Distributed optical fiber sensors (DOFS) is one of the most promising and exciting technologies under research to be applied in the structural health monitoring (SHM) of civil engineering infrastructures. Therefore, in this paper, the authors present a laboratory experiment where a reinforced concrete beam was instrumented with a 5-meter-long polyimide DOFS in a way that four equal segments were bonded to the bottom surface of the beam using for each segment a different type of adhesive. Three strain gauges were also used to compare the results. The beam was then loaded, generating expected equal levels of strain for each of the segments allowing for a direct comparison between them. In this exercise, additionally to the comparison with the other instrumented sensors, it is also important the consideration and analysis of the associated Spectral Shift Quality (SSQ) values of the DOFS measurements.

1. Reducing Uncertainty in Structural Safety
Special Session SS6
Ghent, Belgium
28-31 October 2018

2. António Barrias, Joan R. Casas and Sergi Villalba
On the bonding performance of
distributed optical fiber sensors (DOFS) in
structural concrete

3. ▪ Introduction – Distributed Optical Fiber Sensors (DOFS)
▪ Motivation
▪ Experimental test
▪ Test results and discussion
▪ Conclusion
Outline
2

4. Introduction – Optical Fiber Sensors
Cylindrical symmetric structure that is composed by a central
“core” with a diameter between 4 and 600 µm and a uniform
refractive index
Advantages:
▪ Immunity from electromagnetic interferences
▪ Small size and lightweight
▪ High sensitivity
▪ Withstand high temperatures
▪ Chemically inert – free from corrosion
4

5. Introduction – Optical Fiber Sensors
All possible crack openings are covered by the extension of the sensor
5

6. Optical Backscatter Reflectometry - OBR
1. OBR system measures the
Rayleigh backscatter as a function
of length in an optical fiber with
high spatial resolution (1mm)
2. External stimulus
3. Temporal and spectral shifts in the
local backscattered pattern
4. Shifts can be measured and scaled
to give distributed strain or temperature
measurements
Swept wavelength interferometry
(fiber is divided in small windows)
▪ Sensing range – up to 70 m
▪ Spatial resolution – up to 1 mm
▪ Strain resolution – ± 2 µε
▪ Measurement range – ± 13000 µε
6

7. Motivation
7
▪ Lack of knowledge is still present regarding the choice of the optimal bonding adhesive,
especially in its use when deploying these sensors in reinforced concrete structures
▪ This is even a more critical issue when applying DOFS without any protective thick coating, such
as the case of the deployed fiber in this research
▪ An additional point of interest in this study was the assessment of the influence of different spatial
resolution used in the DOFS system

8. Experimental Test
8
• Reinforced concrete beam with 60x15x15
[cm] dimensions;
• 5,2m polyimide DOFS was instrumented
to the beam
• Four different adhesives were used:
Silicone, Polyester, Epoxy and
Cyanoacrylate.

9. Experimental Test
9
View of loading arrangement
Test cycle
number
Spatial
Resolution
Sampling Rate
1 1 cm
0.2 Hz
2 3 cm
3 1 mm
4-rupture 1 cm
It was decided to initially perform three
separate but identical load cycles where a
different spatial resolution was used;
A final load was applied inducing cracking and
continuing until failure of either the specimen or
the DOFS in test cycle 4.

10. 10
Applied load in test cycles 1 to 3 Applied load in test cycle 4
Properties
𝑓𝑐𝑚 𝑓𝑐𝑡𝑚 𝐸𝑐 𝜀 𝑐𝑡
[MPa] [MPa] [MPa] [µε]
Specimen 48.027 3.944 37886.64 104.10
Experimental Test

11. 11
Test results and discussion - Comparison of spatial resolution input

12. 12
Test results and discussion - Comparison of spatial resolution input
Spatial
Resolution
Comparison
Statistical
Properties
Segments
average
Δ(1mm-3cm)
mean (µ) -1.07
std (σ) 19.19
Δ(1mm-1cm)
mean (µ) -1.23
std (σ) 19.34
Δ(1cm-3cm)
mean (µ) 0.29
std (σ) 3.38

13. 13
Test results and discussion - Comparison of spatial resolution input

14. 14
Test results and discussion - Comparison of spatial resolution input
Spatial
Resolution
Comparison
Statistical
Properties
Segments
average
Δ(MA_1mm-
3cm)
mean (µ) -1.05
std (σ) 4,74
Δ(MA_1cm-
3cm)
mean (µ) 0.22
std (σ) 2.31
Δ(1cm-3cm)
mean (µ) 0.29
std (σ) 3.38

15. 15
Test results and discussion - Comparison of bonding adhesives under elastic behavior

16. 16
Testresultsand discussion- Comparisonof bondingadhesives underelasticbehavior

17. 17
Testresultsand discussion- Comparisonof bondingadhesives underelasticbehavior

18. 18
Testresultsand discussion- Comparisonof bondingadhesives wheninducingcracking
▪ Cracking initiated at 24 kN load level;
▪ All bonded segments detect and locate the
developed crack ;
▪ All segments, except the silicone bonded one,
vary its readings between positive and
negative values after cracking at its location;
▪ Fiber broke completely at minute 62 which
corresponded to a load of 58.2 kN but in
practical terms the data from the fiber was
deemed unusable from minute 50 (Load of 31
kN)

19. 19
Testresultsand discussion- Comparisonof bondingadhesives wheninducingcracking

20. 20
Testresultsand discussion- Comparisonof bondingadhesives wheninducingcracking
Spectral Shift Quality - SSQ
Spectral Shift Quality =
max 𝑈𝑗 𝜐 𝑈𝑗 𝜐 − Δ𝜐𝑗
Σ𝑈𝑗 𝜐 2
▪ 𝑈𝑗(ν) is the baseline spectrum for a given segment of data;
▪ 𝑈𝑗(𝜈 − Δ𝜈𝑗) is the measurement spectrum under a strain or temperature change;
▪ symbol is used to represent the cross-correlation operator.
SSQ is a qualitative measure of the strength of the correlation between the conducted
measurement (at any point and time) and the original baseline reflected spectra.
The value of the SSQ should theoretically be between 0 and 1, where 1 is obtained when a
perfect correlation is achieved and 0 when it is uncorrelated.
The manufacturer advises to disregard any measurements with a SSQ below 0.15.

21. 21
Testresultsand discussion- Comparisonof bondingadhesives wheninducingcracking

22. 22
Testresultsand discussion- Comparisonof bondingadhesives wheninducingcracking
▪ Remove strain data with
corresponding SSQ below 0.15;
▪ Remove strain data outside of
acquisition system
measurement range ±13000 µε;
▪ Remove data with negative
strain data (all segments were
under tension);
▪ Interpolate surface with
remaining data.

23. 22
Conclusions
▪ The results show how the use of a spatial resolution of 1 mm presented an undesired high
spatial variability. This does not occur with a spatial resolution of 1 cm, which presented a
good correlation with the data measured by the strain gauges. Therefore, the use of a spatial
resolution of 1cm is deemed suitable and recommended for this type of applications
▪ It was observed how in un-cracked condition, all segments compared fairly well with the data
measured by strain gauges. Nevertheless, the silicone bonded segment presented a
smoother and more uniform data than the other bonded segments
▪ When the tensile capacity of the concrete was exceeded, it was verified how all the segments
were able to detect and locate the crack formation. Notwithstanding, the associated SSQ
values of the measurements quickly dropped below an acceptable threshold
▪ The use of silicone adhesive provides measurements, which seem to be less influenced by
the decrease of the associated SSQ values and therefore provide coherent information for
further stages of the load but presenting less spatially accurate measurements. The other
researched adhesives, especially the cyanoacrylate, provide more precise spatially data but
are limited earlier to the effect of the decrease of the SSQ values

24. The TRUSS ITN project (http://trussitn.eu) has received funding
from the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie grant agreement
No. 642453
Thanks for your attention