Energy Engineering Seminar: Daniel Kammen, University of California, Berkeley

Energy Engineering Seminar: Daniel Kammen, University of California, Berkeley

October 14, 2019 0 By Stanley Isaacs


Daniel Kammen, who’s a professor of energy
at University of California, Berkeley. Professor Kammen also serves
as the chair of the Energy and Resources Group, and
is the director of the Renewable and Appropriate Energy Laboratory at Berkeley. He is originally from Ithaca, New York. And did his undergrad degree,
I guess in 1984, if I remember right, in physics, here at Cornell.>>Like when dinosaurs
were the main topic.>>When most of us probably, well, not
most of us, some of us were still young. And then went on to do
a PhD at Harvard in 1988. Following his stint at Harvard,
he went to do a Post Doc at Caltech and I think went back to Harvard for
a little bit. But then joined the faculty in
the Woodrow Wilson school at Princeton, where he spent a few years. And then moved on to Berkeley about
20 years ago where he has done just phenomenal things. And so
we’re really pleased to have him here. Now his research focus is generally in
the area of science and technology policy with an interest in energy, developmental
and environmental management. He’s authored over 300 technical
papers and twelve books. He encompasses subjects
ranging from the energy, poverty, climate mixes,
a very interesting topic, bio fuels, portable dates and
smart grid standards. He has also helped found over 10
companies including End Phase that went public in 2012, and Renew Financial
which also when public is 2014. Since 1999, he has served as
a contributing lead author for the Intergovernmental Panel
on Climate Change. And you remember in 2007,
that panel here the Nobel Peace Prize. In 2010, he was appointed the first
energy fellow of the environment and climate partnership for the Americans by
then Secretary of State, Hillary Clinton. He then stayed on to serve as
the energy envoy for John Kerry. And resigned in a letter that went viral,
so if you want to know, go online, quite recently from that post. Now it’s my pleasure to
have him here today, in particular because I think
his mom is in the audience.>>[LAUGH]
>>And she was the one who introduced us. And she sent me an email saying,
my son is in town you need to meet him.>>[LAUGH]
>>And this was almost seven months ago. I reasoned that, well, if mom is onboard if I made
the call Daniel could only say yes. And he did, so
it’s my pleasure to have Daniel here. And before he takes the podium, I wanna
just point out that after the lecture, there’s a lunch and you’re invited. As many of you who can attempt
to attend because that gives us an opportunity to kind of pick his brain
a little more to learn what he knows. Because the purpose of this
class is to educate all of us. So, Daniel, thanks for coming.>>Layden, thank you so
much, appreciate it. [LAUGH] Thanks a lot. I know you have to slip out,
right, at one point?>>No.
>>You don’t? Well, okay, honored, all right. Well, thank you so much. And, obviously, I didn’t know the Layden,
mom back story was as strong as it was, but I appreciate both of them for doing
this and thank you all for coming as well. I am going to try to do a mix of
the technical inputs as you’ll see but really focus on some of the output
side of what we’ve done. Largely because we’re at a moment
that I’ll get to in a few slides, where we need to move much faster than
we’ve been doing in terms of innovation and implementation of
a whole variety of projects. So I’ll start with a short introduction
to my laboratory, because the range of projects we’re able to do was
a function entirely of the students in post docs who do some pretty
interesting and self-directed things. I’ll make a pitch which probably
I don’t need to make here but, I think I have to make it anyway on
a regular basis, about how critical it is that we accelerate the pace of
transition to a cleaner energy economy, not that that should be
a surprise to anyone. But I wanna highlight what at least I feel
we’re learning about innovation, both in terms of new materials and in terms
of some of the new data science tools. And then end that up and I, well, I’m sure I won’t get to this point
cuz by 1:15 we’re done, right?>>Yes.
>>Okay, so yeah, I’ll be racing at the point that I talk
about what we’re currently working on, which is really trying to use the mix
of science and technology that we do, to think hard about what we’re increasing
calling not just the energy transition, but the just transition. So a transition that has equity and
inclusiveness built into it in a way that certainly I credit a lot of things to
my physics training here at Cornell. I did not learn anything about just and equitable transitions in
my regular coursework. Not that they weren’t great,
but there was a lot more to do. So, in terms of the thinking,
I’ll start off with the laboratory, and it’s easy just to start off with
a kind of the picture of the team. It’s a fairly large,
at least by our standard research group, with people from all kinds of industry and
places. Navajo, Glenn, Chile, Columbia, Nigeria,
France, Nicaragua, from the developing country of Washington DC, and
all kinds of various spots along the way. And as Layden mentioned, I’ve been running
this lab at Berkley now for 20 years. I moved there from Princeton and
we set it up, and a lot of the initial work was on both
detection of climate change topics and increasingly on sort of the role of
what was then very expensive technology, solar and energy storage,
a variety of features. But we did get to work on a whole
variety of aspects of the IPCC, the Intergovernmental Panel
on Climate Change. And as you heard, we got to share
the 2007 Nobel Peace Prize, which was pretty cool, pretty great event. You got this great plaque. But as all of you know who read the books
about the Nobel Prize story, and particularly the book,
Losing the Nobel Prize, only three people can win a Nobel Prize,
that’s the law. And that means there were several
hundred of us that were involved in the IPCC at the time and
so you get credit for it. You don’t get the check, but
more importantly, certainly, at Berkeley campus. And since I drove here
today from Syracuse and was prevented by your high quality
gatekeepers from getting onto campus. I think you have the same
problem we do and that is that there is something more
valuable than the check on an academic campus if you get to be part of
a Nobel Prize team, and that is parking. And at Berkeley, anyone who’s visited
knows that we drive around UC parking spaces with a big,
huge NL on them, Nobel Laureates, are the only parking places
that are inside campus. And so, pretty exciting, and
Berkeley thought about it. There were four Berkeley faculty
who were part of the effort. And so, we thought we could get
parking spaces, and we were like, rubbing our little hands together. Inez Fung, who some of you know,
in atmospheric science. Kirk Smith, who does public health issues. And Bill Collins,
who does mesoscale climate, and so we were really kind of licking
their chops about that. And Berkeley did, nine years later, finally come through in
a very Berkeley way. We did receive parking space for
the big NL on it. It’s bicycle parking. And if you read the fine print. Parking not reserved for
IPCC contributors, all cyclists welcome. So it was a very Berkely-esque response,
and so Berkeley does its thing. So our lab has been involved in that. And we’ve been involved
in a lot of non-physics, non-engineering things along the way. I can only describe this best as
financial engineering, we worked on and came up with an idea to basically help
cities lend money to businesses or individuals to invest
in energy efficiency, solar projects, all variety of things. And I’ll come back to both the good
side of that and the bad side later. It works great, it got lots of press,
it’s now called PACE. We tried to call it Berkeley first, that name didn’t really
translate well out of Berkeley. It’s now called
Property Assessed Clean Energy, an investment mechanism that’s used in a
number of states and has done pretty well, and over the last Decade. And the thing that I’ll spend more time
in this talk talking about is that we’ve started to build really
data-heavy models of current power systems around the world and chart
out what they would need to do to transit quite dramatically to almost
entirely decarbonized lines. And so some of our current projects
With China, Morocco, Kenya, Malaysia are part of this kind of mix,
and it’s part of a bigger collaboration. So our lab has a whole variety
of community group partners, some non-profits. We work with a number of businesses, and I’ll come back to the projects
with Google in particular. Natel is a microenergy company
Virunga Power is the energy offshoot of Virunga National Park,
Africa’s oldest national park, and then the nations for which we built
these models are the ones shown here. And I of course put California down cuz we
can never figure out if we’re a country or a state or not, so
it’s confusing to us and to others. But we do this cuz we’ve really had this
huge range of students who have gone on to various jobs from President Obama’s science advisor to the first
person to have held the California public utilities commission that
basically sets the rates and laws. And also commissioner on the energy
commission that decides who gets the money. Magid is actually one of my early students
who is director of epidemiology at Imperial College, so really interesting
and diverse group of places. They’ve gone to PhD students,
there, masters students, as an undergrad examples here, and so all
the things that I’ll try to run through, you can get both on the website
of the laboratory renewable, inappropriate energy,
Berkeley EDU and then the Twitter. We try to do kind of one tweet a day,
about the things the lab is up to. So let me jump into the energy
systems transition story, and just because of all the greats of doing
here, I’m not going to spend time making the argument why we need
to adjust our climate system. I won’t do the climate history, but
I do wanna remind us kind of where we are, or were, or are going depending
how pessimistic you are about making good on the Paris Climate Accords. But basically the agreements
that went into the Paris Accord would take the global
trajectory from warming by over four degrees celsius too if everyone
did what they promised in Paris, we would cut about a degree
off that total, amazing start. Not as far as we have to go,
but a really good start. So this is what everyone’s committed to
and the sad story is that we’re now two months out from the COP24 meeting
in Poland, and only one country is on phase to meet the 80-plus percent
decarbonization that they caused. And since it’s so odd and surprising,
who knows what that country is?>>Is it Denmark?>>It is not Denmark. Denmark is behind schedule because
Maersk is 40% of Danish emissions and they do not have a solid plan. The one country is good guess though.>>Iceland.>>Nope, Iceland has way too many overseas financial holdings in places
they don’t control the emissions. That’s a good question.>>It is Morocco, the only country. So it’s an African country,
that’s the only country on pace and Morocco has many challenges, but they have the most successful large scale
renewable energy program in place so far. Now lots of things can go wrong,
but there’s a lot interesting in that statement that
Morocco, right now, is the leader. They’re the only one who was rated out
as fully on pace for their 2050 goals. The challenge, of course,
is that many of us are not on pace, and that that would only get us just about
halfway to this 2 degree C or less target. Recognizing that we’ve already warmed by
one degree, see, so we don’t have a lot of head room left, so this is the needed
innovations part of the story. And one very interesting
piece of science and policy, it got a lot of attention is that in the excitement after the Paris Accord
calling to keep us under 2 degree C. I’ll get to the 1.5 in a second, but
under the 2 degree C, Secretary Kerry, my former boss was the lead in the US
on getting the Kigali Accords passed, six months after Paris,
everyone signed onto a second round, largely industrial emissions and
refrigerants. And that would cut off
about another half degree. So the things that we’re now promised, but
arguably only one country’s really on pace for, would get us more or
less halfway with the complex story that two degrees obviously was gonna bring
us down to very close to zero emissions. And the next thing of course is
that now there’s this excitement around the 1.5 degree targets, and
I’ll get to them in a little bit. But what that means is that you have to
think about first bending the curve to get us off of this growth path, and then
think about bending the curve negative. And those are inflections that
are hard for business engineers to see happening easily, but
we have some interesting data, right? In California and the United States,
this is electricity use per capita, and if we look at, the California and
US were more or less growing the same. The OPEC Oil crises happened, different parts of the US adopted
different strategies in terms of thinking through clean energy transitions,
what we might do differently. California through a series
of different measures by, developing a state agency working
with university groups and industry to double and
triple down on efficiency for industrial, commercial and residential buildings have
done this remarkable transformation. So electricity use per person in
California has stayed flat for decades despite all of the incredible
proliferation of electronics. That was something that,
I won’t say my economist friend, if I say this way I won’t have
anyone anymore afterwards. But in the 60s and 70s, you could
not find a mainstream economist and say it was possible to
the couple growth and energy and had the GDP keep going, so
this is a remarkable statement of course. I’m cheating a little bit
because I’ve said over and over, I’ve said twice, that this is kilowatt,
this is energy use per person, right? But of course, California, like a lot of
places, has more people than we used to. So per capita is great, but
when you grow your capitas, it doesn’t mean our total
energy budget is flat. But it means that we’ve done remarkably
well per person by investing in efficiency and some of the things
will come out of this process or not just kind of better windows,
better HVAC systems, better lighting. A whole variety things there
are things that we passed that were derided at the time as the impossibly
aggressive and now for anyone under 30 or 35 will seem mundane. So in 2006, one of the things that we
got through The energy commission and then the public utilities commission in
California was a requirement that we can build no more homes that
aren’t net energy generators. And that was seen as a huge
bet on efficiency and largely rooftop solar, although if your
in a rural area could be rooftop wind. Nearby the barn window, or whatever else, but this took us,
well, didn’t work quite right. This worked, California passed
this into law late 2018, and so now, every home in California builds to
a lot of homes, almost 200,000 a year. All new homes have to be net energy zero
by 2020 I’m going to get this was seen as impossible the time when it was written
banalg and many of you have rooftop solar. Yes?>>On the scale of killowatt hours,
is that electric only?>>What’s that?>>The California case?>>Do you mean this or the-
>>The previous slide.>>So actually-
>>Is this electricity only?>>It’s complicated. When it was first done, it was
electricity, but now it’s total energy and so it’s a bigger bill.>>Public transportation, is that-
>>Closer observation, but it does not include industrial use of energy that
doesn’t have an end use in state. So if you use industrial
energy to do export products, they’re outside the scope
of this right now. So that gets to a much more complicated
story about international trade which I will do nothing on during this talk. So this is now law in the state and since we build in the order of 2
million homes nationwide a year. If programs like this were to
scale that would be pretty gutsy. At the same time California
passed this one, we passed another one we
kind of snuck in there. And we knew how to do the first one but
the second one, all office buildings, irrespective of height by 2030 have to
be carbon, have to be energy generation. We do not know how to do that today
at all, we’re hoping the next ten and a half years, ten and
three-quarter years, we’ll make some serious innovations we need those
new batteries like today, to get there. But I would say that half of my
academic friends in California and people at some of
the companies would not of pushed as hard as they had done
without having this in place. And so this kind of interesting back and forth between what science can do or
what we don’t think we can do yet, and what we’re requiring is
an interesting part of the story. Of course California’s Clean
Electricity Mandate went from 20% in 2010 that we missed it by a few years,
we hit our 2010 target in 2013. Our 2020 target is one-third
power by renewables, and we are already about 36%,
the estimate by the end of 2020. Cuz we always go at the end of that
year will be something like 37% or 38%, and the 25th goal was 50%, but
then we passed Senate Bill 100. They require 100% clean energy by 2045 and up to 2030 target to 60%. So it’s an interesting set of kind of
doubling down on what we think we can do, but also passing some regulations for things that we don’t know even
today how we’re gonna get done. And there is some subtlety in here. These are renewable energy standards,
but this says clean energy, because we have not figured out, we have not made
peace yet with large hydro or nuclear. Right now, in California, neither big
hydro or nuclear counts in the target, California is down to one nuclear power
plant, Diablo Canyon that’s going to close now in under eight years, which
means it’s closing three decades early. And this nod to clean energy is
because there’s a lot of uncertainty in how we think about this. And if you saw the first slide, I am
a faculty in the department of nuclear engineering, although
the piece of it which I am likely to dedicate a lot of the next part
of my career will be nuclear, but fusion. So this is consumption based, and
it’s a bit complicated, because we also have tensions over electricity
generated in state versus out of state. But it’s with the utilities must
procure to meet demand, and then there’s a whole second issue
about the credits you get for in state generation versus the out
of state, but this on the utilities. Of course, one of our utilities is now
going bankrupt and so who knows, so that was a very quick bid in terms of
how I often think about these things in terms of California’s path to
try to meet these kinds of targets. And the piece of it that I glossed over,
but should not be glossed over is that now
that 1.5 degree C is on the table. You saw how close to zero net
emissions the two degree path is. That means that most of the scenarios that
we look at in terms of IPCC work, and many states and countries individually,
gets to negative emissions in small, or very insignificant amounts. By sometime, mid-century, and that opens
up this incredibly complicated, instead of broader and more interesting in some
ways, process to think about electricity. Also about heat, but also about how do we manage our
food systems, our land systems so that we can achieve this without
essentially destroying the remaining parts of the planet that we’d actually
like to preserve in the process? So it’s an interesting kind of
much more complex picture and it’s why some of the projects
on everything from biochar to some of the most aggressive smart
agriculture projects, to some of the forest conservation efforts now
have a much different seat at the table. Then they had even five years ago in terms
of thinking about the climate story, so I will not delve into
it in too much detail. But I wanna mention one case that
we worked on that, in my view, kind of highlighted how much those
policy objectives are critical, but if you don’t build good enough
analytic material science techno capacity, you can really box yourself in. That’s all, I’ll do it very briefly, but
as the world or as I should say the world, the US was getting excited about
biofuels in the late 90s and early 2000s, we ended up building
a life-cycle calculator tool. And it’s really at this point that
I learned life-cycle analysis and my kind of evolution, so that we could
compare all of the different fuels, fossil, biofuel and
non-liquid fuels on a comparative basis. And so
we built this whole calculator tool. It got a lot of interesting attention at
the time in terms of some people liking at some people thinking we’d given short trip
to their favorite fuel along the way. But what was interesting to
me looking back on it is that we did this right at
the point that California and the European commission were trying
to think through, how do they both quantify and regulate all of the different
fuels for transportation they will use? And so we built this model,
made it freely available, and interesting little detailed senses that
we put in to be technically correct, but didn’t think they were gonna be the core
of the story, ended up being critical. So life-cycle calculators, we looked at the energy to make fuels,
to grow biofuel crops. The energy to make the equipment etc and a small part kind of a foot note in
the article became the whole story. California adopted our model
within the first year we ran this model called EBEM for all fuels and
states to quantify them, and what came out of it was something
called the low carbon-fuel standard. California basically said the same
way that we rate electricity, we’re going to rate fuels for
transportation. Gasoline was our reference, and
it’s about 94 grams of CO2 per megajoule. And then if we look at midwestern
grown corn, so corn that has kind of the typical mix of fossil fuel-based
fertilizers and irrigation. If you just look at the blue, you get the
direct inputs to the life cycle analysis. Brazilian sugarcane quite a bit better. Why California grows ethanol,
it does not make sense, but they do. These were the fuels that were of
interest to regulators at the time. And at the time,
we only built the blue bars. That was the direct lifecycle analysis
like you do in kind of civil engineering types of processes. But what came out of this one
little sentence in the paper was the recognition that if
we stop growing food, and start growing fuel, or
if we expand land use for biofuels, then there is a signal
that’s through the market. It’s not through the direct
lifecycle of making the tractors and growing the crops and
putting fertilizer on. And this is the controversial,
still to this day, part of the story. The green in each of these graphs
is the indirect land use signal. And that means that if everyone
in the front row is a farmer, one of you grows only on your organic
farm, wearing Birkenstocks, and you never have any sort
of other materials. Everything is wonderful, locally. And you over here are a huge industrialist
owning two-thirds of Midwestern states I won’t mention. And you buy lots of petrochemicals. Each of your decisions may
look different politically. But what happens in the global food and fiber market is something that you
don’t have a direct control over. Whether you’re thinking
you’re super organic or you’re doing, kind of, some traditional
fossil fuel intensive agriculture. And so this is what we calculated with
our partners at Purdue using the global, it’s called the GTAP model. It’s a global trade model
to look at the impacts. California chose to adopt those numbers as the additional impact of these choices,
and some of them are quite interesting. There are parts of the US where corns
is grown with almost no external fertilizer or irrigation, so very small. But there’s also places where because
we’re converting corn production into biofuel that used to be for food,
that we’re now seeing huge expansion. Whether it’s soy or
corn in the Amazon and elsewhere. And there are places where tropical
forests are being cut down. And you would need to grow that crop
successfully for hundreds of years, on the order of 3 to 400 years,
to pay back for all the carbon you lost by cutting down, and deforesting,
and changing the ecosystem. That’s why this unobservable green
continues to this day to be very, very controversial. California, finally the European
Commission, and the EPA under a prior, happier administration, also adopted this. And so this low-carbon fuel standard
has been the way that a lot of places look at this process overall. And because biofuels for many of these
problematic reasons are no longer as popular for transportation as
electricity from cleaner sources. This lower carbon fuel standard
also provided the way to quantify the metrics for electricity if it’s
being produced by wood waste, or we’re doing hydrogen, a variety of things. And so
that’s the way that we assess our fuels. And right now, electric vehicles are
getting all the attention, but you know, in ten years,
maybe it’s hydrogen, who knows? And we battle onward to this day. So just recently the European Commission
was presented a proposal to count any biofuel imported to Europe
in any form as carbon neutral. Biofuels grown within Europe,
they have to apply this metric. But if you ship in wood
from South America, or the US Southeast, which is a big provider,
it doesn’t have to go on. So we’ve been battling away, and
there’s been signing ceremonies and 60 Minutes episodes to kind
of battle through the story. The Guardian is tracking the degree to
which the European system should really account for these kind of impact. So spend the time on this long example. Both because it’s an interesting in
itself in terms of thinking about how you do the quantification for
different biofuels in the economy. But also because it led us to think,
in my lab, much more broadly about energy systems. And so go on from that effort around
lifecycle analyses of fuels for transportation to look at power
systems much more broadly. And so what I wanna spend a little
bit of time on next is a model we started to build ten years ago. It’s called SWITCH,
which roughly stands for solar and wind integrated with transmission and
conventional sources. And SWITCH is a big linear program. It’s a model of the power system. We started by building it for California. Then we built it for
all of western North America. And now all of the places in green
are official partners of the lab where we have university
partners in country. And we have at least one government
office that talks to us. They don’t always listen, but
at least talks to us about the efforts. So we, after we built the western
North America, we built it for Chile. We built in it for Nicaragua. Mexico will come online later this year. East Africa is building a regional
power market modeled quite a bit after their experiences studying what’s
gone on in west North America and the PJM network here. And so we’ve currently built and
released models for Uganda and for Kenya, and we’re working on the others. We finished Bangladesh now
a little bit over a year ago. And the biggest team, and
I’ll come back to in a bit, is looking at the Chinese power market. So this is a very classic
linear programming project. What we do, I’ll only spent a second on
the on the math part, is that what we just do is we minimize the net present value
of all investments in the power sector. So all existing power generation plants,
and all transmission projects, and then we look at all of
the fuel costs for current plants. And what new transmission or upgrades we would need to build to bring
on new plants, whether they’re large, nuclear, or solar plants in the desert,
whether they’re new natural gas plants. And we basically analyze all of that. We look at the capacity factor for
each plant. We keep a reserve margin. Different parts of the world
have reserved margins, meaning, extra generation capacity. Some places, it’s negative,
like Nigeria, not so good plan. Other places, it’s 5 or 10%. In western North America,
we have a reserve margin of around 15%, so generation capacity
could be brought online. And we do this in a large programming
tool where we look through what would be the scenarios? And so
just kind of to jump to an output picture. If all of west North America
were to meet that 2050 target of being under two degrees,
this is what our base case looks like. And just to kind of decode it,
the black line is a forecast of demand. It’s one scenario of growth and demand. The colors hopefully are representative. Light blue is wind,
dark blue is hydro Yellow here is solar, this is gas with capture,
whether it’s bio gas or natural gas. This here is geothermal, this incredibly
thin line of purple is nuclear, and then these negative going spikes
are energy generated in the mix, that’s put into storage, and
then, is brought back out. And that storage can be both
central station fixed plants, large battery systems, pumped hydro,
compressed air underground, or electric vehicles, if they are smart
and vehicle-to-grid connected. So when you’re not using it,
the battery can be used in the system. And so, just wanna highlight this
picture because anyone whose a power systems expert in the room, whose my
age or older, this is a new world. Because in the old days, we thought of the world as largely
powered by base load generation. You have a very steady,
maybe some scheduled down times. And almost none of this,
all this kind of crazy noise, is the new world of energy where we have
to deal with very intermittent sources, the solar and
the wind that go up and down. And hence, the need for storage to be
integrated centrally to the storing. And so we build scenarios like this for all of those countries
that you saw listed there. And just to kind of walk
through an example. This is the scenarios that
the governors of the states in West North America, in the US side and
Alberta, in British Colombia. Alberta is an observer,
I don’t even know what that means. British Colombia is an actual partner,
and then Baja California, Mexico were all partners. So all of these scenarios here are ones
that were either dreamed up by us or requested additionally by the partners. And hopefully, that kinda makes sense. So the reference scenario is
the one I just showed you. Then we have scenarios, Sunshot is where
solar becomes actually as cheap as it’s become, and if batteries also become
cheap, there’s another scenario there. If solar is cheap and batteries are cheap, this is my favorite personal scenario,
that’s this one here. If nuclear prices are what they were
at the end of the US construction. We haven’t built a plant since ’96,
and we use TCS as a case here. If we decide to throw
away energy efficiency, I don’t why, I won’t say which state
requested a low energy efficiency state. It was a state that
hadn’t done much on it. But all of these are just
kind of different versions. They all get you to that 2050 target. The problem is, of course, that some
of them require you to build lots of transmission, others require you to build
lots of distributed generation sources. And so, while the scenarios are really
illustrative of where you might go to get there, they don’t actually tell you
in the end as much as you would like. They’re helpful, but if you want
to go strongly into large-scale distributed solar or wind, you need to
build transmission to get it online. And since building transmission
is as painful as building nuclear plants in the United States, for example,
you have to really do things in advance. So an interesting kind of, you probably
didn’t need the model to know that, but if you optimized where you would,
for example, do a nuclear heavy future, an interesting feature is that
you’d have very little other costs. Because basically, anywhere in the US
where you have a current coal plant, you probably have
a reasonable supply of water. And so dropping a nuclear plant in
there means you can more or less, think about a plug and
play kind of swapping these in and out. However, nuclear’s not doing so well,
for a variety of interesting reasons. CCS, in my view,
does not exist as a technology. It exists as a practice in the lab,
but there is no large-scale CCS. Current US CCS is essentially the Decatur,
Illinois ethanol plant, one megaton a year, and
we need to be on the gigaton scale. So it exists in lab. Some people may disagree here,
but I would say, this scenario,
you should put quotes around it, at best. And so thinking through the investments
you need, both in terms of research and in terms of planning to make these work,
is a really big story. And it’s partially why this is my personal
favorite, but that doesn’t mean much. And it also means a really
interesting mixture of where you build generation and
the transmissions lines you need. And so again,
the colors hopefully kind of make sense, some of them are hard to
read down here a little bit. But so, for example, if we go from 2020
all the way to this very decarbonized 2050, we’re in a world where we
have lots of solar in the west and the south, no surprise. We have hydro in the places
where we have a good supply. We have wind in the middle of the country,
and then a mixture of geothermal, and again, heavy amount of investment and
transmission that bring this all together. So we kinda build these kinds of pictures
for all of those partner countries. And that gives you a picture of how we
kind of think through the technologies that we have today and
reasonable amounts of evolution. So everyone in the room who’s
studied solar in detail has seen this incredibly
complicated graph. This NREL, the National Renewable
Energy Lab, produces it every year. And this is showing you
the efficiency of solar cells. This is, I think, the most complicated
graph I think I’ve ever seen, in some sense, but the colors code
different types of technologies. Single crystalline cells,
multi crystalline cells, amorphous solar,
interesting emerging technologies, proscovites, quantum dots,
organic solar cells. All showing the unit, university, or
laboratory, or business laboratory that had the record for the maximum
efficiency at any given point in time. And if you put it up this way,
kind of gives you an interesting picture of what’s going on in the field,
right, at one crude level. This is efficiency versus time, so the slope of this is somewhat interesting,
but it doesn’t tell you the whole story. But it at least shows you that some
of these older ones have progressed pretty steadily. And while these new,
exciting technologies are not as efficient as the older, old standbys,
at least the slopes are higher here. Of course, you might also ask things like well,
does actually deficiency matter, right? Aren’t we really more interested about
cost and survivability overtime? So there’s lots of interesting metrics,
but analyzing these kinds of trends industry by industry, has gotten
us all kinds of really interesting insights that we then put back into models
like that switch model I showed you. So just to kind of highlight, before we go on to that kind of
thinking about technology evolution, what’s going on now in solar is as
revolutionary as it was years back. So you know,
the neatest new things that I look at are, I work with some colleagues on
some transparent solar material. So you can think about office buildings
where you have actual windows you can look through, which is good. And they’re also generating power. We just hired two new faculty
at Berkeley who are doing fabric integrated solar to power your devices. Someone’s gonna get a shock at some point,
and that’ll be that. But powering up all your personal devices,
not with old clunky things like solar panels strapped
into the back of the jackets, as was very popular in certain movements. But now, integrated materials,
floatovoltaics, kind of a new,
trendy thing on the horizon. All of these kind of illustrate
some of the neat things going on in one of these areas. And so, a whole range of people
in operation research and economics have been looking over time
at What has this level of innovation given us in terms of improving these
technologies in terms of lower costs, but also in terms of performance? Because this goal of 2 degrees or 1.5 degrees is gonna require some
really dramatic changes to get there. And so the basic story over and
over again, even if it really doesn’t appeal to many of us, has been
the classic picture of the learning curve. Right, the idea is that if you look
at the data I just showed you, look at the level of innovation, it
would be nice if the cost in the future, let’s say C2 versus C1,
followed some simple form. And it turns out that that simple
form appears again and again. And that is, if you look at cost,
it’s in log scale. So this is cost not versus time, but
by total deployed volume of technology. Not just built and left in a warehouse,
but installed in the field. We see this wonderful power
law where the slope of this line is essentially showing you
that rate of innovation, right? So this is showing you cost in the future as a function of the volume of
production of that product. And I have never liked this. I have always felt this doesn’t
capture any things that I want but technology after technology we see
this appearing again and again. These learning curves for solar,
it’s about a 20% consistent over decades, reduction in cost for
every doubling of the amount of solar. It’s about about 13 or 14% for wind,
we have a whole variety of these. So I wanna talk a little bit about
that in the last part of the talk to kind of highlight what is going on. And so when we look at this innovation for
decarbonization, how can we both learn what it takes to
make these technologies not just cheaper, but also better, more reliable,
longer lifetime, etc. The pushes that we’ve seen in the solar
and wind areas in particular are dramatic. And so, recently we had solar
installed in Saudi Arabia and then in Dubai at under three
cents a kilowatt hour. I got calls saying, it’s not possible,
they must have cheated in some way. Just recently, we had solar come in
at under two cents a kilowatt hour at large scale, and so
these learning curves are interesting. We’re seeing this progression. And if I were to put up an older curve, of learning curves, here’s one, just don’t
look over on the right for a second. So Knox controls,
something you scale up, wind, solar. The slopes aren’t exactly the same but there’s a lot of interesting
information in here. We need to think hard about
if these are learning curves, I don’t even want to say as
a professor nuclear engineering, what I would have to call that
a forgetting curve, I guess or something. But, this gives you a really
interesting picture of what’s going on. And if we look at the area that I’ve been
studying most intensively recently in terms of this kind of,
the techno-economic perspective, this is the current was as 2017 for
batteries. And so lithium ion batteries around here,
nickle metal hydride batteries, lead acid batteries are on here,
flow batteries are on here. Interesting enough they kinda
show the same shape curve. And of course the two exceptions to
the rule are pretty interesting right? So here we have pumped hydro. Shouldn’t be too surprising that pumped
hydro hasn’t seen the same kind of cost decline from manufacturing
because pumped hydro is pretty simple. If you have the right geography,
you pump water uphill. Pumps have not changed much. It’s a function of potential
energy equals MGH, so the learning hasn’t seen a whole
lot in the last several decades. And then interestingly as well, lead
acid batteries haven’t also seen a lot, because lead acid batteries are not super
useful for these new types of devices. The big push for
better lead acid batteries for vehicles. Designed to give you a starting
jolt not to do long term powering. So, these two exceptions
are quite interesting but all of these with different slopes show
kind of similar types of behavior. So, I’ll tell you in a second
why I really don’t like the idea that this simple
heuristic worked so well. And I don’t like it because I
would like to have thought, consider first principles, we needed
to know more about the investment, or research, and deployment, and various things to get an understanding
of how these technologies are evolving. But those curves are fairly compelling. So we’ve been doing the last two years
was to take this as a starting point and start building a little
more complicated models. And so
the one that shows the best fit to data so far is one that appeals to me greatly. And that is our now two
factor learning curves, have the traditional term due
to the volume of production. But they also have a measure of R&D. And in fact this is the best fit,
I won’t go through the details here, but the best fit for the solar and
the wind and the energy storage data is not that old one factor curve,
but it’s two factor curves. So I regard that, I mean I was
pretty excited when we did this. In fact, it got my student
a pretty neat job as a result. But I actually want to highlight to
you why this is kind of problematic. It shouldn’t surprise to you if we just
drop this term, why this model, A, it fits the data pretty well. B, it’s also wonderfully easy. You can measure quite easily
the total volume of manufacturing and production of materials. So if we just drop this term,
right, set alpha to 0, it’s compelling in terms of the real
world to think things through. We find a better fit if we also
include a term measuring R&D. Now, the way we measured it is in this
case to look not at dollars spent but we use patents generated as our measure. And lots of interesting things
go on when you do that. One is of course that patents are not
always designed to move the field ahead. A number of patents
are defensive to black others. And so patents,
no question are a poor overall predictor. And in fact not to get into the stats but we find that the fit is better not just
because we have the second variable. So we actually find the feed
is significantly better by doing this then you would expect alone. However, what measure
of R&D should we use? We use patents because that’s
a data set that is available. But I’ve got students chasing
a whole variety of other measures. And one of them just proposed to us six
months ago at a big summit we had around this with Google was
actually that they argued that we shouldn’t be using patents or
dollars spent on R&D. We should be using the metric of students
that did not complete their PhD. Which really pained me, right? But the argument made was
students that get their PhD, I guess advisors in the room and
students shouldn’t listen. If you get your PhD,
you’re gonna often do what I do. You’re gonna be an academic and
be semi-useless.>>[LAUGH]
>>But if you drop out of the PhD, but you stay in the field, you’re gonna put
your PhD to use in an accelerator and starting up companies. And you’re actually kind of
a valuable member of society. As opposed to what some of us do. So I’ve resisted that measure. But I highlight it because one of
the things that I think we critically need to do is to think through how do we,
Understand and value R&D, not just because we like
bigger research contracts and grants, and we do believe that it spins off
into companies and useful things. But with this era of kind of fake news,
and this complete Disrespect for
science and engineering. It is clear that if we can do a better job
of understanding this more intricate form, we can actually do quite a good job of
highlighting what is the value of what many of us in the room, maybe all of us,
think is a valuable part of the process. And so we’ve been, again,
working with patents right now, but I would argue there’s
lots of other metrics. I’m loathe to start gathering data on
the dropouts from the PhD programs, but I suspect Google’s gonna give us
that data whether we want it or not. One of the examples of combining the story
together is that we use the switch model to examine those scenarios that
I mentioned for West North America. And in terms of,
These negative going spikes, we convinced state of California largely,
cuz they paid for the model and they are the biggest consumer
of energy in West North America. That we were gonna have to dramatically
accelerate both R&D on batteries and storage, but also deployment. So some of you, anyone in the room
funded directly by ARPA-E? Really, that’s interesting. Does anyone in the room,
who knows what ARPA-E is? Okay, so ARPA-E, the Advanced
Research Project Agency for Energy, I thought half of you would
have been ARPA-E funded. Used to be, okay. It means you’ve moved on. All right, this is an interesting
conversation, by the way, right? So one of my students ran the energy
storage program, someone named Ilan Gur. Who after his term running it came back to
not Berkeley but to Lawrence Berkeley lab, where he now runs a clean tech innovator
startup accelerator called Cyclotron Road. So this scenario requires
a lot of storage. And so we worked closely with
the prior Governor in California, Governor Brown’s team, to argue that
the model said clearly for all of those scenarios, essentially, we needed
significant amount of storage online. And the R&D pipeline is doing pretty well,
thank you guys for working on it, but we needed more storage in place. And so we passed a law for
years ago that said by 2020 California must have storage
available to meet 2% of peak demand. And I won’t get into the fact that some
people apparently don’t understand kilowatts and
kilowatts hours are a different thing. This is complicated, but it’s why we
started off with a mandate not based on duration of storage but
on peak capacity, 2% by 2020. And right now we’re working on
a storage mandate for 2024 or 25 that will be somewhere double that,
4 or 5 or 6%. And this is a physical version that
came out of our first mandate. So Southern California Edison, it’s not in
bankruptcy, contracted with EnerVault to build a flow battery system powered by PV,
because you can’t be powering with fossil, if you’re gonna get
the benefits out of the story. So here we have a facility. It gives you a quarter of
a megawatt of capacity, four-hour duration, and
they’re using it in the mix in the grid. And so we are hoping that as we
get the 2024 mandate rolled out, that has to be rolled out by end of
2019 to give industry enough time, that we will have built up more
storage capacity in California. Germany has a similar feature, and China is working on one that I’ll
talk about at the very last minute. But one of the interesting features that’s
come out of all of this technology mix, not just for California, but now we’re starting to see it
in parts of eastern China. We’re seeing it very
strongly in Australia, that has the most rooftop solar per
capacity, any Australians in the room? You should be proud. I mean, what’s going on Australia, for
all of the challenges over water and cooking fish in the oceans,
solar, doing great. So a number of places are coming up with
a challenge which is looking more and more like this. So this says, for
anyone who works on western grid issues, this is the famous duck curve. And so this is time of day, and
this is the overall net generation load. And so of course, most of us, except for students, are supposed to be asleep during
this period of time, so demand is lower. And during the day solar comes online, so if you have a lot of solar,
you start to hollow out the demand curve. But then we heard the sun still sets,
that kinda stuff. And so at the end of the day you get this
challenge where the solar generation’s going down and demand is highest,
or used to be highest. Because industry is still on,
and people are coming home, so we have the maximum amount
of stuff turned on. So you get this evening peak, and
then you get this horrific ramp rate. And that’s why we built the switch model,
to understand how much generation, how much storage you need. So that this doesn’t blow out
all your transformers, and your system doesn’t work. So interesting mix, right? So one of the features is
that we wanna double, triple, quadruple down on solar and
wind to really maximize our generation. And so we have forecasts that
our generation in the peak generation times in
the afternoons could be so large that our overall demand curve
has been completely eroded away. There is very poor business model for
industry, because of all of our rooftops
are generating power. And then we critically need all this
power to come back online, right? So one of the things we were doing in
California, which has a mandate to have 1 million electric vehicles on the road by
2020, we probably won’t get quite there. Is to start thinking about what mixture
of that central storage I just showed you plus our vehicles, or storage in the
basement of your home or your school or something, could be used to smooth out and basically bring the power
to this kind of peak time. And it’s this kind of use of the models
that’s gotten so interesting. And it’s gotten particularly interesting,
of course, for the EV reasons I mentioned before. And so I think, yeah,
I used 350 a gallon on here. So at 350 a gallon your vehicle
is running at $0.14 a mile, if it’s a more efficient vehicle, be it
hybrid or just more efficient, half that. If you’re using solar from the grid, and I used California prices here,
you’re driving at $0.3 a mile. And at that $0.1 that I showed you before,
you’re driving at $0.1 per mile. So at the point that
the cost of the vehicles, their performance gets comparable, you
don’t find many situations where you have more than an order of magnitude
difference in the operating costs. And since now we have EVs that
are remarkably inexpensive, BYD, a Chinese company, just announced a $9,000
electric vehicle that has 70 mile range. The average American
drives 40 miles a day. So going from the fanciest Tesla, or Prius
or wherever else, it’s a remarkable story. And it brings me back
to what I began with. And that is, of course,
if we look at just the United States, this is the mile per gallon of the same
electric vehicle across the country based on the mixture of clean or
dirty fuel, right? So places that have lots of hydropower, no matter what we think about
what it’s doing for salmon. Hydropower, whether it’s in New York State
or taken from the Cree Nation in Quebec. So you get this interesting mix,
where that same EV is worse than a hybrid vehicle in some places, and
incredibly better in others. And of course, every time you clean
the mix, this number goes up. And so Texas,
which has just announced new mandates, you get these kind of remarkable changes. And now that Tesla and others are thinking they’ll have
these amazing vehicles on the road, 500 mile range, charge it up in 30
minutes, if your batteries won’t blow up. We’ve got this kind of
interesting picture. And so China has just
announced that they are now in the planning process using our switch
model and then the model run by NDRC. NDRC it’s hard to describe,
NDRC the national planning agency and it rolls many different
features into one mega company. So China is now in a process in
this conversation with us to plan, first, the phase out of manufacturing
of internal combustion vehicles. Right now they’re the world’s
largest producer. So the phase out of manufacturing is
following what’s happened with Volvo, who said that after 2021 or 22,
no more combustion vehicles. Ireland and
France have also said they have plans. But China’s now said, no more internal combustion vehicle
manufacturing after sometime in the 2030s, and no more combustion vehicles
on the road after 2040. So our switch modeling team is
really busy doing those scenarios. And we’re working on all kinds of
things like mobile charging pads, you don’t move the cars in the garage,
the charging pads move between vehicles. And of course my current favorite and that
is inductive charging while you drive. And so in Sweden and
in the Netherlands and in 2021 in North Beijing near
North Transfer Power University, there will be a high density roads,
highly used roads where there’s a lane where you can
recharge inductively while going. And so
this is this interesting collaboration. There’s NCPU, the university that’s installing
that first section of road there. And just to give you a feeling for
how envious I was of the data. This is the electric vehicle
control room in Beijing. This is the 2.3 million EVs on the road,
the 279,000 charging stations. And just cuz I walked in they
flipped the control quickly, and they put up the USEU in Japan just to
kind of embarrass us in terms of numbers. But it’s a really
interesting overall process. So I’ve highlighted in kind of this
mix the range of how these kind of technology models I think have played a
critical role in thinking through how you set standards and how you try to plan
your investment in various technologies. And I hope that I know,
that we’re right at time. And I hope that most key lesson was not
that these models tell you what to do, but if you can work on both
the information generation side, and on the implementation,
you can really accelerate the process, as I think we’ve seen in
some of these places. So thank you all very much.>>[APPLAUSE]
>>So we can take a couple of quick questions.>>So in your, cuz I was curious
about your learning curve model. So what does alpha look like
relative to the [UNKNOWN]?>>Yep, [COUGH] So again, if you drop this term entirely, the
classic learning curve has a power law. So that slope is like negative 0.2. And we see that,
roughly that 20% learning here. What we find, and again,
the metric you use is critical because measuring innovation is something
that none of us can do, right? So if you use patents and
in the data we have for batteries where we don’t have as much data yet of people
who clearly patented to block others. Maybe that’s going on but we can’t find
a lot of evidence of patterns that were, Worked on, the patent was generated,
and then the company or its spin offs did nothing on it. Which the classic economic definition
of a patent that was to block others. We don’t find a lot of data on that yet. If we do, we’ll change it. But what we find is that this explains
about 60 to 70% of the behavior. And this is that 30 to 40. And again, if I had walked in here and
said, R&D is not important and I think that the color of your
rooftop is some other metric right, who knows what we would have found? So I think I can’t guarantee we
picked the best second metric, but the fact that it explains
that much the behavior and it does better than just simply another
dummy variable to explain the rest. We’re doing pretty well.>>So a is about 0.1-ish?>>0.1, 0.15 is what we find right now. But, no one has good data on,
almost no one tracks investment in R&D in easy to digest ways.>>[INAUDIBLE]
>>Exactly, right.>>[INAUDIBLE]
>>So, our first paper on this was in
1999 when we first tried this out. And we built the model where we’d looked
at just patents, and we assumed that patents that dollars would flow and
then pounds would go so there be a lag. What we actually found was that
patents preceded dollars spent. And the only explanation that sort
of held up in the conversations is that in areas that were new,
university researchers and those at big laboratories,
had work they’re working on it. Cuz we all know we do shuffle money
between grants a little bit in terms of supporting people, and that if you
know there’s going to be a new call, people get their new products out so
they can. And so
we haven’t followed up on that sense. But in the push of R&D data that came
from, largely from the Carter efforts, there was clearly the patents came
before the spending, so tricky.>>So one other quick question, yes.>>[INAUDIBLE]>>Great, great question. So when it was initially written, going
all the way back to it somewhere here. When it was initially designed, we were thinking entirely your
building is the footprint. You put solar on the roof, put a battery in the basement
you’re doing energy efficiency. Because this only passed into
law less than a year ago, industry, of course, blew up. They said, wait a second, we didn’t know
we had to start building all of these net zero energy homes,
even though net zero is a terrible term. But we’ll leave it for now. And so there’s been all
these interesting pushbacks. Some of the lawsuits, so not so fun. Bbut the interesting ones have
been companies that said, well look, we think we can do it but
exactly what you said. There’s no reason why the generation
needs to be tied to the site. And so one of the largest supermarket
chains in California, called Safeway, has already committed to own 100% renewable
energy, which they buy from off site. So they have now said that for
communities, they will offer for fee to host solar on their rooftop. So that your condominium or home or
business doesn’t have it onsite but you have a share in that and
it’s part of the purchase price, so when the home changes hands
it has to go with it. We and a group at UCLA are working with them to
now say you could do storage off site. And so that big storage battery, you could
have an ownership share in that as well. And so
this is not happening physically yet because it doesn’t require
happen until the end of 2020. But it is where the rules are going
to accommodate this difference. So it’s a big topic in California now.>>Okay, so let’s thank Daniel [UNKNOWN].>>[APPLAUSE]
>>Thank you.