-
maintenance of energy-efficient buildings made of prefabricated components) [
2
]
aims to reduce the discrepancy in quality and energy performance between design and
real behaviour by applying self-inspection techniques.
For that end, INSITER relies on Building Information Modelling (BIM)-based
construction projects, which supports the decision-making process and allows easier
management along the building lifecycle [
2
]. BIM is complemented by intuitive and
cost-effective AR (Augmented Reality), which connects the virtual building model
and the physical building (or the building under construction) in real-time [
3
]. During
construction or commissioning phases, both techniques are very helpful to prevent
errors in-advance, which usually imply high repair costs, or when left unattended to,
they can have large, negative consequences on the building’s technical, and
subsequently, financial performance [
3
].
Nevertheless, the concept of BIM is applied and understood in many ways (e.g., a
simple information repository, a complete data model, etc.) [
4
]. When dealing with
self-inspection processes, BIM application needs to be adjusted. In practice, common
BIM models are too complicated; there is a plethora of information that is not
structured in a step-by-step process for on-site construction workers. It is difficult to
inspect details and technical quality by comparing the BIM model and the realized
building. Moreover, data for comparison with the BIM models are acquired via
threedimensional (3D) measurement devices adjusted for self-inspection [
5
][
6
], which
present issues with robustness, and user-friendliness. Other challenges also arise in
the form of differences in geolocations, building typologies and materials, working
cultures, methodologies, software tools and data registration [
5
]. Complementary,
industry can also benefit from this BIM concept in order to increase the reusability of
data between all the stakeholders involved in the construction process.
BIM model needs also to be enriched with dynamic information so as to apply
selfinspection methodologies, which require dynamic data to calculate KPIs [
7
].
Nevertheless, the main challenge is the implementation of the appropriate interoperability
standards for data exchange [
8
], such as it is also being proven in PERFOMER
project [
9
]. Their approach is similar to the one presented in this paper in terms of
dynamic data. That is to say, integrating data in-situ from sensors through a TimeSeries
server and provide the data through BIM viewers where static sensor and dynamic
data are linked. Also DIMMER project integrates BIM with real-time data and
extends to the district level, leading to District Information Models (DIM) [
10
].
For the aforementioned purposes, this paper presents an interoperable framework
to facilitate the integration of heterogeneous data sources under the Industry
Foundation Classes (IFC) standard [
11
], an open industry standard format. Then, under the
INSITER approach, the various necessary data and information can be merged to
provide useful data; the combination of all these resources improves the construction
process by reducing the costs and time, while concurrently and pre-emptively
detecting potential errors. In this way, by means of the IFC ontology, the various linked
building data may be integrated to provide a homogeneous understanding.
2
Interoperable framework
As stated before, self-inspection tools support the evaluation of building construction
and energy performance. For that end, data are required to assess the building
facilities, although multiple and heterogeneous data formats are available depending on the
device vendor [
8
]. Hence, it is required the harmonization of the information in a
common language that increases the interoperability level in the system-to-system
communication (e.g. BIM and database in the INSITER context). In order to face this
challenge, it is proposed the framework depicted in Fig. 1 [
3
][
12
] where four layers
are intended to ensure interoperability and harmonize the information flowing through
the multiple entities and enabling self-inspection procedures.
From bottom, the data acquisition layer represents the devices for obtaining data
related to the building [
7
]. In this case, it is associated with hardware and documents
for applying the self-inspection methodology. Basically, sensor network to capture
dynamic information of the building (i.e. energy and comfort parameters), existing
databases of the building, hardware (laser scanner, thermography and sound probes
for quality checks [
5
]) and other sources (e.g. data sheets) are included [
3
].
Adaptation layer is the next one and is dedicated to harmonize data coming from
the data acquisition layer [
3
]. The idea behind this layer is the implementation of a
middleware able to translate the specific data formats from the devices into the
common IFC representation [
8
] used in INSITER. It is important to remark data-sheets
and documents are not translated because they do not provide structured information,
but they are included within the BIM platform, as it will be explained later. Basically,
the main translations are based on how to adapt the communication protocols from
building automation to the IFC database that is described in section 2.2. Additionally,
middleware also treats acquisition hardware data though combination of
measurements, importation into Revit and extraction to IFC, as explained in section 3.
The main layer in the INSITER architecture framework is the BIM layer that
contains the data repositories. It is composed by three repositories: the Open Source BIM
Server for static building data, a Postgres database for dynamic data and the
sharepoint to store the documents [
12
]. This three-entities repository is implemented under
the concept of extended BIM (xBIM) or dynamic BIM (dBIM) [
8
], which combines
static and dynamic data under the same format (i.e. IFC). That is to say, it is a “single
data-resource” containing both the static information linked to the building and the
dynamic sensor data. Additionally to the BIM Server and its extended functionality,
BIM services are also contemplated, such as clash detection and model checks.
At top level, the applications are deployed. This Application Layer integrates
information from the services that check building quality, as planning and cost tools,
local monitoring visualizers, or BIM viewers. It also allows the construction workers
to perform quality analysis of the building by extracting information from the BIM
model, as well as performance monitoring on dashboards and interaction with experts
via remote communication, effectively cross-checking solutions in a timely manner.
This layer obtains data from BIM Server in IFC format to be exposed and exchanged.
2.1
Static data model
As stated before, the main topic is how to integrate static and dynamic data. Starting
with the first data model, within the world of Information Technologies (IT), there is
an ever-growing number of information exchange standards, each with its own
purpose and scope. The INSITER project selected standards take into account the
possibility that existing standard(s) may not fulfil the needs and requirements, and could
possibly require the combination of several standards. Mature and often-used
exchange formats, such as DraWinG (DWG), AutoCAD and Revit native file formats,
are closed (or proprietary) formats which introduce a major dependency on software
vendors. Standard alternatives are also available, like Drawing Exchange Format
(DXF) and Green Building Extensible Markup Language (gbXML) [
13
], however,
both standards have a much more limited scope. For instance, with gbXML, the scope
is only related to energy performance. For many requirements of the INSITER
framework, this is not sufficient. Consequently, the standard IFC [
9
] was the most
relevant candidate for INSITER selection [
3
].
Within the building and construction industry, the choice regarding exchange
standards is straightforward. There is only one open standard (IFC, also called
ISO16739) that is widely accepted, mature and covers most of the disciplines within
the sector [
9
]. From the first real practical version, i.e. IFC2x3 TC1 (first technical
corrigendum), major improvements have come with more recent releases. In this way,
IFC4 [
14
] has been available for several years and has many new features of
importance for selection by INSITER, including key defining factors for being selected.
These factors include, but are not limited to:
- Improved representation of curves and surfaces through non-uniform rational
B-spline (NURBS), B-Spline and Bezier algorithm definitions (Rational).
- Improved scheduling capabilities, such as storing content representing
scheduling information very close to the capabilities or freedom available with
commercial packages like Microsoft Project.
- Usable texture embedding capabilities.
- Newer, more compressed ways of storing geometrical data through vertex
index arrays and sets of triangles, with or without normal information.
2.2
Database model: Building automation ontology
Still, the question remains, does the static data model cover all requirements?
Although IFC is very complete, the INSITER framework is still required to offer support
for multiple file formats, as well as dedicated semantic structures. This is the case of
dynamic data, whose aim is to integrate periodic data measured from devices. For that
end, an IFC-compliant database has been used [
12
]. Basically, the database schema
follows the relationships between IFC classes, mapping the IFC classes and their
relationships to create a repository, where the object TimeSeries is re-defined to represent
all the data samples coming from the sensor network. Then, the relationship between
the sensor and the timeseries objects generates the way to store dynamic data.
Moreover, the concept of sensor and timeseries is replicated to KPI entities, being thus able
to store not only raw data, but also aggregated data.
The mechanism to link static and dynamic data relies on the GlobalIDs that each
object has in the BIM. That is to say, they are used as primary keys in the database to
represent sensors, creating a single access to the object. There is a direct correlation
between the objects in the BIM and the elements inserted in the database.
Having this approach in mind, building automation systems can be directly linked
to the BIM of the building, extending the concept of BIM. Hence, the usable ontology
is IFC to represent the objects in a dynamic database. Thanks to this re-definition of
the IFC classes, building data is linked between the static and the dynamic “worlds”.
3
Data integration and interoperability mechanisms
One of the challenges to overcome for buildings self-inspection is to interconnect the
different measurement instruments with BIM [
15
]. This is an interoperability
challenge due to the multiple formats that need to be combined in the common IFC
standard. A specific methodology was developed to merge useful digital information for
self-inspection purposes within AutoDesk REVIT as BIM worldwide representative
software. Suitable digital information sources are 3D point clouds (laser scanning or
photo-based scanning), and also 2D images (thermal cameras and sound brushes) [
3
].
Hence, a specifically tailored plug-in for REVIT has been programmed and
supported by the Point Cloud Library (PCL) as standalone, large scale, open software
project for image and point cloud processing. The plug-in is a dynamic-link library
(DLL) named INSITER-DLL and is able to precisely display the PLY files by layers
in a unique working project. The method to integrate this information is drawn in Fig.
3 [
14
], where the user loads the base file into REVIT, and the INSITER-DLL
automatically loads all related files meeting the requirements previously described. The
PLY files serve as 3D templates to extract useful parametric features for
selfinspection through fully manual/semi-automatic delineation, or just using
complementary AEC (architecture, engineering and construction) plug-ins.
CARTIF-III building is an office building located in Boecillo, Spain made of
prefabricated models and according to nZEB (near Zero Energy Buildings) premise. It is
4.075 m2 of floor space for CARTIF Technology Center activities. The main
construction errors detected in this building are related to thermal bridges, lack of
insulation in some of the zones (mainly one industrial area) and overheating in summer.
4.1
BIM enrichment and dynamic data integration
The integration was two-fold: firstly, field-tested for inspection tasks directly linked
to the long term maintenance planning and, secondly, related to the dynamic data
integration. With respect to field tests, 3D laser scanning of representative elements of
this building were carried out both by a Leica HDS-3000 (TOF (time-of-flight)
technology), and a FARO Focus3D X330, trying to evade voids. As illustrated in Fig. 2,
data consist of geometry (XYZ coordinates); colour (RGB); reflectance (L index:
materials and humidity) and reflectivity through the algorithms described in [
16
].
Thermography (to assess heat gain/loss and thermal bridges); and acoustic imaging
were also obtained in parallel and inserted in MeshLab (see Fig. 2). The resulting
enriched 2D/3D blending is illustrated in Fig. 3 [
16
] where the Revit BIM model of
CARTIF-III is shown and the overlapping of the graphical information, being the left
part the reflectance values and the right side the thermal information.
With regard to the dynamic data from sensors, the aforementioned framework was
deployed in the building to collect data from the sensor networks in the building.
Moreover, the BIM serves as data source for mapping the IFC objects into the
database elements. After several months of data collection, information collected from
sensor with IFC-ID = “0VE1OxL6b0nAnaUiygQBij”, which corresponds to
temperature sensor, is shown in Fig. 4, as well as the associated KPIs. The approach is, as
explained, connect sensor network, map sensor into the IFC and store its data.
During this paper an interoperable framework has been described whose aim is to be
able to integrate heterogeneous data sources in a common data model. Interoperability
is a key aspect nowadays and it is a technology challenge. Furthermore, building
context is especially critical where data are necessary to assess performance and multiple
vendors are involved in the process. For this aim, IFC, as widely standard used in the
AEC industry, has been selected as the INSITER ontology for linked-data. Thanks to
the presented framework, the interoperability capabilities are increased by means of
integrating data into BIM and, then, apply self-inspection techniques to avoid
construction errors during the lifecycle of the building. The major benefit is being able to
predict, prevent and solve errors in advance, and, then, increase the energy efficiency.
Related to future work, dynamic time series are expected to be continuously
gathered in order to provide a full data repository. As well, additional buildings are
planned to be included so as to validate the presented approach.