and increase their efficiency. The Shockley–Queisser limit for solar panels with concentrators is 41% efficiency.

The only way to beat the Shockley–Queisser limit is to make fancy photovoltaic devices that split the light into different wavelengths, processing each wavelength-range with its own personalized band-gap. These are called multiple-junction photovoltaics. Recently multiple-junction photovoltaics with optical concentrators have been reported to be about 40% efficient. [2tl7t6], www.spectrolab.com. In July 2007, the University of Delaware reported 42.8% efficiency with 20-times concentration [6hobq2], [2lsx6t]. In August 2008, NREL reported 40.8% efficiency with 326-times concentration [62ccou]. Strangely, both these results were called world efficiency records. What multiple-junction devices are available on the market? Uni-solar sell a thin-film triple-junction 58 W(peak) panel with an area of 1 m2 . That implies an efficiency, in full sunlight, of only 5.8%. 40 Figure 6.5: Solar PV data. Data and photograph kindly provided by Jonathan Kimmitt. – Heliodynamics – www.hdsolar.com. See figure 6.19. A similar system is made by Arontis www.arontis.se.

Figure 6.17. Part of Shockley and Queisser’s explanation for the 31% limit of the efficiency of simple photovoltaics. Left: the spectrum of midday sunlight. The vertical axis shows the power density in W/m2 per eV of spectral interval. The visible part of the spectrum is indicated by the coloured section. Right: the energy captured by a photovoltaic device with a single band-gap at 1.1 eV is shown by the tomato-shaded area. Photons with energy less than the band-gap are lost. Some of the energy of photons above the band-gap is lost; for example half of the energy of every 2.2 eV photon is lost. Further losses are incurred because of inevitable radiation from recombining charges in the photovoltaic material.

amorphous silicon multi-crystalline silicon single crystal silicon Sunpower WHT Sanyo HIP

Triple-junction limit

600 infrared

500

Shockley– Queisser limit

600

Suntech poly-crystalline thin-film triple junction 0

10%

20% 30% efficiency

40%

50%

Figure 6.18. Efficiencies of solar photovoltaic modules available for sale today. In the text I assume that roof-top photovoltaics are 20% efficient, and that country-covering photovoltaics would be 10% efficient. In a location where the average power density of incoming sunlight is 100 W/m2 , 20%-efficient panels deliver 20 W/m2 .

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

48

Sustainable Energy – without the hot air

41 The Solarpark in Muhlhausen, Bavaria. On average this 25-hectare farm is expected to deliver 0.7 MW (17 000 kWh per day). New York’s Stillwell Avenue subway station has integrated amorphous silicon thin-film photovoltaics in its roof canopy, delivering 4 W/m2 (Fies et al., 2007). The Nellis solar power plant in Nevada was completed in December, 2007, on 140 acres, and is expected to generate 30 GWh per year. That’s 6 W/m2 [5hzs5y]. Serpa Solar Power Plant, Portugal (PV), “the world’s most powerful solar power plant,” [39z5m5] [2uk8q8] has sun-tracking panels occupying 60 hectares, i.e., 600 000 m2 or 0.6 km2 , expected to generate 20 GWh per year, i.e., 2.3 MW on average. That’s a power per unit area of 3.8 W/m2 . 41 The solar power capacity required to deliver 50 kWh/d per person in the UK is more than 100 times all the photovoltaics in the whole world. To deliver 50 kWh/d per person in the UK would require 125 GW average power, which requires 1250 GW of capacity. At the end of 2007, world installed photovoltaics amounted to 10 GW peak; the build rate is roughly 2 GW per year. – . . . paving 5% of this country with solar panels seems beyond the bounds of plausibility. My main reason for feeling such a panelling of the country would be implausible is that Brits like using their countryside for farming and recreation rather than solar-panel husbandry.

Another concern might be price. This isn’t a book about economics, but here are a few figures. Going by the price-tag of the Bavarian solar farm, to deliver 50 kWh/d per person would cost e91 000 per person; if that power station lasted 20 years without further expenditure, the wholesale cost of the electricity would be e0.25 per kWh. Further reading: David Carlson, BP solar [2ahecp]. 43 People in Britain throw away about 300 g of food per day. Source: Ventour (2008). – Figure 6.10. In the USA, Miscanthus grown without nitrogen fertilizer yields about 24 t/ha/y of dry matter. In Britain, yields of 12–16 t/ha/y are reported. Dry Miscanthus has a net calorific value of 17 MJ/kg, so the British yield corresponds to a power density of 0.75 W/m2 . Sources: Heaton et al. (2004) and [6kqq77]. The estimated yield is obtained only after three years of undisturbed growing. – The most efficient plants are about 2% efficient; but the delivered power per unit area is about 0.5 W/m2 . At low light intensities, the best British plants are 2.4% efficient in well-fertilized fields (Monteith, 1977) but at higher light intensities, their conversion efficiency drops. According to Turkenburg (2000) and Schiermeier et al. (2008), the conversion efficiency of solar to biomass energy is less than 1%. Here are a few sources to back up my estimate of 0.5 W/m2 for vegetable power in the UK.

The Royal Commission on Environmental Pollution’s estimate of the potential delivered power density from energy crops in Britain is 0.2 W/m2 (Royal Commission on Environmental Pollution, 2004). On page 43 of the Royal Society’s biofuels document (Royal Society working group on biofuels, 2008), Miscanthus tops the list, delivering about 0.8 W/m2 of chemical power.

Figure 6.19. A combined-heat-and-power photovoltaic unit from Heliodynamics. A reflector area of 32 m2 (a bit larger than the side of a double-decker bus) delivers up to 10 kW of heat and 1.5 kW of electrical power. In a sun-belt country, one of these one-ton devices could deliver about 60 kWh/d of heat and 9 kWh/d of electricity. These powers correspond to average fluxes of 80 W/m2 of heat and 12 W/m2 of electricity (that’s per square metre of device surface); these fluxes are similar to the fluxes delivered by standard solar heating panels and solar photovoltaic panels, but Heliodynamics’s concentrating design delivers power at a lower cost, because most of the material is simple flat glass. For comparison, the total power consumption of the average European person is 125 kWh/d.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

6 — Solar In the World Energy Assessment published by the UNDP, Rogner (2000) writes: “Assuming a 45% conversion efficiency to electricity and yields of 15 oven dry tons per hectare per year, 2 km2 of plantation would be needed per megawatt of electricity of installed capacity running 4,000 hours a year.” That is a power per unit area of 0.23 W(e)/m2 . (1 W(e) means 1 watt of electrical power.) Energy for Sustainable Development Ltd (2003) estimates that short-rotation coppices can deliver over 10 tons of dry wood per hectare per year, which corresponds to a power density of 0.57 W/m2 . (Dry wood has a calorific value of 5 kWh per kg.) According to Archer and Barber (2004), the instantaneous efficiency of a healthy leaf in optimal conditions can approach 5%, but the long-term energystorage efficiency of modern crops is 0.5–1%. Archer and Barber suggest that by genetic modification, it might be possible to improve the storage efficiency of plants, especially C4 plants, which have already naturally evolved a more efficient photosynthetic pathway. C4 plants are mainly found in the tropics and thrive in high temperatures; they don’t grow at temperatures below 10 ◦ C. Some examples of C4 plants are sugarcane, maize, sorghum, finger millet, and switchgrass. Zhu et al.

(2008) calculate that the theoretical limit for the conversion efficiency of solar energy to biomass is 4.6% for C3 photosynthesis at 30 ◦ C and today’s 380 ppm atmospheric CO2 concentration, and 6% for C4 photosynthesis. They say that the highest solar energy conversion efficiencies reported for C3 and C4 crops are 2.4% and 3.7% respectively; and, citing Boyer (1982), that the average conversion efficiencies of major crops in the US are 3 or 4 times lower than those record efficiencies (that is, about 1% efficient). One reason that plants don’t achieve the theoretical limit is that they have insufficient capacity to use all the incoming radiation of bright sunlight. Both these papers (Zhu et al., 2008; Boyer, 1982) discuss prospects for genetic engineering of more-efficient plants. 43 Figure 6.11. The numbers in this figure are drawn from Rogner (2000) (net energy yields of wood, rape, sugarcane, and tropical plantations); Bayer Crop Science (2003) (rape to biodiesel); Francis et al. (2005) and Asselbergs et al. (2006) (jatropha); Mabee et al. (2006) (sugarcane, Brazil); Schmer et al. (2008) (switchgrass, marginal cropland in USA); Shapouri et al. (1995) (corn to ethanol); Royal Commission on Environmental Pollution (2004); Royal Society working group on biofuels (2008); Energy for Sustainable Development Ltd (2003); Archer and Barber (2004); Boyer (1982); Monteith (1977).

44 Even just setting fire to dried wood in a good wood boiler loses 20% of the heat up the chimney. Sources: Royal Society working group on biofuels (2008); Royal Commission on Environmental Pollution (2004).

49

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

7

Heating and cooling

This chapter explores how much power we spend controlling the temperature of our surroundings – at home and at work – and on warming or cooling our food, drink, laundry, and dirty dishes.

Domestic water heating The biggest use of hot water in a house might be baths, showers, dishwashing, or clothes-washing – it depends on your lifestyle. Let’s estimate first the energy used by taking a hot bath. The volume of bath-water is 50 cm × 15 cm × 150 cm ≃ 110 litre. Say the temperature of the bath is 50 ◦ C (120 F) and the water coming into the house is at 10 ◦ C. The heat capacity of water, which measures how much energy is required to heat it up, is 4200 J per litre per ◦ C. So the energy required to heat up the water by 40 ◦ C is

Figure 7.1. A flock of new houses.

4200 J/litre/◦ C × 110 litre × 40 ◦ C ≃ 18 MJ ≃ 5 kWh. So taking a bath uses about 5 kWh. For comparison, taking a shower (30 litres) uses about 1.4 kWh.

Figure 7.2. The water in a bath.

Kettles and cookers Britain, being a civilized country, has a 230 volt domestic electricity supply. With this supply, we can use an electric kettle to boil several litres of water in a couple of minutes. Such kettles have a power of 3 kW. Why 3 kW? Because this is the biggest power that a 230 volt outlet can deliver without the current exceeding the maximum permitted, 13 amps. In countries where the voltage is 110 volts, it takes twice as long to make a pot of tea. If a household has the kettle on for 20 minutes per day, that’s an average power consumption of 1 kWh per day. (I’ll work out the next few items “per household,” with 2 people per household.) One small ring on an electric cooker has the same power as a toaster: 1 kW. The higher-power hot plates deliver 2.3 kW. If you use two rings of the cooker on full power for half an hour per day, that corresponds to 1.6 kWh per day. A microwave oven usually has its cooking power marked on the front: mine says 900 W, but it actually consumes about 1.4 kW. If you use the microwave for 20 minutes per day, that’s 0.5 kWh per day. A regular oven guzzles more: about 3 kW when on full. If you use the oven for one hour per day, and the oven’s on full power for half of that time, that’s 1.5 kWh per day. 50

230 V × 13 A

=

3000 W

Microwave: 1400 W peak

Fridge-freezer: 100 W peak, 18 W average

Figure 7.3. Power consumption by a heating and a cooling device.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

51

7 — Heating and cooling Device Cooking – kettle – microwave – electric cooker (rings) – electric oven Cleaning – washing machine – tumble dryer – airing-cupboard drying – washing-line drying – dishwasher Cooling – refrigerator – freezer – air-conditioning

power

3 kW 1.4 kW 3.3 kW 3 kW 2.5 kW 2.5 kW

time per day

energy per day

1/3 h

1 kWh/d 0.5 kWh/d 1.6 kWh/d 1.5 kWh/d

1/3 h 1/2 h 1/2 h

0.8 h

2.5 kW 0.02 kW 0.09 kW 0.6 kW

24 h 24 h 1h

Table 7.4. Energy consumption figures for heating and cooling devices, per household.

1 kWh/d 2 kWh/d 0.5 kWh/d 0 kWh/d 1.5 kWh/d 0.5 kWh/d 2.3 kWh/d 0.6 kWh/d

Hot clothes and hot dishes A clothes washer, dishwasher, and tumble dryer all use a power of about 2.5 kW when running. A clothes washer uses about 80 litres of water per load, with an energy cost of about 1 kWh if the temperature is set to 40 ◦ C. If we use an indoor airing-cupboard instead of a tumble dryer to dry clothes, heat is still required to evaporate the water – roughly 1.5 kWh to dry one load of clothes, instead of 3 kWh. Totting up the estimates relating to hot water, I think it’s easy to use about 12 kWh per day per person.

Hot water: 12 kWh/d Figure 7.5. The hot water total at both home and work – including bathing, showering, clothes washing, cookers, kettles, microwave oven, and dishwashing – is about 12 kWh per day per person. I’ve given this box a light colour to indicate that this power could be delivered by low-grade thermal energy.

Hot air – at home and at work Now, does more power go into making hot water and hot food, or into making hot air via our buildings’ radiators? One way to estimate the energy used per day for hot air is to imagine a building heated instead by electric fires, whose powers are more familiar to us. The power of a small electric bar fire or electric fan heater is 1 kW (24 kWh per day). In winter, you might need one of these per person to keep toasty. In summer, none. So we estimate that on average one modern person needs to use 12 kWh per day on hot air. But most people use more than they need, keeping several rooms warm simultaneously (kitchen, living room, corridor, and bathroom, say). So a plausible consumption figure for hot air is about double that: 24 kWh per day per person. This chapter’s companion Chapter E contains a more detailed account of where the heat is going in a building; this model makes it possible to

Figure 7.6. A big electric heater: 2 kW.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

52

Sustainable Energy – without the hot air

predict the heat savings from turning the thermostat down, double-glazing the windows, and so forth.

Hot air: 24 kWh/d

Warming the outdoors, and other luxuries There’s a growing trend of warming the outdoors with patio heaters. Typical patio heaters have a power of 15 kW. So if you use one of these for a couple of hours every evening, you are using an extra 30 kWh per day. A more modest luxury is an electric blanket. An electric blanket for a double bed uses 140 W; switching it on for one hour uses 0.14 kWh.

Figure 7.7. Hot air total – including domestic and workplace heating – about 24 kWh per day per person.

Cooling Fridge and freezer We control the temperatures not only of the hot water and hot air with which we surround ourselves, but also of the cold cupboards we squeeze into our hothouses. My fridge-freezer, pictured in figure 7.3, consumes 18 W on average – that’s roughly 0.5 kWh/d.

Air-conditioning In countries where the temperature gets above 30 ◦ C, air-conditioning is viewed as a necessity, and the energy cost of delivering that temperature control can be large. However, this part of the book is about British energy consumption, and Britain’s temperatures provide little need for airconditioning (figure 7.8). 35 30 25 20 15 10 5 0 -5 -10

Figure 7.8. Cambridge temperature in degrees Celsius, daily (red line), and half-hourly (blue line) during 2006.

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

An economical way to get air-conditioning is an air-source heat pump. A window-mounted electric air-conditioning unit for a single room uses 0.6 kW of electricity and (by heat-exchanger) delivers 2.6 kW of cooling. To estimate how much energy someone might use in the UK, I assumed they might switch such an air-conditioning unit on for about 12 hours per day on 30 days of the year. On the days when it’s on, the air-conditioner uses 7.2 kWh. The average consumption over the whole year is 0.6 kWh/d. This chapter’s estimate of the energy cost of cooling – 1 kWh/d per person – includes this air-conditioning and a domestic refrigerator. Society

Cooling: 1 kWh/d Figure 7.9. Cooling total – including a refrigerator (fridge/freezer) and a little summer air-conditioning – 1 kWh/d.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

53

7 — Heating and cooling

kW h/ d

20 15

40

gas used (1000 kWh)

condensing boiler and thermostats installed 61 47 36

33

95

96

43

41

50

44

50 35

34

32

10 5

Figure 7.10. My domestic cumulative gas consumption, in kWh, each year from 1993 to 2005. The number at the top of each year’s line is the average rate of energy consumption, in kWh per day. To find out what happened in 2007, keep reading!

0 93

94

97

98

99

2000

01

02

03

04

05

06

also refrigerates food on its way from field to shopping basket. I’ll estimate the power cost of the food-chain later, in Chapter 15.

Total heating and cooling Our rough estimate of the total energy that one person might spend on heating and cooling, including home, workplace, and cooking, is 37 kWh/d per person (12 for hot water, 24 for hot air, and 1 for cooling). Evidence that this estimate is in the right ballpark, or perhaps a little on the low side, comes from my own domestic gas consumption, which for 12 years averaged 40 kWh per day (figure 7.10). At the time I thought I was a fairly frugal user of heating, but I wasn’t being attentive to my actual power consumption. Chapter 21 will reveal how much power I saved once I started paying attention. Since heating is a big item in our consumption stack, let’s check my estimates against some national statistics. Nationally, the average domestic consumption for space heating, water, and cooking in the year 2000 was 21 kWh per day per person, and consumption in the service sector for heating, cooling, catering, and hot water was 8.5 kWh/d/p. For an estimate of workplace heating, let’s take the gas consumption of the University of Cambridge in 2006–7: 16 kWh/d per employee. Totting up these three numbers, a second guess for the national spend on heating is 21 + 8.5 + 16 ≃ 45 kWh/d per person, if Cambridge University is a normal workplace. Good, that’s reassuringly close to our first guess of 37 kWh/d.

Notes and further reading page no. 50 An oven uses 3 kW. Obviously there’s a range of powers. Many ovens have a maximum power of 1.8 kW or 2.2 kW. Top-of-the-line ovens use as much as 6 kW. For example, the Whirlpool AGB 487/WP 4 Hotplate Electric Oven Range has a 5.9 kW oven, and four 2.3 kW hotplates. www.kcmltd.com/electric oven ranges.shtml www.1stforkitchens.co.uk/kitchenovens.html

Heating, cooling: 37 kWh/d

Jet flights: 30 kWh/d

Biomass: food, biofuel, wood, waste incin’n, landfill gas: 24 kWh/d

PV farm (200 m2/p): 50 kWh/d

PV, 10 m2/p: 5

Car: 40 kWh/d

Solar heating: 13 kWh/d

Wind: 20 kWh/d

Figure 7.11. Heating and cooling – about 37 units per day per person. I’ve removed the shading from this box to indicate that it represents power that could be delivered by low-grade thermal energy.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

54

Sustainable Energy – without the hot air

51 An airing cupboard requires roughly 1.5 kWh to dry one load of clothes. I worked this out by weighing my laundry: a load of clothes, 4 kg when dry, emerged from my Bosch washing machine weighing 2.2 kg more (even after a good German spinning). The latent heat of vaporization of water at 15 ◦ C is roughly 2500 kJ/kg. To obtain the daily figure in table 7.4 I assumed that one person has a load of laundry every three days, and that this sucks valuable heat from the house during the cold half of the year. (In summer, using the airing cupboard delivers a little bit of air-conditioning, since the evaporating water cools the air in the house.) 53 Nationally, the average domestic consumption was 21 kWh/d/p; consumption in the service sector was 8.5 kWh/d/p. Source: Dept. of Trade and Industry (2002a). – In 2006–7, Cambridge University’s gas consumption was 16 kWh/d per employee. The gas and oil consumption of the University of Cambridge (not including the Colleges) was 76 GWh in 2006–7. I declared the University to be the place of work of 13 300 people (8602 staff and 4667 postgraduate researchers). Its electricity consumption, incidentally, was 99.5 GWh. Source: University utilities report.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

8

Hydroelectricity

To make hydroelectric power, you need altitude, and you need rainfall. Let’s estimate the total energy of all the rain as it runs down to sea-level. For this hydroelectric forecast, I’ll divide Britain into two: the lower, dryer bits, which I’ll call “the lowlands;” and the higher, wetter bits, which I’ll call “the highlands.” I’ll choose Bedford and Kinlochewe as my representatives of these two regions. Let’s do the lowlands first. To estimate the gravitational power of lowland rain, we multiply the rainfall in Bedford (584 mm per year) by the density of water (1000 kg/m3 ), the strength of gravity (10 m/s2 ) and the typical lowland altitude above the sea (say 100 m). The power per unit area works out to 0.02 W/m2 . That’s the power per unit area of land on which rain falls. When we multiply this by the area per person (2700 m2 , if the lowlands are equally shared between all 60 million Brits), we find an average raw power of about 1 kWh per day per person. This is the absolute upper limit for lowland hydroelectric power, if every river were dammed and every drop perfectly exploited. Realistically, we will only ever dam rivers with substantial height drops, with catchment areas much smaller than the whole country. Much of the water evaporates before it gets anywhere near a turbine, and no hydroelectric system exploits the full potential energy of the water.

We thus arrive at a firm conclusion about lowland water power. People may enjoy making “run-of-the-river” hydro and other small-scale hydroelectric schemes, but such lowland facilities can never deliver more than 1 kWh per day per person.

Figure 8.1. Nant-y-Moch dam, part of a 55 MW hydroelectric scheme in Wales. Photo by Dave Newbould, www.origins-photography.co.uk.

1344

670 km2 between 800 m and 1344 m

800

20 000 km2 between 400 m and 800 m

400 40 000 km2 between 200 m and 400 m 200

63 000 km2 between 100 m and 200 m 72 000 km2 between 50 m and 100 m 50 000 km2 between 0 m and 50 m

100 50 0

55

Figure 8.2. Altitudes of land in Britain. The rectangles show how much land area there is at each height.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

56

Sustainable Energy – without the hot air

Let’s turn to the highlands. Kinlochewe is a rainier spot: it gets 2278 mm per year, four times more than Bedford. The height drops there are bigger too – large areas of land are above 300 m. So overall a twelve-fold increase in power per square metre is plausible for mountainous regions. The raw power per unit area is roughly 0.24 W/m2 . If the highlands generously share their hydro-power with the rest of the UK (at 1300 m2 area per person), we find an upper limit of about 7 kWh per day per person. As in the lowlands, this is the upper limit on raw power if evaporation were outlawed and every drop were perfectly exploited. What should we estimate is the plausible practical limit? Let’s guess 20% of this – 1.4 kWh per day, and round it up a little to allow for production in the lowlands: 1.5 kWh per day. The actual power from hydroelectricity in the UK today is 0.2 kWh/d per person, so this 1.5 kWh/d per person would require a seven-fold increase in hydroelectric power.

Hydro: 1.5 kWh/d

Heating, cooling: 37 kWh/d

Jet flights: 30 kWh/d

Biomass: food, biofuel, wood, waste incin’n, landfill gas: 24 kWh/d

PV farm (200 m2/p): 50 kWh/d

Notes and further reading PV, 10 m2/p: 5

55 Rainfall statistics are from the BBC weather centre. 56 The raw power per unit area [of Highland rain] is roughly 0.24 W/m2 . We can check this estimate against the actual power density of the Loch Sloy hydro-electric scheme, completed in 1950 (Ross, 2008). The catchment area of Loch Sloy is about 83 km2 ; the rainfall there is about 2900 mm per year (a bit higher than the 2278 mm/y of Kinlochewe); and the electricity output in 2006 was 142 GWh per year, which corresponds to a power density of 0.2 W per m2 of catchment area. Loch Sloy’s surface area is about 1.5 km2 , so the hydroelectric facility itself has a per unit lake area of 11 W/m2 . So the hillsides, aqueducts, and tunnels bringing water to Loch Sloy act like a 55-fold power concentrator.

Car: 40 kWh/d

Solar heating: 13 kWh/d

Wind: 20 kWh/d

Figure 8.3. Hydroelectricity.

– The actual power from hydroelectricity in the UK today is 0.2 kWh per day per person. Source: MacLeay et al. (2007). In 2006, large-scale hydro produced 3515 GWh (from plant with a capacity of 1.37 GW); small-scale hydro, 212 GWh (0.01 kWh/d/p) (from a capacity of 153 MW). In 1943, when the growth of hydroelectricity was in full swing, the North of Scotland Hydroelectricity Board’s engineers estimated that the Highlands of Scotland could produce 6.3 TWh per year in 102 facilities – that would correspond to 0.3 kWh/d per person in the UK (Ross, 2008). Glendoe, the first new large-scale hydroelectric project in the UK since 1957, will add capacity of 100 MW and is expected to deliver 180 GWh per year. Glendoe’s catchment area is 75 km2 , so its power density works out to 0.27 W per m2 of catchment area. Glendoe has been billed as “big enough to power Glasgow.” But if we share its 180 GWh per year across the population of Glasgow (616 000 people), we get only 0.8 kWh/d per person. That is just 5% of the average electricity consumption of 17 kWh/d per person. The 20fold exaggeration is achieved by focusing on Glendoe’s peak output rather than its average, which is 5 times smaller; and by discussing “homes” rather than the total electrical power of Glasgow (see p329). Figure 8.4. A 60 kW waterwheel.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

9

Light

Hydro: 1.5 kWh/d Light: 4 kWh/d

Lighting home and work The brightest domestic lightbulbs use 250 W, and bedside lamps use 40 W. In an old-fashioned incandescent bulb, most of this power gets turned into heat, rather than light. A fluorescent tube can produce an equal amount of light using one quarter of the power of an incandescent bulb. How much power does a moderately affluent person use for lighting? My rough estimate, based on table 9.2, is that a typical two-person home with a mix of low-energy and high-energy bulbs uses about 5.5 kWh per day, or 2.7 kWh per day per person. I assume that each person also has a workplace where they share similar illumination with their colleagues; guessing that the workplace uses 1.3 kWh/d per person, we get a round figure of 4 kWh/d per person.

Heating, cooling: 37 kWh/d

Jet flights: 30 kWh/d

Biomass: food, biofuel, wood, waste incin’n, landfill gas: 24 kWh/d

PV farm (200 m2/p): 50 kWh/d

PV, 10 m2/p: 5

Street-lights and traffic lights Do we need to include public lighting too, to get an accurate estimate, or do home and work dominate the lighting budget? Street-lights in fact use about 0.1 kWh per day per person, and traffic lights only 0.005 kWh/d per person – both negligible, compared with our home and workplace lighting. What about other forms of public lighting – illuminated signs and bollards, for example? There are fewer of them than street-lights; and street-lights already came in well under our radar, so we don’t need to modify our overall estimate of 4 kWh/d per person.

Car: 40 kWh/d

Solar heating: 13 kWh/d

Wind: 20 kWh/d

Figure 9.1. Lighting – 4 kWh per day per person.

Lights on the traffic In some countries, drivers must switch their lights on whenever their car is moving. How does the extra power required by that policy compare with the power already being used to trundle the car around? Let’s say the car has four incandescent lights totalling 100 W. The electricity for those bulbs is supplied by a 25%-efficient engine powering a 55%-efficient generator, so the power required is 730 W. For comparison, a typical car going at an average speed of 50 km/h and consuming one litre per 12 km

Device

Power

Time per day

Energy per day per home

10 incandescent lights 10 low-energy lights

1 kW 0.1 kW

5h 5h

5 kWh 0.5 kWh

57

Table 9.2. Electric consumption for domestic lighting. A plausible total is 5.5 kWh per home per day; and a similar figure at work; perhaps 4 kWh per day per person.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

58 has an average power consumption of 42 000 W. So having the lights on while driving requires 2% extra power. What about the future’s electric cars? The power consumption of a typical electric car is about 5000 W. So popping on an extra 100 W would increase its consumption by 2%. Power consumption would be smaller if we switched all car lights to light-emitting diodes, but if we pay any more attention to this topic, we will be coming down with a severe case of every-little-helps-ism.

Sustainable Energy – without the hot air £170 £160 £150 £140 £130 £120 £110 £100

The economics of low-energy bulbs Generally I avoid discussing economics, but I’d like to make an exception for lightbulbs. Osram’s 20 W low-energy bulb claims the same light output as a 100 W incandescent bulb. Moreover, its lifetime is said to be 15 000 hours (or “12 years,” at 3 hours per day). In contrast a typical incandescent bulb might last 1000 hours. So during a 12-year period, you have this choice (figure 9.3): buy 15 incandescent bulbs and 1500 kWh of electricity (which costs roughly £150); or buy one low-energy bulb and 300 kWh of electricity (which costs roughly £30). Should I wait until the old bulb dies before replacing it? It feels like a waste, doesn’t it? Someone put resources into making the old incandescent lightbulb; shouldn’t we cash in that original investment by using the bulb until it’s worn out? But the economic answer is clear: continuing to use an old lightbulb is throwing good money after bad. If you can find a satisfactory low-energy replacement, replace the old bulb now. What about the mercury in compact fluorescent lights? Are LED bulbs better than fluorescents? Researchers say that LED (light-emitting diode) bulbs will soon be even more energy-efficient than compact fluorescent lights. The efficiency of a light is measured in lumens per watt.

I checked the numbers on my latest purchases: the Philips Genie 11 W compact fluorescent bulb (figure 9.4) has a brightness of 600 lumens, which is an efficiency of 55 lumens per watt; regular incandescent bulbs deliver 10 lumens per watt; the Omicron 1.3 W lamp, which has 20 white LEDs hiding inside it, has a brightness of 46 lumens, which is an efficiency of 35 lumens per watt. So this LED bulb is almost as efficient as the fluorescent bulb. The LED industry still has a little catching up to do. In its favour, the LED bulb has a life of 50 000 hours, eight times the life of the fluorescent bulb. As I write, I see that www.cree.com is selling LEDs with a power of 100 lumens per watt. It’s projected that in the future, white LEDs will have an efficiency of over 150 lumens per watt [ynjzej]. I expect that within another couple of years, the best advice, from the point of view of both energy efficiency and avoiding mercury pollution, will be to use LED bulbs.

£90 £80 £70 incandescent

£60 £50 £40 £30 £20

low-energy

£10 0

2

4

6 8 10 12 years

Figure 9.3. Total cumulative cost of using a traditional incandescent 100 W bulb for 3 hours per day, compared with replacing it now with an Osram Dulux Longlife Energy Saver (pictured). Assumptions: electricity costs 10p per kWh; replacement traditional bulbs cost 45p each; energy-saving bulbs cost £9. (I know you can find them cheaper than this, but this graph shows that even at £9, they’re much more economical.)

Figure 9.4. Philips 11 W alongside Omicron 1.3 W LED bulb.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

59

9 — Light

Mythconceptions “There is no point in my switching to energy-saving lights. The “wasted” energy they put out heats my home, so it’s not wasted.” This myth is addressed in Chapter 11, p71.

Notes and further reading page no. 57 Street-lights use about 0.1 kWh per day per person. . . There’s roughly one sodium street-light per 10 people; each light has a power of 100 W, switched on for 10 hours per day. That’s 0.1 kWh per day per person. – . . . and traffic lights only 0.005 kWh/d per person. Britain has 420 000 traffic and pedestrian signal light bulbs, consuming 100 million kWh of electricity per year. Shared between 60 million people, 100 million kWh per year is 0.005 kWh/d per person. – There are fewer signs and illuminated bollards than street-lights. [www.highwayelectrical.org.uk]. There are 7.7 million lighting units (street lighting, illuminated signs and bollards) in the UK. Of these, roughly 7 million are street-lights and 1 million are illuminated road signs. There are 210 000 traffic signals. According to DUKES 2005, the total power for public lighting is 2095 GWh/y, which is 0.1 kWh/d per person. – 55%-efficient generator – source: en.wikipedia.org/wiki/Alternator. Generators in power stations are much more efficient at converting mechanical work to electricity.

Bulb type

efficiency (lumens/W)

incandescent halogen white LED compact fluorescent large fluorescent sodium street-light

10 16–24 35 55 94 150

Table 9.5. Lighting efficiencies of commercially-available bulbs. In the future, white LEDs are expected to deliver 150 lumens per watt.

Copyright David JC MacKay 2009. This electronic copy is provided, free, for personal use only. See www.withouthotair.com.

10

Offshore wind The London Array offshore wind farm will make a crucial contribution to the UK’s renewable energy targets. James Smith, chairman of Shell UK Electric power is too vital a commodity to be used as a job-creation programme for the wind turbine industry. David J. White

At sea, winds are stronger and steadier than on land, so offshore wind farms deliver a higher power per unit area than onshore wind farms. The Kentish Flats wind farm in the Thames Estuary, about 8.5 km offshore from Whitstable and Herne Bay, which started operation at the end of 2005, was predicted to have an average power per unit area of 3.2 W/m2 . In 2006, its average power per unit area was 2.6 W/m2 . I’ll assume that a power per unit area of 3 W/m2 (50% larger than our onshore estimate of 2 W/m2 ) is an appropriate figure for offshore wind farms around the UK. We now need an estimate of the area of sea that could plausibly be covered with wind turbines. It is conventional to distinguish between shallow offshore wind and deep offshore wind, as illustrated in figure 10.2. Conventional wisdom seems to be that shallow offshore wind (depth less than 25– 30 m), while roughly twice as costly as land-based wind, is economically feasible, given modest subsidy; and deep offshore wind is at present not economically feasible. As of 2008, there’s just one deep offshore windfarm in UK waters, an experimental prototype sending all its electricity to a nearby oilrig called Beatrice.

Shallow offshore Within British territorial waters, the shallow area is about 40 000 km2 , most of it off the coast of England and Wales. This area is about two Waleses. The average power available from shallow offshore wind farms occupying the whole of this area would be 120 GW, or 48 kWh/d per person. But it’s hard to imagine this arrangement being satisfactory for shipping. Substantial chunks of this shallow water would, I’m sure, remain off-limits for wind farms. The requirement for shipping corridors and fishing areas must reduce the plausibly-available area; I propose that we assume the available fraction is one third (but please see this chapter’s end-notes for a more pessimistic view!). So we estimate the maximum plausible power from shallow offshore wind to be 16 kWh/d per person. Before moving on, I want to emphasize the large area – two thirds of a Wales – that would be required to deliver this 16 kWh/d per person. If 60

Figure 10.1. Kentish Flats – a shallow offshore wind farm. Each rotor has a diameter of 90 m centred on a hub height of 70 m. Each “3 MW” turbine weighs 500 tons, half of which is its foundation. Photos © Elsam (elsam.com). Used with permission.