Experiments Set-up and Test
Knowing the type of materials being tested is important to obtaining accurate results from the experiments. Also the more known about a material, more conclusions can be made from experimental data. With this it is important to know the equipment being used and how to set it up properly. Combining these two things will help lead to greater success in finding defects and also will help insure the accuracy of the project.
Materials
The sample materials in the lab are of woven carbon fiber reinforced by a polymer, where the carbon fibers add strength and rigidity. The panels are made of different layers, the number of layers range from 2 to 32 layers of carbon fiber weave. Since CFRP are made layer by layer there is possibility that defects will occur in between the layers. The most common defects that are seen in CFRP are delaminations, voids, inclusions, porosity, and regions having non uniform distribution of fibers [4]. When the panels were laid out, Teflon inserts were placed in between some layers at known locations and sizes to represent defects, mainly delaminations. To find the range of defects that can be found, each panel had the Teflon parts inserted at different depths and sizes, each panel has different shaped defects. Figure 7 is schematic of the CFRP panels that were used in testing.
The actual material properties of any fiber reinforced plastic depend on several parameters such as fiber density, layer thickness, matrix material, shape of the fibers, and quality. The fiber density is how tight the weave of the CFRP is. Since carbon fibers have a much higher thermal conductivity along the length of the fibers as opposed to normal direction, a different fiber density will change the thermal conductivity along the length and width of the panel. Continuing with this would be the layer thickness, how large are the fibers that are being used and what the spacing between the layers is, and the amount of epoxy between the layers. The type of matrix (epoxy) can be different as well. Then the shape of the fiber section, it was briefly mentioned that fibers could be unidirectional or woven but even the woven fibers can have different patterns depending on the application.
Figure 19: (a) Schematic used in experimental testing. The Teflon inserts are equal in size throughout the pane, and are show in there approximates locations. (b) Schematic of a panel used in testing, showing the locations of the Teflon inserts.
Experimental Set-up
The panels were tested experimentally using the equipment provided by the Intelligence Measurement and Evaluation Laboratory (IMEL) and the Center for Advanced Friction Studies (CAFS). Figure 21 is a schematic showing the basic testing set up for the CFRP panels. Figure 22 shows a picture of a typical test set up, where the infrared camera is sitting on top of the hood which encloses the four 1000W heat lamps aimed downward towards the surface of the panel. The heat lamps are computer controlled using a microcontroller that was developed in house, where the camera is controlled using the MikroSpec R/T software that was provided with the camera. The infrared camera being used is a MikroScan 7600PRO manufactured by MIKRON INFRARED. The camera can be used as a hand held camera, and has a built-in visual camera. For infrared cameras, this is considered high resolution at 320x240pixels and uses an uncooled focal plane array microbolometer. This detector can measure temperature differences up to 0.06°C at 60Hz at 30°C, and is accurate within 2% or ±2°C. The temperature range can be set between -40°C-120°C, 0°C-500°C, or 200°- 2000°C, and can focus on an object as close as 0.3m. With an emissivity range that can be adjusted from 0.1 to 1.0 in .01 increments [5].
In general CFRPs have a relatively shiny surface because of the epoxy which results in low emissivity. Although the infrared camera can adjust for different emissivity values there is a greater likely hood of the panel reflecting other light sources resulting in a false temperature reading by the camera. To help reduce this error the surface of the objects are typically painted with high emissivity paint. When the paint is used it is considered as a uniform thin layer that does not affect the result of the temperature profile [6]. The testing done in the IMEL used a temporary black paint manufactured by Dupli-Color. This paint provided a flat black surface reducing the reflectivity effectively raising the emissivity to approximately 0.97. It is easy to apply, takes very little time to dry and can be easily removed without affecting the material itself, making it ideal for non-destructive testing [7].
Figure 8: Schematic of the testing equipment used
Figure 9: Actual photo of the testing equipment and set up used for experiments
Results and Discussion
There are many different types of post processing methods, and they have been shown to help improve the detection of defects. Depending on the type of inspection process will dictate some of the post processing methods that can be used. An image is made up of tiny squares and each square has a number associated to it, typically between 0 and 255, the higher the number the whiter the color (on a gray scale). Many post processing methods find a unique way to make the pixel value of a defect further from the pixel value of the non-defect region.
The subtract function is another popular method used and is built into the MikroSpec software. The subtract function lets the user pick a frame that they would like to subtract from the rest of the images. The software takes the frame that is selected and finds the temperature value for each pixel, then it subtracts that value from the temperature value of all the other frames. For example if the image to be subtracted was a uniform color with every pixel value equal to 100°C, and the frame to be analyzed had pixel values ranging from 10-255°C. The new range of pixel values would be from -90-155°C.
There are many options when it comes to using the subtract function since any frame can be chosen, however certain frames should prove to be better than others. Using ANSYS the ideal points for using the subtract function are evaluated. Figure 10(a) shows a typical temperature vs. time graph with the COP at around 10.5sec. In this figure there are three ovals labeled as 1, 2, 3, these are the three different type of frames that could be used in the subtract function, one before the COP, at the COP and one after the COP. Figure 10(b) shows the temperature vs. time graph using region one, it can be seen that the temperature contrast is 13.9°C this is 11.4°C less than the original maximum temperature contrast of 25.3°C. However, the amount of contrast over the entire experiment above 1°C (about the amount of contrasted need to identify a defect) is about 6% higher than the original. Figure 10(c) is the new graph when the frame at which both the defect and non-defect region temperatures are the same, the temperature contrast for this has not changed significantly and the number of frames above 1°C of contrast did not change either. Choice three, shown in Figure 10 (d), has an increase in temperature contrast by 5.3°C, but the number of frames that the contrast is over 1°C has dropped by 4.2%.
Figure 10: (a)Three locations that can be chosen to be used for the subtract function.(b) Temperature difference when Frame 1 is subtracted from the peak temperature, Frame 0. (c) Temperature difference when Frame 2 is subtracted from the peak temperature, Frame 0. (d) Temperature difference when Frame 3 is subtracted from the peak temperature, Frame 0.
To show the change in thermal contrast another way line profiles have been taken of the original frame and the three manipulated frames. A line profile finds the temperature value of each pixel along the length of a line (length determined by user). For this case a line was drawn across an area with no defect while also crossing though the defect region as shown in Figure 11 (a) by the red line. MikroSpec R/T was used to do this evaluation and since the camera is mounted to the hood and all the subtract function images are from the same experiment there is no error in the line profile location. When the subtract function is used the temperature values change from image to image, so to help better understand the results the temperatures were adjusted so that they all started off as the same temperature. Since this is done by addition there it has no effect on the overall temperature contrast. Using a line profile shown in Figure 11(a), Figure 11(b) shows the adjusted values, and the thick blue line represents the original image without any post processing.
Figure 11: (a) Location of where the line profile is taken so the thermal contrast can be Evaluated (b)Line profiles of a defect and non-defect regions for different subtract function images
In this case it is not an issue but in cases where there is a much smaller temperature difference the extra 1.4°C could lead to finding a defect not previously seen. Another set of experiments were done on a different CFRP panel. The sample was heated for 5sec with four 1000W bulbs and allowed to cool. When the raw data was analyzed only 4 out of the 9 defects could be identified in the data collected, and one frame is shown in Figure 12(a). A frame before the maximum temperature was captured and subtracted from the rest of the images. This frame is Figure 12(b), and only shows 3 out of the 4 defects. When subtracted from the rest of the images more defects are revealed, frame 99 shows the best contrast and is shown in Figure 12(c), where an additional 3 defects are revealed and shown inside of the black circles. The subtract function is not always repeatable but can be very beneficial. There can be some noise and perceived temperature difference in CFRP because of the weave and the inherent nature of thermography. This noise will cause information to be lost and or masked, because of this sometimes some frames work better than others. In theory if a point is taken after the temperature regions cross over there is a better chance of increasing the temperature contrast. However, due to noise and reflections this is not always the case as seen below and will many times come down to the operator’s experience.
Figure 40: (a) Original image with no post processing, showing four defects.(b)Image of the frame that was subtracted from Figure 40 (c)Result of subtracting Figure 41 from Figure 40, 3 more defects can be seen
Conclusion
It is concluded that the amount of heat and the duration is important to the temperature contrast of the material. If there is not enough energy put into the material then the intensity will be decreased, also if the amount of time is too short the change in temperature will be too small for the camera to detect. Increasing both of these will increase the contrast however, if too much energy is input into the material it will start to burn and change the material properties. Since this is nondestructive testing this needs to be avoided. The maximum temperature will depend on the epoxies used in the CFRPs so will depend on a case to case basis but around 120°C will typically be the limit. This limit should not be pushed too far, this is because if the known amount of heat to raise a non defect panel to 120°C is used and the defect creates a hot spot on the surface then this temperature limit compromised. A smaller amount of energy over a longer time period increases the chances of finding deeper defects in thicker materials. Where shorter higher amounts of energy will increase the contrast and allow for faster results in thinner panels. This research has started a broad foundation of information that can be expanded. It has reviewed the temperature trends in the defect material allowing one to better understand the materials being used in the experiment. Future work should look at different post processing methods and combine them experimentally with different types of defects to see if there are other large temperature contrasts that can be detected. Also a study of the COP and its relationship between time and material properties could serve as useful information for processing image data and knowing the ideal amount of time a material should be evaluated. In addition, the variables for the FEA model should be refined so that more accurate predictions can be made between the models and the experimental data. This additional information will help refine CFRP testing even more and the concepts would likely be transferable to other materials as well.
