In the prestessed beam there are upper surface tensile stresses (ó_pt) and lower surface compressive stresses (ó_pt) caused by prestress forces (shown in Figure 1(a)). At the in-service stage, the beam stress caused by service loads is inversely related with that due to prestress force. The tensile stress (ó_qt) increases with service load. During the initial stage, e.g., the value of tensile tress (ó_qt) is equivalent to that of compressive stress (ó_pt), the stress of lower surface in prestressed beam is zero, and then it becomes tensile stress. When the tensile stress of the lower surface reaches the ultimate tensile strength of concrete (f_ct), the first crack appears, defined as the initial cracking stage (shown in Figure 1(b)). In such avstate, the beam has cracks (one or two) with a width of approximately 0.05 mm. Because of the elastic mechanical properties of beams, the cracks will be completely closed and the stress will recover to the no-load state, so the prestress losses can be neglected in the initial cracking state.

With increasing service load a series of vertical cracks will appear on the surface of the beam, while the width of existing cracks increases. When the crack width reaches the maximum allowable value (the tensile stress of the lower surface is equivalent to the allowed stress (ó_ct) corresponding to crack width limits), the prestressed concrete beam is under the limit state, defined as the normal service limit stage (shown as Figure 1(c)). Considering that the prestress tendons are the main load-bearing components and brittle materials with high elastic stress segments, the prestressed beam displays elastic mechanical behavior in the normal service limit stage, so when the service load is unloaded, most of the existing cracks will be closed, but because of concrete chipping off nearby the cracks, the prestress loss that will result from that the beam becomes shorter. The prestress losses are non-uniformly distributed along the beam and increase as more cracks occur.

Optical fiber sensors have been developed for a number of years and many optical fiber-based sensing techniques have been established for structural health monitoring because of their advantages such as immunity to electrical noise, long-term measurement stability and resistance to corrosion. In this study, the smart steel strand is a steel strand with sensing capability for structural condition assessment using BOTDA sensors. The optical fiber (OF) sensor was embedded into a fiber reinforced plastic (FRP) rebar of 5 mm in diameter, named as FRP-OF rebar, to enable its sensing capability. The smart FRP rebar was covered with copper foils and then embedded in the middle of a steel strand, which consisted of six common steel wires around (shown in Figure 2). As the FRP rebar deforms together with the remaining six steel wires, the deformation of the steel strand can be directly measured by the optical fiber sensors embedded in the rebar.

Instead of a normal strand, the smart steel strand was anchored in a prestressed concrete structure. Then the prestress loss of the strand can be monitored by optical fiber sensors in the smart steel strand using BOTDA technology. The principle diagram of fiber optic BOTDA technology is shown as Figure 3.

When an optical pulse is launched into the optical fiber in the smart steel strand, some backscattered signals return to the input end. There are three main types of scattering, and Brillouin scattering is one of them. The pumping pulse light is launched at one end of the fiber and propagates in the fiber, while the continuous wave (CW) light is launched at the opposite end of the fiber and propagates in the opposite direction. When the power of optical pulse signal, which propagates along the single-mode optical fiber, is larger than the Brillouin threshold power, the backward stimulated Brillouin scattering signal is generated. Stimulated Brillouin scattering signal can be described as a parametric interaction among the incident light, the Stokes light, and an acoustic wave. The Brillouin frequency shift ν_B of the backward scattering light of the propagating light in an optical fiber is given by [33] as bellow: (1) v B = 2 n v A λ pwhere ν_A is the velocity of light wave propagating in vacuum, n is the effective refractive index of the optical sensing fiber and λ_p is the wavelength of the input optical pulse.

When a strain (ε) is applied or the temperature (T) is varied on the fiber, this Brillouin frequency shift v_B changes linearly with the applied strain and temperature differences [34,35]. It is expressed as: (2) v B ( ɛ , T ) = v B 0 + C ɛ ɛ + C T T

And for a temperature compensating sensor without applied strains, its frequency shift v_BT (0, T) only depends on the temperature differences. Thus: (3) v B T ( 0 , T ) = v B 0 + C T T

Then the applied strains on the optical sensing fiber yields to: (4) ɛ = Δ v B ( ɛ ) C ɛ = v B ( ɛ , T ) - v B T ( 0 , T ) C ɛwhere Δv_B(ε) is Brillouin frequency shift with strain; v_B(0,T) is Brillouin frequency shift without strain and about 11 GHz; C_ε is the sensing coefficient of strain and about 0.5 GHz/% (strain) at the wavelength k = 1.55 μm. According to Hooke's law, the stress (ó) of the stand can be shown as: (5) σ = E Δ v B / C ɛwhere E is the equivalent Young's module of the smart strand.

Finally, the prestress loss (ó_l) of the strand can be expressed as: (6) v ( ɛ , T ) = v ( 0 , 0 ) + C ɛ ɛ + C T Twhere ó_con is the controlling stress of the strand.

Figure 4 shows the manufacturing process of the proposed smart steel strands. To protect the FRP-OF sensor and increase the bonding between the smart rebar and the steel bars, copper coils were fixed on the FRP-OF rebar using No. 502 glue. For each cross section of the rebar, two or three layers of copper coils were clipped. The common steel strands were then cut to a certain length according to the design. The six steel wires are separated clockwise or anticlockwise, and the sequences of the wires are remembered. The smart steel strand was finally completed by winding the six steel strands with the FRP-OF rebar in turn with the contrary sequence.

To characterize the sensing properties of the proposed smart steel strand, one smart steel strand was manufactured following the procedure described in the last section and tested in the laboratory for calibration. Figure 5 shows the experimental setup for the calibration test. The test specimen had a length of 3 m and was fixed on a reaction frame. The hydraulic jack used in calibration test has a limited loading range, and the purpose of the calibration test is to investigate the sensing performance of the smart steel strand, so the load used was 30 kN, divided into nine loading steps of 3.33 kN for each step. Three loading cycles were repeated. The strain of strand (measured by a DiTeSt STA2000 apparatus produced by Omnisens, Switzerland) was recorded.

The smart steel strands are sufficient and easy to install. Figure 6(a,b) shows the calibration results of the smart steel strand obtained from the BOTDA sensors. The full-scale strain distribution of the strands can be obtained by the BOTDA sensors. The BOTDA sensor in the smart steel strands has a good linearity, with linear coefficients of 99.948%. The load sensitivity of the BOTDA sensor yields 38.824 με/kN. The resolution along the length of the fiber was 10 cm. Thus, the proposed FRP-OF smart steel strands gives promising results and could be applied for further structural property investigation of the steel strands.

For static performance indexes not only showing sensing properties of the sensors, but also affecting dynamic indexes, it is particularly important to analyze the static characteristics. Main static indexes such as hysteresis, linearity, repeatability and overall accuracy etc. are used to describe the application of a sensor under actual conditions and evaluate the merits of sensors. According to the Chinese National Standard “Methods for calculating the main static performance specification of transducers” (GB/T 18459-2001), the main static performance indexes of the smart strand was computed using the calibration test data. Table 1 shows the main static performance indexes of the smart steel strand, which are linearity of 3.9% FS (full-scale), hysteresis of 1.3% FS, repeatability of 1.7% FS and overall accuracy of 3.3% FS.

Two unbounded prestessed concrete beams were designed and cast with the same dimensions and materials. The concrete beams (compressive strength is 26.8 MPa) tested in this series of experiments had a span of 3 m and a cross-section of 100 mm × 200 mm. One proposed smart steel strand was implemented as prestressed reinforcement in the concrete beam. The smart strand had a diameter of 15.12 mm and a standard strength (f_ptk) of 1,660 MPa. Four common steel reinforcements (yielding strength is 310 MPa) with a diameter of 10 mm were distributed in the tension region and compression area (shown in Figure 7). Steel plates and spiral reinforcements were embedded in the tension and anchoring region of the beam to eliminate the stress concentration. The thickness of the steel plate was 10 mm, and the spiral reinforcement was 4 mm in diameter, 50 mm in spiral inner distance. The concrete beams were cured for 28 days in room temperature after casting.

In the fiber reinforced plastic (FRP) optical fiber (OF) rebar, one optical fiber sensor had been applied in the full-length of the beam. To ensure the stability of the sensor, there were no weaknesses (solder joints, etc.) on the optical fiber sensors. To validate the smart steel strand, a load cell was installed at the tension end for comparison, as shown in Figure 8.

Figure 9 shows the test setup. The smart strand was tensioned by a hydraulic jack and anchored by a single-hole anchorage. The control stress applied in this test was 0.70f_ptk, resulting in a max tension load of 160 kN. After 20 days, two concentrated loads provided by the hydraulic jack were applied to a location 500 mm from the center of the loaded beam. At same time, the other beam (unloaded beam) did not bear any loads. The pressure sensor was implemented for the loading control.

Here, two damage conditions (initial cracking and normal service limit state) of prestressed beam were investigated. To analyze time-dependent prestress loss of the concrete beam at the stage of first crack occurrence, all sensors started to record data at the stage of crack initiation and unloading to zero. Data were also recorded by the load cell and the smart steel strand as the limiting crack width then unloading to zero for analyzing time-dependent prestress loss of the concrete beam in the normal service limit state.

Figure 10 shows the strain results during the tensioning operation as monitored by the BOTDA sensors. In order to save channel numbers of Brillouin demodulator and test time, the optical fiber sensors of loaded and unloaded beams were connected in the data acquisition process. From Figure 10(a,c), the test length of strands in the loaded beam (x axis 5–8 m) and unloaded beam (x axis 13–16 m) can be clearly distinguished. From Figure 10(b,d), the strain measured by the optical fiber sensors in the two beams increased linearly with increasing loads, and the correlation coefficient is 99.8%. The sensitivity coefficient of optical fiber sensors in two beams is approximate, and the sensitivity coefficients in loaded beam and unloaded beam are 40.6 με/kN and 37.6 με/kN, respectively.

A Figure 11 shows the 3-D time-dependent prestress losses of the beam measured by the BOTDA sensors in the smart steel strand at the stage of first crack occurrence. The full-scale strain (location 5–7 m) distribution of the strand can be obtained by the BOTDA sensors. The minor damages of the beam do not change the prestress loss significantly, except for a single catastrophe point (the peak near the location of 6 m and at the time of 7,000 min) due to the boundary effect of the BOTDA testing apparatus.

Figure 12 shows the time-dependent prestress loss results at the mid-span cross section of the beam measured by the BOTDA sensors and the load cell at the stage of the first crack occurrence. It shows that the stress measured by the BOTDA sensor agrees well with that determined by the load cell. With a lower resolution, the results monitored by BOTDA sensor fluctuate along that measured from the load cell. The time-dependent prestress loss of the beam with small damages (cracks) fluctuates slightly with time, indicating that minor damages are negligible for prestress loss.

Figure 13 shows photos of the cracks distribution and the measurement of crack width. When the load was 10 kN, the first vertical crack appeared on the left loading position of the prestressed concrete beam. With the increasing loads, the width and number of vertical cracks increased, then the diagonal cracks appeared on two loading positions, one after another. When the load was 50 kN, the maximum crack width of lower edge of prestressed concrete beam reached 0.2 mm.

Figure 14 shows the 3-D time-dependent prestress loss of the loaded beam measured by the BOTDA sensors in the smart steel strand at the normal service limit state. The full-scale strain (location 5.5–7.5 m) distribution of strand can be obtained by the BOTDA sensors. The change of prestress loss is highly related to the distribution of cracks. The prestress losses are not uniformly distributed along the beam, with the locations of 6 m and 7 m being larger than other locations. This is because the loads were directly placed at these locations and more cracks were observed in this region. The prestress losses then are indicated to be increased as more cracks occur.

Figure 15 shows the test results of time-dependent prestress loss at the mid-span cross section of two beams measured by the BOTDA sensors after the occurrence of the maximum allowable crack width (0.2 mm). The prestress losses of the loaded beam monitored by the BOTDA sensors increases with time, during the initial stage, e.g., in the first day, a higher prestress loss rate is observed, and that of the unloaded beam do not change with time.