·Home ·Table of Contents ·Materials Characterization and testing

Effects of porosity on the Mechanical Strength and Ultrasonic Attenuation of CF-Peek fibre Placed Composites A. Demma, Department of Mechanical Engineering, Imperial College, Exhibition Road, London, SW7 2BX, England B.B. Djordjevic, Center for Nondestructive Evaluation, The Johns Hopkins University, Baltimore, MD, 21218, USA Contact

Introduction

Significant progress has been made in recent years in the development of advanced polymer composites, especially for structural applications in aerospace. Many studies have focussed on the development of thermoplastic composites. This category of composites offers significant advantages in quality control, manufacturing, toughness and damage tolerance compared with thermosetting. The market cost of thermoplastic composites and the high efficiency required in their structural applications make NDE techniques indispensable in quality control: defects due to imperfections in the fabrication process such as delamination or porosity can all be detected using NDE techniques.
Because of the detrimental effects of voids on mechanical properties, many studies have been conducted to understand their formation and to characterize their morphological characteristics and related effects. The presence of porosity in CF-Peek composites may be due to many causes. Both the material constituents and the fabrication process may be contributing sources of voids. Porosity is dependent on variables such as temperatures, temperature rates and pressure applied during the process. Previous studies showed that in thermoset laminates, voids tend to be small and spherical at low porosity volume fractions (less than 1.5%) and tend to be bigger and cylindrical at higher percentages (Stone and Clark, 1975). Using the deply technique it has been shown that voids tend to be flat and elongated along the adjacent fiber directions in unidirectional and quasi-isotropic laminates (Hsu and Uhl, 1988).
Since a single pore represents a scatter for elastic waves, ultrasonic attenuation has been commonly used to determine porosity level. Previous investigators used different approaches depending on the size of the average scatter. Three regimes of scattering behavior have been identified (Papadakis, 1965; Fecko and Steiner, 1994), depending on the dimension of the wavelength compared to the dimension of the scatter. In many other works ultrasonic assessment of porosity starts from the knowledge of the morphological characteristics of the voids. Porosity estimation using the frequency dependence of the ultrasonic attenuation (Hsu and Nair, 1987; Nair et al., 1989; Reed et al., 1993; Jeong, 1997) has been attempted using the cylindrical model for single scatter (Tan, 1976). The ultrasonic attenuation as a function of frequency has been measured with broadband techniques (Hsu and Nair, 1987; Nair et al., 1989; Reed et al., 1993; Fecko and Steiner, 1994; Jeong, 1997).
The effects of porosity on mechanical properties depend on the characteristics of the constituent materials but especially on the matrix behavior. Many studies have shown that brittle and ductile resins behave differently (Withney, 1985; Johnson and Mangalgiri, 1987). In addition, the effect of porosity is different, depending on the type of stressed state and strain direction. The most noticeable effects can be seen in the resistance to shear loads and transverse loading whereas the resistance to loading in the reinforcement direction is less sensitive to the presence of porosity. The mechanical response depends on porosity microeffects such as local reduction of the fiber-matrix contact and crack initiation, and macroeffects such as matrix effective cross section reduction. Experimental investigations have shown that voids are detrimental to composite integrity and strength ( Tang et al., 1987; Uhl et al., 1987; Bowles and Frimpong, 1992; Rubin and Jerina, 1993). Porosity estimated by ultrasonic attenuation has been correlated with composite strengths (Almeida and Neto, 1994). However, little information is available on porosity characteristics and effects in the case of tow placed thermoplastic composites, which is one of the most common fabrication processes. Moreover, no narrow band ultrasonic generation has been attempted in the case of porosity assessment.

The aim of this study is to contribute to improvements in the ultrasonic assessment of porosity. The purpose of this paper is to determine whether porosity effects on the mechanical strength can be correlated with the porosity induced ultrasonic attenuation of unidirectional CF-peek composites obtained from tow placement of prepregs.

Experimental method

Samples
The samples used in this investigation were 12 specimens cut from a CF-Peek composite ring containing a wide range of void content. The specimens were cut from locations where the ultrasonic attenuation was fairly uniform; the locations were chosen with the aim of covering the broadest porosity range. The composite ring was obtained from tow placement of prepregs using hot air. The composite was unidirectional and the fibers were placed along the circumference of the ring. The dimensions of the ring were:
f=615 mm, t=5,7mm, L=86mm
where f is the inner diameter, t is the thickness and L is the length

Fig 1: Ring dimensions

The coupons dimensions were:
1"x 0.25"x 0.25"
where 1" is the length along the fibers direction.

Ultrasonic Testing
The composite ring was tested along a length of L1 = 38mm (fig.1) over the entire circumference. An ultrasonic bridge system in water jet configuration using the through transmission method was used to scan the cylinder. The ring was clamped to a turntable so that the transducers were motionless during the scans. A 1/4" nozzle orifice was used. The pixel dimensions on the C-scan final image were 0.38 mm x 0.38 mm. The scans were carried out using narrow band generation at three different frequencies (1 MHz, 2.25 MHz, and 5 MHz). No signal average was performed.

Image processing
The C-scan image relative to a single frequency was divided into four sectors with 90º coverage. Fig. 2 shows the sector 0º-270º at 2.25 MHz. The images were subsequently processed in order to obtain five grey level maps (0-24, 25-49, 50-74, 75-99, 100-127) from each image. The 12 specimen locations were chosen using a visual analysis of the discretized images. Two criteria were used to choose a suitable area: the first was the need to cover the broadest grey levels range possible and the second was to take areas showing uniform ultrasonic attenuation.

Fig 2: C-scan image at 2.25 MHz from 0º to 270º

Microscopy
The most common method to obtain morphological data of the pores is microscopy.
Before carrying out the microscopy experiments all the specimens were polished finely along the fiber direction on one face. A morphological analysis was performed on 144 microscopy images (12 for each sample) with a dimension of 1.7 mm x 1.3 mm.

Image processing


Fig 3: Shear load diagram

The images obtained using the microscope were processed. Only areas bigger than 0.126 mm² were examined. Using a blob analysis tool, we were able to acquire the following data relative to single pores:

• minor axis dimension
• major axis dimension
• angle between major axis and fiber direction
• area.

Mechanical testing
The presence of pores affects the shear strength more than the axial strength. Consequently a short beam shear test (SBS) [ASTM-D 2344, 1997] was chosen to excite shear stress on the composite coupons. The shear test specimen is center-loaded as shown in Fig.3. In Fig. 3 and Table 1, l is the length of the sample, a is the span, b is the width, h is the thickness, c is the loading nose width and r is the relative radius.

	l	a	b	h	c	r
mm	25.4	17.8	5.7	5.7	3.2	6.4
Table 1: Setup and specimen dimensions

The h and l dimensions of the specimen were carefully chosen to obtain the required failure mode (interlaminar shear crack) [Demma, 1999]. An extra sample was tested to verify the geometrical parameters chosen. Some of the experiments (specimens 4-5-6-8-10-11) were run putting rubber under the loading nose to reduce the severity of the stress concentration (Cui et al., 1994).

Results

The results of the ultrasonic tests are presented in Table 2.
It appears that when the frequency increases from 1MHz to 5 MHz, the difference between the maximum and the minimum values of the average also increases.

No area	1	2	3	4	5	6	7	8	9	10	11	12
1 MHz	69	52	36	42	48	45	71	57	42	71	65	49
2.25 MHz	102	97	86	96	99	88	98	100	89	97	103	98
5 MHz	51	41	40	46	56	46	67	35	35	75	70	42
Table 2: Average grey level

Table 3 summarizes the statistical analysis results obtained from the blob analysis. l_Mstat is the value of the minor axis and it corresponds to the value of a generic cord of circumference.
The radius value (a) is obtained considering (1):

(1)

	MIN	MEAN	MAX
lMstat (mm)	0.025	0.106	0.498
a (mm)	0.0199	0.0832	0.3912
Table 3: Dimension main axis and radius

Fig. 4a shows the concentration distribution of the pores. The distribution shows a strong tendency for the voids to align parallel to the plies. Fig 4b shows the aspect ratio distribution of the pores (ratio between length and diameter). The average value of the aspect ratio is 11. Therefore, the pores can be modelled as infinite elliptical cylinders whose axis is aligned with those of the fibres. No clear correlation between microscopy and ultrasonic attenuation was found, probably because little information was acquired about an individual specimen during microscopy.

Fig 4a: Histogram of pore orientation

Fig 4b: Aspect ratio distribution

The reduced scattering cross section G for our case was obtained based on the theory on diffraction by cylindrical obstacles (Tan, 1976) and its application in porosity estimation (Nair et al.1989). The hypotheses usually used to simplify the two-dimensional elastodynamic diffraction problem in the case of a cylindrical obstacle inside an infinite elastic transversely isotropic medium were thereby verified (Nair et al., 1989).
The reduced scattering cross section G in the simple case of a circular cylindrical cavity is shown in Fig. 5 (k is the wave number and a is the pore radius).

Fig 5: Reduced scattering cross section G for circular cavity

Fig. 5 is plotted for a value G equal to 0.533 where h is the ratio between shear velocity and longitudinal velocity in the plane perpendicular to the fiber direction (Nayfeh, 1995).
Fig. 5 shows how ultrasonic attenuation increases with frequency and demonstrates that the biggest slope corresponds with the 2.25 MHz frequency.
The presence of interlaminar cracks was noticed in all the specimens after the mechanical tests. Five out of six of the specimens tested using rubber under the loading nose broke during the test. In the composite coupons, load increased slowly when the load was about 1500 N. At this critical value cracks and delaminations were propagating, which was audible during the tests.
From Fig. 6 a, b, and c it can be seen that the frequency used during the ultrasonic tests affected the correlation between ultrasonic and mechanical tests. Ultrasonic attenuation and therefore porosity is higher when the C-scan image becomes darker. We did not compare the correlations at different frequencies. In order to avoid the complications involved in estimating frequency corrections we considered the narrow band generation as though it was a single frequency generation. Fig. 6a (correlation at 1 MHz) presents a data concentration due to the low energy scattered. No apparent correlation was encountered in this case. At 2.25 MHz a broading of the grey levels range can be seen. That is because at this frequency the scattered energy is higher than in the previous case. Fig. 6.b shows how well ultrasonic and mechanical tests correlate at this frequency. At 5 MHz we observed a further broadening of the grey levels range although at this frequency we lost the correlation found at 2.25 MHz. This can be explained by examining the reduced scattering cross section (Fig. 5). At 5 Mhz the scattered energy is almost constant with the porosity dimension whereas at 2.25 MHz the slope of the scattered energy as a function of ka becomes steeper. It appears that porosity detection sensitivity changes with the signal frequency used in the test and in our case it was optimized at 2.25 MHz.

Fig 6a:	Fig 6b:	Fig 6c:
Fig 6: Correlation between ultimate load and grey levels

Discussion

The theory on diffraction by cylindrical obstacles has been applied to the ultrasonic assessment of porosity. Narrow band generation enabled us to avoid frequency dependent attenuation corrections. Good correlation was found between ultrasonic and mechanical tests at 2.25 MHz whereas no correlation was present at 1 MHz and 5 MHz. This result can be explained by observing that porosity detection sensitivity maximized at about 2.25 MHz in our case. This implies that shear resistance depends on pore dimension and that therefore, morphological analysis is indispensable in the ultrasonic assessment of porosity.
It can be beneficial to extend the approach outlined in this study to the porosity detection of other composite materials.

Acknowledgements.

The authors would like to thank Prof. Robert Green Jr. for his support. We would also like to acknowledge the contributions of Dr. F. Lanza di Scalea in the ultrasonic tests and the invaluable help of Dr. L. Rouch with the mechanical tests.

References

1. Almeida S.F.M. and Neto Z.S.N., Composite structures, Vol.28, p.139 (1994)
2. ASTM-D 2344 (1997)
3. Bowles K.J. and Frimpong S., Journal of composite materials, Vol.26, No. 10, p.1487 (1992)
4. Cui W., Wisnom M.R. and Jones M., , Composites engineering, Vol.4, No.3, p.299 (1994)
5. Demma A., M.S. Thesis, University of Palermo (1999)
6. Fecko D.L. and Steiner K.V., Review of progress in quantitative nondestructive evaluation, Vol. 13, p.1189 (1994)
7. Hsu D.K. and Nair S.M., Review of progress in quantitative nondestructive evaluation, Vol. 6B, p.1185 (1987)
8. Hsu D.K. and Uhl K.M. , Review of progress in quantitative nondestructive evaluation, Vol. 6B, p.1175 (1987)
9. Jeong H., Journal of composite materials, Vol. 31, No. 3 (1997)
10. Johnson W.S. and Mangalgiri P.D., Symposium on Toughened composites (Houston, 1985), ASTM special technical publication 937, p. 295 (1987)
11. Nair S.M., Hsu D. K. and Rose J.H, Journal of Nondestructive Evaluation, Vol. 8, No. 1 (1989)
12. Nayfeh A.H., "Wave propagation in layered anisotropic media", Elsevier, (1995)
13. Papadakis E.M., "Polar backscattering", J. Acoustic Soc. Am., No.37, p.711 (1965)
14. Reed F.A., Batzinger T.J. and Reed R.W., Review of progress in quantitative nondestructive evaluation, Vol. 12, p.1265 (1993)
15. Rubin A.M. and Jerina K.L., Composites engineering, Vol. 3, pp. 601-618 (1993)
16. Stone D.E.W. and Clark B, Non-destructive Testing, p.137 (1975)
17. Tan T.H., , Applied scientific research, No.32, p. 97 (1976)
18. Tang J.M., Lee W.I. and Springer G.S., Journal of composite materials, Vol. 27, p.421 (1987)
19. Uhl K., Lucht B., Jeong H. and Hsu D.K., Rewiev of progress in quantitative nondestructive evaluation, Vol. 7B, p. 1075 (1988)
20. Whitney J. M., Composites science and technology, Vol. 22, p.167 (1985)

© AIPnD , created by NDT.net

|Home|    |Top|