Journal of Accident Investigation

MATTHEW R. FOX, CARL R. SCHULTHEISZ, JAMES R. REEDER, AND BRIAN J. JENSEN
the surface for matrix rollers (pieces of fractured matrix
material rolled into cylindrical shapes by the relative motion of
the fracture surface during cyclic loading), which would have
indicated fatigue. Fracture surfaces of the remaining samples
were cleaned ultrasonically in water before being coated with a
conductive layer of gold and palladium. 6 Typically, delamination
samples about 2 inches square were taken from widely spaced
areas on the exposed fracture surfaces in an effort to identify
overall trends. (See table 1.) Samples were also taken from
areas where the delamination surface morphology changed
(mostly at the ends of plies in the lay-up) to explore for local
differences in stress state or crack propagation direction.
Investigators took more than 300 SEM photographs of
translaminar fractures in the main attachment areas of the
vertical stabilizer and examined more than 1 0 square inches
of the delamination surfaces at high magnification. For
translaminar fractures intersecting the lug attachment hole, they
examined the entire fracture surface at high magnification, and
for translaminar fractures above the lugholes, they examined
several inches of the total extent of the fracture.
One challenge facing investigators during the fractographic
analysis was the relatively small amount of reference material
dealing specifically with fractographic examination of
fabric-reinforced composites. Most of the literature describing
fractography of composites focuses on unidirectional tape
lay-ups. However, fabrics have unique characteristics, such as
variation in resin content on delamination surfaces and less
fiber pullout in translaminar fractures relative to tape-reinforced
materials, as investigators found in the accident airplane. For
example, in the unidirectional lay-ups reported in the literature,
river marks were typically only observed in Mode I (opening
displacement between fracture faces) loading. However, in the
fabric construction of the accident airplane where evidence of
Mode II (sliding displacement between fracture faces) loading
was observed, river marks were also found in the matrix-rich
areas near the bundle crossings, and in the base of hackles 7 where
a bundle at one orientation transitioned to a perpendicular
crossing bundle. River marks in the bundle crossings were used
to identify a general direction of fracture propagation upward
and aftward for both of the large delaminations (at the forward
left and aft left attachments). (Investigators also explored the
river marks at the base of the hackles during their examination
6 A. Sjögren, L.E. Asp, and E.S. Greenhalgh, Interlaminar Crack Propagation
in CFRP: Effects of Temperature and Loading Conditions on Fracture
Morphology and Toughness, in Composite Materials: Testing and Design, and
Acceptance Criteria ASTM STP 1416, Nettles and Zureick, eds., 2002.
7 “Hackles are matrix fracture features that indicate a significant
component of shear across the fracture surface. Hackles are formed
when matrix microcracks that are spaced fairly regularly along planes of
maximum tension join together.” National Transportation Safety Board,
In-Flight Separation of Vertical Stabilizer, Aircraft Accident Report NTSB/
AAR-04/04, NTSB Public Docket (Washington, DC: NTSB, 2004).
of the delaminations at the forward right lug as described later
in this paper.) Because manufacturers are increasing their use
of composites with fabric reinforcements in airplane structures,
more research is needed to characterize fracture surfaces
generated under controlled laboratory conditions to help failure
analysts in interpreting fractographic details.
Fracture Surface Observations and Discussion
During the visual examination, investigators found that the
vertical stabilizer was largely intact with no significant areas of
skin buckling. An overall view of the vertical stabilizer as it
was being recovered from the water of Jamaica Bay is shown in
figure 4. At the lower end, each of the six attachment locations
had separated from the fuselage either by fractures that
intersected the lug attachment hole or by fractures through the
structure above the hole. A schematic of the lower end of the
vertical stabilizer is shown in figure , which shows a general
fracture location for each lug, pointing to overall views of each
lug fracture. Portions of rib 1, the rib 1 rib-to-skin attach angle,
and the lower end of the forward spar also were fractured. In
addition, the trailing edge panels were damaged in several
locations. 8
Description of Main Lug Fractures
Translaminar fractures on the right aft, right forward, and left
forward main lugs intersected the attachment hole. For the
remaining three main lugs, translaminar fractures intersected
the structure above the lug. Each of the lugs had delaminations
in the lug area and/or in the structure above the lug. Safety
Board Materials Laboratory factual reports contain details of
the fractographic examination. 9 Some of the delaminations
extended into the main portion of the vertical stabilizer,
and the extent of these delaminations was determined using
nondestructive inspection (NDI), including ultrasonic
inspection and x-ray-computed tomography scanning and
imaging. Safety Board Materials Laboratory factual reports
contain the results of the NDI. 10
MACROSCOPIC FRACTURE FEATURES
On the right side of the vertical stabilizer, the roughness of the
main lug translaminar fractures was in general consistent with
overstress fracture in primarily tensile loading. Delaminations
8 National Transportation Safety Board, Materials Laboratory Factual Report
02-083, NTSB Public Docket, 2002.
9 (a) NTSB, Materials Laboratory Factual Report 02-083, NTSB Public
Docket, 2002; (b) NTSB, Materials Laboratory Factual Report 03-018,
NTSB Public Docket, 2003.
10 (a) NTSB, Materials Laboratory Factual Report 02-078, NTSB Public
Docket, 2002; (b) NTSB, Materials Laboratory Factual Report 03-033,
NTSB Public Docket, 2003.
14 NTSB JOURNAL OF ACCIDENT INVESTIGATION, SPRING 2006; VOLUME 2, ISSUE 1