by Jon R. Lesniak, Daniel J. Bazile, and Michael J. Zickel*

 

This month's article is a good example of how modifications to existing techniques can continually improve the quality of inspection. This solution improves flaw signal detectability while reducing sensitivity to undesired coating variability. The result is a more robust field application of thermographic techniques. G.P. Singh Associate Contributing Editor

 

Figures 1-2
Figures 3-4

Introduction
A very sizable investment has been made in this nation's steel highway bridges, the integrity of which is often taken for granted. Statistical studies from the Federal Highway Administration (FHWA) reveal that of the 470,000 road carrying bridges in the US, approximately 110,000 are classified as structurally deficient. Sixty percent of those bridges have steel superstructures, and the cost to repair them is estimated to be in the billions of dollars. The FHWA is responsible for overseeing the maintenance of these structures, and must be able to lead the way in health monitoring and prioritizing of repairs. Therefore, it is vital that the FHWA have the tools necessary to assess and monitor the health of key structures quickly and accurately.

Thermal methods have been recognized for rapid inspection of large structures (Cramer and Winfree, 1992). Thermal methods are safe, convenient, and relatively inexpensive tools; however, the techniques' dependence on emissivity makes them sensitive to field conditions such as chipped paint.

This article describes an exciting new crack detection technology: coating tolerant thermography, which is specifically designed to meet the demands of challenging field conditions. The paper shows that coating tolerant thermography exhibits high probability of detection, low false signals, easy interpretation, speed, and portability.

---

Coating tolerant thermography is designed for inspection of large steel structures that are heavily and nonuniformly coated.

---

 

Theory
Thermal methods correlate structural integrity with thermal diffusivity. If the molecular structure is altered, impairing transfer of forces, the conduction of heat energy is also impeded (Osiander et al., 1996). Coating tolerant thermography is based on a technique called "forced diffusion," which projects a pattern of dynamic heat to force flow across cracks, thereby optimizing the measurable thermal gradient (Lesniak et al., 1995, 1996, 1997). Heat travels from "hot" stripe to "cool" stripe as the stripes slowly comb the structure for cracks. Heating techniques in standard thermography usually involve projecting a uniform pattern of heat onto a surface and observing as the heat diffuses into the object or along the plane of the surface.

The in-plane heat flow is impeded by a structural discontinuity, such as a crack, creating a gradient in the thermal image (Figure 1a) that clearly defines the crack (Figure 1b). The direction of the heat flow (from the left or from the right) defines the sign of the gradient as either positive or negative. In Figure 1c the heat flows from the right, causing the sign of the gradient to be negative.

 

Separation of Anomalies
The success of thermal methods for in-field inspections depends on the method's ability to distinguish between pseudosignals and critical data. Coating tolerant thermography locates structural discontinuities to a high degree of accuracy, and with great efficiency, providing results in an intuitive and easily understandable format. False signals like reflections and emissivity gradients are automatically reduced in the final form of the data.

 

Reflections
Reflections from ambient sources are reasonably constant over short periods of time and can be eliminated. A thermal image is collected before the incremental heating of the specimen and is subtracted from the image collected after heating. A reflection and initial condition reduced image I(x,y) is in the form

(1)

where DF(x,y) is the photon flux resulting from the added heat or heat lost since the capture of the initial image, and e(x,y) is the emissivity map. This method of reducing reflections permits work with even high gloss paint.

 

Emissivity Effects
If all structures had perfectly uniform coatings, the simple unidirectional gradient method depicted in Figure 1 would be sufficient to locate cracks. However, gradients caused by emissivity variances can often be misconstrued as cracks. The total gradient in the x-direction Ix(x,y) of the corrected image I(x,y) from Equation 1 is, by the product rule

(2)

At this point in the development of the technique, an emissivity gradient has the same effect as a critical structural anomaly. The fundamental principle behind coating tolerant thermography is that only the thermal spatial derivative of a true structural anomaly will change sign with opposing heat flow (Figures 1b and 1c). When the heat is flowing from the left, the gradient is positive (as defined) because the heat builds up behind the crack on the left. When heat is flowing from the right, the gradient changes sign to become negative, because heat now builds up behind the crack on the right side. Only a true structural flaw has this characteristic.

Gradients caused by paint chips or other emissivity changes on the surface of the structure are not sensitive to conduction direction. In other words, the sign of the differential is independent of the direction of heat flow. As a material (e.g., steel) is heated from the left (Figure 2), the region under the paint edge is uniformly heated because there is no flaw in the steel substrate. On the left edge of the paint chip the differential is apparently hot on the left and apparently cold on the right. Similarly, if the heat conducts from the right, the area under the left edge is heated uniformly and the left side of the edge remains apparently hot and the right side stays apparently cold. Therefore, the gradient is independent of the direction of the heat flow. This is the cornerstone of coating tolerant thermography.

Mathematically this can be thought of as separating the thermal distribution from the emissivity variables. If we normalize the gradient image Ix(x,y) of Equation 2 by the reflection corrected image I(x,y) of Equation 1

(3)

we separate the influences of thermal distributions and emissivity gradients into two terms. If two sets of images are collected, one corresponding to heat from the left and the other corresponding to heat from the right, and processed in this manner

(4)

(5)

the difference of these images Q(x,y) eliminates emissivity as a variable.

(6)

 

The physical result of this process is to virtually eliminate the detection of emissivity effects and to simultaneously increase the response from a critical structural anomaly. If the gradient is flowing from opposite directions, then the second term will augment the first. In order to ensure this augmentation and hence detection of the critical anomaly, the denominator must not be allowed to switch sign.

To demonstrate this ability a 305 ´ 50 ´ 6.35 mm (12 ´ 2 ´ 0.25 in.) steel specimen coated with a rather reflective white paint was prepared with a rusted paint chip near a fatigue crack. The rusted area and the high gloss paint create a potentially misleading emissivity gradient. Figure 3 demonstrates the ability of the normalized subtraction process to minimize false readings at anomalies like paint chips. Comparing the line plots from Figure 1b and 1c to those of Figure 3a and 3b clearly shows how the emissivity gradient across the paint chip could be mistaken for a critical structural anomaly.

Notice the difference in the maximum values between the first two line plots (Figure 3a and Figure 3b) and the third line plot (Figure 3c). The emissivity gradient caused by the paint chip is almost completely eliminated from the final data by performing the subtraction of the right and left images. The crack could not be included in the same line plot as the paint chip and is not shown in the figure. However, reviewing the math and recalling the results shown by the line plots in Figure 1, it is clear that the subtraction process would only enhance the response of the crack.

 

Results from Turner-Fairbanks Demonstration
Stress Photonics demonstrated the coating tolerant thermography method at the FHWA Turner-Fairbanks Laboratory. At this site, a large steel girder had been fatigue tested to establish visible and invisible cracks. Steve Chase of Turner-Fairbanks picked a series of bridge details that represented typical field scenarios including cracks originating at a gusset plate, cracks in a T-weld, and an "invisible" crack in the flange of the girder.

Only a simple stationary pattern projector was used at the demonstration. The projected hot stripe was manually moved from side to side as thermal images were collected. Both an oscillating mirror method and an all-digital method of gradient measurement were used depending on the situation. Absolutely no surface preparation was performed on any of the details on the girder.

In the most challenging example, a crack had grown in the flange above a stiffener (Figure 4a). The structure was heated to the right of the crack, which induced a large thermal gradient at the face of the crack. Figure 4b shows the result of the full process and clearly indicates the crack, which, after being coated with bridge paint, was invisible to the naked eye.

 

Preferred Methodology
Although the data demonstrating the fundamentals of coating tolerant thermography were collected by projecting stationary radiation adjacent to cracks, a more complete system involving moving patterns of thermal radiation is envisioned. The final product will project slowly moving line patterns that comb the structure for cracks. The gradient data will be collected only in the region where gradients are induced. The positive slope data will be normalized and added to the image buffer; the negative slope data will be normalized and subtracted from the image buffer. The resultant thermal data will be placed in an image buffer for delamination or soil detection. Some design decisions remain; for example, a choice between a continuously moving line pattern or four discrete steps must be made. In the event that four discrete locations are used, the optimal order of positions must be determined. Also, the number and timing of reflection subtraction images must be examined.

 

Conclusion
Coating tolerant thermography is an adaptation of forced diffusion technology, specifically designed for inspection of large steel structures that are heavily and nonuniformly coated with paints and perhaps debris. To summarize coating tolerant thermography, the difference between a left heated image and a right heated image will accentuate a crack and minimize signals related to emissivity variance. The gradient data is normalized by the magnitude of the reflection reduced thermal image so that emissivity effects are separated from thermal effects. Emissivity variances are completely eliminated in the final image through a subtraction process.

The coating tolerant thermography method can flag the presence of paint delaminations. A metal test coupon coated with a common industrial paint was tested at the Turner Fairbanks Lab, and after only a few seconds of heating, the delamination was clearly defined. In the event that a critical structural anomaly is hidden beneath delaminated paint, the delaminated area will be highlighted and recorded, so that the loose paint can be removed and the structure can be properly inspected.

There is a clear path to an inexpensive hand held system that will be portable, robust, rapid and widely distributable. Most importantly, the resulting images are easy to interpret, because the existence of a crack is as clear as black on white.

 

Acknowledgments
Stress Photonics would like to acknowledge the supporters of this research, which included the Federal Highway Administration and the National Aeronautics and Space Administration. There are several individuals without whom this project would not have been possible: Steve Chase of the FHWA, Elliott Cramer of NASA LaRC, and Phil Fish of the Wisconsin DOT.

 

References
Cramer, K.E., and W.P. Winfree, "Thermographic Imaging of Cracks in Thin Metal Sheets," in Thermosense XIV, ed. by K. Eklund, Proceedings of SPIE, Vol. 1682, Jan. 1992, pp 162-170.

Lesniak, J.R., D.J. Bazile, and M.J. Zickel, "Structural Integrity Assessment via Coating Tolerant Forced Diffusion Thermography," to be published in the proceedings of the 1997 ASCE Structures Congress, Portland, OR.

Lesniak, J.R., and D.J. Bazile, "Forced-Diffusion Thermography Technique and Projector Design," in Thermosense XVIII, ed. by D. Burleigh and J. Spicer, Proceedings of SPIE, Vol. 2766, Apr. 1996, pp 210-217.

Lesniak, J.R., and B.R. Boyce, "Forced-Diffusion Thermography," in Thermosense XVII, ed. by Sharon A. Semanovich, Proceedings of SPIE, Vol. 2473, Apr. 1995, pp 179-1895.

Osiander, R., J.W.M. Spicer, and J.C. Murphy, "Analysis Methods for Full-Field Time-Resolved Infrared Radiometry," in Thermosense XVII, ed. by D.D. Burleigh and J.W.M. Spicer, Proceedings of SPIE, Vol. 2766, Apr. 1996, pp 218-227.

 

* Stress Photonics Inc., 3002 Progress Rd., Madison, WI 53716; (608) 224-1230; fax (608) 224-1233; e-mail info@stressphotonics.com.

Copyright © 1997 by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]