The Best Way to Fail: Smart Charging vs. Everything Else

April 14, 2016 | 1:56 pm
Peter O'Connor
Former contributor

In my last post, I noted the role of competition in selecting between the various low-carbon energy solutions. Competition also applies to my own research topic. It might or might not win.

I look at “smart charging” of electric vehicles as a means to provide several grid services, including managing the short-term variations in renewable energy supply. However, other competing solutions for grid integration of solar power could be even cheaper. This would be a good problem to have.

A solution in search of a problem?

The concepts of “vehicle-to-grid” and “smart charging” serve to address the problem of matching electricity supply and demand. Wind and solar power are variable energy resources (VER), not typically dispatchable, so when supply and demand are not aligned on the short timescale you typically need to do one of three things:

  • Generate or curtail electricity
  • Increase or decrease demand
  • Use short-term storage such as flywheels or batteries for “smoothing”

One-way “smart charging” and other flexible loads are the second option; two-way “vehicle-to-grid” is more like the third. Dedicated storage is not the only option, nor is it necessary for moderate levels of variable renewables. At very high levels of VER, we might need dedicated energy storage systems to smooth out the variations. But what are “very high levels”? Actual experience keeps showing that threshold to be higher than we used to think it was. (There are somewhat different solutions for variability on the timescale of hours, as I’ll discuss in another post.)

At low levels, you integrate VER using the exact same mechanisms that you use to account for variations in load or the intermittency of traditional generators (which go offline for all sorts of reasons).

At moderate levels, you rely on improved prediction of wind and solar power output. Backup power doesn’t need to be kept running all the time; that myth implies that the short-term variability is totally unpredictable, that wind and solar power could drop instantly to zero at any second. We have no idea what makes the wind blow and the clouds form! It could be the whims of Aeolus! Actually, we know these things pretty well.

Wind is not, in fact, due to the whims of Aeolus.

Wind is not, in fact, due to the whims of Aeolus.

Power grids typically employ several rounds of scheduling the runtimes of power plants, refining the dispatch schedule as the operating time gets closer. The weather affects demand for electricity, so weather forecasting is already part of this process. We can predict wind farm power output with about 80% accuracy one day ahead, and 93-95% accuracy one to two hours ahead.

To account for this forecast error, we can keep a small amount of power on “spinning reserve.” Or, as I’ve suggested, we address this mismatch from the demand side, using flexible loads such as electric vehicles. In an alternative way of managing electric demand, maybe water will be the key—water heaters, pumps, and ice-based commercial AC systems—and EVs won’t play much of a role. Maybe demand-side resources will all compete against dedicated storage or flexible generation.

This competition already happens happen in a regulation markets where generators and end users can compete to provide the short-term small-scale adjustments needed by the grid.

Technologies compete in the regulation market to manage short-term variations in supply and demand.

As we add more variable energy on the grid, it is reasonable to expect that the market for regulation will increase; a 5% forecast error in wind or solar output requires more backup when those resources are 50% of your generation than when they are 10%. However, improved geographic diversity and improved prediction ability will counter that to some degree (perhaps we will have only a 2% forecast error by the time we reach 50% VER). And, it’s possible that we will have enough dispatchable renewable generation (such as hydropower, biogas, or geothermal) to fill in those gaps.

An insurance policy when forecasting costs

The value of smart charging as a flexible load is determined by how difficult it is to integrate large amounts of variable renewables into the grid.

If VER integration is easy, then there isn’t much financial value in providing a flexible load, so smart charging might not be worth the incremental cost of communication and control capability. If VER integration is difficult, then there is value to be had in providing that service. I expect flexible EV charging to be capable of providing this service at a lower cost than dedicated storage, thereby reducing the need for dedicated storage in a high-renewables future (as noted here). But it might be higher-cost than other options like “smart water heaters.”

Let’s say we have a range of estimates for the integration cost of solar, ranging from $1 to $10 per megawatt-hour (MWh). Suppose “smart charging” lets us integrate solar at $3 per MWh. If the actual integration cost comes in at the low end, then “smart charging” isn’t needed. If the integration cost comes in at the higher end, we have a product that can provide that balancing at a lower cost.

So in this sense smart charging is like an insurance policy. Better to have this option and not need it than need it and not have it.

What is the “best way to fail”? If my option is a low-cost and effective way of balancing short-term variations of renewables on the grid, but something else comes along that is even better.