Content uploaded by Harrison Balsdon
Author content
All content in this area was uploaded by Harrison Balsdon on Mar 08, 2017
Content may be subject to copyright.
Balsdon, Khan, Park 1
Application of Smart Grid Technology to the City of Palo Alto
L&S 126 - Energy and Civilization
University of California, Berkeley
Authors:
Harrison Balsdon
Aman Khan
Matt Park
Balsdon, Khan, Park 2
Table of Contents
Abstract 4
I The Growing Necessity of Smart-grid Technology and the Nature of the Renewable
Energy Transition (Park) 5
II Analyzing Smart Grid Technology (Khan) 10
III Understanding and Evaluating Utilities (Park) 15
Potential Adopters 16
IV The Case for Palo Alto (Khan) 18
Summary of Technology 21
V Extrapolating [Applicable] Techniques, Innovations in Artificial Intelligence to a
Smart Grid-Centric Renewable-Power Architecture (Palo Alto, Ca) (Balsdon) 22
i. Introduction 22
ii. Artificial Intelligence 24
iii. AI Techniques Applicable to the Smart Grid Power Paradigm 25
A. Paritional Cluster Analysis Implementation in Wind-Energy Generation and
Distribution Systems 25
B. Sorting “Big Data” of the Smart Grid Using Complex-Event Processing 26
C. Utilizing Fuzzy Logic in Smart Grid Decision-Modeling and Frequency
Control 27
D. Applying Artificial Neural Networks to the Smart Grid 28
VI Conclusion 30
*For the purpose of identifying paper sections and member contributions, as per rubric
requirements
Balsdon, Khan, Park 3
Abstract
The purpose of this paper is to examine various Smart Grid technologies, and analyze
locations within California that would benefit most from upgrading their grid with this
technology. We believe that due to the growing production of renewable energy and our
inefficient and outdated grid infrastructure, smart-grid technology would provide us the best
opportunity for efficient energy production, distribution, transmission and storage. This paper
will analyze multiple locations, and technologies, and select the most optimal cases for adoption.
As will be explored, Palo Alto is a suitable candidate for Smart Grid technology incorporating
machine learning based demand response and transmission systems, and energy storage.
Balsdon, Khan, Park 4
I
The Growing Necessity of Smart-grid Technology and the Nature of the Renewable Energy
Transition
This paper aims to evaluate a number of potential candidates (municipalities) for the
adoption of Smart Grid technology in the state of California. This paper will evaluate the
economic and technological potential of smart-grid adoption for a specific location and
recommend strategies to implement the identified technologies. For this proposal, we define a
smart-grid to be a computer controlled system that efficiently generates, stores, and distributes
electricity over a particular area using upgraded yet conventional transmission methods. The
purpose of these technologies is to bring 21st century computation and automation to the aging
electric grid infrastructure. Historically, utility companies have had to send workers to manually
read electricity meters to formulate approximations for demand. These technologies come in the
form of continuous communication devices connected to a network that can send sensor data in
real time to a utility’s network operations center, giving utilities faster access to the information
they need, and allowing them to more efficiently distribute their electricity
An example of Smart Grid technology comes in the form of using hardware to monitor
and potentially control a homeowner’s electricity usage, and react to changes in behavior in real
time. For instance, using data collected regarding an individual’s energy usage, a utility company
may be able to propose the best time to turn on a dishwasher, or control the home’s HVAC
(heating, ventilation, and air conditioning) system, to maximize returns for the consumer while
reducing demand during peak load. This technology is especially pertinent considering the
potential savings and environmental benefits Smart Grid networks offer, as well as providing
robustness to the grid from a systems standpoint. It is estimated that Smart Grid technology
could increase the U.S. grid’s efficiency by 9%. This efficiency boost could provide savings of
nearly $600 per household, providing a compelling case to modernize a specific community’s
existing local grid.1 Furthermore, U.S. spending on electric grid automation is expected to reach
$1.7 billion in 2016, providing a massive market opportunity for utilities and privately held
companies to develop and execute technology that will modernize the grid.2,3 However, the
incentives to modernize the grid are not just economical. Upcoming energy technologies will
soon be implemented on a wide enough scale that will change the electricity landscape. Because
of products such as renewable energy and batteries, we believe that our current grid will be
incapable of acceptably distributing energy.
1 Smart Grid Consumer Collaborative. 2016. Smart Grid: Where Power Is Going. Web.
2 U.S. Department of Energy (DOE). 2014. 2014 Smart Grid System Report. Web.
3 Statista Inc. 2008-2017. Smart Grid Expenditures Outlook in U.S. by Project 2017. Web.
Balsdon, Khan, Park 5
According to the U.S. Energy Information Administration, every year since 2011, the
amount of energy produced solely by solar has increased by 150%-280% year-over-year.4 In
mathematical terms is exponential growth. Every year our energy portfolio is becoming more
renewable, and every year it is becoming renewable faster. The cost of renewable energy and
technologies, namely solar, wind and batteries, are decreasing at an exponential rate5, and
investments in renewable energy are also higher than they are for fossil fuels.6 Contributing to
this growth are the laws enacted by Governor Brown, which require all utilities to provide 50%
of their electricity from renewable sources by 2030.7 Most Californian utilities are on track to
meet this goal, which itself accelerates renewable energy growth. However, there are a few
utilities that will most likely surpass expectations, accelerating renewable energy adoption even
faster.
Batteries will become a more relevant factor in coming years. The overall costs of energy
storage have been reduced dramatically.8 It is estimated that energy storage prices by 2020 will
be half of the price of in 2015.9 All of this is to say that in the very near future, it will be more
economical to generate power from renewable sources and store it than to source it from a fossil
fuel power plant. Our energy infrastructure will need to adapt to how the market changes for our
energy generation. Renewables present two key challenges to the grid. One is intermittency, or
non-continuous energy generation, because renewable energy sources do not produce power at
all times. The “Duck Curve” describes the disparity where peak generation of renewable energy
does not match with peak use during the course of a day.10 Batteries can help mitigate this
disparity, to an extent.
4 U.S. DOE. 2016. Electric Power Monthly with Data for September 2016. Web.
5 U.S. DOE: Energy Efficiency & Renewable Energy. 2016. Revolution Now 2016. Web.
6 Randall, Tom. “Wind and Solar Are Crushing Fossil Fuels.” Bloomberg.com. 2016. Web.
7 California Energy Commission. 2016. California Energy Commission- Tracking Progress. Web.
8 Randall, Tom. “Tesla Wins Massive Contract to Help Power the California Grid.” Bloomberg.com. 2016. Web.
9 Lazard Ltd. 2016. Lazard Levelized Cost of Storage Analysis-Version 1.0. Web.
10 California Independent System Operator (California ISO). 2016. Fast Facts. Web.
Balsdon, Khan, Park 6
Fig 1. The Duck Curve
The second upcoming challenge is the effective decentralization of grid components. If
green energy technologies continue to fall as they have, California will see multiple small power
plants in the form of rooftop solar as well as solar and wind farms. However, given the current
market trends, it will become more economical for the average customer to have decentralized
power generation and energy storage. This decentralization means that as a whole, the grid will
have more independent yet connected components and be more complicated to manage. Smart
management can make the intermittency and decentralization a mere afterthought. With the sheer
amount of separate parts, from rooftop solar, to smaller solar and wind farms, and batteries, it is
unreasonable to assign a human or group of humans the task of managing our electricity
distribution. In the coming pages, we hope to establish an economic timeframe or condition set
where smart-grid adoption is economical.
The transition from fossil fuels to renewable energy is fundamentally different to the first
widespread adoption of fossil fuels. Instead of looking at the economic trend of industrialization,
we should study adoption of technologies more similar to renewable energy and batteries, for
example modern day necessities such as refrigerators, televisions and smartphones. The closer
one looks at the current renewable energy transition, the harder it is to see similarities to initial
industrial energy adoption. The only real similarity between renewable energy and fossil fuel
power plants is their output: energy.
However, that is where the similarities end. In terms of scale, all fossil fuel plants need to
output a few tens of megawatts to be economical.11 The fact that rooftop solar and consumer-
scale batteries are successful technologies that provide energy on the scale of several kilowatts,
which is smaller by a factor of at least a thousand, means that those technologies are
economically flexible in ways that fossil fuel plants simply are not. Another difference is the
stability of price of energy derived from fossil fuels and renewable sources. Simply put, the price
of fossil fuels fluctuates wildly, especially compared to the stable decline of prices of renewable
11 Wikipedia. “List of Power Stations in California.” 2016. Web.
Balsdon, Khan, Park 7
energy. Various factors cause these fluctuation, from international politics, energy demand,
access to the resource, among others. A 2006 report by the California Energy commission stated
“The best assumption about all forecasts for commodities as volatile as natural gas is that they
will be wrong.”12
Over the past 10 years, the price of oil has been quite erratic, ranging from roughly $150
per barrel to $28.50 per barrel, and it is currently around $50 per barrel.13 The primary
characteristic of the price of oil is its unpredictability. In a single year, from October 2015 to
October 2016, the coefficient determination for the price compared to the month is 0.292,
meaning that a year's worth of economic data is almost useless to predict the price of the next
month. A similar unpredictability is seen in the natural gas market. Compare this uncertainty to
the trend of the price of renewable energy technologies.14 While the price of a fossil fuel can go
up or down from year to year, solar, wind, and batteries consistently go down in price.
With such a dramatic difference in both scale and price patterns, it is easy to think that
there is something fundamentally different between fossil fuels and renewable energy
technology. Indeed, that does seem to be the case. Technologies get cheaper at an exponential
rate. Most famously is Moore’s law, which describes how from 1975 to 2012, the cost of
computing power halved every 18 months, after which, according to Gordon Moore, namesake
of said law, “No longer can we depend on making things smaller and higher density.”15 Solar, in
particular, as well as wind energy, batteries, and electric cars, follow economic trends much more
similar to computer technologies.3 Therefore, looking at the economic history of unpredictable
fossil fuels is less helpful than looking at the economic history of consumer-scale technology,
especially because solar, batteries, and electric cars fall into that category. We can learn more
from the adoption of cars, color tv’s, and refrigerators than from the economics of fossil fuel
plants by looking at a theory penned by Professor Everett Rogers.
Professor Rogers details the phenomenon of “Diffusion of Innovations”. It describes the
socioeconomic pattern where once a given technology or product reaches a certain market
permeation, adoption of said product grows exponentially fast from a combination of societal
pressure and economic viability.16 There is reason to believe that solar power will very soon see
widespread adoption in certain U.S. markets, namely Hawaii, Arizona, and largest of all,
California.17
We believe that the oncoming adoption of solar power in California is reason enough to
adopt smart-grid technologies within the state, as solar adoption does put significant strain on the
12 California Energy Commission. 2007. Integrated Energy Policy Report Update 2006. Web.
13 Macrotrends LLC. 2010-2016. “Crude Oil Prices- 70 Year Historical Chart.” Web.
14 U.S. DOE: Energy Efficiency & Renewable Energy. 2016. Revolution Now 2016. Web.
15 IEEE Spectrum: IEEE.org. 2015. “Gordon Moore: The Man Whose Name Means Progress.” Web.
16 Hoffmann, Volker. 2007. Farmers and Researchers: How Can Collaborative Advantages Be Created In
Participatory Research and Technology Development? ResearchGate.net. In Agriculture and Human Values. Web.
17 McMahon, Jeff. “Hawaii Sitting On The Life of A Solar Explosion.” Forbes.com. 2016. Web.
Balsdon, Khan, Park 8
grid as it reaches 20% penetration.6 The goal of our proposal is to push certain areas of California
toward early widespread adoption of smart-grid technologies to accelerate its diffusion of
innovation.
Balsdon, Khan, Park 9
II
Analyzing Smart Grid Technology
Successful Smart Grid networks and technological implementation can be seen in a
number of areas around the country, strengthening the argument that Smart Grid adoption is
increasingly widespread with a number of measured benefits. These examples are of varying
scale, yet all attempt to improve grid efficiency at some level thus providing economic savings in
the long term. The Department of Energy (DOE) lists a number of these case studies on their
website, as well as a range of classifications of the type of technology deployed for specific
projects. These initiatives are largely the result of the American Recovery and Reinvestment Act
(Recovery Act) of 2009, which provided the DOE with $4.5 billion to modernize the electric
grid, hence leading to its two largest initiatives: The Smart Grid Investment Grant (SGIG) and
Smart Grid Demonstration Program (SGDP). The SGDP in particular highlights a number of
potential examples similar to the model this paper hopes to validate, with a focus on regional
demonstrations of Smart Grid technology and energy storage demonstrations.
Many of the SGDP projects are currently in the process of being analyzed. As a result,
these projects serve as a relatively accurate and recent source of information regarding the
effectiveness of Smart Grid initiatives within the US. One such report highlights a community in
Kansas City that underwent significant power substation and feeder upgrades as the result of $24
million in funding from the Recovery Act, amounting to a total project value of $49.8 million.
This project attempted to “demonstrate, test and assess feasibility of integrating new and existing
technologies in an end-to-end Smart Grid”.18 Kansas City Power and Light (KCPL) implemented
this Smart Grid across a 5-square-mile area, servicing 13,427 customers. A major aspect of this
initiative was the implementation of an updated substation relay system, whereby operators can
effectively monitor and control equipment across the grid with ease, hence improving worker
safety and ability to react to outages. Furthermore, by implementing an Advanced Metering
Infrastructure, with a much higher data resolution than the system previously implemented
(every 15 minutes as opposed to once a day), the utility company was able to reduce monitoring
costs by $104,120. Lastly, KCPL’s updated feeder system saw an estimated 1.6% reduction in
grid-wide energy use through a reduction in voltage (by an average of 1.64%) on peak days.
Within the Bay Area alone, there is a significant amount of monetary investment in Smart
Grid technology development, with a reported $254.8 million in 2013, which accounted for 29%
of total national investment in Smart Grid companies.19This is largely due to the Green Vision
initiative launched by San Jose which hopes to increase investment in clean energy technologies.
However, as investment in clean-tech has decreased (in 2013) compared to previous years, Smart
Grid investment has increased, providing precedence for a number of startups and established
companies, such as IBM20, which are invested in improving grid technology to have a key role in
the development of an efficient and exemplary grid.
18 U.S. DOE: Electricity Delivery and Energy Reliability. 2015. Renovating the Grid and Revitalizing a
Neighborhood: Successes from Kansas City Power & Light’s Smart Grid Demonstration Project. Web.
19 Johns, Chris. “Silicon Valley Continues Strong Lead in Smart Grid Jobs.” PGECurrents.com. 2013. Web.
Balsdon, Khan, Park 10
Among the abundance of case studies supporting Smart Grid networks, highlighted by
the SGDP, there are a number of specific technologies referenced that have the potential to
maximize economic benefit and energy reductions for the communities highlighted. These
specific technologies have been charted below with their corresponding frequency.
Fig 2. Source: Smartgrid.gov
Advanced Metering Infrastructure (AMI) is charted as one of the highest technologies
utilized during the SGDP program. An AMI system operates on the basis of two-way, always on
communication of commands between the utility company (either at a distribution or
transmission level) and the end user. The appeal of AMI is largely centered around the identified
need for highly efficient utility demand response systems. This is an area of concern identified
by a number of utility companies around the country, largely for the implementation of a
dynamic pricing model.21 This would not only maximize profits for utility companies, on the
principle of supply-demand economics, but also incentivize lower energy consumption during
peak load times, thus flattening out the “duck curve” (identified in earlier parts of this paper).
Demand response (DR) would also allow for a more sustainable, renewable energy baseline,
effective forecasting, and support smarter energy storage methods.
However, as the SGDP began implementing AMI across the nation, a number of potential
pain points were quickly identified. Though AMI is effective in monitoring and tracking energy
20 IBM Industries. “Energy and Utilities: Case Studies.” IBM Smart Grid Case Studies. 2016. Web.
21 Conca, James. “‘Demand Response’ Is How the Smart Grid Will Save Us Billions.” Forbes.com. 2015. Web.
Balsdon, Khan, Park 11
usage, it suffers from latency issues and falls shorts in providing rapid response to demand.22
Bandwidth is the core issue for AMI, especially due to the number of demand response events
required for effective feedback control in the real world. Though AMI has demonstrated benefits
in settings that require lower end use devices, scaling up to cities increases response times from
minutes up to periods of hours. This may further strengthen the need for decentralized energy
distribution and transmission, especially in the case of renewables. Furthermore, utilizing a pre-
emptive demand response system based on AI and machine learning would leverage existing
AMI technology in a number of areas, and allow for a dynamic pricing model to be instituted
without an entire infrastructure revamp, where AMI has already been implemented.
Beyond demand response, a number of further Smart Grid technologies were identified
that provided innovation at incremental stages of the electricity generation, distribution,
transmission, and energy storage process. One such transmission improvement is the adoption of
dynamic line ratings (DLR), which is a success story highlighted by the New York Power
Authority (NYPA).23 Line ratings determine specific physical properties of a power line, such as
the maximum amount of current that the line’s conductors can carry, and is generally a static
property. These properties are dependent on a number of variables such as weather conditions,
conduction temperature and sag (power lines droop, which can affect transmission properties and
lead to potential faults). The issue with static line ratings is that they are inherently dependent on
parameters that are constantly changing. Wind speeds and external temperature are not constant,
and as such the factor of safety for line ratings is overly conservative, leading to potential
transmission losses. By implementing DLR technology, the NYPA was able to increase overall
usable transmission capacity by 25%.
Operational efficiency can be improved on a distribution level through voltage
optimization, which was observed in an earlier highlighted SGDP case study of KPC&L. Volt
Var Optimization (VVO) improves overall grid efficiency and reliability, hence allowing for
conservation voltage reduction measures to be implemented. As described in an SGDP paper on
the topic, “voltage optimization consists of two steps, control of power quality and voltage
extremes by putting capacitors and voltage regulators…on a line; and using reduced voltages to
conserve energy.”24 A few of the potential challenges and motivations of VVO technology can be
seen in Table 1.
Challenge Motivation
Power Factor financial Charge for load requiring excessively leading or lagging power
22 Ng, Howard. “Does AMI Have What it Takes for Demand Response?” Comverge Blog. 2013. Web.
23 U.S. DOE: Electricity Delivery and Energy Reliability. 2015. Improving Efficiency with Dynamic Line Ratings:
Successes from New York Power Authority’s Smart Grid Demonstration Project. Web.
24 U.S. DOE: Electricity Delivery and Energy Reliability. 2015. Voltage and Power Optimization Saves Energy
and Reduces Peak Power. Web.
Balsdon, Khan, Park 12
penalties
Resistive Line Losses Higher currents associated with higher I2R losses
Lost Capacity Extra inductive current wastes conductor and transformer
capacity
Voltage Drop Excess current results in excess voltage drop along line
Preparation for Conservation
Voltage Reduction (CVR)
End of line voltage must have sufficient margin to allow
reduction
Table 1. Challenges and Motivations for Voltage Optimization
The economic viability of voltage optimization is clear as well. Given an initial
investment $20,000 and savings of approximately $4,000 a year, VVO technology typically pays
for itself within a 5-year period.
A major aspect of the SGDP that will be explored in this proposal is the demonstrated
cases of energy storage technologies implemented across the country, and their implications.
Energy storage is crucial to effective demand response systems and overall grid efficiency. This
technology plays a major role in leveling out the duck curve and creating a stronger baseline for
renewables.25 By storing energy at times that renewables are most effective, for example peak
solar, and saving that energy until peak load, we can more readily produce renewable energy
throughout the day, rather than resort to fossil fuels sources to fill this gap. Furthermore, this
allows fossil fuel plants to produce energy at a constant rate, rather than ramp up or down
depending on demand, which encourages production to achieve optimal efficiency. This market
is projected to reach $13.5 billion dollars annually, providing a huge opportunity for progressive
economic development. 26 Though this technology comes in a number of forms, ranging from
batteries, to fly wheels, the SGDP explores a number of cases that highlight energy storage
applications, as well as the costs and benefits of each of these measures.
Many demonstrations of battery technology with application to grid storage are still
ongoing and in development, and provide potential for a high density, low cost solution. This is
evident in projects such as Tesla’s recent partnership with SoCal Edison's Mira Loma substation
in Ontario, CA to build a 20 MW/80MWh storage facility which could power 2500 homes for a
25 U.S. DOE; National Renewable Energy Laboratory (NREL). 2015. Overgeneration from Solar Energy in
California: A Field Guide to the Duck Chart. Web,
26 Yole Developpement: I-Micronews. 2016. Energy Management for Smart Grid, Cities and Buildings:
Opportunities for battery electricity storage solutions. Web.
Balsdon, Khan, Park 13
day.27 This project is part of the California Public Utility Commission’s aggressive push to add
1.3 GW of energy storage to the grid by the end of the decade.28
The flywheel solution, which is an emerging technology, is explored in detail in a report
by Hazle Spindle.29 Their technology, the Beacon Gen4 flywheel, is designed to provide 100 kW
of output and store 25 kWh of energy. By combining 200 flywheels in parallel to provide 20MW
of capacity at the Humboldt Industrial Park in Hazle Township, Pennsylvania, this system can
respond to grid imbalances almost instantaneously, in less than four seconds. This solution is
entirely mechanical hence requiring little-to-no maintenance, with 85% efficiency and built to
last 20 years, or 100,000 cycles. As of December 31, 2014 the plant has delivered 11,655 MWh
of electricity. Unlike batteries, flywheels can operate at any temperature, and have a much longer
operating life (20 years compared to roughly 36 months in the case of lithium ion polymer
batteries). Gene Berry, a researcher from Lawrence Livermore National Laboratory, estimates the
capital cost of flywheel technology to be ~$200-500/kWh compared to ~$100-200/kWh of
batteries, however given the longer (cycle) lifetime of flywheels, and the potential of reduced
costs through innovation, flywheels have the potential to be a sustainable source of energy
storage rivaling existing solutions.30
27 Golson, Jordan. “Tesla is building an 80MWh battery pack to supply LA with power.” The Verge. 2016. Web.
28 John, Jeff St. “California Passes Huge Grid Energy Storage Mandate.” Greentech Media. 2013. Web.
29 Amber Kinetics, Inc. 2015. Technical Report Smart Grid Demonstration Program: Flywheel Energy Storage
Demonstration. 2015. Web.
30Berry, Gene. Present and Future Electricity Storage for Intermittent Renewables. Lawrence Livermore National
Laboratory. Web.
Balsdon, Khan, Park 14
III
Understanding and Evaluating Utilities
To decide which location in California is best suited for adopting smart-grid technologies,
we first must establish what kinds of organizations, businesses, and government services control
electricity distribution. There are four types of electric utilities in California: investor-owned
utilities, publicly owned utilities, joint powers authorities, and electric cooperatives.
“Investor-owned utilities (IOUs) are private electricity and natural gas providers.
California Public Utilities Commission (CPUC) oversees IOUs. Pacific Gas and Electric, San
Diego Gas and Electric, and Southern California Edison comprise approximately three quarters
of electricity supply in California. Publicly owned utilities (POUs) are subject to local public
control and regulation. POUs are organized in various forms including municipal districts, city
departments, irrigation districts, or rural cooperatives. Municipal districts may include territories
outside city limits or may not even serve the entire city. Cooperatives are owned by the
customers they serve usually in rural areas. There are more than 40 POUs in the state that
account for approximately a quarter of electricity supply in California. Most POUs are smaller
than IOUs in the electricity sales and the number of customer accounts.”31
The California government provides this infographic summarizing the differences
between IOUs and POUs (see Appendix E). “‘Joint powers authorities’ is a term used to describe
government agencies that have agreed to combine their powers and resources to work on their
common problems.”32 These are essentially agreements between separate utilities in which all
members contribute resources to work toward a common goal. For the sake of our proposal, that
goal is to provide electricity. Electric cooperatives are similar to joint powers authorities in the
sense that they are a collaboration between multiple parties, and IOUs in the sense that portions
of the organization are owned by individuals. In electric cooperatives, all those who pay for its
service own a portion of the cooperative.33
When evaluating which of the 50+ Californian electric utilities would be the best fit for
early adoption of smart-grid technologies, we must look at the compatibility of smart-grid
technology and the nature of each kind of utility. Immediately, joint power authorities seem
ineligible, as they are by far the most decentralized such as the Power and Water Resources
Pooling Authority, which spans over 200 miles without covering the whole area in its range.34 It
is also rather futile to evaluate the utilities within joint powers authorities, as their individual
31 CEC: Energy Assessments Division. 2016. Differences Between Publicly and Investor-Owned Utilities. Web.
32 California State Legislature. 2007. Governments Working Together: A Citizen’s Guide to Joint Power
Agreements. Web.
33 MJM Electric Cooperative. 2015. “What Is a Cooperative.” Web.
34 Power and Water Resources Pooling Authority (PWRPA). 2004. “About Us.” Web.
Balsdon, Khan, Park 15
information does not seem to be available online, as the California government maintains records
of joint powers authorities as a whole rather than individually.
There are some factors to consider while evaluating investor owned utilities. First is their
financial obligations. IOUs exist maximize gains for the shareholders. If our hypothesis that
smart-grid technology adoption will result in lower prices for the consumer, that is a direct
contradiction with the goal of IOUs. They are under no obligation to provide cheap electricity,
and to suggest this to IOUs may be an exercise in futility. Furthermore, to push a city to require
utilities to adopt smart-grid technologies may result in lengthy negotiations between the
government and the corporation, delaying any pilot programs. Secondly, many IOUs have
already started investing smart-grid technologies, although not to the degree we are proposing
(see Appendix D). It is our opinion that to design our pitch to investor owned utilities will be
either futile, as they will be stubborn to adopt the technologies, or pointless, as they may have
already started.
Electric cooperatives are in a sense the diametric opposite to investor owned utilities.
Their purpose is to provide electricity, and, in principle, all money made from its customers, who
are by definition shareholders, go back into investing in the infrastructure. Cooperatives have the
incentive to adopt smart-grid technologies, as they will bring down the cost of electricity over
time. However, we remain skeptical on the economic feasibility for cooperatives to adopt the
technologies we propose. The largest electric cooperative we identified, The Plumas-Sierra Rural
Electric Cooperative, sold 139,304 MWh in 2015. We are not confident that this volume of
electricity sold has generated the funds necessary for the utility to proceed confidently toward
smart-grid technology adoption.
Finally, publically owned utilities appear to be the best candidates when evaluating the
most willing and capable utilities for smart-grid adoption. They are run by the local government,
and so are incentivized to provide the cheapest electricity to its citizens, much like cooperatives.
Unlike cooperatives, publically owned utilities tend to be much wealthier. Utilities for cities such
as Anaheim or Riverside sell as much as 3,110,488 Mwh per year, more than 30x that of the
largest electric cooperative. Additional funds in the form of tax dollars are also potentially
available. We believe that a publically owned utility will be the most willing and capable to
adopt smart-grid technologies.
Potential Adopters
We have compiled information on every utility in California to determine which would be
best for early adoption. We prioritized many factors in our evaluation including:
●Interest in smart-grid technology, as those utilities are more likely to be receptive to our
proposal.
●Type of utility and reliance on renewables, as those grids will benefit more from smart-grid
technology.
●Future climate and energy goals of the cities the utilities provide. This gives us an indication on
the willingness of local governments to contribute resources to adoption.
Balsdon, Khan, Park 16
●Volume of electricity sold in megawatt-hours. This variable gives us insight into the scale of the
utilities’ holistic infrastructure. Too large of a scale may seem unfeasible to otherwise interested
parties, while too small a scale may seem uneconomical for the time being.
●Cost of electricity. If a smart-grid makes energy cheaper in the long run, it is more attractive to
consumers of more expensive electricity.
Considering these criteria, we narrowed down our list of potential adopters to a few
utilities. We constructed a matrix of all utilities containing relevant information, and have
included it in appendix D for reference. The utilities described below stood out in multiple or all
criteria.
●Silicon Valley Power
●San Francisco City and County Utility
●City of Industry Public Utility
●Port of Oakland
●Pasadena Water and Power
●The Sacramento Municipal Utility District
●The City of Palo Alto.
After comparing all the utilities described above with each other, we have determined
that Palo Alto is best fit to be a pilot program for a smart-grid.
Balsdon, Khan, Park 17
IV
The Case for Palo Alto
Most importantly, the city of Palo Alto runs its own electric utility, which we believe will
be the most receptive and capable type of utility for this pilot project. The city itself covers a
rather small area of around 25 square miles, mitigating the cost of long range infrastructure. Palo
Alto is also the most reliant on renewable energy out of all California utilities; it is on record as
deriving all its electricity from renewable resources. It prides itself on being the first carbon-
neutral city in America, and it is pushing to become carbon free.35 We believe investment in
Smart Grid technology will help them achieve that goal.
As for the question of comfortable economic feasibility, Palo Alto has the financial
resources. The citizens of Palo Alto are among the wealthiest and most educated in California,36
and many tech companies, from Amazon, Google, Tesla, and so on, are based in Palo Alto. Along
with the high living cost37 and electricity prices, private individuals, private companies, and the
city all have incentives to lower the cost of electricity. However, it is not necessary for the city to
be the sole bearer of the financial responsibilities of Smart Grid implementation. Palo Alto can
also fall onto the resources of not just the city government, but also San Mateo County, the
California State government, and perhaps most interestingly, Palo Alto can look to the Northern
California Power Agency. Due to the unpredictability of the impact that President Trump’s
administration will have on the Department of Energy (DOE), we do not know how helpful the
DOE will be financially. We see this as all the more reason for our initial adoptee to have the
most financial resources available.
The Northern California Power Agency (NCPA) is “a California Joint Action Agency,
[which] was established in 1968 by a consortium of locally owned electric utilities to make joint
investments in energy resources that would ensure an affordable, reliable, and clean supply of
electricity for customers in its member communities.”38 Palo Alto is a member of NCPA, along
with these utilities:
●Alameda Municipal Power
●Bay Area Rapid Transit
●City of Biggs
●City of Gridley
●City of Healdsburg
35 City of Palo Alto. 2016. “Electric Renewable Resources.” Web.
36Advameg, Inc. City-Data.com. 2016. “Top 101 Cities with the Most People Having Master's or Doctorate
Degrees.” Web.
37 Dunn, Jeff. “Here's how expensive it is to live in the heart of Silicon Valley.” Business Insider: Tech Insider.
2016. Web.
38 Northern California Power Agency (NCPA). 2016. Web.
Balsdon, Khan, Park 18
●City of Lompoc
●City of Ukiah
●Lodi Electric Utility
●Port of Oakland
●Redding Electric Utility
●Roseville Electric
●Silicon Valley Power
●Truckee Donner PUD
●Turlock Irrigation District
Assessing the resources and authorities available to Palo Alto, we are confident that it has
the financial capabilities of implementing smart-grid technology. Furthermore, the city of Palo
Alto has shown interest in Smart Grid technology multiple times in the past decade, indicating
potential for adoption in the near future.
In 2009, the CPAU began investigation into Smart Grid technology applicability in Palo
Alto, which led the council to hire an outside consultant (EnerNex Corporation) in 2011 to
survey infrastructure and gauge interest in potential Smart Grid applications.39 A summary of
their findings concluded that the city could benefit from AMI “smart meter” integration, whereas
Advanced Distribution systems such as Distribution Automation (DA) would likely not improve
efficiency, due to existing reliability. Instead, the consultant recommended waiting at least 2
years before adopting modern Smart Grid technology, due to the potential of lowered costs and
higher return on investment. The consultant also recommended a number of action items to
further strengthen Palo Alto’s case for Smart Grid adoption in the future, including developing
Demand Response (DR) pilot programs, a Distribution System Automation roadmap, and
analyzing volt/var energy conservation and AMI technology. In fact, many of the action items
identified have been explored in this paper. Although a re-evaluation date of 2014 was set at the
time of this discussion, to date the city has not publically proposed any follow up measures after
the two-year strategy that it plans to implement, indicating potential for our proposal to fill this
void.
The consultant’s apparent hesitancy to suggest AMI as an immediate measure, and
suggestion to instead investigate DR software further at the time of engagement echoes a
sentiment explored in this paper, due to AMI’s shortcomings. The implementation of smart
meters would undoubtedly improve efficient metering in the city, though without necessarily the
need for AMI specifically. Furthermore, DA has been developed further in previous years,
particularly with regards to an Artificial Intelligence (AI) layer being applied over distribution.
These two aspects, DR and AI applied to DA, will be explored further with respect to their
application to CPUA, and the specific forms these technologies might take.
Expanding on the energy storage technologies identified, our initial hypothesis was that it
would be beneficial for Palo Alto to adopt a method of storing the renewable energy it generates
to lower costs associated with ramp-up during peak demand. The CPAU is clearly aware of the
39 City of Palo Alto: City Council Staff Report. 2012. Assessment of Smart Grid Applications. Web.
Balsdon, Khan, Park 19
potential benefits of energy storage, which is evident in a well-detailed memorandum released by
the authority in December 2013 covering the technologies identified in this paper as well as a
number of alternative technologies.40 This initiative is further reinforced by the statewide
“California’s energy storage law” (Assembly Bill (AB) 2514)41, which requires all publicly
owned utilities (POU) to evaluate and procure cost effective energy storage systems.
After evaluating these technologies, the city then moved to not set energy storage
procurement targets due to lack of cost effective options, in February 2014.42 AB 2514 requires
that the city reevaluate its target every three years, and this deadline is approaching rapidly.
Interestingly, the two technologies highlighted by the CPAU that had the highest potential for
cost effective integration were thermal energy storage (which is, in a sense a type of battery for
HVAC systems43), and vehicle-to-grid battery storage. However, further investigation is needed
into the method of quantifying the cost/kWh of batteries utilized by the CPAU. There is some
discrepancy between the ~$350/kWh number listed by the CPAU and the ~$100-200/kWh figure
identified earlier. This may be due to the CPAU accounting for capital costs associated with the
installation of batteries.
In their report, the CPAU also identified the potential benefits consumers may experience
by installing their own personal energy storage, especially when paired with the generation of
rooftop solar. This is a model championed by Tesla, especially considering the company’s recent
purchase of Solar City, which opens the door for potential joint consumer and utility partnerships
with the company.
The report by the CPAU is extremely thorough in its findings, and clearly the evidence
they have accumulated points against immediate energy storage procurement as of 2014.
However, with the preset AB 2514 three-year reassessment nearing, it may be beneficial for
CPAU to revisit the possibility of energy storage in a manner that benefits them. We recommend
that the committee further investigate energy storage in the form of batteries, or as an emerging
technology in the form of flywheels.
Summary of Technology
The technologies identified and suggested so far will undoubtedly lead to increased
capital investments and initial costs that will be paid back in the near future through efficiency
savings. Though it is difficult to assess the exact costs of all the technologies proposed and their
return on investment (ROI) (as this information is not readily available, and is often specific to
the technology and context with which it is applied), by analyzing a few of the top contenders, a
sanity check can be conducted regarding whether Smart Grid adoption in some form is beneficial
for the city of Palo Alto and its citizens, as a whole.
40 City of Palo Alto: Utilities Advisory Commission. 2013. Memorandum. Web.
41 California State Legislative Information. Assembly Bill No. 2514. 2009-2010. Web.
42 City of Palo Alto: City Manager. 2014. Do Not Set Energy Storage Procurement Targets. Web.
43 Calmac. 2014. “How Thermal Energy Storage Works.” Web.
Balsdon, Khan, Park 20
One such technology that presents massive benefits for load shifting and optimizing
utility efficiency is the concept of demand response automation, a program that the CPAU has
been investigating and conducting pilots on already. AutoGrid, a startup located in the Bay Area,
has been a partner of the CPAU since 2012, and provides a scalable, fully software approach to
demand response. By gathering petabytes of historic and live data from the CPAU, the startup is
able to implement programs ranging from “power management programs, such as direct load
control, critical peak pricing, peak-time rebates and demand bidding”.44 According to a case
study conducted by AutoGrid in conjunction with CPAU, the “DROMS™” software
implemented was able to “shed an average of 1.2 megawatts of peak demand, saving 3.5
megawatt hours of electricity, per event (complex-event processing is defined in section V.iii.B).”
This may provide savings of up to an estimated $127,000/year at an assumed nominal rate of
$.11/kWh*.45 This figure is difficult to interpret due to a lack of complete information regarding
event duration and frequency, however the savings are clear and should be investigated further. A
request for information for clarification is currently being constructed. Though AutoGrid
provides a compelling case for distribution automation with their software, we believe these
efficiencies extend to production and transmission as well, with massive potential for savings
from the implementation of Artificial Intelligence.
44 Autogrid Systems Inc. “AutoGrid Introduces Big Data Analytics Demand Response Solutions.” RenewGrid.
2012. Web.
45 City of Palo Alto Utilities, Monthly Retail Electric Charges (by Rate Class)." n.d. Web. 7 Nov. 2016
*This estimation was conducted by approximating 1 hr of peak load per day acting as a single event, over the course
of a year, at a single rate set by the CPAU. This estimate may be considered, at most, an order of magnitude
approximation, as events can occur every few seconds, up to every few minutes, and peak load is often assumed to
occur more frequently than 1 hr per day. Hence, the rate of electricity saved per hour is equal to the amount of
electricity saved per day, which is converted to a dollar figure, over 365 days in a year. This approximation was
conducted due to lack of sufficient information from AutoGrid and other sources.
Balsdon, Khan, Park 21
V
Extrapolating [Applicable] Techniques, Innovations in Artificial Intelligence to a Smart
Grid-Centric Renewable-Power Architecture (Palo Alto, Ca)
[Geometric evolution of Artificial Intelligence, its technology and procedures, increases its
utility for a multiplicity of agents across myriad disciplines. AI is a boon to Smart Grid power
paradigms. This paper posits local, regional, and national adoption/appreciation of nascent
Smart Grid power paradigms is contingent upon an algorithm-optimized power generation
system that can autonomously sort and filter mammoth quantities of cloud-based data, harvested
by the grid’s meters, to produce and distribute power more equitably and efficiency.
This section defines Artificial Intelligence, and marshals examples of AI techniques implemented
in the Smart Grid paradigm that resolve inefficient processes.
The foci of this section are Artificial Intelligence and the Smart Grid.]
Keywords: Artificial Intelligence, Smart Grid, Cluster-Analysis, Complex-Event
Processing, Fuzzy Logic, Predictive Analytics, Artificial Neural Networks.
i. Introduction
[The antiquated U.S. Energy Power Plant Infrastructure is a Function of Economic
Malaise and Environmental Spoliation. This section expounds on the state of U.S. energy
infrastructure.]
The U.S. electricity grid is a centrally-planned, unidirectional power production
and distribution model that caters to a heterogeneous consumer class. These consumers,
with disparate aims, intentions, and objectives, preemptively solicit variable quantities of
electricity at variable times for variable durations, 24 hours a day, 365 days a year, ad
infinitum.46 Demand is inelastic and random; “peaker” plants are employed to oversupply
the energy market with electricity to satisfy demand, fully.47 Permanent energy surpluses
is an environmentally hazardous market failure.48 Without predictive analytics that
anticipate demand, the grid must rely on steady-state resources including oil and
petroleum. This model is dependent on the extraction of energy resources derived,
predominantly, from oil and petroleum assets.49 The U.S. nurses a predilection for
captured oil and petroleum, foreign or domestic, because the non-renewable energy assets
46 U.S. Energy Information Administration(EIA). 2016. Energy in Brief: What Is the Electric Power Grid and
What Are Some Challenges It Faces? Web.
47 U.S. DOE. 2016. The Smart-Grid: An Introduction. Web. p. 13 & 14.
48 U.S. DOE. 2016. The Smart-Grid: An Introduction. Web. p. 8 & 9.
49 Department of Homeland Security (DHS). 2015. Energy Sector: Sector Overview. Web.
Balsdon, Khan, Park 22
are abundant, ubiquitous, and cheap.50 Mature technologies (i.e., commodities with
existing infrastructure, supply chains, and quantifiable demand, (e.g., petroleum, gas, and
coal),)—bedrocks of the American energy economy—, collectively, represent 80% of
U.S. energy generation (<81 Quadrillion BTU), while renewable energy systems
currently generate approximately 9 Quad BTU (Appendix A).51 The gross
disproportionality between U.S. energy consumption of conventional fuels and renewable
energy power plants is a signal that the grid is handicapped by obsolescence.
Approximately 65% of gas transmission systems were built prior to the modern epoch, in
the 1950s and 1960s, and do not have the capacity for interoperability with alternative,
renewable generation systems.52 The Quadrennial Energy Review, a U.S. government
sanctioned task force that recommends domestic/federal energy policy, advocates grid
upgrades:
“More than a decade ago, a Department of Energy (DOE) report pronounced the
U.S. electricity grid ‘aging, inefficient, congested, and incapable of meeting the
future energy needs of the information economy without significant operational
changes and substantial public-private capital investment over the next several
decades.’ Although significant improvements have been made to the grid since
then, the basic conclusion of 1-4 QER Report: Energy Transmission, Storage, and
Distribution Infrastructure | April 2015 Chapter I: Introduction the need to
modernize the grid remains valid... The increase in renewable electricity has
changed demands on TS&D infrastructure. Some significant renewable resources
are located far from population centers, and construction of adequate TS&D
infrastructure is key to accessing those resources.”53
Existing energy generation structures are incompatible with renewable energy
technologies.54 The antiquated, 20th century energy grid infrastructure thwarts the massive
deployment of renewable energy systems.55 The Smart Grid is a means of solving
modernization challenges.56 Its compatibility with existing energy infrastructure ensures
that future energy demands of the power, transport, and industrial sectors will be met, and
50 EIA. 2016. Petroleum & Other Liquids: Data. Web.
51 Glover, Ratner. U.S. Energy: Overview and Key Statistics. Congressional Research Service (CRS). 2014. Web.
p. 5-7.
52 Ramchurn, Vytelingum, Rogers, Jennings. Putting the ‘Smarts’ into the Smart Grid: A Grand Challenge for
Artificial Intelligence. Communications of ACM, vol. 55, no. 4, p. 86-97.
53 U.S. DOE: Quadrennial Energy Review (QER). 2014. Energy Transmission, Storage, and Distribution
Infrastructure. Web.
54 U.S. DOE: QER. p. S-14.
55 Glover, Ratner. U.S. Energy: Overview and Key Statistics. CRS. 2014. Web. p. 29.
56 Harley, Liang. Computational Intelligence in Smart Grids. 2011. Web.
Balsdon, Khan, Park 23
net energy intensity will be reduced.57 Bidirectional communication between demand and
supply will provide real-time—“live”—metrics for power plants, improving their
operating efficiency.58 But, the Smart Grid will still face challenges. Incorporating
techniques pioneered in artificial intelligence that address problems with mechanisms in
power system engineering, will enhance Smart Grid processes.59
ii. Artificial Intelligence
[This section thoroughly defines artificial intelligence for the reader. It is imperative the reader
understand the rudimentary mechanics of AI in order to determine its merit for incorporation in
the Smart Grid.]
Artificial Intelligence, (“AI”), is the facility of a machine to think autonomous of
intervention.60 Thinking is the ability to formulate an idea by comprehending, problem-solving,
reasoning, and learning.61 Comprehending is the act of understanding the essence of something.
A machine comprehends if it can give a thorough treatment of data, and parse between what is
relevant and what is irrelevant. Problem-solving is the systematic, sometimes mathematical,
deductive analysis of sundry, unstructured, (i.e., haphazardly ordered), data with the intention of
finding and generating a solution. A machine problem-solves if it can produce y
(output/conclusion) from x (input/initial variables, factors, elements, or information) with the
intention of achieving a goal; namely, a solution. Reasoning is the capacity to contrive a logical
judgement. A machine can reason if it employs a cogent methodology to discriminate data.
Learning is the capability to procure knowledge from experience and instruction. A machine
learns if it evolves; if it augments its comprehension of something by marshalling reason to
justify its problem-solving of said something, and remembers. To display all these qualities is
intelligence. A machine is intelligent if its impetus to comprehend, problem-solve, reason, and
learn, is automatic. It is intelligent if it has an internal locus of control, and knows it. To
articulate:
“A computer program is said to learn from experience ‘E’ with respect to some
class of tasks ‘T’ and performance measure ‘P’, if its performance at tasks in ‘T’,
as measured by ‘P’, improves with experience ‘E’.62
57 U.S. DOE. The Smart Grid: An Introduction. 2016. Web. p. 17.
58 U.S. DOE. The Smart Grid: An Introduction. 2016. Web. p. 17.
59 Ramchurn, Sarvapali et al. Putting the ‘Smarts’ into the Smart Grid: A Grand Challenge for Artificial
Intelligence. Web. p. 86-97.
60 McCarthy, John. What Is Artificial Intelligence? Basic Questions. Stanford University. 2016. Web.
61 McCarthy, John. What Is Artificial Intelligence? Stanford University. 2016. Web.
62 Mitchell, Tom. What Is Machine Learning, and Where is It Headed? 2016. Web.
Balsdon, Khan, Park 24
The goal of the machine is to learn from experience. Experience is derived from real and
artificial simulations in which the machine undergoes an extra-physical metamorphosis, (i.e., a
change in its reasoning procedures), in order to adapt its ability to quantify and qualify data in
proportion to the gravity of the task/ problem stemming from some said given amount of data.
This entails adjusting the initial parameters, (i.e., the metrics in a model needed to issue
predictions), invariably refining the model, (i.e., essentially, an equation generator that makes
predictions on the objective field of data from the input fields), via gradient-descent, (i.e., the
series of iterations involved in a training cycle that are necessary to inform a machine’s ability to
comprehend patterns), to yield a heightened ‘discernment’ for the causal relationship between
bits of data in a dataset.
In sum, AI becomes any mass functioning in a mechanical-physical or extra-physical or
digital matrix, (“environ”), that can harbor, recall, propagate, and conjure data and information,
emulating human neural-chemical processes in the cerebral domain, with a clear, pre-defined,
though not necessarily pre-programmed, goal. This synthetic mimicry of sentience evinced by
and manifest in computers in referred to, generally, as Artificial Intelligence.
Optimization of the U.S. energy grid is conditional. AI techniques that introduce
automated methods to reduce system variability is a requisite for efficiency.
iii. AI Techniques Applicable to the Smart Grid Power Paradigm
[This section presents a plethora of applicable AI techniques for the Smart Grid Power
Paradigm.]
A.
Paritional Cluster Analysis Implementation in Wind-Energy Generation and Distribution Systems
[Cluster-Analysis, (“Clustering”), partitions data into groups or classifications based on initial
input signals and criterion that are meaningful and useful. This technique has function in the
three stages of energy creation, production, and distribution. The purpose of introducing
Cluster-Analysis is to reduce variability throughout the system. Currently, Clustering is being
employed to determine predictive customer segmentation using, “smart metering data focused
on the customer's’ time-based consumption behavior.”63It is also being used to distinguish high-
output wind turbines from low-output wind turbines operating adjacently or within a local wind-
farm.64]
Wind-energy power plants depend on an exogenously generated supply of naturally-
occurring wind. Supply is intermittent; the wind-energy power plant system must cope with
arbitrariness that cannot be omitted from the equation. “First of all, since wind speed fluctuates
63 Flath, Christoph, and Conte, Neumann, Nicolay, Dinther. Cluster Analysis of Smart-Metering Data. Business and
Information Systems Engineering, vol. 4, no. 1. 2012. Web.
64 Ma, Yong, and Jiang, Runolfsson. Cluster Analysis of Wind Turbines of Large Wind Farm. 2009. Web.
Balsdon, Khan, Park 25
sharply from minute to minute, the power output of wind turbines varies fast.”65 Wind-energy
power plants’ energy output is directly proportional to the kinetic intensity and velocity of local
wind currents. The utility of wind-energy power plants is mitigated by wind itself. Devoid of
[artificially] intelligent computer control procedures, the system is inefficient, because its
paradigm does not accommodate for changing wind patterns during high-output periods. AI
innovation remedies this predicament through Partitional Cluster-Analysis (“Clustering”).
Clustering is an algorithm-based, trial-by-error, iterative procedure that parses data in a dataset
for patterns and [relative] similarities, conjures a classification schema, and partitions data into
groups, based on the initial input signals and criteria, that are meaningful and useful. The
premise of introducing Clustering to wind-energy power plants is to reduce variability in the
system’s design and equation model(s). The central parameter of this Clustering model will be
the proximity of high-output turbines to other high-output turbines. Their similarity is measured
by their efficiency and their spatiality:
“The behavior of the total power output of all wind turbines of
entire wind farm is not a simple aggregation of behaviors of
individual turbines. That is, the system output is
partitioned into different groups on the basis of the proximity of individual
dynamics of each group. In order to measure the likelihood of any two turbines
having similar output dynamics, a similarity matrix and the corresponding
Markov transition matrix needs to be constructed.”66
(Keyword: Markov Chain]: Stochastic model that portends the probability of mutually exclusive
events occurring.)
This is a protracted process; clustering requires extensive system training. The effects
will be made incrementally as the algorithm learns which wind-turbine groups to prioritize using
informative metrics that calculate which turbine groups have not met, have met, or have
exceeded their energy output quota. This offers a potent solution to inefficient wind-energy
power plants.
B.
Sorting “Big Data” of the Smart Grid Using Complex-Event Processing
[Complex-Event Processing, (“CEP”), is a failsafe mechanism that derives a criticality rating
from a torrent of data streams—events—accelerated through multiplex sources in a system, in
real-time, and determines, in real-time, whether the events portend an opportunity or a threat,
whether the events are meaningful, useful, insightful, all three, or none (Appendix B).67
65 Ma, Yong et al. Cluster Analysis of Wind Turbines of Large Wind Farm. 2009. Web.
66 Ma, Yong, et al. Cluster Analysis of Wind Turbines of Large Wind Farm. 2009. Web.
67 Schmerken, Ivy. “Deciphering The Myths Around Complex Event Processing.” 2008. Web.
Balsdon, Khan, Park 26
Currently, CEP tools are being utilized to sort and filter swaths of data generated from Smart
Grid’s Advanced Metering Infrastructure, (“AMI”).68 ]
AMI is the data bridge in a Smart Grid-utility company feedback loop.69 AMI analytics
allows the utility company to effectively regulate surpluses by enhancing their prediction models
for energy demand.70 The opportunity cost of using analytics is data volume. Large amounts of
data can be deleterious.71 CEP resolves complexities in the system by monitoring data streams,
inferring patterns that signal a coordinated event is taking place (e.g., weather fluctuation, price-
change deviation, so forth):
“Instead, in complex event processing, information is computed in real-time,
and pushed to a user (or an agent) rather than being pulled. CEP is a set of
technologies and practices, which enable users to receive information as soon as it
is published (rather than requiring periodic updates). Hence, there is no need to
pull information, it will be delivered to users nearly at the moment it is
published.”72
CEP data management involves static queries supervising large data streams, and fast-tracking
datasets that communicate complex events in order to curb system inefficiencies in the short-
term and the long-run. CEP reduces friction in the system by prioritizing data based on its
criticality rating, requiring less human supervision, intervention and regulation.
C.
Utilizing Fuzzy Logic in Smart Grid Decision-Modeling and Frequency Control
[Fuzzy Logic, (“FL”), is a precision-mapping method that assumes truth is continuous for
values between the interval [0,1].73 0 (lower limit) is logic that is false; 1 (upper limit) is logic
that is true; range 0-1 is logic that is partially true (i.e., truthful to a degree).74 Logic is seldom
an ideal Cartesian product (i.e., binary: 0 or 1), rather, logic is partially ideal (i.e., ideal to a
degree).75 FL is a computer schema that assigns membership, along the interval [0,1], to a data-
proposition based on its syntax, proof systems, and semantics.76 In sum, “FL provides a simple
way to arrive at a definite conclusion based upon vague, ambiguous, imprecise, noisy, or missing
68 Liu, Guangyi, and Zhu, Saunders, Gao, Yu. Real-time Complex Event Processing and Analytics for Smart Grid.
2015. Procedia Computer Science. vol. 61. p. 113-119. Web.
69 Electric Power Research Institute, Inc. Advanced Metering Infrastructure. 2007. Web.
70 Electric Power Research Institute, Inc. Advanced Metering Infrastructure. 2007. Web.
71 Liu et al. Real-time Complex Event Processing and Analytics for Smart Grid. 2015. p. 113-119. Web.
72 Wagner, Andreas, and Anicic, Stuhmer, Stognovic, Harth, Studer. Linked Data and Complex Event Processing
for the Smart Energy Grid. 2014. Web.
73 Cintula, Petr, and Fermuller, Noguera. Stanford Encyclopedia of Philosophy. Fuzzy Logic. 2016. Web.
74 Ibid.
75 Ibid.
Balsdon, Khan, Park 27
input information.”77 Currently, FL is being employed as an intelligent problem-solving
controller scheme in the Smart Grid generation and storage system.78]
Existing energy infrastructure integrates multifarious fuels and technologies to generate
electricity.79 The operational efficiency of each fuel and/or technology is variable.80 Conventional
controllers, or power system stabilizers, (“PSS”), regulate excitation in automatic voltage
regulators, (“AVR”), devices that attempt to stabilize oscillation of output voltage created in a
power generator, to within the system’s parameters.81,82 Research conducted by Yaser Qudaih et
al. postulates conventional controllers are inefficient.83 Yaser et al. used Fuzzy Logic in a remote
power distribution system, generating power from a diesel engine and PV power plant (see
Appendix B).
“The control strategy is basically implemented to approach two objectives. One is
to keep the storage system within the permissible limits and to reduce the
generation of the diesel unit to minimum as the other. Diesel unit will receive the
signal via communication network which carries out all the required data from the
fuzzy logic control scheme prepared in one of the available computers in the
network. Once the diesel unit receives the signal from the fuzzy controller the
additional signal will help in the governor action to support the batteries which
will operate in their optimal operation due to the fuzzy controller algorithm as
shown in Figure 3. Making the diesel work in the same loop with the storage
system result in reducing the dependency on the diesel as a fossil fuel and allows
the system to work more stable when the storage system absorbs the fluctuations
caused by the PV generators.”84
76 Ibid.
77 Qudaih, Yaser Soliman, and Ali, Mitani. Journal of Power and Energy Engineering. Microgrid Design Including
Fuzzy Logic Controlled Storage System. 2014. Web.
78 Ibid.
79 Glover Carol, Michael Ratner. U.S. Energy: Overview and Key Statistics. 2014. Congressional Research Service
(CRS). 2014.
80 U.S. DOE. Smartgrid.gov. Energy Storage. 2014. Web.
81 Mitsubishi Electric Corporation. Power System Stabilizer. 2010. Web.
82 Electrical Engineering Centre. Basics of Automatic Voltage Regulators. 2011. Web.
83 Qudaih, Yaser Soliman, and Ali, Mitani. Journal of Power and Energy Engineering. Microgrid Design Including
Fuzzy Logic Controlled Storage System. 2014. Web.
84 Ibid.
Balsdon, Khan, Park 28
.
Fuzzy Logic mitigates the usage of traditional fuels, (e.g., diesel), by improving the
performance and reliability of unconventional, renewable technologies.
D.
Applying Artificial Neural Networks to the Smart Grid
[Artificial Neural Networks, (“ANN”), are non-deterministic, massively parallel
processors replicating mammalian neuro-chemical processes in a tiered network of
interconnected chains of weighted simple-calculation units (or nodes, neurons) that
(synaptically) fire an input across a threshold to precipitate an output.85,86ANN are a biologically
inspired, silicon imitation of the original function approximator, the human prefrontal cortex.87
ANN can be arithmetically expressed by this equation: The output vector is a function of the
input vector, weight vector, and threshold vector.88 ANN are self-supervising expert approximator
systems that extract patterns from complex datasets.89 Currently, ANN is being used in load
forecasting of the Smart Grid.90]
Weather intermittence influences electricity demand load and inhibits energy prices from
stabilizing.91 Conventional, econometric and statistical-based modeling of future-short and
future-long electricity prices makes prognostications using agglomerated historical data, and,
“estimates the relationships between energy consumption (dependent variables) and factors
influencing consumption.”92 This method is protracted, intensive, and byzantine:
85 Wang, Sun-Chong. Interdisciplinary Computing in Java Programming. Artificial Neural Network. 2003. Web.
86 Stergiou, Christos and Dimitrios Siganos. Neural Networks: Why Use Neural Networks? 2016. Web.
87 Wang, Sun-Chong. Interdisciplinary Computing in Java Programming. Artificial Neural Network. 2003. Web.
88 Ibid.
89 Ibid.
90 Zhang, Hao-Tian, and Fang-Yuan Xu, Long Zhu. Ninth International Conference on Machine Learning and
Cybernetics, Qingdao. Artificial Neural Network for Load Forecasting in Smart Grid. 2010. Web.
91 Feinberg, Eugene A., and Dora Genethliou. Applied Mathematics for Restructured Electric Power Systems.
Load Forecasting. 2005. Web.
92 Ibid. p. 273-274.
Balsdon, Khan, Park 29
“The end-use and econometric methods require a large amount of information
relevant to appliances, customers, economics, etc. Their application is
complicated and requires human participation. In addition, such information is
often not available regarding particular customers and a utility keeps and supports
a pro-file of an “average” customer or average customers for different type of
customers.”93
Weather is volatile and subject to intense flux. Weather exercises unfettered hegemony
over the acquisition, transmission and distribution of energy captured by renewable power plants.
For example, wind power production is beholden to the vagaries of wind’s wake turbulence.
Photovoltaic power is predicated on the presence of sunlight and the absence of cloud
layers/darkness/ any natural or man-made opposition/barrier to capturing sunlight. Tidal power is
reliant on the kinetic energy of the ocean currents. Because weather is variable, mechanisms for
forecasting demand and grid load cannot employ a comprehensive rubric for determining
whether power harvested or captured from these resources is liable to underperform their
theoretical or actual output quota in future-short and/or future-long.
ANN reduce the error function between the input vector and the output vector using
synaptic weights (i.e., strength between simple-calculation unit (or nodes, neurons)), improving
upon a conventional forecasting model that generalizes and stereotypes consumer-side, end-use
data to justify its implicit paucity of real information.94 The ANN model circumvents this
predicament by constantly training its nodes on the data being forecasted, unsupervised and in
real-time, autonomously adjusting its computation processes in accordance.95 ANN is not static;
ANN is a dynamic approximator function that significantly improves prediction performance in
forecast models.
93 Ibid. p. 274.
94 Filik, Ummuhan Basaran, and Mehmet Kurban. A New Approach for the Short-Term Load Forecasting with
Autoregressive and Artificial Neural Network Models. 2007. Web.
95 Ibid.
Balsdon, Khan, Park 30
VI
Conclusion
Throughout this paper, we have examined the purpose and properties of Smart Grids, and
their potential benefits. By identifying a number of potential adopters of these overarching
technologies, we recognize that specific technologies within the realm of those identified may be
better suited for specific locations. Furthermore, we acknowledge that a number of utilities have
a higher potential for adoption than others, leading to the conclusion that CPAU is the ideal
candidate for the technologies proposed. This is reaffirmed due to a number of identified factors
including Palo Alto’s progressive energy policies. With this in mind, the case for Palo Alto was
further analyzed, as well as the specific technologies that Palo Alto has already considered, and
others that may require further examination. Advances in software have led to automation
becoming increasingly pertinent, leading us to research and define the framework for a grid that
is fully optimized through Artificial Intelligence.
AI upgrades the Smart Grid by utilizing automation to simplify the system in each of the
three stages of energy creation, production, and distribution. AI techniques explored in this paper
including, Cluster Analysis, Complex-Event Processing, Fuzzy Logic, Artificial Neural
Networks, and their concomitant sub-methods/-techniques, represent the totality of AI
investment in the Smart Grid. The impetus for introducing AI technology to the Smart Grid was
to reduce or omit variability throughout the system. Clustering, CEP, FL, and ANN are testament
of how advanced, unconventional techniques can improve system performance, prediction, and
energy power potential.
Looking ahead, it is evident that the general case for Smart Grid adoption, by individual
utility companies, is the necessary future for our outdated and inefficient grid. The potential
savings, economically as well as environmentally, are a testament to the necessity of this
transition, with AI software at the forefront of this optimization. Every city, and in fact every
neighborhood and household, is unique; as such a blanket approach to energy (creation,
production, distribution) can no longer be considered efficient using existing infrastructure.
While individual aspects of the SGDP that were analyzed including energy storage, AMI
hardware and software upgrades may have higher potential for specific areas, we believe the
approach taken in this paper of identifying technologies and locations is general enough to be
applied for a number of potential adopters. Smart Grid technology, as well as existing
infrastructure, should continually be analyzed critically, with the intention of achieving
maximum grid efficiency, for the benefit of utility companies, their customers, and in fact our
environment and energy infrastructure as a whole.
Balsdon, Khan, Park 31
Letter to Palo Alto
Dear Mr. Shikada and Mr. Cook:
We are undergraduate students at the University of California, Berkeley. In our class,
Letters and Science 126, led by professors James Rector and Christine Rosen, we were tasked
with writing a proposal in the context of energy, civilization, and climate change. We received
your information from Vice Mayor Scharff upon inquiry on whom to submit our proposal
regarding the City of Palo Alto's energy initiatives. In our research, we investigated the
economics and viability of smart-grid technology, and discovered that your municipality has
invested heavily into the technology, performed a few pilot programs, and contracted Autogrid
Inc. to transform Palo Alto’s electricity distribution system. We are pleased to inform you that
our research supports your decision to adopt smart-grid technology.
We understand that Palo Alto prides itself in leading the state in energy efficiency, carbon
neutrality, and environmental protection. From our research, we believe that Palo Alto’s choice
for adopting a smart-grid allows you and your city to continue to set an example for the country,
benefit financially, and ease the transition into a carbon-free energy system.
For your reference, in this letter we have enclosed our proposal. In it you will find
validation of your municipality’s investments and actions, an explanation as to why your city is
most likely the best initial adopter of smart-grid technology, and an examination as to what
technologies a “smart-grid” entails. We hope that you find our proposal and research useful, and
we hope in the coming years you and your government will encourage the rest of California to
decarbonize and transform their electric grid.
On behalf of UC students and citizens of California, we would like to thank you for
taking these important steps to move away from fossil fuels. We will be sending our proposal to
other municipalities that control utilities which have yet to adopt grid technologies to the scale of
Palo Alto.
Harrison Balsdon
Aman Khan
Matt Park
Balsdon, Khan, Park 32
Letter to Other Cities
Dear Mayor _____ and Members of the City Council of ______:
We are undergraduate students at the University of California, Berkeley. In our class,
Letters and Science 126, led by professors James Rector and Christine Rosen, we were tasked
with writing a proposal in the context of energy, civilization, and climate change. In our research,
we investigated the economics and viability of smart-grid technology, and discovered that the
city of Palo Alto has invested heavily into the technology, performed a few pilot programs, and
have contracted Autogrid Inc. to transform their electricity distribution system. We would like to
inform you that our research supports their decision to adopt smart-grid technology, and we
encourage you to follow a similar path.
The reasons for transforming your electric grid are numerous, including, but not limited
to, easing the transition to renewable energy and lowering the cost(s) of distribution. Enclosed is
our proposal and research, in which we explore these topics much deeper. In it you will find
validation of Palo Alto’s investments and actions, an explanation as to why your city and your
citizens would benefit from smart-grid adoption, and a holistic examination as to what
technologies a “smart-grid” entails. We hope that you find our proposal and research useful, and
ultimately consider investing in the technologies we describe. Thank you.
Harrison Balsdon
Aman Khan
Matt Park
To be sent to:
Lisa Gillmor, Santa Clara
Ed Lee, San Francisco
Paul Philips, City of Industry
Terry Tornek, Pasadena
Kevin Johnson, Sacramento
Balsdon, Khan, Park 33
Appendix A
Balsdon, Khan, Park 34
Appendix B
Balsdon, Khan, Park 35
Appendix C
Balsdon, Khan, Park 36
Appendix D: City Selection Matrix
Balsdon, Khan, Park 38
Appendix E
Balsdon, Khan, Park 39
Works Cited
[Due to the number of references in this paper, for ease of use, we have altered the format of the
bibliography from an alphabetical arrangement to a chronological arrangement (i.e., as the
sources appear in the text.]
—. @SGCONSUMER. "Smart Grid: Where Power Is Going." Smart Grid: Where Power Is
Going. Smart Grid Consumer Collaborative, n.d. Web. 06 Dec. 2016.
<http://www.whatissmartgrid.org/>.
United States. Department of Energy. Electricity Delivery and Energy Reliability. 2014 Smart
Grid System Report. Washington, DC: Department of Energy, 2014. Web. p. 1-35.
<http://energy.gov/sites/prod/files/2014/08/f18/SmartGrid-SystemReport2014.pdf>.
President, BCG Senior Vice. "Smart Grid Expenditures Outlook in U.S. by Project 2017 |
Statistic." Statista. Statista Inc., n.d. Web.
<http://www.statista.com/statistics/504297/outlook-for-smart-grid-spending-in-the-us-by-
project/>.
United States. Department of Energy. U.S. Energy Information Administration. Electric Power
Monthly with Data for September 2016. Washington, DC: Department of Energy, 2016.
Web. <http://www.eia.gov/todayinenergy/detail.php?id=24852>.
United States. Department of Energy. Energy Efficiency & Renewable Energy. Revolution Now
2016. Washington, DC: Department of Energy, 2016. Web.
<http://energy.gov/sites/prod/files/2016/09/f33/Revolution%20Now-%20One
%20Pager_1.pdf>.
Randall, Tom. "Wind and Solar Are Crushing Fossil Fuels." Bloomberg.com. Bloomberg, 6 Apr.
2016. Web. <http://www.bloomberg.com/news/articles/2016-04-06/wind-and-solar-are-
crushing-fossil-fuels>.
United States. California Energy Commission. Renewable Energy Information. California
Energy Commission-Tracking Progress. By Energy Commission Staff. Sacramento: CEC,
2016. Web.
<http://www.energy.ca.gov/renewables/tracking_progress/documents/renewable.pdf>.
Randall, Tom. "Tesla Wins Massive Contract to Help Power the California Grid."
Bloomberg.com. Bloomberg, 15 Sept. 2016. Web. 2016.
<https://www.bloomberg.com/news/articles/2016-09-15/tesla-wins-utility-contract-to-
supply-grid-scale-battery-storage-after-porter-ranch-gas-leak>.
—. "Lazard Levelized Cost of Storage Analysis-1.0." (2015): 1-30. Lazard Ltd. Web.
<https://www.lazard.com/media/2391/lazards-levelized-cost-of-storage-analysis-10.pdf>.
Balsdon, Khan, Park 40
"Fast Facts." (2016): 1-4. California Independent System Operator. Web.
<https://www.caiso.com/Documents/FlexibleResourcesHelpRenewables_FastFacts.pdf>.
—. "List of Power Stations in California." Wikipedia. Wikimedia Foundation, n.d. Web.
<https://en.wikipedia.org/wiki/List_of_power_stations_in_California>.
United States. California Energy Commission. Integrated Energy Policy Report Committee.
Integrated Energy Policy Report Update 2006. By Pamela Doughman, Melissa Jones, Bill
Knox, Suzanne Korosec, Suzanne Phinney, and Heather Rait. Sacramento: CEC, 2007.
Web. http://www.energy.ca.gov/2006publications/CEC-100-2006-001/CEC-100-2006-001-
CMF.PDF
—. "Crude Oil Prices - 70 Year Historical Chart." MacroTrends. Macrotrends LLC, n.d. Web.
<http://www.macrotrends.net/1369/crude-oil-price-history-chart>.
Courtland, Rachel. "Gordon Moore: The Man Whose Name Means Progress." IEEE Spectrum:
Technology, Engineering, and Science News. IEEE.org, 30 Mar. 2015. Web.
<http://spectrum.ieee.org/computing/hardware/gordon-moore-the-man-whose-name-
means-progress>.
Hoffmann, Volker. "Farmers and Researchers: How Can Collaborative Advantages Be Created In
Participatory Research and Technology Development?" Agriculture and Human Values
24.3 (2007): 355-68. ResearchGate.net. Web.
<https://www.researchgate.net/profile/Anja_Christinck/publication/225616414_Farmers_a
nd_researchers_How_can_collaborative_advantages_be_created_in_participatory_research
_and_technology_development/links/00b4953a92931a6fae000000.pdf#page=37>.
McMahon, Jeff. "Hawaii Sitting On The Lid Of A Solar Explosion." Forbes. Forbes Magazine, 2
Oct. 2016. Web. <http://www.forbes.com/sites/jeffmcmahon/2016/10/02/hawaii-sitting-on-
the-lid-of-a-solar-explosion/#67593837df06>.
United States. Department of Energy. Electricity Delivery and Energy Reliability. Renovating the
Grid and Revitalizing a Neighborhood: Successes from Kansas City Power & Light's Smart
Grid Demonstration Project. Washington, DC: Department of Energy, 2015. Web.
John, Chris. "Silicon Valley Continues Strong Lead in Smart Grid Jobs." PGE Currents. PGE
Corporation, 13 Dec. 2013. Web. <http://www.pgecurrents.com/2013/12/18/silicon-valley-
continues-strong-lead-in-smart-grid-jobs/>.
—. "Energy and Utilities: Case Studies." IBM Smart Grid Case Studies - Energy and Utilities
Case Studies. IBM Industries, n.d. Web. <http://www-
935.ibm.com/industries/ae/energy/case_studies.html>.
Conca, James. "'Demand Response' Is How The Smart Grid Will Save Us Billions." Forbes.
Forbes Magazine, 24 Feb. 2015. Web. <http://www.forbes.com/forbes/welcome/?
toURL=http%3A%2F%2Fwww.forbes.com%2Fsites%2Fjamesconca
Balsdon, Khan, Park 41
%2F2015%2F02%2F24%2Fsolving-americas-energy-future-requires-a-demand-response
%2F>.
Ng, Howard. "Does AMI Have What It Takes for Demand Response? | Comverge Blog."
Comverge Blog. Comverge Blog, 20 Nov. 2013. Web.
<http://blog.comverge.com/intelligent-energy-management/does-ami-have-what-it-takes-
for-demand-response/>.
United States. Department of Energy. Electricity Delivery and Energy Reliability. Improving
Efficiency with Dynamic Line Ratings: Successes from New York Power Authority’s Smart
Grid Demonstration Project. Washington, DC: Department of Energy, 2014. Web.
https://www.smartgrid.gov/document/NYPA_Improving-Efficiency-Dynamic-Line-
Ratings.html
United States. Department of Energy. Electricity Delivery and Energy Reliability. Voltage and
Power Optimization Saves Energy and Reduces Peak Power. Washington, DC: Department
of Energy, 2015. Web. https://www.smartgrid.gov/document/Voltage-Power-Optimization-
Saves-Energy-Reduces-Peak-Power.html
United States. Department of Energy. National Renewable Energy Laboratory. Overgeneration
from Solar Energy in California: A Field Guide to the Duck Chart. By Paul Denholm,
Matthew O'Connell, Gregory Brinkman, and Jennie Jorgenson. Washington, DC:
Department of Energy, 2015. Web. Task No. TM13.5020.
http://www.nrel.gov/docs/fy16osti/65023.pdf
—. "Energy Management for Smart Grid, Cities and Buildings: Opportunities for Battery
Electricity Storage Solutions." I-micronews. Yole Developpement, n.d. Web.
<http://www.i-micronews.com/power-electronics-report/product/energy-management-for-
smart-grid-cities-and-buildings-opportunities-for-battery-electricity-storage-
solutions.html>.
Golson, Jordan. "Tesla Is Building an 80MWh Battery Pack to Supply LA with Power." The
Verge. Vox Media, Inc., 15 Sept. 2016. Web.
<http://www.theverge.com/2016/9/15/12928390/tesla-utility-battery-pack-powerpack-
california
John, Jeff St. "California Passes Huge Grid Energy Storage Mandate." Greentech Media. N.p., 17
Oct. 2013. Web. <https://www.greentechmedia.com/articles/read/california-passes-huge-
grid-energy-storage-mandate>.
Technical Report (Final) Smart Grid Demonstration Program: Flywheel Energy Storage
Demonstration. Tech. no. DE-OE0000232. N.p.: Amber Kinetics, 2015. Web.
https://www.smartgrid.gov/files/Amber_Kinetics_Final_Technical_Report.pdf
Berry, Gene D. "Present and Future Electricity for Intermittent Renewables." The 10-50
Solution: Technologies and Policies for a Low Carbon Future. Proc. of The Pew Center on
Balsdon, Khan, Park 42
Global Climate Change and the National Commission on Energy Policy, Virginia,
Arlington. N.p.: n.p., 2004. N. pag. Web. http://www.c2es.org/docUploads/10-50_Berry.pdf
United States. California Energy Commission. Energy Assessments Division. Differences
Between Publicly and Investor-Owned Utilities. Sacramento: CEC, 2016. Web.
http://www.energy.ca.gov/pou_reporting/background/difference_pou_iou.html
United States. California State Legislature. Senate Local Government Committee. Governments
Working Together: A Citizen's Guide to Joint Powers Agreements. By Trish Cypher and
Colin Grinnell. Sacramento: Senate Publications, 2007. Web.
http://www.calafco.org/docs/Senate_LG_JPA_Report.pdf
—. "What Is a Cooperative?" What Is a Cooperative? | MJM Electric Cooperative. Touchstone
Energy Cooperatives, 2015. Web. <https://www.mjmec.coop/content/what-cooperative>.
"PWRPA - Power and Water Resources Pooling Authority." PWRPA - Power and Water
Resources Pooling Authority. PWRPA, n.d. Web. <http://www.pwrpa.org/pwrpa
%5Forg/index.CFM?q_webaction=ABOUTUS>.
"Electric Renewable Resources." Electric Renewable Resources - City of Palo Alto. Palo Alto
City Hall (Powered by Civica), 16 Apr. 2016. Web.
<http://www.cityofpaloalto.org/gov/depts/utl/residents/resources/pcm/default.asp>.
—. "Top 101 Cities with the Most People Having Master's or Doctorate Degrees (population
50,000+)." Top 101 Cities with the Most People Having Master's or Doctorate Degrees
(population 50,000+). Advameg, Inc. (Powered by CartoDB, OpenStreetMap), 2016. Web.
<http://www.city-data.com/top2/h182.html>.
Dunn, Jeff. "Here's How Expensive It Is to Live in the Heart of Silicon Valley." Business Insider.
Business Insider, 01 Sept. 2016. Web. <http://www.businessinsider.com/silicon-valley-
palo-alto-2016-9>.
"NCPA | Northern California Power Agency." NCPA. NCPA, 2016. Web.
<http://www.ncpa.com/>.
United States. City of Palo Alto: City Manager. Utilities Advisory Committee. Assessment of
Smart Grid Applications. Palo Alto: City Council Staff, 2012. Web.
http://www.cityofpaloalto.org/civicax/filebank/documents/41569
United States. Utilities Department. Utilities Advisory Commission. Memorandum. Palo Alto:
Utilities Department, 2013. Web.
https://www.cityofpaloalto.org/civicax/filebank/documents/37995
Assembly Bill No. 2514, California State Legislative Information § 469-9620 (California State
Legislature 2010). Web. http://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?
bill_id=200920100AB2514
Balsdon, Khan, Park 43
United States. City of Palo Alto. City Manager. Do Not Set Energy Storage Procurement Targets.
Palo Alto: City Council Staff, 2014. Web.
http://www.energy.ca.gov/assessments/ab2514_reports/City_of_Palo_Alto/2014_Palo_Alto
_report.pdf
—."How Thermal Energy Storage Works." What Is Energy Storage and How Does Thermal
Energy Storage Work? CALMAC, 2014. Web. <http://www.calmac.com/how-energy-
storage-works>.
—. "AutoGrid Introduces Big Data Analytics Demand Response Solutions (Renew Grid)."
AutoGrid Introduces Big Data Analytics Demand Response Solutions (Renew Grid).
RenewGrid, 29 Oct. 2012. Web. <http://www.auto-grid.com/wp/news/autogrid-introduces-
big-data-analytics-demand-response-solutions-renew-grid-2/>.
"U.S. Energy Information Administration - EIA - Independent Statistics and Analysis." EIA's
Energy in Brief: What Is the Electric Power Grid and What Are Some Challenges It Faces?
U.S. Energy Information Administration, n.d. Web. 06 Dec. 2016.
<http://www.eia.gov/energy_in_brief/article/power_grid.cfm>.
"The Smart Grid: An Introduction." The Smart Grid: An Introduction | Department of Energy.
Litos Strategic Communication, n.d. Web. 2016. <http://energy.gov/oe/downloads/smart-
grid-introduction-0>.
United States. Department of Homeland Security. Infrastructure Protection. Energy Sector-
Specific Plan 2015. Washington, DC: DHS, 2015.
Web.https://www.dhs.gov/sites/default/files/publications/nipp-ssp-energy-2015-508.pdf
"Petroleum and Other Liquids Data." Energy Information Administration. EIA, 30 Nov. 2016.
Web. <http://www.eia.gov/petroleum/data.cfm>.
United States. Library of Congress. Congressional Research Service. U.S. Energy: Overview and
Key Statistics. By Carol Glover and Michael Ratner. Washington, DC: CRS, 2014. Web.
https://fas.org/sgp/crs/misc/R40187.pdf
Ramchurn, Sarvapali D., Perukrishnen Vytelingum, Alex Rogers, and Nicholas R. Jennings.
"Putting the 'smarts' into the Smart Grid." Communications of the ACM 55.4 (2012): 86.
Web. <http://cacm.acm.org/magazines/2012/4/147362-putting-the-smarts-into-the-smart-
grid/fulltext>.
—. "Quadrennial Energy Review: Energy Transmission, Storage, and Distribution
Infrastructure." (2015): 1-384. U.S. Department of Energy. Web.
<http://energy.gov/sites/prod/files/2015/04/f22/QER-ALL%20FINAL_0.pdf>.
Harley, Ronald G., and Jiaqi Liang. "Role of Proactive Behaviour Enabled by Advanced
Computational Intelligence and ICT in Smart Energy Grids." Proceedings of the 2nd
International Conference on Smart
Balsdon, Khan, Park 44
McCarthy, John. "WHAT IS ARTIFICIAL INTELLIGENCE?" WHAT IS ARTIFICIAL
INTELLIGENCE? N.p., 12 Nov. 2007. Web. 01 Dec. 2016. <http://www-
formal.stanford.edu/jmc/whatisai/>.
Mitchell, Tom. "Tom Mitchell." Tom Mitchell's Home Page. Tom Mitchell, n.d. Web.
<http://www.cs.cmu.edu/~tom/>.
Flath, Christoph, David Nicolay, Tobia Conte, and Lilia Filipova-Neumann. "Cluster Analysis of
Smart Metering Data." ResearchGate. Business & Information Systems Engineering, Feb.
2012. Web.
<https://www.researchgate.net/publication/257790595_Cluster_Analysis_of_Smart_Meteri
ng_Data>.
Ma, Yong, John N. Jiang, and Thordur Runolfsson. "Cluster Analysis of Wind Turbines of Large
Wind Farm." 2009 IEEE/PES Power Systems Conference and Exposition (2009): n. pag.
IEEE XPLORE. Web. 5 Dec. 2016. <http://ieeexplore.ieee.org/stamp/stamp.jsp?
arnumber=4840148>.
Schmerken, Ivy. "Deciphering the Myths Around Complex Event Processing - Wall Street &
Technology." Wall Street & Technology. UBM, 15 May 2008. Web.
<http://www.wallstreetandtech.com/latency/deciphering-the-myths-around-complex-event-
processing/d/d-id/1259489>.
Liu, Guangyi, Wendong Zhu, Chris Saunders, Feng Gao, and Yang Yu. "Real-time Complex
Event Processing and Analytics for Smart Grid." Procedia Computer Science 61 (2015):
113-19. Science Direct. Web.
<http://www.sciencedirect.com/science/article/pii/S1877050915029993>.
—."Demand-Response and the Advanced Metering Infrastructure." Smart Grid (2014): 235-58.
Electric Power Research Institute. EPRI, Feb. 2007. Web.
<https://www.ferc.gov/CalendarFiles/20070423091846-EPRI%20-%20Advanced
%20Metering.pdf>.
Wagner, Andreas, Darko Anicic, Roland Stuhmer, and Andreas Harth. "Linked Data and
Complex-Event Processing for the Smart Energy Grid." ResearchGate. N.p., n.d. Web.
<https://www.researchgate.net/publication/267230352_Linked_Data_and_Complex_Event
_Processing_for_the_Smart_Energy_Grid>.
Cintula, Petr, Fermüller, Christian G. and Noguera, Carles, "Fuzzy Logic", The Stanford
Encyclopedia of Philosophy (Winter 2016 Edition), Edward N. Zalta (ed.), forthcoming
URL = <https://plato.stanford.edu/archives/win2016/entries/logic-fuzzy/>.
Qudaih, Yaser Soliman, Raheel Ali, and Yasurnori Mitani. "Microgrid Design Including Fuzzy
Logic Controlled Storage System." Journal of Power and Energy Engineering. Microgrid,
Apr. 2014. Web. <http://file.scirp.org/pdf/JPEE_2014042114353990.pdf>.
Balsdon, Khan, Park 45
—. "Energy Storage Demonstration Projects." Energy Storage Demonstration Projects. U.S.
Department of Energy, n.d. Web. 05 Dec. 2016.
<https://www.smartgrid.gov/recovery_act/program_impacts/energy_storage_technology_pe
rformance_reports.html>.
—."Mitsubishi Electric Power System Stabilizer." Mitsubishi Corp. (2009): 1-7. Mitsubishi
Corporation. Mitsubishi Coproration, Sept. 2010. Web.
<http://www.meppi.com/Products/GeneratorExcitationProducts/Static%20Excitation
%20System/Power%20System%20Stabilizer.pdf>.
Lemau. "Automatic Voltage Regulator." Blog post. Electrical Engineering Centre. EEC, 13 Aug.
2011. Web. <http://www.electricneutron.com/generator-2/basic-of-automatic-voltage-
regulator/>.
Wang, Sun-Chong. "Artificial Neural Network." ResearchGate. The Springer International Series
in Engineering and Computer Science, Jan. 2003. Web.
<https://www.researchgate.net/publication/289636954_Artificial_Neural_Network>.
Stergiou, Christos, and Dimitrios Siganos. "Neural Networks." Neural Networks. N.p., n.d. Web.
05 Dec. 2016.
<https://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/cs11/report.html#The Back-
Propagation Algorithm>.
Zhang, Hao-Tian, Fang-Yuan Xu, and Long Zhu. "Artificial Neural Network for Load
Forecasting in Smart Grid." IEEE Xplore. Ninth International Conference on Machine
Learning and Cybernetics, Qingdao, 14 July 2010. Web.
<http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5580713>.
Feinberg, Eugene A., and Dora Genethliou. "Load Forecasting." SpringerLink. Applied
Mathematics for Restructured Electric Power Systems, 2005. Web.
<http://link.springer.com/chapter/10.1007%2F0-387-23471-3_12>.
Filik, Ummuhan Basaran, and Mehmet Kurban. "A New Approach for the Short-Term Load
Forecasting with Autoregressive and Artificial Neural Network Models." ResearchGate.
N.p., Jan. 2007. Web.
<https://www.researchgate.net/publication/228625376_A_New_Approach_for_the_Short-
Term_Load_Forecasting_with_Autoregressive_and_Artificial_Neural_Network_Models>.
Chang, C. S., Zhaoxia Wang, Fan Yang, and W. W. Tan. "Hierarchical Fuzzy Logic System for
Implementing Maintenance Schedules of Offshore Power Systems." IEEE Transactions on
Smart Grid 3.1 (2012): 3-11. IEEE XPLORE. Web.
<http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6112195>.
—. Grids and Green IT Systems (2013): n. pag. Web. 5 Dec. 2016.
<http://www.brain2grid.org/documents/mypaper/EFRI_publication_1298568050.pdf>.
Balsdon, Khan, Park 46
Hiyama, Takashi. "Fuzzy Logic Power System Stabilizer Using Polar Information." Electric
Power Applications of Fuzzy Systems (2009): 1-29. IEEE XPLORE. Web.
<http://ieeexplore.ieee.org/xpl/ebooks/bookPdfWithBanner.jsp?
fileName=5266280.pdf&bkn=5266264&pdfType=chapter>.
Santofimia-Romero, Maria-Jose, Xavier Del Toro-Garcia, and Juan-Carlos Lopez-Lopez.
"Artificial Intelligence Techniques for Smart Grid Applications." Green ICT: Trends and
Challenges. CEPIS UPGRADE, Oct. 2011. Web. 5 Dec. 2016. <http://www.bing.com/cr?
IG=12E0BED329DD4753A2F36E1E786527EB&CID=0A36B031E437637C0277B9D2E5
0662FF&rd=1&h=nSOIA03hKDTka_Uo4Qcx-l3O0V_tG2ypTgkGwSmr04o&v=1&r=http
%3a%2f%2farco.esi.uclm.es%2fpublic%2fpapers%2f2011-
UPGRADE.pdf&p=DevEx,5084.1>.