|NDT.net June 2006 Vol. 11 No.6|
Distributed Brillouin Sensor Based on Brillouin Scattering for Structural Health MonitoringXiaoyi Bao (firstname.lastname@example.org), Fabien Ravet, Lufan Zou (email@example.com)
Department of Physics, University of Ottawa, Ottawa, ON K1N 6N5, Canada
AbstractWe used the Distributed Brillouin Sensor (DBS) to monitor the structure changes in a steel pipe and a concete/FRP (Fiber Reinforced Polymer) column subjected to heavy loads. We have shown that the DBS is capable of providing strain distribution of both column and pipe under the various loads at structural and local level for the spatial resolution of 10-15cm. The DBS can detect deformations, cracks and buckling before they became fatal to the structure. This demonstrates that it is a promising tool for Structural Health Monitoring (SHM).
1. IntroductionSHM is used to identify early signs of problems, allowing prevention of disasters and then the repair of these damages. It is also used to provide guidelines for new building materials, reducing the need for repair over the structures lifetime. Currently, sensors used for SHM are punctual devices that give only partial information about the stresses affected to the structure. Their localized nature gives incomplete information about the structure health. They fail to locate defects in the early stage, such as cracks or buckling, which requires cm spatial resolution over large structural coverage. There is a need for a technique that detects faults and assesses the severity of the damage of the whole structure. Such sensor must perform distributed temperature and strain measurements in real time over tens meters to kilometers. The DBS has the advantage to combine all these characteristics . Recent studies have been conducted to monitor strain of composite and concrete beams under limited load, i.e. the structure responds to the load linearly ,. Here we demonstrate that DBS is able to fulfill SHM task on structures subjected to high loads such that the strain is not elastic anymore. Therefore it leads to the formation of localized defects.
Two study cases emphasize DBS capabilities: 1) The buckling monitoring on natural gas transit steel pipe. The pipe was subjected to axial and bending loads in order to provoke buckling, which was successfully anticipated by our DBS. This experiment is collaboration between the University of Ottawa (Physics Department), C-FER Technologies and Trans-Canada Pipeline Limited . 2) To monitor a concrete column strengthened with Fiber Reinforced Polymer rods and sheets. The column was subjected to an axial load while successively bended back- and forward with increasing loads. We monitored the structure deformation but also correlated the readings with the applied stresses and evidenced the possible de-bonding of the FRP and concrete as well as FRP ruptures. This experiment is in collaboration with Civil Departments Engineering of the University of Ottawa .
2. Distributed Brillouin SensorThe sensing medium is an SMF-28 optical fiber. Two beams are launched into the fiber in a counterpropagating configuration. The two beams interact when the frequency difference of the two lasers matches the frequency of the acoustic phonons of the optical fiber. Energy is transferred from the high frequency beam (pump) to the low frequency beam (probe). The amount of the loss of the pump is recorded at one end of the fiber as a function of the frequency difference in the form of Brillouin loss spectrum (BLS). The maximum loss occurs at the Brillouin frequency (Vpk), which is the phonon frequency. The distributed sensor can record the BLS at different location along the fiber because the probe is a pulsed created with an electro-optic modulator (EOM). The spatial resolution (w) is determined by the pulsewidth (Dt), which is 1.5ns equivalent to 15cm, with relation of w=vgDt/2 where vg is the pulse group velocity in the optical fiber. Measurement is taken every 5cm. Brillouin frequency varies with strain when temperature remains unchanged as epk=(Vpk-V0B)/Ce where epk is the tensile strain, Vpk is the strained fiber Brillouin frequency, rob is the unstrained fiber Brillouin frequency of the and Ce is the strain coefficient (Ce=0.05941MHz/µe) .
Common DBS are limited in frequency resolution when Dt < tB due to spectral broadening (tB is the phonon lifetime). Our sensor has a unique feature to overcome that limitation. In fact, the broadening is mitigated by coherent probe (pulse)- pump (continuous wave) interaction made possible by the finite extinction ratio (ER) of the EOM . BLS is enhanced and narrower than spectra produced by pulse-pump interaction only (infinite ER case). We used an extinction ratio of 15dB, which keeps the BLS linewidth within a few percent of the Brillouin natural linewidth, achieving simultaneous high spatial and frequency resolutions.
3. Pipeline buckling monitoringThe tested specimen is a sample of natural gas transiting steel pipe with a length of 2.58 m and diameter of 0.75 m, as shown in Fig.la. The pipe was sealed on both ends. To simulate natural gas transit, the inter pressure was kept at 18.4 MPa during pipe-wall buckling test. Longitudinal directions of 6 and 12 o'clock along the pipe were subjected to tension and compression, respectively, by applying simultaneous vertical and lateral forces. The monitoring fiber was composed of 10 sections located from 6 to 12 o'clock around the pipe and glued on the external, as shown in Fig. lb. Each section is separated by a 1 m of loose fiber. Strain profiles along the 6 o'clock section of the pipe (tension side) under vertical and bending loads conditions are presented in Fig.2a and Fig.2b. The biggest tensile strain happens around 140 cm. Compression side (12 o'clock section) results are shown in Fig.2c and 2d. It shows the difference of Vpk under the same vertical and bending load conditions. Obviously 6 and 12 o'clock sections experienced tension and compression (associated with a negative Brillouin frequency shift difference). When vertical and bending loads were further increased, sections of the fiber located on the pipe compressive part were eventually broken and interrupted the measurements. However according to Fig.2, we predicted that a localized wall buckling would happen around 140 cm, and it did, after visual inspection of the pipe .
4. Concrete/FRP column monitoringA concrete column strengthened with an FRP casing, was tested under simulated seismic loading (Fig.3a). The optical fiber is glued horizontally at 10 distinct cross-sections of the column (from bottom, layer 1, to the top, layer 10) as shown in Fig.3b. Each section is 1 m long and separated by lm of loose fiber. The column was tested under an axial load of 1880 kN and increased lateral loading. Pull face corresponds to the column side facing the observatory. The opposite face is associated with the push action. The push and pull amplitudes are quantified by the drift, defined as the ratio of the lateral displacement (dN) to the height of the column (h), varied from 3 to 8% as shown in Fig.3.c. Measurements with the DBS are taken at each drift step. Strain gauges were also placed on the surface of FRP casings.
Measured axial strain profiles (epk) are presented in Fig. 3d and Fig. 3e. Large strains are concentrated at the bottom of the column (layers 1 to 5) with a peak value at layer 2 confirmed by structural engineering theory predictions. The most extensive damage occurred at approximately 100 mm to 160 mm above the column-footing interface, which coincided with the location of first FRP rupture in all columns. The move of the critical section from the interface is attributed to the confining effect of the footing . The progressive column degradation is analyzed with strain, both epk and eSG (strain gauge reading), vs drift plots (Fig.4.). epk and eSG experience a monotonic increase with a bend in the curves. The region of smaller epk slope corresponds to elastic condition without deformation. The damage in concrete is not significant and it is still capable of resisting applied loads, without a significant contribution from FRP casing. The change in the slope is associated with the appearance of local stresses indicating the start of local crushing of concrete at column critical region. Concrete crushing was evidenced in post-mortem analysis of the column. The bend in epk is followed by a slope rise, which is due to the extensive damage in concrete. At this point, FRP casing and undamaged concrete are no longer in contact (FRP debonding). The column requires a significant contribution from FRP casing to maintain its integrity. Load is transmitted to the FRP through the crushed concrete now capable to follow any load variation. At large drifts (>7%), epk continues to increase threatening the column safety. The FRP, the only supporting element of the structure, is strained significantly and then started to rupture locally. These local ruptures, visually observed on the column, release the tension inducing a local strain reduction as illustrated in Fig.4c where eSG drops, which affects the spectrum symmetry .
5. ConclusionsWe conducted two structure deformation experiments on a concrete column reinforced by FRP was monitored with the DBS during a trial reproducing earthquake like conditions, and on pipe buckling process. The DBS also provided detailed information on the structure health at a local and global level. The DBS allows the detection of deformations, cracks and buckling. The DBS is the only sensor offering global and local strain measurements, demonstrating that it is a powerful candidate to fulfill SHM tasks.