© 2007 IBM Corporation
IBM Research
© 2009 IBM Corporation
IBM Smarter Energy Management Systems for Intelligent Buildings
June 10, 2009
Dr. Jane L. Snowdon
IBM T. J. Watson Research Ctr
IBM Research
© 2009 IBM Corporation2
IBM Research Strategic Goals for Energy
• Smart Grid
• Smart Buildings including Data Centers
• Smart Energy
IBM Research
© 2009 IBM Corporation3
Vision: End-to-End System for Smarter Energy Management
Energy Management Software
Servers, Storage, Networks
Data Center or Office Personalization
Raised Floor
Office Space
Building Infrastructure Systems
HVAC, Lighting, Electrical, phones, etc.
Building Brick and Mortar
.
Power Grid
Cloud , Virtualized Computing
Industry Specific Solutions
Power Grid
Data Centers
OfficeBuildings
Intelligent Buildings
Today:Separate Systems
Tomorrow:Fully Integrated
Systems
IBM Research
© 2009 IBM Corporation4
Buildings represent one of the largest areas for energy efficiency gains – particularly in the US and other developed countries
Source: Canaccord Adams; IBM Corporate Market Insights Analysis, EPA Report to Congress, August 2, 2007
CO2 Emissions in the US
Transportation
33%
Buildings
39%
Industrial
28%
§ Buildings in developed countries contribute approximately 40% of the greenhouse gas (GHG) emissions that are forcing climate change
– GHG contributions surge to 50% when all the energy-related factors necessary to serve buildings and their occupants are included
– Over the next 25 years, CO2 emissions from buildings are projected to grow faster than those from any other sector
Energy and Resource Consumption for All Types of Buildings in the US
39% of US primary energy use (includes production-related fuel input)
Energy
136M tons building related debris/yrWaste
40% of raw materials globally (3B tons/yr)Materials
12% of all potable water (15T gallons/yr)Water
70% of all consumptionElectricity
Commercial and Data Center increasing dramatically
Data Centers in US consumed 62 billion kWh in 2006, about $4.5B or 1.5% of the total, doubling since 2000. At current trends, it will double again by 2011
IBM Research
© 2009 IBM Corporation5
IBM Shares Mutual Areas of Interest for Research with CITRIS andIBM Shares Mutual Areas of Interest for Research with CITRIS and HiPerBRICHiPerBRIC
6 © 2009 IBM Corporation
IBM Research
© 2009 IBM Corporation6
Today:
Multiple “proprietary“ systems
24/7 Monitoring
FIRE & SAFETY
ACCESS
ENERGY HVAC
ELEVATOR
SECURITY
LIGHT
COMMUNICATION
Separate Control Systems
Future:
“One Managed Network of Optimized Systems” for all services
24/7 Monitoring
FIRE & SAFETY
ACCESS
ENERGY HVAC
ELEVATOR
SECURITY
LIGHT
COMMUNICATION
Distributed, but Coordinated, Monitoring & Control
Intelligent Building Transformation
WATERWATER
IP TELEPHONY
IP / DIGITAL TV
HIGH SPEED INTERNET HIGH SPEED
INTERNET
IP TELEPHONY
IP / DIGITAL TV
IBM Research
© 2009 IBM Corporation7
The Intelligent Energy Efficiency Building Solution project proposes to enable real-time measurement, monitoring and management of building systems through the development of new algorithms, analytics of real-time external events and building system state, and implementation of control mechanisms to better optimize energy so consumption can be reduced.
High-Speed Internet
Backbone
Wireless
Te
na
nt
Se
rvic
es a
nd
Tech
no
log
ies
HVAC sensors
Fire
24/7 monitor
Video surveillance
Lighting
Energy
Elevators
Access
Bu
ildin
g S
erv
ices
an
d T
ech
no
log
ies
Digital / IP TV
Visitor management
Touch ScreenKiosk
Audio and video conferencing
IP Telephony
Real-time external events
-Weather forecast
-Energy prices
-Human location information,- e.g., People leaving or coming to work & home
-Utility
Analytics
-Demand Management System-Are any building systems out of spec?-Gas-Water-HVAC-Pumps-Chillers
Control Mechanisms- Automatically
control the equipment
IBM Research
© 2009 IBM Corporation8
Intelligent Building High Level Architecture
Data Center Systems• IT & Infrastructure Interfaces
• Threshold Control
Building Control Systems•HVAC, Lighting
• Fire & Security Systems• Utilities & Site Services
Systems Integration and Optimization• Data Warehouse
• On-Site and Remote Operations• Resource Optimization Rules
•Fault Isolation
Premises Management
External Systems Inputs
•Utility & ServiceProviders
•Weather Systems• Emergency Services
• City Controls
Physical Systems•Sensors / Meters
• Monitors•Actuators
•Natural Systems
Resource Management•Data Correlation
•Analytics•Systems Integration
•Resource Optimizations
Business Integration•Information Presentation
•Business Optimization• Property Management
Communications Layer Communication Interconnect & Protocol Translation
Function Mgmt •Information Presentation
• Process Control
•Dashboards• Business Rules
•Decision Support •Business Optimization
Services Mgmt• Presentation Services
• User Services• Recovery Services
• Asset Services
Process Mgmt• Process Automation
• Change Control
InformationControl
IBM Research
© 2009 IBM Corporation9
Integrated Management of Energy, Air, Water for Buildings
Mobile Measurement Technology (MMT) for Buildings
§Create a robotized version capable of understanding temperature variations and flow patterns in buildings
§Detect HVAC leaks and airborne contaminants
Monitor Airborne Contaminants
§Swap re-circulating air for outside air and vice-versa
§ Integrate with shut down controls for HVAC and access control systems
§Ensure safety and health of occupants and equipment – Risk/Reward Models
Scope: City > Campus > Building
Local Weather and Damage Prediction
§Develop damage prediction model to estimate number of outages
§ Incorporate GIS infrastructure data into weather and outage models
§Support for siting and operation of renewable generation capabilities
Framework, Architecture & Platform
§Develop a framework, extensible architecture, reference semantic model, and platform to support Intelligent Building operational needs
§Supports integration of sensors, actuators, and building sub-systems
Integrated Management
IBM Research
© 2009 IBM Corporation10
Energy Management Analytics
Energy Procurement Planning
§Select optimal supply of energy blocks (size, duration, time-of-use)
§Based on stochastic optimization over energy demand forecast and energy market prices (futures and spot)
Green Portfolio Planning
§Plan investments in green assets over a time-line
§Based on analysis of trends in demand, energy price, and technology maturity under uncertainty
Scope: City > Campus > Building
Demand Response Elicitation
§Software platform for estimating demand shifting and shedding as a function of control signals (info, price)
§Statistical analysis of demand elasticity under various load attributes
Dynamic Demand Management
§Schedule discrete loads (water pumping, PHEV charging, energy storage charge/discharge)
§Generate control signals (info, price, caps) for optimal demand shaping
Energy Usage Model
IBM Software Group | Information Management Software
11© 2008 IBM Corporation
New! InfoSphere Streams analyzes extreme volumes of structured and unstructured information-in-motion
All statements regarding IBM's plans, directions, and intent are subject to change or withdrawal without notice. Any reliance on these Statements of Direction are at the relying party's sole risk and will not create any liability or obligation for IBM.
Statement of Direction IBM intends to make available in the first half of 2010, an offering, IBM InfoSphere Streams, which will help customers continuously analyze massive volumes of information at extreme speeds to improve business insight and decision making. This product will be based on an ongoing stream computing project in IBM's Research Division.
Based on IBM Stream Processing Research§ Stream processing systems research in
IBM’s Watson Labs
§ Developed in conjunction with large scale government initiative
§ Research project continues in parallel with productization
Key Features§ Parallel and high performance stream
processing software platform
§ Analysis of structured and unstructured in information
§ Scalable over a range of hardware
Customers already using Streams§ Early Access Program
§ Over 10 customers in 3 continents
Extreme Volumes
Extreme Analysis
Extreme Speed
•Stream Computing is an extreme evolution in data analytics
IBM Software Group | Information Management Software
12© 2008 IBM Corporation
Single Server Deployment
InfoSphere Streams is scalable on commodity hardware from singlenode to blade centers to high performance multi-rack clusters
Adapts to changes in resources, workload, data rates
For US equity electronic trading brokerage- processing 1.6M events / second
Optimizing scheduler continually manages resource allocation
Deployment on blade servers§ Cell Blades§ Intel Blades§ x Series§ …
Deployment on High Performance Cluster
IBM Blue Gene
millisecond latency
Big Green Innovations
© Copyright IBM Corporation 2007
13
Deep Thunder – Forecasts for Weather-Sensitive Operations§Problem: weather-sensitive business
operations are often reactive to short-term (3 to 72 hours), local conditions (city, county, state) due to unavailability of appropriate predicted data at this scale
– Energy, transportation, agriculture, insurance, broadcasting, sports, entertainment, tourism, construction, communications, emergency planning and security warnings
§Solution: application of reliable, affordable, weather models for predictive & proactive decision making & operational planning
– Numerical weather forecasts coupled to business processes
– Products and operations customized to business problems
– Competitive advantage -- efficiency, safety, security and economic & societal benefit
Big Green Innovations
© Copyright IBM Corporation 2007
14
§Weather causes damage and outages
§Outages require restoration (resources)
§Restoration takes time, people, etc.
§Build stochastic model from weather observations, storm damage and related data
–Outage location, timing and response
–Wind, rain, lightning and duration
–Demographics of effected area
–Ancillary environmental conditions
§Can this model be coupled to a weather model to enable a forecast of impact and response?
Storm Impact and Response Prediction
Weather prediction
Damage prediction
Restoration time
prediction
Resource requirement prediction
Business Partner Transformation
© 2006 IBM CorporationIBM Confidential
Big Green Innovations
© Copyright IBM Corporation 2007
15
Deep Thunder Damage Prediction -- 18 January 2006
Actual OutagesForecasted Outages
Business Partner Transformation
© 2006 IBM CorporationIBM Confidential
Big Green Innovations
© Copyright IBM Corporation 2007
16
Uncertainty in Damage Prediction -- 18 January 2006
Forecasted Outages with Probability of 51 to 100 Outages
Forecasted Outages with Probability Exceeding 100 Outages
17
© Copyright IBM Corporation 2008
Thermal Analysis and Optimization
Mobile Measurement Technology (MMT)
hot air is sucked
into cold aisle
Optimize to eliminate hot spots, reduce energy consumption, and extend the life of the data center
Utilizing MMT
Hot spot at long aisle
13.0oC
54.5oC
33.7oC
@ 5.5 feet
Ø Unique cart-based design for measurement collectionØ Creates a High Resolution, 3D temperature map of the
data center – real measurements not CFD guessesØ Samples 100+ temperatures at one time and thousands
of temperatures in the data centerØ Monitors positionØ Surveys large areas in a short time
18
© Copyright IBM Corporation 2008
cold aisle
hot aisle z=0.5 feetz=1.5 feetz=2.5 feetz=3.5 feetz=4.5 feetz=5.5 feet
Hot spot at long aisle
z=6.5 feetz=7.5 feetz=8.5 feet
13.0oC
54.5oC
33.7oC
yz
x
- rapid survey capabilities using MMT (5000 square feet per hour)- post-processing and thermal models for DC optimization
Mobile Measurement Technology (MMT): Thermal Scans
Confidential © 2005 IBM Corporation19
2009 Energy and Environment Conference
Enhancement of MMT tool§ robotization of MMT cart for autonomic data collection
§ enhance MMT sensing capabilities
§ integrate laser scanners for 3D layout mapping
§ implement barcode & RFID scanning
(air flow, acoustics, other environmentals)
Robotize MMT and enhance sensor technologies –apply to buildings
Real-time sensor systems and instrumentation§ further develop wired sensing solutions (one-wire protocol)
§ develop appropriate (modular) sensor deployment strategies
§ develop means & techniques (RFID, barcode) for sensor management
§ wireless sensor systems (Zigbee etc..)
§ integration of MQTT-S messaging protocol
IBM Research
© 2009 IBM Corporation20
Integrated Management Energy ManagementControl of every resource with energy management software
IT Assets
SystemsStorage
Facility Infrastructure Assets
Data Center Infrastructure Assets
New IBM Servers
iPDU/External
Native
Active EnergyManager
Alerts View
Data Center Conditions
Systems Utilization
IBM Values:• Integrated Solution• Solution/Value Growth• Footprint Generation • Reduced Development
Alerts/Data/ActionsAlerts
Alerts/Data/Actions
Alerts
POC Box Data
Confidential © 2005 IBM Corporation21
2009 Energy and Environment Conference
Airborne Contaminants and MMT
§ Airborne contamination in data centers is an important contemporary problem
– Corrosion of silver to silver sulfide
– Galvanic attack of copper by silver over-plating leading to so-called “copper creep corrosion.”
– IBM Lab Services Contamination Filtration Offering Addresses these issues.
– Many chemicals and gases can be culprits in corrosion:
• Chlorine
• Sulfur Compounds, esp. Sulfur Dioxide
• Hydrogen Chloride, Hydrogen Peroxide
• Ozone
• Nitrogen Dioxide
• Ammonia
– These compounds are particularly prevalent in countries getting many new data centers
§ Same issues exist in more general indoor facilities such as manufacturing and storage facilities
– Possibly different airborne contaminants of concern in these facilities:
• Formaldehyde
• Mold
• Radon
• viral agents (e.g. hospitals)
• total particulates
• Others
IBM Research
© 2009 IBM Corporation22
Energy Management Analytics
Energy Procurement Planning
§Select optimal supply of energy blocks (size, duration, time-of-use)
§Based on stochastic optimization over energy demand forecast and energy market prices (futures and spot)
Green Portfolio Planning
§Plan investments in green assets over a time-line
§Based on analysis of trends in demand, energy price, and technology maturity under uncertainty
Scope: City > Campus > Building
Demand Response Elicitation
§Software platform for estimating demand shifting and shedding as a function of control signals (info, price)
§Statistical analysis of demand elasticity under various load attributes
Dynamic Demand Management
§Schedule discrete loads (water pumping, PHEV charging, energy storage charge/discharge)
§Generate control signals (info, price, caps) for optimal demand shaping
Energy Usage Model
IBM Research
© 2009 IBM Corporation23
1. Energy Procurement Planning
12AM 9AM Noon 6PM 9PM 12AM
Hedge for a Summer Wednesday (say)
offpeakoffpeak
midpeak midpeak
PEAK
kW
§ Energy Hedge is an “Energy Block” of a particular size, covering a particular duration & time-of-use, at a known predetermined price
§Load over and above the “Hedge” is the exposure to the “spot market”
§ How do you decide what is the optimal Contracted Base Load? (Regulated Markets)
§ Similarly, what is your optimal portfolio of “Energy Blocks” ? (Deregulated Markets)
§Operationally handle peaks using demand shifting using price signals
IBM Research
© 2009 IBM Corporation24
Optimal Energy Procurement Under Risk§ We can address the Energy Procurement Planning problem as a Stochastic
Optimization question: What is the optimal hedge sizing for the fixed price purchase component?
§ Given (Input)
– A horizon of interest (say 1 month, 1 Quarter, Half-Year, Full-year, etc.)
– Above described Regime-switching Stochastic Models for Real Time Price, Day Ahead Price, and Load, pertinent to the above horizon
– Rate Structure Details
– Candidate set of Energy Hedge-Blocks, along with business constraints, such as Minimum Block Size, Minimum Duration of Purchase, etc.
§ Compute (Output)
– The optimal set of Hedge-Blocks, with size and duration of coverage
– Real-time & Day-Ahead exposure that is recommended with the above Hedge Solution
– Such that the Overall Energy Expenditure Distribution, has an acceptablerisk of exceeding a user-defined, known, threshold (tolerance level)
§ Use Stochastic Mathematical Programming techniques to solve this problem
Energy Procurement Planning
IBM Research
© 2009 IBM Corporation25
2. Green Portfolio Planning
Green Portfolio Planning
Energy Demand• Current energy load types
and consumption pattern• Demand growth trends
Current Infrastructure• For each major load types (HVAC, pumps, lighting, etc.)• Equipment age (remaining lifetime, salvage value)• Energy efficiency
Portfolio Analysis• Timeline for asset
investments• Capacity planning for
local power generation and energy storage
Green Asset Options• Energy efficient technologies• Energy load control systems• Local renewable power generation• Energy storage options• Asset costs and efficiency trends
Energy Supply• Energy cost trends
(peak, off-peak)• Demand response rebates
and retainers• Carbon caps/taxes
IBM Research
© 2009 IBM Corporation26
Energy-star appliances
Compact fluorescent light bulbs
Geothermal heat pump
Equipment tune-up
Building weatherization
Solar thermal water heating
Analysis of Energy Efficiency
§ Identify options for improving energy efficiency
§ For each option:
– Estimate cost of upgrade
– Estimate usage (peak/off-peak)
– Estimate energy savings (peak/off-peak)
– Calculate cost/KWhr saved (peak/off-peak)
§ Compare with peak and off-peak pricing of electricity
– Current and future variability
§ Select cost effective options
$C
ost
/KW
hr
Save
d
MWhr Saved/Year (cumulative)
Projected Energy Cost
Example: Building Energy Efficiency Options
Green Portfolio Planning
IBM Research
© 2009 IBM Corporation27
Analysis of Energy Load Elasticity § Provides analytical support for
– daily peak load management
– demand response
– dynamic control systems
§ Identify loads that can be shed or shifted depending on:
– time of day
– type and level of incentives
§ Based on price elasticity models and discrete choice models
§ Model parameter extraction approach (assumes energy monitoring capability)
– Initial pilots and experiments
– Ongoing monitoring and data mining
Cooling reduction
Lighting Sweep
Reschedule pumps
Pre-cooling
Thermal Energy Storage
$C
ost
/KW
hr
Shift
ed
Peak KWhr Saved/Year (cumulative)
Cost Rebate
VFD Ventilation
Dimmable Ballasts
Example: Building Demand Response Options
Green Portfolio Planning
IBM Research
© 2009 IBM Corporation28
3. Demand Response Elicitation
§ Demand Response Metrics:
– Price Elasticity: %increase in demand per %increase in price
– Elasticity of Substitution: %shift of demand per %price difference (peak vs. off-peak)
§ Past studies are at the macro-economic level
– Skewed distribution of elasticity suggests need for microeconomic study
§ Types of demand response signals to study:
– Price
– Information of energy use and carbon emissions (comparison to anonymous peers)
– Curtailment advisories during critical peak periods (with rebate on curtailment)
§ Granularity and stratification of the analysis:
– By type of pricing contract, time-of-day, type of load, end-user profile, weather
§ Benefits:
– Evaluate effectiveness alternative demand response policies
– Parameterize energy demand model for operational decision-making
IBM Research
© 2009 IBM Corporation29
Demand Conditioning based on Dynamic Price Controls
§ The GridWise Olympic Peninsula Project
– 112 households
– Thermostats programmed to response to price signals
– Customer can choose between ‘more comfort’ and ‘more economy’
– 3 options for pricing contract:
• Fixed price
• Time-of-use (with manually initiated critical peak price periods)
• Real-Time pricing (double auction)
Results
§ Real-Time Pricing group:
– Peak decreased 15-17% (pre-heating/cooling)
– Overall consumption 4% higher than fixed-price group
§ Time-of-Use group:
– -0.17 price elasticity
– Overall consumption 20% lower than fixed-price group
Demand Response Elicitation
IBM Research
© 2009 IBM Corporation30
Software Platform for Demand Response Elicitation
Building Management System
BMS DB
Gateway
DRDB
Room nnn Room nnn Room nnn Room nnn Room nnn
Statistical Analysis
Room nnn
Web Services
Experiment Management• Set price (static or dynamic)• Set temp, light set points• Read energy use, set points• Create portal content
Demand Response Model• By time-of-day, resident type, contract, price,
ambient temperature
Supply constraints
§§ Light dimmers Light dimmers –– sense and controlsense and control
§§ Thermostat Thermostat –– sense and controlsense and control§§ Interface device Interface device –– information display, device control, preference settingsinformation display, device control, preference settings§§ Web Web –– information display, preference settingsinformation display, preference settingsB
MS
Ven
do
rIB
M S
oft
ware
Experiment Administration• Enrollment• Surveys• Contract Assignment• Student Management and
Support
Portal Server
(if web services do not exist)
Experiment Results• Demand elasticity by
contract type, time-of-day, end-use, resident type
Exp 1
Exp 2
Exp n
Demand Response Elicitation
IBM Research
© 2009 IBM Corporation31
4. Dynamic Demand Management
Operational decision-making
–Demand management: determine signal sent to energy load components
• Demand response (episodic) during critical peak periods
• Continuous demand management in response to:
– Real-time pricing
– Carbon caps
– Variability of on-site power (and stored energy)
–Scheduling of discrete energy loads:
• Energy storage
• Water pumps
• PHEV charging
• Discrete loads (e.g., washers, roof-top HVAC units)
Energy inputsImport price
Local generationCarbon cap/tax
ObjectiveEnergy cost
Carbon emissionsOpportunity cost
SignalsPrice, Info,
Alerts, Caps
EnvironmentConstraints
Energy Usage Model
IBM Research
© 2009 IBM Corporation32
Example of Decision Support Scenario for Energy Management:Decision maker imposes price premiums based on predicted impact on peak demand
Peak demand reduction
Dynamic Demand Management
IBM Research
© 2009 IBM Corporation33
Room/Zone Sense & Control
Interaction of analytics with building automation
Ethernet TCP/IP
Monitor
Network Manager/AdapterBuilding Automation Network
Programmable Controllers
Room/Zone Sense & Control
Programmable Controllers
§HVAC: Ventilation/AHU/Chillers§Lighting§Water pumps
§Temperature§Humidity§Motion
ControlDB
Energy Use Model
Energy Optimization
Energy Use Prediction
Report
Model parametersCurrent state (building, environment)
Future energy demand
Usage/Outage Alerts
Equipment Failure Alerts
Demand Response Signals
Discrete Load Schedules
Ve
nd
or
Te
ch
no
log
ies
IBM
An
aly
tic
s
§On-site generation§Storage systems
Programmable Controllers
Programmable Controllers
Forecasting§Weather§Prices§Tech trends§Demand
Data Mining§Demand response§Statistical process control
Dynamic Demand Management
IBM Research
© 2009 IBM Corporation34
WASHER 1
WASHER 2
WASHER 3
WASHER N-1
WASHER n
:
ENERGY
• Finally at an operational level (daily/weekly) how do we stay within the contracted load? (i.e. min spot purchases and control operational budgets)
• smart devices such as washing machines• given the desired time window (based on price signal) schedule washing
machines to level the load • This requires a higher level of individual commitment • Incentive alignment through pricing
• Can we rise to this challenge of energy management and bring discipline to our every day activities
Finally… Energy Reservation System?
Dynamic Demand Management
IBM Research
© 2009 IBM Corporation35
On Smart Cities we can already deliver…Public Safety
- S3 Surveillance System- Emergency Management
Integration- Deep Thunder Micro-
Weather Forecasting
Intelligent Transportation Systems
- Integrated Fare Management
- Road Usage Charging- Traffic Information
Management
Energy Management- Network Monitoring &
Stability- Smart Grid – Demand
Management- Intelligent Building
Management
Water Management- Water purity monitoring- Water use optimization- Waste water treatment
optimization
Environmental Management- City-wide Measurements- Key Performance Indicators- Scorecards- Reporting