Structure and Infrastructure Engineering

Maintenance, Management, Life-Cycle Design and Performance

ISSN: 1573-2479 (Print) 1744-8980 (Online) Journal homepage: https://www.tandfonline.com/loi/nsie20

Improved visualization of infrastructure

monitoring data using building information
modeling

Chanakya Boddupalli, Ayan Sadhu, Ehsan Rezazadeh Azar & Scott Pattyson

To cite this article: Chanakya Boddupalli, Ayan Sadhu, Ehsan Rezazadeh Azar & Scott Pattyson
(2019) Improved visualization of infrastructure monitoring data using building information modeling,
Structure and Infrastructure Engineering, 15:9, 1247-1263, DOI: 10.1080/15732479.2019.1602150

To link to this article: https://doi.org/10.1080/15732479.2019.1602150

Published online: 22 Apr 2019.

Submit your article to this journal

Article views: 124

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=nsie20
STRUCTURE AND INFRASTRUCTURE ENGINEERING
2019, VOL. 15, NO. 9, 1247–1263
https://doi.org/10.1080/15732479.2019.1602150

Improved visualization of infrastructure monitoring data using building

information modeling
Chanakya Boddupallia, Ayan Sadhub, Ehsan Rezazadeh Azara and Scott Pattysonc
a
Department of Civil Engineering, Lakehead University, Thunder Bay, Canada; bDepartment of Civil and Environmental Engineering, Western
University, London, Canada; cWSP Canada Inc., Thunder Bay, Canada

ABSTRACT ARTICLE HISTORY

With growing complex infrastructure, autonomous condition assessment of large-scale structures has Received 28 October 2018
garnered significant attention over the past few decades. Data-driven structural health monitoring Revised 11 March 2019
(SHM) techniques offer valuable information of existing health of the structures, maintain the safety Accepted 12 March 2019
and their uninterrupted use under varied operational conditions by undertaking timely risk and hazard
KEYWORDS
mitigation. Traditional approaches, however, are not enough to monitor a large amount of SHM data Structural health monitoring
and conduct systematic decision making for future maintenance. In this article, building information (SHM); building information
modeling (BIM) is utilised as a promising computing environment and integrated digital representation modeling; system
platform of SHM that can organize and visualise a considerable amount of sensor data and subse- identification; long-term
quent structural health information over a prolonged period. A BIM-enabled platform is utilised to monitoring; visualisation
develop the proposed visualisation tool for a long-span bridge and enable automated sensor data tool; Revit; big data
inventory into the BIM environment. Such automated tool facilitates systematic maintenance and risk
management, while avoiding manual errors resulting from visual inspection of the structures. The pro-
posed method can be considered as a user-friendly and economic framework for condition assessment
and disaster mitigation of structures from long-term monitored data.

1. Introduction preprocessed data, during the assessment of the structure,

can add up hundreds of gigabytes of information which can
Structural health monitoring (SHM) (Carden & Fanning,
lead to misalignment of data. A substantial amount of big
2004) is considered as a valuable approach to detect critical
data and images resulting from long-term SHM set further
responses of a structure, and subsequently track and evalu-
challenges for data interpretation. The data processing
ate any unusual symptoms, serviceability, and safety of the
framework, including storage, transmission and manipula-
structure. A structure deteriorates with time due to natural
tion associated with big data becomes a significant challenge
disasters, excessive operational loads, cyclic freezing and
thawing, corrosion and many other factors. Early identifica- (Cai & Mahadevan, 2016). In this article, an attempt is
tion of damages facilitates prevention of catastrophic failures made to develop a visualisation tool for SHM that would
(Lynch, Farrar, & Michaels, 2016). Maintenance engineers enable easier and systematic tracking of overall evolution of
typically assess these structures manually, but a network of structural conditions over a long period of time.
sensors can automatically evaluate the integrity of the struc- There are numerous examples of long-term SHM of
ture and locate the damage which could significantly reduce large-scale structures such as bridges. For example, Zhou,
the operational costs while improving the public safety Li, Xia, Yang, and Zhang (2017) proposed advanced bridge
(Somwanshi & Gawalwad, 2016). Among many methods, safety evaluation and technologies for a long-span bridge
output-only vibration testing (Sadhu, Narasimhan, & located in China. Catbas, Susoy, and Frangopol (2008)
Antoni, 2017, Friesen & Sadhu, 2017) uses response data instrumented and monitored a long-span bridge under traf-
obtained from the accelerometers placed optimally (Sadhu & fic, wind, and temperature, and determined the structural
Goli, 2017) throughout the structure to determine existing reliability under environmental effects. Omenzetter and
conditions of the structure. Brownjohn (2006) used several existing time series analysis
Apart from critical inspection, continuous monitoring of (e.g., univariate/multivariate models) to understand the
the structure is of paramount importance to detect the strain data recorded from a long-span bridge. Orcesi and
changes of structural parameters and early detection of Frangopol (2011) explained the importance of monitoring
unsafe condition using real-time data (Abe, 1998). With the fatigue life of different components of a bridge. Hu, Wang,
advent of modern sensing technology (Sony, Laventure, & and Ji (2013) discussed the long-term bridge monitoring
Sadhu, 2019), it is now possible to collect a significant using the fusion of acceleration, temperature and strain
amount of rich quality data in ease. However, all raw and data. Li et al. (2006) employed SHM system for a highway

CONTACT Ayan Sadhu [email protected]

ß 2019 Informa UK Limited, trading as Taylor & Francis Group
1248 C. BODDUPALLI ET AL.

bridge with a large number of sensors to provide informa-

Architect
tion for planning and maintenance activities and alert the
future construction policies for cable-supported bridges. Ko
Building Structural
and Ni (2005) discussed the current practices towards bridge engineer
owner
health monitoring and addressed the existing challenges of
the SHM. Above mentioned literature highlights the need
for an integrated tool for robust and accurate data fusion
Construction BIM Contractor and
and visualisation tool for SHM. manager model sub-contractors
In the past few years, there have been significant devel-
opments of efficient and cost-effective visualisation of SHM
systems using augmented reality (AR) or virtual reality HVAC Facility
(VR). For example, Malkawi and Srinivasan (2005) demon- engineer operator
strated a framework towards real-time computational fluid
dynamics immersive data visualisation. The human building Figure 1. Conceptual framework of BIM.
interaction model created in this study can visualise and
interact with buildings and their thermal environments. extraction for multiple users. The users were able to store
Highly interactive and immersive AR environment allows relevant data on their servers while the overall data was vir-
for efficient processing and data transfer. Napolitano, Blyth, tually integrated through cloud computing platform. Anil,
and Glisic (2017, 2018) transformed the current practices of Akinci, Kurc, and Garrett (2015) analysed a structure that
SHM data visualisation by creating a digital workflow that was heavily impacted by the earthquake and used BIM to
can visualise and access SHM data and metadata in 3D. provide accurate system information of entire structure. The
Such an interactive method utilised spherical imaging and transformation of the structure caused by the seismic activ-
information modeling to integrate SHM data into a VR ity was developed as 3D images using BIM. Fanning (2014)
environment. In Ballor et al. (2019), the authors investigated documented the current utilisation of BIM practice by pro-
the potential of AR to minimise the drawbacks of visual viding a case study of two bridge construction projects, one
infrastructure inspections. It enhanced the ability of human with BIM and another with traditional manner. It was
inspectors to capture high-resolution 3D measurements of shown how BIM improves the efficiency and effectiveness of
infrastructure. Glisic, Yarnold, Moon, and Aktan (2014) design and construction of bridges. The integration viability
proposed a novel solution for efficient SHM data manage- of bridge information modeling with different modules of
ment. Principles for live and historic visualisation of SHM bridge management system was explored by Marzouk and
data and how different attributes of data users influence Hisham (2012). The authors developed a C# program that
visualisation were examined. creates inspection spreadsheets and bridge component data-
In this article another powerful object-oriented modeling base directly from bridge model by combining multiple soft-
tool, Building Information Modeling (BIM), is explored to ware packages such as Ansys, Navisworks and Tekla.
develop a visualisation tool for SHM data. BIM is a digital Simulating the sensor data of bridges using BIM models
representation of physical and functional characteristics of a will visualise all the sensor information which can make the
structure (NBIMS 2010). It is a platform that shares the proper advantage of data interpretation and facilitates timely
information of a facility from its design till its demolition. damage detection in the model. BIM uses a static data
BIM is not only a computer-aided design tool, but also an source to assess the structure. Therefore, extending the
object-oriented 3D modeling and information management application of BIM model from static to dynamic model by
platform which enables engineers and architects to work linking the sensor data will provide effective information to
seamlessly on the same project remotely. BIM integrates assess and interpret the ongoing performance of the struc-
various aspects of engineering design through three-dimen- ture. Acquiring more extensive and detailed data in a real-
sional spatial representation. Each BIM model can carry time inspection can result in accurate decision-making.
attributes by providing their key features and perform cost McGuire (2014) developed a BIM technique to investi-
estimation and material ordering. Figure 1 provides a basic gate the effects of damage through electronic media that can
layout of information sharing through BIM for a complex prioritise better decision-making and public safety.
structure. It benefits owners and contractors more as the Shakhramanyan, Kuckartz, and Kolotovichev (2012) dis-
model can foresee potential problems before the construc- cussed the three stages of a real-time structural monitoring
tion phase. Traditional BIM models primarily focus on system that followed the design, construction and mainten-
design and life-cycle analysis of a new building. (Bhusar & ance stages using sensors and performed analysis using
Akhare, 2014) conducted environmental impact-based life ‘SODIS building M’ software. Delgado, Brilakis, and
cycle assessment and developed a model which connects the Middleton (2016a) used BIM for visual sensor information
BIM and structural analysis software. for an effective interpretation of data obtained from a con-
Even though BIM data is stored and transformed through crete bridge. The values of the strain attributes of each sen-
Autodesk RevitV files, Zhang, Liu, Yu, and Zhao (2014) pro-
R
sor entity were linked with the material attributes of the
posed a framework where the BIM integrated service plat- BIM model. Banfia, Barazzettia, Previtali, and Roncoronib
form was used for data integration and model information (2017) included 3D digital surveys, parametric modeling
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1249

and monitored data sets for documentation, and visualisa- Most of the existing BIM-based SHM work-flow are pri-
tion of SHM information. A 3D web share cloud technology marily focussed on metadata of the sensors (i.e., sensor spe-
was used to develop a 3D model of a bridge. A semantic cification and raw data) as opposed to condition assessment
rich BIM model can facilitate more streamlined approaches data (i.e., system identification results) that can be used for
to SHM and data management, and can identify changes in progressive tracking of the structural performance. Also, the
structural behavior (Delgado et al. 2016b). existing BIM implementation of SHM is mostly limited to
Despite the increasing trend in intelligent SHM, it is still small-scale structures. There are very limited researches on
not possible to digitally represent the sensor information detailed documentation to track progressive maintenance
from long-term monitored data. Kay and Eike (2015) dis- strategies through system identification and their subsequent
cussed the opportunities and challenges towards the digital effectiveness in improving the overall health of structures.
representation of SHM systems using Industry Foundation Based on the above gap areas, in this article, it is proposed
Class (IFC) which was used to bridge the gap between sen- to integrate system identification component of SHM with
sor models and BIM. Maximilian and Kosmas (2016) BIM for improved visualisation of diagnosis and prognosis
explained the different methodologies to perform SHM information, thereby pursuing better damage assessment of
within BIM. By integrating and monitoring the information critical structures. By comparing the current SHM data with
into BIM helps in categorising, documenting and updating the predicted response from the Finite Element (FE) model,
monitoring related to information throughout the life cycle the proposed visualisation tool will expand decision-making
of the structure. Traditional approaches require a longer capabilities of SHM within the BIM to identify the struc-
time to exchange the data with limited access as well as tural performance under various weather and operational
errors associated with manual data. Park and Kim (2015) conditions. It offers an integrated tool that shows the com-
focused on resolving building safety issues through BIM. plete sensor information as well as systematic diagnostic
For long-span structures, visual inspection techniques results of the entire life-cycle of the structure that would be
beneficial for the infrastructure owners. Finally, the pro-
facilitate real-time maintenance to improve the overall safety
posed visualisation tool is illustrated using a case study asso-
of an aging structure. Chan et al. (2016) proposed an
ciated with a full-scale bridge subjected to various traffic
approach for bridge asset management systems using an
loads under different seasons.
unmanned aerial vehicle. Using the images captured by the
UAV and BIM models, the bridge inspector identifies dam-
age location and rate of progressive deterioration of the 2. Proposed framework
structure. The study by Theiler, Dragos, and Smarsly (2017)
This section provides an overview of the proposed research
enabled the description of monitoring-related information
and the methodologies implemented to visualise the SHM
using BIM. Since the existing IFC schema does not support
information within the BIM through Autodesk’s RevitV. The
R

the full description of modelling related information, the

applications of RevitV are explained in the context of tool
R

authors extended IFC schema as ‘IFC Monitor’ facilitating

development. The approach uses RevitV and MATLABV as
R R

the documentation of SHM systems and validated the work-

base platforms to integrate the sensor information with the
flow using a simplistic laboratory model. A 3D BIM model
diagnostics results. RevitV has a capability of designing 3D-
R

was expanded by adding time data (4D), cost data (5D),

models and can perform facility maintenance through time-
facility management (6D) and sustainability (7D) for better
to-time monitoring of the structure with the updated model
visualisation and improved effectiveness (Del Grosso, Basso, of the selected infrastructure. The dynamic behaviour of the
Ruffini, Figini, & Cademartori, 2017). The authors demon- structure is then analysed using the sensor data
strated the linking of data to sensor representations within in MATLABV.
R

the BIM model and concluded that the process of linking The proposed framework harnesses the relative merits of
BIM model and SHM data is highly platform dependent SHM and BIM to develop the visualisation tool for monitor-
and required customised interfaces. ing of large-scale infrastructure. The data and system identi-
Valinejadshoubi, Bagchi, and Moselhy (2017) investigated fication information of SHM are systematically embedded
the use of BIM in data management systems and showed with the BIM software such that long-term health monitor-
how to link the virtual sensors of the BIM model with exter- ing information can be visualised and used for maintenance
nal data captured by the real sensors. Minehane, and decision making purposes. As shown in Figure 2, the
O’Donovan, Ruane, and O’Keeffe (2012) used laser scanning proposed research is consisted of four key steps: (a) drawing
that offers safety and remotely accesses comprehensive sheet, (b) BIM model, (c) sensor data, and (d) data process-
information without any direct interface with the structure. ing. First, RevitV is used to develop 3D object-oriented
R

It also allowed rapid decision-making and performance building information models. For the approach presented in
evaluation to satisfy building’s safety requirements. For this paper, a structural model is developed that closely rep-
effective collaborative design, construction, and mainten- resents a real structure. The data represented in attributes
ance, a 3D model with metadata was developed to share define the physical, geometrical and abstract properties of
information of the design (Shim, Yun, & Song, 2011). This the structure. As shown in Figure 3, with the help of 2D
interoperability enhances the design revision process and drawings, the structure is developed into a 3D model with
all the generic parameters and properties using RevitV. This
R
communication for better productivity.
1250 C. BODDUPALLI ET AL.

Sensor
Drawing Revit System information
sheet model identification and data
processing
BIM SHM

BIM based visualization tool of

SHM

Figure 2. Layout of the proposed framework.

Figure 3. Graphical representation of the test structure in RevitV with its key features.
R

model provides the real-time appearance of the structure different operating conditions. The collected data is proc-
essed using MATLABV and modal responses are extracted
R
with all the material properties as well as model updating
during its service period. using a popular system identification method, namely
The BIM platform reduces the working space for struc- Second-order Blind Identification (SOBI) method
tural analysis by linking its model to other structural ana- (Belouchrani, Abed-Meraim, Cardoso, & Moulines, 1997). A
lysis software. The integration between RevitV and Robot
R
brief description of SOBI can be found in the appendix. The
structural analysisV (RSA) allows bi-directional data
R
modal identification is facilitated with the FE model of the
exchange between both the platforms. RSA as shown in structure, and the FE model is compared with the experi-
Figure 4 is one of those software that is linked to the cur- mentally obtained modal parameters. All the results are pro-
rent BIM platform. With the integration tool, the results of jected into the BIM model. These results help in monitoring
static analysis and the calculated deflections are transferred the performance of the structure during the different period
to the RevitV model. Modeling a structure in RevitV helps to
R R
of data collection. The comparison between the recorded
visualise the relationship between real structure. The modal frequencies and the FE model gives a better under-
obtained results from RSAV are saved as readable files for
R
standing of the evolution of structural damage.
the parental platform. Following is a summary of the proposed visualisation
Apart from performing the static analysis, the results tool which is briefly illustrated in Figure 5.
from dynamic analysis help in monitoring the structure dur-
ing its service period. Sensors connected to the bridge are 1. Convert the existing 2D drawing of a structure into a
utilized to collect vibration response of the structure under 3D BIM model.
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1251

Figure 4. Robot structural analysisV and the RevitV model.

R R

Figure 5. Flowchart of the proposed integrated tool.

Export the designed BIM model into RSAV to perform

R
2. information of the SHM. All the results are shown in a
static analysis. user-friendly format and integrated with the visualisation
platform of RevitV. The results of the structural analysis
R
3. Collect the vibration data of the real structure using
obtained from RSAV are compared with the system identifi-
R
appropriate instrumentation.
Assign the required sensor information in RevitV to
R
4. cation results of the SHM data.
visualise the metadata and process information.
5. Analyse the collected data using SOBI method
through MATLABV.
R 3.1. Details of the bridge and instrumentation
6. Compare the static and dynamic analysis of the struc- The selected large-span bridge is located in Thunder Bay,
ture using the FE model. Ontario as shown in Figure 6. The reinforced concrete
bridge is approximately 1000 ft long and 24 ft wide. The
bridge was instrumented with vibration sensors in winter
3. Case study
2017 and spring 2018 respectively, and the collected long-
In this article, the proposed method is demonstrates using a term data and the evolution of structural health is demon-
long-span bridge located in Thunder Bay, Ontario. The strated here to illustrate the proposed tool. The vibration
model developed in RevitV is integrated with the sensor
R
testing was performed on the bridge on 20 January 2017
1252 C. BODDUPALLI ET AL.

Figure 6. Long-span bridge.

Figure 7. Location of the sensors and layout of the bridge.

between 9 am and 12 pm and on 23 April 2018 between 9 frequency was set to 200 Hz. As shown in Figure 8, sensors
am and 12 pm. The temperature during the test was around were installed on the walkway, located close to one side of

5  C and 15  C in 2017 and 2018, respectively. Ten sensors the bridge. As shown in Table 1, 12 tests were performed
were placed along the walkway on the North side of the with a different class of vehicles such as cars (3000 lbs),
bridge, and the sensors were set up to measure uniaxial trucks (20,000 lbs) and large trucks (40,000 lbs) at various
vibration in the vertical direction. Each sensor has a sensi- speeds. The vehicles and the tests are documented in
tivity of 10 V/g. The spacing of the sensors and the location chronological order of occurrence.
of the data acquisition (DAQ) system is shown in Figure 7.
The data collection was performed through DAQ by con-
necting it with sensors using BNC cables and with a laptop 3.2. Data analysis
using a USB cable.
Various number of vehicles travelled across the bridge at Measured sensor data is shown in Figure 9(a) under a typ-
different speeds during the tests. The duration of each test ical traffic condition. The time-domain data is processed
was between 30 seconds and 2 minutes, and the sampling through a Fast Fourier Transform (FFT) and is shown in
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1253

Figure 8. Sensors (highlighted using yellow box) placed on the side walk and the vehicles running on the bridge during the test.

Table 1. Test details. Table 2. Comparison of the identified frequencies (Hz) with the FE model.
Large Total no. Total FE 2017 2018
Test # Car Truck truck of vehicles weights (lbs) 2.34 2.33 2.33
1 4 1 1 6 72,000 2.4 2.5 2.42
2 6 1 2 9 118,000 3.7 3.9 4.05
3 2 0 0 2 6000 5.8 5.5 5.5
4 0 1 0 1 20,000 7.6 7.8 7.7
5 7 1 2 10 121,000
6 3 0 1 4 49,000
7 14 2 3 19 202,000 information into a user-friendly software which can provide
8 2 1 0 3 26,000
9 0 0 1 1 40,000
easy access of long-term monitored information. For the
current case study, RevitV is utilised as the BIM platform to
R
10 4 2 0 6 52,000
11 6 1 1 8 78,000 integrate the SHM information. All the physical, geometric
12 2 1 0 3 26,000
and material properties are approximately assigned to the
model according to the data of the real bridge. The sensor
Figure 9(b). The frequencies corresponding to peak response parameters are assigned to determine the type, location,
amplitude are selected using the peak picking method. In extent and rate of damage in the structure. The current
RevitV extension creates a predefined list of colour scheme
R
this way, the preliminary modal frequencies are identified.
As shown in Figure 10, vibration response for sensor 5 is system for areas and keynote text files for piers, decks, bear-
plotted for a maximum and minimum number of vehicles. ings and provides user-defined families to represent each
The results clearly reflect that the response increases signifi- structural object and generates fundamental documents for
cantly with more number of vehicles. the entire model as shown in Figure 14.
In order to correlate the amount of vehicular weight that
passed by the bridge during each test, vehicle’s weights were
assumed for each type of vehicle from the website of the 3.4. Robot structural analysis
Ministry of Transportation of Ontario. As shown in Figure 11, Once the model is developed in RevitV, the structural ana-
R

the estimated maximum acceleration against vehicle weight lysis is performed using the RSA. RSA is Autodesk’s FEM
for sensor 5 and sensor 6 are plotted. The root-mean-square structural analysis program that has the capability to analyse
(RMS) values of the accelerations of each test are plotted in complex models with the FE auto-meshing modeling algo-
Figure 12. The RMS value is higher for heavier vehicles, which rithms. The designed RevitV model is exported to RSAV as
R R

caused more excitation, across the bridge during the test. shown in Figure 15 to perform analysis. The nonlinear static
Once the preliminary data analysis is performed, the vibra- analysis through RSAV considers the second-order effects,
R

tion data is processed through the SOBI method. The SOBI due to change in stiffness and moments generated due to
method has shown significant potential in ambient system the applied vertical forces at nodes. The current software
identification (Sadhu et al., 2017) to identify modal parame- allows basic structural analysis of many types of structural
ters of the structures, without requiring any knowledge of elements such as tension and compressions members, plastic
excitation. The SOBI method is employed to perform condi- hinges, cables, supports. A combined live load, dead load,
tion assessment of the bridge to identify natural frequencies and thermal load are applied on the model to extract the
based on both 2017 and 2018 data as shown in Table 2. maximum deflection under different seasons. The thermal
Typical resulting modal responses are shown in Figure 13. load effect is applied by assigning average temperature of
every month. Upon further analysis, the software generated
the graphical information which includes the forces, stresses,
3.3. Development of the BIM model
displacements, reactions and moments. The graphical repre-
After the data analysis and system identification are per- sentation of results allows viewing of the specific nodes,
formed, the primary task is to integrate the sensor bars, and members. In this article, the maximum deflection
1254 C. BODDUPALLI ET AL.

Figure 9. (a) Typical vibration data and (b) Fourier spectra of the measured data.

Figure 10. Time-history of mid-span sensor under maximum and min- Figure 12. RMS response of mid-span sensors under different traffic.
imum weights.

maximum deflection is observed at the panels with larger

span lengths.

3.5. BIM-integrated visualisation tool

Visualisation of sensor networks and structural data is one
of the central aspects of SHM. An appropriate visualisation
tool communicates efficiently and effectively between dispar-
ate groups working on a single project. The proposed
approach fosters communication by creating sensor proper-
ties through scheduled parameter table. Despite its signifi-
cant advancements, RevitV does not have the ability to read
R

all the outputs from other simulation software (e.g.,

MATLABV and RSAV). For successful integration of the
R R

information, all the output results are converted into the

corresponding readable files. The sensor data, FFTs, and
Figure 11. Peak vibration response of mid-span sensors under different traffic. system identification results of the sensor data are converted
into excel and jpeg files for the better interpretation and
results of the RSAV are converted into pdf files. The SHM
R

from the applied load combination is identified for each information including sensor number and location, time,
panel of the bridge to compare the results with the identi- data sheet and identification results are assigned to the sen-
fied frequencies from SHM. As shown in Figure 15, the sor properties. The identities of the structural objects such
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1255

Figure 13. Sample modal responses of the measured data.

The acquired sensor data includes all the sensor entities

that are featured directly in the BIM model to enable the
data interpretation as shown in Figure 18. In addition to the
sensor data, users can add the metadata of the installed sen-
sors, such as sensor type and model, to the BIM model
using the similar procedure that was explained and access
anytime. Figure 18 shows a sample of extracted sensor
metadata from the BIM model which are used in this case
study. The sensor information during the testing is obtained
in the form of text files. To make it user-friendly, the infor-
mation is exported into an excel file. Upon selecting the file-
name, the URL of the parameter links to the source file and
the application generates the sensor data as shown in Figure
19. In the analysis of the bridge, several parameters are eval-
Figure 14. RevitV model of the bridge.
R

uated such as natural frequencies and maximum acceler-

ation. The natural frequencies are traced by processing the
vibration data using FFT in MATLABV. Since, RevitV is
R R

as type, image, manufacturer, URL, description, cost, assem-

unable to read a MATLABV output file, the FFT plots are
R

bly code are inserted in the model such that these parame-
ters appear in the sensor properties. For each type of entity, linked to the scheduler parameter table through the url. On
the various attributes are defined depending on the data clicking the cells, the FFT plots are displayed in the parent
and attributes. The sensors are organised in their prescribed platform. The converted file is linked to the sensor entity in
RevitV to monitor the structural performance.
R

locations. RevitV visualises the sensor schedule table includ-

R

ing all the sensors as well as their entitled properties.

As shown in Figure 16, the SHM data is visualised with 3.6. RSA-enhanced comparison
all information. The corresponding results from the
MATLABV and RSAV are shown in the table through URL. The enhanced features of RSAV software enable to simulate
R R R

As shown in Figure 17, the properties of the sensor are and analyse the structures with enormous algorithms. The
simulation can be run directly in RSAV, or the structure can
R
assigned through ‘scheduled parameter’ command. Through
the shared parameter command, the sensor properties are be exported to Autodesk Simulation. The load simulation
added by selecting ‘Add parameter’ command, which allows feature allows to simulate the wind flow around the struc-
assigning all the required information to the sensor entity. ture and also allows the live load and dead load simulations.
For the current case study, different structural properties The enhanced feature of Robot is useful for large structures
including forces, sampling frequency, MATLABV figures,
R
with complicated geometry. The results of this simulation
sensor data, time of the test, deflections and other informa- are used for further analysis. The structure is designed
tion are added to the sensor properties. One of the unique under different load combinations. The loads are categorised
features of the sensor schedule table is that, when a specific based on the regular seasons in a year. In January, a lot of
sensor is selected on the model, the selected sensor automat- snow occurs which also adds up along with the live loads
ically highlights all the essential parameters assigned to the and dead loads. Whereas, for the other seasons (May,
specific sensor. October), the snow load is replaced with heavy wind loads
1256 C. BODDUPALLI ET AL.

Figure 15. Transformation of the BIM model from RevitV to RSAV.

R R

Figure 16. Integrated visualisation tool.

 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1257

Figure 17. Sensor properties added to RevitV model.

R

and traffic loads. Based on these assumptions and load com- results are separated based on the outputs such as reaction
binations, structural analysis is performed. Once the analysis forces, moments and deflections. The converted pdf format
of the RSAV file is linked to the sensor entity to characterise
R
is performed, the software generates a model to identify the
structural behavior. After the analysis is performed, the the performance of the structure. The structural integrity is
developed by exporting the images and reports of the RSAV
R
results are converted into a pdf format which provides a
work-space into RevitV. The output source files are format-
R
better understanding of the structural behavior and elimi-
nates the requirement of another design software. These ted as pdf format and are linked to the cells of reaction
1258 C. BODDUPALLI ET AL.

Figure 18. Visualisation of the sensor data.

Figure 19. Visualisation of the analysed data.

 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1259

Figure 20. Details of deflection results obtained from RSAV.

R

the RevitV provide the primary basis of collected informa-

R
forces and deflections. Upon clicking the cells, the scheduled
parameter table generates the necessary outputs that are tion and analysed results. The results replicate the behavior
required to access the performance of the structure. of the structure and mobilise the necessary remedies to
Figure 20 shows the representation of deflections in the avoid fatalities in the structure. The given approach identi-
structure. The deflection results are obtained to give the fies the damage states during the structural diagnosis and
prolonged characterisation of the structure regarding static also characterised the evolution phase, based on the analysis
analysis. The resulted deflection values are useful in finding and interpretation through a numerical model of the struc-
the stiffness of the structure throughout its length. Thus, the tural system through the given sensor parameters.
obtained results are converted into pdf format and linked to To show the effectiveness of the proposed visualisation
the sensor entity. Stiffness values from the sensor informa- tool, resulting time-efficiency is compared with the manual
tion and the static analysis help in analysing the perform- documentations. Considering just typical 1 hour duration of
ance of the structure during its service period and also assist bi-weekly data measured over a period of 6 months, Table 3
in monitoring the structure throughout its life period. All compares the total duration that is needed for decision mak-
the results from RSAV are reported in pdf formats as shown
R
ing. In this case study, it takes around 30 minutes by SOBI
in Figures 20 and 21. The obtained results imported into to process one hour data sampled at a rate of 200 Hz. In
1260 C. BODDUPALLI ET AL.

Figure 21. Details of reaction forces obtained from RSAV.

R

Table 3. Performance evaluation (time in minutes) of the proposed visualisa- 4. Conclusions

tion tool.
Week # 2 4 6 8 10 12 14 16 In this article, an attempt is made to develop a user-friendly
Proposed 30 30 30 30 30 30 30 30 visualisation tool using BIM and facilitate decision making
Manual 30 60 90 120 150 180 210 240 on maintenance and rehabilitation of large-scale structures
using long-term SHM data. For bridge maintenance engin-
case of manual documentation, it is assumed that the previ- eer, it is always a challenge to track the progressive deterior-
ous data sets will be processed to compare the evolution of ation of highway bridges during their service life. Although
frequencies over time. Whereas, in case of the proposed SHM serves as a valuable data-driven condition assessment
visualisation tool, the real-time system identification results methodology, it is a daunting task to analyse and interpret
will be stored and available with no additional run-times. big data resulting from the long-term monitored structures.
Such time-savings can be considered as a significant The proposed BIM-based visualisation tool solves above
improvement over existing manual documentation where a unique challenge of SHM by utilising Autodesk’s
bridge engineer had to prepare separate tables of maximum Revit platform.
response and modal frequencies of the bridge from the The proposed tool is implemented and validated using a
long-term monitored data. real-life bridge through SHM data collected under varied
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1261

temperature conditions. In the proposed approach, the sen- Catbas, F. N., Susoy, M., & Frangopol, D. M. (2008). Structural health
sor system and BIM are integrated to improve data manage- monitoring and reliability estimation: Long span truss bridge appli-
cation with environmental monitoring data. Engineering Structures,
ment and visualisation of big SHM data. The proposed
30(9), 2347–2359. doi:10.1016/j.engstruct.2008.01.013
BIM-based models enable a valuable digital representation Chan, B., Guan, H., Hou, L., Jo, J., Blumenstein, M., & Wang, J.
of SHM information of the structural systems throughout (2016). Defining a conceptual framework for the integration of
their entire life cycle, thus enhancing the quality and assess- modeling and advanced imaging for improving the reliability and
ment of the infrastructure. To demonstrate the integrated efficiency of bridge assessments. Journal of Civil Structural Health
visualisation tool, the modal frequencies from the sensor Monitoring, 6(4), 703–714. doi:10.1007/s13349-016-0191-6
Delgado, J. M. D., Brilakis, I., & Middleton, C. (2016a). Modelling,
data are compared with identified frequencies of the FE management, and visualization of structural performance monitor-
model. The results show the usefulness of the proposed ing data on BIM. In: Proceedings: Smart Infrastructure and construc-
approach in detecting and visualising the potential hazards tion (pp. 543–549).
during its service period. Delgado, J. M. D., Butler, L. J., Gibbons, N., Brilakis, I., Elshafie, M., &
The novel contribution of this research lies in unique Middleton, C. (2016b). Management of structural monitoring data
of bridges using BIM. Proceedings of the Institution of Civil
integration of SHM and resulting dynamic information such Engineers - Bridge Engineering, 170(3), 204–2018. doi:10.1680/
as system identification results within the BIM model which jbren.16.00013
benefits infrastructure owners to decide upon future main- Del Grosso, A., Basso, P., Ruffini, L., Figini, F., & Cademartori, M.
tenance strategies in a cost-effective manner. The proposed (2017). Infrastructure management integrating SHM and BIM pro-
idea would enable systematic visualisation of condition cedures. In: Proceedings of the 4th Conference on Smart
Monitoring, Assessment and Rehabilitation of Civil Structures,
assessment data that are collected in a continuous basis.
Zurich, Switzerland.
Apart from direct integration of system identification with Fanning, B. (2014). Impacts and benefits of implementing BIM on
BIM, the proposed work-flow also compares results of exist- bridge and infrastructure projects (Master’s Thesis). Colorado State
ing SHM data with the RSA to evaluate the structural deg- University.
radation. Finally, it is demonstrated using a full-scale bridge Friesen, P., & Sadhu, A. (2017). Performance of tensor decomposition-
based modal identification under nonstationary vibration. Smart
under various traffic and environmental conditions.
Materials and Structures, 26(3).
Glisic, B., Yarnold, M., Moon, F., & Aktan, A. (2014). Advanced visu-
alization and accessibility to heterogeneous monitoring data..
Acknowledgements Computer-Aided Civil and Infrastructure Engineering, 29(5),
This research was funded by NSERC Discovery and Engage grants of 382–398. doi:10.1111/mice.12060
the second author. In-kind contributions and assistance from WSP Hu, X., Wang, B., & Ji, H. (2013). A wireless sensor network-based
Canada Inc. during the bridge instrumentation were greatly structural health monitoring system for highway bridges. Computer-
acknowledged. Aided Civil and Infrastructure Engineering, 28(3), 193–209. doi:
10.1111/j.1467-8667.2012.00781.x
Kay, S., & Eike, T. (2015). IFC-based monitoring information model-
References ling for data management in structural health monitoring. In
International Conference on the Application of Computer Science and
Abe, M. (1998). Structural monitoring of civil structures using vibra- Mathematics in Architecture and Civil Engineering.
tion measurement: Current practice and future. In: I. Smith (Ed.), Ko, J. M., & Ni, Y. Q. (2005). Technology developments in structural
Artificial intelligence in structural engineering (pp. 1–18). Berlin: health monitoring of large-scale bridges. Engineering Structures, 27,
Springer. 1715–1725.
Anil, E. B., Akinci, B., Kurc, O., & Garrett, J. H. (2015). Building-infor- Li, H., Ou, J., Zhao, X., Zhou, W., Li, H., Zhou, Z., & Yang, Y.
mation-modelling-based earthquake damage assessment for rein- (2006). Structural health monitoring system for the Shandong
forced concrete walls. ASCE Journal of Computing in Civil Binzhou yellow river highway bridge. Computer-Aided Civil and
Engineering, 30(4), 04015076. Infrastructure Engineering, 21(4), 306–317. doi:10.1111/j.1467-
Ballor, J., Maclain, O., Mellor, M., Cattaneo, A., Harden, T., Shelton, 8667.2006.00437.x
P., Martinez, E., Narushof, B., Moreu, F., & Mascarenes, D. Lynch, J. P., Farrar, C. R., & Michaels, J. E. (2016). Structural health
(2019). Augmented reality for next-generation infrastructure monitoring: Technological advances to practical implementations.
inspections. Model Validation and Uncertainty Quantification, 3, Proceedings of the IEEE, 104(8), 1508–1512. doi:10.1109/
185–192. JPROC.2016.2588818
Banfia, F., Barazzettia, L., Previtali, M., & Roncoronib, F. (2017). Malkawi, A. M., & Srinivasan, R. (2005). A new paradigm for human-
Historic BIM: A new repository for structural health monitoring. building interaction: the use of CFD and augmented reality.
International Archives of the Photogrammetry, Remote Sensing and Automation in Construction, 14(1), 71–84. doi:10.1016/
Spatial Information Sciences, XLII-5/W1, 269–274. j.autcon.2004.08.001
Belouchrani, A., Abed-Meraim, K., Cardoso, J., & Moulines, E. (1997). Marzouk, M. M., & Hisham, M. (2012). Bridge information modeling
A blind source separation technique using second-order statistics. in sustainable bridge management. In ICSDC 2011: Integrating
IEEE Transactions on Signal Processing, 45(2), 434–444. doi:10.1109/ Sustainability Practices in the Construction Industry
78.554307 Maximilian, S., & Kosmas, D. (2016). BIM-Based Modelling of struc-
Bhusar, A., & Akhare, A. R. (2014). Application of BIM in structural tural health monitoring systems using the IFC standard. In Forum
engineering. International Journal of Civil Engineering, 1(5), 12–20. Bauinformatik.
Cai, G., & Mahadevan, S. (2016). Big data analytics in online structural McGuire, B. M. (2014). Using building information modeling to track
health monitoring. International Journal of Prognostics and Health and assess the structural condition of bridges (Master’s Thesis).
Management, 7, 2153–2648. Colorado State University.
Carden, E. P., & Fanning, P. (2004). Vibration-based condition moni- Minehane, M. J., O’Donovan, R., Ruane, K. D., & O’Keeffe, B. (2012).
toring: a review. Structural Health Monitoring, 3(4), 355–377. doi: The use of 3D laser scanning technology for bridge inspection and
10.1177/1475921704047500 assessment (Master’s Thesis) University College Cork, Ireland.
1262 C. BODDUPALLI ET AL.

Musafere, F., Sadhu, A., & Liu, K. (2016). Towards damage detection
using blind source separation integrated with time-varying auto- M€x ðtÞ þ C_x ðtÞ þ KxðtÞ ¼ FðtÞ; (1)
regressive modeling. Smart Materials and Structures, IOP, 25(1),
015013. where x(t) is a vector of displacement coordinates at the degrees of
Napolitano, R., Blyth, A., & Glisic, B. (2017). Virtual environments for freedom. M, C and K are the mass, damping and stiffness matrices of
structural health monitoring. In Conference: Structural Health the system respectively, and FðtÞ is the input excitation to the system
Monitoring. that is assumed to be Gaussian and broadband. The solution to
Napolitano, R., Blyth, A., & Glisic, B. (2018). Virtual environments for Equation (1) can be written in terms of superposition of vibration
visualizing structural health monitoring sensor networks, data, and modes with the following matrix form (Musafere, Sadhu, & Liu, 2016):
metadata. Sensors, 18(1). x ¼ As; (2)
National Building Information Modeling Standard (NBIMS), online at
http://www.wbdg.org/pdfs/NBIMSv1_p1.pdf. where, x e m  N is the measurement matrix composed of the sampled
Omenzetter, P., & Brownjohn, J. M. (2006). Application of time series components of x; s e n  N is a matrix of the corresponding modal
analysis for bridge monitoring. Smart Materials and Structures, coordinates, Amn is the modal transformation matrix and N is the
15(1), 129–138. doi:10.1088/0964-1726/15/1/041 number of data points of the measurements. The column of A matrix
Orcesi, A. D., & Frangopol, D. M. (2011). Optimization of bridge are linearly independent, and they represent the modeshape matrix of
maintenance strategies based on structural health monitoring infor- structure. The SOBI method (Belouchrani et al., 1997) aims to formu-
mation. Structural Safety, 33(1), 26–41. doi:10.1016/ late two covariance matrices Rx ð0Þ and Rx ðpÞ evaluated at the time lag
j.strusafe.2010.05.002 zero and p from the observed measurements, then simultaneously diag-
Park, S., & Kim, I. (2015). BIM-based quality control for safety issues in the onalise them in order to find unknown mixing matrix, A: The key
design and construction phases. International Journal of Architectural steps of the SOBI method are as follow;
Research, 9(3), 111–129. doi:10.26687/archnet-ijar.v9i3.881 h

T i
Poncelet, F., Kerschen, G., Golinval, J., & Verhelst, D. (2007). Output Rx ð0Þ ¼ E xðnÞ xðnÞ
only modal analysis using blind source separation techniques. ¼ ARs ð0ÞAT
Mechanical Systems and Signal Processing, 21(6), 2335–2358. doi: h

 i (3)
10.1016/j.ymssp.2006.12.005 Rx ðpÞ ¼ E xðnÞ xðn
pÞ T
Sadhu, A., Narasimhan, S., & Antoni, J. (2017). A literature review of
output-only structural modal identification employing blind source ¼ ARs ðpÞAT
separation methods. Mechanical Systems and Signal Processing, 94,
415–431. doi:10.1016/j.ymssp.2017.03.001
where
h

 i
Sadhu, A., & Goli, G. (2017). Blind source separation-based optimum Rs ðpÞ ¼ E sðnÞ sðn
pÞ T ¼ I (4)
sensor placement strategy for structures. Journal of Civil Structural
Health Monitoring, 7(4), 745–758. The measured responses xðnÞ are zero-mean, and the whitening is
Shakhramanyan, A., Kuckartz, J., & Kolotovichev, Y. A. (2012). obtained as follows. The singular value decomposition is used to diago-
Structural health monitoring system for high-rise and unique build- nalise Rx ð0Þ :
ings. In SODIS-Research and Production Association of Modern h

T i
Diagnostic Systems. Rx ð0Þ ¼ E xðnÞ xðnÞ ¼ Vx kx Vx T (5)
Shim, C. S., Yun, N. R., & Song, H. H. (2011). Application of 3D
bridge information modeling to design and construction of bridges. where, Vx is the eigenvectors matrix and kx is the eigenvalue matrix.
Procedia Engineering, 14, 95–99. doi:10.1016/j.proeng.2011.07.010 The whitened signals are then computed as shown in Equation (6)
Somwanshi, S., & Gawalwad, B. (2016). Monitoring civil structures where, Q is realised as whitening matrix,
with a smart wireless sensor network. International Journal of
x ðnÞ ¼ k
1=2
x Vx T xðnÞ
Engineering and Applied Sciences, 3, 34–39. (6)
¼ Q xð n Þ
Sony, S., Laventure, S., & Sadhu, A. (2019). A literature review of next-
generation smart sensing technology in structural health monitoring, The reason for whitening is to remove the correlation between the
Journal of. Structural Control and Health Monitoring, 26(3), e2321. measured responses.
doi:10.1002/stc.2321 h

T i
Theiler, M., Dragos, K., & Smarsly, K. (2017). BIM-based design of Rx ¼ E xðnÞ xðnÞ ½A ¼ I: (7)
structural health monitoring systems. In Proceedings of the 11th
Due to whitening process Rx ðpÞ becomes Rx ðpÞ; which is given by:
International Workshop on Structural Health Monitoring, Stanford, " #
CA, USA. 1 XN
T
Valinejadshoubi, M., Bagchi, A., & Moselhy, O. (2017). Managing R 
x ð p Þ ¼ 
x ð n x
Þ ð n
p Þ ¼ QRx ðpÞQT ¼ QARs ðpÞAT QT : (8)
N n¼1
structural health monitoring data using building information model-
ing. In Proceedings of the 2nd World Congress and Exhibition on Equation (8) reveals that whitened covariance matrix at a particular
Construction and Steel Structure, Las Vegas, NV, USA. time-lag can be diagonalised; therefore the product QA is realised as
Zhang, J. P., Liu, Q., Yu, Z. Z., & Zhao, W. (2014). A framework of unitary matrix and can be determined. During the orthogonalisation
cloud-computing-based BIM Service for Building Lifecycle. In process, the whitened covariance matrix Rx ðpÞ is diagonalised whose
Computing in Civil and Building Engineering (pp. 1514–1521). eigenvalue decomposition satisfies
Zhou, J. T., Li, X. G., Xia, R. C., Yang, J., & Zhang, H. (2017). Health
monitoring and evaluation of long-span bridges based on sensing Vx Rx ðpÞVx T ¼ kx (9)
and data analysis: A survey. Sensors, 17(3), 603. The diagonal matrix kx has distinct eigenvalues, thus the mixing
matrix can be estimated by the following equation,
^ ¼Q
1 Vx ¼ Vx kx T Vx :
A (10)
Appendix
In order to find the unitary matrix QA that diagonalises the whit-
Second-order blind identification ened covariance matrix Rx ðpÞ at one or several non-zero time lags, the
SOBI carries out an approximate joint diagonalization approach based
The basic dynamics of a structure can be considered as a linear, clas- on Givens rotation technique. Consequently, the associated problem
sically damped, and lumped-parameter n degrees of freedom system, becomes to find minimum performance index I; such that unitary
subjected to an excitation force, FðtÞ: diagonalisation satisfies D ¼ VT Rx ðpÞV (Belouchrani et al., 1997),
 STRUCTURE AND INFRASTRUCTURE ENGINEERING 1263

X X
IðV; pÞ ¼
p
jjDij jj2 ; (11) ^
1 xðnÞ;
^s ðnÞ ¼A (12)
p 1i6¼jns

where, V is the unitary matrix and also the joint approximate diago-
naliser for all p-shifted covariance matrices Rx ðpÞ (Belouchrani et al., where, ^s ðnÞ contains the modal responses in time-domain from which
1997). Therefore, the estimated sources can be obtained once the esti- the modal frequencies and damping ratio can be obtained (Poncelet,
mated mixing matrix is calculated, Kerschen, Golinval, & Verhelst, 2007).