Because no two buildings are alike, IoT Building Insights uses AI (Augmented Intelligence) and Brick Uniform Building Metadata Schema to deliver an up-to-date and accurate view of every building in an enterprise. Whether it be for energy diagnostics and prediction or future occupancy insights, IoT Building Insight’s platform is designed to consolidate, store, analyze, and learn from the things and people in buildings.
2. Today
Global Real estate is valued at over
$250 Trillion, with $7 Trillion added
annually. This represents 13% of
global GDP, increasing to 15% in 2020.
Watson IoT / Building Insights / June 20182
Tomorrow
Smarter Building Investments will reach
$30BN annually by 2022 with
estimates of over a billion sensors
deployed.
Historically
Human beings spend 90% of their
lives indoors.
3. Operations consume 70% of
the total cost of ownership
(TCO) for a facility.
Of that, the largest
component is utilities; with
commercial buildings
consuming nearly 40% of
Total U.S. energy
26%
24%
17%
18%8%
7%
Utilities
Repairs and maintenance
Administrative
Cleaning
Security
26%
Utilities
24%
Asset Maintenance
and Repairs
25%
Administration and
cleaning
Source: BOMA 2016 Office Experience Exchange Report (Office EER)
Watson IoT / Building Insights / July 20183
4. The Evolution of Building Management
Automated buildings
(1980 – 2000)
Smart buildings
(2000 – 2015)
Cognitive buildings
(> 2015)
Phased investment has evolved infrastructure, rate of change is accelerating
Visualize KPI
Good for ratings
Allows identifying general issues
Bad for identifying energy waste
Analyze energy consumers
Understand consumption of rooms
and central assets
Only primary datapoints are analyzed
Learn behavior
Predictive control down to desk level
Understand energy flow and building occupancy
Consider comfort preferences of users
Collect context such as weather and meetings
Too many data points even for advanced analytics
Watson IoT / Building Insights / July 20184
5. Light
Utilities
Air quality
Noise
Hot desks
Power
consumption
HVAC
Meeting rooms
Temperature
Smart meters
Energy
optimization
Staff
productivity
Space
occupancy and
optimization
Predictive
maintenance
Energy optimization is a leading smart building demand
Aggregate and
store data
Watson IoT / Building Insights / July 20185
6. The challenge in IoT
?
We have:
Billions of different IoT devices
Buildings creating Zetabytes of data
We want:
Analytic workflows
distributed across the fog
Watson IoT / Building Insights / July 20186
7. Putting AI to work for Energy Management
Artificial
Intelligence
Augmented
Intelligence
Machine
Learning
Natural
Human
interaction
Reason
insights &
actions
Watson IoT / Building Insights / July 20187
8. Energy Management without AI
⎯ Manual approaches require experienced operators and scale poorly
⎯ Rule-based approaches require deterministic behaviour and maintenance
⎯ Statistical approaches do not consider exogenous variables
⎯ Data mining models are black-box models that are hard to interpret and trust
Analysing energy data is promising to reduce building operation expenses. But, energy consumption
is influenced by many aspects and it is hard to detect and diagnose abnormal consumption:
?Building
Data
How can I
resolve abnormal
behaviour?
How can I detect
abnormal behaviour?
Watson IoT / Building Insights / July 20188
9. 9
“We can’t make
sense of our smart
building data.”
• Drowning in data - We have invested
in sensors and smart meters but the
is raw and not meaningful.
• Skill siloes - Our energy managers
aren’t data scientists and our IT
professionals aren’t energy experts.
• Data siloes - We have multiple legacy
systems that do not communicate
each other or new IoT technology.
• Enterprise scalability - Only one
building is ‘smart’ or a few floors
our buildings are connected, but we
know how to scale.
• False positives / manual changes –
We have a rules-based energy
that sends unreliable alerts and
manual adjustments.
• New investment – We’ve made
significant investments in new
designed for performance not energy
efficiency.
• Spreadsheet jail – Reporting on
energy consumption is time
only occurs quarterly or annually, and
doesn’t provide detailed insights
to make impactful decisions.
Pain point caused by energy
management without AI
10. Introducing IBM IoT Buildings Insights
Creating Proactive Insights with AI and IoT
Watson IoT / Building Insights / July 201810
Understanding
Get full view of all
buildings, assets, and
sensors across entire
enterprise
Reasoning
Know which assets are
consuming excessive
energy, as well as when
and for how long the
spikes occurred
Learning
Know which buildings are
consuming the most
energy and use AI to
predict future
consumption
Make a long term investments that can scale and extend to any building, IoT technology,
and system
12. 12
A fully cloud based offering that connects:
Legacy building systems (BMS, IWMS)
IoT sensors
External solutions
in every building across an entire enterprise.
IoT Building Insights is…
Responsive web application that…
displays AI driven data insights in an user-friendly interface,
which building owners and operators can leverage within
improvement action plans to reduce costs and improve
occupancy experiences.
The first release of this application focuses on energy.Watson IoT / Building Insights / July 2018
14. Collect
Client building
sensors connect and
send data to Watson
IoT Platform
Lighting
Chiller
HVAC
Refrigerators
Connect
A knowledge graph of
client buildings and
assets with KITT
semantic metadata
layer
Understand
External
Systems
Any energy sensor connected to
a main meter or sub meter
Key Capabilities
Release 1.0
IoT Partners,
e.g. Weather
Company
Building
Management
Systems
Visualize
Release 1.0
Store & Archive
Data is securely
stored and archived
Learn
Analytical models
learn the knowledge
graph, identify
anomalies, and
predict energy usage
IBM IoT Building Insights
Watson IoT / Building Insights / July 2018
16. IoT Building Insights works by…
Creating a knowledge graph of each building in an
enterprise…
https://www.youtube.com/watch?time_continue=2&v=aj9dU3vtWiQ
Using AI models to diagnose and predict abnormal
behavior in the building and its systems…
Watson IoT / Building Insights / July 2018
17. Enterprise level view
Know which buildings are
consuming excessive energy
Learn if the current energy
consumption of the building is
more than the same 30 days
during previous year
Based on current issues and
previous predictions, know how
much energy is being wasted
Predict if energy consumption will
increase or decrease in the next 48
hours
Know if any sensors or smart
meters are not sending data
Easily view and drill down into
individual buildings consuming
excessive energy
Watson IoT / Building Insights / July 2018
18. Building level view
Know which assets, e.g. refrigerator,
HVAC, etc. are consuming excessive
energy
Learn how long the unusual behavior
occurred, could indicate ‘spikes’
Know how much the building is
contributing to total enterprise
consumption
Learn if the current energy consumption of
the building is more than the previous year
Based on current issues and previous
predictions, know how much energy is
being wasted
Predict if energy consumption will increase
Know if any sensors are not sending data
Easily view and navigate to other buildings
consuming excessive energy
Watson IoT / Building Insights / July 2018
19. IBM IoT Buildings Insights allows you to…
Measure real-time energy consumption and run diagnostics
⎯ Overall energy health of real estate on any date in last month, last year or previous year
⎯ Average energy consumption for all sites v/s average energy consumption for specific site
⎯ Detailed information on metering outage / site outage
⎯ Energy consumption across entire enterprise and specific building
⎯ A correlation of energy consumption with weather data
⎯ Energy consumption status of the enterprise on a map, with alerts to focus on sites that need attention
⎯ Details of assets causing excessive energy consumption
Access AI models for energy prediction
⎯ Predicted energy consumption across enterprise for next 48 hours and next month
⎯ Predicted energy consumption at specific for next 48 hours and next month
Get Prescriptive Information on energy waste and cost avoidance
⎯ Overall energy savings across enterprise and a building for last month and last six months
⎯ Overall energy wastage in enterprise and a building for last month and last six months
Watson IoT / Building Insights / July 201819
20. Strong investment in montioring infrastructure of their stores to
make them more energy efficient
160
Stores
110.956
Data Objects
678TB
Data per annum
£20M savings pa
20%
reduced cooling
Case Study: Large Grocery Store
Watson IoT / Building Insights / July 201820
22. What is IoT Building
Insights?
A cloud based platform that
connects legacy building systems,
modern sensors, and external
products AND a responsive web
application that reveals AI driven
insights in all buildings across an
entire enterprise.
22
IoT Buildings Insights puts AI to work
How does the AI work?
It creates a knowledge graph of
each building, which understands
large datasets of concepts and
objects in a building and how they are
related, then AI models use the
knowledge graph, as well as data
coming from IoT sensors and building
systems to detect and predict
abnormal behavior.
What does it do?
It connects data from sensors, securely stores and archives it,
creates a knowledge graph with that data and other data
coming from external sources and building management
systems, uses AI models to learn from the data, and
presents visualizations of the data analysis and predictions
from the AI models.
Watson IoT / Building Insights / July 2018
23. CONNECT with us
#1 #2
REACH out to your IBM
contact
#3
What should I do NEXT?
[put your own twitter/linkedIn account here]
ASK questions now
#4
CHECK us out
https://www.ibm.com/us-en/marketplace/iot-building-insights
Watson IoT / Building Insights / July 2018
Lighting accounts for nearly 20% of electric bills in commercial buildings and price increase 2-3% CAGR
IBM believes that automated and smart buildings are increasingly giving way to cognitive buildings. In the 1980s and 1990s, building automation allowed real estate and facility management teams to visualize their buildings’ key performance indexes through dashboards.
This helped to understand overall trends, and allowed better rating of buildings for energy reporting, for example. However, these dashboards were static, historical and aggregated, and did not provide actionable insight. They could tell us which buildings produced most waste, but not why, or what to do about it.
As technology has matured and the penetration of instrumentation and analytics has increased, smarter buildings have become the norm. In the period 2000 – 2015, it has been possible to link sensor specific information with analytical tools to create actionable insights at the room and asset-specific level. However, as it is only possible to analyze primary data points, and as few organizations have implemented tools to be able to analyze large amounts of unstructured data, insights are still at an aggregate level and limited to comparisons with historical metrics.
IBM believes that the era of cognitive buildings is now here. Cognitive buildings autonomously integrate IoT devices and learn system and user behavior to optimize performance and have three core capabilities:
The ability to provide insight, from applying advanced analytics to near real-time data from IoT sensors and instrumented devices, enriched with contextual information from application programming interfaces (APIs), such as weather data.
The ability to learn, reason with purpose and interact naturally with humans, with speech and text. Buildings learn normal operational patterns based on weather, occupancy and history, and can autonomously identify and diagnose abnormal patterns and recommend an action to be executed by a human, robot or system. The recommendation is added to the knowledge base and its effectiveness will be used to further improve the recommendation of future actions.
The ability to act and deploy changes to building operations, within agreed boundaries and delegations, subject to human intervention and override.
What we are hearing from our are 4 key areas where they want to focus their IOT activities, because that is where they see the most value. The first 3 are very tangible and quanitifiable, and they include Energy Management, Space Management and predictive maintenance. The 4th as described previously is not so easy to quantify, but it generally agreed that employee productivity IOT use cases are the key to ensuring the attraction and retention of the top class employees, whilst ensuring that they constantly, deliver output to the best of their ability.And we do this by using the same IOT data that delivers the operational efficiencies and making it available in a format that is useable and actionable for the employees. This involves using the WIOT platform to ensure that the disparate data sources are bought into one place where the data can be managed, analysed and acted upon in a secure platform.Organizations are at dfferent stages of their journey in the cognitive era. Early adopters have digitized processes, instrumented buildings and assets, and have learned from cognitive pilots to make operations more efficient, responsive and agile.They understand their end users’ customer experiences and are working on initiatives to improve satisfaction and loyalty. These early adopters are launching new services to capture new revenue streams and are making plans to enable and rebalance their workforce to make them ready for the cognitive era, equipping them with the needed solutions and tools. Accordingly, they have also thought through and are realigning delivery models and ecosystems.
IBM Research developed an energy efficiency diagnoser as one example for a complex analytics. Energy becomes an important goal in building operation and is often required to target specific energy goals in daily operation. However, even for experts it is hard to predict the correct energy consumption as it depends on multiple influences such as the operation hours, occupancy and occupant behavior as well as external variables such as public holidays or the weather. This renders also common approaches such as rule systems or statistical outlier detection are not able to deal with this flexibility. Data-mining models on other hand consider these effect but provide only black-box models that are not interpretable by an operator, which is why he also does not trust the predictions.
Energy wastage: The offering creates what is essentially an energy signature of each asset and compares it to the energy signature of each system. By doing this, IoT Building Insights identifies which assets
are wasting energy. For instance, in the case of one client, some systems had a bad habit of switching to manual mode after an error. It was a very subtle glitch that was only able to be detected with the insights gained through IoT Building Insights.
Visualize: For the first time, clients will be able to visualize the refrigeration systems, HVAC, lighting, IoT sensors, etc. as a whole. This offering stitches every system
together and reporting on their vitals in real time. This means the client report and predict energy consumption for every system in every building.
IoT Building Insights connects data from sensors, securely stores and archives it, creates a knowledge graph with that data and other data coming from external sources and building management systems, uses AI models to learn from the data, and presents visualizations of the data analysis and predictions from the AI models.
Using “Uniform Metadata Schema” for buildings (called Brick), the knowledge graph understands large datasets of concepts and objects in a building and how they are related (drastically reducing need for human interaction).
AI models are combined with the knowledge graph and learn from historical and sensor data, as well as information from building management systems, while normalizing it for weather, seasonality, time of day, occupancy, etc.
This client was part of the prototype for IOT Building Insights
By acquiring and harvesting the readily available data from one store and applying big data statistical approaches such as predictive analytics to a large data set, Tesco Ireland was able to validate refrigeration performance and identity case anomalies against standard policies and control strategies within its chain of grocery stores. This deeper knowledge not only yields significant energy reduction and savings in the short term but can also have a positive impact on future refrigeration control strategies.
Real Business Results
Tesco Ireland gained clear results demonstrating that optimizing the performance of its in-store freezers would help it save a significant portion of its total energy costs. Consequently, ensuring that its freezers operate at the right temperature helped the company cut refrigeration energy costs by up to 20 percent, which could realize savings of USD25 million a year across the wider UK and Ireland estate. Maintenance teams in the future should be able to assess the freezers remotely, diagnose any problems and arrive onsite with the right parts for repairs.
KITT: Will benefit clients by connecting the dots between assets, values, location, measurements, etc. It helps them see the bigger picture of how all their individual systems and devices are connected, which in turn helps the client understand
what things are driving the most energy spend.
WEATHER: Other competitors in this space do not have out-of-the-box access to weather data. With this data the offering is able to give a client a precise picture of how weather was affecting energy consumption. Linking weather data to energy models allows us to predict how much energy will be used and which systems are struggling under those conditions, particularly on hot days when the systems have to work harder
to keep up. More energy to keep things cool on hot days? Seems like a no brainer. But when you don’t have a solution capable of connecting the dots, even the obvious is hard to see from a store level, let alone enterprise level.
AI models for predictive analytics: For many IBM clients faults in a system happens at annual audits. By using AI models, IoT Building Insights can spot indicators of potential failure in near real time. By spotting the tiniest anomolies in energy consumption patterns, clients will have insights at his or her finger tips, allowing them to be proactive before those tiny anomalies became bigger problems.