Conference PaperPDF Available

Automatic generation of second level space boundary geometry from IFC models

Authors:

Abstract

Building Information Modeling aims at the use of digital building data in all planning processes. However, several challenges arise when using BIM data in the IFC file format to create building simulation models. Complex, detailed or erroneous data and the lack of software interfaces complicate and delay the simulation preprocessing. One way to reduce the simulation expert's effort is to provide space boundary data, which can be applied in both CFD and BEPS. This paper presents an algorithm for the generation of space boundaries based on the IFC data model. The focus is on model error resistance and simultaneous support of CFD and BEPS to accelerate simulation preprocessing, avoid redundant geometric model setup, and simplify the simulation result exchange. Key innovations • Introduction of an automated preprocessing algorithm generating space boundaries • Application in the setup of simulation models for CFD and BEPS • Reduction of the required preparation time of a simulation model and removal of modeling errors Practical implications This paper introduces an algorithm for space boundary generation based on IFC building models. The proposed method simplifies and accelerates the application of IFC data in numerical building simulation software.
Automatic generation of second level space boundary geometry from IFC models
Eric Fichter, Veronika Richter, J´erˆome Frisch, Christoph van Treeck
Institute of Energy Efficiency and Sustainable Building (E3D),
RWTH Aachen University, Germany
Abstract
Building Information Modeling aims at the use of dig-
ital building data in all planning processes. However,
several challenges arise when using BIM data in the
IFC file format to create building simulation mod-
els. Complex, detailed or erroneous data and the lack
of software interfaces complicate and delay the sim-
ulation preprocessing. One way to reduce the sim-
ulation expert’s effort is to provide space boundary
data, which can be applied in both CFD and BEPS.
This paper presents an algorithm for the generation
of space boundaries based on the IFC data model.
The focus is on model error resistance and simultane-
ous support of CFD and BEPS to accelerate simula-
tion preprocessing, avoid redundant geometric model
setup, and simplify the simulation result exchange.
Key innovations
Introduction of an automated preprocessing algo-
rithm generating space boundaries
Application in the setup of simulation models for
CFD and BEPS
Reduction of the required preparation time of a
simulation model and removal of modeling errors
Practical implications
This paper introduces an algorithm for space bound-
ary generation based on IFC building models. The
proposed method simplifies and accelerates the appli-
cation of IFC data in numerical building simulation
software.
Introduction
Building Information Modeling (BIM) refers to a co-
operative working methodology based on digital data
that contains all information relevant to a building’s
life cycle. An essential component of BIM is the 3D
building model, which is continuously enriched with
additional or updated information as the planning
process progresses. Typical planning information in-
cludes geometries and properties of architectural and
technical building components as well as time, cost,
and maintenance planning data. To enable the ex-
change of building data among project participants,
the international organization buildingSMART has
developed the open data model Industry Foundation
Classes (IFC) (ISO 16739-1:2018).
In numerical building simulations, the use of existing
digital data can contribute to improved and simpli-
fied processes. For example, time-consuming geomet-
ric remodeling could be avoided. However, since the
data is usually not designed and exported simulation-
specific, extensive preparation steps may have to be
carried out to eliminate modeling and export er-
rors. To automate the model preparation, various
programs have been developed in academic research.
Some of them are out-dated or no longer accessible.
The algorithm presented in this paper is part of a
toolchain, which aims to provide the basic methods to
derive simulation models from digital building data.
Therefore, the proposed algorithm generates Space
Boundaries (SBs) based on IFC files for the applica-
tion in Computational Fluid Dynamics (CFD) and
Building Energy Performance Simulation (BEPS).
The paper first gives a brief overview of SBs, their
generation, and challenges in simulation applications.
Subsequently, the requirements for the simulation
and the SB generation algorithm are explained. Fi-
nally, the results are shown and applied in simulations
using an academic IFC example model.
Space Boundary Definition
SBs are the boundary surfaces of spaces and the in-
terfaces of building elements in contact with air. In
building-related CFD, SBs are usually required to in-
vestigate the air volume for temperatures, flow veloc-
ities, and gas concentrations by imposing boundary
conditions on them. In BEPS, SBs are the heat trans-
fer surfaces between zones of different temperatures.
A distinction is made between first level and second
level SBs, see figure 1. At the first level, SBs are
pure space delimiters. At the second level, SBs take
changes in building elements or spaces on the other
side into account, which makes them well suited for
the application in BEPS. Regardless of the level, SBs
cannot be explicitly extracted from a purely geomet-
ric IFC model.
Within the IFC standard in version 4.1, published by
buildingSMART in 2018, SBs are represented by the
class IfcRelSpaceBoundary. In addition to the geo-
metric representation and the assignment to a space
and a building element, SBs can be characterized by
their boundary type. For instance, second level SBs
can be internal or external, physical or virtual, and
of type 2a or 2b. While 2a SBs allow heat exchange
between two spaces, 2b SBs are in front of a heat
blocking building element (figure 1).
Spaces separated by walls First level
Second level, type 2a Second level, type 2b
Figure 1: Distinction between first and second level
SBs (based on buildingSMART (2018)).
State of the art
Challenges using IFC data
According to the authors’ research, the use of 3D
geometries from IFC files in CFD has only re-
cently been applied in industry. Among other rea-
sons, that is because IFC files are not provided
and software to handle them is unknown or un-
available. Meanwhile, an increasing number of soft-
ware providers are creating IFC import interfaces in
their tools. These include, for example, the com-
mercial programs BIMHVACTool, a graphical pre-
processor for OpenFOAM (Weller et al., 1998), and
BIM inside ANSYS, an interface for the ANSYS sim-
ulation environment. Besides, there are free and par-
tially web-based services for converting IFC geome-
tries into various geometry data formats such as Stan-
dard Triangle Language (STL) and Wavefront OBJ,
for example IfcConvert and BIMvision.
The use of IFC files can be challenging with or with-
out commercial software, especially for huge models
with unnecessary objects, corrupted geometries, and
a high level of detail. Also, modeling issues and ex-
port errors such as overlapping objects or gaps be-
tween objects can cause difficulty. Moreover, when
converting the IFC into a geometry format, BIM
data such as object attributes, inheritance informa-
tion, and relations of objects to other entities is lost.
Hence, the motivation of the proposed algorithm is to
automate parts of the preprocessing steps that have
to be conducted for CFD mesh creation.
Multiple BEPS tools provide an interface to IFC.
Nevertheless, according to the buildingSMART Inter-
national Standards Implementation Database (2020),
the majority of them only supports the prior version
IFC2x3. This concerns, for example, the commer-
cial tools Simergy (2020) and IDA ICE as well as
the BimServer plugin OsmSerializer (2016) used in
the open source software OpenStudio (2020). The
EnergyPlus supporting middleware SimModel inte-
grated into Simergy (O’Donnell et al., 2011; Nytsch-
Geusen et al., 2019) and the toolchain BIM2Modelica
(Nytsch-Geusen et al., 2019) are restricted to IFC2x3,
too. According to Mediavilla et al. (2018), the Reno-
BIM software, a preprocessor for EnergyPlus devel-
oped within the BERTIM (2019) project, is capa-
ble of reading IFC4 data. The toolchain suggested
by the BIM2SIM project supports EnergyPlus sim-
ulation based on IFC4 files. For this software, the
availability of SBs is crucial as their representation
is used as geometry input for the Building Energy
Model (BEM). In the workflow proposed by Andria-
mamonjy et al. (2018), SBs must be defined as well.
Despite SBs are defined within the IFC standard,
several geometric and non-geometric issues can arise
when using SBs from IFC files in BEPS. Issues can be
caused by incorrect export settings, insufficient IFC
support in the authoring tool, and modeling that was
not adapted for BEPS. Typical geometric errors are
missing, duplicate, deformed, incorrectly oriented, or
overlapping SBs, that are not enclosing a space cor-
rectly. Also, SBs with missing or wrong semantic
information, e. g. concerning boundary types and re-
lated elements, can occur. A detailed description of
occurring issues was given by Maile et al. (2013).
Because of the potential errors mentioned, an IFC
model check with a focus on SBs should be performed
beforehand. However, there are only a few tools sup-
porting this. Visual verification can be performed in
IFC viewers. Also, syntactic correctness in compar-
ison with the IFC standard can be checked. Tools
that support such rule checking are the IfcCheck-
ingTool (2020), the Solibri Model Checker and Ifc-
Doc. The latter can also be used to implement SB-
specific custom rules, as Wimmer et al. (2017), Pin-
heiro et al. (2018) and Ying and Lee (2017) proposed.
Unfortunately, there is no comprehensive software for
checking geometric and complex BEPS-specific rules.
Though, Ying and Lee (2020) introduced a tool, de-
tecting geometric errors with a focus on watertight
geometry using a Monte Carlo ray tracing approach.
From the challenges described in this section, a need
for algorithms and tools generating correct SBs can
be derived.
Space boundary generation
SBs can be written to IFC at two different stages
in BIM. A generation can be done within the CAD
software during the export of the IFC file, which is
shared with the simulation expert. This would be
the preferred way since all information of the para-
metric modeled BIM objects and their relationships
among each other remains available. However, the
IFC export interfaces of the CAD tools may be un-
reliable. In such a case, the SB generation needs to
be based on the geometric IFC data. Due to the
IFC schema definition and the export of the CAD
software, information on the relationships between
the objects is largely lost. Hence, it must be algo-
rithmically regenerated. Another challenge for the
downstream generation of the SBs is the quality of
the IFC data, which is influenced by errors and the
level of geometry. Several algorithms to generate
SBs based on three-dimensional architectural build-
ing models were presented in academic research (van
Treeck and Rank, 2007; Rose and Bazjanac, 2013;
Jones et al., 2013; Ladenhauf et al., 2016; Lilis et al.,
2016; Nytsch-Geusen et al., 2019). They are summa-
rized in table 1 listing important prerequisites and
methods. Most of the implementations that arose
from the algorithms are not actively supported any-
more, are based on prior IFC versions, or are not
publicly accessible.
Typically, SB generation is done in three main steps.
In the first step, the fulfillment of the semantical and
geometric prerequisites of the algorithm must be en-
sured by preprocessing. Especially to reduce clashes
between objects beforehand, Boolean operations have
to be used. This is for example demonstrated in Lilis
et al. (2015), proposing methods to ensure the needs
of their CBIP algorithm (Lilis et al., 2016). Impor-
tant requirements of the algorithms mentioned above
are summarized in table 1.
In the second step, first level SB geometry is gen-
erated. For this, some approaches rely on the pres-
ence of IfcSpace geometry. If IfcSpaces are used, first
level SBs are generated by intersecting the faces of
spaces and building elements by Boolean operation,
which requires a correct and consistent model without
gaps and intersections. Algorithms that do not rely
on IfcSpaces must extract faces in contact with air
by other approaches. Van Treeck and Rank (2007)
create an edge-face graph of the building by mutu-
ally intersecting solid objects using Boolean opera-
tions. Evaluating the graph’s adjacency information
allows detecting first level SBs and ensures watertight
spaces. Jones et al. (2013) identify spaces by ana-
lyzing view factors between polygons calculated by
ray-face intersection.
Second level SB geometry is generated in the third
step. Except for van Treeck and Rank (2007), the
mentioned algorithms use a projection approach. In
this, polygon pairs are determined according to var-
ious criteria such as parallelism, distance, and the
presence of reverse-oriented surface normals. SBs are
then created by mutual projecting and clipping the
paired faces. In the approach of van Treeck and Rank
(2007), imprints on faces created by extruded solids
describe a subset of second level SBs.
A combination and adaptation of the presented ap-
proaches has the potential to make the SB genera-
tion more error resistant and to fulfill the semantic
and geometric requirements of CFD and BEPS. This
can lead to acceleration of simulation preprocessing,
avoidance of redundant geometric model setups and
simplification of the exchange of simulation results.
Simulation demands
For a CFD simulation of a zone based on STL input
files, a manifold watertight triangular surface mesh
is needed, which contains only objects relevant to set
boundary conditions. To simplify the meshing pro-
cess, the faces should be consistent and correctly ori-
ented, and triangulated in a way that a vertex is con-
nected to three edges. Inner, double, and overlapping
coplanar surfaces, as well as surface intersections, are
not allowed. The above requirements can be satis-
fied by the proposed algorithm. However, remeshing
of the geometry must be done in separate meshing
software, for example, to eliminate faces with high
perimeter to area ratio or faces with very acute an-
gles. Also, a collision of one or more correct geome-
tries in a vertex or edge can cause problems when
generating a surface mesh, even though it is geometri-
cally valid. Such cases must be solved either manually
or by a wrapping algorithm in the meshing software.
For BEPS, the requirements of the IFC standard must
be met in particular. Additional requirements must
also be fulfilled:
SBs form a closed shell around the space
SB normals pointing outward of the space
Spaces and SBs shall not intersect each other
Simplifying container elements (IfcCurtainWall)
Link to IfcExternalSpatialElement for facade SBs
The SBs of a void filling object are coplanar to the
SBs of the parent element
Methods
The proposed algorithm aims at generating SBs that
meet both, the stated requirements for CFD and for
surface-based IFC SBs. To be independent of data
and model correctness, IfcSpace geometry should not
be used for first level SB generation. For this rea-
son and to generate watertight adjacent first level
SBs, the approach of van Treeck and Rank (2007) is
suitable. However, it is adapted to a purely surface-
based approach. Thus, no closed BREP solids with
correct normals are assumed, which are necessary in
Table 1: Space boundary generation algorithms. References to columns from left to right: van Treeck and Rank
(2007), Jones et al. (2013), Rose and Bazjanac (2013), Ladenhauf et al. (2016), Lilis et al. (2016)
Tre07 Jon13 Ros13 Lad16 Lil16 Proposed
Prerequisites for algorithm
IfcSpaces for 1st level SBs × × ×
Absence of clashes/gaps between space and construction × × ×
Absence of clashes/gaps between constructions × × ×
Watertight solid BREPs × × (×)×
Correct surface normals × × × ×
Geometric representation solid polygon polygon polygon polygon all faces
Method for 1st level SB generation
Clipping of space and construction faces × × ×
Detection by view factors ×
Evaluating edge-face graph created by Bool. operation × ×
Method for 2nd level SB generation
Projection and clipping of faces × × × × ×
Extension of solids/faces before 1st level SB generation × ×
Could generate geometry and properties of IFC4 SBs × × × × ×
other algorithms, for example, to eliminate collisions
between components. Furthermore, virtual faces or
incorrectly modeled objects, e. g. unrealistically thin
objects, can be handled. Based on the extracted
first level SBs, projection and clipping allow the gen-
eration of all second level subdivisions required by
IFC. Ray tracing operations ensure the distinction of
type 2a and 2b faces and recognition of correspond-
ing faces, to address changes in material, building el-
ement or space. The algorithm involves several steps:
1) Parsing IFC data and conversion to BREP
2) Geometric healing (partially optional)
3) Geometric simplification (partially optional)
4) Creation of face-only data structure
5) Offsetting faces
6) Creation of edge-face graph by Boolean operation
7) Traversing graph to identify e. g. first level SBs
8) Export for CFD as STL (optional)
9) Projection and clipping of faces
10) Identification of SB properties by ray tracing
11) Merging of SBs with same properties
12) Export for BEPS as IFC
The algorithm is currently being implemented in C++.
The IFC data is parsed and geometrically interpreted
using IfcOpenShell (2020), a software library support-
ing IFC’s geometric resources. It is based on the open
source 3D modeling kernel Open Cascade Technology
(OCCT, 2020), which is later used for the geomet-
ric operations of the presented algorithm. Using Ifc-
OpenShell, OCCT shapes are created for the three-
dimensional building elements of relevant IFC classes.
The shape geometry is stored as BREP, with no re-
strictions on the curvature of edges and faces.
Procedure
Especially for complex objects with high levels of de-
tail, the shapes can be faulty. The reason for this
is usually an incorrect geometric description in the
IFC file. A common problem is the presence of non-
closed shells and inner, overlapping, or wrongly ori-
ented faces. As long as the gaps in the hull of the
shape do not exceed a critical value (a few centime-
ters), this kind of geometry does not affect the algo-
rithm. Yet, to speed up later processes, an attempt
is made to heal faulty shapes. For this, smaller gaps
are first closed by face extension. Then, the envelop-
ing surface of the shape is extracted according to the
geometric analysis proposed by van Treeck and Rank
(2007), ensuring a consistent face orientation. If pos-
sible, coplanar adjacent surfaces are unified.
The result of a healing process is exemplified in fig-
ure 2. The staircase is taken from the IFC build-
ing presented later (figure 5). It consists of a set
of faces that form partially closed and partially non-
closed volumes. Moreover, the staircase has inner and
wrongly oriented faces as well as faces that are unnec-
essarily triangulated. As figure 2 (b) shows, the errors
can be eliminated by the integrated healing process.
(a) before (b) after
Figure 2: Healing of an IfcStair’s geometry. Inner
surfaces are colored red, faces with incor-
rect normals (arrows) are shown in blue.
Window and door objects are often described geo-
metrically detailed, which is not practical for sim-
ulation. Therefore, the geometry of associated
IfcOpeningElements is used instead. Because of the
face-based approach, geometric correctness of a wall
is irrelevant, as long as the usually simple opening
element is correct. Thus, an opening can be ex-
pressed as two faces lying on two wall faces, reducing
the geometric complexity. Assemblies of components,
such as IfcCurtainWalls, are simplified by rectangu-
lar enveloping surfaces as required by BEPS. Figure 3
shows some simplifications and indicates that holes
can be closed within the process, too.
(a) before (b) after
Figure 3: Geometric simplification of windows and a
revolving door.
Regardless of healing and simplification, all surfaces
are collected from the BREPs. Subsequently, the
faces are offset by a few centimeters to fill gaps within
and between components. Boolean intersection of the
faces and adjacency evaluation results in an edge-face
graph of the building, in which faces are connected
to other faces via edge pairs. Faces with unconnected
edges are removed. The orientation of a face’s surface
normal is not important for the subsequent processes.
However, if the faces were part of a healed shape, the
correct orientation is known.
To find first level SBs and to detect inner faces, the
graph is traversed starting from a randomly selected
face, collecting neighbor faces recursively. If the cor-
rect face normal of the neighbor face is known but not
compatible, the search is interrupted and restarted
with another initial face or another face orientation.
If it is unknown, the orientation of the face is adjusted
to be compatible. If a surface has several neighbors
at one edge, the surface with the smaller angle is se-
lected. The search ends with a closed space if no
more new faces are found. Checks follow to verify
the correctness of the found space. Spaces must not
overlap, the surface normals must point into the en-
closed volume, and component surfaces may only be
used once. For CFD-based applications, the clustered
faces are forwarded to the triangulation and export
process along with information about a SB’s related
space and building element.
To achieve a geometric second level subdivision, faces
that do not exceed a maximum distance, are mutu-
ally projected and clipped. The maximum distance is
a user-defined value, which is also used to distinguish
the boundary types (2a or 2b). For BEPS, attributes
like boundary type, corresponding face, or material
composition between the face pair need to be as-
signed. This is done based on a face-line-intersection
(ray tracing) as depicted in figure 4. Here, lines are
generated based on SB surface normals and examined
for contact with other faces. The first level SB that
is hit first, represents the corresponding SB. If no SB
is hit within the maximum distance, a 2b boundary
type is assigned. Changes in the material are derived
from the order in which the faces are hit. Finally,
SBs are merged with their neighbors if they share all
attributes.
2a SB
Wall Column
1
2
3, 4
Figure 4: Face-line-intersection between two 2a SBs
(red) seen from above. There are four in-
tersection points: wall face (1), column
face (2), wall and column face (3, 4).
Data export
As described in the requirements, a triangulated mesh
should be created that just needs to be stitched before
the surface meshing can be performed. During stitch-
ing, adjacent triangles are topologically connected via
geometrically identical edges. Nonetheless, surface
wrapping of the geometry is required, if additional
objects are added after the export. The geometry
is exported as a single STL file. The file contains
STL solids, which are composed of faces belonging to
the same space and the same building element. This
approach maintains control over all objects. Thus,
individual faces, objects, IFC classes, and spaces can
be filtered via regular expressions e. g. in the CFD
software. The algorithm ensures consistent triangula-
tion, but mesh quality criteria such as skewness (ratio
of internal angles against optimal angle) and aspect
ratios cannot be influenced. This is an acceptable
procedure since geometry is remeshed anyway when
creating surface meshes suitable for CFD purposes.
To export the created objects into the IFC data for-
mat, instances of the classes IfcRelSpaceBoundary,
IfcSpace and IfcExternalSpatialElement are gener-
ated first. In addition to the geometric representa-
tion, attributes such as name, description, and the
globally unique identifier are then assigned to the en-
tities. For this purpose, a conversion of the geomet-
ric classes of OCCT into IFC-based instances is per-
formed using IfcOpenShell. Finally, the entities are
linked together to describe the related objects.
Results
For this paper, the IFC4 Phantasy office building of
the Institute for Automation and Applied Informatics
(IAI) of the Karlsruhe Institute of Technology (KIT)
is processed (figure 5). It was chosen because of its
good representability and the presence of SBs, which
are used for validation of the proposed algorithm.
Figure 5: The example IFC building.
STL result
Figure 6 (a) shows a clipped visualization of the re-
sulting STL file. It can be seen that the assignment to
the IFC products is preserved despite tessellation and
a consistent triangulation exists. However, the trian-
gulation is not optimized for angles or other mesh
parameters, which is up to the meshing tool. Fig-
ure 6 (b) depicts the staircase zone selected by text
filtering based on the naming convention.
(a) clip (b) staircase zone
Figure 6: Filtered STL result file.
IFC result
The enriched and exported IFC file contains an
IfcExternalSpatialElement, 78 IfcSpaces and almost
2000 IfcRelSpaceBoundaries. The added entities
are visualized in figure 8. While subfigure (c)
shows the external SBs belonging to IfcSpaces,
the subfigures (d) and (e) depict the SBs of the
IfcExternalSpatialElement. The typing of the SBs in
contact with the ground is not yet complete, since the
IfcSite is not yet considered by the algorithm. Unlike
the original office model, the enriched model provides
SBs of type 2b, creating watertight spaces.
Application
In this section, the generated STL and IFC files will
be applied to the simulation process.
CFD
The application of the STL file is shown by means of
a steady-state airflow simulation in the staircase, in-
vestigating a smoke protection pressure system. The
simulation is performed for summer conditions with
closed doors. For this purpose, the STL file to be sim-
ulated was imported into Fluent Meshing. By naming
the surfaces according to zone, IFC class, and related
building element, it was possible to delete all building
areas except the staircase and to impose the bound-
ary conditions on the desired surfaces in an easy and
straight forward manner.
Due to the appropriate preparation, wrapping was
not necessary. Instead, stitching was performed,
which was based solely on vertex merging. Figure 7
(a) shows the used surface mesh. For volume mesh-
ing, the unstructured triangular surface mesh was
converted to a hexagonal mesh consisting of 2.8 mil-
lion cells. The simulation was performed in Fluent
using the SST k-ωturbulence model. As one of three
flow variables, the resulting air pressure is shown in
figure 7 (b).
(a) Surface mesh (b) Relative pressure in Pa
Figure 7: Indoor air flow simulation of a stairway us-
ing the exported STL geometry.
BEPS using EnergyPlus
The simulation of the office model enriched with the
SBs was performed in EnergyPlus. The needed Ener-
gyPlus Input Data File (IDF) was generated using the
BIM2SIM Python toolchain. This toolchain is based
on IfcOpenShell and GeomEppy (2016) to convert the
SBs objects to IDF surface geometries. According to
the IFC file, the IDF surfaces were positioned on the
building component surfaces, to correctly represent
the thermal mass of the building. SBs of type 2b
were used to close the spaces and set to be thermally
adiabatic. Each IfcSpace was transferred as a sin-
gle zone and all non-geometric simulation parameters
such as heating and cooling set points, loads, sched-
ules, construction, and materials were chosen accord-
ing to standardized templates.
Before importing the IFC file into the BIM2SIM
toolchain, it was checked for syntactic errors using
the IfcCheckingTool and a Python script inspecting
the corresponding rules proposed by Ying and Lee
(2017). Both inspections reported no issues. Addi-
tionally, a visual inspection of the generated IDF in
OpenStudio was carried out, examining interior 2a SB
(a) Internal, 2a (b) Internal, 2b (c) External related to IfcSpaces
(d) External Earth (e) External (f) IfcSpaces
Figure 8: Generated space boundaries highlighted by type.
pairings, surface normals, EnergyPlus boundary con-
ditions, thermal zones and constructions. As shown
in figure 9, the geometry is well-defined and all zones
appear to be watertight. According to the Energy-
Plus error log, no errors attributable to the enriched
SBs were reported during the simulation.
Figure 9: Visualization of the generated IDF based on
the enriched IFC.
The positions and areas of the SBs were compared
with the SBs of the original office model. The quan-
titative evaluation showed agreement in the number
and position of all zones except the stairwell. This
is due to the different zoning, as there are no virtual
SBs in the proposed IFC file.
Conclusion
This paper presented an algorithm for generating
first and second level space boundaries based on IFC
data. The need for this tool was identified during
research work and cooperations with industrial part-
ners. Based on two application examples, the func-
tionality of the algorithm and the correctness of the
generated data were successfully demonstrated. A
significant simplification and acceleration could be
observed in the creation of simulation models. The
obtained results are a good basis for further improve-
ments of the tool. Future developments may include
the support of virtual space boundaries, the increase
of the overall robustness, and the subdivision of faces
in contact with the site. This tool contributes to an
improved BIM work process for building simulations.
This includes investigations in which a BEPS as well
as a CFD simulation must be performed. By support-
ing both simulation geometries, model setup can be
simplified and boundary conditions can be exchanged
more easily between the simulations.
Acknowledgments
The authors gratefully acknowledge the financial
support of the German Federal Ministry for Eco-
nomic Affairs and Energy of Germany in the project
“BIM2SIM - Development of methods for the genera-
tion of simulation models using Building Information
Modeling data” (project number 03ET1562A).
The authors also thank the experts of the work pack-
age Building Information Modeling of the IBPSA
Project 1 for the detailed and helpful discussions.
References
Andriamamonjy, A., D. Saelens, and R. Klein (2018).
An automated ifc-based workflow for building en-
ergy performance simulation with modelica. Au-
tomation in Construction 91.
BERTIM (2019). Building energy renovation through
timber prefabricated modules. www. bertim. eu .
buildingSMART (2018). https: // standards.
buildingsmart. org/ IFC/ RELEASE/ IFC4_ 1/ FINAL/
HTML/ copyright. htm .
buildingSMART International Standards Im-
plementation Database (2020). https:
// technical. buildingsmart. org/ resources/
software-implementations/ .
GeomEppy (2016). www. github. com/ jamiebull1/
geomeppy .
IfcCheckingTool (2020). www. iai. kit. edu/ 1649. php .
IfcOpenShell (2020). www. ifcopenshell. org .
International Organization for Standardization
(2017). Industry Foundation Classes (IFC) for
data sharing in the construction and facility
management industries — Part 1: Data schema
(ISO 16739-1:2018).
Jones, N. L., C. J. McCrone, B. J. Walter, K. B.
Pratt, and D. P. Greenberg (2013). Automated
translation and thermal zoning of digital building
models for energy analysis. In Proceedings of Build-
ing Simulation 2013: 13th Conference of IBPSA.
IBPSA.
Ladenhauf, D., K. Battisti, R. Berndt, E. Eggeling,
D. W. Fellner, M. Gratzl-Michlmair, and T. Ullrich
(2016). Computational geometry in the context of
building information modeling. Energy and Build-
ings 115, 78–84.
Lilis, G., G. Giannakis, and D. Rovas (2015). De-
tection and semi-automatic correction of geomet-
ric inaccuracies in ifc files. In Building Simulation
Conference 2015: 13th Conference of IBPSA.
Lilis, G., G. Giannakis, and D. Rovas (2016). Au-
tomatic generation of second-level space boundary
topology from ifc geometry inputs. Automation in
Construction 76.
Maile, T., J. O Donnell, and V. Bazjanac (2013). BIM
– geometry modelling guidelines for energy perfor-
mance simulation. In Proceedings of Building Sim-
ulation 2013: 13th Conference of IBPSA.
Open Cascade Technology (2020). www. opencascade.
com .
Mediavilla, A., Y. Sebesi, and P. Philips (2018). De-
liverable 4.6: Integration with external tools (en-
ergy plus). In TECNALIA (Ed), BERTIM - Build-
ing Energy Renovation through Timber Prefabri-
cated Modules.
Nytsch-Geusen, C., J. R¨adler, M. Thorade, and C. R.
Tugores (2019). BIM2Modelica - an open source
toolchain for generating and simulating thermal
multi-zone building models by using structured
data from BIM models. In Proceedings of the 13th
International Modelica Conference. Regensburg,
Germany, March 4–6, 2019.
O’Donnell, J., R. See, C. Rose, T. Maile, and V. Baz-
janac (2011). Simmodel: A domain data model
for whole building energy simulation. In Proceed-
ings of Building Simulation 2011: 12th Conference
of International Building Performance Simulation
Association, Sydney, 14-16 November.
OpenStudio (2020). www. openstudio. net .
OsmSerializer (2016). https: // github. com/
BIMDataHub/ BIMServerOsmSerializer .
Pinheiro, S., R. Wimmer, J. O’Donnell, S. Muhic,
V. Bazjanac, T. Maile, J. Frisch, and C. van Treeck
(2018). MVD based information exchange between
BIM and building energy performance simulation.
Automation in Construction 90, 91 – 103.
Rose, C. M. and V. Bazjanac (2013). An algorithm to
generate space boundaries for building energy sim-
ulation. Engineering with Computers 31 (2), 271–
280.
Simergy (2020). www. d-alchemy. com/ products/
simergy .
van Treeck, C. and E. Rank (2007). Dimensional re-
duction of 3d building models using graph theory
and its application in building energy simulation.
Automation in Construction 76.
Weller, H. G., G. Tabor, H. Jasak, and C. Fureby
(1998). A tensorial approach to computational
continuum mechanics using object-oriented tech-
niques. Computers in Physics 12 (6), 620.
Wimmer, R., S. Pinheiro, J. O’Donnell, S. Muhic,
V. Bazjanac, T. Maile, J. Frisch, and C. van Treeck
(2017). Realizing openBIM: Development of a BIM
model view definition for advanced building energy
performance simulation. Geb¨audetechnik in Wis-
senschaft Praxis 138 (4), 276–291.
Ying, H. and S. Lee (2017). A framework for rule-
based validation of IFC space boundaries for build-
ing energy analysis. In Computing in Civil Engi-
neering 2017. American Society of Civil Engineers.
Ying, H. and S. Lee (2020). Automatic detection of
geometric errors in space boundaries of IFC-BIM
models using monte carlo ray tracing approach.
Journal of Computing in Civil Engineering 34 (2).
... Different approaches for the generation of SBs are compared by [5], where also a new SB generation algorithm is proposed. This algorithm is applied to generate SBs for some of the use case IFC files described in Section 6.2. ...
... For this purpose, the IFC schema enables the attribution of all geometric and semantic information relevant for building simulation (cf. [5] for more detailed explanation of IFC SB syntax). This includes information about the type (2a, 2b), the relating space, and the related building element as well as the position (internal or external) and the nature of the boundary (physical or virtual). ...
... For this study, IFC4 SBs are evaluated from files created by Graphisoft ArchiCAD Version 20 (AC20) and Revit Versions 2019 (RV19) and 2022 (RV22) as authoring tools, created by the STREAMER Early Design Configurator [7], as well as SBs generated from the SB generation tool [5]. The proposed algorithms are tested on IFC4 input files. ...
Conference Paper
Full-text available
Energy performance simulation plays an important role for minimizing the energy consumption of buildings, thus reducing CO2 emissions and helping to enable the energy transition process. Within the design stage of buildings, the Industry Foundation Classes (IFC) standard can be used as a common data exchange format between all stakeholders within the design process. For applications in energy modeling, the IFC standard includes information on Space Boundaries (SBs). Those can be converted to their corresponding building surfaces as input for building energy performance software, such as EnergyPlus (EP). However, the quality of the SBs provided by the IFC varies depending on the modeler's decisions and differences in the implementation of the IFC authoring software. This often results in geometric and semantic inconsistencies and errors. Examples of those errors are missing SBs, misaligned openings, and incorrect assignments of SB properties (e.g. confounded internal and external classification). This paper addresses two types of challenges that arise when converting SBs from IFC to valid EP input: first, geometric simplification, and second, the detection and correction of geometric and semantic errors. At the beginning, a variety of common errors are listed and discussed. Then, multiple algorithms are presented to overcome these issues. Finally, the results are evaluated based on the best-practice EP input model of the analyzed buildings. The algorithms presented in this paper are capable of increasing the quality of generated EP input. This leads to high-quality EP input files even if the initial quality of the provided IFC SBs is poor.
... In addition to the manual conversion process in the viewer, work is underway on a fully automated algorithm that converts the IFC building model into a consistently triangulated surface mesh (Fichter et al., 2021). The resulting Standard Triangle Language (STL) geometry can be used directly to generate a surface mesh while preserving the IFC classifications of the building elements. ...
Preprint
Full-text available
The quality of BIM models is often insufficient for direct mapping into simulation models. Also, the IFC standard still has gaps, particularly in the area of representation of HVAC systems. Enrichment and advanced analysis functions nevertheless make it possible to generate simulation models semi-automatically and reduce effort and errors compared to the conventional process of modeling.
Conference Paper
Die Bauindustrie gehört zu den weltweit größten Verbrauchern von Material und Energie und trägt stark zur CO2-Emission bei. Die Holzbauweise bildet eine Möglichkeit nach-haltiger zu bauen. Dies betrifft nicht nur die Produktion sondern auch die Wiederverwertung beim Rückbau. Diese Bauweise stellt jedoch Architekten und Planer vor vielen planerischen Herausforderungen, da sie sich in vielen Punkten vom klassischen Betonbau unterscheidet. Aber auch hier spielen die korrekten Planungsentscheidung eine wichtige Rolle für den Erfolg des Projekts und die Zufriedenheit der späteren Nutzer. Eine der schwierigsten konstruktiven Entscheidungen im Holzbau betrifft den Schallschutz. Während bei der Statik und dem Brandschutz die Einhaltung von Vorschriften der Sicherheit dient, betreffen Entscheidungen zum Schallschutz neben der Zufriedenheit der späteren Nutzer auch deren Gesundheit. Die WHO hat eine Verbindung zwischen Lärm und verschiedenen Krankheiten gefunden, wie kardiovaskuläre Erkrankungen, Schlafstörungen, kognitive Beeinträchtigung bei Kindern, Tinnitus und viele mehr. Um die des Schallschutzes zu erleichtern mangelt es bisher an Planungswerkzeugen, die die Vielfältigkeit der Holzbauweise berücksichtigen. Der Trend hin zu mehr Digitalisierung bietet hier eine Lösung: indem der Holzbau für die Planung des Schallschutzes auf Bauwerkinformationsmodelle zugreift und mit diesen Daten anhand der Konstruktionsdetails relevante Kennzahlen aus bewährten Berechnungen oder Datenbanken ausliest.
Article
Full-text available
Steadily increasing use of Building Information Modeling (BIM) in all phases of building's lifecycle, together with more attention for openBIM and growing software support for the most recent version of the Industry Foundation Classes (IFC 4) have created a very promising context for an even broader application of Building Energy Performance Simulation (BEPS). At the same time, an urgent need for modeling guidelines and standardization becomes evident. A well-defined BIM-based workflow and a set of tools that fully exploit and extend the possibilities of the openBIM-technology can make the difference when it comes to reliability and cost of BEPS to design, build and operate high-performance buildings. This paper describes the essential elements of this integrated workflow, explains why openBIM comprises much more than just a standardized file-format and what is achieved with the already available technology, namely the Information Delivery Manual (IDM) and a newly developed Model View Definition. This MVD is tailored to the needs of Building Energy Performance Simulation (BEPS) that uses the Modelica language together with a specific library (IDEAS) and can easily be adapted to other libraries. In this project, several tools have been developed to closely integrate BEPS and IFC4. The simulation engine now gets the vast majority of the required input directly from the IFC4-file. For the implementation of the tools, the PYTHON language and the open source library IfcOpenShell are used. A case study is presented, that was used for extensive tests of the proposed approach and the implemented tools. The essential benefits of this new workflow are illustrated, and the feasibility is demonstrated. Opportunities and remaining bottlenecks are identified to encourage further development of BIM software to fully support IFC4 as an information source for BEPS. Besides some improvements of the proprietary class structure and functionality, enabling the export of IFC4 files based on custom MVDs is one required key feature.
Article
Full-text available
Abstract The process of preparing building energy performance simulation (BEPS) models involves repetitive manual operations that often lead to data losses and errors. As a result, BEPS model inputs can vary widely from this time consuming, non-standardised and subjective process. This paper proposes a standardised method of information exchange between Building Information Modelling (BIM) and BEPS tools using the Information Delivery Manual (IDM) and Model View Definition (MVD) methodologies. The methodology leverages a collection of use cases to initiate the identification of exchange requirements needed by BEPS tools. The IDM/MVD framework captures and translates exchange requirements into the Industry Foundation Classes (IFC) schema. The suggested approach aims to facilitate the transfer of information from IFC based BIM to either conventional or advanced BEPS tools (e.g. EnergyPlus and Modelica) through the development of a specific MVD that defines a subset of the IFC data model that deals with building energy performance simulation. By doing so, the potential of BIM-based simulation can be fully unlocked, and a reliable and consistent IFC subset is provided as an input for energy simulation software.
Article
Full-text available
The Industry Foundation Classes (IFC) is a semantically rich data model providing necessary information to support extraction of information necessary for the setup of building energy simulations. Often, 2nd-level space boundary data contained in IFC, are missing or incorrect. To facilitate the connection between BIMs and energy simulation programs, the Common Boundary Intersection Projection (CBIP) algorithm is introduced. CBIP uses the geometric representations of building entities obtained from IFC files to generate the building's 2nd-level space boundary topology. A prototypical implementation of the CBIP algorithm is used in a complex geometry building, as a verification of the capability of the algorithm to properly identify space boundaries.
Conference Paper
Full-text available
Accurate representation of geometry-related parameters is a prerequisite for the generation of building thermal simulation models. In case geometric data are extracted from IFC files, three types of geometrical errors are often encountered: clash errors where two building solids intersect, space definition errors where a building space volume is incorrectly defined leaving volume gaps between the space and neighboring building entities (walls, slabs,...) and surface orientation errors, where the normal vector of some boundary surfaces of building solids points to the wrong direction , generating a misconception on which area is inside or outside the solid. These errors are introduced either in the design process or during export from the BIM authoring tool. In this paper, error detection and correction algorithms are introduced to address each individual error type. The algorithms are tested on the geometrical data of the IFC files of two office buildings , where errors of the types mentioned above were detected and automatically corrected.
Conference Paper
Full-text available
Building energy simulation is valuable during the early stages of design, when decisions can have the greatest impact on energy performance. However, preparing digital design models for building energy simulation typically requires tedious manual alteration. This paper describes a series of five automated steps to translate geometric data from an unzoned CAD model into a multi-zone building energy model. First. CAD input is interpreted as geometric surfaces with materials. Second, surface pairs defining walls of various thicknesses are identified. Third, normal directions of unpaired surfaces are determined. Fourth, space boundaries are defined. Fifth, optionally, settings from previous simulations are applied, and spaces are aggregated into a smaller number of thermal zones. Building energy models created quickly using this method can offer guidance throughout the design process.
Article
In Industry Foundation Classes (IFC) building information modeling (BIM), the objectified concept of a space boundary (SB) provides a means to define building space geometries with surface entities. Such building-geometry definitions are widely used for various engineering applications such as energy simulation, lighting analysis, and facility management. However, quality issues (i.e., geometric and nongeometric issues) of SBs have been widely reported, which makes it necessary to validate the SBs before retrieving them from IFC models for relevant applications. Unfortunately, there is still a lack of reliable mechanisms/tools to automatically evaluate the quality of SBs, especially the geometric quality. This study proposes a Monte Carlo ray tracing approach to automatically detect geometric errors in SBs. The approach checks SBs space by space in terms of whether each space is correctly bounded by its SBs. The geometric errors in the set of SBs of a space that the approach can detect include gaps, overhangs, and overlaps between SBs as well as incorrect surface normal directions of SBs. To accelerate the ray tracing process in the approach, the axis-aligned bounding box (AABB) tree is implemented to spatially index SBs of each space. The approach is evaluated with extensive performance tests in terms of robustness and efficiency. The results show that the approach can robustly and efficiently detect all four types of geometric errors even in extreme cases and that the AABB tree helps speed up the approach significantly for large-scale IFC models with many complex spaces.
Conference Paper
In an industry foundation classes (IFC) building information model (BIM), proper space boundaries (SBs), especially 2nd-level SBs, are needed to properly construct building surface geometry for building energy analysis. This study aims to develop a framework to achieve a rule-based automatic validation of IFC 2nd-level SBs. Two types of errors, namely semantic and syntactic errors and geometric errors, which 2nd-level SBs in IFC instance files may often encounter, are first analyzed. Corresponding to the two types of errors, the framework consists of two modules: 1) the semantic and syntactic validation module, which checks whether the IFC model complies with semantic constraints on 2nd-level SB definition and the IFC schema; 2) the geometric validation module, which checks whether the 2nd-level SBs are geometrically correct. For the syntactic and semantic validation, validation rules are identified and supporting validation tools are investigated. Rules for the geometric validation are also identified and the rule interpretation and execution method are discussed.
Article
Building energy analysis has gained attention in recent years, as awareness for energy efficiency is rising in order to reduce greenhouse gas emissions. At the same time, the building information modeling paradigm is aiming to develop comprehensive digital representations of building characteristics based on semantic 3D models. Most of the data required for energy performance calculation can be found in such models; however, extracting the relevant data is not a trivial problem. This article presents an algorithm to prepare input data for energy analysis based on building information models. The crucial aspect is geometric simplification according to semantic constraints: the building element geometries are reduced to a set of surfaces representing the thermal shell as well as the internal boundaries. These boundary parts are then associated with material layers and thermally relevant data. The presented approach, previously discussed at the International Academic Conference on Places and Technologies [1], significantly reduces the needed time for energy analysis.