Some Economics Of Solar Power

Suppose we wanted to generate a significant proportion of Australia’s electricity from solar power. What would it take and how much would it cost?

We could replace a significant proportion of our coal fired generation for roughly the cost of the NBN, create an entire export industry, profit from carbon trading instead of losing out from the current tax and keep the coal for much more valuable uses like adding to iron for steel exports, or eventually, making carbon nanofibres.

I don’t intend here to compare the current cost of solar power with that of coal, gas, nuclear, hydro, wind etc., because with proper investment, the costs of solar will greatly reduce. What I want to do is work out how much it would cost to substitute a large percentage of our current coal fired electricity generation, both in terms of dollars and land area.

Firstly, how much electricity do we actually produce in Australia?

In 2009, it was 261 TWh (terawatt hours). Of this, 19 TWh was used by the power stations themselves, 17 TWh was lost in transmission and distribution and 11 TWh was consumed by the energy sector itself. This leaves 214 TWh available for all commercial, industrial, government and private consumption ie. only 82% of electricity produced ends up being available for end user consumption.

That’s in 2009. In recent times, electricity production has been increasing at around 3% p.a., which implies that 2012 production will be approximately 285 TWh, of which 233 TWh will available for non energy sector consumption, rounding to the nearest integer and assuming no material increases in the efficiency of production and distribution.

Electricity consumption should increase at a bit less than the increase in real GDP, which is about 3 - 5% (about 1.5 - 2% from population growth and the rest from productivity gains due to better technology etc.).

Our total electricity generation capacity in 2009 was 56 GW (gigawatts). This doesn’t mean we actually use that full capacity. In fact, we rarely do. On average we use about 70% of our electricity capacity ie. an average output of 39 GW.

Of the 56 GW of power generating capacity in 2009, 30.3 GW came from coal fired power stations. These are at nearly full capacity pretty much all the time, so in 2009, we actually got 78% of our total power produced from burning coal, despite coal fired power stations providing only 54% of total generating capacity.

As far as renewables go, in 2009, we had 7.1 GW of hydroelectric generation capacity and 2.5 GW from solar, wind and biomass. The rest is gas and similar. The gas and hydro power can be on almost all of the time, but solar and wind cannot.

Suppose we decided to replace coal with solar and wind, in equal measures. Recall that 78% of our actually generated power (as distinct from theoretical capacity) comes from coal, so we would need (in 2012) 91 TWh per year from each of solar and wind available for end users.

A significant percentage of solar power is consumed at the place of generation. There would be a significant reduction in the 52 TWh or 18% of total production lost to inefficiencies. Let’s say we could reduce losses by two thirds in the case of solar ie. to 6%. That means we would need to generate 97 TWh from solar in 2012, increasing at approximately 3 TWh each subsequent year.

For the sake of definiteness, I’ll work with the 2013 target of 100 TWh. Being slightly more than one third of the total electrical energy requirement, this is about the upper limit of what can be generated from solar, since even spreading the grid across the country and using batteries (chemical and thermal), it's very difficult to get more than 8 effective hours per day of solar generation.

Let’s start with photovoltaic cells. I’ll discuss solar thermal power later. How much electricity do solar cells generate?

At 100% efficiency, about 1 KW / m². The most efficient, commercially available solar panels are the Sunpower E20, with an efficiency of approximately 20%. Let’s suppose that with improvements in technology, we could get to 25% in a few years. That’s 250 W / m².

Solar cells don’t generate electricity at full capacity all day. The best ones produce about 5 hours worth each day. That’s 1.25 KW / m² / day, or 456 KW / m² / year, a significant percentage of which would need to be stored in batteries for later use.

So, if we could store all the electricity produced by solar cells, or if we just fed the power not used at source into the grid and generated our nocturnal power from hydro and gas, we would need 220 km² of solar cells to hit our target.

Could we even do this?

The advantage of solar electricity generation is that it is easily distributed. The idea would be to make residential, government, recreational and commercial buildings generate as much power as possible from their rooftops, then add in banks of medium and large solar photovoltaic or solar thermal arrays in rural locations.

Large, industrial users such as aluminium smelters (which account for more than 10% of Australia’s entire electricity demand) could draw some of their daytime power from solar fed back into the grid, but realistically, they would continue to derive their electricity from large, gas fired stations.

There are approximately 9 million residential dwellings in Australia, of which approximately 7 million are free standing houses and 800,000 are semis or townhouses¹. Most free standing houses can fit 10 normal size solar panels on the roof. That’s 16m² per house, with semis and town houses about half that. Suppose you achieved a 50% take up rate for houses and 25% for semis and townhouses.

That’s 57.6 km² of solar panels, just on residential housing. Increase the take up rate to 75% for houses and 32% for semis and townhouses and we get to 86 km².

Now add in all the rooftops at schools, office buildings, local shops, factories, sporting pavilions etc. I don’t have figures on the number of each, however it should take us to halfway: 110 km².

The remaining 110 km² would need to be generated by large solar cell arrays, or their solar thermal equivalents. Most likely, this would be split between say, 60 x 1 km² large arrays and 5,000 x 0.01 km² smaller arrays. Allowing for spacing between the panels, the larger arrays would each need about 2 km² = 200 ha = 500 acres of dedicated land and the smaller ones 2 ha = 5 acres each. 5,000 might seem like a large number of 5 acre plots to find, but when you realise there are 15,000 suburbs and country towns in Australia, it puts the number into perspective. There would be plenty of 5 acre plots not suited to farmland not far from many country towns. Alongside irrigation channels would be sensible as it would lower the evaporation rate.

Even if we were only able to meet 20% of the 220 km² target from rooftop arrays, we would need 12,000 medium size (5 acre) arrays instead of 5,000. It is only feasible to substitute solar thermal generators for the large photovoltaic arrays.

Logistically, the placement of the solar generation capacity is feasible … and that’s replacing half of our current coal fired electricity generation with solar. Even finding the extra 13 km² of land each year for the extra 3% capacity will be fairly easy for a few generations, by which time we should have finally cracked the fusion puzzle.

What about the cost?

The Sunpower E20 panels cost around $1,000 / m². Add at least 50% for frames, wiring, batteries, DC à AC inverters and installation. Without economies of scale, the cost would be $330 billion. Totally infeasible.

But there would be huge economies of scale. We would need to produce something like 140 million solar panels just for domestic consumption, plus another 4 million per year, plus exports.

The panels could easily come down to $100 / m². Labour costs would not scale much, apart from efficiencies gained by installing all the cells in the one place in the large arrays. A Spanish paper comparing solar photovoltaic and thermal generation shows the current cost of thermal power is significantly lower than photovoltaic at Australian latitudes. Conversely, photovoltaic production costs are more responsive to economies of scale. Still, we might be able to reduce the all up installation cost to something like $250 / m², including the costs of hooking all arrays into the existing grid.

That’s $55 billion … over many years … to replace half our current coal fired electricity generation. About the same cost as the NBN and probably more valuable.

Who would pay for all this?

Firstly, the government would need to venture capitalise multiple solar cell manufacturing plants. If the entire project took 20 years, Australia would need to quickly develop a manufacturing capacity of 11 million solar panels per year (remember that we need 4 million per year just to meet the 3% increase in demand). Any exports would require extra capacity again.

The only way to raise the capital for such large manufacturing plants is with the federal government at least as a partner. It could use part of the future fund, or issue government bonds.

There is nothing wrong with the government acting as a venture capitalist. It happens all the time: Telstra, the Commonwealth Bank, CSL. The real debate should be about when the government should sell its shares to free up the capital for future investment and return profits to the taxpayers in the form of better services and / or reduced taxes.

By legislation, the government could guarantee the success of the industry. The government would build and operate the large arrays. It could sell them later if the public agreed. All new buildings would be required to have solar generating capacity. Existing buildings wishing to install solar panels could qualify for government loans at the government bond rate.

It is not wrong for the government to pick winners in terms of future industries and technologies, provided there are strong, independent arguments implying their benefits: the export and tax revenue, the employment creation, the use of non-arable land, the shielding of irrigation channels, the profit from carbon trading. Free movement of private capital does not always make the correct decisions and often cannot respond to the demands of the scale of both investment size and horizon.

Additionally, if you want to give Aboriginal communities real benefits, independence and land rights, install some of the arrays on their land, funnel part of the royalties into providing services and let them do what they want with the rest.

Afterword: Germany is producing a solar power output of 22 GW right now. They probably get closer to 4 effective hours per day, but if we installed the same capacity in Australia, at 5 effective hours per day, that's 40 TWh per year. The Germans are already at 40% of the very high solar target discussed above, even with the current high cost of solar energy.

¹ Source: RP Data