Evaluation of Defects in CHi

242

Evaluation of Defects in Carbon/Carbon Composites by Using Thermal Diffusivity Mapping Distribution

Yicheng Peter Pan, Richard A. Miller, Tsuchin Phillip Chu and Peter FilipSouthern Illinois University Carbondale , Department of Mechanical Engineering and Energy Process,

75 B Patrick Ln, Carbondale, IL 62901(618)453-7039; fax (618) 354-7658; e-mail panyicheng@gmail.com

ABSTRACTIn order to develop new NDE capabilities to determine defects and measure thermal diffusivity in Carbon/Carbon (C/C) composites at the same time, a study was conducted of NDE methods for identifying subsurface defects in C/C composites by thermal diffusivity measurement using step heating method. Since traditional NDE methods are effective in inspection of thin carbon/epoxy composites, but they are not very effective in detecting defects in thick C/C composites. In addition, there is substantial industrial demand for the nondestructive, rapid, on-line evaluation of the thermal diffusivity of an entire C/C composite parts surface. Therefore, this paper applied step heating infrared thermography method to do through-thickness thermal diffusivity measurements and NDE inspection for whole fi eld carbon/carbon disk brake. In this work a brief description of the theory behind step heating method and a sample application are given. FEA analysis with the use of ANSYS was also used to compare with the experimental results and found that they were in good agreement with one another. As the result, this method is capable of evaluating defects in C/C composite materials. With more research, this method is effi cient, economically feasible, easy implementation, and rapid assessment of detecting defects in C/C composite materials that will be able to be incorporated into a manufacturing process quality control system.

INTRODUCTIONIn recent decades, carbon/carbon composites (C/C) have been preferably employed in several different industries [1]. The main reason is due to the advantages of the composites, such as slight weight, fracture toughness, high strength, and high stiffness. Besides, high-fatigue resistance and heat resistance up to 3000 K are the important advantages of C/Cs for high-temperature applications, for instance, race car disk brakes, surfaces of hypersonic vehicles, and refractory tiles [2-3]. The capability to determine subsurface defects in C/C materials is important to many industries producing and using C/C composite disk brakes, and an inexpensive and easy to implement and operate method is a crucial component to the industry [8]. In order to keep the high reliability of C/C materials, most applications use NDE methods to evaluate the quantitative information about fl aw and defect size to serve as an input to fracture mechanics based predictions of remaining life. The NDE techniques became to mainly important method to detect the reliability and performance of C/C materials. The traditional NDE techniques, which include eddy current, ultrasonic, X-ray and acoustic emission, etc. have faced serious impenetrability in NDE of the C/C, due to the unique properties of C/C materials [3-4]. These methods are often not adequate for detecting fl aws and defects at an early stage because of the lack of the space and depth resolution. Moreover, the traditional NDE methods in detecting fl aws and defects in thick and/or multi-layered C/C structures are almost unattainable [5]. Because of these problems, it is very diffi cult to monitor the performance, condition and quality of C/C. That is why the C/C can not be popularized and rarely used in primary structures which require high reliability. In general, many C/C composite materials are used for their thermal characteristics, as in the case of the C/C brakes and thermal protections for hypersonic vehicles. Thermal diffusivity, as a fundamental property of the material, is one of the important parameters when heat transfer phenomenons are involved. The current standard test method by industry is the (ASTM) test method for thermal diffusivity of solids by the Flash Method, which measures diffusivity at one point. This method also requires the destructive removal of a small sample for examination, special machinery for surface scanning, and is not suitable for quality assessment in a manufacturing environment.

The step heating method is an alternative technique to the known fl ash method, which can measure non-destructively the thermal diffusivity easily and quickly especial for thick whole scale C/C composite [6]. Additionally, a new method to potentially detect defects in C/C was discovered. Preliminary results showed that defects areas have lower through-thickness thermal diffusivity of the carbon-carbon disk brake [7]. Therefore, this paper built the through-thickness thermal diffusivity measurements and NDE inspection system by using step heating method for whole fi eld carbon-carbon disk brake for meeting those demands. By utilizing relatively simple infrared thermography equipment, through-thickness diffusivity measurements can be made quickly. The through-thickness thermal diffusivity mapping of each whole fi eld carbon-carbon disk brake can be obtained within 10 minutes, simultaneously evaluate the defect in C/C composite by thermal diffusivity mapping distribution.

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

243

In addition, the through-thickness thermal diffusivity measurements may be a new method to potentially detect defects in C/C that was discovered from previous research. Preliminary results showed that defected areas have lower through-thickness thermal diffusivity values of the carbon-carbon disk brake. In order to develop new NDE capabilities to determine defects in C/C composite disk brakes, this study conducts NDE methods for identifying subsurface defects in carbon-carbon C/C composites by through-thickness thermal diffusivity measurement. This method is capable of evaluating defects in C/C composite materials. With more research, this method can become effi cient, economically feasible, easy implementation, and rapid assessment of detecting defects in C/C composite materials that will be able to be incorporated into a manufacturing process quality control system.

METHODOLOGY FOR STEP HEATING METHOD Step heating method is based on the application of an instantaneous constant heat fl ux on the front face of the sample and measurement of the temperature response at the rear face, and then the thermal diffusivity can be obtained by calculation. The step heating method allows for constant heat fl ux for a semi-infi nite plate, which implies that the heat fl ow is unidirectional and normal to the imposed surface. Therefore, incorporated into the step-heating method are ideas proposed in the earlier works [9]. Substituting step-heating for the laser pulse was used to test samples with relatively large dimensions in comparison to those used in the fl ash method [10]. The step heating is a promising photothermal technique for measurement of thermal diffusivity of solids (steel and insulators) with a relatively small size at ambient temperature from 25-500 C [11]. The step-heating method can be viewed as an extension of the well known fl ash method, based on measurement and analyzing the temperature response at the rear face after application of an instantaneous heat pulse. Another advantage of the step-heating method is the relatively low intensity of the imposed heat fl ux compared with that necessary for the fl ash heating techniques. The sample is therefore less likely to exhibit a phase transition or decompose as a result of a sudden large temperature increase at the front face.

Theory Background The ideal heat transfer model is based on the behavior of a homogeneous, thermally insulated, semi-infinite slab with uniform and constant thermal properties and density, subjected to a constant heat flux, uniformly applied since the time origin, over its front face (x = 0) of sample. The transient temperature T = T(L, t) at the rear face (x = L) of the sample can be obtained by solving the one-dimensional heat conduction equation

tlxxTa

tT ;0,2

2

ddww w

w ! 0 (1) With the initial and boundary conditions as below:

T(x, 0) = 0, lx dd0 (2)

tkq

xtT ,),0( w

w ! 0 (3) t

xtlT ,0),( w

w ! 0 (4) Where is the through-thickness thermal diffusivity, k is the thermal conductivity and Q is the heat flux. The expression for the sample temperature as a function of position x = L and time t is:

' f 2

2)(

1222)1(2

61),( L

tn

n

n

r enL

tk

QLTtLTTDS

SD (5)

Where Tr is the initial reference temperature. Figure 1 shows the thermal response based on this analytical solution. The observe parameter V is a ratio of rear face surface temperature changes and is defined as:

r

r

TtLTTtLTV

),(),(

2

1 (6)

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

244

The secant iteration method is used to determine diffusivity values. For a given measured value of V, the diffusivity () is varied until the absolute value of the difference of the left and right hand sides of Equation 6 become less then 0.003%. When this condition is met, the corresponding diffusivity value is taken to satisfy 6.

Figure 1: Analytical solution of temperature rise.

The scheme of secant iteration method needs the defi nition of a value G in Equation 7.

Assumed values for the diffusivity are systematically substituted into Equation 7 until the absolute value of G becomes less than 10-5. The secant iteration method is used to determine G. When G is met the restriction, the diffusivity can be obtained.

EXPERIMENTS AND EXPRIMENTAL SETUP

MaterialThere were two materials used in this study. Sample one is a non-heat-treated 3-D Ex-PAN Needle Felt Needle stitched in the z-direction, CVI infi ltrated C/C composite disk brake material that was provided by the Center for Advanced Friction Studies (CAFS). Sample two is heat treated 3-D Ex-PAN Needle Felt Needle stitched in the z-direction, CVI infi ltrated C/C composite disk brake material that are used on aircraft braking systems and provided by certain C/C composite disk brake manufacturer company (XXX). The samples are shown in Figure 2 and listed in Table 1:

Sample A Sample B Figure 2: Image of samples used in experiment.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 0.5 1 1.5 2 2.5 3 3.5

at/L2

f

f

222

212

)(

1222

2

)(

1222

1

2

1

)1(261

)1(261

),(),(

Lt

n

n

n

Ltn

n

n

r

r

enL

t

enL

t

TtLTTtLTG DS

DS

SD

SD

(7)

T

/(Q

L/k

)

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

245

Table 1: The samples used in experiment.

Manufacturer Type DescriptionThickness /Size or

Diameter Label No. FEA material properties

CAFS C/C 3DFully processed without heat

treat, no visible defectsT:30 mm

S:114 x 88 mmCAFS 0001

NHTDensity:1810 kg/m3

Heat Cp: 573 J/kg CConductivity:40.9(X)

36.3(Y)24.3(Z)w/m CXXX C/C 3DFully processed, heat treat,

no visible defectsT:25.4 mm

D:317.5 mmXXX 0002

HT

Theoretical and Experimental of Diffusivity Measurement

Thermal Diffusivity Prediction by FEA Models

To properly model the effects of the C/C composites of the heat transfer, a Finite Element Analysis (FEA) was performed using ANSYS Workbench 11. The material properties for the 3-D C/C composite materials were found from a donated non-heat treated needle felt, ex-PAN fi ber, CVI carbon matrix Honeywell disk brake. The model simulated a homogeneous anisotropic material, instead of the true fi ber and matrix mix for ease of simulation. The model also used a constant convective heat loss on the surfaces exposed to air with the constant given as 20 W/m2 C. Also, the heat input was a heat fl ux applied to the surface opposite the simulated defect. The heat input was simulated as a heat fl ux of 1.25 x 104 W/m2. This heat fl ux was on for 10 to 160 seconds. The result of FEA was analyzed to determine the thermal diffusivity of all C/C composite samples.

Heat Flux Measurement and Uniform CheckIn order to know the accurate heat fl ux from the surface of the hood used in the experiments, this research did a simple measurement. Since all material properties are known for AL 6061-T6, an accurate measurement of the heat fl ux applied to the surface of the sample can be determined. The experiment uses the MikroSpecRT infrared camera. An initial temperature of the rear surface is found using the IR camera and then the Al sample is heated for approximately 150 sec, fi nally temperature is recorded at the end of the 150th sec. Heat fl ux of 1.25 x 104 W/m2 applied to surface for 150 sec. This data was used in FEA models to determine the heat fl ux applied to the surface of the samples by the four linear halogen lamps (500W) positioned in the hood. Furthermore, uniform heat fl ux was necessary for the step heating method. In order to check the consistency of heat fl ux on the surface of the hood, this research did a simple check. Al sheet is heated for approximately 15 sec and then the temperature distribution of the Al sheet, which was painted with dry graphite fi lm, was checked. The results showed that the standard deviation of the temperature distribution for the Al sheet was 1.8472, which is suitable for step heating method. Figure 3 shows photos of the actual experimental setup at SIUC.

Figure 3: Experimental setup for Al sample.

IR Camera Foam Insulations

AL 6061-T6 sample Hood with four

500W halogen bulbs

Al sheet

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

246

Step Heating Infrared Thermography In order to meet the objectives of this research, the step heating infrared system was setup to determine what experiments needed to be performed. The experiments were based on the modeling work conducted at Southern Illinois University Carbondale (SIUC). Experiments were completed using four 500 Watt halogen linear tubes as the heat source and an infrared camera to determine baseline experimental temperature distribution data. The equipment used in these experiments for obtaining the thermal images from heated whole fi eld areas on the C/C composite sample was the MikroSpecRT system. The infrared camera used in these experiments to record the thermal images was a MikroSpecRT thermal imaging camera with a resolution of 0.06 C at 30 C, a measurement accuracy of + 2 C of reading, and 320 x 240 dpi. The thermal image infrared camera unit incorporated a black and white or (grayscale) image viewing screen as well as a colored viewing screen for more refi ned and sharper imaging. The infrared camera was linked to a Dell computer, which used the MicroSpecRT software that came with the IR camera. This software allowed for even further rendering and processing of the images. The software could record in real time and create video recordings or take snap shots as the sample was heated. A model of the test setup is shown in Figure 4. The measurement process is described as the following: 1. Turn on the light and wait for 10-15 seconds. 2. Open the shutter and start recording the thermal response with the IR camera. 3 After the measurement is complete turn off the heat fl ux source.

Figure 4: Transient thermal diffusivity measurement system.

Through-Thickness Thermal Diffusivity MeasurementTo obtain thermal diffusivity values for the C/C materials, the IR camera, samples and hood (with heat source) were used. Foam insulation was used to create a 1-D conductive heat transfer model. The IR camera was placed on the opposite side of the heat source. In this experiment, video was recorded at four frames a second for 150 seconds. A region of whole fi eld C/C composite samples was inspected, and the average temperature was considered. This data was analyzed by through-thermal diffusivity measurements and NDE system to determine the thermal diffusivity of all C/C composite samples. The system was coded by MATLAB. Due to the data acquisition error, the temperature distribution in time domain is non-linear. Therefore, polynomial curve fi tting is necessary to fi t the data trends. The black line is raw data and blue line is curve fi tting line. The temperature profi les selected and extracted were 5 5 pixels regions. In the sample test, the total measurement time was 150 seconds, in which time the maximum temperature increase of the rear face reaches 33.2-35 C. To verify that accurate data was collected, it is compared to theoretical sources. The collected data should bear resemblance to the theoretical temperature time (T-t) curve. Additionally, due to the convergence of Equation 6, when time1/time2 is plotted against Temperature1/Temperature2 with a fi xed interval, the temperature curve should remain below the time curve as in Figure 5. Once the data was reviewed and determined accurate, hypothetical diffusivity values are generated and tested in an iterative process. With diffusivity being the only unknown in Equation 6, the hypothetical values are inserted and the left and right hand sides of the equation are evaluated against one another. This method allows for quick acquisition of the actual diffusivity value.

Control Unit(Laptop)

Target

FT/STPower Supply

IR Camera

Hood

Four Halogen Lamps

A Transient Thermal Diffusivity Measurements System

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

247

Figure 5: Theoretical temperature time curve and temperature curve for rear surface.

MEASUREMENT RESULTS AND DISCUSSIONBoth the results from the step heating experiments and that of the FEA results were modeled and setup in the same manner. Both methods had the same heat fl ux applied to one surface with all four sides insulted to simulate a perfect insulation. A convective heat transfer coeffi cient was determined for a forced convection in the FEA models. Along with those parameters, the heat fl ux and amount of time that was recorded was identical for both the experiment and the FEA modeling. There are two models for each sample created in FEA, that was a defect sample and a non-defect model. The FEA results are shown in Figure 6 and the values for the diffusivity are calculated by Equation 7. The red line is the measuring profi le.

Figure 6: (a) FEA result for sample 1 in defect and (b) no defect and (c) sample 2 in defect and (d) no defect.

Comparison Between FEA Results and Step Heating Experimental ResultsFigure 7 shows the FEA results and the experimental results for sample one. The diffusivity was measured every 2 mm across the sample for a total of 50 points. In FEA, the through-thickness thermal diffusivity was 0.15746cm2/s for non-defected

(A) (B)

(C) (D)

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

248

samples shown as a yellow line in Figure 7. For the defected sample, the thermal diffusivity value varies from (0.15746-0.1056 cm2/s) and is shown as a light blue line in Figure 7 and defected areas have lower thermal diffusivity distribution. In experiments, the range of thermal diffusivity at non-defected area is from 0.14645 to 0.121 cm2/s and is shown as a dark blue line in Figure 7. In the defected areas, thermal diffusivity is signifi cant changed and ranges from 0.14645 to 0.08978 cm2/s and is shown as a pink line. Both the FEA results and the experimental results have the same trend of thermal diffusivity mapping distribution. Also, defected areas have lower through-thickness thermal diffusivity values in the C/C sample. The results from the experiments have a lower thermal diffusivity than the FEA results, this could be due to the environment in which the experiment were created in, or could be contributed to the sample in the experiment not being perfectly insulated on the sides and hood was not seal up, where in the FEA models the four sides of the sample are perfectly insulated. This would explain the small drop in thermal diffusivity due to convection around the boundaries. In addition, the results from the experiments have a higher thermal diffusivity on the sides. This is contributed by the sample in the experiment not being perfectly insulated on the sides due to gaps between the sample and the foam insulator. This would explain the small increase in thermal diffusivity in sides.

Figure 7: Thermal diffusivity mapping for Sample 1.

Figure 8: Thermal diffusivity mapping for Sample 2.

For the second experiment, the diffusivity was measured around the circumference of the circle at every 30 degrees. Figure 8 shows the FEA results and the experimental results for sample two. The FEA results follow the experimental data relatively close, with the detected thermal diffusivity change being slightly lower than the predicted FEA results. Results of the FEA models, showed a through-thickness thermal diffusivity of 0.26 cm2/s for non-defect sample, and is shown as a yellow line

Thermal Diffusivity vs Angle

0

0.05

0.1

0.15

0.2

0.25

0.3

0 50 100 150 200 250 300 350

Theta (Degrees)

cm^2/s

measurementFEA DefectFEA No Defect

Thermal Diffusivity VS position

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0 20 40 60 80 100mm

cm^2/s

Defect area No defect areaFEA No defect

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.

249

in Figure 8. For the defected sample, thermal diffusivity varies from (0.26-0.08745 cm2/s) and is shown as a pink line. In experiments, the range of thermal diffusivity is from 0.213 to 0.089 cm2/s. In the defected area, the thermal diffusivity is signifi cant less. The abrupt change in diffusivity values between 90 and 150 degrees could be indicative of a defect within the material between those two locations. Upon inspection of the disk, a crack through the material along the side section is easily visible.

CONCLUSIONThe results from the experiments conducted at SIUC show that the step heating method can rapidly determine the thermal diffusivity values of a whole fi eld C/C composite disk brakes. This method can detect the defects in C/C disk brakes by using thermal diffusivity mapping distribution. The through-thickness thermal diffusivity mapping of each whole fi eld C/C composite can be obtained within 10 minutes. It could be with in 5 minutes if the system uses an automatic control to measure the thermal diffusivity. Step heating as a method, is a cheap and cost effi cient method for the use of measuring thermal diffusivity and determining defects in C/C samples. With a higher resolution IR camera, step heating as a means of measuring thermal diffusivity and determining defects in C/C disk brakes can be very effective. REFERENCES1. Ilcewicz, L.B., D.J. Hoffman, and A.J. Fawcett, Comprehensive Composite Materials, Elsevier Science, Amsterdam, The

Netherlands. 2000.2. Ruosi, A., Nondestructive detection of damage in carbon fi bre composite, Journal of Physica Stat., Vol, 2(5), March,

pp 1153-1155. 2005.3. Hatsulade, Y., T. Inaba, N. Kasai, Y. Maruno, A. Ishiyama, and S. Tanaka, Detection of deep-lying defects in carbon

fi ber composites using SQUID-NDE system cooled by a cryocooler, Journal of Physica C, 412-414, June, pp 1484-1490. 2004.

4. Dobiaova, L., V. Stary, P. Glogar, and V. Valvoda, X-ray structure analysis and elastic properties of a fabric reinforced carbon/carbon composite, Journal of Carbon vol, 40, pp 1419-1426. 2002.

5. Yang, H.C., J.H. Chen, S.Y. Wang, C.H. Chen, J.T. Jeng, J.C. Chen, C.H. Wu, S.H. Liao, and H.E. Horng, Superconducting Quantum Interference Device: The Most Sensitive Detector of Magnetic Flux Superconducting Quantum Interference Device: The Most Sensitive Detector of Magnetic Flux, Tamkang Journal of Science and Engineering, Vol. 6, No. 1, pp. 9-18. 2003.

6. Vozar, L. and T. Sramkova, Two Data Reduction Methods for Evaluation of Thermal Diffusivity from Step-heating Measurements, International Journal of Heat Mass Transfer, v 40, n 7, pp. 1647-1655. 1997.

7. Zhang, Y. and T.P. Chu, Thermal Diffusivity Evaluation on C/C Brake Using Step-heating Method, Proceedings of the 2007 SEM Annual Conference & Exposition on Experimental and Applied Mechanics, Springfi eld, Massachusetts, June 3-6, Session 29, No. 289, CD ROM. 2007.

8. Miller, R., T.P. Chu, P. Filip and J. Don, Detection of Defects in C/C Composites Using Infrared Thermography, Proceedings of the 2008 SEM Annual Conference & Exposition on Experimental and Applied Mechanics, Orlando, Florida, June 2-5, CD ROM. 2008.

9. Butler, C.P. and E.C.Y. Inn, Thermal Diffusivity of Metals at Elevated Temperature, Thermaldynamic and Transport Properties of Gases, Liquids and Solids, Transports, ASME, New York, pp. 377-390. 1959.

10. Balageas, D.L. and A.M. Luc, Transant thermal behavior of directional reinforced composites: applicability limits of homogeneous property model, AIAA J., v 24, pp. 109-114. 1986.

11. Brittle, R.R. and R.E. Taylor, Step-heating technique for thermal diffusivity measurements of large-grained heterogeneous materials, Journal of American Ceramic Societies, v 67, pp.186-190. 1984.

ASNT Fall Conference and Quality Testing Show 2008 [Charleston, SC, November 2008]: pp 242-249. Copyright 2008, 2011, American Society for Nondestructive Testing, Columbus, OH.