• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Evaluation of white light Fabry-Perot interferometry fiber-optic gages for small strains

Posted on: Wednesday, 24 September 2003, 06:00 CDT

An experimental study was conducted to evaluate whether fiber optic strain gages (FOSG) may be better suited than classic foil gages (FG) for small strain measurement applications. A particularly attractive feature of FOSG is their specified resolution of 0.01% of full-scale (0.1 [mu]strain for 1000 [mu]strain full-scale). This feature would make FOSG practical as tank level sensors, by measuring very small strains on the support structure of a tank. A specific application in mind was to measure liquid oxygen tank level, by measuring strain in support beams that were predicted to contract approximately 22 [mu]strain as the contents change from empty to full. Among various fiber optic technologies currently available, Fabry-Perot Interferometry using white light was selected. This technology exhibits highly desirable features such as absolute strain measurement, linearity over its full-scale, and temperature compensation. Experimental results show that the resolution is 0.8 [mu]strain, at best; calibration from one sensor to another can be off by 2.4-11.2%; and temperature compensation is not fully predictable, with errors of up to -8.2 [mu]strain over an 11[degrees]C range. Hence, when compared with classic FG, FOSG possess less accuracy, similar resolution and repeatability (precision), and are linear over their entire operating range. FOSG are also immune to EMI and their signals suffer minimal degradation traveling over long distances. Based on operating principles of these technologies, one may also expect that drift with time will be minimal in FOSG whereas the gage factor of foil sensors will probably experience more significant changes over time when operating in outdoor environments. In conclusion, FOSG are better suited than FG for small strain measurements as long as the application allows calibration of individual units as installed for operation. In this case, the accuracy has the same value as the resolution, and the special qualities of fiber optic technology provide improvements in linearity, signal quality, EMI immunity, and resistance to the elements.

BACKGROUND

Although much has been done to apply fiber optic technology to design a variety of sensors, few are truly commercially available. Classical temperature, pressure, and strain sensors have not been displaced by this new technology to any significant extent. There could be important advantages to using fiber optic (FO) sensors over resistance, piezoelectric, or thermoelectric sensors, but these cannot be realized until FO sensors are demonstrated to be "better" than classical sensors, if only for some applications. As a way to determine the suitability of FOSG for an application that requires measurement of very small strains, a series of experiments were carried out to evaluate the performance of FOSG against classic FG, focusing on accuracy, resolution, precision, drift, and temperature compensation.

Foil Gages are widely used to measure strain. They are small, and have gained wide acceptability across the scientific and engineering community. However, FG have some shortcomings where FO sensor technology may offer, perhaps, a better alternative. A brief discussion follows.

A change in ambient temperature produces four effects on a

These effects produce a thermally induced mechanical strain in the gage that does not occur in the specimen. In contrast, the fiber optic sensing element is simply an air cavity that strictly follows dimensional changes in the specimen. A combination of specially engineered materials and circuit designs have been developed to deal with the apparent strains in FG, but they have the effect of reducing the operating temperature range and the sensitivity of the strain gage-Wheatstone bridge. The temperature effects exhibited by FG may also be produced by self-heating of the resistance element.

Other considerations for use of fiber optic technology include: (1) Fiber optic signals are immune to electromagnetic interference (EMI), and add little noise when long leads are needed before the signal reaches a processing unit; (2) Long leads have significant resistance/capacitance that produces loading on the foil strain gage measurement system. Using a third or fourth wire as compensation leads mitigates this problem, but reduces sensitivity. Fiber optic cables do not exhibit this problem. (3) Three or four wires must be connected to a strain gage, while only one fiber optic cable is needed per FOSG. If long leads are involved, a reduction in the overall number of leads has important benefits from the operating/ maintenance perspective.

Additional advantages of FOSG include corrosion resistance and practically zero probability to generate sparks. These are very attractive attributes for applications in critical environments.

OPERATION OF WHITE LIGHT FABRY-PEROT FIBER OPTIC STRAIN GAGES (FPFOSG)

These sensors consist of a multimode optical fiber used to transport white light, with the sensing element at the tip. The sensing element is defined by a micro capillary tube that holds the end of the fiber close to another small piece of the same fiber, leaving a cavity in between2-4 (Fig. 1). The fiber-ends that define the cavity are deposited with mirrors, so that the white light entering the cavity is reflected, and hence frequency-modulated in accordance to this length. When the sensor is bonded to a surface, the length of the cavity in the micro capillary expands or contracts exactly by the same amount of strain experienced by the surface

The modulated light returning from the sensing element is interpreted using a white light cross-correlator. This device matches the sensor's cavity-length to the thickness of a specific location in the variable-thickness lens4 (Figs. 2 and 3). The light transmitted through this specific location in the lens contains the highest level of energy as a result of modulation in the sensor cavity. The light is detected by a CCD array, where the pixel receiving the highest amount of energy corresponds to the sensor cavity-length. Each pixel of the array corresponds to a specific cavity length.

Temperature compensated FOSG are also available. These units null- out strain on the measurand due solely to temperature variations. Compensation is accomplished by the use of a metallic fiber with the same thermal coefficient of expansion as that of the material being measured (Fig. 4).

White-light Fabry-Perot interferometry fiber-optic strain-gages are robust, exhibiting a design that leaves little room for variation/degradation in performance. Measurement of the cavity length is encoded by light-frequency rather than amplitude, thus significant variations in performance by the light sources do not affect the sensor's performance.

TEST SETUP

Two carbon-steel test-beams were prepared for tests. The first beam (Beam 1) was instrumented with two FOSG; one non-compensated and one compensated, one foil gage, and one Type K thermocouple. The un-compensated FOSG, was faulty, and so data were not collected from it. The FOSG and foil gage were mounted in accordance to instructions from the manufacturers. In fact, the procedures are very similar, and the cleaning and bonding materials used are the same. The second beam (Beam 2) was instrumented with two FOSG (compensated), two foil gages, and one Resistance Temperature Detector (RTD) to measure temperature. In addition, a Reference Beam (small piece of the same material as the beams) was instrumented with one FOSG, one foil gage, and one RTD. The reference beam was subjected to the same temperature conditions as Beam 2, but did not experience stress, thus providing data necessary to compensate experimentally the sensor-measurements in Beam 2 for any temperature effects.

Fig. 1: Fabry-Perot FOSG: Schematic (Courtesy of FISO Technologies, Inc.7)

Fig. 2: FOSG Signal Conditioning Components (Courtesy of FISO Technologies, Inc.7)

Fig. 3: FOSG Cross-Correlator to extract cavity length (Courtesy of FISO Technologies, Inc.7)

Fig. 4: Temperature compensation (Courtesy of FISO Technologies, Inc.7)

Carbon steel beams were used to match the material of the structure supporting the Liquid Oxygen (LOX) tank of interest. The compensated FOSG were specified to match the thermal expansion coefficient of carbon steel (12 [mu]strain/[degrees]C). Beam 1 was 2.5 in x 13 in x 0.1 in, mounted so that the cantilevered portion would measure 12 in. Beam 2 was similar, except for the thickness, which was increased to 0.128 in.

A thermocouple was used with Beam 1, but to improve accuracy, RTDs were used with Beam 2 and the Reference Beam.

Figure 5 shows Beam 2, and Fig. 6 shows the experimental setup. A FOSG bus-unit and individual conditioning modules were used for the FOSG (Bus-System Fiber Optic Signal Conditioner, FISO Technologies, Ste-Foy Quebec-Canada). Analog signals from each conditioning unit were sampled using National Instruments' SCXI 1100 Module (32- channel Analog Input Module), filtered at 4 Hz. Foil gage signals were conditioned and sampled using National Instruments' SCXI 1520 Module (Universal Strain Gage Input Module), filtered at 10 Hz. All data were sampled at 100 samples/second.

Fig. 5: Beam 2 (2 FOSG, 2 foil gages, and I RTD) and Reference Beam (I FOSG, I foil gage, and I RTD)

Fig\. 6: Experiment Instrumentation: FOSG conditioning rack and modules (below), rack and conditioning modules for foil strain gages (above), and current source for RTD's (on top)

Temperature experiments were done using a variable transformer (VARIAC) to control the intensity of a light source from 0% to 100% of its rated wattage (60 watts). To reach and maintain stable environmental conditions, the beams were placed inside an enclosure. A computer program written in Lab View5 was developed to integrate and automate data acquisition.

TEST PROCEDURES AND RESULTS

All sensors and instrumentation were kept ON for a warm-up period longer than that prescribed by the manufacturers. Experiments to determine linearity, resolution, and precision, were done using a set of weights that were hung at the end of the beam. Applicable elements of ASTM E 251-92 Test Standard Methods for Performance Characteristics of Metallic Bonded Resistance Strain Gages were considered in the design and performance of the experiments. For the temperature experiments, a period of two hours was determined to be more than sufficient to allow stabilization of the temperature throughout the beam every time the level of intensity of the heat lamp was changed.

Each data point shown in the plots (Figs. 7-14) represents an average of 400 to 500 measurements taken at 100 samples/s. Assuming a symmetric statistical distribution, the standard deviation of the mean was calculated to be 0.002 [mu]strain for the foil gage and 0.008 [mu]strain for the FOSG. The respective statistical probable errors are 0.001 and 0.005 [mu]strain. These values are more than two orders of magnitude smaller than the measured resolution of the sensors, hence establishing the appropriate statistical significance.

Experiments with Beam 1 show that FOSG have very good Linearity (Fig. 7). Standard deviations are on the order of 0.04 [mu]strain for the foil gage and 0.15 [mu]strain for the FOSG. Errors were determined with respect to theoretical calculations. For the FOSG, the theoretical strain value is larger, because the diameter of the sensor had to be considered (250 microns). The FOSG error is -5.1%, and the Foil error -2.7%. Two consecutive experiments (load/unload cycles), where the temperature varied minimally (less than 0.14[degrees]C), were compared to assess precision. The foil sensors achieve 0.2 [mu]strain precision and the FOSG 0.06 [mu]strain precision (Fig. 8).

An experiment to determine resolution shows that both sensors can resolve about 0.8 [mu]strain at best (1 [mu]strain for practical purposes). Figure 9 illustrates the effect of adding small increments of strain (approximately from 0.2 to 3.3 [mu]strain) at various levels of total strain (approximately from 0 to 180 [mu]strain). Values are compared with theoretical predictions, which do not consider the diameter of the FOSG, hence they are shown only as a reference.

An experiment to identify drift was done by monitoring the sensors for an extended period of time (Fig. 10). Measurements were taken every two hours, for a total of 34 measurement sets (during a 68-hour period). Two FOSG exhibit good stability, remaining within 0.5 [mu]strain. One FO sensor (Fiber 2) follows closely the temperature curves, and drifts further at the end to -1.7 [mu]strain. Foil sensors change by as much as 1.5 [mu]strain.

Fig. 7: Linearity Experiment with Beam I

Fig. 8: Comparison of two consecutive tests with Beam I

Fig. 9: Resolution experiment with Beam I

Fig. 10: Drift experiment with Beam 2

A second experiment using the weights was done to further investigate precision and linearity (Fig. 11). The experiment consisted of two load-unload cycles with the weights. Linearity and precision are very good. Fiber 2 exhibits the largest discrepancy of about 1.2 [mu]strain at around 10 [mu]strain of total deformation. All other sensors have discrepancies of less than 0.5 [mu]strain.

Fig. 11: Linearity and precision experiment with Beam 2

Since Fiber 1 exhibited values farther from those of the foil sensors, it was removed and replaced by another sensor Fiber 1R. Later, a third fiber was added to Beam 2, Fiber 3, to further investigate variations from one FOSG to another. Measurements show that these variations range from 2.4% to 11.2%.

A temperature experiment was performed by heating the beams using a light bulb. The test and reference beams were placed in an enclosure and the temperature was changed by controlling the light intensity. Temperature was increased by a total of 11[degrees]C from ambient (around 25[degrees]C), and measurements were taken every two hours. Foil sensors, after subtraction of the values from the reference beam foil sensor, should exhibit near zero readings. The maximum error is 1.6 [mu]strain at [Delta]T = 11[degrees]C (Fig. 12). FOSG, since they are compensated, should exhibit near zero readings. Fiber 1 does not appear to be temperature compensated at all. Fiber 2 and the reference exhibit some degree of compensation (maximum errors of 3.1 and 2.2 [mu]strain), but they are not linear (Fig. 13). The only explanation for the nonlinear and uncompensated behavior of the FO sensors is that the temperature compensation technology is not yet well developed. Thus, when using FO sensors, it may be advisable to use uncompensated sensors, and mount one unit on a specimen subjected purely to temperature strains to compensate for this effect.

Results describing accuracy and resolution agree with similar experiments carried out at NASA Langley Research Center, where the focus was to evaluate foil and fiber optic sensors experiencing large deformations.6

Fig. 12: Compensated foil gages performance with increasing temperature from ambient

Fig. 13: FOSG performance with increasing temperature from ambient

ERROR ANALYSIS

Alignment and location of sensors were considered as possible causes of experimental error. However, it was determined that the measurements were not very sensitive to these errors. In order to account for the observed variation in values from one FOSG to another, a location error along the beam of at least 1/4'' would be needed. Experimental sensor location was at least ten times better than that value. Alignment errors of the order of 5 degrees cause only 0.4% strain error, and again, sensors were aligned at least ten times better than that value. Other errors caused by the electronic processing systems were minimized by ensuring appropriate warm up periods and zeroing of biases at the beginning of each experiment.

COST CONSIDERATIONS

A traditional foil sensor and conditioning unit cost approximately $3,000. This does not include costs of data acquisition and analysis. A data acquisition system may add another $625 (Assuming a $5,000 data acquisition system with 8 channels). The cost of a POSG and signal conditioner from FISO Technologies7 comes to approximately $3,850, plus approximately $400 for data acquisition). Luna Innovations,8 using probably a somewhat different technology, charges approximately $6,900 for a FOSG and signal conditioner. Hence, at least for one brand of FOSG, costs are comparable to those of traditional foil sensors.

Table 1.-Summary of Experimental Data

CONCLUSIONS AND RECOMMENDATIONS

Table I summarizes experimental data and other considerations. Experimental results are limited to strain measurements from zero to 180 [mu]strain, and temperature variations of approximately 11[degrees]C above ambient. Both types of sensors have equivalent linearity and precision at quasi-constant temperature conditions (ambient). Load/unload cyclic strains show a variation of less than 0.5 [mu]strain on all sensors, except for one FOSG that exhibited a 1.2 [mu]strain variation. Both technologies have equivalent resolution of approximately 0.8 [mu]strain, independent of the total level of strain. For small temperature variations (less than 1[degrees]C), both technologies exhibit small drifts (less than 1.5 [mu]strain). The FOSG were temperature compensated, but compensation appears to be deficient. Two FOSG exhibited non-linear response to temperature variations. We can only conclude that the temperature compensation technology of the fiber optic sensors is not yet well developed. Foil sensors, when subjected to pure temperature changes, exhibited a maximum error of 1.6 [mu]strain at [Delta]T = 11[degrees]C. Finally, measurement variations from one FOSG to another were significant, from 2.4% to 11.2%. Therefore, individual calibration of FOSG is necessary to obtain accurate absolute measurements.

Given the deficient temperature compensation exhibited by the compensated FO sensors it is recommended to use uncompensated units to improve calibration/accuracy. Compensated FOSG have additional sources for errors as they ' have to be fitted with a metallic fiber of very accurate dimensions and coefficient of thermal expansion. FOSG may be considered "better" than FG for some applications based on inherent properties that include high immunity to RFI and corrosion, spark-less operation, robust signal conditioning with low probability of making mistakes and easy integration into a data acquisition system. Also, fiber optic signals can be transported long distances (miles) without deterioration. FOSG were also easier to operate (there was a smaller number of variables to track). The processing software and electronics were easy to use and provided diagnostics that ensured high integrity of the data. The signal processing (hardware) for the foil sensors required more attention.

FOSG are "worse" than foil sensors in that variation of the sensor gage factor from one unit to another is much larger. Therefore, each unit must be individually calibrated if one needs accurate absolute measurements. In the LOX tank application, FOSG can be easily calibrated by recording two strain values, when it is empty and full (or any set of two distinct known levels). If the FOSG is encapsula\ted as a component of a sensor, then it can be calibrated in the laboratory. Foil gages technical experts from Measurements Group, Inc.9 indicated that accuracy better than 1% is attainable in laboratory environments and around 5% in field installations.

Over a small temperature range about ambient, FOSG are "equivalent" to foil sensors in linearity, precision, and resolution.

ACKNOWLEDGMENTS

The authors wish to thank Mr. Lester Langford for his assistance in acquiring and setting up the sensors and instrumentation (hardware and software) that made possible this work. We are also thankful to Lester for his valuable recommendations and suggestions on matters related to sensors and instrumentation. We thank, as well, Mr. Charles Thurman for his support and assistance concerning circuit theory. Finally, this work would not have been possible without the encouragement and funding support provided by Dr. Shamim Rahman.

References

1. Dally, J.W., Riley, W.F., and McConnell, KG., "Instrumentation for Engineering Measurements," Second Edition, John Wiley & Sons, Inc., 1993.

2. Trouchet, D. et al., "Prototype Industrial Multi-Parameter F. O. Sensor Using White Light Interferometry," Springer Proceedings in Physics, Vol. 44, 1989, pp. 227-233.

3. Lefevre, H.C., "White Light Interferometry in Optical Fiber Sensors," 7th Optical Fibre Sensors Conference, December 2-6, 1990, Sydney, New South Wales, pp. 345-351.

4. FISO Technologies, Inc., "Fiber Optic Strain Gages," Application Note, FISO Technologies, Inc., 2014 Jean-Talon N., Suite 125, Sainte-Foy (Quebec) Canada G1N-4N6.

5. National Instruments, Austin, Texas, USA.

6. Hare, David A., and Moore, Thomas C., "Characteristics of Extrinsic Fabry-Perot Interferometric (EFPI) Fiber-Optic Strain Gages," NASA/TP-2000-210639, December 2000.

7. FISO Technologies, Inc., Sainte-Foy (Quebec) Canada G1N-4N6.

8. Luna Innovations, Blacksburg, VA, USA.

9. Measurement Group, Inc., Raleigh, NC, USA.

F. Figueroa and W. St. Cyr are affiliated with NASA Stennis Space Center, MS. D. Van Dyke, G. McVay and M. Mitchell are affiliated with Lockheed Martin Space Organization, Stennis Space Center, MS.

Copyright Society for Experimental Mechanics, Inc. Jul/Aug 2003

More News in this Category