When you are dealing with intermittent, unpredictable, sources of energy, you can no longer apply the same categories as with the currently prevalent modes of generation. If you want solar power to be competitive without subsidies, grid parity is not enough – at least not if you define grid parity as the average cost of generation being equal to the average price.
In order to be truly competitive they must not only deliver the same quantity of energy at the same average price, but they must compete at actual market prices. This is very likely to be a relevant factor even for private consumers, if plans to introduce a “smart grid” work out.
It is impossible to sell solar power on long-term contracts for most applications, unless those can accommodate an unsteady supply of power – such as pumping water into reservoirs or land drainage, a field on which it competes with much cheaper wind power. Wind power has a very successful history in those areas, literally millions of wind-driven water pumps were installed in the USA in the early 20th century to pump water for cattle and other purposes – and of course there are the famous wind mills of Holland used to drain the polders.
But even the success of wind power in those very convenient applications is based on using overpowered pumping stations that are simply shut off when enough water has been pumped. This is necessary to ensure that enough water will be pumped at all times, despite the unreliable supply of power. Thus a part of the potential energy simply went to waste, which is a recurrent theme in this area.
Grid parity must not be reached with respect to the cost of energy that could potentially be generated, but with respect to the amount of that energy that is actually worth using. Any installation using an unreliable and/or fluctuating supply of energy must be oversized and will be more expensive and less efficient than one that has access to a constant reliable supply.
The Principle of Continuous Operation
Continuous operation is one of the key elements of energy efficiency in almost all walks of life. An express-train running non-stop requires less time and less energy than a train making several stops. In the example of the water pump, even if you had reliable and predictable supply of energy for the first half of the day and none during the second (a luxurious condition compared to the realities of wind or solar power), this would still mean that instead of one pump you would need two in order to do the same amount of work.
More complicated processes also need a lot of energy just to stop or to start up. A simple example is an industrial bakery. Ideally, you would want to keep running the ovens continously, because once you turn off the power, the heat in the oven below a certain temperature is lost – just as the heat required during the initial heating process. The same applies over all of the rest of the production line and again, you would need two bakeries to do the job of one. (Ok, more realistically the ratio might be 5:3 instead of 2:1, considering that maintenance would be done during the downtime.)
Just because of that effect, using such a kind of energy supply will not be worthwhile for many applications – it will, in most cases, even be unreasonable if the energy were free of cost. It is rarely the case that energy accounts for the majority of the cost of running a process and even rarer that the expense can justify additional investments.
Even were the investment is worth it, it will lead to a loss of efficiency. I’m quite sure this process could be modelled as something akin to a Markov Chain process, similar to the way that Shannon modelled the maximum information that a given channel could transmit. Unfortunately, this kind of math is not my strong suit.
Wind power is certainly the cheapest of the renewables that can be deployed on a large scale, but it is also the least reliable. Load factors (the ratio of average to maximum power generation) vary between 20% on land and 30-40% in off-shore parks.
Unfortunately, the times at which maximum power generation is reached, varies all over the place. There can be several weeks of lulls without power generation. Worst of all, there is no good co-relation between power generation and power demand. Dutch windmill owners during the Dutch golden age were exempt from the religious prohibition of work on Sundays. Basically, they worked whenever the wind was blowing, including at night.
You can do this to small group well-paid people, but telling a whole nation to concentrate on working when the wind provides energy is simply not feasible. (No holidays for you, there’s an atlantic depression moving in next week!)
Of course, this is not an argument not to use wind energy, but you do run into problems doing so. Germany had an installed wind power of 27 GW, providing on average 4.5 GW of power. As you can see on the graph, since 2007 the amount of power generated has stayed constant, even dropped, despite a constant increase of installed a capacity. We’ll come to talk about this fact.
Power consumption in Germany varies roughly between a minimum of 40GW at night and a maximum of 100GW. The average is 70GW. On a windy night, wind power could provide more than half, even two thirds of the power. But it won’t. That’s because coal fired and nuclear power plants can’t be shut down over night and started up during day. They can, however, reduce their power output and they will, because the supply of wind power is driving prices down during that time, below the cost of generation. If the supply of wind power, that is not actually sold on the market, is so large that it outstrips the demand left even after all other power stations are running at their minimum, wind turbines have to be shut down to avoid overloading the grid.
So is it all the fault of coal and nuclear power being so inflexible? Well, just ask yourself what would happen if Germany were to triple its wind power capacity. Then the installed peak power would be about 80GW. On a windy night, half of them would have to be shut down for lack of demand. The average power generation, assuming load factors of 20-25% would be 16-20GW, about one quarter of the average German consumption of electricity.
In 2010, however, the load factor had already dropped below 17% because of what was just discussed. A higher capacity of generation would reduce load factors even further – we are dealing with yet another case of diseconomies of scale. A doubling of installed capacity does not double the amount of useful energy generated.
I will now discuss the problem for solar power. Energy storage will follow afterwards.
The one saving grace of Solar power is that at least in warmer climates, it provides a lot of power when it is needed most for air conditioning (in summer, at noon). Northern latitudes, that are dominated by need of heating in cold weather, are not so lucky. They need energy most when solar power is at its minimum (in winter, at night).
There is also a co-relation between the time the sun shines and the time most power is needed, since people understandably prefer to work during the day. This means that maximum power is reached when power demand and prices, ceteris paribus, are above average. But, as I like to say, ceteris is rarely paribus. (Circumstances rarely stay the same when you change something.)
However, capacity factors of solar power are even worse than for wind power. They vary between about 10% in the sunnier places of Germany and a theoretical maximum somewhere below 25% in a desert. That’s because not only 50% of the day is night, but the power during the day depends on the angle of the sun. The math is somewhat involved – it depends on latitude, season, the direction the solar panel is facing and whether or not it can be moved. But roughly speaking it is zero at night and a sine function during the day, modulated by the vagaries of weather and celestrial mechanics.
If Germany wanted to install enough solar power capacity to provide 20% of its total electric energy needs (about 15GW on average), it would not be sufficient to install a capacity of 150GW, as a load factor of 10% would imply. The argument is the same as for wind power. The maximum power demand of Germany is 100GW. Not only would 50GW go to waste at noon of a sunny day, even if all other power stations were switched off, but this amount is also entirely dependent on the weather.
It isn’t much of a hyperbole to say that solar power would either literally be too cheap to meter (hence, not worth selling) or just not there. Of course this means that under such circumstances prices will no longer be highest, but lowest at noon just when solar power generates most of its energy. You drive up supply and prices plummet. The price of grid parity with solar power generation is much lower than what you would expect without it, even if solar power only makes up a few percent of total power generation.
3% of total average power generation with solar power means that you will get up to 30% of peak power generation from solar power. This makes for great propaganda. Hey you sceptics, see, we can do it! But actually *this* is the problem.
300GW of installed capacity could nominally provide over 40% of German electric energy, but only up to 100GW are actually needed – and even that only in winter. It’s somewhat less in summer. Trying to use the surplus will result in an inefficient use of energy, thus increasing the total demand of electric energy. But all scenarios for sustainable supply of electricity in Germany require *more* efficiency and a decrease in total use of electricity.
But up to here we have only considered daily variations. Depending on where you live, there are also seasonal variations. There is precious little sun to go around in Europe right now, because it’s the shortest day of the year. Worse yet, the sun is shining at a very shallow angle, providing even less energy than the reduced time of daylight alone would suggest.
This makes relying upon solar power as the main mode of energy generation a questionable proposition, unless the surplus of energy in summer could be retained for use in winter.
All the problems outlined above stem from the fact that there is no mechanism to store energy in the huge quantities involved. Again, using Germany as an example, we find that it has no more than 40-50 GWh of storage. Two hours on a windy day are enough to fill up all pumped storage reservoirs with wind energy only, if they all the reservoirs were empty before. The energy stored is way would last about half an hour during the day and one hour at night, if they had sufficient turbine capacities to actually generate 40 or 100 GW and pumps to use that power. (Which they don’t.)
If we wanted to use this storage to store an easily imaginable power surplus of 200GW between 9:30 and 14:30 for later use, we would not only need 1000GWh = 1TWh of storage capacity, 50 times the current amount – but of course also enough pumps capable of dealing with this amount of power without being turned into a smoldering heap of scrap metal.
Turning our eyes away from the fact that building pumped storage is frowned upon, protested against and sued to oblivion by German environmentalists (who keep pointing out that such things must be done in the Alps or in Norway – go figure), you may also be unpleasantly surprised that letting all the water run back from the reservoirs will only yield about 700-800GWh of electricity out of 1000GWh used to pump it up. The generation capacity that goes beyond the maximum power demand will have to be increased accordingly, if you want to keep using the same amount of electricity. So, we would have to increase the installed capacity of about 300GW by another 50GW just to compensate for losses in storage compensating daily variations – in a scenario in which solar power contributes less than half the total electricity required. (Again, those are not exact values, but they do show the kind of problem we’re facing.)
An efficiency of 70-80% is actually a pretty good value, as we will soon see. But given the sheer size of the required capacity (20-25 times the current amount), it is doubtful that even this much can be obtained. And we’re still only talking about compensating for daily variations and a partial supply of renewable energy. The larger the proportion wind and solar power, the larger the proportion of the generated energy that needs to be stored. Getting 30GW from wind energy and 30GW from solar yields installed capacities of 120GW and 300GW, even with 100% efficient storage. The figures will be more favourable for solar thermal power stations in the desert, which have higher capacity factors and are already using salt for heat storage up to several hours – but the cost of those is one of the main reasons for Solar Millenium going bust …
Longer term variations, both seasonal and the occasional lull in wind power, require much larger amounts of energy storage. We are no longer talking about 1TWh, but on the order of over 100TWh. It is reasonable to expect that in the best of all worlds, only half of the energy generated with solar or wind power could be used directly without storage – most likely less.
What can we do to store such massive amounts of energy? Pumped storage is out of the question. Highly efficient flywheels or batteries, mass produced in unprecendented quantities, could barely provide for daily needs. Unfortunately, high-efficiency means fly right out of the window when you have to store truly massive amounts of energy, unless new ones are put into practice (as opposed to merely being announced).
Such amounts of energy can be stored chemically, most plausibly using electrolysis of water to generate hydrogen. In this process, about 70% of the electricity is turned into chemical energy. Turning chemical energy into electricity, no matter if you’re using the most efficient fuel cells or gas turbines, is at best 50% efficient. This means we are losing two thirds of the energy just through conversion.
But that’s not all. Hydrogen gas is not actually storable, because it has a miserable energy density by volume. It must be liquified, losing at least another 20% of the energy for a total efficiency of 28%. Or hydrogen could be turned into another chemical compound. A prototype I recently saw on TV that turns hydrogen and CO2 into methane (for which we already have an infrastructure) was said to have 60% efficiency of converting electricity into chemical energy. But the problem is the CO2, it needs to be taken from the exhaust of the turbine. This process has already been developed for CCS, it is called oxyfuel and consumes on the order of 15% of the energy. But the methane must also be compressed, consuming another few percentage points – but who is counting? Combined with a turbine efficiency of 50%, the total efficiency is 25.5% already.
Similar things can be said for plans to use the well-developed Haber-Bosch-Process to turn hydrogen and nitrogen into ammonia and use this as storage. (Certainly the most convenient to store … and least convenient to be around.)
I must admit I haven’t done the math on the proportions of the energy that can be used directly, be used within a day or must be stored long-term. Especially since the requirements for windpower are even harsher than for solar power, I will assume that between one third and two thirds of the energy must be taken from longterm storage. I will assume storage for daily variations to be perfect and long-term chemical storage to reach an efficiency of 33%. (This is not an easy task, especially since turbines are already operating near the physical boundaries and energy losses in creating chemical compounds are unavoidable.)
In this case, compensating the losses of chemical energy storage would require average energy generation to be increased by 100% in the case of two thirds being taken from longterm storage and by 66% in the case of one third.
There is one question left that I cannot answer and this is the cost of the installations required to store energy. But again, we might assume those to be free and err on the side of being too favourable towards renewable energy.
And finally, to bring this long text to a conclusion, we can say that in order for renewable energy like wind power and solar power to be competitive with the more readily and reliably available forms of energy generation, their cost must not merely reach parity – they must be at most half as expensive as those.
But must they be? Preferably yes. The feed-in tariffs used today are distorting the markets and were a major contributor to the solar bubble currently bursting. Other forms of special subsidies are very likely to lead to different kinds of distortions, perhaps more benign ones.
Any effort to internalize externalities would certainly be appreciated, so long as this happens in an unbiased process. There is no need to talk about the impact of mining, oil drilling etc. they obvious and obviously an externality not reflected in the price of the energy. But the huge installations required by solar power (which I will detail some time in a hopefully much shorter post) or pumped storage have non-trivial impacts on the environment. (And biomass is another rant altogether.) We are talking about installations very much larger than the areas occupied by coal strip mines or the Alberta oil sands, for example. And while I’m convinced that no country has, of yet, build wind power on a scale that would do significant damage or be more than a local nuisance (although the population of Danemark or parts of Northern Germany might beg to differ), its potential is not unlimited either.