You are on page 1of 12

OTC 3742

INVESTIGATION OF RANGER I COLLASPE


METALLURGICAL AND FRACTURE ANALYSIS

by R.C. Jerner, John A. Harcourt, and Charles W. Powell,


Engineering & Materials Technology Corp.

©Copyrlght 1980 Offshore Technology Conference

This paptlr was presented at the 12th Annual OTe In Houston, Tex., May 5-8, 1980. The material Is subject to correction by the author. Permission to copy Is restricted to an abstract of not more than 300 words.

ABSTRACT shown in Figure 1. The starboard forward leg


failed in bending at the 38 foot level. The port
The May 10, 1979, collapse of Ranger I, south forward leg separated at the 68 foot level. Figure
of Galveston, Texas, occurred as a result of initi- 2 is a top schematic plan of the mat, leg nomencla-
ation and growth of a fatigue crack in the aft leg/ ture, and the orientation referenced herein. The
stiffener fillet weld runout. The 23 inch fatigue aft leg was completely separated and hairpinned
crack apparently developed and grew over the struc- over the mat structure. The lower portion of the
ture's lifetime. After the fatigue crack became aft leg, its fractured end partially buried in bot-
critical and complete leg separation occurred, the tom mud, was inside the "moon pool". The upper aft
leg moved laterally and dropped to the mat surface. leg portion, originally near the hull, was buried
The leg apparently remained in this general posi- outside the mat structure. A diagram showing the
tion for a period of time while settling into an approximate leg location and attitude is shown in
attitude approximately 9· to II· off vertical. Ap- Figure 3. The final mat attitude, with respect to
plied loading became critical and the aft leg col- the ocean floor, can be pictured by reference to
lapsed at the 44 foot level. This in turn allowed Figure 1.
the aft end of the upper hull to fall. Hull side
failure of the aft leg together with bending and Forward Starboard Leg Recovery
separation of the forward legs followed.
The two forward legs were retrieved prior to
attempting aft leg recovery. The forward starboard
INTRODUCTION leg was secured with a lifting line to the barge
derrick and separated from the mat by circumferen-
The authors were requested by the United tially cutting at approximately the 24 foot level.
States Coast Guard Marine Board of Investigation to The freed leg was then lifted to the barge for
travel to the site of the collapse of Ranger I in close inspection. The fracture surfaces were pho-
the Gulf of Mexico approximately twelve miles south tographed, macroscopically examined, washed with
of Galveston, Texas. The purpose of this trip was clear water, and oiled to inhibit corrosive action.
to provide on-site advice and consultation concern- The upper leg separation, at the 89 to 90 foot
ing the recovery, significance, and preservation of level, occurred predominately in shear. The frac-
fractured members which might have contributed to ture surface associated with the collapse, which
structural collapse. They were also requested to occurred at the 38 foot level, was macroscopically
conduct detailed metallurgical studies of pertinent examined. Tensile overload fractures extended from
fractures and to provide a report of factual find- 270· through O· to 90·. The fracture probably
ings. Ranger I, a self-elevating mobile drilling initiated in the upper corner of the 270· pinhole
unit, was located in about sixty feet of water when and propagated above the O· pinhole to approximate-
the mishap occurred on May 10, 1979. ly the 30· position. Another fracture initiated at
the 90· pinhole and propagated toward the O· pin-
hole but veered downward at termination. The leg
INVESTIGATION also experienced clockwise torsional deformation.
It was determined that this leg failure occurred
Wreckage Recovery and Macroscopic Examination subsequent to the aft leg collapse.

A view of the lower or mat section wreckage as Forward Port Leg Recovery
initially observed from the recovery vessel is
A similar recovery procedure was followed for
the forward port leg. Failure at the 68 foot level
References and illustrations at end of paper. occurred primarily in the aft direction. Maximum

59
tension was noted from 270· to 90·. The fracture (MSPF). Deformation of the fractured upper leg and
was primarily slant and had some torsional clock- the mat surface conformed perfectly. Two peaks in
wise component. Again, it is believed that this the mat surface resulting from support given by
leg failed following the aft leg collapse. underlying vertical bulkhead and structural sup-
ports were observed to match with HSPF deformation.
Aft Leg Recovery Leg loads were then transferred to the leg/mat con-
tact area and other HSPF and MSPF contact points.
A line was then secured to the submerged aft The HSPF and MSPF subsequently sheared at about the
leg near the collapse (44 foot level) and the leg 180° and 330· locations (see Figure 6).
lifted to the recovery vessel deck. A long strand
of cable embedded in bottom mud and partially con- As the 180° and 330· shearing occurred (by
tained within the fractured leg was also recovered. compressive loading), increased loading, deforma-
The lower aft leg primary fracture and entrapped tion, and fracture were experienced by the leg in-
cable are shown in Figure 4. The aft leg hull side terior annular baffle (see Figure 5). Some defor-
failure occurred primarily by tensile tearing; how- mation of the remaining aft leg stub, from 30· to
ever, some shear component was noted. The hull 120°, probably resulted as the upper leg settled
side fracture occurred after the lower leg/mat off vertical. Prior to final collapse, the leg sat
separation and (44 foot level) collapse. The lower at an off-vertical attitude of approximately 9° to
aft leg wall near the primary fracture surface, 11° (see Figure 7).
shown in Figure 4, was thinned and heavily abraded.
Massive thinning of approximately eight inches of
the primary fracture negated any analysis in that NON-DESTRUCTIVE LNSPECTION
region. This thinned area was probably the result
of rubbing contact with the mat following aft leg Magnetic Particle and Ultrasonic Inspection
collapse. Subsequently, this area was confirmed to
contain the main fracture origin. Barnacles were removed in the stiffener area
from both forward legs and the aft leg stub which
Mat and Aft Leg Stub Recovery and Examination were then non-destructively inspected using ultra-
sonic and magnetic particle techniques. Briefly,
The salvage operation then focused on mat and this inspection indicated the presence of cracks at
mat/aft leg stub recovery. The mat was surfaced, the leg/stiffener and/or stiffener/mat welds at
towed to the Houston International Terminal, and approximately the 0°, 90·, and 180· positions on
visually inspected. Slight deformation was ob- the aft leg stub. Crack indications were also
served at the O· position in both forward legs identified near structural and baffle joints at
about four feet above the mat surface. various locations in the forward port and starboard
leg stiffening ring area.
An overall view of the aft leg stub is shown
in Figure 5. The reference directional orienta-
tions are noted and conform to those indicated in LABORATORY MACRO EXAMINATION
Figure 2. The fracture, now recognized as the pri-
mary fracture, initiated at approximately the 270· Aft Leg Primary Fracture
location in the upper toe of the leg/stiffening
ring fillet weld and propagated in both clockwise In order to facilitate a more detailed inspec-
and counterclockwise directions. Frac ture topo- tion, examination, and sampling, the aft leg stub,
graphy was flat with no readily identifiable shear surrounding mat area, and the intersecting vertical
lip, indicating either progressive fatigue failure bulkheads were cut and removed to a controlled lab-
or rapid crack growth under restraint conditions oratory environment. Figure 8 shows the aft leg
adequate to develop brittle fracture. As crack stub and surrounding mat after removal but prior to
growth progressed, the crack tip stress state was cleaning and sectioning. The circumference of both
altered and the propagation direction departed from stub and leg fractures were divided into approxi-
the fillet weld toe upward into the leg. Final mately equal segments, numbered, and photographed.
separation of the primary fracture occurred at The segments were numbered clockwise from the 0°.
about the 45· position, approximately 15 inches position. Primary fracture origin, (PFO), was con-
above the mat surface. After separation of the leg firmed in the 270° region.
near the mat surface, the upper leg segment moved
laterally in the 270° direction and dropped about 270· Secondary Fracture
three inches, impacting the mat surface. As a re-
sult of this shift and three to four inch drop, the The mat side aft leg stub 270° area contained
primary fracture origin (being the "lowest" point not only the main fracture origin, but also several
and also located at approximately 270°) was the re- secondary fractures which exhibited evidence of
gion of primary contact between the dropping, frac- progressive or fatigue failure. Relationship and
tured upper leg, and mat. The mat surface near nomenclature of the secondary fracture is shown in
this location was heavily deformed and, as a re- Figure 8.
sult, the brittle barnacle layer was fragmented and
detached in the area around the 270· location. 1. Aft leg primary fracture or1g1n (PFO)
2. Secondary S-l (stiffener/mat weld runout)
During the aft leg examination, the hull side 3. Secondary S-2 (stiffener/mat weld runout)
of the primary fracture (HSPF) was cut from the re- 4. Secondary S-3 (stiffener/mat weld runout)
mainder of the leg in an attempt to "fit-up" with
the stub or mat side of the primary fracture

60
The fracture surface near 270· was cut and The section ultimate strength was calculated to be
cleaned for close macro- and microscope examina- approximately 76,000 psi.
tion. The lower half of the S-l fracture, showing
a flat, smooth fracture, typical of progressive or
fatigue failure, was also cut for examination. FRACTURE MECHANICS TESTING AND ANALYSIS
Initiation occurred at the stiffener/mat weld run-
out. Three distinct stop marks were observed, in- Fracture mechanics testing and analysis of re-
dicating at least three significant changes in load sults from actual leg material led to the conclu-
spectrum were experienced during the propagation sion that the most realistic range for the plane-
time frame. stress fracture toughness (K c ) was BO to 110 ksi.
Assuming the lower range value of fracture tough-
270· Secondary S-4 Fracture ness and a tensile stress level of 20 ksi, results
in a critical crack size of 10.2 inches. Converse-
A section was then cut from the primary or~g~n ly, if the flaw size at failure can be established,
area, including the stiffener/leg fillet weld and the tensile stress level can be calculated. For
the upper half of the S-l fracture. As the main example, if it is assumed that the critical crack
fracture surface was being removed, a previously size is 23 inches, a tensile stress level of only
unidentified cracked region, emanating from the 13.3 ksi is required for unstable propagation.
leg to stiffener fillet weld root and propagating Electron microscopic examination, discussed below,
along the weld/ stiffener interface, was revealed. revealed the existence of a 23 inch fatigue crack.
This fracture, S-4, was cut for scanning and trans-
mission electron microscopy.
METALLOGRAPHY AND MICROHARDNESS

CHEMICAL ANALYSIS A section was cut from the aft leg/stiffener/


mat joint at the forward or O· position. Longitu-
Aft Leg Quantitative Analysis dinal metallographic sections were prepared con-
taining the leg/upper fillet weld/stiffener and the
Quantitative chemical analysis was performed stiffener/weld/mat joints.
on specimens cut from various leg and mat loca-
tions. Resul ts indicated that the material con- Close examination of the stiffener/leg joint
formed to all initial and present (some leg materi- weld toe runout and root area revealed several
al had been recently replaced) specifications. small cracks. It should be noted that the primary
fracture origin initiated at an identical location
at the 270· position. Microstructure of the leg,
MECHANICAL TESTING stiffener, and mat material was examined and ap-
i
peared normal.
Tensile Tests
Standard 0.505 inch diameter tensile test ELECTRON MICROSCOPY
specimens were cut from various leg locations.
Specimen axis was parallel to the leg axis and Scanning and transmission electron microscopic
transverse to the final rolling direction. Results examinations were performed on several aft leg
obtained are tabulated in Table 1 • fractures to ascertain failure mode and extent of
any progressive fracture.
Charpy Impact Tests
Fracture Mechanics Test Specimens
Standard Charpy V-notch (CVN) impact specimens
(ASTM A370, Type A) were tested to determine upper Preliminary macroscopic evidence strongly sug-
and lower shelf fracture energies and transition gested that the primary failure mode was fatigue.
temperature behavior. Specimens were cut trans- Documentation of fatigue and overload failure modes
verse to the final rolling direction and notched in A-36 steel with the crack growth orientation
perpendicular to the plate surface. Results foi similar to that experienced in the subject aft leg
each temperature including impact value, percent was deemed imperative. Several three-point slow
shear area, and lateral expansion were obtained. bend test specimens used for fracture toughness
testing were precracked at various levels of stress
A plot of impact value vs. temperature indi- intensity and then were subjected to scanning and
cated that the 15 foot-pound transition temperature transmission electron microscopy. The nature and
for material near the primary fracture was approxi- topography of fatigue in A-36 leg material subject-
mately 40·F. The 50 percent shear transition tem- ed to a IJ. K range of 30 ksi is observed in the
perature was approximately BO·F. scanning micrograph in Figure 9. Figure 10 is a
transmission micrograph from the same region.
Full Leg/Stiffener Weld Section Test
Secondary Fracture S-l
The leg/stiffener weld crack indication previ-
ously identified by magnetic particle inspection at Secondary fracture, S-l, was chosen for exami-
the 90· location was removed for full section ten- nation as it appeared to contain a fatigue region
sile testing at 50·F. A full section tensile spec- and a fast fracture zone created as the upper aft
imen, pulled to failure, contained a 0.06 inch deep leg dropped to the mat surface. The section chosen
fatigue crack and exhibited very limited ductility. for SEM and TEM studies is shown in Figure 11 where

61
the fast fracture and fatigue zones are noted. inches, has an origin probabiy located near the
Results from the fast fracture region indicated 240” region.
that the failure mode was quasi-cleavage. These
micrographs, from an area of known overload frac- Through-the-wall fatigue, as determined by a
ture, subjected to the same post collapse environ- TEM indication of fatigue at the I.D. surface, ex-
ment as the primary fracture, provide a basis for tends from sections O to 6 and has an origin near
comparison with primary fracture surface TEM 270”. Fatigue termination in the counterclockwise
studies. Such comparison then allowed determina- direction probably occurred near the clockwise
tion of the primary fracture failure mode and the limit (between sections B and C) of the adjacent
fatigue zone circumferential limits. 240° fatigue crack. It is possible that the two
fatigue cracks joined at this location. Apparent
Primary Fracture Surface crack growth rate, as determined by localized fa-
tigue striation counting, was somewhat increased in
In order to accommodate the sample in the SEM Sections 5 and 6 near the clockwise fatigue bounda-
chamber, the aft mat side leg primary fracture was ry.
cut into sections approximately three inches in Thus, TEM results confirm the existence of two
length. The origin area, located at approximately fatigue crack# which possibly joined near the 255”
270°, was divided into two sections. The half to- region. The through-the-wall crack centered near
ward 180° was noted as number 1 and the half toward 270” was approximately 23 inches in length. A
0° was designated as number 2. Sections toward the second fatigue crack, 50 to 70 percent through the
180° position, fro~ number 1, were subsequently wall over an appreciable distance, was approximate-
designated O, C, B, A, D, E, F, etc.; those toward ly 14 inches in length and centered near 240”. A
o“, from number 2, were numbered 3, 4, 5, 6, etc. map of the identified fatigue region is shown in
A photograph showing the cut and referenced sec- Figure 16.
tions is shown in Figure 12.
A sample from the primary fracture, cut near
Initially, extensive electron microscopic the 90° location, was also studied using the TEM.
studies were conducted on the primary origin left Results clearly confirmed the fracture mode at that
side, Section 1. For reference purposes, Figure location as quasi-cleavage.
13 shows a closeup of the Section 1 fracture sur-
face and locations where detailed TEM studies were
performed. Evidence from this series of studies on SUMMARY AND CONCLUSION
Section 1 indicate that the fracture mode waa fa-
tigue and that a through-the-wall fatigue crack did Results of this investigation permit develop-
exist prior to final collapse. Because of heavy ment of a most probable sequence of events prior to
oxide and surface deposits, scanning electron the collapse of Ranger I on May 10, 1979.
microscopic studies failed to reveal topography
which could unequivocally be identified as fatigue. Magnetic particle and ultrasonic testing,
However, transmission electron microscopy was able metallography, drop weight testing, and leg/s~if-
to conclusively establish the fatigue failure mode fener joint tensile test results all indicate the
(see Figures 14 and 15). presence of pre-existing surface cracks. These
cracks, all in the. region of leg/stiffener and
Transmission electron microscopy was conducted stiffener/mat weld runout, were found to be coinci-
on various circumferential sections in order to de- dent with leg/mat bulkhead and structural joints.
fine the fatigue crack growth limits. Results from This suggests that considerable rigidity at these
selected sections are summarized in Table 2. locations resulted in load and stress concentra-
tion. Stress fluctuations, resulting from wave
The extent of identified fatigue crack growth action and rig operation, appear to be of suffi-
in the counterclockwise direction was found to oc- cient magnitude and nature to initiate and propa-
cur in Section F. TEM results and visual examina- gate aft leg fatigue cracks near the 0°, 90°, 180°,
tion confirmed that this fatigue crack existed ap- and 270° regions. Crack indications were also ob-
proximately 50 to 70 percent through the wall. served near these same locations on the forward
Termination of this crack in the clockwise direc- legs. Asymmetric loading and/or increased stress
tion probably occurred between sections B and C. intensity range resulted in greater fatigue crack
This crack, having a length of approximately 14 propagation n~ar the 270” rather than 90° region.
TABLE 1

TENSILE TEST RESULTS

Leg Yield Tensile Reduction


Elevation SCrength Strength Elongation of Area
Specimen
(ft.) (ksi) (ksi) (% in 2 in.) (%)
Identification

2671 12 38.4 71.2 31.0 61.0

2671 12 38.6 71.1 32.0 61.0

2681 20 34.9 66.2 33.0 61.0

2681 20 35.2 66.0 35.0 59.5

2682 39 39.6 68.5 33.0 59.5

39 40.2 69.0 31.5 61.0


2682
2683 88 38.4 75.0 29.0 58.0

88 38.6 74.6 31.0 61.5


2683

TABLE 2

SUMMARY OF TRANSMISSION ELECTRON MICROSCOPIC STUDIES PERFORMED


ON PRIMARY FRACTURE
Crack
Area General Growth
Section # Designation Location Results (cyclin) Photo #

F Center Mw Fatigue H.48


F Center Mw Fatigue H.47
E Right 314W Fatigue H.46
D Left Mw Fatigue H.45
c Center 314W Fatigue H.42
c Right IDS Quasi-Cleavage H.29
c Left IDS Quasi-Cleavage H.30
B Center 3/4w Fatigue H.43
B Right IDS Quasi-Cleavage H.31
B Right IDS Quasi-Cleavage H.32
A Center Mw Fatigue H.44
A Right Mw Fatigue 1 .2X105 H.33
o H Mw Fatigue 4.1X105 H.27
0 J 114W Fatigue H.28
1 ICX .Mw Fatigue H.18
1 2Y Mw Fatigue 1 .3X105 H.19
1 YM l14w Fatigue 2.1X105 H.20
1 3x 114W Fatigue H.21
1 3x 114W Fatigue H.22
1 YR IDS Fatigue H.23
2 R Mw Fatigue H.24
2 s 114W Fatigue H.25
2 T Mw Fatigue 3.5X105 H.26
4 Left IDS Fatigue 8.3x104 H.34
4 Left IDS Fatigue H.35
5 Left IDS Fatigue H.36
6 Left IDS Fatigue 1.3X105 H.37
6 Left IDS Fatigue 9 .OX1O4 R.38
7 Right ODS Quasi-Cleavage H.39
7 Right ODS Quasi-Cleavage H.40
8 Right Mw Quasi-Cleavage H.41

m = Mid wall IDS = I.D. surface


l/4W = 1/4 wall from ID surface ODS = O.D. surface
3/4W = 3/4 wall from ID surface
Fig. 1- Viewof mat sectionwreckagefrom the recoveryvessel.

110’–0“

94’-o“
o o—
Forward Port
\. o

270”
Aft
‘o 0“
“ 1 m
I
a.
II Hul
0
*

Forward Starboard~ -0
0
0 ‘/ 0J
Fig.3-Approximate leg locationand attitude.
Fig.2-Top schematicplanof mat showingnomenclature
and or-
ientation.
Fig.4-Lower aft leg primaryfracture.

Fig.5-Overall viewof aft leg stub.


Fig.6-Fit-up of aft leg primaryfracture.

- 9’toll”

Fig.7- Viewtowardaft showingapproximate


pre-collapse
attitude.
Fig. 8-Aft leg stuband surrounding
mat (courtesyof Andersonand Associates).

Fig.9-Scanning micrographof fatiguein low~K regionof slowbendspecimen- Fig. 10- Transmission


micrographoffatiguein low~K regionof slowbend
originalmsg.2000X. specimen— originalmsg.2080X.
Fig. 11- Secondaryfracture,S-1,showbothfatigueand fastfractureregions.

~&&&a& ,

% -&$&
>
%
/
*
—.
\
-+
&
~
../ .\

Fig. 12 -View of sectionscut for electronmicroscopy.


3

m mi
Fig. 13-Closeup viewof section1 fracturesurface. Fig. 14- Transmission
micrographof fatigueat I.D.surface,section1 –
originalmsg. 7670X.
Fig. 15 - Transmission micrograph of fatigue at quarter-wall location,
section 1 -- original mw. 7~7~X.

Fig. 16 - View of 270° region with fatj~~e zone delineated.

You might also like