CHAPTER

Fleet impact resulting from a

space shuttle Columbia main
engine controller wire failure
during Mission STS-93
4
Steven J. McDanels
National Aeronautics and Space Administration (NASA), Kennedy Space Center, Florida, USA

CHAPTER OUTLINE
1 Space Shuttle Columbia Wiring Hardware Overview .............................................. 76
2 Investigation ...................................................................................................... 77
3 Conclusion ......................................................................................................... 84
Acknowledgments .................................................................................................... 86
References .............................................................................................................. 86

Space Shuttle mission STS-93 is notable for being the first NASA mission with a
female commander, Eileen Collins, and for launching the Chandra X-ray Observa-
tory, the most advanced X-ray observatory in existence at the time of its launch. It is
also notable for bringing to light a concern over the condition of electrical wiring
used throughout the orbiter fleet. The Space Shuttle Columbia lifted off from NASA
Kennedy Space Center Launch Complex 39B on July 23, 1999 (Figure 4.1). Approx-
imately 5 s after lift-off, NASA flight controllers observed a drop in voltage in one of
Columbias electrical buses [1]. The voltage drop caused a disabling of the primary
digital control unit (DCU-A) for the center engine and the backup digital control unit
(DCU-B) for the right engine. However, with built-in primary/backup redundancy
for each engine, the ascent was able to proceed and a contingency abort was averted
despite the loss of redundancy. Once orbit had been achieved, NASA flight control-
lers requested that the crew verify that the circuit breaker for the suspect main engine
controller had indeed opened; the astronauts verified that the breaker had opened,
indicating that the issue was local and not wider ranging [1].
Following a successful mission, the Columbia returned to Earth on July 28, 1999;
post-flight inspection revealed that a main engine controller wire had arced against
the head of an adjacent screw (Figure 4.2). Likewise, a second area of exposed con-
ductor was observed on the same wire approximately 2 in. away from the arced
region. The second area of damage was also located above a nearby screw.

Handbook of Materials Failure Analysis With Case Studies from the Aerospace and Automotive Industries.
http://dx.doi.org/10.1016/B978-0-12-800950-5.00004-1
Copyright 2016 Elsevier Ltd. All rights reserved.
75
76 CHAPTER 4 Fleet impact of space shuttle wire failure

FIGURE 4.1
Launch of STS-93. The Columbias three main engines functioned nominally despite the loss
of redundancy of the primary digital control unit (DCU-A) for the center engine and the
backup digital control unit (DCU-B) for the right engine (SSMEs during lift-off, left; SSMEs
after touch-down, right).
Credit: NASA.

FIGURE 4.2
Arced wire (solid white arrow) and nearby damaged area (solid black arrow) with associated
screws (open white arrows) in the midbody aft port-side lower wire tray 11/12 frame.

1 SPACE SHUTTLE COLUMBIA WIRING HARDWARE OVERVIEW

The space shuttles each had over 200 mile of wiring, weighing over two-and-a-half
tons, in addition to associated cables, conduits, and trays as well as hardware to house,
route, and contain all of the materials. Although the initial investigation focused on
only several inches of the wiring, the ramifications extended to all 200-plus miles
of wiring aboard the Columbia, and impacted the entire Shuttle fleet [2].
 2 Investigation 77

The subject wire provided power to the digital control units and was a 14 Amer-
ican wire gauge (AWG) polyimide-insulated twisted three-wire conductor, with
Kapton polyimide insulation surrounding the nickel-plated copper conductor.
The insulation was topcoated with an aromatic polyimide resin color coded to indi-
cate AWG size, with green corresponding to 14 AWG and red to 20 AWG. Although
the proximate cause of the loss of signal redundancy was found in the form of a short
circuit caused by the wire arcing to an adjacent screw, the root cause of the arcing
event had to be ascertained.

2 INVESTIGATION
A failure analysis was initiated to determine the cause of the main engine controllers
short circuit. The shorted main engine controller wire and nearby damaged wire were
located in the midbody aft port-side lower wire tray 11/12 frame (Figure 4.3). Both
the arced screw and the screw beneath the second damaged area had exposed base
metal where the screw head slots had been deformed. It should be noted that the wires
180 from the damaged areas appeared undamaged.
The shorted and nearby damaged wires were harvested along with their corre-
sponding screws. Further inspection revealed that an additional seven other wires
in the bundle also displayed varying degrees of damage; these wires were likewise
removed from service for analysis.
Several intact wires were also removed for laboratory testing in order to simulate
various forms of mechanical damage and generate exemplars. This testing consisted
of the following methods: a step test, where a person literally stepped on a wire bun-
dle that rested upon a screw head; a shear test that forced a screw head and a torque
tip against a wire in an attempt to damage the insulation and conductor; a vibration
test to determine if a wire bundle would abrade when vibrated against a screw head;
an abrasion test to determine if the polyimide insulation would become damaged by

FIGURE 4.3
Shorted wire bundle with arced screw removed (arrow) in the midbody aft port-side lower wire
tray 11/12 frame.
78 CHAPTER 4 Fleet impact of space shuttle wire failure

rubbing against helical convolex convoluted tubing; and an impact test to determine
if convoluted tubing would protect a wire bundle from a foreign object strike, such as
by a torque tip or screwdriver bit.
Visual and low-magnification examinations of the various wires were performed.
The arced conductor wire (Figure 4.4) displayed several areas of mechanical damage
adjacent to areas that appeared melted, corresponding to the region of arc tracking.
Dark deposits of apparently burned material were evident over much of the exposed
areas. Deposits were also evident covering some of the nearby mechanically damaged
area. The topcoat adjacent to the shorted area appeared deformed. Similar examination
of the screw head that the wire arced against revealed a definite region of arc-tracking
damage (Figure 4.5). Raised areas of bare base metal were also evident on the sides of
the slots in the screw head; these raised areas were devoid of paint.
Macroscopic examination of the second damaged area revealed that the insula-
tion had been pushed up (Figure 4.6). Mechanical gouges on the conductors were
observed. The directionality of the mechanical damage in this area corresponded
well to that of the mechanical damage in the arced region.

FIGURE 4.4
Rotational view of shorted wire displaying arced region (white arrows) and mechanically
deformed region (black arrows).

FIGURE 4.5
Screw head displaying arc tracking and raised areas.
 2 Investigation 79

FIGURE 4.6
Mechanically damaged area approximately 2 in. from the shorted section of the main engine
controller wire displaying deformation of the conductors and topcoat.

FIGURE 4.7
SEM micrograph of the damaged region of the shorted wire. Mechanical damage is evident on
the conductors (denoted MECH). The melted region is between the "MELT" and "MVC"
zones. Magnification: 27.

Next, the wires were analyzed via scanning electron microscope (SEM), energy-
dispersive X-ray spectroscopy (EDS), and electron spectroscopy for chemical anal-
ysis (ESCA). SEM analysis of the shorted wire verified that the polyimide insulation
was missing in the arced area, with the wires conductors clearly evident (Figure 4.7).
It should be noted that the damaged insulation was even and smooth, as opposed to
serrated. Three distinct topographical features were evident on the shorted wire: a
mechanically damaged zone, a melted zone, and a zone displaying microvoid coa-
lescence (MVC). The mechanically damaged zone appeared gouged (Figure 4.8).
The damage was directional, roughly tangential to the length of the wire. The central
portion of the exposed area contained a mixture of mechanical damage and melted
metal (Figure 4.9). The melting was confined to the outer layer of wire strands and
did not appear to penetrate to the inner strands. The final region of the exposed area
displayed MVC (Figure 4.10), typical of ductile overload. The overload area corre-
sponds to the region where the wire actually melted and fused with the screw head
FIGURE 4.8
SEM micrograph of shorted wire displaying mechanical damage to the conductors.
Magnification: 90.

FIGURE 4.9
SEM micrograph of shorted wire illustrating mechanical damage and thermal damage.
Magnification: 370.

FIGURE 4.10
SEM micrograph of shorted wire displaying MVC, typical of ductile overload. Magnification:
1500 .
 2 Investigation 81

FIGURE 4.11
SEM micrograph of shorted screw head displaying area of resolidified metal and MVC.
Magnification: 15.6 .

FIGURE 4.12
SEM micrograph of a second damaged area from the shorted wire. The conductors appeared
mechanically damaged, with the topcoat pushed back. Magnification: 12.

(Figure 4.11). SEM examination of the damage 2 in. from the shorted region of the
controller wire showed evidence of mechanical damage but no arcing (Figure 4.12).
EDS and ESCA revealed the presence of nickel and copper on the arced screw
head in the region of resolidified metal. ESCA, which typically has a surface sensi-
tivity of approximately 10 nm, revealed the presence of an oxide layer on the shorted
wire. The depth of this layer was compared to that of three laboratory exemplars: a
freshly nicked portion of copper wire, a 2-week-old exposed piece of copper, and
a sample that had been exposed a year previously during another investigation.
82 CHAPTER 4 Fleet impact of space shuttle wire failure

The depth of the oxide layer on the shorted wire was substantially thicker than that of
either freshly exposed exemplar and was approximately four times thicker than the
year-old samples. ESCA was then performed on an area of damage near the shorted
section of wire that also had damaged conductors. A precise timeframe for when the
damage to this region occurred could not be established with certainty, but based
upon the topography, directionality, and extent of damage, it likely occurred concur-
rently to the mechanical damage found in the arced region. ESCA of this additional
damage disclosed a substantially thicker oxide layer than that of the fresh exemplars
and similar to that of the arced region. The thickness of the oxide layer indicated an
approximate exposure time of 4-5 years [3,4].
Of the seven additional wires removed for analysis, neither damage to the con-
ductor nor serration of the insulation was noted (Figures 4.13 and 4.14).
Mechanical tests were performed to duplicate the launch environment and dam-
age characteristics of the service failure. Step tests, even under increasing effort and
force, did not damage the conductor when forced against a smooth screw head. SEM
examination of the step-test specimens showed only topcoat/insulator damage. No

FIGURE 4.13
Macrograph of typical damage observed on additional wires.

FIGURE 4.14
SEM micrograph of additional wire typifying topcoat and insulation damage.
Magnification 37.
 2 Investigation 83

conductor damage was observed (Figures 4.15 and 4.16). However, subsequent step
testing performed after the primary investigation was concluded did cause damage to
the conductor. Vibrational testing at frequencies ranging from 20 to 2000 Hz and
loads varying from 600 to 2100 g damaged the conductor in the 2100 g range. Single
event drop testing meant to simulate a sharp object impacting the bundle carried out
with drop heights of 1-3 ft and loads of 900-20,000 g did not damage the conductor
until 19,050 g were added to the impacting tool. Additional vibrational abrasion and
drop testing verified that the convoluted tubing would be damaged before any dam-
age to the conductor would occur.

FIGURE 4.15
Typical step-test damage.

FIGURE 4.16
SEM micrograph of step-test exemplars. Mechanical deformation of insulation is evident with
no damage to conductor. Magnification: 75, 95.
84 CHAPTER 4 Fleet impact of space shuttle wire failure

3 CONCLUSION
The subject main engine controller wire displayed evidence of mechanical damage,
both in the arced region as well as the damaged area 2 in. away. ESCA results showed
that the oxide layers in these regions were substantially thicker than the oxide layers
of recently exposed pieces of conductor. The depths of the oxide layers, along with
the fact that the oxide layer was deposited above the mechanical damage, indicated
that the arced wires deformation was present prior to the short. Although oxide
would have been deposited during the arcing event, the fact an area which displayed
similar mechanical damage to the arced section yet was 2 in. away from the arced
region also contained an oxide layer implied that the damage likely occurred at
approximately the same time. Maintenance records indicated that the last time these
wiring trays were exposed was 4 years prior, which corresponds well to the estimate
of the age of the oxide accumulation [4].
It is likely that the exposed base metal of the screw resulted from deformation of
the screw head during insertion or removal of the fastener during its service life, pos-
sibly during maintenance 4 years earlier. Contact between the exposed conductor and
the base metal of the screw head likely resulted due to launch-induced vibration.
Once the circuit was energized, a short developed. Chemical testing verified the pres-
ence of copper and nickel in the melt zone on the iron-based screw head, indicating
that the wire and screw did make contact at some point. The wire fused with the
screw until a subsequent applied mechanical force separated them. The MVC on
the fused area indicates that the force was a single event, as no evidence of a pro-
gressive separation was observed.
Additionally, the fact that the subject wire was damaged in two places, each
above screw heads with raised, bare base metal, indicated that similar events may
have caused damage to both areas. The mechanical simulation tests showed that
vibration and physically stepping on the components when pressed against smooth
screw heads would likely have not caused the damage. Although the step tests per-
formed during the investigation did not damage the conductor, even against burred
screw heads, subsequent testing was able to damage the conductor by stepping on a
simulated bundle using additional orientations of the hardware. If something had
been dropped on the wire bundle, the adjacent wires immediately surrounding, as
well as those 180 from, the damaged areas might not have been damaged because
the bundle could absorb the shock due to its compressibility, however the bundle as a
whole would still be forced onto the screw heads, causing localized damage. The two
most probable scenarios for the damage occurrence are either a strike with a tool,
such as with a torque tip or pliers, or something being dropped or placed on the wire
bundle, causing the bundle to be forced against adjacent burred screw heads with
raised, bare metal, and damaging the conductors; stepping on the bundle and pressing
it against a burred screw head would fall into the latter category.
Three previous occurrences of wires shorting had been reported during the shuttle
program to that point. Due to the incidence of arc tracking on STS-93, an
 3 Conclusion 85

FIGURE 4.17
Orbiter midbody before inspection and during wiring inspection. Tags indicate areas of
potential damage to the wiring.

unprecedented inspection of wiring throughout the orbiter fleet was undertaken.

Accessible areas as well as critical hard-to-access locations were examined using
inspection mirrors, flashlights, and 10  magnification. Manipulation of the wiring
was kept to a minimum to lessen the likelihood of introducing any further damage.
Between the four orbiters, the Atlantis, the Columbia, the Discovery, and the
Endeavour, approximately 3000 suspect indications were identified (Figure 4.17)
[5]. Of the 3000 discrepancies, nearly 500 involved exposed conductor. Of the
exposed conductors, approximately 150 had mechanical damage, although no addi-
tional instances of arc tracking were reported.
Following the inspection of the orbiter fleets wiring, several lessons were
learned. It was recommended that similar screws adjacent to wire bundles be exam-
ined to make sure no metal was raised on the screw heads, and also that no screws
have bare metal exposed. If any were found, the screws were to be replaced or the
raised areas smoothed and the screws repainted. The use of convoluted tubing was
recommended as beneficial, both as a physical protection for the bundles, as well as
an insulator for any conductors which might become exposed.
Strict adherence to procedures and processes to minimize potential damage dur-
ing installation and maintenance of wiring and surrounding hardware is imperative.
Ongoing training was initiated to ensure that proper precautions were taken when
working in or around wiring. Handling and access was also addressed, with the
use of temporary workstands and covers to protect vulnerable components being
emphasized. Also, the use of convolex convoluted tubing was recommended in
high-traffic and critical areas to minimize the likelihood of damage, whether through
vibration or impact, as well as to provide protection against similar short circuits.
The lessons learned, both through the failure analysis of the STS-93 main engine
controller wire, as well as the fleet-wide inspection which followed, will be incor-
porated into all subsequent programs. Whether returning to the Moon, landing
humans on Mars, or visiting other extra-terrestrial bodies, electrical wiring is a crit-
ical component of any spacecraft; ensuring the integrity of such vital hardware is of
paramount importance.
86 CHAPTER 4 Fleet impact of space shuttle wire failure

ACKNOWLEDGMENTS
The author would like to acknowledge the talents and contributions of:
Peter Marciniak, NASA, Kennedy Space Center
Kirk Scammon, University of Central Florida/Advanced Materials Processing and Analysis
Center (AMPAC)

REFERENCES
[1] STS-93 Mission Control Center Status Report # 1. http://www.nasa.gov/centers/johnson/
news/shuttle/sts-93/STS-93-01.html.
[2] Managing Electrical Connection Systems and Wire Integrity on Legacy Aerospace Vehi-
cles. S. Sullivan, Kennedy Space Center, National Aeronautics and Space Administration,
Kennedy Space Center; George A. Slenski, Materials Directorate, Air Force Research
Laboratory Research Laboratory, AFRL/MLSA, Wright-Patterson AFB. Federal Aviation
Authority Principal Inspectors and Engineers Aging Aircraft Workshop, Seattle, WA;
2001.
[3] Failure Analysis of a 115 VAC 400 Hz Main Engine Controller Wire From the Midbody
Port Side Lower Wire Tray 11/12 Frame From OV-102 That Shorted During the Launch of
STS-93, KSC-MSL-07291999, S. McDanels, National Aeronautics and Space Adminis-
tration, Kennedy Space Center; 1999.
[4] ESCA Analysis of Shuttle Columbia Wiring final report. K. Scammon, University of Cen-
tral Florida/Advanced Materials Processing and Analysis Center (AMPAC); 1999.
[5] Fleet Wire Inspection & Repair Summary. D. White, United Space Alliance Final Report;
October 1999.