Generating power for the entire country comes with one fascinating quirk: the national grid produces the energy that is demanded at that particular moment. Phrased another way, power is used as it is created. In Xcel’s Pawnee coal plant in Brush, CO, a piece of coal is burned to drive a steam turbine, and some microseconds later, that energy is used to boil a kettle in nearby Wiggins.
From this reality of the grid, several consequences follow. The
country has to generate more energy when the Superbowl goes to commercial and
Americans turn on stoves and microwaves, so we have to have power plants that
can quickly ramp up their production to meet the sudden demand. This need has
been problematic in the UK when unexpected breaks in soccer games drive
millions to energy-hungry tea kettles.
The power plants that have the capability to ramp up and
down their production in response to demand are called load following power
plants. These load followers carefully track the energy needs of the grid they
are supplying and ramp their production up and down to meet the demand at that
moment. When the Superbowl starts, these are the plants that ramp up or turn on
to supply electricity to millions of TVs.
As well as the load followers, other forms of energy
generation are the more laid-back base load power plants. A great example of a
base load plant is nuclear energy. If you asked a nuclear power plant to
quickly ramp up production, they would ask how you got in, then say they can’t
easily change their output. These plants are kept on 24/7 to meet the base
load: the lowest expected daily energy demand.
The extreme example of load following plants are peaking
power plants, known as “peakers.” These plants are by default not producing
power and only turn on when the base load cannot be met by the base load power
plants. Their sole purpose is to smooth out intermittencies and ensure that
just enough power is being produced. These plants operate inefficiently to
prioritize a fast ramp up and ramp down time, and so the power they provide
commands a premium price, as demand is necessarily high.
So what happens if this system goes wrong and there is more
or less energy on the grid than is demanded? You might not have heard of the
electricity wars between Edison and Tesla (the inventors, not the companies),
but you have certainly heard of the band AC/DC. AC/DC got their name after
someone saw the initials on a sewing machine. It stands for Alternating Current
or Direct Current. Direct current is the kind you get from batteries, and
alternating current comes from wall outlets. It is called alternating because
it switches direction 60 times a second – 60Hz.
When more or less energy is produced than is needed, this
frequency changes, which could begin to wreak havoc on things plugged into the
national grid that are expecting a particular frequency. The grid is expected
to deliver within 1% of 60Hz frequency to businesses and residences here in the
US. One solution to this problem of shifting frequency is to have consumer devices like
refrigerators monitor the incoming AC current and when it deviates too much from
60Hz, the device responds by powering down to reduce the demand. Enough
consumer devices doing this will reduce demand back to the operating limits,
and save consumers a few dimes a year to boot.
At this point it is helpful to look at some data. The
California Independent System Operator (CAISO) is the main grid operator in
California, responsible for about 80% of the energy usage of the state.
Helpfully, they put all their data online, and update it every five
minutes, and keep records going back months.
To start, here is the demand curve: how much
electricity is needed throughout the day. Note that the y axis does not start
at 0:
No surprises here. Overnight, less energy is needed, and
there is a peak in the morning as people are getting up and in the evening
after getting back from work. A lot of this is from energy-intensive appliances
that heat up food and water, and relatively less from lights and TVs, as might
be expected.
The purple here represents the non-renewable resources. As
expected, there is a dip starting around 7 am, lasting until about 5pm. This midday dip in non-renewable energy demand comes from the sun hitting solar panels.
One thing that can be seen by looking at the Net Demand graphs from
many days is that California has been heavily investing in megawatts of solar
energy, and less so in wind energy. The effect of heavy investment in solar over wind is that there is a narrow margin of
renewables available when the sun isn’t out. December 16, 2019 (the day the data from these graphs was collected) wasn’t a
particularly windy day, but the point still holds when looking at relatively
overcast and windy days (look at Sep 9, 2019 for an example of a cloudy and
windy day).
This graph shows how California met its demand on December
16th. I didn’t include coal, batteries, or “other” as they are
insignificant to the overall supply, and I didn’t include Imports for clarity. The imports curve mirrors the natural gas curve, reduced by about 50%.
First to note is nuclear. The Diablo Canyon Nuclear Power
Plant provides a constant 1,100 MW to California’s overall supply 24/7. On December 16, its output varied between 1,102 and 1,106 MW. This nuclear plant is a base load plant that cannot ramp. The Diablo Canyon Plant is
due to shut down in 2024 or 2025.
Next is large hydro, or water behind dams. These can be
ramped quickly, but not indefinitely. The water level cannot vary too much, so
there is a natural limit to how long these plants can be ramped up - once the water level falls, it's time to shut down. On the other hand, if snowmelt
from the neighboring Sierra Nevadas comes through the dam, operators have no choice but to
let the water through. This leads to cheap energy prices in spring months. Large hydro is
somewhere between a base load and a load following power plant.
Renewables including wind, geothermal, biomass, biogas, and
small hydro account for a about 2,000 MW, and these can be considered
base load plants. Solar is intermittent, and so can be load-following, but the manner in which they can be load following seems a shame. To make a solar farm load following, build many more solar panels than you need, and connect or disconnect them as the load changes. This seems wasteful, however solar panels do have an unmatched ramp up and
ramp down speed; turning them on and off takes nearly no
time. California routinely overproduces and curtails solar power at certain
facilities during summer months.
Lastly is the hero of California’s load following energy
production: natural gas. Natural gas plants can ramp up and down production to follow the load far quicker than coal or nuclear plants. They are comparatively nimble and reactive. Natural gas is at this point absolutely indispensable. Without
it, California would be entirely reliant on energy imports from Nevada to match
demand.
One more curiosity from the CAISO database. There are times when the cost
of energy is negative. Before you get excited, you're not going to see a check from the energy company unless you are a wholesale buyer of energy, which you aren't. This negative cost arises due to excess energy being produced, and the
fact that energy is expensive to transport. In some cases, it is cheaper for the grid operator to pay
middlemen to offload excess energy than it is to transport or ramp down production.
Living nearby a large hydro installation in springtime benefits an energy
consumer. A typical large hydro plant might have fifteen room-sized turbines, 5 of which
are usually on, but while snowmelt is coming through all fifteen might be
producing power to get the water through the dam. This energy is cheaper to
offload locally than transport elsewhere. Note to supervillains with
energy-hungry doomsday weapons: find a valley in California and run your tests
in the springtime to save money on a big electric bill.
The whole reason to have base load plants, load-following plants, and consumer appliances that turn of and off depending on incoming AC frequency
is that we don’t have any way to store large amounts of energy. This is analogous
to trying to run a restaurant with no refrigerators, or a bar with no kegs – a logistical
nightmare. The challenge of large-scale energy storage is what stands between
our current power grid, and one that runs off intermittent sources of energy like
wind and solar.
There are some ideas floating around that may be able to
scale to meet some of the eventual demand for energy grid storage:
1)
Batteries. Creating batteries that power cities
rather than cars, watches, or laptops is difficult. Just wiring a lot of LiPo
batteries together isn’t going to cut it. These city batteries will need to be charged in the
summer, and hold that charge until winter with a near zero failure rate, and be
made from earth abundant elements.
<sidenote> "Rare earth elements" are poorly named. Most aren’t rare at all. Every "rare earth element" is more common than gold, and there is about as much selenium as there is nickel. The thing that makes them hard to extract (rare) is that they don’t
clump in exploitable deposits like gold or gemstones – they are more evenly distributed. To get an appreciable
amount, you need to dig up an process a lot of earth. </sidenote>
2)
Pump Hydro Storage. This is essentially pumping water uphill and
behind a dam during the day when you have energy, and letting it flow through a
turbine at night when you need it. You can also pump up extra water in the
summer and it stores perfectly well until it is needed in the winter. There is
a related idea that replaces the dam in a valley with an underground reservoir,
and trades out gravity for pressure: pump water underground at high pressure,
and release the pressure to drive a turbine when you want the energy back.
3)
Molten Salt. Using power generated during the
day to heat up salt until it is molten, as well as chilling antifreeze. This
creates an extreme thermal gradient and you can use this to drive a heat
exchanger to get energy out. This method doesn’t have a long shelf life – heat up
salt during the summer and it will be cold by winter.
The three above are highlighted in this video.
4)
Flywheels. During the day, use solar power to spin
up an enormous circular weight called a flywheel, storing energy in the form of
angular momentum. At night, apply the brakes to spin up a turbine.
5)
Vehicle-to-Grid. We are starting to heft around
pretty big batteries in cars and sometimes in wall mounted batteries in
garages. The idea is to use these batteries for ‘peak shaving:’ sending energy
to the grid when demand is high, and ‘valley filling:’ charging batteries at
night when demand is low.
Without large scale and reliable energy storage solutions, the
only way to move fully to renewable resources would be to fully supply our power needs even in the dreary January. Meeting the demand in January using
California’s current ratio of renewable energy sources would result in overproduction in
summer months several times over.
In the meantime, we need to meet the base load, follow
the load as it changes, all while not releasing too much CO2. The
path ahead seems clear:
·
Innovate and install energy storage solutions
·
Continue to invest in diverse renewable
power sources
·
Build more nuclear plants to meet the base load
·
Keep our natural gas infrastructure in place
until it’s no longer needed to follow the load
Cheers,
- Scott
