POSTIMPACT BEHAVIOR OF AEROSPACE COMPOSITES FOR HIGH-TEMPERATURE APPLICATIONS - EXPERIMENTS AND SIMULATIONS

Authors
Citation
G. Sala, POSTIMPACT BEHAVIOR OF AEROSPACE COMPOSITES FOR HIGH-TEMPERATURE APPLICATIONS - EXPERIMENTS AND SIMULATIONS, Composites. Part B, Engineering, 28(5-6), 1997, pp. 651-665
Citations number
32
Categorie Soggetti
Materials Sciences, Composites",Engineering
ISSN journal
13598368
Volume
28
Issue
5-6
Year of publication
1997
Pages
651 - 665
Database
ISI
SICI code
1359-8368(1997)28:5-6<651:PBOACF>2.0.ZU;2-W
Abstract
The post-impact performance of different carbon-fabric-reinforced comp osite materials were studied experimentally and analytically. Three ty pes of thermosetting matrix were considered: conventional epoxy, high- temperature curing epoxy and epoxy-isocyanate. Experimental testing co nsisted of impacting rectangular specimens at different energy levels by using a spring-driven impact apparatus that was able to impart velo cities of up to 5 m s(-1) to masses of 0.5, 1.0, 2.5 and 5.0 kg travel ling horizontally. After impact, coupons were nondestructively inspect ed by means of opaque-enhanced dye-penetrant X-radiography and tested in static compression to correlate impact energy, damage extent and re sidual strength. Epoxy composites contain damage within a narrow regio n, while epoxy-isocyanate materials propagate the damage far away from impact point. Epoxy composites show an asymptotically decreasing fail ure strength with impact energy up to a lower threshold (0.3-0.4 times that of the undamaged material), while epoxy-isocyanate material show s a trend of over-decreasing residual strength. An analytical study wa s performed by means of the finite element code PAM-FISS, used to simu late the compression-after-impact (CAI) tests. Type, size and location of damage, as well as the mechanisms leading to final failure, were r eproduced quite well by the finite element analysis (FEA), while some discrepancies between FEA and experimental CAI residual strength tests were found (7% for undamaged specimens and 10% for blister-delaminate d specimens); higher errors were found in the case of completely delam inated specimens, mainly owing to the inability of the present softwar e and hardware to conveniently model the complete state of damage. (C) 1997 Elsevier Science Limited.