Impact and subsequent fatigue damage growth in carbon fibre laminates

Impact and subsequent fatigue damage growth in carbon fibre laminates

Impact and subsequent fatigue damage growth in carbon fibre laminates W. J. C a n t w e l l , P. T. C u r t i s a n d J. M o r t o n The influence of ...

2MB Sizes 0 Downloads 44 Views

Impact and subsequent fatigue damage growth in carbon fibre laminates W. J. C a n t w e l l , P. T. C u r t i s a n d J. M o r t o n The influence of the ply stacking sequence on the impact resistance and subsequent 0-tension fatigue performance of carbon fibre laminates has been investigated. Drop-weight impact tests were conducted on a range of 16 ply carbon fibre laminates with either all non-woven plies or mixtures of woven and non-woven plies. Damaged coupons were tested in 0-tension fatigue for up to 106 cycles, scanned using an ultrasonic probe and then loaded in tension until failure. The impact resistance and subsequent fatigue performance have been found to be sensitive to the ply stacking sequence. The non-woven composites showed a marked sensitivity to impact loading, but increases in residual static strength were noted after cycling. The inclusion of a woven fabric served to improve the impact resistance of the laminates. Fatigue cycling resulted in considerably improved residual static strengths; by 106 cycles any effect of the impact damage

had been removed. Key words: fatigue; composite materials; drop tests; zero-tension fatigue; residual static strength; laminate stacking sequence

Advanced carbon fibre-reinforced composite materials are finding increasing application in aerospace structures. The high specific stiffness and strength of these materials enable considerable savings in structural weight and potentially large fuel economies to be made. However, these composites can be sensitive to impact by foreign bodies, 1-s since they absorb energy mainly through fracture mechanisms rather than elastically or plastically, as metals do. Carbon fibre composites are used mainly in the form of laminates in which the fibre direction varies from lamina to lamina. Material properties can be altered significantly by changing the stacking sequence, and several programmes have been aimed at gaining an understanding of the r61e that the ply stacking sequence plays in determining the impact resistance of a composite panel. 6-8 Stellbrink and Aoki 6 concluded that changing the stacking sequence of a laminate does not alter its impact performance. Dorey, 8 however, found that placing 45 ° plies on the surface of a laminate can enhance the subsequent residual properties of an impactdamaged composite since they would tend to protect the main load-carrying 0 ° fibres. Impact energy absorption in carbon fibre composites occurs by means of delaminations between plies, fibre cracking and matrix shear cracking. The relative importance of each of these depends upon the lay-up, thickness and impact energy. 9 In structural composites impact damage itself may not cause immediate failure. This damage may, however, give rise to subsequent failure under service loads. In aircraft applications fatigue loading is a major design consideration. Results of zero-tension fatigue tests on impacted coupons reported by Dorey 1° indicated that, under certain conditions, subsequent fatigue may mitigate the effects of impact in composites. Bishop and Curtis 11 ' 12 drew attention to the possible improvement in impact tolerance of carbon fibre laminates through the selective incorporation of woven fabric. Woven fabric was used to replace two opposing 45 ° plies which would normally provide sites for extensive delamination. In the present work the post-impact fatigue performance of

three 16 ply mixed-woven and non-woven laminates was assessed.

Experimental procedure The laminates studied were manufactured from preimpregnated sheets of Torayca T300 high strength, surfacetreated carbon fibres in Ciba-Geigy BSL 914C epoxy resin, cured to the manufacturer's recommendations. Details of the laminates are given in Table 1. Laminates NA, NB and NC possessed a balanced layup: each layer was approximately 1 mm thick, resulting in panels with a nominal thickness of 2 mm and a fibre volume fraction of approximately 60%. Laminates WA, WB and WC were manufactured by replacing the individual +-45° plies with a +45 ° five shaft satin weave fabric, which resulted in slightly thicker laminates. In this report these laminates are referred to as mixed-woven laminates. After manufacture the quality of the laminates was assessed using an ultrasonic C-scan facility. Impact testing was conducted using a drop-weight rig consisting of a 12.7 mm spherically-nosed carriage falling freely through a height of one metre to strike the specimen centrally. The panels were clamped between two 100 mm diameter supports. Impact energies up to 7 J were obtained by adding weights to the carriage. However, this upper limit

Table 1. Summary of the laminates studied Lam i nate*

Lay-u p

NA NB NC WA

[+45, - 4 5 , 0, 0, --45, +45, 0, 0] s [+45, --45, 0, 0, +45, --45, 0, 0] s [0, 0, +45, --45, 0, 0, ---45, +45] s [-+45, 0, 0, T-45, 0, 0]s [-+45, 0, 0, -+45, 0, 0] s [0, 0, +-45, 0, 0, -T-45]s

WB WC *N

-- non-woven

; W

--

mixed-woven

0142-1123/84/020113--06 $3.00 © 1984 Butterworth & Co (Publishers) Ltd Int J Fatigue Vol 6 No 2 April 1984

113

of impact energy was reduced to 6 J when the extension of damage was such that it spread into parts of the panel reserved for other specimens. After impact the carriage was caught to avoid secondary impact. The impact-damaged panels were then cut into individual specimens with dimensions 250 × 50 × 2 mm using a diamond slitting wheel and rescanned so that an appreciation of the type and extent of the damage could be gained. Aluminium end-plates, 50 × 50 × 2 mm in size, were subsequently bonded to the specimen ends in preparation for mechanical testing. The cyclic stresses for fatigue were determined from a series of monotonic tests on selected specimens conducted on a screw driven test machine with a crosshead speed of 2 mm rain -1. 0-tension fatigue cycling was undertaken at a stress corresponding to 80% of the undamaged quasi-static strength of the specimen and at a frequency of 20Hz. Tests were interrupted after 10 a, 10 5 or 10 6 cycles to rescan the specimens and then residual tensile strength tests were performed.

of impact damage growth, which tended to be in the +45 ~ fibre direction. The effect of impact on the laminates was such as to reduce the residual strength. Figs 3 to 8 show the variation of residual strength with incident impact energy for each of the laminates. It was observed that subsequent ()-tension fatigue cycling did not give rise to further loss of strength; indeed the post-fatigue residual strengths were generally higher than the post-impact strengths.The effects of varying numbers of fatigue cycles are also shown in Figs 3 to 8, in which each point is the result of a single test. The impact damage, residual strength and postimpact fatigue performance of each laminate will now be summarized.

Results Microscopic sections of impact-damaged coupons were used to characterize the damage and its extent. Fig. l a shows typical impact damage consisting of considerable delamination and matrix shear cracking, which radiates from the point of impact in a manner similar to that observed in CFRP ballistic impact by Card and Rhodes. 13 The C-scan technique provides a non-destructive method of assessing impact damage. It does not, however, distinguish between the different types of damage nor readily identify the through-thickness location of delamination damage. With the C-scan technique no damage was observed in any laminate subjected to an impact energy of 1 J or less but above this energy the damage (thought to be mainly delamination) grew rapidly with increasing impact energy. Generally, subsequent O-tension fatigue did not increase the extent of the impact damage (as indicated in the C-scans). In Fig. 2a, however, laminate NA does show a considerable increase in damage zone size after 1 0 6 cycles. It should be noted that this growth was not apparent after 10 s cycles and that growth tends to occur in the direction of the applied load rather than in the direction

0 cycles

106 cycles

Fig, 2 Ultrasonic C-scans of non-woven and mixed-woven specimens subjected to an impact energy of 4J and 0 or 106 cycles 0-tension fatigue: (a) laminate NA, (b) laminate WA, (c) laminate NC, (d) laminate WC

Fig. 1 Micrographs o f lay-up NB taken directly under the point of impact: (a) impact energy 4 J, 0 cycles, (b) impact energy 4 J, 10 ~ cycles 0-tension

114

Int J Fatigue A p r i l 1984

900

I000

800

9OO

700 ~ " ~ • ~ , ~ ,

/I

8OO 1

,,%

"" 6OO

• <%

700

,

......................

-..,j~-~...
.<:..........



"G (3-

.=

500

600

I ~ , ~'" ~ , ~ .

*g m

\

400

-~

500

~

. 400

\

,% %.

o= Number of cycles

n-

30O

• . . . . .

0



10 4

• . . . . .

105

Number

300

x ................... iOs

200

200 I00

of cycles



. . . . .



~

0 i0 4



. . . . .

jO 5

x .................. I0~

tO0

I I

O0

I 2

I I 3 4 Iml:x]~ energy (,J) 00

Fig. 3 Residual tensile strength vs impact energy for laminate NA I000

I I

I 2

I I 3 4 Impact energy (J)

Fig. 5 Residual tensile strength

vs

I 6

impact energy for laminate NC

/X,

900 / I

800'

\

900

/

• .,...../IX,\ .\. ..... /...., ..~.~\ \

xt/

8001

......: ~ k

,--

"*'..%°

700

,'L

%-\\..;,<.~ k ...... . ........................... . / . .j,. .\-........~,-..... x..."~ \ / ~ .........

/ , 1 -

~\ " " " x ' . ~

\

'" .......~ -

~ v

~

60O

....,

x

/, "%% \

~a

500

-.\

700

~....X_'x

60O

\

.....

"~}..~.

o

I 5

%.

'% ~N

f jf "%

x•/

5OO

/ f

f

f

.% %.

•~ 400

"o cr

Number of cycles

300



200

. . . . .

I0 4

• . . . . .

10 5

Number of cycles

300

0



400

x .................... l0 s

200



. . . . .

0



-

-

t0 4



. . . . .

i0 5

x ................... i0 6 IO0

I00

0 0

I I

I 2

Fig. 4 Residual tensile strength

I n t J Fatigue April 1 9 8 4

I I 3 4 Impact energy (J) vs

I 5

impact energy f o r l a m i n a t e N B

I 6

0 0

I I

I 2

I I 3 4 Impact energy (d)

I 5

I 6

Fig. 6. Residual tensile strength L,s i m p a c t energy f o r laminate WA

115

900 {



"¢"-.

i./ ~n ...-"

.

carrying 0 ° surface fibres which are vulnerable to impact by foreign bodies. Fatigue cycling this laminate increased the static tensile strength. In this case the residual strength of the coupons continued to increase with cycling up to 106 reversals by which stage the residual strength of the laminate appeared to be independent of any form of initial defect. Examination of the C-scans indicated that a considerable extension of the damage zone had occurred in that specimen subjected to an impact energy of 2 J and 106 cycles of fatigue. The damage had extended between parallel splits along the length of the specimen to the aluminium end-plates. This damage zone extension resulted in a 40% increase in static strength. However, the consequence of this extensive damage after perhaps 107 cycles is not known and further work is required to clarify this.

I.-.

~

.""'~o

~

`% -%

600 EL

_w

~ 4oo

300

Mixed-woven laminates

Number of cycles -

200 -

Laminate WA

• . . . . .

O

• •. . . .

i04

x

IO6

IO5

....................

IOO

O 0

I I

I 2

l 3

I 4

I 5

I 6

Impact energy (d) Fig. 7. R e s i d u a l t e n s i l e s t r e n g t h vs i m p a c t e n e r g y f o r l a m i n a t e WB

Non-woven laminates

The results displayed in Fig. 6 show considerable scatter and trends are not absolutely clear. The reason for this unpredictable behaviour is not fully understood; the ultrasonic scans did not indicate substandard quality. The reduction in strength over the energy range considered was substantially less than that noted in the equivalent nonwoven laminate. Cycling this laminate once again resulted in improved residual strengths. After six decades of cycling the stress concentrating effect of the impact damage appeared to have been eliminated. The presence of widespread microdelaminations appeared to have no effect on the residual strength of the laminate and in fact these specimens were found to possess superior residual strengths compared with similarly damaged uncycled specimens.

Laminate NA

This stacking configuration exhibited considerable sensitivity to impact damage; a 50% reduction in static strength was noted over the energy range considered (see Fig. 3). Interestingly, no reduction in static strength resulted from an impact energy of 2 J, even though quite considerable delamination was noted in the C-scans. Cycling this laminate resulted in increases of up to 40% in the static strength. Fig. 2 indicates a 100% increase in delaminated area after 106 cycles of fatigue for a specimen which had been subjected to an impact energy of 4 J. This increase in delamination area is associated with an increase in residual strength.

900

800 7...~.~c/II

....."'............ .:...~

700



""--.... ~ ~

6oo

500

Laminate NB

The damage incurred by this laminate was found to be very similar in appearance and size to that observed in the previous laminate. Correspondingly, the cycled and uncycled static properties were akin to those noted in laminate NA (see Fig. 4). Improvements in static strength of up to 50% were noted after 104 cycles, but continued cycling did not result in further strength increases. A similar effect was noted by Ramani and Williams14 during a series of tests on centrenotched 0 ° + 45 ° laminates.

4oo-

:300-

Number of cycles • . . . . .

200-

0

• ~

iO 4

• . . . . .

t05

x .................. iO6

IO0

Laminate NC

The uncycled curve shown in Fig. 5 exhibits an almost linear decline in static strength with increasing impact energy. This feature is attributed to the fact that the laminate has load-

116

0 O

l I

I 2

I I 3 4 Impact energy (d)

I 5

I 6

Fig. 8. R e s i d u a l t e n s i l e s t r e n g t h i~ i m p a c t e n e r g y f o r l a m i n a t e WC

Int J Fatigue April 1984

Laminate WB The nature of the damage and its consequences on the residual properties of this laminate were similar to that observed in the previous laminates. 0-tension fatigue initiated a series of microdelaminations throughout the test specimens which again did not appear to degrade their residual strength. After six decades of fatigue all specimens yielded failure stresses that were comparable with the undamaged coupons. Laminate WC The reduction .of uncycled static strength with increasing impact energy was linear over the range of energies considered and similar to that noted in the corresponding non-woven composite. These similarities were carried through into the post-fatigue static performances. Cycling the laminate resulted in improved static performance and once again initiated the growth of microdelaminations.

Laminates NA and NB

Irnpoctenergy(J) Fig. 9 Schematic variation of residual tensile strength with impact energy

Discussion Impact damage tolerance

The laminates were impacted while supported between two rings so that the central impacting head gave rise to plate flexure. It is not thought, however, that elastic energy absorption mechanisms are significant. Rather, three processes - fibre cracking, matrix shear cracking and delamination between plies - are important. In general all three may occur but the relative extent of each depends upon the impact energy and laminate stacking sequence. Each process can have a different effect upon the residual strength of coupon specimens. Fibre cracking of 0 ° fibres results in an irreversible loss of strength and, for small amounts of cracking in a notchinsensitive laminate, the loss of strength is proportional to the number of cracked fibres and, in the absence of other absorption mechanisms, to the impact energy. Laminates NC and WC have 0 ° plies on their outer surfaces. In these laminates the tensile face under impact loading suffers fibre cracking as the major energy absorbing mechanism. The fibre cracking does not, however, extend beyond the 0 ° plies, so that the cracks are contained in, essentially, a notch-insensitive lamina. The variation of residual strength with incident energy in these laminates is dominated by 0 ° fibre cracking. Matrix shear cracking occurs in the + / - 4 5 ° plies. The location of the 0 ° plies in the outermost faces causes the flexural stiffness of the C-laminates to bc about 48% greater than the A-B laminates. Impact of the A-B laminates causes large tensile strains in the outermost surface +/--45 ° plies and shear cracking in these plies, together with delamination between these outermost two +/--45 ° plies for lay-ups NA and NB. The increasing impact energy increases the depth of the shear cracks until they approach the next 0 °plics. The effect of these cracks near the main load-bearing 0 ° fibres is to cause local stress raisers which promote notch sensitivity in the laminates. 11 The residual strengths of impacted laminates NA and NB are then expected to drop off rapidly with increasing impact energy, at least until the impact energy is sufficient to cause 0 ° fibre cracking in the third and fourth plies in from the outer faces, when the rate of decrease of strength would decrease. It should be noted that impact cracks ending within 0 ° plies tend to cause fibre splitting, which is a process which promotes notch insensitivity in residual strength tests. Schematic

Int J Fatigue April 1984

plots of the expected residual strength variations with impact energy would then have the forms shown in Fig. 9. Delamination is most likely to occur between plies in which the fibre orientation mismatch is greatest. This would be between + / - 4 5 ° plies in the non-woven laminates. The woven fabric has short-range sites for delamination within the weave but extensive delamination would be expected to take place between the +45 ° fabric and 0 ° plies. As well as providing a mechanism for absorbing impact energy, delamination can have an important r61e in determining residual strength levels. In particular, the site of delamination can affect local notch sensitivity.Is Delamination between + / - 4 5 ° does not change the notch sensitivity of the normally notch-sensitive 00/+45 ° laminate. However, delamination between 0 ° and 45 ° plies effectively uncouples the notch-insensitive 0 ° plies from the constraining effect of the 45 ° plies or fabric. Such delamination then deflects or blunts the matrix cracks. Hence, with laminates WA and WB, impact energy is absorbed by matrix cracking and delamination, the latter being in the form of microdelaminations within the weave and, more extensively, between 0 ° and +45 ° fabric. The resin-rich areas in the weave are sites for matrix cracks which lie at the interface between the fabric and 0 ° plies. The stress-raising effects of the matrix cracks are then not transmitted to the load-bearing 0 ° fibres. As a result, the impact damage tolerances of laminates WA and WB improve much more than those of NA and NB. In laminate WC, however, impact damage is absorbed also by 0 ° fibre cracking so that the behaviour of laminates NC and WC would be expected to be similar, at least until the impact energy is great enough to cause the cracking on the tensile face to extend through the outermost two 0 ° plies and into the adjacent + / - 4 5 ° plies or -+45° fabric. The experimental evidence presented here, however, suggests that the range of impact energies used was not sufficient to cause complete 0 ° fibre cracking in the outer plies. Fatigue growth of impact damage

The effect of subsequent fatigue on the residual strength of impact-damaged coupons depends upon the fatigue growth characteristics of the three main damage types. Matrix cracks propagate within + / - - 4 5 o plies after relatively small numbers of fatiguc cycles. However these cracks tend to be deflected at ply interfaces and become local delaminations.

117

Subsequent growth of these fatigue-induced microdelaminalions and impact delaminations depends, generally, upon the stacking sequence and particularly on the sign and level of normal stresses within the laminates. The C-scans indicated that delamination damage grew significantly between l0 s and 106 fatigue cycles in lay-up NA but not in NB. While impact delamination extended in the 45 ° fibre direction, subsequent fibre growth was mainly along the 0 ° fibre direction. Fatigue crack growth normal to 0 ° fibres is not usually observed in CFRP. The increase in strength of impacted coupons after fatigue results from fatigue growth of matrix cracks and subsequent deflection when they approach ply interfaces. The stress-raising effects of these cracks, which gave rise to large reductions in residual strength in impact laminates NA and NB, are then mitigated by fatigue. In lay-ups NC and WC the major part of the strength loss results from 0 ° fibre cracking. There is still, however, an improvement in residual strength after fatigue which suggests that matrix shear cracking occurs along with the fibre cracking and plays a similar r61e as in lay-ups NA, NB, WA and WB in increasing the residual strength. In the mixed-woven fabric the matrix cracks starting at the resin-rich areas grow rapidly along the interface between the 0 ° ply and fabric under fatigue loading. The residual strength of these laminates is then significantly increased by fatigue.

strengths showed little dependence on the incident impact energy. It is concluded that mixed-woven composites offer both improved impact resistance and post-impact fatiguc performance and, coupled with the ease of manufacture, it is concluded that mixed-woven laminates offer clear advantages over their non-woven counterparts.

Acknowledgement This work has been carried out with the support of the Procurement Executive, Ministry of Defence.

References 1.

Bradshaw, F.J., Dorey, G. and Sidey, G. R. 'Impact resistance of carbon fibre reinforced plastic' RAE TR 72240 (Royal Aircraft Establishment, UK, 1973)

2.

Chamis, C. C., Hanson, M. and Serafini, T. T. 'Designing for impact resistance with unidirectional fibre composites' NASA TN-D 6463 (1971)

3.

Adams, D. E. 'Impact response of polymer-matrix composite materials' Composite Materials Testing and Design 4th Conference, ASTM STP 617 (American Society for Testing and Materials) pp 409--423

4.

Husman, G. E., Whitney, J. M. and Halpin, J. C. 'Residual strength characterization of laminated composites subjected to impact loading' Foreign Object Impact Damage to Composites, ASTM STP 568 (American Society for Testing and Materials, 1973) pp 92--113

5.

Beaumont, P. W. R., Riewald, P. G. and Zweben, C. 'Methods for improving the impact resistance of composite materials' ibid pp 134--158

6.

Stellbrink, K. K. and Aoki, R. M. 'Effect of defects on the behaviour of composites' 4th Int Conf on Composite Materials Tokyo, Japan, October 1982

7.

Lifshitz, J. M. 'Impact strength of angle ply fibre materials' J Composite Mater 10 pp 9 2 - 1 0 1

8.

Dorey, G. 'Fracture of composites and damage tolerance' Agard LS124

9.

Sykes, J. and Stoakley, S. 'Impact penetration studies of graphite/epoxy laminates' National SAMPE Conf 23

10.

Dorey, G. "Damage tolerance in advanced composite materials' AHS/NASA-Ames Conf, 1977

11.

Bishop, S. i . and Curtis, P. T. 'An assessment of the potential of woven carbon fibre reinforced plastics for aerospace use' RAE TR 83010 (Royal Aircraft Establishment, UK. 1983)

12.

Bishop, S. M. and Curtis, P. T. 'Fibre reinforced composites containing woven and non-woven layers' Patent Application 8306653 (April 1983)

13.

Card, M. F. and Rhodes, M. D. 'G r aphi t e- epoxy panel compression strength reduction due to local impact' Agard CP 288, 1980 paper 9

14.

Ramani, S. V. and Williams, D. P. 'Notched and unnotched fatigue behaviour of angle ply graphite-epoxy laminates' ASTM STP 636 (American Society for Testing and Materials, 1976) pp 27--46

15.

Bishop, S. M. 'Effect of moisture on the notch sensitivity of carbon fibre composites' Composites 14 No 3 (1983) pp 2 0 1 - 2 0 5

Comparison of n o n - w o v e n and woven laminates

The experimental evidence clearly indicates that the inclusion of a woven fabric in a multidirectional composite can significantly enhance its impact resistance. This is particularly evident in lay-ups A and B where the mixedwoven composites possessed residual strengths of up to 50% greater than their non-woven counterparts. Reasons for the improved impact performance have been attributed to the nature of the woven fabric. Lay-up C offers a considerably superior impact damage tolerance to that of lay-ups A or B within the experimental range of impact energies. The inclusion of a woven fabric does little more to improve the impact resistance of this laminate. 0-tension fatigue cycling resulted in improved strengths in all laminates considered. This was particularly evident in the mixed-woven composites WA and WB; indeed, the strengths of the impact damaged coupons after 106 cycles were comparable to similar undamaged specimens. It is clear from the experimental evidence that replacing individual -+45° plies with a woven fabric can significantly enhance the impact resistance and subsequent 0-tension fatigue performance of an orthotropic composite. Combining this with thc superior drape and manufacturing capability, it appears that woven fabrics offer considerable potential for use in aeronautical structures.

Conclusions All the laminates examined exhibited reductions in tensile strength as a result of low velocity impact. The impact resistance of a composite has been found to be sensitive to the stacking sequence. Replacing the individual +-45° plies with a woven fabric resulted in panels displaying an improved impact performance. Cycling the non-woven composites led to improvements in static strength of up to 30%. Fatiguing the mixedwoven laminates resuhcd in improvements of up to 40% in the residual strengths and by 10 6 cycles the residual

118

Authors Mr Cantwcll and Dr Morton are with the Department of Aeronautics, Imperial College of Science and Tcchnoh)gy, Prince Consort Road, South Kensington, London SW7 2BY, UK. Dr Cures is with the Royal Aircraft Establishment, Materials and Structures Department, Farnborough, Hants ( ; U I 4 6TD, UK. Inquiries should bc addressed to Mr Cantwell.

Int J F a t i g u e A p r i l 1 9 8 4