July-September 1997 Issue


Using Optical Fibers to Monitor the Health of Concrete Structures

ridges and other reinforced concrete structures inevitably crack before they fail. To look for cracks, owners of such structures perform periodic visual inspections--a technique that is both expensive and unreliable. Energy Laboratory and Brown University researchers are now developing a sensor that can automatically monitor concrete structures for cracking. The sensor involves an optical fiber placed on or within the concrete structure. When a crack forms, the fiber bends and light rays passing through it escape. By analyzing reductions in the light signal, the sensor indicates not only the existence of new cracks but also their sizes and locations. The sensor can detect multiple cracks all along the length of the fiber--a major advantage over other sensors being developed, which typically monitor only what happens at a particular point. Both laboratory experiments and computer simulations demonstrate the capabilities of the new sensor. Other experiments show that wrapping optical fibers around a rod produces a crack sensor that can be embedded into buried structures such as barriers for hazardous waste containment. In related work, the researchers are using chemical deposition techniques to produce optical fiber sensors that can detect strain (stretching or shortening) at a local point on a concrete structure under stress. The novel strain sensor is highly sensitive and a fraction the cost of optical strain sensors produced by conventional means.

Optical Sensor for Detecting Cracks

A good way to monitor the condition of bridges and other reinforced concrete structures is to look for cracks. For example, the advent of widely opened cracks (several millimeters or more) after an earthquake indicates severe damage and urgent need for repair. Smaller cracks can also cause problems. Over time, water and road salt can seep in and corrode the reinforcing steel bars embedded in the concrete, shortening the lifetime of the structure. To prevent such damage, bridge owners are required by law to inspect their bridges every other year. At present, the only means of inspection is by looking. An inspector climbs over the bridge, examining the top, sides, and bottom for cracks--a method that is time-consuming, expensive, and unreliable because some cracks are missed. For buried concrete structures such as underground containers, there are no reliable techniques for detecting cracks.

Various researchers have worked to develop sensors that can automatically detect and monitor cracks in concrete. Approaches have involved transducers, optical fibers, and other devices. But the sensors produced have limited usefulness largely because they can only detect cracks that run through them. In other words, they must be placed precisely where cracks will appear--something that no one can predict.

Now Christopher K.Y. Leung, associate professor of civil and environmental engineering at MIT, Theodore F. Morse, professor in the Division of Engineering at Brown University, and their teams of researchers have used optical fibers to develop a sensor that can detect cracks anywhere along the entire length of the fiber, not limited to a single spot. The only information needed is the general area that is vulnerable and the probable orientation of cracks--factors that often can be predicted. For example, bridges crack because downward pressure causes the surface of the bridge to compress and the bottom to stretch. Since concrete is strong under compression but weak under tension, cracks will tend to form on the underside of the bridge, running crosswise.

Key to the sensor is how a beam of light traveling down an optical fiber behaves when the fiber bends. An optical fiber is a solid glass structure consisting of two parts: a core surrounded by a cladding that has a lower refractive index. (Refractive index is a measure of how fast light travels through a material.) When a fiber is straight, a light ray travels down it by reflecting off first one side of the core and then the other. But when the fiber is bent, some of the light will hit the outside of the core at an angle such that it will not all reflect. Some of it will pass through, escaping from the fiber.

Based on years of work in the field of composite mechanics, Professor Leung and his colleagues knew that if a fiber is embedded in a material and a crack opens along the fiber, the fiber stretches in order to span the new crack. But as shown in the figure below, it must also bend twice, once on each side of the crack, unless the crack intersects the fiber at exactly a right angle. Thus, the intensity of light passing through a fiber embedded in a concrete structure will remain essentially constant unless a crack occurs. Then the fiber will bend, and the amount of light transmitted will abruptly drop.

A loss in signal thus indicates the formation of a crack--but not its location. To determine location, the sensor borrows a technique from the communications industry that involves measuring not the "forward" signal but "backscatter." As light passes through an optical fiber, a small amount is reflected backwards by nonuniformities in the glass structure. A bend in the fiber likewise reduces the backscattered signal. By feeding in quick pulses of light and monitoring the backscattered signal as a function of time, the researchers can calculate (based on the speed of light in the glass) where the loss, hence the crack, has occurred and how large the crack is. Based on subsequent losses in the signal, they can identify and locate additional cracks along the same fiber.

Optimizing the sensitivity of the optical sensor is critical. A submillimeter crack must produce a drop in intensity large enough to detect. But if that drop is too large, a few small cracks or even a single large one will reduce the signal so much that additional changes, thus additional cracks, cannot be detected. The sensitivity of the sensor depends on many parameters, including the size and mechanical and optical properties of the fiber, the thickness and stiffness of the protective plastic coating around the fiber, and the fiber layout in the structure. Achieving an optimal design by trial and error is thus impractical. Therefore, the researchers have developed a computer model that helps. Predictions from the model show good qualitative agreement with results from preliminary experiments with cracks in an epoxy block in which fibers were embedded.

Another challenge is how to use the sensor on concrete structures that are already in place. Since embedding fibers in existing structures is impractical, the researchers have developed a technique for mounting them on surfaces. They enclose the fiber in a thin polymeric sheet and glue the sheet tightly onto the concrete--tightly enough that when the concrete cracks the sheet will also crack rather than peel away. The fiber, on the other hand, is loosely attached inside the sheet so that it will slide, stretch, and bend rather than break when a crack opens.

The researchers have performed a series of experiments that demonstrate the feasibility of that technique. They glue a polymeric sheet containing the optical fiber to the bottom of a beam specimen. They then cut a notch across the specimen so the crack position is known, and they place an instrument across the notch to measure the opening directly. When a load is applied to the beam, the instrument reading and the intensity of light passing through the fiber are both recorded.

The figure below shows the results as a plot of light signal loss versus crack opening. Signal loss is clearly detectable at crack openings below 0.2 mm. The loss steadily increases as the crack grows until the crack reaches about half a millimeter. As the crack opens further, the loss grows more slowly and then levels off at a crack width of about 1 mm. At that width a relatively long expanse of the fiber extends straight across the crack. Since losses occur only where the fiber bends, further expansion of the opening causes little additional loss--an advantage, as the formation of a single large crack will not necessarily mean a total loss of signal.

The researchers are using a similar concept to develop optical fiber sensors that can be installed in buried structures formed in situ. Of particular interest are protective barriers around hazardous waste sites. The Idaho National Engineering and Environmental Laboratory (INEEL) is developing a way to build a structure around buried waste. They drill closely spaced holes around and under the waste and then inject columns of grout. The grout penetrates the soil and forms a continuous wall and bottom. Such a structure will effectively contain any hazardous waste that moves outward--unless the structure develops cracks.

The MIT researchers are designing a sensor for detecting cracks in such structures. The new sensor is a rod wrapped with optical fibers and inserted into the holes before the grout is injected. Cracks in a column generally run crosswise, the result of the column's having been bent. The optical fibers are therefore wrapped around the rod in a spiraling fashion so cracks will intersect them at an angle and be detected. Monitoring of backscattered light will again indicate the formation and location of cracks so that failing sections can be regrouted. Using an experimental sensing rod cast into a mortar beam, the researchers have demonstrated the crack-sensing ability of their new device.

Optical Sensor for Monitoring Strain

In related work, the researchers are using optical fibers to detect another worrisome behavior in concrete structures: when stressed, structures may stretch rather than crack. A measure called strain expresses the fractional change in the length of a concrete structure due to stress. Detecting strain is vital in critical members such as prestressed concrete sections. But developing an effective strain sensor is difficult, in part because small changes are difficult to detect and even a small change in overall length may be important structurally.

One method of detecting strain uses a fiber made up of alternating segments of glass (or other materials) with differing refractive indexes. When white light, which contains many wavelengths, passes through such a fiber, some of each wavelength is reflected from the interfaces between the segments. But a narrow band of wavelengths is reflected more intensely than other wavelengths are, and the particular band of "peak reflectivity" depends on the spacing between the interfaces. If the fiber is attached to an underlying structure that stretches, the fiber will also stretch. The thickness of the alternating segments will change, so the spacing between the interfaces will change. The wavelength at which peak reflectivity occurs will then shift, and the shift can be measured. Such a fiber is an ideal strain sensor. Since changes in temperature will alter the thickness of the segments through thermal expansion or contraction, the fiber also serves as a temperature sensor. Indeed, a separate, uncoupled fiber must be used along with the strain sensor so that temperature effects on the sensor can be quantified and separated from the strain effects.

A major challenge with this approach has been how to create the segments with differing refractive indexes. The standard technique uses high-energy lasers and is very expensive. Professor Leung, Professor Brown, and their coworkers are developing a novel method that is potentially much cheaper. They use chemical deposition techniques to deposit thin layers of silicon nitride and silicon-rich silicon nitride--materials with different refractive indexes--onto the tip of the fiber. By stacking up forty alternating layers they focus the reflected light to a narrower and narrower band of wavelengths, thereby making the sensor sensitive to smaller and smaller changes in the spacing of the layers. Experiments on a layered optical fiber embedded in a polymer specimen show that the system behaves as predicted in its responsiveness to strain and temperature.

The most promising way to use this layered-tip fiber on an existing structure is to glue the fiber firmly onto a steel strip and then glue the steel strip onto the concrete section under strain. In that design, the strain over the length of the steel strip is transmitted directly to the fiber tip and detected. Results to date suggest that this novel strain sensor should be able to measure changes in length smaller than a thousandth of a millimeter.

This research was supported by the University Research Consortium of the Idaho National Engineering and Environmental Laboratory. Further information can be found in references.