REPORT No.
MARCH 1968
SSL 170
SHIP STRUCTURES LABORATORY
DELFT UNIVERSITY OF TECHNOLOGY
Background of some investigations in the Ship Structures Laboratory
The danger of brittle fracture In connection to fatigue In ships
REPORT No.
SSL 170
SHIP STRUCTURES LABORATORY
Background of sorne investigations in the Ship Structures Laboratory
The danger of brittle fracture in connection to fatigue In ships
by J. J. W. Nibbering
Background of some investigations in the Ship Structures Laboratory
The danger of brittle fracture in connection to fatigue in ships
by J. J. W. Nibbering
General.
Up to World War. Il
a peculiar situationexisted in ship structural design. Relatively little was
known about the loads working on a ship In a seaway
and about the lod.bearIng capacity of the structure.
Yet the safety of ships was great and what is more, the
safety margin was not excessive.
This sounds surprising for a branch of industry which
was often regarded as béing conservative. The cause is that ships are. supervised by Classification Societies from
the cradle to the grave. The collection, statistical treat-ment.and interpretation of all theempirical information thus obtained represents. one of the most realistic and
effective methods of research, Indeed each ship is used as a full-scale specimen In loading conditions which are
as representative as can be. This situation was quite satisfactory till the second World War when the
expe-rience and knowledge gained fell short of the mark
because riveting was rapidly replaced by welding. Nu-merous.fractures, - mostly brittle, - and serious bottom damage (corrugations) occurred.
Research immediately started, especially in the U.S. In a few years the danger of brittle fracture was reduced
considerably. Nevertheless, ships stili fractured
inciden-tally and partial brittle cràcks stili developed relatively frequently. This could not be explained by the results of tests notwithstanding the fact that full scale structures
and very wide notched 'plates. were used.
The insight that these experiments were not sufficiently
realistic has come only very gradually.
The first big steps forward were made by Greene and Wells in Great Britain, Kihara in Japan and Mylonas In the U.S. They led to the realization of the tremendoUs
damage inflicted to notched. steel by welding and to the
understanding of the. role of residual stresses. Later
Nibbering proved that the loathng conditions in the
U.S.-tests did not conform (sufficiently) to reality. He
included In his tests in the Deift Ship Structures
Labora-tory the effects of fatigue and shock loadings ánd suc-ceeded in demonstrating their importance.
About 1955 the needs for better insight became acute
when thesizeof tankers started to increase enormously.
later followed by the growth of bulk carriers. Other
revolutionary developments were the drastic reduction in the number of tlransverse bulkheads, unusual cargo
distributions and large hatch openings. This led to the
use of thick plates and higher strength steels and to the
introduction of modern welding processes. lt will be clear that no longer reliable extrapolations from the
experience acquired with ships of moderate size could
be made. Therefore in the first place research on wave-induced bending moments at sea was accelerated and intensified. This was effectuated by measuringstresses on ships at sea, by model testing and by the development of theoretical calculations as promoted by. Ben net, Lewis, Jacobs,and others. Most ofthem directed their efforts to
the estimation of long term distributions of
,wave..ben-ding loads. For some time the extremevalues were
more or less considered asa by-product of the.frequency
distributions. But in 1962 Yuille:p.ut fgrvar.i the idea
that the.estimation of the whole long .term.distribution of loads was of limited value and that researchshouId
be concentrated on the extreme values. His arguments were that wave induced loads of certain magnitudes do not occur sufficiently frequently to cause fatigue-cracks. In the extensive discussions to his R.l.N.A.-paper curious
contradictions came to light. Some people stated that shundred of fatigue-cracks have been found in ships.
while:others maintained that cracks developveryseldom
and, if so, are mostly brittle.
lt is at this stage that the Ship Structures Laboratory
started to contribute to the solution ofthe problem.
The approach of the Ship Structures Laboratory. About that time. the S.S.L. had started low cycle fatigue-tests with full-scale structural elements In the
6OO-tonsma-chine. (See section 4e).
One of the questions raised was whether it was neces-sary to test the specimens to complete fracture or only tó the formation of cracks of restricted length.
lt wassoon discovered that for the configurations tested (specimens containing severe discontinúitles) cracks did
Initiate after relatively small numbers.of cycles. On the other hand, it was found that these cracks propagated very slowly. On account of this itseemed pretty certain that small fatigue-cracks will often occur in ships. But the question remained how long may become these
cracks. Even more important Was whether the presence
of fatigue-cracks in ships constitutes a danger or only a nuisance (leakage). In connection with the former,
extrem-ely sharp notches which might be a' favourable start-ing point. for brittle fractures. The latter consideration actually determined the whole set-up ofthe full-scale. tests In the S.S.L. All specimens were pulled to rupture at low températUres after small fatigue-cracks had
de-veloped. (See section 4f). But 'first It had to be decided how long thesefatigue-cracks had to be.as crack-length
Is avery important parameter In brittle fracture
re-search. This is not only due to the dimenslona effect, but'also to thefact that small cracks generally end in welds or heat-affected zones, while larger cracks end in the adjacent plating. To achieve a good 'estimation of a realistic crack length, loading spectra of ships had
to be available.' Unfortunately, most of the stress spectra
given in-literature for various types ofships only
com-prised longitudinal wave bending stresses; although that
represents the mäln part of the.complete spectrum of strèsses in ships' structural detaIls, the difference be
tween the completespectrum and the longitudinal wave spectrum ¡s appreciable.
This was' demonstrated in 1963 by Nibbering who in-troduced corrections for slamming corrosion, tempe-rature changes, local Ioad etc. À typical 'diagram is given In figure 4 fromwhlch lt couk be deduced that
many-cracks of.a length equal to about plate thickness may be expected at the end oía ship's life. The conse-quence was that the. influenceof cracks *ith a length
from zero to late thickness had tè ,be Investigated.
On the whole It was'obvlous that YuiIle's.statement was
too optimistic. Of course this was In a way comforting
to all those who had, dedicated a largè part of their
scientific life to the estImation of the whole spectrum
of wave bending stressès Instead ofto the soleesti matlon
of extreme values. On the othe(hand,the limited value
In practiCf .f wave-bending spectra had been made
clear. lt was evidènt that statistical information about
slamming-Induced stresses and stress-alternations due to differences In water-pressure onshell and. bottom were.
particularly indispensable. The latter want-is presuma :bly for the first timò upplI'ed with the aid of the multi-channel high speed,. digital' recorder of the S.S.L. In 1966 the wave-induced stresses in the double bottom-structure ofaibulkcarrler were measured oñ 40
measur-Ing points during a North Atlantic trip (fig. 24, 25).
t) Fractures which start beròre general yielding.
The Whole experIment has been a.great succes; the circle can now be closed not only m'a qualitative way.. but alsó quantitatively at least for ships vi comparable size and type. It is hoped that 'other research institutes
will start to Investigate different types of shIps ès soon
as possible in the same way;
After these optimistIc statements there Is stIll room for
some. reserve. lt is not absolutely certain that'the already highly realistic experiments with fatique-cracked welded
structures at low temperatures are fully representative
of the ways fractures devélop' In ships. Firstly, of course,
fractures can very well start at other minor .defects'Iike weld-flaws and hèataffected-zone cracks. Oher causes may 'bé large deformations at critical places induced'
during fabrication, which giye the so-called Mylonas'
effect. . .
But ali these.cases, hówever important they maybe for brittle, fracture.research,. have little to do with the.
cyclic' loads.áctlng on ships. .And .it was espeÌaliy the roleof these cycllè.Iòads with respect, to fracture-whith hadbeen made theoblect ostudy in thé S.SL.
Return-in to the ¿ase, of fatigue-çracks, it was Intrigulngto''
'see that real low-stress fractures..') had not: 'ocured in the full-scale specimens tested, although the tempera
turès used were very low Ì0°C)' for the mâtriai
concerned (fìg, 19a) The cracks only caused tremendousreduction in ductility. Two condusioñs were :pbsbli:
fatigue-cracks are nôt,a prlmaiy causeoffractûres'In
ships;
the testing condItions have still not been sufficiently
representative for sea-conditions.
The firstcônclùslon prove'd 'to be, dóubtful after añother
investigation of the S.S.L. in which thé occurrence of a
large brittle fracture In a ship,coùld be attrIbuted to
fAtigue '(fig. 5). The fracture had started at a small
fatigue-crack. '
The quèstion now remalned.howbrittle fractures could
InItIate 'at fatigue-cracks. wlthòut prior - more than local - yielding. The only remaining possibility was
,Impact. A p,rellmlnary'lnve.stigátlon wascarried out with small plates of full thickness contaIning notdies
of various degrees of sharpness In welds or close to thm
(fig. 6). It emerged that under staic.loadlng the danger
Fig. 4. Actual wave-bending stresses of the M.S. "Canada" (lower line) corrected for slamming, corrosion etc. Thestress-line thus obtalnèd is used for estimating the fatigue-life of ship structural details with the aid of representative "Wähler" curves. [39].
DOUBLE AMALITUDE OF STRESS(1) KG/mm2 35 34 33 32 31 -30 30
29-
2827 26 -25O INITIAL CRACK (25,,,n) COMPLETEMAI.URE
2L-
23
-- HORIZONLLY WELDED MACHINED
-22 -
'
---m---
-21-
-1 20 20',
19 - b' .?e .-1ÔI
18 L . L -4'..ÇF'S 4ODF O T2-CDNNECIION HORIZONTALLY WELDED:UNMACHINED
17 - . . .
-16 . 15 15 0 UNMACHINED -. 14 - 1f\
\
\
WELDED IN DWFICULT -13 -.-"\
. 't . .'
--J--CJ
)--
12 -NM AC HINE D 11 -- '. . N, WITH SMALL WELD -. ORIGMAI. 1CONNECTIDM __________ FAULTS SEE 115.10 N, -______________
oT
4 -
. . . .+
CLJMULATIYE FRED. ,. -I I 10 io2 io6 108 N. 10000 .Lbs/O' 20000 Lbs/o"from mechanically-made notches and it is much smaller
than that of notches due to welding.
But under impact-loading the fatigue-cracks were by far
the most dangerous. The main reason for this seemed to be the high sharpness of these cracks as compared
with mechanically-made notches.
lt seemed worthwhile also to investigate this
phenom-enon in the fullscaie structures. lt was equally im-portant to Investigate whether comparable impact
FIg. 5. Ship fracture started at small fatigue-crack at about +4 oc [47J.
BILGE KEEL
loads can occur in ships. Apart from explosions the only
possible source of high-speed stressing Is slamming.
Whether slamming really can cause brittle fractures at appreciable distances from the bow is not known
be-cause up to now slamming-induced stresses have mainly
been measured with low-frequency apparatus which only allows two-knot vibrations to be recorded. Higher knot vibrations and high-velocity stress-waves are
nor-mally filtered out.
8 6 2 o 80 60 40 20
RESULTS OF CHARPV-V IMPACT TEST
o MATERIAL OF LONG., AND TRANSVERSE BUTT-WELD---.. D.9
IIp1i
/
Dg-
/
MATERIAL APD TRANSVERSE OF LONG. //
TID
-40 -20 0 +20 +40 TEST TEMPERATURE ICENTIGRADE)+60 BILGE PLATING
RESULTS OF NOT. Temp. Isotherm Arrest Tamp.
BRITTLE FRAC- DROP-WEIGHT LE.?...+30C E.? .35C TURE AND RO8ERT54 (E.8...t30C E.8 +40C
TESTS I D.9...+20C 0.9 +35C TRANSVERSE BUTT- WELD -40 -20 0 + 20 +40 +60 60 50 40 30 .0 20 15 10
Fig. 6. Illustration of danger of fatiguecracks under impact-loading as compared to that for other types of notches. Note especially the difference with,static loading. [44], [48].
In order to obtai nan idea about this, the S.S.L. equipped
a ship with a number of strain gauges which were
connected to high frequency recorders. Therebefore Impact-tests with full-scale fatigue-crackéd specimens
were carried o:UJt. The procedure.conslsted in subjecting
the specimens to a small static load at low tmperature
and submitting them to light impact loads; when the
specimens did not fracture the static load was increased.
The result of the investigation is thought to be of great
o .0 E 5tatIc: -35 C 19 Tons .Fatigue crack Width of shear lip 0.2mm '4--Brittle part
importance. Two out of four specImens fractured at a
load which was nearly one third of yield load. The other
two fractured at about 60% of yield load. lt should
be realized that this signifies a high probability of frac-ture for any ship coñtaining fatigue-cracks and liable tò
slamming and navigating at a low enough
tempera-ture.
In many cases temperatures between 0°C and +10°C will be low enough.
MODIF (WELLS NOTCH")
t
3mm SAWCUT + 1mm PRESSED NOTCH lED DROP-WEIGHT ('PELLINI NOTCH)I
NOTCHED BRITTLE WELD TEST ('FATIGUE NOTCH") ¿Weight: 1 22kg .2mm SAWCUT+ 4mm FATIGUE CRACK. -20° -10° 0° +1O +20jN.D.T.
i
1'
4;'
..---..
./
-
- -
uuPIP1.. .1'-.
I..ØNo sheartip Width of shear
tip: 0.2mm
-60° - 50° -40° -30°
oc
Light Impact: - 20 C Heavy impact: + 20 eC
Low-cycle fatigue tèstswithfull-scale.
structur-al specirnefls made of Mjld Steel and High
Tensile Steél ' ' .
The need to dIspose of fatigue-strength data on
full-scale structures has often been stressed, both by struc-.
turai engineers and by fatigue specialists. The first like
to base their designs on fatigue-data valid for cases which
are as realistic as possible. while the second need test cases for checking theories based on results obtained
'with simple test bars.
For the present investigation, specimens representing
stiffened plate structu reswere used. They were of a
con-servative dèslgn in order. to avoid results too optimistic
for practice. Axial loading was applied of a type in
between' pure alternating and repeating. (Pmin/Pmax=
-'/2). With the aid of a large number of strain gauges
on thespecimens, the distribution of the elastic stresses and plastic strains were determined. The fatigue-results
could thus be given as a function of the total load and
of the load in each part of the-specimens separately and
also as a function of the stresses (or strains) in the
im-mediate vicinity f the origin of the cracks. This 'was
done for 3 different crack-areas. Comparisons could be
made with specime of higher strength steel and. of
silghtly different'design. It clearly emerged that, for
the type of loadIng concerned, no distinct advantage exists for the stronger steel. (FIg. 17.)
Interesting observations could be made with regard to crack propagation. In tje specimens containing severe discontinuities, a crack starts early In the test bu prop-. agates relatively slowly. In continuous speèimens the reverse Is true. Due to thls. the latter corresponds to a "safe-life" concept and the former to a "fail-safe" one; This might be a new and Important consideration when
deciding which concept should be adhered to.
Literature:
J. J. W. Nlbbering [45] with J. van Lint E3].
Brittle fracture tests with full-scale specimens
damaged by fatigue
Twenty full-scale mild steel specimens used for the fa-tigue.tests described above were pulled to rupture at low temperatures (fig. 2).
They contained small low-cycle fatigue-cracks mainly situated-in heat-affected zones fig. 18). The Charpy 15
ft lb temperature of the material was 10 .°C. the
N.D.T. 8 °C and the Isotherm Robertson
arrest-témperature '+17 °C.
The nomiñal fractures-stress of the specimens was
gen-erally ònly .75% of yield point at temperatures
be-tween .10e and 40°C. The correspondIng strain
over the wholelength of the specimens was about 0,15%.I
i
gI
Fig. 17. Fatigue-strength of brackèt > and bottomdetalls for 100 mm2 crack-area shown as a function of nominaland local strains; [43], (45]. 2000 8000 4 o EE(AVO) t-.-
\
IBOITOMPLATES)-\ ISt 52) t I ttill
Io 3000-o N -6 (1) .14 N 10' EE lAVO 1/2)-\
EE(AVO 1/2)\
N
N
5. I tt litt
t tlt tít
toi 6000 5000 £ \. \..-.iBOTTOMFtATES-St $2) EE(23\\"
N
t lt I i i i i i L_t III i i t t t lit io NORDINATES ARE TOTAL RANGES OF CYCLIC STRAIN.
CURVES 'APPLY TO IOOnrn2 ÇRACK AREA,
IA TYPE - 24-TYPE t t
ttttti
I itttttt
03 i I t t . I0 N (E (AVG 6/7)fl
EEtAVG6/lt7
-s.. tBOT14S-2At\.
i ¡IBOTTOMS-Si (BJUCI( 52) IS-IA) -t ttttitt
tBOTTOMR-1A s... (BOACKETS-2A t t ttttttt
IA2F
066-TWO SPECIMENS WITHOUT FATIGUE CRACKS
:.''ï
- -... TEST TEMPERATURE (°F)
-50 -40 -30 -20 -lO
-50 -40 -30 -20
TEST TEMPERATURE (CENTIGRADE)
(Fig. 19a.) The cause of this small degree of plastic de-formabi!ity was simple. The test section contained too little material because a part of it was ineffective as a
result of bending (fig. 19b). Below 40 oc a transition
to low-stress fractures was apparent. This transition did
not correlate with results obtained with usual
accept-ancetests. In an additional investigation with y. d.
Veen-type specimens containing fatigue-cracks It could be proved that this was due to the pure.static character of
thefull-scale tests (fig .6). In static bending thesespeci mens
fractured at 70 oc at a load as high as 22 tons. In con-tract these specimens could be fractured by a
hand-driven hammer at 10°c. This enormous difference
proved to be mainly due to the extreme sharpness of the fatigue-cracks and not to damage at the notch root.
*10 .20 -10 15 ft. b. CHARPY-V N DT. ? CHARPY-V vd.VEEN 100% CRISTALLINE
Q
AVERAGE STRAIN AT Z Z IN REGION A+30 *40 *50 *60 .70
.10 --- .20
ROBERT ARREST (ISOTHERM TEST
Fig. 19a. Strain at fracturefor differentgauge.lengthsofspeci-
-mens containing fatigue-cracks [44], [45].
Fig. 19b. Illustration of inefficient use of bracket material.
+ ADRACKET
rl__
b
PBCITOM fxABo1Tcl STRAIGHT SYMBOLS I .. . CUR VEO SYMBOLS Z-Z Z-Z Z -Z - D1.02r3 o
o
Nibbering, J. J. W.
Enige beschouwingen over breukverschlJnselen, sterktecriteria en constructieve vormgeving. Schip en Werf 26 (1959) no 12.
Vermoeiing van scheepsconstructies
Schip en Werf 30 (1963) no.10.
Fatigue of ship structures.
Report NSS-TNO no. 55S (1963) and l.S.P 10(1963) no. 109.
The influence of structural design on the brittle
fracture strength of ships
Proceeding Intern. Ship Structures Congress Delft 20.-24 July 1964. Chapter I of Report Comm. 3d.
Die Materialprüfung im Schiffbau.
In: Materialprüfungs- und Versuchswesen in der
Schweiz und im Ausland. Neugut 1965.
Sôme practical conclusions from strenght research
In shipbuilding.
Holland Shlpbuiidlng 14 (1.965) no. 1.
Low-cycle fatigue of steel structures. (With J. van Lint).
Report NSS-TNOno. 82S (1966) and l.S.P. 13. Sept. 1966 and Lastechniek 32, Aug. 1966.
Brittle fracture. of full scale structures damaged by fatigue. (With J. van Lint and R. T. van Leeuwen).
Report NSS-T.NO no. 85S (1966) and LS.P. 13, Nov. 1966.
An experimental investigation in the field of low-cycle fatigue and brittle fracture of ship structural
components.
Trans. R.I.NA. vol. 109, Jan. 1967.
De cost en de boot.
Lastechniek 32, Dec. 1966.
Partial fracture of bilge- and bottomplating of an oil-tanker.
Doc. l.l.S./l.l.W. XIII-409-65; Welding In the World
I 968.
Brosse breuk van vermoeide constructies.