18-Month, Real-World Trial Proves Electric Vehicles Can Stabilize The Grid


BMW i3

Demand Response, Duck Curves and other Jargon

After driving my i3 for about a year I got an email from BMW; would I like to participate in a study? After taking a few moments to understand the details I realized I was being offered $1500 to drive my car and there wasn’t much in what might be labeled “effort.”

Additionally there would be an app on my phone and a box in my garage, but really all I had to do was charge my car and drive it. BMW would monitor the charging remotely and occasionally pause the charging, but for no more than an hour, and I could use the app to “opt out” any time I thought I might need the electrons. My participation in this study would be for 18 months and would help BMW and its partner PG&E learn how electric vehicles can stabilize the grid. Since I did the vast majority of my charging while I slept, this seemed low to no risk.

Me In My I3

Eighteen months later I can unequivocally state that participation was transparent, hassle-free and mind-numbingly dull to the point that I mostly forgot about it. The only reminders were occasionally meeting other participants over complimentary sandwiches during a group interview. Just the way you would want it.

So, What’s This All About?

One criticism about Electric Vehicles is that they will cause grid-instability. All those electric cars plugging in at once on an antiquated grid! Pandemonium will ensue, dogs and cats will be living together and civilization will plunge into darkness.

Except that’s not even remotely true.

EVs are a great way to stabilize the grid; it isn’t even difficult. It can (and has been) done on grid scale with 2015 technology. During this study I was one of 100 participants; when I’m one of a million participants it will truly be evolutionary.

The purpose of this study, a joint effort between PG&E and BMW, was to demonstrate what a lot of smart people believed but had never proven. Before I explain what that is, it will be instructive to explain why it is important.

One of the most fundamental laws of physics states that energy can be neither created nor destroyed; it can only be converted from one form to another. I wouldn’t be surprised if every engineer at California Independent System Operator (CAISO) has to get this tattooed on the inside of their eyelids as a condition of employment. That’s because it’s their job to ensure just the right number of electrons get pumped onto the grid as get consumed. In real-time. If they screw it up the grid goes unstable.

If your eyes are already glossing over, let me give you the CliffsNotes version of Demand Response in one chart.

Figure 1. California uses 5% of its grid capacity for only 50 hours per year. Courtesy PG&E.

An airline executive once said you don’t build an airline based on demand over the Thanksgiving holiday. Essentially the idea of Demand Response is to flatten the curve shown in Figure 1 by giving grid operators the ability to turn off loads. Going back to the airline analogy, think of Demand Response as overbooking; instead of asking for volunteers to be compensated for giving up their seats, the electric utility is asking for volunteers to be compensated for giving up charging their car (or running the AC and so forth).

Adding to the complication, not only does CAISO have to cope with cars plugging in but they also have to contend with legions of renewable energy sources that wax or wane depending on, of all things, the weather! As a bona fide rocket scientist, the engineers who have to manage this have nothing but my respect.

This phenomenon has been explained by the infamous “duck curve” (Figure 2): a juxtaposition of a day’s energy demand with generation from renewable sources superimposed. This net is the energy generation required by traditional energy sources (large hydro and what is known as thermal being two of the most common). Anticipating ever-increasing renewables coming online and extrapolating out a few years, the family of curves is said to look like the profile of a duck.

Figure 2 — The Duck Curve is the result of actual power demand with renewable sources superimposed. Courtesy of CAISO.

Just as the sun is setting and significant solar production is winding down for the day, people are rushing home, plugging in their cars, cooking dinner and turning on the air conditioning. In simple terms, renewable production is going down just as demand is starting to peak in the late afternoon and early evening, and a lot of power needs to come online, sometimes too fast to bring on new energy production. Figure 2 highlights a case from February 2016 of an almost 11 GW increase over just 3 hours.

There have been studies about EVs supplying energy to the grid (known as Vehicle to Grid or V2G). That’s not what we are talking about here. While V2G could be an important technology at some point in the future, it has a set of logistical headaches to deal with that frankly may never get worked out. Studies have shown, however, that giving control to grid operators over EV charging is 80% as effective as V2G but with far lower costs and fewer logistical headaches as acknowledged by Michael Liebreich in his address to the Bloomberg New Energy Finance Summit.

Demand Response programs are nothing new. Hydro-Quebec has been using them for decades giving financial incentives to residents to allow the utility to control their heating and air conditioning during periods when the grid is struggling to meet demand. Many other utilities participate in various other residential Demand Response programs.

What is new and different about the BMW-PG&E Demand Response partnership are manifold:

1- EVs create an interesting opportunity because the loads are higher than anything else in the average home. Multiplied by many homes, EVs gives the utility a very powerful tool at their disposal to curb peak consumption at critical times.

2- EVs create an interesting risk because consumers are not going to be happy if they go out to their cars expecting to be able to go about their business only to find that the utility has delayed charging on their vehicle.

3- BMW gave PG&E a powerful tool with the i3 because of the i3’s real-time telematics and the ability to pause and restore a charge session remotely.

4- BMW also gave PG&E a powerful tool because they guaranteed 100 kW of power reduction at PG&E’s demand.

5- Last but not least, BMW gave the ability for consumers to opt out of any Demand Response event, giving i3 owners the peace of mind they needed to participate in meaningful large numbers.

The BMW-PG&E Demand Response Program

The BMW-PG&E Demand Response program, formally known as the BMW i ChargeForward program, consists of two major components. The first is 225 kWh of battery storage at BMW’s Mountain View Technology Center, capable of delivering 100 kW of power back to the grid on demand. This battery storage is made of BMW Group 2nd Life batteries from the MINI-E program that were repurposed specifically for the BMW i ChargeForward program.

The second major component is nearly one hundred customer-owned BMW i3s in the San Francisco Bay Area. The real-world complication that this added was drivers may or may not be charging their cars when PG&E has a Demand Response, and those drivers may or may not have opted out of a Demand Response for that day.  But this was believed to be manageable and the program ultimately verified it was manageable.

After 18 months of the BMW i ChargeForward program, BMW and PG&E have now published an extensive report describing the results and also characterizing the drivers who participated. You can read PG&E’s blog entry summarizing this study here and obtain a full copy of their report here.

What Was It Like to Participate in the BMW i ChargeForward Program? What was Learned?

As previously mentioned, BMW gathered the participants together for group interviews several times over the 18 months of Phase One. There were several participants whom I would classify as “like me,” meaning nerdy to the point of unreasonable giddiness about numbers. But there were also plenty of people who just drove their cars neither knowing nor caring about the technology that made them work.

The program was designed to let people go about their business without worry or overhead. It excelled at this. During the group interviews, most admitted to forgetting that BMW may stop the charge on their car. It was just never an issue.

BMW gave us an app on our phones so we could see the Demand Response and opt out if we were charging. My experience was that initially I checked the app daily to see if there was an active Demand Response. But eventually I got bored and went weeks without peeking at the app. Most in the program went the entire 18-month duration of the program without ever being aware their car’s charging was being remotely managed.

And yet the data clearly shows that the program was a raging success.

Over the 18 months of the study there were 209 Demand Response events. Putting that into context, PG&E’s SmartRate residential Demand Response program is capped at 15 per year. PG&E is giddy at the prospect of broadening the program.

Figure 3 – The target for each Demand Response event was 100 kW back to the grid. This figure shows that on average, 20% came from reduced load by way of delayed charging and 80% came from BMW second life batteries. Courtesy PG&E.

Each Demand Response event was to deliver 100 kW back to the grid. Figure 3 shows that on average, 80% of this came from BMW’s second life battery storage and the remainder came from delayed charging of the 100 i3s in the test.

As one could imagine, the most participation in the program came from Demand Response events that occurred in the wee hours of the morning when most cars were actually charging. Figure 4 shows both the average number of cars participating as well as the power contributed, distributed by hour of day for the events.

Figure 4 – Participation in Demand Response events varied greatly depending on the time of day. Courtesy PG&E.

Most importantly, the program demonstrated that opt out rates were astonishingly low, as shown in Figure 5.

Figure 5 – Courtesy PG&E.

The program enlisted ~100 i3 drivers from around the Bay Area, but the number of vehicles participating in any given event was far fewer, as shown in Figure 6. Conversely, on average the typical vehicle participated in only 8 out of the 209 Demand Response events with one vehicle never participating and another participating in 70, as shown in Figure 7.

Figure 6 – Courtesy PG&E.

Figure 7 – Courtesy PG&E.

The full report gives far more detail than can be summarized here.

The BMW i ChargeForward program was so successful that Phase 2, now underway, has been expanded to 250 vehicles and includes not only the i3, but all BMW plug-in models. In Phase 1 of the program, charging was only monitored and delayed if the vehicle was at the owner’s home. It was assumed if the car was charging away from home that it was doing so out of necessity. Phase 2 of the program extends to all charging whether at home or in public, including workplace charging, and still gives owners the option to opt out.

During the final group interview it was made clear that PG&E wanted to incentivize participants to charge during daytime hours. One glance at the Duck Curve and it is easy to see why. Any utility is hugely incentivized to get EVs to charge during periods of large solar production and to stop charging as that production begins to wane for the day. Phase 2 is designed to learn the behaviors of drivers while they are charging away from home, perhaps with the hope of discovering how to incentive drivers to charge during the day. During my “exit interview” from Phase 1 I made it pretty clear that access to a plug during the day was what was required, and no driver incentive was likely to change that.

As an EV enthusiast I was excited to participate in the BMW i ChargeForward program, but even more importantly, I was satisfied knowing that BMW and PG&E were developing ways to make EVs the solution instead of castigating them as the problem.

Categories: BMW, Charging

Tags: , ,

Leave a Reply

42 Comments on "18-Month, Real-World Trial Proves Electric Vehicles Can Stabilize The Grid"

newest oldest most voted

Well done, John! Thanks for a) participating and b) taking the time to explain the program here for the InsideEVs community. It’s great to see BMW and PG&E partnering to do this!

This is actually pretty incredible. If they can solve for balancing the 10% of the time that is the highest peak level, they could get rid of 25% of their power plants and imported power. Considering that peek power is by far way more expensive than base, this could greatly reduce electricity costs for all customers.

The idea that EVs should charge during the day is already strange (and logistically difficult) and will become stranger as more storage comes online. I look at the CAISO charts frequently (especially on today’s hot day) and the duck belly is does not really look like what that graph shows 2017 would look like. Although some portion of this can be explained by solar and wind curtailments. In the summer in particular, the belly doesn’t fall (on average) below the low of the middle of the night and there are also no curtailments. http://www.caiso.com/outlook/SystemStatus.html We are not at the point where the net demand bottom at 9AM(ish) during summer merits moving charging away from night time. If you have smart charging (demand response) and your vehicle is plugged in all day you can deal with charging in the morning or the night. Although customers will always prefer to leave for work with a full battery. But right now most EVs are not plugged in at work. They just sit in an unpowered space. It would take a very large pilot program to change that. On an additional note, the Chevy Bolt has built-in a demand response mechanism. GM doesn’t specify… Read more »

Why is charging during the day logistically difficult? With workplace charging, it is often easier. Especially for those who live in apartments.

Because most parking spots at work aren’t connected to the grid. Even if companies have charging spots, they don’t have enough. A charging spot at work charges 2-3 cars a day. And that means 2/3rds of EVs aren’t plugged in at any given time. And that doesn’t even count the EVs (like mine) that don’t plug in at all at work because finding an EVSE spot (and moving when done) is too much trouble.

Pardon me if I’m skeptical about workplace charging. I’ve seen the problems it causes, all the fighting about who gets to plug in and who doesn’t. Unless you can afford to electrify every spot in all your work lots it’s going to have problems. And electrifying every spot has cost-efficacy issues since average spot will not have EVs in it between 16 and 24 hours a day.

The biggest bar to workplace charging are Plug-In Prius (PiP).

Workplace charge stations are by necessity, close to the work building in order to have access to electricity. This in turn causes any driver of a vehicle with a plug to hog the few available charging spots. PiP(s) are the worst offender.


When are we going to curtail plug hogging by gas/electric cars?

Our household has 2 cars – S and Leaf.

I have to say when a Volt or PIP has one of the work spots, I am pleased because the alternative maybe using gas. Sure, I don’t get to save a few pennies. My wife is always annoyed but they have small batteries – so need to charge at work more than I do. A few paid spots for emergencies would be helpful some day.

+1000 God how I hate the Plug in Prius AND the ford fusions. I swear they know their cars are done charging after an hour yet they stay in the charging spot for 12 hours at a time.

If only someone made a grid-scale battery that could be depleted during those hours to meet peak demands. In order for it to be cost-effective, it would have to be modular, self-contained, and built on existing, high-volume consumer manufacturing channels.

…..and 1/5th the current price.

It’s obvious that EV could stabilize the grid, but it’s interesting that the experiment actually proves it. Installing bunch of EVSE and promoting EV would be far cheaper than installing multi-megawatt gas turbine generator.

Gas turbines cost about $300/kW depending on size while 7.2 kW EVSE would be about $2000 to install include some trenching ($277/kW), making them comparable in install cost. But unlike gas turbine, EVSE doesn’t need fuel. Utilities could install EVSE for free to customers, and it could still be more cost effective than installing / running peaker generators.

This might be problematic for 80 mile-ish range EV, but 200+ miles EV can tolerate not charging to full every day. In fact, even 80 miles range EV could work well with the right pricing model and manual over-ride and wide availability of DCFC. Grid problem solved!

Very interesting, but ultimately there is a mismatch between car charging and solar/wind power, unless charging can be done during the day. It still seems that storage of energy for ~12 hours is going to play a key role in taming this.

Not really. If the majority of cars are EV, and they are plugged in when not driven (ie, >90% of the time), that’s practically limitless sink for power at any time.

That sink is somewhat limited by driver’s desire to have their cars be charged up when they go to drive them.

I’m not saying BEVs can’t be part of this solution, but when considering the cost and logistics it is important to also include a derating factor for batteries which have other reasons to store energy beyond balancing the grid. After doing this and considering the price of powering up every parking spot across a city we may find that dedicated battery installations make more sense.

And other smart grid applications may best them both.

eMotorwerks partners with commercial suppliers to allow this demand charging at work locations. Leave your car plugged in all day. It needs two hours, somewhere during the eight.

Great article John.
It brings into focus that its not just more renewables that are needed for a max renewable grid and the important role that EV’s and when they charge will play.

There was also a report from SoCal Edison that stated that it was better for the grid for the vehicle’s timer to be programmed based on departure time. The idea was that it would randomize the start time across the vehicle population because nobody would have the same amount of time required to fill the car, even if they were all set for the same departure time. If you take this thought to its logical conclusion, the driver should set what State of Charge they need by their departure time and the utility should schedule the power level to charge with and the hours when it should charge. However, this requires more data input because the scheduling algorithm needs the starting and ending SOC for each charging event. This would definitely require cooperation with the automaker to provide telematics data.

Wait, wouldn’t then by default the highest usage be in the morning, say when TOU rates expire?

How is everyone stopping at 6AM any different than say everyone starting at Midnight when TOU rates. Seems to me there would be a fairly equal demand at the highest point regardless of how you did it.

It’s more stressful on the grid for many cars to start charging all at once, say midnight, than to finish at the same time. Of course, your point is well taken that the TOU windows would also have to be considered in the scheduling. If the EVSE was smart and separately metered or sub-metered, you could also put it on a separate rate schedule that is always cheap as long as the utility gives the green light.

How so? Not many people have high power DCFCs so why would 1M cars starting charging at say 6.6 at midnight be more stress than 1M cars still charging and finishing at 6AM? Same amount of cars and electricity draw, just later in the AM and frankly when people are up and using more electricity so I would think it is an even higher demand.

Interesting. The LA Times had a story just now about how California sometimes PAYS other states (albeit mostly in winter) to take their excess solar-generated power during the day in order to keep the grid stable. Might it be more sensible for them to install remotely-managed free Level 2 or even Level 1 chargers at office parks and the like? Kick them on to draw down excess power. Kick them off if demand spikes. Drivers could sign up for a plan that enables them to pay nothing to charge at work, and pay a reduced price to charge with demand-based curtailments at home. Everybody wins: the grid is stabler, the consumer’s tank is fuller when they leave work (reducing range anxiety), and the utility and consumer both save money.

It’s not just solar. California has to curtail wind generation at times too.

I don’t know if your suggestion makes sense. Most of the time California is not curtailing wind or solar. Who is going to hook up to an EVSE that doesn’t charge their car most of the time?

Workplace EVSE at very low or free prices that only use “surplus” electricity could definitely have its place if it was mixed with normal EVSE that will always charge immediately, but has a higher cost. People would gladly leave their car there all day to eventually get a charge if they didn’t need the range boost quickly. The problem with most workplace chargers in Silicon Valley is that they are way over-subscribed so there is a lot of pressure to go move your car as soon as it’s done charging so somebody else can use it. Chargers subsidized by demand response could be installed in much larger numbers. They should definitely do some pilot projects like this.

What you describe is more similar to how this program works. It’s not similar to how California uses curtailment. California may go an entire summer without curtailing renewables. So people would be plugging in to get no charge at all for weeks.

I’m not saying BEVs are no part of this leveling system, but I don’t see this EVSE system described as a system to replace curtailment even partially. Sequencing EVSEs based upon demand would really more just another portion of a smart grid.

It’s great a role can exist for DR without the draw-downs of V2G. Depending upon the algorhythm, I’d probably volunteer for V2G as well. The problem, as you I think are getting at. People charge at night, for lack of daytime price or access. I’ve seen schools do simple Level 1 outlets, in great number, but that doesn’t seem to give ChargePoint or other venders what they want. It doesn’t show up in utility dockets because I believe the call on equipment is cheaper, and therefore less profitable for the utilties & vendors to chase. Practically speaking, 8 hour work parking at 1.5kw gets most people home, many both ways. Sorry to mini rant. AG’s of both R’s and D’s attacked VW and what do many want? Urban chargers, for people who have no home to plug into. As “plug in culture”, we see the logic of this article, and how it might extend to multiple, perhaps lower, watt-rated opportunities. I even see virtue in this, but political motion is instead attempting to chicken-or-egg lower income non-EV owners into battery cars. The irony? We might see VW’s 150KW chargers, and their expense, in urban areas where a material number of… Read more »

I think this and V2G could be a killer combination. Sure there was a study recently where V2G was detrimental to the battery however when EVs are ubiquitous the batteries will probably have become better and much larger. If you take just a couple of kWh from a million cars, each with 100+ kWh batteries it really can’t affect the car all that much but it will make a huge difference for the grid. Of course there should be a proper compensation model for this, that is one reason why I think utilities should be the ones installing chargers everywhere.

This duck curve is exactly why i charge when the sun is up. what goes into my ev doesn’t even show on my electric bill (neither as generation or consumption), it’s straight from the solar panels…

With Net Metering, I save money by using the utility as my battery. They give me more $ credits for my solar generation than I pay charging overnight.

Same here.

Another reason why net metering really is more of a way to make solar affordable (subsidy) than anything related to grid management. As solar becomes affordable without subsidy and we have more and more solar power put into the grid we should be reducing net metering subsidy and instead using money to encourage energy storage (home and grid). That’s how we get to where we can use renewable energy all night and start eating into the base load CO2 emissions.

Hey, for $1,500 I’d mess up my leased EVs battery too 😀

How would this mess up the battery?

No kidding. Someone didn’t read the article.

I read it and I do understand it doesn’t include V2G. But despite this sometimes the phrasing used is confusing:

“Each Demand Response event was to deliver 100 kW back to the grid.”

The cars don’t deliver any energy back into the grid. The batteries at the BMW center do so, but the cars do not do so. Each event makes available 100kW of power that wouldn’t otherwise be available, but strictly speaking the 20% of that from cars (on average) is not being delivered back to the grid, it isn’t being drawn from the grid.

You will find it expressed like this, where “DR” bids into a grid market as supply, even though that supply is effectively from their own curtailment.

the number of people that reply to the headline grows – More consistently since March of 2016.
Usual Suspects is very clear and present here, you get to know -very Soon- who to always scroll past (three strikes takes -at most- a week).
The Ignore button would be a welcome addition here, as the number of hourly self-impressed commenters has not lessened.

On a lighter and positive note, Thank you Mr. Higham for a Very down-to-earth summary of what appears to be a great program. Trust that Many people actually Did read your entire article and appreciate the effort!

Add Solar shingles and power wall to this mix. A majority of new homes, and over time replacement roofs will have major positive impact.

Solar balanced with workplace charging makes a lot of sense.

Sure CA isn’t curtailing much now but 5 years from now when solar is a higher percentage, they need somewhere for that electricity to go during an April day.

The renewable generation demand imbalance is a big deal and utility scale batteries are expensive even if getting cheaper.

A cool sunny day is a time to use solar and charge during the day. It helps the ROI on utility based solar. And that makes the utility happy and encourages them to keep building more solar. Win win.

The idea that to best help the Grid, and the renewables energies, is to charge the EVs in the afternoon is very important to stress. That means charging at the Work CarPark location. I would be great that Governments start incentiving the enterprises so they enable this charging at work carpark,… the sooner the better.

Excellent! Any idea why PG&E was chosen as the utility over say SCE or SDG&E in southern CA? Perhaps location relevant to BMW Mountain View?

The ever-growing duck curve shows California has a potential problem they are not really addressing.

And there is no reason for it because of the ubiquitous time-of-use metering installed. They simply need to disincentivize production, or greatly incentivize usage during this time.

2 cents /kwh from 9 am to 12 pm would alleviate this problem – of course greedy utilities don’t want to give up the money, and its of no concern to PG&E what the central stations wear problems are. Therefore CalISO has to act to provide proper pricing.

The ‘loads’ will then take care of themselves.