347 Optical Fiber S 14. Optical Fiber Strain Gages Chris S. Baldwin 14.1 Optical Fiber Basics ............................... 348 14.1.1 Guiding Principals for Optical Fiber 348 14.1.2 Types of Optical Fibers ................. 349 14.2.3 14.2.4 14.2.5 14.2.6 Advantages of Fiber Optic Sensors . Limitations of Fiber Optic Sensors .. Thermal Effects ........................... Introduction to Strain–Optic Effect 352 353 354 354 14.3 Interferometry ..................................... 14.3.1 Two-Beam Interference ............... 14.3.2 Strain–Optic Effect....................... 14.3.3 Optical Coherence ....................... 14.3.4 Mach–Zehnder ........................... 14.3.5 Michelson .................................. 14.3.6 Fabry–Pérot ............................... 14.3.7 Polarization................................ 14.3.8 Interrogation of Interferometers ... 354 355 355 356 357 357 357 358 358 14.4 Scattering ............................................ 359 14.4.1 Brillouin Scattering...................... 359 14.4.2 Strain Sensing Using Brillouin Scattering................................... 359 14.5 Fiber Bragg Grating Sensors .................. 14.5.1 Fabrication Techniques ................ 14.5.2 Fiber Bragg Grating Optical Response ................................... 14.5.3 Strain Sensing Using FBG Sensors .. 14.5.4 Serial Multiplexing ...................... 14.5.5 Interrogation of FBG Sensors, Wavelength Detection.................. 14.5.6 Other Grating Structures ............... 361 361 362 363 364 14.6 Applications of Fiber Optic Sensors ......... 14.6.1 Marine Applications..................... 14.6.2 Oil and Gas Applications .............. 14.6.3 Wind Power Applications ............. 14.6.4 Civil Structural Monitoring ............ 367 367 367 368 368 366 367 14.2 General Fiber Optic Sensing Systems....... 351 14.2.1 Strain Sensing System Concept ...... 351 14.2.2 Basic Fiber Optic Sensing Definitions ................................. 351 References .................................................. 369 Many optical fiber sensors are based on classical bulk optic arrangements, as discussed in Chap. 18 – Basics of Optics. These arrangements are named after the in- ventors who are credited with their invention. Other optical fiber sensors are based on physical phenomenon such as scattering effects that are inherent to the optical 14.7 Summary ............................................. 368 Part B 14 Optical fiber strain sensing is an evolving field in optical sciences in which multiple optical principles and techniques are employed to measure strain. This chapter seeks to provide a concise overview of the various types of optical fiber strain sensors currently available. The field of optical fiber strain sensing is nearly 30 years old and is still breaking new ground in terms of optical fiber technology, instrumentation, and applications. For each sensor discussed in the following sections, the basic optical layout is presented along with a description of the optical phenomena and the governing equations. Comprehensive coverage of all aspects of optical fiber strain sensing is beyond the scope of this chapter. For example, each sensor type can be interrogated by a number of means, sometimes based on differing technology. Furthermore, these sensors are finding applications in a wide variety of fields including aerospace, oil and gas, maritime, and civil infrastructures. The interested reader is referred to the works by Grattan and Meggit [14.1], Measures [14.2], and Udd [14.3] for more-detailed descriptions of interrogation techniques and applications of the various sensors. 348 Part B Contact Methods fiber material. In general, fiber optic sensors function by monitoring a change in an optical parameter as the optical fiber is exposed to the strain field. The three main types of optical fiber strain gages are • • • interferometry (changes in optical phase) scattering (changes in optical wavelength) fiber Bragg grating (changes in optical wavelength) Other optical phenomena such as intensity variations are exploited for optical fiber sensors measuring other quantities such as chemical concentrations, applied loads/pressure, and temperature. The strain sensing mechanisms listed above are explored in the sections below, but first an introduction to optical fibers and a general overview of the basic layout, advantages, and disadvantages of optical fiber strain sensors is provided. 14.1 Optical Fiber Basics Part B 14.1 The first thought many people have when introduced to optical fiber is of a thin, fragile piece of glass. On the contrary, the ultrapure manufacturing that goes into producing the low-loss optical fiber of today creates a thin glass structure essentially free of defects with tensile strength values near 800 ksi (5.5 GPa) [14.2]. The following sections discuss the optical guiding properties of optical fiber and provide a brief description of the various types of optical fiber commercially available for sensing applications. 14.1.1 Guiding Principals for Optical Fiber Before any discussion of fiber optic strain sensors can be realized, an understanding of the light guiding principles of optical fiber should be presented. The basic material for optical fiber is fused silica [14.2]. Standard optical fiber consists of two concentric glass portions. The inner portion is called the core and the outer portion is called the cladding. The core region of the optical fiber typically contains dopants such as germanium (Ge) or boron (B) to increase the refractive index of the core (n core ) to a slightly higher value than that of the cladding (n clad ), which is pure fused silica. Optical fiber also uses protective layers of polymer coatings to protect the glass surface from damage. For most strain sensing applications, one of two coatings is applied to the optical fiber. Acrylate-coated optical fibers have a final outer diameter of 250 μm. Polyimide-coated optical fibers are available in a range of final outer diameters, typically around 180 μm. Standard telecommunications optical fiber has a germanium-doped core. The optical transmission spectrum for Ge–Si is shown in Fig. 14.1 [14.2]. The transmission spectra displays the characteristic decay of the Rayleigh curve (1/λ4 ); at higher wavelengths the transmission is limited by the absorption of energy by the silica structure. The multiple peaks in the attenua- tion curve are due to OH− (hydroxyl) scattering in the optical transmission curve. The effect of the hydroxyl peaks is the creation of transmission windows for optical fiber. The telecommunications industry makes use of the wavelength regions around 1550 nm due to the lowloss properties and the availability of optical sources and detectors. This is commonly referred to as the Cband. The region around 1300 nm, termed the S-band, is also used for data transmission, but typically is reserved for shorter-length applications such as local area networks (LANs). A schematic diagram of standard single-mode optical fiber is shown in Fig. 14.2. The size of the core (radius = a) is dependent on the wavelength of light the optical fiber is designed to guide. For single-mode operation in the C-band, this value is around 4 μm. The cladding diameter is standardized at 125 μm for many optical fiber types. This allows for the use of common optical fiber tools such as cleavers, strippers, and fusion splicers to be used on different optical fiber types from different optical fiber vendors. In the simplest sense, light is guided via total internal reflection (TIR) between the core and cladding Attenuation (dB/km) S-band 0.6 0.8 1 1.2 C-band 1.4 1.6 1.8 Wavelength (μm) Fig. 14.1 Optical transmission windows for optical fibers Optical Fiber Strain Gages 1 250 μm a J0 (V ) 0.75 349 V = 2.405 J1 (V ) 0.5 125μm 14.1 Optical Fiber Basics 0.25 0 Fig. 14.2 Schematic of typical optical fibers –0.25 Therefore, the modal propagation of light in a stepindex optical fiber is dependent on the core size (a), the refractive index difference between the core and cladding, and the wavelength of light propagating in the core. The V number value to ensure single-mode operation is 2.405. If the V number is larger than 2.405 then Acceptance cone α Cladding Core Fig. 14.3 Total internal reflection between core and cladding –0.5 –0.75 –J1 (V ) 0 2 4 Singlemode region 6 8 10 Fig. 14.4 Bessel solution space for modal operation of op- tical fiber light in the optical fiber is not propagating as singlemode. Based on the formulation given in (14.1), a cutoff wavelength value can be determined for the optical fiber, which defines the lower limit of single-mode operation with respect to wavelength. The cutoff wavelength is given by: 2πa n 2core − n 2clad . (14.2) λcutoff = V Any wavelength of light lower than the cutoff wavelength will not propagate as single-mode in the optical fiber. Likewise, to determine the appropriate core size to ensure single-mode operation at a given wavelength, (14.1) can be written as (14.3) a = 2.405λ/2πV n 2core − n 2clad . The vast majority of optical fiber strain sensors employ standard single-mode optical fiber or similar constructed optical fiber. The following section provides a brief description of optical fiber types that may be encountered in the field of strain sensing. 14.1.2 Types of Optical Fibers For the most part, fiber optic strain sensors make use of readily available, standard optical fiber and components. Optical fiber sensing has experienced a new thrust of interest and development in recent years. Much of the current research has been focused on the field of biosensing with the development of new optical fiber types and sensing techniques. Optical fiber strain sensing has been steadily finding and being integrated Part B 14.1 of the optical fiber, as illustrated in Fig. 14.3. TIR was introduced in Chap. 18. Figure 14.3 shows a slab waveguide description for optical fiber. Based on the slab waveguide formulation, the standard Snell’s law can be employed to determine the guiding principles for the structure. This analysis leads to an acceptance cone where light entering the core within this angle will result in a guiding condition. In optical fiber terms, this is related to the numerical aperture (NA). Of course, the explanation of TIR for the guiding principles of optical fiber is very limiting. The derivation for the propagation of light through an optical fiber is based on a solution of the electromagnetic wave equation (18.1), as discussed in Chap. 18. This derivation is beyond the scope of this chapter. The important result from the derivation is that the propagation of light through the core is based on Bessel functions (Ji (x)). Single-mode propagation in optical fiber is defined by a parameter known as the V number, which is the argument of the Bessel function solution for guided core modes [14.2]. The solution plots of the first few Bessel functions are shown in Fig. 14.4. The V number for standard single-mode optical fiber is expressed as 2πa n 2core − n 2clad . (14.1) V = ka n 2core − n 2clad = λ 350 Part B Contact Methods a) Singlemode b) Multimode c) PM-elliptical core d) PM-bowtie Part B 14.1 Fig. 14.5a–d Cross-section views of various types of opti- cal fiber into more application areas including civil infrastructure monitoring, oil and gas application, embedded composite sensing, and maritime sensing. As discussed in Sect. 14.2.3, optical fiber sensing possesses some inherent advantages over traditional sensing techniques for these and other application areas. The development of low-loss optical fibers in the 1970s began a period of innovation in both telecommunications and sensing applications. The first optical fibers designed for telecommunication applications were manufactured by Corning in 1970 and possessed attenuation values just less than 20 dB/km [14.2]. Current attenuation values for telecommunication grade optical fiber are of the order of 0.2 dB/km [14.2]. Although the vast majority of fiber optic sensors use standard communications-grade single-mode optical fiber, it should be noted that multiple varieties of optical fiber exist and are being explored for various sensing applications. Figure 14.5 shows a cross-section view of several types of optical fiber. For example, a variety of polarization-maintaining optical fibers exist that possess an induced birefringence throughout the fiber length, thus maintaining two principal polarization axes for optical transmission. Various dopants and fiber structures have been explored throughout the past few decades to decrease the influence of bend losses and improve signal quality within optical fiber. Using Fig. 14.2 as a reference, the optical fibers displayed in Fig. 14.5 can be compared to the standard single-mode optical fiber. Multimode optical fiber (illustrated in Fig. 14.5b) possesses a much larger core size compared to single-mode fiber. This aspect was discussed in Sect. 14.1.1 in terms of the V number dictating single-mode operation. Multimode optical fiber may either have a step index core (like the single-mode optical fiber) or a graded refractive index core. If the multimode fiber is step index, then the various propagation modes travel different path lengths due to the oscillatory nature of the propagation. Graded-index optical fibers allow the various modes propagating in the fiber to have equivalent path lengths. Figure 14.5 also displays two types of polarizationmaintaining fiber (PM fiber). Polarization of light is covered in both Chaps. 18 and 25 of this Handbook. The polarization-maintaining aspect is induced either through a geometric structure of the core (elliptical core fiber – Fig. 14.5c) or by inducing a permanent mechanical load within the fiber structure (such as bow-tie fiber – Fig. 14.5d). PM fiber develops two polarization states (fast and slow axis) whose responses to mechanical loads are slightly different. PM fiber has found applications in strain sensing, as discussed in Sect. 14.3.7. Most recently, the development of photonicbandgap (PBG) fiber and photonic-crystal fiber (PCF) has generated great interest in the sensing community, particularly in the field of chemical and biosensing. These optical fibers are commonly referred to as “holey fibers” because the core and cladding regions are developed by manufacturing a structured pattern of voids (holes) that traverse the entire length of the optical fiber. For more information on this type of optical fiber, the reader may visit www.crystal-fibre.com. In general, PCFs use a patterned microstructure of voids along the length of the optical fiber to create an effective lower-index cladding and guide light in a solid core by modified total internal reflection (M-TIR)[14.4]. PBG fibers guide light within a low-index region (longitudinal void) by creating a photonic bandgap based on the structure and pattern of the voids [14.4]. Optical Fiber Strain Gages 14.2 General Fiber Optic Sensing Systems 351 14.2 General Fiber Optic Sensing Systems 14.2.1 Strain Sensing System Concept Interrorgation unit Fiber optic sensor Source Optics Feedback Data acquisition Demodulator Fig. 14.6 Basic configuration of an optical fiber sensor sys- tem a) Distributed b) Discrete Fig. 14.7a–c Fiber optic sensor classifications 14.2.2 Basic Fiber Optic Sensing Definitions Due to the wide variety of fiber optic sensors, some definitions of optical fiber strain sensing are required. Fiber optic sensors can be classified into distributed, discrete, or cumulative strain sensors, as illustrated in Fig. 14.7. A distributed strain sensor provides a measure of strain at potentially every point along the sensing optical fiber. Distributed sensors typically have a spatial resolution of approximately 1 m due to the resolution of the measurement systems. A discrete strain sensor provides a strain measurement at one location often based on a smallgage-length fiber optic sensor. By serially multiplexing these sensors, a distributed sensor array can be fabricated with the strain measurements being at discrete points instead of an average of the spatial resolution. A cumulative strain sensor provides a measure of strain that is an average strain value over the entire sensing length of the optical fiber. The selection of a fiber optic strain sensor for a particular application is driven by the sensing requirements for the application. For example, measurement of the hoop strain of a pressure vessel may use a distributed sensor to obtain strain values at multiple locations around the circumference. A discrete sensor can provide a single strain measurement at a single location on the circumference or multiple measurements if serially multiplexed. A cumulative strain sensor would provide a single strain measurement based on the overall strain induced in the fiber bonded to the circumference of the vessel. Cumulative sensors provide the highest sensitivity and are often employed when very small strain signals are of interest such as the measurement of acoustic fields. In many cases, the optical fiber serves as both the conduit for the optical signal and the sensing mechanism. In this situation, the sensor is referred to as an intrinsic sensor. An extrinsic sensor is the case where the optical fiber delivers the optical signal to the sensc) Cumulative Part B 14.2 The basic building blocks of a fiber optic strain sensor system can be compared to the building blocks of a resistance strain gage (RSG) system discussed in Chap. 12. A resistance strain gage circuit requires a voltage supply, Wheatstone bridge, electrical wires connecting to the RSG, a voltmeter to monitor the circuit output, and a data-acquisition system to record the voltage changes. In terms of fiber optic sensors, an equivalent arrangement can be drawn where the voltage supply is equivalent to the light source (laser, light emitting diode, or other optical source). The Wheatstone bridge is equivalent to the optics that guide the light to and from the sensors or sensing region of the optical fiber. The wires that connect to the RSG are simply the lead optical fiber to the fiber optic sensors. In some cases, a single strand of optical fiber is used to monitor multiple (potentially thousands) fiber optic strain sensors (see multiplexing below). This is vastly different from RSG circuits where each uniaxial gage requires at least two conductors. The voltmeter for a fiber optic sensor system is the interrogation unit (also called a demodulator). The interrogation units typically have one or more photodetectors to transfer the optical signal to a voltage signal proportional to the optical intensity. The data-acquisition system is the one component that may be an exact duplicate between the RSG circuit and the fiber optic sensor system. This is typically a personal computer (PC) or laptop device with appro- priate interface cards/boards to record the data from the interrogation unit. 352 Part B Contact Methods A A/2 B/2 B A/2 B/2 Fig. 14.8 Bidirectional optical fiber coupler Part B 14.2 ing mechanism such as an air cavity. In both cases, the applied strain induces a physical change in the geometry of the sensor. For the case of an intrinsic sensor, a strain–optic effect is also induced and must be considered when determining the calibration coefficient (gage factor) of the sensor. Another differentiation between types of optical fiber sensors is whether the sensor works as a reflective or a transmission element. The majority of fiber optic strain sensors function in reflection mode, where light traveling in the optical fiber is reflected back towards the optical source from reflective elements or from inherent scattering of the optical fiber material. In order to direct the back-reflected light to the interrogation instrumentation, an optical fiber coupler device is employed. A bidirectional coupler is typically used in these applications, as shown in Fig. 14.8. The commercially available standard bidirectional coupler is termed a 3 dB coupler because it divides the optical signal equally into the two output arms as indicated in Fig. 14.8. 14.2.3 Advantages of Fiber Optic Sensors Optical fiber sensors offer many advantages over traditional electrical-based strain sensors. Many of these advantages stem from the pure optical nature of the sensor mechanisms. Traditional sensors depend on the measurement of variations of resistance or capacitance of the electrical sensors. Optical fiber sensors depend on changes to optical parameters of the optical fiber or the light traveling within the optical fiber. This difference leads to many advantages, including • • • • immunity to electromagnetic interference long transmission lead lines no combustion danger serial multiplexing The following discussion expands on each of the advantages stated above. Immunity to Electromagnetic Interference Optical fiber sensors function by measuring changes to the light that is traveling within the optical fiber. Unlike low-level voltage signals, resistance changes, or other electrical phenomena used for traditional sensors, electromagnetic influences do not cause measurable levels of noise. Therefore, when dealing with optical lead lines and cabling, no special shielding is required, thus greatly reducing the cabling weight compared to traditional sensors. Long Transmission Lines With the advances made in the telecommunications field for optical fiber transmission, current standards for optical signal transmission allow the light signal to be transmitted many kilometers without a detrimental level of signal degradation. Optical power loss in current communications grade optical fiber are on the order of 0.2 dB/km. Other factors that affect the transmission of optical signals in fibers such as chromatic and material dispersion have also experienced improvements. No Combustion Danger In some applications, there exists a risk of ignition or explosion of combustive materials from sparks or electrical potentials from sensors. The optical power guided within the core for fiber optic sensors is typically less than 1 mW, essentially eliminating any potential of spark or ignition danger. Without the requirement of a voltage potential at the sensor location and the ability to have long lead lines, optical fiber sensors are regarded as being immune to combustion danger. This is one reason why fiber optic sensors are being heavily explored for various oil and gas industry applications [14.5]. Multiplexing Some of the optical fiber sensors discussed in this chapter have the ability to be multiplexed (multiple sensors interrogated by a signal instrumentation system). The two main types of multiplexing are parallel and serial multiplexing. In parallel multiplexing (also called spatial multiplexing), the light source is separated into multiple optical fiber channels, with each channel containing an optical sensor. Typically, each sensor in a parallel multiplexing scheme will have its own detector and processing instrumentation. If the light source is guided to the multiple channels via a fiber optic switch component (as shown in Fig. 14.9), then the optical sensor signals are not monitored simultaneously but sequentially depending on the timing of the optical switch. In this case, a common detector and processing instru- Optical Fiber Strain Gages Optical switch Source Coupler Demodulation and data acquisition Sensor arrays Switch control signal Fig. 14.9 Parallel multiplexing employing an optical switch Coupler #1 #2 #3 #n–1 #n Sensor array Instrumentation Fig. 14.10 Serially multiplexed sensors mentation may be employed for all the sensors. If the optical source is divided to all the sensors via a fiber optic coupler arrangement, then all the fiber optic sensors can be interrogated simultaneously through each sensor’s individual processing instrumentation. Serial multiplexing is a major advantage of some types of optical fiber sensors (in particular the fiber Bragg grating, Sect. 14.5). In serial multiplexing techniques, multiple sensors are positioned along a single optical fiber lead (in serial fashion), as illustrated in Fig. 14.10. The strain signals from each of these sensors are separated through optical means or via the processing instrumentation. The sensing system may also contain a secondary fiber sensor to provide reference and/or gating signals to decouple the serially multiplexed signals. 14.2.4 Limitations of Fiber Optic Sensors In most texts dealing with optical fiber sensors, the reader can find a similar list of advantages as displayed in Sect. 14.2.3. What is often omitted from these texts is a list of their limitations. For the reader (and potential user of these systems), it is imperative that these factors be discussed and explained so the strain sensing community achieves a common understanding of the present limitations of fiber optic strain sensors. The following lists the major limitations of fiber optic sensing technology. The degree to which each limitation affects fiber optic sensing technology is dependent on the particular style of fiber optic sensor 1 1.5 2 2.5 Bend radius (cm) Fig. 14.11 Attenuation due to bend radius for SMF-28 optical fiber with one complete wrap 1550 nm (marks data, solid line theory) • • • • limited bend radius precise alignment of connections cost varying specifications Limited Bend Radius In order to maintain optimum transmission of the optical power, the limited bend radius of the optical fiber must be adhered to. This is much more restricted compared to traditional electrical sensors where bending the electrical wires through a 90◦ turn does not damage the electrical wire. Thus placing optical fiber sensors near physical boundaries requires some consideration. A general rule of thumb is to keep the bend radius greater than 2 cm. When the optical fiber bend radius decreases below 2 cm, the level of attenuation sharply increases, as shown in Fig. 14.11. Manufactures have been developing new classes of optical fiber for various applications. Some of these are considered bend insensitive. Precise Alignment with Connections Unlike traditional electrical sensors, where twisting two copper conductors together is all that is needed to make a connection, optical fiber requires precise alignment of the two end-faces to ensure proper transmission of the optical power. For single-mode optical fiber, the alignment of the adjoining cores is critical. The end-faces of the fiber optic connectors must also be polished and cleaned to ensure low losses at the connection. Issues with misalignment can cause signal loss and also reduce signal-to-noise levels with the introduction of backscat- 353 Part B 14.2 Source Power loss (%) 100 90 80 70 60 50 40 30 20 10 0 0 0.5 14.2 General Fiber Optic Sensing Systems 354 Part B Contact Methods tered light from the interface. Fortunately, a number of fiber optic connectors have become industry standards and these provide quality connections. Part B 14.3 Cost In some cases, the cost of the optical fiber sensor is equivalent to the price of standard telecommunication fiber and components. With the high-volume production of optical fiber and certain optical components, these items are reasonably priced. However, the instrumentation systems required to interrogate these sensors are high-precision electro-optic systems that are relatively expensive, especially compared with RSG instrumentation. Varying Specifications As mentioned earlier, there are multiple choices for instrumentation systems for each fiber optic strain sensor type. This level of variability in the marketplace has led to a lack of industry standards, with companies continuing to market a particular technology to various application areas. This issue is of greatest importance for the fiber Bragg grating sensor discussed in this chapter where various types of the FBG sensors can be used to measure strain with different interrogation techniques. respect to temperature. With RSG sensors, proper selection of materials and fabrication of the foil allows for the manufacture of sensors that are thermally compensated for a particular material. This same process of developing thermally self-compensating sensors cannot be implemented with fiber optic sensors. There is a definite limit on the material property selection for optical fibers, and the fabrication procedures for optical fibers do not allow for the development of material property variations that occur with metal coldworking processes. There are some athermal packages of fiber optic sensors available commercially, but these are not designed for strain measurement. The sensors are bonded into self-contained housings (not suited for bonding to other structures) that are typically composed of materials of differing thermal expansion coefficients, thus counteracting any expansion or contraction due to thermal variations. For the most part, thermal compensation is accomplished by using a secondary sensor to measure the temperature and subtract this effect from the measured strain reading of the fiber optic sensor. Often, the secondary sensor is another fiber optic sensor of similar design, but isolated from the strain field such that the sensor thermal response is well characterized for the application. 14.2.6 Introduction to Strain–Optic Effect 14.2.5 Thermal Effects Most fiber optic sensors also experience issues of thermal apparent strain readings. Extrinsic sensors, whose strain sensing is accomplished in an air gap or comparable structure, may have negligible thermal issues. Intrinsic sensors suffer from thermal influences due to the thermal expansion of the optical fiber and changes in the refractive index with respect to temperature. These quantities are comparable to the elongation and the resistance change of electrical resistance strain gages with Central to many fiber optic strain sensors is the influence of the strain–optic effect. As a mechanical load is applied to the optical fiber, the length of the optical fiber changes and the refractive index also changes as a function of the strain field. The change in the refractive index is similar to the change in specific resistance of electrical strain gages with respect to the applied strain. Similar to the thermal effects, extrinsic sensors do not experience this influence due to the use of an air cavity as the sensing region. 14.3 Interferometry As low-loss optical fiber became more readily available, researchers in the late 1970s and 1980s began exploring the uses of optical fiber as a sensing medium. Some of the first sensors to be explored were optical fiber equivalents of classical interferometers such as those of Michelson and Mach–Zehnder (MZ) [14.1]. Another interferometer, the Sagnac fiber optic sensor, has enjoyed commercial success as the fiber optic gyroscope [14.6]. Many of the basic concepts employed in fiber optic interferometry are discussed in Chap. 18 in this Handbook. Each interferometer discussed in this section makes use of fiber optic components to create two optical paths that interfere to provide a measurement of strain. Similar to full-field strain measurement discussed in other chapters of this Handbook, fiber optic interferometers must extract the strain data from the resulting fringe information. In the case of fiber op- Optical Fiber Strain Gages tic interferometers, the resulting interference signal is detected via a photodetector and transformed to a voltage signal, which is processed via an instrumentation system. 14.3.1 Two-Beam Interference where nd is the optical path length for the wave, ω is the frequency of the light wave, and φ is the phase of the optical wave. Interference is generated by the combination of two coherent light waves with different optical phases. In general, the phase difference is created by having the two optical waves travel different paths forming a path length difference (PLD) as discussed in Chap. 18. The derivation of the intensity response of a general interferometer begins with two coherent light waves that differ by the optical phase term, as given in (14.6). E1 = A1 cos(φ1 − ωt) , E2 = A2 cos(φ2 − ωt) . (14.6) The intensity that results from the interference of these two electric field vectors is given by I = |E1 + E2 |2 . identity: 2 cos(α) cos(β) = cos(α + β) + cos(α − β). Applying these two considerations to (14.8), the intensity function becomes I = Ā + A1 A2 cos(φ1 + φ2 − 2ωt) + . . . A1 A2 cos(φ1 − φ2 ) . I = A21 cos2 (φ1 − ωt) + A22 cos2 (φ2 − ωt) + . . . (14.8) 2A1 A2 cos(φ1 − ωt) cos(φ2 − ωt) . The first two terms in (14.8) are the intensity of the two individual electric field vectors and only contribute an effective constant to the measurement signal as described in Chap. 18. The final term in the above expression can be replaced by the product of cosines (14.9) The second term in (14.8) is at twice the optical frequency and only adds to the constant term in the equation. The final term in (14.8) provides a means for measuring the difference in optical phase between the two coherent light waves with intensity. Equation (14.8) can now be written in the more familiar form for interferometers I = A + B cos(Δφ) , (14.10) where the phase difference (Δφ) contains the strain information of interest in the terms of a PLD. The differences between the types of fiber optic interferometers are based on how the two coherent beams are derived from the system and where the two optical paths are recombined. In classical interferometers, the output of a light source is split via a beam-splitter and the two paths are recombined on a screen. In fiber optic interferometers, the beam splitting and recombining are accomplished within the optical fiber components so that the optical signals are always guided. The interference signal is then coupled to one (or more) photodetectors for signal acquisition and data processing. 14.3.2 Strain–Optic Effect The following derivation examines how the change in optical phase is related to the strain on the optical fiber. Where the stress–optic law discussed in Chap. 25 examines the optical effect on a two-dimensional structure (i. e., plate) under a state of plane stress, the strain–optic effect discussed here examines the influence of a strain state on an optical fiber. Light traveling in an optical fiber can be classified by an optical modal parameter called the propagation constant β given by [14.7] (14.7) Substituting (14.6) into (14.7) yields: 355 β = nkw = 2πn core , λ (14.11) where n core is the refractive index of the core of the optical fiber and kw is the wavenumber associated with the wavelength of light propagating in the optical fiber. The optical phase (φ) of light traveling through a section of optical fiber is given by φ = βL = 2πn core L. λ (14.12) Part B 14.3 As discussed in Chap. 18, the basic equation for the electric vector of a light wave is given by (18.1) and is rewritten here for convenience in (14.4): 2π (14.4) (z − vt) , E1 = A cos λ where A is a vector giving the amplitude and plane of the wave, λ is the wavelength, and v is the wave velocity. Since the wave is traveling in the optical fiber, the refractive index of the core is included in the path length and wave velocity as 2π c nd − t E1 = A cos λ n 2π 2π nd − ct = A cos λ nλ (14.5) = A cos(φ − ωt) , 14.3 Interferometry 356 Part B Contact Methods When a mechanical load is applied to the optical fiber, a phase change is experienced due to changes in the length of the optical fiber (strain) and due to changes to the propagation constant (strain–optic). These influences are the basis for interferometric strain sensors. The change in optical phase (Δφ) can be written as [14.8] Δφ = βΔL + LΔβ . (14.13) Part B 14.3 The influence of the βΔL term is directly related to the strain on the optical fiber (ε), as a change in length and can be written as βεL. The change in the propagation constant is attributed to two sources. One is a waveguide dispersion effect due to the change in the optical fiber diameter. This effect is considered negligible compared to the other influences on the optical phase change [14.8]. The other source is the change in the refractive index, written as dβ (14.14) Δn . LΔβ = L dn The dβ/ dn term is reduced to the wavenumber, kw , but the Δn term is related to the optical indicatrix, Δ[1/n 2 ]2,3 , as [14.8, 9] n3 1 (14.15) , Δn = − Δ 2 2 n 2,3 where light is propagating in the axial direction of the optical fiber as indicated in Fig. 14.12. For the case of small strains, the optical indicatrix is related to the strain on the optical fiber as: 1 = pij ε j , (14.16) Δ 2 n eff i where pij is the strain–optic tensor of the optical fiber and ε j is the contracted strain tensor. In the general case, the strain–optic tensor will have nine distinct elements, termed photoelastic constants related to the normal strains (shear strain is neglected). Fortunately, for a homogenous, isotropic material (such as optical fiber), the strain–optic tensor can be 2 Fig. 14.12 Indicial directions for optical fiber For the case of a surface-mounted sensor, it has been shown that the contracted strain tensor with zero shear strain may be written as [14.2, 9] ⎛ ⎞ 1 ⎜ ⎟ (14.18) ε j = ⎝−ν ⎠ εz , −ν where ν is Poisson’s ratio for the optical fiber. If transverse loads are of interest, they may be incorporated through the strain tensor. Due to the symmetry of the optical fiber, the optical indicatrix is equivalent for i = 2 or 3. Incorporating the contracted strain tensor into (14.16) leads to the following expression for a surfacemounted optical fiber sensor 1 = εz [ p12 − ν( p11 + p12 )] . (14.19) Δ 2 n eff 2,3 After substituting (14.19) into (14.13), the result is Δφ = βεz L − 2π L n 3 εz [ p12 − ν( p11 + p12 )] . λ 2 (14.20) Reducing the above equation leads to: n2 Δφ = βL 1 − [ p12 − ν( p11 + p12 )] εz . 2 (14.21) Normalizing (14.21) by (14.12) yields the standard form of the phase change with respect to the original phase: n2 Δφ = 1 − [ p12 − ν( p11 + p12 )] εz . (14.22) φ 2 14.3.3 Optical Coherence 1 3 represented by two photoelastic constants p11 and p12 as [14.9] ⎞ ⎛ p11 p12 p12 ⎟ ⎜ pij = ⎝ p12 p11 p12 ⎠ . (14.17) p12 p12 p11 Another important concept for fiber optic interferometers is optical coherence. Interference of optical waves can only occur if the two waves are coherent, meaning that the phase information between the two waves is related. When light is emitted from a source, it is composed of coherent packets of energy. Coherence of Optical Fiber Strain Gages Fig. 14.13 Optical source spectrum λ 20 Lc ≈ Δλ 357 Sensing path Δλ λ0 14.3 Interferometry Reference path Fig. 14.14 Basic Mach–Zehnder fiber optic sensor arrangement λ 14.3.5 Michelson Lc ≈ λ20 Δλ . (14.23) 14.3.4 Mach–Zehnder As optical fiber and optical fiber components became available for sensing purposes, the recreation of classic interferometers was the first choice for early researchers. The first reported fiber optic strain gage by Butter and Hocker was an optical-fiber-based Mach– Zehnder [14.8]. In this early version, the output from two optical fibers was recombined on a screen to monitor the resulting interference pattern. As displayed in Fig. 14.14, light from a coherent light source is split into two optical paths via a 3 dB coupler. A second 3 dB coupler is used to recombine the optical paths and create the interference signal. The interference signal passes through both outputs of the second 3 dB coupler to the photodetectors. The use of both outputs is beneficial to some interrogation techniques. The path length difference for the Mach–Zehnder interferometer is dependent on the strain applied to the sensing path. The history of the Michelson interferometer was discussed in the basic optics chapter of this handbook (Chap. 18). Like the Mach–Zehnder interferometer, the Michelson interferometer is constructed in a similar fashion to its classical namesake. The Michelson interferometer, illustrated in Fig. 14.15, uses a 3 dB coupler to divide the optical source into two optical paths. Each optical path has a mirror that reflects the optical signal back to the 3 dB coupler where the two paths interfere. The intensity of the interference signal is then monitored via the photodetector. The reference path is usually isolated from strain, so changes to the sensing path due to the applied load create an optical phase change with respect to the reference path. 14.3.6 Fabry–Pérot The basis of a Fabry–Pérot interferometer is the interference of a light wave reflected from a cavity. French physicists Fabry and Pérot analyzed this type of optical structure in the 19th century. As displayed in Fig. 14.16, the basic fiber optic Fabry–Pérot sensor functions from the interference of a reflection from a first surface, fiber to air, and a second surface, air to fiber. The air–glass interface leads to a reflection of approximately 4% of the input intensity. With such a low reflectivity percentage, the Fabry–Pérot interferometer can be assumed to be a two-beam interferometer. If higher-reflectivity coatings are applied, then a multipass interferometer is created. The particular Fabry–Pérot sensor displayed in Fig. 14.16 is an extrinsic Fabry–Pérot interferometer Sensing path Reference path Fig. 14.15 Basic Michelson fiber optic sensor Part B 14.3 an optical source is a measure of the length of the optical wave packets. Each wave packet can be thought to have a different starting optical phase; thus, light from two different wave packets have different starting phases and will not produce an interference pattern. For very narrow-band sources such as lasers, the coherence length is very long. For broadband optical sources such as light bulbs, the coherence length is very short. In order for an interferometer to function and produce an interference signal, the coherence length of the optical source must be greater than the resulting PLD of the interferometer. A general rule of thumb for determining the coherence length (L c ) of an optical source is to divide the square of the center wavelength (λ0 ) by the bandwidth of the source (Δλ), as given by (14.23). A depiction of the optical source spectrum is shown in Fig. 14.13. 358 Part B Contact Methods signal is generated when the output of the PM fiber is launched back into standard single-mode fiber. It should be noted that PM fiber and PM optical components are also used to make Michelson, Mach– Zehnder, and Fabry–Pérot interferometers with the light being restricted to a single polarization axis. This type of construction alleviates issues due to polarization fading that are common among the optical fiber interferometers. Fig. 14.16 Schematic of a fiber optic Fabry–Pérot sensor 14.3.8 Interrogation of Interferometers Part B 14.3 (EFPI). The sensor is extrinsic because the sensing region is the air cavity formed by the two optical fiber end faces. There are intrinsic Fabry–Pérot sensors where the sensing cavity is formed between partial mirrors fused within an optical fiber. These sensors are much more difficult to fabricate and have not experienced the same commercial availability as the EFPI sensors. The low-finesse interferometer allows for a twobeam interferometer to be assumed for mathematical analysis. With the sensing region taking place in an air cavity for the EFPI, the influence of the strain–optic effect can be neglected and the refractive index is assumed to be equal to 1. Thus, the optical phase change with respect to applied strain can be written as Δφ = βΔL = 4π L FP ε. λ (14.24) For the case of intrinsic Fabry–Pérot sensors, the strain– optic effect will be included in the phase change formulation. 14.3.7 Polarization As discussed in Sect. 14.1.2, there exist specialty optical fibers that are designed to preserve the polarization of light as it is guided along the optical fiber. These polarization-maintaining (PM) fibers induce a birefringence state in the optical fiber where the refractive index along the two principal axes of polarization are different. Thus, light traveling in the optical fiber will travel faster along the lower index axis (fast axis) than the higher-index axis (slow axis). In the simplest form, the polarization interferometer functions by taking light from a standard single-mode optical fiber and launches it into a length of PM fiber such that light travels down both the fast and slow axes. These two paths represent the two paths for the interferometer. Any applied load to the PM fiber will induce an optical phase change between the two paths due to the difference in the refractive index between the two axes. The interference Phase Detection There exist a number of means for interrogating interferometric fiber optic strain sensors. Each of these techniques is focused on extracting the optical phase change from the cosine function given in (14.10). Some techniques use an active modulation scheme where the reference path is perturbed by a sinusoidal or ramp function. Other forms of modulation include varying the optical source wavelength (optical frequency modulation). These active modulation techniques have different attributes in terms of frequency response and strain range. The selection of the phase detection technique is dependent on the selected interferometric sensor and the sensing application requirements. Passive interrogation techniques for interferometric sensors are concerned with optically deriving a sin(Δφ) and cos(Δφ) function from the sensing system. Then, an arctangent operation can be used to obtain the phase change. Another form of phase detection is based on low-coherence interferometry, also called white-light interferometry. These techniques use a broadband optical source with a coherence length shorter than the optical path difference (OPD) of the interferometer. In order to achieve an interference signal, the optical output of the sensing interferometer is passed through a reference interferometer, also called a readout interferometer. The resulting OPD between the two Intensity Strain Time Fig. 14.17 Phase ambiguity with respect to change in strain (dashed line) Optical Fiber Strain Gages interferometers is less than the coherence length of the optical source leading to an interference signal. The reference interferometer can then be tuned to the sensing interferometer as a means of interrogation. This method is advantageous because it allows the use of lower cost and longer-life broadband optical sources such as lightemitting diodes (LEDs). Phase Ambiguity One of the main difficulties with interferometric detection is the issue of phase ambiguity. As an increasing strain field is applied to the sensor, the intensity signal is a cosine function given by (14.10). With this 14.4 Scattering type of response, differentiating between an increasing strain and a decreasing strain is difficult. As shown in Fig. 14.17, the intensity response looks the same for increasing and decreasing strain. The crossover point can be identified by the change in the intensity pattern. If this change occurs at a maximum or minimum of the intensity function the crossover becomes difficult to detect. Some interrogation techniques limit the intensity response (and thus the phase change) to a single linear portion of the cosine function. Other interrogation techniques must account for multiple fringes in the response function. This is similar to fringe counting as discussed in Chap. 25. where Va is velocity of sound in the optical fiber and λ is the free-space operating wavelength. The Brillouin frequency shift is dependent on the density of the optical fiber, which is dependent on strain and temperature. 14.4.2 Strain Sensing Using Brillouin Scattering Measuring strain with Brillouin scattering requires the generation of a narrow-line-width short-duration pulse. As the pulse travels down the optical fiber, scattering occurs at the Brillouin frequency. The location along the optical fiber is determined by gating the optical pulse and measuring the time of flight of the returned scattered signal. The standard spatial resolution for these sensor systems is approximately 1 m. This level Intensity E = hν = hc λ 3 – 5 orders of magnitude Anti-Stokes 14.4.1 Brillouin Scattering Brillouin scattering is a result of the interaction between light propagating in the optical fiber and spontaneous sound waves within the optical fiber. Because the sound wave is moving, the scattered light is Doppler-shifted to a different frequency. The Brillouin frequency shift (ν B ) is determined by: Temperature dependent Pump wavelength Rayleigh scattering Brillouin scattering Fluorescence Stokes Raman scattering Wavelength Fig. 14.18 Picture of various scattering effects with respect to the νB = 2nVa /λ , (14.25) pump wavelength Part B 14.4 14.4 Scattering As light travels through optical fiber, scattering mechanisms displayed in Fig. 14.18 come into play that can be exploited for sensing purposes. In order to achieve measurable scattering effects, a relatively high-powered optical source is required. Rayleigh scattering results from elastic collision between the phonons and the optical fiber material. The scattered wavelength (which will be guided back towards the optical source) is equivalent to the pump or excitation wavelength. Rayleigh scattering is extensively used with pulsed laser sources and time-gated detectors to monitor long lengths of optical fiber for attenuation sources such as connections, optical splices, tight bend locations, and breaks in the optical fiber. Raman scattering involves the inelastic scattering effect of the optical energy interacting with molecular bonds. Raman scattering is used extensively for distributed temperature monitoring. The Raman scattering effect is immune to strain effects, making it ideal for these applications. Brillouin scattering results from reflections from a spontaneous sound wave propagating in the fiber [14.10]. The Brillouin scattering effect is sensitive to both strain and temperature variations. 359 360 Part B Contact Methods a) Brillouin gain 1.016 1.012 Data acquisition 1.008 1.004 1 0.996 –100 0 Part B 14.4 100 200 300 400 500 600 700 Position along the fibre (m) b) Gain 1.02 1.01 1 12.7 c) Brillouin frequency shift (GHz) 12.83 12.82 12.81 12.8 12.79 12.78 12.77 12.76 12.75 12.74 –100 0 12.75 12.8 12.85 700 600 500 Distance (m) 400 300 200 100 12.9 0 Frequency (GHz) Data analysis 100 200 300 400 500 600 700 Position along the fibre (m) Fig. 14.19 Illustration of a typical measurement from a Brillouin scattering system [14.11] (Courtesy of OmniSensTM ) of spatial resolution is adequate for applications on long structures such as pipeline monitoring. In order to construct the Brillouin frequency response along the optical fiber, the optical pulse is scanned through the frequency range of interest. In this manner, each pulse provides a measurement of the scattered intensity along the optical fiber at a particular frequency, as depicted in Fig. 14.19. A three-dimensional plot is generated with respect to optical frequency and position along the optical fiber, with each pulse producing an intensity response at a single frequency. The Brillouin frequency shift is then computed from this plot by taking the max- imum gain intensity at each position along the optical fiber. Since Brillouin scattering is sensitive to both strain and temperature, a means for performing temperature compensation is required. In most cases, a secondary optical fiber is deployed in a loose tube fashion such that no strain is coupled and temperature can be monitored along this fiber and compensated for in the strain measurement, as discussed in Sect. 14.2.5. In these cases, an alternative optical method such as Raman scattering may also be used to monitor the temperature. Optical Fiber Strain Gages 14.5 Fiber Bragg Grating Sensors 361 14.5 Fiber Bragg Grating Sensors Because of these difficulties, sensing applications with fiber gratings were not realized until a novel manufacturing technique was developed in 1989 [14.15]. In the case of Bragg grating formation, the change of the refractive index of an optical fiber is induced by exposure to intense UV radiation (typical UV wavelengths are in a range from 150 to 248 nm). To create a periodic change in the refractive index, an interference pattern of UV radiation is produced such that it is focused onto the core region of the optical fiber. The refractive index of the optical fiber core changes where the intensity is brightest in the interference pattern to produce a periodic refractive index profile [14.12]. The manner in which the refractive index change is induced has been linked to multiple mechanisms over the past 15 years of research. The various mechanisms are dependent on the intensity of the UV source, the exposure time, and any pretreatment of the optical fiber. The resulting FBG from each mechanism displays slightly different properties related to the optical spectrum response, sensitivity to strain and temperature, and performance at elevated temperatures. The most common type of FBG sensor (type I) is a result of short exposure times to relatively low-intensity UV radiation to standard telecommunications optical fiber. There exist other FBG sensor types (type IIa, type II, chemical composition) that have been researched for sensing applications. The discussion in this section will be limited to the standard type I FBG sensor. The length of a single period for the grating structure is called the grating pitch, Λ, as shown in Fig. 14.20. The pitch of the grating is controlled durCladding 14.5.1 Fabrication Techniques Fiber Bragg grating sensors are unique compared to the strain sensors discussed thus far in this chapter. The sensors up to this point have been fabricated from a combination of standard optical fibers or making use of inherent properties of optical fibers. The FBG sensor is fabricated by altering the refractive index structure of the core of an optical fiber. The refractive index change is made possible in standard communications-grade optical fiber due to a phenomena known as photosensitivity discovered in 1978 by Hill [14.13]. Hill-type gratings are limited in practicality because they reflect the UV wavelength that was used to induce the refractive index change and are difficult to manufacture. Λ Core n neff ncore z Fig. 14.20 Schematic of an FBG sensor Part B 14.5 Fiber Bragg gratings (FBGs) have become an integral part of the telecommunications hardware, used in applications such as add/drop filters, fiber lasers, and data multiplexing [14.12]. Since the discovery of the photosensitive effect in optical fiber [14.13] by which ultraviolet (UV) light is used to induce a permanent change in the refractive index of optical fiber, researchers have been discovering new applications for this unique optical phenomena. Because of the widespread use and development of this technology, several textbooks dedicated to FBGs and their applications have been published [14.2, 12, 14]. In recent years, the FBG sensor has become a popular choice for fiber optic sensor applications for many reasons. As will be seen in this section, the optical response of the FBG sensor is wavelength-encoded, allowing many FBG sensors to be serially multiplexed via wavelength division multiplexing techniques. FBG sensors may also be time division multiplexed, allowing approximately 100 FBG sensors to be monitored along a single fiber strand. There exist other multiplexing techniques for FBG sensors that allow thousands of FBG sensors to be serially multiplexed and monitored with a single instrumentation system. Many commercially available interrogation systems have also come to market, making these types of sensors and systems readily available and relatively less complex for the enduser. Unfortunately, many of these systems use different technologies for interrogation of the FBG sensors, leading to different sensor specifications as well as different requirements for the fabrication of the FBG sensor. An overview of the different interrogation technologies is presented in Sect. 14.5.5. 362 Part B Contact Methods Lloyd mirror Focused UV Focused UV Optical fiber Phase mask Optical fiber –1 Interference pattern +1 Interference pattern Reflective surface Fig. 14.21 Phase mask fabrication technique Part B 14.5 ing the manufacturing process, and is typically on the order of ≈ 0.5 μm, while the amplitude of the index variation is only on the order of 0.01–0.1% of the original refractive index [14.12]. Many methods are used to create the periodic interference pattern (e.g., phase masks [14.16], Mach–Zehnder interferometers [14.15], and Lloyd mirrors [14.14]). Phase masks are corrugated silica optical components, as shown in Fig. 14.21. As laser radiation passes through the phase mask, the light is divided into different diffraction beams. The diffracted beams in the example are the +1 and −1 beams, which create an interference pattern that is focused on the optical fiber core. Other orders of diffraction are designed to be minimized in these passive optical devices. The interference pattern induces a periodic refractive index change along the exposed length of the optical fiber, thus creating the Bragg grating. Phase masks provide a stable, repeatable interference pattern and are used for high-volume production of FBG sensors at a common wavelength. The Mach–Zehnder interferometer technique uses a bulk optic Mach–Zehnder interferometer to create an interference pattern on the optical fiber. The optical arrangement for a basic Mach–Zehnder technique is shown in Fig. 14.22. Light from the laser emission is passed through focusing optics and is split via an optical beam splitter. The two divergent laser beams are Focused UV Mirror Interference pattern Beam splitter Mirror Optical fiber Fig. 14.22 Mach–Zehnder interferometer technique Fig. 14.23 Lloyd mirror technique redirected via mirrors to combine at the optical fiber location, resulting in an interference pattern from the combination of these two optical paths. The focusing optics are designed to focus the laser energy at the optical fiber location. This technique allows for the fabrication of FBG sensors with different pitches through adjustment of the angle of incidence of the interference beams. The Lloyd mirror technique uses a Lloyd mirror to create an interference pattern on the optical fiber, as shown in Fig. 14.23. Light from the laser emission is passed through focusing optics and transmitted to the Lloyd mirror arrangement. The Lloyd mirror causes the input laser beam to split into two beams. These beams are recombined and focused at the optical fiber location, which creates an interference pattern and forms the FBG sensor. 14.5.2 Fiber Bragg Grating Optical Response Light traveling in an optical fiber can be classified by an optical modal parameter β, as discussed in Sect. 14.3 and given by [14.7] 2πn core (14.26) , λ where n core is the refractive index of the core of the optical fiber and kw is the wavenumber associated with the wavelength of light propagating in the optical fiber. The function of the Bragg grating is to transfer a forward-propagating mode (β1 ) into a backwardpropagating mode (−β1 ) (i. e., reflect the light) for a particular wavelength meeting the phase-matching criterion. The phase-matching condition is derived from coupled-mode theory and is given by β = nkw = βi − β j = 2πm , Λ (14.27) Optical Fiber Strain Gages 14.5 Fiber Bragg Grating Sensors 363 Fig. 14.24 Schematic of an FBG sensor with reflected and transmitted spectra Reflected spectrum Incoming spectrum Transmitted spectrum Λ λB = 2n eff Λ , (14.29) (ΔT ), so that the Bragg wavelength shifts to higher or lower wavelengths in response to applied thermalmechanical fields. For most applications, the shift in the Bragg wavelength is considered a linear function of the thermal-mechanical load. The treatment of FBG sensors here will ignore the thermal effects, because the thermal effects can be modeled as an independent response of the Bragg grating. The shift in the Bragg wavelength due to an incremental change of length (ΔL) is given by [14.9]: ∂Λ ∂n eff (14.30) + n eff ΔL . ΔλB = 2 Λ ∂L ∂L Assuming that the strain field is uniform across the Bragg grating length (L), the term ∂Λ/∂L can be replaced with Λ/L. Likewise, the term ∂n eff /∂L can be replaced by Δn eff /ΔL in (14.30). The terms Λ and L are physical quantities that are determined by the interference pattern formed during fabrication and are known. The change in the effective refractive index (Δn eff ) can be related to the optical indicatrix, Δ[1/n 2eff ], as discussed in Sect. 14.3 [14.9]. Δn eff = − n 3eff 1 . Δ 2 2 n eff (14.31) where the subscript ‘B’ defines the wavelength as the Bragg wavelength. Equation (14.29) states that, for a given pitch (Λ) and average refractive index (n eff ), the wavelength λB will be reflected from the Bragg grating, as illustrated in Fig. 14.24. In the case of small strains, the optical indicatrix is related to the strain on the optical fiber as: 1 Δ 2 = pij ε j , (14.32) n eff i 14.5.3 Strain Sensing Using FBG Sensors where pij is the strain–optic tensor of the optical fiber and ε j is the contracted strain tensor. The directions associated with the indices in (14.32) are illustrated in Fig. 14.25. In the general case, the strain–optic tensor will have nine distinct elements, termed photoelastic constants. Fortunately, for a homogenous, isotropic material (such as optical fiber), the strain–optic tensor can be represented by two photoelastic constants p11 and Bragg gratings operate as wavelength selective filters reflecting the Bragg wavelength λB which is related to the grating pitch Λ and the mean refractive index of the core, n eff , given by (14.29). Both the effective refractive index (n eff ) of the core and the grating pitch (Λ) vary with changes in strain (ε) and temperature Part B 14.5 where m is an integer value representing the harmonic order of the grating [14.12]. From (14.27), multiple propagation modes can be used to satisfy the phasematching condition. Therefore, a single FBG sensor will reflect multiple wavelengths with respect to the order of the integer m. Most optical sources do not have a large enough bandwidth to excite multiple wavelength reflections from a single FBG sensor. Some research has been performed using multiple optical sources to excite more than the first-order Bragg condition as a means of providing a temperature and strain measurement with a single FBG sensor [14.17]. For all practical applications to date, only the first-order Bragg condition of m = 1 is employed for standard uniform Bragg gratings. For this case, the backward-propagating mode (−β1 ) is substituted into (14.27) for β j , and the following equation is derived π 2πn eff (14.28) β1 = = . Λ λ The index of refraction is now noted as an effective (or average) refractive index, n eff , for the Bragg grating, due to the periodic change across the length of the optical fiber. Solving (14.28) for the wavelength (λ) provides the Bragg wavelength equation 364 Part B Contact Methods 2 1 3 Fig. 14.25 Indicial directions for an FBG optical fiber Part B 14.5 p12 as [14.9]: ⎞ ⎛ p11 p12 p12 ⎟ ⎜ pij = ⎝ p12 p11 p12 ⎠ . p12 p12 p11 Transverse loads may also be modeled using this analysis by choosing the proper contracted strain tensor formulation. For example, the measurement of a uniform pressure field on the optical fiber can be modeled with the following contracted strain tensor ⎛ ⎞ εz ⎜ ⎟ (14.39) ε j = ⎝εr ⎠ . εr Using (14.39) in the above formulation leads to the following response function for the FBG sensor response: (14.33) For the case of a surface mounted FBG sensor, it has been shown that the contracted strain tensor may be written as [14.2, 9]: ⎛ ⎞ 1 ⎜ ⎟ (14.34) ε j = ⎝−ν ⎠ εz , −ν where ν is Poisson’s ratio for the optical fiber. Due to the symmetry of the optical fiber, the optical indicatrix is equivalent for i = 2 or 3. Incorporating the strain tensor into (14.32) leads to the following expression for a surface-mounted optical fiber sensor: 1 (14.35) Δ 2 = εz [ p12 − ν( p11 + p12 )] . n eff n2 ΔλB = εz − eff [εz p12 + εr ( p11 + p12 )] . λB 2 (14.40) For other strain states, such as diametric loading, the analysis of the FBG response becomes more complicated. This is due to the induced polarization effects and nonuniform loading on the optical fiber. In extreme diametric loading cases, the FBG sensor response will split due to the reflection from the different polarization axis and nonuniform strain on the grating structure [14.18]. Research is continuing in this field for the development of transversely sensitive FBG sensors with emphasis on using FBG sensors written into PM optical fiber to take advantage of the pre-existing polarization properties. 14.5.4 Serial Multiplexing One of the main advantages of FBG sensors is the ability to measure multiple physical parameters. This ability combined with serial multiplexing of FBG sensors alAfter substituting (14.31) and (14.35) into (14.30), the lows for multiple parameters to be monitored not only result is by a single instrument, but also with all the data transmitted on a single piece of optical fiber [14.19]. This is n3 ΔλB = −2Λ eff εz [ p12 − ν( p11 + p12 )] + 2n eff Λεz . advantageous in applications where minimal intrusion 2 into an environment is required. Many applications of (14.36) FBG sensors concern measuring strain and/or temperNormalizing (14.36) by the Bragg wavelength demon- ature. In these applications, many sensors (from fewer strates the dependence on the wavelength shift of the than ten to over a hundred) are multiplexed to provide FBG sensor on the refractive index, strain–optic coeffi- measurements across the structure. Examples of these applications are monitoring civil infrastructure [14.20], cients, and Poisson’s ratio for the optical fiber. naval/marine vessels [14.21, 22], and shape measure n 2eff ΔλB ment of flexible structures [14.23]. = 1− [ p12 − ν( p11 + p12 )] εz . (14.37) The ability to serially multiplex FBG sensors is arλB 2 guably the most prominent advantage of this sensor The terms multiplying the strain in (14.37) are constant type. With over a decade of development, researchers over the strain range of the Bragg grating, and (14.37) have devised three main techniques for serial multiis often written in simplified form as plexing FBG sensors: wavelength-domain multiplexing (WDM), time division multiplexing (TDM), and optical ΔλB = Pe εz . (14.38) frequency-domain reflectometry (OFDR). λB Optical Fiber Strain Gages Wavelength response of FBG #2 Λ1 Λ2 Λ3 Thermomechanical load Reflected spectrum Fig. 14.26 Serial multiplexing of FBG sensors from which sensor. The default is to place the lowerwavelength signal with the previous lower-wavelength sensor signal. If the lower-wavelength signal completely overwhelms the higher-wavelength signal, as is depicted in Fig. 14.27, then the instrumentation records the wrong wavelength data for each of these sensors after the overlap event. Knowledge of the potential measurement range for each sensor is required to prevent sensor overlap issues. When this is not provided, conservative estimates should be used to select sensor wavelength separations. In cases where the FBG sensors are expected to experience similar responses, such as thermal measurements, the wavelength spacing of the sensors may be more tightly spaced. Time Division Multiplexing TDM uses a time-of-flight measurement to discriminate the FBG sensors on a fiber. The basic TDM architecture is shown in Fig. 14.28. A light source generates a short pulse of light that propagates down the optical fiber to a series of FBG sensors. All the FBG sensors initially have the same wavelength and reflect only a small portion of the light at the Bragg wavelength. The light source has a somewhat wide spectrum that is centered about the unstrained Bragg wavelength of the FBG sensors. Each FBG sensor generates a return pulse when it reflects the light at its particular Bragg wavelength, which is dependent upon the strain/temperature state of the FBG. For the example shown in Fig. 14.28, there are five return pulses generated by the FBG sensors. These pulses propagate back along the fiber and are coupled to a high-speed photodetector, which measures when each pulse was detected. The FBG sensors must be separated by a minimum distance along the fiber to allow accurate measurement of the difference in arrival time of the reflected pulses, often by 1 m or more. An additional wavelength measurement system is needed to measure the Bragg wavelength of each sensor as it is detected. TDM systems have become more popular in ε, T λB1 λB2 λB1 λB2 λ λ ε, T FBG #1 FBG #2 λB2 λB1 λ Fig. 14.27 Wavelength shift description and sensor overlap for WDM systems 365 Part B 14.5 Wavelength Division Multiplexing As shown in Fig. 14.26, WDM is accomplished by producing an optical fiber with a sequence of spatially separated gratings, each with different grating pitches, Λi = 1, 2, 3, . . .. The output of the multiplexed sensors is processed through wavelength selective instrumentation, and the reflected spectrum contains a series of peaks, each peak associated with a different Bragg wavelength given by λi = 2nΛi . As indicated in the figure, the measurement field at grating 2 is uniquely encoded as a perturbation of the Bragg wavelength λ2 . Note that this multiplexing scheme is completely based on the optical wavelength of the Bragg grating sensors. The upper limit to the number of gratings that can be addressed in this manner is a function of the optical source profile width and the expected strain range. WDM was the first form of multiplexing explored for the FBG sensor and is common in commercially available systems. A major concern for WDM FBG systems is sensor wavelength overlap. This occurs when two neighboring sensors in wavelength space experience loads that cause the reflected Bragg wavelengths from each sensor to approach each other and overlap. During the overlap event, the monitoring instrumentation cannot process both sensors independently and will record only a single senor. After the overlap event when both sensors can be resolved, the instrumentation cannot distinguish which of the reflected Bragg wavelengths is 14.5 Fiber Bragg Grating Sensors 366 Part B Contact Methods I (t) Light source FBG sensors t S-1 S-2 S-3 S-4 S-5 3 dB coupler I (t) t1 Detector 1 t2 t3 t4 t5 t Time delayed signals from each of the sensors Light dump Fig. 14.28 Time division multiplexing technique Part B 14.5 recent years due to advances in optical sources, data acquisition, and computing technology. determine the Bragg wavelength of that particular sensor [14.24]. Optical Frequency-Domain Reflectometry OFDR is a more complex interrogation technique then the other systems discussed. The basic concept of an OFDR system is to create a response function from the serially multiplexed FBG sensors that spatially isolates the individual FBG sensor response in the frequency domain of the system response [14.24]. A schematic of a typical OFDR optical system layout is shown in Fig. 14.29. Light from a tunable laser source is coupled into an optical arrangement containing the FBG sensors and a reference interferometer. The reference interferometer is used to trigger the sampling on detector 1, thus ensuring that the response from the FBG sensors is sampled at a constant wavenumber interval. The FBG sensors are low reflectivity, typically at a common wavelength, and can be very closely spaced (on the order of 1 cm). Each sampled set of FBG data from detector 1 is initially processed via a Fourier transform. The frequency spectrum obtained from the Fourier transform displays a collection of peaks at the physical distance of each FBG along the sensing array. Each peak is then bandpass filtered in the frequency spectrum and an inverse Fourier transform is used to 14.5.5 Interrogation of FBG Sensors, Wavelength Detection As the popularity of using FBG sensors has grown in recent years, so has the number of techniques to perform interrogation. Of course, the method of interrogation used depends on the type of multiplexing, the sampling rate requirements, and the measurement resolution requirements. The existing technologies for interrogating FBG sensors include scanning lasers, tunable filters, linear optical filters, and spectroscopic techniques. The end-user must ensure the instrumentation system chosen meets the requirements for the application. In terms of sampling rate, the response of FBG sensors is very fast; unfortunately, all instrumentation systems are limited by the speed of the electronics, data acquisition, and data processing. This is especially true for serially multiplexed systems. In most cases, the sampling rate for individual sensors is not selectable. The instrumentation records data at one sampling rate for all sensors interrogated by the system. The end-user may design additional data-acquisition software to store the data into separate files with different Detector 1 Coupler Tunable laser source FBG sensors Coupler R Li Coupler R R Detector 2 Lref Reference interferometer Fig. 14.29 Schematic of typical OFDR interrogation design Optical Fiber Strain Gages 14.5.6 Other Grating Structures Although not commonly employed for the measurement of strain, there do exist other fiber grating structures that are founded upon the variation of refractive index in the core of the optical fiber. Two structures in particular are the chirped grating and the long-period grating. The chirped grating is an FBG with a linear variation of the periodicity (Λ(z)). Without a uniform periodicity, the reflected spectrum of the grating structure becomes spread out. This device is commonly used as a linear wavelength filter. Although the chirped grating has similar sensitivities to strain as the FBG sensor, it is almost never used as such because of the difficulty in obtaining a wavelength peak detection value, as well as increased fabrication costs. A long-period grating (LPG) is similar to the Bragg grating such that the periodicity is uniform, but the periodicity is much larger, on the order of Λ = 1 mm. Examination of the phase-matching condition (14.27) for the Bragg grating permits interaction between allowable optical modes (β y i). For the case of an LPG, the forward-propagating optical mode in the core is transferred to cladding modes, which are attenuated from the optical fiber, leading to wavelength-dependent attenuation in the transmitted signal. Once the light gets beyond the LPG structure, the various modes are coupled back to the core region and the wavelength attenuated signal is transmitted through the optical fiber. The attenuation mechanisms for the cladding modes are highly dependent on the boundary condition of the cladding. Therefore, LPG sensors have been examined for potential use as chemical sensors with the presence of a chemical species inducing wavelength attenuation shifts [14.25]. 14.6 Applications of Fiber Optic Sensors Although fiber optic strain sensors are not widely used in any industry, they have been tested in many applications. Often, these applications have attributes that make traditional sensing technologies ineffectual or difficult to implement. With advantages of long lead lines, embedding, and multiplexing, fiber optic sensors have proven their ability to perform measurements were traditional sensors cannot. The following sections provide a brief overview of some of the applications. References [14.2, 3, 14] provide additional applications. With new applications of fiber optic sensors occurring regularly, a quick internet search will provide information on the latest and greatest applications. 14.6.1 Marine Applications In a marine environment, fiber optic sensors have the advantage of not requiring extensive waterproof- ing for short-duration tests. Also, many sensors can be multiplexed to achieve a high sensor count for large structures such as ships and submarines. FBG sensors have been used to monitor wave impacts and loads on surface ships [14.21, 22] and for American Bureau of Shipping (ABS) certification of a manned submarine [14.26]. 14.6.2 Oil and Gas Applications Fiber optic sensing in the oil and gas industry has made great strides in recent years. Of particular use are distributed temperature sensing (DTS) systems based on Raman scattering [14.27, 28]. For structural monitoring, the area that has witnessed use of fiber optic strain sensors is in riser monitoring. Risers are long structural components used on offshore platforms for many applications including drilling, water injection, and col- 367 Part B 14.6 sampling rates. Alternatively, postprocessing the oversampled data through averaging or filtering techniques may provide improved results. Some WDM multiplexed instrumentation systems require the FBG sensors to have certain nominal wavelengths to fit within the instrument’s wavelength filtering windows. This limits the flexibility of the wavelength spacing discussed previously and limits the number of FBG sensors that can be serially multiplexed with these instrumentation systems. Other WDM instrumentation systems allow for the wavelength spacing of neighboring FBG sensors to be less than 1 nm, allowing for over 100 FBG sensors to be serially multiplexed, but issues of sensor wavelength overlap must be taken into consideration. For TDM and OFDR systems, the special design of the FBG sensors in terms of low reflectivity and physical spacing requirements limits the availability of commercial vendors to provide these sensing arrays. The end-user can typically only purchase FBG sensors that function with the systems from the system vendors themselves. 14.6 Applications of Fiber Optic Sensors 368 Part B Contact Methods lection of the oil. Tidal currents are known to induce vibrations in these structures, and fiber optics sensors have recently been used to gain an understanding of the structural effects the vibrations have on the risers [14.29, 30]. 14.6.3 Wind Power Applications Part B 14.7 With the upswing in oil prices, more effort is being expended on research and development of alternative energy sources, including renewable energy sources such as wind power. Wind power turbines are large structures that make use of composite blades. Wind turbine manufacturers are examining ways in which to improve the efficiency of these devices and are turning to fiber optic sensing to provide load monitoring and control data [14.31]. Other wind turbine applications using fiber optic sensors include health monitoring of the wind turbine [14.31] and shape measurement of the wind turbine blade [14.32]. Attributes that make fiber optic sensors attractive for these applications include multiplexing, immunity to radiofrequency (RF) noise, and immunity to damage caused by the electrical charge from lightning strikes. Fiber optic sensors also have the ability to be embedded into these composite structures. 14.6.4 Civil Structural Monitoring The application area that has received the most interest in fiber optic strain sensing has been civil structure monitoring. Using the attributes of long lead fibers and sensor multiplexing, monitoring the loads and structural health of these large structures has been an ideal application for fiber optic sensors. FBG sensors have been used to monitor bridge structural components including bridge decks [14.2], composite piles [14.20], and stay-cables [14.2]. In each of these cases, the FBG sensor has been embedded into the structural component. As the growth of FBG sensor applications continues, many more examples of FBG sensors being employed in civil structural monitoring will be realized. Michelson-interferometer-based systems have also been used for civil structural monitoring purposes including bridges [14.33], dams [14.34], and buildings [14.35]. The Michelson interferometer has a long sensor gage length, which is beneficial when measuring the deformation of these large structures. The reference arm of the Michelson can also be packaged alongside the sensing arm, allowing for straightforward temperature compensation of the measured signal. Brillouin scattering sensor systems are also finding applications in civil structures, including pipeline monitoring and dam monitoring [14.36]. Again, the ease of embedding these sensors directly into the structure is leveraged for these applications. Furthermore, the distributed sensing nature of Brillouin-based sensors allows for a strain measurement at approximately every meter along the optical fiber, making Brillouin scattering appealing for these applications. 14.7 Summary Optical fiber sensing is a growing field, full of potential. In general, fiber optic sensing is a viable technology for strain monitoring applications in many industries. Advancement of this technology is dependent on leveraging its many advantages over traditional sensors. Arguably the key aspect of fiber optic sensors is the ability to multiplex many sensors or obtain distributed measurements using a single strand of optical fiber. This allows for a vast number of sensing points with reduced cabling weight and minimal intrusion into the application environment. The ability to embed optical fiber into composite structures is also a primary driver for fiber optic sensing in the composite structures field. As discussed in this chapter, there are many different types of fiber optic strain sensors. Some of these sensor types have their roots in classical optics, while others take advantage of the special properties of optical fiber. There exist single-point sensors (EFPI, FBG), cumulative strain sensors (MZ, Michelson), and distributed strain sensors (Brillouin). The selection of a strain measurement technique is highly dependent on the application requirements. Just as there are multiple types of fiber optic strain sensors, there are often multiple interrogation techniques for each of the sensor types. This leads to a very complex design space for the novice engineer tasked with selecting the appropriate sensing system. Careful consideration of measurement resolution, sampling rate, and number of sensors will assist in making an informed decision. Optical Fiber Strain Gages References 369 References 14.1 14.2 14.3 14.4 14.5 14.7 14.8 14.9 14.10 14.11 14.12 14.13 14.14 14.15 14.16 14.17 14.18 14.19 14.20 14.21 14.22 14.23 14.24 14.25 14.26 14.27 14.28 14.29 14.30 14.31 14.32 14.33 14.34 C. Baldwin, T. Salter, J. Niemczuk, P. Chen, J. Kiddy: Structural monitoring of composite marine piles using multiplexed fiber Bragg grating sensors: Infield applications, Proc. SPIE 4696, 82–91 (2002) A. Kersey, M.A. Davis, T.A. Berkoff, A.D. Dandridge, R.T. Jones, T.E. Tsai, G.B. Cogdell, G.W. Wang, G.B. Havsgaard, K. Pran, S. Knudsen: Transient load monitoring on a composite hull ship using distributed fiber optic Bragg grating sensors, Proc. SPIE 3042, 421–430 (1997) C. Baldwin, J. Kiddy, T. Salter, P. Chen, J. Niemczuk: Fiber optic structural health monitoring system: Rough sea trials testing of the RV Triton, MTS/IEEE Oceans 2002, 3, 1807–1814 (2002) C. Baldwin, T. Salter, J. Kiddy: Static shape measurements using a multiplexed fiber Bragg grating sensor system, Proc. SPIE 5384, 206–217 (2004) B.A. Childers, M.E. Froggatt, S.G. Allison, T.C. Moore Sr., D.A. Hare, C.F. Batten, D.C. Jegley: Use of 3000 Bragg grating strain sensors distributed on four eight-meter optical fibers during static load tests of a composite structure, Proc. SPIE 4332, 133–142 (2001) Z. Zhang, J.S. Sirkis: Temperature-compensated long period grating chemical sensor, Techn. Dig. Ser. 16, 294–297 (1997) J. Kiddy, C. Baldwin, T. Salter: Certification of a submarine design using fiber Bragg grating sensors, Proc. SPIE 5388, 133–142 (2004) P. E. Sanders: Optical Sensors in the Oil and Gas Industry, Presentation for New England Fiberoptic Council (NEFC) (2004) P. J. Wright, W. Womack: Fiber-Optic Down-Hole Sensing: A Discussion on Applications and Enabling Wellhead Connection Technology, Proc. of Offshore Technology Conference, OTC 18121 (2006) D.V. Brower, F. Abbassian, C.G. Caballero: Realtime Fatigue Monitoring of Deepwater Risers Using Fiber Optic Sensors, Proc. of ETCE/OMAE2000 Joint Conference: Energy for the New Millennium, ASME OFT-4066 (2000) pp. 173–180 L. Sutherland: Joint Industry Project, DeepStar, www.insensys.com, Insensys Ltd. (2005) Insensys: Wind Energy, www.insensys.com (2006) J. Kiddy: Deflections Measurements of Wind Turbine Blades using Fiber Optic Sensors, WINDPOWER 2006, Poster (2006) A. Del Grosso, D. Inaudi: European Perspective on Monitoring-Based Maintenance, IABMAS ’04, International Association for Bridge Maintenance and Safety (Kyoto 2004) P. Kronenberg, N. Casanova, D. Inaudi, S. Vurpillot: Dam monitoring with fiber optics deformation sensors, Proc. SPIE 3043, 2–11 (1997) Part B 14 14.6 K.T.V. Grattan, B.T. Meggitt (Eds.): Optical Fiber Sensor Technology, Vol. 1-4 (Kluwer Academic, Boston 1995) R.M. Measures: Structural Monitoring with Fiber Optic Technology (Academic, San Diego 2001) E. Udd (Ed.): Fiber Optic Smart Structures (Wiley, New York 1995) Crystal Fibre, Birkerød, Denmark (www.crystalfibre.com) N.G. Skinner, J.L. Maida Jr.: Downhole fiber-optic sensing: The oilfield service provider’s perspective, SPIE Fiber Optic Sensor Technol. Appl. III, 5589, 206–220 (2004) B. Culshaw: Optical fiber sensor technologies: Opportunities and – perhaps – pitfalls, J. Lightwave Technol. 22(1), 39–50 (2004) A. Snyder, J. Love: Optical Waveguide Theory (Chapman Hall, New York 1983) C.D. Butter, G.B. Hocker: Fiber optics strain gauge, Appl. Opt. 17(18), 2867–2869 (1978) S.M. Melle, K. Lui, R.M. Measures: Practical fiberoptic Bragg grating strain gauge system, Appl. Opt. 32, 3601–3609 (1993) B. Hitz: Fiber sensor uses Raman and Brillouin scattering, Photonics Spectra 39(7), 110–112 (2005) M. Nikles: Omnisens, www.omnisens.ch R. Kashyap: Fiber Bragg Gratings (Academic, San Diego 1999) K.O. Hill, Y. Fujii, D.C. Johnson, B.S. Kawasaki: Photosensitivity in optical fiber waveguides: Application to reflection filter fabrication, Appl. Phys. Lett. 32, 647–649 (1978) A. Othonos, K. Kalli: Fiber Bragg Gratings, Fundamentals and Applications in Telecommunications and Sensing (Artech House, Boston 1999) G. Meltz, W.W. Morey, W.H. Glenn: Formation of Bragg gratings in optical fibers by a transverse holographic method, Opt. Lett. 14, 823–825 (1989) K.O. Hill, B. Malo, F. Bilodeau, D.C. Johnson, J. Albert: Bragg gratings fabricated in monomode photosensitive optical fiber by UV exposure through a phase mask, Appl. Phys. Lett. 62, 1035–1037 (1993) P. Sivanesan, J.S. Sirkis, Y. Murata, S.G. Buckley: Optimal wavelength pair selection and accuracy analysis of dual fiber grating sensors for simultaneously measuring strain and temperature, Opt. Eng. 41, 2456–2463 (2002) R.B. Wagreich, J.S. Sirkis: Distinguishing fiber Bragg grating strain effects, 12th International Conference on Optical Fiber Sensors, OSA, Vol. 16 (Technical Digest Series, Williamsburg 1997) pp. 20–23 C. Baldwin: Multi-parameter sensing using fiber Bragg grating sensors, Proc. of SPIE 6004, Fiber optic sensor technologies and applications IV, 60040A-1 (2005) 370 Part B Contact Methods 14.35 B. Glisic, D. Inaudi, K.C. Hoong, J.M. Lau: Monitoring of building columns during construction, 5th Asia Pacific Structural Engineering and Construction Conference (APSEC) (2003) pp. 593–606 14.36 D. Inaudi, B. Glisic: Application of distributed fiber optic sensory for SHM, 2nd International Conference on Structural Health Monitoring of Intelligent Infrastructure (2005) Part B 14
© Copyright 2025 Paperzz