Smart Models for Smart Cities - Modeling of Dynamics, Sensors, Urban Indicators, and Planning...

Technische Universität München Lehrstuhl für Geoinformatik

Smart Models for Smart Cities –

Modeling of Dynamics, Sensors,

Urban Indicators, and Planning Actions

Thomas H. Kolbe Chair of Geoinformatics Technische Universität München

[email protected] 29th of October 2015 Joint International Geoinformation Conference JIGC 2015, Kuala Lumpur


2 29.10.2015

Model Entities

(Resources, Objects)

Actors (Agents, Stakeholders,

Citizens)

Processes (Activities,

Actions, Flows)

City System Modeling

T. H. Kolbe – Smart Models for Smart Cities

represented by

City System


3 29.10.2015

Today: Separate Modeling by Sectors


Energ

y

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

Mo

bili

ty

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning E

co

log

y

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

Eco

no

my

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

City System


4 29.10.2015

Linking Sectors creates a Lattice of Models


Energ

y

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

Mo

bili

ty

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning E

co

log

y

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

Eco

no

my

• Commu-nity

• Models

• Indicators

• Evalua -tion

• Planning

City System


Lattice of Sector Models

29.10.2015 T. H. Kolbe – Smart Models for Smart Cities 5

► n Sectors potentially n2 connections!

► difficult to express, to maintain, and to keep consistent

Energy

Economy

. . . Ecology

Mobility


What if we could link to One Common Model?


► n Sectors n connections!

► Sector models can be linked via the Common Model

► Sector models need to be aligned with the Common City

System Model high degree of coherence required

Common City

System Model

Energy

Economy

. . . Ecology

Mobility


7 29.10.2015

Is there such an integrative model? Candidates?


City System

Common City

System Model

Energy

Economy

. . . Ecology

Mobility

repre-

sented

by


29.10.2015

Semantic

3D City Models


3D Decomposition of Urban Space

► City is decomposed into meaningful objects with clear

semantics and defined spatial and thematic properties

● buildings, roads, railways, terrain, water bodies, vegetation, bridges

● buildings may be further decomposed into different storeys

(and even more detailed into apartments and single rooms)

● application specific data are associated with the different objects

Image: Paul Cote, Harvard Graduate School of Design



City Geography Markup Language – CityGML

Application independent Geospatial Information Model for semantic 3D city and landscape models

► comprises different thematic areas (buildings, vegetation, water, terrain, traffic, tunnels, bridges etc.)

► Internat‘l Standard of the Open Geospatial Consortium

● V1.0.0 adopted in 08/2008; V2.0.0 adopted in 3/2012

► Data model (UML) + Exchange format (based on GML3)

CityGML represents

► 3D geometry, 3D topology, semantics, and appearance

► in 5 discrete scales (Levels of Detail, LOD)



Energy

Heat energy demand

Energy demand for warm water

Electric power demand

Noise immission

Noise levels on the facade

Number of inhabitants

Economy

Assessed real estate value

Provided support for rents

Information Integration within the 3D City Model

T. H. Kolbe – Smart Models for Smart Cities 11 29.10.2015


New: CityGML Model of New York City in LOD 0&1


> 1,000,000 buildings

> 866,000 land lots

> 149,000 streets

> 16,000 parks

> 9,500 water bodies

> DTM with 1m resolution

• fully-automatically generated

from the 2D geodata

published in the NYC Open

Data Portal

• semantic and geometric

transformations

• all objects have 3D geometry

• rich semantic information

(5 - 75 attributes per object

resulting from combining

different NYC datasets)

• integrated within 1 dataset! The 3D CityGML model is Open Data! Download:

www.gis.bgu.tum.de/en/projects/new-york-city-3d/

[Barbara Burger, Berit Cantzler 2015]


Web-based 3D Visualization & Data Inspection

► Using the Open Source 3DCityDB + the new Webclient

● www.3dcitydb.net & https://github.com/3dcitydb/3dcitydb-web-map


http://www.3dcitydb.net

https://github.com/3dcitydb/3dcitydb-web-map







29.10.2015

Current Challenges

in the light of

Smart City Projects


3D City Models – State of the Art + Challenges (I)

► Semantic 3D City Models

● Standardization (CityGML) provides a common vocabulary &

common ways to represent the many urban objects

● Semantic 3D city models are provided by official authorities

high reliability, stability, full coverage

● Objects of a semantic city model are a good platform to organize

and integrate data & sensors

► Today, 3D city models are mostly being used to describe

the current / a specific state of the city

● But: cities are constantly changing and there are many

dynamic aspects (moving objects, time variant attributes)

● Some of the time varying properties are provided by sensors

● Dynamics and processes not addressed much so far



3D City Models – State of the Art + Challenges (II)

► 3D City Models are used as a data source for simulations

and decision support in multiple application sectors

● these are interested in (computing) their specific indicators

● different application sectors have their own models and rules how

to compute indicator values (e.g. in the energy or mobility sectors)

► In planning & decision support it is important to have

immediate impact analyses of planned actions

● 3D City Model needs be modified according to some planned action

(like the energetic retrofitting of a building)

● Then, the (change of) relevant indicators should be derived from the

modified city model

● Planned actions mean complex transactions on the 3D city model

with specific meanings semantic modeling of actions



Modeling City Systems (MCS)

► Climate-KIC Innovation Project

► Project partners: ETH Zürich (iA, CVL), Imperial College,

TU Berlin, TU München, SmarterBetterCities, TNO, ESRI

► Project duration: 1. 1. 2014 – 31. 12. 2015 (2 years)

► EIT Funding (total): 2.4 Mio €



New Frameworks developed in the MCS Project

► General Indicator Model (GIM)

► General Planning Actions Model (GPAM)

● GIM and GPAM are based on Model Driven Engineering (MDE)

concepts defined in Software Engineering

► Dynamics in CityGML 3.0

● Two frequencies: low frequency changes evolution of the city

presentation of Kanishk Chaturvedi this morning

● Dynamic properties and behaviours of city objects (like the current

energy consumption, solar power production, traffic density)

introducing “Dynamizers“



29.10.2015

General Indicator

Model (GIM)


City (and its parts)

Measuring City Performance


Energy

Indicators Ecological

Indicators

Financial

Indicators Social

Indicators

Mobility

Indicators

► Evaluation is typically based on indicators,

the most relevant are called Key Performance Indicators (KPIs)

Source: shuttersock.com


Indicators Geobase data


Energy

Indicators

Mobility

Indicators

Ecological

Indicators

Social

Indicators

Financial

Indicators

CityGML Data

Data from National

Topography Models

LADM Data

INSPIRE Data

BIM Data


Observations

1. Geobase data are available for entire countries and can

be used for computing indicator values

● (however, typically additional domain specific data are required)

2. All these geospatial information are based on

standardised semantic data models / ontologies

● e.g. 3D City Models: CityGML; European SDI: INSPIRE; BIM: IFC

3. So far, indicators are typically not formally modelled

using a standardised framework

4. Furthermore, no systematic model exists yet for linking

indicators and geobase data



Model Driven Engineering (MDE)

► … is a software engineering paradigm which began to

evolve in the 1980s

► MDE puts the “model” in the form of formal specifications

in the center of software analysis and design

● Application relevant structures are represented by formal data

models (e.g. using Unified Modeling Language, UML)

● Program code is automatically derived from models

► MDE also addresses the linking of different models

● This is called Model Weaving

● Different models are linked by a weaving model which takes care

of data transformation across the models




General Feature

Model

ISO 19109

CityGML

Application

Schema

M1: Model

M2: Metamodel

X Y Z

This is the general schema which

all geospatial data models follow

(e.g. CityGML, INSPIRE, LADM,

national cadastre & topogr. models)

This is the data model of the

3D city model (here: CityGML)

It defines the structures of all

possible 3D city models

3D city model data, e.g. the objects

of the 3D city model of Berlin M0: Instance

Geospatial Information Modelling



General Feature

Model

ISO 19109

CityGML

Application

Schema

General

Indicator

Model

Energy Related

KPIs Application

Schema

Climate Related

KPIs Application

Schema

KPI A Building Y

KPI B Building Z

M1: Model

M2: Metamodel

X Y Z

M0: Instance

Indicator Modelling

Domain specific

indicators follow a

General Ind. Model

These are the

indicator models

from different

application

domains

Concrete indicators

for concrete city /

landscape objects


Requirements for Indicator Models

► Different types of indicators need to be distinguished

(i.e. numerical, textual, categorical indicators)

► Complex indicators can be composed & computed from

● attribute values from associated city / landscape model objects

● constants

● mathematical expressions (unary / binary arithmetic operations)

on other indicators

► Indicator value aggregation (e.g. summation, average,

maximum, etc.) of other indicators

► Augment indicator values with meta information like

accuracy, lineage / source etc.

● allowing for automatic sensitivity analysis



Domain Specific Indicator Modelling (I)


HeatDemand

+ value

Numeric

Indicator

General Indicator

Model

Domain

Indicators

Energy Planner

Where do I get the data from?

Domain of the stakeholder/application specialist

Energy Planner




-volume

Building

District

HeatDemand

+ value

Numeric

Indicator

DistrictHeat

EnergyDemand

+ compute()

BuildingHeat

EnergyDemand

OCL Rule 2

General Indicator

Model

Domain

Indicators

Object Related

Domain Indicators

Reference

Objects

«Aggregation»num

Energy Planner



**

Domain Specific Indicator Modelling (II)

Many of the

reference objects in

the context of urban

indicators are

spatial objects

Energy Planner



Linking Geospatial and Indicator Models

Building CityObject

GroupBuilding

Connector

District

Connector

-volume

Building

District

HeatDemand

+ value

Numeric

Indicator

CityObjectDistrictHeat

EnergyDemand

+ compute()

BuildingHeat

EnergyDemandSolid

OCL Rule 2

General Indicator

Model

Domain

Indicators

Object Related

Domain Indicators

Reference

Objects

«Aggregation»num

geometry

Geospatial Application Model

(e.g. CityGML)

Energy Planner


City Modeler

What can we do with our data?

Weaving

Model

Domain of the geodata provider Domain of the stakeholder/application specialist

**

*

OCL Rule 1


City Modeler Energy Planner

Domain of the stakeholder/application specialist Domain of the geodata provider


General Indicator Modeling Framework

► Each Indicator Application Model is defined purely from

the viewpoint and requirements of the domain specialist

● data is modeled and structured according to application domain

needs only – and not according to a given geospatial data model

► The data model is separated into 5 consecutive sections

1. Abstract Indicator classes (e.g. numeric indicator)

2. Domain specific indicators (e.g. heat demand)

3. Object-related domain specific indicators (e.g. building heat demand)

4. Reference Objects for the indicators (e.g. building)

► The 5th section addresses linking of the indicator model with

a geospatial application schema (like CityGML)

● Weaving Classes relate Reference Objects with Feature Types from

the geospatial application schema



Linking Geospatial and Indicator Models

Building CityObject

GroupBuilding

Connector

District

Connector

-volume

Building

District

HeatDemand

+ value

Numeric

Indicator


EnergyDemand

+ compute()

BuildingHeat

EnergyDemandSolid

OCL Rule 2

General Indicator

Model

Domain

Indicators

Object Related

Domain Indicators

Reference

Objects

«Aggregation»num

geometry


(e.g. CityGML)

Energy Planner


City Modeler


Weaving

Model


**

*

OCL Rule 1


City Modeler Energy Planner

Domain of the stakeholder/application specialist Domain of the geodata provider

1 2 3 4 5


Linking of an Indicator Model to

different Geospatial Application Models and BIM

Reference Object Classes

Weaving Classes 1

Weaving Classes 2

Weaving Classes 3

CityGML

INSPIRE

BIM / IFC

Object Related Indicators Domain A

Object Related Indicators Domain B

General Indicator Model


Domain of the stakeholder/application specialist Domains of

the geodata /

BIM providers

Model Weavings

Reference Object Classes

Weaving Classes 1

Weaving Classes 2

Weaving Classes 3

CityGML

INSPIRE

BIM / IFC

Object Related Indicators Domain A

Object Related Indicators Domain B

General Indicator Model

HeatDemand

+ value

Numeric

Indicator

General Indicator

Model

Domain

Indicators

Energy Planner



Building CityObject

GroupBuilding

Connector

District

Connector

-volume

Building

District

HeatDemand

+ value

Numeric

Indicator


EnergyDemand

+ compute()

BuildingHeat

EnergyDemandSolid

OCL Rule 2

General Indicator

Model

Domain

Indicators

Object Related

Domain Indicators

Reference

Objects

«Aggregation»num

geometry


(e.g. CityGML)

Energy Planner


City Modeler


Weaving

Model


**

*

OCL Rule 1

Building CityObject

GroupBuilding

Connector

District

Connector

-volume

Building

District

HeatDemand

+ value

Numeric

Indicator


EnergyDemand

+ compute()

BuildingHeat

EnergyDemandSolid

OCL Rule 2

General Indicator

Model

Domain

Indicators

Object Related

Domain Indicators

Reference

Objects

«Aggregation»num

geometry


(e.g. CityGML)

Energy Planner


City Modeler


Weaving

Model


**

*

OCL Rule 1

We can analyse & compare how good /

easy an indicator model fits to a

specific geospatial application model!


29.10.2015

General Planning

Actions Model

(GPAM)


present

t0

past

t-1

future

t1

Reality t-1

City Model t-1

KPIs t-1

Reality t0

City Model t0

KPIs t0

registration/

update registration

Reality t1 ?

City Model t1

KPIs t1

City Model t1‘

Reality t1‘ ?

KPIs t1‘

NO data

collection

possible

calculation

change

calculation

T. H. Kolbe – Smart Models for Smart Cities 29.10.2015 35


Formalization of Action Plans

100 %

0 %

political text regulation ontology for

actions


Aim: making action plans virtually executable on 3D city models!

29.10.2015 36


Properties of Planning Actions (I)

► Actions cause a change of the geometry or the

attributes of the city objects

● they are planned modifications / operations on the entities of a city

► Actions always pursue a specific goal

● that is of different nature / motivation (e.g. monetary, cultural,

personal) and is politically intended

● can be measured by the impact on some key performance

indicators (KPIs)

► Types of actions

● extend existing objects (by new parts, properties, relations)

● change existing objects (update attributes, relations)

● remove existing objects (delete whole & parts, properties, relations)



Properties of Planning Actions (II)

► Relations and dependencies between actions – they can

be composed of others, competing, conflicting, coherent

► Actions can be applied to different reference units from

the city model

● administrative boundaries buildings building parts

● correspond to concrete decision levels

► Actions consist of an ordered sequence of operations

● insert / delete / update of geometries, attributes, and relations

► Actions require resources

● time, cost and goods




Data Model

Aus dem Dissertationsvorhaben

von Maximilian Sindram



General Planning Actions Model (Draft)




Prototypical Action

Specific Action

Semantic 3D City Model

Fra

me

P

ara

mete

r

Instance



German regulation

11 and 12 main street

Buildings (11, 12 and 13 main street)

• refurbishment (energy) (A): policy measure

(1bn Euro)

• sub action (SA): renovation of a building

• facade renovation (S1): insulation of a wall causes a

change of U-value (of the wall)

• window renovation (S2): exchanging windows causes a

change of UW-values (of the window)

11 12 13

S1 S2

A

SA

• (A): political funding

(1m Euro in Munich)

• (SA): renovation of all buildings in main street

• (S1) and (S2): insulation of all buildings with:

bricks (24cm) U-value > 0,8 W/m2K and

double-glazed windows UW-Wert > 1,1 W/m2K

• 11 main street: U-value = 0,7 and UW-value = 2,0

• 12 main street: listed building (age: 200)

• 13 main street: U-value = 1,8 and UW-value = 3,9

11 partial renovation / 12 no renovation / 13 renovation

I

I

I

I

I

I I



29.10.2015

Bringing it

all together


General Feature

Model

ISO 19109

General Indicator

Model General Planning

Actions Model

Energy Related

KPIs Application

Schema

Climate Related

KPIs Application

Schema

Energy Planning

Application

Schema

Traffic Planning

Application

Schema

KPI

building X

KPI

building Y Facade

retrofitting

building X

CityGML

Application

Schema



29.10.2015

Outlook: Dynamics

in CityGML 3.0

&

Conclusions


An Outlook to CityGML 3.0

► CityGML 3.0 is currently under development in the OGC

● release of the new version expected in 2017

► New CityGML 3.0 features (subject to voting)

● data model based on ISO 19136 (GML 3.2.1); automatic derivation

of the exchange format (i.e. the CityGML application schema)

● more flexible LOD concept – separate indoor & outdoor LODs

possible; the previous LOD 0-4 concept is retained as a profile

● new feature types (e.g. for non-building constructions) &

refinements (e.g. building units)

● versioning and historization

● dynamic, i.e. time-dependent object properties e.g. for energy

consumption, energy production, moving objects

● direct linking of sensor data to object properties

● tabulated & interpolated values; simple & complex patterns



Dynamic Data in CityGML 3.0

► Two distinct approaches

● For slow changes / city model evolution: versioning & historization

● For fast changes: time-variant properties (attributes, geometry)


CityGML

V2

V3

How to organize

the state of our

world

Dynamizer

Dynamic variations

Periodic patterns

Mechanism

required to

ensure that

within one

version we

have a

consistent

city model

Provides

replacers /

overriders,

replacing the

attributes in

the static

CityGML

model

Dynamic Data Schema

Versioning Schema City Model

V2

V3

V1 V1

[Kanishk Chaturvedi 2015]


Conclusions

► Semantic 3D city models are good platforms to structure

and organize urban data – but mostly in a static way so far!

► We propose two new frameworks complementing the

General Feature Model (GFM) ISO 19109

► General Indicator Model (GIM)

● allows to specify domain specific indicator models independent from

geospatial application schemas

● concept for linking indicator models to geospatial application schemas

● programs for indicator computations are automatically derivable

► General Planning Actions Model (GPAM)

● allows to formalize planning actions in different application domains

● consideration of affected KPIs, modified objects, transactions



References

► Elfouly, Mostafa; Kutzner, Tatjana; Kolbe, Thomas H.: General Indicator Modeling for Decision

Support based on 3D City and Landscape Models using Model Driven Engineering. Peer Reviewed

Proceedings of Digital Landscape Architecture 2015 at Anhalt University of Applied Sciences,

Wichmann, 2015

Click for article download

► Sindram, Maximilian; Kolbe, Thomas H.: Modeling of Urban Planning Actions by Complex

Transactions on Semantic 3D City Models. Proceedings of the International Environmental Modelling

and Software Society Conference 2014 (iEMSs), San Diego, International Environmental Modelling

and Software Society (iEMSs), 2014


► Chaturvedi, Kanishk; Kolbe, Thomas H.: Dynamizers - Modeling and implementing dynamic

properties for semantic 3D city models. 3rd Eurographics Workshop on Urban Data Modelling and

Visualisation (UDMV 2015), 2015 (in print)


► Chaturvedi, Kanishk; Smyth, Carl Stephen; Gesquiere, Gilles; Kutzner, Tatjana; Kolbe, Thomas H.:

Managing versions and history within semantic 3D city models for the next generation of CityGML.

Selected papers from the 3D GeoInfo 2015 Conference (Lecture Notes in Geoinformation and

Cartography), Springer, 2015 (in print)



http://mediatum.ub.tum.de/doc/1247168/526945.pdf

http://mediatum.ub.tum.de/doc/1224665/886628.pdf

http://mediatum.ub.tum.de/node?change_language=en

http://mediatum.ub.tum.de/node?id=1276238


Credits

► The projects Energy Atlas, Modeling City Systems,

Smart Sustainable Districts have been funded

by Climate-KIC of the European Institute

for Innovation and Technology (EIT)


Smart Models for Smart Cities - Modeling of Dynamics, Sensors, Urban Indicators, and Planning...

Science

Transcript of Smart Models for Smart Cities - Modeling of Dynamics, Sensors, Urban Indicators, and Planning...