it's a great time to be in California if you believe in clean energy the last
five years I've been watching all my neighbors slowly one by one started putting solar panels on their roof and
now they all started buying electric cars so it's fantastic yeah the problem
is we all go home Suns going down we plug our cars in and then what happens
so steel has been thinking about problems like that for for the last more than the last ten years
originally it'll be Earl and then Google and then slack and all of that and still
100% clean well I started by saying
welcome did you hear that okay good so I want to talk a little bit about the
motivation behind transitioning to clean energy today I also want to talk a little bit about how the electricity
grid works today I want to talk a little bit about how we're transitioning the
electricity grid and particularly I want to focus on some of the issues with the transmission system the distribution
system and the consumer side technologies and I want to talk a little bit about what our team gismo grid
integration systems and mobility is doing in this in this area so whenever I
say gizmo you should clap or something my entire team is in the back with their t-shirts thank you all for coming
all right so during this sort of our I also want to take you from my home country Turkey to our lab in building 27
as well so let me start with a little bit about me I grew up in Istanbul in
Turkey I was a terrible student in elementary school at first grade especially so I had to change schools
because I didn't learn how to read and write so my parents took me to this other school and I actually had a really
fun time the first couple of days I would escape from the classroom and start playing on water in the in the
bathroom and one day the teacher actually followed me and came to the bathroom and said listen
I understand you don't want to be in the classroom but Earth's resources are
limited and I don't like you wasting water and instead please come to the
classroom we can talk about other activities that you can do so of all the
things in elementary school that really stuck wouldn't stuck with me all these years and it really shaped me in terms
of the way I look at energy not just water energy and really drove me in the
path of efficiency and conservation so that brings me to my next slide which is
this picture of the earth that I really really like and when people look at it the first time they always look at these
bright spots but today I want you to concentrate on the dark spots because there are over 1 billion people who do
not have access to electricity and those are the kind of people that we need to bring up we need to make sure that they
have access to electricity but we need to do it in a very thoughtful very sustainable way not the same way that we
have been using our resources over the last hundred years
so another picture of the earth because I really like it and this time I want to
talk a little bit about the climate change and of course climate changes the observed changes in the century century
scale rise in the earth's temperature and it's related effects in general and
this number and I do have a clicker so I'll use that this number plus 2 degrees
Celsius is a very important number it's the limit in the change of the average temperature increase from the compared
to the pre-industrial levels so a lot of people talk about this this number and it is an important number but for me
it's a very difficult number to sort of wrap my arms around and start thinking doing something about it
so instead I focus on these two numbers 300 parts per million this is the
concentration of a limit concentration of co2 in the atmosphere and then 500
parts-per-million is another limit to 350 being the lower limit 350 being the
upper limit and today we're somewhere around 410 okay
maybe not today we're somewhere around 4:00 time this number is important because if you think about 350 parts per
million as a healthy person let's not talk about what 550 gets you but 410 is
someone who went to the doctor and the doctor said you're overweight you have high blood pressure you have high
cholesterol you need to quit smoking and you need to lose weight so that's where we are as a planet today we need to do
something about it and we need to do something about it soon so here are sort of the co2 contributions in the
greenhouse mix globally carbon dioxide's contributes 60 first 65% to greenhouse
gas emissions and in the u.s. it's the number is more like 81% so these numbers
are huge tackling this this gas is an important first step for us the second
thing is that we at gizmo are really focusing on buildings electricity grid
and transportation so when you look at those things like the electricity
so when you look at the combination of electricity transportation and buildings
globally we have about 45 percent of the global greenhouse gas emissions in the
u.s. this number is much larger close to 67% so in our research we're trying to
tackle a large portion of the problem that we hope so let me talk a little bit
about electricity grid our electricity grid is designed to run one way the
power flows one way generators generate they use different resources they burn
them and and generate electricity this electricity is transferred over
high-voltage long lines over to distribution centers distribution
substations eventually step this electricity down to where the buildings
can consume so again all right
I think I turned it on this time so transmission lines high voltage long
distances and then eventually we get the voltages in our homes 120 240 volts now
for a distribution system operator the voltages at the grid edge is important because that's how they actually manage
the electricity current they care about these voltages because our appliances are you know things that run in our
homes really are sensitive to these voltages about a 5% plus and minus
change is is okay but anything beyond that could be detrimental to some of the
appliances that we have and overall the system runs on 60 Hertz or actually in
the u.s. it runs on 60 Hertz in in the in Europe it's 50 Hertz in Japan it's 60
and 50 Hertz depending on where you are this frequency of the system is an important metric because when the
transmission system operators are looking at the grid and running the grid they really look at this number and
gauge what the supply and demand balance is like so if the if it drops there's
definitely a problem with the system whether there is a generation that's down or loads that are up or a
combination of things so this number is extremely important for grid operators so let me just tell you the four basic
rules in in the electricity grid the first one well I'll start from the
bottom and work my way up how about that that will keep you awake the system must
maintain voltage within acceptable limits and that's the 5% plus and minus at the grid age that's what I was
talking about the system must be maintained steady state a steady frequency that's the 60 Hertz I was
talking about system must be able to address variability and uncertainty now
our system this system is designed such that generators generate so that if I flip a switch that I can have
electricity where I am so that's sort of the variability and uncertainty that we
need to think about when we're when we're thinking about running the grid I'll
talk a little bit I'll come back to this issue because renewables add much more variability and uncertainty to the
electricity current causing huge problems but we're talking about the grid as it was and again variability and
uncertainty are big portions of the risk management then the generation transmission must be able to meet peak
demand that's a big big problem because let me just show you illustrate this a
little bit this is a graph called load
duration graph and this is the real data from the California system between September 5 2005 and September 2006 so
it's a little blurry but let me try to explain it on this axis we have the
hours of the year 1 to 8760 and then on
this side we have the magnitude of the peak so fear for each hour we plot the
magnitude of the peak and then we order them so here that year we had a new peak
in California over 50 gigawatts of peak but but we had all of these resources
about 10% of the resources ready to go just so we can mean to meet the needs of
this peak and that peak only occurred 60 hours less than 60 hours that year so
building for a big capacity is important because that's how we maintain reliability of the system but it's also
important that if we can cut down to speak that our system can be much more efficient and effective so that's
something to think about going forward I'll go back to the variability and peak issues in the next few slides I wanted
to give you this actually URL in the bottom because it's a great place to
look at and visualize the Electress u.s. electric secret and also I pulled this
this visualization just to show you that the u.s. grid is made up of three interconnections the Western
interconnect in green the eastern interconnection that's in pink and Texas in yellow
arkhat is the region that's in yellow and each of these regions are little bow
ties here are interconnections they're connected interconnected to each other so if there's a problem in one
interconnection we can isolate that whole interconnection so that it doesn't affect the rest of the electricity grid
in the US and this slide also this picture also shows the transmission
systems that are already there and the size of those transmission systems but
also the planned transmission of the of the electricity grid in the US so I
really recommend that you go to infrastructure USA Oregon take a look at some other visualizations are very very
interesting for the you describing the electricity grid okay so that was a
great day as we know our grid is changing right now part of it is driven by these renewable
portfolio standards that are adopted in 29 states so we may be outside the
federal government may be outside of some of the Accords and thinking but the
states continue to maintain their renewable portfolio standards and go ahead with their renewable adoption
goals the highest is Hawaii with 100% renewables by 2045 and then I don't know
maybe Lois's Pennsylvania with 85 percent by 2020 I know that some cities
in Pennsylvania have renewed goals going all the way up to 50 percent by 2030 so
even if the states may not have renewable portfolio standards or mandates that there are some cities that
are taking action into their own hands and making sure that they purchase their energy from clean some resources so at
the same time that we are adopting these renewable portfolio standards and having all these goals we also look at the cost
of of these systems so Green is the cost of wind
blue is the cost of solar this is the increase in capacity of these resources
and the price of these resources are changing because part of it is because
all these mandates and people are going in purchasing more solar power and wind power but part of it is driven by other
parts of the world where there's a lot of attention to clean energy adoption
and there's a lot more production of solar units and wind units when we look
at natural gas and in China coal here's the actual price points today actually
these are probably price points from a year ago and then if we look at us coal and nuclear here are the price points so
when you think about solar and when the adoption is it can be even cost justified going forward not just
something that we do for the good of the society and good of the earth but really economically something that makes sense
okay oh I wanted to say one more thing in this slide and that is when you think
about coal jobs there are about 60,000 70,000 coal jobs in the US but the solar
industry has created 250,000 jobs so not
only does it make sense in terms of the cost of these systems but it also makes
sense in terms of uplifting our economy as well to invest in these clean energy technologies okay so let me talk a
little bit about what's happening in California first we had a million
dollars or so million solar not million dollar but million solar rooftops by 2020 18 as one of the goals where we're
slowly getting there we have 1.3 gigawatts of storage by 2020 and this is
sort of distributed to as a mandate to all of the utilities in California PG&E
has to have about 560 or somewhere
around 600 megawatts of storage of which 85 megawatts has to be on the consumer
side the others and distribution and transmission systems saying with
Southern California Edison and San Diego's gases Sandia Gas and electrics
goals are somewhere around 160 megawatts so they're not only having these goals
but the PUC is making incentives available for folks to actually support
transitioning to storage devices now storage is still expensive we know that so these incentives really
help build a storage base that can help with the renewable adoption especially
with variability of renewables there's a twelfth gigawatts of distributed
generation goal by 2020 we're on our way to achieve that achieving that as well
we have a 5 million zero emission vehicles by 2030 we have about what
300,000 electric vehicles today in California so we have a long way to go on this one and it's zero emission
vehicles not just electric vehicles these could be hydrogen fuel cell cars and other clean vehicles and of course
over the last decade we have installed 1.8 million meters smart meters and
what's so smart about them well there are a couple of things right someone
doesn't have to go to your house to read the meter they can just roll a truck and read that meter the other thing is that
it can capture data energy consumption data granularly so it helps us to think
about dynamic pricing and using dynamic pricing in the future the other thing is
that it has an antenna that can communicate with the in-home devices so you can pull really fast granular data
from your smart meter almost up to five-second data from your smart meter if you have the right equipment and the
right equipment is costing somewhere around $20 to $100 today you can plug it
into your PC and grab real-time data from your meter we don't have a lot of tools that allow us to do something with
that data but we're working on it okay so meeting California is fifty
percent clean energy goal by 2030 requires several pathways several things that we need to explore and work on the
first one of course near and dear to my heart is efficiency in conservation there is a lot more we can do we asked
California we've done really well in terms of efficiency and conservation but we can do a lot more the other one is
reducing the carbon in electricity production so going to solar or wind and
other resources that are available and low carbon another pathway that was
identified is a fuel switching so today we have heating a lot of our heating is
with gas we have a lot of cooking with gas one of the sort of the thinking is that
if you switch your heating and cooking to electricity and then use electric
clean energy resources for electricity production then now you have a win-win situation and then the final is reducing
carbon content of the fuels that we utilize today so those are sort of the pathways do I take questions now or
later oh the transmission systems that carry
high voltage is about 6% HVDC is much more efficient but let's Park that I'll
get back to you so our group is really focusing on these two pieces how can we
switch fuels and then how can we make our electric system cleaner ok so here's
how our grid is changing we talked about this grid for the last century we're adding wind and solar at the
transmission scale these are really large-scale some distributed some concentrated wind
and solar generation resources we're adding wind solar and storage to our distribution system which is adding much
more variability on the distribution system and we're adding wind solar and
storage in our homes we all have to deal with and sort through a lot of incentives
today to figure out what is the right combination of things that we may want to put in our homes and then of course
we're buying electric cars so lots of choices for consumers today lots of
distributed new generation and new storage technologies and lots of variability of the system which is a
major problem and in some cases we're making the power flow in the reverse
direction which can cause problems for the system the system has protection
devices that chip off certain areas and by having a power flow different
direction we could actually have some consequences and on top of it we're
adding a lot of sensors to our system we are adding we've added about 75
high-resolution sensors in our transmission system in California about 300 in the whole Western interconnection
region we have added really fast distribution system sensors without it
smart meters we have AV charging stations that monitor the AV charging we
have some solar generators that monitor solar so there's a lot of more data
that's coming out from the system and that means that we can actually use new
techniques like data science and machine learning to better utilize these resources to better plan and operate the
electricity grid and I'll talk a little bit about what we do in that space okay
so imagine if we had more than 50% renewables and distribute energy resources today well we actually had a
very short period of time we had about 80 percent renewables well the the the
small line here says 67 because hydro does not count into our renewable
portfolio standards so only wind and solar counts so eighty percent with
hydro we had some moments in California where we had all of but in the long run it wasn't
sustainable we won't be able to operate the grid if we had it more than several hours it would be detrimental we would
have reliability issues stability issues we wouldn't know how to deal with these
resources because we don't have markets that can take these resources at the
level that they're in we would have a lot of problems with security we don't
have really well-defined security of the distributed resources so if we wanted to do it today for a long time we wouldn't
be able to do it unfortunately so the solutions to you know taking all this
80% of the state's power from renewable resources lies in the fact that we need
to think about technologies and tools but we need to think about policies and markets as well because technologies
that we developed today need to find value in the marketplace and need to be
supported by some of these policies and here at SLAC we're working on technologies and tools we're partnering
with university Stanford University on the policies and market side and
Outsiders in California and outside of California on those on the experts on these areas too ok so how can we
minimize the cost of integration when we have more than 50% renewables well we
have several options I want to talk briefly about these options the first one is having long distance high-voltage
DC lines where the transmission losses are small but not many people want these
transmission lines in their backyard so it's expensive the permitting is long
and so we need to plan way ahead of time for these things to be available we can
wrap natural gas and storage capacity and as I mentioned earlier storage is expensive and natural gas is not
necessarily carbon free so there are some issues with those too but but certainly there are there there are an
option and there are now for the transition for sure and then we can use flexible load to follow
generation so this is probably one of the lower-cost options and I'll talk a little bit about what that means but the
truth is we'll have to deal with a combination of the three we can't just pick one and say that's what we're going
to do that's the only thing that we're going to do oh and I had some pictures that didn't show up so who here knows
about the stunt curve I know you do okay so if you don't take anything else
from today's talk please learn about this duck curve the dock curve was
developed by the California Independent System Operator and it was developed in
2013 and here you have the megawatts oh so it is designed or developed or it's
the data from a March day in 2013 and they also plotted a March day in 2012 so
it's a spring day where there's a lot of solar but there isn't a lot of loads there isn't a lot of air conditioning
loads to take on that solar generation so in 2012 the system looked like that
on that March day it had a little peak here but peaked in the afternoon in 2013
this was the actual data it peaked a little bit more in the at the end of the day and then from that and with the
goals that we have and also the models show the penetration rates of these
solar they said that these are the typical March days that we may expect
going forward from 2013 to 2020 so that's why it's doc called the curve
this is the belly of the doc this is the head and this is the tail they make it
look like a duck so but but let me talk a little bit about what this signifies
this means that there's going to be some huge ramps in the morning and in the
afternoon and this also means that there's going to be a new peak and the middle well when there's no Sun and in
the evening from 6 to 10 period now when we think about our rates today we still have rates that
disincentivize folks were using electricity here and incentivizes people
to use electricity here so are our rates and our system is not well aligned and
so we need to think about how we can get our policymakers and utilities move a little faster in terms of addressing the
problems of the grid now this was a projection that was done in 2013 in 2016
on a February day we actually saw this huge ramp so be even before 2020 we we
actually experienced the California critics to be experienced this man and then in in 2016 in May we actually sort
of saw this really low load period in in
in May so things that they have predicted that would happen in 2020 has
already started happening in 2016 and we need to think about how our system can
respond to these changes given the impact of the renewables okay so this is
actually a screenshot from the Cal ISO portal here is the sort of the URL for
anyone who wants to take a look at it daily you can go back to two months I think this was on an April day
this shows the peak expected peak this is how much capacity we have on that day
actual peak is something like this so we built another 8 gigawatts or we have
resources that are ready spinning at 60 Hertz and some and some not so much
waiting to be called in case there's a peak like that so not so efficient
necessarily this is the system peak I was telling you about and took that happened in 2006 on this hot summer
afternoon and this is the actual consumption and then the hourly these
faint lines are the hourly forecasts that the ISO makes day ahead
so one of the things that we have been testing is this load flexibility and I
want to talk about how the loads can be flexible and what it means and when we
talk about loads being flexible we think about automating them so that there's no human in the loop and I'll talk a little
bit about that we think about shifting loads shedding loads and shaping loads so I want to show you what each other.we
what only each of these mean so let's say this is a typical load profile of a
Safeway or a target I don't know let's say Safeway the peak is around 300
kilowatts and this is the time of day for the entire 24 hour period and the
peak happens around 6 or 7 o'clock now
this nicely coincides with the Ducker so I made this up so that it would but in
terms of shifting we would need to take this peak and move it to an earlier time or even or a later time in the case of a
Safeway it would probably mean that they would either pre-call cool their facility so really sort of change their
set points down so that they can prequel the entire facility so that they could ride through that that period when it's
very crowded or it could mean that they could buy a storage device and use that storage device it could mean that there
could be other thermal storage capability in the building whether it's the product themselves they may they may
use those so basically shifting the load from a peak period to another period but
using exactly the same energy basically sheddings a little bit different because
now we're not only saving energy but we're also reducing the peak so shedding
typical strategies is if you're operating you're building at 72 degrees typically you increase that setpoint to
74 now your chillers back off your fans back off and so now you can actually
save energy for a period of time now in the case of the Safeway it could be a combination of things and this is a
hypothetical Safeway it could be that have turned off their some of their
equipment some of their lighting it could be some of the HVAC systems it could be some of the cooling elements or
anti sweat heaters on their coolers so it could be a bunch of things but this
is sort of what it looks like shedding is you reduce energy and you reduce your
peak and you never recovered it from that and then shaping is really sort of
setting a limit or you know thinking about your peak and saying I'm not going
to exceed this level today and I'll do everything I can to make sure that I can maintain my building at that level these
these strategies are important because if we automate a bunch of buildings not
just single safeway and let's talk about all the safe ways in California if we
can automate their reaction to the duct curve then we can actually deliver these
services that could support some of the issues related to the doctor let me talk
a little bit about the automation so whenever I talk about automating loads
automating buildings people think oh my god no the in you know utility is going
to reach out into my building and control my things and it's quite a different thing what I'm talking about
is so the utility or the Independent System Operator publishes price or
reliability information from the electric secret these are published on the Internet in a machine readable
format so any computer can grab this information and bring it to the facility
and the facility site a site B there is a building control system that has P
programs strategies and this is very important because these are strategies that are pre-programmed by the facility
operator and brings the building into a low-power mode what I mean by that that
it automatically adjusts temperatures by 2 degrees it could automatically turn off some of the lights it could do
a bunch of things but it is designed by the building operator and it's you know
programmed by the building operator so no utility does anything to the building all they
want all they do is provide these machine readable signals to the homes or
the buildings and the buildings respond to those I like automation it's very dear to me
because I spend about 10 years building this infrastructure and the reason for
that is you can automate once and use it many many times so the only cost is the
initial cost of automation and it's much less than some of the generation sources
that we have today the other piece of that is you can use it for short periods
or longer periods you can have a very flexible sort of calling of these events
and and reaction to those events and then the nice the third thing is that there's no human in the loop so if the
facility operator wants to take the day off go play golfing do something else they
can't they don't need to be there to be able to trigger these events and this is
what a couple of buildings like 25 of them that have participated in an automated event look like so we stacked
all of their loads on top of each other we have an event a reliability event
starting at noon and it continues until 6 at 3 o'clock the price of electricity
goes five times so between noon and 3:00 p.m. they do one thing they realize that
the prices increased significantly they do another shed and a deeper cut during
that 3 hours and at the end of the event period some people go home so the
buildings don't even come up some continue to be on but then go home at later times but in general this is sort
of what are we you know an event looks like from a bunch of buildings and this
black line here is what the load would have been if there was no demand
response event and on this day and this is an actual day we saved about two
megawatts of power from these 25 buildings which is significant especially if they're co-located for
that distribution grid it's a it's a big big significance in fact so let me talk a little bit so
that was sort of transmission how the loads can participate in transmission system operators there's all sorts of
policy implications a lot of market implications a lot of costs associated I'm not covering any of those but I want
to talk a little bit about the problems with the distribution that we are facing with all of these renewables on the
system so the okay so the first one that doesn't show up here is that there are
more active devices on the on the distribution network we have inverters that take electricity
that is generated by solar generation converted into alternating current that
we can use and we have control over those inverters but we don't have very good models for running the grant with
these inverters so lots of active devices that can be controllable that are not modeled appropriately the
utility is unaware of small deployments so you want to put solar and they do do an interconnection agreement with your
utility utility takes that paperwork and nicely puts it in the drawer none of the folks in the planning department or the
operation 2-pound Department really know beyond you know 10 kilowatts of solar
what people have in their homes so but if you have a lot of less than 10
kilowatts of solar it can add up to be a lot on a distribution network and can be destructive and and so they'd like to
know if there are customers and keep track of those and of course the
bi-directional power flow and over voltages are things that the utilities are concerned about and that these
renewables at the grid age could potentially cause problems so this is
one device that is an active controller which was in the first bullet point but oh well so I'm going to talk a little
bit about the utility being unaware and what we're doing in terms of providing them information about solar in their
territory from each home and then I'll talk a little bit about some of the machine learning applications that we're
doing so that we can actually detect bi-directional power flows or over voltages on the system very quickly
oh no it comes so one solution that
we've been developing and my colleagues in the room have been working on this for a while is solar disaggregation and
it's using data science and machine learning to tell the utility which customers have solar how much they're
producing and when so in order to do that we're using something called
unsupervised source separation algorithm which is a mouthful I know but what it
does is actually this if this is the net load we're seeing from a smart meter
this is the actual consumption in the home and this is the solar generation in
the same home so it takes that signal and divides it into what portion of it is load and what portion of it is
generation in this home we have about 7.5 kilowatts of peak on this day and
then we take that solar that we disaggregated and we compare it with the data that's actually on that home so we
can see if our solar disaggregation algorithms are working properly and in this case the true solar the measured
solar is in the dark lines and the yellow lines show the disaggregated solar so we're doing okay in an
individual home level it's very difficult this is a very difficult problem to solve we're doing ok but at
the aggregate we're doing considered a considerably much better the other one I
want to talk a little bit about is this detecting over voltages or detecting voltage deviations now let me walk you
through this this is a very interesting thing to me so I hope you'll get to
follow me as well so this is a voltage actual voltage readings at a given time
and after this time we'd like to know what the voltage is going to be so that
if there's a deviation if it's above a certain limit we want to detect that and we want to make sure that we fix that
but taking lots of voltage data and predicting voltages is a very difficult thing and
we can't typically do it because it depends on a lot of variables and we can't possibly predict that what we can
do and we can do really well is that we can take a power data or demand data
from a home or from a neighborhood and we can do we can forecast that demand
data we have been doing this for a long time and we can do this really really well what we've been working on is
developing a learning mapping between the power and the voltage magnitude so that if we can predict power really well
then we can use the reverse mapping to generate voltage readings and then detect this voltage violation so I know
it's a little bit sort of difficult to think about but it is actually in a very
important piece going forward for the distribution system planning and operations and we're really excited to
be working on this problem and in finding solutions that includes data science includes machine learning okay
so how am i doing in time okay so let's
talk a little bit about consumers as consumers we make decisions all the time we make daily energy efficiency as a
core piece of our operations we think about you know we have a lot of
incentives we think about replacing light bulbs we think about using more efficient appliances and we control
better control our devices and this is a daily activity and the objective is to again as much service from each kWh that
we use each kilowatt hour that we use we want to get as much service as we can when you think about time abusive energy
it's really about avoiding high electricity price periods and really consuming during times where there's
electricity is less expensive and the wedge shows like the hours of engagement
with the with the grid so we spend a lot of time on the energy efficiency somewhat on the time of use and probably
less so on the inter increase interactions of the grid those loadshedding shifting
and shaping type of activities the important thing about that is that we need better measurement more control and
and we need to be able to meet multiple objectives and let me talk a little bit about that if you want to save
electricity if you if you have a home today if you want to save electricity you can turn off your appliances but the
equation gets a little bit more complicated when now you have solar you have electric vehicles and you have
storage so now you have to manage a lot more things in behind the same meter so
maybe even get the same type of savings on top of it let's say you decide to provide grid services to the
transmission system operator this ramping service or whatever it may be the objective space becomes even more
complicated so what we're doing at Stanford and SLAC is to develop three
pieces of technology to automate that decision-making and to make sure that you can participate in any kind of a
profitable market and benefit from that transaction so at the lower timescales
at the milliseconds to microseconds we have a technology called smart dimming fuse this is a fuse that sits in your
panel and what it does is very simple things it turns on and off the entire circuit in your home or it can change
the voltage in that within those limits to make sure that you can save electricity power at a higher level at
the minutes to milliseconds level we have a device called the home hub and this home hub maintains the stability of
the system especially if you have solar and you know storage and other things
and then its purpose is to collect all this data from your appliances and your smart meter and your smart dimming fuse
and send it to the cloud coordinator the cloud coordinator is a machine sitting
somewhere having some control algorithms that are embedded in it and it makes
sure that you can share your resources optimally among these homes and also provides some great services so I will
now show you a quick video and I will walk you through this how these systems
are supposed to work together we actually build this video before we actually build the system to just as get a sense
of how it may work for us so let me see
okay so let me stop I told you I was going to take you to my lab our lab this
is our lab this is the lab in building 27 and we're really excited to have the
space to be able to experiment that's like okay so this is a power net
demonstration the objective is to build pin the principle of connecting information networks into power networks
here's what the home hub interface looks like and I'll stop this and talk a little bit
in each of the home so we have two set of lights we have one storage device and
we have one fan those sort of illustrative loads in a typical home
this is the interface for the cloud coordinator and it shows what the price
of electricity is what's the current consumption of all of the homes and some other fancy things now let me know I
knew this was going to happen let me bring it here okay we have two sets of
homes two sets of light lights for the economy home and a fan and battery
storage two sets of lights and a battery storage and a fan for the comfort home
so this is our economy home and this is our comforter yep that's the comfort
home so let's look at the dynamic pricing scenario the electricity price goes from three point three three cents
to three point one cents will you do this is the what we see on the cloud coordinator the lights in the economy
mode one of them turns off and then the storage feeds the electricity back to
these lights so both lights are on but one of them is fed through the storage device
this is what the drop in electricity looks like let's say there's a loss of
generation again this is the economy home this is the cloud coordinator that
says we have a loss of generation notified two units each home gets this message that says loss of generation now
we are economy mode home turns up a set of lights and then uses these lights the
the storage to power this light and turns off the fan in our comfort home the lights are not fed through the
storage and one set of lights turn off and in in our cloud coordinator we see
the economy home going down and then there is a also a another one that goes
down as well but you can't see it it's too faint so these are the goals of the project minimize information exchange
even though we don't have any communication or a very little communication we want the system to
operate we want to enable low costs load control so making sure that we can embed
some of these controllers and our devices increase consumer quality of service at the same time so this is a
project that is funded by the California Energy Commission and the Department of Energy arpa-e program we are partners
with Stanford University Google and University of Florida and we are going to be testing this in the Navy homes in
San Diego so a couple of other things that I want to talk about but I'll skip
this one I'll take you to this one exciting project that we have launched
this year last year and they here is the
the the goal of this project is to really think about if we can deploy
artificial intelligence to improve Grid resilience and we want to use machine
learning and artificial intelligence for anticipating impact of weather events we want to absorb great events by virtual
islanding and that means that changing in controlling modes and recover fast
from the outages and here is a scenario that I want to sort of leave you with let's say
we know all the tilt location of the electricity poles and a tilt and direction of the poles and we know that
there is going to be a heavy wind event coming up there's going to be really strong winds tomorrow we can identify
those polls that have tilt and they could potentially be vulnerable to the
heavy wind we can actually simulate our system to figure out well where the vulnerabilities are but also what the
impact of those vulnerabilities are and we can come up with recovery scenarios and we can automate this entire process
so there's no human-in-the-loop so that's sort of what we're going to be doing in the next couple of years
in this project and we're very excited about it our team is about three years old
our vision collectively we came up with is enabling hundred percent clean energy for all unlike other labs that are
strong in power grids buildings and mobility we're really focusing on the intersection of all of these areas we
want to look at vehicle to grid vehicle to building you know those kinds of
application areas where people haven't really spent a lot of time thinking here
is our team those with black t-shirts on the back they're here to support me so
if you have any questions you can certainly grab them and ask them questions - we're lucky to have a very
strong team and we get a lot of visitors international visitors from Europe and other places we work closely with
Stanford faculty and students and we also have a group of students who come
here and work with us from Carnegie Mellon Silicon Valley campus every year in the fall so we're lucky to have them
so that's all I have thank you so much [Applause]
Sylla but before you guys ask questions
I think I use that part of my question with a loss on the network but if it's a
little off-topic you can say no but how vulnerable is the grid to sabotage the
grant is very vulnerable to sabotage there has been some sabotage attempts we
had people shooting substations and making significant damage especially in
PG nice territory we had several of them we had people stealing copper from some
of these substations yeah I mean our substations our distribution network not
very safe in terms of your duck curve if
you go from 350,000 electric vehicles right now to 5 million it factor 10
increase how is that going to affect the duck curve at 5:00 p.m. and 6:00 or 8:00
p.m. when everybody comes home wants to charge their vehicle and it's going to depend item in terms of efficiency of
the vehicle the range of the vehicle the typical commute distance etcetera so
there's a lot of variables there that I think you're going to be rather interesting to try to handle so there's going to be a lot of incentives to make
sure that there's workplace charging for those vehicles that they can charge in the middle of the day in public charging
stations and workplace charging areas again the right now the incentives or
disincentives are not well aligned my when I take my car home I can charge it
around 10 because that's where the price of electricity go goes down but in the future the price of electricity is going
to be really high during those peak periods descents incentive this incentivize
folks to charge during those times so we need to make sure that our incentives are aligned with the technologies that
we have but unless we do that there's going to be a big problem you're right
in the in the projections that you're making for the future for the state I
use how do you include the cost of making the panels of making the
batteries which are very intensive intensive in terms of use of energy and
of course if you assume that made in Nevada or in China and you don't have to worry about California but in the
overall picture this is pretty important so I don't know if you're familiar with
this new si si decision that every home that's going to be built after a certain
year is going to have to have a solar panel on on its roof I don't model these
costs so I depend on experts who do these calculations and I can send you
some papers that are available but I good there is a professor at Caltech who
wrote a number of papers about this and he said that actually solar panels have
a payback period in energy not in cost an energy of about four years so if you
install a solar panel on your house for the first four years you're not doing
anything it may be three point five years but it's not one or two years whereas in the case of wind it's much
better but you have to take into account this the cost of the energy cost of making the device and the batteries are
much worse than the salt of the pounds
there seems to be a lot of effort on putting individual installations in
houses which are very small-scale and efficient to install not maintained by anybody with
competence why are we not just building square miles of solar panels in the
desert we're doing that too I think it's a combination there's a lot of first
costs for those kinds of large installation yeah no actually the cost
per kilowatt hour is smaller yes but when you think about the scale that you're building at and and this is
something that individual companies private investment has to take on the
solar adoption in the homes is driven by individual household owners or who are
taking it upon themselves and they want to you know install these systems on their roof yes but sort of the will to
participate in yes but the nice people but they're being subsidized by tax
credits which don't apparently go to the utility company so I'm missing the point
here you talked about security with your shape-shifting internet connections and
such what about the other kind of sabotage yeah so when we were developing
this automation system we looked at security really really hard we're using
the same kind of security that you use in your bank transactions so if people are willing to and can get into your
back bank transactions through the connectivity options they can get into your home management not home management
system but this this load shedding system what they can send is nothing
more than a signal that your system can understand and act upon and of course at
a large scale they could be detrimental to the grid as well but you know there
hasn't been a lot of attacks and these demand response systems today I think they still think that attacking banks is
a more profitable so you know breach
with California's new mandate to install solar panels on new buildings do you
think that's being done in a responsible way that's well thought out or is there
some some like vulnerability associated with that with the new mandate to
install solar panels on on buildings is that being done in a well-thought-out
way or is it more ad hoc or can you talk about some of the challenges that that
proposes yeah so it's so new that I actually haven't looked at the details
there's going to be a lot of controversy around this topic the there are some
assumptions that were made in making this decision and I haven't even looked at those assumptions yet unfortunately
this is maybe a couple of weeks old so let's talk in a couple of weeks after I
gone through it I'm at one time there
was I don't know whether to call it a plan or a lot of optimism about using
banks of electric cars during the day to actually feed power back I don't so you
mentioned several parameters of storage and there I don't know if you were including that idea and that storage and
and where is that proposal or idea or wishful yes no I have done one of the
first vehicle to grid demonstration projects in California with at the LA
Airport space in Southern California there we have seen a lot of issues with
vehicle to grid integration first it was extremely costly to get these resources into the market then it was really
costly to have bi-directional charging stations then it was a lot of problems
in terms of power quality issues that these bi-directional inverters were causing at the facility then we realized
that there isn't a lot of money because the expectation was that each of the cars would receive
around $200 per month and could make up for the for the lease for their lease
but at the end these cars made a portion a very small portion of these expected
savings or opportunities so we we looked at it there are a lot of issues maybe it
was before its time but currently we don't have a lot of value in the
marketplace to be able to do that and it's costly to participate and costly to run these systems so the smart meters
the they emit a wireless signal I presume and is there a public API for
how to communicate with them API you have to ask her I think they have turned
all of the antennas but if you can't catch a signal you can ask your utility
to turn the antenna on it is a there is a public API you can pull data using
green button and green button connect options and and yeah you can have access
to your own electricity to a much more granular what's the what's the current
state of the art for utility scale storage I don't know current state of
that art I know that there are a lot of companies that have different types of chemistry's that have been playing in
that area that have been doing installations with utilities there are flowe batteries or lithium ion batteries
or lead acid batteries there's sort of a lot of combinations of these systems so
I'm not really sure maybe someone else can we had a public lecture talking on
batteries yes you a was the person from Stanford University who talked about it and talked about
batteries on all scales including grid integration grid storage batteries you
could probably go to our website and look for our past public lectures and make one of that out so can you talk
about about the life cycle of each different renewable energy technologies
next how long do you need to replace your solar panel when when it becomes straight really ineffective and
batteries and I guess the wing will be long lasting but I'm not sure everyone
claims that their system can go over 25
the 30 years the problem is that some of these battery chemistry some of these batteries haven't been around that long
so well there's a lot of sort of claim that these the lifespan of these
technologies can be long we haven't really had these technologies and long
enough times to sort of really understand their lifetimes so we'll wait
and see I guess do you have any
information on the problem that appears to be happening is we're going to more
efficient equipment which is constant power the lighting and computers always
been that way and the grid relies on loads basically if you lower the voltage
they draw a little less power but actually now they draw more current yeah
is there any work being done on that yeah so the smart Deming fuse that we're
working on is actually looking at extracting you sound like a technical person so I'll go a little bit technical
it extracts the IV curves for each of the sub circuits and then when you have
that you can actually determine if changing voltage is going to save you any energy if you have inductive loads
that may not to be the case if you have resistive loads you could actually changing the voltage you can save energy
now if I have to choose between energy efficiency LEDs versus controllable I
don't know incandescent lightbulbs I'll take LEDs low consumption energy efficiency every
day always and then think about variable resources load flexibility with other
types of things but for me energy efficiency is always first flexible load
come second only could you actually
start thinking about making kind of any
sort of Giveaways from energy efficiency if there is any value that the consumer
can extract from other things and can help with taking on low cost clean
technologies right that's the only time that I would say maybe we want to think about rethink this if if it can operate
on clean energy low-cost clean energy maybe want to take that energy on and
help the grits in some days but overall energy efficiency always first yeah do
you guys take into consideration a reactive power for the inductive and
capacitive loads in controlling the overall situations and Peaks yeah
so I didn't go there I didn't go explaining and talking about the reactive power but the new inverter
technology is an electricity rule 21 that was passed really pays attention to control of the reactive power to these
inverters have capabilities not only control the bolts but also bars they reactive the power component of the
system so it is a big issue reactive you know dealing with reactive power is a
big issue and these new technologies are positioning themselves in a way that we
would be able to handle some of the reactive power issues as well you're right
[Music] a few minutes and gives more team will
be around [Applause]
[Music]