Automation in Construction 50 (2015) 16–28

Contents lists available at ScienceDirect

Automation in Construction
journal homepage: www.elsevier.com/locate/autcon

Developing a physical BIM library for building thermal energy simulation

Jong Bum Kim, WoonSeong Jeong, Mark J. Clayton, Jeff S. Haberl, Wei Yan ⁎
Texas A&M University, United States

a r t i c l e i n f o a b s t r a c t

Article history: Insufficient interoperability resulting from complex data exchange between architectural design and building en-
Received 30 December 2013 ergy simulation prevents the efficient use of energy performance analyses in the early design stage. This paper
Received in revised form 4 August 2014 presents the development of a Modelica library for Building Information Modeling (BIM)-based building energy
Accepted 21 October 2014
simulation (ModelicaBIM library) using an Object-Oriented Physical Modeling (OOPM) approach and Modelica,
Available online xxxx
an equation-based OOPM language. By using the ModelicaBIM library, our project investigates system interfaces
Keywords:
between BIM and energy simulation, which can perform semi-automatic translation from the building models in
Building Information Modeling (BIM) BIM to building energy modeling (BEM) using a BIM's authoring tool's Application Programming Interface (API).
Building energy modeling (BEM) The ModelicaBIM library consists of OOPM-based BIM classes and OOPM-based BIM structure. OOPM-based BIM
Object-Oriented Physical Modeling (OOPM) classes represent building component information. OOPM-based BIM structure consists of test case models that
Interoperability demonstrate (i) how building information in BIM can be transformed to OOPM and (ii) how design operations in
BIM, such as changing a building geometry and editing building components, can be translated into BEM. A case
study for simulation result comparisons has been conducted using (i) OOPM-based BIM models in the
ModelicaBIM library and (ii) LBNL Modelica Buildings library (a Modelica-based building thermal simulation li-
brary developed by Lawrence Berkeley National Laboratory). Our implementation shows that the ModelicaBIM
library enables (i) objects in BIM to be translated into the OOPM-based energy models and (ii) existing OOPM
library to be utilized as a simulation solver for BIM-based energy simulation.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction imperative computer language such as FORTRAN, C, and C++ [7]. The
GUIs facilitate rapid input and output processing as well as simulation
Since 1996, more than four hundred software tools for building energy running when the semantics of input files, parameters of simulation set-
simulation have been listed in the “Building Energy Software Tools Direc- tings, and the formats of output files are dependent on the simulation en-
tory” provided by U.S. DOE (2013). Among the tools, a few are dominantly gines. In many cases, simulation engines and GUIs were developed
used in education and industry [1–5]. For example, DesignBuilder, DOE-2, separately. For instance, Hevacomp and DesignBuilder are independently
eQuest, Ecotect, Energy-10, EnergyPlus, Green Building Studio, HEED, and developed GUIs of EnergyPlus.
IESVE are widely used in the United States [6]. In education, Ecotect, Some existing simulation tools and their GUIs were modified for
Energy-10, Radiance, CONTAM, and eQUEST are often used [3]. data exchange capabilities between BIM and energy simulation through
Recently, linking Building Information Modeling (BIM) and energy standard data schemas such as IFC and gbXML, which contain building
simulation either with standard data schemas such as the Industry geometry information and other information of internal loads, occupan-
Foundation Classes (IFC) or with common data formats such as Green cy, zone assignments, system configuration, and utilization schedules
Building XML (gbXML) is a developing area in research. Some tools [4]. Currently, both schemas are supported by BIM tools including
were modified to utilize BIM and others were developed to be compat- Revit, Bentley, and ArchiCAD, as well as energy simulation tools such
ible with BIM authoring tools [5]. In this section we trace approaches as Green Building Studio, Ecotect, Hevacomp, eQUEST, HAP, and
and limitations of current BIM-based energy simulation. IES b VEN [8].
RIUSKA, a GUI of the DOE-2.1E engine, has an IFC import feature that
can obtain building geometry information from BIM. eQUEST supports di-
1.1. Generating energy simulation models from BIM rect imports of DWG and gbXML with limitations: only two-dimensional
building footprints can be obtained from DWG [9]; and some eQUEST fea-
Most energy simulation tools consist of the simulation engine and the tures are not usable for the gbXML model [9].
graphical user interfaces (GUIs). The engines were often written in Many applications and GUIs are developed for EnergyPlus such as
Simergy, DesignBuilder, OpenStudio, CYPE-Building Services, Demand
⁎ Corresponding author at: 3137 TAMU, College Station, TX 77843, United States.
E-mail addresses: [email protected] (J.B. Kim), [email protected]
Response Quick Assessment Tool, Easy EnergyPlus, EFEN, AECOsim,
(W. Jeong), [email protected] (M.J. Clayton), [email protected] (J.S. Haberl), Hevacomp, and SMART ENERGY [10]. OpenStudio is a collection of appli-
[email protected] (W. Yan). cations: a SketchUp plug-in as a building geometry editor; an OpenStudio

http://dx.doi.org/10.1016/j.autcon.2014.10.011
0926-5805/© 2014 Elsevier B.V. All rights reserved.
 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 17

application as a main energy modeling interface; RunManager as a simu- do not indicate which and how building components need to be modi-
lation interface; and the ResultsViewer [11]. OpenStudio can import fied to achieve a specific energy performance.
gbXML in company with material, construction, surface, space, building, Although the functionalities of the building energy simulation
and schedule information from BIM. DesignBuilder can import three di- tools have been developed, further work is expected to improve
mensional geometry information through gbXML and two dimensional the interoperability between BIM and the energy simulation tools
building footprints through DXF. OpenGL solid modeler of DesignBuilder [22]. For BIM-based energy simulation in early design, software
visualizes building façade design and solar studies [10]. Simergy, a GUI of tools need to have intuitive GUIs [2], seamless data exchange capa-
EnergyPlus, supports editing building geometries and HVAC systems, bilities [19], reliable BIM compatibilities [5], and automated rule-
generating EnergyPlus input (IDF) files, running EnergyPlus simulations, based translation capabilities from BIM to BEM [18]. In addition,
and reporting simulation results [12]. Building geometry information in tools' good maintainability that allows modification and customiza-
CAD/BIM models can be imported into Simergy through IFC and gbXML tion of calculation modules may improve the use of building energy
or can be generated using the Simergy GUI. Simergy is developed based simulation [23].
on Simulation Domain Model (SimModel) that is a XML-based data
model [13]. Hevacomp, another GUI of EnergyPlus, can load building ge- 2. ModelicaBIM library, Revit2Modelica framework, and PBIM
ometry and construction information through gbXML, DWG, and DXF, research
and then generate the EnergyPlus input files.
Our research investigates a new interface for BIM-based building en-
1.2. Integrating energy simulation tools with BIM authoring tools ergy simulation, integrating the architectural design and the energy
simulation process. The objectives are: (i) enhancing the interoperabil-
Some simulation tools can be plugged into BIM authoring tools as ity among BIM and BEM, (ii) enabling more reliable BEM generation
add-ins such as IESVE and Green Building Studio, which allow energy from BIM, (iii) enhancing the integration and the coordination among
simulation within the BIM environment. Green Building Studio is a multi-domain BEM, and (iv) enabling BIM as a common user interface
web-based energy analysis environment based on DOE-2.2. Energy for the multi-domain energy simulations. To do so, we investigated a
models are generated from Autodesk Revit models via certain manual BIM-based energy modeling and simulation framework, which inte-
preparation processes. Then, simulation is performed by a cloud service grates BIM and multi-domain energy simulations, and named it as Phys-
and the results are reported to users. IESVE has a Revit plug-in that can ical BIM or PBIM [24]. Building thermal, daylighting, and BIPV analyses
generate an energy model based on gbXML in Revit, but editing the are being implemented in the PBIM framework, and this paper describes
exported energy model can be done in the IESVE interface. our research on the BIM to thermal BEM in detail.
Ecotect performs various analyses for solar, daylighting, acoustic, Revit2Modelica is a framework of BIM-based building thermal simu-
thermal, and airflow using multiple engines such as Radiance, DAYSIM, lation (Fig. 1). Revit2Modelica consists of Revit Application Program-
EnergyPlus, and Ecotect built-in engines. Ecotect can import gbXML, ming Interface (API) programs written in C# programming language
IFC, DXF, etc. from BIM [14]. In addition, Vasari [15], a parametric concep- and the Object-Oriented Physical Modeling (OOPM)-based library
tual building design tool, can generate a Green Building Studio thermal (ModelicaBIM library) written in Modelica. Revit API programs have
model and perform solar studies, solar radiation analysis, and wind stud- functions that add data sets to BIM, access the BIM data structure, and
ies. In sum, these tools and their GUIs can read partial information of a produce Modelica models using the ModelicaBIM library and Lawrence
building model from BIM through IFC or gbXML. Berkeley National Laboratory (LBNL) Modelica Buildings library. The
ModelicaBIM library consists of a ModelicaBIM Class package and a
1.3. Limitations ModelicaBIM Structure Example package. The ModelicaBIM Class pack-
age contains building element classes that use the classes in the LBNL
While energy modeling using data schemas such as IFC and gbXML Modelica Buildings library. The need of wrapping classes is driven by
has been implemented for many simulation programs, reliable energy the different object semantics and structures between BIM and BEM.
models can be acquired through manual model checks and modification Detailed wrapping processes applied in the ModelicaBIM library devel-
[16]. Although energy simulation tools have been modified or devel- opment will be explained later.
oped to support the energy model generation process using BIM, the en- A workflow of Revit2Modelica is illustrated in Fig. 2. Revit2Modelica
ergy simulation analysis in design is still regarded as cost and labor translates BIM into OOPM in Modelica, semi-automatically generates
intensive [17]. BEM (Modelica models) from BIM (Revit models), immediately per-
Generating an energy simulation model from BIM is time consum- forms building thermal analyses, and reports simulation results.
ing, error-prone, and not intuitive [18]. Incongruent information stored
• First, the prototype accesses models in BIM to read model information
in BIM and BEM requires a certain level of translation from BIM to BEM.
such as geometry, materials, and location information.
First, all model information of BIM does not need to be translated into
• Second, the prototype transforms the obtained information by follow-
BEM. For instance, a room in an architectural model does not always in-
ing the model structure of the ModelicaBIM library.
dicate a zone in an energy simulation model. Second, some values re-
• Third, by using the transformed model information, the prototype
quired for energy modeling are not imbedded in BIM. For example,
outputs a Modelica BEM and calls Dymola, an Integrated Development
neither thermal zone information nor boundary conditions are stored
Environment (IDE) for Modelica, to run simulation.
in BIM. Often, model information in BIM is abstracted in BEM. A wall, a
• Fourth, the prototype reports simulation results in the BIM user in-
roof, and a floor in BIM are simplified as surfaces in energy simulation.
terface. The prototype currently analyzes heat flow of each build-
When energy modeling is conducted for a building with complex geom-
ing component, free floating indoor temperatures, accumulated
etry, manual model inspections become more complex [19]. Studies on
heating and cooling loads, as well as peak heating and cooling
BIM-based energy modeling describe that standards or guidelines are
loads. Plots of the simulation results are generated and opened in
required to relieve the problems in building geometry translation
the Revit user interface.
from BIM to BEM [20,21]. On the other hand, incorporating simulation
results into the design stage is not always obvious [17]. For designers, Three key milestones in the Revit2Modelica development are:
what the simulation results imply and how building design can corre-
spond to such results are not clear. For instance, whole building level • developing an OOPM library (ModelicaBIM library) including OOPM-
simulation results such as annual energy consumption and peak loads based classes and OOPM-based models to link BIM and existing OOPM
are often provided in a format of charts, spreadsheets, and plots. They simulation solvers,
18 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

PBIM Research

Integrated Simulator and

Performance Analyzer

Revit2Modelica Revit2Radiance Revit4BiPV

(Solar Thermal) (Daylighting) (BiPV)

ModelicaBIM Applications Radiance BIM Applications BiPV BIM Applications

(Revit API) (Revit API) (Revit API)

ModelicaBIM Library Radiance / DaySIM

LBNL Modelica
Buildings Library

Fig. 1. PBIM research framework. The ModelicaBIM library is a part of BIM-based solar thermal research.

• prototyping the system interfaces using BIM API, and imbedded information to create a building product model [25,26]. BIM
• validating the interfaces and the library through simulation result stores both geometric descriptions and non-geometric attributes in its
comparisons. objects and parameters. Semantically rich information can be stored
into the shared database. Parameterization capabilities in BIM enable
This paper describes the development of the ModelicaBIM library. The the rapid generation of complex forms and interactive model changes.
following sections explain the framework of the library development, BIM authoring tools provide API that enables software developers to
goals and challenges, methods and tools, descriptions of the library com- extend the tools, e.g. Autodesk Revit's API in C# and VB.NET; Graphisoft
ponents, as well as validation of test case models in the library. ArchiCAD's API in C and C++; Bentley MicroStation's macros in VB.NET,
C++, and C#, as well as MicroStation Development Libraries in C. It al-
3. ModelicaBIM library development for BIM-based energy simulation lows complex geometric modeling, accessing information of BIM ob-
jects, comprehensive database creation, and presentation of analysis
This section describes tools and data, challenges and objectives, as output associated with corresponding BIM objects. In our project, in-
well as methodologies in the library development. stead of using IFC or gbXML, we propose a BIM API method to access
the BIM data directly from the BIM authoring tool in order to take ad-
3.1. Tools and data vantages of the parametric modeling capability of BIM and a more
seamless integration with less data conversions between BIM and BEM.
3.1.1. BIM and the BIM authoring tools
BIM aims to support information creation, exchange, and applica- 3.1.2. OOPM and Modelica
tions in the buildings' lifecycle in the Architectural Engineering Con- OOPM is a fast-growing modeling and simulation approach, offering
struction (AEC) industry. BIM ties all the building components with a structured and equation-based modeling [27]. Modelica is an OOPM

Fig. 2. A workflow of Revit2Modelica.

 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 19

language that models the dynamic behaviors of technical systems using used to calculate the surface area in the LBNL Modelica Buildings library,
differential algebraic equations (DAE)-based simulation [27]. It has been which is limited to rectangular surfaces. We use the area value instead
used in the complex physical system design of mechanic, electric, ther- of width and height to represent diverse geometry and the values are
mal, and control systems. A component connection diagram in Modelica obtained from BIM directly.
can present physical system topology of energy models [27]. The object-
oriented modeling approach of Modelica may enable an intuitive map-
3.3. Methodology
ping from the object-based BIM structure to OOPM.
The use of Modelica needs a Modelica library offering model compo-
Our methodologies in the ModelicaBIM library development are
nents and solvers as well as a Modelica simulation environment. In this
(1) object mapping for semantics and behaviors, (2) preprocessing
research, we use Dymola [28] as a simulation environment and the LBNL
BIM, and (3) wrapping existing solver classes with new library classes.
Modelica Buildings library as a simulation solver.

3.1.3. LBNL Modelica Buildings library 3.3.1. Object mapping for semantics and behaviors
The LBNL Modelica Buildings library contains dynamic models and Object mapping allows us to identify required information and a
control systems for building energy simulations [23]. It supports the structure of the ModelicaBIM library. Mostly, we addressed object
simulation of heating and cooling systems, controls, heat transfer semantics and behavior mismatches that can impede an automatic
through building envelope, as well as airflow models. The library is model transformation from Revit to Modelica. The two mismatches
being developed and some packages and components have been vali- we defined in this paper are (1) semantic mismatches of building
dated. The HeatTransfer and Rooms packages in the library are major re- components in BIM and BEM and (2) behavior mismatches between
sources for the building thermal analysis, and they have been tested and architectural design and energy modeling. While we implemented
validated through benchmark simulation models [29,30]. The execution an object mapping between Revit and Modelica, we describe se-
time test of Modelica simulation shows that it can be comparable with mantic mismatches using generic concepts that are not software-
TRNSYS [31]. dependent.
To construct a building energy model with the LBNL Modelica Build- Semantic mismatches of building components can delay data ex-
ings library, a user needs to manually retrieve a building's data and change of building objects and their attributes between two domains.
write Modelica codes that represent a building's energy model. Once For instance, a building envelope in BIM is commonly decomposed
the BIM model is preprocessed manually, our system automatically con- into walls, roofs, and floors/slabs, while it can be recognized as exterior
verts models in BIM into Modelica-based energy models by using the or interior surfaces in BEM. An interior space is recognized as a room in
ModelicaBIM library and then performs thermal analysis using the BIM, while it can be represented with a thermal zone in BEM. Precise
LBNL Modelica Buildings library. This automated transformation from geometry information of building components is stored in BIM, but a
BIM to BEM will speed up the design and the simulation process signif- wetted surface area is required for one-dimensional heat transfer calcu-
icantly and substitute operators' subjective interpretations to minimize lation in BEM.
human errors in energy simulation. Object mapping for the roof object is exemplified in Table 1. To
map the roof object, some values can be directly used, but others
3.2. Challenges and objectives need to be either translated or created. For instance, the origin
and normal vectors are used in BIM, while the tilt and azimuth
3.2.1. Apply a coherent object relationship from BIM to BEM values are needed in BEM for the coordinate system; therefore,
In the energy modeling process, object relationships and semantics the coordinate system is transformed. When different units are
of architectural models are often abstracted when such model informa- used such as thermal resistance, unit conversion is carried out.
tion does not substantially affect simulation results. For example, build- Some missing values in BIM such as absorptance and emittance
ing components such as walls, roofs, and floors are frequently simplified are added in BEM.
as surfaces and then heat transfer through the surfaces are simulated in Behavior mismatches during the modeling process can defer inter-
the energy models. When designers modify building design according active energy model update in accordance with building design chang-
to the abstracted simulation results, energy performance of each surface es. For example, adding a door on the exterior wall can be interpreted as
needs to be mapped to corresponding building components again. Such adding an opening that causes air flow and air infiltration. Separating
abstraction can delay the result interpretation in building design. one room with an interior wall in BIM can be translated into BEM as
Consistent object classifications and semantics between BIM and BEM adding an interior surface that performs heat transfer calculation be-
are applied in the ModelicaBIM library development. Our Modelica-based tween two thermal zones.
simulation will perform component-level analyses and the simulated per- The object mapping method establishes a set of rules for an auto-
formance will be forwarded to Revit to inform design. Applying consistent matic data transformation, which will be applied to the Revit2Modelica
object classifications of building components from BIM to BEM would be prototype development. This object mapping method enables the
significant to pair simulation results with BIM.

3.2.2. Create parameters to receive values from BIM

In the LBNL Modelica Buildings library, predefined materials such as Table 1
An example of roof object mapping.
brick, concrete, and insulation boards are used. To add new materials
that are not defined in the library, users need to create new material ob- Category BIM ModelicaBIM
jects prior to the use. The use of manually-predefined material classes Object Roof PBIM.BIMPackage.Roof
makes both modification and creation of materials time consuming. In Orientation Origin/normal (vector) Tilt/Azimuth
Revit2Modelica, we transform material properties stored in BIM into Area Area Area
Thickness Width Thickness
the parameters of the Modelica energy models. To do so, classes in the
Material DefaultRoof PBIM.BIMPackage.Material
library are created to have parameters that can store properties in BIM. Thermal resistance Resistance (ft2·°F·hr/BTU) R (m2·K/W)
Number of layers CompoundStructure.LayerCount nLay
3.2.3. Create objects for diverse building geometry Other physical Roughness Roughness
New objects are created to have parameters flexible for diverse properties N/A Absorptance
N/A Emittance
building geometry. For example, width and height of a surface are
20 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

Revit2Modelica prototype to (1) transform each BIM object (that is ther- Modelica Buildings library. Primary building components for building
mally significant to building energy consumption) into a Modelica ener- design have been implemented. For instance, the Room class of the
gy object by using the ModelicaBIM library classes; (2) generate ModelicaBIM library is a wrapper class of the MixedAir class of the
ModelicaBIM models by applying the ModelicaBIM Structure; and LBNL Modelica Buildings library. The door class of the ModelicaBIM li-
(3) allow the ModelicaBIM to call the publically available LBNL brary wraps DoorDiscretizedOpen class of the Buildings library. Detailed
Modelica Buildings library as a simulation engine. Below we describe implementations are explained in the following section.
the detailed development process of the ModelicaBIM library and the
ModelicaBIM Structure. 4. ModelicaBIM library
The significance of comprehensive information exchange between
stakeholders has been investigated in research on the Model View Def- The ModelicaBIM library consists of a ModelicaBIM Class package
inition (MVD). MVDs have been widely accepted in information ex- and a ModelicaBIM Structure Example package. The Class package pro-
change using IFC, so they are often understood as subsets of the IFC vides wrapper classes of existing Modelica library — LBNL Modelica
model specification [32]. For instance, a set of MVDs are proposed in Buildings library. Currently, the Structure Example package provides 5
“Concept Design BIM 2010”, enabling four types of analyses from an ar- example Modelica models that represent Modelica BIM structures. In
chitectural model: spatial program validation, circulation and security this section we explain each class in the Class package to describe the
analysis, energy performance analysis, as well as quantity takeoff and object mapping process from BIM to BEM through the class wrapping
cost estimating [33]. A more generic definition of MVDs is a comprehen- process. Then, we explain the Structure Example package to describe
sive representation of the information concepts required for particular the required modeling process for creating a two-zone model with win-
information exchange among multiple domains [34]. Therefore, the dows and an interior door. The findings here will be a foundation to for-
general concept can be captured when our library is created for multi- mulate a rule-based automated model translation from BIM to BEM.
domain energy simulations. Dymola is used as a Modelica development and a simulation interface
as shown in Fig. 4.
3.3.2. Preprocessing BIM The left side pane is a package browser that shows a tree structure of
We preprocessed BIM by using the modeling capabilities of the BIM the ModelicaBIM package. The model diagram in the right side shows Test
authoring tools in this paper. The automated preprocessing of BIM will Case 4 that has two rooms, an interior door, and two windows. The yellow
be implemented in the future research. Some of the numerical values in boxes are material, structure, and construction information of building
BIM are directly usable, but others need to be transformed according to components, which are modeled with the classes in the ModelicaBIM
the appropriate data transformation rules between BIM and Modelica. class package.
In our development, BIM is preprocessed by addition, translation, and In the diagram, a series of icons are connected to each other by com-
calculation: addition is to add data sets that are absent in BIM but need- position lines. The icons represent physical components and the compo-
ed for physical modeling such as solar and infrared absorptance of sition lines represent connections among these components. Between
building materials; translation is to translate data between BIM and two room icons, a red line links two small red dots. The small red dots
BEM such as the room-to-thermal zone translation; and calculation is are connectors and the line is the connect. The connector is a Modelica
to populate new values from included values in BIM such as a class that creates a physical flow between components. The connect
window-frame ratio. class is a communication interface between connectors of the compo-
Some simple preprocesses in BIM enable building geometry abstrac- nents. A connect between two connectors is established as an equation
tion. For example, a building in Fig. 3 consists of four exterior walls, one in Modelica using the following code: connect (connector1, connector2).
interior wall, one floor, and a roof. The colored wall attached to two For instance, the door icon has four blue dots that are connected to
rooms is one piece in BIM, but it needs to be recognized as two pieces two room icons by four cyan lines. In the Modelica terminology, a
in energy modeling. A required preprocess in BIM is splitting the wall door component has four connectors. Two connects between the door
into two pieces and the same preprocesses can be applied to the roof and each room are created to link two of four door connectors and two
and the floor. While the preprocessing in this paper is performed man- room connectors.
ually in BIM, they will be substituted with automatic preprocessing in
our future research based on a rule based conversion for building topol- 4.1. ModelicaBIM Class package
ogies and space boundaries [20,21,35].
The ModelicaBIM Class package holds OOPM-based BIM classes that
3.3.3. Wrapping solver classes with the ModelicaBIM library represent building geometries, topologies, and material properties. Cur-
A class wrapping method enables an intermediate Modelica energy rently, basic classes for building components such as room, wall, roof,
model not only to follow data structures and semantics of BIM but floor, window, and door are implemented for heat transfer through
also to use existing Modelica libraries as a simulation engine. It facili- building envelopes and airflows between multi-zones.
tates object mapping from BIM to Modelica energy models. Object map- While models in Revit and in Modelica are based on object-oriented
ping from BIM to the LBNL Modelica Buildings library is accomplished modeling concepts, they are independently created for respective imple-
by the wrapper classes created in the ModelicaBIM Class package. mentation considerations with specific object semantics and relation-
The ModelicaBIM Class package contains wrapper classes of the LBNL ships. Such object-relational and semantic differences or mismatches
can cause conceptual and technical difficulties in the object-oriented
programming process when creating a translator between object-based
applications [36]. Compared to a translator between object-based appli-
cation (e.g. Revit) and procedural-programming-based application (e.g.
ASHRAE Toolkit for Building Load Calculations), the translator between
two object-based applications (Revit and Modelica-based thermal
models) is a more natural selection that reduces the interoperability
problem during implementation [24]. The method to wrap predefined
classes of the LBNL Modelica Buildings library results in new classes in
our Class package so that (1) required parameters in Modelica can be
transformed from Revit to Modelica thermal models and (2) building to-
Fig. 3. Preprocessing in BIM of a building envelope. pology in Revit can be translated into Modelica object relationship.
 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 21

Fig. 4. Schematic model view of Test Case 4 in the ModelicaBIM library.

Each class is created to contain (1) a set of parameters that can trans- buildings, rooms, and building components that can facilitate
fer parameters from Revit in Modelica, (2) instantiated classes of the transformation of object hierarchy and classification from Revit to
LBNL library that hold equations, and (3) connects that link the wrapper Modelica.
class and the wrapped class. A building consists of a series of spaces (Room class), each room con-
Energy model generation from BIM will translate building topolo- tains roofs, floors, and walls (Construction class), each construction
gies, geometries, and properties in Revit to Modelica. In creating new li- consists of a single or multiple material layer(s) (Structure class), and
brary objects, we apply a common object relationship (Fig. 5) among each layer of the structure has material properties (Material class).

Fig. 5. Schematic object hierarchy of building components in the ModelicaBIM library.

 22
Table 2
Test case models in the ModelicaBIM Structure Example package. The table also shows the workflows of designers (application users) and developers (the authors of the paper) in relation to the application (Revit2Modelica).

Test Case 1 Test Case 2 Test Case 3 Test Case 4 Test Case 5

Building Description

J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

• One room model • Two room model • Two room model having windows • Two room model having
windows and an interior door
• Three room, two story model having
windows and an interior door
Create BIM objects (A1 in Fig. 2, • Create four walls, a floor, and a • Create an interior wall based on • Create windows on the exterior • Create a door on the interior wall • Create a room below the room of
by designers manually) roof object Test Case 1 walls of Test Case 2 of Test Case 3 Test Case 4
• Create one room • Create two rooms
Generate Modelica instances in the • Generate instances of materials, • Generate interior surface • Generate window instances • Generate a door instance • Generate an instance of a lower
ModelicaBIM (A2, B1 in Fig. 2, layers, and instances • Assign new boundary condition of • Generate an association of room- floor thermal zone
by our system automatically) constructions • Split two exterior surfaces two walls having windows (from opening-room for air flow calcu- • Generate a connection between
• Generate thermal zone instances intersected by the interior wall exterior surfaces to exterior lation two zones
for a single zone model • Split a floor and a roof into two surfaces having windows). • Assign an opening schedule
pieces.
• Divide one zone into two zones.
• Build an association of room-
wall-room for heat transfer
calculation
Develop Modelica classes in the • Develop classes of material, layer, • Develop material, layers, and • Develop material, layer, and • Develop material, layer, and con- • Develop material, layer, and
ModelicaBIM library (B2 in Fig. 2., by and constructions of wall, roof, and construction classes of the construction classes of the window struction classes of the door construction classes of the lower
our researchers/developers) floor interior wall • Develop Modelica structure for • Develop Modelica structure for floor thermal zone.
• Develop a room class • Develop Modelica structure for opaque constructions having the multi-zone air flow model • Develop Modelica structure for the
• Develop Modelica structure for the the multi-zone heat transfer windows multi-zone heat transfer model
single zone model model containing multiple stories.
Use Modelica classes and call solvers in • Use classes for single zone thermal • Use classes for multi zone • Use classes for the window system • Use classes for the • Use classes for multi-zone thermal
the LBNL Modelica Buildings library simulation thermal simulation • Call solvers for energy simulation multi-zone airflow model simulation
(B3 in Fig. 2., by our system • Call solvers for energy simulation • Call solvers for energy • Call solvers for energy • Call solvers for energy simulation
automatically) simulation simulation
 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 23

Constructions such as walls can host openings such as windows (Win- association); the parameters are created to use door geometry informa-
dow Structure classes) and doors (Door class). Material classes are in- tion from BIM; the wrapped class supports a closed door, but an opera-
stantiated as parameters in a Structure class, the Structure class is ble and opened door can be simulated by using other classes directly
instantiated as parameters in a Construction class, and they are instan- from the LBNL Modelica Buildings library.
tiated as parameters in a Room class in the Modelica energy model. The
hierarchical components are described below. 4.2. ModelicaBIM Structure Example package
Room: the Room class is a single-zone model to calculate heat ex-
change through a building envelope such as convection, conduction, in- The ModelicaBIM Structure Example package holds OOPM-based
frared radiation, and solar radiation. Parameters such as floor area, room BIM models that present the energy modeling procedure by using the
height, and latitude are included in the Room class. To create the Room ModelicaBIM Class package. Five test cases have been created in the
class, we wrap the MixedAir model in the LBNL Modelica Buildings li- Structure Example package to demonstrate how building design in
brary, which is a model of a thermal zone with completely mixed air. BIM can be translated into the energy modeling in Modelica. Test Case
The application and validation of this model have been conducted [29, 1 presents a room model having six exterior surfaces. From this single
30,37,38]. zone model, other models are sequentially created to demonstrate sim-
By wrapping existing classes, we assign new parameters and define ple operations in building design such as changing room geometry,
new object relationships among building materials, components, and adding an interior wall, and installing windows and doors as shown in
topologies in the Room class. To minimize an error-prone process in Table 2.
the energy modeling, hard coded values in the existing LBNL Modelica Table 2 describes detailed modeling processes that can correspond
Buildings library are replaced with parameters in the Room class to the Revit2Modelica workflow in Fig. 2. The associations between
declaration. Table 2 and Fig. 2 are as follows.
The Room class can be used for a single thermal zone. When we in-
vestigate thermal modeling for a single zone building, a room in BIM is • The second row, creating BIM objects, is related to A1 in Fig. 2. In
modeled as a single zone model in Modelica by using one Room object. Revit2Modelica, the development of extended BIM is manually done
By connecting multiple Room class objects, multi-zone thermal models by designers and some BIM objects are manually preprocessed.
can be created. One of our test cases demonstrates a multi-story build- • The third row, generating Modelica instances, is related to A2 and B1
ing by using two Room objects, which calculates conduction through a in Fig. 2. The ModelicaBIM Structure Example package defines re-
slab construction between two zones. In another test case, airflow be- quired rules of Modelica object instantiations and model creation so
tween two rooms is modeled by using a Door object and two Room that our system can translate BIM to ModelicaBIM models automati-
objects. cally.
Material: the Material, Structure, and Construction classes are creat- • The fourth row, developing Modelica classes in the ModelicaBIM li-
ed for material properties and geometry information of opaque con- brary, is related to B2 in Fig. 2. The results of this task are the
structions. The Construction class instantiates the Structure class as a ModelicaBIM class package.
parameter and the Structure class instantiates the Material class as a pa- • The fifth row, using Modelica classes and calling solvers of the LBNL
rameter as shown in Fig. 5. Modelica Buildings library, is related to B3 in Fig. 3. The ModelicaBIM
The Material class represents thermal properties of a single material Structure Example package defines how to use existing Modelica clas-
layer of opaque constructions. Parameters of this class include the mate- ses and call solvers so that our system can generate ModelicaBIM
rial thickness and physical properties such as thermal conductance, spe- models and run the simulation automatically.
cific heat capacity, and thermal resistance. Multi-layer constructions can
be created by using multiple Material objects in the Construction class.
Structure: the Structure class represents thermal properties of 4.2.1. Creating a basic building model with a single thermal zone (Test Case 1)
opaque constructions having single or multiple material layers. The Test Case 1 presents the Modelica model structure and the modeling
Structure class uses the Material class objects as parameters. For a procedure to create a single-zone model. To create a thermal zone,
multi-layer construction, the corresponding number of Material objects (i) building envelope information, (ii) boundary conditions of the
can be arrayed as parameters from the outside to the inside surface.
Then the thermal resistance of all material layers is calculated in the
Structure class. Other physical properties of outer and inner surfaces
such as absorptance and emittance are included as parameters in the
Structure class.
Construction: the Construction class represents geometry informa-
tion and physical properties of building components based on the Struc-
ture class. Construction Classes for roof, floor, and wall have been
implemented to create building geometry. Geometry information such
as area, tilt, and azimuth values are added as parameters and the values
will be obtained from BIM.
Window: the Window class represents a window system with glass
panels and window frames. Parameters for window geometry are in-
cluded and the portion of a frame to the window system is included to
store the net frame area. A single pane glass is implemented currently
and multi-layer glasses and window shading devices will be added to
our library later.
Door: the Door class performs airflow simulation between two
zones. We have implemented a closed door that calculates heat transfer
through the door and infiltration through the door assemblies. The door
class consists of parameters of door geometry, a wrapped class of the
multi-zone airflow class in the LBNL Modelica Buildings library, and
connect classes that present the building topology (room-door-room Fig. 6. A Modelica code block for a thermal zone modeling.
24 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

Fig. 7. A Modelica code block to connect two zones.

building envelope, and (iii) room geometry information need to be de- conduction through window frames, convection caused by glass
fined as shown in the Modelica code blocks in Fig. 6. panes, infiltration through window assembles, and heat distribution
First, building envelope information is defined by using the Material, for infrared radiative heat gains).
the Structure, and the Construction classes. In the code line 1, a roof ma- In the Modelica model, (i) the boundary condition of the exterior
terial object is instantiated and values for physical properties are given surface having windows is updated from an opaque surface (datConExt)
to the parameters. Then a roof object is instantiated (line 3) and the to an opaque surface with windows (datConExtWin) within the Room
roof material instance is used for a parameter (line 4). instance parameters, (ii) glass material objects are instantiated by
Second, a boundary condition of the roof object is defined based on using a Glass class in the LBNL Modelica Buildings library, and (iii) win-
the rules of the LBNL Modelica Buildings library. Types of boundary con- dow structure instances are created by using the Window class in the
ditions we apply are opaque surfaces (datConExt), opaque surfaces with ModelicaBIM Class package.
windows (datConExtWin), interior walls between two thermal zones
(conBou, surBou), and interior partitions in a thermal zone (conPar).
Third, a room object is instantiated (line 7) and building envelope in- 4.2.4. Adding an interior door (Test Case 4)
formation is given as parameters of the room object. Six surfaces of the An interior door in BIM can imply a room-door-room association
room are categorized as opaque surfaces (line 8) and their layer infor- causing airflow between two zones through the door assembles and
mation (line 9), area (line 10), a tilt angle (line 11), and an azimuth the door openings in the energy model.
angle (line 12) are provided. In the Modelica model, a door class in the LBNL Modelica Buildings
library is wrapped, parameters for the door dimensions are added, and
4.2.2. Adding an interior wall (Test Case 2) Modelica connects are created. Our door class implements a closed
Adding an interior wall in BIM splits a single zone into two zones in door that simulate airflow through a door assembles. The door object
the Modelica model. From an energy simulation perspective, (i) an ex- calculates bi-directional airflow between the door and a room, so two
terior surface enclosing two zones needs to be divided into two surfaces Modelica connects are required for linking a door and each of the two
as shown in Fig. 3 and (ii) heat transfer through the interior surface be- rooms; therefore four Modelica connects are created for two rooms
tween two zones needs to be modeled. To split one zone into two zones, (Fig. 8).
the second Room object is added. New wall, roof, and floor objects that
are intersected with the interior wall are also instantiated and then
added to the two Room objects. 4.2.5. Adding stories (Test Case 5)
To build an interior surface in the Modelica model, materials, struc- In BIM, vertical stacking of multiple rooms in Test Case 5 and hori-
ture, and construction information of the interior wall are added. Then, a zontal expanding of multiple rooms in Test Case 2 are performed
boundary condition of the interior wall is assigned as an interior surface upon different approaches: creating a multi-story building in BIM
in the two Room objects (lines 3 and 6 in Fig. 7). Physical properties of needs new floor plans at different levels with new building envelope el-
the interior wall are added in the two Room objects. Lastly, the two ements. However, in BEM, heat transfer through an opaque surface be-
Room objects are connected by using a Modelica connect that links tween two stories needs to be simulated, which is similar to the heat
two rooms' connectors (line 11). transfer between two rooms in Test Case 2. While two thermal zones
in the Test Case 2 are connected by an interior wall, two rooms in the
4.2.3. Adding a window (Test Case 3) Test Case 5 are connected by the slab between the two stories.
When a window is installed in an exterior wall in BIM, the boundary In the Modelica model, the roof in the lower story and the floor in the
condition of the exterior surface in the energy model needs to be up- upper story are modeled as two opaque surfaces to perform conduction
dated to calculate heat flows through the window (e.g., solar radiation heat transfer. Then a Modelica connect is defined to link the two opaque
through glass panels, solar and infrared radiative heat exchange, surfaces.

Fig. 8. Modelica connects for a door between two rooms.

 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 25

Table 3 5.1. Building-level energy simulation

Annual peak temperature of test cases.

Cases Room name Max temperature (°C)/Date Min temperature (°C)/Date In sum, the simulation results of all test case models created by using
1 Room 33.88/Jul. 27th — 7 pm −20.90/Jan. 4th — 10 am
the ModelicaBIM library agree with results from the models created
2 East 33.00/Jul. 27th — 7 pm −19.46/Jan. 4th — 10 am with the LBNL Modelica Buildings library: two models of each test
West 33.01/Jul. 27th — 7 pm −19.41/Jan. 4th — 10 am case show identical simulation results of annual indoor air temperature
3 East 34.38/Jul. 27th — 5 pm −19.83/Jan. 4th — 7 am and annual heat flow through conduction, convection, and radiation of
West 33.39/Jul. 27th — 8 pm −19.68/Jan. 4th — 10 am
all five test cases. In Table 3, the highest temperature is obtained around
4 East 34.22/Jul. 27th — 5 pm −19.18/Jan. 4th — 7 am
West 32.83/Jul. 27th — 7 pm −17.96/Jan. 4th — 10 am 5–8 pm on July 7th and the lowest temperature is found around 7–
5 Upper 33.77/Jul. 27th — 5 pm −18.22/Jan. 4th — 7 am 10 am on January 4th. The maximum temperature is 34.38 °C at the
Lower 32.02/Jul. 27th — 8 pm −18.21/Jan. 4th — 10 am East room of Test Case 3 on 5 pm, July 27th. The minimum temperature
is −20.90 °C in the room of Test Case 1 on 10 am January 4th. These re-
sults also agree with those samples that we created using LBNL
Modelica Buildings library classes and their model structure.
5. Results and applications Figs. 9 and 10 show an indoor air temperature of Test Case 4, a dry
bulb outdoor temperature, and a global horizontal radiation in the
To validate the mapping method between BIM and ModelicaBIM, we first week of February and August, respectively. Both our model and
conducted simulation result comparisons between our library and the the LBNL model have an east room with two windows and a west
LBNL Modelica Buildings library. Five test cases are paired with two room without windows. Each room of both models shows an identical
models: one is created by using the classes and model structures of indoor temperature variation: two data series of each room are over-
the ModelicaBIM library and the other is modeled by using classes and lapped on each other. While we present the Test Case 4, other test
model structures of the LBNL Modelica Buildings library. cases show the same indoor temperature change between ModelicaBIM
We hypothesize that the two Modelica models of each test case can library and the LBNL Modelica Buildings library.
generate close or identical simulation results if the ModelicaBIM classes The two rooms have similar patterns in the indoor temperature
and the ModelicaBIM Structure Example package are correctly created curves, but the west room curve shifts to right side of the east room
in two conditions. curve. When the curves of global horizontal radiation and outdoor dry
bulb temperature are upward sloping, increasing of indoor temperature
• First, our ModelicaBIM classes accurately wrap solver classes of the starts from east rooms because of their two windows. On the other
LBNL Modelica Buildings library by using the new parameters and hand, dropping of indoor temperature also starts from east rooms due
the new object hierarchy. to heat loss through their windows.
• Second, our ModelicaBIM structure Example package can be used to While these result comparisons are performed for validation of basic
build the same architectural models as the LBNL Modelica Buildings cases, more extensive validation of the simulation results was also con-
library. ducted. We implemented a test case model using ANSI/ASHRAE Stan-
dard 140–2007 that has been used for the building simulation tool
Coherent model conditions across five test cases are as follows:
validation [24]. Annual heating and cooling loads as well as their peak
• The floor is not attached to the soil but above the ground level. days are compared to five simulation tools and the LBNL Modelica Build-
• One room has a single thermal zone. ings library. All the results are very close to LBNL's. In addition, the LBNL
• The building is located in the same location (Denver, Colorado, USA). Modelica Buildings library has been tested and validated [29] and it is
• The building has no internal heat gains from occupancy and equip- concluded that the simulation results of Buildings library are within the
ment. range specified in the ANSI/ASHRAE standards. Overall, the ModelicaBIM
• The door and windows are closed. library generates similar results to the Buildings library.
• No HVAC systems are included.

All simulations have been performed using Dymola 2012 simulation 5.2. Component-level energy simulation
program, LBNL Modelica Buildings library version 1.3, a solver tolerance
of 10−5, and a simulation interval of 3600 s for one year. First, building- Energy simulation using the ModelicaBIM library can calculate
level simulation results are compared. Then, building component-level component-level energy performances. We compared heat transfer
simulation results are compared for the five test case models. through individual building components of the five test cases, and the

15 1000
900
Temperature ( C)

10
800
° 5 700
Wh/m2

0 600
500
-5 400
-10 300
200
-15
100
-20 0
1 2 3 4 5 6 7
Date
West room East room Dry Bulb Temperature
West room (LBNL) East room (LBNL) Global horizontal radiation

Fig. 9. Indoor air temperature of Test Case 4 in the first week of February.
26 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

40 1000
35 900

Temperature ( C)
800
° 30
700

Wh/m2
25 600
20 500
15 400
300
10
200
5 100
0 0
1 2 3 4 5 6 7
Date
West room East room Dry Bulb Temperature
West room (LBNL) East room (LBNL) Global horizontal radiation

Fig. 10. Indoor air temperature of Test Case 4 in the first week of August.

same simulation results are obtained. Figs. 11 and 12 show convective associated with individual BIM objects, which can inform design-
heat transfer values of Test Case 4 in a single day (February 7th). Be- decision making to improve the overall building performance.
cause of the same simulation results of each test case models in both li-
braries, only the results of the ModelicaBIM library are visible in figures. 5.3. Discussion on simulation results
The test case has two rooms and each room has six surfaces as
shown in Table 2, so two plots present both the west and the east Our library and the LBNL's library yield identical simulation results
room temperatures. Convective heat transfer values of individual build- for the five test cases presented in this paper. However, different but
ing components are categorized as heat gain and heat loss and then they very close simulation results may occur when we change the order of
are accumulated respectively. a room's components used as arguments in Modelica thermal calcula-
In each bar chart, taller bars above zero imply higher convective heat tion functions.
gain and shorter bars below zero imply lower convective heat loss. As shown in the room object in Fig. 6, for example, we assign the
Overall, the east room has higher heat gain and heat loss values than building component data as the room's parameters. In the five test
the west room. At the component level, one having a taller bar above cases, we input the roof data first, then floor, exterior walls, and interior
zero contributes more to heat gain of a thermal zone than other compo- walls — the same order as in LBNL's models. When we switch the order of
nents. For instance, the blue bars in Fig. 12 imply that the south wall of the components, different but very close simulation results are obtained.
the east room contributes more to heat gain through convective heat As an example, we switched the order of the roof and the floor and cal-
transfer than other surfaces. The east room having windows drives culated the annual convective heat transfer differences between our
more heat gain and loss than the west room due to the windows. The models and LBNL's models. The difference is defined as the area between
heat flow of east room fluctuates more than the west room; the highest the two models' curves, representing the time series of heat transfer, di-
and lowest values are shown in the east room; and the maximum value vided by the area of the LBNL's model. For the roof of the west room in
is found around 11 a.m. in the east room and around 4 p.m. in the west Test Case 4, the difference is 5.7%. For the floor of the same room, the dif-
room. ference is 8.3%. In the building level in Test Case 4, the difference is 6.2%.
Fig. 13 summarizes the annual convective heat transfer in accordance Reducing model tolerance in Dymola will reduce these errors.
with building components; 12 surfaces of two rooms in Test Case 4 are
presented. Both the south and the east wall having windows in the 6. Conclusions
east room contribute to heat gain and loss mostly, and both north walls
of the two rooms cause heat loss in general. The east room floor shows This paper presents the development and validation of the
more heat gain than the west room floor because of solar radiation ab- ModelicaBIM library for BIM-based building energy simulation using
sorption from two windows. While we present only a couple of plots, Modelica in the scope of the building envelope. Based upon our method-
more diverse component level simulation results can be generated and ology, building topology, geometry, and materials in BIM can be

Fig. 11. Convective heat transfer of the west room of Test Case 4 on February 7th.
 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28 27

Fig. 12. Convective heat transfer of the east room of Test Case 4 on February 7th. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

translated into Modelica building energy models. The library is used to parameters in Modelica and then the parameters can be used in energy
create Modelica building energy models that can represent object seman- simulation. Using the LBNL Modelica Buildings library written in
tics of BIM. Classes of the ModelicaBIM package can store object attributes equation-based object-oriented modeling makes our implementation
of BIM as their parameters and then use the information for energy sim- faster than our previous approach on using Fortran-based ASHRAE
ulation. The ModelicaBIM structures of our library can represent object se- Loads Toolkit [24].
mantics of BIM with same or similar object hierarchy among building BIM and Modelica have been developed not only by different parties
components. but also with different perspectives. We emphasize the benefits and the
Modelica energy models using our classes and model structures gen- potentials of integrating them for BIM-based energy performance anal-
erate identical simulation results with models using the LBNL Modelica ysis in the design stage. Our research will contribute to the advanced
Buildings library in result comparison (e.g. indoor temperatures and system development for sustainable building and community design.
component-level heat transfers). For future work, new object classes and model structures will be in-
Our methodology enables to produce component-level simulation vestigated for complex building geometry and the HVAC systems. A set
results that can indicate the performance of heat transfer of individual of parameters within the ModelicaBIM Class package and a series of
buildng components. Coherent model structures and object hierarchy modeling rules in the ModelicaBIM Structure Example package will be
between BIM and BEM allow direct feedback from energy analyses to utilized as new system interfaces between BIM and energy simulation
building design without complex result interpretation. Through the vi- that perform automatic translation from BIM to BEM.
sualization capabilities that BIM authoring tools provide, each
component's performance can be visualized in the Revit models and Acknowledgements
time-series simulation results can also be animated through custom ap-
plications written with the Revit API [39]. This material is based upon work supported by the National Science
As an intermediate library between BIM and Modelica, our Foundation under Grant No. 0967446. Any opinions, findings, and con-
ModelicaBIM library facilitates translation from BIM to Modelica clusions or recommendations expressed in this material are those of the
BEM; attributes of Revit models can be converted into energy model authors and do not necessarily reflect the views of the National Science
Foundation. We appreciate the valuable input from Sandeep Kota, Jose
Luis Bermudez Alcocer, and the BIM-SIM Group, and the contribution
from Mateo Aviles at Texas A&M University to the project.

References
[1] D.B. Crawley, J.W. Hand, M. Kummert, B.T. Griffith, Contrasting the capabilities of
building energy performance simulation programs, Proceedings of Building Simula-
tion 2005: Ninth International IBPSA Conference, Montreal, Quebec, Canada, 2005,
pp. 231–238.
[2] S. Attia, State of the art of existing early design simulation tools for net zero energy,
Buildings: A Comparison of Ten Tools (Technical Report). Louvain La Neuve,
Belgium, 2011.
[3] J.S. Haberl, SIMBUILD survey: academic use of simulation software, Presented at
SIMBUILD 2008, Berkeley, California2008.
[4] T. Maile, M. Fischer, V. Bazjanac, Building energy performance simulation tools — a
life-cycle and interoperable perspective, CIFE Working Paper #WP107, Stanford
University, 2007.
[5] A. Aksamija, BIM-based building performance analysis: evaluation and simulation of
design decisions, Fueling Our Future with Efficiency. Presented at the 2012 ACEEE
Summer Study on Energy Efficiency in Buildings, Pacific Grove, California, 2012.
[6] S. Attia, L. Beltrán, A.D. Herde, J. Hensen, “Architect friendly”: a comparison of ten
different building performance simulation tools, Presented at the Building Simula-
tion 2009, Glasgow, Scotland2009. 204–211.
[7] J.A. Clarke, Energy Simulation in Building Design, Taylor & Francis, 2001.
[8] B. Dong, K.P. Lam, Y.C. Huang, G.M. Dobbs, A comparative study of the IFC and
gbXML informational infrastructure for data exchange in computational design sup-
port environments, Proceedings of Building Simulation 2007: Tenth International
IBPSA Conference, Beijing, China, 2007, pp. 1530–1537.
[9] State-of-the-art of digital tools used by architects for solar design, in: M.C. Dubois,
M. Horvat (Eds.), IEA SHC Task 41: Solar Energy and Architecture, ST‐B: Methods
Fig. 13. Annual convective heat transfer in the component level. and tools for solar design, T.41.B, 2010.
28 J.B. Kim et al. / Automation in Construction 50 (2015) 16–28

[10] U.S. Department of Energy (U.S. DOE), Building Energy Software Tools [26] R. Sacks, C.M. Eastman, G. Lee, Parametric 3D modeling in building construc-
DirectoryRetrieved from http://www.energytoolsdirectory.gov2012. tion with examples from precast concrete, Autom. Constr. 13 (3) (2004)
[11] NREL (National Renewable Energy Laboratory), OpenStudioRetrieved from http:// 291–312.
openstudio.nrel.go2013. [27] P.A. Fritzson, Principles of object-oriented modeling and simulation with Modelica
[12] Lawrence Berkeley National Laboratory, Simergy, Computer SoftwareRetrieved 2.1. IEEE, 2004.
from http://energy.lbl.gov/bt/simergy2013. [28] Dassault Systèmes, Dymola, Computer SoftwareRetrieved from http://www.3ds.
[13] J. O'Donnell, R. See, C. Rose, T. Maile, V. Bazjanac, P. Haves, SimModel: a domain data com/products-services/catia/portfolio/dymola2013.
model for whole building energy simulation, 12th International Conference of the [29] T.S. Nouidui, K. Phalak, W. Zuo, M. Wetter, Validation and application of the room
International Building Performance Simulation Association (IBPSA), November model of the Modelica Buildings library, Proceedings of the 9th International
14–16, 2011. Sydney, Australia, 2011, pp. 382–389. Modelica Conference, Munich, Germany, 2012, pp. 727–736.
[14] S. Kumar, Interoperability Between Building Information Models (BIM) and Energy [30] M. Wetter, W. Zuo, T.S. Nouidui, Modeling of heat transfer in rooms in the Modelica
Analysis Programs (Master's Thesis), University of Southern California, Los Angeles, ‘Buildings’ library, Proceedings of the 12th IBPSA Conference, Sydney, Australia,
California, 2008. 2011, pp. 1096–1103.
[15] Vasari Autodesk, Computer SoftwareRetrieved from http://autodeskvasari.com2013. [31] M. Wetter, C. Haugstetter, Modelica versus TRNSYS — a comparison between an
[16] General Services Administration, 3D-4D Building Information ModelingRetrieved equation-based and a procedural modeling language for building energy simulation,
from http://www.gsa.gov/portal/category/210622012. Proceedings of the 2nd SimBuild Conference, Cambridge, MA, USA, 2006, pp.
[17] V. Bazjanac, IFC BIM-based methodology for semi-automated building energy 262–269.
performance simulation, Presented at CIP-W78, Santiago, Chile2008. [32] M. Venugopal, C.M. Eastman, R. Sacks, J. Teizer, Semantics of model views for infor-
[18] J. O'Donnell, T. Maile, C. Rose, N. Mrazovic, E. Morrissey, C. Regnier, P. Parrish, V. mation exchanges using the industry foundation class schema, Adv. Eng. Inform. 26
Bazjanac, Transforming BIM to BEM: generation of building geometry for the (2) (2012) 411–428.
NASA Ames sustainability base BIM, LBNL Report #LBNL-6033E, University of Cali- [33] R. See, J. Karlshøj, D. Davis, An integrated process for delivering IFC based data
fornia, Berkeley, 2013. exchangeRetrieved from http://bips.dk/files/bips.dk/integrated_idm-mvd_
[19] V. Bazjanac, A. Kiviniemi, Reduction, simplification, translation and interpretation in the processformats_14_0.pdf2011.
exchange of model data, Presented at CIB-W78, Maribor, Slovenia2007. 163–168. [34] C.M. Eastman, I. Panushev, R. Sacks, M. Venugopal, V. Aram, R. See, A guide
[20] M.J. Clayton, J. Haberl, W. Yan, S. Kota, F. Farías, W. Jeong, J. Kim, J. Bermudez Alcocer, for development and preparation of a national BIM exchange standard,
Development of a Reference Building Information Model (BIM) for Thermal Model Technical Report, PCI, Charles Pankow Foundation, Georgia Tech, Technion,
Compliance Testing, RP-1468, ASHRAE, 2013. 2011.
[21] T. Maile, J. O'Donnell, V. Bazjanac, BIM — geometry modelling guidelines for energy [35] C.M. Rose, V. Bazjanac, An Algorithm to generate space boundaries for building en-
performance simulation, 13th International Conference of the International Building ergy simulation, Eng. Comput. 29 (4) (2013).
Performance Simulation Association (IBPSA), August 25–28, 2013. Chambéry, [36] S.N. Woodfield, The impedance mismatch between conceptual models and imple-
France, 2013, pp. 3242–3249. mentation environments, Proceedings of the ER'97 Workshop on Behavioral Models
[22] Y.N. Bahar, C. Pere, J. Landrieu, C. Nicolle, A thermal simulation tool for building and and Design Transformations: Issues and Opportunities in Conceptual Modeling,
its interoperability through the Building Information Modeling (BIM) platform, UCLA, Los Angeles, California, 1997.
Buildings 3 (2) (2013) 380–398. [37] M. Wetter, Multizone airflow model in Modelica, Proceedings of the 5th Interna-
[23] M. Wetter, Modelica-based modeling and simulation to support research and develop- tional Modelica Conference, Vienna, Austria, 2006, pp. 431–440.
ment in building energy and control systems, J. Build. Perform. Simul. 2 (2) (2009) [38] M. Wetter, W. Zuo, T.S. Nouidui, X. Pang, Modelica Buildings library, J. Build.
143–161. Perform. Simul. 7 (4) (2014) 253–270.
[24] W. Yan, M. Clayton, J. Haberl, W. Jeong, J. Kim, S. Kota, J. Bermudez Alcocer, M. Dixit, [39] W. Jeong, J. Kim, M.J. Clayton, J.S. Haberl, W. Yan, Visualization of building energy
Interfacing BIM with building thermal and daylighting modeling, 13th International performance in building information models, The ACADIA (The Association for
Conference of the International Building Performance Simulation Association Computer-Aided Design in Architecture) 2013 International Conference, October
(IBPSA), August 25–28, 2013. Chambéry, France, 2013, pp. 3521–3528. 21–23, Cambridge, Ontario, Canada, 2013.
[25] C.M. Eastman, P. Teicholz, R. Sacks, K. Liston, BIM Handbook: A Guide to Building In-
formation Modeling for Owners, Managers, Designers, Engineers, and Contractors,
Wiley, Hoboken, N.J., 2011