Keynote Sensys 11-7-2012

Sustainable Energy Networks - a Sensys grand opportunity
David E. Culler University of California, Berkeley
Sensys 2012 Keynote

11-7-2012
11/7/2012
Sustainability?
Sustainable development should meet the
needs of the present without compromising the ability of future generations to meet their own needs
Our Common Future, World Commission on Environment and Development, United Nations, 1987
11/7/2012
My Goal today
Suggest ways SenSys community is and can
be more engaged in achieving sustainable energy networks
11/7/2012
AB 32
Quantifying Sustainability California Law
Reduce GHG emissions to 1990 levels by 2020
Governors executive order S-3-05 (2005)

80% reduction below 1990 levels by 2050
Renewable Portfolio Standard

33% renewables by 2020, 20% biopower procurement
480 => 80 mmT CO2e in 40 years

Population: 37 => 55 million Economic growth
4
11/7/2012
CA2050: GHG 90% below 1990
But,
GHG Emissions (MtCO2e/yr)
1,000 800 600 400 200 0

1990 2005 2020 2050 Historical
5 Energy emissions Non-energy emissions 2020 Target 2050 Target
BAU
11/7/2012
CA2050 4 Part GHG Reduction Plan

Eciency
11/7/2012

Eciency Electrify
11/7/2012

Eciency Electrify Decarbonize
the electricity Decarbonize the fuel
11/7/2012

Eciency Electrify Decarbonize
the electricity Decarbonize the fuel
11/7/2012
All required for even 60% reduction
but still fall short of 80%

10
11/7/2012
Scenarios
New Nuclear plant New CCS facility every 14 months every 9 mo. for 40 years Exceeds saline aquifer
11
11/7/2012
Resources exist - 1.4 % of CA land - 43% agriculture - 3.4% urban
The Problem: Supply-Demand Match

Baseline + Dispatchable Tiers Oblivious Loads
Generation
Transmission
Distribution
Demand
12
11/7/2012
An Engineering Marvel
NA: 3 synchd regions Lots of wires
170k miles of >200 kv
transmission 6m miles of distribution 3k miles of 500 kv DC
3,200 retail
distributors 147 M customers

125M res, 17.6 M com,
0.78 M ind.
10 ISO/RTO cover 2/3 Little communication

http://www.nerc.com/docs/oc/rs/BA_BubbleDiagram_2011-10-03.jpg
13
11/7/2012
Towards an Aware Energy Network

Baseline + Dispatchable Tiers Oblivious Loads
Generation
Transmission
Distribution
Demand
Non-Dispatchable Sources
Aware Interactive Loads Communication Communication
14
11/7/2012
Classical view of the Energy Challenge

Markets
Imports
Transmission
Distribution
Peaker Single cycle nat. gas Intermittent Combined cycle nat. gas
Baseline
Electronics Appliance Lighting HVAC Motors

Lines Transformers Power Supplies Meters VFD Control Circuits
Geothermal Hydro Nuclear, Coal Generators
Efficiency
15
Supply
Transport
11/7/2012
Load
Modern Energy Network Challenges

Information Plane
Storage
Fluctuating
Imports
Pumped
Storage Fluctuating Renewable
Storage
Renewable
Wind Photovoltaic
Personal Electronics Appliance Facilities
Transmission
Distribution
Peaker Single cycle nat. gas / bio Intermittent Combined cycle nat. gas / bio
Baseline
Lighting HVAC Motors Industrial
Geothermal Hydro
Transportation
Nuclear, Generators Coal Physical Plane

16
Lines Transformers
Meters
Power Supplies VFD
Circuits Control Efficiency
Usage
Pricing
Schedule
Supply
Transport
11/7/2012
Load
Consumption
Sensys? Creating the info plane
17
11/7/2012
Smart Meters the WSN Killer App?

Markets
Imports
Transmission
Distribution
Peaker Single cycle nat. gas Intermittent Combined cycle nat. gas
Baseline
Electronics Appliance Lighting HVAC Motors

Lines Transformers Power Supplies Meters VFD Control Circuits
Geothermal Hydro Nuclear, Coal Generators
Efficiency
18
Supply
Transport
11/7/2012
Load
Smart meter rollouts

Open IPv6,
Proprietary / Zigbee
http://www.edisonfoundation.net/iee/Documents/IEE_SmartMeterRollouts_0512.pdf
19
11/7/2012
Much more than meters and billing
http://www.openthegrid.com/docs/ipv6_in_smart_grid_field_area.pdf
20
11/7/2012
Key WSN Research Developments

Event-Driven Component-Base Operating System
Applica tion Applica tion
Framework for building System & Network abstractions Low-Power Protocols Hardware and Application Specic
FF lus h lus h Golden Ga te Bridg Golden Ga te Bridg ee
Delug Delug ee
Drip Drip
R edwoods R edwoods North ea North SSea
SSPP IN IN Broa dca t torm Broa dca ss tSStorm TinyOS 0.6 TinyOS 0.6
CODA CODA
BVR BVR
CTP CTP
MintR oute MintR oute
SSPP
Link Link
1999
Idle listening
2000 2001
2002
2003
2004
2005
CP -MAC SSCP -MAC PP E DAMACS E DAMACS X-MAC X-MAC
Wis eMAC Wis eMAC
WooMAC WooMAC
-MAC SS-MAC Aloha S Aloha PP S
B -MAC B-MAC
T-MAC T-MAC
2006
2007
2008
All the energy is consumed by listening for a packet to receive
=> Turn radio on only when there is something to hear
Reliable routing on Low-Power & Lossy Links

Power, Range, Obstructions => multi-hop Always at edge of SNR => loss is common
=> monitoring, retransmission, and local rerouting
Trickle dont ood (tx rate < 1/density, and < info change)
designer. never naively respond to a broadcast re-broadcast very very politely
21
11/7/2012
Connectivity is determined by physical points of interest, not network
SS44
Network Network
MultihopLQI MultihopLQI FF us ion us ion
PP SFF Q S Q
R MS T R MS T
FF P S PS
Dra in Dra in
Volca no Volca no
GAF GAF
GDI GDI
Dip Dip
Tra ns port Tra ns port
The Mote/TinyOS revolution

zeevo BT BTNod Eyes Mote inside Intel e trio uP => Arm Cortex cf-mica Radio => 802.15.4g narrow=band SmartDust Mica Telos Ren WeC frequency hopper XBOW e XBOW XBOW XBOW XBOW mica mica2 cc-dot TinyOS too rene2 micaZ digital sun SOC from here Bosch rain-mica cc-mica Dust Inc blue cc-TI
LoWPAN/IPv6
Intel Intel/UCB rene dot
Intel iMOTE
Intel MOTE2
Epic
98
99 SENSIT Expedition
00 8 kB rom kB ram
01
02
03 CENS STC
04 48 kB rom 10 kB ram 802.15.4
05 NETS/ NOSS
06
07 CyberPhysical
10 11
NEST
NSF
11/7/2012
22
IETF RPL
Idle Listening: 3 Basic Solutions

Scheduled Listening

Arrange a schedule of communication Time Slots Maintain coordinated clocks and schedule Listen during specic slots Many variants:
Aloha, Token-Ring, TDMA, Beacons, Bluetooth piconets, S-MAC, T-MAC, PEDAMACS, TSMP, FPS,
Sampled Listening
Listen for very short intervals to detect eminent
transmissions On detection, listen actively to receive DARPA packet radio, LPL, BMAC, XMAC, Maintain always on illusion, Robust
Listen after send (with powered infrastructure)

After transmit to a receptive device, listen for a short time Many variants: 802.11 AMAT, Key fobs, remote modems,
Many hybrids possible

11/7/2012
23
Internet WSN assimilated

Diverse Object and Data Models (HTML, XML, , BacNet, )
7: app 4: xport 3: net 2: link
Application (Telnet, FTP, SMTP, SNMP, HTTP) Transport (UDP/IP, TCP/IP) Network (IP) Link
RPL RFC6550
COAP
6LoWPAN RFC6282 802.11 802.11a WiFi 802.11b WiFi 802.11g WiFi 802.11n WiFi WiFi 802.15.4 LoWPAN 802.15.4e 802.15.4g P1901.2
1: phy
Serial X3T9.5 802.3 802.5 802.3a Modem FDDI Ethernet Token Ring 802.3i Ethernet ISDN Sonet 802.3y Ethernet 10b2 DSL 802.3ab Ethernet GPRS 10bT 802.3an Ethernet 100bT Ethernet 1000bT 1G bT
11/7/2012
24
Or in industry speak
25
11/7/2012
At incredible scale
26
11/7/2012
ROLL & Sensys Role

RPL retained fundamental routing diversity
Multiple DODAGs with selection
Incorporated trickle density awareness Many questions for which the research was simply
not there
Piece-wise source routing, routing stretch versus protocol
complexity & state, routing metrics,
Many issues for which the analysis is seriously
incomplete
Local / global repair, loop formation, routing staleness,
scaling, state management, trickle timers
IETF, mobile, and industry rediscovering Where is the research community?

27
11/7/2012
Creating an Energy Network ???

Information Plane
Storage
Fluctuating
Imports
Pumped
Storage
Renewable
Wind Photovoltaic
Transmission
Distribution
Baseline
Geothermal Hydro
Transportation

28
Lines Transformers
Meters
Power Supplies VFD
Usage
Pricing
Schedule
Supply
Transport
11/7/2012
Load
Consumption
How do we start?
29
11/7/2012
Grid Exists
Generation Transmission Distribution Load Conventional Electric Grid
30
11/7/2012
Internet Exists
Generation Transmission Distribution Load Conventional Electric Grid
Conventional Internet
31
11/7/2012
Intelligent Energy Network as Overlay on Both

Intelligent Energy Network
Source IPS energy subnet Load IPS Intelligent Power Switch
Generation Transmission Distribution Load Conventional Electric Grid Conventional Internet

32
11/7/2012
Where to Start?
Buildings
72% of electrical consumption (US), 40-50% of total consumption, 42% of GHG footprint US commercial building consumption
Electricity source
doubled 1980-2000, 1.5x more by 2025 [NREL]

Renewable energy consumption
Where Coal is used Prime target of opportunity for
renewable supplies
33
11/7/2012
Building Power Consumption

1MW 883 kW 11%
34
11/7/2012
MVP (most valuable principle) of WSN

Do nothing well
35
11/7/2012
PowerProportional Buildings ?
Cory Hall: Office + Semiconductor + IT
950 KW
1150 KW
Min = 82% of Max
36
11/7/2012
Stanley Hall:
Office + BioScience - 13 NMRs 1.45 MW 2.02 MW
Min = 72% of Max
37
11/7/2012
756 KW Koshland Hall: Office + ??? 1030 KW
Min = 69% of Max
38
11/7/2012
LeConte Hall: Office
202 KW 62 KW
Min = 31% of Max
39
11/7/2012
Power Proportionality
Consumption
Productivity
40
11/7/2012
Power Proportionality
Consumption
Productivity
41
11/7/2012
Doing Nothing Well ???
42
11/7/2012
Power Proportional Buildings?

50 Ton Chiller 200 Ton Chiller
10 months
Scott McNally Bldg Manager
11/7/2012
2 months
43
The Building Challenge

Operations and Environment
ree ctric Load T Ele
Climate Plant
CT: mains power monitoring

A B
1 5 9 13 17 21 25 29 33 37 41 3 7 11 15 19 23 27 31 35 39 2 6 10 14 18 22 26 30 34 38 42 4 8 12 16 20 24 28 32 36 40 B B
Panel 1
Panel 2
1 5 9
3 7
2 6
4 8
13 17 21 25 29 33 37 41
11 15 19 23 27 31 35 39
10 14 18 22 26 30 34 38 42
12 16 20 24 28 32 36 40
Vibra;on
panel level power monitoring
Humidity Temperature Pressure

11/7/2012
ACme: plug load energy monitor and controller 44
CEC B2G Testbed - An Energy Transparent Building

DOP HVAC
MCL equip
Whole Bldg
MCL infra Central vent
MCL vac
office HVAC
servers Plug loads

45
Lighting Parking Lot

11/7/2012
inst Lab 199 HVAC
The Other Energy Usage

Wireless plug meters on 611 of 1200 loads
Bldg 90
46
11/7/2012
sMAP simple Monitoring and Action Protocol Uniform Access to Diverse Physical Information
Applications
Modeling Control Storage Visualization Location Debugging Personal Feedback Continuous Commissioning
JSON Objects REST API HTTP/TCP
Actuation Authentication
sMAP
Physical Information
Electrical Geographical Occupancy
Water
Structural
Weather
Environmental
Actuator
47
11/7/2012
sMAP Virtual Energy-Lab Ecosystem

sMAP Resources
California ISO sMAP Gateway
Applica;ons
sMAP sMAP
AC plug meter Vibra;on / Humidity Proxy Server
Database
sMAP
Weather
Google PowerMeter
EBHTTP / IPv6 / 6LowPAN Wireless Mesh Network sMAP sMAP
EBHTTP Transla;on
Edge Router
Internet
Every Building Light switch Temperature/PAR/TSR Dent circuit meter
sMAP
sMAP
Modbus
sMAP Gateway
RS-485
sMAP Gateway Cell phone
48
11/7/2012
Factoring is critical
sMAP
6lowpan networks
Public interfaces
sMAP control sMAP
RS-458 bus
models
mgmt
web
BacNET/IP
Archiver
RDBMS
TSDB
Represent, transmit data and metadata Abstract underlying heterogeneity into simple data model
49
Provide access to archived data Manage views, data cleaning

11/7/2012
Application-specific functionality built on exposed interfaces
sMAP is
Universal information representation for physical
data
Self-describing, compact JSON schema, transportable
over UDP/TCP Integrated metadata
Software Architecture for physical data processing
and actuation

Real-time and archival data, time-series DB Adapters/Drivers for legacy and direct streams Subscription, syndication, distillates Query processing, visualization interface
Resource-oriented web-service framework for
embedded applications
11/7/2012
http://code.google.com/p/smap-data
50
All physical info

Name ISO Data ACme devices EECS submetering project EECS steam and condensate UC Berkeley submetering feeds Sutardja Dai, Brower Hall BMS UC Davis submetering feeds Weather feeds CBE PMP toolkit NOA Weather Forecast SDH Air Quality Sensor Type CAISO, NYISO, PJM, MISO, ERCOT, BPA Plug-load electric meter Dent Instruments PowerScout 18 electric meters Cadillac condensate; Central Station steam meter ION 6200, Obvius Aquisuite; PSL pQube, Veris Industries E30 Siemens Apogee BMS, Legrand WattStopper, Johnson Control BMS Misc., Schneider Electric ION Vaisala WXT520 rooftop weather station; Wunderground Dust motes; New York Times BMS Meteorological (window, solar, cloud, etc) CO2, Temp, TSR, PAR, Hum Access Method Web scrape Wireless 6lowpan mesh Modbus Modbus/TCP Mosbus/Ethernet, HTTP BACnet/IP OPC-DA SDI-12, LabJack/Modbus, web scrape CSV import; serial Web Wireless 6lowpan mesh Channels 1211 344 4644 13 4269 4064 34 (+) 33 874 77740 50
total: 93,242
www.openbms.org
51
11/7/2012
Living Lab Approach: Innovate in a Virtual Private Grid
Generation Transmission
VPG
Distribution Load
52
11/7/2012
sMAP Energy Markets, Ops, $,
BPA
http://www.isorto.org
53
11/7/2012
sMAP Solar, Wind, Meteorology
1450 Met feeds covering Cal Solar Initiative: 130,275 proj. 1370 MW Cal Utility Solar: 60 facilities 695 MW Cal Utility Wind: 134 facilities 4295 MW
54
11/7/2012
sMAP generation
CA generation plant locations, type, and
rated power (> 0.1 MW) [CEC] Hourly output from each type of CA generation source for > 1 year [CAISO]
[CEC] http://energyalmanac.ca.gov/powerplants/Power_Plants.xls [CAISO] http://www.caiso.com/green/renewableswatch.html

55
11/7/2012
Modern Energy Network Challenges

Information Plane
Storage
Fluctuating
Imports
Pumped
Storage
Renewable
Wind Photovoltaic
Transmission
Distribution
Baseline
Geothermal Hydro
Transportation

56
Lines Transformers
Meters
Power Supplies VFD
Usage
Pricing
Schedule
Supply
Transport
11/7/2012
Load
Consumption
What would the CA Grid be like @ 60% renewables?

G-20 Investment ($B)
Whos Winning the Clean Energy Race? 2010 Pew Charitable Trust www.PewEnvironment.org/CleanEnergy Bloomberg New Energy Finance, Liebreich April 5, 2011
57
11/7/2012
CA grid today - Supplies

Source Geothermal Biomass/ Biogas Small Hydro Wind Solar3 Nuclear Hydro Imports Thermal Total
1 2 58
Rated (GW) 2.600 1.145 1.380 2.812 0.403 4.456 12.574 11.0552 44.339 80.764
Capacity Factor1 38.7% 43.5% 31.7% 29.1% 28.7% 85.9% 27.7% 66.6% 19.7% 32.6%
Total Energy (TWh) 8.68 4.30 3.77 7.06 1.00 33.00 30.05 63.43 75.43 226.71
% of Total Energy 3.8% 1.9% 1.7% 3.1% 0.4% 14.6% 13.3% 28.0% 33.3% 100.0%
Mean delivered power divided by rated power (excl. import) For imports, rating is the maximum observed power 3 Residential net factored into demand
11/7/2012
CA grid today Supply Challenge
59
11/7/2012
More views time and blend
60
11/7/2012
CA grid today - Supplies

Source Geothermal Biomass/ Biogas Small Hydro Wind Solar3 Nuclear Hydro Imports Thermal Total
1 2 61
Rated (GW) 2.600 1.145 1.380 2.812 0.403 4.456 12.574 11.0552 44.339 80.764
Capacity Factor1 38.7% 43.5% 31.7% 29.1% 28.7% 85.9% 27.7% 66.6% 19.7% 32.6%
Total Energy (TWh) 8.68 4.30 3.77 7.06 1.00 33.00 30.05 63.43 75.43 226.71
% of Total Energy 3.8% 1.9% 1.7% 3.1% 0.4% 14.6% 13.3% 28.0% 33.3% 100.0%
Mean delivered power divided by rated power (excl. import) For imports, rating is the maximum observed power 3 Residential net factored into demand
11/7/2012
A year in the todays grid
Seasonal, Weekly, Daily variations Many underlying factors
Peak: 47.1 GW Min: 18.8 GW Mean: 26.3 GW

62
11/7/2012
A year daily averages

70 60
IMPORTS: 28.0% THERMAL: 33.3% WIND: 3.1% SOLAR: 0.4% BIOMASS/BIOGAS: 1.9% GEOTHERMAL: 3.8% HYDRO: 13.3% SMALL HYDRO: 1.7% NUCLEAR: 14.5% CA GRID DEMAND
Total CA Power (GW)
50
40
30
20
10
0 Sep
Nov
Jan
Mar
May
Jul
63
11/7/2012
A mid-summers week
45 40 35
Total CA Power (GW)
30 25 20 15 10 5 0 07/08
70
60
Total CA Power (GW)
50
40
07/09
07/10
07/11
07/12
07/13
30
20
10
0 Sep
Nov
Jan
Mar
May
Jul
64
11/7/2012
A winter weeks tale

45 40 35
Total CA Power (GW)
30 25
70
20
60
15 10 5 0 12/19
Total CA Power (GW)
50
40
12/20
12/21
12/22
12/23
12/24
30
20
10
0 Sep
Nov
Jan
Mar
May
Jul
65
11/7/2012
The Demand Duration Curve
66
11/7/2012
A Simple what if
Take current demand, current activity, current
technology, current deployment At a crude top-level scale (by category)

Represented by the time series
Scale up the renewable portions

Preserve the seasonal, weekly, daily, hourly eects
of mother nature *
Scale back the fossil fuel based supplies With current demand as a reference
67
11/7/2012
Example: Solar
68
11/7/2012
Example: Solar Scaled
69
11/7/2012
Example: Wind
70
11/7/2012
Wind - Scaled
71
11/7/2012
Joint Wind/Solar Scaling

Source Unscaled Rated (GW) Wind Solar TOTAL 2.81 0.40 3.21 Scaled Rated (GW)
57.1 29.8 86.9
72
11/7/2012
A Year in CA grid @ 60%

70 60
EXCESS IMPORTS: 2.6% THERMAL: 9.7% WIND: 34.4% SOLAR: 18.2% BIOMASS/BIOGAS: 1.9% GEOTHERMAL: 3.8% HYDRO: 13.3% SMALL HYDRO: 1.7% NUCLEAR: 14.5% CA GRID DEMAND
Total CA Power (GW)
50
40
30
20
10
0 Sep
Nov
Jan
Mar
May
Jul
73
11/7/2012
A Summer Week @ 60%

80 70 60
Total CA Power (GW)
50 40 30 20 10 0 07/08
70
60
07/09
07/10
07/11
07/12
Total CA Power (GW)
50
40
30
20
10
74
11/7/2012
0 Sep
Nov
Jan
Mar
May
Jul
A Winter Week @ 60%

80 70 60
Total CA Power (GW)
50 40
70
30
60
20 10 0 12/19
Total CA Power (GW)
50
40
12/20
12/21
12/22
12/23
30
20
10
75
11/7/2012
0 Sep
Nov
Jan
Mar
May
Jul
CA Grid @ 60%
76
11/7/2012
What Can we do to Make it Work?

Design for deep penetration Eciency for shaping
especially at night
Optimize for the whole, not peak production Use your o-grid intuition Poor power proportionality of buildings and other loads,
Storage
Move energy in time Precooling, preheating, guardband adjustment Deferral, acceleration Utilize resources in concert with non-dispatcables
Load scheduling (continuous DR)
Integrated Portfolio Management Curtailment

77
11/7/2012
Load shifting to follow supply

70 60
Total CA Power (GW)
50
40
30
20
Windrush? Sunrush? - energy agile industry?
10
0 Sep
Nov
Jan
Mar
May
Jul
78
11/7/2012
A Day
35 30
Total CA Power (GW)
25
20
15
10
0 00:00
06:00
12:00
18:00
79
11/7/2012
The Day with +/- 3 hours of shift

35 30
Total CA Power (GW)
25
20
15
10
0 00:00
06:00
12:00
18:00
80
11/7/2012
How to match demand to supply?
81
11/7/2012
Modern EE & CS Energy Challenges

Information Plane
Storage
Fluctuating
Imports
Pumped
Storage
Renewable
Wind Photovoltaic
Transmission
Distribution
Baseline
Geothermal Hydro
Transportation

82
Lines Transformers
Meters
Power Supplies VFD
Usage
Pricing
Schedule
Supply
Transport
11/7/2012
Load
Consumption
Interesting Sensor Networks

151 Temperature Sensors
ense 6,000 S > oints ontrol P and C
312 Light Relays
50 Electrical Sub-meters 6 Variable Speed Pumps
12 Variable Speed Fans
138 Air Dampers
121 Controllable Valves
Sutardja Dai Hall Built in 2009

83
140k sq. ft.

11/7/2012
Controls are Widely Available

Bancroft Library:
Built in 1949 100k sq. ft. 5,000 points
>70% of large buildings have digital controls
U.S. Energy Information Administration, 2009

84
11/7/2012
HVAC Model Predictive Control
Whole Bldg
inst Lab 199 HVAC
85
11/7/2012
Measure => Model => Mitigate

Mathematical model from Newtons law of cooling
time constant of room change in temperature over time weather
dT/dt = -krT - kcu(t) + kww(t)) + q(t)

AC cooling
heating from occupants and equipment
Model identied using semi-parametric regression

Temperature: Experimental (blue) Simulated (red)
23
(C)
22 21 11AM 12PM 1PM 2PM 3PM Time 4PM 5PM 6PM 7PM 8PM
(C/s)
Heating from occupants and equipment
0.029 0.0285 11AM 12PM 1PM 2PM 3PM Time 4PM 5PM 6PM 7PM 8PM
86
11/7/2012
Learning-Based Model Predictive Control

Experimental LBMPC: 12.6kWh Consumed Temperature
23
(C)
22 1PM 3PM 5PM 7PM 9PM 11PM 1AM Time 3AM 5AM 7AM 9AM 11AM
Control Action
AC State
1 0 1PM 3PM 5PM 7PM 9PM 11PM 1AM Time 3AM 5AM 7AM 9AM 11AM
Simulated Hysteresis Control: 29.7kWh Consumed (estimated)

23
Temperature Control Action
(C)
22 1PM 3PM 5PM 7PM 9PM 11PM 1AM Time 3AM 5AM 7AM 9AM 11AM
AC State
1 0 1PM 3PM 5PM 7PM 9PM 11PM 1AM Time 3AM 5AM 7AM 9AM
6 Transient Power Steady State Power
11AM
LBMPC adjusts for internal dynamics, avoids over-cooling, trades off duty cycle and switching frequency
kW
(Aswani, Master, Taneja, Culler, Tomlin, 2011)

87
11/7/2012
4 Minutes
Supply-Following Computational Loads Background Processing (shiftable)
Requests Availability Forecasts
IPS
QoS (fidelity & latency)

Power
Controllable Storage
88
11/7/2012
Energy-Availability Driven Scheduling

Non-dispatchable, variable supply
14 12 10 x 10
4
Power (W)
8 6 4 2 0
200
400 Time (hrs)
600
800
Pacheco wind farm

x 10 5
4
x 10 5 4
greedy pacheco 2.0x
Power (W)
Power (W)
0 200 400 Time (hrs) 600 800
4 3 2 1 0
3 2 1 0
Power proportional, grid-aware loads
200
250 Time (hrs)
300
Scientific computing cluster

89
11/7/2012
NREL Western Wind and Solar Integration Study Dataset http://wind.nrel.gov/Web_nrel/
Energy Slack
Thermostatically Controlled Load IPS
Set Point
Guard band
90
11/7/2012
Everyday Slack and Control

6 kg. NH4Cl in 19.1% aqueous solution
91
11/7/2012
From Auto Demand Response

Sutardja Dai Hall
92
11/7/2012
To Personalized Automated Lighting Control

HTTP
Python Control Process MySQL Python Django
sMAP
BACnet
Three controllable
ballasts per xture
Gateway
93
11/7/2012
~5 zones per oor
Real Energy Savings
SDH 4th Floor Lighting Energy Usage

4 3.5 3 2.5 KW 2 1.5 Colab Savings 1 Floor kW 0.5 Collab kW 0 0 0 58% 53% 68% 69% 63% 65% 68% 75% 80% 71% 70% 61% 60% 50% 40% 30% 20% 10% 0%
94
11/7/2012
A Macroscope ???
100% Cooling
100 80 SDH.S101 SDH.S204 SDH.S404 SDH.S508 SDH.S510 SDH.S608 SDH.S610 SDH.S713 SDH.S710 SDH.S219 SDH.S218 SDH.S701 SDH.S518 SDH.S705 SDH.S304 SDH.S317 SDH.S601 SDH.S216 SDH.S704 SDH.S215 SDH.S501 SDH.S716 SDH.S319 SDH.S307 SDH.S211 SDH.S418 SDH.S413 SDH.S512 SDH.S316 SDH.S708 SDH.S311 SDH.S602 SDH.S308 SDH.S603 SDH.S406 SDH.S502 SDH.S615 SDH.S205 SDH.S607 SDH.S217 SDH.S315 SDH.S702 SDH.S306 SDH.S416 SDH.S409 SDH.S103 SDH.S104 SDH.S202 SDH.S206 SDH.S207 SDH.S212 10/14 00:00 SDH.S405 SDH.S301 SDH.S707 SDH.S611
60
40
20
Dead band
20
40
60
80
100% Heating
95
100 10/11 00:00
10/11 12:00
10/12 00:00
10/12 12:00
10/13 00:00
10/13 12:00
11/7/2012
Network of Control Loops

Outside Air Intake
Cooling Tower Loop

Condenser Water Pump (CWP) Always on max Chiller 45 F setpoint
Chilled Water Pump (CHP) Always on max
Economizer Loop
Air Handling Loop

Cooling Coil & Valve 55 F setpoint Supply Fan (Air Handler Unit AHU) 336 Pa setpoint
VAV Loop
VAV Loop
VAV Loop
Dampers (VAVs)
Return Fan
Exhaust Air
Heating Coils & Valves 70 F setpoint
96
11/7/2012
Hot Water Loop
Cold Water Loop
Script the building
97
11/7/2012
98
11/7/2012
Theres a bldg app for that
99
11/7/2012
Holistic HVAC optimization

Empirical models, including 280 external factors, and 260 monitoring, provide rapid, focused feedback 240
220
Savings: 16.52% 25.4 kW
Power (kW)
200
180
160
140
Measured Baseline
12/11 12/12
RMSE = 5.6%
12/13 12/14 12/15 12/16 12/17 12/18 12/19
120 12/10
10
x 10
Dynamically set economizer, supply air temp, min
Air Flow (CFM)
8 6 4 2 0 08/10 Fresh air Return Air 08/10 100 08/11 08/11 08/12 08/12 08/13 08/13 08/14 08/14
airow, reheat
11/7/2012
Where is this going: BOSS and Software Defined Buildings

application logic
control process
Model Training
Runtime Logic
Time-series Client
Transaction Client
system libraries
portable, robust applications for the physical environment

1. Hardware presentation layer: sMAP" 2. Hardware abstraction layer: device-specic logic" 3. Time-series service: the archiver" 4. Reliable control inputs: the transaction manager" 5. Security: the authorization service"
Authorization token
BOSS boundary
request historical data authorize verify callback submit
5
Auth Service
2
HAL Service
Time Series Service
Transaction Manager
3
publish command
sMAP
sMAP
sMAP
sMAP
sMAP
XML/HTTP OPC-DA
6loWPAN
RS-485
BACnet/IP
101
October 26, 2012
CyberPhysical Building Systems
External
Planning Visualiza;on
Security, Fault, Anomaly Detect &Management Mul;-Objec;ve Model-Driven Control
Control and Schedule
Occupant Sa;sfac;on
Activity/ Usage Streams Pervasive Sensing
Empirical Models Physical Models
Building Integrated Operating System drvrs drvrs drvrs Local Controllers
Occupant Demand
Transport Process Loads
Electrical
BMS
HVAC
Legacy Instrumentation & Control Interfaces
Cyber
102
11/7/2012
Physical Building
Light
BIM
Modern EE & CS Energy Challenges

Information Plane
Storage
Fluctuating
Imports
Pumped
Storage
Renewable
Wind Photovoltaic
Transmission
Distribution
Baseline
Geothermal Hydro
Transportation

103
Lines Transformers
Meters
Power Supplies VFD
Usage
Pricing
Schedule
Supply
Transport
11/7/2012
Load
Consumption
Synchrophasors
Synchronized Phasor Measurement Units (PMUs)
distributed across the transmission system

Synchronized (<< ms) via GPS
30 Hz sampling of voltage, current, frequency & phase Aggregated at Phasor Data Concentrator (PDC) wide area situational awareness
104
11/7/2012
The power of observability

Power Systems Applications

Enhanced state estimation, Operator visualization Black Start visibility, Line impedance derivation Oscillatory mode detection & damping Phase Difference before the Blackout Post-disturbance analysis, Islanding Power network model validation
Aug 14, 2003
105
11/7/2012
Currently ~500 PMUs in US grid
106
11/7/2012
The issues to solve?

Sensor placement
Huge complex grid, few expensive PMUs
Time synchronization
Higher delity requires tighter synchronization Especially as we move to the distribution tier
Latency Data rates (!), storage, query Continuous, unattended operation

Especially during crises
Analytics, Prognostics,
107
11/7/2012
DIY Synchrophasors
108
11/7/2012
What happens when ?

Push observability throughout distribution tier? Into consumption tier ? Electrify more and more of the transportation ? The loads become ecient, power
proportional, and grid responsive?
109
11/7/2012
From a Grid to an Energy Network

Availability
Pricing Planning
Source IPS energy subnet Intelligent Power Switch Load IPS
Forecasting
Tracking Market
Monitor, Model, Mitigate

Deep instrumentation Waste elimination Efficient Operation Shifting, Scheduling, Adaptation
110
11/7/2012
IT and the 4 Part GHG Reduction Plan

Eciency Electrify Decarbonize Decarbonize
Monitoring, Analysis, Modeling, Waste Elimination, Power Proportionality, Optimal Control Intelligence, Communication, adaptation in Everything ZELB. Supply-Following Loads, Energy SLA, Cooperative Grid
the electricity the fuel
111
11/7/2012
What made me think it was time for WSN reseatch?
112
11/7/2012
Why the next tier of internet is here?
113
11/7/2012
Thanks
114
11/7/2012

Keynote Sensys 11-7-2012

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Keynote Sensys 11-7-2012

Hochgeladen von

Copyright:

Verfügbare Formate

Sustainable Energy Networks - a Sensys grand opportunity

David E. Culler University of California, Berkeley

Sensys 2012 Keynote

be more engaged in achieving sustainable energy networks

Quantifying Sustainability California Law

Reduce GHG emissions to 1990 levels by 2020

Governors executive order S-3-05 (2005)

Renewable Portfolio Standard

480 => 80 mmT CO2e in 40 years

CA2050: GHG 90% below 1990

1,000 800 600 400 200 0

CA2050 4 Part GHG Reduction Plan

CA2050 4 Part GHG Reduction Plan

CA2050 4 Part GHG Reduction Plan

the electricity Decarbonize the fuel

CA2050 4 Part GHG Reduction Plan

the electricity Decarbonize the fuel

All required for even 60% reduction

but still fall short of 80%

Resources exist - 1.4 % of CA land - 43% agriculture - 3.4% urban

The Problem: Supply-Demand Match

transmission 6m miles of distribution 3k miles of 500 kv DC

distributors 147 M customers

10 ISO/RTO cover 2/3 Little communication

Towards an Aware Energy Network

Aware Interactive Loads Communication Communication

Classical view of the Energy Challenge

Electronics Appliance Lighting HVAC Motors

Geothermal Hydro Nuclear, Coal Generators

Modern Energy Network Challenges

Storage Fluctuating Renewable

Personal Electronics Appliance Facilities

Lighting HVAC Motors Industrial

Nuclear, Generators Coal Physical Plane

Power Supplies VFD

Circuits Control Efficiency

Sensys? Creating the info plane

Smart Meters the WSN Killer App?

Electronics Appliance Lighting HVAC Motors

Geothermal Hydro Nuclear, Coal Generators

Smart meter rollouts

Much more than meters and billing

Key WSN Research Developments

FF lus h lus h Golden Ga te Bridg Golden Ga te Bridg ee

R edwoods R edwoods North ea North SSea

MintR oute MintR oute

CP -MAC SSCP -MAC PP E DAMACS E DAMACS X-MAC X-MAC

Wis eMAC Wis eMAC

-MAC SS-MAC Aloha S Aloha PP S

All the energy is consumed by listening for a packet to receive

=> Turn radio on only when there is something to hear

Reliable routing on Low-Power & Lossy Links

=> monitoring, retransmission, and local rerouting

Connectivity is determined by physical points of interest, not network

MultihopLQI MultihopLQI FF us ion us ion

Tra ns port Tra ns port

The Mote/TinyOS revolution

Intel Intel/UCB rene dot

04 48 kB rom 10 kB ram 802.15.4

Idle Listening: 3 Basic Solutions

Listen for very short intervals to detect eminent

Listen after send (with powered infrastructure)

Many hybrids possible