You are on page 1of 10

Field Monitoring of RC-Structures under Dynamic

Loading Using Distributed Fiber-Optic Sensors


Zachary E. Broth 1 and Neil A. Hoult, M.ASCE 2

Abstract: The ability to properly assess existing reinforced concrete (RC) structures provides an opportunity to prevent costly rehabilitations
or replacements and aid in the optimization of future designs. Distributed fiber-optic sensing (DFOS) is a promising option in the assessment
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

of these complex structures. Previous uses of this technology, however, have been limited to static measurements, prohibiting the assessment
of structures exposed to dynamic loads. This research intends to assess the dynamic sensing capabilities of a DFOS system and, in turn,
improve the current understanding of the dynamic behavior of an existing RC beam tested in situ through a case study. Based on the
data provided by the dynamic distributed fiber-optic sensors (DDFOS), dynamic measurements of distributed strains, distributed deflections,
and crack widths can all be provided. Further details regarding the use of the DDFOS system in assessment, including the determination
of support conditions and dynamic response factors, are presented and discussed. DOI: 10.1061/(ASCE)CF.1943-5509.0001488. © 2020
American Society of Civil Engineers.
Author keywords: Fiber-optic sensors; Distributed strain measurement; Dynamic loading; Reinforced concrete (RC); Crack widths;
Deflections; Monitoring; Serviceability.

Introduction Techniques have also been developed by Brault and Hoult (2019b)
to measure distributed concrete surface strains and use that data to
Current assessment methods for complex and heavily redundant simultaneously measure RC beam deflections, crack spacing, and
reinforced concrete (RC) structures often require conservative as- crack widths. These methods were also applied to the field assess-
sumptions, potentially leading to unnecessary costly rehabilitation ment of statically loaded RC beams during the construction of a
or reconstruction (Middleton 1997; Bentz and Hoult 2016). Addi- shopping center in Ottawa, Canada (Brault et al. 2018). Most pre-
tionally, the inability to fully understand the behavior of these vious uses of DFOS, including these studies, have been limited to
existing structures prohibits the opportunity for optimization in new static measurements. This prohibits the measurement of dynamic
designs (Orr et al. 2014). This can have negative economic and strains, and the corresponding deflections and crack widths, as the
environmental effects on new construction (Orr et al. 2014). Con-
result of cyclic or moving loads. By combining DFOS with dy-
sequently, the need exists for monitoring technologies that can pro-
namic sensing capabilities, it could then be possible to better under-
vide accurate and robust data sets to enable a better understanding
stand the behavior of these complex structural systems, as well as
of the behavior of these structures, specifically when exposed to
provide sufficient data for the proper assessment and future opti-
both dynamic and static loads.
mization of RC-structures.
Most current structural assessment and monitoring techniques are
Dynamic distributed fiber-optic sensing (DDFOS) is a new tech-
limited to discrete measurements of individual structural metrics,
such as electrical resistance-based strain gauges and displacement nology that can measure strain along the length of a fiber-optic
transducers (Gastineau et al. 2009). These discrete measurements cable at acquisition rates as high as 250 Hz (LUNA Technologies
often prohibit the ability to monitor the global behavior of a structure, 2017). Existing studies utilizing DDFOS technology are limited;
such as the deflected shape, as well as localized behavior in locations however, it has been shown that external RC strains can be recorded
where the sensors are not present, including the presence and size of during laboratory fatigue tests of RC beams (Barrias et al. 2018a).
cracks (Brault and Hoult 2019a). Other examples of DDFOS in use do exist in the transportation
Distributed fiber-optic sensing (DFOS) technologies improve (Wheeler et al. 2018) and aerospace (Davis et al. 2016) industries;
on traditional monitoring methods by providing the ability to re- however, to date, there are no studies that exhibit the use of DDFOS
cord accurate strain data (∼1 microstrain) at a dense spatial reso- for the full-scale assessment of existing RC-structures.
lution (<5 mm) (Kreger et al. 2007). Regier and Hoult (2014) In order to both utilize and evaluate the capabilities of DDFOS
provided evidence that DFOS can measure detailed external RC technology for structural assessment, a field study has been con-
strains throughout a fiber-optic cable when applied to an RC bridge. ducted. The study investigates the behavior of an existing RC beam,
which was instrumented and tested, in situ, under both static and
1 dynamic loads. The objectives of this research program were to
Researcher, Dept. of Civil Engineering, Queen’s Univ., 58 University
Ave., Kingston, ON, Canada K7L 3N6. Email: z.broth@queensu.ca (1) monitor existing RC beams using both discrete and dynamic dis-
2
Associate Professor, Dept. of Civil Engineering, Queen’s Univ., tributed sensors under varying load cases, (2) evaluate the results of
58 University Ave., Kingston, ON, Canada K7L 3N6 (corresponding DDFOS measurements by comparing them with those from discrete
author). Email: neil.hoult@queensu.ca
sensors, and (3) utilize DDFOS data to determine dynamic measure-
Note. This manuscript was submitted on June 13, 2019; approved on
March 13, 2020; published online on May 20, 2020. Discussion period ments of deflection, crack widths, and dynamic response factors as a
open until October 20, 2020; separate discussions must be submitted for result of the maximum expected dynamic loading on a structure.
individual papers. This paper is part of the Journal of Performance of The following section of this paper provides a brief background
Constructed Facilities, © ASCE, ISSN 0887-3828. on structural assessment methods and FOS technologies, followed

© ASCE 04020070-1 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


by details of the field study. The results of this investigation are then Distributed RC Strain Measurement
presented and discussed.
Fiber-optic sensing (FOS) technologies are available in two forms,
discrete and distributed. Similar to traditional strain gauges, discrete
fiber-optic sensing methods can only measure strain data at a single
Background point within a fiber-optic cable. Fiber Bragg Gratings (FBG), for ex-
ample, are discrete sensors that can be multiplexed but are also more
Structural Assessment Methods expensive than electrical-based strain gauges and are ultimately lim-
ited by the number of discrete measurements that they can provide
Structural assessment is often required to determine the remaining
(Gebremichael et al. 2005).
integrity of RC structures. While the technology and methods em-
DFOS technologies offer an improvement on discrete strain mon-
ployed by engineers have improved with time, these advancements
itoring techniques by providing continuous strain measurements
have primarily focused on select assessment properties, including
throughout the length of a fiber-optic cable. One example of a DFOS
the determination and monitoring of load, strain, and displacements technology is Brillouin optical time domain reflectometry (BOTDR).
as well as detecting the presence and size of cracks (Chang et al.
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

BOTDR is most commonly used in the assessments of large struc-


2003). Structures can be monitored under normal working condi- tural systems due to its ability to monitor several kilometers of fiber-
tions; however, the application of maximum or unique loading optic cable with an accuracy of 30 microstrain (Mohamad et al.
cases can provide a better understanding of the future performance 2011). Similar to BOTDR, Brillouin optical time domain analysis
of the structure. These loads are most commonly applied statically (BOTDA) can also offer long-range sensing capabilities but requires
by placing objects of known weight overtop of the assessed struc- access to both ends of the fiber-optic cable, which can be difficult
tural members. Brault et al. (2018), for example, utilized the weight to achieve in a field assessment setting (Kurashima et al. 1990).
of six scissor lifts on an RC beam during the construction of a shop- Rayleigh backscattering is another form of DFOS that can measure
ping center. Dynamic loads have also been applied to structures in strains at a higher accuracy (∼1 microstrain) and smaller gauge spac-
situ often in the form of a moving vehicle or other moving load. For ing (5 mm) than Brillouin techniques. However, unlike BOTDR and
example, Tennyson et al. (2001) used large, multi-axle trucks to BOTDA, Rayleigh backscattering systems are limited to smaller
apply a moving load to the Taylor Bridge in Manitoba, Canada. sensing lengths up to 100 m (Kreger et al. 2007).
Repeated or cyclic loads, however, are less common in the field Previous studies have explored the application and capabilities
assessment of these structures. Salawu and Williams (1995) pro- of DFOS and Rayleigh backscatter technology in the assessment of
vide a review of dynamic loading techniques for the assessment existing RC-structures. Barrias et al. (2018b), for example, utilized
of bridge structures. These techniques are used to induce vibrations both DFOS and backscattering technologies to measure strains dur-
in order to monitor a structure’s dynamic response. Impactors, for ing the renovation of two structures in Barcelona, Spain. The DFOS
example, are one method for applying forced excitation to a struc- technology used in that study was able to accurately monitor the
ture by providing a repeated impact from an instrumented hammer effects of the renovations on a structure, as well as the increased
device, similar to cyclically applied load. These methods, however, construction loads applied to the structures during this process.
can be destructive and are not feasible for indoor applications of Similarly, Regier and Hoult (2014) used Rayleigh-based DFOS
structural assessment. technologies during the assessment of the Black River Bridge in
The ability to adequately monitor the effects of these applied Ontario, Canada, and found DFOS to be a feasible technology for
loads is often determined by the availability of monitoring and field assessment applications. This study also explored the use of
sensing technologies. Displacements within a structure are often DFOS data to estimate the full deflected shape of the structure,
measured using displacement transducers, such as linear variable which had previously only been done with Brillion based systems
differential transformers (LVDT) and linear potentiometers (LP). (Ohno et al. 2001). Data from the Black River Bridge study were
These devices can provide discrete measurements of displacement later used by Bentz and Hoult (2016) to perform model updating,
at the point of installation but are only useful when fixed to a known which included estimates of crack widths based on the DFOS strain
point of reference, reducing their applicability in large scale assess- data. The methods used for the determination of the deflected shape
ments (Lee and Shinozuka 2006). An alternative is a noncontact and crack widths were validated by Brault and Hoult (2019b)
displacement measurement method, such as digital image correla- through a series of laboratory RC beam tests. These theories were
tion (DIC). DIC has not only proven to be an effective method for applied by Brault et al. (2018), who monitored the behavior of RC
accurately measuring displacements within a structure, but has also beams during the construction of a shopping center.
been successfully used to locate and monitor cracks on the surface
of RC beams (Hoult et al. 2016). This method, however, requires a
line of sight between the camera and the element being monitored, Dynamic RC Strain Measurements
which can often be difficult in a field setting or during the assess- Dynamic RC strain measurements are most commonly taken using
ment of large structures. electrical-based strain gauges, which are limited to discrete mea-
Strain measurements for field assessments of structures are most surements. FBG can also measure dynamic strains at acquisition
commonly taken with electrical-based strain gauges, which, similar rates up to several thousand samples per second (Majumder et al.
to displacement transducers, can only provide discrete measure- 2008). Over 60 FBG sensors were used in the dynamic monitoring
ments of strain at the point of installation. The use of electrical- of a moving truckload during the assessment of the Taylor Bridge
based strain gauges, however, is often limited due to environmental in Manitoba, Canada (Tennyson et al. 2001). These sensors offered
conditions, project scale, and installation restrictions (Gebremichael a general understanding of the structure’s behavior due to the mov-
et al. 2005). Advancements in wireless and smart sensing technol- ing load; however, the discrete nature of the sensor type failed to
ogies have improved on the traditional strain gauge by simplifying provide details regarding localized behavior, such as cracking.
sensor installation and allowing for remote monitoring of a structure Recent advances in FOS technology have led to the development
(Spencer et al. 2004). However, the discrete nature of these measure- of DDFOS that can provide dynamic strain measurements continu-
ments once again restricts the extent to which these methods can be ously throughout a fiber-optic cable. The optical distributed sensor
used for assessment. interrogator (ODiSI-B) by LUNA Technologies (Roanoke, VA) is a

© ASCE 04020070-2 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


Rayleigh-based dynamic distributed analyzer that was used in the campus in Kingston, Ontario, Canada, which was designed and
present study. The ODiSI-B has acquisition rates up to 250 Hz, de- constructed in the late 1950s. The following section provides de-
pending on the total sensing length. For a maximum sensing length of tails regarding the RC beam that was tested as well as the instru-
20 m, however, the acquisition rate is 50 Hz, with a gauge spacing of mentation and testing procedure used for the investigation.
2.61 mm, gauge length of 5.2 mm, and accuracy of 25 microstrain
(LUNA Technologies 2017). As discussed later in this paper, an im-
proved accuracy of 5 microstrain was observed during this study. Field Site and Instrumentation
Although limited, research using the ODiSI-B and other The monitored site consists of a cast-in-place T-beam located below
DDFOS systems have found that this technology can be applied to a multi-purpose room between an external column and a perpen-
a variety of dynamic sensing applications and can provide high- dicularly intersecting beam. The instrumented beam had an acces-
resolution strain data at high acquisition rates (Kreger et al. 2013). sible length (i.e., wall-to-column) of 9.24 m. In addition to the
Some of the earliest applications of the ODiSI-B include a series of profile view of the specimen, Fig. 1(b) also provides a cross-section
small-scale laboratory tests performed by Kreger et al. (2013), such of the T-beam, as well as the location of the beams in relation to the
as the measurement of high-resolution dynamic strains in a swing-
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

room presented [Fig. 1(c)].


ing golf club. Examples of field applications of the ODiSI-B The schematics in Fig. 1 also provide details regarding the in-
include the work of Wheeler et al. (2018), who monitored dynamic strumentation used during this case study. Fiber-optic sensors were
longitudinal strains in railroad tracks and Wong et al. (2018) who installed along both the top and bottom of the beam. Starting
performed damage detection of in-service pipelines. To the authors’ 50 mm away from the wall face, the fiber-optic cable was installed
knowledge, however, there is currently no evidence of the use of 40 mm above the bottom edge of the beam until reaching the col-
DDFOS or Rayleigh backscatter systems to dynamically measure
umn face. The fiber was then looped back 40 mm below the slab
strains, deflected shapes, and crack widths during the assessment of
soffit to an end position approximately 0.62 m away from the wall.
in situ RC-structures.
The difference in the horizontal start and end position is due to the
sensing length limitation of the DDFOS system used (i.e., 20 m).
Experimental Program The installation of the fiber-optic cable involved a three-step pro-
cedure, including (1) sanding of the fiber pathway, (2) cleaning of
This study provides the results of a case study performed on an the fiber pathway, and (3) bonding of the fiber to the concrete sur-
existing RC beam within Ellis Hall on the Queen’s University face (Brault et al. 2018). Rotary sanding wheels were used to

Beam Center Midspan Column Center


Floor Slab

End FOS FOS 470 mm 40 mm


Start FOS

LP 3 40 mm LP 2 LP 1
Wall Face

9.24 m
9.10 m
6.93 m
4.62 m

(a)
Floor Slab

FOS Beam
550 mm 702 mm Location

356 mm

(b) (c)

Fig. 1. Beam schematics, including (a) beam profile and instrumentation; (b) beam cross-section; and (c) beam location within the 3rd floor floorplan
of Ellis Hall.

© ASCE 04020070-3 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


remove paint and smooth the concrete surface. The surface was to the two loading orientations as the point load (PL) and uniformly
then cleaned with water and 99% isopropyl alcohol before the fiber distributed load (UDL).
was bonded using a two-part epoxy (Loctite E-20HP). In addition Before the testing began, the students were positioned away
to the fiber optics, three LPs were installed at the midspan (LP 3), from the beam’s tributary area, on the opposite side of the room.
quarter-span (LP 2), and 0.14 m away from the column face (LP 1), The students were then asked to slowly walk toward the beam and
as shown in Fig. 1(a). It is worth noting that the fiber needs to be form their position in the UDL orientation. Once in place, the stu-
continuously bonded if the full surface strain profile is required, dents were then asked to jump in unison for 30 consecutive jumps.
although even with continuous bonding, crack locations result in Following the jumping cycles, the students walked off the beam
strain peaks being measured rather than zero strain, as noted in pre- back toward the other side of the room, unloading the specimen.
vious research studies (Regier and Hoult 2014). However, it is also After a 1 min and 30 s break, this process was then repeated for
possible to measure the average strain between bonded points if the PL test. By having the students walk on and off the beam before
sufficient tension in the fiber is maintained between bonded loca- and after each test, the unloaded (i.e., off the beam), statically loaded
tions to prevent sagging of the cable. (i.e., at rest on the beam), and dynamically loaded (i.e., jumping)
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

state of the beam could be measured. It should be noted that the data
acquisition rate of both the DDFOS and displacement data was
Testing Procedure
50 Hz, and the data was logged continuously throughout the tests.
The application of the dynamic loads applied during this investi- Therefore, any long-term effects caused by the UDL test prior to the
gation was provided by 38 student volunteers who were asked to PL test would be seen in the data. This is discussed further in the
jump in unison over the instrumented specimen. The use of human results section.
loading allowed for easier management of the load placement,
which resulted in the completion of two separate tests, including
a distributed and concentrated load test. The distributed load test Results and Discussion
was conducted by aligning the students across the full span of
the classroom (wall-to-wall) directly above the instrumented beam.
Deflection Behavior
For the concentrated load test, the students were gathered in a group
over the midspan of the beam. This concentrated load was intended Deflection data can be provided for the complete duration of the
to represent a point load. However, the group of students filled an two tests from the three LPs that were used during the investigation,
area that was approximately 3 by 3 m area centered over the mid- including the midspan, quarter-span, and column support. Fig. 2
span position within the tributary area of the beam. Based on an provides the deflection versus time data from two of the three
average recorded mass of the volunteers of 75 kg, the applied con- LPs (i.e., midspan and quarter-span). The data from the third LP
centrated load was determined to be approximately 28 kN, while were omitted due to the minimal deflection measured at the column
the distributed load was approximately 3 kN=m. This study refers support. The deflection measured by each LP began at zero, which

0 0
Displacement (mm)

Displacement (mm)

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 20 40 60 80 100 0 20 40 60 80 100
(a) Time (Seconds) (c) Time (Seconds)

0 0
Displacement (mm)

Displacement (mm)

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 20 40 60 80 100 0 20 40 60 80 100
(b) Time (Seconds) (d) Time (Seconds)

Fig. 2. Displacement results from the LPs, including (a) quarter-span UDL test; (b) midspan UDL test; (c) quarter-span PL test; and (d) midspan
PL test.

© ASCE 04020070-4 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


represents the unloaded condition of the beam. As the test began, could have resulted in the formation of new cracks or the growth of
the deflection slowly increased as the volunteers walked onto the existing cracks. Additionally, the results show that the magnitude of
beam. Once positioned over the beam, the deflections due to the the strain peaks at cracks is greater during the PL test, specifically
static load can be measured. This occurred at approximately 30 s within the middle third of the fiber length. This is likely due to
into each test. As a result of the static loads, the beam experienced the concentration of the load within this area during the PL test.
0.25 mm under the UDL and 0.34 mm during the PL test at Information regarding the observed cracking behavior during the
midspan. two tests, including the possibility of crack growth, is discussed
Once the jumping cycles began, the results show that the deflec- in greater detail later in this section.
tion data followed a repeating pattern that includes the deflection As expected, strains within the top fiber are small due to the top
response of the beam due to the push-off force of the students, time fiber location near the neutral axis of the beam. A few strain peaks,
the students were airborne (i.e., unloaded), impact of the students, however, can be observed within the top fiber data. Similar to the
and beam’s response during the time the students were at rest be- cracks measured along the bottom fiber, these strain peaks suggest
tween jumping cycles. During the UDL test, there was an acciden- that some cracks originating from the bottom tensile region of the
tal interruption of the jumping cycles seen at approximately 38 s
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

beam have grown to the height of the top fiber-optic cable. This is
within the deflection data. Within the jumping cycles, the unloaded supported by the alignment of the measured cracks within the top
phases of each jump do not always return to zero. This is attributed and bottom fibers, as it is expected that the flexural cracks would
to the timing of the volunteers’ jumps where it is unlikely that all grow vertically through the beam’s depth.
the volunteers were airborne at the exact same moment. Addition- The bottom fiber data also provide insight into the boundary
ally, the amount of time between the unloaded and impact phases conditions of the instrumented beam, including inflection points
may have not been long enough for the beam to respond to the near the supports of the beam (at approximately 2 and 8 m). Beyond
push-off force and return to the zero position before the impact the two inflection points, compressive strains are measured in the
occurred. Overall, the maximum dynamic deflections under the bottom fiber, suggesting that there is moment transfer at the sup-
two loading cases at midspan were 0.54 and 0.84 mm for the UDL ports. A single negative strain peak can be observed at approxi-
and PL tests, respectively. mately 1 m within the bottom fiber from each test. This peak likely
Following the jumping phase of the two tests, the beam returned represents the closing of a crack at this location due to the com-
to the statically loaded position while the volunteers were at rest. pressive stresses experienced in this area.
The beam was then unloaded as the volunteers walked off the beam
and away from the tributary area of the specimen. After the com- Dynamic Response Factor
pletion of the test, when the beam was unloaded, there are remain- By comparing the maximum distributed strains as a result of the
ing deflections of 0.03 and 0.04 mm at the midspan for the UDL dynamic jumping loads with the average static strain distribution,
and PL tests. Due to the intensity of the loading applied to the struc- it might be possible to quantify the dynamic response factor (DRF)
ture during the two tests (compared to the normal working loads on for the two loading arrangements. Fig. 4 provides both the static
the structure), the remaining deflections that can be measured are as and maximum dynamic strain distributions for both the UDL and
the result of crack development or growth. These residual deflec- PL tests. Based on this data, a distribution of DRF values was de-
tions, however, could also be the result of minor shifts in the LP termined throughout the fiber length. DRF values are important in
sensors due to the vibrations caused by the repeated impact loads or
because their supports moved. Temperature can also have an impact
on the fiber-optic measurements using Rayleigh backscatter for 400
Bottom Fibre
experiments occurring over a longer period of time (e.g., Davis Top Fibre

et al. 2017); however, it was not expected to impact the current tests 300
Strain (Microstrain)

Crack 1
that occurred over a duration of less than 5 min within a building.
200
This behavior is discussed in greater detail later in the results
section. Top Fiber
100
Ends

Distributed Fiber-Optic Strains 0


Inflection Point Inflection Point

Maximum Strain Profile -100


0 1 2 3 4 5 6 7 8 9
Utilizing the data from the DDFOS system, the exact point in time (a) Fibre Length (m)
at which the maximum strains occurred can be identified. Fig. 3
provides the maximum distributed strain profiles for the top and 400
Bottom Fibre Crack 1
bottom fiber that occurred during each test. Note that the top fiber Top Fibre
data does not cover the entire span of the beam due to sensing 300
Strain (Microstrain)

length restrictions; therefore data are not displayed beyond that


point (approximately 0.62 m). 200

Multiple strain peaks can be observed within the bottom fiber


100 Top Fiber
data provided in Fig. 3 that represent the location of cracks that Ends
intersect the bonded fiber-optic cable. As noted previously, strain
0
peaks occur at crack locations as the fiber-optic cable debonds Inflection Point Inflection Point
locally and bridges the crack (Regier and Hoult 2014). It is also -100
possible that the strain change could be due to fiber debonding in 0 1 2 3 4 5 6 7 8 9
the absence of cracking; however, if this were the case, one would (b) Fibre Length (m)
expect to measure the average strain between the two bonded points
Fig. 3. Maximum strain profiles along the top and bottom of the beam
rather than a strain maximum. The strains at these cracks exceed
for (a) the UDL test; and (b) the PL test.
the expected cracking strain of concrete (∼100 microstrain), which

© ASCE 04020070-5 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


structural assessments, as they allow for a good approximation of could occur at the location of cracks even if no change in the crack
the maximum dynamic load when applied to the results of a static geometry or material properties occurred. Note that they did not see
test. Although this information may not be useful for the assess- this behavior in the results from other sensor locations positioned
ment of an RC beam within an academic building, the ability to away from or in between cracks, as the strains in these cases re-
accurately determine a DRF value would be valuable in the assess- turned to zero after the beam was unloaded. This suggests that these
ment of structures that are exposed to dynamic loads, such as residual strains likely only occur as a result of peak strains at the
bridges, culverts, and arenas. location of cracks. Further discussion regarding this phenomenon is
The average of the distributed DRF results was determined to be provided later. There is also an apparent linear increase in the
1.8 and 1.6 for the UDL and PL test, respectively. However, it is strains throughout the jumping cycles during the two tests. As a
clear from the DRF profile that there is significant variation in these result, the static and unloaded strains following the jumping cycles
profiles due to the comparison of small strains in the uncracked are larger than the strains measured before the cycles began.
areas of the fiber length, which, combined with an approximate
5 microstrain accuracy, has a significant effect on the results as Crack Widths
the range of calculated DRF goes from under 1 to over 3. An alter- To further investigate the cause of the residual strains shown in
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

native measure of the beam’s DRF can be calculated based on the Fig. 5, as well as observe the dynamic behavior of cracks, an exami-
static and dynamic deflection measurements from the LP data pro- nation of the crack widths with time for the crack labeled in Fig. 3
vided in Fig. 2. The results of this calculation are 2.16 and 2.47 for was conducted. Brault and Hoult (2019b) proposed a method of
the UDL and PL test, respectively. It is clear that there is a signifi- estimating the area under the concrete surface strain plot in the
cant discrepancy between the DRF results and the DDFOS and LP vicinity of a crack to calculate the crack width. They proposed
data. Previous studies, such as Bakht and Pinjarkar (1989) and the two methods, and in this study, the method used was where the
Highway Research Board (1962), have discussed similar trends crack width is calculated using the area of a triangle with the maxi-
when comparing DRF results based on deflection measurements mum strain at the crack as the peak of the triangle and the slopes of
to those based on strain. In each case, the DRF values determined each side were defined by the minimum strains on either side of the
using deflection measurements were found to be greater than those crack. Based on this method provided by Brault and Hoult for static
for the strain-based results. However, in the current study, the crack width estimations, dynamic crack widths could be calculated
amount of variation in the DDFOS measurements means that it for each sampling of the DDFOS data and are provided in Fig. 6.
is not clear whether the difference in DRF values is a function The data display the opening and closing of cracks as a result of the
of the measurement technique or the sensor measurement accuracy. applied loads. Based on these results, both the static crack width
Thus, while using DDFOS data to calculate DRF values may be and maximum dynamic crack width can be determined. The two
possible if larger strains are present, it is not recommended for cracks grow throughout the duration of both tests, resulting in
the level of strain measured in this work (i.e., average values less an increased crack width due to static loading, as well as a residual
than 50 microstrain). crack width after the beam was unloaded. This is an important re-
sult since though locked-in strains can affect the minimum crack
Maximum Strain with Time width measurement, they would not affect the maximum crack
Fig. 5 provides the strain versus time data for the sensor located at width measurement and so it appears the cracks did grow during
the maximum strain location (4.22 m) throughout the UDL and PL the test. This behavior would result in both the residual strains seen
tests. Note that the maximum strains occurred at the location of in Fig. 5 and the remaining deflections shown in Fig. 2 after the
Crack 1, as labeled in Fig. 3, during both tests. It is important beam was unloaded. Following the UDL tests, the crack continues
to note that the strains at the beginning of the PL test are not zero. to grow throughout the static loading and jumping cycles of the PL
The UDL test was performed first followed by the PL test (1 min test, resulting in further crack growth after the beam was unloaded
and 30 s later), and the residual strains as a result of the UDL test the second time. The total crack growth measured after the com-
were carried over into the PL test. These residual strains observed at pletion of both tests is approximately 0.0025 mm. Although the
Crack 1 could be caused by the widening and growth of the crack, crack widths when the beam is unloaded could have been affected
resulting in an irrecoverable strain after the beam was unloaded by the residual strain phenomena explored by Brault et al. (2018),
during the UDL test. Creep could also contribute to the develop- these phenomena should not impact the maximum strains and thus
ment of these strains. Since it is assumed that the dynamic loads crack widths measured.
applied throughout the jumping cycles are likely the highest loads
experienced by the beam within its lifetime, they could have caused Deflected Shape
the cracks to extend and widen, contributing to the remaining de- Utilizing the strain data from both the bottom and top fiber-optic
flections measured at the LP locations along the RC beam after it cables and the distance between them, it is possible to calculate the
was unloaded. It is most likely that both mechanisms were occur- curvature along the fibers’ length. By applying Bernoulli–Euler
ring in the beam as short-term displacement increases have also elastic beam theory and the validated methods provided by Brault
been observed in the lab due to creep under constant load and a and Hoult (2019b), the curvature results can be double-integrated to
decrease in tension stiffening at the cracks under cyclic loading determine distributed bending displacement throughout the length
(Broth and Hoult 2020). However, isolating the effects of each of the instrumented section, as shown in Fig. 7. Though Brault and
mechanism on the behavior is not possible with the available data Hoult did this using static measurements, the method is applied for
as both mechanisms are highly variable, nonlinear, and time and the first time here using dynamic measurements. Fig. 7 shows the
load dependent. Both crack growth and concrete creep could also calculated deflected shape for the statically loaded case, maximum
explain the remaining displacements measured by the LP following dynamically loaded case, and unloaded case following the jumping
each test, as shown in Fig. 2. cycles for both load cases. Note that for each deflected shape to be
Another explanation for these residual strains, as investigated by calculated, two boundary conditions must be used. For all of the
Brault et al. (2018), could be due to locked-in strains within the results provided in Fig. 7, both the midspan and support LP data
fiber-optic cable or the epoxy used to bond the fiber-optic cable were used as boundary conditions, while the quarter-span LP was
to the concrete. Brault et al. indicated that irrecoverable strains used to verify the results. Also, note that deflection data could not

© ASCE 04020070-6 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


400 400
Maximum Dynamic Strain Maximum Dynamic Strain
Average Static Strain Average Static Strain

Strain (Microstrain)

Strain (Microstrain)
300 300

200 200

100 100

0 0
0 2 4 6 8 0 2 4 6 8
(a) Fibre Length (m) (c) Fibre Length (m)
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

5 5

Dynamic Impact Factor


Dynamic Impact Factor

4 4

3 3

2 2

1 1

0 0
0 2 4 6 8 0 2 4 6 8
(b) Fibre Length (m) (d) Fibre Length (m)

Fig. 4. Dynamic response factor results, including (a) the dynamic/static strain profiles from the UDL test; (b) resulting DRF distribution for the UDL
test; (c) the dynamic/static strain profiles from the PL test; and (d) resulting DRF distribution for PL test.

400 400
Strain (Microstrain)

Strain (Microstrain)

300 300

200 200

100 100

0 0
0 20 40 60 80 0 20 40 60 80 100
(a) Time (Seconds) (b) Time (Seconds)

Fig. 5. Strain with time data for (a) UDL test at maximum strain location (x ¼ 4.22 m; Crack 1); and (b) PL test at maximum strain location
(x ¼ 4.22 m; Crack 1).

0.02 0.02
Crack Width (mm)

Crack Width (mm)

0.015 0.015

0.01 0.01

0.005 0.005

0 0
0 20 40 60 80 0 20 40 60 80 100
(a) Time (Seconds) (b) Time (Seconds)

Fig. 6. Crack width measurements with time for Crack 1 identified in Fig. 3, including (a) Crack 1—UDL; and (b) Crack 1—PL.

© ASCE 04020070-7 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


0.2 LP-Static LP-Dynamic LP-Unloaded
FOS-Static FOS-Dynamic FOS-Unloaded To determine the exact cause, a direct measurement of the deflec-
Displacement (mm) 0 tion at this support would be required. It is also noted that the maxi-
-0.2 mum deflection under the dynamic load is positioned to the right of
Difference = 0.004 mm
-0.4
Difference = 0.013 mm
the midspan LP. This could be the result of the uneven positioning
-0.6 Difference = 0.034 mm of the load but could also be an effect of the left-hand support
-0.8
behavior.
Similar to the statically loaded condition, the unloaded deflected
-1
0 1 2 3 4 5 6 7 8 9 shape is as expected. What is most important about the unloaded
(a) Beam Span (m) deflected shape is that the two sensor types agree that there are
residual displacements within the beam after the loads are removed.
0.2 LP-Static
FOS-Static
LP-Dynamic
FOS-Dynamic
LP-Unloaded
FOS-Unloaded
This in turn helps explain the residual strains observed in Fig. 5 and
0 verify the crack growth estimations provided in Fig. 6.
Displacement (mm)

-0.2
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

-0.4 Difference = 0.013 mm


Difference = 0.025 mm
Applications and Limitations
-0.6
Difference = 0.043 mm
-0.8 Although presented for a specific structure in this case, the tech-
-1 niques developed in this paper can be applied much more broadly.
0 1 2 3 4 5 6 7 8 9
Dynamic distributed strain profiles enable the average strain to be
(b) Beam Span (m)
determined along a member, which can then be used with analysis
Fig. 7. Maximum integrated deflections from DFOS results for (a) the techniques, such as the modified compression field theory (Vecchio
UDL test; and (b) the PL test. and Collins 1986) or finite element analysis (Bentz and Hoult 2016),
to assess existing structures. In addition, crack width measurements
are critical for assessing the durability of a structure [ACI Committee
224 (ACI 2001)] and so understanding the range of crack widths that
be calculated within the first 0.62 m of the instrumented span due to occur under live load would allow for the likelihood of deterioration
the lack of data from the top fiber. to be assessed more effectively. Finally, the data from this study,
The results provided in Fig. 7 allow for a direct comparison be- along with others to ensure accuracy and repeatability, can be used
tween the two sensor types (i.e., FOS and LP). While the midspan along with a finite element model updating approach to develop
and support LP are used in calculating the DFOS deflections, the potentially more accurate design models (Bentz and Hoult 2016).
difference between the quarter-span LP and the integrated deflec- These design models can then be evaluated using the results of fur-
tion at that location enables a comparison between both of the de- ther field monitoring on structures designed using these approaches,
flection measurements. For each deflected shape shown in Fig. 7, leading to eventual improvements in RC design.
the difference between these two points is provided. In each case, The three main limitations of the current technique are the
these differences are within the expected accuracy of the LP sensors (1) maximum sensing length, (2) maximum data acquisition rate,
used (approximately 0.1 mm). It is noted that there is an increase in and (3) the inability to measure reinforcement strains. The maxi-
the discrepancy with an increase in the load. This is potentially due mum sensing length of the system is 20 m, limiting the length of
to the level of noise measured within the DDFOS data as the loads members that can be completely instrumented to approximately
are applied, especially within the top fiber. As previously found by 9 m since some fiber is lost connecting to the beam and joining
Van Der Kooi et al. (2018), this particular measurement system has the top and bottom fibers. This is shorter than the beam instru-
higher measurement noise, resulting in fewer useable data points in mented in this study and led to the need for supplemental displace-
the last 5 m of the fiber [the length of fiber near the point marked ment measurements to determine the boundary conditions. For
“End FOS” in Fig. 1(a)]. This means that the double integration beams less than 9 m in length, an assumption of zero displacement
of curvatures in this region to get displacements is based on fewer at the ends of the beam can be made to calculate the displacement
points and that an increased likelihood that local strain variations and avoid the need for supplemental displacement measurements.
due to cracking might not have been included when calculating the Alternatively, measurements can be taken of the top and bottom
curvature. However, these measurements were not required to cal- fiber independently during different tests under the same loading
culate the maximum displacements and so only the left side of the to calculate the location of the neutral axis and allow for approx-
imately 19 m long beams to be monitored using the bottom fiber
beam deflection diagram is impacted.
and the neutral axis location to calculate the curvature. In addition,
The general deflected shape behavior under static loading ap-
the maximum data acquisition rate of 50 Hz for the 20 m sensing
pears to be as expected. Although the calculated deflected shape
length restricts the maximum frequency of loading to the 25 Hz
does not cover the entire span, it appears that the deflection at each
Nyquist frequency to prevent aliasing, which was not an issue for
support is zero, with a maximum deflection located at the midspan.
the tests conducted here but could be for some dynamic events such
This behavior, however, is not seen within the deflected shape
as blast loading. Finally, for existing structures where the fiber-optic
under the maximum dynamic load measured during the UDL and
sensors have to be mounted on the concrete surface, the internal
PL tests. In this case, the right-hand support (i.e., column) displace-
reinforcement strains cannot be measured directly although they
ment does remain close to zero, but the deflection toward the left-
can be measured in new structures allowing for dynamic reinforce-
hand support (i.e., the beam) is not close to zero, with a measured ment strains to be measured (Broth and Hoult 2020).
deflection of 0.28 and 0.58 mm at the leftmost point for the UDL
and PL tests, respectively. Although deflections were not directly
measured at the beam-beam connection, the supporting beam may Conclusions
have experienced some deflection as a result of the dynamic loads.
However, it is also possible that this behavior is due to noise within This research program was conducted in order to establish a
the FOS data near the end of the sensing length along the top fiber. better understanding of the dynamic performance of existing RC

© ASCE 04020070-8 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


structures and improve on previous static assessments by utilizing structures in Barcelona.” Struct. Infrastruct. Eng. 14 (7): 967–985.
DDFOS. To test the abilities of DDFOS, a case study was con- https://doi.org/10.1080/15732479.2018.1438479.
ducted on an existing RC beam. The beam was tested in situ under Bentz, E. C., and N. A. Hoult. 2016. “Bridge model updating using dis-
static and dynamic loads and monitored with discrete sensors and a tributed sensor data.” Proc. Inst. Civ. Eng. Bridge Eng. 170 (1): 74–86.
DDFOS. Based on the results of the study, the following conclu- https://doi.org/10.1680/jbren.15.00030.
Brault, A., N. Hoult, T. Greenough, and I. Trudeau. 2018. “Monitoring
sions can be made:
of beams in an RC building during a load test using distributed sensors.”
• By utilizing the ODiSI-B DDFOS system, both static and dy-
J. Perform. Constr. Facil. 33 (1): 04018096. https://doi.org/10.1061
namic concrete strains could be successfully measured during /(ASCE)CF.1943-5509.0001250.
dynamic loading tests on an existing RC beam in situ. Brault, A., and N. A. Hoult. 2019a. “Distributed reinforcement strains:
• The recorded DDFOS data enabled critical performance metrics Measurement and application.” ACI Struct. J. 116 (4): 115–127.
to be measured within the instrumented beams including maxi- Brault, A., and N. A. Hoult. 2019b. “Monitoring reinforced concrete serv-
mum strains, support conditions, and cracking behavior. iceability performance using fiber-optic sensors.” ACI Struct. J. 116 (1):
• The observed trends in the data from both the discrete and dis- 57–70.
tributed sensors were in good agreement, suggesting that the Broth, Z., and N. A. Hoult. 2020. “Dynamic distributed strain sensing to
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

DDFOS was able to accurately capture the true behavior of assess reinforced concrete behaviour.” Eng. Struct. 204 (Feb): 110036.
the beam. https://doi.org/10.1016/j.engstruct.2019.110036.
• DDFOS data provided the ability to determine the dynamic re- Chang, P. C., A. Flatau, and S. C. Liu. 2003. “Review paper: Health
sponse factor of the instrumented structure, which was found to monitoring of civil infrastructure.” Struct. Health Monit. Int. J. 2 (3):
be lower than the deflection-based DRF results. Both are con- 257–267. https://doi.org/10.1177/1475921703036169.
Davis, C., M. Knowles, N. Rajic, and G. Swanton. 2016. “Evaluation of a
sidered valid DRF values, as similar comparisons are found in
distributed fibre optic strain sensing system for full-scale fatigue test-
previous studies.
ing.” Procedia Struct. Integrity 2: 3784–3791. https://doi.org/10.1016/j
• Distributed crack widths and deflected shapes, previously lim- .prostr.2016.06.471.
ited to static measurements, could be determined dynamically Davis, M. B., N. A. Hoult, S. Bajaj, and E. C. Bentz. 2017. “Distributed
utilizing the recorded strain data from the DDFOS system, sensing for shrinkage and tension stiffening measurement.” ACI
including the identification of the maximum results due to Struct. J. 114 (3): 753–764.
the applied dynamic load. Gastineau, A., T. Johnson, and A. Schultz. 2009. Bridge health monitoring
• In the future, the data from these experiments can be used in and inspections: A survey of methods. Rep. No. MN/RC 2009-29.
conjunction with a finite element model to undertake model up- St. Paul, MN: Minnesota Dept. of Transportation.
dating to develop more accurate approaches to design. However, Gebremichael, Y. M., et al. 2005. “A field deployable, multiplexed Bragg
it should be noted that data from multiple structures will be grating sensor system used in an extensive highway bridge monitoring
required to ensure the reliability and accuracy of these finite evaluation tests.” IEEE Sens. J. 5 (3): 510–519. https://doi.org/10.1109
element models for design. In addition, structures designed us- /JSEN.2005.846185.
ing these approaches should have sensors installed to confirm Highway Research Board. 1962. Special report 61D: The AASHO road
test. Rep. No. 4: Bridge Research. Publication 953. Washington,
the accuracy of these models.
DC: National Academy of Sciences, National Research Council.
Hoult, N., M. Dutton, A. Hoag, and W. A. Take. 2016. “Measuring crack
movement in reinforced concrete using digital image correlation: Over-
Data Availability Statement view and application to shear slip measurements.” Proc. IEEE 104 (8):
1561–1574. https://doi.org/10.1109/JPROC.2016.2535157.
Some or all data, models, or code generated or used during Kreger, S. T., D. K. Gifford, M. E. Froggatt, A. K. Sang, R. G. Duncan,
the study are available from the corresponding author by request M. S. Wolfe, and B. J. Soller. 2007. “High-resolution extended distance
(i.e., raw fiber-optic strain data and LP data). distributed fiber-optic sensing using rayleigh backscatter.” In Proc.,
Sensor Systems and Networks: Phenomena, Technology, and Applica-
tions for NDE and Health Monitoring. Bellingham, WA: SPIE.
Kreger, S. T., A. K. Sang, N. Garg, and J. Michel. 2013. “High resolution,
Acknowledgments
high sensitivity, dynamic distributed structural monitoring using optical
frequency domain reflectometry.” In Proc., Fiber Optic Sensors and
The authors would like to acknowledge the Natural Sciences
Applications X. Bellingham, WA: SPIE.
and Engineering Research Council of Canada for their financial
Kurashima, T., T. Horiguchi, and M. Tateda. 1990. “Distributed-
support. The authors would also like to thank Sara Nurmi, Eric temperature sensing using stimulated Brillouin scattering in optical
Pannese, Paul Thrasher, and Andre Brault from Queen’s University. silica fibers.” Opt. Lett. 15 (18): 1038–1040. https://doi.org/10.1364/OL
Finally, the authors would like to thank all of the volunteers who .15.001038.
were involved in the experiments. Lee, J. J., and M. Shinozuka. 2006. “A vision-based system for remote sens-
ing of bridge displacement.” NDT & E Int. 39 (5): 425–431. https://doi
.org/10.1016/j.ndteint.2005.12.003.
References LUNA Technologies. 2017. ODiSI-B optical distributed sensor interroga-
tor: User’s guide. Blacksburg, VA: Luna Innovations Incorporated.
ACI (American Concrete Institute). 2001. Control of cracking in concrete Majumder, M., T. K. Gangopadhyay, A. K. Chakraborty, K. Dasgupta, and
structures. ACI 224R. Farmington Hills, MI: ACI. D. K. Bhattacharya. 2008. “Fibre Bragg gratings in structural health
Bakht, B., and S. G. Pinjarkar. 1989. “Dynamic testing of highway monitoring—Present status and applications.” Sens. Actuators, A 147 (1):
bridges—A review.” Transp. Res. Rec. 1123: 93–100. 150–164. https://doi.org/10.1016/j.sna.2008.04.008.
Barrias, A., J. R. Casas, and S. Villalba. 2018a. “Fatigue testing of rein- Middleton, C. R. 1997. “Concrete bridge assessment: An alternative
forced concrete beam instrumented with distributed optical fiber sensors approach.” Struct. Eng. 75 (23/24): 403–409.
(DOFS).” In Proc., Civil Engineering Research in Ireland. Dublin, Mohamad, H., K. Soga, A. Pellew, and P. J. Bennett. 2011. “Performance
Ireland: Civil Engineering Research Association of Ireland. monitoring of a secant-piled wall using distributed fiber optic strain
Barrias, A., G. Rodriguez, J. R. Casas, and S. Villalba. 2018b. “Application sensing.” J. Geotech. Geoenviron. Eng. 137 (12): 1236–1243.
of distributed optical fiber sensors for the health monitoring of two real https://doi.org/10.1061/(ASCE)GT.1943-5606.0000543.

© ASCE 04020070-9 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070


Ohno, H., H. Naruse, M. Kihara, and A. Shimada. 2001. “Industrial appli- with fiber optic sensors.” Smart Mater. Struct. 10 (3): 560–573.
cations of the BOTDR optical.” Opt. Fiber Technol. 7 (1): 45–64. https://doi.org/10.1088/0964-1726/10/3/320.
https://doi.org/10.1006/ofte.2000.0344. Van Der Kooi, K., N. A. Hoult, and H. Le. 2018. “Monitoring an in-service
Orr, J. J., T. J. Ibell, A. P. Darby, and M. Evernden. 2014. “Shear behaviour railway bridge with a distributed fiber optic strain sensing system.”
of non-prismatic steel reinforced concrete beams.” Eng. Struct. 71 (Jul): J. Bridge Eng. 23 (10): 05018007. https://doi.org/10.1061/(ASCE)BE
48–59. https://doi.org/10.1016/j.engstruct.2014.04.016. .1943-5592.0001281.
Regier, R., and N. Hoult. 2014. “Distributed strain behavior of a reinforced Vecchio, F. J., and M. P. Collins. 1986. “The modified compression-field
concrete bridge: Case study.” J. Bridge Eng. 19 (12): 05014007. https:// theory for reinforced concrete elements subjected to shear.” ACI J.
doi.org/10.1061/(ASCE)BE.1943-5592.0000637. 83 (2): 219–231.
Salawu, O. S., and C. Williams. 1995. “Review of full-scale dynamic test- Wheeler, L. N., E. Pannese, N. A. Hoult, W. A. Take, and H. Le. 2018.
ing of bridge structures.” Eng. Struct. 17 (2): 113–121. https://doi.org “Measurement of distributed dynamic rail strains using a rayleigh back-
/10.1016/0141-0296(95)92642-L. scatter based fiber optic sensor: Lab and field evaluation.” Transp. Geo-
Spencer, B. F., M. E. Ruiz-Sandoval, and N. Kurata. 2004. “Smart sensing tech. 14 (Mar): 70–80. https://doi.org/10.1016/j.trgeo.2017.10.002.
technology: Opportunities and challenges.” Struct. Control Health Wong, L., W. K. Chiu, and J. Kodikara. 2018. “Using distributed optical
Monit. 11 (4): 349–368. https://doi.org/10.1002/stc.48. fibre sensor to enhance structural health monitoring of a pipeline sub-
Downloaded from ascelibrary.org by Heriot-Watt University on 10/09/21. Copyright ASCE. For personal use only; all rights reserved.

Tennyson, R. C., A. A. Mufti, S. Rizkalla, G. Tadros, and B. Benmokrane. jected to hydraulic transient excitation.” Struct. Health Monit. 17 (2):
2001. “Structural health monitoring of innovative bridges in Canada 298–312. https://doi.org/10.1177/1475921717691036.

© ASCE 04020070-10 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(4): 04020070

You might also like