Siemens Building Technologies cooperates with building owners, tenants and facility managers on creating digital twins of buildings and connecting data from building devices to AWS cloud. Their platform digitaltwin.siemens.com is one of Siemens’ key projects and built jointly with AWS Professional Services. The digital twin of a building is a combination of the digital construction twin (i.e. the digital floor plans of the building), the digital product twin (i.e. the digital twin of the products installed in the building) and the digital performance twin (i.e. the timeseries data from the installed devices which is mapped to the floor plans). The customer project is in early adoption/pilot phase. Among the first buildings connected are Siemens’ Global Headquarters in Germany. Further expansion in Asia and Americas are on the roadmap for 2019. With this project Siemens shows how static data (floor plans) and dynamic data (device timeseries) can be connected to offer value-added services to customers, like optimization of energy efficiency and square meter usage, situational awareness during/after intrusion or fire alarm, utilization-based cleaning scheduling, or optimization of canteen visit times. By connecting buildings to the AWS cloud, Siemens gives a guarantee of 50% reduction in district heating, 20% maintenance cost reduction, 20% higher occupancy rate, and 15% increased customer satisfaction. AWS services used: AWS IoT, Amazon Kinesis, Amazon Neptune, Amazon AppStream 2.0, AWS Lambda, Amazon ECS, Amazon EC2, Amazon S3.
Designing IA for AI - Information Architecture Conference 2024
Connecting Buildings with AWS
1. AWS Pop-up Loft Berlin 2018
Page 2
Siemens Digital Lifecycle Platform
Targets
• Add 500,000 buildings to Digital Lifecycle Platform
• Engineering process improvement by 50%
• Faster tendering process to win more customers
2. AWS Pop-up Loft Berlin 2018
Page 3
What is the business of
Siemens Building Technologies (BT)?
5. AWS Pop-up Loft Berlin 2018
Page 7
Does digitalization matter in
Building business?
6. AWS Pop-up Loft Berlin 2018
Page 8
Why the construction industry needs IoT
Productivity increase in the Construction industry1 Customers know – Digitization will affect every process2
Source: 1 McKinsey Global Institute “Reinventing Construction”; February 2017 | 2 Siemens customer survey, 2014, 2015
of customers want
visualization of data80%
of customers expect an
improved service process69%
of customers location-
independent access to their data65%
of customers expect new digital
services and business models50%
7. AWS Pop-up Loft Berlin 2018
Page 9
Building IoT Data Market
Building in transformation1 Big data analytics2
Outcome economy3 Connected World4
Source: 1 Memoori Report 2016 | 2 World Economic Forum 2016 | 3 Memoori Report 2016 | 4 World Economic Forum 2016
60%annual growth in data collected form smart
buildings year over year, volume doubles every two years
50%of the world’s data, in the history
of mankind, was created in less than the last year
Building IoT Market will grow from US$23.5 bn in 2015 to
US$75.5 bnin 2021 with 20.7% CAGR
8 billiondevices connected to the Internet
today; by 2030 it is forecast that there will be 1 Trillion
8. AWS Pop-up Loft Berlin 2018
Page 10
Building Data
Example Office Building
~200 Gigabytes static data
~60 Sensor types
~2,000 Datapoints
>500 MB data per day or ~200 GB per year
Siemens Requirements
500,000 commercial buildings 100 Petabytes per year
~30 years operation phase 3 Exabytes
Availability in 4 regions, 160 countries
Access for 12,000 service engineers and millions of users
16. AWS Pop-up Loft Berlin 2018
Page 18
Creating perfect places based on Services – User Centric focus as a
holistic approach to the modern workplace …
Customer Interest Relevant KPI’s
Cost per space unit
Employee satisfaction
CO2 emissions
Employee productivity
Optimizing CAPEX and OPEX
Energy and
asset efficiency
Space
efficiency
Individual efficiency
and comfort
Workplace Utilization
Revenue per space unit
Vacancy Rate
Asset Performance/Useful Life
17. AWS Pop-up Loft Berlin 2018
Page 19
How do you create a
Digital Twin of a building?
18. AWS Pop-up Loft Berlin 2018
Page 20
BT Digital Twin – Four steps from isolated data silos
to an integrated building knowledge base
Digital Twin APIs
Data
Federation
Rules and Predictions
Knowledge Graph
4
3
2
1Access
Link
Enhance
Provide
• Rules and constraints
• Machine learning
• Graph databases
• Semantic data models
• Virtual or physical data integration
• Query rewriting
• (Micro) Services
19. AWS Pop-up Loft Berlin 2018
Page 21
A semantic data model enables flexible linking and an integrated,
intuitive API for applications
Semantic Data Model 2
3
1
new new
Planning
tools
Simulation
tool
Building
Operations
Security aaS Building
Performance
Service Portal Location Ba-
sed Services
Novel BT
Services
3rd party
Services
Connect
to data
Drive
applications
Knowledge
Graph
ETL or
Virtual Integration
Customer Data
Weather Data
Public Energy Data
MindSphere (PoC)Building Structure
(e.g. BIM IFC)
Product Data
www
4
Data API
IFC import/
GraphicsAPI
Product API
20. AWS Pop-up Loft Berlin 2018
Page 22
Amazon Neptune Graph Database maintains the logical data model
and links to other data sources
Load Property Graph and RDF Data Store billions of relationships Fast graph queries
https://aws.amazon.com/de/neptune
Amazon S3
Property Graph
CSV
Resource Description
Framework (RDF)
Turtle
N-Triples
N-Quads
RDF/XML
Bulk
Load API
SPARQL
Endpoint
Gremlin
Server
21. AWS Pop-up Loft Berlin 2018
Page 23
Translate structural building information from IFC to a standard
semantic data model supported by Amazon Neptune
IFC OWL OntologyIndustry Foundation Classes (IFC)
• File format with an object-oriented data model
• Open standard ISO/PAS 16739 to increase interoperability
• Developed by buildingSMART
• Available converter for .ifc files to W3C RDF (IFC2RDF)
22. AWS Pop-up Loft Berlin 2018
Page 24
Example Query – Combine structural and device information
List all Siemens
devices in break
rooms
23. AWS Pop-up Loft Berlin 2018
Page 25
One query, multiple sources – The semantic federation system
provides an integrated mechanism
SELECT ?object ?temperature ?timeStamp
FROM NAMED :building_A
WHERE {
?space rdf:type ifcowl:IfcSpace .
?space ifcowl:description_IfcRoot ?descr .
?descr express:hasString "BREAK ROOM" .
?contained_in ifcowl:relatingStructure_IfcRelContainedInSpatialStructure
?space .
?contained_in ifcowl:relatedElements_IfcRelContainedInSpatialStructure
?object .
?object rdf:type :MultiFireDetector .
?object iot:ID ?object_iot_ID .
?object iot:propertySet ?object_prop_set .
?results iot:hasTemperature ?temperature .
?results iot:timeStamp ?timeStamp .
SERVICE mdsp:MindsphereTimeseries {
?results mdsp:timeseriesAPI :_blank
:_blank mdsp:iotID ?object_iot_ID .
?api mdsp:propertySet ?object_prop_set .
}
}
Local Graph Store
Timeseries REST API
Named
Graph
:building_A
Default
Graph
Named
Graph
:building_A
Mindsphere API Endpoint
https://somegateway.com/iot/{propertySet}/{iotID}
24. AWS Pop-up Loft Berlin 2018
Page 26
Outlook – The graph infrastructure provides services to ensure data
quality of the integrated building knowledge base
Data validation, completion and
prediction services are implemented
using rules-based system, constraint
solving and machine learning.
Data validation
services
Data completion
services
Link prediction
service
Data Connectors
Digital Twin API
Example: Device Placement Rules
IFCSITE(?site), locationInside(?site, „Germany“),
composedOf(?site, ?building), IFCBUILDING(?building),
composedOf(?building, ?storey),
IFCBUILDINGSTOREY(?storey),
composedOf(?storey, ?space), IFCSPACE(?space),
Name(?space, „Treppenhaus“),
Area(?space) > 5
IFCRELSPACEBOUNDARY(?space, ?boundary),
IFCRELFILLSELEMENT(?space, ?relfillselement),
IFCOPENING(?relfillselement, ?opening),
IFCRELVOIDSELEMENT(?opening, ?wall)
Device(?dev), name(?dev, „FDMH291-R“), inRoom(?dev, ?space), attachedTo(?dev,
?wall)
Domain Knowledge
25. AWS Pop-up Loft Berlin 2018
Page 27
What is the value of
Digital Twins?
26. AWS Pop-up Loft Berlin 2018
Page 28
Digital Twin and Knowledge Graph enables Data Analytics and
Machine Learning
• Use exisiting HVAC sensors to predict room
occupancy
• Data Mining
• Visual Analysis
• Descriptive Statisitcs
• Machine Learning
• Get more information out of exisiting sensor data
• Improve data quality
• Detect anomalies
• Enable predicitve maintenance
• Combine knowledge of building structure with
sensor location and time series data
27. AWS Pop-up Loft Berlin 2018
Page 29
Visualizing Timeseries Sensor Data
29. AWS Pop-up Loft Berlin 2018
Page 31
Building Model Synchronization
30. AWS Pop-up Loft Berlin 2018
Page 33
How do you integrate Digital Twins
into an IoT platform?
31. AWS Pop-up Loft Berlin 2018
Page 34
MindSphere IoT Platform for Building Technologies
MindApp
Building
MindSphere
connected devices
Gateway
MindConnect Integration
TimeseriesAsset
Management
MindApp Backend
32. AWS Pop-up Loft Berlin 2018
Page 35
Siemens Building Technologies Cloud-based System Concept
Owner Operator Tenant Visitor TechnicianPlanner …others
IFC
BACnet devices
Digital Construction Twin
(Building structure data)
MindSphere
Digital Performance Twin
(Events and Time series data)
Semantic Link (Graph database)
(performance & structure data)
Construction Data API Performance Data API
Amazon S3 storage Amazon Neptune
MindApp:
Space
utilization
33. AWS Pop-up Loft Berlin 2018
Page 36
What challenges did you face
when connecting buildings?
34. AWS Pop-up Loft Berlin 2018
Page 37
Challenges with onboarding buildings
• Unclear responsibles
• Connectivity approvals
• No device naming conventions
• Data polling vs. streaming
• Heterogeneous device landscape
• Device misconfigurations
• Determine device position
35. AWS Pop-up Loft Berlin 2018
Page 38
Call out
We want to hear your smart ideas for
automated geo positioning of installed devices!
36. AWS Pop-up Loft Berlin 2018
Page 39
How did you manage the move from
innovation to production stages?
37. AWS Pop-up Loft Berlin 2018
Page 40
Moving from Innovation to Production stages
Principles that helped in the Innovation stage:
• „No Regrets Move“ corporate project culture
• Promote agile development practices
• Iterate fast to get pilot customer feedback
Principles that helped to scale to Production:
• Promote DevOps culture
• Establish best practices for infrastructure management
• Implement CICD/stages
• Automate everything (incl. testing)
38. AWS Pop-up Loft Berlin 2018
Page 41
Markus Winterholer
Innovation Manager / Chief Product Owner
BT SSP TIA TI
Siemens Schweiz AG
Theilerstrasse 1a, 6300 Zug, Switzerland
Phone: +41 798540007
E-mail:
markus.winterholer@siemens.com
Internet
buildingtechnologies.siemens.com
Intranet
intranet.siemens.com/bt
39. AWS Pop-up Loft Berlin 2018
Page 42
Don‘t miss the following AWS Loft sessions
40. AWS Pop-up Loft Berlin 2018
Page 43
Don‘t miss the following AWS Loft sessions
41. AWS Pop-up Loft Berlin 2018
Page 44
Don‘t miss the following AWS Loft sessions
Our customers embark with us on journey of creating perfect places and
bringing buildings into the digital age –
everywhere in the world and in relevant vertical markets
(animation goes)
And talking about real estate now, I would like to welcome Dr. Sluitner on stage,
the head of a large real estate business, our own Siemens corporate Real Estate unit,
also one of our important customers
some facts about buildings and their environment.
80%: chance to address those costs also after buildings are constructed.
30%: optimizing space of course, interested to provide the users with positive and productive working space.
Such workplaces are especially important to millennials (50%)
All together = huge potential to optimize buildings to become perfect places.
Asset Tracking
ProblemMedical staff spends significant time looking for hospital equipment, leading to overstocking and low employee productivity
Solution Digital construction twin as calculation, asset type and user interface basis, different tracking technologies and asset data as input for the digital performance twin
Space utilization
ProblemLow efficiencies due to lack of good occupancy and usage data
Solution Digital construction twin as calculation and user interface basis, occupancy sensors based on different technologies as input for digital performance twin
Load Property Graph and RDF Data
You can import data from S3. For the Resource Description Framework (RDF) graph model, Neptune supports Turtle, N-Triples, N-Quads, and RDF/XML serializations. For the Property Graph model, Neptune supports a CSV format
Store billions of relationships
Amazon Neptune is a fast, reliable, fully managed graph database service that efficiently stores and navigates highly connected data. Its query engine is optimized for leading graph query languages, Apache TinkerPopTM Gremlin and the W4C’s RDF SPARQL
Fast graph queries
You can easily build and run applications that work with highly connected datasets using a SPARQL endpoint for RDF or a Gremlin Websocket Server for Property Graphs. You can provision up to 15 low latency read replicas for high throughput applications
Endpoint url flexibel halten?
Der Einsatz verschiedener Werkzeuge und Services im Bereich der Datenanalyse und Machine Learning, wie das Python scikit-learn Framework oder die Webservices Elasticsearch, Kibana und AWS Sagemaker, ermöglichten ein umfassendes Bearbeiten der Data Mining Problemstel-lung. Als besonders entscheidend erwies sich die Phase «Assess model» im ML-Modelling Prozess. So ist ein strukturiertes Vorgehen, wie auch die Anwendung eines geeigneten Testdesigns (vgl. k-folds Crossvalidation) beim Vergleich der Modell Qualitätseigenschaften zentral um fundierte Erkenntnisse über die Qualität der entwickelten Modelle zu gewinnen. Als überraschend positiv wurde die einfache und schnelle Verwendung vieler ML-Frameworks empfunden. Jupyter Notebooks stellten sich als eine unverzichtbare Hilfe im Data Preparation und Modeling-Prozess heraus
some facts about buildings and their environment.
80%: chance to address those costs also after buildings are constructed.
30%: optimizing space of course, interested to provide the users with positive and productive working space.
Such workplaces are especially important to millennials (50%)
All together = huge potential to optimize buildings to become perfect places.
Architekur NRM3
Mapping auf Physikalische Gebäudedaten