Why the "duck curve" created by solar power is a problem for utilities

https://www.vox.com/2016/2/10/10960848/solar-energy-duck-curve
lanjutannya: https://www.vox.com/2016/2/12/10970858/flattening-duck-curve-renewable-energy

Let's talk about the "duck curve," shall we? Everyone who cares about solar energy should know about the duck curve. Plus, it's fun to say. Duck curve. Duck curve.
The long story is below, but the short story is: The duck curve refers to the effect that solar power has on demand for utility electricity.
For many, many decades, demand for electricity followed a fairly predictable daily course, allowing utility grid managers to become experts at predicting and satisfying it.
The addition of large amounts of solar to the grid promises to fundamentally change the shape of that daily demand profile — in ways that make grid operators nervous about maintaining power and reliability. And in ways that make it look like a duck.
The duck curve is a problem, albeit a solvable problem. In my next post, I'm going to get into all the various ways to solve it. But for today, I'm just going to lay it out. And possibly make a bunch of duck jokes.
Let's start at the beginning.

Electricity demand used to have a predictable, manageable shape (namely a camel)

Demand for electricity varies throughout the day, but it does so in fairly predictable ways. It rises in the morning to a little hump before noon, levels out over midday, and then rises to a higher hump in the evening, when everyone gets home from work and turns on their TVs and stoves.
Here's a typical "load curve," from the New England region in October 2010:

(EIA)

If you squint, you can kind of see a camel's back, with its two humps.
The exact shape of this curve varies from place to place and season to season, obviously. In some times and places the humps are more pronounced; in temperate climates, with less heating and cooling demand, they're a little flatter.
But in most cases, load curves share a few key characteristics. There are two daily humps. Demand never gets too high or too low, meaning it stays within a reasonably manageable range. And the ramp-ups and ramp-downs of demand are fairly gradual.
For nigh a century, that's the demand utilities met, and they got really good at it.
For that baseline amount of energy that's always needed — "base load" — they run big power plants, usually nuclear and coal, around the clock. These plants are typically slow (and expensive) to start or stop, but cheap once they are running.
Then there's "intermediate load," with the next-cheapest tier of power plants, and at the top of that second hump, "peak load," satisfied by (usually natural gas) "peaker plants" that are expensive to run but easy to ramp up and down quickly.
It all worked out fine until wind and solar came along. They do different things to the load curve, though, and today we're focusing on solar.

Hello, Newman.

(Shutterstock)

The thing about solar power is you can't schedule it like you can a power plant. The sun shines when the sun shines, typically from morning to mid-afternoon. When the sun is out and a customer's solar panel's are generating energy, that customer is using less of the energy put on the grid by the utility.
In other words, from the grid operator's point of view, solar energy doesn't look like a power plant at all (those are controllable, or "dispatchable," in the jargon), it looks like a reduction in demand. It's a reduction in demand for the power supplied by the grid operator's power plants — a somewhat predictable reduction, but not a controllable one.
So now grid operators no longer have to supply total demand. They have to supply total demand minus solar power. Total load minus solar power is known a "net load." That's the new target utilities have to hit.
And when solar starts getting big, net load starts looking a lot different from old-fashioned load.

With lots of solar, load curves start looking like ducks

Which brings us to the duck curve.
A few years ago, the California independent system operator (ISO), or CAISO, put out a short paper on the duck curve that got a lot of attention. California has experienced the highest penetration of solar PV of any state and expects enormous growth in years to come.
In 2012, California's load curve was bopping along like normal, looking like a camel. To illustrate, I give you this delightful sketch from journalist Jordan Wirfs-Brock (from her great article on the duck curve):

(CAISO/Jordan Wirfs-Brock)

That light blue line tracing the camel's humps is the shape of California's actual 2012 load curve.
But in 2013, as solar ramped up, things started changing. Demand was suppressed more during the day, when the sun was up. And in coming years, CAISO expects the effect to become more and more pronounced, until the load curve starts looking like … a duck:

(CAISO/Jordan Wirfs-Brock)

Okay, fine, here it is without the duck:

(CAISO)

What's wrong with a duck?

One notable thing about the duck curve is that it wreaks havoc on the revenue of power producers and utilities. That gives them every reason to exaggerate its inevitability and its danger — remember that, we'll return to it later.
From the point of view of the grid operator, worries about the duck curve are threefold:

1) Steep, tall ramps

The ramps, those times when net load is rising or falling, no longer look like the gentle slope of a camel's hump. They get steep and tall (like a duck's back) and relatively quick.
That means grid operators are forced to take a bunch of power plants offline, or put a bunch online, rapidly.
What's especially unfortunate is that the sun tends to go down just before the evening peak of demand, which means net load goes from very low to very high, very quickly (13,000 MW in three hours, in the CAISO example), and then down low again.
Grid operators don't like steep ramps. It is expensive and highly polluting to turn a bunch of plants down (or off) and then crank them back up again all at once. It also makes voltage and frequency management more difficult.
Coal is not good in this role, as it is slow to ramp. Nuclear is proving a little more flexible in some places, but not so much in the US yet. For the most part, for fast-responding power plants, utilities turn to natural gas.
So California needs enough natural gas capacity to supply the evening peak, but for most of the midday, it doesn't need any of it. That amounts to a lot of natural gas plants sitting around a lot of the time, with low "capacity factors," but being ramped up and down frequently, increasing operating and maintenance costs.
That all makes grid operators grumpy.

Not an ideal load profile.

(Shutterstock)

2) Overgeneration and curtailment

When the duck gets really fat, its belly starts hanging closer to the bottom of the chart — net load gets closer and closer to zero around midday. That means all the peaker plants get shut down, all the intermediate plants get shut down, and some of the base load plants start to get ramped down too.
And then a few hours later, they all get ramped back up.
For one thing, that's expensive. For another, grids need a certain amount of reserve power online at all times as a buffer in case of accident or disruption. If so much solar comes online that it starts to eat into those reserves, solar will be "curtailed," i.e., the grid will stop accepting it. (Curtailment also happens for economic reasons.)
In Hawaii, where 10 percent of customers now have rooftop solar, the duck's belly has hit bottom a few times, as this story by Jeff St. John details. Check out the red line:

(GTM)

As you can see, net load was negative there for a few hours on August 8 — there was "backfeed" into the grid, which can mess with voltage and stability.
In Hawaii, the duck's back is so low, and the ramp up to its head so high, they've started calling it the "Nessie curve," after the Loch Ness Monster.
These worries have led Hawaiian Electric Co. (HECO) to pull back on solar and institute new interconnection standards. (Right now, somewhat insanely, the grid has no communication with most of those solar panels and no ability to control or predict them.)

3) Frequency response

For stability, the grid must closely balance supply and demand, second by second. Frequency is maintained at around 60 hertz. In case of a sudden disruption — the unexpected loss of a power plant, transmission line, or large load — the grid needs resources capable of ramping up or down quickly to compensate.
This is done by automated frequency response systems, usually on conventional power plants. If solar starts shutting down all those plants in the middle of the day, the grid loses those resources, and with it some stability.
(Right now, most solar systems do not have automated frequency response, but they are capable of it — more on that in the next post.)

Flattening the duck

Remember, solar is screwing up utilities' business model, but "we're making less money" does not move the hearts of regulators, especially when it's a response to customer choice.
So there's some incentive on their part to exaggerate the duck curve's status as a technical problem.

A flatter sort of duck.

(Shutterstock)

So it's worth stressing: There are lots and lots of ways to flatten the duck. From an engineering perspective, the problem is solvable, at least for the foreseeable future. There may be some level of wind and solar penetration where the cost of integrating them into the grid exceeds the benefit, but we are nowhere close to that level yet.
The next steps to accommodate more solar on the grid are clear. I'll cover some of them in my next post. Get excited, for duck's sake.