Open Access
Add to BibSonomyAbstract
This paper presents a fiber-optic quasi-distributed in-line twist/rotation sensor. The presented sensor consists of a polarization maintaining lead-in fiber and series of in-line fiber polarizers with integrated semi-reflective mirrors that are interconnected by a standard single mode fiber. Sections of interconnecting single mode define multiple twist/rotation sensitive sensor segments, which can be interrogated individually by a simple optical time domain reflectometer. The presented system exploits the inability of the standard single mode fiber to change the polarization state, i.e. E-field vector orientation of a wave propagating down twisted/rotated fiber. Thus, twisting of the fiber segments between consecutively located polarizers causes modulation of the transmitted optical power along each sensor segment, which is further determined by observation of back-reflected signals from semi-reflective mirrors integrated together with in-line polarizers. A quasi-distributed sensing system with nine sensing segments is demonstrated with twist/rotational resolution better than 0.3 degrees and low cross-talk. The entire system is built out of a small number of simple and efficient optoelectronic components.
© 2016 Optical Society of America
1. Introduction
On-line monitoring of civil, aerospace, industrial and other structures is gaining importance over the last two decades. Safety issues, new developments in active structure’s control, and reduction of down-time are the main drivers behind this development. Torsional twist/rotation is one of the important structure-related parameters that shall be monitored in many of these applications. For example, hindrances, such as torsional buckling, are present in high-rise buildings, where the effect is caused due to irregularity of the structural layout, eccentricity in mass distribution [1] or in turbulent winds [2, 3]. Torsional buckling also occurs in bridge constructions [4] and advanced composite structures/materials [5, 6]. Efforts have also been devoted to a research into the reduction of torsional vibration in rotary drilling operations i.e. oil and gas well drilling [7, 8]. Different methods for torsional structure monitoring have therefore been developed, for example, by using coherent radar [9, 10], a series of accelerometers, global positioning systems (GPS) [11–14], multiple laser displacement sensors [15], etc. On the other hand, Fiber Optic Sensors (FOS) have already proved in structural monitoring and oil and gas sector to be a robust and efficient sensing technology. FOS twist/rotation sensors were therefore investigated intensively recently. Currently there are several possible approaches that allow for twist/rotation FOS design. Among them are, for example, sensing systems that utilize measurements of circular birefringence [16–19] in twisted fibers, change of twisted fiber’s linear birefringence [20–30], systems that exploit different effects related to fiber Bragg gratings (FBG) [31–34], tilted FBGs [35–39] and long period FBGs [40–45], E-filed vector displacement in circularly symmetric fibers [46–48], multimode interference [49], and other systems based on specialty fibers and waveguides [50]. None of these systems are however suitable for multipoint or quasi-distributed twist/rotation sensing, which is often required in modern structural monitoring.
This paper presents a quasi-distributed twist/rotation sensor system. The proposed sensor system is constructed from individual sensing segments, where each segment, when exposed to torsion, has the ability to measure twist/rotation angle independently. The prosed approach is straightforward for practical implementation and can potentially yield a very cost-efficient sensing system.
2. Sensor design and operation
The proposed quasi-distributed sensor setup is depicted in Fig. 1. The sensor consists of a series of Micro-Optics Segments (MOS), which are interconnected by a standard Single-Mode Fiber (SMF), connected to the interrogation unit through a 3-dBm PM coupler and PMF lead-in fiber. The MOS consists of a linear polarizer and a semi-reflective mirror with about 5% reflectance Fig. 2(a). Both components are coupled to a standard Single-Mode Fiber through appropriate fiber-micro-collimators and packed in a standard optical-fiber micro-component coaxial ϕ3x60 mm case. The design and functionality of MOS are identical to a standard in-line micro-optics fiber polarizer, except that a semi-reflective mirror is added behind the linear polarizer in order to provide controlled partial back-reflection for the incoming radiation, which allows for sensor interrogation using the time domain reflectometry approach, as described further below.
The operation of the sensor can be explained by observation of the optical radiation E-field vector behavior when passing through a single sensor segment consisting of a pair of MOS as depicted in Fig. 2(b). If we disregard the semi-reflective mirrors, this setup consists of a pair of linear polarizers that are interconnected by a standard fully circularly-symmetric SMF. When linearly polarized optical radiation propagates along a short and straight section of SMF, the E-field vector maintains its spatial orientation, regardless of whether the fiber is subjected to a twist/rotation around its longitudinal axis [51]. When there is no torsion applied to the segment, the direction of the E-field vector remains aligned with both the polarizer’s axis and the transmission and back-reflection is maximized from the second mirror as depicted in Fig. 3(a). After applying twist/torsion to the sensor segment, as depicted in Fig. 3(b), misalignment between the polarizers occurs, which reduces transmission through the second polarizer in a forward direction of radiation propagation. Thus, the setup consisting of two fiber polarizers that are interconnected by a short section of straight SMF, is equivalent to an ordinary Malus arrangement consisting of a conventional linear polarizer and analyzer, which are placed in series. Twisting of SMF, while rotating the second polarizer, thus modulates the transmitted optical intensity according to Malus` law. A semi reflective mirror, added behind each polarizer, returns back a predetermined portion of the transmitted radiation to allow for measurement of the magnitude of intensity modulation caused by fiber twisting/rotation Fig. 2(a). This back-reflected radiation then passes the SMF in a backward direction and is modulated again by the first polarizer. The ratio between intensities reflected from the two neighboring mirrors/MOSs (i.e. IN and IN + 1) can thus be expressed as:
where RN and RN + 1 are the reflectance of preceding and subsequent MOS mirrors, while k represents a correction factor, which is introduced to compensate for insertion losses in each sensor segment. Factor k consists of an insertion loss (α(N)out) of an output collimator of the preceding MOS, losses in SMF (αSMF) (many due to the splice between two fibers interconnecting MOS), insertion loss of an input collimator (α(N + 1)in) of a subsequent MOS and insertion loss of linear polarizer (α(N + 1)pol) in subsequent MOS and semi-reflective mirror loss (αR(N)), in the preceding MOS and semi-reflective mirror loss (αR(N + 1)) and in subsequent MOS Fig. 2(b). Correction factors k were determined experimentally for each sensor segment in an initial calibration process when all MOSs transmission axes were aligned.The ratio of intensities reflected from the neighboring mirrors (N and N + 1) is, thus, in proportion to the twist/rotation angle θ of the sensor segment interconnecting neighboring MOS:
Reflections/intensities from individual MOS mirrors are measured by a simple Optical Time Domain Reflectometry system (OTDR), built out of standard telecom components and a digital oscilloscope as shown in figure Fig. 1. About 2 ns wide current pulses are fed to a telecom FP Laser diode operating at 1550 nm. The diode is coupled through a 3-dB coupler to the sensor’s lead-in fiber. The detection system consists of a PIN photodiode with Transimpedance Amplifier (TIA), Programmable Gain Amplifier (PGA), digital oscilloscope for signal acquisition, and a Personal Computer (PC) for data/signal processing. The oscilloscope used was a Tektronix MDO4104B-6 with 11bit vertical resolution and Gigabit Ethernet interface for data transfer to the PC. A LabVIEW code was developed for signal acquisition, calculation of segments’ angles according to Eq. (2) and data presentation.The entire sensor was packaged into a U-shaped aluminum profile with cross-sectional dimension of 40 x 40 x 2 mm as shown in Fig. 4(a). MOSs were fixed into this profile by compact precision rotational stages and solid aluminum holders/blocks with dimensions of 40 x 11 mm as shown in Fig. 4(b). Rotational stages were used to align the MOSs’ axis precisely during initial calibration. All MOSs were spaced nearly equidistantly along the structure with about 1.5-1.7m spacing. The fibers in-between MOSs were fusion spliced and were pre-tensioned slightly with tension forces of around 1 N (this pretension was non-critical for sensor operation). The entire test structure was 16 m long (which we believe might be a typical practical dimension for a proposed sensor in typical civil engineering systems). The entire structure was then suspended by eleven small pillars that supported the sensor assembly. Connections between pillars and the sensor assembly were carried through slide bearings to allow for controlled twisting of individual sensor segments. The entire test setup is shown in Fig. 5.
The sensor can be configured to operate in two modes/configurations. In the first configuration all MOSs axis are aligned initially. This provides maximal initial transmission through all sensors` segments. The cumulative twist/rotation of all sensors in the array is limited to about 90 degrees in this configuration (when cumulative rotation of all segments reaches 90 degrees the array becomes opaque). This configuration, however, does not allow for twist/rotation direction discrimination. Rotation of an individual segment in a clockwise or a counterclockwise direction yields the same optical signal modulation depth (change in the ratio among neighboring mirror reflections due to the segment rotation), thus segments within a sensor array can provide only information on the magnitude of rotation of each sensor segment. In the second, direction discrimination mode, MOS axes are initially brought out of the parallelism (i.e. pre-aligned) for a predetermined (known) rotation angle. Twisting of such a pre-aligned section in one direction (for example in clockwise direction) will, for example, increase the ratio of neighboring reflections, while twisting it in the opposite direction (e.g. counterclockwise) will decrease the ratio of neighboring reflections. The increase/decrease of neighboring reflections from an initial preset value thus carries information on directional change in the twist/rotation of individual segments. In the latter configuration, ambiguous measurement ranges of the segment (and the entire array) are however reduced by the pre-set angle(s).
The presented sensor design also provides uniform torsional distribution along the interconnecting fibers. While this is probably not a significant for sensor operation in most situations, it can prevent potential localized bending of the fiber in case rotation occurs suddenly/abruptly due to, for example, a crack formation in the measured object, which might lead to unexpected polarization transformation effects.
Finally, it should be mentioned that twisting of a fiber also induces circular birefringence, which causes E-field vector rotation in the twisted fiber. While this effect is small [49] for rotation up to 45 degrees, the double pass of the optical wave in opposite directions through the same twisted fiber actually cancels out this effect in full and is of no concern in the presented configuration.
3. Experimental results
The experimental sensor consisted of 10 MOS, which defined 9 twist/rotation sensitive segments. Sensor experimental investigations were started by the sensor’s initial calibration. The calibration process included spatial alignment of all polarizers’ axes (i.e. by setting the twist angle to zero at all sensors` segments), while recording sensors` response (OTDR trace) using oscilloscope/PC. An example of this recording is shown in Fig. 6. The peaks in the recording represent reflections from individual MOS mirrors. Due to the variation of MOS insertion losses, splice losses and mirror reflectivity, the amplitude of the peaks decayed unevenly with the distance from the laser/detector. In order to be able to use Eq. (2) for further determination of individual segment’s twist angles, we determined individually coefficients k for each sensor segment. These coefficients were determined simply by calculating the ratio of neighboring peaks` amplitudes in the OTDR trace when all polarizers axes were in aligned positions as indicated in Fig. 6. These coefficients k were then stored in PC memory for further use in twist angle calculation.
Fig. 6 Timed domain back-reflected signal as recorded by digital oscilloscope when all polarizers’ axes were aligned (zero twist angle position). To improve clarity of the signal presentation in the following figures, we omitted primary time and voltage axis designations as presented on Fig. 6. However, the gridlines on all further recordings with their respective values are equal as in Fig. 6.
The experimental setup for angular testing consisted of a single precision motorized rotational stage with angular resolution of 0.01 degrees that was used to rotate a single selected segment controllably. This motorized stage setup was used to obtain the result shown below. When the entire sensor (consisting of all 9 segments) was tested, the rotational angles were preset manually using a precision mechanical angular measurement protractor. The mechanical angular setting accuracy was, in this case, around 0.2°.
After initial calibration, the sensor was ready for operation in a direction-non-discriminating mode of operation. To demonstrate a sensor’s capability to measure twist of individual sensor segments we submitted the sensor to a series of randomly chosen twist/rotation configurations as shown by the inserts in Fig. 7. All OTDR traces in Fig. 7 show the recorded sensor’s responses, together with calculated ratios between neighboring peaks in the OTDR trace, corrected by initial calibration factors k. These corrected ratios can be inserted directly into Eq. (2) for twist angle calculation of individual sensor segments. Figure 7(a) shows the sensor in an initial non-twisted position. Corrected/calibrated ratios among neighboring peaks correspond to 1, which is the equivalent of zero-degree rotation. Figure 7(b) demonstrates the situation when 30 degree sensor twists/rotation occurred only at the fifth segment, while all other segments remained aligned/untwisted. The calibrated ratio of neighboring peaks corresponded to 1 at all segments except at segment five, where this ratio reduced to 0.564. By inserting this value into Eq. (2) we obtained the rotation angle corresponding to 29.94 degrees, which is in a good agreement with the mechanically preset value. The example shown in Fig. 7(c) demonstrates rotation of the entire 3rd and 7th segments, which causes bending/twisting of the 2nd, 4th, 6th and 8th segments. The entire 3rd segment was rotated by 15 degrees, causing peak reduction on the 2nd segment from 1 to 0.871 and on the 4th segment from 1 to 0.87. By inserting the latter values into Eq. (2) we obtained computed angles corresponding to 14.97° and 15.03° respectively. Similarly, we rotated the 7th segment by 35 degrees, causing peak reduction on the 6th segment from 1 to 0.45, and on the 8th segment from 1 to 0.451 which, when inserted in Eq. (2) yields angles of 35.01 and 34.97° respectively. The 2nd and the 4th, and, the 6th and the 8th segments are subjected to the opposite direction of rotation; however, this situation is indistinguishable in the presented direction-non-discriminating mode of operation. Figure 7(d) depicts a situation when the entire sensor is exposed to a uniform twist along its entire length. Corrected/calibrated ratios among neighboring peaks correspond on average to 0.941. Small discrepancies among individual neighboring peak ratios are due to the slight segment length differences and two point clamping of the sensor during the test, which resulted in small differences in the rotation applied to individual segments (for example the 4th and 6th segments were slightly longer that the other segments and, since the tested sensor structure was clamped only in two points when performing this experiment, the rotation of the 4th and 6th segments was somewhat larger than the rotation of other segments). According to Eq. (2), the average value of neighboring peaks` ratio of 0.941 corresponded to an average rotation of 9.996 degrees per segment. This is well consistent with 1/9th of the 90-degree twist which was imposed to the sensor array. In the last demonstration Fig. 7(e), segments were exposed consecutively to the same but oppositely signed rotations i.e. to −20, + 20, −15, + 15, −10, + 10, −15, + 15, −20 degrees, resulting in peak reductions from: 1 to 0.781 for the 1st segment, 1 to 0.78 for the 2nd segment, 1 to 0.872 for the 3rd segment, 1 to 0.87 for the 4th segment, 1 to 0.942 for the 5th segment, 1 to 0.941 for the 6th segment, 1 to 0.87 for the 7th segment, 1 to 0.871 for the 8th segment and 1 to 0.777 for the 9th segment. By processing the latter values with Eq. (2), we attained angles that corresponded to 19.93, 19.98, 14.91, 15.03, 9.88, 9.97, 15.03, 14.97, 20.13 degrees respectively.
Fig. 7 Timed domain back-reflected signal as recorded by digital oscilloscope for different arbitrary chosen twist/rotation states of the sensor array.
When the sensor array is configured for operation in direction-discriminating mode, the initial calibration procedure is similar as before: All MOS polarizers’ axes are first aligned, while calculating neighboring peaks` amplitudes to obtain correction factors k. Then those segments (one, few or all) that shall be able to discriminate the direction of the rotation are pre-set (pre-twisted) for a known angle. While the choice of this pre-twist angle is arbitrary, this angle defines the maximum twist angle that can be resolved unambiguously at a given segment, while the sum of all angles on the pre-twisted segments within the array cannot exceed 90 degrees. An example of such a configuration is presented Fig. 8(a), where we pre-set only the segment 5 for direction discrimination. Here the 6th MOS polarizer axis was pre-aligned (displaced in a counterclockwise direction) by 15 degrees. This resulted in the 6th to 5th peak ratio reduction from 1 to 0.871 which, from Eq. (2), corresponds to the calculated angle of 14.97 degrees. This prepared segment 5 for direction discrimination as shown in Fig. 8(b). The 6th segment was then rotated in a clockwise direction (CW) and counterclockwise direction (CCW) by 15 degrees. When the 5th segment was rotated in a CW direction, the 6th to 5th peak ratio increased from 0.871 to 0.99999 which resulted in an angle of 0.12° as calculated by Eq. (2), while CCW rotation of the same segment caused reduction of the 6th to 5th peak ratio from 0.871 to 0.561, resulting in an angle of 30.06°, when calculated by Eq. (2). Thus, to compute rotation angles of the object to which MOS are attached, the difference between the calculated and pre-set calculated angles, i.e. 15°, shall be calculated: θ5 = 0.12° - 15° = −14.88° for CW and θ5 = 30.06° - 15° = 15.06° for CCW rotation. Figure 8(c) shows rotation of the fifth segment as obtained by an on-line interrogation system using an OTDR circuit, oscilloscope and PC with LabVIEW code, when we gradually rotated the fifth segment from its initial position in a CCW direction for 15 degrees and then back to its initial position followed by the same segment rotation in a CW direction for another 15 degrees and back to the initial position.
Fig. 8 (a-b) Direction-discrimination mode timed domain back-reflected signal, recorded by digital oscilloscope when one segment was pre-aligned and (c) calculated angle.
Another example of direction-discrimination configuration and it use is depicted in Fig. 9. In the latter, we pre-set, i.e. pre-aligned, all segments in a counterclockwise direction by 5 degrees, resulting in neighboring peak ratio reductions from 1 to an average of 0.985 as shown in Fig. 9(a). Twisting of the 1st, 5th and 9th segments by 5 degrees in a CCW direction caused neighboring peak ratio reduction from 0.985 to an average value of 0.941 on the respective segments, while twisting of the 3rd and the 7th segments by 5 degrees in a CW direction caused neighboring peak ratio to increase on these segments from 0.985 to 1. In this sensor state, the 2nd, 4th, 6th, and the 8th segments were exposed to the oppositely signed rotations, i.e. the 2nd and 6th in a CW direction, and the 4th and the 8th segments in a CCW direction Fig. 9(b). The algorithm for calculating the segment’s rotation angles is the same as described in the previous example and, thus, if we insert neighboring peaks` ratios corresponding to the 1st, 2nd, 3rd, 4th, 5th, 6th, 7th, 8th and 9th from Fig. 9(b) in Eq. (2) while subtracting the pre-set angle of 5°, we obtain segments’ rotation angles of 4.98°, −4.97°, −5.02°, 5.16, 5.07°, −5.14°, −4.98°, 4.91°, 4.94° respectively, which is consistent with mechanical preset configuration of the sensor.
Fig. 9 (a) Direction-discrimination mode timed domain back-reflected signal, recorded by digital oscilloscope when all segments are pre-aligned and (b) The response when they are exposed to twist/rotation.
Angular resolution is also affected by the nonlinearity of Eq. (2). For small angles (below a few degrees), Eq. (2) has flat-top dependency (which results in low sensitivity of Eq. (2) to the input change), while it becomes fairly linear for angles between about 5 and 65 degrees. Pre-setting the sensors’ segments by about five or more degrees might thus be beneficial, not only in allowing for directional discrimination, but also for measurements of small changes in rotational/twist angles.
From the results in Figs. 7-9 we can estimate the discrepancy between set and measured angles, which did not exceed 0.12 degrees. If we add to this a mechanical setting angle uncertainty (about 0.2°), we can conclude that the presented system`s absolute error did not exceed 0.3°. This might not be the actual limit and characterization on a more sophisticated and customized precision multi-segment angular measurement setup would be required to investigate this further (unfortunately not available at the time of characterization).
Another demonstration of the proposed sensor operation is shown in Fig. 10. The sensor was configured in direction-non-discriminating mode and then each segment was twisted individually and sequentially by 10 degrees in a CCW direction as shown in Figs. 10(a)-10(i) (sequence of time progressing sensor’s configuration change is indicated by black solid arrows). After twisting all segments, the twist sequence was reversed, and the sensor was returned to its initial position by sequential twisting of segment 9 to 1 in a CW direction, i.e. we changed the sensor from the state described in Fig. 10(i) to the state described in Fig. 10(a) (this time-sequence of sensor’s segments` twisting is indicated by red dashed arrows). During the experiment described above, we performed on-line interrogation of all nine segments (using oscilloscope, PC and LabVIEW code) and the results are shown in Fig. 10(j), which demonstrates successful on-line resolution/interrogation of arbitrary/complex sensor states.
Fig. 10 (a-i) Twist of each segment one by one with (j) On-line interrogated twist angle of each segment.
Since macro-bending induces linear birefringence within SMF, which might influence its polarization state and, thus, cause additional errors in angle read-out, we investigated the influence of fiber curvature interconnecting two MOS on the angle read-out. We performed this test on a single sensor segment. The initial distance among two MOS was decreased gradually to obtain different fiber curvatures (fiber radius was approximated from a fiber sag, as shown in Fig. 11(a). The input polarization E-vector (input MOS) was also aligned at 45 degrees relative to the plane of the fiber curvature (this shall represent worst case as the optical axis lies in the plane of the curvature). Figure 11(b) shows sensor readout change as a function of fiber curvature (i.e. inverse value of fiber radius). It appears that bending of a fiber with a radius above 2 m does not have a significant influence on the sensor performance. This is a curvature range that can be guaranteed relatively easily by intrinsic sensor design/packing. The bend induced sensor readout error follows a parabolic relationship with the fiber curvature, which is consistent with the appearance of bend-indicted fiber birefringence magnitude within the fiber [52].
Fig. 11 Investigation of macro-bend effects on the single sensor segment performance: a) Test setup, b) Change of angle readout as a function of fiber curvature.
A single sensor segment was also tested for micro-bend effects. Firstly, the fiber between two MOS was exposed to micro-bending, using the setup shown in Fig. 12(a). The fiber was depressed in a deformer consisting of three 0.8 mm diameter pins with different forces. The input polarization vector was aligned at 45 degrees relative to the cylinders’ longitudinal axis and the sensor segment was also pre-twisted by 45 degrees (this shall mimic worst-case configuration). The deformer was then loaded with various weights. Results are shown in Fig. 12(c) – Curvature U. The sensors started to show significant (more than 0.2 degrees) deviation in the readout angle when the fiber depression force exceed 0.3 N. To investigate the origin of this effect, we replaced the MOS in the test setup with semi-reflective mirrors (i.e. no polarizers in the setup). We also set the second mirror’s reflectance to a lower value to mimic 45-degree segment rotation Fig. 12(b). This setup shall show how the sensitivity of the single segment is sensitive to micro-bend induced losses along the sensing segment (rather than to any polarization effects), i.e. the torsion angle is calculated from neighboring OTDR peak ratios. Results for this second test are also shown in Fig. 12(c) – Curve V. Both curves U and V show the same trend, but with different rates of angle deviation (the curve U is caused by loss and polarization effects, while the curve V is only due to the loss effects). The rate of angle deviation versus asserted force is, however, similar for both curves, indicating that when the micro-bending is sufficient to cause polarization related readout angle change, the same perturbation will also already provoke significant losses in the same fiber causing comparable effects. This indicates that a significant (loss causing) micro-bend deformation is required to induce significant polarization related effects. A three pin deformer, loaded by 0.3 N, actually asserts relatively large perturbation on the fiber, which is unlikely in properly packaged fiber. While care shall be taken during packaging of the fiber to avoid micro-bending, it seems that limited local micro-bend perturbations shall be of no concern for the proposed sensor design.
Fig. 12 Investigation of micro-bend effects on the single sensor segment performance: a) Test consisting of a single sensor segment, b) Reference setup without polarizers c) Change of angle readout for both setups.
In another set of experiments, we investigated the static characteristics and resolution of the proposed sensing system. The system was configured in direction-non-discriminating mode, while turning the 1st, 5th and 9th segments individually over 0° to 90° in 10-degree steps (each segment was turned individually from 0° to 90°, while all other segments remained in 0° position during this test). The results, as obtained by the proposed interrogation setup, are shown in Fig. 13(a). Results on the 1st and the 5th segments show very close matching to the physical twist in practically all steps. On the 9th segment there is a small discrepancy in steps close to 90°. The latter can be attributed to a limited signal-to-noise ratio obtained for the last peak in the OTDR trace and the amplification circuit offset, as the amplitude of the last peak in the back-reflected trace gets very low near the 90° twist angle. Improvements in amplification circuit, introduction of automatic offset control and optimization of mirrors’ reflections might contribute further to the improvement of this discrepancy (a very simple amplification circuit was utilized in the experimental setup).
Fig. 13 (a) Demonstration of sensors` static characteristics and (b) Resolution by left and right movement of the 9th segment by 0.28°.
The measurement of resolution varies among individual segments due to the OTDR pulse power decay that is caused both by system (insertion) and modulation (sensor rotation induced) losses within the system, i.e. more distant segments receive less pulsed laser diode power, which results in a poorer signal to noise ratio of received back-reflected pulses. Since the rotation of individual segments changes modulation losses arbitrarily, the resolution can also change during the sensor operation. The achievable resolution for a particular segment is, thus, a complex function of OTDR parameters (initial pulse power, pulse duration, measurement system bandwidth and applied post processing) and optical pulse power available at input of the observed segment. This is, however, not a vital limitation as reflectance from MOS mirrors can be set arbitrarily and since even simple laser diode generates sufficient optical power to achieve reasonable high signal to noise ratios even in the presence of high losses in the system. To demonstrate minimum achievable resolution with the current experimental version of the proposed system (using a very simple OTDR system) we performed rotation of the 9th segment (10th MOS) periodically in CW and CCW directions, by using a precision rotational stage (the resolution obtained for the 9th segment shall represent the worst case, as the signal-to-noise ratio is the worst for peaks originating from the last two MOS mirrors). Figure 13(b) shows the response of the system to the 9th segment CCW/CW rotations with physical angle amplitude of 0.28°. The cumulative angle of all preceding segments in the sensor network was set to 60°. (This shall demonstrate about the absolute worst case encountered in practical situations). Peak-to-peak output noise corresponded to less than 0.08° and 0.09° respectively, which also indicates an achievable resolution of the presented system while using the current experimental setup.
Finally, we examined crosstalk among individual segments of the array. In this experiment we observed the influence of the rotation of the 1st, 2nd, 3rd, and 4th segments on the measurement (interrogated) output of the 5th segment, which was fixed and did not rotate during this test (i.e. we fixed the 5th segment, while we rotated segments 1 to 4 consecutively by 45 degrees in CCW and CCW directions). The results are shown Fig. 14 and indicate that the cross-talk between segment 5 and any of the segments 1 to 4 was below 0.1 degree (0.06 on average). The origin of this small crosstalk can be probably attributed to electronic circuits non-idealities and limited polarizers` extinction ratios.
Fig. 14 Response of segment 5 redout to successive 45° CW and CCW twisting of the 4th, 3rd, 2nd and 1st segments.
4. Conclusion
This paper presented a fiber-optic quasi-distributed in-line twist/rotation sensor. The sensor utilizes the inability of the standard single mode fiber to change the polarization state (E-field vector orientation) of the wave propagating down the twisted/rotated fiber. Multiple linear polarizers with integrated semi-reflective mirrors were set along standard single-mode fiber to form multiple rotation sensitive segments that can be interrogated by a simple OTDR system. The entire experimental system was built out of simple and low-cost optical and optoelectronic components. A measurement resolution better than 0.3°, unambiguous total measuring range of 90°, and the crosstalk lower than 0.1° among individual segments was demonstrated experimentally. The sensor can operate in two modes i.e. in non-direction discrimination mode and in direction-discrimination mode. Direction discrimination requires sensor pre-alignment and reduces the unambiguous range of operation. The sensor system also enables the possibility of creating no-sensitivity zones by swapping interconnecting SM fiber with a PM fiber on sections where there is no necessity for sensing twist/torsion. The sensor is also intrinsically insensitive to temperature and longitudinal strain.
The fibers between the MOS must be straight, which might introduce certain limitations in mechanical design of the sensor’s packing and sensor installation conditions. Experimental results, however, indicate Fig. 11 that deviation from the system’s “straight” configuration with fiber bend radius of greater than a couple of meters is tolerable, which shall be achievable straightforwardly by different practical sensor packaging designs. Mild local micro-bends also proved to be of no concern.
The demonstrated performances might likely be improved further by optimization of the mirror reflections along the sensor length and improvement of the electronic amplification circuits, which were very elementary in the present experimental investigation.
Funding
This work was supported by Slovenian Public Research agency under P2-0368 and 6316-3/2012-883.
Acknowledgments
We also thank Borut Preložnik for his support with LabVIEW code development.
References and links
1. T. Koh, H. Takase, and T. Tsugawa, “Torsional problems in aseismic design of high-rise buildings,” in Proceedings of the Fourth World Conference on Earthquake Engineering, Santiago de Chile (1969), pp. 71–88.
2. I. Venanzi, F. Cluni, V. Gusella, and A. L. Materazzi, “Torsional and across-wind response of high-rise buildings,” in Proceedings of the 12th International Conference on Wind Engineering AWES (2007).
3. A. M. Aly, A. Zasso, and F. Resta, “Dynamics and control of high-rise buildings under multidirectional wind loads,” Smart Mater. Res. 2011, 549621 (2011). [CrossRef]
4. A. L. Materazzi and F. Ubertini, “Eigenproperties of suspension bridges with damage,” J. Sound Vibrat. 330(26), 6420–6434 (2011). [CrossRef]
5. O. A. Bauchau, T. M. Krafchack, and J. F. Hayes, “Torsional buckling analysis and damage tolerance of graphite/epoxy shafts,” J. Compos. Mater. 22(3), 258–270 (1988). [CrossRef]
6. I. L. Howard, “Practical approaches for evaluating bending and torsion of fiber reinforced polymer components using instrumented testing,” J. Appl. Sci. Eng. Technol. 2, 1–7 (2008).
7. P. D. Spanos, A. M. Chevallier, N. P. Politis, and M. L. Payne, “Oil and gas well drilling: a vibrations perspective,” Shock Vib. 35(2), 85–103 (2003). [CrossRef]
8. J. Rajnauth and T. Jagai, “Reduce torsional vibration and improve drilling operations,” Int. J. Appl. Sci. Technol. 2, 109–123 (2012).
9. D. Dei, M. Pieraccini, M. Fratini, C. Atzeni, and G. Bartoli, “Detection of vertical bending and torsional movements of a bridge using a coherent radar,” NDT Int. 42(8), 741–747 (2009). [CrossRef]
10. M. Pieraccini, M. Fratini, F. Parrini, and C. Atzeni, “Monitoring of bridges using coherent radar: detection of longitudinal and torsional modes,” in Proceedings of the European Radar Conference, EuRAD 2008 (2008), pp. 176–179.
11. H. S. Park, H. G. Shon, I. S. Kim, and J. H. Park, “Monitoring of structural behavior of high-rise buildings using GPS,” in Proceedings of the CTBUH 2004 Seoul Conference (2004).
12. W. A. A. W. Akib, S. K. Kok, and Z. Mat Amin, “High rise building deformation monitoring with GPS,” in Proceedings of Symposium and Exhibition on Geoinformation (2004).
13. V. H. S. Khoo, Y. K. Tor, and G. On, “Monitoring of High Rise Building using Real-Time Differential GPS,” in Proceedings of FIG Congress - Facing the Challenges – Building the Capacity, Sydney, Australia (2010).
14. S. W. Choi, I. S. Kim, J. H. Park, Y. Kim, H. G. Sohn, and H. S. Park, “Evaluation of stiffness changes in a high-rise building by measurements of lateral displacements using GPS technology,” Sensors (Basel) 13(11), 15489–15503 (2013). [CrossRef] [PubMed]
15. A. Maekawa, M. Noda, M. Shintani, and M. Suzuki, “Development of noncontact measurement methods using multiple laser displacement sensors for bending and torsional vibration stresses in piping systems,” Int. J. Press. Vessels Piping 137, 38–45 (2016). [CrossRef]
16. A. Galtarossa and L. Palmieri, “Measure of twist-induced circular birefringence in long single-mode fibers: Theory and experiments,” J. Lightwave Technol. 20(7), 1149–1159 (2002). [CrossRef]
17. X. Xi, G. K. L. Wong, T. Weiss, and P. S. Russell, “Measuring mechanical strain and twist using helical photonic crystal fiber,” Opt. Lett. 38(24), 5401–5404 (2013). [CrossRef] [PubMed]
18. D. E. Ceballos-Herrera, I. Torres-Gomez, A. Martinez-Rios, L. Garcia, and J. J. Sanchez-Mondragon, “Torsion Sensing Characteristics of Mechanically Induced Long-Period Holey Fiber Gratings,” IEEE Sens. J. 10(7), 1200–1205 (2010). [CrossRef]
19. H. H. Liang, Y. X. Jin, Y. Zhao, and J. F. Wang, “Twist sensor by using a pressure-induced birefringence single mode fiber based Sagnac Interferometer,” Proc. SPIE 8311, 83112H (2011). [CrossRef]
20. O. Frazao, C. Jesus, J. M. Baptista, J. L. Santos, and P. Roy, “Fiber-optic interferometric torsion sensor based on a two-LP-mode operation in birefringent fiber,” IEEE Photonics Technol. Lett. 21(17), 1277–1279 (2009). [CrossRef]
21. W. G. Chen, S. Q. Lou, L. W. Wang, H. Zou, W. L. Lu, and S. S. Jian, “Highly sensitive torsion sensor based on Sagnac interferometer using side-leakage photonic crystal fiber,” IEEE Photonics Technol. Lett. 23(21), 1639–1641 (2011). [CrossRef]
22. B. Song, H. Zhang, Y. Miao, W. Lin, J. Wu, H. Liu, D. Yan, and B. Liu, “Highly sensitive twist sensor employing Sagnac interferometer based on PM-elliptical core fibers,” Opt. Express 23(12), 15372–15379 (2015). [CrossRef] [PubMed]
23. Y. Yu, Q. Huang, X. Li, X. Chen, and C. Du, “An in-line fiber-optic modal interferometer for simultaneous measurement of twist and ambient temperature,” Sens. Biosensing Res. 2, 38–42 (2014). [CrossRef]
24. H. Y. Fu, S. K. Khijwania, H. Y. Tam, P. K. A. Wai, and C. Lu, “Polarization-maintaining photonic-crystal-fiber-based all-optical polarimetric torsion sensor,” Appl. Opt. 49(31), 5954–5958 (2010). [CrossRef]
25. O. Frazão, S. O. Silva, J. M. Baptista, J. L. Santos, G. Statkiewicz-Barabach, W. Urbanczyk, and J. Wojcik, “Simultaneous measurement of multiparameters using a Sagnac interferometer with polarization maintaining side-hole fiber,” Appl. Opt. 47(27), 4841–4848 (2008). [CrossRef] [PubMed]
26. O. Frazão, R. M. Silva, J. Kobelke, and K. Schuster, “Temperature- and strain-independent torsion sensor using a fiber loop mirror based on suspended twin-core fiber,” Opt. Lett. 35(16), 2777–2779 (2010). [CrossRef] [PubMed]
27. R. M. Silva, M. S. Ferreira, and O. Frazao, “Temperature independent torsion sensor using a high-birefringent Sagnac loop interferometer,” Opt. Commun. 285(6), 1167–1170 (2012). [CrossRef]
28. H. M. Kim, T. H. Kim, B. Kim, and Y. Chung, “Temperature-insensitive torsion sensor with enhanced sensitivity by use of a highly birefringent photonic crystal fiber,” IEEE Photonics Technol. Lett. 22(20), 1539–1541 (2010). [CrossRef]
29. P. Zu, C. C. Chan, Y. X. Jin, T. X. Gong, Y. F. Zhang, L. H. Chen, and X. Y. Dong, “A temperature-insensitive twist sensor by using low-birefringence photonic-crystal-fiber-based Sagnac interferometer,” IEEE Photonics Technol. Lett. 23(13), 920–922 (2011). [CrossRef]
30. M. S. Ferreira, J. L. Santos, P. Mergo, and O. Frazao, “Torsion sensor based on a figure-of-eight cavity fibre laser,” Laser Phys. Lett. 10(4), 045105 (2013). [CrossRef]
31. Y.-L. Lo, B.-R. Chue, and S.-H. Xu, “Fiber torsion sensor demodulated by a high-birefringence fiber Bragg grating,” Opt. Commun. 230(4-6), 287–295 (2004). [CrossRef]
32. W. Yiping, M. Wang, and X. Huang, “In fiber Bragg grating twist sensor based on analysis of polarization dependent loss,” Opt. Express 21(10), 11913–11920 (2013). [CrossRef] [PubMed]
33. Y. P. Wang, X. Q. Huang, and M. Wang, “Temperature- and strain-independent torsion sensor utilising polarisation-dependent loss of Hi-Bi FBGs,” IEEE Electron. Lett. 49(13), 840–841 (2013). [CrossRef]
34. W. G. Zhang, G. Y. Kai, X. Y. Dong, S. Z. Yuan, and Q. D. Zhao, “Temperature-independent FBG-type torsion sensor based on combinatorial torsion beam,” IEEE Photonics Technol. Lett. 14(8), 1154–1156 (2002). [CrossRef]
35. C. Y. Shen, Y. Zhang, W. J. Zhou, and J. Albert, “Au-coated tilted fiber Bragg grating twist sensor based on surface plasmon resonance,” Appl. Phys. Lett. 104(7), 071106 (2014). [CrossRef]
36. C. Mou, K. Zhou, R. Suo, L. Zhang, and I. Bennion, “Fibre laser torsion sensor system using an excessively tilted fibre grating and low-cost time domain demodulation,” Proc. SPIE 7503, 75033E (2009). [CrossRef]
37. Y. F. Lu, C. Y. Shen, D. B. Chen, J. L. Chu, Q. Wang, and X. Y. Dong, “Highly sensitive twist sensor based on tilted fiber Bragg grating of polarization-dependent properties,” Opt. Fiber Technol. 20(5), 491–494 (2014). [CrossRef]
38. X. Chen, K. Zhou, L. Zhang, and I. Bennion, “In-fiber twist sensor based on a fiber Bragg grating with 81 degrees tilted structure,” IEEE Photonics Technol. Lett. 18, 2596–2598 (2006). [CrossRef]
39. T. Guo, F. Liu, B. O. Guan, and J. Albert, “Polarimetric multi-mode tilted fiber grating sensors,” Opt. Express 22(6), 7330–7336 (2014). [CrossRef] [PubMed]
40. C. Y. Lin, L. A. Wang, and G. W. Chern, “Corrugated long-period fiber gratings as strain, torsion, and bending sensors,” J. Lightwave Technol. 19(8), 1159–1168 (2001). [CrossRef]
41. Y. J. Rao, T. Zhu, and Q. J. Mo, “Highly sensitive fiber-optic torsion sensor based on an ultra-long-period fiber grating,” Opt. Commun. 266(1), 187–190 (2006). [CrossRef]
42. Y. P. Wang and Y. J. Rao, “Long period fibre grating torsion sensor measuring twist rate and determining twist direction simultaneously,” IEEE Electron. Lett. 40(3), 164–166 (2004). [CrossRef]
43. M. Pulido-Navarro, J. Álvarez-Chávez, D. Ceballos-Herrera, and P. Escamilla-Ambrosio, “Optical fiber torsion sensor with mechanically induced long period fiber gratings in rare-earth doped fibers,” Opt. Photonics J. 04(06), 129 (2014). [CrossRef]
44. Y. P. Wang, J. P. Chen, and Y. J. Rao, “Torsion characteristics of long-period fiber gratings induced by high-frequency CO2 laser pulses,” J. Opt. Soc. Am. B 22(6), 1167–1172 (2005). [CrossRef]
45. L. L. Shi, T. Zhu, Y. E. Fan, K. S. Chiang, and Y. J. Rao, “Torsion sensing with a fiber ring laser incorporating a pair of rotary long-period fiber gratings,” Opt. Commun. 284(22), 5299–5302 (2011). [CrossRef]
46. D. Lesnik and D. Donlagic, “In-line, fiber-optic polarimetric twist/torsion sensor,” Opt. Lett. 38(9), 1494–1496 (2013). [CrossRef] [PubMed]
47. T. Guo, F. Liu, F. Du, Z. Zhang, C. Li, B. O. Guan, and J. Albert, “VCSEL-powered and polarization-maintaining fiber-optic grating vector rotation sensor,” Opt. Express 21(16), 19097–19102 (2013). [CrossRef] [PubMed]
48. V. Budinski and D. Donlagic, “Miniature, all-fiber rotation sensor based on temperature compensated wave plate,” IEEE Photonics Technol. Lett. 27(1), 85–88 (2015). [CrossRef]
49. B. Song, Y. Miao, W. Lin, H. Zhang, J. Wu, and B. Liu, “Multi-mode interferometer-based twist sensor with low temperature sensitivity employing square coreless fibers,” Opt. Express 21(22), 26806–26811 (2013). [CrossRef] [PubMed]
50. C. G. Askins, G. A. Miller, and E. J. Friebele, “Bend and twist sensing in a multiple-core optical fiber,” in 2008 Conference on Optical Fiber Communication/National Fiber Optic Engineers Conference (2008), pp. 1545–1547. [CrossRef]
51. A. M. Smith, “Birefringence induced by bends and twists in single-mode optical fiber,” Appl. Opt. 19(15), 2606–2611 (1980). [CrossRef] [PubMed]
52. R. Ulrich, S. C. Rashleigh, and W. Eickhoff, “Bending-induced birefringence in single-mode fibers,” Opt. Lett. 5(6), 273–275 (1980). [CrossRef] [PubMed]
