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

Temperature Calibration Of Efpi Fiber Optic Strain Sensors

Posted on: Thursday, 11 March 2004, 06:00 CST

KEY WORDS: Fiber optic strain sensor, apparent strain calibration, thermal expansion coefficient, temperature compensation

In recent years, optical fiber sensing systems have become potentially attractive due to their inherent salient technical features over the conventional sensors. The advantages of a fiber optic sensor over a conventional sensor are, small physical size and weight, flexibility, immunity to electromagnetic interference, resistance to corrosion, high resolution, large bandwidth of signal, practically no noise, high sensitivity, ability to multiplex sensors, ability to distribute sensors over a single fiber and high performance. These fiber optic sensors can sense a variety of physical parameters such as pressure, strain, temperature and displacement. Their small size and flexible nature allow them to be integrated into structural materials without sacrificing structural integrity.

Fiber optic strain sensors are considered to be very stable for long term strain monitoring, but it is necessary to correctly interpret the influence of various parameters on the fiber optic strain gage output, during health monitoring of critical structures. Variation in temperature is an important factor that influences the fiber optic strain gage output. Currently, temperature compensated fiber optic strain sensors are not commercially available for many structural materials. Even those, which claim to be temperature compensated, are not fully compensated. Hence, it is important to clearly differentiate between the gage output due to free thermal expansion of the structural material from that of strain. During health monitoring of civil engineering structures like bridges, the output from the fiber optic strain sensor may have a component of apparent strain due to a temperature effect, which must be properly accounted for. This apparent strain is caused due to the tern- perature difference between the reference temperature and the temperature at any other time of measurement. The variation in temperature of structures between extreme seasons can be as much as 30 to 40C, and this variation may cause an apparent strain of large magnitude. This paper describes the details of generating calibration curves for temperature induced apparent strain for fiber optic strain sensors, for different structural materials, such as steel, aluminum and concrete. These calibration curves are useful for correcting the apparent strain due to temperature effect, for identical sensors and structural materials.

Fig. I: Fabry-Perot fiber optic strain sensor configuration

FIBER OPTIC SENSORS

A fiber optic sensor system basically consists of a light transmitter, a receiver, an optical fiber or bundle, a modulator element and a signal processing unit1. Light is transported to the measurement point (Modulator) using optical fibers or bundles and such a scheme is generally termed as extrinsic modulations. If the fiber itself acts as a sensitive element, then intrinsic modulation takes place.

Fiber optic sensors are fabricated using high strength silica, which possesses an inherent immunity to corrosion and electromagnetic interference. The properties of optical fibers allow innovative approaches for the design of optical sensors. For this reason, a number of fiber optic sensor types have been developed. Fiber optic sensors can be classified under different categories. Localized, distributed and multiplexed sensors are based on "sensing" methods while, intensity, interferometric, polarimetric and spectrometric are based on transduction mechanism.

Extrinsic Fabry-Perot Interferometric (EFPI) sensors are reported to be good for strain sensing in civil engineering applications. A cavity comprising two mirrors (reflectors), which are parallel to each other and perpendicular to the axis of the optical fiber, form the localized sensing region in an EFPI type of sensor. Here the reference and sensing optical fibers are one and the same up to the first mirror (partial mirror), which constitutes the start of the sensing region. The Fabry-Perot cavity is formed between the air- glass interface of two fiber end faces aligned in a hollow core fiber (Fig. 1). Changes in the separation between the two fiber end faces, known as air gap length, cause interferometric fringe variations. The interference pattern generated is sinusoidal in shape and is directly related to the intensity of the applied strain field. The period of the waveform constitutes a fringe and by proper calibration the magnitude of the strain can be determined.

TEMPERATURE INDUCED APPARENT STRAIN

The change in strain of an optical fiber sensor subjected to a change in temperature alone is an important issue that has to be addressed when considering its application in practical structures. Strain measurements made with an optical fiber sensor are complicated by the presence of apparent strain due to differential thermal expansion of the fiber optic sensor and the structural material.

It may be noted that in the case of electrical resistance strain gages, the errors due to temperature effects can be eliminated by employing an identical compensating or "dummy" gage connected suitably (half bridge) to the Wheatstone bridge circuit2. The self- temperature-compensated strain gages are also used to minimize the thermal output of electrical resistance strain gages. The metallurgical properties of certain strain gage alloys can be processed to minimize the thermal output over a wide temperature range when bonded to a test material with thermal expansion coefficients for which they are intended. Strain gages employing these specially processed alloys are referred to as Self- Temperature-Compensated (STC)3.

EFPI strain sensors are the only known fiber optic strain sensors that can be self temperature compensated. Two advantages of EFPI sensors are that they are not sensitive to thermal variation and to transverse strain4. Self-temperature compensation of EFPI fiber optic strain sensor is theoretically possible, primarily because the EFPI strain sensor itself contributes very little to thermal strain. Self-temperature compensation of the EFPI is achieved by replacing the standard optical fiber reflector with a metal reflector. The length and thermal coefficient of the metal reflector are carefully chosen to compensate the thermal expansion of a structural material. For example, to create a self-temperature compensated EFPI strain sensor for stainless steel, a metal wire reflector is chosen which possesses a higher thermal expansion coefficient than stainless steel. Design calculations are performed to determine the EFPI air gap displacement vs. temperature relationship for an uncompensated EFPI sensor mounted to stainless steel. Further calculations are then performed to determine what type and length of metal reflector wire provides the same, (but opposite) EFPI air gap displacement vs. temperature relationship5.

APPARENT STRAIN CALIBRATION

When making strain measurements in a variable temperature environment, the indicated strain is equal to the sum of stress- induced strain in the test specimen and the temperature induced apparent strain of the gage bonded to the test specimen. With the thermal output expressed in strain units, correction for this effect can be made by simply subtracting (algebraically) the apparent strain from indicated strain.

To correct the measured strain, the apparent strain must be established separately. In order to correct the temperature effects, temperature calibration was carried out from laboratory experiments on three structural materials namely steel, aluminum and concrete, using commercially available EFPI fiber optic strain sensors which are either not compensated for temperature effects or partly compensated for steel6.

A steel specimen of size 300 20 3mm was prepared and ' two fiber optic strain sensors, one temperature compensated for steel and the other without any temperature compensations, were bonded adjacent to each other. A temperature sensor (electrical resistance type) was also bonded (adjacent to fiber optic strain sensors) using suitable adhesive to measure the surface temperature of the specimen. The instrumented test specimen was placed inside a temperaturecontrolled oven and the temperature was raised in steps from ambient temperature to a maximum of 80C to 90C. The temperature of the test specimen was allowed to stabilize at each stage, before measurements were carried out. Strain from the fiber optic strain sensor and temperature from the temperature sensor were recorded for each temperature setting. In a similar fashion, temperature calibration studies were conducted for obtaining temperature correction data on an aluminum specimen.

While conducting temperature calibration studies for concrete, a fiber optic strain sensor (without any temperature compensation) was also embedded inside the concrete cylinder. For studies in the concrete specimen, a temperature controlled water bath was used instead of a temperaturecontrolled oven to eliminate the drying shrinkage effect. Also the concrete specimen was soaked in water for a sufficient period to obtain a saturated condition.

Figs. 2 to 4 show the Temperature vs Strain plots, from which appropriate temperature correction coeffici\ents can be obtained. The slope of the plots gives the apparent strain per degree Celsius for the particular sensor and structural material.

DISCUSSION

Experiments were conducted in the laboratory to identify the problems related to temperature effects on fiber optic strain sensors for strain measurements. Fiber optic sensors, which are claimed to be fully temperature compensated for steel, also show a little apparent stain for steel specimen. From the experiments using non temperature compensated fiber optic strain sensors, it is seen that the apparent strain per degree Celsius is very close to the thermal expansion coefficient of these materials used in the experiments. Hence using a non-compensated EFPI fiber optic strain sensor in a test specimen, one can directly measure the thermal expansion coefficient of any material.

Fig. 2: Temperature calibration curves for apparent strain correction-steel specimen

Fig. 3: Temperature calibration curves for apparent strain correction-aluminum specimen

Fig. 4: Temperature calibration curves for apparent strain correction-concrete specimen

From the experiments conducted on the concrete specimen, it was found that the fiber optic strain sensor which is temperature compensated for steel specimen, gives nearly zero apparent strain and the fiber optic strain sensor without any temperature compensation gives apparent strain very close to the thermal expansion coefficient of this particular concrete mix. Concrete is one of the most important engineering materials, which has varying thermal expansion coefficient. The thermal expansion coefficient varies depending on the concrete mix, type and size of aggregate, type of curing, type of admixtures and water cement ratio. Hence it is important to conduct separate experiments to obtain apparent strain due to temperature for each type of fiber optic strain sensor and concrete used in the structure.

Since the temperature compensated fiber optic strain sensors are not fully compensated, it is advisable to use nontemperature compensated fiber optic strain sensors. In such case the apparent strain correction can be carried out theoretically by knowing the thermal expansion coefficient of the structural material and the temperature at the location of measurement. However it is clear from the above experiments that the temperature induced apparent strain is very significant and should be properly accounted for, in order to conduct an accurate strain analysis during long term monitoring of structures using fiber optic strain sensors.

References

1. Udd, E. Fiber Optic Smart Structures, John Wiloy & Sons, Inc., New York, 1995.

2. Kesavan, K., Ravisankar, K., Narayanan, T., Parivallal, S. and Narayanan, R. Generation of Apparent Strain Calibration Curves for Structural Materials at Elevated Temperature, EXPERIMENTAL TECHNIQUES, Nov/Dec 1999, pp 39 to 41.

3. Strain Gage Temperature Effects, Technical Note-504-1, Measurements Group Inc., Vishay, USA.

4. Choquet, P., Juneau, F. and Dadous, F. New Generation of Fiber Optic Sensors for Dam Monitoring, Proceedings of the International Conference of Dam Safety and Monitoring, China, 1999.

5. EFPI Strain Gage Self-Temperature Compensation, Technical Note- Revision B-8/99, Luna Innovations, USA.

6. Kesavan, K., et al, Experimental Studies on Temperature Calibration of Fiber Optic Strain Sensors, SERC Research Report SERC, EML-RR-2002-1, 2002.

K. Kesavan, K. Ravisankar, T. Narayanan, S. Parivallal, and P. Sreeshylam, are Scientists with the Structural Engineering Research Center, CSIR Complex, Chennai, India.

Copyright Society for Experimental Mechanics, Inc. Jan/Feb 2004

More News in this Category