I’m an entrepreneurial electrical engineer

(iEEE): from academia to startupsMads R. Almassalkhi

Department of Electrical & Biomedical Engineering (EBE)

University of Vermont (UVM)

IEEE PES Young Professionals

April 27th, 2019

Overview

1

2008 2013 2014

BSEE + Math3 years in image processing,

classification, detection

PhD in EE:SControls, Power Systems, and Optimization

Thesis (w/ Ian Hiskens): Contingency mitigation in bulk power systems

Startup “Post-doc”$1M VC-funded energy

optimization SaaS company

Developed optimal energy

plant dispatch tech and

software for customers with

$40M+ energy budgets

University of VermontResearch: grid optimization with feedback

Lead $3M DOE/ARPA-E NODES project

Co-founder of energy tech startup

Lead $2.M SETO ENERGISE project

2016

First lesson learned

2

You can do more than

what you already know

EE as an undergraduate/co-op

Dual major in mathematics (applied math)

Took too many courses and tinkered too little

>300 credit hours at graduation, needed only 186 hours.

Should have taken more programming courses → software is critical!

First Co-operative Education program in U.S. (1906)

Worked for 3 years (50% full time / 50% part-time) in small dynamic, R&D firm: Etegent Tech.

Focused on algorithm development for signal/image processing (Medical/DoD applications)

Learned image processing, Matlab programming, pattern detection and classification (i.e., ML)

3

Find that which

is hidden

Next lesson

4

Change is constant,

so learn to adapt

EE as a grad student

PhD student in EE: Systems (control theory focus)

5-year funding guarantee with a 2-year fellowship: gave me freedom to pursue projects/advisors

Spent 1st year working on autonomous vehicles (Dr. Del Vecchio’s lab)

Took math + control theory courses

April, 2009: “I cannot imagine that I find this ‘smart grid’ interesting at all...”

June, 2009: started working with Prof. Ian Hiskens on corrective grid control

Took optimization and system theory courses, but no power systems courses

Was really fortunate to be able to work on

Multi-energy system modeling (energy hubs)

Coordinated EV charging (predictive control)

Corrective grid control with energy-constrained resources

Last year of PhD (2012): consulted / moon-lighted for MI/IL energy startup

5

EE as a grad student

6

Multi-energy systems

EV charging

A small multi-energy system example

Electric

Natural Gas

Hydro

Wind

Distributed

optimization

and control

Modeling

EE as an grad student

7

Corrective control

“Co-op 2.0” = part-time consultant

Economics & reliability

Resilient response

• Multi-energy optimization

• Adapt unit commitment

• Adapt economic dispatch

A lot of midnight oil was burned in 2012

Next lessons

8

Scalability requires

modularity

Software is the lingua

franca of tech

Learn to communicate your

work with MBAs

EE as an entrepreneur

9

Root3 raised $1M from VCs in CA, MI, and IL.

Joined Root3 after multiple post-doc offers from MIT, CMU, ETHz, and other top schools.

EE as an entrepreneur

10

Focused on optimization, data science, and software UI/UX work

Economic dispatch/unit commitment for C&I energy plants ➔ chillers, boilers, CHPs, TES.

EE as an entrepreneur

11

Focused on optimization, data analysis, and software UI/UX work

Each customer became a research project➔ not scalable

We needed to stop and re-assess approach ➔ no time

Received an offer to join UVM as TT faculty

Next lessons

12

Mathematics and software are

the lingua franca of research

Faculty job is 5 rolled into 1 with

freedom to pursue any EE interest

Find bridges between

academia and industry

My tenure-track experience

13

Undergrad

Grad school

Faculty

© EHT

As an undergraduate student:

1. Take courses (Student)

As a grad student:

1. Perform actual research (Researcher)

2. Take some courses (Student)

As a faculty:

1. Lead research project (Project manager)

2. Teach + curriculum development (Teacher)

3. Research advisor for your grad student (Mentor)

4. Do actual research (Researcher)

5. Advise undergrads on career choices (Advisor)

14

Advanced Control

Systems

Controlling DERs

Valuing DERs

Wind/Solar Growth

Transmission

Planning

T&D planning

Renewables

Integration

The Energy Systems Lab at

the University of Vermont

Stochastic systems

Distributed Charging

of EVs

Control Systems

Power

Optimization &

Control

Multi-period OPF

Communications

Power

Systems

Power Systems

Resilience

Battery

optimization

Recent and ongoing industry-research projects with

Recent and ongoing federal collaboration with

The Energy & Systems Laboratory (TESLa) is growing

15

16

Years of federal R&D and IAB feedback

Network Optimized Distributed Energy Systems

ARPA-E PROJECT PARTNERS, LED BY THE U OF VERMONT

UTILITY PARTNERS

SOLUTION PROVIDERS

GOVERNMENT & POLICY

TECH 2 MARKET

Identify three key challenges

Variable supply Aging infrastructure Distributed energy

Increasing the need for demand that can follow generation

Rising demand for alternatives to avoid expensive capital expenditures

New generation leading to financial and engineering challenges

Connect trends

Turning connected things

into virtual batteries

100% Connected

100% Clean

20

Daily evening peaks

due to utility’s

(timed) demand

response program

~15,000 electric water

heaters

2015

2016

2017

Days in April (2015-2018)

Up to a 50% increase in demand from storms/clouds

2018

Overcast (2018)

Duck says what?

21

VT load curves for two consecutive days in 2019

22

Solutions?Solutions?

cliparts.co

Controllable, but

expensive & dirty

Coordinate Loads

How do these loads behave in aggregate?

How well can we control them?

Quality of service guarantee?

Install more generation

Free & “Clean”

Leverage key tools to coordinate at scale

23

Packetization of

data on internet

Randomization to

desynchronize

Supply

Time

Demand

MW

24

Packetized energy management (PEM) at scale

Uncoordinated demand

Uncoordinated packets

Supply

Demand

Before PEM

25

Packetized energy management (PEM) at scale

Packetization

Algorithms enable the coordinator to

follow a dispatch schedule,

just like a battery!Packetization +

Randomization

Load choreographed with PEM

After PEM

26

Supply

Demand

Packetized energy management (PEM) at scale

Tracking a time-varying signal (real-time comms)

27Desrochers, Khurram, et al., Real-world, Full-scale Validation of Power Balancing Services from Packetized Virtual Batteries, IEEE ISGT, 2019

Comparing : diversity increases flexibility

28

TCL-only offers less flexibility!

Diversity is key to unlock VPP flexibility!

1500-device VPPs

Almassalkhi M., Espinosa L.D., et al. (2018) Asynchronous Coordination of Distributed Energy Resources with Packetized Energy Management. In: Meyn S.,

Samad T., Hiskens I., Stoustrup J. (eds) Energy Markets and Responsive Grids. The IMA Volumes in Mathematics and its Applications, vol 162. Springer, New

York, NY

Traditional vs. Packetized

CONVENTIONAL THERMOSTAT (long on/off times)

TIME (MINUTES)

Black = Device is OFF

DEV

ICE

ID

PACKETIZED THERMOSTAT (multiple short on/off times)

DEV

ICE

ID

60 90 120 150 180 210 240 270 300 330 360

30

Packetized energy management (PEM) at scale

Power consumed by

5000 packetized water

heaters controlled to

match renewable

energy baseline.

Temperature distribution

of 5000 packetized water

heaters

Dispatchable demand

Turning connected devices into virtual batteries (VBs)

31

+

_

Every home, neighborhood, feeder,

or city can be a virtual battery

Less than ½ the cost of physical

batteries for same kW/kWh rating

Dispatchable Demand

Optimally coordinating networked DERs at scale

Manage resources economically Manage grid physics optimally Manage resources dynamically

Key idea: adapt wide-area control concepts to distribution grid operations

Key challenge: resources have finite energy constraints (not a generator)

32

Optimally coordinating networked VBs at scale

Manage resources economically

Key idea: adapt wide-area control concepts to distribution grid operations

Key challenge: resources have finite energy constraints

33

Quantifying techno-economic benefits of advanced inverter and battery functionality

Interactive in-

browser 3ph power

flow solver

Ejecting utilities from the death spiral with DERs

34

The smart solution:

packetized virtual batteries

Value to utility

Cost to utility

Consumer inconvenience

Full stack of grid services, including the ability to manage distribution network constraints through distributed, heterogeneous grid edge devices.

Flexibility solutions at about half the cost of batteries. Innovative deployment solutions will lead to even more affordable programs in the future.

Device-driven approach ensures that consumers see no difference in their energy services.

Devices request energy when needed

Also in the real world (crushing peaks)

ABOUT 60 WATER HEATERS, VERMONT ELETRIC CO-OP (raw kW data)

Also in the real world (arbitraging)

ABOUT 60 WATER HEATERS, VERMONT ELETRIC CO-OP (raw kW data)

Most large energy appliances can be packetized

Water Heaters

The Mello smart thermostat

EV Chargers

Packetized WebastoLevel 2 EV charging system

HVAC ThermostatsHeat Pumps(mini splits)

Pool Pumps Irrigation PumpsGrid Edge Batteries

PV Inverters

Refrigeration

Advantages inherent to PEM

Set it and forget it

Smart design makes our software easy to use for both end users and utilities

Built on ideas that run the Internet, our solutions increase in value as they scale

Device-driven solutions enable flexibility without impacting customer comfort

Scalability Consumer comfort Privacy & security

Bottom-up design minimizes data collection and reduces security threats

Momentum

2013-2015

2016

IP DEVELOPMENT

Initial R&D, first patent disclosure applied to EVs

TECH ADVANCEMENT

$2M ARPA-E project, company founded, second patent disclosure, awarded first pilot

2017

2018

CUSTOMER ADOPTION

Launched 2 new projects, UL listed smart device for water heaters, new DOE and NSF grants awarded

COMMERCIAL VALIDATION & PARTNERSHIPS

Launching new projects in CA, new OEM partnerships, system deployment

2019

SCALING

Proving value, sales, moving from demonstration projects to full-scale deployment

“Game changing startups of 2019”

41

”I cannot imagine that I find this

‘smart grid’ interesting at all.”

Mads Almassalkhi

April, 2009

Questions? Comments? Thank you!

42

Contact info

Mads Almassalkhi

malmassa@uvm.edu

@theEnergyMads

Batteries that never run out!

Questions? Comments? Thank you!

43

Contact info

Mads Almassalkhi

malmassa@uvm.edu

@theEnergyMads

Optimization Methods for Unbalanced Power Distribution Systems

Enabling Advanced Grid Operations with DER coordination

Advanced Grid Architectures to support scalable DER integration

Dates to be set shortly

Join me in Atlanta, GA!

Chairing three sessions at PES GM 2019: state of the art of DOPF and DERs